Five French soldier companies (n = 751, mean age ± SD: 25.3 ± 4.8 years), who travelled during a five-month period in tropical Africa (Gabon or Ivory Coast, from 2002 to 2007), were included in this study. Individuals used mandatory anti-malaria prophylaxis including anti-vectorial equipment, such as permethrin-impregnated bed nets, repellents and long-sleeve battle dress at night and chemoprophylaxis (100 mg of doxycycline per day). Blood samples were collected one week after the return to France and sera were stored at -80°C. Sera from individuals living in France (n = 31, mean age ± SD: 31.2 ± 6.5 years, Caucasians), who have never been exposed to malaria vectors and parasite were used as negative controls. Sera from highly exposed individuals (HEI) to malaria from Congo (n = 9), gratefully provided by Pr. Hovette (Centre médical de secours Total, Pointe Noire, Republic of the Congo) and from Senegal (n = 5), were used as positive controls. All participants gave their written informed consent to take part in the study and Marseille-2 University ethical committee has approved the protocol (Ethics Statement).

The RP8 P. falciparum cytoadherent strain was selected from the Palo Alto (FUP)1 strain by panning on Human Umbilical Vein Endothelial Cells (HUVEC, ATCC number: PCS-100-010), as described by Fried and Duffy 23. This cell line express CD36 and ICAM-1 which are well known host receptors involved in cytoadherence of iRBC. Parasites were maintained in continuous culture in candle jar as previously described 2425. Briefly, parasites were cultured at 5% haematocrit of type O⁺ human RBC suspended in RPMI 1640 (Invitrogen, Paisley, UK) supplemented with 25 mM HEPES (Invitrogen), 6 mM D-glucose (Sigma-Aldrich, St Louis, USA), 23 mM NaHCO₃ (Invitrogen), 32 μg/ml neomycine (Sigma-Aldrich), 0.2 mM L-glutamine (Invitrogen), 0.25 μg/ml orotic acid (Sigma-Aldrich), 0.5 μg/ml hypoxanthine (Sigma-Aldrich) and 10% of heat-inactivated type O⁺ human serum, at 37°C in a gas mixture of 3% CO₂, 17% O₂. Parasitaemia and erythrocytic cycle stages proportion were monitored daily microscopically by examination of blood smear stained with Giemsa-stained (RAL^® 555, France).

RBC and iRBC from in vitro cultures with about 10% parasitaemia were used to perform L-IFA as previously described 2627. Briefly, 50 μl of cell suspension were washed two times in RPMI 1640 at room temperature and incubated in DAPI (40 μg/ml) (Invitrogen, USA) for 30 min at 37°C. After 3 washing in PBS (Invitrogen), immuno-staining was started by incubating cells with 50 μl of diluted sera for 30 min on ice. Each sera was individually analysed using dilutions of 1:40, 1:80, 1:160, 1:320, 1:640 and 1:1,280 in PBS pH 7.4. Positive and negative control sera were used at 1:40 dilution. After two washes with RPMI 1640, alexa-fluor 488 goat anti-human immunoglobulin G (H+L) (Molecular Probes, USA) was added to the cell suspension and incubated for 30 min on ice. A droplet of cell suspension was laid on a microscopy slide, and antibody binding and DNA staining were assessed by fluorescence microscopy on a Nikon eclipse E800 fluorescence microscope (100 × magnification). Sample titres were determined by the maximum dilution where the fluorescence could be detected.

Plasmodium falciparum adherent mature stages (trophozoite and schizonte stages) were enriched by Plasmion flotation (Fresenius France Pharma, Louvier, France) as described elsewhere 2528. Infected erythrocytes were washed two times in PBS medium and lysed in cold hypotonic medium (H₂O-saponin 0.1%, Sigma) for 2 min. Free parasites were discarded by centrifugation (9 300 × g for 4 min). Erythrocyte membranes were recovered from the supernatant, and this cell fraction was submitted to repeat steps of ultra centrifugations (100 000 × g for 1 h at 4°C) followed by washing until iRBC ghost were colourless, and stored at -80°C. The same protocol was used to collect membrane proteins from uninfected erythrocytes. Membrane protein extracts were then suspended in 4% (w/v) CHAPS (Sigma). Membrane aggregates were then disrupted by ultrasonication (Vibracell 72412, Bioblock Scientific, Illkirch, France) for 5 min on ice at maximum amplitude and precipitated in 100% acetone (Sigma) to discard lipids. Protein concentration was estimated using Lowry-based DC assay (Biorad, Hercules, CA, USA) according to the manufacturer's instruction. Membrane protein extracts were suspended in a buffer containing 8 M urea (Sigma), 2 M thiourea (Sigma), 4% (w/v) CHAPS (Sigma) and 30 mM Tris (Sigma), adjusted to pH 8.5 in order to obtain a protein concentration adjusted to 2.5 μg/μL.

The RBC and iRBC membrane samples were reduced in a Tris buffer containing dithiothreitol (1% w/v, Sigma), and 15 μg of each sample were loaded per well onto a 10% polyacrylamide gel before to be separated by SDS-PAGE in a Mini PROTEAN II (BioRad, Hercules, CA, USA). After SDS-PAGE, gels were transferred onto a nitrocellulose membrane (0.45-μm, GE Healthcare) by semidry blotting 29. Each membrane was cut into 18-22 strips (3-4 mm wide), before incubation with a saturation buffer (5% non-fat dried milk in PBS containing tween-20 (0.1% v/v, Sigma)). Immunoblots were carried out with human sera, which were diluted at 1/100 in saturation buffer. After overnight incubation, blots were incubated with mouse anti-human Fcg/IgG horse radish peroxidase (HRP) conjugated antibody (1/5 000, Beckman Coulter, San Jose, CA, USA), and revealed using an ECL Plus western blotting (WB) detection system (GE Healthcare). Densitometric analysis of autoradiographs was performed using Diversity database™2.2 software (Biorad) to align and compare the IgG immune patterns. Two sera, one from a NEI and another one from a HEI which were used as controls, were tested on each blot in order to assess the quality of each immunoblot and to improve the accuracy of the alignment of antibody reactivities intra- and inter-gels 3031. Molecular weights were estimated by comparison with standard molecular weight (Biorad).

Membrane proteins from iRBC were minimally labelled with CyDye according to the manufacturer's recommended protocols (GE Healthcare). Briefly, membrane protein extracts (50 μg) were labelled with 400 pmol of Cy5, freshly dissolved in anhydrous dimethyl formamide (Sigma), and incubated on ice for 30 min in the dark. The reaction was quenched with 1 μL of free lysine (10 nM, Sigma) by incubation 10 min on ice. An equal volume of 2× sample buffer (8 M urea, 2 M thiourea, 4% (w/v) CHAPS, 10 mM DTT, 1% (v/v) IPG Buffer 3-10 (GE Healthcare) was added to the Cy5 labelled sample before submission to two-dimensional electrophoresis (2-DE).

Reagents that were not purchased from GE Healthcare, are indicated. Destreak buffer containing 1% (v/v) IPG buffer 3-10 was used for IPG strips (18-cm) overnight rehydratation. The samples were laid at the acidic end of the IPG strip using a cup-loading technique. IEF was carried out on an Ettan IPGphor II electrophoresis unit at 20°C, for a total of 45 kVh. Prior to separation in the second dimension, IPG strips were reduced and alkylated in a equilibration buffer containing 50 mM Tris-HCl, pH 8.6 buffer, 6 M urea, 2% SDS and 30% glycerol supplemented with 1% (w/v) DTT or 2.5% (w/v) iodoacetamide instead of DTT for 10 min. IPG strips were then placed on the top of 10% uniform polyacrylamide gels. Strips were overlaid with 0.5% agarose in 1× running buffer containing bromophenol blue and the proteins were further separated by SDS-PAGE (15 W per gel) at 20°C, in Ettan DALT Six electrophoresis system. Two-dimensional gels were either stained with Imperial™Protein Stain (Thermo Scientific, Rockford, IL, USA) or transferred onto nitrocellulose membranes and incubated with human sera as described above (see Materials and Methods section: immunoblots and analysis procedures). For nitrocellulose membranes with Cy5-labelled proteins, antibody response was revealed using goat-anti-human IgG FITC-conjugate (diluted at 1:2000) (Invitrogen). Immunoblots were directly digitalized using a Typhoon™Trio Image scanner. Images were analysed with Decyder v6.5 software, allowing an accurate spot matching between 2-D protein patterns and 2-D antigenic patterns from gels and immunoblots respectively. Matched spots were selected for excision and further identification by mass spectrometry (MS).

Based on Decyder analysis, spots of interest were excized using Shimadzu Biotech Xcise System (Champs sur Marne, France). Protein spots were digested overnight at 37°C with sequencing-grade trypsin (12.5 μg/mL; Promega Madison, WI, USA) in 50 mM NH₄HCO₃ (Sigma). The resulting peptides were extracted with 25 mM NH₄HCO₃ for 15 min, dehydrated with acetonitrile (ACN) (Sigma), incubated with 5% acid formic (Sigma) for 15 min under agitation, then dehydrated with ACN, and finally completely dried using a SpeedVac. Samples were then stored at -20°C before analysis by mass spectrometry (MS).

The samples were analysed on a LCQ DecaXPplus (ThermoFinnigan, San Jose, CA) ion trap. Nano-liquid separation of peptides was carried out using an Ettan MDLC chromatographic system (GE Healthcare) in high throughput configuration. Ten microliters of the digest were first trapped on a zorbax 300SB-C18 5 × 0.3 mm column and eluted at a flow rate of approximately 200 nL/min on a zorbax 300SB-C18, 3.5 μm, 150 × 0.075 mm by a linear gradient of eluant B (0.1% Formic acid, 84% ACN) in eluant A (1% Formic acid). Chromatographic system was piloted by Unicorn 5.01 software (GE Healthcare). MS measurements were performed using a LCQTM Deca XP Plus ion trap mass spectrometer (ThermoFinnigan) equipped with a LCQTM nanospray ionization source. A spray voltage of 1.8 kV was applied to the liquid junction via an in-union high voltage contact coupled to a silicaTip emitter (New Objective). Operation of the mass spectrometer was fully automated during the entire procedure using Excalibur 1.3 data system (ThermoFinnigan). Continuous cycles of one full scan (m/z 500 to 1700) followed by three data-dependent MS/MS measurements at 35% normalized collision energy were performed. MS/MS measurements were allowed for the three most intense precursor ions with a maximum rejection time limit of 1 min. All MS/MS spectra were sequence database searched using Mascot Daemon v2.2.2 software (Matrix Science, London, UK).

The data were searched using Mascot software, against Homo sapiens and P. falciparum National Center for Biotechnology Information non-redundant (NCBInr, NIH, Bethesda, MD) protein database (June 10^th, 2009). Search parameters are set in order to allowed for one missed tryptic cleavage site, the carbamidomethylation of cysteine, and the possible oxidation of methionine; precursor and product ion mass error tolerance was <0.2 Da. All identified proteins have a Mascot score greater than 41 (Mixed: H. sapiens + P. falciparum, 237,583 sequences), corresponding to a statistically significant (p < 0.05) confident identification. Moreover, among the positive matches, only protein identifications based on at least two different non-overlapping peptide sequences with a mass tolerance <0.05 Da were accepted. For single peptide-based identification, in addition to Mascot score significance, were considered only peptide sequence with at least six consecutive amino acids detected on MS spectra (Additional file 1). These validation criteria were added in order to limit the number of false positive matches without missing real proteins of interest.

For IgG antibody patterns analysis, the data were expressed in binary mode (0, absence of antigenic band; 1, presence of an antigenic band). The Mann-Whitney and the Fisher exact test were used as appropriate. All statistical analyses were done with the SPSS 12.0 software.

IgG antibody reactivity from 751 BEI sera sampled after a five months journey in endemic area was assessed by liquid-indirect immunofluoresence assay (L-IFA). Considering reactions with titres over 1:80 as positive, 38 BEI sera (5.0%) shown an IgG reactivity directed against the surface of iRBC (Figure 1). Among the 38 selected BEI, 37 were males with a mean age ± SD of 25 ± 3.5 years. Eight BEI had a P. falciparum clinical episode during the journey. These positive BEI sera were selected for 1 D immunoblotting analysis. Sera from individuals never exposed to malaria (NEI, n = 31) were also tested by L-IFA, and none of them presented IgG reactivity against iRBC at 1:40 dilution (Figure 1). Sera from highly exposed individuals (HEI, n = 14) presented a high IgG reactivity against the surface of iRBC until dilution to 1:320.

The IgG antibody response of selected sera from NEI (n = 31), BEI (n = 38) and HEI (n = 14), was tested successively against RBC and iRBC membrane protein extracts using immunoblots (Figure 2A & 2B). The densitometric analysis of the different patterns of proteins recognized by IgG obtained with regard to the molecular mass, were performed for quantitative comparisons. Against RBC membrane protein extracts, 3.4 ± 2.6 (means ± SD), 3.5 ± 2.1 and 4.9 ± 1.5 antigenic bands were detected per strip for NEI, BEI and HEI sera, respectively (Figure 2C). The number of antigenic bands recognized by each sera group, against RBC membrane protein extracts, was low and no significant difference was observed between these groups (Mann-Whitney test, p > 0.05). Against iRBC membrane protein extracts, 3.4 ± 2.0 (means ± SD), 8.1 ± 4.4, 19.9 ± 4.0 antigenic bands were detected per strip for NEI, BEI and HEI sera, respectively (Figure 2C). In this case, the number of antigenic bands recognized by BEI and HEI groups were significantly more important compared to NEI group (Mann-Whitney test, p = 1.6 × 10^-7 and p = 4.5 × 10^-7 respectively). Moreover, the number of antigenic bands detected between RBC and iRBC membrane protein extracts is significantly increased for BEI (Mann-Whitney test, p = 8.2 × 10^-8) and HEI (Mann-Whitney test, p = 4.9 × 10^-8). For NEI group, between RBC and iRBC membrane protein extracts, the number of antigenic bands detected was not significantly changed (Mann-Whitney test, p > 0.05). Collectively, these results suggest that individuals exposed to P. falciparum parasites (BEI and HEI) present an IgG immune response specifically directed against iRBC membrane protein extracts. It is interesting to note that no antigenic bands could be detected above 200 kDa.

In order to define band identity of antigens recognized by BEI, a qualitative analysis of the IgG immune response was performed between NEI and BEI groups. A mapping and alignment of the IgG patterns allowed us to detect a total of 23 and 36 distinct antigenic bands against RBC and iRBC membrane protein extracts, respectively. Against RBC membrane protein extracts, no antigenic band was significantly detected more frequently in one group than to the other one (Fisher exact test, p > 0.05). Conversely, against iRBC membrane protein extracts, eight antigenic bands were significantly detected more frequently in BEI group compared to NEI group (Fisher exact test, p < 0.05; Figure 2D). Each of these antigenic bands was detected by at least 20% of BEI (Figure 2D). Additionally, each BEI sera recognized at least two of the eight selected antigenic bands. The molecular weights of these antigenic bands ranged from 18 kDa to 130 kDa, and a schematic representation of these antigenic bands was depicted on an immunoreactivity diagram (Figure 2E). Among the BEI group, five sera were selected with regard to their IgG response on 1-D immunoblots for subsequent antigens identification. Each selected BEI sera recognized at least three of the eight discriminatory antigenic bands and the pool covered the whole spectrum of these antigenic bands against iRBC membrane protein extracts. The pool was composed of these five selected BEI sera.

To further characterize these discriminatory antigens, a 2-D immunoproteomic approach was adopted. A major difficulty of this approach is to perform a perfect match between antigenic spots detected on 2-D immunoblot and their counterparts on the preparative gel, on which matched spots will be excized and further analysed by mass spectrometry (MS). As the results of these protein identifications are the starting point of further studies, it is necessary to limit miss matching. Thus, 2-D immunoblot using a fluorescence-based method allows to increase considerably the confidence in spot matching between preparative gel and immunoblots (Figure 3).

The pool of BEI sera allowed to detect 42 antigenic spots onto the 2 D immunoblot (Figure 4). To facilitate the alignment of the antigenic bands with the corresponding spots between 1 D and 2 D immunoblot, iRBC membrane protein extracts were loaded onto 2 D gel beside the IPG strip, prior electrophoresis, as described previously 30. For seven out of eight antigenic bands selected by 1 D analysis, their corresponding spots were detected onto 2 D immunoblot (Figure 4). No antigenic spot corresponding to the molecular weight (MW) of band "V" was detected onto 2 D immunoblot. This could be attributable to the extremes pI or low solubility of this antigenic protein in the buffer for 2DE.

As some bands split into several protein species, 30 spots were assigned to the seven antigenic bands (Figure 4, Additional file 2). A typical example is presented by the antigenic band number "I" which can be resolved into nine different spots on the 2-D immunoblot. The 42 antigenic spots detected onto the immunoblot, were excized and submitted to MS for identification (Figure 4). The resulting fragment ion spectra were searched against H. sapiens and P. falciparum-protein databases (NCBInr). Nineteen protein spots (45%) were successfully identified and correspond to 13 distinct proteins according to their NCBI accession number (Additional file 2). Among them, four were identified as P. falciparum proteins and nine as H. sapiens proteins. Selected bands "I", "II" and "VII" were identified as H. sapiens antigens and selected bands "III", "IV" and "VI" as P. falciparum antigens. However, MS did not identify protein spots corresponding to the selected band "VIII", despite their apparent high abundances. This could be attributable to the low MW of these spots giving spectra with low complexity. Thus, spots corresponding to six out of eight antigenic protein bands were identified.