Development of a high-throughput flexible quantitative suspension array assay for IgG against multiple Plasmodium falciparum ant

Study samples

Plasma samples used in the establishment of the qSAT assay were obtained from children and adult volunteers naturally exposed to malaria, participating in two studies conducted at the Centro de Investigação em Saúde da Manhiça (CISM), southern Mozambique. One set of samples were from a clinical trial of intermittent preventive treatment during pregnancy with sulfadoxine-pyrimethamine (IPTp-SP) (Clinicaltrials.gov identifier NCT00209781) conducted in 2003–2005 [17]. A unique operator in 22 consecutive days and in triplicates assayed plasmas from this study by qSAT to assess precision of the assay. Another set of samples were from a clinical trial of different chemoprophylaxis schedules to selectively control first exposure to P. falciparum in infancy (Clinicaltrials.gov identifier NCT00231452) conducted in 2005–2009 [18]. Two operators assayed samples from this second study in 5 different days. A pool of hyperimmune plasmas (HIP) from Mozambican adults life-long exposed to malaria was used as a polyclonal positive control and antigen-specific reference standard [19]. Negative controls from malaria-naïve volunteers were obtained from ISGlobal repository. The analysis of all the samples was covered under protocols approved by the National Mozambican Ethics Review Committee and the Hospital Clínic of Barcelona Ethics Review Committee, and written informed consent was obtained from all participants or their parents/guardians before collection of specimens.

Antigens and antibodies

A combination of 11 antigens expressed during the pre-erythrocytic and erythrocytic stages of P. falciparum life cycle was selected for the qSAT multiplex panel. The apical membrane antigen (AMA)-1 of the 3D7 parasite strain [20,21,22], the F-2 region of the erythrocyte binding antigen (EBA)-175 [22, 23], the Duffy binding-like (DBL)3x domains and DBL-α of the erythrocyte membrane protein (PfEMP)-1 were produced at ICGEB (Delhi, India) [24, 25]. The AMA-1 and the 42 kDa fragment of the merozoite surface protein 1 (MSP-1₄₂) from the FVO strain were provided by WRAIR (Walter Reed Army Institute of Research, MD, USA) [26,27,28]. The liver stage antigen (LSA)-1 [29, 30], the sporozoite surface protein 2 (SSP2 or TRAP) [31], the circumsporozoite protein (CSP) [32] and the cell traversal-ookinete surface antigen (CelTOS) [33] were purchased from Protein Potential, LLC (Rockville, MD, USA).

Sp3C6 monoclonal antibody (mAb) produced in mice was gifted to this study by the Pluschke lab at the Swiss TPH Institute (Basel, Switzerland). Sp3C6 mAb gives specific responses to CSP of the 3D7 P. falciparum strain [34].

Microsphere coupling

A qSAT multiplex panel was constructed to quantify IgG responses to P. falciparum antigens using Luminex xMAP™ technology (Luminex Corp., Austin, TX, USA) and the Bio-Plex 100 platform (Bio-Rad, Hercules, CA, USA). MagPlex polystyrene 6.5 μm COOH-microspheres (Luminex Corp, Austin, TX, USA) of different ID regions were selected for each antigen, including one for bovine serum albumin (BSA). For the standard curve, microspheres were coupled to anti-human IgG F’ab antibody (Sigma-Aldrich, Madrid, Spain). For comparison of the anti-IgG standard curve with a curve generated from the anti-CSP Sp3C6 mAb, microspheres were coupled to anti-mouse IgG F’ab antibody (Jackson ImmunoResearch Inc. PA, USA). Briefly, microspheres were washed, sonicated and activated with Sulfo-NHS (N-hydroxysulfosuccinimide) and EDC (1-Ethyl-3-[3-dimethylaminopropyl]carbodiimide hydrochloride) (Pierce, Thermo Fisher Scientific Inc., Rockford, IL, USA). Microspheres were washed and resuspended in cold Dulbecco’s PBS (dPBS) pH 7.2 (Invitrogen, Carlsbad, CA, USA) or 50 mM MES pH 5.0 (Sigma, Tres Cantos, Spain), depending on the optimal buffering system for each individual antigen. The recombinant proteins were added to the tubes in concentrations ranging from 30 to 50 μg/mL and left at 4°C on a shaker overnight. Coupled microspheres were resuspended in PBS with 1% BSA and 0.05% sodium azide (PBS-BN) to block. Microspheres recovery was quantified on a Guava PCA desktop cytometer (Guava, Hayward, CA, USA). Equal amounts of each antigen-coupled microspheres were combined in multiplex tubes and stored at 1000 microspheres/μL at 4°C, protected from light. Anti-IgG-coupled microspheres were stored at 2000 microspheres/μL at 4°C, protected from light in singleplex. Bead blocking agent BSA in the coupling buffers to covalently ‘block’ the free carboxylic group (-COOH) from the microspheres was included, absorbing most of the non-specific binding to secondary or tertiary antibodies during assay steps [12] and heterophilic antibody binding seen in previous systems [13]. Also a BSA-coated microsphere in the multiplex panel was included, to determine non-specific ‘bead binders’ of serum IgG to the BSA.

qSAT assay

Antigen-coupled microspheres were added to a 96-well μClear^® flat bottom plate (Greiner Bio-One, Frickenhausen, Germany) in multiplex (1000 microspheres per analyte per well) in a volume of 50 μL of Luminex Buffer (PBS-BN). Anti-IgG-coupled microspheres were added to the plate in singleplex (2000 microspheres per analyte per well) in 50 μL of Luminex Buffer. 50 µL of test plasma samples diluted 1:250 and 1:10,000 in Luminex Buffer were added to the plates in duplicates (final dilutions of 1:500 and 1:20,000, respectively). The HIP pool was used as a positive control and included on each assay plate diluted 1:150,000. Technical blanks consisting of Luminex Buffer and microspheres without samples were added in duplicate wells to detect and adjust for non-specific microsphere signal. Plates were incubated for 1 h at room temperature in agitation and protected from light. Then, washed three times with 100 μL PBS-T (0.05% Tween 20 in PBS) on a Bio-Plex Pro wash station with magnetic platform (Bio-Rad, Hercules, CA, USA). 100 μL of biotinylated anti-human IgG (Sigma-Aldrich, Tres Cantos, Spain) diluted 1:2500 in Luminex buffer was applied to all wells and incubated for 45 min as before. For the assay of anti-CSP Sp3C6 mAb and mouse IgG standard curves, biotinylated anti-mouse IgG (Sigma-Aldrich, Madrid, Spain) was used. After washing plates, 100 μL of streptavidin-conjugated R-phycoerythrin (Invitrogen, Carlsbad, CA, USA) diluted 1:1000 (1 μg/mL) in Luminex Buffer was applied to all wells and incubated for 25 min as before. Plates were washed and microspheres resuspended with 100 μL of Luminex Buffer, and covered with an adhesive film and stored at 4°C overnight to be read the next morning. Data were acquired on a Bio-Plex 100 reader using Bio-Plex Manager version 4.0 (Bio-Rad, Hercules, CA, USA). At least 50 microspheres per analyte were acquired, and median fluorescence intensity (MFI) was reported for each analyte.

Standard curve

A heterologous reference standard [35,36,37] for estimating concentration of antibodies in plasma was constructed using microspheres coupled to anti-human IgG F’ab region and dilution series of IgG purified from human serum (Sigma-Aldrich, Tres Cantos, Spain). The commercially available purified human IgG was incubated with the anti-human IgG F’ab microspheres in a 10-step dilution series (twofold) starting at 250 ng/mL and producing an 11-point curve. For comparison of a heterologous standard with a homologous standard, a curve was generated with purified mouse IgG (ThermoFisher, Spain) and microspheres coupled with anti-mouse IgG F’ab region (Jackson ImmunoResearch Inc. PA, USA) plus biotinylated anti-mouse IgG (Sigma-Aldrich, Madrid, Spain). This curve was contrasted with a curve generated with a dilution series of anti-CSP Sp3C6 mAb with CSP-coupled microspheres in singleplex. The 5-PL regression was the selected method for fitting curves due to its superior fit to antibody data:

where A is the lower asymptote (Emin), B is the slope at the inflection point (Hill), C is the concentration at the inflection point (EC50), D is the upper asymptote (Emax), and G is a factor of asymmetry added in the 5-PL regression model [38]. If the 5-PL regression model did not converge, then, a 4-PL method without asymmetry factor G was fitted instead. To obtain AU/mL, corresponding MFI values were adjusted by their corresponding blank values before curve fitting.

An r² cut-off value of 0.994 was used for each standard curve as acceptability criteria. Additionally, blank values of all antigen-coated microspheres had to be below 200 MFI, and for the anti-IgG-coated microspheres had to be below 300 MFI. The parameters of the anti-IgG standard curve were used in a Microsoft Excel template to calculate antigen-specific AU/mL, respectively. Using the derived parameters of the standard curve, the estimates of concentration were multiplied by corresponding dilution factors to calculate antigen-specific AU/mL.

Determinations of limit of blank (LOB) and limit of quantification (LOQ)

LOBs were estimated by measuring replicates of a technical blank (well without sample) and calculating the log₁₀ MFI mean and the SD (LOB = mean blank + 1.645 × (SD blank)). To calculate concentration in AU/mL for each sample and antigen tested, we first adjusted each MFI value by its corresponding blank values. Then, the parameters from a non-linear 5-PL regression model obtained from the singleplex IgG standard curve were fitted in the inverse 5-PL equation for each sample and antigen. We established a ‘good range’ of quantification where percent change in AU/mL does not exceed 5% for a 1% change in MFI, and these were considered the assay LOQ. Only AU/mL measurements were adjusted by their corresponding blanks values.

Assessment of precision

For the determination of repeatability and intermediate precision, antibody levels (in log₁₀ MFI or AU/mL) were measured against BSA, AMA-1 3D7, AMA-1 FVO, MSP-1₄₂ 3D7, MSP-1₁₉ 3D7, EBA-175, LSA-1 and CSP in 10 malaria-exposed and 9 non-exposed individuals spanning a large range of immunogenicities. Samples were measured in triplicates on 22 different days, equivalent to 1254 measurements for the 8 antigens included. Repeatability between replicates for each antigen and day was assessed by the Intra-class Correlation Coefficient (ICC) [39], one way ANOVA [40] and CV (\(\frac{SD}{Mean} \times 100\)) [3, 15, 41]. Bland–Altman plots were also used to assess ‘within-day reproducibility’ [40].

To assess reproducibility, two operators performed the assay on 5 different days in the same laboratory and using the same apparatus. Operator and day effects were assessed, but inter-laboratory variation could not be assessed. IgG (in log₁₀ MFI levels and AU/mL) to 11 P. falciparum antigens (AMA-1 3D7, AMA-1 FVO, MSP-1₄₂ 3D7, MSP-1₄₂ FVO, EBA-175, CelTOS, LSA-1, SSP2, DBL-α, DBL3x and CSP) and BSA were measured in 282 samples from malaria-exposed individuals, and 52 malaria-naïve individuals. Positive controls and blanks were included in all plates, and all samples were measured in duplicates.

Assessment of accuracy

Since samples in this study had unknown antibody concentrations, standard curves with known concentrations of total IgG measured on 22 different days were used to assess accuracy and the observed and expected concentrations for each day of analysis compared. Also, as performance of total IgG measurement might not be representative of the performance of antigen-specific IgG measurements, parameters from the curve fitting of standard curves with mouse IgG were compared with anti-CSP mouse mAb Sp3C6 IgG standard curves.

Statistical methods

All MFI or AU/mL measurements were log₁₀-transformed for statistical analysis. Means and 95% confidence intervals (CI) were calculated for repeated measures. T-tests were used to assess differences between means. One-way ANOVA and Levene’s test [42] were used to assess differences between replicates. Agreement was assessed by performing Bland–Altman plots [40, 43], and reliability by the ICC from psych R package [39, 44]. SD and CVs were calculated for precision measurements. A p value < 0.05 was considered statistically significant. Data were analysed using R software version 3.4.1.