R32LR is a recombinant protein produced in AR58 Escherichia coli strain with a temperature induction process and purified by three precipitation steps, followed by reversed-phase high performance liquid chromatography and size-exclusion chromatography, as already described . The final sample buffer was 0.2 M phosphate buffer, pH 6.5. The protein was stored in aliquots at -80°C until use.
His-R32LR was constructed with six histidine residues at the N-terminus. Briefly, a plasmid encoding an E. coli codon-optimized R32LR DNA sequence preceded by a histidine-tag was obtained from GENEART AG (Regensburg, Germany). His-R32LR was subcloned in a pET29a plasmid by insertion between NdeI and SacI sites, and transformed in the BLR(DE3) E. coli strain. Expression of the recombinant protein was obtained by addition of isopropyl β-D-1-thiogalactopyranoside (1 mM) before the temperature was shifted to 39.5°C for 4 h. For purification purpose, bacterial cells were grown in fermenter using fed-batch method and the same induction strategy. The bacterial pellet was suspended in 50 mM phosphate buffer, pH 7.5 containing 300 mM NaCl, 5% glycerol (v/v), 0.1% sodium deoxycholate (w/v), supplemented with complete protease inhibitor cocktail (Roche, 1 tablet/50 ml buffer). The cells were lysed by three French press extractions at 15,000 psi. After refrigerated centrifugation for 30 min at 17,400 × g, the supernatant was harvested and filtered through a 0.22-μm membrane.
His-tagged R32LR was further purified through a nickel column (Ni NTA superflow, Qiagen) with elution by imidazole gradient. The pooled positive fractions were dialysed against 50 mM Tris, pH 7.5, containing 5% glycerol (v/v) and subjected to ion-exchange chromatography (Mono-Q column, Pharmacia). The positive fractions were pooled and dialysed against 10 mM phosphate buffer, pH 6.6, containing 150 mM NaCl. R32LR (and His-R32LR) is neither detected by Coomassie blue nor by UV absorbance at 280 nm , but high-performance liquid chromatography coupled with detection at 205 nm was used to control batch homogeneity. Protein purity was evaluated by Coomassie blue-stained electrophoresis gel to visualize any contaminating protein, and on western blot membrane probed with rabbit anti-serum targeting E. coli lysate proteins. Protein identity was evaluated by western blotting using a murine monoclonal antibody recognizing NANP (in-house; R3G12 antibody). Endotoxin concentration in the final product was determined by Limulus amoebocyte lysate -kinetic QCL assay (Cambrex). R32LR and His-R32LR concentrations were determined by dosage of total nitrogen content. Briefly, after pyrolysis, nitrogen compounds were converted into nitrogen monoxide which then reacted with ozone to produce NO2*. NO2* emitted a photon which was detected after signal amplification in an Antek 9000 device (Alytech, France). After purification, both antigens were kept in aliquots at -80°C.
The assay was developed by using serum samples from healthy subjects in different clinical trials, and taken at various time points after vaccination with RTS,S. RTS,S is composed of a polypeptide chain of 19 NANP repeats and a C-terminal region of P. falciparum (3D7) CSP encompassing amino acids 207–395, fused to hepatitis B surface antigen (HBsAg), and an unfused (S) polypeptide of 226 amino acids of HBsAg, spontaneously forming a virus-like particle [4, 32, 33].
Negative control serum samples were obtained from healthy adults living in malaria non-endemic areas and who were thus considered as anti-CSP repeats antibody seronegative. Positive control samples were pools of post-vaccination serum samples.
As no international standard preparation is available for the determination of anti-CSP repeats antibody concentration in serum, a human sample from the recipient of an exploratory malaria vaccine was taken to constitute the reference standard for the assay, to which a concentration of 109 ELISA units per ml (EU/ml) was attributed (IR2 reference standard, kindly provided by the Walter Reed Army Institute of Research). All study participants and serum donors have given written informed consent for the use of their serum for anti-CSP repeats test development.
Reference human monoclonal antibody
Hybridomas producing anti-R32LR antibodies were generated using a method described previously . In brief, human peripheral blood mononuclear cells (PBMC) collected from an individual vaccinated with the RTS-S vaccine candidate were injected in the spleens (1-2 × 107 cells per animal) of three SCID mice (C.B-17 Prkdcscid/Prkdcscid). Six days later the mice were bled and their plasma anti-R32LR antibodies were measured using the in-house ELISA. On day 7, the mouse displaying the highest anti-R32LR concentration was sacrificed, the spleen was removed and a cell suspension was prepared. Human PBMC-SCID spleen cells were mixed with K6H5/B5 heteromyeloma cells at a 5:1 ratio, and polyethylene gly-col 1500 (50% v/v; Boehringer Mannheim, Mannheim, Germany) was added for 2 min before being washed away. Fused cells (5 × 104) were cultured in microtiter plate wells in 200 μl of medium supplemented with human recombinant insulin (10 μg/ml, Boehringer Mannheim), ouabain (1 μM, Sigma, St. Louis, MO), hypoxanthine-aminopterin-thymidine (Life Technologies, Belgium) and 10% v/v BM Condimed HI (Boehringer Mannheim). Cultures were replenished with fresh medium every other day and individual wells were checked for cell growth first and anti-CSP antibody production subsequently. Seven anti-R32LR antibody-producing cultures were selected, subcloned and further expanded.
The seven hybridomas, producing anti-R32LR IgG1, were cultured in CELLine 1000 (Integra Bioscience, Chur, Switzerland), a membrane based disposable cell culture system. For optimal production level the device was inoculated with 50 × 106 cells in Dulbecco’s modified Eagle medium at high glucose concentration supplemented with glutamine, sodium pyruvate, essential and non-essential amino acid, a cocktail of antibiotics and 5% fetal bovine serum. Eighty percent of the production medium and the entire nutrition medium were changed twice a week.
Systems were stopped after 37 days of culture including preculture phase, and 2 hybridomas (MAL 1C and MAL 2A) showed a productivity of 19 mg/month/150 million cells. Monoclonal antibodies were purified by affinity chromatography on Protein A-sepharose Fast Flow (GE Healthcare), aliquoted and stored in phosphate-buffered saline at -20°C before being used as standard in anti-CSP repeats ELISA.
IgG antibodies directed against the CSP repeat region were measured as follows. R32LR was coated (100 μl/well of a solution of 1.25 μg/ml prepared in 0.05 M carbonate/bicarbonate buffer, pH 9.4-9.8) onto a 96-well polystyrene plate (F96 MaxiSorp, Nunc) for 14-16 h at 5 ± 3°C. After coating, plates were washed 3 times with phosphate-buffered saline (PBS), pH 7.4, supplemented with 0.1% Tween-20 (v/v; Sigma, ref P1379). Nonspecific binding sites were satura-ted with 200 μl/well of blocking buffer, consisting of PBS (pH 7.4) containing 0.1% Tween-20 (v/v) and 0.5% skimmed milk (Becton Dickinson, ref 232100), for 2 h at room temperature (RT) on an orbital shaker. No washing step occurred after blocking but the plate was turned upside down to remove blocking buffer, and gently tapped down on clean blotting paper. Then, eight serial twofold dilutions of standard, controls and samples were added (100 μl/well) and incubated for 2 h at 37°C.
The plates were washed thrice before peroxidase-conjugated anti-human IgG rabbit antibody (Dako, ref PO214) diluted in blocking buffer was added for 30 min incubation at RT. After another washing step, the chromogen substrate mix was added and incubated for 30 min at RT. The chromogen substrate mix was prepared extemporaneously and consisted of four volumes of substrate buffer [Na2HPO4 (Merck 1.06586), citric acid (Merck 1.002441000) supplemented with 0.006% H2O2 (v/v) and adjusted to pH 4.1-4.5] mixed with 1 volume of 3,3’,5,5’-tetramethylbenzidine solution (Sigma, ref T-0440). The colorimetric reaction was stopped by the addition of 50 μl of 1 N sulphuric acid, before reading the assay plate at 450 nm in a microtiter plate reader.
An un-weighted 4-parameter logistic (4-PL) fitting algorithm  was applied to the standard curve, allowing the determination of antibody concentration in the samples, expressed in ELISA Units per milliliter (EU/ml).
Anti-hepatitis B antigen ELISA
Anti-HBsAg antibodies were quantified using Abbott’s AxSYM Micro-particle Enzyme ImmunoAssay (MEIA) platform with the dedicated reagent-pack AUSAB, using a standard curve as described by the manufacturer.
For the validation of the assay, the following parameters were evaluated according to the The International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) guidelines: precision, linearity, specificity, robustness, interference and stability. Accuracy could not be determined, as no reference analyte of known purity is available. Furthermore, the proposed ELISA could not be compared with another well-characterized gold-standard assay, as no such assay is available.
Precision can be defined as the closeness of agreement between independent test results obtained under stipulated conditions, called intermediate precision and repeatability conditions. Intermediate precision expresses the precision obtained under conditions mimicking the routine assay conditions. Repeatability expresses the precision under minimal variable conditions. For the determination of repeatability and intermediate precision in the ELISA, 12 serum samples from RTS,S-vaccinated subjects spanning a large range of anti-CSP repeats antibody concentrations were used, measured in duplicate by three different operators on four different days, equivalent to 288 measurements. For the determination of repeatability, every sample was tested in duplicate on the same plate.
Precision was considered acceptable if it was in line with the descriptive statistics derived from our in-house database compiling repeatability and intermediate precision for 138 ELISAs over the last five years, with application of the same experimental design, that is: the 90th percentile of coefficient of variation (CV) are 12.93% for repeatability and 31.92% for intermediate precision, respectively.
The linearity of an analytical procedure is defined as its ability (within the analytical range) to provide a measurement directly proportional to the analyte concentration. Linearity is confirmed if the degree of underestimation and overestimation does not exceed 20%. Sixty sera from RTS,S-vaccinated subjects, of pre-defined anti-CSP repeats antibody concentration, were used. Sera with a high anti-CSP repeats antibody concentration were pre-diluted in order to obtain a concentration situated in the upper part of the standard curve. All samples were serially diluted twofold in the microtiter plate. The last dilution step for each sample was expected to give an anti-CSP repeats concentration situated in the lower part of the standard curve. Then, anti-CSP repeats antibody concentration in all serially diluted samples was measured according to the assay procedure.
The specificity is the ability to assess unequivocally the analyte in the presence of components that may be expected to be present. Two types of experiments were conducted to assess the specificity of the proposed ELISA. First, the specificity was investigated by testing 14 serum samples from individuals naïve for malaria but vaccinated with HBsAg  to demonstrate that the ELISA does not detect any anti-HBsAg IgG. Second, a competition experiment with AMA-1 (Plasmodium falciparum apical membrane antigen-1, kindly provided by the Walter Reed Army Institute of Research), HBsAg, and R32LR was conducted. The sera containing specific anti-CSP repeats antibodies were pre-mixed with either of these antigens (final concentration: 10 μg/ml in blocking buffer) and incubated for 24 h at room temperature before being processed further according to the assay procedure.
The robustness of an analytical procedure is the measure of its capacity to remain unaffected by small but deliberate variations in method parameters, which provides indication of its reliability during normal usage. First, three timings for the coating with R32LR were evaluated: 14, 16 and 18 h (standard timing being 16 h). Second, the impact of freeze-thaw cycles from -80°C to room temperature of the coating antigen R32LR was assessed by evaluating the anti-CSP repeats geometric mean concentration and intermediate precision after 1, 2 and 3 freeze-thaw cycles.
Interferences are defined as artefactual increase or decrease in apparent concentration of an analyte due to the presence of a substance or a treatment.
Analyte stability at various temperatures and impact of freezing/thawing cycles
Ten samples from RTS,S-vaccinated subjects were subjected to the following conditions: overnight storage at -20°C, or 3 cycles (24 h duration each) of freeze-thawing (-20°C to RT), or storage at room temperature for 24 h, before anti-CSP repeats ELISA.
Heat-inactivation of the serum samples
The effect of complement-inactivation on the measurement of CSP repeats-specific IgG concentration was assessed on 100 sera (30 negatives, 70 positives). Each of the 100 sera tested was divided equally into two vials, one for control and one for heat-inactivation. Paired samples were tested by the same operator but on different days. Complement-inactivation consisted of incubation at 56°C during 30 minutes.
To evaluate the stability of the anti-CSP repeats ELISA, two samples (a low-concentration positive control and a high-concentration positive control, prepared by pooling, homogenizing and aliquoting samples previously demonstrated to be positive) were tested in each plate on every run performed by the laboratory since the start of the routine testing in 2007. The limits of these controls were computed on log10-transformed concentrations using a minimum of 40 independent values.
To complete the assessment of the stability of the anti-CSP repeats ELISA, an additional internal blind proficiency panel consisting of 50 samples that have been previously tested has been re-analysed every six months, starting in 2008.
Determination of assay characteristics
The calculation of the limit of detection (LOD) and the limit of quantification (LOQ) was based on a methodology from The Centers for Diseases Control and Prevention . Each standard curve was fitted using a four parameters logistic (4PL) and the 95% confidence interval (95% CI) was considered. For a standard curve, the LOD is the concentration corresponding to the interpolated intersection of the upper 95% CI of the lower asymptote with the 4-PL fit of standards data. The LOQ is the concentration corresponding to the interpolated intersection of the upper 95% CI of the lower asymptote with the lower 95% CI of the 4-PL fit of standards data. LOD and LOQ were computed for each curve and 95th percentile of LODs and 95th percentile of LOQs were assimilated respectively as the final LOD and the final LOQ.
The cut-off of the ELISA was based on the upper limit of the 99.9% one-sided confidence interval of the anti-CSP repeats response in a naïve malaria-population. The experiment was performed on 108 serum samples obtained from healthy adults living in a non-endemic malaria region, hence assumed not to contain anti-CSP repeats antibody. The log10 -transformed optical density (OD) corresponding to the first dilution of each sample were averaged and the standard deviation was determined. In order to get a value in ELISA Units/ml, the upper limit of the one-sided 99.9% confidence interval of the mean was interpolated from the 4PL function built on the average ODs of the standard curves incubated on the 20 plates. Finally this value was multiplied by 50, the minimum dilution factor of the sample.
For the assessment of the analytical range, the lower limit was set at the LOQ and the upper limit was defined as the upper limit of linearity.
Statistical analyses were performed on log10-transformed data.
For the validation of assay precision, analysis of variance was performed on the transformed data by using the MIXED procedure of the SAS system, with all factors considered as random. Computations were done on the transformed values but the variability in terms of CV was expressed relative to the non- transformed concentrations.
For linearity, the ratio between successive dilutions was evaluated for each sample. This ratio was estimated by linear regression and computed from the slope of this regression (log (concentration) versus log (dilution)). These ratios were summarized by calculating the 5th percentile, median and 95th percentile.
To evaluate the effect of three freeze-thaw cycles or the effect of 24 h storage at room temperature, analysis of variance was performed by using the MIXED procedure of the SAS system, with the factor condition considered as fixed. The heat-inactivated condition was compared with the non-inactivated condition by means of a paired t-test.
The robustness of the assay was evaluated by analysis of variance using the MIXED procedure of the SAS system, with the factor incubation considered fixed, for the effects of coating time. To study the conditions of storage of the coating antigen, analysis of variance was used (MIXED procedure of the SAS system) with the factor condition considered fixed, followed by one-sided Dunnett test. The effect of sample inactivation was analysed using a concordance analysis, Deming regression [38–40] and correlation analysis.
Stability was evaluated using on the one hand two positive controls and on the other hand a blind proficiency panel of 50 samples. The quality control (QC)-specifications of each of the two controls were defined by computing standard deviation through all data and uncertainty of the mean was taken into account with an alpha risk set at 1% for each of the two positive controls. The stability results using the panel of 50 samples were analysed with a Deming’s regression and a concordance analysis.