The commonly-used yeast species Saccharomyces cerevisiae and Pichia pastoris have been exploited extensively for the heterologous expression of plasmodial proteins, both for preparative purposes and functional characterization. S. cerevisiae is a popular choice as host owing to the ease with which it can be manipulated genetically and to the extraordinary amount of information accumulated about its molecular biology and physiology 70. The use of this model organism varies from vaccine production 71 to functional characterization, drug discovery and resistance studies and elucidating parasite protein interaction networks ("interactomes") through high-throughput 2-hybrid analyses (e.g. 5572).

Expression in S. cerevisiae is typically controlled by inducible promoters, e.g. ADH2 (a glucose-repressed alcohol dehydrogenase promoter), while N-terminal fusion to a pheromone results in efficient secretion of recombinant protein from the yeast cells 7374. Coupled with carefully controlled growth in fermenters, impressive yields of 10 – 75 mg/L have been reported 7576. Secretion into the culture medium represents a major advantage of the yeast system over E. coli, as it circumvents formation of inclusion body precipitates and simplifies subsequent purification. Furthermore, trafficking through the yeast secretory compartments facilitates appropriate disulphide bond formation and protein folding, which is especially important for malaria vaccine production 767778. However, the latter advantage may have a sting in its tail: disulphide bonding in individual proteins may be heterogeneous, resulting in a population of structural conformers, not all of which contain the conformationally correct epitopes 76. In addition, the yeast secretory pathway may result in spurious O- and N-linked glycosylation of the heterologous proteins 79. The latter is especially relevant for plasmodial proteins, since these are rarely (if ever) glycosylated in their native state. The problem may be circumvented to some extent by altering the coding sequence to remove potential N-glycosylation motifs or inclusion of N-glycosylation inhibitors 80. Limitations in yeast systems that can reduce the final yield include the codon usage of the expressed gene, the gene copy number, efficient transcription by using strong promoters leading to the depletion of precursors and energy, translocation, processing and folding in the endoplasmic reticulum and Golgi, as well as protein turnover by proteolysis 81. Cells which produce high levels of proteins can accumulate unfolded protein in the ER, which aggregates, overwhelms and eventually shuts down the secretory pathway 82. The yeast proteins that assist in folding and disulphide bond formation differ from their counterparts in higher eukaryotes, which may affect folding of foreign proteins 83. Saccharomyces cerevisiae moreover recognizes distinct A+T containing codons as termination signals which could lead to the truncation of encoded proteins especially of the P. falciparum genome. Due to this codon incompatibility, synthetic genes incorporating a yeast codon bias are often used 84.

Recombinant vaccine production in S. cerevisiae has focused largely on the preparation of transmission-blocking vaccines 76. The principal components of these experimental vaccines that block transmission of malaria parasites to Anopheles are the epidermal growth factor (EGF)-like domains of Pfs25 (of P. falciparum) or Pvs25 (of P. vivax), a glycosylphosphatidyl inositol (GPI)-anchored surface protein of the mosquito-stage zygotes and ookinetes of the parasite 858687. Additional vaccine expression studies in S. cerevisiae have centred on the C-terminal domain of the major GPI-anchored surface protein of merozoites, merozoite surface protein 1 (MSP1₁₉) 8889. Despite these promising efforts, S. cerevisiae has largely been superseded by Pichia pastoris as the host cell of choice for vaccine production (see discussion below). Nonetheless, S. cerevisiae has remained relevant to malaria research through its highly characterized biology and amenability to genetic manipulation, which allows it to be exploited for functional characterization through complementation experiments. In these studies, the ability of the heterologously expressed plasmodial proteins to rectify or modify aberrant phenotypes of conditional yeast deletion mutants is typically assessed. The studies are aided by the fact that relevant yeast mutants may be obtained from extensive repositories e.g. EUROSCARF 90, the Saccharomyces Genome Deletion Project 91 and the ATCC 92. Often, the plasmodial coding sequence is modified to increase G+C content and improve yeast codon compatibility, single-copy yeast episomal plasmids are used to keep heterologous protein expression to physiologically relevant levels and expression is driven by constitutive (e.g. ADH1) or inducible (e.g. GAL1) promoters. Several classes of plasmodial proteins have been characterized in this fashion including integral membrane transporters 939495; phospholipid synthesis and modifying enzymes 96979899; cell cycle, transcriptional and translational regulators 100101102103; and possible drug targets e.g. Vitamin B6 synthesis enzymes (PfPdx1 and PfPdx2, 104) and folate metabolism proteins, dihydrofolate reductase-thymidylate synthase, (DHFR-TS) and dihydrofolate synthase-folylpolyglutamate synthase (DHFS-FPGS, 105). A highly informative series of experiments focused on DHFR-TS demonstrated how yeast complementation may be used as a versatile alternative to preparing soluble purified protein for the characterization and exploitation of parasite drug targets. The complementation system was employed to determine the effects of DHFR mutations in parasite field isolates on sensitivity/resistance to antifolate drugs 106107, the contribution of various DHFR mutations to antifolate resistance and predict future resistance trends by using a PCR-based random mutagenesis approach 108, and to screen novel antifolate compounds for activity against sensitive and resistant alleles of DHFR 109110.

Lastly, S. cerevisiae is widely used to determine protein binding partners by yeast two-hybrid analysis which has resulted in a characterization of the P. falciparum interactome and identification of a network of 2,846 interactions between 1,312 plasmodial proteins which have minimal similarity to interaction networks of other model organisms 5572.

Some of the non-conventional yeasts, such as Kluyveromyces lactis and the methylotrophs Pichia pastoris and Hansenula polymorpha (now known as Pichia angusta, 111) have also been developed as expression systems. Pichia pastoris is now the most frequently used yeast species for heterologous protein expression in general. It shares the underlying features and methodology of S. cerevisiae, with one notable exception: its biology is considerably less well characterized. Hence, it is used almost exclusively for preparative expression as opposed to the large number of functional studies that have been carried out in S. cerevisiae. However, the same principles apply: plasmodial sequences are altered to incorporate a yeast codon bias; potential N-glycosylation sites in the plasmodial sequences are removed by mutation; expression is driven by an inducible promoter (the methanol-inducible alcohol oxidase promoter, AOX1); efficient secretion into the culture medium is afforded by N-terminal fusion to the S. cerevisiae prepro-α-factor leader peptide and C-terminal tags are incorporated to facilitate purification from the medium. The redox environment and co-factors in the yeast secretory pathway should enhance correct folding, solubility and intra-molecular disulphide bonding of vaccine proteins that rely critically on conformational epitopes to generate a protective immune response; the latter may be compromized by spurious and unpredictable O-glycosylation of the heterologous protein, as well as the formation of disulphide bond conformers with reduced protective immunogenicity 112113. It appears that the use of Pichia pastoris may improve heterologous protein yield, compared with S. cerevisiae, and levels of 30 mg/L to more than 1 g/L of purified protein have consistently been reported, especially when combined with batch-fed fermentation 113114115116117118119120121. In addition, the ratio of desirable/undesirable conformers may be improved in Pichia pastoris 113. In one study, this was further enhanced by co-expressing the heterologous protein with P. falciparum or Pichia pastoris protein disulphide isomerase (PDI), which improved yield as well as reducing O-glycosylation 121. Exoprotease truncation of heterologous proteins may also be encountered in Pichia pastoris, but was circumvented in one case by alteration of the N-terminal sequence of a P. falciparum circumsporozoite protein (CSP)-derived vaccine product prone to digestion 122.

A number of plasmodial vaccine proteins have been successfully expressed in Pichia pastoris, notably merozoite proteins required for erythrocyte invasion (AMA-1, EBA-175, MSP-1, MSP-3), the sporozoite-stage antigen CSP, and transmission-blocking mosquito-stage antigens (Pfs25) 41112113114115116117118119120123124125126127. A notable exception to the vaccine preparation studies in which Pichia pastoris has been used almost exclusively in malaria research is the functional characterization of P. falciparum transporters thought to be involved in quinoline drug resistance, PfCRT (P. falciparum chloroquine resistance transporter) and a P-glycoprotein homologue, Pgh-1 128129130131. In addition to optimizing the codon usage for yeast expression, the plasmodial sequences were altered to remove poly(A) stretches, potential yeast-specific premature termination signals and predicted stem-loop structures. A 5' Kozak sequence was added, as well as C-terminal fusions to bacterial biotin acceptor domains to enable subsequent detection and/or purification. Successful membrane-localized expression of the transporters was followed by functional characterization of these proteins.

The use of baculovirus-infected insect cells to express recombinant proteins is well established 132. A lytic virus, Autographa californica nuclear polyhedrosis virus, is most commonly used. The gene of interest is cloned into a transfer vector, which is co-transfected with viral DNA into Spodoptera frugiperda (usually Sf9 or Sf21) insect cells. In a homologous recombination event, the foreign gene, under the control of a strong viral promoter such as polh or p10, is integrated into the viral genome and recombinant protein is produced in the infected cells. A C-terminal or N-terminal fusion tag is incorporated into the protein to facilitate purification. The large size of the viral genome makes it amenable to the insertion of large segments of foreign DNA. Numerous baculovirus systems, including a variety of transfer vectors, are now commercially available and technological improvements have simplified the production and isolation of recombinant baculovirus.

This approach has been successfully applied to several P. falciparum proteins. In one large study, 17 proteins that were expressed as insoluble inclusion bodies in E. coli were transferred to a baculovirus/Sf21 system, which resulted in high levels of soluble protein in one case (11.7 mg/500 ml culture) and lower levels of six other proteins, ranging from 0.6 to 7.2 mg/500 ml culture 5. The low yield can easily be overcome by scaling up the procedure.

The baculovirus system uses eukaryotic insect cells to express recombinant proteins and thus the folding and assembly of newly synthesized polypeptides may be similar to the natural process in Plasmodium. As a result, soluble recombinant baculovirus P. falciparum proteins are likely to fold correctly to produce immunologically active proteins that are recognized by conformation-specific monoclonal antibodies. This feature is vital when assessing the immunogenic response of the host to a recombinant Plasmodium protein that is considered to be a vaccine target and several studies on P. falciparum antigens illustrate this aspect. Recombinant domains of merozoite surface protein MSP-1 133134, CSP 135136 and a erythrocyte membrane protein PfEMP1 variant 137 retained their native conformational epitopes and elicited antibody responses. In contrast, recombinant PfMSP-1 expressed in yeast had a different conformation to the native protein and was immunologically inactive 133. Erythrocyte binding antigen (PfEBA-175) has also been expressed in a baculovirus system in soluble form that was functionally active 138139. Additional examples are apical membrane antigen (PfAMA-1), which was successfully purified from baculovirus-infected insect cells 140 and recombinant serine repeat antigen (PfSERA), which was subsequently processed into several fragments that reflect the in vivo situation in the parasite 141.

The genes coding for two P. falciparum aminopeptidases were chemically synthesized using codons optimized for the yeast Pichia pastoris, but expression in yeast was unsuccessful. In contrast, when these constructs were transfected into Sf9 insect cells, soluble recombinant proteins that were enzymatically active were produced in sufficient quantities to perform kinetic and biochemical analyses 142143. A subtilisin-like protease (PfSUB-1), involved in merozoite invasion, was expressed and secreted in a baculovirus system and used in functional studies to assess the interaction with its cognate propeptide 144. A micronemal protein present in Plasmodium berghei ookinetes (von Willebrand factor A domain-related protein, PbWARP) was produced in a baculovirus system and retained its function and oligomeric structure 145. This protein plays a role in the motility and attachment of the ookinete in the mosquito midgut and homologues of this protein exist in P. falciparum and P. vivax 145.

An important aspect of the baculovirus/insect cell system is that it recognizes eukaryotic targeting signals, which allows expression and processing of different classes of proteins, including those that are secreted, localized to the plasma membrane or nucleus, and cytoplasmic proteins 135146. The host insect cell can also perform most post-translational modifications, such as phosphorylation, acylation and glycosylation. The high expression levels of recombinant proteins in the late stages of the viral life cycle can however lead to incomplete post-translational processing, resulting in, for example, hypophosphorylation of some proteins. Glycosylation can also be problematic 132, especially since N- and O-glycosylation only seem to occur at very low levels in P falciparum 147. Native PfEBA-175 is essentially unglycosylated, despite the presence of putative sites, whereas 20% of the recombinant baculovirus protein was N-glycosylated, although this did not affect its immunogenicity 139. The major carbohydrate modification in Plasmodium is the addition of a glycosylphosphatidylinositol (GPI) anchor to the C-terminal amino acid of a protein 147. GPI-anchored proteins play an important role in numerous cellular functions and are essential for parasite survival. Baculovirus may not be ideal for expressing GPI-anchors since recombinant CSP was devoid of the GPI moiety, despite the presence of a potential GPI cleavage/attachment site in the DNA.

A modification of the baculovirus system is to use a virus specific for silkworm larvae, as opposed to cultured insect cells. The yield of biologically and immunologically active PfMSP-1₄₂ recombinant protein from these larvae was > 100-fold higher than from Sf cells 148. However, expertise in handling silkworm larvae is currently centred mainly in China.

The baculovirus system is a valuable tool for the expression of Plasmodium proteins and represents a viable alternative to explore, especially since it is difficult to predict which heterologous host will be optimal for a specific protein and also because a universally applicable method to express recombinant malaria proteins is not available (Figure 1). In addition, the system lends itself to automation and scaling up of protein production. However, it requires a considerable investment in time and financial resources and it is technically more demanding than manipulating E. coli.

The single-celled amoeba Dictyostelium discoideum has become an important and accessible model organism for studying numerous aspects of eukaryotic cell biology, such as cell division, locomotion, signal transduction and vesicular trafficking, 149. From the point of view of plasmodial expression, D. discoideum offers several attractive features: 1) The relative ease with which it can be cultured and manipulated genetically may be second only to yeast amongst eukaryotic systems, hence its popularity as a model organism 149; 2) It contains the canonical organelles and cell biological features of eukaryotic cells, and could be used to infer the functional attributes of heterologously expressed plasmodial proteins by their interaction with endogenous proteins. It may be particularly useful for integral membrane protein expression. For example, expression and targeting of PfCRT to the digestive compartments of D. discoideum provided evidence for the role of mutations in this membrane protein in reducing chloroquine (CQ) accumulation and hence mediating CQ resistance in parasites 150; 3) The relative ease with which endogenous D. discoideum genes may be disrupted by homologous recombination or silencing offers the potential to create mutants for use in the functional characterization of heterologous proteins by complementation. For example, adenylate cyclase null mutants were used to demonstrate the properties of PfACα, an integral membrane protein possibly involved in gametocytogenesis 151; 4) Plasma membrane anchoring via endogenous signal sequences and GPI-anchoring motifs could be used to study the adhesion properties of malaria proteins, e.g. CSP 152. Additionally, such recombinant D. discoideum strains may be explored as potential live, non-toxic, non-pathogenic "sporozoite-mimicking" vaccines 152; 5) Dictyostelium discoideum originally became an attractive vehicle for heterologous plasmodial protein expression due to the unusual A+T bias of its gene coding sequences: a feature it shares with P. falciparum 153. The organism was first used for the preparative expression of malaria vaccine candidates including P. falciparum CSP and, subsequently, the C-terminal portions of CSP from P. yoelii and P. falciparum and a fragment of MSP1 of P. vivax 152154155. Although codon optimization was not performed, it was noted that expression was considerably improved by the N-terminal addition of D. discoideum leader sequences or fusion partners. Unfortunately, the paucity of reports on the use of D. discoideum for the preparative expression of plasmodial proteins prohibits an assessment of the advantages conferred by the organism's A+T bias and its utility in this regard over the more conventional yeast expression systems. Nonetheless, it is possible that D. discoideum is underexplored as an alternative to mammalian cells and Xenopus oocytes for the functional expression of plasmodial membrane proteins and transporters (see discussions below).

Toxoplasma gondii is an important parasite pathogen in immuno-compromized individuals and during pregnancy. Like malaria parasites, it belongs to the phylum Apicomplexa, with the advantage that it is considerably more susceptible to experimental manipulation by standard molecular genetic approaches 156. Phylogenetic relatedness notwithstanding, the differences in host cell niche, pathogenesis and life-cycles (in addition to a G+C vs. A+T codon bias) makes it debatable to what extent the cell biology and metabolic pathways of T. gondii are related to those of the various life-cycle stages of malaria parasites. Nonetheless, Toxoplasma and Plasmodium share the same mechanism of locomotion as well as the specialized secretory organelles and mechanisms required for host cell invasion 157. Consequently, T. gondii may be a useful host for the heterologous expression and characterization of malaria proteins involved in locomotion, invasion and biogenesis of the specialized secretory organelles (rhoptries, micronemes and dense granules). Per illustration, expression of a dominant negative mutant form of PfVps4 led to the identification of a multi-vesicular body endosomal compartment involved in rhoptry biogenesis in Toxoplasma and suggested the existence of similar organelles and their functions in malaria parasites 158. A further potential use of T. gondii expressing plasmodial sequences is as an attenuated live vaccine. Experiments in Rhesus monkeys and mice have demonstrated the promising utility of T. gondii strains expressing P. knowlesi and P. yoelii CSP in this regard 159160. From a preparative point of view, the P. falciparum cGMP-dependent protein kinase (PfPKG) was expressed and purified from T. gondii to characterize its enzymatic properties and susceptibility to inhibitors 161. Nonetheless, it is unlikely that T. gondii will challenge E. coli or yeast systems as a first choice for this type of application.

To a significant extent, mammalian cells share with S. cerevisiae the features of being well characterized, having versatile reagents and protocols for transgene expression and analysis as well as potentially aiding correct folding of heterologous proteins in secretory compartments and imparting post-translational modifications (e.g. lipid modifications, GPI-anchoring, etc.). However, the substantially lower yields of recombinant proteins minimize their use for preparative expression. An advantage of mammalian cells is the ability to target heterologous membrane proteins to subcellular compartments and the cell surface. As a result, the large majority of studies reporting the expression of plasmodial proteins in mammalian cells have focused on the identification and characterization of parasite surface ligands that mediate host cell binding. These experiments have significantly contributed to our understanding of erythrocyte invasion by helping to identify merozoite proteins and their domains that mediate erythrocyte binding, notably EBA-175, AMA-1, Duffy-binding protein, JESEBL, MAEBL, MSA-1 and PfAARP (e.g 162163164165166167168169170). In addition, the studies have characterized regions of PfEMP-1 involved in binding to host chondroitin sulphate A (CSA), ICAM-1, CD36 and uninfected erythrocytes, thus mediating parasite sequestration to the placenta and vascular endothelium as well as erythrocyte rosetting in infected individuals 171172173174. A concomitant advantage of surface expression has been to map out conserved epitopes recognized by inhibitory monoclonal antibodies and human sera (e.g. 175176177). To facilitate the surface binding experiments, parasite protein domains are modified by N- and C-terminal fusion to mammalian or viral secretory signal sequences and transmembrane domains/GPI-anchoring motifs, respectively, and tagged with an epitope tag and/or GFP to monitor expression. The COS-7 green monkey kidney cell-line has generally been used for these (and other plasmodial expression) experiments, but the well-known CHO-K1 and HeLa cell lines have also been employed. Interestingly, with few exceptions (e.g. 174177), the plasmodial sequences were rarely altered to rectify A+T-codon bias and remove potential N-glycosylation sites.

The second most numerous application of mammalian cell expression in malaria research is to support the development of DNA plasmid-based vaccines. Transfection of mammalian cells with the vaccine plasmids is used to verify expression of the relevant plasmodial sequences prior to proceeding with animal vaccination. In addition, mammalian cell expression may also be used to assess the generation of the desired specific antibodies subsequent to vaccination (e.g. 178179180181). Again, codon optimization is rarely reported. This raises the possibility that mammalian cells may be more accommodating for the expression of the A+T-biased plasmodial sequences.

Despite the advantages offered by the extensive characterization of mammalian cell biology, the cells appear to be rarely used for functional characterization of plasmodial proteins, beyond the ligand binding studies described above. One likely reason is the comprehensive difficulty of preparing functionally compromized mutants, which hampers the complementation studies that have been extensively done in S. cerevisiae (and potentially D. discoideum). A prospective (unreported) alternative may be to use post-transcriptional silencing (e.g. RNAi-mediated) to generate deficient cells for complementation assays. In addition, provided the functional readout is sufficiently sensitive and robust, overexpression of plasmodial proteins above the endogenous background may yield useful results. A further advantage for functional characterization is the high transient transfection efficiencies that can be achieved with current transfection reagents and cell types (e.g. COS cells), which may increase the turnaround time from transfection to analysis. Examples have included the cytoplasmic expression of CDP-diacylglycerol synthase (PfCDS), mannosyl transferase (PfPIG-M), PfVPS4 and histidine rich protein II HRPII, which has shed light on the potential roles of these parasite proteins in phospholipid synthesis, GPI-anchor synthesis, endocytic trafficking and pH-sensing, respectively 158182183184.

Both Plasmodium membrane transporters implicated in CQ resistance, Pgh1 and PfCRT, have been functionally expressed in mammalian cells. Although the intracellular location of Pgh1 expressed in CHO cells could not be determined accurately, it increased CQ uptake and sensitivity in these cells, whilst Pgh1 incorporating CQ resistance-associated mutations did not 185. After synthetically introducing a yeast codon bias into the PfCRT cDNA sequence, the protein was expressed in HEK 293 cells and faithfully targeted to lysosomal membranes 186. Surprisingly, in contrast to the results obtained in D. discoideum (150; see earlier discussion), PfCRT did not alter CQ accumulation in the host cells but increased acidification of lysosomes, an effect which was enhanced when the CQ resistance-conferring mutation K76T was included in the protein. The latter contradiction highlights the possibility that conclusions drawn from characterization in heterologous cell types may not always translate to accurate functional prediction in the parasite and does not substitute for characterization in the parasite itself.

Once the technical and logistical challenges of Xenopus oocyte micro-manipulation and cRNA injection have been overcome, the cells may be extremely useful for the functional expression and characterization of membrane proteins, notably transport proteins. It has become the method of choice for characterizing Plasmodium transporters, particularly since the oocytes appear to be more tolerant than other eukaryotic cells to the peculiar codon preferences of Plasmodium genes 187188. The expression system has been used most extensively to characterize PfHT1, the principal hexose transporter of P. falciparum 188, in order to define its substrate specificity, probe structure-function relationships by mutational analysis with a view to therapeutic inhibitor design and confirm it as a potential drug target 189190191. The substrate specificity of PfENT1, one of the four equilibrative nucleoside transporters encoded by the P. falciparum genome, was also determined using Xenopus oocyte expression and compared favourably with results obtained from transport experiments using isolated parasites, thus validating the oocyte system as a more convenient surrogate for studying this transporter 192193194195. A similar comparison of properties between isolated parasites and heterologous expression in Xenopus oocytes was obtained for Na⁺- and membrane potential-dependent transport of inorganic monovalent phosphate by PfPiT 196. Additional transporters characterized by expression in Xenopus oocytes coupled with oocyte swelling assays include the aquaglyceroporins of P. falciparum and P. berghei (PfAQP and PbAQP) 197198.

A significant contribution of the Xenopus oocyte expression system to malaria drug discovery was the identification and characterization of the foremost target of the artemisinins, PfATP6, a SERCA-type Ca²⁺ ATPase 199. The study followed on the characterization of PfATP4, a P. falciparum P-type membrane Ca²⁺-ATPase, vis-à-vis its susceptibility to inhibitors and Ca²⁺ concentration-dependent activation 200. Of further relevance to drug discovery was the use of Xenopus oocyte expression to investigate the role of PfCRT in conferring CQ resistance. The findings led the authors to propose that PfCRT affects CQ susceptibility in parasites by a modulation of food vacuole transporters, rather than by directly affecting vacuole pH or transporting CQ 201.

It should be noted that, although the Xenopus oocyte expression system has been used almost exclusively to characterize malaria membrane transporters, it is a functional cell and can be a useful system for studying other cellular functions as well, for example, cell signalling. After expression in oocytes, the parasite leucine-rich repeat antigen (PfLRR1) was found to bind to the endogenous Xenopus protein phosphatase PP1 and disrupt the oocyte G2/M cell cycle arrest, observed as germinal vesicle breakdown. By inference, PfLRR1 probably regulates protein phosphatase activity and cell cycle progression in malaria parasites 202. In addition, measurement of cAMP levels in oocytes was used to characterize the properties of a X. laevis codon-optimized version of the parasite adenylyl cyclase (PfACα) and confirm results obtained by complementation assays in Dictyostelium 151.

The use of plants as expression hosts for foreign proteins has received increasing attention. Several heterologous proteins have been expressed in plants due to their ability to successfully express, fold and post-translationally modify foreign proteins. The real potential lies in the possibility of transgenic plants as bioreactors for the production of vaccines since the feasibility and low cost have been validated in several human and animal disease vaccines. It is therefore attractive to speculate that plants may be a host for 1) the transient production of functional plasmodial proteins or 2) the transgenic production of e.g. malaria vaccine targets. The latter has been under investigation with reports by Ghosh et al. (2002) in which they successfully obtained constitutive expression of a 19 kDa proteolytic product of P. falciparum MSP1 in tobacco plants resulting in the same immunogenic properties as an E. coli expressed form of the protein 203. More recently, the concept of edible vaccines derived from antigen production in plants was supported by evidence from Hepatitis-B trials in humans and mice and has now been extended to the P. yoelii MSP4/5 expressed in transgenic tobacco 204. This protein was codon optimized for production in plants and resulted in the high level expression of the antigen that could elicit protective antibodies in mice fed orally. These reports therefore indicate the feasibility of exploiting plants for the production of malaria proteins.

Cell-free transcription-translation systems have been exploited with varying degrees of success for numerous proteins including membrane proteins and cytotoxic proteins (reviewed in 205206). Advantages include production of proteins that undergo proteolysis or that usually accumulate in inclusion bodies, as well as allowing selective labelling. Of recent interest is the wheat germ cell-free platform 206, which allows the high-level expression of eukaryotic proteins that can not be expressed in E. coli cells/cell-free systems. Optimized wheat germ systems have been created by removal of ribosomal protein synthesis inhibitors (tritin) from the wheat embryo to allow the robust production of proteins. This technology has been adopted by structural genomics consortia (e.g. Centre for Eukaryotic Structural Genomics, Madison, WI). The difficulties experienced over many years in obtaining significant soluble expression of PfDHFR-TS are well-documented 207. In a recent study it was shown that a wheat germ cell-free expression system yielded bioactive protein as an obligate dimer without the associated toxicity observed in cell-based systems 207. Moreover, several malaria vaccine targets (Pfs25, PfCSP and PfAMA-1) were successfully expressed in both native and codon-optimized forms which did not seem to influence the efficiency of expression, at least of Pfs25 208. Yields between 50 and 200 μg/ml of soluble protein were obtained which was subsequently shown to be immunogenically active in raising specific antibodies in mice. Ambitiously, this study was expanded to include 124 additional P. falciparum blood stage proteins that were expressed in their native forms. Of these, 75% of the proteins were produced with an average yield of 1.9 μg/150 μl reaction mixture and 65% solubility. Inverse correlations were observed between soluble protein production and protein size, frequency of low-complexity regions and pI of the proteins.