Molecular & Biochemical Parasitology 144 (2005) 109–113

Short communication

Further definition of PfEMP-1 DBL-1� domains mediating rosettingadhesion ofPlasmodium falciparum�

Clare Russella, Odile Mercereau-Puijalonb, Cecile Le Scanfb,Michael Stewardc,1, David E. Arnota,∗

a Institute of Immunology and Infection Research, Division of Biological Sciences, Ashworth Laboratory,University of Edinburgh, King’s Buildings, West Mains Road, Edinburgh EH9 3JT, UK

b Parasite Molecular Immunology Unit, CNRS URA 2851, Pasteur Institute, Paris, Francec Adprotech, Little Chesterford Research Park, Saffron Walden, Essex, UK

Received 11 March 2005; received in revised form 22 June 2005; accepted 29 June 2005Available online 11 August 2005

Keywords: PfEMP-1; Rossetting;Plasmodium falciparum

he-ntsni-ran-

ry-ics

llscids

ood

s to

nce

ro-

-ph-

The adhesion ofPlasmodium falciparum binding pro-teins to host receptors is vulnerable to inhibition by anti-bodies and an interesting target for malaria vaccine devel-opment although the binding reaction is not yet under-stood in the requisite molecular detail[1,2]. Receptor-binding sequences on merozoite erythrocyte binding proteins(EBPs), and on infected erythrocyte surface adhesion recep-tors such as PfEMP-1 antigens consist of related, cysteine-rich domains. These are termed Duffy-Binding Like (DBL)domains because of their similarity in sequence toP. vivaxandP. knowlesi proteins that bind a chymotrypsin-sensitiveepitope on the Duffy blood group surface antigen[3]. Malariaparasites thus use an evolutionarily conserved protein domainto attach themselves to diverse host receptors.

Little is known about the specific requirements forPfEMP1-mediated attachment and in vitro assays to studythe relationship between structures and functions are needed.A notable example of a PfEMP-1-mediated adhesion interac-tion is ‘rosetting’, whereP. falciparum infected erythrocytesstick to uninfected red blood cells by binding receptors suchas complement-receptor 1 (CR1)[4] and the trisaccharideglycan of blood group A antigen[5].

The evidence that the DBL-related sequences form adsion domains for such functions derives from experimeinvolving the heterologous expression of truncated miproteins and measurement of their substrate binding. Tsient expression of truncated DBL1� domains on the surfaceof COS-7 cells leads to the acquisition of an uninfected ethrocyte binding phenotype by the COS cells that mimthe rosetting adhesion seen in some patient-derivedP. falci-parum isolates. Expressing PfEMP-1 fragments in COS ceto carry out adhesion assays established that 309 amino acontaining 14 cysteines from the R29var1 DBL1� containedthe binding domain that mediated adherence to red blcells with the CR1+ phenotype[4,6]. Further mapping of thebinding region localised the essential binding sequencewithin 233 amino acids of the R29var1 DBL1� domain[7].DBL1� domains have also been shown to mediate adhereto endothelial cell ligands such as heparan sulfate[8].

We have cloned fragments from a distinct PfEMP-1 ptein from a differentP. falciparum line showing rosettingadhesion, the serologically definedvar O variant derived fromthe Palo Alto isolate[9]. The var O gene has been identified as the major PfEMP-1 transcript expressed in perieral blood parasites from infected, splenectomisedSaimiri

� Note: Sequences described in this paper have been deposited inEMBL/GenBank/DDBJ under the accession DQ027014.

Cam-b

sciureus monkeys[10]. RNA was extracted from high para-sitaemia infections (>20% with 70–90% mature schizonts),and RT-PCR was carried out using PfEMP-1 (UNIEBP)p lye redt fer-

0 d.d

∗ Corresponding author. Tel. +44 131 650 8656.E-mail address: d.e.arnot@ed.ac.uk (D.E. Arnot).

1 Present address: Domantis Ltd., 315 Cambridge Science Park,ridge, UK.

166-6851/$ – see front matter © 2005 Elsevier B.V. All rights reserveoi:10.1016/j.molbiopara.2005.06.009

rimers[11]. A PCR product was identified that is highnriched invar O parasite RNA preparations, as compa

o RNA from parasites in an infection induced by a dif

110 C. Russell et al. / Molecular & Biochemical Parasitology 144 (2005) 109–113

ent Palo Alto-derived antigenic variant, the var R parasites[8,9]. Sequencing indicated that this variant-specific tran-script encoded a PfEMP-1 DBL�-type domain[9]. RT-PCRusing the broad specificity�AF/�BR primers[12] was sub-sequently used to amplify DBL1� domains from this RNA.Sequencing of 49 cDNA clones indicated that the majority ofthese transcripts (60%) had the same sequence, again derivedfrom thevar O gene. Primers based on these two fragmentshave been used to sequence the entire gene (manuscript inpreparation).

To test whether thisvar O DBL1� domain encodes theRBC binding phenotype, COS cell expression constructswere made and their capacity to bind uninfected CR1+,blood group O erythrocytes was assayed. To overcome mam-

malian cell expression problems caused by the codon usagebias of P. falciparum genes a codon-optimised version ofthis segment of thevar O DBL1� gene was made. A + Trich P. falciparum codons that are relatively rare in mam-malian or insect cells (frequency <15/1000) were mutatedto a more G + C rich synonym. PotentialN-glycosylationsites were removed to minimizeN-glycosylation, rareor absent in P. falciparum [13], which could inter-fere with protein folding. PredictedN-glycosylation sites(NXS/TN) were identified using the ‘NetNGlyc’ program(http://www.cbs.dtu.dk/services/NetNGlc/). When modify-ing these sites the serine or threonine position was replacedwith alanine. Restriction sites that could hinder cloningmanipulations were also removed. TheHomo sapiens codon

FpapttcR

ig. 1. COS-7/erythrocyte binding assay. Cells were viewed under phase-Display-transfected cells showing COS cells densely covered with closely asnd in the spaces between cells. (An average of 45% of these cells strongly flositive control pRE4-P. vivax DBL Region II-pRE4-transfected cells. (An aver

hese positives bound erythrocytes.), (C) ‘empty’ pDisplay-transfected cells.han 2% of these bound erythrocytes.) (D) ‘Mock’ transfected cells. (An aveells bound erythrocytes.) (E) un-transfected cells. (An average of 1% of cellsesults are the average of three transient transfection experiments.

contrast microscopy at 40× magnification. (A) codon optimisedvar O DBL1�-sociated erythrocytes (arrow). A few erythrocytes have settled on thecell monolayeruorescenced after transfection and 30% of these positives bound erythrocytes.), (B)age of 25% of these cells strongly fluorescenced after transfection and 15% of(An average of 25% of these cells had positive fluorescence after transfection, lessrage of 1% of these cells fluorescenced after transfection and less than 1% of all

fluorescenced after transfection and less than 1% of all cells bound erythrocytes).

C. Russell et al. / Molecular & Biochemical Parasitology 144 (2005) 109–113 111

usage table was taken to be representative of the COS-7 greenmonkey genome usage (http://www.kasusa.or.jp/codon).

The codon-optimised gene was assembled using anunpublished methodology, based on earlier methods[14,15].A set of 46, 5′ phosphorylated 40mers were designed tocover the whole of the 906 bp gene fragment with 20 basepair sequence overlaps in complementary strands. Oligonu-cleotides were synthesised by MWG Biotech (0.01�molsynthesis, HPLC purified) and assembled in a single reactionusing bothPfu polymerase andTaq ligase. The reaction prod-uct was then amplified in a conventional PCR reaction withPfu polymerase and purified by agarose gel electrophore-sis. Correctly sized product was extracted and re-amplifiedwith Taq polymerase using terminal primers withBam HIandBg1 II sites to facilitate cloning into the pCR®4-TOPOvector. Plasmids obtained after bacterial transformation withthe cloned fragment were sequenced to confirm productionof the desired sequence.

The pDisplay (InvitrogenTM) vector was used for theexpression of the codon-optimised domain. COS-7 cells weretransfected with appropriate controls and after 48 h cellswere harvested and dissolved in SDS sample buffer. Sam-ples of cell lysates and culture supernatants were separatedby SDS-PAGE and Western blotted. Rabbit anti-influenzahaemaglutenin antibody was used to detect epitope-tagged

protein. An intense band of the expected size was seen onlyin the cell lysate of the optimisedvar O DBL1�–pDisplaytransfected COS cells. Surface expression on the COS cellswas verified by fluorescence microscopy using the rabbitpolyclonal anti-haemaglutenin antibody on fixed COS cells.Background was negligible and untransfected cells (aroundhalf of the cultures) did not fluoresce.

To determine if the COS cell-expressedvar O DBL1� sub-domain mediated binding of uninfected red blood cells adhe-sion assays were carried out. Un-transfected cells, ‘mock’transfected cells, cells transfected with the pDisplay vec-tor without insert and cells transfected with aP. vivax DBLRegion II gene fragment[7] in the pRE4 expression vectorwere set up in parallel as controls. Thevar O DLB1� stronglybound multiple uninfected RBCs (Fig. 1C1) and this was alsoobserved in the positive control transfections with the DBLRegion II gene fragment (Fig. 1 C2). No binding was seenin any of the controls. Binding tovar O DBL1� expressingcells is sufficiently strong that the erythrocytes are lined up ontheir sides at the surface of the transfected COS-7 cell (indi-cated by arrow inFig. 1C1). Erythrocytes settling on the cellmonolayer are easily distinguished from densely packed sur-face binding blood cells. Invar O DBL1� expressing cells,dense surface binding of erythrocytes was seen in up to 50%of the COS cells.

Fot[u

ig. 2. Comparison of the expressedvar O sequence with two Pf EMP1 DBL 1� af thevar O DBL-1� sequence is compared to the two otherP. falciparum sequenc

o bind either erythrocyte complement receptor 1 (R29) or heparan sulfate (F17], the R29 clone is derived from the It A4 clone[6] and the FCR3S1.2 clone ising the CLUSTALW program. Residues that are conserved in all three sequ

mino acid sequences of known binding specificities. The protein sequencees that have been shown in in vitro assays with recombinant gene fragmentsCR3S1.2). Thevar O clone is derived from the Palo Alto strain ofP. falciparum

s of the FCR 3 lineage[18]. The multiple sequence alignment was carried outences are shaded black, those conserved in at least two sequences are shaded grey.

112 C. Russell et al. / Molecular & Biochemical Parasitology 144 (2005) 109–113

The expression optimisedvar O erythrocyte bindingsequence is shown inFig. 2 in comparison with the ery-throcyte binding sequences of the complement receptor 1-binding R29 P. falciparum line and the heparan sulfate-binding FCR3S1.2P. falciparum line. The figure shows thatrelative positions of all cysteine residues in the DBL1� seg-ments are exactly maintained betweenvar O and R29, but notbetween these sequences and the FCR3S1.2 sequence.varO and R29 are identical at 68% of comparable amino acidpositions (88/256 positions vary). These sequences are thussignificantly more closely related to each other than eitheris to the FCR3S1.2 heparan-binding sequence (39% aminoacid identity with bothvar O and R29). Thevar O constructused here is 30 amino acids shorter at each end than the R29construct used by Rowe et al.[4] but 50 amino acids largerthan the R29 construct made by Mayor et al.[7]. The dif-ferences between the R29 andvar O-derived sequences havea marked tendency to occur at presumably less selectivelyconstrained, highly variable positions. Substitutions at suchpositions include charge and polarity changes as well as con-servative replacements.

All three P. falciparum PfEMP-1 DBL 1� sequencesare more distantly related to theP. vivax Duffy antigenerythrocyte binding sequence, whose capacity to toleratevariation while retaining erythrocyte-binding capacity hasbeen systematically investigated. Region II of the Duffya riticale lace-mT inoam siduei otiff herf itutedi teb

tedbr thea theP her tk n tor tor-1[ vea thehr bseto

A

anda s on

the manuscript. Alison Creasey, Mercy Sowa and Lisa Shar-ling are thanked for their assistance and advice. The workwas supported by a grant from the Commission of the Euro-pean Union (QLK2-CT-2002-01197 EUROMALVAC2) toD.E.A. and O.M.-P. and by a BBSRC-CASE studentshipto C.R.

References

[1] Smith JD, Craig AG, Kreik N, et al. Identification of aPlasmodiumfalciparum intercellular adhesion molecule 1 binding domain: a par-asite trait implicated in cerebral malaria. Proc Natl Acad Sci USA2000;97:1766–71.

[2] Chitnis CE. Molecular insights into receptors used for erythro-cyte invasion by malaria parasites. Curr Opin Hematol 2001;8:85–91.

[3] Wertheimer SP, Barnwell JW.Plasmodium vivax interaction with thehuman Duffy blood group glycoprotein: identification of a parasitereceptor-like protein. Exp Parasitol 1989;69:340–50.

[4] Rowe JA, Moulds JM, Newbold CI, Miller LH.P. falciparum roset-ting mediated by a parasite variant erythrocyte membrane proteinand complement-receptor 1. Nature 1997;388:292–5.

[5] Barragan A, Kremsner PG, Wahlgren M, Carlson J. Blood group Aantigen is a co-receptor inPlasmodium falciparum rosetting. InfectImmun 2000;68:2971–5.

[6] Rowe JA, Rogerson SJ, Raza A, et al. Mapping of the region ofcomplement receptor (CR) 1 required forPlasmodium falciparumrosetting and demonstration of the importance of CR1 in resetting

uesinlood

reng of

reau-

andJ

[ jalonse ton

[ n ofnti-

Cell

[ -m

[ auseal

[ in-mbers

[ M,ficientEng

ntigen binding sequence has been defined as the crythrocyte-binding region, using systematic alanine repents of surface exposed amino acids in this region[16].he P. vivax Region II corresponds to around 100 amcids in the central region of theP. falciparum align-ents shown, starting at a highly conserved cysteine re

mmediately before the characteristic LARSFADIG mound in almost all PfEMP-1 proteins. Surprisingly, ratew (20/77) of the surface-accessible residues substn the P. vivax Region II proved essential for erythrocyinding.

Plasmodium rosetting adhesion can clearly be mediay parasite adhesion to different red cell ligands[4–7]. Cur-ent evidence indicates that variation in over a third ofmino acid positions in the central binding region offEMP-1 DBL1� domain is compatible with adhesion. T

eceptor binding specificity of thevar O parasites is not yenown although monoclonal antibodies previously showeverse rosetting parasites by binding complement recep6] do not disruptvar O rosettes, indicating that it may hadifferent binding specificity to the R29 parasites. Giveneterogeneity in red cell ligands employed byP. falciparum itemains uncertain if a specific rosetting motif defines a suf PfEMP-1 DBL1� domains.

cknowledgments

We thank Alex Rowe and Ahmed Raza for helpdvice on COS cell expression and for their comment

in field isolates. J Immunol 2000;165:6341–6.[7] Mayor A, Bir N, Sawhney R, et al. Receptor-binding resid

lie in central regions of Duffy-binding-like domains involvedred cell invasion and cytoadherance by malaria parasites. B2005;105:2557–63.

[8] Vogt AM, Barragan A, Chen Q, Kironde F, Spillmann D, WahlgM. Heparan sulfate on endothelial cells mediates the bindinPlasmodium falciparum-infected erythrocytes via the DBL1� domainof PfEMP-1. Blood 2003;101:2405–11.

[9] Fandeur T, Le Scanf C, Bonnemains B, Slomianny C, MercePuijalon O. Immune pressure selects forPlasmodium falciparumparasites presenting distinct red blood cell surface antigensinducing strain-specific protection inSaimiri sciureus monkeys.Exp Med 1995;181:283–95.

10] Le Scanf C, Fandeur T, Bonnefoy S, Guillote M, Mercereau-PuiO. Novel target antigens of the variant-specific immune responPlasmodium falciparum identified by differential screening of aexpression library. Infect Immun 1999;67:64–73.

11] Smith JD, Chitnis CE, Craig AG, et al. Switches in expressioPlasmodium falciparum var genes correlate with changes in agenic and cytoadherent phenotypes of infected erythrocytes.1995;82:101–10.

12] Taylor HM, Keyes SA, Harris D, et al. A study ofvar gene transcription in vitro using universalvar gene primers. Mol BiocheParasitol 2000;105:13–23.

13] Dieckmann-Schuppert A, Bender S, Odenthal-Schnittler M, BE, Schwarz RT. Apparent lack ofN-glycosylation in the asexuintra-erythrocytic stage ofPlasmodium falciparum. Eur J Biochem1992;205:815–25.

14] Stemmer WP, Crameri A, Ha KD, Brennan TM, Heyneker HL. Sgle step assembly of a gene and entire plasmid from large nuof oligodeoxyribonucleotides. Gene 1995;164:49–53.

15] Withers-Martinez C, Carpenter EP, Hackett F, Ely B, SajidGrainger M, Blackman MJ. PCR-based gene synthesis as an efapproach for expression of the A + T-rich malaria genome. Prot1999;12:1113–20.

C. Russell et al. / Molecular & Biochemical Parasitology 144 (2005) 109–113 113

[16] Van Buskirk KM, Sevova E, Adams JH. Conserved residues inthe Plasmodium vivax Duffy-binding protein ligand are criticalfor erythrocyte receptor recognition. Proc Natl Acad Sci USA2004;101:15754–9.

[17] Fandeur T, Bonnefoy S, Mercereau-Puijalon O. In vivo and in vitroderived Palo Alto lines are genetically unrelated. Mol Biochem Par-asitol 1991;47:167–78.

[18] Chen Q, Heddini A, Barragan A, Fernandez V, Pearce SFA,Wahlgren M. The semi-conserved head structure ofPlasmod-ium falciparum erythrocyte membrane protein 1 mediates bindingto multiple independent host receptors. J Exp Med 1999;192:1–10.