Optimization of incubation conditions of Plasmodium falciparum antibody multiplex assays to measure IgG, IgG_1–4, IgM and IgE using standard and customized reference pools for sero-epidemiological and vaccine studies

A customized multiplex panel with 33 BS and 6 pre-erythrocytic (PE) P. falciparum antigens was established (Table 1). The glycan α-Gal (Gala1–3GalB1–4GlcNAc-R), detected in the surface of sporozoites, was also included, as anti-α-Gal IgM antibodies have been associated with malaria protection [30]. In addition to P. falciparum antigens, the hepatitis B surface antigen (HBsAg, a component of the RTS,S vaccine) was added, as the assays were intended to be used with samples from this vaccine trial. Also, bovine serum albumin (BSA) and glutathione S-transferase (GST) were added to the panel to control for background signal coming from unspecific binding to the BSA used to block the coupled beads, and to the GST present in some of the fusion proteins.

Coupling of carboxylated polystyrene microspheres was carried out as described elsewhere [26]. Briefly, MagPlex^® microspheres (Luminex Corp., Austin, Texas) with different spectral signatures selected for each antigen, were washed with distilled water and activated with Sulfo-NHS (N-hydroxysulfosuccinimide) and EDC (1-ethyl-3-[3-dimethylaminopropyl]carbodiimide hydrochloride) (Pierce, Thermo Fisher Scientific Inc., Rockford, IL), both at 50 mg/mL, in activation buffer (100 mM Monobasic Sodium Phosphate, pH 6.2). Microspheres were washed with 50 mM MES (4-morpholineethane sulfonic acid, Sigma, Tres Cantos, Spain) pH 5.0 or dPBS (Dulbecco’s Phosphate Buffered Saline, Lonza) pH 7.0 to a 10,000 beads/µL concentration, and coated with antigens at a concentration previously established in MES or PBS and incubated in a rotatory shaker overnight (ON) at 4 °C and protected from light. Microspheres were blocked with PBS-BN [PBS with 1% BSA and 0.05% sodium azide (Sigma, Tres Cantos, Spain)] and re-suspended in PBS-BN to be quantified on a Guava PCA desktop cytometer (Guava, Hayward, CA) to determine the percentage recovery after the coupling procedure. Antigen-coupled beads were validated in singleplex and multiplex by measuring IgG in serial dilutions of a positive control. Similar IgG MFI levels were obtained in singleplex and multiplex measurements, with a strong correlation for all antigens assessed (R² > 0.98; p < 0.05) (Additional file 1). Coupled beads were stored multiplexed at a concentration of 1000 beads/µL/antigen at 4 °C and protected from light.

WHO Reference Reagent for anti-malaria (P. falciparum) human plasma (10/198) (referred as WHO reference reagent). A pool derived from plasma donations collected at the Blood Bank from individuals based in Kisumu, Kenya, with a history of malaria. This reference reagent presents IgG reactivity to P. falciparum AMA-1, MSP-1₁₉, MSP-1₄₂, MSP-2 and MSP-3 [25]. The reagent has been defibrinated and diluted (1:5) with deionized sterile water and filled into 1 mL/ampoules. Each ampoule has been lyophilized comprising a freeze-dried residue of diluted human plasma.

RTS,S vaccine positive control (referred as WHO-CSP pool). An RTS,S pool prepared with plasmas from 10 Mozambican children vaccinated with RTS,S/AS02 with known high IgG titres to CSP at peak response [65] was added to the WHO reference reagent (1:50 WHO reference reagent + 1:100 RTS,S pool), creating a CSP and HBsAg antibody enriched WHO reference reagent.

Malaria primo-infected plasma pool (referred as IgM pool). A customized pool prepared with plasmas from 20 malaria naïve European adults with known high anti-malaria IgM levels after being experimentally infected with P. falciparum in a controlled human malaria infection (CHMI) trial [66]. To prepare the pool, we first selected the time point that elicited the highest IgM breadth of response to a panel of 20 BS and 1 PE antigens from the CHMI trials conducted in Barcelona (day 35) and Tübingen (day 84). Ten individuals from each trial with the highest IgM breadth of response were selected and pooled.

RTS,S samples. Three samples from individuals participating in the RTS,S malaria vaccine phase 2b trial conducted in Mozambique [65] were randomly selected. High, medium and low responders were defined by tertiles.

IgG, IgG_1–4 subclasses, IgM and IgE levels were measured in the WHO reference reagent and other customized pools against multiplexed P. falciparum antigens using the xMAP™ technology (Luminex Corp., Austin, Texas). Fifty microliter of multiplexed antigen-coupled beads were added to a 96-well μClear^® flat bottom plate (Greiner Bio-One, Frickenhausen, Germany) at 1000 beads/analyte/well. To assess the optimal temperature and duration of sample incubation for IgG and IgG_1–4 assays, 50 µL of WHO reference reagent at 11 serial dilutions (1:3, starting at 1/150) and the negative control at 4 serial dilutions (1:2, starting at 1:50) were incubated against a panel of 14 P. falciparum antigen-coated beads in a 96-well plate (Table 1). Plates were incubated in a rotatory shaker at 600 rpm and protected from light under three conditions: (i) 37 °C for 2 h; (ii) 4 °C ON and (iii) room temperature (RT) for 1 h. For the IgM assay, 50 µL of the WHO reference reagent or the IgM pool were assayed in 15 serial dilutions (1:3, starting at 1/50) against a panel of 40 P. falciparum antigens plus HBsAg (Table 1). Plates were incubated at two different conditions: 37 °C for 2 h and 4 °C ON. IgE levels in the WHO reference reagent assayed at 8 serial dilutions (1:2, starting at 1/10) were also measured under two different incubation conditions: 37 °C for 2 h and 4 °C ON. Finally, using the WHO-CSP pool, 23 standard curves for IgG, IgG1, IgG3 and IgM were constructed; and 12 standard curves for IgG2 and IgG4 all of 18 serial dilutions (1:2, starting at 1:50). The standard curves were incubated at 4 °C ON against a total of 40 antigens (Table 1). Beads coupled with BSA and GST were included in the panel as background controls to assess unspecific binding to BSA and GST. After the incubation, plates were washed with PBS-0.05% Tween 20 buffer using a manual magnetic washer platform (Bio-Rad, Hercules, CA, USA). Secondary antibodies were added as previously described [23]. Briefly, biotinylated anti-human IgG at 1:2500 (Sigma B1140, polyclonal), anti-human IgM at 1:1000 (Sigma B1265, polyclonal) anti-human IgG3 at 1:1000 (Sigma B3523, clone HP-6050), and anti-human IgG1 at 1:4000 (Abcam ab99775, clone 4E3). For the IgG2, IgG4 and IgE assays, the secondary antibodies were unconjugated to biotin: mouse anti-human IgG2 at 1:500 (Thermo Fisher MA1-34755, clone HP6014), mouse anti-human IgG4 at 1:8000 (Thermo Fisher MA1-80332, clone HP6025), and mouse anti-human IgE at 1:500 (Abcam ab99834, clone HP6029). All secondary antibodies were incubated 60 min at RT and washed. In IgG2, IgG4 and IgE assays, a tertiary biotinylated goat anti-mouse IgG (Sigma B7401, polyclonal) was added and incubated 60 min at RT. Plates were washed as before and streptavidin-R-phycoerythrin at 1:1000 (Sigma, Tres Cantos, Spain) was added to all wells and incubated 30 min at RT. Plates were washed and beads re-suspended in 100 μL/well of PBS-BN, protected from light and stored ON at 4 °C to be read the next day. Plates were read using the Luminex xMAP^® 100/200 analyser (Luminex Corp., Austin, Texas) and at least 50 microspheres per analyte were acquired per well. Results are expressed as Median Fluorescence Intensity (MFI).

RTS,S-induced antibodies were measured in 3 samples from RTS,S-vaccinated children with known high, medium and low responses, together with serial dilutions of the WHO reference reagent or the IgM pool (1:3 starting at 1:50 for IgG, IgG_1–4 and IgM, and 1:2 starting at 1:10 for IgE) and incubated at 4 °C ON. Samples were assayed in 4 serial dilutions (1:10) starting at 1:500 for IgG, in 3 serial dilutions (1:10) starting at 1:100 for IgM and IgG1, and in 2 serial dilutions (1:10) starting at 1:50 for IgG2 and IgG4. Samples were not assayed for IgG3 or IgE.

To stabilize the variance, the analysis was done on log₁₀-transformed values of the MFI measurements. The correlation and reliability between the different sample incubation conditions for the IgG and IgG_1–4 subclasses measured in the positive control, the negative control and the blanks were evaluated. After the Shapiro–Wilk normality test was applied, differences between conditions were assessed by Kruskal–Wallis test with posthoc Tukey test. Reliability was assessed by the interclass correlation coefficient (ICC) [67]. Titration curves of antibody concentrations vs. MFIs per antigen were fitted using a five-parameter (5PL), a 4PL or an exponential logistic equation depending on the best yield, following the formula MFI = Emax + ((Emin − Emax)/((1 + ((Conc/EC₅₀)^Hill))^Asym)), where EC₅₀ is the half maximal effective concentration, Emin is the minimum response, Emax is the maximum response, Asym is the asymmetry factor and Hill is the slope factor [68], using the drLumi package [69]. We calculated the coefficient of variation (CV) of Emin, Emax and used the goodness of fit model to assess the fitting of the curves. All analyses were done using R version 3.4.1.

To assess the suitability of the WHO reference reagent as a positive control to capture all responses in the context of RTS,S vaccine studies, levels of IgG, IgG_1–4, IgM and IgE against the RTS,S-specific antigens (CSP full length, CSP NANP repeat and CSP C-terminus) were measured and compared to levels in sera from RTS,S-vaccinated children from a phase 2b trial with known IgG CSP titres [65] (Fig. 1). The WHO reference reagent and RTS,S-vaccinees antibody responses to the whole antigenic panel (Table 1) are shown in the Additional file 2. The IgM pool and RTS,S-vaccinees IgM levels were also compared (Fig. 1h and Additional file 2). The WHO reference reagent presented lower IgG, IgG1, IgG2, IgG4, and IgM levels to RTS,S antigens than samples from RTS,S-vaccinated children who had high CSP responses (Fig. 1a–d, g). Comparisons of IgG3 and IgE levels were not possible because these data were not available for RTS,S samples. The IgM pool presented higher IgM levels to RTS,S antigens than the WHO reference reagent and the RTS,S samples. Consequently, we decided to prepare a customized positive control for the RTS,S immunological studies, containing 1:50 of the WHO reference reagent plus 1:100 of pooled plasma from RTS,S-vaccinated children with high CSP titres (WHO-CSP pool). IgG responses were compared between the WHO-CSP pool and the WHO reference reagent (Additional file 3). In addition, the EC₅₀ ratio between positive controls (EC₅₀ WHO reference reagent/EC₅₀ WHO-CSP) was calculated for RTS,S-specific antigens as a proxy measure of relative potency of the WHO-CSP pool to the WHO reference reagent (Additional file 4). The EC₅₀ ratio for the 3 CSP antigens was between 0.44 and 0.58 for IgG, IgG1 and IgG3, and close to 1 for IgG2 against CSP full length.

Fifteen proteins in the multiplex panel were GST-fused (Table 1). The WHO-CSP pool was reactive to GST because the sera from RTS,S vaccinees 1 month after primary vaccination had antibodies that cross-reacted with GST. However, even if the samples contained equal or higher levels of antibodies to GST, this did not impede to accurately measure anti-malarial antibodies to the GST fusion proteins, as shown in correlation analyses of GST vs. GST fusion proteins in plasmas from RTS,S vaccinees (Additional file 5).

The WHO-CSP pool was used to generate IgG, IgG_1–4 and IgM titration curves incubating at 4 °C ON in the context of an RTS,S immunology study. The level of response was antigen-dependent; the most immunogenic proteins (AMA-1 3D7 and FVO, MSP-1₄₂ 3D7 and FVO) gave saturated signals even at the 1:6.5 × 10⁶ dilution (Fig. 2).

To further characterize the IgG subclass composition of the WHO-CSP pool, the ratios of IgG_1–4 subclasses to total IgG [MFI IgG subclass at dilution (i)/MFI total IgG at dilution (i) × 100] were measured (Fig. 3). The predominance of IgG subclasses also varied depending on the antigen. For example, IgG1 responses were dominant for HBsAg, LSA-1, MSP-5, P41, RH1, RH2, PTRAMP, RH4.2, RH4.9 and SSP2, whereas MSP-2 full length, MSP-1 block 2 and RH4 induced mainly IgG3. IgG subclass responses to AMA-1 (3D7 and FVO), CSP (C-terminus and NANP repeat), EXP-1, MSP-1₄₂ (3D7 and FVO), MSP-3 and RH5 were dominated by IgG1 and IgG3.

To assess the optimal temperature and time of incubation for the measurement of IgG and IgG_1–4 subclasses, the assay performance of the WHO reference reagent against a panel of 14 P. falciparum antigens (Table 1) under three different incubation conditions (4 °C ON, 37 °C 2 h and RT 1 h) was compared. IgG and IgG_1–4 assays varied depending on the incubation procedure, with the largest difference between 4 °C ON and RT 1 h (p < 0.001) for IgG. No differences were found between these two incubation conditions for IgG2, IgG3 and IgG4. Differences between 4 °C ON and 37 °C 2 h were only observed for IgG (p = 0.026). IgG and IgG_1–4 levels against BSA and blanks were not affected by the incubation conditions. The MFI levels of IgG and IgG_1–4 measured in the negative control only varied when comparing 4 °C ON vs. RT 1 h (p < 0.001) for some of the antigens. Figure 4 shows examples of the results for IgG1, and the complete data set is in the Additional file 6. The incubation at 4 °C ON, on average, showed the highest MFIs in the first dilution, except for IgG4 and reached blank levels at the lowest dilution (Fig. 4 and Additional file 6). Negative control MFI levels were also higher at 4 °C ON compared to other conditions, however the difference with the WHO reference reagent at same dilution was high enough to establish a positivity threshold (Fig. 4).

Correlations between incubation conditions for IgG and IgG_1–4 subclasses measured against all antigens in the WHO reference reagent and negative control showed a r² > 0.93 for all IgG and IgG_1–4 subclasses. The ICCs between incubation conditions for IgG and IgG_1–4 measured in the WHO reference reagent showed overall good reliability, being 0.91 (0.89–0.93) for IgG3, 0.88 (0.87–0.89) for IgG1, 0.83 (0.79–0.86) for total IgG, 0.79 (0.74–0.83) for IgG2 and 0.63 (0.53–0.72) for IgG4. However, as seen in Fig. 4 and Additional file 6, ICCs in the negative control were of lower reliability, being of 0.85 (0.78–0.9) for IgG4, 0.74 (0.64–0.82) for IgG2, 0.38 (0.23–0.53) for total IgG, 0.39 (0.23–0.54) for IgG1 and 0.11 (− 0.03–0.14) for IgG3. Blank levels were similar between incubation conditions (Fig. 4 and Additional file 6). Taking together these results, we chose the incubation at 4 °C ON as the optimal for the IgG assays.

Incubation conditions to measure IgM and IgE responses against a panel of 38 P. falciparum antigens plus HBsAg, α-Gal, BSA and GST (Table 1) were tested using the WHO reference reagent and an alternative IgM pool. The IgM pool gave higher IgM responses and of higher range compared to those obtained with the WHO reference reagent for most of the antigens, especially AMA-1s, MSP-1s and CSPs (Fig. 5). Incubation of the IgM pool at 4 °C ON showed higher responses compared to incubation at 37 °C 2 h (Additional file 7A), with 80% of the antigens studied (35/43) presenting a higher EC₅₀ (i.e. AMA-1 3D7 EC₅₀ 4 °C ON 3.64 ± 0.66 and EC₅₀ 37 °C 2 h 2.62 ± 0.96). The IgM responses of the negative control measured at first dilution were higher than those of IgG and IgG subclasses, but levels dropped quickly after the first dilution. Overall, IgM pool responses showed higher difference to the negative control than those obtained with the WHO reference reagent (Fig. 5). Similar differences in IgM responses between incubation conditions were obtained with the WHO reference reagent, measuring higher levels when incubating at 4 °C ON than at 37 °C 2 h (Additional file 7B). IgM technical blanks were not affected by incubation conditions (Additional file 7A, B). Correlations for IgM responses between incubation conditions were r² = 0.96 for both WHO reference reagent and IgM pool. For the IgE assay, there were no differences between incubation conditions (Additional file 7C).

The ICCs between antibody responses measured in the two incubation conditions with the WHO reference reagent were 0.92 (0.91–0.93) for IgM and 0.82 (0.79–0.85) for IgE; and the ICC between conditions for the IgM assay using the IgM pool was 0.91 (0.9–0.92). However, IgM responses of negative controls showed moderate reliability between incubation conditions, having an ICC of 0.66 (0.57–0.73).

When comparing antibody levels measured in the WHO reference reagent vs. the IgM pool, there was moderate reliability, with ICC of 0.65 (0.61–0.769) at 4 °C ON, and 0.66 (0.61–0.7) at 37 °C 2 h, meaning that there was 35% of variability between reference pools. Considering the strong correlation and reliability of the two incubation conditions, but the higher IgM levels and MFI ranges obtained at 4 °C ON, this incubation was also chosen for the IgM assay.