Plasmodium falciparum Transcriptome Analysis Reveals Pregnancy Malaria Associated Gene Expression

Background Pregnancy-associated malaria (PAM) causing maternal anemia and low birth weight is among the multiple manifestations of Plasmodium falciparum malaria. Infected erythrocytes (iEs) can acquire various adhesive properties that mediate the clinical severity of malaria. Recent advances on the molecular basis of virulence and immune evasion have helped identify var2csa as a PAM-specific var gene. Methodology/Principal Findings The present study presents a genome-wide microarray transcript analysis of 18 P. falciparum parasite isolates freshly collected from the placenta. The proportion of PAM over-expressed genes located in subtelomeric regions as well as that of PAM over-expressed genes predicted to be exported were higher than expected compared to the whole genome. The identification of novel parasite molecules with specificity to PAM and which are likely involved in host-pathogen interactions and placental tropism is described. One of these proteins, PFI1785w, was further characterized as the product of a two-exon PHIST gene, and was more often recognized by serum samples from P. falciparum-exposed women than from men. Conclusions/Significance These findings suggest that other parasite proteins, such as PFI1785w, may contribute beside VAR2CSA to the pathogenesis of PAM. These data may be very valuable for future vaccine development.


Introduction
Clinical severity of Plasmodium falciparum malaria is linked in part to alterations of the infected erythrocyte (iE) induced by parasite proteins exported to the iE membrane during blood stage development. Some of these proteins confer cytoadherence properties to the iE, leading to parasite sequestration from the blood stream. The best characterised parasite adhesin is P. falciparum erythrocyte membrane protein 1 (PfEMP1), encoded by the large polymorphic var gene family. This adhesin undergoes antigenic variation, a phenomenon that is thought to contribute to parasite evasion from the host immune response [1]. Certain relatively conserved sub-families of PfEMP1 molecules have been associated to differences in disease outcome [2,3,4].
Malaria during pregnancy and particularly first pregnancies is associated with dramatic adverse effects on fetal growth (reviewed in [5]). Primigravidae lack protective antibodies, which suggests that pregnancy-associated malaria (PAM) parasites express novel surface molecules to which women have not been exposed previously [6,7,8,9]. PAM parasites show placental tropism and specifically adhere to chondroitin sulphate A (CSA) [6,10]. Plasma inhibitory activity to iE binding CSA is not dependent on the geographical origin of parasites or plasma [7], indicating that proteins mediating placental tropism are conserved between parasite genotypes or contain shared epitopes and that a vaccine against this form of malaria can be developed.
PfEMP1 proteins encoded by parasite var genes and expressed on the surface of iEs were extensively shown to mediate various binding activities hence representing an important target for antibodies associated with protective immunity against malaria. Several studies have shown that a single var gene, var2csa, is highly transcribed and expressed in both in vitro CSA panned P. falciparum iEs and parasites from infected placentas [4,11,12,13,14]. The finding that antibody levels to VAR2CSA are dependent on sex and gravidity and are associated with reduced consequences of PAM [15,16] has further strengthened the candidacy of VAR2CSA as the CSA adhesin in PAM. However, other genes may be involved either in facilitating the VAR2CSA expression and binding or in other unknown PAM associated mechanisms. Identification of these putative genes would be crucial for our understanding of PAM and the development of a vaccine against this particular form of malaria.
In the present study, DNA microarray analysis of genome-wide expression profiles from 18 fresh P. falciparum isolates directly collected from human placentas at delivery allowed the identification of novel genes associated with placental tropism.

Results
Identification of differentially transcribed genes in placental parasites compared to 3D7 reference To assess the steady-state of mRNA levels of P. falciparum infecting pregnant women, parasites were collected from the placenta of delivering women in Senegal. Microscopic examination demonstrated a domination of late stage iEs (trophozoites and shizonts) in all isolates. Due to the variable distribution of late stages, samples were divided into 3 groups: (pool 1 included 3 isolates: #5% rings, 70% trophozoites, $25% schizonts; pool 2 with 7 isolates: #5% rings, $25% trophozoites, 70% schizonts; pool 3 with 8 isolates: #5% rings, 0% trophozoites, $95% schizonts). Total RNA was isolated from each individual sample and RNA from individuals corresponding to each defined group was pooled by mixing equal quantities of RNA. We chose to explore placental parasite gene expression versus a common 3D7 reference and preferentially select the genes differentially expressed in all 3 groups of placental parasites. Reference parasite population was made-up of 10% rings, 45% trophozoites and 45% schizonts. RNA pooled samples were fluorescentlylabeled prior to hybridization to an array of high-density oligonucleotides.
Using the Bonferroni adjustment method, 84 genes consistently displayed similar differential expression profiles in 2 out of 3 groups (over-expressed N = 38; under-expressed N = 46) ( Figure 1, Table 1). These genes generally had highly abundant transcripts either in the test samples or in the 3D7 reference pool as judged by spot fluorescence values. Var2csa was the only var gene unambiguously over-expressed. Over-expression in placental parasites was observed in 14 out of 18 different var2csa oligonucleotides (data not shown). Other var hybridising probes showing transcript abundance differentials represented either truncated or pseudo-genes or var exon2, [17].
The proportion of genes in subtelomeric regions over-expressed by placental parasites was higher than expected compared to the whole genome (p = 3.2e-6) (figure 2, panel A); no such bias was observed with placental parasites under-expressed genes. Similarly, the proportion of over-expressed genes predicted to be exported [18]was higher than expected compared to the whole genome (p = 4.5e-8) (figure 2, panel B), whereas no bias was observed with placental parasites under-expressed genes.

Validation of microarray data by real-time PCR
To confirm the accuracy of the microarray results, the transcript abundance of 13 selected genes was carried out. The genes were selected based on over or under-expression, abundance of transcripts and the presence of at least one TM, export prediction, and/or signal sequence motifs (see supplement table S1). Realtime PCR was performed first with the three pooled RNA/cDNA samples used in the microarray analysis, then on RNA/cDNA from each individual parasite sample included in the pool. Overall, real-time PCR confirmed the arrays (Spearman correlation test after log-transformation of the data, r = 0.79, p,.0001) ( Table 2). Plots of log2 ratio of expression (M) against average log intensity (A) for three different pools of placental isolates (pool 1 included 3 isolates: #5% rings, 70% trophozoites, $25% schizonts; pool 2 with 7 isolates: #5% rings, $25% trophozoites, 70% schizonts; pool 3 with 8 isolates: #5% rings, 0% trophozoites, $95% schizonts) versus common 3D7 reference (10% rings, 45% trophozoites, 45% schizonts). Only statistically significant data according to Bonferroni p value correction are shown. A single dominant var gene (var2csa) is detected in placental isolates of all groups examined (red dots; each dot represents one oligonucleotide). Non var genes over-expressed or under-expressed are shown as black (when the difference was present in all 3 pools), green (in 2 of the 3 pools). Blue dots correspond to oligonucleotides corresponding to genes only differentially expressed in 1 out of 3 experiments (these were not taken into account in our analysis). doi:10.1371/journal.pone.0001855.g001

Biological validation of array data
To further assess biological relevance of the microarray findings, the transcription abundance of 13 genes selected for real-time PCR was determined in parasites associated with different malaria clinical backgrounds. Transcription abundance of the 13 genes was also measured in laboratory-adapted parasite lines selected or not for PAM phenotypes. By examining the transcription profiles of parasites from different clinical groups, four categories of transcription profiles were defined: (A): highest expression in placental parasites (B): highest expression in parasites from symptomatic women and asymptomatic children (C): highest expression in parasites from symptomatic women and pregnant women but not from asymptomatic children ( Figure 3) and no expression difference among parasites from all 3 clinical groups (not shown). Overall 9 of the 13 tested genes behaved as predicted regarding differential expression between parasites from PAM and non-PAM donors. In group A, in addition to var2csa, one gene, PFI785w, encoding a hypothetical protein displayed a placental parasite-specific profile similar to that of var2csa. PFI1785w encodes a 39.4 kD protein that is predicted to be exported to the red blood cell (table S2). In group B (Figure 3), the two RT-PCR tested genes found to be PAM under-expressed by microarray analysis, PFD1120c and PF10_0350 were also found under-expressed in placental parasites when compared to non pregnant women and children parasites. In group C, the four genes explored were over-expressed in symptomatic adults compared to asymptomatic children.. Microarray detected PAM over-expression of PF14_0757 was not confirmed by real-time qPCR. However, although microarray data were confirmed by RT-PCR for PF14_001, PFL0260c, no differential expression was observed among the different groups of clinical isolates analyzed.
When transcript abundance of the 13 genes was measured in laboratory-adapted parasite lines selected or not for CSA binding phenotype ( Figure 3, left panel), only var2csa, displayed a marked specificity for CSA-binding [13,19].

Characterization of PFI1785w
The specific over-expression of PFI1785w by fresh placental isolates, and not by non-PAM parasites or by in vitro CSA-selected laboratory lines, raised a particular interest for further characterization. An interesting aspect of PFI1785w is the presence of a PHIST (Plasmodium helical interspersed subtelomeric family) domain, and thereby a predicted role in erythrocytic trafficking or remodeling [18]. PFI1785w conforms to the canonical gene structure of Pexel-containing proteins in which the signal peptide is encoded by the first exon, and the Pexel motif and the remainder of the protein are encoded on the second exon. However, PFI1785w differs from almost all Pexel-encoding genes in that it possesses a third exon according to the 3D7 annotation in the current parasite database. The PFI1785w genomic sequences amplified in this study confirmed the presence of the in-frame stop codon at aminoacid position 146 in the 3D7 gDNA, which is not present in the alleles from the Ghanaian (Sanger Institute Browser) or HB3 (Broad Institute Browser) strains. However the spliced structure as determined by PCR and sequencing of the cloned 3D7 PFI1785w gDNA and corresponding full size cDNA from 3D7 and from 2 placental isolates revealed existence of only one The number of oligos used and the number of those genes encoding hypothetical or predicted exported proteins are also shown. doi:10.1371/journal.pone.0001855.t001 intron on this gene and the splicing site was defined between nucleotide position 189 and 358 (figure 4). All cDNA sequences data have been submitted to the GenBank with the following accession numbers: EU181220 (for 3D7), EU181221(for placental isolate NTN42), EU181222 (for placental isolate NTN14). The nucleotide sequence was highly conserved between 3D7 and the two placental isolates cDNA. Only two point mutations were observed. Our data indicate that the PFI1785w mRNA was predicted incorrectly. This raises the necessity to confirm the current annotation of the parasite genome with experimental data. Part of the corresponding protein encoded by PFI1785w was expressed in baculovirus transfected insect cells and purified as a secreted HIS protein (that we called NP561). NP561 was recognized in ELISA by serum samples from individuals living in endemic areas (Ghana). Sera from women had significantly higher levels of antibodies against NP561 compared to sera from men of the same area in Ghana (p = 0.04). Recognition of another parasite recombinant protein (GLURP) was also measured and no significant difference was observed between sera from both genders ( Figure 5).

Discussion
Many studies have generated convincing evidence of the unique antigenic and binding properties of P. falciparum iEs collected from the placenta. Previous analysis of the parasite variant surface antigens has pointed to PfEMP1 variants as the major target of antigenic variation and mediator of iE adhesion properties. Our prior studies have shown that VAR2CSA PfEMP1 accounts for CSA binding by iEs originating from the placenta [4]. The present study has identified differences in gene expression compared to in vitro 3D7 parasite reference, using the whole-genome microarray analysis of P. falciparum parasites directly collected from the placenta. First, this analysis confirmed var2csa (represented by 16 oligonucleotides) as the only var gene clearly over-expressed in all placental parasite isolates. This however does not exclude possible over-expression of other var genes, which were not represented on the microarray due to sequence diversity. Second, along side with the var2csa, a significant number of genes were over-expressed by placental parasites, including genes predicted to encode exported proteins and thus possibly contributing to parasite placental tropism. For this whole-genome analysis of in vivo transcription of placental parasites, stringent selection criteria were used i.e. differential expression in at least 2 out of 3 pooled samples after stringent Bonferroni p value adjustment. A common reference containing all parasite erythrocytic stages with an emphasis on schizonts and trophozoites, forms predominantly found in placental isolates, was used. Although there are clearly defined morphologic differences between the schizont, merozoite, and ring stages, biological processes may be shared in contiguous stages. Using a common reference was preferred to three separate controls attempting to match the composition of each pool, as such matching on morphological criteria is very difficult on late stages, if not impossible. This allowed identification of 38 and 46 genes over-or under-expressed in placental parasites respectively.
Of special interest is gene PFI1785w that was highly and specifically over-expressed in placental parasites. The transcription of PFI1785w displayed a clearly defined profile, similar to that of var2csa, suggesting a specific association with placental parasites. However, no evidence of its relation to the CSA binding phenotype was seen in parasite lines biologically selected for in vitro CSA binding. Several other genes were also over-expressed in placental parasites, but not in in vitro selected CSA binding parasites. This suggests either that these genes are selectively overexpressed during pregnancy but are not directly involved in CSA binding, or that conditions required for in vivo binding to proteoglycans expressed by the placenta and in vitro binding to CSA differ. A recent microarray study on individual parasites by Francis et al. [20] also observed high transcription of PFI1785w by placental parasites, and the expression of this particular PHIST gene at the protein level was described by Fried et al. [21] in membrane extracts from placental parasites. Although CSA binding property represents the major specific phenotype of placental parasites that can be reproduced in laboratory-adapted parasite lines in vitro, some characteristics of parasites causing PAM might require in vivo conditions for expression. We hypothesise that this could be the case for PFI1785w expression.
The presence of a stop codon (...KKP-stop-KKPHP...) within the 3D7 PFI1785w explains why the gene was miss-annotated as an atypical three-exon PHIST gene and suggests that the sequence downstream the stop codon is not necessary for 3D7 survival in vitro. However in other strains such as those causing PAM the entire protein structure may be needed for placental tropism of the parasite.
The recombinant protein (NP561) corresponding to part of the protein encoded by this gene was more highly recognized by serum samples from women living in malaria-endemic areas as compared to serum from men. The relation to gender was less strong than that of VAR2CSA, raising the question of this gene's exact role in the mechanisms governing P. falciparum placental tropism. However, the importance of this relation was similar to the one observed with recombinant PFD1140w in the Francis et al. study [20], and the presence of significant levels of antibodies to VAR2CSA has been reported among some men and children [22]. Whether antibodies to PFI1785w express functional activity against PAM parasites will be addressed in future studies. Differential acquisition of these antibodies between men and pregnant women should also be confirmed in a dedicated seroepidemiological survey.
PFI1785w is a single copy gene that encodes a small protein likely exported to the red blood cell as judged by its predicted signal sequence and export motif. The gene is found conserved (.98%) at the nucleotide level in all 4 species of Plasmodium spp Blocks A to C, left panels correspond to higher transcription levels in: (A) placental parasites from pregnant women at delivery (PW); (B) non placental parasites from children (C) and non placental parasites from non pregnant symptomatic women (NPW); (C) parasites from women (PW or NPW). Shown are transcription levels for 22 parasites isolated from the placenta of women (PW), 9 parasites from asymptomatic children (C) (mean age 10 years) and 4 parasites from symptomatic non-pregnant women (NPW) living in the same area, as well as by (in the right side of each panel) the in vitro cultured P. falciparum strains 3D7 (black dots), FCR3 (red dots), and NF54 (blue dots). Each of these 3 strains were used either non-selected or after selection for CSA binding using panning on soluble CSA (3D7), Bewo cells (FCR3), or VAR2CSA IgGs (NF54). Hb3 (yellow dots) was also studied after selection on Bewo cells. These parasite lines were highly synchronized and aged 24 hours post invasion Missing points correspond to lack of detection of the corresponding gene (in particular for PFI1785w in FCR3). Horizontal bars indicate statistically significant differences between groups. On top, (+) indicates over-expressed in PAM parasites according to array data; (2) indicates under-expressed in PAM parasites according to array data. doi:10.1371/journal.pone.0001855.g003 infecting humans as well as in the Plasmodium reichenowi genome, suggesting that the product of this gene plays an important role for the survival of Plasmodium, but argues against a direct role in antigenic variation on the surface of the infected red blood cell. No evidence has been presented that the PHIST proteins in general are surface antigens, and none of them have TM domains. In contrast to PfEMP1 and RIFIN proteins, they exhibit no amino acid diversity between isolates; for example, comparison of PFI1785w alleles between 3D7 and HB3 (Broad Institute browser), the Ghanaian isolate (Sanger browser) or placental isolates from our study shows essentially 100% sequence conservation.
A recent study identified a ,22 KDa protein as being involved in the binding of iEs to CSA [23]. In the present study, high transcription of a number of genes encoding low molecular weight proteins with predictive characteristics for red blood cell trafficking was specifically observed in placental isolates. Notable examples include, along with PFI1785w (44 KDa corrected here), PFA0700c (12.83 KDa), PF14_0757 (24.95 KDa), and PFB0105c (35.3 KDa). In our study, PF14_0757 with its theoretical molecular weight of 24.95 KDa is the closest in size to this potential protein, but its transcription was not shown to significantly differ between placental and non-placental parasites. Among the 5 genes found by Francis et al. [20] to be specifically co-transcribed with var2csa by placental parasites when analyzing individual isolates, four were also identified in our study (PFI1785w, PFL0050c, PFD1140w, PFB0115w). However only PFI1785w met our own criteria of interesting genes (high expression in at least 2 out of the 3 test pools). PFD1140w, the main candidate gene characterized in the Francis et al. study [20] was also detected here, but only in the pool with domination of trophozoites. This apparent discrepancy suggests a stage-dependent expression of the PFD1140w gene. Proteins from 2 (PFI1785w and PFA700c) genes with interesting scores in our study were also detected by Fried et al [21]. For 3 other genes (PFB0115w, PF14_0016, PF14_0260) for which proteins were detected in the same study, differential expression was only recorded in 1 of the 3 test pools. With PF14_0016 and PF14_0260 rather showing PAM under-expression in our study.
Our results suggest that many sub-telomeric genes in P. falciparum genome play a role in host-parasite interactions. As sub-telomeric genes, PHIST appears dissimilar to any known structure, making it difficult to assign possible functions to members of this family. A published microarray study of field isolates has shown that most genes from PHISTb and PHISTc families are over-expressed in P. falciparum parasites directly collected from infected children compared to 3D7 [24]. As shown in table S2, in the present study, 4 genes belonging to the three PHIST families, i.e. PF14_0757, PFA110w, PFI1785w and PFB105c were found over-expressed in PAM parasites compared to 3D7. Thus, it is all the more interesting that proteomic data support consistent presence of proteins belonging to PHIST family in iE membrane preparations [21,25,26,27].
These findings suggest that other parasite proteins beside VAR2CSA may contribute to the pathogenesis of PAM. Prospective clinical studies are needed to estimate the importance of those new proteins in the mechanism controlling acquisition of distinct binding and serological phenotypes by P. falciparum and thus, identification of critical targets for intervention.

Patients, sample collection and storage
Samples were collected in November 2003 in Guediawaye, a suburb of Dakar, Senegal. The human experimentation guidelines of both French and Senegalese governments were followed. Women were explained the nature of the project, and informed verbal consent was obtained in the presence of an external and independent witness. The study design, the sampling protocol, as well as the way to collect informed consent were approved by the ethical committee of the Ministry of Health, Senegal. Notably verbal consent was considered valid and recommended in this context of poorly literate study population. Pregnant women, admitted for delivery, were enrolled when blood smears and/or immuno-chromatography test were positive for P. falciparum infection. Non-pregnant women who presented with fever and whose blood smears were positive for P. falciparum infection, were also enrolled as controls. Clinical data were registered and 10 ml of peripheral blood were collected in sodium heparin. Peripheral ring stages iEs were matured for 18-20 h at 37uC in a candle jar [28]. Placental parasites were obtained by perfusing the placenta with 0.1% sodium heparin in PBS [29]. Parasites were conserved in Trizol (Invitrogen) and stored at 280uC, or as spots on Whatman 3MM filter that were dried and stored at room temperature, until subsequent RNA and DNA extractions. Plasma was separated from the peripheral blood and stored at 220uC until use.
During the same period, venous blood samples were collected from asymptomatic infected children in a paralleled on-going cross-sectional study conducted in the neighboring area of Thiès [30]. Samples were collected following the same protocols described above.
Parasite culture and selection for specific phenotypes P. falciparum strains were cultured as described [31]. The 3D7 strain was sorbitol-synchronized 4 times with an interval of 8 hours to obtain tightly synchronous stages. Strains FCR3 and HB3 were previously panning-selected for adhesion to CD36 or Bewo cells as described [32]. The NF54 strain was selected using antibodies specific to the var2csa-encoded PfEMP1 as described [33]. The 3D7 was also panning-selected for adhesion to immobilized CSA. After selection, parasites were cultured 5-6 cycles to generate sufficient parasite density. Mature stages were enriched by Macs purification, as described [34]. In vitro panningselected lines were characterized by reactivity with IgG from multiparous women and lack of reactivity with IgG from P. falciparum-exposed men [6,9,35].

DNA and total RNA extractions
Genomic DNA was extracted from filter spots using the chelex procedure [36]. Total RNA from iE containing trophozoite/ schizont stage parasites from placental and matured peripheral blood was prepared using organic extraction from Trizol. RNA quality was assessed with Agilent 2100 Bioanalyser.

Microarrays
The microarrays used here are described elsewhere [12]. Briefly, glass slides were spotted with 8870 70-mer oligonucleotides originating from the Malaria Oligo Set (Qiagen-Operon) and custom design oligonucleotides, covering most of all P. falciparum genes. Of the 5542 identified genes, the array covers 5396 genes (97.4%) including protein coding, rRNA, tRNA, snRNA and pseudogenes. Due to the variable distribution of late stages in different isolates after placental perfusion (thin smears reads against 200 parasites), RNA samples from 18 women were pooled into 3 groups. The 3D7 reference was made-up of 10% rings, 45% trophozoites and 45% schizonts. Integrity of the RNA transcripts from each test sample was assessed with an Agilent 2100 Bioanalyser. RNA samples labelling and hybridization were performed as described previously [12].
For each of the three pools of placental parasites, two dye swaps were performed to compensate dye effect and to assess technical reproducibility, leading to four hybridized slides. Data from each of the three pools were analyzed separately using R function (The R project) and Bioconductor package [37] as described previously [12]. After normalization, a paired Student t test was used to assess differentially expressed spots and the raw p values were then corrected using the Bonferroni method with a type 1 error of 0.05. Log2 ratios of pooled placental parasites versus 3D7 pooled reference was obtained for genes which were found differentially transcribed in at least 2 of 3 test pools and presented in S2. Our data have been submitted to the publicly available ArrayExpress database with the following accession number: E-MEXP-1087. Primer design and quantitative real time rt-PCR Gene-specific real time PCR primers were made based on the 3D7 sequences for 13 selected genes (supplement Table S1). Primer amplification efficiency was tested on serial dilutions of gDNA from three laboratory isolates (3D7, FCR3, and HB3) and three different field isolates. Melting curve analysis and agarose gel electrophoresis ensured specificity. The var2csa specific primers used are described previously [4].
For real-time PCR, RNA Trizol extraction was followed by 15 min DNase 1 (Sigma) treatment at 37uC. Absence of DNA contamination was assessed by 40 cycles of real time PCR using primers targeting a previously validated endogenous control gene (fructose-biphosphate aldolase) [13]. DNA-free RNA was reverse transcribed using random hexamer primers with Superscript II enzyme (Invitrogen) at 25uC for 10 min, and 42uC for 50 min followed by 70uC for 15 min.
Quantitative real time PCR was performed on cDNA using a Rotorgene thermal cycler system (Corbett Research) [13]. Reactions were done in 20 ml volumes using Quantitect SYBR Green PCR Master Mix (Qiagen) and 0.5 mM primers [13]. Quantitative analysis of the gene-expression level was done using Rotorgene software version 4.6. Transcripts abundances were compared by DCt values calculated using fructose-biphosphate aldolase as endogenous control.
Differences between groups of samples studied by real-time rt-PCR were tested by the appropriate non-parametric test (Mann-Whitney two samples rank sum test). Correlations were tested by the Spearman rank sum test. The significant limit was P = 0.05. STATA software (version 7.0, STATA Corporation) was used.

Sequencing and recombinant proteins generation
The PlasmoDB predicted protein encoded by PFI1785w in the P. falciparum clone 3D7 genome, was selected for expression as a soluble protein. The product of this gene is predicted to be exported according to Sargeant et al. [18]. Primers were designed to amplify and clone the full size cDNA and also gDNA, and a fragment of 672 bp from the start codon in order to contain both predicted introns in the 3D7 gene. The forward primer used in the generation of the recombinant protein, NP561, was designed downstream of the apparent stop codon, to avoid a potential unspliced stop. The sequence from 391 bp to 967 bp was amplified from 3D7 gDNA using the following primers that included EcoR1 and Not1 restriction sites in 59 and 39 respectively. Fw: CAATCAAGATATAATAGATCA; Rv: CGGTTTTACCCT-TATAATG. All inserts were cloned into a modified pAcGP67A vector containing EcoR1 and Not1 restriction sites, and recombinant proteins produced in baculovirus-infected insect cells and purified as described [15]. The purified recombinant protein named NP561, corresponding to about 70% of the full size of the protein was subsequently used as antigen in this study.

Enzyme-linked Immunosorbent Assay
Plasma levels of P. falciparum-specific IgG were measured in ELISA using recombinant VAR2CSA (DBL5e) or NP561 protein for coating. GLURP recombinant protein was used as control antigen. Two panels of plasma samples from P. falciparum thick blood smear negative Ghanaian individuals (30 women of various gravidity and 30 sympatric males) exposed to hyperendemic and seasonal malaria transmission were used [15]. Sera from the two groups were previously characterized for their sex-specific and parity-dependent recognition of VAR2CSA DBL5e and CSAadherent parasites [15]. A pool of plasma samples from unexposed Danish women was used as a negative control. Antibody levels were analyzed and expressed as OD values.