Deletion of the SARS-CoV-2 Spike Cytoplasmic Tail Increases Infectivity in Pseudovirus Neutralization Assays

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the etiologic agent of the COVID-19 pandemic. The development of a high-throughput pseudovirus neutralization assay is critical for the development of vaccines and immune-based therapeutics. In this study, we show that deletion of the cytoplasmic tail of the SARS-CoV-2 spike leads to pseudoviruses with enhanced infectivity. This SΔCT13-based pseudovirus neutralization assay should be broadly useful for the field.


RESULTS
Deletion of the last 13 amino acids in spike protein enhances SARS-CoV-2 pseudovirus infection. Two main forms of lentiviral vector-based pseudovirus production strategies have been reported, involving a two-plasmid system or a three-plasmid system (8). The former requires a backbone plasmid that expresses a full set of HIV-1 gene products except for the HIV-1 envelope (Env) and a second plasmid that encodes heterologous Env of interest. In contrast, the three-plasmid system separates the backbone plasmid into two individual constructs, one packaging plasmid and one transfer plasmid (8). The three-plasmid system was generally regarded as a safer method, since multiple virulent genes have been removed (27). For this reason, we chose the threeplasmid system to generate the SARS-CoV-2 pseudovirus, with which we established a pseudovirus-based neutralization assay. The HIV core (backbone) is formulated by the packaging plasmid psPAX2 that expresses the HIV gag and pol proteins and the transfer plasmid pLenti-puro cytomegalovirus (CMV)-Luc that harbors the reporter firefly luciferase gene and essential HIV genomic cis-acting regulatory elements allowing proper encapsulation into the HIV capsid (28). During preliminary experiments, the SARS-CoV-2 full-length S demonstrated a rather modest viral entry (2-to 20-fold above cell-only background; data not shown), which limited its utility in evaluating neutralizing activity and thus warranted assay optimization.
Deletion of the last few amino acids in the cytoplasmic tail of SARS-CoV-1 S generated pseudoviruses with greater infectivity (29). We thus sought a similar strategy, deleting the last 13 amino acids in the cytoplasmic tail (CT) of the SARS-CoV-2 S (termed CoV-2 SDCT13) and evaluating it in HEK293T-human ACE2 (hACE2) cells (Fig.  1A). Interestingly, we noticed significantly increased viral entry for CoV-2 SDCT13 CoV-1 S, CoV-1 SDCT13, CoV-2 S, CoV-2 SDCT13, CoV-2 SDCT6, VSVg, or empty (pcDNA3.1) were used to infect HEK293T-hACE2 cells, and the firefly luciferase activity was quantified 48 h postinfection. Data are presented as means 6 SDs from 6 to 12 replicates. (B) S, CoV-1 S, CoV-1 SDCT13, CoV-2 S, CoV-2 SDCT13, CoV-2 SDCT6, VSVg, or empty (pcDNA3.1) were cotransfected with an HIV-1 Tat-expressing plasmid into HEK293T cells; meanwhile, pQCXIP-hACE2 was transfected into TZM-bl cells. Cells were mixed in compared to that for CoV-2 S (Fig. 1A). Similarly, SARS-CoV-1 full-length (CoV-1 S) and the cytoplasmic tail-truncated version (CoV-1 SDCT13) were included in parallel (Fig.  1A). Both SARS-CoV-1 and SARS-CoV-2 demonstrated augmented virus infection when the last 13 amino acids were ablated, suggesting a possible general mechanism of the enhancement. In particular, SARS-CoV-2 appeared to rely more on CT deletion than SARS-CoV-1, with 24-fold and 2.4-fold increases of infectivity, respectively. The SARS-CoV-1 S showed more than 200-fold enhancement of viral entry compared with that of SARS-CoV-2 S, despite the two sharing the same entry receptor, hACE2. To further explore the determinants of the CoV-2 S, we generated an additional CT variant with the last 6 amino acids deleted (CoV-2 SDCT6). As shown in Fig. 1A, CoV-2 SDCT6 displayed similar viral infection efficiency to that of CoV-2 S, suggesting that a potential spike incorporation motif exists in the heptad peptides "SEPVLKG." We further evaluated the spike protein mediated cell-cell fusion between HEK293T/spike and TZM-bl/ hACE2 cells. Deletion of the last 13 amino acids significantly improved CoV-1 S-mediated cell-cell fusion, while CoV-2 SDCT13 did such to a less extent (Fig. 1B). Consistent with the viral infection result, CoV-2 SDCT6 induced cell-cell fusion similarly to CoV-2 S (Fig. 1B). To gain mechanistic insight, we performed Western blot analysis of the proteins expressed in virus-producing HEK293T cells. As shown in Fig. 1C, the transfected cells showed a similar expression level of loading control b-actin and HIV core structural protein p24, suggesting comparable transfection efficiency. However, the CoV-1 S plus CoV-2 S (S11S2) precursors appeared comparable for CoV-1 S, CoV-1 SDCT13, CoV-2 S, and CoV-2 SDCT13, whereas S1 expression in CoV-2 SDCT13 appeared higher than in CoV-2 S, indicating the CT region may affect the spike processing.
We purified viral particles by ultracentrifugation, and analysis by Western blot showed that p24 was generally comparable in all the groups, reflecting similar virus production. However, the S11S2 and S1 demonstrated a sharp increase (Fig. 1D) for both CoV-1 SDCT13 and CoV-2 SDCT13. To confirm, we also collected the virion-containing supernatants and measured the concentration of HIV-1 p24 and CoV-2 spike by enzyme-linked immunosorbent assay (ELISA). In line with Western blot data, more spike proteins as well as a higher spike/p24 ratio were detected in CoV-2 SDCT13 supernatants (Fig. 1E). The data support a model in which deletion of CT facilitates S incorporation into pseudovirions, thus leading to elevated viral entry efficiency.
HIV assembles on the plasma membrane, while coronaviruses bud predominantly through ER (30). We tested whether the enhanced S incorporation into pseudovirions with the DCT constructs resulted from the higher expression of S on the plasma membrane. We performed cell surface staining of SARS-CoV-1 and SARS-CoV-2 spike proteins. Interestingly, the S and SDCT13 showed similar expression levels, as reflected by the geometric mean of S intensity ( Fig. 2A and B), suggesting that the surface expression level unlikely explains the distinct spike incorporation efficiency. Taken together, the results show that deletion of the last 13 amino acids of S substantially facilitated S incorporation into pseudovirions and viral entry of SARS-CoV-2 in cells. Based on these data, we selected SARS-CoV-2 SDCT13 for further study.
SARS-CoV-2 S and SDCT demonstrated similar sensitivities to neutralization and inhibitor blocking. We next established a neutralization assay with this optimized lentivirus-based pseudovirus (see Materials and Methods for a detailed protocol). To address the question whether the SDCT13-based pseudovirus neutralization assay would correlate with the live virus neutralization assay, we therefore performed a direct comparison between the pseudovirus-mediated and live virus-mediated neutralization assay results from a total 67 rhesus monkeys and observed a strong correlation The cell lysates from virus producer cells in panel A were analyzed by Western blotting. (D) Meanwhile, the viral particles in supernatants were purified, concentrated, and analyzed by Western blotting. Internal control b-actin, HIV-1 p24, and coronavirus spike precursor (S11S2) and S1 proteins were examined in both cell lysates and viral particles. (E) The concentrations of HIV p24 and SARS-CoV-2 spike of indicated pseudoviruses were assayed by HIV p24 ELISA and SARS-CoV-2 S ELISA. Data from CoV-2 SDCT13 were set as 100%. The spike density was quantified as Spike/p24. NS, not significant; ***, P , 0.001. we included two additional SARS-CoV-2 S variants, D614G and V367F, which are prevalent virus variants in the global pandemic (31,32). All groups demonstrated similar 50% neutralization titers (NT 50 values) (Fig. 3B). Soluble ACE2 also similarly inhibited CoV-1 S, CoV-1 SDCT13, CoV-2 S, and CoV-2 SDCT13 pseudoviruses (Fig. 3D). Moreover, these data show the flexibility of this assay in evaluating neutralization of SARS-CoV-2 variants as they emerge.
Qualification of the SARS-CoV-2 neutralization assay. As qualification experiments, we assessed the following parameters: linearity, limit of detection (LOD), specificity, and reproducibility. Linearity between input virus and luciferase activity was analyzed with increasing amounts of pseudovirus. As shown in Fig. 5A, solid linearity was achieved between 3.125-and 50-ml virus inoculums (R = 0.9894, P , 0.0001). Of note, 1 ml of input virus was equivalent to 4 pg of HIV-1 p24, as determined by HIV-1 p24 ELISA. To determine sample dilutional linearity combined with intermediate precision, a pooled, heat-inactivated, high-titer COVID-19 human convalescent-phase serum sample was tested undiluted (neat) and in 2-fold serial dilutions ranging from 1:1 to 1:32. These assays were performed by three different operators over three different days. Linear regression of NT 50 values plotted as a function of the serum dilution factor suggested strong linearity when the serum concentration was between undiluted and 1:16 dilution (R = 0.9984, P , 0.0001) (Fig. 5B).
Intra-assay and intermediate assay precision was assessed using the dilutional linearity sample data. Using prespecified acceptance criteria of #25% coefficient of variation (%CV) to define the upper limit of quantification (ULOQ), the assay ULOQ was established at an NT 50 of 3,153 ( Table 1). The limit of detection (LOD) was assessed by measuring the neutralizing activity of human serum starting with a dilution of 1:20. Sera from three other model species, rhesus macaque, hamster, and mouse (naive serum), were included for comparison. The LOD was established to 1:23, 1:28, 1:22, and 1:20 for human, rhesus macaque, hamster, and mouse, respectively (Fig. 5C). Due to limited availability of high-volume COVID-19 human convalescent-phase serum  samples at the time of assay development, undiluted and 1:16 diluted samples were used to approximate intra-assay precision at the high and low ends of the assay range with samples meeting predefined criteria of CV of #25% at each level for each operator ( Table 2). Development of discrete quality control samples that fall in the high (;80%), medium (;50%), and low (;20%) ranges for established assay range will subsequently be developed as high-volume high-neutralizing antibody (nAb)-titer reagents become available. The lentivirus-based neutralization assay demonstrated an overall assay precision of 15.7%. With this established SARS-CoV-2 neutralization assay, we tested human serum from individuals prepandemic (n = 28), at an early stage of COVID-19 (within 10 days since symptom onset, n = 50), and at a late stage of COVID-19 (greater than 10 days since symptom onset, n = 50). There was a clear enhanced neutralizing antibody titer upon infection, with median titers of 20, 218, and 776, respectively ( Fig. 5D and E).

DISCUSSION
Pseudoviruses are versatile and valuable tools for both basic and applied virologic research and have particular advantages when live virus requires higher biosafety BSL3 or BSL4 facilities. High-throughput assays using pseudoviruses in BSL2 facilities can facilitate rapid testing and development of SARS-CoV-2 countermeasures such as vaccines. As such, SARS-CoV-2 pseudovirus neutralization assays are being established. In the manuscript, we describe the development and qualification of an optimized SARS-CoV-2 pseudovirus neutralization assay for preclinical studies and clinical trials. We found that the full-length S of two closely related human coronaviruses, SARS-CoV-1 and MERS-CoV, were readily translatable into a standard pseudovirus system (15,16). Full-length SARS-CoV-2 S mediated a low level of infection, but this can be improved by deleting 13 amino acids in the CT. We addressed several key issues in the development of this assay. First, expression of the virus-derived coding sequence in mammalian cells may be suboptimal due to potential codon usage bias (36). We therefore codon optimized the entire spike DNA sequence by replacing the GC stretches with Homo sapiens-biased codons and validated its robust expression profile in HEK293T cells (Fig. 1C). Second, the viral entry efficiency of SARS-CoV-2 into target cells appears to be intrinsically inferior to that of SARS-CoV-1. SARS-CoV-1 exhibited a .100-fold higher entry efficiency than SARS-CoV-2 (Fig. 1A). This may be ascribed to the stronger binding affinity to the receptor hACE2 (37), greater spike incorporation, accelerated entry kinetics, or other factors. Regardless, we sought to improve SARS-CoV-2 pseudovirus entry into target cells to improve the assay performance. Because partial deletion of the SARS-CoV-1 CT was reported to increase the viral infectivity (29), we deleted the last 13 amino acids of the cytoplasmic tail for both SARS-CoV-1 and SARS-CoV-2. Both SARS-CoVs demonstrated greater viral entry, but this effect was more pronounced for SARS-CoV-2 (Fig. 1A), with a 20-fold enhancement (22). We also observed a marked increase in SDCT incorporation into pseudovirions compared to that of S for both SARS-CoV-1 and SARS-CoV-2, which correlated with enhanced viral entry. The observed SARS-CoV S protein expressed on the cell surface (Fig. 2) may be due to leakiness of the ER retention upon overexpression (29). We hypothesize that the enhanced incorporation of SDCT may reflect higher cell surface expression as a result of the removal of a reported ER retention motif KxHxx in the last 13 amino acids (26,38). However, we observed the same cell surface expression levels for both SDCT13 and S. In addition, the SDCT6 deleted the KxHXX motif but maintained a similar viral entry profile to that of the S, arguing against the hypothesis that the spike cell surface expression makes a difference. Interestingly, when we performed a cell-cell fusion assay, where the CoV-2 S and CoV-2 SDCT13 had comparable expression levels ( Fig. 1C and 2), cell-cell fusion events were similarly induced (,2-fold difference) (Fig. 1B), suggesting that the functioning of the spike may not be altered by CT deletion. A direct measurement of viral entry kinetics (Fig. 4B) similarly implicated that truncation of the 13 amino acids unlikely altered the viral entry properties.
One question regarding the use of SDCT is whether the C-terminal deletion may have changed the conformation of the protein, thus potentially altering neutralization sensitivity. There is currently a lack of direct structural evidence for the CT region of SARS-CoV-2 (39). We observed a strong positive correlation between the pseudovirusand live virus-based neutralization assays (Fig. 3A), suggesting the utility of the pseudovirus neutralization assay. Interestingly, we noticed that a subset of samples with modest NT 50 values were detected with the pseudovirus assay but not with the live virus assay, suggesting different assay sensitivities. Additional studies are warranted to test this hypothesis. Moreover, immunologic and virologic comparisons between SARS-CoV-2 pseudoviruses with S and SDCT13 suggested similar sensitivities to polyclonal antibody, hACE2-mIgG, and BafA1. Therefore, it is probably quantitative rather than qualitative changes of the spike in the virion that determine the increased viral infection rate. Further investigation including detailed structural insight may be warranted. Currently, VSV-based and lentiviral vector-based pseudovirus systems stand for the primary assays for neutralizing antibody titer determination (17)(18)(19)(20)(21)(22)26). The intrinsic properties of rapid replication of infectious VSV accelerate the data generation, while the lentiviral vectors morphologically resemble SARS-CoV-2 and may more genuinely reflect the entry process of SARS-CoV-2.
To define performance characteristics of this SARS-CoV-2 pseudovirus neutralization assay, we performed a series of qualification experiments. The limits of detection for four species appear to be similar (approximately 1:20). Intra-and intermediate precision demonstrated precision levels of ,25%. Overall, this optimized and qualified assay has been developed to evaluate SARS-CoV-2 neutralizing antibody responses in both preclinical and clinical studies. Moreover, this pseudovirus neutralization assay is readily adaptable to incorporating S sequences from SARS-CoV-2 variants, such as currently circulating UK and South African variants.
Rhesus and human serum samples. Rhesus macaque anti-SARS-CoV-2 serum was obtained after an adenoviral vector-based vaccination (3). Serum samples from 16 animals were collected 4 weeks postsingle-shot vaccination. The convalescent-phase serum IgG from our early study (40) was purified and pooled. Twelve deidentified SARS-CoV-2 prepandemic human serum samples from Boston, MA, were obtained. All human studies were conducted in compliance with all relevant local, state, and federal regulations and were approved by the Partners Institutional Review Board (IRB).
Generation of HEK293T-hACE2 stable cells. The retroviral pseudotypes expressing human ACE2 (hACE2) were generated by cotransfecting HEK293T cells with pHEF-VSVG, pBS-CMV-gag-pol, and pQCXIP-hACE2 at a ratio of 0.5:1:1. Viruses in the supernatants were collected every 12 h. The collected viruses were used to transduce HEK293T cells, and positive cell populations were selected by using 1 mg/ml puromycin.
Lentiviral luciferase-based neutralization assay. The SARS-CoV-2 pseudoviruses neutralization assay was generated with an approach similar to that described previously (2,40,41). To determine the neutralization activity of the serum, plasma, or IgG samples from cohorts, HEK293T-hACE2 cells were seeded in 96-well tissue culture plates at a density of 1.75 Â 10 4 cells/well overnight. Threefold serial dilutions of heat-inactivated plasma samples were prepared and mixed with 50 ml of pseudovirus. The mixture was incubated at 37°C for 1 h before adding to HEK293T-hACE2 cells. Forty-eight hours after infection, cells were lysed in Steady-Glo luciferase assay (Promega) according to the manufacturer's instructions. SARS-CoV-2 neutralization titers were defined as the sample dilution at which a 50% reduction in relative light unit (RLU) was observed relative to the average from the virus control wells.
Live virus neutralization assay. A full-length SARS-CoV-2 virus based on a Seattle, WA, isolate was designed to express luciferase and green fluorescence protein (GFP) and was recovered via reverse genetics. The titer of the virus was determined in Vero E6 USAMRID cells to obtain a relative light units (RLU) signal of at least 10Â the cell-only control background. Vero E6 USAMRID cells were plated at 20,000 cells per well the day prior in clear-bottom black-walled 96-well plates. Neutralizing antibody serum samples were tested at a starting dilution of 1:40 and were serially diluted 4-fold up to eight dilution spots. Antibody-virus complexes were incubated at 37°C with 5% CO 2 for 1 h. Following incubation, growth medium was removed, and virus-antibody dilution complexes were added to the cells in duplicates. Virus-only controls and cell-only controls were included in each neutralization assay plate. Following infection, plates were incubated at 37°C with 5% CO 2 for 48 h. After the 48 h of incubation, cells were lysed, and luciferase activity was measured via a Nano-Glo luciferase assay system (Promega) according to the manufacturer's specifications. SARS-CoV-2 neutralization titers were defined as the sample dilution at which a 50% reduction in RLU was observed relative to the average from the virus control wells.
Cell-cell fusion assay. HEK293T cells were transfected with spike-encoding plasmids along with HIV-1 Tat optimized expression vector (AIDS Reagent Program, cat. no. 827). Meanwhile, TZM-bl cells were transfected with human ACE2 plasmid. Twenty-four hours after transfection, cells were cocultured with TZM-bl cells at 1:1 ratio for 12 h, lysed, and measured for firefly luciferase activity. All samples were tested in duplicates, and the results were averaged.
Virus entry kinetics. Briefly, 1.5 Â 10 4 HEK293T-ACE2 cells/well were seeded in a flat 96-well plate at a density of 1.5 Â 10 5 cells per ml. The second day, 50 ml virus was added into each well, and the plate was spun at 1,680 Â g for 30 min at 4°C. The cell culture plates were transferred into a 37°C incubator immediately after the spin to initiate viral internalization and infection. Fifteen microliters DMEM containing 50 nM BafA1 was added into each well immediately after spinoculation (0 h) and 0.5, 1, 2, 3, 4, and 6 h post-temperature shift. Twelve hours after culturing, cells were replenished with fresh DMEM. Fortyeight hours postspinoculation, cells were lysed for luciferase activity quantification.
Cell surface staining. Cells were washed twice with cold 1Â DPBS plus 2% FBS, detached with 1Â DPBS containing 5 mM EDTA, and incubated on ice with the appropriate primary antibodies for 1 h. After three washes with PBS plus 2% FBS, cells were further incubated with fluorescein isothiocyanate (FITC)-conjugated secondary antibodies for 45 min. After two washes, cells were fixed with 2% formaldehyde and analyzed in a BD LSRII flow cytometer.
Intra-assay precision. In the absence of predefined, precalibrated quality control samples, two dilutions within the linear range established in the intermediate precision and dilutional linearity assay were selected to approximate high and low levels of the range. These values subsequently were used to determine intra-assay precision.
Limit of detection. The LOD was determined by testing multiple COVID-19 prepandemic negative serums. To establish the LOD, 12 prepandemic negative human serum samples, 28 naive rhesus monkey serum samples, 30 naive hamster serum samples, and 30 naive mouse serum samples were utilized to set the LOD. The LOD of the assay was determined as the mean of the negative serum samples plus the standard deviation (SD) on the assay nominal scale.
Statistical analyses. All statistical analyses were carried out in GraphPad Prism 8 with Student's t tests or one-way analysis of variance unless otherwise noted. Typically, data from at least three independent experiments were used for analyses.