A Novel In-Cell ELISA Assay Allows Rapid and Automated Quantification of SARS-CoV-2 to Analyze Neutralizing Antibodies and Antiviral Compounds

The coronavirus disease 2019 (COVID-19) caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is currently the most pressing medical and socioeconomic challenge. Constituting important correlates of protection, the determination of virus-neutralizing antibodies (NAbs) is indispensable for convalescent plasma selection, vaccine candidate evaluation, and immunity certificates. In contrast to standard serological ELISAs, plaque reduction neutralization tests (PRNTs) are laborious, time-consuming, expensive, and restricted to specialized laboratories. To replace microscopic counting-based SARS-CoV-2 PRNTs by a novel assay exempt from genetically modified viruses, which are inapplicable in most diagnostics departments, we established a simple, rapid, and automated SARS-CoV-2 neutralization assay employing an in-cell ELISA (icELISA) approach. After optimization of various parameters such as virus-specific antibodies, cell lines, virus doses, and duration of infection, SARS-CoV-2-infected cells became amenable as direct antigen source for quantitative icELISA. Antiviral agents such as human sera containing NAbs or antiviral interferons dose dependently reduced the SARS-CoV-2-specific signal. Applying increased infectious doses, the icELISA-based neutralization test (icNT) was superior to PRNT in discriminating convalescent sera with high from those with intermediate neutralizing capacities. In addition, the icNT was found to be specific, discriminating between SARS-CoV-2-specific NAbs and those raised against other coronaviruses. Altogether, the SARS-CoV-2 icELISA test allows rapid (<48 h in total, read-out in seconds) and automated quantification of virus infection in cell culture to evaluate the efficacy of NAbs and antiviral drugs using reagents and equipment present in most routine diagnostics departments.


INTRODUCTION
By the time of writing, more than 29 million people experienced a laboratory confirmed infection with the severe acute respiratory syndrome coronavirus (SARS-CoV)-2 and more than 930,000 people died while having coronavirus infectious disease . First surveillance studies and calculations of excess mortality rates indicate that the precise number of infections and the true number of fatalities exceed above-mentioned numbers by far.
Coronaviruses (CoVs) are positive strand RNA viruses widespread among various vertebrate hosts including bats and rodents (1). Together with four seasonal human CoVs (HCoVs) and the two other emerging HCoVs SARS-CoV-1 and MERS-CoV, SARS-CoV-2 is the seventh HCoV causing widespread human diseases (2). In December 2019, SARS-CoV-2 was first recognized in the Hubei province in China (3) from where it rapidly spread throughout the world. In addition to its genetic similarity, SARS-CoV-2 not only shares some clinical characteristics with SARS-CoV-1 (4) but also exhibits some highly relevant particularities such as an increased spreading efficacy and the length of the course of disease (5). On January 31, the WHO declared the SARS-CoV-2 outbreak a Public Health Emergency of International Concern. On March 11, 2020, WHO started to denote the outbreak a global pandemic. The center of the pandemic shifts between regions, posing danger of repetitive local and temporal reintroduction circles. Thus, even countries that managed the first wave must prepare in terms of diagnostics capacities for potential future re-emergences.
Most SARS-CoV-2 infections lead to mild or moderate illnesses. However, a considerable fraction of cases proceeds to severe pneumonia or life-threatening acute respiratory distress syndrome. Elderly individuals and people with pre-existing comorbidities such as impaired immunity, chronic respiratory diseases, cardiovascular diseases, and cancer are more prone to suffer from severe COVID-19. The case fatality rate (CFR) is difficult to calculate in the midst of the pandemic. Depending on the age, CFR estimates of up to 18.4% for individuals older than 80 years and 1.38% (range 1.23-1.53) for the general population have been reported (6).
Given the extent, pace, and severity of the COVID-19 pandemic, diagnostics departments even in countries with highly developed medical systems struggle to provide sufficient and timely test capacities. Since nucleic acid-based pathogen detection has a very short window of opportunity, assays detecting long-lasting immune responses such as antibodies are required to monitor virus spread and to estimate potential herd immunity.
With some delay, most infected individuals raise a detectable humoral immune response including specific immunoglobulins (Ig) M, IgA, and IgG (7)(8)(9). Neutralizing antibodies (NAbs) bind and abrogate the function of viral proteins such as the SARS-CoV-2 Spike (S) protein that are essential for virus entry into host cells, e.g., through recognition of the cognate receptor ACE2 (10). Accordingly, monoclonal NAbs exhibit strong therapeutic and prophylactic efficacies in SARS-CoV-2-infected human ACE2-transgenic mice (11). A recent vaccination study conducted in non-human primates identified NAbs as correlate of protection (12), indicating that a human SARS-CoV-2 vaccine should also be capable to elicit potent NAb responses. Accordingly, first human vaccine studies focussed on the question if NAbs were induced (13). Additionally, monoclonal NAbs and NAb-containing hyper-immunoglobulin preparations may be applicable as treatment against COVID-19 (14). NAbs are also the backbone of convalescent plasma (CP) therapies (15)(16)(17) that are one of seven clinical recommendations of the IDSA (18). Based on the havoc COVID-19 causes to the global economy, immunity passports and vaccination certificates, documenting protection through NAbs, have been discussed [e.g., (19)]. Taken together, SARS-CoV-2-specific NAbs and their quantification appear to be of central importance for the medical and socio-economic management of the pandemic.
Different types of neutralization tests (NT) have been developed for SARS-CoV-2. However, to our knowledge, these assays rely on laborious microscopic counting of virus plaques or antibody-stained foci by trained personnel (20)(21)(22) or on genetically modified viruses such as transgenic SARS-CoV-2 mutants (23) or pseudo-typed viruses [e.g., vesicular stomatitis virus (VSV) or human immunodeficiency virus (HIV)] expressing the S protein of SARS-CoV-2 (24,25). Genetically modified viruses are generally prohibited in routine diagnostics laboratories and usually not applicable in less developed regions. Due to the central importance of NAbs and the limitations of the currently available methodology, we established a cheap, simple, fast, reliable, and automatable in-cell ELISA (icELISA)-based icNT applicable in routine diagnostics departments with access to BSL3 laboratories.

SARS-CoV-2 icELISA and icNT
For the analysis of neutralizing antibodies, serum samples were inactivated at 56°C for 30 min. A detailed icELISA/icNT protocol is provided in Supplementary File 1. Briefly, defined doses of SARS-CoV-2 were incubated with different serum dilutions for 1 h at 37°C prior to Vero E6 infection. At 16-24 h p. i., cells (~5 × 10 4 /well of a 96-well plate) were fixed with 4% (w/v) paraformaldehyde/PBS. Cells were permeabilized with 1% (v/v) Triton-X-100/PBS and blocked with 3% (v/v) FCS/PBS. The primary antibody was added and incubated for 2 h at room temperature or overnight at 4°C. Peroxidase-labelled secondary antibody was incubated for 1-2 h. Washing steps were performed with 0.05% (v/v) Tween-20/PBS. Tetramethylbenzidin (TMB) substrate was added to visualize the enzyme reaction. The reaction was stopped with 0.5 M HCl. Subsequently, the absorbance was measured using a microplate multireader (Mithras2 LB 943; Berthold Technologies). The a-N mAb1 (ABIN6952435), a-N mAb2 (ABIN6952433), a-E Ab (ABIN1031551), and POD-coupled secondary antibodies (Dianova) were used. In the case that an absolute quantification is necessary, the relative quantification of the icELISA can be transformed into an absolute quantification in terms of the virus dose using virus calibration curves (like in Figure 1) or using commercially available recombinant N preparations.

Immunoblot Analysis
Immunoblotting was performed as described previously (26) using antibodies recognizing the SARS-CoV-2 N protein (ABIN6952435) or GAPDH (sc-25778, Santa Cruz). Lysates were inactivated by sequential heat incubation (10 min at 70°C and 10 min at 95°C) before they were discharged from the BSL3 laboratory. Proteins were visualized using peroxidase-coupled secondary antibodies (Dianova) and the ECL chemiluminescence system (Cell Signaling Technology).

Relative Quantification of SARS-CoV-2 Replication and its Inhibition by Antiviral Compounds Using icELISA
We hypothesized that virus-encoded proteins expressed by infected cells should be amenable as source of viral antigens for the detection and quantification by ELISA. We optimized the experimental conditions such as cell type, virus dose, infection period, cell fixation method, blocking reagent, amd type and concentrations of primary and secondary antibodies ( Figure 1 and data not shown). As described in the Materials and Methods section and provided as detailed laboratory protocol in the supplementary information (Supplementary File 1), we compared different commercially available antibodies for the icELISA-based quantification of SARS-CoV-2 proteins. Based on existing data on virus entry (10), we infected human Caco-2 cells with graded virus doses (ranging from 0.03125 to 2 PFU/cell). At 3 days post infection (d p.i.), we fixed and stained the cells with different antibodies either recognizing the E or the N protein of SARS-CoV-2. In accordance with high expression level of the N protein (27,28), certain N-specific mouse IgGs exhibited a signalto-noise ratio favourable for icELISA ( Figure 1A). A comparison of Vero E6 and human Caco-2 cells revealed that the icELISA is applicable to different cells ( Figure 1B).
Since the icELISA signal directly correlated with viral replication and viral antigen expression, we tested its ability to determine antiviral effects. Human Caco-2 cells were treated with graded concentrations of human interferon (IFN) a2 or IFNb and infected 3 h later with SARS-CoV-2. At 3 d p.i., viral antigen amounts were quantified by icELISA. In accordance with a recent clinical phase 2 trial (29), IFNb exhibited strong and dosedependent antiviral activity against SARS-CoV-2 in human cells ( Figure 1C), indicating that the icELISA is applicable for future experiments addressing the efficacy of potential antiviral drugs in human cells.
Despite different start MOIs, similar icELISA signals were observed at 72 h p.i. in Vero E6 ( Figure 1B), indicating multiple rounds of virus amplification and extraordinary fast replication kinetics of SARS-CoV-2 in Vero E6 cells, consistent with previous studies (30). To test if shorter infection periods might result in virus dose-dependent signals, we analyzed infected Vero E6 cells after 6, 15, and 22 h. SARS-CoV-2 was readily detectable in Vero E6 cells by icELISA already at 6 h p.i. (Figures 1D-F). To evaluate if the icELISA faithfully reports on the amounts of SARS-CoV-2-expressed antigens, we infected cells with graded infectious doses and performed immunoblots and icELISAs in parallel. As expected, icELISA and immunoblot signals correlated very well ( Figure 2A). Since we observed icELISA signals already at 6 h p.i. (Figures 1A, D) and the N protein is an abundant component of SARS-CoV-2 particles (31)(32)(33), the N protein derived from input viruses could have contributed to icELISA signals. To evaluate this, we infected cells in the presence and absence of the translation inhibitor cycloheximide ('CHX'). Consistent with the notion that the icELISA signal is dominated by de novo N protein expression, the signal was significantly diminished by CHX ( Figure 2B).
Taken together, these data indicate that the icELISA allowed rapid identification and relative quantification of SARS-CoV-2 replication in Vero E6 and human cells and its inhibition by antiviral compounds.

Neutralization Tests Based On icELISA Allow the Quantification of SARS-CoV-2 NAbs as Early as 6 Hours Post Infection
Since the icELISA allowed simple and automated quantification of SARS-CoV-2-dependent antigen expression, we tested if an icELISA-based neutralization test (icNT) can be established. We infected Vero E6 cells for 6, 15, and 24 h with graded SARS-CoV-2 infection doses (500, 5,000, or 50,000 PFU/well) in the absence or presence of two convalescent sera in three different concentrations (1/20, 1/40, and 1/80 dilution). Using the high infectious dose, viral antigens became dose dependently detectable by icELISA as early as 6 h p.i. ( Figure 3A). Both infection. Even the strong icELISA signal at 24 h p.i. was dose dependently neutralized by both sera ( Figure 3F). Please note that the neutralizing capacity of given sera dilutions was less pronounced at higher virus doses as compared to lower virus doses, as indicated by residual icELISA signals. Taken together, the icELISA resulted in a time-and virus dose-dependent signal constituting a surrogate for SARS-CoV-2 infection and replication. The fact that the infection and the resulting icELISA signal were neutralized by NAbs present in immune sera indicated that the fast and automated icELISA format is applicable for icNTs.

The icNT Results Correlate With the Standard SARS-Cov-2 Neutralization Test Conducted by Staining of Virus Foci and Microscopic Counting
Although most SARS-CoV-2 NTs have not been formally validated and certified, classic plaque reduction neutralization tests (PRNT) are currently considered to represent the gold standard for the detection of SARS-CoV-2-specific NAbs. Various commercially available IgM, IgA, and IgG ELISAs have been compared to PRNTs [e.g., (30)]. To validate the novel icNT, 53 sera-24 positive for SARS-CoV-2-specific IgG as determined using the Euroimmun ELISA (ELISA ratio: 1.13-9.92) and 29 ELISA-negative sera (ELISA ratio < 0.9)-were compared side-by-side by icNT and standard PRNT using 200 PFU/well ( Figure 4A). One set was processed by icNT, while for the other set virus foci were stained using an antibodybased AEC staining method and manually counted by microscopy.

The icNT Discriminates SARS-CoV-2-Neutralizing Sera From Non-Immune Sera and Provides Superior Resolution When Increased Virus Doses are Used
Standard SARS-CoV-2 NTs base on microscopic counting of plaques or antibody-stained virus foci. To enable plaque/foci recognition and individual counting by trained personnel, a countable number of PFU must be applied in PRNTs. Depending on the PRNT protocol, 100 (20) to 400 PFU (21) are used to infect each well. Based on previous experiences with virus neutralization experiments (34,35), we suspected that lower infectious doses might be more susceptible to NAbs than higher virus doses, e.g., through altered ratios of NAbs and antigenic regions determining neutralization, such as the receptor binding domain (RBD) of the S protein of SARS-CoV-2. Accordingly, graded infectious doses (1,250-10,000 PFU per well) showed virus dose-dependent susceptibilities to the same serum sample (Supplementary Figure S1). Before we assessed clinical specimens, we compared icNT signals upon a high MOI infection with viral progeny as determined by parallel TCID50 titration. As expected, we observed that icNT signals and residual virus numbers correlated very well (Supplementary Figure S2).  Figures 5A-C), the neutralizing capacity of the same sera considerably differed at the more restrictive high virus dose ( Figures 5D-F  infections of cell cultures with well-defined virus preparations in the presence or absence of compounds that may impair the infection. Thus, the NT specificity results from the choice of virus applied to the cells. However, antiviral compounds can either be specific for one particular virus species (e.g., most monoclonal neutralizing antibodies), broadly active against virus families (e.g., receptor blocking agents) or even bigger genera (e.g., hyperimmunoglobulin preparations). In addition to SARS-CoV-2, other HCoVs such as the alphacoronaviruses HCoV-229E and HCoV-NL63 and the betacoronaviruses HCoV-OC43 and HCoV-HKU1 circulate autochthonously in the human population, resulting in high sero-prevalence rates (36). To address if the experimental SARS-CoV-2 infection in the icNT might be diminished by cross-reactive antibodies, we tested sera of individuals who have had infections with the alphacoronaviruses HCoV-229E or HCoV-NL63 or betacoronaviruses such as HCoV-HKU1 ( Figure 6A) or a combination of HCoV-229E and HCoV-OC43 ( Figure 6B). In contrast to the positive control containing SARS-CoV-2-specific NAbs, all tested serum samples recognizing other coronaviruses did not neutralize SARS-CoV-2. Although the limited sample size prevents definitive conclusions regarding crossneutralizing capacities of antibodies raised by other HCoVs, in conjunction with consistent publications (37,38) it indicates the specificity of the icNT assay.

DISCUSSION
We established a novel icELISA-based test principle for detection and relative quantification of SARS-CoV-2. Given the excellent signal-to-noise ratio between infected and uninfected cells, the test was applicable to quantify the efficacy of antiviral compounds, here shown for IFNb, and SARS-CoV-2-specific NAbs present in immune sera. Compared to icELISA and icNT, standard virus titrations and PRNTs are far more laborious, time consuming, and expensivenot to speak from subjective and expectation biases upon usage for research. The entire icNT can be processed in less than 2 days including the infection period. Given that the protocol includes an early fixation step using 4% (w/v) paraformaldehyde which inactivates SARS-CoV-2, all subsequent steps can be processed outside the biosafety level 3 laboratory. The actual data acquisition is conducted within seconds, using standard ELISA plate readers present in most routine diagnostics departments. We believe that this is advantageous compared to other NT assays that rely on more sophisticated and more expensive devises, e.g., for imaging cytometry (39). The icELISA and icNT provide increased data quality and precision by generating continuous data sets. Since the detection antibodies can be applied in icELISA and icNTs in relatively high dilutions (1/5,000 to 1/10,000 and 1/2,000 of primary and secondary antibody, respectively), the assay is relatively cheap (in our case, around 0.10 € per well for both antibodies and the TMB peroxidase ELISA substrate). The specificity of the icELISA and icNT is provided by the defined SARS-CoV-2 added to the cell cultures on purpose. The primary and secondary detection antibodies just serve to visualize and quantify viral antigens. Thus, SARS-CoV-2-specific antibodies can be applied for icELISA detection notwithstanding potential cross-recognition of other CoVs such as HCoV-HKU1 or HCoV-OC43 -simply because these viruses are not present in the culture. Obviously, such antibodies recognizing conserved residues cannot be used for classic antigen-recognizing ELISAs due to their inability to discriminate coronaviruses. More virus (>1PFU/cell) can be applied to icNTs. Such high PFU icNTs scrutinize the virusneutralizing capacity of sera more strictly, enabling a higher resolution compared to PRNT assays that all rely on low virus numbers. It is tempting to speculate that sera exhibiting superior neutralization in high PFU icNT might be more beneficial in CP therapies.
NTs based on pseudo-typed viruses using heterologous expression of the SARS-CoV-2-encoded S by non-related viruses may have certain advantages. However, genetically modified viruses are inapplicable by law in various routine diagnostics departments and usually unavailable in less developed countries. Recently, an elegant surrogate assay has been designed that determines if antibodies are capable to prevent the interaction of immobilized hACE2 with a reporter enzyme-coupled receptor binding domain (RBD) of S (40). This assay showed a good correlation with conventional NTs (r 2 = 0.8591; p value < 0.0001) and NTs based on pseudo-particles (r 2 = 0.8374). However, without the use of genuine infectious viruses, assays relying on pseudo-typed viruses do not fully interrogate the full spectrum of antiviral effects for example if other viral proteins influence the system, e.g., by complement activation (41). Accordingly, the icNT showed a superior correlation compared to PRNT (r 2 = 0.987; p value 5.81E-50).
Taken together, we propose the icELISA and icNT for the quantification of SARS-CoV-2 replication and its inhibition by NAbs and antiviral compounds. By changing the detection antibody and, if necessary, the cells according to the viral infection system, the test principle is transferable to all other viruses.

DATA AVAILABILITY STATEMENT
All datasets presented in this study are included in the article/ Supplementary Material.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Ethics committee of the medical faculty of the University of Duisburg-Essen (20-9208-BO, 20-9511-BO, and 20-9512-BO). The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
LS, VTKL-T, ME, and DM performed research. OA, AK, and AH provided essential reagents. LS, VTKL-T, and MT analyzed the data. LS, VTKL-T, UD, and MT interpreted the data. VTKL-T and MT supervised the project. LS, VTKL-T, and MT wrote the manuscript. All authors contributed to the article and approved the submitted version.

FUNDING
The authors received support by the Stiftung Universitätsmedizin Essen, Kulturstiftung Essen, the Else-Kröner Promotionskolleg ELAN, and the Deutsche Forschungsgemeinschaft (DFG) through grants RTG 1949/2, TR1208/1-1, and TR1208/2-1. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu.2020.573526/ full#supplementary-material SUPPLEMENTARY FIGURE 1 | The infectious virus dose strongly influences the neutralization capacity of serum samples. Graded amounts of SARS-CoV-2 were incubated with indicated dilutions of a serum sample (chosen to exhibit a low neutralizing capacity) for 1 h before Vero E6 cells were infected. Neutralization was evaluated by icELISA.

SUPPLEMENTARY FIGURE 2 |
The results of the high MOI icNT correlate with residual virus after neutralization. SARS-CoV-2 was incubated with indicated dilutions of serum samples for 1 h before Vero E6 cells were infected. (A) Neutralizing capacity was evaluated by icELISA. (B) Neutralizing capacity was evaluated by back titration of residual virus by TCID50 in 2-fold serial dilutions.