Widespread functional specialization of ... - Wiley Online Library

Molecular Microbiology (2003) 47(5), 1265–1278

Widespread functional specialization of Plasmodium falciparum erythrocyte membrane protein 1 family members to bind CD36 analysed across a parasite genome Bridget A. Robinson, Teresa L. Welch and Joseph D. Smith*† Department of Microbiology, Immunology, and Pathology, Colorado State University, Fort Collins, CO 80525, USA.

Summary Plasmodium falciparum-infected erythrocytes sequester from blood circulation by binding host endothelium. A large family of variant proteins mediates cytoadherence and their binding specificity determines parasite sequestration patterns and potential for disease. The aim of the present study was to understand how binding properties are encoded into family members and to develop sequence algorithms for predicting binding. To accomplish these goals computational approaches and a binding assay were used to characterize adhesion across Plasmodium falciparum erythrocyte membrane 1 (PfEMP1) proteins in the 3D7 parasite genome. We report that most family members encode the capacity to bind CD36 in the protein’s semiconserved head structure and describe the sequence characteristics of a group of PfEMP1 proteins that do not. Structural and functional grouping of PfEMP1 proteins based upon head structure and additional domain architectural properties provide new insights into the protein family. These can be used to investigate the role of proteins in malaria pathogenesis and potentially tailor vaccines to recognize particular binding variants.

Introduction The clinical manifestations of a Plasmodium falciparum malaria infection depend on a combination of various parasite, host, geographical, and social factors (Miller et al., Blackwell Science, LtdOxford, UKMMIMolecular Microbiology0950-382XBlackwell Publishing Ltd, 200347Original ArticleB. Robinson, T. L. Welch and J. D. SmithGenome-wide characterization of PfEMP1 binding to CD36

Accepted 15 November, 2002. *For correspondence. E-mail [email protected]; Tel. (+1) 206 284 8846 ext. 384; Fax (+1) 206 284 0313. †Present address: Seattle Biomedical Research Institute, 4 Nickerson Street, Seattle, WA 98109, USA.

© 2003 Blackwell Publishing Ltd

2002a). Among these, immunity to P. falciparum and the various adhesion properties of the mature infected erythrocyte has a major impact on disease outcome. All Plasmodium species studied undergo a process of clonal antigenic variation of proteins at the infected erythrocyte surface (Kyes et al., 2001). Antigenic variation allows parasites to establish chronic infections and enhance opportunities for transmission to mosquitoes. The P. falciparum erythrocyte membrane protein 1 (PfEMP1) family, encoded by var genes, are clonally variant antigens that also encode adhesive properties (Smith et al., 2001). As the parasite matures in erythrocytes, the red cell undergoes changes in shape and membrane rigidity. PfEMP1 proteins are responsible for the sequestration of mature, infected erythrocytes in microvascular spaces and parasite avoidance of spleen-dependent killing mechanisms. Infected erythrocytes bind a surprisingly large number of host receptors and interact with different cell types. The variable interaction of PfEMP1 proteins with different sets of receptors may determine parasite ability to cause disease. Plasmodium falciparum infections are characterized by the slow build-up of resistance to disease after repeated infections. One component of protective immunity is believed to be the acquisition of variant-specific antibodies against the surface of infected erythrocytes (Bull and Marsh, 2002). Adults, because they have experienced more infections than children, generally recognize a wider proportion of the circulating variants. However, first-time pregnant mothers, including those with significant degrees of malaria protective immunity, are more susceptible to infections characterized by placental sequestering parasites (Steketee et al., 1996). Pregnancy malaria places both the mother and fetus at increased risk of severe disease and death. Significantly, the parasite variants that infect children and pregnant women are functionally distinct (Fried and Duffy, 1996; Beeson et al., 1999). Whereas childhood infections are typified by CD36binding parasites, infected erythrocytes recovered from placenta do not bind CD36, but instead bind the receptors chondroitin sulphate A (CSA) (Fried and Duffy, 1996; Achur et al., 2000; Alkhalil et al., 2000; Beeson et al., 2000), hyaluronic acid (Beeson et al., 2000) and non-

1266 B. Robinson, T. L. Welch and J. D. Smith immune immunoglobulin (Flick et al., 2001). From these findings, the model has emerged that the placenta is a unique sequestration niche that supports the clonal expansion of parasite variants that do not thrive in the non-pregnant and to which immunity has not yet completely developed. Importantly, pregnant mothers do eventually develop protective immunity, including antibodies to CSA-binding parasite variants. Remarkably, these antibodies are broadly cross-reactive and recognize placental isolates worldwide (Fried et al., 1998; Ricke et al., 2000; Staalsoe et al., 2001). These findings and initial vaccine developmental studies against PfEMP1 (Gamain et al., 2001a; Baruch et al., 2002; Lekana Douki et al., 2002) have encouraged hope that it will be possible to protectively immunize against the adhesion ligands at the infected erythrocyte surface. Toward the goal of a vaccine or other disease interventions, there have been intensive research efforts to dissect the function of the PfEMP1 protein family. The PfEMP1 binding region consists of multiple receptor-like domains called the Duffy binding-like (DBL) domain and the cysteine-rich interdomain region (CIDR). Although PfEMP1 proteins differ substantially in sequence, adhesion domains can be grouped according to sequence similarity (Smith et al., 2000). Sequence analysis has revealed that PfEMP1 structure and domain architecture is not entirely random (Smith et al., 2000; Gardner et al., 2002a) and might also offer important insights into binding function. For instance, there are now several different DBL-g domains that bind CSA (Buffet et al., 1999; Reeder et al., 2000; Gamain et al., 2002; Vazquez-Macias et al., 2002), and many different CIDR-a type domains were found to bind CD36 (Baruch et al., 1997; Gamain et al., 2001b). However, the total number of PfEMP1 domains that have been functionally characterized is limited and it is not yet possible to predict PfEMP1 binding function from sequence inspection alone. An important question for vaccine design is whether parasite variants that bind the same receptor have similar antigenic or structural features that might be targets of broadly protective immunity. Analysis of large sets of PfEMP1 domains can define the properties required for binding and may be key for vaccine development. Each parasite genome contains approximately 50 var genes, or PfEMP1 proteins (Su et al., 1995; Thompson et al., 1997). Before the Malaria Genome sequencing effort, progress on understanding cytoadherence had been hindered by the large size and complexity of the PfEMP1 protein family. Recently, the nearly complete genome sequence and annotation for the P. falciparum strain 3D7 was published (Gardner et al., 1998; Bowman et al., 1999; Gardner et al., 2002a; Gardner et al., 2002b; Hall et al., 2002; Hyman et al., 2002) setting the stage for genome-wide functional characterizations of PfEMP1 binding. CD36 is

considered a major endothelial receptor for parasite sequestration and this binding property has been mapped to the CIDR1 domain of the PfEMP1 head structure (Baruch et al., 1997). To study how CD36 binding is distributed between PfEMP1 proteins, we investigated the CIDR–CD36 interaction across proteins in the 3D7 genome. This approach defined structural and functional PfEMP1 protein groups that could be systematically distinguished. The implication of these findings for malaria pathogenesis, parasite transmission and vaccine development are discussed.

Results PfEMP1 proteins can be structurally grouped, based upon type of semi-conserved head structure and binding region domain composition In the 3D7 parasite genome, 59 different predicted PfEMP1 proteins were identified along with several pseudogenes (Gardner et al., 2002a). The PfEMP1 extracellular binding region mainly consists of four different types of domains: (i) an N-terminal segment (NTS) at the beginning of the protein; (ii) DBL domains; (iii) CIDR domains; and (iv) C2 domains (Su et al., 1995; Smith et al., 2000). In the publication of the 3D7 genome, the protein architectures of the 3D7 PfEMP1 were described and for this study a sequence analysis of 3D7 PfEMP1 proteins was also performed. To study protein architectures, domain boundaries were defined and CIDR and DBL trees were constructed. For tree construction, PfEMP1 sequences from the 3D7 genome and 14 non-3D7 PfEMP1 proteins analysed previously (Smith et al., 2000) were investigated. There was good agreement between our findings and those reported (Gardner et al., 2002a) so we will mainly discuss some additional observations that were made that provided important insight into PfEMP1 structure and function. Within the DBL tree all domain assignments were similar except for some sequences within the newly recognized DBL-X group (Gardner et al., 2002a). In this study, DBLX sequences were drawn from widely branching limbs including a DBL-b subtype noted previously (Smith et al., 2000) and a group of sequences distinct from the major DBL-d branch (data not shown). Thus, DBL-X sequences included domains that are either tandemly arrayed with a CIDR domain and those that are not (Gardner et al., 2002a). In addition, the CIDR tree included some domains that had previously been characterized for CD36 binding and we noticed that a subset of CIDR-a sequences grouped separately from the main group (bootstrap value 100%) and included the FCR3-CSA CIDR1 domain, which does not bind CD36 (Buffet et al., 1999) (Fig. 1). Sequences © 2003 Blackwell Publishing Ltd, Molecular Microbiology, 47, 1265–1278

Genome-wide characterization of PfEMP1 binding to CD36 1267

Fig. 1. Sequence classification of cysteine-rich interdomain region (CIDR) domains and CD36 binding function. A CIDR tree was constructed based upon Plasmodium falciparum erythrocyte membrane 1 (PfEMP1) from the 3D7 genome (Gardner et al., 2002a) and CIDR domains from the A4var, A4tres, Malayan Camp, FVO and FCR3-CSA PfEMP1 proteins previously characterized for CD36 binding (Gamain et al., 2001b). CIDR domain sequences were clustered according to similarity using the neighbour-joining method. Domains are identified by accession no. and their position in the protein. A circle indicates distinct tree branches, bootstrap support greater than 95%. CIDR sequences are grouped into three major types (a, b, and g) and the CIDR-a1 subtype. The scale bar, 0.1 changes, refers to substitutions per site along a branch. CIDR recombinant proteins were tested for binding to CD36. Blue sequences bound CD36, red sequences did not, and black sequences were not tested. Weaker CD36-binding CIDR sequences are indicated by asterisks.

related to the FCR3-CSA PfEMP1 have been implicated in placental malaria (Buffet et al., 1999; Lekana Douki et al., 2002). Because subtype classification might provide additional predictors of PfEMP1 structure, function, or antigenic similarity the ‘FCRvar CSA-like’ sequences were designated the CIDR-a1 subtype (Fig. 1). Interestingly, nearly all CIDR-a1 subtype sequences were paired © 2003 Blackwell Publishing Ltd, Molecular Microbiology, 47, 1265–1278

downstream of particular DBL sequences, the DBL-a1 subtype, that formed their own branch within the DBL-a type (data not shown) and shared distinct amino acid features. Ten residues, including three cysteines, were highly discriminatory between the DBL-a1 subtype and other DBL-a sequences (Fig. 2). Together, the DBL-1 domain and CIDR-1 domain form

1268 B. Robinson, T. L. Welch and J. D. Smith

L b

.

h– ChA GN LV.

V

C –

RC G+

Fig. 2. Consensus residue features of the DBL-a and DBL-a1 subtype. A multiple alignment was performed on 57 DBL-a and nine DBL-a1 subtype sequences from the 3D7 genome and other parasite genotypes. The 80% consensus for each type is shown with conserved residues indicated by capital letters of the single amino acid code. Residues that were conserved for similar amino acid character are indicated as: c, charge: D, E, H, K, R; +, positive: H, K, R; h, hydrophobic: A, C, F, I, L, M, V, W, Y; p, polar: C, D, E, H, K, N, Q, R, S, T; s, small: A, C, D, G, N, P, S, T, V; u, tiny: A, G, S; b, big: E, K, R, I, L, N, S, Y, W. Within the alignment, periods represent positions that were not conserved and dashes (–) indicate gaps that were introduced to maintain the alignment. Shaded residues were highly discriminatory and present in at least 80% of one type of sequences but absent or present in less than 5% of the other type. Previously defined DBL homology blocks (Smith et al., 2000) are overlined and labelled A to J.

a tandem unit that has been called the PfEMP1 protein semi-conserved head structure (Su et al., 1995; Chen et al., 2000). Based upon sequence classification, distinct PfEMP1 head structures types could be defined (Fig. 3). The main sequence characteristics of the different head structure components are partially summarized in Figs 2 and 4. Overall, the majority of 3D7 PfEMP1 proteins had a type 1 head structure consisting of a DBL1-a and CIDR1-a domain (Fig. 3, and data not shown). In contrast, a subset had a type 2 head structure in which a CIDR-a1 subtype or a CIDR-g sequence replaced the CIDR-a sequence and the DBL-1 domain was frequently the DBLa1 subtype (Fig. 3). Using sequence classification and domain composition, principles have begun to emerge to divide PfEMP1 proteins into different structural groups. In reporting on the 3D7 genome, 16 different types of PfEMP1 binding region architectures were described (Gardner et al., 2002a). Structural groupings could have important implications for function if different domain types encoded distinct binding properties. For this study, head structure classification was additionally applied to examine the 3D7 protein architectures. In the 3D7 genome, 38 out of 59 predicted proteins had a type 1 binding region architecture, consisting of an NTS domain followed by a duplicated arrangement of the DBL/CIDR tandem (Gardner et al., 2002a). All of these had a type 1 head structure (Fig. 3) as did the 3D7 var 2T.2, which has a type 16 binding region architecture (Gardner et al., 2002a). Thus, most small 3D7 PfEMP1 proteins had similar head structure architectures except

for three predicted proteins that substituted two DBL domains, instead (the type 3 binding region architecture) (Gardner et al., 2002a). The remaining 3D7 PfEMP1 proteins were larger and could be categorized into 13 different binding region architectures based upon domain composition (Gardner et al., 2002a). Seven large PfEMP1 proteins had a type 1 head structure, whereas nine had a type 2 head structure (Fig. 3, and data not shown). There was also one predicted large protein, the type 13, which did not have a typical head structure (Gardner et al., 2002a). Thus, head structure was another criterion for classifying PfEMP1 proteins. Because a defect in infected erythrocyte adhesion to CD36 has been mapped to the CIDR1 domain (Gamain et al., 2001b) we next investigated whether CIDR domain or head structure classification had broader implications for CD36 binding function. CIDR sequence classification is highly predictive for CD36 binding CD36 is a pivotal receptor for infected erythrocyte cytoadherence because it is used in most infections but is not recognized by the parasite variants that cause disease in pregnant mothers. Thus, there is a functional dichotomy in parasite adhesion to CD36 that impacts infected erythrocyte organ tropism and disease outcome. Previously, it has been shown that CD36 binding is functionally conserved between diverse CIDR domains (Baruch et al., 1997), but binding could not be predicted nor has it been © 2003 Blackwell Publishing Ltd, Molecular Microbiology, 47, 1265–1278

Genome-wide characterization of PfEMP1 binding to CD36 1269 analysed on a genome-wide scale. For this study, CIDR1 and CIDR2 recombinant proteins from 3D7 PfEMP1 proteins were expressed at the cell surface of Cos-7 cells and tested for binding to CD36-coated dynal magnetic beads.

Binding was quantified as the percentage of transfected cells associated with five or more CD36-coated beads and controlled with non-CD36 coated beads (Table 1; Supplementary material, Table S1).

Fig. 3. Classification of PfEMP1 binding region architectures by head structure, domain composition, and CD36 binding. At the top is a generalized schematic diagram of a PfEMP1 binding region in which DBL and CIDR domains are numbered according to their position in the protein and coloured according to adhesion domain type. The N-terminal segment (NTS) and C2 domain are labelled. Shown are type 1 and type 2 semiconserved head structures defined according to domain composition. Within the 3D7 genome, there were also four predicted PfEMP1 proteins that do not have a classic head structure (Gardner et al., 2002a) not shown. Below the line, the binding region architectures of PfEMP1 proteins tested for CD36 binding in this study are shown. Proteins are divided by the number of domains in the binding region and separated according to head structure (A-C). Individual proteins are identified by accession number. The PFD1235w and MAL7P1.1 sequences are duplicated genes from chromosomes 4 and 7 that have an identical binding region. 3D7 PfEMP1 proteins that had a chimeric structure are indicated by brackets surrounding the region of identity. One protein with a chimeric structure was not tested (nt) for CD36 binding. All sequences shown are predicted to encode proteins except for two pseudogenes (P). © 2003 Blackwell Publishing Ltd, Molecular Microbiology, 47, 1265–1278

Fig. 4. Sequence comparison of CIDR domains that bind and do not bind CD36. At the top is a generalized schematic diagram of a CIDR domain divided into M1, M2, and M3 regions. The M2 region is the minimal CD36 binding region for the Malayan Camp var1 (Baruch et al., 1997) that was mutagenized in this study. The alignment is organized into CD36 binding and non-binding sequences according to sequence type. Sequences in the alignment begin at conserved CIDR cysteine number 9 (C9) located in the M2 region and ends at C12. The area encompasses the region that was mutagenized for this study. Asterisks above the CIDR-a1 subtype sequence alignment highlight discriminatory residues that were present in at least 90% of one type of sequences but were absent or present in less than 5% of the other type. The CS2-CSA sequence is listed as a CD36 binding sequence or as a non-CD36 binding sequence depending on a three-amino-acid substitution from DIE to GHR (Gamain et al., 2001b). The position of the ‘DIE’ motif that supports binding is shown above the CIDR-a alignment. Beneath the alignments are the 90% amino acid consensus and predicted secondary structures (ss). Residues are classified according to Fig. 3 legend.

1270 B. Robinson, T. L. Welch and J. D. Smith

© 2003 Blackwell Publishing Ltd, Molecular Microbiology, 47, 1265–1278

Genome-wide characterization of PfEMP1 binding to CD36 1271 Table 1. Expression and CD36 binding characteristics of CIDR-a constructs. Construct expressed

% Positive cellsa

Intensityb

% Positive cells with beadsc

No. of beads/100 positive cells

PF08_0142 2T.1 PFL0005w PF10_0406 PFL0935c PF13_0364 PFL1955w PFD0625c PFD0995c PF07_0049 PFD0005w 3T.1 2T.2 PFL0020w PFI1830c PFL1960w PF07_0048 MAL6P1.252 PFD1245c PF07_0050 PF10_0001 PFL1950w 3T.2d Ch4.6 Ch7.7 Ch12.8 Ch13.2 Ch8.2 Ch12.7

7.3 ± 6.8 2.2 ± 0.9 21.6 ± 0.6 10.3 ± 5.5 10.7 ± 3.0 7.1 ± 0.4 11.6 ± 2.5 6.4 ± 4.5 14.5 ± 0.8 4.2 ± 0.1 12.3 ± 5.4 5.3 ± 4.0 7.5 ± 2.3 5.7 ± 0.6 6.4 ± 0.0 9.5 ± 1.0 16.4 ± 15.1 4.5 ± 3.3 4.0 ± 0.4 3.3 ± 0.0 5.3 ± 3.9 4.2 ± 1.2 7.3 ± 2.6 10.0 ± 0.4 8.4 ± 5.6 11.5 ± 3.3 7.5 ± 1.6 14.1 ± 10.7 20.5 ± 6.4

Moderate Low Bright Moderate Moderate Low Moderate Moderate Moderate Moderate Bright Moderate Moderate Moderate Moderate Moderate Bright Moderate Moderate Moderate Moderate Low Bright Moderate Moderate Moderate Bright Moderate Very bright

63 ± 14.0 74 ± 1.0 70 ± 6.0 86 ± 6.0 79 ± 4.0 87 ± 6.4 95 ± 2.1 96 ± 3.0 96 ± 6.0 94 ± 3.0 94 ± 9.0 98 ± 3.0 95 ± 5.0 96 ± 0.0 97 ± 2.0 100 ± 0.0 97 ± 4.0 98 ± 3.5 95 ± 1.4 96 ± 6.0 98 ± 2.8 96 ± 0.7 98 ± 2.0 95 ± 4.2 99 ± 0.7 97 ± 4.0 100 ± 0.0 100 ± 0.0 100 ± 0.0

643 ± 191 1102 ± 19 1418 ± 264 1542 ± 40 1612 ± 221 1661 ± 391 1977 ± 334 2077 ± 397 2117 ± 94 2189 ± 550 2351 ± 143 2388 ± 163 2495 ± 511 2328 ± 72 2376 ± 276 2417 ± 63 2530 ± 325 2561 ± 45 2664 ± 559 2779 ± 144 2759 ± 624 2760 ± 247 2829 ± 314 2997 ± 97 2877 ± 414 3075 ± 119 3134 ± 13 3196 ± 805 3362 ± 322

a. The percentage of surface positive transfected cells in 300 DAPI-labelled cells. b. Visual score of the intensity of surface fluorescence using an antibody against the recombinant protein. c. Percentage of 100 transfected cells with five or more CD36-coated beads. d. From a total of eight experiments, the coefficient of binding variation was 0.05. CIDR recombinant proteins were expressed at the surface of Cos-7 cells and tested for binding to CD36-coated beads. Recombinant protein surface expression was confirmed by immunofluorescence. Fluorescent cells with five or more beads attached were considered positive for binding. For all constructs, CD36 binding was controlled with non-CD36-coated beads and was less than 3%. CIDR constructs were tested in batches of 5–10 sequences at a time. The 3T.2 CIDR1-a sequence was included as a control for each batch assay. Results are expressed as the mean and SD of two independent experiments.

An important conclusion was that CIDR sequence classification was highly predictive of CD36 binding. Within the 3D7 PfEMP1 series, CIDR-a type sequences bound CD36, whereas CIDR-b- and CIDR-g-type sequences did not (Fig. 1). Indeed, CIDR-a domains bound CD36 despite extreme sequence divergence (Fig. 4), including the PFL1960w CIDR1-a domain that adhered to CD36 even though it contained a large tandem insertion of ‘GGGGSG’ repeated 16 times. This insert, nearly one-fifth of the residues of the entire domain, occurred just upstream of the M2 area that has been defined as the minimal binding domain for the Malayan Camp CIDR1-a domain (Baruch et al., 1997). Although CIDR polymorphism more typically manifest as small insertions of charged, polar, or small residues (Fig. 4), this large insert illustrates the resiliency of the CIDR module for binding. Overall, many conserved CIDR residues are hydrophobic (Fig. 4) and could serve as a structural core for folding. In contrast, variant residues are typically small or hydro-


philic, suggesting that they are surface exposed and could sometimes occur in loops owing to size variation between CIDR sequences (Fig. 4). The minimal sequence requirements for CIDR domains to fold properly as recombinant proteins to bind CD36 can vary between sequences (Miller et al., 2002b). Systematic removal of repetitive and variable residues may provide a complementary approach to engineer less complex CIDR immunogens for vaccine trials. CIDR-a recombinant proteins could be divided into weak or strong CD36 binders based upon the number of beads associated per cell. The binding measured, although only semi-quantitative, could not always be explained by recombinant protein expression level. For example, the PFL0005w CIDR1-a recombinant protein was a low binder but among the best expressed at the Cos-7 cell surface (Table 1). When mapped on the tree, the weakest binding CIDR domains localized to either of two branches (Fig. 1). It is not clear whether these

1272 B. Robinson, T. L. Welch and J. D. Smith sequences shared related substitutions that led to lower binding or was only coincidental because neighbouring sequences bound CD36 strongly. Therefore, further analysis will be required to investigate differences in binding affinity. By examining the protein context of CIDR domains tested in the binding assay other clues about PfEMP1 structure and function emerged. For instance, CIDR-a type sequences were uniquely localized to the PfEMP1 head structure, a position that may favour their interaction with CD36. In contrast, CIDR-b- or CIDR-g-type sequences did not bind CD36 and were more typically located in the second DBL/CIDR tandem. The different CIDR sequence types had overlapping but also distinct signature residue features, including different numbers of conserved cysteine residues. Some of these changes could contribute to the inability of these domains to bind CD36 as, for instance, the third cysteine in a cysteine-rich motif (CX7-9CX3-5CX3CXCX3WX3KX3EW) was essential for MC var1 CIDR1a binding to CD36 (Baruch et al., 1997), but absent in some CIDR-b type sequences (data not shown). In terms of head structure, all PfEMP1 proteins tested with a type 1 head structure encoded the capacity to bind CD36 while none with a type 2 head structure did. Thus, the PfEMP1 protein head structure distinguished different functional groups. To investigate this functional dichotomy in greater molecular detail, a sequence and binding analysis was performed on CD36-binding and non-binding CIDR1 domains.

Residues contributing to the inability of CIDR domains to bind CD36 map to a region that is divergent between type 1 and type 2 head structures Previously, two different PfEMP1 proteins, the FCR3-CSA and CS2-CSA, have been characterized for their inability to bind CD36 (Gamain et al., 2001b). The FCR3-CSA PfEMP1 has a type 2 head structure containing a CIDR1a1 subtype sequence and the CS2-CSA is the only studied example of a CIDR-a type sequence that does not bind CD36. Heterologous replacement studies between these sequences and the MC var1 CIDR1-a that binds CD36 have demonstrated that critical CD36 binding residues are non-continuous throughout the CIDR domain and include the less conserved, second half of the minimal binding region (Gamain et al., 2001b). To further localize modifications that impact binding, signature residues of the CIDR-1a1 type were substituted from the FCR3-CSA CIDR1-a1 into the MC var1 CIDR1-a. Wild-type and mutant MC var1 CIDR1-a recombinant proteins were expressed at the surface of Cos-7 cells and tested for binding to CD36-coated beads (Fig. 5). Alone, none of the substitutions affected binding. However, a combination of changes reduced the number of CD36-coated beads to 63% of wild type, indicating their contribution to binding. Other differences between CIDR-a and CIDR-a1 sequences mapped to an area located between the sites mutated above that was not amenable to precise mutagenesis owing to lower amino acid conservation that

Fig. 5. CD36 binding properties of wild-type and mutagenized MC var 1 CIDR1-a recombinant proteins. MC CIDR recombinant proteins (aa 401–853 in the protein) were expressed at the surface of Cos-7 cells and tested for binding to CD36 coated beads. The signature residues of CIDR-a and CIDR-a1 subtype sequences are shown from Fig. 4, along with binding results from mutants generated by substituting non-CD36 binding FCR3-CSA CIDR1-a1 residues into the MC CIDR1-a. The percentage of transfected cells binding five or more beads and the number of beads per 100 transfected cells was quantified. It was previously shown that an FCR3-CSA CIDR-a1 recombinant protein did not bind CD36-coated beads (Gamain et al., 2001b) and also found to be the case in this study (data not shown). Transfected cells were identified with an antibody against an epitope tag in the recombinant protein. Results are the mean and standard deviation (SD) of two independent experiments. © 2003 Blackwell Publishing Ltd, Molecular Microbiology, 47, 1265–1278

Genome-wide characterization of PfEMP1 binding to CD36 1273 made it difficult to line up residues to be exchanged. In this region, CIDR-a and CIDR-a1 subtype sequences had a different hydrophobic character and the CIDR-a1 subtype sequences shared a charged lysine and a bulky tryptophan residue that could change the orientation of a predicted alpha helix (Fig. 4). Substitution of this entire region abolished MC var1 CIDR1-a binding to CD36 (Fig. 5), confirming that critical residues reside in this region (Gamain et al., 2001b) and localizing them to a smaller area. Intriguingly, changes to the same predicted helix noted above controlled the CS2-CSA CIDR1-a domain binding to CD36 (Gamain et al., 2001b). A CS2-CSA construct with the residues ‘DIE’ bound CD36, whereas a ‘GHR’ substitution abolished binding. The same ‘GHR’ substitution into MC var1 CIDR1-a reduced binding to 50% of wild type (Gamain et al., 2001b). Although directly implicated in binding, it is clear from the panel of CIDR domains tested that a ‘DIE’ motif is not essential for binding (Fig. 4). Rather the CIDR–CD36 interaction is sensitive to substitutions to this area but binding outcome is influenced not only by the substituted residues but also by the surrounding CIDR framework. Discussion The dual nature of PfEMP1 proteins acting as variant antigens and cytoadherent receptors means that the parasite repertoire is shaped by functional pressures for binding and diversifying pressures to evade immunity. Under this dynamic situation, many parasite cytoadherent properties are highly variant, whereas a few are more common. CD36 is an example of a common infected erythrocyte adhesion property (Barnwell et al., 1985; Ockenhouse et al., 1991a; Newbold et al., 1997). Besides the CIDR1 domain of PfEMP1, other molecules have also been reported to bind or support CD36 adhesion (Ockenhouse et al., 1991b; Crandall et al., 1994; Trenholme et al., 2000). However, PfEMP1 proteins appear to have a critical role in adhesion because non-binding parasite variants express CIDR1 domains that do not bind CD36 (Gamain et al., 2001b). Even though CD36 is considered a major endothelial receptor for infected erythrocyte sequestration, before this study the genome-wide distribution of this property was not known nor was it possible to predict binding from sequence inspection alone. This analysis demonstrates that CIDR sequence classification offers specific predictors for CD36 binding. From a large panel of domains tested, CIDR-a domains bound CD36 and other sequence types did not. Although sequence classification predicted binding, there are exceptions such as CS2-CSA CIDR-a domain that do not bind CD36 (Reeder et al., 1999; Gamain et al., 2001b). © 2003 Blackwell Publishing Ltd, Molecular Microbiology, 47, 1265–1278

Sequence and structure-functional analysis of non-binding domains indicates that critical CD36 binding residues are non-continuous throughout the CIDR1 domain (Gamain et al., 2001b) and highlight a functionally sensitive region that is divergent between type 1 and 2 head structures (Figs 4 and 5). These results can be incorporated into improved algorithms for predicting the CIDR–CD36 interaction. To better understand the binding potential of a parasite genome, a functional genomics map would be useful. Besides CD36, genome-wide domain binding characterizations of CIDR and DBL recombinant proteins have great promise for experimentally annotating PfEMP1 adhesion properties and offer the significant advantage of allowing a higher throughput of analysis. For example, CIDR domains can bind other receptors in addition to CD36. Indeed, Chen and colleagues characterized a PfEMP1 variant, FCR3S1.2, which bound at least five different receptors (Chen et al., 2000). By sequence criteria, the head structure of the FCRS1.2 PfEMP1 was type 1 and the CIDR1-a domain bound CD36, IgM, and CD31 (Chen et al., 2000). Not all parasite variants bind the same broad spectrum of receptors as FCRS1.2, however, panels of CIDR and DBL expression constructs can be used to explore binding conservation. One caveat to defining binding properties in individual domains is that their activity may be modified or suppressed by the surrounding protein context (Gamain et al., 2002). Thus, some binding properties defined in recombinant proteins might be latent properties that require additional mutation or recombination into another protein context to be activated. In general, recombination promotes diversification of adhesive phenotypes among proteins in the family. Although a higher rate has been measured for subtelomeric var genes (Freitas-Junior et al., 2000; Taylor et al., 2000), several chimeric var genes in the 3D7 genome (Fig. 3) were part of two different ‘central’ var gene clusters in chromosome 4 (Hall et al., 2002). Ultimately, high-throughput domain binding studies can be combined with more laborious investigations of receptor-selected infected erythrocyte lines that may be more limited in scope but provide further insight into PfEMP1 structure and function. From its strong conservation, CD36 is a kind of ‘core’ PfEMP1 protein adhesion property that apparently is functionally selected even as immune pressure diversifies the protein family. The CIDR–CD36 binding interaction is remarkable because the CIDR domain tolerates extensive sequence divergence without eliminating CD36 binding function (Baruch et al., 1997; Table 1). In addition, the natural antibody response to CD36 binding parasites is highly variant specific (Bull and Marsh, 2002). This is in contrast with that observed for placental isolates (Fried et al., 1998) and means that different CD36 binding par-

1274 B. Robinson, T. L. Welch and J. D. Smith asite variants can be used to establish repeated and chronic infections. At least one aspect of the functional selection for CD36 binding is sequestration and CD36 was shown to be a critical receptor for infected erythrocyte sequestration in a SCID mouse/human skin transplant sequestration model (Ho et al., 2000). However, CD36 has a widespread cellular distribution and offers the parasite numerous opportunities for host interaction. Besides endothelial cytoadherence, infected erythrocyte adhesion to CD36 causes the downmodulation of the antigen presenting function of dendritic cells in vitro (Urban et al., 1999), activates a respiratory burst in monocytes (Ockenhouse et al., 1989; McGilvray et al., 2000), and bridges clumps of infected cells through platelets (Pain et al., 2001a). The latter is only a property of some CD36 binding parasites and associated with more severe disease (Pain et al., 2001a). Thus, the CIDR–CD36 interaction has complex roles in parasite survival, immune evasion, and malaria pathogenesis. A high degree of CD36 variability exists in individuals from malaria-endemic areas, including complete deficiencies in surface expression. In one study CD36 deficiency was protective (Pain et al., 2001b), but in another was associated with cerebral malaria, perhaps because variants with tropism for cerebral vessels grew out (Aitman et al., 2000). Given the strong genomic investment in CD36 binding, it will be important to perform longitudinal malaria investigations of individuals completely deficient in CD36 surface expression to evaluate infection outcomes and to determine the parasite binding function and form of PfEMP1 variants selected in the absence of CD36. The demonstration that CD36 binding is broadly encoded into different PfEMP1 proteins has implications for how the parasite evolved to enhance transmission opportunities in children and adults while increasing host range with novel binding variants. Plasmodium parasites require both mosquito and vertebrate hosts for their survival. Antigenic variation allows parasites to establish chronic infections and facilitates Plasmodium survival during intermittent transmission periods, such as the dry season, when mosquito numbers are reduced (Kyes et al., 2001). For P. falciparum, these transmission-enhancing benefits are modified because the PfEMP1 variant antigens are also sequestration receptors. By sequestering, mature, infected erythrocytes avoid the spleen and its killing mechanisms (Miller et al., 2002a). However, virulent parasite adhesion traits, such as cerebral sequestration have evolved that can cause host death before parasite transmission. Thus, there are trade-offs in cytoadherence and transmissibility. An additional complication is that malaria transmission requires sexual (gametocyte) forms of the parasite to infect mosquitoes. Early stage gameto-

cyte-infected erythrocytes sequester from blood circulation (Smalley et al., 1981) and are reported to express PfEMP1 proteins at the infected erythrocyte surface (Hayward et al., 1999). Although not directly established that PfEMP1 proteins have a role in gametocyte sequestration, anti-PfEMP1 antibodies could potentially recognize the transmissible parasite forms. It is possible that by sharing the responsibility for sequestration across a protein family encoding core adhesion properties the parasite has greater freedom to experiment with novel binding variants. Core properties, like CD36, may generally favour transmission, even though occasionally associated with more severe disease (Pain et al., 2001a). Simultaneously, rare and latent properties could evolve to exploit specialized niches, such as the placenta, or even predispose to lethal complications. Thus, a distributed cytoadherence responsibility might contribute to the non-pregnant acting as disease reservoirs for the pregnant and the increased transmission fitness of parasite genotypes harbouring lethal adherence variants. At the genomic level, distinct structural and functional groups of PfEMP1 can be defined. An emergent question is whether these groups will have different roles in binding and disease? A striking observation was that the majority of 3D7 PfEMP1 are small and predicted to bind CD36. It will be interesting to test whether these variants typify most infections not causing severe disease. In contrast, large 3D7 PfEMP1 were less common, but have more complex adhesion domain composition including DBL-b and DBL-g domains that have been linked to disease. Intriguingly, there is a report that individuals with cerebral malaria are infected by parasites expressing larger molecular weight PfEMP1 than those with non-severe disease (Bian and Wang, 2000) but more study is needed. Other types of PfEMP1 in the 3D7 genome did not have a classic head structure (Gardner et al., 2002a) or had a type 2 head structure that would be predicted not to bind CD36. These may be under selection for other adherence traits. The structural and functional grouping of PfEMP1 proteins opens new avenues to investigate the PfEMP1 variants contributing to malaria disease. To define var/ PfEMP1 proteins contributing to disease a combination of gene expression profiling and antibody studies have been used (Khattab et al., 2001; Fried and Duffy, 2002; Lekana Douki et al., 2002; Rowe et al., 2002; Vazquez-Macias et al., 2002). One difficulty of interpreting var profiling studies has been to know when an amplified ‘sequence tag’ marks a PfEMP1 protein that is actively participating in sequestration or when it derives from circulating, immature, infected erythrocytes (Chen et al., 1998; Kyes et al., 2000). Knowledge about the predicted binding properties of var sequences can aid analysis in several ways. In


Genome-wide characterization of PfEMP1 binding to CD36 1275 some cases, it may be used to predict the binding properties of a tagged gene and used to prioritize tags for more complete cloning and additional functional and immunological characterizations. In other instances, it could give new insight into disease when binding is not known. As an example, an unusual form of DBL-a tags was detected in severe malaria infections of the non-pregnant (Kirchgatter and Portillo Hdel, 2002). Based upon their characteristics, these tags were DBL-a1 subtype, which could implicate non-CD36 binding PfEMP1 proteins. Finally, if the binding properties of parasite variants are known then this can be used to work backwards to predict var gene profiles and test for them. For instance, placental variants do not bind CD36 but do bind CSA. Already, var genes with type 2 head structures have been detected in infected placentas (Rowe et al., 2002) and others could possibly contribute because of their ability to bind CSA (Gamain et al., 2002). In the 3D7 genome, PfEMP1 with a type 2 head structure are associated with one of two largely distinct 5¢-UTR sequences. The first of these has been associated with placental isolates (Vazquez-Macias et al., 2002) but the possible involvement of the other in placental malaria has not been reported. Thus, probes can be developed to deliberately investigate the role of PfEMP1 protein functional groups in malaria disease. To date, many vaccine efforts focused on pregnancy malaria have concentrated on the DBL-g domain because of its role in CSA binding (Buffet et al., 1999; Reeder et al., 2000; Khattab et al., 2001; Gamain et al., 2002; Lekana Douki et al., 2002). However, it was recently reported that PfEMP1 that do not bind CD36 might have a different protein structure from those that bind CD36 (Gamain et al., 2002). This interpretation was based upon an observation that CD36 and CSA binding were mutually exclusive properties of a combined recombinant protein containing both CIDR1 and DBL2-g domains. Remarkably, a very subtle three amino acid substitution in the CIDR1 domain regulated binding for both domains (Gamain et al., 2002). This modification occurred at a site that was distinct between type 1 and 2 head structures. Thus, although more structural investigation of PfEMP1 proteins is needed, it is possible that proteins with type 2 head structures have evolved a slightly different protein structure under selection for binding properties that are distinct and incompatible with CD36 binding head structures. This would have important implications for vaccine design. In conclusion, PfEMP1 proteins are responsible for diverse binding activities of infected erythrocytes. Adhesive domain sequence classification is a powerful tool to investigate PfEMP1 structure and function. PfEMP1 structural and functional groupings offer a new approach to understand the function of this protein family and to potentially prevent disease caused by cytoadherence.


Experimental procedures PfEMP1 sequence analysis To obtain a representation of the Plasmodium falciparum erythrocyte membrane 1 (PfEMP1) protein family from the 3D7 parasite genome sequenced by Stanford, Sanger, and TIGR sequencing facilities, BLAST analysis was used to identify genes initially. During the latter half of the genome sequencing effort and again at the completion of the malaria genome publications, chromosomal contigs were downloaded from the Plasmodium genome consortium PlasmoDB website (http://PlasmoDB.org) and analysed by the GENEQUEST program within DNASTAR. Our analysis of PfEMP1 proteins and domain architecture was compared with the genome publication (Gardner et al., 2002a) and var genes were named according to the published accession number. Plasmodium falciparum erythrocyte membrane 1 domain boundaries were defined using published criteria (Su et al., 1995; Smith et al., 2000). DBL and CIDR trees were constructed using CLUSTALX for multiple alignments and PAUP*4.0b8 (*Phylogenetic Analysis Using PARSIMONY and other methods) for tree generation. Trees were constructed and compared using both the neighbour-joining and maximum parsimony methods, accompanied by bootstrap analysis with 1000 replicates. Conserved amino acid features were identified using the CONSENSUS program at (http://www. bork.embl-heidelberg.de/Alignment/consensus.html) and secondary structure predictions at the Predator website (ftp:// ftp.edi.ac.uk/pub/software/dos/predator).

CIDR domain expression constructs and mutagenesis Cysteine-rich interdomain region (CIDR) constructs were amplified from 3D7 strain genomic DNA by polymerase chain reaction (PCR) and cloned into the T8(12CA5) vector (Affymax). The T812CA5 vector provides a signal sequence and a glycoslyphosphoinositol (GPI) anchor for mammalian cell surface expression and epitope tags for recognition by the 12CA5 and 179 mAbs (Whitehorn et al., 1995). Oligos were designed to express full-length CIDR domains. If the original CIDR construct was not surface positive, or there were not optimal sites for priming because of ‘AT’ richness typical of malaria genes, smaller versions of the full-length domain were constructed. Domain and construct boundaries are in Supplementary material, Table S2. The MCvar1/FCR3-CSA chimeric CIDR construct was generated by amplifying each part separately using PfuTurbo (Strategene), ligating the PCR products, and amplifying from the ligation mix to clone the chimeric product in frame into the T8 (12CA5) vector. CIDR site-directed mutants were produced with the Quikchange XL site-directed mutagenesis kit (Stratagene). All CIDR constructs, mutant and wild type, were sequenced in their entirety.

CIDR-CD36 binding assay Cysteine-rich interdomain region recombinant proteins were expressed at the surface of transiently transfected Cos-7 cells grown on coverslips. Binding was assessed by incubat-

1276 B. Robinson, T. L. Welch and J. D. Smith ing transfected cells with dynal magnetic beads coated with a soluble CD36 recombinant protein. After incubation, cells were washed by flipping the coverslip and letting unbound beads settle by gravity. Co-localization of CD36-beads and transfected cells was assessed using mAb 179 against an epitope tag in the CIDR recombinant protein. For immunofluorescence, cells were fixed with a 2% paraformaldehyde/ phosphate-buffered saline (PBS) solution (Electron Microscopy Sciences). Binding was quantified by counting the percentage of transfected cells associated with five or more CD36-beads and number of beads per 100 transfected cells. Transfection efficiency was measured by staining Cos-7 nuclei with 4¢,6-diamadino-2-phenylindole (DAPI) (Sigma) and quantifying the percentage of transfected cells in 300 DAPI cells. Surface expression of constructs was visually inspected and given an overall score of low, moderate or bright.

Acknowledgements The authors thank Sue Kraemer and Leia Smith for critical reading of the manuscript, and Dror Baruch for helpful comments on the manuscript and supplying the soluble CD36 recombinant protein. In addition, we thank Erik Whitehorn of Affymax Research Institute for providing the T8(12CA5) vector and mAb 179. We also wish to thank the scientists and funding agencies comprising the international Malaria Genome Project for making sequence data from the genome of P. falciparum (3D7) public before publication of the completed sequence. The Sanger Centre (UK) provided sequence for chromosomes 1, 3–9 and 13, with financial support from the Wellcome Trust. A consortium composed of The Institute for Genome Research, along with the Naval Medical Research Center (USA), sequenced chromosomes 2, 10, 11 and 14, with support from NIAID/NIH, the Burroughs Wellcome Fund, and the Department of Defense. The Stanford Genome Technology Center (USA) sequenced chromosome 12, with support from the Burroughs Wellcome Fund. The Plasmodium Genome Database is a collaborative effort of investigators at the University of Pennsylvania (USA) and Melbourne University (Australia), supported by the Burroughs Wellcome Fund. This work was supported by an Ellison Medical Foundation New Investigator Award in Infectious Disease and an award from the National Institutes of Health Grant (RO1 AI47953–01A1).

Supplementary material The following material is available from http://www. blackwellpublishing.com/products/journals/suppmat/ mole/mole3378/mmi3378sm.htm Table S1. Expression and CD36 binding characteristics of CIDR-a1, -b and -g constructs. Table S2. Domain and construct boundaries of CIDR recombinant proteins.

References Achur, R.N., Valiyaveettil, M., Alkhalil, A., Ockenhouse, C.F., and Gowda, D.C. (2000) Characterization of proteoglycans

of human placenta and identification of unique chondroitin sulfate proteoglycans of the intervillous spaces that mediate the adherence of Plasmodium falciparum-infected erythrocytes to the placenta. J Biol Chem 275: 40344– 40356. Aitman, T.J., Cooper, L.D., Norsworthy, P.J., Wahid, F.N., Gray, J.K., Curtis, B.R., et al. (2000) Malaria susceptibility and CD36 mutation. Nature 405: 1015–1016. Alkhalil, A., Achur, R.N., Valiyaveettil, M., Ockenhouse, C.F., and Gowda, D.C. (2000) Structural requirements for the adherence of Plasmodium falciparum-infected erythrocytes to chondroitin sulfate proteoglycans of human placenta. J Biol Chem 275: 40357–40364. Barnwell, J.W., Ockenhouse, C.F., and Knowles, D.M. (1985) Monoclonal antibody OKM5 inhibits the in vitro binding of Plasmodium falciparum-infected erythrocytes to monocytes, endothelial, and C32 melanoma cells. J Immunol 135: 3494–3497. Baruch, D.I., Ma, X.C., Singh, H.B., Bi, X., Pasloske, B.L., and Howard, R.J. (1997) Identification of a region of PfEMP1 that mediates adherence of Plasmodium falciparum infected erythrocytes to CD36: conserved function with variant sequence. Blood 90: 3766–3775. Baruch, D.I., Gamain, B., Barnwell, J.W., Sullivan, J.S., Stowers, A., Galland, G.G., et al. (2002) Immunization of Aotus monkeys with a functional domain of the Plasmodium falciparum variant antigen induces protection against a lethal parasite line. Proc Natl Acad Sci USA 99: 3860–3865. Beeson, J.G., Brown, G.V., Molyneux, M.E., Mhango, C., Dzinjalamala, F., and Rogerson, S.J. (1999) Plasmodium falciparum isolates from infected pregnant women and children are associated with distinct adhesive and antigenic properties. J Infect Dis 180: 464–472. Beeson, J.G., Rogerson, S.J., Cooke, B.M., Reeder, J.C., Chai, W., Lawson, A.M., et al. (2000) Adhesion of Plasmodium falciparum-infected erythrocytes to hyaluronic acid in placental malaria. Nat Med 6: 86–90. Bian, Z., and Wang, G. (2000) Antigenic variation and cytoadherence of PfEMP1 of Plasmodium falciparum-infected erythrocyte from malaria patients. Chin Med J (Engl) 113: 981–984. Bowman, S., Lawson, D., Basham, D., Brown, D., Chillingworth, T., Churcher, C.M., et al. (1999) The complete nucleotide sequence of chromosome 3 of Plasmodium falciparum. Nature 400: 532–538. Buffet, P.A., Gamain, B., Scheidig, C., Baruch, D., Smith, J.D., Hernandez-Rivas, R., et al. (1999) Plasmodium falciparum domain mediating adhesion to chondroitin sulfate A: a receptor for human placental infection. Proc Natl Acad Sci USA 96: 12743–12748. Bull, P.C., and Marsh, K. (2002) The role of antibodies to Plasmodium falciparum-infected-erythrocyte surface antigens in naturally acquired immunity to malaria. Trends Microbiol 10: 55–58. Chen, Q., Fernandez, V., Sundstrom, A., Schlichtherle, M., Datta, S., Hagblom, P., et al. (1998) Developmental selection of var gene expression in Plasmodium falciparum. Nature 394: 392–395. Chen, Q., Heddini, A., Barragan, A., Fernandez, V., Pearce, S.F., and Wahlgren, M. (2000) The semiconserved head structure of Plasmodium falciparum erythrocyte membrane © 2003 Blackwell Publishing Ltd, Molecular Microbiology, 47, 1265–1278

Genome-wide characterization of PfEMP1 binding to CD36 1277 protein 1 mediates binding to multiple independent host receptors. J Exp Med 192: 1–10. Crandall, I., Land, K.M., and Sherman, I.W. (1994) Plasmodium falciparum: pf adhesin and CD36 form an adhesin/ receptor pair that is responsible for the pH-dependent portion of cytoadherence/sequestration. Exp Parasitol 78: 203–209. Flick, K., Scholander, C., Chen, Q., Fernandez, V., Pouvelle, B., Gysin, J., et al. (2001) Role of nonimmune IgG bound to PfEMP1 in placental malaria. Science 293: 2098–2100. Freitas-Junior, L.H., Bottius, E., Pirrit, L.A., Deitsch, K.W., Scheidig, C., Guinet, F., et al. (2000) Frequent ectopic recombination of virulence factor genes in telomeric chromosome clusters of P. falciparum. Nature 407: 1018–1022. Fried, M., and Duffy, P.E. (1996) Adherence of Plasmodium falciparum to chondroitin sulfate A in the human placenta. Science 272: 1502–1504. Fried, M., and Duffy, P.E. (2002) Two DBLgamma subtypes are commonly expressed by placental isolates of Plasmodium falciparum. Mol Biochem Parasitol 122: 201–210. Fried, M., Nosten, F., Brockman, A., Brabin, B.J., and Duffy, P.E. (1998) Maternal antibodies block malaria. Nature 395: 851–852. Gamain, B., Miller, L.H., and Baruch, D.I. (2001a) The surface variant antigens of Plasmodium falciparum contain cross-reactive epitopes. Proc Natl Acad Sci USA 98: 2664–2669. Gamain, B., Smith, J.D., Miller, L.H., and Baruch, D.I. (2001b) Modifications in the CD36 binding domain of the Plasmodium falciparum variant antigen are responsible for the inability of chondroitin sulfate A adherent parasites to bind CD36. Blood 97: 3268–3274. Gamain, B., Gratepanche, S., Miller, L.H., and Baruch, D.I. (2002) Molecular basis for the dichotomy in Plasmodium falciparum adhesion to CD36 and chondroitin sulfate A. Proc Natl Acad Sci USA 99: 10020–10024. Gardner, M.J., Tettelin, H., Carucci, D.J., Cummings, L.M., Aravind, L., Koonin, E.V., et al. (1998) Chromosome 2 sequence of the human malaria parasite Plasmodium falciparum. Science 282: 1126–1132. Gardner, M.J., Hall, N., Fung, E., White, O., Berriman, M., Hyman, R.W., et al. (2002a) Genome sequence of the human malaria parasite Plasmodium falciparum. Nature 419: 498–511. Gardner, M.J., Shallom, S.J., Carlton, J.M., Salzberg, S.L., Nene, V., Shoaibi, A., et al. (2002b) Sequence of Plasmodium falciparum chromosomes 2, 10, 11 and 14. Nature 419: 531–534. Hall, N., Pain, A., Berriman, M., Churcher, C., Harris, B., Harris, D., et al. (2002) Sequence of Plasmodium falciparum chromosomes 1, 3–9 and 13. Nature 419: 527–531. Hayward, R.E., Tiwari, B., Piper, K.P., Baruch, D.I., and Day, K.P. (1999) Virulence and transmission success of the malarial parasite Plasmodium falciparum. Proc Natl Acad Sci USA 96: 4563–4568. Ho, M., Hickey, M.J., Murray, A.G., Andonegui, G., and Kubes, P. (2000) Visualization of Plasmodium falciparum– endothelium interactions in human microvasculature: mimicry of leukocyte recruitment. J Exp Med 192: 1205– 1211. Hyman, R.W., Fung, E., Conway, A., Kurdi, O., Mao, J., © 2003 Blackwell Publishing Ltd, Molecular Microbiology, 47, 1265–1278

Miranda, M., et al. (2002) Sequence of Plasmodium falciparum chromosome 12. Nature 419: 534–537. Khattab, A., Kun, J., Deloron, P., Kremsner, P.G., and Klinkert, M.Q. (2001) Variants of Plasmodium falciparum erythrocyte membrane protein 1 expressed by different placental parasites are closely related and adhere to chondroitin sulfate A. J Infect Dis 183: 1165–1169. Kirchgatter, K., and Portillo Hdel, A. (2002) Association of severe noncerebral Plasmodium falciparum malaria in Brazil with expressed PfEMP1 DBL1 alpha sequences lacking cysteine residues. Mol Med 8: 16–23. Kyes, S., Pinches, R., and Newbold, C. (2000) A simple RNA analysis method shows var and rif multigene family expression patterns in Plasmodium falciparum. Mol Biochem Parasitol 105: 311–315. Kyes, S., Horrocks, P., and Newbold, C. (2001) Antigenic variation at the infected red cell surface in malaria. Annu Rev Microbiol 55: 673–707. Lekana Douki, J.B., Traore, B., Costa, F.T., Fusai, T., Pouvelle, B., Sterkers, Y., et al. (2002) Sequestration of Plasmodium falciparum-infected erythrocytes to chondroitin sulfate A, a receptor for maternal malaria: monoclonal antibodies against the native parasite ligand reveal pan-reactive epitopes in placental isolates. Blood 100: 1478–1483. McGilvray, I.D., Serghides, L., Kapus, A., Rotstein, O.D., and Kain, K.C. (2000) Nonopsonic monocyte/macrophage phagocytosis of Plasmodium falciparum-parasitized erythrocytes: a role for CD36 in malarial clearance. Blood 96: 3231–3240. Miller, L.H., Baruch, D.I., Marsh, K., and Doumbo, O.K. (2002a) The pathogenic basis of malaria. Nature 415: 673–679. Miller, L.H., Hudson-Taylor, D., Gamain, B., and Saul, A.J. (2002b) Definition of the minimal domain of CIDR1alpha of Plasmodium falciparum PfEMP1 for binding CD36. Mol Biochem Parasitol 120: 321–323. Newbold, C., Warn, P., Black, G., Berendt, A., Craig, A., Snow, B., et al. (1997) Receptor-specific adhesion and clinical disease in Plasmodium falciparum. Am J Trop Med Hyg 57: 389–398. Ockenhouse, C.F., Magowan, C., and Chulay, J.D. (1989) Activation of monocytes and platelets by monoclonal antibodies or malaria-infected erythrocytes binding to the CD36 surface receptor in vitro. J Clin Invest 84: 468–475. Ockenhouse, C.F., Ho, M., Tandon, N.N., Van Seventer, G.A., Shaw, S., White, N.J., et al. (1991a) Molecular basis of sequestration in severe and uncomplicated Plasmodium falciparum malaria: differential adhesion of infected erythrocytes to CD36 and ICAM-1. J Infect Dis 164: 163–169. Ockenhouse, C.F., Klotz, F.W., Tandon, N.N., and Jamieson, G.A. (1991b) Sequestrin, a CD36 recognition protein on Plasmodium falciparum malaria- infected erythrocytes identified by anti-idiotype antibodies. Proc Natl Acad Sci USA 88: 3175–3179. Pain, A., Ferguson, D.J., Kai, O., Urban, B.C., Lowe, B., Marsh, K., et al. (2001a) Platelet-mediated clumping of Plasmodium falciparum-infected erythrocytes is a common adhesive phenotype and is associated with severe malaria. Proc Natl Acad Sci USA 98: 1805–1810. Pain, A., Urban, B.C., Kai, O., Casals-Pascual, C., Shafi, J., Marsh, K., et al. (2001b) A non-sense mutation in Cd36

1278 B. Robinson, T. L. Welch and J. D. Smith gene is associated with protection from severe malaria. Lancet 357: 1502–1503. Reeder, J.C., Cowman, A.F., Davern, K.M., Beeson, J.G., Thompson, J.K., Rogerson, S.J., et al. (1999) The adhesion of Plasmodium falciparum-infected erythrocytes to chondroitin sulfate A is mediated by P. falciparum erythrocyte membrane protein 1. Proc Natl Acad Sci USA 96: 5198–5202. Reeder, J.C., Hodder, A.N., Beeson, J.G., and Brown, G.V. (2000) Identification of glycosaminoglycan binding domains in Plasmodium falciparum erythrocyte membrane protein 1 of a chondroitin sulfate A-adherent parasite. Infect Immun 68: 3923–3926. Ricke, C.H., Staalsoe, T., Koram, K., Akanmori, B.D., Riley, E.M., Theander, T.G., et al. (2000) Plasma antibodies from malaria-exposed pregnant women recognize variant surface antigens on Plasmodium falciparum-infected erythrocytes in a parity-dependent manner and block parasite adhesion to chondroitin sulfate A. J Immunol 165: 3309– 3316. Rowe, J.A., Kyes, S.A., Rogerson, S.J., Babiker, H.A., and Raza, A. (2002) Identification of a conserved Plasmodium falciparum var gene implicated in malaria in pregnancy. J Infect Dis 185: 1207–1211. Smalley, M.E., Abdalla, S., and Brown, J. (1981) The distribution of Plasmodium falciparum: the peripheral blood and bone marrow of Gambian children. Trans R Soc Trop Med Hyg 75: 103–105. Smith, J.D., Subramanian, G., Gamain, B., Baruch, D.I., and Miller, L.H. (2000) Classification of adhesive domains in the Plasmodium falciparum erythrocyte membrane protein 1 family. Mol Biochem Parasitol 110: 293–310. Smith, J.D., Gamain, B., Baruch, D.I., and Kyes, S. (2001) Decoding the language of var genes and Plasmodium falciparum sequestration. Trends Parasitol 17: 538–545. Staalsoe, T., Megnekou, R., Fievet, N., Ricke, C.H., Zornig, H.D., Leke, R., et al. (2001) Acquisition and decay of antibodies to pregnancy-associated variant antigens on the

surface of Plasmodium falciparum-infected erythrocytes that protect against placental parasitemia. J Infect Dis 184: 618–626. Steketee, R.W., Wirima, J.J., Slutsker, L., Heymann, D.L., and Breman, J.G. (1996) The problem of malaria and malaria control in pregnancy in sub-Saharan Africa. Am J Trop Med Hyg 55: 2–7. Su, X.Z., Heatwole, V.M., Wertheimer, S.P., Guinet, F., Herrfeldt, J.A., Peterson, D.S., et al. (1995) The large diverse gene family var encodes proteins involved in cytoadherence and antigenic variation of Plasmodium falciparuminfected erythrocytes. Cell 82: 89–100. Taylor, H.M., Kyes, S.A., and Newbold, C.I. (2000) Var gene diversity in Plasmodium falciparum is generated by frequent recombination events. Mol Biochem Parasitol 110: 391–397. Thompson, J.K., Rubio, J.P., Caruana, S., Brockman, A., Wickham, M.E., and Cowman, A.F. (1997) The chromosomal organization of the Plasmodium falciparum var gene family is conserved. Mol Biochem Parasitol 87: 49–60. Trenholme, K.R., Gardiner, D.L., Holt, D.C., Thomas, E.A., Cowman, A.F., and Kemp, D.J. (2000) clag9: a cytoadherence gene in Plasmodium falciparum essential for binding of parasitized erythrocytes to CD36. Proc Natl Acad Sci USA 97: 4029–4033. Urban, B.C., Ferguson, D.J., Pain, A., Willcox, N., Plebanski, M., Austyn, J.M., et al. (1999) Plasmodium falciparuminfected erythrocytes modulate the maturation of dendritic cells. Nature 400: 73–77. Vazquez-Macias, A., Martinez-Cruz, P., Castaneda-Patlan, M.C., Scheidig, C., Gysin, J., Scherf, A., et al. (2002) A distinct 5¢ flanking var gene region regulates Plasmodium falciparum variant erythrocyte surface antigen expression in placental malaria. Mol Microbiol 45: 155–167. Whitehorn, E.A., Tate, E., Yanofsky, S.D., Kochersperger, L., Davis, A., Mortensen, R.B., et al. (1995) A generic method for expression and use of ‘tagged’ soluble versions of cell surface receptors. Biotechnology (NY) 13: 1215–1219.
