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

Background Plasmodium falciparum is the aetiological agent for malaria, a deadly infectious disease for which no vaccine has yet been licensed. The proteins displayed on the merozoite cell surface have long been considered attractive vaccine targets because of their direct exposure to host antibodies; however, progress in understanding the functional role of these targets has been hindered by technical challenges associated with expressing these proteins in a functionally active recombinant form. To address this, a method that enables the systematic expression of functional extracellular Plasmodium proteins was previously developed, and used to create a library of 42 merozoite proteins. Methods To compile a more comprehensive library of recombinant proteins representing the repertoire of P. falciparum merozoite extracellular proteins for systematic vaccine and functional studies, genome-wide expression profiling was used to identify additional candidates. Candidate proteins were recombinantly produced and their integrity and expression levels were tested by Western blotting and ELISA. Results Twenty-five additional genes that were upregulated during late schizogony, and predicted to encode secreted and cell surface proteins, were identified and expressed as soluble recombinant proteins. A band consistent with the entire ectodomain was observed by immunoblotting for the majority of the proteins and their expression levels were quantified. By using sera from malaria-exposed immune adults, the immunoreactivity of 20 recombinant proteins was assessed, and most of the merozoite ligands were found to carry heat-labile epitopes. To facilitate systematic comparative studies across the entire library, multiple Plasmodium proteins were simultaneously purified using a custom-made platform. Conclusions A library of recombinant P. falciparum secreted and cell surface proteins was expanded by 20 additional proteins, which were shown to express at usable levels and contain conformational epitopes. This resource of extracellular P. falciparum merozoite proteins, which now contains 62 full-length ectodomains, will be a valuable tool in elucidating the function of these proteins during the blood stages of infection, and facilitate the comparative assessment of blood stage vaccine candidates.


Background
Plasmodium falciparum is the aetiological agent of the most deadly form of malaria, an infectious tropical disease that accounts for up to one million deaths annually [1,2]. The vast majority of malaria fatalities (85-90%) occur in sub-Saharan Africa, primarily in pregnant women and children under the age of five [2,3]. While anti-malarial drugs exist, the emergence of drug-resistant parasite strains remains a global health concern and no vaccine has been licensed to date.
The asexual blood stages of malaria are initiated when a form of the parasite, called a merozoite, invades, replicates and synchronously ruptures host erythrocytes [4] releasing up to 32 progeny merozoites that can invade new erythrocytes. This cyclical phase causes the recurrent fevers and chills that are characteristic of malaria infection [5]. Merozoites are ovoid cells containing apically located secretory organelles that release proteins which are required for the invasion of new erythrocytes [6,7]. While erythrocyte invasion is a rapid process, the brief extracellular exposure of merozoites outside of their intra-erythrocytic niche places them in direct contact with host antibodies, which contribute to naturally acquired immunity to malaria [8,9]; therefore, merozoite cell surface and secreted proteins have long been considered attractive targets for rational vaccine development.
The publication of the P. falciparum genome project in 2002 [10] identified the full complement of parasite proteins but progress in understanding the function of these proteins, including those displayed on the merozoite cell surface, has been hindered by the technical difficulties in expressing Plasmodium proteins in a functionally active form [11]. Although the reasons why Plasmodium proteins are difficult to express in heterologous expression systems are not clear, several protein characteristics, such as high molecular mass (>60 kDa), presence of export motifs, and atypical signal peptide sequences negatively impact recombinant expression [12]. In addition, the remarkably high (~80%) A + T content of parasite genes can result in long stretches of repetitive amino acids [13], and codons that are not frequently used by organisms popular for heterologous protein expression. Extracellular vaccine candidates, in particular, present an additional challenge because they often require structurally critical disulfide bonds for correct folding and contain transmembrane domains that make them difficult to solubilize in detergents that retain their native conformation [14][15][16].
Despite these challenges, recombinant expression of Plasmodium proteins has been attempted in a number of expression systems [12,17] ranging from bacteria [18], yeast [14,19], Dictyostelium discoideum [20], plants and algae [21,22] to mammalian cells [23,24] and cell-free systems [13]. Among them, Escherichia coli is the most popular [17], but the systematic expression of functional P. falciparum proteins remains difficult, with success rates as low as just 6% [25], and often requires subsequent laborious and complex refolding procedures with uncertain outcomes [26]. Consequently, the functional characterization of extracellular parasite proteins has typically been restricted to smaller subfragments that can be expressed rather than the full-length protein or entire ectodomain, which is more likely to be representative of the native protein.
The development of a standardized method to express large panels of P. falciparum cell surface and secreted proteins in their native conformation would enable comprehensive protein libraries to be systematically screened in parallel so that direct comparisons between antigens can be made in functional assays such as vaccine screening and immuno-epidemiology studies. To achieve this, Crosnier and colleagues recently developed a method of expressing the entire ectodomains of functional recombinant Plasmodium proteins and used it to compile a large library of 42 proteins [27]. Working towards a comprehensive library of cell surface and secreted P. falciparum merozoite proteins, this manuscript describes the identification and characterization of an additional 20 proteins.

Recombinant protein design and expression
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].

Parallel protein purification
His-tagged recombinant merozoite proteins were purified from spent tissue culture supernatants by using a custombuilt, 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).

Enzyme-linked immunosorbent assay (ELISA)
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.

Western blotting
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).

Immunogenicity study
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.

Identification of candidate Plasmodium falciparum merozoite cell surface and secreted proteins
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][39][40][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][29][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][47][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].

Expression and purification of an expanded Plasmodium falciparum merozoite protein library
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. Protein sequences from 3D7 reference P. falciparum strain were truncated to remove endogenous signal peptides and sequences corresponding to GPI anchors. The region of each protein that was targeted for recombinant expression is shown. Expression levels are given as a guide only because of the significant batch-to-batch variability of transient transfection, and grouped into "high" (between 30 μΜ and 100 μΜ after purification), "medium" (1 μΜ and 30 μΜ) and "low" (0.01 μΜ and 1 μΜ). No detectable expression (N/D) was obtained for PFA0445w, PF14_0044 and SERA1. ¥Proteins selected from [51]. *Proteins included to complete the set of proteins belonging to the MSP3, MSP7-like and SERA families. #Proteins chosen from [50]. All other proteins were short-listed from bioinformatic analysis of transcription microarray data [38,39]. GPI, GPI-anchored.

Members of the recombinant merozoite protein library proteins are immunoreactive and carry heat-labile epitopes
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  The merozoite recombinant proteins are immunoreactive against hyperimmune sera. The immunoreactivity of the recombinant P. falciparum merozoite proteins was tested using pooled sera from malaria-exposed Malawian adults (red bar) or malaria-naïve adults (green bar). The reduced immunoreactivity of immune sera to heat-denatured antigens (blue bar) demonstrates the presence of heat-labile (conformational) epitopes. All proteins except one (MSRP4) were identified as being immunoreactive as assessed by immunoreactivity >3 SD above negative control (green bar). AMA-1 and Cd4 were used as the positive and negative control, respectively. Data points are shown as mean ± s.d.; n = 3. data show that the merozoite recombinant proteins are correctly folded and at least in part mimic the native protein conformation.

Discussion
The technical difficulties in expressing Plasmodium proteins in a recombinant functional form has presented difficulties both for basic malaria research and vaccine development [12]. Here, an approach based on a mammalian expression system has been utilized to significantly expand a recombinant library consisting of recombinant P. falciparum merozoite extracellular proteins from 42 to 62 members [27]. A high-throughput, custom-made purification platform was successfully used to purify recombinant proteins from large volumes of tissue culture supernatants to permit systematic comparative studies of purified antigens [33].
While the overall success rate of expression was high, consistent with previous experience of the HEK293 system [27,[53][54][55][56][57], a few merozoite proteins still failed to express, or were expressed at low levels. Although the genes were codon-optimized for human cells, P. falciparum proteins are unusual because they are enriched in asparagine, glutamic acid and lysine, and often contain homopolymeric stretches of amino acids. Indeed, it has been recently demonstrated that the stability of several Plasmodium proteins depends upon their association with heat shock proteins which act as molecular chaperones [58]. Therefore, recombinant protein expression in the HEK293 system could be further enhanced by the presence of PfHsp110c, which has been proposed to be a protein-stabilizing chaperone [58].
Using sera from malaria-immune adults, it has been shown that the expressed recombinant proteins contained heat-labile epitopes suggesting that they adopt their native conformation and are likely to be biochemically active. A number of proteins (e g, PF10_0166, PFA0135w) showed little serological response in comparison to the negative control. While these proteins may not contain epitopes present in the native protein, the possibility that the native proteins are poorly immunogenic and do not induce a strong antibody response in vivo, cannot be excluded. For example, previous studies reported that only 23% of immune sera examined contained specific serum IgG antibodies against PFA0135w, suggesting that this protein does not normally elicit strong humoral responses [59]. Interestingly, RIPR, which binds to PfRH5 [60], is also among the group of generally low responders. PfRH5 is a high priority vaccine target as it plays an essential and universal role in erythrocyte invasion [53,57,[61][62][63]], yet it is poorly immunogenic in vivo [61], possibly due to its late release onto the merozoite surface during erythrocyte invasion. RIPR may be similarly masked from the host immune system, but further work with this antigen is clearly required.

Conclusion
A library of recombinant P. falciparum proteins has been expanded and characterized with the eventual aim of compiling a set that is representative of the merozoite surface. These plasmids, which are freely available to the global research community through Addgene [36], will be a valuable resource for basic research and aid the efforts to develop an effective malaria vaccine.