Towards a comprehensive Plasmodium falciparum merozoite cell surface and secreted recombinant protein library

Proteins were expressed essentially as described previously[27]. Briefly, full-length secreted molecules and the entire ectodomains of membrane-embedded proteins were identified using transmembrane[28], GPI-anchor[29], and signal peptide[30] prediction software. To prevent the inappropriate addition of glycans, which are absent from Plasmodium proteins[31], all potential N-linked glycosylation sites (N-X-S/T, where X is not proline) were systematically mutated by substituting alanine for serine/threonine at these sites. Gene constructs were made by gene synthesis (GeneartAG) using sequences that were codon-optimized for expression in human cells. Protein coding sequences were flanked with unique NotI and AscI restriction sites and subcloned into a derivative of the pTT3 expression plasmid between a 5’ mouse variable κ light chain signal peptide[32], and a 3’ tag consisting of the rat Cd4 domains 3 and 4 followed by an enzymatic biotinylation sequence and a hexahistidine tag[33]. Proteins were expressed as soluble monobiotinylated proteins by transient cotransfection of HEK293E cells with the BirA biotin ligase[34] and harvested six days post transfection[35]. All expression plasmids are openly available from Addgene[36].

His-tagged recombinant merozoite proteins were purified from spent tissue culture supernatants by using a custom-built, piston-driven, sample-loading apparatus as described previously[33]. Briefly, a 96-well His MultiTrap HP filter plate (GE Healthcare) was pre-equilibrated with binding buffer (20 mM sodium phosphate, 40 mM imidazole, 0.5 M NaCl, pH 7.4) at a flow rate of 1 mL/min. The harvested tissue culture supernatants (~200 mL for each protein) were supplemented with imidazole (10 mM) before loading each well at 1 mL/min. The plate was washed with 1.6 mL of binding buffer and proteins eluted with 0.2 mL of elution buffer (20 mM sodium phosphate, 0.4 M imidazole, 0.5 M NaCl, pH 7.4).

ELISAs were performed as previously described[37]. Briefly, purified recombinant biotinylated proteins were serially diluted in PBS-T (PBS, 0.1% Tween-20) with 2% BSA, and captured on streptavidin-coated, 96-well plates (NUNC) for one hour before washing and incubating with the anti-Cd4 monoclonal antibody OX68 (1 μg/mL in PBS-T, 2% BSA) for another hour. Plates were washed and incubated with an anti-mouse IgG (Sigma) secondary antibody conjugated to alkaline phosphatase for one hour before further washes and incubation with p-nitrophenyl phosphate (Substrate 104; Sigma) at 1 mg/mL. Absorbance was measured at 405 nm on a PHERAstar plus (BMG Labtech). Concentrations were calculated by comparison to known standards.

Each purified recombinant protein was resolved by SDS-PAGE under reducing conditions before blotting onto Hybond-P PVDF membrane (GE Healthcare) for one hour at 30 V. Membranes were blocked with 2% BSA, in PBS-T and incubated with 0.02 μg/mL of streptavidin-HRP (Jackson Immunoresearch) diluted in PBS-T, 0.2% BSA and detected with the Supersignal West pico chemiluminescent substrate (Pierce).

Proteins were heat treated at 80°C for 10 minutes or left untreated and immobilized on streptavidin-coated, 96-well plates (NUNC), at concentrations sufficient for complete saturation of the available binding surface/well (as determined by ELISA). Following three washes in PBS-T, plates were incubated with pooled sera from malaria-exposed Malawian adults or malaria-naïve UK individuals at a 1:1,000 dilution in PBS-T, 2% BSA followed by an alkaline phosphatase-conjugated anti-human IgG secondary antibody (Sigma) and detected as above.

With the aim of expanding an existing recombinant P. falciparum merozoite cell surface and secreted protein library, publicly available, genome-wide transcription microarray data of P. falciparum intra-erythrocytic stages were analysed[38, 39]. To compile a list of merozoite extracellular proteins with possible roles in erythrocyte invasion, the transcription profiles of four well-established P. falciparum merozoite ligands (RH5, AMA1, EBA140, EBA175) that have all been previously implicated in erythrocyte invasion were examined[11, 38‐41]. It was observed that they all follow a similar expression pattern, passing through a minimum 20 ± 6 hours post invasion and peaking at approximately 42 ± 6 hours after invasion[38, 39]. Of the 465 candidate blood stage genes that showed similar expression time windows, 207 encoded a predicted signal peptide and/or a single transmembrane domain/GPI anchor[28‐30], suggesting they are likely cell surface or secreted proteins. Multi-pass membrane proteins were excluded from this list because these proteins are unlikely to be expressed in a soluble, secreted form. Within the 207 candidates, 31 were already represented in the existing merozoite library[27] and another 120 were excluded because of their predicted function (e g, involvement in lipid metabolism or nuclear localization) or protein domain content (e g, RNA or DNA binding motif) following gene ontology analysis using PlasmoDB[42], and protein domain mapping using Pfam[43]; also, 42 proteins were excluded due to their large size (>1,400 amino acids). In total, this bioinformatics analysis identified 14 putative secreted or membrane-tethered merozoite proteins. To this list, other members of the MSP3[44, 45], MSP7-like[46‐48] and SERA paralogous protein families[49] were added, which were not present in the existing P. falciparum merozoite recombinant protein library[27]. Finally, the available literature was scanned to identify an additional seven merozoite cell surface and secreted proteins[50, 51]. In total, 25 putative merozoite secreted or cell surface proteins were chosen for recombinant expression, two of which contained putative GPI anchors, and 23 contained no predicted membrane anchor (Table 1).

To design the expression constructs, the entire predicted extracellular domain was selected between the signal sequence and the GPI-anchor, if present. Any predicted N-linked glycosylation sites were systematically removed by substituting alanine for serine/threonine at these sites. Expression constructs were made by gene synthesis and codon optimized for expression in human cells. All expression plasmids are publicly available through Addgene, a not-for-profit, open access plasmid repository[36].

All proteins were expressed in HEK293E cells as soluble fusion proteins that contained a C-terminal rat Cd4(d3 + 4)-hexahistidine tag for purification and could be optionally monobiotinylated by cotransfecting a secreted version of the E. coli BirA enzyme[33]. To purify many proteins in parallel for comparative screening, a custom-built protein purification system was employed, that enables the simultaneous purification of up to 96 His-tagged proteins, even from large (>50 mL) tissue culture volumes[33].

Following purification, recombinant proteins were quantitated by ELISA and while expression levels varied significantly between individual proteins, most were purified to micromolar levels (Table 1). Overall, detectable expression was obtained for 22 out of 25 (88%) proteins. Both GLURP and SERA6, although detected by ELISA, were expressed at levels that were too low to include in further analysis and three proteins (PF14_0044, PFA0445w, SERA1) remained undetectable by ELISA despite repeated transfections.

To assess their integrity, the recombinant proteins were resolved by SDS-PAGE, and the presence of the C-terminal biotin tag detected by Western blotting (Figure 1). A band consistent with the full-length ectodomain was obtained for 20 out of 25 proteins (80%); in some cases, smaller bands were also evident, most likely due to proteolytic processing. For SERA7 and PF08_0006, bands of ~30 and ~25 kDa were detected, respectively, suggesting complete cleavage at the C-terminus of the protein. Collectively, these observations establish the successful expression of 18/25 (72%) recombinant P. falciparum merozoite proteins at a size consistent with a full-length recombinant protein at usable amounts.

In nature, protective antibodies largely recognize proteins in their native conformation; therefore, to examine whether the recombinant merozoite library proteins were correctly folded, their immunoreactivity against hyperimmune sera from adults living in malaria-endemic regions was tested. All 20 proteins from the library expressed at useable levels were arrayed on a streptavidin-coated, microtitre plate and their relative immunoreactivity to pooled sera from Malawian adults was compared to that from malaria-naïve individuals[52]. All but one (MSRP4) of the proteins were immunoreactive (Figure 2). Strikingly, strong immunoreactivity was observed for most of the SERA proteins, consistent with previous observations[49].

To demonstrate that serum antibodies were recognizing conformational epitopes within the protein library, all recombinant proteins were denatured by heat treatment before being captured via their biotin tag. For 16 of the 20 proteins (all but MSRP4, SERA7, PFA0210c, and PFB0475c) the immunoreactivity decreased significantly when the proteins were heat inactivated, establishing that the antigens contain heat-labile epitopes. These data show that the merozoite recombinant proteins are correctly folded and at least in part mimic the native protein conformation.