Structural comparison of apical membrane antigen 1 orthologues and paralogues in apicomplexan parasites

Molecular & Biochemical Parasitology 144 (2005) 55–67

Structural comparison of apical membrane antigen 1 orthologues and paralogues in apicomplexan parasites Marie-Laure Chesne-Seck a , Juan Carlos Pizarro a , Brigitte Vulliez-Le Normand a , Christine R. Collins b , Michael J. Blackman b , Bart W. Faber c , Edmond J. Remarque c , Clemens H.M. Kocken c , Alan W. Thomas c , Graham A. Bentley a,∗ b

a Unit´ e d’Immunologie Structurale, CNRS URA 2185, Institut Pasteur, 25 rue du Dr. Roux, 75724 Paris, France Division of Parasitology, National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 1AA, UK c Department of Parasitology, Biomedical Primate Research Centre, 2280 GH Rijswijk, The Netherlands

Received 4 May 2005; received in revised form 25 July 2005; accepted 25 July 2005 Available online 16 August 2005

Abstract Apical membrane antigen 1 (AMA1) is a membrane protein present in Plasmodium species and is probably common to all apicomplexan parasites. The recent crystal structure of the complete ectoplasmic region of AMA1 from Plasmodium vivax has shown that it comprises three structural domains and that the first two domains are based on the PAN folding motif. Here, we discuss the consequences of this analysis for the three-dimensional structure of AMA1 from other Plasmodium species and other apicomplexan parasites, and for the Plasmodium paralogue MAEBL. Many polar and apolar interactions observed in the PvAMA1 crystal structure are made by residues that are invariant or highly conserved throughout all Plasmodium orthologues; a subgroup of these residues is also present in other apicomplexan orthologues and in MAEBL. These interactions presumably play a key role in defining the protein fold. Previous studies have shown that the ectoplasmic region of AMA1 must be cleaved from the parasite surface for host-cell invasion to proceed. The cleavage site in the crystal structure is not readily accessible to proteases and we discuss possible consequences of this observation. The three-dimensional distribution of polymorphic sites in PfAMA1 shows that these are all on the surface and that their positions are significantly biased to one side of the ectoplasmic region. Of particular note, a flexible segment in domain II, comprising about 40 residues and devoid of polymorphism, carries an epitope recognized by an invasion-inhibitory monoclonal antibody and a T-cell epitope implicated in the human immune response to AMA1. © 2005 Elsevier B.V. All rights reserved. Keywords: Apical membrane antigen 1; MAEBL; Structural homology; Polymorphism; Proteolytic processing

1. Introduction Apical membrane antigen 1 is a malarial surface protein found in all characterized Plasmodium species [1] and orthologues exist in at least two other apicomplexan parasites, Toxoplasma gondii [2,3] and Babesia bovis [4]. AMA1 is stored in the microneme organelles immediately after synthesis and is transported to the parasite surface just prior to, or during, host-cell invasion. It comprises an N-terminal ∗

Corresponding author. Tel.: +33 1 45 68 86 10; fax: +33 1 40 61 30 74. E-mail address: [email protected] (G.A. Bentley).

0166-6851/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.molbiopara.2005.07.007

ectoplasmic region, a single transmembrane segment and a small cytoplasmic domain. Sixteen invariant Cys residues are encoded in the ectoplasmic region of all characterized Plasmodial AMA1 genes. AMA1 is currently being developed as a malaria vaccine [5,6]. The precise function of AMA1 is not known but many lines of evidence point to a direct role in host cell invasion. Monovalent Fab fragments derived from invasion inhibitory monoclonal anti-AMA1 antibodies can also inhibit erythrocyte invasion, implying direct interference with AMA1 function rather than cross-linking of the antigen or immune clearance of the parasite as the mechanism of protection [7]. In COS-7

56

M.-L. Chesne-Seck et al. / Molecular & Biochemical Parasitology 144 (2005) 55–67

cells expressing different recombinant ectoplasmic domain combinations of AMA1 from P. yoelii, a construct comprising domains I and II was able to specifically bind rodent erythrocytes [8]; however, others have failed to demonstrate erythrocyte binding by either cell surface-expressed, soluble recombinant, or naturally processed AMA1 from P. falciparum [9,10] or P. vivax (C. Kocken, Unpublished results). In vitro studies with an invasion-inhibitory monoclonal antibody suggest that AMA1 is probably involved in the invasion phase after the initial non-specific reversible attachment of the merozoite to the erythrocyte [11]. These studies implicate AMA1 in some way in the formation of the tight junction that forms between the merozoite and erythrocyte surfaces as the parasite penetrates the host cell. Consistent with this, a recent study in T. gondii using an inducible promoter to obtain conditional knock-down of AMA1 expression has shown that the molecule is not involved in initial attachment to the host cell but is required for intimate attachment and efficient triggering of rhoptry secretion [12]. A recent study with different AMA1 domain combinations expressed on the surface of CHO cells showed that domain III interacts with the erythrocyte membrane protein Kx, but only after treatment with trypsin [13]. While these diverse observations are consistent with a receptor-binding function for AMA1, conclusive confirmation for such a role is still lacking. Although first found in the Plasmodium merozoite, AMA1 is also expressed in the sporozoite of P. falciparum [14] and monoclonal anti-AMA1 antibodies that inhibit erythrocyte invasion by the merozoite can also inhibit hepatocyte invasion by the sporozoite. AMA1 could therefore have identical or very similar functions in the pre-erythrocytic and asexual blood stages of the parasite life cycle, and might serve as an effective vaccine against these two critical phases. We have recently reported the crystal structure of the complete ectoplasmic region of AMA1 from P. vivax (PvAMA1) from the strain Sal I [15], showing that it is divided into three domains as previously predicted from the pattern of cystinebridge formation [16,17]. Here, we discuss the implications of the PvAMA1 crystal structure for the three-dimensional structure of AMA1 from other Plasmodium species and its orthologues in T. gondii and B. bovis. Merozoite antigen erythrocyte-binding ligand (MAEBL), a rhoptry membrane protein identified in several Plasmodium species [18,19], is a paralogue of AMA1. It contains two closely related tandemrepeat segments, M1 and M2, in the N-terminal moiety of the protein, each homologous to the combined ectoplasmic domain I/II motif of AMA1, and a Duffy-binding like (DBL) domain immediately upstream from the transmembrane and cytoplasmic regions. We compare the M1 and M2 regions of MAEBL with PvAMA1 and show that some of the invariant and highly conserved residues in domains I and II of AMA1 are also likely to play an important structural role in this paralogue. We also examine the distribution of polymorphic sites in PfAMA1 and PvAMA1, and discuss the consequences for vaccine development.

2. Methods and materials 2.1. Sequence alignment Sequence alignments were made using the program MUSCLE [20]. One representative AMA1 sequence was taken for each Plasmodium species. Minor manual adjustments were made to the alignment of AMA1 sequences with that of MAEBL from P. vivax using criteria based on the PvAMA1 crystal structure. Sequences used in the alignments are given in Figs. 2 and 5. 2.2. Structure modelling The structure of the P. falciparum orthologue, PfAMA1, of the FVO strain [21] was modelled by introducing the 135 amino acid changes that exist in the ordered regions of the PvAMA1 crystal structure (an additional 48 changes occur in disordered regions). Atomic coordinates of PvAMA1 were taken from the Protein Data Bank (PDB) entries 1w81 and 1w8k. Dihedral angles of the side chains were adjusted on the PvAMA1 main-chain scaffold using the mean force field algorithm [22]. No changes were introduced into the mainchain conformation. The limited insertions in the PfAMA1 sequence with respect to PvAMA1 are located on disordered regions and were thus not modelled.

3. Results and discussion 3.1. The structure of the AMA1 ectoplasmic region in Plasmodium species The secondary structure and domain organisation of PvAMA1 is illustrated in Fig. 1. Based on the crystal structure [15], the domains of PvAMA1 can be defined as follows: domain I extends from residues 43 to 248, domain II from 249 to 385, and domain III from 386 to 487. Residues are numbered from the first residue of the signal sequence and, from alignment with the PfAMA1 sequence, the prosequence of PvAMA1 is assumed to terminate at residue 42 [9]. Domain I has an N-terminal extension (residues 43–62) in which strand ␤1 hydrogen bonds in parallel to ␤20 of domain III and helix ␣1 lies across domain II. The core of domains I and II is based on the PAN folding motif (Plasminogen, Apple, Nematode [23]), a protein fold describing domains commonly associated with carbohydrate- or protein-receptor binding functions. The PAN folding motif consists of a central five-stranded anti-parallel ␤-sheet with strand order 1-5-34-2, an ␣ helix connecting the second and third strands of the sheet, and two anti-parallel ␤-strands hydrogen bonded together, one connecting the first ␤-sheet strand to the second and the other connecting the fourth ␤-sheet strand to the fifth. The helix and double ␤-strand lie against opposite sides of the central ␤-sheet to one another. PAN domains are also characterized by six Cys residues linked in the order

M.-L. Chesne-Seck et al. / Molecular & Biochemical Parasitology 144 (2005) 55–67

57

Fig. 1. Secondary structure and domain organisation of PvAMA1. ␤ strands are shown as arrows, ␣ helices are shown as rectangles and the 3.010 helix ␪1 as a wavy rectangle. Cystine bridges are drawn in thick line with residue numbers indicated. Domain I secondary structure elements are indicated in white, domain II in light grey and domain III in dark grey.

Cys1-Cys6, Cys2-Cys5 and Cys3-Cys4, although in a number of cases the Cys1-Cys6 bridge is absent [23]. In the PvAMA1 structure, only the Cys2-Cys5 bridge, connecting the ␣ helix to the fourth ␤-sheet strand, conforms to the PAN pattern (Cys162-Cys192 in domain I and Cys282-Cys354 in domain II); the other cystines are thus not critical for maintaining this protein fold in AMA1. The structural homology between domains I and II suggests that they are probably the product of gene duplication and have since diverged significantly. Indeed, both domains – domain I in particular – have significant additional secondary structure elements and loop regions appended to the PAN core. Domain III does not belong to any currently known fold. The structure of the isolated domain III from PfAMA1 has been studied in solution by nuclear magnetic resonance (NMR) [17]. Much of the NMR structure is disordered because domains I and II interact with, and presumably stabilize, these regions of domain III. Indeed, only ␤22, ␤23, the first turn of ␣8 and the three cystine bridges of the NMR structure are similar to the crystal structure of PvAMA1. The root-mean-square difference between 26 equivalent C␣ positions of the crystal and NMR ˚ suggesting that even structures after superposition is 3.5 A, these regions depart significantly from the native structure in the absence of domains I and II. Several Plasmodium AMA1 ectoplasmic sequences, which are highly conserved, are compared in Fig. 2, with the secondary structure components observed in the crystal structure of PvAMA1 indicated in the alignment. The overall sequence identity is 39%, with 40% in domain I, 38% in domain II and 33% in domain III for the sequences chosen for this comparison. The most significant conserved feature in Plasmodium AMA1 sequences is the set of 16 Cys residues, underlining the importance of the disulphide bridges in defining the polypeptide fold. A number of polar and apolar

interactions in the PvAMA1 crystal structure are made by residues that are invariant or conserved in character between Plasmodium orthologues and include both inter- and intradomain contacts. Polar interactions in PvAMA1 between the side chains of species-invariant residues, listed in Table 1, Table 1 Polar interactions in the PvAMA1 crystal structure between the side chains of invariant residues in Plasmodium AMA1 Intra-domain contacts Domain I

Ser70 Arg73 Glu78 Glu78 Lys99 Tyr147 Gln200 Lys237

Asp75 Glu201 Lys237 Asn238 Gln200 His165 Tyr234 Asn238

Asn283 Asp293 Thr374

Thr373 Ser337 Ser377

Lys392 Glu453 Arg455

Asp426 Arg455 Tyr474

Ser70 Asp79 Lys225

Asp293 Lys336 Asn283

Domain I–III

Trp55

Tyr391

Domain II–III

Asp262 Asn366

Arg420 Ser442

Domain II

Domain III

Inter-domain contacts Domain I–II

*

*

*

*

*

*

*

*

*

*

Several invariant side-chain/main-chain interactions that also occur in the structure are not shown. Those interactions where residues are also invariant with TgAMA1, BbAMA1 are indicated by *. Intra- and inter-domain contacts are separated according to domain. Using the coordinates from the ˚ PDB entry 1w8k, the inter-atomic distance cut-off was set at 3.5 A.

58

M.-L. Chesne-Seck et al. / Molecular & Biochemical Parasitology 144 (2005) 55–67

Fig. 2. Sequence alignment of Plasmodium AMA1 ectoplasmic regions. Strictly conserved residues are indicated with black background and conservative differences with grey background. Residue numbering is given for P. vivax and secondary structure assignation from the PvAMA1 crystal structure (calculated by the program DSSP [51]) is indicated as “e” for ␤ strand, “h” for ␣ helix and “g” for 3.010 helix. Disordered regions that were not modelled are shown with “+” and the domain II loop is indicated. The following sequences were selected for comparison: P. vivax, GenBank entry CAA76546; P. knowlesi, AAA63444; P. cynomolgi, CAA60053; P. fragile, AAA29474; P. reichenowi, CAB66387; P. falciparum, AJ277646; P. chabaudi, AAB36509; P. yoelii yoelli, EAA20929; P. berghei, AAC47192.

M.-L. Chesne-Seck et al. / Molecular & Biochemical Parasitology 144 (2005) 55–67

59

Fig. 3. Conserved polar interactions in Plasmodium AMA1. The ectoplasmic region of PvAMA1 (coordinates from the Protein Data Bank entry 1w81) is shown as a stereo view. Domain I is green, domain II is blue and domain III is mauve. Polar interactions between the side chains of residues that are invariant in Plasmodium are indicated by dotted lines: red for intra-domain interactions and yellow for inter-domain interactions. Disulphide bridges are indicated by solid yellow lines.

form a network extending over the entire molecule (Fig. 3). A larger number of polar contacts are made between invariant side chains and the polypeptide backbone (not shown); these should contribute equally to the structural similarity of AMA1 from the different species. The sequence of PfAMA1 from the FVO strain was modelled on the PvAMA1 (Sal I) crystal structure (58.9% sequence identity). Sequence differences between these two species are distributed over the entire ectoplasmic region. Five occur at solvent-inaccessible sites in the PfAMA1 model and are conservative changes involving aliphatic side chains (Val, Leu and Ile), whereas the remaining 183 differences are located on the surface of the molecule (Fig. 4A). The sidechain conformation of the non-conserved residues could be easily adapted to the fixed PvAMA1 main-chain conformation, implying that the structure of the two orthologues should be very similar to each other. This probably also applies to other Plasmodium species because the limited sequence insertions and deletions occur in disordered regions of the PvAMA1 crystal structure, with the exception of one inserted residue in the P. chabaudi orthologue at a position aligning with ␤19 (see Fig. 2). Domain I and, to a lesser extent, domain II have long polypeptide segments that are either convoluted with no regular secondary structure or are disordered in the PvAMA1 crystal structure. One notable feature in the structure is the presence of 13 buried water molecules in domain I. These molecules are located in cavities that are lined with conserved

residues and accordingly can be expected in all Plasmodium orthologues. Three of these buried water molecules are in contact with each other and contribute importantly to a network of internal hydrogen bonds with several mainchain atoms and the side chains of six invariant buried polar residues: His68, Ser70, Asp75, Lys99 and Gln200 from domain I, and Asp 293 from domain II. This highly polar cavity is separated by the invariant residue, Tyr234, from another cavity with two buried solvent molecules, also maintained in an environment of invariant residues. Other buried solvent molecules occur singly, often interacting with mainchain atoms and invariant or highly conserved residues. 3.2. Polymorphism in PfAMA1 and PvAMA1 AMA1 is significantly polymorphic in both P. falciparum [24–26] and P. vivax [27]. The predominance of nonsynonymous over synonymous mutations in the PfAMA1 gene implies that sequence diversity is selected for, and maintained, to counter the host’s immune response [28–32]. Although polymorphic sites are distributed throughout the length of the ectoplasmic region, domain I is the most variable [29–32]. Using the crystal structure of PvAMA1 and our model of PfAMA1, we examined the three-dimensional distribution of polymorphic sites in these two species. Taking 356 PfAMA1 sequences that were available in the EMBL, GenBank and DDJB databases, we considered a site to be polymorphic if

60

M.-L. Chesne-Seck et al. / Molecular & Biochemical Parasitology 144 (2005) 55–67

Fig. 4. Model of PfAMA1 showing the three-dimensional distribution of (A) PfAMA1(FVO)/PvAMA1(Sal I) sequence differences in magenta and (B) threedimensional distribution of PfAMA1 polymorphic sites shown in green; unaffected residues are in grey. The two views in both (A) and (B) are related by a 180◦ rotation about a vertical axis so that both sides of the molecule are shown.

at least two sequences differed from the consensus amino acid at that position. All polymorphic sites are surface exposed; a total of 32 are located in domain I (15.5% of the residues comprising this domain), 11 in domain II (8.0%) and 9 in domain III (8.6%), thus showing domain I to be the most polymorphic region of PfAMA1, as previously noted by others [24–26]. Most sites are dimorphic, but 13 sites have three different amino acids (however, at five of these sites the least frequent substitution appears in only one sequence), two sites have four substitutions (the least frequent substitution appears only once at one site) and one site shows seven substitutions

(see Table 2). Unlike the distribution of PvAMA1/PfAMA1 sequence differences, the distribution of the PfAMA1 polymorphic sites is highly biased to one side of the molecule (Fig. 4B). This suggests that this part of the ectoplasmic region may be more exposed to the exterior on the parasite surface and/or less susceptible to functional constraints. Polymorphism has been less studied in P. vivax, and because current data give only limited coverage of domains II and III, our analysis is restricted to domain I. From a comparison of 232 partial PvAMA1 sequences taken from the same data bank sources as PfAMA1, we found 17 polymorphic

M.-L. Chesne-Seck et al. / Molecular & Biochemical Parasitology 144 (2005) 55–67

61

Table 2 Comparison of polymorphic sites in PfAMA1 and PvAMA1 PfAMA1 Domain I 121 162 167 172 173 175 185 187 188 189 190 195 196 197 199 200 201 204 206 207 244 225 227 228 230 242 243 244 245 248 265 267 269 273 282 283 285 296 300

PvAMA1 EK NK TK EG NKE DY P NEK P LPH MI L DNY QGDEHRV RK DHRL FLSV DN EK YD MI NI D NK KEQ YD KNE DNY KN H K EQ KI M KI SL QE DH KE

(230/3) (274/81) (262/93) (194/164) (305/49/2) (319/37) – (138/133/85) – (327/24/5) (240/116) – (265/90/1) (116/85/84/32/26/7/6) (352/4) (158/144/27/26) (281/66/8/1) (206/150) (278/78) (298/58) (351/5) (243/113) – (345/10) (239/97/20) (183/173) (223/72/61) (328/14/14) (328/28) – – (216/140) (341/15) – (238/117) (249/106) (280/75) (282/70) (251/95)

66 107 112 117 118 120 130 132 133 134 135 140 141 142 144 145 146 149 151 152 169 170 172 173 175 187 188 189 190 193 210 212 214 218 227 228 230 241 245

R DA KTR G D RKS NK DN DN H I IL AE N K EAG R D V E K V AT G Q E K EKN KEQ HY SP A N VL EV SD E N K

Domain II 308 325 330 332 393 395 404 405 407 435 439

EQK HDR SP NI HRY KR RT KE QH INT HND

(108/59/1) (162/3/1) (152/15) (137/29) (115/51/1) (152/13) (95/71) (85/82) (141/26) (131/34/2) (98/64/4)

253 270 275 277 338 340 349 350 352 380 384

G Y E E K R S V K Q L

Domain III 448 451 485 493 496 503 505 512 544

DN KM KI DA IM NRH FY KR KN

(136/31) (98/69) (97/70) (131/36) (94/73) (97/69/1) (159/7) (88/79) (95/48)

393 396 427 435 438 445 447 454 486

D E K E R N Y K L

– (11/2) (131/93/8) – – (219/7/6) (157/75) (119/113) (223/9) – – (124/108) (185/47) – – (170/60/2) – – – – – – (229/3) – – – – (113/91/27) (148/83/1) (229/3) (181/51) – (197/35) (165/67) (165/67) – – –

Sequences of PfAMA1 and PvAMA1 are compared at all residue positions where at least one species is polymorphic. Polymorphic sites in each species are highlighted in bold and all observed amino acid substitutions are given. A total of 356 PfAMA1 sequences and 232 PvAMA1 sequences were used. For PvAMA1, only domain I polymorphisms are shown since the coverage of domains II and III is not significant. The number of sequences contributing to each polymorphic substitution at each site is given in parentheses in the same order as the amino acid substitutions themselves.

62

M.-L. Chesne-Seck et al. / Molecular & Biochemical Parasitology 144 (2005) 55–67

sites in domain I (8.3%). As with PfAMA1, all polymorphic residues are surface exposed. Polymorphic sites on domain I of PfAMA1 and PvAMA1 have similar distributions, being mostly situated within the long, solvent-exposed, convoluted peptide segment between residues 105 and 136 (PvAMA1 residue numbering), the helices ␣3 and ␣4, and the turns connecting ␤7 to ␤8 and ␤9 to ␤10 (Table 2). Ten of the polymorphic sites in domain I are in common between the two orthologues. In both these species, the presence of a particular amino acid at certain polymorphic sites was found in some cases to be highly correlated with the sequence occurring at other polymorphic sites. Since a number of these are well separated from each other in the tertiary structure, it does not seem that their occurrence is directly caused by structural factors in the protein itself but rather by genetic factors.

3.3. Other orthologues and paralogues of AMA1 in Apicomplexa Orthologues of AMA1 from the apicomplexan parasites T. gondii (TgAMA1) [2,3] and B. bovis (BbAMA1) [4] have been identified and sequenced. Many of the invariant residues and conservative differences in AMA1 across Plasmodium species are also present in TgAMA1 and BbAMA1. These two orthologues show the same pattern of Cys residues in domains I and II of Plasmodium, and several residues that are invariant or conservative in character with respect to Plasmodium sequences play a critical role in the threedimensional structure of PvAMA1 (Fig. 5A). Interactions between invariant polar residues in domains I and II of Plasmodium AMA1 that should be preserved in TgAMA1 and BbAMA1 are noted in Table 1. Domain III in both TgAMA1 and BbAMA1 shows significant sequence differences to Plasmodium AMA1, most notably in the Cys residues (Fig. 5B). Domain III of BbAMA1 has only four Cys residues but their alignment shows that they should form disulphide bridges equivalent to Cys388-Cys444 and Cys432-Cys449 in PvAMA1 (see Figs. 1 and 5B). The cystine bridge corresponding to Cys434-Cys451 in PvAMA1, however, is absent, and the pair of CXC motifs characterizing the cystine knot [33] observed in the PvAMA1 structure is thus not present in BbAMA1. Although TgAMA1 has six Cys residues, as in domain III of the Plasmodium orthologues, alignment only occurs with Cys432 and Cys444 in PvAMA1, which are not linked together in Plasmodium; the pattern of cystine bridges is thus different in TgAMA1. Sequence identity with Plasmodium AMA1 at other residue positions is low for domain III of TgAMA1 (Fig. 5B). Therefore, whereas domain III of BbAMA1 might bear some structural homology to the PvAMA1 crystal structure, this is unlikely to be the case in TgAMA1. The differences in domain III may reflect the proximity of this part of AMA1 to the parasite surface and its consequent interaction with other surface proteins in the different apicomplexan taxa, or perhaps interaction with different host-cell receptors.

Although domains M1 and M2 of MAEBL are each homologous to combined domains I/II of AMA1, their sequence identity to AMA1 is low (Fig. 5). Ten of the 14 Cys residues of M1 and M2 align with the 10 Cys residues of the domain I/II AMA1 consensus sequence (Cys 2, 6, 9, 10, 11, 12 13, 14, 15, 16 of M1 and M2 in Fig. 5) and thus imply the same pattern of cystine bridge formation for these residues in MAEBL. Cys7 and Cys8 in M1 and M2 (Fig. 5) may also be linked together since this would conform to one of the consensus PAN cystine bridges [23]. There is not sufficient information to predict the pairing of the remaining Cys residues (Cys1, Cys3, Cys4, Cys5). Most of the residues in M1 and M2 that are invariant with respect to AMA1 are engaged in conserved hydrophobic interactions in the PvAMA1 structure. Of note here, the invariant hydrophobic pair Phe-Leu (residues 110–111 in PvAMA1) anchors its two flanking sections that form long solvent-exposed loops. These two residues are buried in a hydrophobic pocket lined by several invariant and highly conserved residues. All invariant Gly residues (Fig. 5B), with the possible exception of those that are equivalent to Gly125 in PvAMA1 (occurring in a disordered region in the crystal structure) are buried and would give rise to steric hindrance if substituted by other amino acids. The only predicted conserved polar interaction is made by a Lys residue equivalent to Lys225 in PvAMA1, which is buried and forms charged hydrogen bonds to the main chain. An epitope mapping study with the invasion-inhibitory monoclonal antibody 4G2 has shown that this residue plays an important role in maintaining the domain I/domain II interface in AMA1 [15] and this probably applies to MAEBL as well. The alignment of M1 and M2 shows significant insertions immediately preceding ␤4 and ␣5 (about 30 and 75 residues, respectively, in M1 and M2) with respect to AMA1. 3.4. Natural processing in AMA1 AMA1 in P. falciparum undergoes proteolytic processing in the parasite after translocation from the micronemes to the parasite surface [10,34]. The main chain is cut at the Thr517-Ser518 peptide bond, 29 residues upstream from the transmembrane region, thus cleaving the ectoplasmic region from the merozoite surface. An additional internal cleavage occurs at the Asn464-Asp465 peptide bond in about one third of the shed ectoplasmic polypeptides, but the biological significance of this is not known. Proteolytic cleavage of the AMA1 ectoplasmic region also occurs in P. knowlesi [35] and is likely to take place in other Plasmodium species, although this has not been experimentally confirmed. Cleavage of the AMA1 ectoplasmic region from the tachyzoite has also been observed for TgAMA1 [2,3], although this occurs at an intramembrane site [34]. Antibodies that prevent cleavage of AMA1 in P. falciparum inhibit invasion [36], suggesting that shedding may be a prerequisite for invasion. We have found two partial cleavage sites by N-terminal sequencing of the soluble recombinant PvAMA1 ectoplasmic region expressed in Pichia pastoris, both presumably

M.-L. Chesne-Seck et al. / Molecular & Biochemical Parasitology 144 (2005) 55–67

63

Fig. 5. Sequence alignment of (A) domains I and II of PvAMA1, PfAMA1, TgAMA1 and BbAMA1 with domains M1 and M2 of PvMAEBL (GenBank entries Y16950, AJ277646, AF010264, AY486101 and AY042083, respectively), and (B) of domain III of PvAMA1, PfAMA1, TgAMA1 and BbAMA1. Strictly conserved residues are indicated with black background and conservative differences with grey background. Residue numbering is given for P. vivax and the aligned secondary structure from the PvAMA1 crystal structure (calculated by the program DSSP [51]) is indicated as “e” for ␤ strand and “h” for ␣ helix. The domain II loop is indicated by “+”. Cys residues of regions M1 and M2 of PvMAEBL are numbered sequentially from 1 to 16.

64

M.-L. Chesne-Seck et al. / Molecular & Biochemical Parasitology 144 (2005) 55–67

mediated by a protease(s) from P. pastoris. One site occurs at the Tyr409-Ser410 peptide bond, which is located within the disordered loop between residues 403 and 414 in domain III, aligning with the cleavage site, Asn464-Asp465, in PfAMA1 that is partially proteolyzed during natural processing. The second partial proteolysis site in recombinant PvAMA1 occurs between Lys321 and Ser322 in domain II, and thus lies within the disordered segment 295–334 that we have referred to as the domain II loop [15]. This was not modelled in the PvAMA1 crystal structure. This cleavage site aligns with a partial cleavage site found in the recombinant ectoplasmic domain of PfAMA1, also expressed in P. pastoris [37]. For both these sites, the cleaved polypeptide chains are covalently held together by the disulphide bridges and the proteolyzed PvAMA1 ectoplasmic region thus migrates at essentially the same rate as the intact protein in SDS-PAGE under non-reducing conditions. The natural cleavage site Thr517-Ser518 in PfAMA1, which leads to shedding of the ectoplasmic region during maturation, aligns with the Lys459-Glu460 peptide bond of PvAMA1. This peptide bond is intact in the crystal structure, stabilized by extensive hydrogen bonds within in a ␤-hairpin. In this conformation, we anticipate protection of this peptide bond against proteolysis. Thus, if AMA1 in P. vivax (and other Plasmodium species) is cleaved, like PfAMA1, from the parasite surface during erythrocyte invasion, we suggest that the conformation of this region of the protein differs from that observed in the crystal structure, at least during the proteolysis step. The structure of this part of the molecule may be influenced by interactions with other parasite surface molecules, by the proximity of the parasite membrane or by erythrocyte molecules during the course of invasion. It is possible that the cleavage site is thus protected against premature proteolysis and only becomes exposed to the protease upon interaction between AMA1 and its receptor(s). The two natural cleavage sites identified in PfAMA1 are both in the membrane-proximal domain III and should thus be easily accessible to membrane-bound proteases on the parasite. By contrast, the additional partial cleavage that is found in domain II of soluble recombinant PvAMA1 (Lys321-Ser322) and PfAMA1 (Lys376-Ser377) may be less accessible to such proteases on the parasite surface, explaining its absence in the naturally cleaved ectodomain. The enzyme responsible for cleaving the PfAMA1 ectoplasmic region is a membrane-bound parasite protease indistinguishable, in terms of inhibitor susceptibility, from that involved in secondary processing of the major merozoite surface protein MSP1 [10]. It has been suggested that the enzyme involved belongs to a class of proteases often referred to as sheddases, which are not primarily sequence-specific in cleavage-site recognition but instead target exposed, structurally flexible, polypeptide segments [38]. Processing of MSP1 also occurs at the time of erythrocyte invasion and, as with PfAMA1 [36], is essential for parasite survival.

3.5. The domain II loop In the crystal structure of PvAMA1, a 40-residue segment in domain II, comprising residues 295–334, had no corresponding electron density in the Fourier maps and thus could not be traced. This region is therefore disordered because of conformational heterogeneity or mobility. We have previously shown that the base of the domain II loop includes the epitope of the anti-PfAMA1 invasion-inhibitory monoclonal antibody, 4G2 [15]. Interestingly, this B-cell epitope overlaps with a highly immunogenic T-helper cell epitope in the human response [39]. The location of the 4G2 epitope and currently known structure–function correlations of PAN domains led us to speculate that AMA1 might have a receptor-binding role involving domain II. The loop is strictly conserved in residue length within the different Plasmodium orthologues but has lower sequence identity (21%) compared to the complete ectoplasmic region (39%). Nonetheless, no polymorphic sites have been detected to date in the domain II loop in P. falciparum, suggesting that functional constraints may operate to prevent antigenic variation in this surfaceexposed segment. Of note, secondary structure predictions using PHD [40] consistently assign a high probability to helix formation in the first half of the domain II loop in all Plasmodium orthologues. In BbAMA1, the domain II loop is two residues longer compared to Plasmodium AMA1, whereas in TgAMA1 this region is 17 residues shorter. This may indicate functional differences; whereas Babesia, like Plasmodium, invades erythrocytes, Toxoplasma is able to invade a wider spectrum of cell types. In MAEBL, the equivalent region in M1 and M2 is short and thus closer in length to that present in other PAN domains; the role of this region is most likely different for this paralogue.

4. Conclusion The crystal structure determination of AMA1 [15] has revealed a hitherto unexpected relationship to the PAN motif, a polypeptide fold found in proteins often associated with receptor-binding functions. The generally low sequence similarity between PAN domains, the departure in AMA1 from the usual PAN Cys signature, and the insertion of additional large peptide segments onto the PAN scaffold of domains I and II account for this structural homology not being previously recognized from sequence alone. PAN domains are commonly present as tandem repeats in proteins [23], and gene duplication may indeed have produced domains I and II in AMA1. According to this hypothesis, the two domains would have subsequently co-evolved to form inter-domain contacts that are highly conserved between species, while at the same time developing large sequence insertions on the surface of the protein that contribute directly to the function of AMA1, either in receptor binding or in displaying antigenic variability to escape immune responses by the host. Of note, however, the only insertions/deletions that occur in

M.-L. Chesne-Seck et al. / Molecular & Biochemical Parasitology 144 (2005) 55–67

Plasmodial AMA1 sequences are in domain III. The domain I/II motif is also present in the chimeric Plasmodium paralogue MAEBL as the tandem repeats M1 and M2, implying a second gene duplication; here, insertions into the PAN scaffold are even more extensive than in AMA1, although the region matching the domain II loop is considerably shortened. The similarity between AMA1 and MAEBL suggests that the domain I/II ensemble originated early in apicomplexan evolution and that the M1/M2 duplication occurred prior to divergence of Plasmodium species [41]. Like AMA1, MAEBL also plays a role in hepatocyte invasion by the sporozoite as well as erythrocyte invasion by the merozoite [18,42]. AMA1 is the first protein in Plasmodium shown, via threedimensional structure analysis, to have PAN folding motifs, and by homology, we can also include MAEBL. The PAN module and the closely related Apple domain are found in other micronemal proteins in Apicomplexa [43], hinting that they, AMA1 and MAEBL might be phylogenetically derived from a distant common ancestor. The use of individual domains or sub-domains of the ectoplasmic region could lead to simpler and possibly more stable AMA1-based vaccine products. For example, antibodies that were affinity-purified from the plasma from a donor in a malaria-endemic region using the recombinant PfAMA1 domain III showed significant inhibition of erythrocyte invasion with two different parasite strains [17], thus demonstrating the presence of protective epitopes on this domain. However, a recent study in rabbits using different recombinant PfAMA1 domain combinations as immunogens suggests that the situation may be less straightforward [44]; individual domains were not effective in inducing invasion inhibitory antibodies and only the double domain I/II combination induced growth-inhibitory antibodies levels comparable to the complete ectoplasmic region. Moreover, the activity of the invasion-inhibitory mAb 4G2, which recognizes a reduction-sensitive epitope located at the base of the PfAMA1 domain II loop, requires the presence of domain I [15]. But since the domain II loop also contains a human Tcell epitope [39] and carries no known polymorphisms, peptide mimetics of this region would be of particular interest for vaccine development. Although the 4G2 epitope on domain II requires the presence of domain I, we do not exclude the possibility that such sub-domain mimetics could be designed with knowledge of the conformation of the ordered extremities of the domain II loop in the crystal structure; promising candidates could be selected with mAb 4G2. Polymorphism is a critical factor governing the effectiveness of a vaccine. Although vaccination with AMA1 in animal model systems protects against homologous challenge [45–47], it is less effective against heterologous challenge [29,37,48]. Nonetheless, significant levels of in vitro parasite inhibition are still observed when challenged with strains that have over 20 amino acid differences in the AMA1 ectoplasmic region [37,49] (24 amino acids between FVO and 3D7 [37], and 23 between 3D7 and HB3 [49]), which is close to the maximum found in naturally occurring strains. For example,

65

antibodies from rabbits immunised with AMA1 strain 3D7 gave 20% inhibition of erythrocyte invasion by FVO parasites [37]. This supports the hypothesis that a non-negligible part of the protective immune response induced by AMA1 is against conserved epitopes. Indeed, the loss of inhibition with heterologous challenge is approximately correlated linearly with the number of amino acid differences between the species in question [37]. Moreover, most experiments showing lower inhibition following heterologous challenge have been done with rabbit or mouse antibodies, species whose immune systems do not always adequately represent the human immune system, as is increasingly becoming evident [50]. In support of this notion, preliminary data obtained from an AMA1 vaccination study in rhesus macaques show that the differences in in vitro parasite inhibition levels between homologous and heterologous challenge are much less pronounced when compared to antibodies derived from rabbits under similar experimental conditions (Thomas AW and Remarque EJ, Unpublished data). Irrespective of these considerations, polymorphism in AMA1 will have an impact on vaccine development and thus the challenge for the vaccine developers is to address this issue. For instance, in a recent human phase I trial [6], the use of an equal mixture of FVO and 3D7 AMA1 (strains that differ at 24 of the 52 polymorphic sites in the ectoplasmic region) gave equal ELISA titres for both strains, and invasion inhibition of heterologous challenge was at the same level as for homologous challenge. Thus, multiple-allele immunogens appear to be additive in their protection against heterologous infection [6] and vaccine constructs based on two or more representative strains may offer effective solution to the problem of polymorphism.

Acknowledgements This work was funded by the European Commission (contracts QLK2-CT-1999-01293 and QLK2-CT-2002-01197), the European Malaria Vaccine Initiative, the Pasteur Institute, the Centre National de la Recherche Scientifique, the Biomedical Primate Research Centre and the Medical Research Council, UK.

References [1] Waters AP, Thomas AW, Deans JA, et al. A merozoite receptor protein from Plasmodium knowlesi is highly conserved and distributed throughout Plasmodium. J Biol Chem 1990;265:17974–9. [2] Donahue CG, Carruthers VB, Gilk SD, Ward GE. The Toxoplasma homolog of Plasmodium apical membrane antigen-1 (AMA-1) is a microneme protein secreted in response to elevated intracellular calcium levels. Mol Biochem Parasitol 2000;111:15–30. [3] Hehl AB, Lekutis C, Grigg ME, et al. Toxoplasma gondii homologue of Plasmodium apical membrane antigen 1 is involved in invasion of hosT-cells. Infect Immunol 2000;68:7078–86. [4] Gaffar FR, Yatsuda AP, Franssen FF, de Vries E. Erythrocyte invasion by Babesia bovis merozoites is inhibited by polyclonal anti-

66

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20] [21]

[22]

M.-L. Chesne-Seck et al. / Molecular & Biochemical Parasitology 144 (2005) 55–67 sera directed against peptides derived from a homologue of Plasmodium falciparum apical membrane antigen 1. Infect Immunol 2004;72:2947–55. Saul A, Lawrence G, Allworth A, et al. A human phase 1 vaccine clinical trial of the Plasmodium falciparum malaria vaccine candidate apical membrane antigen 1 in Montanide ISA720 adjuvant. Vaccine 2005;23:3076–83. Malkin EM, Diemert DJ, McArthur JH, et al. Phase 1 clinical trial of apical membrane antigen 1: an asexual blood-stage vaccine for Plasmodium falciparum malaria. Infect Immunol 2005;73:3677–85. Thomas AW, Deans JA, Mitchell GH, Alderson T, Cohen S. The Fab fragments of monoclonal IgG to a merozoite surface antigen inhibit Plasmodium knowlesi invasion of erythrocytes. Mol Biochem Parasitol 1984;13:187–99. Fraser TS, Kappe SH, Narum DL, VanBuskirk KM, Adams JH. Erythrocyte-binding activity of Plasmodium yoelii apical membrane antigen-1 expressed on the surface of transfected COS-7 cells. Mol Biochem Parasitol 2001;117(1):49–59. Howell SA, Withers-Martinez C, Kocken CH, Thomas AW, Blackman MJ. Proteolytic processing and primary structure of Plasmodium falciparum apical membrane antigen-1. J Biol Chem 2001;276:31311–20. Howell SA, Well I, Fleck SL, Kettleborough C, Collins CR, Blackman MJ. A single malaria merozoite serine protease mediates shedding of multiple surface proteins by juxtamembrane cleavage. J Biol Chem 2003;278:23890–8. Mitchell GH, Thomas AW, Margos G, Dluzewski AR, Bannister LH. Apical membrane antigen 1, a major malaria vaccine candidate, mediates the close attachment of invasive merozoites to host red blood cells. Infect Immunol 2004;72:154–8. Mital J, Meissner M, Soldati D, Ward GE, Conditional expression of Toxoplasma gondii apical membrane antigen-1 (TgAMA1) demonstrates that TgAMA1 plays a critical role in hosT-cell invasion. Mol Biol Cell, in press. Kato K, Mayer DC, Singh S, Reid M, Miller LH. Domain III of Plasmodium falciparum apical membrane antigen 1 binds to the erythrocyte membrane protein Kx. Proc Natl Acad Sci USA 2005;102:5552–7. Silvie O, Franetich JF, Charrin S, et al. A role for apical membrane antigen 1 during invasion of hepatocytes by Plasmodium falciparum sporozoites. J Biol Chem 2004;279:9490–6. Pizarro JC, Vulliez-Le Normand BV, Chesne-Seck ML, et al. Crystal structure of the malaria vaccine candidate apical membrane antigen 1. Science 2005;308:408–11. Hodder AN, Crewther PE, Matthew ML, et al. The disulfide bond structure of Plasmodium apical membrane antigen-1. J Biol Chem 1996;271:29446–52. Nair M, Hinds MG, Coley AM, et al. Structure of domain III of the blood-stage malaria vaccine candidate. Plasmodium falciparum apical membrane antigen 1 (AMA1). J Mol Biol 2002;322:741–53. Kappe SH, Noe AR, Fraser TS, Blair PL, Adams JH. A family of chimeric erythrocyte binding proteins of malaria parasites. Proc Natl Acad Sci USA 1998;95:1230–5. Ghai M, Dutta S, Hall T, Freilich D, Ockenhouse CF. Identification, expression, and functional characterization of MAEBL, a sporozoite and asexual blood stage chimeric erythrocyte-binding protein of Plasmodium falciparum. Mol Biochem Parasitol 2002;123:35–45. Edgar RC. Muscle: a multiple sequence alignment method with reduced time and space complexity. BMC Bioinformat 2004;5:113. Kocken CH, Withers-Martinez C, Dubbeld MA, et al. High-level expression of the malaria blood-stage vaccine candidate Plasmodium falciparum apical membrane antigen 1 and induction of antibodies that inhibit erythrocyte invasion. Infect Immunol 2002;70: 4471–6. Koehl P, Delarue M. Application of a self-consistent mean field theory to predict protein side-chains conformation and estimate their conformational entropy. J Mol Biol 1994;239:249–75.

[23] Tordai H, Banyai L, Patthy L. The PAN module: the N-terminal domains of plasminogen and hepatocyte growth factor are homologous with the apple domains of the prekallikrein family and with a novel domain found in numerous nematode proteins. FEBS Lett 1999;461:63–7. [24] Oliveira DA, Udhayakumar V, Bloland P, et al. Genetic conservation of the Plasmodium falciparum apical membrane antigen-1 (AMA-1). Mol Biochem Parasitol 1996;76:333–6. [25] Marshall VM, Zhang L, Anders RF, Coppel RL. Diversity of the vaccine candidate AMA-1 of Plasmodium falciparum. Mol Biochem Parasitol 1996;77:109–13. [26] Kocken CH, Narum1 DL, Massougbodji A, et al. Molecular characterisation of Plasmodium reichenowi apical membrane antigen-1 (AMA-1), comparison with P. falciparum AMA-1, and antibodymediated inhibition of red cell invasion. Mol Biochem Parasitol 2000;109:147–56. [27] Figtree M, Pasay CJ, Slade R, et al. Plasmodium vivax synonymous substitution frequencies, evolution and population structure deduced from diversity in AMA 1 and MSP 1 genes. Mol Biochem Parasitol 2000;108:53–66. [28] Hughes MK, Hughes AL. Natural selection on Plasmodium surface proteins. Mol Biochem Parasitol 1995;71:99–113. [29] Crewther PE, Matthew ML, Flegg RH, Anders RF. Protective immune responses to apical membrane antigen 1 of Plasmodium chabaudi involve recognition of strain-specific epitopes. Infect Immunol 1996;64:3310–7. [30] Hughes MK, Hughes AL. Natural selection on Plasmodium surface proteins. Mol Biochem Parasitol 1995;71:99–113. [31] Escalante AA, Grebert HM, Chaiyaroj SC, et al. Polymorphism in the gene encoding the apical membrane antigen-1 (AMA-1) of Plasmodium falciparum X. Asembo Bay Cohort Project. Mol Biochem Parasitol 2001;113:279–87. [32] Polley SD, Conway DJ. Strong diversifying selection on domains of the Plasmodium falciparum apical membrane antigen 1 gene. Genetics 2001;158:1505–12. [33] Isaacs NW. Cystine knots. Curr Opin Struct Biol 1995;5:391–5. [34] Howell S, Hackett F, Jongco AM, et al. Distinct mechanisms govern proteolytic shedding of a key invasion protein in apicomplexan pathogens. Mol Microbiol 2005;57:1342–56. [35] Deans JA, Thomas AW, Alderson T, Cohen S. Biosynthesis of a putative protective Plasmodium knowlesi merozoite antigen. Mol Biochem Parasitol 1984;11:189–204. [36] Dutta S, Haynes JD, Barbosa A, et al. Mode of action of invasioninhibitory antibodies directed against apical membrane antigen 1 of Plasmodium falciparum. Infect Immunol 2005;73:2116–22. [37] Kennedy MC, Wang J, Zhang Y, et al. In vitro studies with recombinant Plasmodium falciparum apical membrane antigen 1 (AMA1): production and activity of an AMA1 vaccine and generation of a multiallelic response. Infect Immunol 2002;70:6948–60. [38] Schwager SL, Chubb AJ, Woodman ZL, et al. Cleavage of disulfidebridged stalk domains during shedding of angiotensin-converting enzyme occurs at multiple juxtamembrane sites. Biochemistry 2001;40:15624–30. [39] Lal AA, Hughes MA, Oliveira DA, et al. Identification of T-cell determinants in natural immune responses to the Plasmodium falciparum apical membrane antigen (AMA-1) in an adult population exposed to malaria. Infect Immunol 1996;64:1054–9. [40] Bost B. PHD: predicting one-dimensional protein structure by profile based neural networks. In: Carter CW, Sweet RM, editors. Methods in enzymology: macromolecular crystallography Part A, 266. New York: Academic Press; 1997. p. 525–39. [41] Michon P, Stevens JR, Kaneko O, Adams JH. Evolutionary relationships of conserved cysteine-rich motifs in adhesive molecules of malaria parasites. Mol Biol Evol 2002;19:1128–42. [42] Preiser P, Renia L, Singh N, et al. Antibodies against MAEBL ligand domains M1 and M2 inhibit sporozoite development in vitro. Infect Immunol 2004;72:3604–8.

M.-L. Chesne-Seck et al. / Molecular & Biochemical Parasitology 144 (2005) 55–67 [43] Brown PJ, Gill AC, Nugent PG, McVey JH, Tomley FM. Domains of invasion organelle proteins from apicomplexan parasites are homologous with the Apple domains of blood coagulation factor XI and plasma pre-kallikrein and are members of the PAN module superfamily. FEBS Lett 2001;497:31–8. [44] Lalitha PV, Ware LA, Barbosa A, et al. Production of the subdomains of the Plasmodium falciparum apical membrane antigen 1 ectodomain and analysis of the immune response. Infect Immunol 2004;72:4464–70. [45] Deans JA, Knight AM, Jean WC, Waters AP, Cohen S, Mitchell GH. Vaccination trials in rhesus monkeys with a minor, invariant. Plasmodium knowlesi 66 kD merozoite antigen. Parasit Immunol 1988;10:535–52. [46] Collins WE, Pye D, Crewther PE, et al. Protective immunity induced in squirrel monkeys with recombinant apical membrane antigen-1 of Plasmodium fragile. Am J Trop Med Hyg 1994;51:711– 9.

67

[47] Narum DL, Ogun SA, Thomas AW, Holder AA. Immunization with parasite-derived apical membrane antigen 1 or passive immunization with a specific monoclonal antibody protects BALB/c mice against lethal Plasmodium yoelii yoelii YM blood-stage infection. Infect Immunol 2000;68:2899–906. [48] Healer J, Murphy V, Hodder AN, et al. Allelic polymorphisms in apical membrane antigen-1 are responsible for evasion of antibodymediated inhibition in Plasmodium falciparum. Mol Microbiol 2004;52:159–68. [49] Hodder AN, Crewther PE, Anders RF. Specificity of the protective antibody response to apical membrane antigen 1. Infect Immunol 2001;69:3286–94. [50] Mestas J, Hughes CCW. Of mice and men: differences between mouse and human immunology. J Immunol 2004:2731–8. [51] Kabsch W, Sander C. Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers 1983;22:2577–637.