Antibodies to Escherichia coli-Expressed C-Terminal Domains of Plasmodium falciparum Variant Surface Antigen 2-Chondroitin Sulfate A (VAR2CSA) Inhibit Binding of CSA-Adherent Parasites to Placental Tissue

Abstract

Placental malaria (PM) is characterized by infected erythrocytes (IEs) that selectively bind to chondroitin sulfate A (CSA) and sequester in placental tissue. Variant surface antigen 2-CSA (VAR2CSA), a Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1) protein family member, is expressed on the surface of placental IEs and mediates adherence to CSA on the surface of syncytiotrophoblasts. This transmembrane protein contains 6 Duffy binding-like (DBL) domains which might contribute to the specific adhesive properties of IEs. Here, we use laboratory isolate 3D7 VAR2CSA DBL domains expressed in Escherichia coli to generate antibodies specific for this protein. Flow cytometry results showed that antibodies generated against DBL4ε, DBL5ε, DBL6ε, and tandem double domains of DBL4-DBL5 and DBL5-DBL6 all bind to placental parasite isolates and to lab strains selected for CSA binding but do not bind to children's parasites. Antisera to DBL4ε and to DBL5ε inhibit maternal IE binding to placental tissue in a manner comparable to that for plasma collected from multigravid women. These antibodies also inhibit binding to CSA of several field isolates derived from pregnant women, while antibodies to double domains do not enhance the functional immune response. These data support DBL4ε and DBL5ε as vaccine candidates for pregnancy malaria and demonstrate that E. coli is a feasible tool for the large-scale manufacture of a vaccine based on these VAR2CSA domains.

INTRODUCTION

Pregnancy malaria (PM) results when Plasmodium falciparum-infected erythrocytes (IEs) sequester in intervillous spaces, leading to severe clinical sequelae for the mother and her fetus. Malaria infection during pregnancy is related to both maternal and infant anemia, increased risk of abortion, premature delivery, and low birth weight (1, 2). In addition, for HIV-positive women, placental malaria may increase the risk of mother-to-child transmission (MTCT) of the virus (3). Every year, up to 20,000 women die from pregnancy complications and as many as 200,000 infants die from complications related to low birth rate as a result of malaria infection during pregnancy (4). The only malaria drug currently recommended by WHO for use during pregnancy is sulfadoxine-pyrimethamine, and like most antimalarials, drug resistance is a growing problem (5). A vaccine against PM is the best option for preventing illness and death in these women and their children.

Among the human malaria parasites, the ability to sequester in vascular beds is a hallmark of P. falciparum. IEs express parasite proteins on their cell surface, causing them to adhere to endothelial cells as well as to other host cell types (6). These antigenically distinct parasite proteins, called Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1), are encoded by var genes, a group of approximately 60 genes that are variably expressed by the parasite in a mutually exclusive fashion (7, 8). PM is characterized by infected erythrocytes that selectively bind chondroitin sulfate A (CSA), a glycosaminoglycan expressed on the surface of placental syncytiotrophoblasts (9). Variant surface antigen 2-CSA (VAR2CSA), a PfEMP1 protein, is selectively expressed in CSA-binding placental parasites (10, 11) and encodes 6 extracellular domains, of which several have been demonstrated to bind to CSA in vitro (12, 13). Women in regions where malaria is endemic acquire antibodies to VAR2CSA over successive pregnancies as they become resistant to placental malaria (14–16). Importantly, parasites engineered to lose the VAR2CSA gene lose the ability to adhere to CSA (17, 18). VAR2CSA has thus emerged as the primary parasite protein associated with CSA binding in the placenta and as a lead candidate in vaccine research for pregnancy malaria.

The size and complexity of VAR2CSA are a challenge to large-scale vaccine production, and thus, studies have mainly focused on defining smaller regions that can induce a broad antiadhesive antibody response. The six individual Duffy binding-like (DBL) domains of VAR2CSA are involved in the specific adhesive properties of infected cells (6, 12). Recent reports have indicated that antibodies to some of these domains may inhibit parasite binding to CSA on the surface of placental cells (19, 20). Here, we investigate this further by focusing on the domains of the C-terminal half of VAR2CSA, specifically, on DBL4ε and DBL5ε.

Previously, we demonstrated that antisera to the Escherichia coli-expressed and refolded laboratory isolate 3D7 DBL5ε domain cross-react with surface proteins of CSA-binding parasites in lab strains as well as a PM clinical isolate (21). Here, we expand this work to study the ability of antisera to DBL5ε and its immediate neighbors, DBL4ε and DBL6ε, to inhibit parasite binding to CSA and to placental tissue. We show by enzyme-linked immunosorbent assay (ELISA) and Western blotting that pooled plasma from multigravid (MG) women recognizes specific recombinant VAR2CSA domains, including the three domains of the C terminus. We also show by flow cytometry and immunofluorescence assay (IFA) that antibodies generated against these domains bind parasites derived from pregnant women. Importantly, antibodies to DBL4ε and DBL5ε can inhibit binding of IEs derived from pregnant women to placental tissue to a similar degree as that seen with plasma pooled from multigravid women. Furthermore, we also show that, to various degrees, these antibodies inhibit binding of several parasites derived from pregnant women to CSA, whereas they do not inhibit the binding of children's parasites to CD36. We conclude that E. coli expression can yield functional antibodies to DBL4ε and DBL5ε and that such a system would hence be an asset for use in large-scale vaccine production.

MATERIALS AND METHODS

Cloning.

For this study, all constructs were cloned into the pET28b(+) expression vector to express N-terminal His₆-tagged proteins. Each VAR2CSA domain was amplified by PCR from P. falciparum strain 3D7 genomic DNA (gDNA), cloned into the vector, and verified by sequencing. Primers used to generate new clones for this study are listed in Table 1. The amino acid boundaries were as follows: DBL1X, amino acids (aa) 1 to 449; DBL2X, aa 512 to 975; DBL3X, aa 1193 to 1577; DBL4ε, aa 1570 to 1926; DBL5ε, aa 1982 to 2336; and DBL6ε, aa 2325 to 2648. Double-domain constructs containing domains 4 and 5 or domains 5 and 6 contained the boundaries aa 1570 to 2336 and aa 1982 to 2648, respectively. Domains 4, 5, and 6 from a maternal parasite isolate (patient 661) were also amplified and cloned into pET28b(+). For DBL4ε, DBL5ε, and DBL6ε from patient 661, we amplified a protein region corresponding to aa 1570 to 1882 of 3D7 DBL4ε, aa 1997 to 2256 of 3D7 DBL5ε, and aa 2265 to 2587 of 3D7 DBL6ε. For use as a control, we also cloned 3D7 P. falciparum AMA-1 (PfAMA-1) into the pET28b(+) expression vector. Primers used for DBL1X, DBL3X, DBL5ε, AMA-1, and DBL6ε from patient 661 are described in reference 16.

Table 1.

Primers used in this study

Primer name	Orientationa	Sequence
DBL2	F/R	TTTTGCTAGCTCTAGTTCTAATGGTAC/GGGCTCGAGTTATTGATTCATATATTC
DBL4	F/R	TTTTGCTAGCGAGAAAAAAAATAATA/GGGCTCGAGTTATTGTTCCTTTTT
DBL6	F/R	TTTTGCTAGCGAGTATGATAAGGGTA/GGGCTCGAGTTATTCGAGAATGTCACT
DBL4/5	F/R	TTTTGCTAGCGAGAAAAAAAATAATA/GGGCTCGAGTTATTTATTACAAATATA
DBL5/6	F/R	TTTGAATTCTCATATGTTAGATAGATG/GGGCTCGAGTTATTCGAGAATGTCACT
661DBL4	F/R	TTTCCATGGGCGAGAAAAAAAATAATAAATCTCTTTG/GGGCTCGAGTTAGTGGTGGTGGTGGTGGTGAGGTTCCATAATCATTGAATAATCTTT
661DBL5	F/R	TTTCCATGGGCGATTTAATTGGAGATGCTATAGGATG/GGGCTCGAGTTAGTGGTGGTGGTGGTGGTGTGTATTTTTAAATTCATCCATACCC

^aF, forward; R, reverse.

Expression.

E. coli Rosetta 2(DE3)pLysS Singles competent cells (Novagen) were transformed with each construct, induced for protein expression using Overnight Express autoinduction medium (Novagen), and lysed with CelLytic B and/or CelLytic IB reagent for soluble and insoluble proteins, respectively. All single-VAR2CSA-domain proteins were predominantly in the insoluble fraction, while double domains were found in the soluble fraction. Recombinant His-tagged proteins were purified using the Ni-nitrilotriacetic acid His-Bind resin and buffer kit from Novagen. All buffers were brought to 6 M urea.

Protein purification and refolding.

His-tagged proteins were reduced by addition of 2% β-mercaptoethanol (BME) and further purified using reverse-phase high-performance liquid chromatography (RP-HPLC) on an acetonitrile gradient from 5% to 95% at 1 ml/min in 90 min (C₄ column; size, 250 by 4.6 mm; 5 μm; Jupiter 5uC4 300A; Phenomenex). Proteins were then lyophilized, followed by resuspension in 6 M urea with 0.1% BME and 25 mM HEPES, pH 8.0, and analyzed by SDS-PAGE to assess purity. Refolding by dialysis was performed against 1× phosphate-buffered saline (PBS) with 5 mM reduced glutathione and 1 mM oxidized glutathione, as described in reference 21.

Parasite culture.

Clinical strains were isolated from peripheral blood of pregnant Tanzanian women or from children (22) and adapted to in vitro culture. Twelve maternal isolates and 12 children's isolates were used for assays; we present data only where a minimum of 3 isolates were tested. Ethical clearance was obtained from the institutional review boards of the Seattle Biomedical Research Institute (SBRI) and the National Medical Research Coordinating Committee in Tanzania. The binding phenotype of these parasites was assessed by routine methods described in reference 22 to determine affinity for CSA. Maternal isolates bound exclusively to CSA, while children's isolates bound to CD36. Laboratory isolates 3D7, CS2, and NF54 were panned on immobilized CSA as described elsewhere (10) to obtain CSA-binding forms, which have been shown to react specifically to immune sera from African multigravid women but not with immune sera from male subjects, and were confirmed to express the VAR2CSA transcript by quantitative reverse transcription-PCR using published gene-specific primers and conditions (10, 11). Before all seroreactivity assays, IEs were floated on 0.5% porcine gelatin (Sigma) at 37°C for 30 min to enrich for mature trophozoite-stage parasites.

Immunization.

Rats were immunized via inoculation of 30 μg of antigen in complete Freund's adjuvant. Three boosts with the same amount of antigen were done using incomplete Freund's adjuvant (at Antibodies, Inc., Davis, CA). Mice were challenged in-house using 10 μg of antigen per injection in TiterMax adjuvant with three additional boosts. All antisera were tested by ELISA against the proteins that had been used for immunization as well as against AMA-1 and in some cases non-self-VAR2CSA domains to test for background reactivity. Immunizations were performed with the approval of SBRI's Institutional Animal Care and Use Committee.

IgG purification.

IgG was purified from sera using Gammabind Plus Sepharose beads (GE Healthcare) and Pierce spin columns. Binding, wash, and elution buffers were from Pierce/Thermo Scientific; neutralization buffer was from Emerald Biosystems. Purified IgG was then lyophilized, resuspended in PBS, and dialyzed against PBS (3 buffer changes) using Slide-A-Lyzer dialysis cassettes (Pierce/Thermo Scientific).

IFA.

IEs (maternal isolates 918 and 710 and children's isolates 934, 708, and 631) were blocked in 5% goat serum, washed in PBS, and incubated with immune serum at a 1:5 or 1:10 dilution. Cells were washed again and incubated with Alexa Fluor 488-labeled anti-rat IgG (Invitrogen) at a 1:25 dilution. To distinguish between infected and uninfected cells, nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI) at a 20-μg/ml final concentration. All steps were done on ice. Samples were applied to coverslips with p-phenylenediamine (1 to 2 crystals were dissolved in 1× PBS with 10% methanol) as an antifade reagent and viewed by fluorescence microscopy.

Flow cytometry.

IEs (maternal isolates 710 and 736, lab CSA-binding isolate CS2, and children's isolates 934 and 678) were incubated with immune serum at a 1:25 dilution, followed by Alexa Fluor 488-labeled secondary anti-rat IgG (Invitrogen) at a 1:100 dilution. To distinguish between infected and uninfected cells, samples were stained with ethidium bromide at a 1-μg/ml concentration. For each replicate, 10,000 cells were counted. We include only data from studies that were done in triplicate on 3 separate days. Samples were run using an LSRII flow cytometer (BD Biosciences) and analyzed using the FlowJo (Tree Star, Inc.) software program. Plasma from multigravid women and plasma from noninfected individuals were used as positive and negative controls, respectively.

Binding inhibition assay on placental tissue or on CSA-coated petri dishes.

Sections (6 μm) of frozen uninfected placental tissue were prepared using an HM505E cryostat (Microm) and applied to glass slides, and the glass slides were stored dry at −20° until use. Slides were rehydrated in PBS and blocked with 3% bovine serum albumin (BSA) in PBS for 30 min before assay. To make CSA- or CD36-coated plates, 15 μl of CSA or CD36 at 10 μg/ml was coated within 10-mm circles on polystyrene petri dishes, and the dishes were incubated overnight at 4°C in a moist chamber and then washed and blocked with 3% BSA. To test for binding inhibition, IEs (12 maternal isolates and 12 children's isolates) were first gelatin enriched for trophozoites, brought to 20% parasitemia and 0.5% hematocrit, and then washed 3 times in RPMI before blocking with 3% BSA for 30 min at 37°C. Cells were then incubated with immune serum (rodent) or plasma (human) diluted 1/5 and 1/10 (for placental sections) or immune IgG at 1 mg/ml and 0.5 mg/ml (for CSA spots) at 37°C for 30 min and then gently placed on placental sections or CSA-coated spots and allowed to incubate at 37°C for another 30 min. For placental section analysis, slides were washed and then fixed in methanol and stained with 5% Giemsa for viewing under a light microscope. For CSA-binding analysis, cells were fixed with 6% glutaraldehyde and stained with 5% Giemsa or 1 μg/ml DAPI. IEs were counted under ×40 magnification on a TE2000U inverted microscope (Nikon) and analyzed by NIS Elements AR software (Nikon Instruments, Inc.). The level of binding inhibition activity was calculated in comparison to the level of inhibition with the negative control (anti-AMA-1 antibody).

RESULTS

Pooled plasma from MG women recognizes individual VAR2CSA domains.

Women in regions where malaria is endemic develop cross-reactive inhibitory antibodies to placental malaria parasites over successive pregnancies (23). We therefore took plasma from multigravid women to assess the antibody response to various domains of VAR2CSA that they might have. We cloned individual VAR2CSA domains from the 3D7 parasite strain into the pET28b plasmid vector to express them in E. coli for the generation of His-tagged recombinant proteins. After purification over an Ni²⁺ column, each DBL domain was then subjected to further purification by HPLC, refolded in PBS-glutathione (oxidation-reduction), and dialyzed against 1× PBS. All domains were analyzed for proper size and purity on SDS-PAGE. To test for recognition by Western blotting (Fig. 1B), we ran 200 ng of each domain on SDS-polyacrylamide gels (Fig. 1A) and probed with pooled plasma from MG women participating in the MOMS (Mother-Offspring Malaria Study) project in Tanzania. Plasma from this cohort of women clearly recognized several of the VAR2CSA domains. Reaction to PfAMA-1 and to DBLβ2c2 from P. falciparum 11_0521 (Pf11_0521), a non-pregnancy-associated PfEMP1 protein, was also observed. No reaction to human dihydrofolate reductase (DHFR) was seen.

Fig 1.

Single-VAR2CSA-domain reactivities to pooled plasma from multigravid women. For Western blot analysis, 0.2 μg of each VAR2CSA domain was loaded onto an SDS-polyacrylamide gel (A), transferred to a polyvinylidene difluoride membrane (B), and probed with plasma from multigravid women at a 1/500 dilution. Horseradish peroxidase-conjugated goat anti-human IgG at 1/1,000 was used for detection. The assay was done in triplicate, with representative results shown. (C) For ELISA, 100 ng of each domain per well was coated in a 96-well ELISA plate and tested for reactivity by serial dilution of plasma. Reactivity to nonimmune plasma and pooled male plasma (not shown) was negligible for all domains except DBL6, which showed minor reactivity to male plasma. The estimated minimum titer (in thousandths) is shown in bold type.

To estimate the antibody titer, 100 ng of each domain was then coated into individual wells of a 96-well ELISA plate and tested for reaction to the pooled MG plasma (Fig. 1C) as well as to pooled plasma from nonimmune adults (not shown). Again, we observed that MG plasma reacted with PfAMA-1 and Pf11_0521 DBLβ2c2 but not with human DHFR protein. We also observed that MG plasma reacted to all VAR2CSA domains, with appreciably stronger recognition of DBL1X, DBL3X, DBL5ε, and DBL6ε. Nonimmune plasma did not recognize any VAR2CSA domain. These data show that multigravid women in this part of Africa have a strong antibody response to DBL5ε and to DBL6ε but have lower immunogenicity and poorer recognition of DBL4ε, similar to what was previously reported in our work using domains expressed in mammalian cells (16). Another group has reported similar findings that serum antibodies from multigravid women in regions where malaria is endemic show the strongest recognition of DBL3X, DBL5ε, and DBL6ε (24), suggesting that these domains may be important in a protective immune response.

Rodent antisera to DBL4ε, DBL5ε, and DBL6ε react with maternal IEs.

Because we previously observed that the 3D7 DBL5ε domain can induce cross-reactive antibodies, we next generated in rats and mice antisera to each of the three individual DBLε domains and to tandem domains (DBL4 and DBL5 [aa 1570 to 2337] and DBL5 and DBL6 [aa 1982 to 2649]) from the C-terminal region of VAR2CSA. To determine if these antisera had specificity for maternal parasites, we first performed IFA on intact red blood cells infected with parasites collected from either pregnant women or children. With these assays, we observed reactivity of each of the single-domain-specific antisera to the surface of IEs from pregnant women but not to children's parasites (Fig. 2A to C).

Fig 2.

Antisera to VAR2CSA DBL domains recognize maternal (Mat) but not children's (Kid) parasites. (A to C) IFA was performed on live IEs using antibodies generated against individual VAR2CSA domains. IEs were incubated with rat antisera to the individual DBL domains, followed by Alexa Fluor anti-rat IgG secondary antibody to visualize reactivity. (D) Flow cytometry was carried out on all antibodies generated in this study. Only results for reactivity (as measured by mean fluorescence intensity) of rat antisera to single and tandem domains are shown. The y axis values indicate the fold change of binding strength relative to that of anti-AMA-1, used as a negative control. Plasma from multigravid women was also tested, and the results were compared to those for nonimmune plasma (data not shown).

In addition, we performed flow cytometry on all rodent sera (mouse data not shown) to confirm reactivity specific to maternal parasites (Fig. 2D). These data indicate that antibodies generated to these domains of VAR2CSA, including DBL4ε, bind to maternal IEs or lab strains selected for CSA binding. Previously, we reported that mouse or rabbit antisera generated against DBL4ε did not recognize IEs in our flow cytometry assays (21). Here, our anti-DBL4ε sera are based on an expanded domain boundary of DBL4ε: aa 1570 to 1926 here versus aa 1570 to 1881 in the cited work (21). The updated boundaries include cysteine residues that may be important for correct folding and may account for the difference that we observed here, in support of our earlier data (21) which suggested that flexible loops exposed on correctly folded proteins are the most likely epitopes for recognition of live IEs that express this protein. Additionally, we generated antibodies using coimmunization with two domains at once and observed similar results by flow cytometry (data not shown). Of the sera tested, only rat anti-DBL6ε showed some cross-reactivity to children's parasites. These data confirm that antibodies generated against individual domains of VAR2CSA specifically recognize maternal IEs or IEs selected for CSA binding.

Antisera generated against DBL4ε and DBL5ε inhibit IEs from binding CSA.

To determine whether or not these domains generated functional antisera, we performed several binding inhibition assays to ascertain their ability to prevent maternal IE field isolates from binding to CSA. After enrichment for trophozoites, IEs from various maternal parasite lines were incubated with IgG purified from immune sera specific for the C-terminal domain constructs. IEs were then applied to CSA-coated petri dishes to determine their binding capabilities. For comparison, maternal IEs were also incubated with IgG purified from antisera specific to AMA-1 (not shown), to nonimmune adults (not shown), and to multigravid women. As an additional control, children's parasites were incubated with our immune antisera to determine if binding to CD36 could be inhibited (not shown).

We first assayed antisera generated against the last 3 domains of VAR2CSA that had been amplified from the gDNA of a field isolate (isolate 661). Purified rat IgG specific for the isolate 661-based DBL4ε consistently showed stronger binding inhibition of IEs than rat IgG specific to isolate 661-based DBL5ε or isolate 661-based DBL6ε, which showed only minimal inhibition (Fig. 3A). Results for mouse antisera (Fig. 3B) generated against the same domains showed a similar trend in assays done on IEs from field isolate 918. In assays performed on an additional isolate (isolate 755), all mouse IgG antibodies specific for the C-terminal VAR2CSA domains inhibited CSA binding at a level equal to that for pooled sera from multigravid women. For a third isolate tested (isolate 711), only IgG specific for the DBL5ε domain inhibited as well as sera from multigravid women did. These sera did not inhibit children's parasites from binding to CD36.

Fig 3.

Antisera to VAR2CSA DBL domains based on a maternal field isolate inhibit maternal IEs (mIE) from binding to CSA. IEs collected from pregnant Tanzanian women were incubated with purified IgG from immune sera generated against VAR2CSA constructs based on the sequence of maternal isolate 661 and then tested for binding to CSA. Results for rat IgG (A) and mouse IgG (B) are shown. Percent inhibition is on the y axis, with the mean value represented by vertical bars. Data are based on a minimum of 3 replicates. Error bars indicate standard errors of the means. MG, pooled sera from multigravid women.

Purified IgG specific for the 3D7 variant of VAR2CSA showed variable binding inhibition abilities across the 4 field isolates that we tested (Fig. 4A). Overall, IgG specific for DBL4ε showed the strongest binding inhibition ability, with a range of 36% to 67% inhibition. IgG specific for DBL5ε and DBL6ε had ranges of binding inhibition ability of 20% to 56% and 13% to 43%, respectively. Mouse antisera specific for these domains were tested against two field isolates and confirmed a similar trend (data not shown). Again, these sera did not inhibit children's parasites from binding CD36.

Fig 4.

Antisera to 3D7 VAR2CSA DBL domains inhibit maternal IEs from binding to CSA. IEs collected from pregnant Tanzanian women were incubated with purified IgG from immune sera generated against 3D7 VAR2CSA constructs and then tested for binding to CSA. Results are shown for rat antisera generated against single domains (A) and against tandem double domains (noted by slashes) and by coimmunization with single domains (noted by plus signs) (B). Percent inhibition is shown on the y axis, with the mean value represented by vertical bars. Data for a minimum of 3 replicates are shown unless otherwise noted. ND, no data. Error bars indicate standard errors of the means.

We also tested the antisera that were generated against the 3D7 tandem double domains or by coimmunization of two 3D7-based single domains simultaneously (Fig. 4B). In general, we did not observe a significant increase in binding inhibition of the tandem double domains relative to the strength of inhibition of individual domains. Only in the case of coimmunization with DBL5ε and DBL6ε did we observe an increased ability of rat antisera but not mouse antisera (data not shown) to inhibit parasite binding.

Antibodies to DBL4ε and DBL5ε inhibit binding of maternal parasites to placental tissue.

To determine whether or not antisera to these domains could prevent maternal IEs from binding to native receptor, we performed binding inhibition assays similar to those described above using uninfected placental cryosections instead of CSA-coated petri dishes. Here, we tested whole-rat antisera based on our isolate 3D7 DBL domains for the ability to inhibit binding of maternal parasite 710 to placental tissue, allowing us to estimate the possibility that the antisera that we have generated in rodents might be effective ex vivo. The results of this assay (Fig. 5) showed a similar trend in the overall binding inhibition ability of rat antisera based on single 3D7 VAR2CSA domains. Calculating the average of 100 fields of vision, we found that antibodies to DBL4ε and to DBL5ε were able to inhibit IE binding to placental tissue to a degree nearly as strong as that observed for pooled plasma from multigravid women (Fig. 5B). Antibodies to DBL6ε showed less binding inhibition but were stronger than plasma from nonimmune adults. Antibodies to AMA-1 did not inhibit binding, whereas preincubation with CSA, as expected, almost completely inhibited binding. A repeat of this assay showed similar results.

Fig 5.

Antisera to VAR2CSA DBL domains inhibit maternal IE binding to placental tissue. (A) Maternal IEs from patient 710 were incubated with either CSA, plasma from multigravid women (MG), plasma from nonimmune adults (NI), or rat antisera to 3D7 PfAMA-1 or to 3D7 VAR2CSA DBL domains and then tested for binding to placental tissue. (B) Average number of parasites per 100 fields. Error bars show standard errors of the means. Asterisks show statistical significance (***, P < 0.0001 for nonimmune adults versus plasma from multigravid women, determined by t test; **, P < 0.05 fpr AMA-1 versus each DBL domain determined by analysis of variance with Dunn's multiple comparisons). The assay was repeated twice with similar results.

DISCUSSION

In the present work, we have used antibodies generated against E. coli-expressed single and double domains of the C-terminal region of VAR2CSA to determine whether or not we can induce binding-inhibitory antibodies that are functional against multiple field isolates. We show that although the effect is variable across isolates, antibodies to DBL4ε and DBL5ε, which we have based on both a clinical isolate and a lab strain, can inhibit binding of heterologous IEs to CSA and, importantly, to placental tissue. Furthermore, we demonstrate here that immunization with multiple domains does not necessarily lead to increased binding inhibition.

Published data about the inhibitory ability of anti-DBL5ε antibodies have been contradictory. While antibodies raised against DBL5ε expressed in Pichia (25) and baculovirus (20) were not inhibitory or had very limited inhibition activity, antibodies raised against E. coli-expressed FCR3 strain DBL5ε were inhibitory against several laboratory lines selected for CSA binding using a flow binding inhibition assay (26). Our results, obtained in a static assay, support and expand the data of the latter work, further demonstrating not only inhibition of lab CSA-binding strains but also inhibition of maternal field isolates.

Our work here furthermore shows that rodent antisera to DBL4ε have binding inhibition activity comparable to that of DBL5ε against the parasite isolates that we tested. Plasma pooled from multigravid patients in our MOMS study did not demonstrate a high degree of reactivity to DBL4ε, though DBL4ε has the highest sequence conservation among all domains. This, as we suggested earlier (27), may indicate an important role of this domain in protein structure and function, including CSA binding, for the intact molecule. Naturally acquired antibodies to DBL4ε might also reduce CSA binding, and therefore, the immunogenicity of this domain may have decreased during parasite evolution. Neighboring domains DBL3X and DBL5ε, which have higher immunogenicity and are strong targets for host IgG, may have developed other masking mechanisms, such as interaction with nonspecific IgM to prevent IgG opsonization (28). Our data here suggest that although DBL4ε is not a strong target for naturally acquired IgG, antibodies generated against this domain (at least in immunized animals) appear to be sufficient for blocking parasite binding to CSA or placenta and, importantly, that this inhibition is cross-reactive against a variety of field isolates.

Other groups have reported similar findings with regard to the ability of antibodies against DBL4ε to inhibit parasite adhesion to CSA. Nielsen et al., for example, expressed strain FCR3 DBL4ε in baculovirus cells and observed that rat antibodies against this domain were highly inhibitory of FCR3 IE binding to CSA but variably inhibited heterologous lab isolates (20). Magistrado et al. tested FCR3 DBL4ε-based antisera against field isolates from Benin and Tanzania and observed inhibition between 45% and 60%, depending on the DBL4ε-based construct used for immunization (29). These data are similar in range to what we observed here. On the other hand, it should be noted that Bigey et al. (30) did not observe inhibition using mouse antibodies based on their FCR3 DBL4ε construct, even against homologous parasites. Nevertheless, these data highlight the importance of including field isolates in binding inhibition analyses. Although the percentages may be lower overall than those obtained when sera are tested against homologous lab strains, they are a better reflection of what we can expect from a potential vaccine that would be needed against field isolates, whose susceptibility to antibodies may be quite variable. Compared to plasma from multigravid women, which in this study showed inhibition over a range from 45% to 95%, the anti-DBL4ε inhibition that we observed may be considered biologically significant and protective.

Additionally, we observed strong reactivity of pooled plasma from multigravid women to 3D7 DBL6ε protein (Fig. 1). However, neither mouse nor rat antisera based on the 3D7 DBL6ε domain demonstrated appreciable binding inhibition across the maternal isolates that we tested. Fernandez et al. raised mouse antisera against strain FCR3-based DBL6ε expressed in HEK293 cells and showed that these sera were able to partially inhibit homologous parasites from binding to Sc1D cells expressing CSA. Their study found inhibition as high as 64% (19). The difference between these results may be at least partially explained by the fact that in the present study, the sera were tested against heterologous isolates. As the DBL6ε domain is generally considered the most polymorphic of the VAR2CSA domains across global strains, some variability in results is to be expected (31, 32).

We also generated antibodies to two multidomain constructs, DBL4-DBL5 and DBL5-DBL6. In assays using these antisera, we did not observe an appreciable increase in the ability to inhibit parasite binding compared to that for antisera based on the individual domains. We also did not observe an appreciable increase in inhibition when using antisera generated from coimmunization with individual DBL domains. An exception to this was observed for coimmunization with DBL5ε and DBL6ε, where the rat antisera showed an enhanced inhibitory effect compared to that of sera based on the individual domains. However, this trend did not hold true in the mouse-based antisera. Fernandez et al. found that mouse antisera to DBL5-DBL6 from the FCR3 isolate were highly inhibitory of FCR3 IE binding to BeWo cells expressing CSA under flow conditions (26). They also found that mouse anti-DBL6ε had an overall higher inhibitory activity than mouse anti-DBL5ε, which was the inverse of what we observed. These differences may be due to several factors, including the genome upon which the protein used in vaccination was based, the particular amino acid boundaries of the domains used for vaccination, the folding conditions used to generate the final product before vaccination, or differences in expression systems and vaccination protocols.

Recently, the entirety of the extracellular domains of VAR2CSA (DBL1 to DBL6) was successfully produced as one recombinant protein from both the 3D7 strain (33) and the FCR3 strain (34). Antibodies generated to the FCR3-based construct were reported to render complete inhibition of binding of recombinant protein to CSPG (condroitin sulfate proteoglycan), as well as inhibition of binding of homologous IEs to CSPG (34); it is not yet clear if antibodies against this full-length protein can inhibit heterologous IEs. When antibodies to the 3D7 full-length construct were tested for inhibition to heterologous lab isolates, no inhibition was observed (35), leaving open the question of whether or not a larger portion of VAR2CSA will enhance vaccine efficacy. To fully appreciate the role of antibodies to the full-length construct, it will be necessary to test them against field isolates to determine if they can cross inhibit those parasite lines. Because such a large protein may present a greater challenge with regard to efficient production and purification, a smaller protein that is able to generate antibodies with sufficient binding inhibition capabilities across field isolates may be preferred for the purposes of large-scale vaccine production.

Much work regarding the creation of a vaccine against PM has centered on determining which individual domains bind best to CSA and, furthermore, which domain will generate the best titer of antibody to CSA-binding parasites. DBL4ε domains from multiple strains have been implicated in the generation of antiadhesive antibodies (20, 29), despite reports that strain FCR3 and 7G8-based DBL4ε domains do not bind to CSA (12, 36–38). These data indicate that the CSA-binding ability of a domain does not necessarily correlate with the induction of protective antibodies. Moreover, the generation of a strong antibody titer does not necessarily correspond with functional ability. As an example, our ELISA (not shown) and flow cytometry data for rat anti-DBL6ε showed a strong immune response against field isolates 710 and 736 (Fig. 2D), yet the same antibodies showed very weak inhibition of these isolates (Fig. 4A). Our data here suggest that there may be no direct correlation between the ability to bind parasites infecting the mother and the ability to block parasite binding to CSA or placental tissue.

An effective vaccine against PM must confer cross-reactive protection against diverse parasites in the field, possibly by inhibiting parasite binding to placental tissue. Importantly, such a vaccine will need to be amenable to mass production. The complexity of the full-length VAR2CSA protein may thus restrict its use in large-scale vaccine generation. E. coli-expressed and refolded VAR2CSA domains are able to induce antibodies that inhibit binding of CSA-selected laboratory lines to CSA (26), an observation that we have expanded on here by testing the efficacy of these antibodies against field isolates. This report further demonstrates that we can use the highly efficient and economical E. coli system to generate proteins that induce functional antibodies with a significant degree of cross-reactivity to field isolates. These results are a promising step forward for vaccine development.