A haemagglutination test for rapid detection of antibodies to SARS-CoV-2

Abstract

Serological detection of antibodies to SARS-CoV-2 is essential for establishing rates of seroconversion in populations, and for seeking evidence for a level of antibody that may be protective against COVID-19 disease. Several high-performance commercial tests have been described, but these require centralised laboratory facilities that are comparatively expensive, and therefore not available universally. Red cell agglutination tests do not require special equipment, are read by eye, have short development times, low cost and can be applied at the Point of Care. Here we describe a quantitative Haemagglutination test (HAT) for the detection of antibodies to the receptor binding domain of the SARS-CoV-2 spike protein. The HAT has a sensitivity of 90% and specificity of 99% for detection of antibodies after a PCR diagnosed infection. We will supply aliquots of the test reagent sufficient for ten thousand test wells free of charge to qualified research groups anywhere in the world.

Fig. 1. Haemagglutination test (HAT) for detection of antibodies to SARS-CoV-2 receptor binding domain.

A Concept of the HAT. B Sequence of VHH(IH4)-RBD fusion protein. Residues underlined are encoded by cloning sites AgeI (TG) and SalI (AST). The codon-optimised cDNA sequence is shown in Supplementary Information. C SDS-PAGE gel of purified VHH(IH4)-RBD proteins. Three micrograms of protein were run on 4–12% Bolt Bis-Tris under reducing conditions. 1: IH4-RBD produced in house in Expi293F cells. 2: IH4-RBD produced by Absolute Antibody, Oxford, in HEK293 cells. This was done twice with similar results.

Fig. 2. Haemagglutination with human monoclonal antibodies or nanobodies to the SARS-CoV-2 RBD.

A Titration of IH4-RBD and monoclonal antibody CR3022 to RBD. Doubling dilutions of CR3022 and IH4-RBD were prepared in separate plates. 50 µL red cells (O−ve whole blood diluted 1:40 in PBS) were added to the CR3022 plate, followed by transfer of 50 µL titrated IH4-RBD. From this titration, 100 ng/well of IH4-RBD was chosen for detection. Similar results were obtained in five other experiments, performed with three separate batches of IH4-RBD. B Detection of other anti-RBD monoclonal antibodies and ACE2-Fc. Doubling dilutions of monoclonal antibodies were prepared in duplicate in 50 µL PBS (from a stock solution of 20 µg/mL) from left to right, 50 µL of 1:40 O−ve red cells were added, followed by 50 µL of IH4-RBD (2 µg/mL in PBS). The endpoint was defined as the last dilution without teardrop formation on tilting the plate for ~30 s. The binding sites for CR3022, EY6A, and ACE2 on RBD have been defined^,,,. EW-9B and EW-9C are monoclonal antibodies against non-RBD epitopes on the spike protein . ACE2-Fc has been described in ref. . Similar results were obtained in two other titration experiments.

Fig. 3. Titration of antibodies in the agglutination assay.

A Eighteen plasma samples from mild cases were compared for titration in the HAT with 1:40 O−ve whole blood from a seronegative donor and endpoint titre in an RBD ELISA. Four samples were negative in both assays. The data point marked with an arrow on the graph (plasma 2 on the plate, Fig. 3B) was checked with a reagent composed of IH4 without RBD and shown to be dependent on antibodies to the RBD. This sample did score positive for antibodies to full length spike in an ELISA (endpoint titre 1:1123). These results were confirmed in a repeat assay. B An example of titration: positive agglutination endpoints (loss of teardrop) are marked with a black solid-line circle, partial teardrops are marked with a dotted-line circle. C Titration of mAb CR3022 and a high titre serum from a COVID-19 patient show that agglutination in the HAT detected by eye correlates with antibody binding to RBCs, as revealed by FACS analysis. Standard HAT titration was performed by double dilutions of RBC suspension containing 1 µg/mL of the IH4-RBD reagent, bar the first and last well. After the HAT assay, the RBCs were stained for FACS analysis by performing three washes before incubation for 60 min at 4 °C with FITC-labelled goat–anti-human IgG. RBCs were then washed twice before analysis by flow cytometry. Green numbers in the upper right corners of the histograms correspond to the geometric mean fluorescence intensity. Original graphs are provided as Supplementary Fig. 4. Similar results were obtained in three other experiments.

Fig. 4. Operating characteristics of the HAT.

A The test set of 297 randomised plasma samples was diluted 1:40 and mixed with 1:40 O−ve blood in two columns. IH4-RBD (100 ng in 50 µL) was added to the test samples and PBS to negative control wells. The plates were incubated at room temperature for 1 h to allow the red cell pellet to form, then tilted for ~30 s to allow a teardrop to form. Complete loss of teardrop was scored as positive agglutination (marked with a black solid-line circle). Full teardrop or partial teardrop (marked with a dotted-line circle) was scored negative. The samples in columns were re-randomised and tested a second time 2 days later. B Contingency table showing the operating characteristics of the HAT. Source data are provided as a Source Data file.

Fig. 5. Titration of the set of 232 samples in the HAT.

A The collection included 32 samples from 24 critical patients, 62 samples from 48 severe, 39 samples from 32 mild, 20 single samples from healthcare workers (HCW), 54 samples from 43 patients with unrelated sepsis in the pre COVID-19 era, and 25 samples from healthy unexposed controls. Median is indicated by a red line. DD doubling dilutions. B Comparison to Siemens result (anti-RBD) with HAT titre by doubling dilution for 153 samples from critical, severe, mild, and HCW SARS-CoV-2 PCR positive donors. Source data are provided as a Source Data file.

Fig. 6. HAT as a point of care test.

Capillary blood samples were obtained by Lancet. Antibodies to the RBD were detected by HAT on autologous red cells in the sample in “Test” wells (plasma at 1:40) after addition of 100 ng/well IH4-RBD (see “Methods”). NC negative control (PBS replaces IH4-RBD), PC positive control (20 ng/well CR3022, an anti-RBD monoclonal antibody added). In parallel, after removal of red cells, the plasma was tested in a standard ELISA for detection of antibodies to the RBD. Low levels of antibody detected in the ELISA were sufficient to give a positive result in the HAT. Detection of antibodies by HAT on autologous red cells obtained by finger prick has been repeated multiple times in both Oxford and Toulouse, and will be the subject of a future report. Source data are provided as a Source Data file.