Identification of VAR2CSA Domain-Specific Inhibitory Antibodies of the Plasmodium falciparum Erythrocyte Membrane Protein 1 Using a Novel Flow Cytometry Assay

Abstract

VAR2CSA, a member of the Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1) family, is a leading candidate for use in vaccines to protect first-time mothers from placental malaria (PM). VAR2CSA, which is comprised of a series of six Duffy binding-like (DBL) domains, binds chondroitin sulfate A (CSA) on placental syncytiotrophoblast. Several recombinant DBL domains have been shown to bind CSA. In order to identify and develop recombinant proteins suitable for clinical development, DBL2X and DBL3X, as well as their respective third subdomain (S3) from the FCR3 parasite clone, were expressed in Escherichia coli, refolded, and purified. All but DBL3X-S3 recombinant proteins bound to CSA expressed on Chinese hamster ovary (CHO)-K1 cells but not to CHO-pgsA745 cells, which are CSA negative as determined by flow cytometry. All but DBL3X-S3 bound to CSA on chondroitin sulfate proteoglycan (CSPG) as determined by surface plasmon resonance (SPR) analysis. Purified IgG from rats and rabbits immunized with these four recombinant proteins bound homologous and some heterologous parasite-infected erythrocytes (IE). Using a novel flow cytometry inhibition-of-binding assay (flow-IBA), antibodies against DBL3X-S3 inhibited 35% and 45% of IE binding to CSA on CHO-K1 cells compared to results for soluble CSA (sCSA) and purified multigravida (MG) IgG, respectively, from areas in Tanzania to which malaria is endemic. Antibodies generated against the other domains provided little or no inhibition of IE binding to CSA on CHO-K1 cells as determined by the flow cytometry inhibition-of-binding assay. These results demonstrate for the first time the ability to identify antibodies to VAR2CSA DBL domains and subdomains capable of inhibiting VAR2CSA parasite-IE binding to CSA by flow cytometry. The flow cytometry inhibition-of-binding assay was robust and provided an accurate, reproducible, and reliable means to identify blocking of IE binding to CSA and promises to be significant in the development of a vaccine to protect pregnant women.

INTRODUCTION

During pregnancy, Plasmodium falciparum-infected erythrocytes (IE) bearing a preferentially expressed VAR2CSA protein are sequestered on placental syncytiotrophoblast by binding to chondroitin sulfate A (CSA) (1, 2, 3). Primigravid women are more susceptible to higher-density placental infections, resulting in maternal anemia, low birth weight, or death of the fetus (4, 5, 6). However, over multiple pregnancies, women acquire blocking antibodies that bind the parasite-encoded VAR2CSA protein expressed on the surface of parasite-IE and are associated with better pregnancy outcomes (7, 8, 9, 10).

Genetic, immunological, and binding studies identified VAR2CSA as the leading candidate for a placental malaria (PM) vaccine (9, 11, 12, 13, 14, 15). VAR2CSA is a member of the P. falciparum erythrocyte membrane protein 1 (PfEMP1) family encoded by var genes (1, 2, 3). PfEMP1s are large and highly polymorphic proteins, expressed at the surface of IE, which exhibit different adhesive phenotypes and orchestrate the binding of IE to microvascular endothelium through a variety of different host-receptor interactions (16, 17). Expression of var2csa on IE is at high frequencies in laboratory parasite isolates selected for CSA binding and placental isolates (9, 13, 14). Gene disruption has been used to show that var2csa is the primary var gene responsible for CSA binding (11, 15, 17). Furthermore, women develop antibodies that bind to the VAR2CSA protein and correlate with protection from PM disease (8, 9).

VAR2CSA is a large multidomain protein (∼350 kDa) consisting of six Duffy-binding-like (DBL) domains, a cysteine-rich interdomain region (termed CIDR_PAM), and other interdomain regions (13, 18). Whereas several DBL domains (DBL2X, DBL3X, and DBL5ε) have been shown to bind to CSA (7), single DBL domains have much lower binding activity than the full-length VAR2CSA recombinant protein (19, 20). Analysis of various multidomain constructs suggests that the VAR2CSA minimal CSA binding region resides in the N-terminal DBL1-DBL3 region (21, 22). Of these, the DBL2 and flanking interdomain regions have been proposed to comprise a “core binding region” (23). Additionally, both the solved crystal structure of DBL3X (24, 25) and analysis of DBL3X subdomains suggest that the DBL3X subdomain 3 region may contact CSA (26). Taken together with small-angle X-ray scattering (SAXS) structural studies of full-length recombinant proteins (20, 23), these findings suggest that the six DBL domains in VAR2CSA fold to form a more compact protein with high-affinity binding activity centered around DBL1X-DBL3X.

The large size of VAR2CSA and complexity of its CSA binding site have complicated vaccine development. Whereas full-length VAR2CSA recombinant proteins are highly immunogenic, the adhesion blocking activity they elicited was strain specific (19, 27). Conversely, most single-domain antigens, with the exception of DBL4ε and DBL5ε, have elicited limited adhesion blocking activity (28, 29, 30, 31, 32). More recently, larger, multidomain VAR2CSA constructs incorporating the “core” binding region (DBL2 plus flanking regions) was shown to elicit adhesion blocking antibodies that were partially cross-inhibitory on placental isolates (23, 33). However, VAR2CSA polymorphism still poses a challenge to pregnancy malaria vaccine development. To overcome interclonal diversity in VAR2CSA, one approach may be to define smaller CSA-binding fragments that can elicit broadly inhibitory antibodies.

Our current study was focused on the individual DBL2X and DBL3X domains and their corresponding third subdomains. These individual DBL domains and subdomains of approximately 25 to 50 kDa were produced in Escherichia coli and well characterized. Antibodies raised against these domains provided a means to qualify a novel flow cytometry-based inhibition-of-binding assay (flow-IBA) while identifying VAR2CSA domains amenable to antibody blockade.

MATERIALS AND METHODS

Parasite clones.

FCR3, 7G8, 3D7, Pf2004, Pf2006, and HB3A were the VAR2CSA-expressing parasite lines used for the binding studies. A4ultra is a CD36-binding parasite line that does not express VAR2CSA. Only the homologous FCR3 parasite line was used for the flow-IBA.

Design of DBL synthetic genes.

Sequences of the VAR2CSA gene (GenBank AAQ73926) were amplified using sequence-specific primers. DBL2X (residues 543 to 970), DBL2X-S3 (residues 758 to 970), DBL3X (residues 1220 to 1580), and DBL3X-S3 (residues 1445 to 1580) sequences were amplified from a previously produced, codon-optimized EcIT4 DBL2-3XL synthetic gene construct and cloned in frame with a downstream His tag in the pET24a(+) expression vector (Novagen, Madison, WI). The constructs were transformed into BL21(DE3) competent cells (Novagen) following the manufacturer's protocol. Glycerol stocks were prepared with sequence-confirmed positive clones (Integrated DNA Technologies, Coralville, IA) for storage until used for protein expression.

Recombinant protein expression in E. coli.

Five microliters of each DBL glycerol stock was used to initiate 5 ml of bacterial culture in LB medium (MP Biomedicals, Solon, OH) containing 30 μg/ml of kanamycin. Cultures were incubated overnight at 37°C, 250 revolutions per minute (RPM). Overnight cultures were transferred into 1 liter of LB medium containing 30 μg/ml of kanamycin in 2.5-liter baffle flasks and incubated at 37°C, 250 RPM, to 0.80 optical density at 550 nm (OD₅₅₀) units. Cultures were induced with 1 ml/liter of freshly prepared 1 M isopropyl-β-d-thiogalactopyranoside (IPTG) (American Bioanalytical, Natick, MA) and incubated for an additional 3 h. Bacterial cells were harvested by centrifugation at 10,000 × g for 15 min. One gram of bacterial cell pellet was resuspended in 10 ml of lysis buffer (10 mM Tris-HCl, 10 mM EDTA, 100 mM NaCl, and 2.6 mM dithiothreitol [DTT], pH 8.0), mixed completely, and lysed by passage through a Microfluidizer fluid processor (Microfluidics Corporation, Newton, MA). Soluble proteins were separated from the inclusion bodies (IB) by centrifugation at 20,000 × g for 45 min. IB pellets were washed by resuspending in distilled water using a handheld homogenizer and centrifuged at 20,000 × g for 45 min. The recombinant proteins in the washed IB were extracted by solubilization. One gram of IB pellets was dissolved in 5 ml of IB solubilization buffer (110 mM sodium phosphate, 11 mM Tris-HCl, 7 M guanidine-HCl, 20 mM betamercaptoethanol, pH 8.0) and stirred overnight at room temperature. The His₆-affinity-tagged recombinant DBL proteins were isolated from the solubilized IB by Ni-Sepharose affinity column chromatography using the Akta system (GE Healthcare Bio-Sciences Corp., Piscataway, NJ), following the manufacturer's recommendations. Proteins were screened for optimum refolding condition by rapid dilution using 16 different refolding buffers followed by a large-scale column refolding. Briefly, DBL3X and DBL3X-S3 were refolded on a Sephacryl S300 26/60 column (GE Healthcare) with a buffer containing 50 mM Tris-HCl, 50 mM NaCl, 40 mM Sucrose, 500 mM l-arginine–HCl, 2 mM EDTA, 2 mM DTT, and 5 mM cystamine dihydrochloride, pH 7.6, at 0.44 ml per min. DBL2X-S3 was refolded similarly, with a buffer containing 55 mM morpholineethanesulfonic acid (MES), 10.56 mM NaCl, 0.44 mM KCl, 1.1 mM EDTA, 440 mM sucrose, 550 mM l-arginine–HCl, and 1 mM DTT, pH 6.5. DBL2X was refolded with 55 mM Tris-HCl, 264 mM NaCl, 11 mM KCl, 550 mM guanidine-HCl, 1.1 mM EDTA, 1 mM glutathione (GSH), and 0.1 mM glutathione disulfide (GSSG), pH 8.2. Refolded DBL2X was purified by passage through a Q XL column (GE Healthcare) in a buffer containing 20 mM sodium phosphate and 5 mM EDTA, pH 6.6. The DBL2X protein flowthrough from the Q XL column was concentrated by binding to an SP column (GE Healthcare) using 20 mM sodium phosphate and 5 mM EDTA, pH 6.6, and eluted with 20 mM sodium phosphate, 1 M NaCl, and 5 mM EDTA, pH 6.6.

The refolded proteins were further purified by hydrophobic interaction column chromatography (HIC) using phenol-650 resin (Tosoh Bioscience, Tokyo, Japan) with ammonium sulfate buffer, followed by size exclusion chromatography (SEC) using Superdex S200 and S75 columns with phosphate-buffered saline (PBS), pH 7.4, buffer for DBL3X and DBL3X-S3, respectively. DBL2X was purified using Superdex S200 and buffer containing 50 mM Tris-HCl, pH 7.76, and 50 mM NaCl, while DBL2X-S3 used Superdex S75 and PBS, pH 7.4. Protein yields per liter of bacterial culture were approximately 25 mg (each) for DBL2X and DBL3X and 64 mg (each) for DBL2X-S3 and DBL3X-S3.

Biochemical and biophysical analysis.

Edman degradation was performed for N-terminal protein sequencing, and the theoretical masses were verified by electrospray ionization (ESI)-mass spectrometry (Research Technologies Branch, NIH, Rockville, MD). Protein purity was determined by reversed-phase high-performance liquid chromatography (RP-HPLC) (Waters, Milford MA), and solution mass was determined by analytical SEC using a multiangle light scattering (MALS) device (Wyatt Technology, Santa Barbara, CA), following an already-described protocol (34).

Recombinant protein binding to CHO cells.

Both Chinese hamster ovary (CHO)-K1 cells expressing chondroitin sulfate A (CSA) proteoglycan and mutant CHO-pgsA745 cells (deposited by Jeffrey D. Esko in ATCC [Manassas, VA]) deficient in proteoglycan synthesis were cultured in complete F-12K medium (Kaighan's modification of Ham's F-12 medium) containing 10% fetal calf serum (ATCC) and 1% penicillin-streptomycin-glutamine (Invitrogen, Grand Island, NY). Cells were adherent and were grown to 100% confluence in T75 culture flasks at 37°C in 5% CO₂. Adherent cells were lifted after the removal of the spent medium by incubation at 37°C for 10 min with 3 ml of 0.25% trypsin–0.53 mM EDTA in Hanks balanced salt solution (HBSS) without calcium or magnesium (ATCC). The reaction was stopped by adding 7 ml of complete F-12K medium. Detached CHO cells were recovered by centrifugation at 125 × g for 7 min at room temperature and resuspended in flow cytometry staining buffer (FSB) containing PBS and 1% (wt/vol) bovine serum albumin (Sigma-Aldrich, St. Louis, MO). One million CHO-K1 or CHO-pgsA745 cells were incubated with 100 μg of each purified recombinant protein for 30 min at room temperature. The CHO cells were washed twice by centrifugation at 200 × g for 5 min using 200 μl of FSB each and then sequentially incubated with 1 μg of mouse anti-His₆ tag (Invitrogen Corp., Camarillo, CA) and 1 μg of goat anti-mouse IgG-fluorescein isothiocyanate (FITC) conjugate (Imgenex Corp., San Diego, CA) for 30 min. Cells were washed as described above and incubated in 2 μl of propidium iodide (Invitrogen, Eugene, OR) for 2 min. Samples were washed once and resuspended in FSB for FACSCalibur flow cytometer (BD Bioscience, San Jose, CA) analysis. Data analyses were performed using the FlowJo 7.6.5 software program (Tree Star, Inc., Ashland, OR). Propidium iodide stain was used to gate dead from live cells, and the FITC-stained recombinant proteins bound to CHO-K1 cells were visualized at its 488-nm wavelength emission spectra by flow cytometry. Median fluorescence intensities (MFI^⁎) were calculated using the FlowJo software program, and the fold change was obtained by dividing the test MFI^⁎ by control MFI^⁎.

SPR analysis.

A Biacore3000 system (GE Healthcare Inc., NJ.) was used to perform surface plasmon resonance (SPR) experiments. Amine coupling chemistry was used to covalently attach NeutrAvidin protein (Thermo Scientific, Rockford, IL) to carboxymethylated dextran hydrogel on a CM5 sensor chip (GE Healthcare). Briefly, reactive N-hydroxysuccinimide esters were introduced on the surface through modification of the carboxymethyl groups by an 8-min injection of a mixed solution of 0.4 M 1-ethyl-3-(3-dimethyl-aminopropyl) carbodiimide (EDC) and 0.1 M N-hydroxysuccinimide (NHS). A 100-μg/ml amount of NeutrAvidin biotin binding protein diluted in 10 mM sodium acetate (pH 5.0) was immobilized on a CM5 sensor chip by a 10-min injection through the flow cells to the N-hydroxysuccinimide esters. After coupling, any remaining reactive esters were blocked with a 7-min injection of 1 M ethanolamine (pH 8.5). The flow rate and detection temperature were 10 μl/min and 25°C, respectively. The NeutrAvidin-CM5 surface was conditioned with 3 consecutive injections of 1 M NaCl in 50 mM NaOH at a flow rate of 10 μl/min and a contact time of 1 min each. Fifty μg/ml of biotin-labeled bovine chondroitin sulfate proteoglycan (CSPG-b) (Decorine; Sigma-Aldrich, St. Louis, MO) in HBS-EP buffer (0.01 M HEPES, pH 7.4, 0.15 M NaCl, 3 mM EDTA, 0.005% surfactant P20) (GE Healthcare) was immobilized on the NeutrAvidin-CM5 sensor chip by injecting at the rate of 10 μl/min and with a contact time of 1 min over flow cells 2 and 4 until a plateau was reached at 400 response units (RU). Control proteins were similarly immobilized in reference flow cells 1 and 3 by injecting 10 μg/ml of irrelevant biotin-labeled scrambled 47-mer peptide in HBS-EP buffer until a plateau was reached at 600 RU.

Prior to protein analysis, all flow cells were equilibrated by a continuous flow of HBS-EP buffer at 30 μl/min. Serial dilutions of the DBL2X, DBL2X-S3, DBL3X, and DBL3X-S3 proteins in HBS-EP buffer were injected into both flow cells 3 and 4 at 30 μl/min. Specific protein binding to CSA on the immobilized CSPG-b was obtained by subtracting the response on reference flow cell 3 from that on test flow cell 4. Injection of 5 μl of 10 mM NaOH, pH 10 (GE Healthcare), was sufficient to regenerate both flow cells to levels similar to the starting baseline after each binding experiment. Kinetics and affinity determination employed the 1:1 Langmuir binding model to fit experimental data using the BIAevaluation software program (version 4.1.1) (GE Healthcare).

Animal immunization, antibody production, and ELISA.

Rat and rabbit anti-DBL antibodies were produced by immunization with 25 μg and 100 μg, respectively, of each DBL recombinant protein formulated in Montanide ISA720 adjuvant (Seppic), following the manufacturer's recommendations. Rats were boosted three times, while rabbits received two boosts at 3-week intervals using the same amount of proteins used for the priming. Immunization controls received only PBS formulated in Montanide ISA720. All animals were bled prior to each immunization and 2 weeks after the final immunization. Blood samples were processed and stored as sera in frozen aliquots. Antibody titers were quantified by enzyme-linked immunosorbent assay (ELISA). Briefly, the wells of the 96-well ELISA plates were coated with 100 μl of antigen diluted in coating buffer (Na₂CO₃ [pH 9.6]) to a final concentration of 1 μg/ml and incubated overnight at 4°C. Blocking buffer (5% skim milk in Tris-buffered saline [TBS]) was used to block binding of nonspecific antibody to the wells for 2 h at room temperature. Wells were washed four times with TBS–0.5% Tween solution using an automated plate washer (Terra Universal Inc., Fullerton, CA). One hundred microliters of serial dilutions (1:50, 1:500, 1:50,000, and 1:100,000) of sera in diluent solution (1% BSA in TBS–0.5% Tween) were incubated for 2 h. ELISA plates were washed as described above, followed by 2 h of incubation with either alkaline phosphatase (AP)-conjugated goat anti-rat IgG(H+L) or goat anti-rabbit IgG(H+L) (KPL Inc., Gaithersburg, MD). Wells were incubated with AP substrate for 20 min after the final wash. Absorbance was recorded using a Spectra Max 340PC instrument (Molecular Devices, Sunnyvale, CA).

Antibody IgG purification.

Rat IgG antibodies were purified using Gamma-Bind Plus columns (GE Healthcare), while rabbit IgG antibodies were purified using a protein G column (GE Healthcare), following the manufacturer's recommendations. Purified antibodies were neutralized with 100 μl of 1 M Tris-HCl, pH 8.0, per 1 ml solution and dialyzed against PBS, pH 7.4.

Parasite clones, sequencing, and analysis of the var2csa gene.

P. falciparum parasites were grown in RPMI 1640 supplemented with 10% (vol/vol) heat-inactivated human serum, 25 mM/liter sodium bicarbonate, 0.125 mg/ml gentamicin (Invitrogen), and 5% hematocrit of human group O⁺ erythrocytes. CSA binding laboratory isolates, FCR3/IT4-CSA (24), 7G8-CSA, HB3-CSA allele A and HB3-CSA allele B, and NF54-CSA (29), and Pf2004-CSA and Pf2006-CSA (35), were maintained by periodic selection on CSA. For non-CSA binding controls, A4ultra (CD36 binder expressing the IT4var14 gene) was used. Genotyping of parasites was done using MSP1/MSP2 primers according to published approaches (36). var gene transcription was assessed by quantitative reverse transcription-PCR (qRT-PCR) using universal primers as previously described (29). The DBL2X domain portion of var2csa was sequenced for the isolates Pf2004-CSA and Pf2006-CSA. Briefly, the DBL2X domain portion of var2csa was amplified using the primers VAR2CSA DBL2F (5-CATGTAACACACATAGCTC) and DBL2R (5-CACCTGAATTGTTTCCACA), which had been designed to flank the DBL2X domain. PCR products were cloned into Escherichia coli and sequenced using standard methods. Pf2004-CSA and Pf2006-CSA sequences were aligned using the software program Clustal X2.0. Sequences for FCR3/IT4, 3D7/NF54, and HB3 were already available (13). The DBL3X domain portion of Pf2004-CSA and Pf2006-CSA VAR2CSA was recently published (29).

Assay of binding of DBL antibodies to CSA-expressing parasites.

Mature-stage IE grown in O⁺ blood were incubated with rabbit or rat IgG purified from polyclonal sera as described above. For each assay, 10 million erythrocytes infected with 5 to 8% trophozoites were incubated with either 0.2 mg/ml or 0.5 mg/ml of rabbit IgG or 0.5 mg/ml of rat IgG. Bound antibodies were detected by adding Alexa Fluor 488-conjugated goat anti-rabbit IgG (A-11034; Molecular Probes) or Alexa Fluor 488-conjugated goat anti-rat IgG (A-11006; Molecular Probes). LSRII (BD Bioscience, San Jose, CA) was used to collect sample data, analyzed by the FlowJo 8.1 software program (Tree Star Inc.). Binding was calculated as the mean of the adjusted geometric mean of fluorescence intensity (MFI) for samples run in duplicate or triplicate. The adjusted MFI was calculated as (IE_i − UE_i) – (IE_p – UE_p), where IE_i is the MFI of IE following incubation in immune anti-VAR2CSA IgG, UE_i is the MFI of uninfected erythrocytes following incubation in immune anti-VAR2CSA IgG, IE_p is the MFI of infected erythrocytes following incubation in IgG from controls that receive immunization with only adjuvant in PBS, and UE_p is the MFI of uninfected erythrocytes following incubation in IgG from controls that receive immunization with only adjuvant in PBS.

Flow cytometry inhibition-of-binding assay (flow-IBA).

Erythrocytes infected with the FCR3-CSA-expressing parasites at the late trophozoite to early schizont stages were isolated using the magnetically activated cell sorting (MACS) 25LS separation column (Miltenyi Biotec, Auburn, CA), following the manufacturer's recommendations. Briefly, the MACS column was rehydrated with 4 ml of complete prewarmed (37°C) RPMI 1640 medium (RPMI) while mounted on a strong magnetic field before the parasite-IE culture was allowed to flow through the column. Then, the column was washed with RPMI to remove any adhering non-IE. The column was removed from the magnetic field, and the captured parasite-IE at late trophozoite to early schizont stages were eluted with 8 ml of FSB (PBS plus 1% bovine serum albumin). MACS-purified parasite-IE were incubated with CFSE (carboxyfluorescein succinimidyl ester) dye (Molecular Probes, Inc., Eugene, OR) at 1 million cells per microliter of 5 mM CFSE in FSB in a shaker at 250 RPM for 15 min. All incubations were performed at 37°C. Cells were pelleted and washed with RPMI by centrifugation at 530 × g for 3 min, and incubation was continued for an additional 30 min in RPMI. Stained cells were pelleted, and the reaction was quenched by washing with prewarmed FSB. CFSE-stained IE were incubated for 45 min with DBL domain antibodies. CHO cells were harvested using trypsin-EDTA (ATCC or HyClone; Thermo Scientific, South Logan, UT), following the manufacturer's recommendations. One million CHO cells per microliter of 5 mM VFSE450, a fluorescein diacetate (FDA) derivative dye (AAT Bioquest, Sunnyvale, CA), were incubated in prewarmed FSB for 15 min in a shaker at 250 RPM. Cells were washed with prewarmed complete F-12K medium by centrifugation at 530 × g for 3 min. Incubation was again continued for an additional 30 min in F-12K medium. Stained cells were pelleted, and the reaction was quenched by washing with prewarmed FSB. One million VFSE-stained CHO cells were combined with two million CFSE-stained IE previously incubated with anti-DBL antibodies. A control sample containing CFSE-stained IE and VFSE-stained CHO cells was also prepared. The combinations were incubated for 45 min, and flow cytometry data collection was performed using an LSR II or LSR II Fortesa flow cytometer (BD Bioscience, San Jose, CA) before and after 10 min of fixation in 3% paraformaldehyde. To quantify CHO-IE binding, CHO cells were first gated by size, and then doubly positive cells were gated, indicating IE bound to CHO-K1 cells. The FlowJo 7.6.5 software program (Tree Star, Inc.) was used for data analysis.

Western blot assay.

Recombinant DBL domains and subdomains proteins (0.5 μg each) separated on SDS-PAGE gels (Novex by Life Technologies, Carlsbad, CA) were transferred on nitrocellulose paper using the iBlot gel transfer system (Novex). Blots were incubated overnight at 4°C with blocking solution (KPL Inc.). This was followed by 1 h of incubation at room temperature with either purified pooled domain/subdomain-specific rabbit antibodies or purified pooled multigravida IgG antibodies (10 women from Muheza, Tanzania) in blocking buffer. Blots were washed three times, 5 min each, using wash buffer (KPL Inc.). Incubation was continued at room temperature for 1 h with alkaline phosphatase-conjugated anti-rabbit IgG(H+L) or anti-human IgG(H+L) (KPL Inc.). After a final wash, as above, antibodies were visualized using the 5-bromo-4-chloro-3-indolylphosphate–nitroblue tetrazolium (BCIP-NBT) alkaline phosphatase substrate (KPL Inc.).

RESULTS

Recombinant protein production and characterization.

Each of the DBL2X, DBL2X-S3, DBL3X, and DBL3X-S3 regions of an E. coli codon-optimized var2csa gene from the FCR3 parasite was cloned into the pET24a(+) vector for expression in E. coli strain BL21(DE3). Schematics of the constructs were illustrated in Fig. 1A. Each clone was fermented, and cell paste containing induced protein in inclusion bodies was used for the preparation of refolded purified protein. The presence of disulfide bond formation in the refolded protein was assessed by SDS-PAGE, in which a mobility shift was observed upon reduction (Fig. 1B). Analytical size exclusion chromatography with online multiangle light scattering (SEC-MALS) showed that each of the four refolded proteins existed primarily as a monomer in solution, and protein identity was determined by RP-HPLC (see Table S1 in the supplemental material). The molar masses observed by SEC-MALS and masses determined by time-of-flight mass spectrometry (TOF-MS) of all the four DBL proteins were within the expected ranges (see Table S1) and were similar to those of the DBL3X-S3 protein shown as a representative example in Fig. S1. Edman degradation was used to confirm the identity of the N-terminal sequence of each of the four purified proteins.

Fig 1.

Recombinant DBL2X and DBL3X and their respective subdomain 3 constructs. (A) SynEcA4-DBL2-3X gene, showing DBL2X, interdomain (ID) region, and DBL3X (GenBank accession no. AAQ73926). Below are the DBL2X fragment (residues 543 to 970), DBL2X-S3 fragment (residues 758 to 970), DBL3X fragment (residues 1220 to 1580), and DBL3X-S3 fragment (residues 1445 to 1580). All four constructs each contained a C-terminal 6×His tag. (B) Coomassie blue-stained DBL2X, DBL2X-S3, DBL3X, and DBL3X-S3 proteins under nonreduced and reduced conditions.

Binding of DBL recombinant proteins on CSA-expressing CHO cells and SPR analysis.

Since VAR2CSA binds to CSA, the binding activity of each DBL domain and subdomain was assessed using CHO-K1 cells which express CSA (CHO-CSA⁺) or the CSA-deficient CHO-pgsA745 cell (CHO-CSA⁻) control. CHO-CSA⁺ cells bound DBL2X and DBL3X at high median fluorescence intensities compared to results for CHO-CSA⁻ cells (7.8-fold and 8.0-fold, respectively), whereas DBL2X-S3 and DBL3X-S3 showed lower binding (2.2-fold and 1.3-fold, respectively) by flow cytometry (Fig. 2). In order to better characterize the binding observed on the CHO-CSA⁺ line, surface plasmon resonance (SPR) was used. SPR analysis of DBL2X, DBL3X, and DBL2X-S3 binding on CSPG demonstrated that DBL2X and DBL3X bound to CSA with an equilibrium dissociation constant (K_D) in the low nanomolar range (1.47 × 10⁻⁹ M and 8.59 × 10⁻⁹ M, respectively) (Fig. 3A and B). DBL2X-S3 had nearly a 24-fold decrease in binding affinity compared to that of DBL2X (K_D, 3.52 × 10⁻⁸ M) (Fig. 3C). DBL3X-S3 essentially did not bind to CSA on CSPG (K_D, 1.88 × 10⁻⁴ M) (Fig. 3D).

Fig 2.

Binding interactions of DBL2X, DBL2X-S3, DBL3X, and DBL3X-S3 with CSA on CHO cells by flow cytometry. Histograms and median fluorescence intensities of the DBL2X, DBL2X-S3, DBL3X, and DBL3X-S3 recombinant proteins binding to CHO-pgsA745 cells (A) or CHO-K1 cells (B) are shown. Median fluorescence intensities were used to report the differences in binding between the two cell lines.

Fig 3.

SPR kinetics analysis of DBL domains and subdomains on immobilized CSPG. Kinetics fits of DBL2X (A), DBL3X (B), DBL2X-S3 (C), or DBL3X-S3 (D) binding to CSA on GSPG are shown.

Binding of DBL antibodies to CSA-expressing parasites.

To investigate whether antibodies generated against the VAR2CSA DBL domains could recognize native protein, rabbit and rat antisera were produced (Table 1 shows titers at an OD of 1), and IgG were purified and tested on IE. Antibodies to each recombinant protein were generated in rabbits and rats (Table 1). As a control for anti-His reactivity, an orthologous recombinant protein produced with or without a His tag was used for ELISA capture. No significant reactivity was observed against either protein (data not shown). Initially, the homologous FCR3 VAR2CSA-expressing parasite line was evaluated by flow cytometry, followed by a panel of five CSA-binding isolates from diverse geographical locations. A CD36-binding control parasite line, A4ultra, which does not express VAR2CSA (29), was included as a specificity control. All VAR2CSA-expressing parasite lines bound with high affinity to CSA, except the low-affinity-CSA-binding 3D7/NF54 line.

Table 1.

Reactivities of rat and rabbit IgG with recombinant DBL domains and subdomains as determined by ELISA^a

Animals	IDb	Reactivity, ELISA units (OD = 1.0)c
		Anti-DBL2X	Anti-DBL2X-S3	Anti-DBL3X	Anti-DBL3X-S3
Rabbits	1	32,900	23,800	66,300	96,687
	2	23,300	30,700	106,700	63,040
Rats	1	14,640	1,860	57,100	5,927
	2	8,660	7,170	47,000	13,235
	3	12,130	7,460	41,700	23,488
	4	16,510	2,950	52,100	9,309

^aFour rats and two rabbits were immunized with either DBL antigens or adjuvant alone in PBS.

^bIdentification no.

^cEach column gives the dilution of IgG for an observed ELISA OD of 1.0.

Of interest, rabbit anti-DBL3 antibodies showed broader cross-reactivity to heterologous CSA-binding parasites than anti-DBL2 antibodies (Table 2). Rat anti-DBL3 antibodies were also cross-reactive but less so than rabbit antibodies (Table 2). Similar observations have previously been reported for anti-DBL3 antibodies generated against the recombinant Pichia pastoris DBL3 protein (37). To investigate this phenomenon in more detail, we compared VAR2CSA sequence identity between FCR3 and other parasites in the panel (see Fig. S3 in the supplemental material). Overall, DBL3 (mean, 90% identity; range, 85.7 to 92.8%) was more conserved than DBL2 (mean, 80% identity; range, 74.9 to 83.1%) (see Table S2). Curiously, anti-DBL3 antibodies cross-reacted best on the HB3-CSA and 7G8-CSA parasite lines, despite the fact that the DBL3 domain in HB3-VAR2CSA had lower sequence identity than that of other parasites in the panel (see Table S2). By comparison, rat anti-DBL2 antibodies also cross-reacted on HB3-CSA and 7G8-CSA parasites, and these two sequences were more highly related to FCR3-VAR2CSA (see Table S2). In addition, DBL domain-specific antibodies showed broader reactivity against heterologous parasite lines than the subdomain-specific antibodies (Table 2). It is surprising that the DBL3X-S3 antibodies were not more cross-reactive, because this region is highly conserved between VAR2CSA alleles (mean, 97% identity; range, 93.8 to 99.2%) in comparison to subdomain 3 in DBL2X (mean, 82% identity; range‘ 77.5 to 86.3%) (see Table S2). Taken together, these findings suggest that cross-reactive vaccine antibodies did not target highly conserved epitopes in DBL3. Although FCR3 and HB3 appear to share a cross-reactive epitope(s) in subdomain 3, this epitope appears to be polymorphic or may differ in exposure between different VAR2CSA alleles.

Table 2.

Binding of purified rat and rabbit DBL2X NS DBL3X and their respective subdomain antibodies on var2csa-expressing parasite lines

Flow-IBA development and evaluation of antibody blockade of parasite binding to CHO-CSA⁺ cells.

To investigate if antibodies could inhibit the binding of IE to CSA, we developed a novel flow cytometry-based inhibition assay (flow-IBA). CSA-IE were coincubated with CHO-CSA⁺ cells or negative control CHO-CSA⁻ cells. In this assay, CFSE-labeled IE were preincubated with anti-DBL antibodies before mixing with VFSE-labeled CHO cells. In control experiments, CFSE-stained IE bound to the VFSE-stained CHO-CSA⁺ (CHO-K1) cells (8.78%) but not the VFSE-stained CHO-CSA⁻ (CHO-pgsA745) cells (1.18%) (Fig. 4A and B). The percent inhibition of IE binding by antibodies was determined by calculating the percent reduction of the difference between amounts bound to CHO-K1 cells and to CHO-pgsA745 cells (Fig. 5). In control experiments, 0.5 mg/ml of pooled multigravida (MG) IgG inhibited binding by approximately 60%, and soluble CSA inhibited binding by >90%. Of the four vaccine immunogens, the DBL2X, DBL2X-S3, and DBL3X proteins elicited no adhesion blocking activity above that of the adjuvant-alone control (P > 0.3588 by Student's t test). In contrast, rabbit anti-DBL3X-S3 IgG and MG-IgG inhibited IE binding to CSA on CHO-K1 cells by 35% and 75% (P < 0.0181 and P < 0.0210 by Student's t test), respectively (Fig. 5). No significant differences were observed for experiments performed with 0.5 mg/ml and 1.0 mg/ml of antibodies (data not shown). A combination of DBL2X and DBL3X antibodies was tested, but no additive or synergistic effect was observed (data not shown). Fixed samples a week after storage at 4°C gave results similar to those with unfixed samples.

Fig 4.

Flow-IBA data of homologous FCR3csa parasite-IE binding to CHO cells. (A) VFSE-stained CHO-K1 (CSA+) cells combined with CFSE-stained IE; (B) VFSE-stained CHO-pgsA745 (CSA-) cells combined with CFSE-stained IE.

Fig 5.

Inhibition of FCR3csa parasite-IE binding to CSA on CHO-K1 cells by purified preabsorbed rabbit DBL domains and subdomains IgG (antibodies). Percent inhibition of FCR3csa parasite-IE binding to CSA on CHO-K1 cells using various IgG antibodies and soluble CSA. Each dot represents an independent flow cytometry inhibition-of-IE-binding experiment. Diamonds represent mean values, and error bars represent standard errors. DBL3X-S3 antibodies significantly inhibited IE binding to CSA compared to that with adjuvant control antibodies, by 35% (P < 0.0181 by Student's t test), while MG-IgG inhibited IE binding to CSA by 75% (P < 0.0210 by Student's t test).

MG IgG recognized DBL3X and DBL3X-S3 that produced inhibitory antibodies.

The DBL2X domain is thought to be part of a “core” CSA-binding region (23, 33). To better understand the inability of DBL2X and DBL2X-S3 to generate inhibitory antibodies, we compared the abilities of the four recombinant proteins to be recognized by purified IgG from multigravida women or vaccine antibodies. Surprisingly, maternal IgG reacted only with the DBL3X and DBL3X-S3 proteins and not the DBL2X and DBL2X-S3 proteins (Fig. 6A). In contrast, all 4 recombinant proteins were well recognized in Western blots by their respective domain-specific antibodies (Fig. 6B). DBL2X was also poorly recognized by MG-IgG compared to results for DBL3X in ELISA (Fig. 6C). Similarly to results for DBL3X, MG-IgG also reacted with DBL3X-S3 as determined by ELISA (data not shown). Altogether, this indicates that the DBL2X domain is less immunodominant and suggests the possibility that it may be partially hidden from antibodies.

Fig 6.

(A and B) Western blot of recombinant DBL2X, DBL2X-S3, DBL3X, and DBL3X-S3 proteins probed with multigravida (MG) IgG (A) or domain-specific pooled rabbit IgG (B). (C) ELISA of recombinant DBL2X and DBL3X with serial dilutions of pooled MG-IgG. The control columns contain DBL2X- and DBL3X-coated wells without MG IgG but stained with secondary antibodies only.

DISCUSSION

Production of blocking antibodies against the parasite-encoded VAR2CSA protein on IE correlates with reduction in placental malaria disease burden in multigravidae women living in areas to which malaria is endemic (8). Despite this observation, successful vaccines against PM have been elusive due to the large size and polymorphic nature of VAR2CSA (5, 38, 39). In this study, we investigated whether DBL2X or DBL3X single-domain or subdomain 3 constructs, is the minimal CSA binding region (33), could elicit inhibitory antibodies. Previous work has suggested that the DBL2X and DBL3X domains contain CSA-binding activity, and binding activity has been mapped to the DBL3 subdomain 3 (12, 24, 40). However, it has been questioned whether single domains accurately recapitulate the CSA binding site in native VAR2CSA (41), because single domains have much lower binding activity and more promiscuous binding specificity than full-length recombinant proteins (19, 20).

Large-scale production of pure and high-quality DBL domains and subdomains were possible with the E. coli expression system due to their small sizes, thus allowing the generation of specific antibodies that may result in enhanced inhibitory activities. The DBL2X and DBL3X recombinant proteins bound in a specific manner to CHO-CSA⁺ cells and not to CHO-CSA⁻ cells (Fig. 2). In contrast, DBL2X-S3 bound very weakly, while DBL3X-S3 did not bind CSA on CHO-K1 cells by flow cytometry (Fig. 2) and failed to bind CSA on CSPG as determined by SPR analysis (Fig. 3), but antibodies against DBL3X-S3 blocked IE binding to CSA (Fig. 5). Previous nuclear magnetic resonance (NMR) studies have shown that the DBL3X-S3 domain can bind CSA (40), suggesting that on its own this region may have very weak binding affinity for CSA. These weakly interacting DBL3X-S3 antibodies may be sufficiently specific to effectively block parasite-IE binding to CSA.

In immunization studies, both the DBL2X and DBL3X antigens elicited antibodies that could react with homologous FCR3-CSA IE. The reactivities of rat and rabbit antibodies against the DBL2X and DBL3X immunogens differed. Although it is not unusual to find differences in antibody reactivities between different animal species, it is unclear whether these findings have any biological significance. However, while the DBL2X antibody had poor cross-reactivity, the DBL3X antibody was highly cross-reactive on heterologous CSA-binding IE. By comparison to full-length domains, the DBL2X-S3 and DBL3X-S3 antigens elicited lower titers of antibody to homologous parasite isolates. In addition, the DBL3X-S3 antibody was not as broadly cross-reactive as the DBL3X antibody, although it did cross-react on HB3-CSA parasites. This finding suggests that antibody epitopes contributing to DBL3 cross-reactivity are located both within and outside subdomain 3. The sequence of DBL3X-S3 appears to be highly conserved across different parasite alleles (see Fig. S3B in the supplemental material).

The inhibitory differences between antibodies specific for these DBL domains and subdomains were explored during the development and optimization of our novel flow-IBA. Flow-IBA analysis was performed using two different strategies (gating for doubly positive cells or gating for large CHO cells bearing CFSE-IE), yielding identical results. The small amount of IE (2 × 10⁶ cells in 50 μl) required for the assay makes it ideal for direct analysis of maternal parasite-IE field isolates without prior culturing. Using flow-IBA, only antibodies against DBL3X-S3 inhibited about 35% of the homologous FCR3csa parasite-IE binding to CSA on CHO-K1 cells (Fig. 5). None of the other DBL domain- and subdomain-specific antibodies that we tested gave significant CSA inhibition by flow-IBA (Fig. 5). Along with a recently described flow-based assay using syncytiotrophoblast microvillous membrane preparations (21), this provides another new and quantitative approach for assessing inhibitory antibody activity.

It is puzzling why antibodies targeting the presumed minimal VAR2CSA binding region (DBL1X-DBL3X) are able to react with the surface of CSA-binding IE by flow cytometry but are unable to inhibit IE binding to CSA (23). One explanation may be the complexity of the CSA binding site. Single DBL domains and subdomains may not fully recapitulate the high-affinity binding site similar to that created by multiple DBL domains and interdomain regions. In this regard, longer DBL2X recombinant proteins that contain the flanking domains (e.g., NTS-DBL2X) could elicit cross-reactive antibodies that inhibit IE binding (33), and in other work the minimal DBL2X construct that elicited adhesion-blocking antibodies included the flanking interdomain regions (ID1-DBL2-ID2a) (23). Thus, it is possible that the best blocking antibodies are ones that recognize quaternary epitopes between two or more VAR2CSA domains/interdomain regions. Another possibility is that the CSA-binding site may be partially shielded from antibodies due to quaternary interactions within the protein or protein-protein interactions at the IE knob-like protrusions. Also, recombinant DBL3X may be folded in such a way that some critical epitopes present in recombinant DBL3X-S3 may be masked. Of interest, seroepidemiological studies of five of the six VAR2CSA DBL domains suggest that DBL3X and DBL5ε are more immunodominant than other domains (42, 43, 44). However, DBL2X has not been analyzed in previous seroepidemiological studies. It is notable that DBL2X was not well recognized by maternal IgG in comparison to DBL3X (Fig. 6C). This suggests that the “core CSA binding region” in the native protein may not be as accessible to antibodies prior to CSA binding. This suggests that this region of the protein is not accessible in its native form prior to CSA binding.

In conclusion, this study suggests that single DBL and subdomain antigens may encode binding activity capable of generating adhesion-blocking antibodies. In addition, flow-IBA is a highly robust and reproducible inhibition-of-binding assay that will enable the investigation of maternal immunity and the development of a malaria vaccine for pregnant women.