High Heterogeneity of Virus-Neutralizing and RBD-Binding Activities of COVID-19 Convalescent Sera

Abstract

The parameters of the humoral response are an important immunological characteristic of donors who recovered from COVID-19 and vaccinated individuals. Analysis of the level of virus-binding antibodies has become widespread. The most accurate predictor of effective immune protection against symptomatic SARS-CoV-2 infection is the activity of virus-neutralizing antibodies. We determined virus-neutralizing activities in plasma samples of individuals (n = 111) who had COVID-19 from April to September 2020. Three independent methods were used: conventional with live virus, with virus-like particles pseudotyped with spike protein, and a surrogate virus-neutralization test (cVNT, pVNT and sVNT, respectively). For comparison, the levels of IgG, IgA and IgM antibodies against the receptor-binding domain of the SARS-CoV-2 spike protein were also evaluated. The levels of virus-binding as well as virus-neutralizing antibodies in cVNT and pVNT showed high heterogeneity. A comparison of cVNT and pVNT results showed a high correlation, sVNT results also correlated well with both cVNT and pVNT. To the greatest extent, the level of IgG antibodies correlated with the results of cVNT, pVNT and sVNT. These results can be used in the selection of plasmas that are best suited for transfusion and treatment of acute COVID-19. In addition, data on the virus-neutralizing activity of plasma are important for the selection of potential donors, for the isolation of SARS-CoV-2-specific B-lymphocytes, in order to further generate monoclonal virus-neutralizing antibodies.

In the context of the current COVID-19 pandemic, serological analysis of biological samples has great importance. The humoral immune response is an important characteristic for both naturally recovered and vaccinated individuals [1]. Monitoring antibody levels against SARS-CoV-2 is essential to predict the risk of reinfection and assess the effectiveness of vaccination. In addition, these parameters are used for retrospective contact tracing of sick patients, estimates of the number of asymptomatic carriers, and for estimating the level of herd immunity.

In serological assays for SARS-CoV-2, two viral proteins are commonly used as antigens: the nucleoprotein (N) and the spike protein (S). SARS-CoV-2 binds to its receptor on the host cell through the receptor-binding domain (RBD) of the S-protein, therefore RBD is considered the main target for neutralizing antibodies [2]. Thus, the level of S- and RBD-binding antibodies may indirectly indicate the neutralizing activity of the serum. A high level of virus-neutralizing antibodies is a good predictor of effective immune protection against the symptomatic course of COVID-19 [3, 4], so the determination of virus-binding antibodies should be supplemented by direct measurement of the level of virus-neutralizing antibodies. Note that the analysis for virus-binding antibodies is well developed and widely used in laboratory practice, while the determination of virus-neutralizing antibodies is available only to specialized laboratories.

In the early stages of the SARS-CoV-2 pandemic, when there was an acute lack of reliable therapeutic agents for the treatment of COVID-19, plasma transfusion of convalescents was the only possible etiotropic therapy. Transfusion of plasma from donors who have recovered from COVID-19 has the potential to provide immediate passive immunization and protection against SARS-CoV-2 infection. However, the effectiveness of plasma therapy was ambiguous. The most successful method is the use of convalescent plasma in the early stages of the disease ‒ shortly after hospitalization. The use of plasma with a high titer of anti-S protein or RBD antibodies is also a prime factor. Transfusion of convalescent plasma provided a more than twofold increase in patient survival compared to control groups with properly selected parameters [5]. All this indicates that the plasma intended for transfusion must be strictly controlled for virus-neutralizing activity.

There are several methods for determining the virus-neutralizing activity of sera, each of which is divided into several sub-variants. These are, firstly, a method using a native live SARS-CoV-2 virus—a conventional virus neutralization test (cVNT), secondly, using virus-like particles pseudotyped with an S protein (pVNT) [6], and finally surrogate VNT (sVNT) [7].

The aim of the study was to compare the virus-neutralizing activity of convalescent plasma samples determined by three different methods: cVNT, pVNT and sVNT. An attempt was made to assess the variability in the virus-neutralizing activity of plasma samples and compare it with the virus-binding activity of the same samples.

EXPERIMENTAL

Characteristics of patients. The study included volunteers (n = 111) who suffered from COVID-19 between April and September 2020. Most of the patients (86%) received inpatient treatment at the FRCC FMBA of Russia, some patients—in other medical institutions in Moscow or at home. All participants were diagnosed with COVID-19 based on PCR testing and/or clinical findings. 1–3 months after recovery, volunteers donated blood plasma to the FRCC of the FMBA of Russia for subsequent transfusion to sick patients. Plasma harvesting was performed by plasmapheresis, the volume of plasma harvested was 640 mL for each donor. In our study, 100 microliter aliquots of plasma were used.

Patients who recovered from COVID-19 ranged in age from 18 to 52 years (median 38 years, interquartile range (IQR) 21 to 43 years); 68 of them were men (61.2%) and 43 women (38.8%). The comparison group included healthy donors (n = 25). The comparison group in terms of age and the ratio of men and women corresponded to the COVID-19 recovered group. Venous blood from healthy volunteers was collected into heparinized tubes (Sarstedt, Germany), then the tubes were centrifuged for 10 min at 300 g and plasma was collected. Plasma samples (100 µL aliquots) were frozen and stored at –70°C.

ELISA quantification of anti-RBD antibodies. The levels of anti-RBD antibodies in plasma was determined by ELISA using commercial plates coated with RBD (#K153G, Xema Medica, Russia) according to the manufacturer’s instructions. Antibodies against human IgG, IgA, and IgM conjugated with horseradish peroxidase (Cat. nos. 109-036-088, 109-035-129, 109-035-011, Jackson Immuno Research, USA) were used as detection reagents. 100 μL of the peroxidase substrate, 3,3',5,5'-tetramethyl benzidine (TMB), was added to each well and incubated until a blue color appeared. The reaction was stopped by adding 100 μL of stop reagent. The optical density of the solution was measured using a BioRad iMark Microplate Absorbance Reader photometer (Bio-Rad, United States).

All samples were titrated to find the linear region on the plot of absorbance (OD450) versus plasma dilution. A point in the middle of the linear section of the titration curve was used to determine the level of antibodies. The concentration of anti-RBD IgG antibodies was determined in absolute units (ng/mL), using the human monoclonal antibody 34B12 with a known concentration as a standard (kindly provided by Doctor of Biological Sciences A.V. Taranin, Institute of Molecular and Cell Biology SB RAS, Novosibirsk [8]). The levell of anti-RBD IgA and IgM antibodies were determined in relative units, using the plasma of one of the donors as a standard.

cVNT (conventional virus-neutralizing test). The analysis was performed on Vero E6 cells that were infected with authentic SARS-CoV-2 (strain hCoV-19/Russia/Moscow-PMVL-12/2020) at a dose of 100 TCID₅₀ (50% tissue cytopathic infectious dose). The cytopathic effect was assessed visually after 96 h using an inverted microscope. The virus-neutralizing activity of plasma samples was determined by the suppression of the cytopathic effect of the virus. The cytopathic effect of the virus in the absence of plasma was taken as 100%. Plasma was titrated starting from a dilution of 1 : 10. The highest dilution of plasma, at which no cytopathic effect on 100 TCID₅₀ of the virus had yet developed, was taken as the antibody titer. Experiments on neutralization were accompanied by control of the working dose of the virus, the toxicity and the sterility of sera.

pVNT (pseudovirus-based virus-neutralizing test). This assay was performed on HEK-293 cells stably transfected with a plasmid expressing angiotensin converting enzyme-2 (ACE2). Lentiviral virus-like particles (VLPs) pseudotyped with SARS-CoV-2 S‑protein were added to target cells. The VLPs carried the green fluorescent protein (GFP) reference gene, which was expressed in infected cells. The dose of VLPs was selected so that in the absence of antibodies, 50% of target cells were infected—they were detected by GFP fluorescence using a CytoFLEX S flow cytometer (Beckman Coulter, USA). When the infection was completely inhibited, the target cells did not give a fluorescent signal. For each serum, five 5-fold serial dilutions were made, starting at 1 : 4. The results were presented as the serum dilution at which 50% inhibition (ID₅₀) of cell infection was observed, calculated from the Sigmoidal titration plot, 5PL, in the GraphPad Prism program (USA).

sVNT (surrogate virus-neutralizing test). This assay was performed using the SARS-CoV-2 VNAFA K533 kit (#2104, Xema Medica) according to the manufacturer’s recommendations. Plasma samples were diluted 10-fold with sample buffer and a solution of RBD conjugate with horseradish peroxidase (RBD-HRP) was added to them in a ratio of 1 : 10 (v/v). The mixture was incubated for 30 min at 37°C, after which 100 μL of the reaction mixture was transferred to an ACE2-coated plate and incubated for another 30 min at 37°C. The wells were washed 5 times with wash buffer. The reaction was shown and recorded in the same way as described above when performing ELISA. Data were analyzed using Zemfira 4.0 software.

Statistical analysis. The data were processed using the GraphPad Prism software version 8.4.3 (GraphPad, USA). The significance of differences between samples was assessed using the Mann–Whitney test. Differences in the compared parameters were considered statistically significant at p < 0.05. The graphs show medians (middle line), third, and first quartiles (rectangles), whiskers show a 1.5-fold interquartile range. A quantitative assessment of the statistical relationship between the parameters was carried out by calculating the coefficients of linear regression, as well as Spearman’s rank correlation (r).

RESULTS

The study included patients who had COVID-19 between April and September 2020. At that time, the Wuhan strain of SARS-CoV-2 or its variant with the D614G substitution in the S protein circulated in Russia [9–11].

Level of RBD-Binding Antibodies

First of all, the level of RBD-binding antibodies was determined in the plasma of patients who recovered from COVID-19. The results of IgG quantitation and semiquantitative IgA and IgM levels of anti-RBD antibodies are shown on Fig. 1. The results for healthy donors are represented as a negative control.

Fig. 1.

Distribution of IgG (a), IgA (b), and IgM (c) antibodies to SARS-CoV-2 spike protein RBD in patients who recovered from COVID-19. Box plots on the left and violin plots on the right. Each dot represents a plasma sample. Gray dots indicate samples obtained from patients who recovered from COVID-19; white dots are from healthy volunteers. The horizontal dotted lines represent the cutoff threshold.

For IgG antibodies in recovered patients, the median was 1819.7 ng/mL (IQR 635.3–4753.4), while for healthy donors it was 16.9 ng/mL (IQR 2.0–27.4) (p < 0.0001). For IgA antibodies, these values were 118.0 rel. units (IQR 52.72–208.9) and 13.5 rel. units (IQR 3.9–18.3), respectively (p < 0.0001); for IgM antibodies, 251.8 rel. units (IQR 103.5–669.9) and 10.3 rel. units (IQR 1.44–20.41), respectively (p < 0.0001).

The analysis of the distribution of the level of antibodies in a cohort of patients showed high heterogeneity of this parameter. The content of RBD-binding antibodies varied quite widely, within 5, 3, and 4 orders of magnitude, respectively, for IgG, IgA, and IgM antibodies. At the same time, the distribution of RBD-specific IgG and IgM antibodies differed markedly from a normal distribution, which was confirmed by analysis by the D’Agostino-Pearson method (p < 0.0001). We hypothesized that the reason for the deviation from the normal distribution is the presence of several null values of RBD-specific IgG and IgM antibodies. When zero values were excluded from the samples, the distribution of values for IgG and IgM antibodies approached a normal distribution (p = 0.8170 and p = 0.0504, respectively). For IgA antibodies, the initial distribution was close to normal (p = 0.2911).

Level of Virus-Neutralizing Antibodies

The virus-neutralizing activity of plasma samples was determined using three independent methods: cVNT, pVNT and sVNT. Most of the participants had neutralizing antibodies: 101/111 (91%), 91/111 (82%) and 88/111 (79%) patients, respectively.

The levels of virus-neutralizing antibodies determined by the cVNT and pVNT methods were heterogeneous (Fig. 2), while the sVNT results varied within two orders of magnitude. This is understandable, since sVNT results are represented as percentage of virus neutralization, which, by definition, does not exceed 100%. The actual distribution of sVNT results could be wider, which was not observed, since plasma samples with high titers fell into the saturation region.

Fig. 2.

Distribution of levels of virus-neutralizing antibodies in patients who recovered from COVID-19. The results of the cVNT (a), pVNT (b), and sVNT (c) methods are presented. The horizontal dotted lines represent the cutoff threshold.

The distribution of virus-neutralizing activity of plasma samples, as well as the levels of RBD-binding antibodies, deviated from a normal distribution. The presence of points with zero neutralizing activity was one of the characteristics of this distribution. It can be assumed that the zero points appeared as a result of the erroneous inclusion in the study group of individuals who did not have COVID-19. In this case, the plasma samples taken would give zero values at once in several tests, but this was not observed.The appearance of zero points was probably due to the different sensitivity and/or specificity of the test systems used.

Correlation between the Levels of RBD-Binding Antibodies and Virus-Neutralizing Activity

We conducted a correlation analysis in order to assess the relationship between the results obtained by different methods. First of all, we evaluated the relationship between data on the binding of antibodies to RBD and virus neutralizing activity (Fig. 3a). The level of IgG antibodies correlated to the greatest extent with the results of cVNT, pVNT, and sVNT (r = 0.677, p < 0.0001; r = 0.624, p < 0.0001; r = 0.615, p < 0.0001, respectively). A slightly lower correlation was found between cVNT, pVNT, sVNT and the level of IgA antibodies (r = 0.63, p < 0.0001; r = 0.581, p < 0.0001; r = 0.444, p < 0.0001, respectively). Finally, the IgM level weakly correlated with sVNT (r = 0.367, p = 0.001) with good correlation with cVNT and pVNT (r = 0.671, p < 0.0001; r = 0.716, p < 0.0001, respectively).

Fig. 3.

Correlation analysis of the activity of convalescent plasma samples. (a) Comparison of the results of virus neutralization tests: cVNT, pVNT, and sVNT with data on the determination of IgG, IgA, and IgM antibodies against RBD. (b) Comparison of the results of tests for virus neutralization: cVNT, pVNT and sVNT—with each other. (c) Heat map of correlation coefficients according to Spearman. Straight lines represent linear regression trend lines. On the subpanels, the numbers indicate the coefficients of linear regression, as well as the rank correlation according to Spearman (r). Responses with low values when presented on logarithmic scales were taken equal to 1.

Comparison of the results obtained by different modifications of the virus-neutralization tests is shown in Fig. 3b. As expected, the highest correlation of results was found for cVNT and pVNT (r = 0.841, p < 0.0001), and it was somewhat weaker for cVNT and sVNT (r = 0.643, p < 0.0001) and pVNT and sVNT (r = 0.665, p < 0.001).

DISCUSSION

The most adequate approach to determine the virus-neutralizing activity of plasma is using live SARS-CoV-2 virus. The cVNT method is one of the most accurate and informative approaches because it reveals the effect of antibodies that neutralize the virus by various mechanisms [12]. The cVNT method is based on the use of a replication-competent virus, and therefore it is carried out in laboratories with a biosafety level of at least BSL3 (according to the International Biosafety Level Classification), and this greatly limits its use. The disadvantages of cVNT also include subjectivity in assessing the cytopathic effect of target cells and the discrete nature of the received data. While the first drawback is leveled by many years of experience and qualifications of the researcher, the lack of continuity of the results obtained complicates subsequent statistical analysis.

Virus-like particles pseudotyped with the SARS-CoV-2 S-protein mimic the initial stages of the virus life cycle: its binding and fusion with the target cell [6]. The pVNT method is much easier to perform than the cVNT method and allows work in laboratories with a BSL2 biosafety level. As a result of these circumstances, pVNT has become widespread in research laboratories.

It should be noted that both cVNT and pVNT require living cells; in addition, an important factor is the qualification of the performer. Due to these limitations, the above methods are not widely used in clinical diagnostic practice. As an additional method, we used the so-called surrogate virus neutralization test, sVNT [7]. It can be classified as a rapid test, as it allows assessment of the virus-neutralizing effect of plasma samples in just 1–2 hours, while cVNT and pVNT take several days. The disadvantages of sVNT include limited specificity: it can detect only antibodies that work exclusively by the mechanism of blocking the binding of RBD with ACE2 [7].

It is interesting to note that a small proportion of convalescents had practically no virus-neutralizing antibodies. This observation is consistent with previously published data that 2% of patients who recovered from COVID-19 do not produce neutralizing antibodies [13]. Notably, the largest number of such patients (n = 23) was detected using sVNT, a test that detects only RBD-neutralizing antibodies. It can be assumed that some plasma samples from this group of recovered patients contained neutralizing antibodies that bind to the S protein outside the RBD region [14]. Indeed, it has been previously shown that antibodies to the N-terminal domain or the S2 region of the S-protein can also be involved in the neutralization of the virus [15].

Despite the noted differences revealed by us, as well as in some other studies, different methods of virus neutralization correlate well with each other [16–18]. The highest correlation coefficient was found between the values obtained in the cVNT and pVNT tests (r = 0.841). This means that the cVNT test can in most cases be replaced by the pVNT test when assessing the SARS-CoV-2 neutralizing activity of antibodies in donor plasma. With certain reservations, all three tests can be used to evaluate the neutralizing activity of plasma samples. Comparison of the results on virus-neutralizing activity and the content of RBD-specific antibodies makes it possible to evaluate the contribution of different isotypes to the neutralizing ability of a particular plasma sample. For each neutralization test used, the highest correlation coefficients were obtained for IgG antibodies. This indicates the predominant role of IgG antibodies, over IgM and IgA, in blocking the interaction of the virus RBD with its receptor, ACE2. It is interesting to compare these results with published data for other groups of convalescents, as well as for emerging variants of SARS-CoV-2.

Note that recently our compatriots, D. Kolesov et al. [14], published work similar to that presented here. Their study was also performed on the Moscow population and with the participation of patients who had been infected with Wuhan or similar variants of SARS-CoV-2 [14]. However, in our study, we obtained lower correlation coefficients between the results of cVNT, sVNT, and the anti-RBD test. Perhaps this is due to the different sVNT formats. It can also be noted that our correlation coefficient between sVNT and anti-RBD test is comparable to that for vaccinated individuals [17].

On a sufficiently large sample, we found a high heterogeneity of convalescent plasmas. Samples with high virus-neutralizing activity attract the most attention - titers exceeded 1000 in cVNT, and ID₅₀ values were above 1000 in pVNT. These plasma samples are among the most promising for subsequent transfusion in patients in the acute phase of COVID-19. The percent of such samples is small (no more than 5%). Plasma with a moderate amount of virus-neutralizing antibodies (titers in the range from 90 to 300 in cVNT and ID₅₀ values in the range from 100 to 1000 in pVNT) prevailed in the sample. Previously, it was suggested that heterogeneity in the virus-neutralizing activity of plasma/serum in patients with COVID-19 may be associated with severity of disease [19, 20], as well as with different timing of sampling. The variability in the virus-neutralizing activity of plasma samples was most clearly manifested when comparing different mutant variants of SARS-CoV-2 [16]. The low virus-neutralizing activity of blood plasma/serum in some patients with COVID-19 may be one of the reasons for re-infection [21].

Thus, a comparison of tests to assess the virus-neutralizing activity of plasma samples of COVID-19 convalescents allows us not only to answer the question of the mutual substitution of one or another test, but also to identify samples that contain antibodies against SARS-CoV-2 that work through different neutralization mechanisms. It seems important to use such test systems for the preliminary selection of donors for sorting single SARS-CoV-2-specific B-lymphocytes in order to generate human monoclonal virus-neutralizing antibodies.

FUNDING

This work was financially supported by the Ministry of Science and Higher Education of the Russian Federation (Agreement no. 075-15-2021-1086, contract no. RF-193021X0015, 15.IP.21.0015).

COMPLIANCE WITH ETHICAL STANDARDS

Conflict of interests. The authors declare that they have no conflicts of interest.

Statement of compliance with standards of research involving humans as subjects. All research involving humans comply with the ethical standards of the institutional and/or national research ethics committee and the 1964 Declaration of Helsinki and its subsequent amendments or comparable ethical standards. Informed consent was obtained from each of the participants included in the study. The study protocol was approved by the local ethics committee of the FRCC FMBA of Russia (protocol #4-2020 April 28, 2020).