From a recombinant key antigen to an accurate, affordable serological test: Lessons learnt from COVID-19 for future pandemics

Abstract

Serological tests detect antibodies generated by infection or vaccination, and are indispensable tools along different phases of a pandemic, from early monitoring of pathogen spread up to seroepidemiological studies supporting immunization policies. This work discusses the development of an accurate and affordable COVID-19 antibody test, from production of a recombinant protein antigen up to test validation and economic analysis. We first developed a cost-effective, scalable technology to produce SARS-COV-2 spike protein and then used this antigen to develop an enzyme-linked immunosorbent assay (ELISA). A receiver operator characteristic (ROC) analysis allowed optimizing the cut-off and confirmed the high accuracy of the test: 98.6% specificity and 95% sensitivity for 11+ days after symptoms onset. We further showed that dried blood spots collected by finger pricking on simple test strips could replace conventional plasma/serum samples. A cost estimate was performed and revealed a final retail price in the range of one US dollar, reflecting the low cost of the ELISA test platform and the elimination of the need for venous blood sampling and refrigerated sample handling in clinical laboratories. The presented workflow can be completed in 4 months from first antigen expression to final test validation. It can be applied to other pathogens and in future pandemics, facilitating reliable and affordable seroepidemiological surveillance also in remote areas and in low-income countries.

Keywords: COVID-19; Detection of antibodies elicited by infection or vaccination; Dried blood spot for diagnostic testing; Purification of SARS-COV-2 spike protein; Recombinant antigen production in mammalian cells; Serological test development.

Graphical abstract

Fig. 1

S protein production and purification. (A) Left panel: CHO-K1 and HEK293 were transiently transfected with pαH plasmid. Center panel: HEK293 cells were transiently transfected at high (A, 2.0 µg/mL) and low (B, 0.75 µg/mL) pαH plasmid concentration, as well as stably transfected by co-transfecting the pαH plasmid (0.75 µg/mL) along with a second vector (0.15 µg/mL) containing the neo selection marker. Right panel: stability of expression of the secreted protein was confirmed for 100 days post-transfection (dpt). (B) Left panel: high cell viabilities and viable cell densities (VCD) were achieved for the stable cell line grown in shake flasks and 1.5-L stirred-tank bioreactors, in batch and fed-batch mode, using chemically-defined, animal component-free culture media. Right panel: spot blot for detection of S protein in the cell culture supernatant on different days. (C) S protein identity was confirmed by Western blot analysis at ~170 kDa. SDS-PAGE showed a lower purity for the concentrated/diafiltered (UF/DF) sample (lane 3), but a very high purity for the sample purified by affinity chromatography (AC, lane 4). 1: Cell culture supernatant from non-transfected parental HEK293 cells. 2: Molecular-mass standard (protein markers, from top to bottom, have a molecular mass of 250, 150, 100 and 75 kDa). 3: UF/DF sample after 90-fold volumetric concentration and buffer exchange into PBS. 4: Eluate from AC. 5: Supernatant from stable recombinant cell line. (D) A typical chromatogram of the AC purification. Fractions of the flow-through (fr1 to fr8) were collected during sample injection and analyzed by spot blot to evaluate progressive saturation of affinity ligands, showing a high binding capacity and a low degree of S protein leakage. For comparison, spot blots of the injected cell culture supernatant, or a 1:10 dilution, are shown in the box on top. In all immunoblots, presence of S protein was detected using a pooled serum of SARS-COV-2 convalescent patients (1:1000) as primary antibody, followed by incubation with anti-human IgG-HRP and chemiluminescent detection.

Fig. 2

Comparison of ELISAs with UF/DF- or AC-purified S protein. (A) ELISA performance using low-purity (UF/DF) or high-purity (AC) S protein antigen to coat plates. A total of 22 samples were used: 15 sera from COVID-19 convalescent patients (SARS-COV-2 samples, open circle), two control samples collected until 2018 (pre-COVID-19, yellow diamond), one post-COVID-19 control sample from a healthy individual, two samples from SARS-COV-2 infected individuals characterized as PCR+/RDT- (#1 light gray filled circle, #2 dark gray filled circle), and one sample from a SARS- COV-2 convalescent patient who had the severe form of the disease (#3, red square). The cut-off was defined as mean + 3 standard deviations (X + 3 SD) of the O.D. of negative controls. (B) Samples used in (A) were titrated in four serial dilutions (1:40, 1:120, 1:360 and 1:1080), both for UF/DF or AC ELISA. (C) O.D. summation of the titration curves shown in (B); symbols as in (A). (D) Samples used in (A) were titrated in four serial dilutions (1:40, 1:120, 1:360 and 1:1080) for AC ELISA using different S-protein coating concentrations (0.3 – 10 μg/mL, corresponding to 15 – 500 ng/well). (E) Three selected samples from (D) were diluted 1:40 to assess assay robustness by evaluating results from data obtained with different antigen lots and two different operators. Representative data from 3 experiments using 3 different lots of the AC-purified S protein (lots A to C) are shown. Experiment with lot A was performed by one operator, and a second operator performed experiments with lots B and C. RDT: rapid diagnostic test.

Fig. 3

Validation of S-UFRJ ELISA based on AC-purified antigen for early detection and quantification of anti-S IgG antibodies. (A) Anti-S IgG antibody detection in samples from healthy individuals, obtained as pre-pandemic controls (n = 124, yellow diamond) or post-pandemic controls from individuals who tested negative for the virus by PCR (n = 20, open square), and samples from COVID-19 patients who were PCR-positive (n = 66, open circle). Relative levels of antibodies are shown as O.D. ratio of values of individual samples to the [mean + 1 standard deviation (X + 1 SD)] of the O.D. of the negative controls in the same ELISA plate. Sera samples were diluted at 1:40. For this stage of development of the assay, it was conservatively established that an O.D. ratio below 1 indicates a negative result (N), an O.D. ratio above 2 indicates a positive result (P), and an O.D. ratio between 1 and 2 is considered undetermined (U). (B) The PCR-positive samples shown in (A) were tested for anti-SARS-COV-2 IgG by a commercial rapid diagnostic test (RDT) (MedTeste, Medlevensohn, Brazil, imported from Hangzhou Biotest Biotech Co.). Samples were then grouped by IgG reactivity according to the RDT result (RDT IgG-, open circle; RDT IgG+, red filled circle), and their anti-S IgG levels measured by the S-UFRJ ELISA are plotted. (C) Anti-S IgG levels measured by the S-UFRJ ELISA for the RDT IgG-negative samples shown in (B), grouped according to the RDT IgM result as RDT IgM- (open circle) or RDT IgM+ (open red circle). (D) Levels of anti-S IgG in samples grouped according to the timepoint of sample collection given in days after symptoms onset (DASO); symbol legends are indicated in the figure. (E) Positivity rates versus DASO for anti-S IgG measured by S-UFRJ ELISA (open circle), for IgM measured by RDT (open red circle) and for IgG measured by RDT (red filled circle). (F) S-UFRJ IgG titration of 16 samples from COVID-19 convalescent patients (SARS-COV-2 samples); samples with the highest and lowest endpoint titer of the group are indicated in solid red and solid black colors, respectively. (G) Correlation between S-UFRJ ELISA IgG endpoint titers and neutralization (PRNT90) for samples titrated in (F) (n=16). Three samples with PRNT90 values <10 are plotted as 1. Statistical analysis was performed using Pearson’s test.

Fig. 4

Receiver operating characteristic (ROC) curve of S-UFRJ test. The curve is based on calculating the sensitivity and specificity as a function of varying cut-off values (each data point represents one cut-off value). This curve was prepared based on a panel comprising 420 positive and 68 negative samples, and allowed determining the sensitivity and specificity of the S-UFRJ test.

Fig. 5

S-UFRJ test optimization for sample collection by dried blood spots. (A) Dried blood spots (DBS) obtained by finger pricking with commercially available lancing devices: a 2.5 cm (W) x 7.5 cm (L) filter paper with three blood spots from the same volunteer and a commercially available paper hole punching device were used to prepare a DBS disk (arrowhead) from which blood was eluted for ELISA testing. (B) S-UFRJ ELISA comparing the O.D. values for plasma samples in increasing serial dilutions and for the corresponding eluates prepared by incubating the DBS disks in increasing volumes of buffer. (C) O.D. summation of the data shown in (B). (D) Dried blood spots collected in plastic strips containing 1, 2 or 3 pads of filter paper. (E-G) correlations of O.D. ratios between samples collected in the strip pad 1 and either the respective plasma sample (E), the respective DBS disk (F), or the second pad of the same strip (G). In E-G, plasma dilution was 1:40, DBS in filter paper disks were eluted in 100 µL of PBS-1% BSA, and DBS in single pads were eluted in 200 µL PBS-1% BSA. O.D. ratios were calculated as defined earlier, using data from negative control plasma samples (negative controls from DBS disks or pads can be equally used, if available). Statistical analysis was performed using Pearson’s test.

Fig. 6

S-UFRJ test applied to monitor anti-S IgG seroconversion following vaccination. Dried blood spots were collected in plastic strips containing pads of filter paper. For each of the 15 vaccinees (V01 to V15), DBS samples were collected at different time points, before and after the first and second doses of the inactivated vaccine Coronavac (Sinovac), which were administered on days 0 and 28.