key: cord-0794489-96j4g7b5
authors: Mravinacova, Sara; Jönsson, Malin; Christ, Wanda; Klingström, Jonas; Yousef, Jamil; Hellström, Cecilia; Hedhammar, My; Havervall, Sebastian; Thålin, Charlotte; Pin, Elisa; Tegel, Hanna; Nilsson, Peter; Månberg, Anna; Hober, Sophia
title: A cell-free high throughput assay for assessment of SARS-CoV-2 neutralizing antibodies
date: 2021-10-07
journal: N Biotechnol
DOI: 10.1016/j.nbt.2021.10.002
sha: 0f6d8e6b7afdcac1f5f0996fffee3c6edbcabeb3
doc_id: 794489
cord_uid: 96j4g7b5

Highly accurate serological tests are key to assessing the prevalence of SARS-CoV-2 antibodies and the level of immunity in the population. This is important to predict the current and future status of the pandemic. With the recent emergence of new and more infectious SARS-CoV-2 variants, assays allowing for high throughput analysis of antibodies able to neutralize SARS-CoV-2 become even more important. Here, we report the development and validation of a robust, high throughput method, which enables the assessment of antibodies inhibiting the binding between the SARS-CoV-2 spike protein and angiotensin converting enzyme 2 (ACE2). The assay uses recombinantly produced spike-f and ACE2 and is performed in a bead array format, which allows analysis of up to 384 samples in parallel per instrument over seven hours, demanding only one hour of manual handling. The method is compared to a microneutralization assay utilising live SARS-CoV-2 and is shown to deliver highly correlating data. Further, a comparison with a serological method that measures all antibodies recognizing the spike protein shows that this type of assessment provides important insights into the neutralizing efficiency of the antibodies, especially for individuals with low antibody levels. This method can be an important and valuable tool for large-scale assessment of antibody-based neutralization, including neutralization of new spike variants that might emerge.

The severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) emerged during late 2019 in China and since then has spread around the world and caused a global crisis with more than 200 million detected infections and a death toll exceeding 4 million [1] . Due to the wide range of symptoms that overlap with other respiratory infections, confirmation of the based protection against the virus, for which studies to assess the duration of the induced antibodies upon vaccination or natural infection are needed.

The envelope of the coronavirus SARS-CoV-2 consists of a lipid bilayer with three different surface proteins; the spike-, the membrane-and the envelope proteins [6] , all of which are exposed to the host upon entry. While the membrane-and envelope proteins are mainly responsible for virion assembly inside host cells, the spike protein plays an important role in host cell entry. The first identified mechanism by which the SARS-CoV-2 virus enters the host cell is via membrane-bound angiotensin converting enzyme 2 (ACE2) [7] . Through a high affinity interaction between the spike protein and ACE2, the viral and cellular membranes fuse allowing the viral genome to enter the cell [7] [8] [9] .

The SARS-CoV-2 spike protein is a heavily glycosylated trimer constituted by two functional subunits, S1 and S2. The trimeric structure can be stabilized by exchanging the furin cleavage site for two prolines (spike-f) [10, 11] . The S1 unit forms the outer part of the protein containing the receptor binding domain (RBD) which binds to ACE2, while the S2 unit is responsible for anchoring the protein in the bilipid layer of the virus envelope [6] . Serological tests that utilize various representations of the spike protein can assess the ability of antibodies to recognize different conformations and mutations of the spike protein. However, such tests do not evaluate the neutralizing capacity of the antibodies, which is important to assess the protection expected against an infection upon re-exposure to the virus. Hence, it is important not only to assess the amount of antibodies induced against the viral proteins, but also to understand whether they are able to neutralize the virus. This becomes even more necessary in the aftermath of a global vaccination campaign where it will be important to ensure that the vaccines give the expected protective effect.

The standard method to assess the neutralizing capacity of antibodies is by determining the infection rate, either of the virus itself or of a pseudovirus, in cultivated cells, where the level of antibodies needed to inhibit the viral infection of the cells is used as a measure of the neutralization capacity [12, 13] . Although data from cell-based neutralization assays are considered the gold standard for the analysis of virus neutralization capacity, comparison of results from different analyses and different laboratories might be problematic due to varying cell types, cell numbers and virus inoculum. Also, cultivation of cells and viruses requires a high-level safety laboratory environment. Because of this, analysis of many samples in a high J o u r n a l P r e -p r o o f throughput setting demands expensive infrastructure and labour. Therefore, methods based on pseudoviruses measuring the cell infection rate [14] , or binding assays (e.g. enzymelinked immunosorbent assay, ELISA) measuring inhibited binding between recombinantly produced ACE2 and the spike protein, have been developed [15] .

Presented here is the development and evaluation of a novel high throughput bead-based pseudoneutralization (PNT) assay for assessment of spike-specific antibodies with the ability to inhibit the interaction of spike-f with ACE2. The generated data is compared and benchmarked to a conventional neutralization assay, and to spike-targeting antibody levels detected in a serological assay.

Spike-f [11, 12] USA) according to [16] . Cells were harvested by centrifugation up to 13 d after transfection, depending on cell concentration and viability. The supernatant was filtered through a 0.2 µm filter, prior to purification. Spike-f protein was produced with a strep-tag [12] , and therefore a biotin blocking solution of 18.1 µl/ml of supernatant (BioLock, IBA Lifesciences, Göttingen, Germany, 2-0205-050) was added before filtration.

All purifications were performed using ÄKTAxpress chromatography systems (Cytiva, Uppsala, Sweden). ACE2 was purified on HPC4 columns according to [17] and spike-f was purified on StrepTrap (Cytiva) according to the manufacturer's instructions. Protein concentration was determined using absorbance at 280 nm and quality control was performed by SDS-PAGE with Western blotting (WB) and mass spectrometry (MS/MS). After purification, spike-f protein was biotinylated using a 25x molar excess of EZ-link Sulfo-J o u r n a l P r e -p r o o f NHS-LC-Biotin (Thermo Fisher Scientific), according to the manufacturer's protocol. In brief, EZ-link Sulfo-NHS-LC-Biotin was equilibrated to room temperature (RT) and dissolved in amine-free buffer to a 10 mM concentration immediately before use. A volume of biotin-reagent corresponding to a 25x molar excess was added to the spike-f, concentration 1 mg/ml, and the mixture was incubated at RT for 30 min. Unreacted reagent was removed from the sample through a NAP5 desalting column (Cytiva).

Positive control samples included blood samples donated from SARS-CoV-2 seropositive healthcare workers (HCW) after mild COVID19 infection (n=76) and hospitalized qPCRconfirmed COVID-19 patients (n=56) at Danderyd Hospital, Stockholm. The negative control samples were from SARS-CoV-2 seronegative HCW (n=6) and healthy blood donors collected in 2019 (n=163). In addition, 17 seropositive blood samples donated from HCW were utilized for comparison of plasma, serum, heated and non-heated samples. Informed consent was obtained by all participants, and the study was approved by the regional ethical committee (EPN: dnr 2020-01653, dnr 2020-01620).

Two convalescent sera were purified using affinity chromatography. 1 ml HiTrap column coupled with ZCaTetraCys ligand (Cytiva) was used to enable mild antibody elution at almost neutral pH, with less risk of damage to acid-sensitive antibodies compared with the standard Protein A matrices requiring low pH for product elution [20] . Prior to affinity chromatography, both sera were heat-inactivated at 56 C for 30 min, filtered using 0.2 μm filters and spiked with 1 mM CaCl2 to promote interaction with ZCaTetraCys. The purification was performed as described by [18] Affinity chromatography was conducted on an ÄKTA Start instrument (Cytiva) at 25 C and a flow rate of 1 ml/min. The eluate was collected, and buffer exchanged to PBS (20 mM sodium phosphate, 150 mM NaCl, pH 7.5) for subsequent SPR analysis.

To confirm that the in vitro produced proteins (ACE2 and spike-f) interact with each other, an affinity measurement experiment was performed utilising SPR. Spike-f was immobilized via amine coupling on a CM5 chip (Cytiva) at a flow rate of 10 l/min to 750 RU. A J o u r n a l P r e -p r o o f multicycle kinetic analysis was performed on a T200 instrument (Cytiva) using PBS supplemented with 0.05 % (v/v) Tween20 (PBS-T) as running buffer. ACE2 was applied in three different concentrations (250 nM, 125 nM and 62.5 nM) and the purified antibodies were analysed in three concentrations (4 μM, 2 μM and 1 μM). ACE2 and the antibodies were allowed to associate with the spike-based proteins for 400 sec before dissociation was monitored over 600 sec at a flow rate of 30 l/min. The kinetic parameters were determined using the Biacore T200 Evaluation Software (Cytiva) and a 1:1 binding model. All runs were followed by regeneration using 10 mM HCl at 30 µl/ml for 30 sec.

The PNT was performed as a high throughput bead-based assay. ACE2 was immobilized on the surface of colour-coded magnetic beads (MagPlex, Luminex corp., Austin, TX, USA) as described in [19] . In brief, the protein was diluted in 100 mM 2-(N-morpholino) For the assays comparing heated and non-heated samples, the samples were heat-treated in a water bath for 30 min at 56 °C before dilution.

Microneutralization assay was performed as described earlier in [20] . Briefly, heatinactivated serum (56 °C for 30 min) was first tested at a 1:10 dilution. Samples with neutralizing capacity were further titrated using a 2-fold dilution series starting at a 1:20 dilution. All samples were prepared in duplicates. Each dilution was mixed with an equal volume of 200 50% tissue culture infectious dose (TCID50) SARS-CoV-2 (50 μl serum plus 50 μl virus (diluted to 4000 TCID50/mL SARS-CoV-2)) and incubated for 1 h at 37 °C and 5 % CO2. After incubation, the mixtures were added onto confluent Vero E6 cells. Four days later, the cells were inspected for signs of cytopathogenic effect by optical microscopy.

Results are shown as the arithmetic mean of the reciprocals of the highest 50 % neutralizing dilutions for each sample.

Antibodies towards the SARS-CoV-2 virus were analysed as previously described in [10] In brief, plasma and serum samples were diluted 1:50 and antibodies were detected by a beadbased assay using spike-f and a C-terminal fragment of the nucleocapsid protein (NC-C). The In the serology assay, the cut-off for seropositivity for spike-f was calculated as the mean + 6

x standard deviation (6 SD), rounded up, of 12 negative reference samples included in the analysis. The reference samples were carefully selected to represent a wide range of possible background signals as described in [10] .

Here we report the development and evaluation of a bead array-based PNT assay. In the assay, the diluted serum samples are first incubated with biotinylated spike-f, to allow for interaction with spike-f-specific antibodies in the serum sample (Figure 1) . When adding the ACE2-coated beads, remaining free biotinylated spike-f binds to the beads for detection by fluorescently labelled streptavidin. Hence, the more antibodies present in a sample binds the RBD region of spike-f and thereby block its interaction with ACE2, the lower the signal ( Figure 1 ).

To confirm the functionality of the recombinantly produced spike-f and ACE2 proteins, their interaction was evaluated by SPR analysis. First, the affinity between spike-f, immobilized on the SPR-surface and ACE2 as analyte was measured. The affinity was shown to be 1.3 nM (kon= 2.4*10 5 Ms -1 and koff = 3.1*10 -4 s -1 ) (Figure 2A) which is in agreement with prior data [21, 22] . Next, the interaction between spike-f and two polyclonal antibody samples purified from two different sera, one COVID-19 convalescent serum sample and one seronegative serum sample, was assessed. A distinct difference between the two samples was observed, indicating that antibodies from the convalescent serum effectively bind to the recombinantly produced spike-f while no interaction was observed in the seronegative serum sample ( Figure   2 B and C).

To evaluate the performance of the PNT assay, a set of 301 serum samples from different cohorts was assembled. The set comprized samples from HCW following mild COVID-19 disease (n=76), hospitalized COVID-19 patients following severe disease (n=56) and seronegative samples from non-infected HCW (n=6) as well as pre-pandemic control samples from 2019 (n=163). For validation of the PNT method, all samples collected during 2020

were analysed using a cell-based microneutralization assay. Microneutralization data for the seropositive HCW samples revealed neutralization titres ranging from 10 to 320 with a median titre of 60 (Supplementary Figure 2) . However, there were two exceptions within this group displaying titres as high as 2560. Sera from hospitalized COVID-19 patients overall displayed higher titres with a median of 200, albeit within a wide range from 60 to 3640. In Higher dilutions are needed to discriminate between highly neutralizing samples (Figure 4) .

Further, the signals obtained from the PNT assay were compared to the microneutralization titres. As can be seen in Figure 3 and Figure 4A , a 1:10 dilution was sufficient to discriminate between negative and positive samples whereas further dilutions increased the signal for samples with low neutralization capacity resulting in an overlap with the truly negative samples. Samples with higher neutralization capacity, including the hospitalized individuals, showed dilution-dependent dynamics while the distribution among the negative controls were largely unaffected by sample dilution. The linearity within the different neutralization spans was highly dependent on the sample dilution, giving the possibility to clearly discriminate between all samples, regardless of neutralization capacity. When comparing signal intensities with neutralization titres over a larger interval, the 1:50 dilution gives a good separation at microneutralization assay titres between 50 and ~200 and a dilution of 1:500 or more is required for neutralization titres of 200 and above. To control for potential signal variance caused by nonspecific binding of spike-f protein to the beads, beads without ACE2 were included in the PNT assay. As the measurements for all samples in all 3 dilutions were below 30 AU (Supplementary Figure 3) , it was concluded that the background signals did not influence the resulting data. Despite the variation in signal intensity between the dilutions, the technical intra-assay variation was similar, ranging from 2.4 % to 7.9 % with a mean of 4.8 %.

The full set of 301 serum samples was additionally analysed in a serological assay using parallel detection of IgG antibodies to different virus proteins: spike-f and the NC-C.

Classification of the serology data was performed according to [10] and the results are compared to the PNT and microneutralization data in Figure 3 . As can be seen, most of the samples classified as seropositive in the microneutralization assay four months post infection, were found to be seropositive also in the serological assay using a classification panel requiring antibodies against both the spike-f and NC-C [10] . Further, 12 of these samples were now shown to be negative against the NC-C, but positive against spike-f, which is in agreement with other studies showing that antibody signals against the nucleocapsid wane faster than those against spike-f [23] . Using the same classification panel, all negative controls but two were classified as seronegative, although, a few more samples were defined as reactive towards single antigens ( Figure 3) . Interestingly, the neutralization capacity does not entirely follow the serological status, either when the more stringent read out demanding antibodies against two SARS-CoV-2 antigens [17] is used to avoid false positives, or when relying on the read out from one antigen, spike-f or NC-C. However, all sera with neutralizing capacity were shown to also have antibodies binding to spike-f ( Figure 3 ). To investigate this further, signal intensities from the pseudoneutralization and the serological assay against spike-f were compared. Here, dilutions of 1:10 and 1:50 were used in the PNT assay, and 1:50 and 1:500 for the serological assay ( Figure 5) . As expected, a high concordance was found when comparing the relative antibody levels and their neutralizing capacity. However, as previously discussed, there were outliers in the COVID-19 negative group showing low but positive anti-spike IgG levels according to the serological assay, but no neutralizing capacity according to the PNT assay. This indicates that the PNT assay could deliver more reliable results regarding previous infection and antibody protection of individuals compared to serological analysis.

To assess if the assay read-out is dependent on the sample preparation, a set of serum and plasma samples from 17 seropositive HCW was used. To evaluate the effect of heattreatment (HT), an aliquot of each sample was heated for 30 min at 56 °C. The different sample types and pre-treatment conditions were analysed for PNT capacity using a 1:50 sample dilution. Plasma and serum samples displayed comparable results, ensuring the assay J o u r n a l P r e -p r o o f utility for both sample types (Figure 6A,B) . A high correlation between the heated and nonheated samples could also be confirmed, suggesting that the assay can be used for both sample preparations ( Figure 6C,D) .

During the SARS-CoV-2 pandemic, it is of particular importance to assess and map immunity by determining the presence of specific antibodies, both naturally acquired and obtained through vaccination. Such serological measurements are already efficiently performed by clinical laboratories around the world, but due to non-specific interactions within the antigens in the assays the analyses might report erroneously. The high throughput method presented here assesses the ability of the anti-spike antibodies to neutralize the spike protein's interaction with ACE2. For this purpose, our assay would be superior to the less specific serological methods that measure all antibodies against the spike protein, regardless of their capacity to inhibit this specific interaction. In comparison with the gold standard microneutralization method, this novel bead-based method delivers comparable data regarding neutralization capacity, without the need to cultivate live viruses and cells. Due to the high throughput setting, the method could replace the less specific serological analyses mentioned previously and would not only report on earlier infection or vaccination but also provide information on the efficiency of the antibodies. Moreover, there are recent reports regarding new variants of the SARS-CoV-2 virus that show differences within the surface responsible for interacting with ACE2 and thereby affect virus entry into the cell [24]. This creates an emerging need for high throughput methods that can assess these small, but important variations. The method presented here can easily be amended to determine the neutralization capacity of the antibodies towards different novel virus variants. Figure 1 . Schematic of the assay procedure. Serum with antibodies (orange) is preincubated with biotinylated spike-f (green) (I) followed by incubation with ACE2-coupled magnetic beads (blue) (II). Non inhibited spike-f binds to the beads (III). Fluorescently labelled streptavidin is added (yellow) to enable read out of bead-bound spike-f (IV). To discriminate between samples with diverse high signals, various sample dilutions were performed in A) 1:10 PNT and 1:50 serology; B) 1:50 PNT and 1:50 serology and C) 1:50 PNT and 1:500 serology. Negative serum samples collected during 2019 were included in the assay to assess the possible background in the PNT assay. Colours indicates status in the microneutralization assay; grey: negative; orange: positive. The dashed line in A) represents a cut-off between the samples classified as positive or negative in the PNT assay and is defined as 3x the SD of the mean of the pre-pandemic samples (n = 163) and the samples classified as negative in the microneutralization assay (n = 6). The dotted line in A) represents a cut-off for seropositivity for spike-f and was calculated as the mean + 6 x SD (6 SD) of 12 negative

PCR-based diagnostics for infectious diseases: uses, limitations, and future applications in acute-care settings

Reliability of RT-PCR tests to detect SARS-CoV-2: Risk analysis

A serological assay to detect SARS-CoV-2 seroconversion in humans

Antibody responses to SARS-CoV-2 in patients with COVID-19

Insights into SARS-CoV-2 genome, structure, evolution, pathogenesis and therapies: Structural genomics approach

Structure, Function, and Antigenicity of the SARS-CoV-2 Spike Glycoprotein

Pathogenesis of severe acute respiratory syndrome

SARS-CoV-2 Cell Entry Depends on ACE2 and TMPRSS2 and Is Blocked by a Clinically Proven Protease Inhibitor

Systematic evaluation of SARS-CoV-2 antigens enables a highly specific and sensitive multiplex serological COVID-19 assay

Cryo-EM structure of the 2019-nCoV spike in the prefusion conformation

Measuring immunity to SARS-CoV-2 infection: comparing assays and animal models

Establishment and validation of a pseudovirus neutralization assay for SARS-CoV-2

A SARS-CoV-2 surrogate virus neutralization test based on antibody-mediated blockage of ACE2-spike proteinprotein interaction

A high-throughput neutralizing antibody assay for COVID-19 diagnosis and vaccine evaluation

High throughput generation of a resource of the human secretome in mammalian cells

Improvements of a high-throughput protein purification process using a calcium-dependent setup

Highly selective Protein A resin allows for mild sodium chloridemediated elution of antibodies

Array-Based Profiling of Proteins and Autoantibody Repertoires in CSF

Expansion of SARS-CoV-2-Specific Antibody-Secreting Cells and Generation of Neutralizing Antibodies in Hospitalized COVID-19 Patients

Structural basis of receptor by SARS-CoV-2

SARS-CoV-2 D614G spike mutation increases entry efficiency with enhanced ACE2-binding affinity

crossm Changes in SARS-CoV-2 Spike versus Nucleoprotein Antibody Responses Impact the Estimates of Infections in Population