ABSTRACT
Pregnancy-associated malaria is a severe clinical syndrome associated with the sequestration of Plasmodium falciparum-infected erythrocytes in the placenta. Placental binding is mediated by VAR2CSA, a member of the large and diverse P. falciparum erythrocyte membrane 1 (PfEMP1) protein family. To better understand if conserved regions in VAR2CSA can be targeted by antibodies, we immunized rabbits with VAR2CSA-DBL1 and -DBL5 recombinant proteins produced in Pichia pastoris and developed a panel of seven chondroitin sulfate A (CSA)-binding parasites from diverse geographic origins. Overall, no two parasites in the panel expressed the same VAR2CSA sequence. The DBL1 domains averaged 80% amino acid identity (range, 72 to 89%), and the DBL5 domains averaged 86% amino acid identity (range, 83 to 99%), similar to a broader sampling of VAR2CSA sequences from around the world. Whereas antibodies generated against the VAR2CSA-DBL1 recombinant protein had only limited breadth and reacted with three or four parasites in the panel, immunization with DBL5 recombinant proteins elicited broadly cross-reactive antibodies against all or most parasites in the panel, as well as to fresh clinical isolates from pregnant women. These findings demonstrate that the major PfEMP1 variant expressed by placental isolates exposes strain-transcendent epitopes that can be targeted by vaccination and may have application for pregnancy malaria vaccine development.
In regions where Plasmodium falciparum is endemic, pregnant women are at increased risk for malaria, especially during the first pregnancy (12, 54). Pregnancy-associated malaria (PAM) is characterized by the selective accumulation of P. falciparum-infected erythrocytes (IEs) in the placental microvasculature, primarily mediated by IE adhesion to chondroitin sulfate A (CSA) (9, 23). Prior to their first pregnancy, women generally lack antibodies to CSA-binding isolates, which suggests that these are novel variants to which women have limited previous exposure. Over successive pregnancies, women acquire antibodies to the surface antigens expressed by placental IEs, which correlate with improved pregnancy outcomes (7, 25, 33, 44). Antibodies are thought to protect against placental malaria by inhibiting IE binding to CSA (17, 25, 40, 44) or opsonizing IEs for phagocytosis (28, 29, 34, 52).
Placental binding of IEs has been linked to the expression of a unique member of the P. falciparum erythrocyte membrane protein 1 (PfEMP1) family, termed VAR2CSA (45, 46). PfEMP1s are a large and diverse family of clonally variant adhesion proteins, which are expressed in a mutually exclusive fashion at the IE surface (6, 50, 53). Placental isolates and CSA-binding laboratory parasite lines have upregulated var2csa expression (15, 32, 45, 46, 58), and CSA-binding parasites are used as an in vitro model for placental binding isolates. Parasites in which var2csa is genetically disrupted lose the ability to bind to CSA (16, 60), suggesting that VAR2CSA is the only or the major PfEMP1 variant associated with CSA binding. Furthermore, VAR2CSA is a primary target of antibodies at the surface of placental IEs (4), and antibodies to VAR2CSA correlate with improved pregnancy outcomes (45). Taken together, these findings suggest it may be possible to develop a vaccine to protect women from placental malaria, but a key issue is whether VAR2CSA displays conserved epitopes that could form the basis for a vaccine.
VAR2CSA is a large, polymorphic protein (∼300 to 350 kDa), and therefore, a major challenge for vaccine development will be to overcome antigenic diversity. Whereas var2csa is unusually conserved for the var gene family, sequences ranged between 75 and 83% amino acid identity in global sequence comparisons (11, 56). Sequence comparisons have also revealed extensive gene mosaicism, which may contribute to antigenic cross-reactivity between different CSA-binding isolates (3, 8, 10, 20, 25). Although polyclonal plasma and human monoclonal antibodies derived from pregnant women appear to be highly focused on polymorphic regions in VAR2CSA (3, 4, 37, 39), epitope mapping has suggested that conserved regions in some of the VAR2CSA Duffy binding-like (DBL) domains may be accessible to antibodies (1). However, it is not known if these conserved regions are exposed in all VAR2CSA variants or if they can be developed as vaccine targets. Because of its large size, it has not been technically possible to express the complete VAR2CSA extracellular domain as a recombinant protein, and instead, vaccine development has been focused on individual DBL domains (46). It is not yet clear which of the six DBL domains in VAR2CSA would make the best vaccine targets. Prior studies have shown that immunization with Baculovirus-expressed or refolded bacterially expressed DBL5 recombinant proteins elicited cross-reactive antibodies to CSA-binding or placental isolates (32, 38). These findings suggest the potential exposure of strain-transcendent antibody epitopes in native VAR2CSA protein, but it has been difficult to confirm the extent of isolate transcendence because previous studies focused on only a limited number of CSA-binding lines or clinical isolates in which VAR2CSA was not cloned and sequenced.
To investigate if conserved regions in VAR2CSA can be targeted by antibodies, we immunized rabbits with VAR2CSA-DBL1 or -DBL5 recombinant proteins produced in Pichia pastoris by employing new, longer construct boundaries that included additional cysteine residues predicted to be involved in DBL domain disulfide bonding (2, 27, 30, 48). Rabbit plasma samples were assayed against a panel of seven CSA-binding parasite lines from different geographic origins to assess the breadth of antibody reactivity. In contrast to DBL1, immunization with DBL5 elicited broadly cross-reactive antibodies against diverse CSA-binding parasite lines. These findings demonstrate the existence of a strain-transcendent antibody epitope(s) in VAR2CSA-DBL5, which may present opportunities for PAM vaccine development.
MATERIALS AND METHODS
Design of DBL synthetic genes.Synthetic genes were constructed by GenScript Corporation (Piscataway, NJ), and codons were optimized for P. pastoris expression. N-Glycosylation sites were removed by converting asparagine to glutamine, and DNA sequences carrying more than five adenine nucleotides in a row were mutated to avoid any premature termination without changing coding features.
Recombinant protein expression in Pichia pastoris.Cloning of recombinant proteins was done as previously described (3). In brief, VAR2CSA constructs with a His6 tag on the C terminus were amplified from synthetic genes and cloned into SnaBI and NotI restriction sites in the pPIC9K vector (Invitrogen). Constructs were digested with SacI and electroporated into P. pastoris strain GS115. VAR2CSA construct boundaries are indicated in Fig. 1. The IT4var22-DBL3 (GenBank accession number EF158076) recombinant protein went from amino acids G1179 to N1487, numbering from the first methionine in the protein. For protein production, P. pastoris was grown overnight at 20°C with shaking at 250 rpm in 0.9 liter buffered complex medium (BM; 1% yeast extract, 2% peptone, 1% yeast nitrogen base, 1 M potassium phosphate buffer [pH 6.0]) plus 2% glycerol (BMG) and shifted to 0.3 liter BM plus 0.5% methanol for protein induction. His-tagged recombinant proteins were harvested from supernatants on day 4 or 5 by using nickel resin or cobalt-nitrilotriacetic acid-agarose (Sigma-Aldrich). Recombinant proteins were analyzed in 4 to 20% SDS-PAGE gels under reduced or nonreduced conditions. Gels were stained with GelCode blue reagent or transferred to a nitrocellulose membrane and detected by Western blotting using anti-His tag antibodies (Invitrogen). Protein concentrations were determined by Bradford assay (500-0205; Bio-Rad). The identities of recombinant proteins were confirmed by mass spectrometry analyses. Purified proteins were stored at −80°C in 1× phosphate-buffered saline (PBS).
Expression of VAR2CSA-DBL5 recombinant proteins in P. pastoris. (A) Protein schematic of VAR2CSA. The original DBL domain boundaries are numbered and indicated by black rectangles. In gray are new, longer domain boundaries predicted by X-ray crystallographic analysis of DBL disulfide bonds (27, 30). The first and last amino acids in VAR2CSA constructs are indicated below the protein schematic. The arrowhead indicates that the 3D7-DBL5 protein was truncated in the N terminus at the amino acid position 95, which resulted in a loss of ∼10 kDa of the protein. One microgram of recombinant proteins was analyzed under nonreducing conditions in a 4 to 20% SDS-PAGE gel and stained by GelCode blue reagent (B) or detected by Western blotting using anti-His tag antibodies (C). MW, molecular weight.
Animal immunization.Immunizations were performed at R&R Rabbitry (Washington) according to animal immunization guidelines. In brief, rabbits received 25 to 100 μg recombinant protein in complete Freund's adjuvant for the first immunization and were boosted four times with the same amount of recombinant protein in incomplete Freund's adjuvant. Preimmune and immune rabbit plasma samples were heat inactivated for 45 min at 57°C and stored at −20°C.
Parasite lines. Plasmodium falciparum parasites were grown in O+ erythrocytes and 10% human plasma. CSA-binding laboratory isolates FCR3/IT4-CSA (47), CS2-CSA (41), 7G8-CSA (3), HB3-CSA (3), 3D7-CSA and NF54-CSA (3), and Pf2004-CSA (18) and Pf2006-CSA (18) are described in Table 1 and were maintained by periodic selection on CSA. The HB3 parasite encodes two var2csa gene copies (31). Parasite lines predominantly expressing the HB3 var2csa allele A or the HB3 var2csa B allele were cloned from a mixed HB3-CSA parental line by flow cytometric sorting. For non-CSA binding controls, A4ultra and ItG-ICAM-1 were employed. A4ultra expresses the IT4var14 gene, which was maintained by selection with monoclonal antibody (MAb) BC6 (gift from Chris Newbold, Oxford, United Kingdom). ItG-ICAM-1 expresses IT4var16, which was maintained by selection on intercellular adhesion molecule 1. The A4ultra and ItG-ICAM-1 lines are isogenic with FCR3 and CS2. Genotyping of parasites in the reference CSA-binding panel was done with MSP1/MSP2 primers according to previously published approaches (51). The recently adapted clinical isolates 736, 755, and 711 were isolated in Tanzania from the peripheral blood of pregnant women who had placental malaria. Samples were collected at the time of delivery, frozen, and used within a few cycles of in vitro culture without CSA selection, with the exception that the 736 and 711 isolates were selected once by panning on CSA (1xpcsa).
Geographical origins and derivations of the CSA-binding parasite lines
qRT-PCR and flow cytometry assay on IEs.Quantitative real-time PCR (qRT-PCR) and flow cytometry were performed as previously described (3). In brief, var gene transcription was assessed by qRT-PCR by using primers to the DBL4 domain in VAR2CSA (58) or gene-specific primers to ITvar14 and ITvar16 (61). Results are expressed as fold differences in expression relative to that of the housekeeping Pf adenylosuccinate lyase (Pfasl) (PFB0295w) (3). For flow cytometry, mature-stage IEs were grown in O+ blood and incubated with rabbit and human plasma that had been preabsorbed twice on uninfected O+ erythrocytes. For each assay, 10 million erythrocytes infected with between 5 and 8% trophozoites were incubated with a 1/25 dilution of rabbit plasma or a 1/10 dilution of human plasma. Bound antibodies were detected by adding Alexafluor 488-conjugated goat anti-rabbit IgG (A-11034; Molecular Probes) or fluorescein isothiocyanate (FITC)-conjugated goat F(ab′)2 anti-human IgG (PN IM1627; Beckman). Samples were analyzed in an LSR II flow cytometer (Becton Dickinson) and analyzed using FLOWJO 8.1 software (Tree Star Inc.). Binding is presented as the mean of the adjusted geometric mean of fluorescence intensity (MFI) for plasma samples run in duplicate. The adjusted MFI was as follows: (IEi − UEi) − (IEp − UEp), where IEi is the MFI of IEs following incubation in immune plasma, UEi is the MFI of uninfected erythrocytes following incubation in immune plasma, IEp is the MFI of IEs following incubation in preimmune plasma, and UEp is the MFI of uninfected erythrocytes following incubation in preimmune plasma.
Plasma and clinical parasite samples.The human plasma and clinical parasite samples used in these studies were collected from African donors, under protocols approved by the relevant ethics review committees. Study participants gave a written, informed consent before donating samples, and samples included those from adult multigravid women from Tanzania (n = 10) (24) and adult men (n = 10) and multigravid women from Cameroon (n = 10) (gift from Isabella Quakyi). Nonimmune adult plasma samples are from donors in Seattle.
Depletion of anti-plasma antibodies on recombinant proteins.For the depletion assay, 200 μg of His-tagged P. pastoris-expressed recombinant protein 7G8-DBL5 or 7G8-DBL1 was incubated with 0.3 ml of nickel-nitrilotriacetic acid-agarose resin (Ni-NTA; Qiagen) overnight at 4°C with rotation. Unbound protein was removed by centrifugation, and the recombinant protein-Ni-NTA mixture was then incubated 1 h at 4°C with 0.1 ml of preimmune or immune plasma. The supernatant corresponding to the depleted anti-DBL plasma was collected by centrifugation and tested by flow cytometry as described above.
IE binding and antibody binding inhibition assays.IE binding was performed as previously described (3). In brief, IEs were tested for binding to 10-μl spots of 0.1 mg/ml bovine CSA (Fluka Biochemika) or 0.05 mg/ml rCD36 (R&D). Binding assays were performed with 10 μl of 1 × 107 IEs/ml enriched via MACS purification (43). As a control, CSA spots were pretreated with chondroitinase ABC (Sigma-Aldrich). For antibody inhibition assays, IgG antibodies were purified from 0.2 ml of preimmune or immune rabbit plasma, using the Pierce Nab spin kit (89949; Pierce). The purified IgG concentration was determined by a bicinchoninic acid (BCA) assay (23225; Pierce). Antibody binding inhibition assays were performed as described above, except that binding inhibitions were compared at two concentrations of CSA (0.01 mg/ml or 0.1 mg/ml) and in the presence of purified IgGs at a final concentration of 0.5 mg/ml or 1 mg/ml. Binding inhibition was calculated as the percentage of binding in the presence of immune or preimmune IgGs.
Sequence analysis.Phylogenetic analyses were done using PAUP*4.0b10 to generate neighbor-joining trees with 1,000 bootstrap replicates. Gap opening and gap extension penalties were 5.0 and 0.05, respectively.
RESULTS
Expression of P. pastoris recombinant proteins.It has previously been shown that immunization with the VAR2CSA-DBL1, -DBL3, or -DBL6 recombinant proteins produced in P. pastoris elicited variant-specific antibodies to CSA-binding parasite lines, but we were unable to produce the VAR2CSA-DBL5 in that study (3). One reason why the DBL5 domain may not have been expressed previously is that the construct ended shortly after the 10th cysteine residue (C10), and recent structural information indicates that PfEMP1 DBL domains have two additional cysteine residues that are involved in domain disulfide bonding (30, 48), similar to the erythrocyte invasion binding ligands (49, 55). The final two cysteine residues in DBL domains frequently have a CX1-2C motif or CX1-2CX1-2C motif (30). By extending the domain boundaries of DBL5 recombinant proteins to include a CX1-2CX1-2C motif we were able to secrete this domain from P. pastoris and improve the expression of other poorly expressed domains (2). In this study, we investigated the immunogenicity of three VAR2CSA-DBL5 recombinants and a VAR2CSA-DBL1 recombinant protein by using the new construct boundaries that included two additional cysteine residues believed to be important for domain folding (30, 48) (Fig. 1). As a vaccination control, a DBL domain from a non-VAR2CSA protein, IT4var22-DBL3, was also expressed. Overall, 7G8-DBL1 and 7G8-DBL5 were the best-expressed proteins (from 1 to 1.5 mg/liter), while 3D7-DBL5, IT4-DBL5, and IT4var22-DBL3 recombinant proteins were produced at between 0.1 mg/liter and 0.5 mg/liter. Under SDS-PAGE nonreducing conditions, all of the proteins ran at the expected size, except 3D7-DBL5, which was ∼10 kDa smaller than expected (Fig. 1 and data not shown). Edman degradation confirmed that 3D7-DBL5 had undergone N-terminal truncation at amino acid position 95, accounting for the loss in size and giving it nearly the same domain boundaries as those of the other two DBL5 recombinant proteins.
Development of a panel of CSA-binding parasite lines.An obstacle to developing a PAM vaccine is the lack of a standard panel of CSA-binding parasite lines to evaluate antibody breadth. Most analyses have focused on four lab-adapted parasite lines, IT4/FCR3, 7G8, HB3, and NF54/3D7. The first three parasites have been cloned and represent single genotypes, while NF54 is the parent to the 3D7 genome reference isolate and also appears to contain another minor parasite genotype (22). A single var2CSA gene copy is present in all of the lines except for HB3, which has two distinct gene copies (31).
For this study, two new CSA-binding lines were derived from West African isolates (Pf2004 and Pf2006) and a mixed HB3-CSA parasite line was cloned by flow cytometry to generate parasite lines that transcribed either the HB3 A or B alleles to develop a panel of seven CSA-binding parasite lines. By the use of allele-specific qRT-PCR primers, one of the cloned HB3 lines predominantly expressed the HB3 A allele, while the other strongly transcribed the HB3 B allele but also transcribed a low level of the HB3 A allele (data not shown). Altogether, the parasites in the panel included isolates from Central America, South America, and Africa (Table 1), and the two parasite lines from West Africa were isolated several decades after the other parasites in the panel. The geographic origins of IT4/FCR3 and 3D7 are more ambiguous (Table 1), but the IT4/FCR3 genotype is most similar to that of the Southeast Asian isolates based on chromosome-wide single-nucleotide polymorphism (SNP) haplotype (35).
As expected, all of the CSA-binding parasite lines strongly bound CSA but had limited binding to the blood microvasculature receptor CD36 (Table 2). In contrast, the negative-control A4ultra and ItG-ICAM-1 parasite lines had the opposite phenotype. Compared to other parasites in the panel, the NF54-CSA and 3D7-CSA parasite lines (data not shown) were weak CSA binders, even after repeated selection on CSA for more than eight times, and quickly became a mixture of CSA and CD36 phenotypes despite repeated CSA selection. In contrast, the other CSA-binding lines were highly stable for more than 10 parasite cycles (data not shown) and bound CSA to a much greater extent (Table 2).
Characterization of a panel of P. falciparum laboratory isolates
To confirm that parasites in the panel had a phenotype similar to that of placental isolates, they were tested by qRT-PCR for var2csa transcription and with plasma samples from malaria-exposed donors. All CSA-binding lines highly transcribed var2csa compared to the housekeeping reference control gene Pf adenylosuccinate lyase (Asl). In contrast, var2csa was weakly transcribed in the negative-control A4ultra and ItG-ICAM-1 parasite lines, which express var genes different from var2csa (Table 2). Furthermore, all CSA-binding parasite lines, except for Pf2006-CSA, were preferentially recognized by comparing a pool of plasma samples from multigravid Cameroonian women to those from immune men, a characteristic of placental isolates (Table 3). By comparison, the non-CSA-binding A4ultra and ItG-ICAM lines were recognized equally well or better by male plasma samples than by female plasma samples. The non-gender-biased recognition of Pf2006-CSA may represent only a gap in the Cameroon plasma pool, because Pf2006-CSA was highly recognized by a pool of samples from multigravid women from Kenya and highly transcribed var2csa (Table 2). In addition, individual plasma samples from primigravid and multigravid women in Malawi had significantly higher levels of IgG to all of the CSA-binding parasite lines in the panel than did those from the men (M. Hommel and J. Beeson, unpublished observations). Taken together, this analysis confirms that var2CSA transcription is strongly linked to CSA binding and that parasites in the panel have the binding and antigenic phenotype of placental isolates.
Cross-reactivity of DBL5 sera on lab-adapted parasite lines and clinical isolates
Conservation and diversity of VAR2CSA sequences in the CSA-binding panel.To investigate whether parasites in the CSA binding panel differed in var2csa sequences, the DBL1 and DBL5 domains were cloned and sequenced and compared to other full-length VAR2CSA sequences available in GenBank. In every parasite line, except for HB3, only a single var2csa DBL1 or DBL5 sequence was amplified. This includes the Pf2006-CSA and Pf2004-CSA parasite lines that have not been cloned but have been selected multiple times on CSA in the laboratory and appear to now contain a single or major parasite genotype by MSP-1/MSP-2 genotyping (data not shown). We also confirmed that CS2-CSA and FCR3-CSA parasites, which are isogenic, had identical DBL5 sequences and that the DBL5 sequence from NF54-CSA was identical to the 3D7 progeny line.
Based on previous studies of full-length var2csa genes, DBL5 is the second most conserved VAR2CSA domain after DBL4 (11). In that study, VAR2CSA domains ranged from a low of ∼61% amino acid identity (DBL6) to a high of ∼88% identity (DBL4). In the present study, no two parasites in the panel expressed the same DBL1 or DBL5 sequence, except the HB3 A and B var2csa alleles, which encoded identical DBL5 domains (Fig. 2). Therefore, there is extensive diversity of VAR2CSA sequences in the parasite population. Overall, the DBL1 domain in the parasite panel averaged 80% amino acid identity (range, 72 to 89%), and the DBL5 domain averaged 86% amino acid identity (range, 83 to 99%), similar to that of a larger sampling of 17 VAR2CSA sequences from around the world (for DBL1, 79% identity, range of 71 to 89%; for DBL5, 87% identity, range of 83 to 99%) (Fig. 2). Although parasites in the panel differed in DBL1 and DBL5 sequences, there was a significant overlap of polymorphism between globally dispersed isolates and relatively restricted polymorphism at variable regions (see Fig. S1 and S2 in the supplemental material). The most divergent sequence was the DBL1 domain from WR80, which had unique polymorphism in the C terminus or subdomain 3 (see Fig. S1 in the supplemental material). The WR80 parasite was derived from Southeast Asia (35) but is not in the CSA panel. Except for the unique variation in the C terminus of DBL1, the WR80 sequence shares other variable blocks with other P. falciparum isolates. These findings are consistent with earlier work demonstrating extensive gene mosaicism between var2csa alleles (11, 56) and suggest that VAR2CSA polymorphism has an ancient origin that is likely maintained in the parasite population by gene recombination. Together, this analysis confirms that the parasite panel is comprised of different VAR2CSA sequences, which reflects diversity in the parasite population.
Sequence comparison of VAR2CSA-DBL5. Neighbor-joining trees comparing DBL1 (A) and DBL5 (B) sequences from CSA-binding parasites in the panel to global isolates. Bootstrap values above 80% are shown. The geographic origin of parasites is indicated by color: yellow, Southeast Asia; light brown, Africa; metallic brown, Central or South America; gray, unknown. Parasites in the reference panel are indicated by oval-shaped or rectangular shading. DBL5 recombinant proteins were made from parasites with rectangular shading.
Immunization with DBL5 recombinant proteins elicits broadly strain-transcendent antibodies to diverse CSA-binding parasite lines.To investigate the immunogenicity of recombinant proteins, rabbits were immunized with each of the three VAR2CSA-DBL5 allelic forms, as well as the 7G8 VAR2CSA-DBL1 recombinant protein and the negative-control IT4var22-DBL3 recombinant protein. Antibodies raised against the three DBL5 proteins had nearly identical endpoint titers against the 7G8-DBL5 recombinant protein (ranged between 3 × 105 and 5 × 105), showing that the recombinant proteins contained cross-reactive epitopes. To investigate whether cross-reactive epitopes were exposed in the native protein, plasma antibodies were tested against CSA-binding IEs. By flow cytometry, each of the three VAR2CSA-DBL5 immunogens elicited broad antibody responses to all or most of the CSA-binding parasites in the panel but not with the non-CSA-binding control A4ultra or ItG-ICAM-1 parasite lines (Table 3). As expected, the FCR3-CSA and CS2-CSA lines, which have been used interchangeably in previous studies, gave similar reactivity profiles (data not shown).
In contrast to DBL5 plasma, rabbits immunized with the negative-control var22-DBL3 or A4var-CIDR1 (3) did not react with the CSA-binding parasite lines. In addition, anti-VAR2CSA-DBL1 plasma had lower reactivity, and the three rabbit plasma reacted with only three or four of the seven CSA-binding parasite lines, including the homologous 7G8-CSA parasite. Whereas DBL5 plasma had greater breadth than did DBL1 plasma, there were differences in serological profiles between the three DBL5 immunogens. For instance, the Pf2006-CSA line was weakly recognized by the IT4-DBL5 plasma but was not reactive with 7G8-DBL5 plasma. Conversely, the 7G8-CSA parasite was well recognized by 7G8-DBL5 and 3D7-DBL5 plasma but was only weakly reactive with one of the three rabbits immunized with IT4-DBL5. Differences were also observed between rabbits immunized with the same DBL5 immunogen.
DBL5 plasma samples were also tested against three fresh clinical isolates collected from the blood of pregnant women. These parasite isolates were not fully adapted to continuous in vitro cultivation, but two of them were selected once in vitro for adhesion to CSA to increase VAR2CSA expression. At the time of DBL5 plasma assessment, all three clinical isolates were confirmed to bind to CSA and not CD36 (see Table S1 in the supplemental material) and reacted in a gender-specific manner with malaria-exposed plasma (Table 3). Of the three clinical isolates, isolate 736 reacted with all three DBL5 plasma samples, and all three of the clinical isolates were recognized by at least one DBL5 plasma sample. Thus, immunization with VAR2CSA-DBL5 recombinant proteins elicited broad antibodies to diverse CSA-binding parasite lines.
To confirm that anti-DBL5 antibodies were responsible for pan-reactivity, plasmas were preadsorbed against recombinant proteins. Preabsorption of rabbit polyclonal DBL5 plasma on 7G8-DBL5 recombinant protein depleted antibody recognition of both homologous and heterologous CSA-binding parasite lines, while preabsorption with the negative-control 7G8-DBL1 recombinant protein had no effect on serological recognition (Table 4). To investigate adhesion blocking of DBL5 plasma, IgG antibodies were purified from two of the rabbits that displayed the broadest serological cross-reactivity and tested for inhibitory activity against three different CSA-binding parasite lines. While purified IgG antibodies from immune rabbits bound in a specific manner to CSA-binding parasite lines by flow cytometry, the level of CSA-binding inhibition was weak and variable (between 3 and 26%) and was not significantly above that observed with a negative-control ItG-ICAM-1 parasite binding to CD36 (see Table S2 in the supplemental material; data not shown). Thus, VAR2CSA-DBL5 immunogens elicited pan-reactive antibody responses to diverse CSA-binding parasite lines but limited or no inhibitory activity under this immunization protocol.
Depletion of antibody reactivity by recombinant protein
DISCUSSION
VAR2CSA is the leading candidate for a pregnancy malaria vaccine, but developing an effective vaccine will require the identification of relatively conserved antibody targets and immunogens that generate broadly reactive antibodies to different isolates. Plasma samples collected from pregnant women can inhibit the CSA binding of parasite isolates from distant geographic regions (25), suggesting that placental isolates have a degree of cross-reactive epitopes. However, CSA-binding parasite isolates have significant antigenic diversity (8, 10, 33), and some studies suggest that adhesion-blocking epitopes are polymorphic (10, 57). It is not known whether cross-reactive antibodies in maternal plasma are due to overlapping polymorphism between VAR2CSA alleles (11, 56) or strictly conserved epitopes.
Key questions for pregnancy malaria vaccine development are (i) how many antigen combinations will be required for broad coverage and (ii) will vaccine components need to be tailored for specific geographic regions? VAR2CSA is a polymorphic protein and may evolve more rapidly than other parasite genes leading to antigenic drift between different parasite populations. To date, there has been very limited investigation into antibody cross-reactivity of VAR2CSA immune plasma (3, 21, 37, 38), but one study showed that anti-DBL5 antibodies made against a refolded bacterial protein were able to cross-react on two different CSA-binding parasite lines and with a placental isolate (38), and anti-DBL5 antibodies have been used to show that VAR2CSA is expressed at the surface of placental IEs and that anti-DBL5 antibodies were reactive with some, but not all, placental isolates tested (32). These promising results suggested that the VAR2CSA-DBL5 domain may expose cross-reactive epitopes; however, the VAR2CSA sequences of the placental isolates used were not characterized, and the study was performed in a single geographic region. Therefore, the extent of antibody cross-reactivity that may be achieved against a single DBL domain in VAR2CSA remains unclear.
In this study, we developed a panel of seven different CSA-binding parasite lines from different regions of the world and used this as a basis to assess the breadth of anti-VAR2CSA plasma. We show that parasites in the panel have an amount of sequence diversity that is similar to that of a broader sampling of VAR2CSA sequences from around the world and therefore represent a starting point to assess antibody breadth. Using this panel, we demonstrate that anti-DBL5 plasmas were broadly cross-reactive to most or all of the parasites in the panel, while anti-DBL1 plasma had limited cross-reactivity. Our results also suggest that a small number of DBL5 variants in a vaccine may be sufficient to give broad population coverage since all isolates were recognized to some extent by at least one of the anti-DBL5 plasmas.
It is surprising that DBL5 plasma had much greater breadth than did DBL1, since the two domains have relatively similar levels of conservation (79% versus 87%). The variant surface antigens of P. falciparum IEs are important targets of acquired immunity; however, acquired antibodies appear to recognize primarily variable epitopes (13). This study demonstrates highly strain-transcendent antibody epitopes in the native VAR2CSA protein and shows that they can be targeted by DBL5 vaccination. Remarkably, each of the three different VAR2CSA-DBL5 allelic variants was able to elicit broad antibody responses to most of the CSA-binding parasite lines in the panel, although the serological profiles differed slightly between immunogens. This suggests that the fine specificity of cross-reactive antibodies may differ between DBL5 immunogens; therefore antibodies may not be targeting strictly conserved epitopes, or the exposure of conserved epitopes on native VAR2CSA may differ between variants. The DBL5 domain is naturally immunogenic, and antibodies are acquired following placental infection with P. falciparum (5, 39, 59). However, maternal antibodies purified on DBL5 recombinant proteins were not cross-reactive to different placental isolates (32), indicating that the breadth of reactivity of acquired antibodies may be substantially different for vaccine-induced responses. Therefore, it will be important to investigate the basis for anti-DBL5 antibody cross-reactivity and investigate if such antibodies are acquired during natural infections and whether vaccine coverage can be expanded by combining DBL5 recombinant proteins and employing human-compatible adjuvants.
Pregnancy malaria vaccine development has been highly focused on binding inhibitory antibodies. However, vaccine design strategies have been complicated because several different recombinant DBL domains from VAR2CSA bind CSA (14, 26, 27, 48), but it has been questioned whether all of these are functional in the native protein (42). It has previously been shown that immunization with single VAR2CSA domains can induce antibodies that react with CSA-binding IEs (3, 5, 21, 38); however, none of the antibodies could effectively inhibit CSA binding. A recent study suggested that this may relate to the species of animal used for immunization. In that report, immunization of rats with any of the six VAR2CSA-DBL domains could elicit adhesion-inhibitory antibodies, but the best inhibitory antibodies were made against a VAR2CSA-DBL4 recombinant protein and were partially inhibitory against two heterologous parasite isolates (36). The mechanism of inhibition is unknown, and it is still unclear whether adhesion-blocking epitopes are conserved or how to consistently target adhesion-inhibitory epitopes by vaccination.
Although DBL5 did not induce adhesion-inhibitory antibodies in rabbits, it may still be a good selection for vaccine development. Monocytes and macrophages are commonly seen in infected placentas and frequently contain IEs or hemozoin pigment (62). IgG1 and IgG3 are the predominant antibody isotypes in placental infections (19), indicating that antibodies may opsonize IEs for phagocytosis as well as to inhibit binding. Although inhibitory antibodies have been correlated with protection in some studies (17, 25), this correlation has not been observed in all reports, and opsonizing antibodies have also been proposed to have a protective role (28, 29, 34). It is not known whether women ever achieve sterile immunity against placental infections, but they do acquire a substantial degree of protective immunity after repeated exposure (12). The finding that VAR2CSA-DBL5 displays highly strain-transcendent epitopes suggests a role for additive or synergistic vaccine strategies that would combine both broad adhesion-blocking and opsonizing antibody responses to prevent high-density placental infections associated with disease.
ACKNOWLEDGMENTS
We are grateful to Yuko Ogata from the Proteomic Core at SBRI, and we thank Karen Callahan (SBRI) and Karine Reiter (MVDB) for excellent lab assistance.
This work was supported by a grant from the Bill and Melinda Gates Foundation (P.E.D., M.F., J.D.S.). This research was supported in part by the Intramural Research Program of the NIH, Malaria Vaccine Development Branch, National Institute for Allergy and Infectious Diseases. J.G.B. and M.H. are supported by the National Health and Medical Research Council of Australia.
FOOTNOTES
- Received 13 April 2009.
- Returned for modification 6 July 2009.
- Accepted 22 February 2010.
- Copyright © 2010 American Society for Microbiology