Different Neutralization Sensitivity of SARS-CoV-2 Cell-to-Cell and Cell-Free Modes of Infection to Convalescent Sera

The COVID-19 pandemic caused by SARS-CoV-2 has posed a global threat to human lives and economics. One of the best ways to determine protection against the infection is to quantify the neutralizing activity of serum antibodies. Multiple assays have been developed to validate SARS-CoV-2 neutralization; most of them utilized lentiviral or vesicular stomatitis virus-based particles pseudotyped with the spike (S) protein, making them safe and acceptable to work with in many labs. However, these systems are only capable of measuring infection with purified particles. This study has developed a pseudoviral assay with replication-dependent reporter vectors that can accurately quantify the level of infection directly from the virus producing cell to the permissive target cell. Comparative analysis of cell-free and cell-to-cell infection revealed that the neutralizing activity of convalescent sera was more than tenfold lower in cell cocultures than in the cell-free mode of infection. As the pseudoviral system could not properly model the mechanisms of SARS-CoV-2 transmission, similar experiments were performed with replication-competent coronavirus, which detected nearly complete SARS-CoV-2 cell-to-cell infection resistance to neutralization by convalescent sera. These findings suggest that the cell-to-cell mode of SARS-CoV-2 transmission, for which the mechanisms are largely unknown, could be of great importance for treatment and prevention of COVID-19.


Introduction
SARS-CoV-2 is a respiratory virus, a causative agent of COVID-19. The primary target of the virus is the airway epithelium of the upper respiratory tract [1][2][3]. During the course of the disease, the virus can descend to the lower respiratory tract, infecting bronchial epithelium and type II pneumocytes [4]. The main receptor for SARS-CoV-2, angiotensin-converting enzyme 2 (ACE2) [1,5,6], determines the viral tropism, which is not restricted to the respiratory epithelium and in certain cases can infect enterocytes, as well as kidney, heart, brain, and other cell types [7,8]. Molecules other than ACE2 have been reported to be involved in SARS-CoV-2 entry, such as neuropilin-1 [9,10], AXL [11], and CD147, although the role of the latter is speculative [12,13].
SARS-CoV-2 entry is mediated by the spike (S) protein [14]. The S protein belongs to trimeric class I fusion proteins [15] that undergo substantial conformational changes when bound to a cellular receptor, leading to fusion between viral and cell membranes [16,17]. The extracellular portion of the spike consists of two subunits: S1 binds to ACE2 and S2 mediates the viral fusion [17]. A newly synthesized spike exists in a metastable prefusion state [17]. Following attachment to permissive cells, the receptor-binding domain (RBD) in the S1 subunit transitions between the inactive 'down' position and the accessible 'up' position for interaction with ACE2 [18][19][20]. Using single-molecule fluorescence resonance

Human Serum Samples
All serum samples were derived from the human serum biobank of the Gamaleya Center for Epidemiology and Microbiology. Study was approved by the local ethics committee of the Moscow First Infectious Disease Hospital (Protocol #2 dated 22 January 2021).

Plasmid Construction
The plasmid pCG1-SARS-2-S coding for the codon-optimized S-protein was kindly provided by Prof. Dr. Stefan Pöhlmann (Infection Biology Unit of the German Primate Center, Leibniz Institute for Primate Research). C-terminal truncation of the S-protein (∆C19), addition of 8 amino acids from the HIV gp41 (H2), and mutation of the furin cleavage site PRRA¨A (∆F) were introduced by PCR with Pfu polymerase (Sibenzyme, Russia) and verified by sequencing. The HIV-1 (strain NL4-3) packaging plasmids pCMV-dR8-2 (# 12263) and vector pCMV-VSV-G for expression of the protein G from vesicular stomatitis virus (# 8454) were obtained from Addgene; reporter plasmids pUCHR-inLuc-mR and pUCHR-IR-GFP were described previously [38,39]. The plasmid pUCHR-hACE2 was generated by subcloning the ACE2 coding sequence from the pCG1-hACE2 plasmid obtained from Prof. Dr. Stefan Pöhlmann (Infection Biology Unit of the German Primate Center, Leibniz Institute for Primate Research) into lentiviral vector pUCHR.

Establishing 293T/ACE2 Target Cell Line
To produce lentiviral particles, 0.35 × 10 6 293T cells were plated in 1 well of a 6-well plate in 2.5 mL of growth medium. The next day, the cells were transfected with 0.66 µg of pCMV-dR8-2, 0.88 µg of pUCHR-hACE2, and 0.22 µg pCMV-VSVG using Lipofectamine 2000 (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instruction. At 48 h post transfection, supernatants with PVs were cleared through 0.45 µm pore size filters and used for transduction. 293T cells (8 × 10 4 per well) were plated in a 24-well plate overnight in 500 µL of growth medium. Serially diluted lentiviral particles were added to the cells, and the percentage of ACE2-positive cells was analyzed by flow cytometry at 48 h post infection. The sample with an approximately 30% level of transduction was selected for further isolation using a Sony MA900 (Sony Biotechnology, San Jose, CA, USA) cell sorter. The cells were expanded and sorted once again to enrich the ACE2-positive population more than 98%.

Detection of Syncytia Formation
A total of 8 × 10 4 293T cells per well were plated in a 24-well plate in 500 µL of growth medium. After 24 h, the cells were transfected with 0.5 µg pCMV-GFPt and 0.3 µg pCG1-SARS-2-S, pCG1-SARS-2-S∆C19, or pCG1-SARS-2-S∆C19-H2 using Lipofectamine 2000 (Thermo Fisher Scientific) according to the manufacturer's instruction. The next day, the transfected cells and 293T/ACE2 cells were detached with 1 mM EDTA, mixed at the ratio of 1:1, and plated in wells of a 24-well plate with the total number of 10 5 cells per well. Images of live cells were acquired by the Nikon eclipse Ti microscope at the ×10 magnification 24 h later.

Single-Cycle Cell-Free Infection
A total of 8 × 10 4 293T/ACE2 cells per well were plated in a 24-well plate in 400 µL of growth medium. The following day, the whole volume was replaced with 400 µL of medium containing PVs. Infection level was determined 48 h later by luciferase assay or flow cytometry. To measure neutralizing activity of sera from COVID-19 patients, sera were serially four-fold diluted in growth medium and preincubated with PVs in the total volume of 400 µL for 1 h at room temperature before addition to target cells.

Single-Cycle Cell Coculture Infection
To generate pseudovirus-producing cells, 9 × 10 5 293T cells were plated in a 6-cm dish in 5 mL of growth medium. The next day, the cells were transfected with 1.67 µg pCMV-dR8-2, 2.22 µg pUCHR-inLuc-mR, and 1.11 µg pCG1-SARS-2-SdFdC19 using Lipofectamine 2000 (Thermo Fisher Scientific) according to the manufacturer's instruction. Then, 24 h post transfection, producer cells were detached with 1mM EDTA and washed twice with PBS. A total of 2.6 × 10 4 cells were mixed with serial four-fold dilutions of sera in the total volume of 200 µL and incubated for 1 h at 4 • C. Next, they were mixed with 5.4 × 10 4 target 293T/ACE2 cells detached with 1 mM EDTA and resuspended in 200 µL of medium. Cell mixture was plated in a 24-well plate and cocultured in the 400 µL volume of medium. Luciferase activity was determined 48 h later.

Cell-Free SARS-CoV-2 Spreading Assay
Vero E6 cells were plated at 8 × 10 4 cells/well into 96-well plates the day prior to experiments. Serum samples were serially four-fold diluted in growth medium, mixed with MOI 0.01 of SARS-CoV-2 and incubated for 1 h at 37 • C. The mixture was then added to Vero E6 cells and incubated for 5 days at 37 • C. Cytopathic effect (CPE) was determined by MTT assay [41,42].

Cell-to-Cell SARS-CoV-2 Spreading Assay
Vero E6 cells were plated into a T25 cell culture flask and infected with SARS-CoV-2 at MOI 0.01. The next day, infected cells were detached with trypsin/EDTA solution (Gibco, USA), washed twice with PBS, mixed with serial four-fold dilutions of sera at 2.6 × 10 4 cells per sample, and incubated for 1 h at 37 • C. The mixture was then combined with 5.4 × 10 4 uninfected Vero E6 cells in 96-well plates in the total volume of 200 µL and incubated for 5 days at 37 • C. CPE was determined by MTT assay.

Flow Cytometry
To measure S protein expression on the surface of 293T PV-producing cells, 3 × 10 5 transfected cells were incubated with the serum from a convalescent donor at the 1:100 dilution in PSB for 30 min followed by the incubation with the secondary anti-human IgG antibodies conjugated with PE (1:250, # H10104, Thermo Fisher Scientific) for 30 min. ACE2 expression was assessed by staining cells with polyclonal rabbit antibodies against human ACE2 (PAB886Hu01, Cloud-Clone Corp, Katy, TX, USA) followed by secondary anti-rabbit antibodies conjugated to PE (1:250, # P-2771MP, Thermo Fisher Scientific). Samples were analyzed on a CytoFLEX S flow cytometer (Beckman Coulter, Brea, CA, USA). FlowJo LLC software (Ashland, OR, USA) was used for histogram visualization.

Data Analysis
The data were analyzed and visualized using GraphPad Prism 8 Software (San Diego, CA, USA). NT 50 values were calculated using a nonlinear regression curve fit to normalized data expressed as % inhibition of infection.

Generation and Optimization of a SARS-CoV-2 Pseudoviral System to Measure Cell-Free Infection
Pseudoviruses are viruses enveloped with a heterologous surface protein that changes their natural tropism. Unlike native systems, heterologous protein envelopes are often incorporated into PV particles at lower efficiency. A number of studies have been focused on optimizing lenti-and retroviral systems pseudotyped with the SARS-CoV S protein. Giroglou et al. showed that C-terminal truncation of the S protein increased retroviral particle titers, explained by the removal of the ER retention signal and exposure of the S protein to the cell surface [43]. Moore et al. found that both C-terminal truncation and substitution of the cytoplasmic portion of the S protein with eight amino acids from the C-terminus of HIV-1 gp41 increased the level of infectious lentiviral particle production [44]. Later on, the spike protein from SARS-CoV-2 with a cytoplasmic portion deleted was used in a number of pseudoviral test systems [6,33,[45][46][47]. The furin cleavage site that was present in SARS-CoV-2 S but not in the SARS-CoV S protein is thought to be involved in spike maturation, virus entry, and syncytium formation [5] and, therefore, can also affect infectivity measured with PVs.
In order to establish an HIV-based infection system, the SARS-CoV-2 S protein was modified by deleting the last 19 amino acids (∆C19) or substituting them with eight amino acids from HIV-1 gp41 (H2). These modifications were either combined with the mutation in the furin cleavage site RRAR to A (∆F), or left uncombined, to generate the six variants of spike protein indicated in Figure 1A. Next, a SARS-CoV-2 permissive HEK 293T cell line was established with a stable expression of the human ACE2 receptor via lentiviral transduction and FACS sorting ( Figure 1B). A cell-free infectivity assay was set up, as schematically illustrated in Figure 1C. PVs were generated by co-transfecting 293T cells with one of the S-protein-coding plasmids, HIV-1 packaging vector pCMV-dR8-2, and an improved intron-regulated reporter vector pUCHR-inLuc-mR, capable of measuring both cell-free and cell coculture infections using the mean of luciferase activity [38,39]. Additionally, the pUCHR-IR-GFP reporter plasmid without an intron was used to evaluate cell-free infectivity levels using flow cytometry. At 48 h post transfection, supernatants containing PVs were harvested and concentrated by centrifugation. Equal amounts of PVs were added to 293T/ACE2 cells for 48 h, and levels of infection were estimated by measuring luciferase activity or percentage of GFP-positive cells, depending on reporter type. The resulting values of infection were normalized to p24 levels, and presented relative to the values obtained for the wild-type S protein. As shown in Figure 1D, ∆C19 moderately increased the level of infection, while the H2 modification had no or little effect on infectivity. By contrast, the ∆F mutation resulted in about a 1.5 log increase in PV infectivity. On ∆F background, however, the improving effect of ∆C19 was much less pronounced than Viruses 2021, 13, 1133 6 of 17 detected without ∆F. PV titration was used to confirm a substantial effect of the ∆F mutation on the level of PV transduction (~20-fold enhancement in many PV dilutions) ( Figure 1E,F). The increased infectivity of the ∆F mutant PVs was not accompanied by an increase in S protein expression on PV-producing 293T cells ( Figure 1G). Thus, it was unclear whether ∆F infectivity was enhanced from S incorporation into PVs or if this was a feature of the 293T cellular system, in which S processing by furin is important during the fusion step of the viral life cycle.
In summary, a SARS-CoV-2 cell-free infection test was developed in a 24-well plate format with a high level of sensitivity. Using the ∆F∆C19 modification of the SARS-CoV-2 S protein, 50-60% GFP transduction and about 4 logs over the background elevation of luciferase activity was achieved, making consecutive inhibitory analysis accurate and reproducible.
of PVs were added to 293T/ACE2 cells for 48 hours, and levels of infection were estimated by measuring luciferase activity or percentage of GFP-positive cells, depending on reporter type. The resulting values of infection were normalized to p24 levels, and presented relative to the values obtained for the wild-type S protein. As shown in Figure 1D, ∆C19 moderately increased the level of infection, while the H2 modification had no or little effect on infectivity. By contrast, the ∆F mutation resulted in about a 1.5 log increase in PV infectivity. On ∆F background, however, the improving effect of ∆C19 was much less pronounced than detected without ∆F. PV titration was used to confirm a substantial effect of the ∆F mutation on the level of PV transduction (~20-fold enhancement in many PV dilutions) (Figures 1 E and F). The increased infectivity of the ∆F mutant PVs was not accompanied by an increase in S protein expression on PV-producing 293T cells ( Figure  1G). Thus, it was unclear whether ∆F infectivity was enhanced from S incorporation into PVs or if this was a feature of the 293T cellular system, in which S processing by furin is important during the fusion step of the viral life cycle.
In summary, a SARS-CoV-2 cell-free infection test was developed in a 24-well plate format with a high level of sensitivity. Using the ∆F∆C19 modification of the SARS-CoV-2 S protein, 50-60% GFP transduction and about 4 logs over the background elevation of luciferase activity was achieved, making consecutive inhibitory analysis accurate and reproducible.

Development of a SARS-CoV-2 Pseudoviral System to Quantify Cell-to-Cell Infection
In order to evaluate SARS-CoV-2 cell-to-cell infection, a one-step transfection-infection assay with the inLuc-mR reporter vector described earlier was set up [38,39]. Briefly, 293T/ACE2 cells were co-transfected with viral vectors, as outlined above, for cell-free infection. In approximately 12-16 h, transfected cells started to produce PVs, which infected nearby 293T/ACE2 cells. At 48 h post transfection, one cycle of replication was complete and luciferase activity could be measured (Figure 2A). Using this assay, the levels of infection with one of the six variants of the S protein were quantified. Wild-type, ∆C19, and H2 proteins did not mediate infection at all; however, all three variants bearing the ∆F mutation supported a good level of infection. The addition of ∆C19 to ∆F increased the level of infectivity by 0.5 log, while the H2 modification had no effect on the signal ( Figure 2B). We have previously demonstrated that intron-regulated reporter vectors do not detect infection in cell syncytia, as the reporter protein can be expressed only in actively replicating target cells [38]. Therefore, the ability of differently modified S protein variants to induce syncytia formation in 293T/ACE2 cells was examined. Consistent with previously reported data [6,23,37], a massive cell-cell fusion upon expression of all three variants of spike bearing the furin cleavage site was detected, and there was no syncytia formation in the samples transfected with ∆F variants ( Figure 2C). This suggests that S-mediated syncytia formation inhibits lentiviral reporter expression in permissive cells; consequently, wild-type S protein cannot be used to assess cell-to-cell infection in the 293T/ACE2 cellular model. In summary, the possibility of measuring SARS-CoV-2 cell-to-cell infection using the intron-regulated luciferase vector was demonstrated, and the ∆F∆C19 mutant of S was selected for further study. luciferase activity measured for a mutant spike was normalized to that obtained for the wild-type S protein. Two independent experiments with two different PV preparations were performed. (E,F) The levels of cell-free infection with indicated PVs were measured using either GFP (E) or inLuc (F) reporter. (G) The levels of S protein expression on PV-producing cells estimated by flow cytometry. To express indicated variants of protein S, 293T cells were transfected and stained with convalescent human serum in 48 hours. Median fluorescence intensity (MFI) level was calculated for every mutant in the gate of transfected cells and normalized to the MFI detected for wild-type S protein.

Development of a SARS-CoV-2 Pseudoviral System to Quantify Cell-to-Cell Infection
In order to evaluate SARS-CoV-2 cell-to-cell infection, a one-step transfection-infection assay with the inLuc-mR reporter vector described earlier was set up [38,39]. Briefly, 293T/ACE2 cells were co-transfected with viral vectors, as outlined above, for cell-free infection. In approximately 12-16 hours, transfected cells started to produce PVs, which infected nearby 293T/ACE2 cells. At 48 hours post transfection, one cycle of replication was complete and luciferase activity could be measured (Figure 2A). Using this assay, the levels of infection with one of the six variants of the S protein were quantified. Wild-type, ∆C19, and H2 proteins did not mediate infection at all; however, all three variants bearing the ∆F mutation supported a good level of infection. The addition of ∆C19 to ∆F increased the level of infectivity by 0.5 log, while the H2 modification had no effect on the signal ( Figure 2B). We have previously demonstrated that intron-regulated reporter vectors do not detect infection in cell syncytia, as the reporter protein can be expressed only in actively replicating target cells [38]. Therefore, the ability of differently modified S protein variants to induce syncytia formation in 293T/ACE2 cells was examined. Consistent with previously reported data [6,23,37], a massive cell-cell fusion upon expression of all three variants of spike bearing the furin cleavage site was detected, and there was no syncytia formation in the samples transfected with ∆F variants ( Figure 2C). This suggests that Smediated syncytia formation inhibits lentiviral reporter expression in permissive cells; consequently, wild-type S protein cannot be used to assess cell-to-cell infection in the 293T/ACE2 cellular model. In summary, the possibility of measuring SARS-CoV-2 cell-tocell infection using the intron-regulated luciferase vector was demonstrated, and the ∆F∆C19 mutant of S was selected for further study.  The levels of infectivity measured using one-step assay with different spike protein mutants. Samples with the intact furin cleavage site produced no signal above background (n/d). The differences between ∆F mutants were calculated by one-way ANOVA with Tukey's multiple comparison test, and were significant at p = 0.0008 (***) and p = 0.0457 (*). Simbols and color codes in (B) are the same as in Figure 1D

Comparative Analysis of the Neutralizing Activity of Convalescent Sera in SARS-CoV-2 Cell-Free and Cell Coculture Pseudoviral Infection Tests
Using the developed pseudoviral infection tests, side-by-side comparisons of the neutralizing activity of convalescent sera from COVID-19 patients in cell-free and cell coculture modes of infection were performed. To avoid possible biases that can be observed when a neutralizing agent is added at the time of infection initiation, neutralization tests were designed to allow either PVs or producer cells to be preincubated with a serum for 1 h prior to the target cell addition (see schematic in Figure 3A). Specifically, cell-free PVs in the amount of 10 ng of p24 were incubated with indicated serum dilutions in a total volume of 400 µL of culture medium, and added to 8 × 10 4 293T/ACE2 cells, seeded overnight in a 24-well plate. The levels of cell-free infection were estimated 48 h later by measuring luciferase activity in cell lysates. In these experimental settings, the results with control samples were consistently reproduced at the level of~10 6 RLU, giving an opportunity to detect a wide range of inhibitory activity. Five COVID-19 convalescent sera with high neutralizing activity were selected and evaluated in the cell-free infection test with ∆F∆C19. As shown in Figure 3A,E, all samples demonstrated NT 50 in a range between 1/1500 and 1/12000 dilution, whereas a non-immune serum had no inhibitory activity. Additionally, in order to determine whether the furin cleavage site mutation or C-terminal truncation influenced neutralization titer, wild-type and ∆C19 variants per se were tested. The inhibition rates against Swt and ∆C19 were similar. The addition of ∆F to ∆C19 moderately reduced serum neutralization capacity in comparison to inhibitory titers measured for wt or ∆C19 S proteins ( Figure 3B), including NT 50 values ( Figure 3C). Thus, the ∆F modification in the S protein slightly changed the level of PV neutralization in the cell-free test, but was absolutely necessary for measuring cell coculture infectivity and making the correct comparison between two types of infection.
The cell-to-cell neutralization test was designed to be as similar as possible to settings used for the cell-free PV inhibition analysis. To generate SARS-CoV-2 producer cells, non-permissive 293T cells were co-transfected with pCMV-∆8.2R, pUCHR-inLuc-mR, and pCG1-SARS-2-S∆F∆C19 plasmids, as described for cell-free infection. After 24 h, cells were gently suspended using ethylenediaminetetraacetic acid (EDTA) and washed once with phosphate-buffered saline (PBS); 2.6 × 10 4 transfected cells in 200 µL culture medium were preincubated with a certain serum dilution for 1 h at +4 • C and mixed with 5.6 × 10 4 293T/ACE2 cells, giving a total of 8 × 10 4 cells in 0.4 mL of culture medium. The cell mixture was placed in the wells of a 24-well plate and incubated for 48 h before luciferase activity measurement ( Figure 3D, schematic). The described format, and the resulting ratio of one producer cell to two target cells, provided the optimal sensitivity for measuring cell coculture infection and comparing it to cell-free infection in control samples. As shown in Figure 3D on the right, inhibition of SARS-CoV-2 cell coculture infection required high concentrations of convalescent sera, with NT 50 detected within the 1/100 to 1/1100 dilution range. Compared to the serum activities against cell-free infection, the neutralization capacities of the same sera against cell coculture infection were more than tenfold lower ( Figure 3E). Nonetheless, NT 50 titers of individual serum samples measured in cell-free and cell-to-cell infection tests correlated with each other (Figure 3F), i.e., sera with a higher inhibitory titer detected in the cell-free infection test more effectively inhibited cell coculture infection.
In summary, by using the developed pseudoviral single-cycle replication assay with the intron-regulated reporter vector, it was demonstrated that SARS-CoV-2 cell coculture infection was much more resistant to neutralization by convalescent sera than infection with purified PVs. In summary, by using the developed pseudoviral single-cycle replication assay with the intron-regulated reporter vector, it was demonstrated that SARS-CoV-2 cell coculture infection was much more resistant to neutralization by convalescent sera than infection with purified PVs.

Neutralization Potential of Convalescent Sera against Replication-Competent SARS-CoV-2
Neutralization tests with PVs, although safe, have serious limitations, since they can only mimic the entry step of the viral life cycle. The mechanisms of viral assembly, egress, and transmission for HIV and coronaviruses are very different, so the HIV-1 core proteins responsible for these processes-and used in this study's pseudoviral tests-cannot model SARS-CoV-2 cell-to-cell transmission. With an understanding of all the drawbacks of the developed tests, an investigation of whether the resistance of cell-to-cell transmission to antibody neutralization could be reproduced with a full replication-competent SARS-CoV-2 was conducted. To this end, Vero E6 monkey fibroblast cells were chosen for virus production, setting up cell-free and cell-to-cell infection. The 293T/ACE2 cells for this purpose were excluded, as they died quickly after infection with coronavirus, making viral stock generation or maintaining multiple cycles of replication impossible. To remain consistent with pseudoviral tests, the number of plated Vero cells were proportionally similar to what was used for 293T cells. Neutralization of cell-free infection was performed by preincubating 0.01 MOI of SARS-CoV-2 strain hCoV-19/Russia/Moscow_PMVL-4 with serially diluted convalescent sera for 1 h, and then adding to the Vero cells, seeded in a 96-well plate overnight. At day five post infection, cytopathic effect (CPE) was measured using the MTT test. As shown in Figure 4A, all sera completely blocked SARS-CoV-2 replication at 1/100 dilution; NT 50 values ranged from 1/400 to 1/1400. These values were lower than the corresponding NT 50 determined in the pseudoviral test. This can be explained by the doses of PVs and virions used for the neutralization assays, which are difficult to compare or normalize. Nevertheless, the results of two cell-free assays correlated well to each other ( Figure 4B).
Cell-to-cell SARS-CoV-2 spreading assay was initiated by infecting 2.6 × 10 4 Vero E6 cells with fully competent virus for 24 h, followed by PBS washing and preincubating with a serum before it was added to 5.6 × 10 4 uninfected Vero E6 cells. The level of CPE was measured five days later using the MTT test. In stark contrast to the cell-free infectious test, the majority of the serum samples, even at minimal dilution, did not prevent the cytopathic effect of fully-competent SARS-CoV-2 ( Figure 4C), with the exception of serum 9611, which at 1/20 dilution displayed~50% inhibitory activity.
In conclusion, it was demonstrated that cell-to-cell spread of the fully competent wild-type SARS-CoV-2 is almost completely resistant to convalescent serum neutralization. This effect was even more pronounced than the resistance detected using PVs.

Neutralization Potential of Convalescent Sera against Replication-Competent SARS-CoV-2
Neutralization tests with PVs, although safe, have serious limitations, since they can only mimic the entry step of the viral life cycle. The mechanisms of viral assembly, egress, and transmission for HIV and coronaviruses are very different, so the HIV-1 core proteins responsible for these processes-and used in this study's pseudoviral tests-cannot model SARS-CoV-2 cell-to-cell transmission. With an understanding of all the drawbacks of the developed tests, an investigation of whether the resistance of cell-to-cell transmission to antibody neutralization could be reproduced with a full replication-competent SARS-CoV-2 was conducted. To this end, Vero E6 monkey fibroblast cells were chosen for virus production, setting up cell-free and cell-to-cell infection. The 293T/ACE2 cells for this purpose were excluded, as they died quickly after infection with coronavirus, making viral stock generation or maintaining multiple cycles of replication impossible. To remain consistent with pseudoviral tests, the number of plated Vero cells were proportionally similar to what was used for 293T cells. Neutralization of cell-free infection was performed by preincubating 0.01 MOI of SARS-CoV-2 strain hCoV-19/Russia/Moscow_PMVL-4 with serially diluted convalescent sera for 1 hour, and then adding to the Vero cells, seeded in a 96-well plate overnight. At day five post infection, cytopathic effect (CPE) was measured using the MTT test. As shown in Figure 4A, all sera completely blocked SARS-CoV-2 replication at 1/100 dilution; NT50 values ranged from 1/400 to 1/1400. These values were lower than the corresponding NT50 determined in the pseudoviral test. This can be explained by the doses of PVs and virions used for the neutralization assays, which are difficult to compare or normalize. Nevertheless, the results of two cell-free assays correlated well to each other ( Figure 4B).
Cell-to-cell SARS-CoV-2 spreading assay was initiated by infecting 2.6 × 10 4 Vero E6 cells with fully competent virus for 24 hours, followed by PBS washing and preincubating with a serum before it was added to 5.6 × 10 4 uninfected Vero E6 cells. The level of CPE was measured five days later using the MTT test. In stark contrast to the cell-free infectious test, the majority of the serum samples, even at minimal dilution, did not prevent the cytopathic effect of fully-competent SARS-CoV-2 ( Figure 4C), with the exception of serum 9611, which at 1/20 dilution displayed ~50% inhibitory activity.
In conclusion, it was demonstrated that cell-to-cell spread of the fully competent wild-type SARS-CoV-2 is almost completely resistant to convalescent serum neutralization. This effect was even more pronounced than the resistance detected using PVs.

Discussion
The fast and global spread of COVID-19 requires unprecedent efforts to control this pandemic. An important parameter of collective immunity, gained after SARS-CoV-2 infection or vaccination, is the neutralization activity of anti-spike antibodies. This reflects the degree to which the studied population is protected from the infection, and provides more adequate information than a titer of anti-spike antibodies measured by ELISA. However, this test requires strict BSL3 conditions in order to work with highly pathogenic full-length SARS-CoV-2, and it has not been widely utilized. In this respect, different PV systems are considered safe and acceptable for many labs, since they allow the completion of only one cycle of viral replication. The PVs have been adapted by many researchers to characterize the SARS-CoV-2 entry process [3,5,6,23,48,49], monitor the dynamics of neutralizing humoral immunity to SARS-CoV-2 [14,50,51], and screen potential inhibitors. [28,37].
The aim of this study was to not just replicate a PV system but rather, develop a lentivirus-based PV test capable of measuring SARS-CoV-2 infection, both with purified PVs and in cell cocultures. The latter has not been appreciated previously or measured accurately. The use of identical vectors to initiate both types of infection makes comparative analysis of PV infectivity more rigorous. First, our system was optimized by pseudotyping PVs with S protein mutants. Since the cytoplasmic portion of the S protein has an ER-Golgi retention signal needed for incorporation into coronaviruses that bud from endosomal membranes [52,53], but which may not be optimal for efficient pseudotyping of lenti-or retroviral particles that assemble predominantly at the plasma membrane, this signal should be removed. An early study on SARS-CoV by Giroglou et al. [43] demonstrated that the C-terminally truncated spike increased PV infectivity, which led to the inclusion of this modification in the many subsequent PV systems developed for SARS-CoV-2 [6,33,[45][46][47]54]. The truncation of the last 18-19 [47,54,55], or even 13 amino acids [45] enhanced PV infectivity from 10-to 100-fold. Consistent with the data reported above, our study has shown that the ∆C19 mutation improved cell-free infection by 15-fold, and one-step infection by 5-fold ( Figures 1D and 2B). Nevertheless, a few studies did not find a substantial influence from ∆C19 [46] or point mutations in the ER retention signal [34,56] on PV infectivity. In agreement with published papers [34,43,45], we did not observe substantial differences between wt and ∆C19 S protein levels expressed on the surface of PV producing cells ( Figure 1G). Thus, the mechanism of enhanced infectivity for ∆C19 PVs remains unclear, and can be related to improved incorporation of this mutant into PVs [44,45,56] and/or stabilization of S1-S2 subunit interaction [45]. Unlike simple truncation, the substitution of C19 with the most membrane-proximal cytoplasmic domain of gp41, ∆C19-H2 [44], did not alter PV infectivity in our tests. Crawford et al. substituted the cytoplasmic portion of the S protein with the intracellular domain from influenza hemagglutinin, and reported no improvement in PV infectivity as well [34].
Unlike SARS-CoV-1 S, the S protein of SARS-CoV-2 contains a furin cleavage site (F), located a little upstream of the S1/S2 boundary [24]. It became clear early on that the presence of F increases Env-mediated cell-cell fusion, at least for in vitro experiments [23,57]. However, the effects of F on virus infectivity were contradictory, i.e., either a decrease [48] or an increase [49] in ∆F PV infectivity relative to the wt S protein was reported. Finally, several groups found that infectivity depends on the cell target and, in particular, the entry site that the virus uses during infection [2,23,58]. The latter is largely dependent on the S protein cleavage at the S2' site by surface protease TMPRSS2 or lysosomal cathepsins [22]. If cells express TMPRSS2 and the virus enters via the plasma membrane, then ∆F decreased PV infection [23], whereas TMPRSS2-negative target cells, such as the widely used 293T/ACE2 cells, were usually infected similarly [2,23], or even better, with ∆F PVs [48,49,56,[58][59][60], depending on the mutations introduced at the F site. Consistent with these reports, a 15-20-fold increase in infectivity with purified ∆F PVs in 293T/ACE2 cells was detected in this study. Strikingly, one-step transfection/infection using all variants of the S protein with intact F was undetectable ( Figure 2B), which was explained by massive syncytia formation induced by the S protein ( Figure 2C) and by blocking inLuc-mR transduction in fused cells [38]. Thus, measurement of cell coculture infection must adhere to the ∆F variant of spike and be used in cell-free infection for comparative purposes. Summarizing this part of the study, the ∆F∆C19 mutant of the S protein was selected as the one providing the highest sensitivity to PV infection, both in cell-free and cell coculture experimental settings.
Next, the generated PV system was validated in a neutralization test with convalescent sera. As the ∆F mutation is localized in the external part of the S protein, it can potentially influence serum neutralization activity. A comparison of the ∆F∆C19 and ∆C19 mutants revealed that ∆F required~twofold higher serum concentration for PV neutralization than without ∆F ( Figure 3C). This is consistent with the study by Johnson et al. [61], and suggests that using the ∆F∆C19 mutant spike slightly underestimated the neutralization potential of sera, but did not overestimate it. Using five selected COVID-19 convalescent sera with high anti-spike titers, inhibitory activity was quantified against the ∆F∆C19 in cell-free and cell coculture modes of infection, and NT 50 was calculated for all tested sera. It was demonstrated that all convalescent sera were at least tenfold less efficient in the neutralization of cell coculture infection, relative to inhibitory activity detected with purified PVs. This became possible by using a previously developed replication-dependent vector in Luc. There is a limited number of studies in which infectivity or fusogenicity of SARS-CoV-2 PVs was measured using original luciferase-based reporter vectors, such as Gaussia luciferase (Gluc and in-Gluc) [21,47] or split-Nanoluc system [62]. Although the authors of these studies accurately quantified the levels of S-mediated virus-cell and cell-cell fusion, the cell coculture infection per se was not measured. Acknowledging that the validity of the results obtained in this study with lentiviral PVs could be heavily criticized, we conducted neutralization experiments on Vero E6 cells with live full-length SARS-CoV-2, at transmission settings that were as close as possible to those developed for the single-round infection tests. The viral multiple replication assays not only confirmed the results with PVs, but also demonstrated the near complete resistance of SARS-CoV-2 cell coculture infection to neutralizing antibodies ( Figure 4C). This phenomenon has been observed for a number of viruses [63][64][65][66][67][68], but has not been reported for coronaviruses. Some respiratory viruses have been shown to utilize cell-to-cell transmission. Examples include induction of intercellular extensions by the influenza virus, PIV5 [69,70], HMPV [63,71], and RSV [72], and usage of intercellular membrane pores by the measles virus, which is also able to infect airway epithelium [73]. Coronaviruses extensively reorganize not only the ER-Golgi network but also change plasma membrane characteristics, inducing formation ruffles and filopodia [74]. Ogando et al. observed that Vero E6 cells infected with SARS-CoV-2 alter their morphology by forming long filopodia with budding viruses [75]. The study on Caco-2 cells by Bouhaddou et al. showed that the N protein of SARS-CoV-2 interacts with the kinase CK2 that leads to cytoskeleton reorganization and filopodia formation [76]. Using scanning electron microscopy, Caldas et al. visualized thin protrusions between Vero cells with adherent virions [77]. Whether the observed intercellular contacts play a role in SARS-CoV-2 cell-to-cell transmission in vivo is not known. Massive cell-cell fusion induced by SARS-CoV-2 has been observed in cell culture [23,57] and in human organoids [78,79], as well as in post-mortem material [80][81][82][83][84]. However, the biological significance of syncytia formation in vivo remains uncertain. The most logical mechanism that would protect SARS-CoV-2 cell-to-cell transmission from antibody neutralization is a tight cellular contact with a synaptic cleft where the virus buds, and where large immunoglobulins may have difficulty penetrating. Our preliminary experiments with much smaller nanobodies that relatively efficiently blocked SARS-CoV-2 cell coculture infection partially support this assumption. This mechanism has been implied for retroviral infectious synapses [85]. However, such structures have not been described for SARS-CoV-2, which transmits in airway epithelium. The neutralizing activity of serum antibodies will also depend on viral inputs used in cell-free and cell coculture infection tests. It cannot be excluded that the continuous production of viral particles and expression of S protein on producer cells during cell-to-cell infection will require more antibodies for efficient neutralization. Prolonged co-incubation of infected and uninfected Vero cells for 5 days, required for a clear CPE detection, likely leads to virus abundance that somewhat explains the substantially lower neutralization of cell-to-cell infection performed with live coronavirus relative to PVs, which support only a single round of replication.
In conclusion, we developed lentivirus-based single-round pseudoviral infection assays suitable for quantitatively measuring SARS-CoV-2 entry in cell-free and cell coculture conditions. Using this system, as well as the SARS-CoV-2 spreading assay, our study has shown that cell-to-cell infection of SARS-CoV-2 is considerably more resistant to serum neutralization than infection with purified viral particles. These results underline the importance of SARS-CoV-2 cell-to-cell transmission for virus biology, immune protection, and the development of entry inhibitors. Although the developed cell coculture PV system has a serious limitation related to the usage of mutated variants of the spike protein and unnatural core proteins, we believe it will help validate antibodies, peptides, and small molecules that would block virus entry more effectively than inhibitors selected by a conventional PV system.  Informed Consent Statement: Informed consent was obtained from all individuals whose serum was tested in the study.
Data Availability Statement: All authors declare that data and materials will be made available according to the guidelines of the journal.

Conflicts of Interest:
The authors declare no financial or non-financial interest.