Quantitation of SARS-CoV-2 neutralizing antibodies with a virus-free, authentic test

Manufacturing of SARS-CoV-2 VLPs

SARS-CoV-2 VLPs, termed S⁺ VLPs, were generated by transient co-transfection of expression plasmids encoding all four structural proteins of the virus: S (Wuhan-2019, D614G or B.1.617.2), M, N and E in defined stoichiometry into HEK293T cells. To trace the S⁺ VLPs, a fifth expression plasmid was co-transfected to express a chimeric reporter protein consisting of the human CD63 tetraspanin protein and, at its carboxy terminus, an activator of split-nano-luciferase (CD63∼HiBiT). Three days after transfection, assembled S⁺ VLPs were present in large quantities in the cell culture medium. Further purification and concentration were optional and applied if needed to characterize the S⁺ VLPs in detail.

cryoEM of SARS-CoV-2 VLPs

Cryo-electron microscopy (cryo-EM) of the S⁺ VLPs revealed spherically shaped vesicles in the range of 60 to 150 nm in diameter with a membrane consisting of an intact lipid bilayer (Fig. 1). Like SARS-CoV-2 virions (Ke et al., 2020; Liu et al., 2020; Yao et al., 2020), the VLPs displayed a characteristic corona of dense, needle-like radial proteins protruding perpendicularly from the membrane. On the distal ends of the protrusions spacious heads sit on slimmer stems, suggesting that these structures correspond to the viral glycoprotein spike of SARS-CoV-2. The shape and dimensions of the protrusions which are about 25 nm in length and have a stem width of 7 nm clearly support this assumption. In addition, elongated structures are observed on certain spike bearing particles (white arrows in Fig. 1). These structures might correspond to spike protein protrusions lying down on the vesicle surface, likely caused by surface tension effects prior to plunging of the sample in cryogen. Also, evaporation might reduce the height of the liquid film causing a partial air contact of the particle’s envelope and a redistribution and flattening of surface components in contrast to spikes from the periphery, which maintain their integrity in the surrounding liquid phase. Alternatively, these elongated structures may also correspond to some elements such as fibrous proteins present in the lumen of S⁺ VLPs.

• Download figure
• Open in new tab

Figure 1. Cryo-electron microscopy images of SARS-CoV-2 VLPs (S⁺ VLPs)

The images show different SARS-CoV-2 virus-like-particles (S⁺ VLPs) of approximately 60 to 150 nm in diameter recorded by cryo-electron microscopy (cryo-EM). The particles bear the characteristic corona of radial, dense spike-like proteins protruding from the envelopes’ intact lipid bilayers, which are characteristic for trimers of the viral glycoprotein of coronaviruses, spike (S), as observed for SARS-CoV-2 virions. The particle in the right panel shows elongated structures (white arrows), which might correspond to spike protein protrusions lying down on the vesicle surface, likely caused by surface tension effects prior to rapid freezing of the sample. White scale bar in all three panels corresponds to 100 nm in length.

SARS-CoV-2 virions contain a complex of ribonucleoproteins (N) and the ∼30 kb RNA genome, but S⁺ VLPs seemingly do not contain a similar luminal mass (Fig. 1) probably because the vRNA genome is absent. Together with N, the large vRNA molecule might act as a sizing factor, which could explain the variability in diameter seen in the S⁺ VLP preparations. Other than that, our S⁺ VLPs seem to mimic SARS-CoV-2 virions structurally (Fig. 1).

Molecular characterization of SARS-CoV-2 VLPs (S⁺ VLPs)

Spike, S, the viral fusion protein (FP) of SARS-CoV-2 is a highly glycosylated type I transmembrane protein which assembles as homotrimers. It encompasses the two domains S1 and S2, which are proteolytically separated by cellular furin protease. After furin cleavage, S1 and S2 remain non-covalently associated (Millet and Whittaker, 2015), but it appears as if furin cleavage is dispensable for infection (Papa et al., 2021).

Because of its central role in viral infection, the correct conformation of S is critical for the tropism and fusogenicity of both SARS-CoV-2 virions and S⁺ VLPs. We used two commercially available antibodies that recognize S1 or S2 together with 43A11, a new, in-house generated high affinity monoclonal antibody (Supplementary Fig. S1), which exclusively recognizes non-dissociated S (but not the single S1 or S2 domains; Fig. 2C) to visualize S1, S2, S full-length (S^FL) and higher order S^FL complexes in S⁺ VLP preparations under reducing and non-reducing conditions. By western blot (WB) analysis (Fig. 2), we confirmed the presence of S1 and S2 domains, S^FL, trimers of S^FL (S^FL₃ or S2₃S1₃) and other additional S complexes in our S⁺ VLPs preparations. Proteolytic cleavage by furin and possible subsequent dissociation of S1 during viral egress (Papa et al., 2021; Zhang et al., 2020), but also dimeric S^FL complexes might generate certain additional higher order complexes (Ou et al., 2020), which we also observed in Fig. 2. In S⁺ VLP preparations from HEK293T cells the majority of S is efficiently cleaved, presumably by furin (Fig. 2B, top panel, reducing conditions), but S1 and S2 domains remain largely complexed in single and higher order S^FL conformations (Fig. 2A,C, top panels, non-reducing conditions). The presence of S2, S^FL and S^FL₃ in lysates of S⁺ VLP producer cells was further confirmed with serum of a COVID-19 convalescent donor (Fig. 2D) and showed a pattern similar to the S2 specific commercial antibody (Fig. 2A, top panel, reducing conditions).

• Download figure
• Open in new tab

Figure 2. Spike western blot analyses of protein lysates from S⁺ VLPs and SARS-CoV-2 virus stock

Western blot analyses of S⁺ VLPs and extracellular vesicles (EVs) produced in or spontaneously released from HEK293T cells, and SARS-CoV-2 virus stock produced from infected Vero E6 cells are shown. Antibodies are directed against the S1 or S2 domains or recognize the intact, full-length (FL) spike molecule S^FL. The analyses confirm the presence of spike protein in various states in S⁺ VLPs and SARS-CoV-2 virion preparations but not in EVs which served as negative control.

(A, B) Commercially available S2 and S1 specific monoclonal respective polyclonal antibodies detect both spike domains in cell-free preparations of S⁺ VLPs as well as S^FL protein (top panels). The S2 domain specific antibody also detects trimeric S^FL (S^FL₃) and spike complexes of higher order under non-reducing (non-red.) conditions. In SARS-CoV-2 virus stock (bottom panels) the antibodies detect S^FL protein and the S1 domain in panels A and B but not the S2 domain.

(C) 43A11, an in-house developed, high affinity monoclonal antibody (Supplementary Fig. S1) recognizes full-length spike (S^FL) exclusively, confirming the presence of mono- and trimeric S^FL protein complexes in S⁺ VLPs (top panel) and SARS-CoV-2 virus stock (bottom panel).

(D) Serum from a convalescent COVID-19 patient contains antibodies directed against the S2 domain, S^FL and S^FL multimers in protein lysates from HEK293T cells transiently transfected with expression plasmids encoding the four structural components (S, M, N, E) of SARS-CoV-2.

Parallel to S⁺ VLPs, we analyzed a heat inactivated (56°C, 15 min) SARS-CoV-2 virus stock harvested from infected Vero E6 cells. Compared to S⁺ VLPs, the virus stock showed a similar S composition, but the S2 specific antibody recognized only S^FL, whereas the S1 antibody detected both S^FL and S1. Using the 43A11 monoclonal antibody (mAb), the SARS-CoV-2 virus stock was found to contain S in its distinct trimeric state, but also monomeric S^FL. Repeated passaging of SARS-CoV-2 on Vero E6 cells can lead to the loss of the furin cleavage site as has been previously reported (Turonova et al., 2020), which might explain the equal fraction of non-cleaved S (Fig. 2B, lower panel) and the absence of the separate S2 domain in the virus stock in Fig. 2A. We conclude that our S⁺ VLP preparations and an authentic SARS-CoV-2 virus stock are similar according to WB analyses but differ with respect to the fraction of furin cleaved S.

We also developed a highly sensitive sandwich enzyme-linked immunosorbent assay (ELISA) to characterize S⁺ VLP and the SARS-CoV-2 virus preparations and to quantify their S content. We used the mAbs 43A11 together with 55E10, a second in-house generated anti-S mAb (Supplementary Fig. S1), which recognize orthogonal, non-overlapping epitopes. As external reference we employed a commercially available recombinant S protein to obtain a calibration curve to assess the amount of S (Fig. 3A). The assay reliably detected S concentrations as low as 3 ng mL^-1 recombinant protein (Fig. 3A) as well as S protein in S⁺ VLP preparations (Fig. 3B), and was found to be highly specific when probed with control samples consisting of extracellular vesicles (EVs) without a viral fusion protein (Δ vFP EVs). Using this ELISA, we quantified our S⁺ VLP productions and heat inactivated SARS-CoV-2 virus stock (RT-qPCR, ct value of 15.3) and found about 626 ± 20 ng mL^-1 and 149 ± 6 ng mL^-1 S protein, respectively, according to the S protein standard (Fig. 3C).

• Download figure
• Open in new tab

Figure 3. Spike-specific, quantitative sandwich ELISA and nanoparticle tracking analysis (NTA) of S⁺ VLP preparations and heat-inactivated SARS-CoV-2 virus stock

Two in-house generated monoclonal antibodies (43A11 and 55E10), which exclusively recognize two orthogonal, non-overlapping epitopes in S^FL protein, were used to establish a sandwich ELISA (enzyme-linked immunosorbent assay) for the quantitation of spike protein in biological samples.

(A) Calibration of the spike-specific sandwich ELISA with a commercially available recombinant (rec.) S protein standard, which encompasses the extracellular domain (ECD) of S. The calibration curve of three independent replicates allows for calculating the amount of S protein in samples within the linear range of optical density (OD) values (0.7 ≤ OD ≤ 1.7, r² > 0.99). The detection limit of this assay was estimated to correspond to 3 ng mL^-1 recombinant S protein.

(B, C) Concentrated (conc.) and non-concentrated S⁺ VLPs from supernatants of transiently transfected HEK293T cells were analyzed for their amount of S^FL protein. Based on the linear regression function in panel A, the concentration (c) of S was calculated (ng mL^-1) of three technical replicates in the linear OD range and compared with inactivated SARS-CoV-2 virus stock with a known ct value (15.3) of its vRNA copies according to RT-qPCR. Controls are solvent (PBS) and extracellular vesicles without a viral fusion protein (Δ vFP EVs) harvested from cell culture medium of HEK293T cells transiently transfected with expression plasmids coding for M, N, E and CD63∼HiBiT but omitting S.

(D) Nanoparticle tracking analyses (NTA) of three independent preparations of unconcentrated and concentrated S⁺ VLPs (S, M, N, E, CD63∼HiBiT) and S⁺ EVs (S, CD63∼HiBiT omitting M, N and E) from cell culture medium of transiently transfected HEK293T cells are shown. For comparison, NTA data from heat-inactivated SARS-CoV-2 virus stock from infected Vero E6 cells are provided.

Particle analysis of SARS-CoV-2 VLPs

Next, we quantified the number of total physical particles in our samples via nanoparticle tracking analysis (NTA). Intriguingly, co-expression of M, N, E together with S increased the number of particles drastically compared to transfections of S alone (Fig. 3D), suggesting that SARS-CoV-2 S⁺ VLPs consisting of M, N, E and S evolve via a self-assembling mechanism and egress without the need of non-structural viral proteins, as described for SARS-CoV VLPs (Siu et al., 2008). NTA of our samples indicated that our S⁺ VLP preparations contained 9.1×10¹⁰ mL^-1 particles, while comparable preparations of EVs obtained from HEK293T cells after transfection with the S encoding expression plasmid (S⁺ EV in Fig. 3D) only yielded 5.0×10⁹ mL^-1 particles. The heat inactivated SARS-CoV-2 virus stock was determined to contain 7.2×10⁹ mL^-1 particles. Total particle counts included 1.5×10⁹ mL^-1 bovine EVs from fetal bovine serum (FBS) contained in cell culture medium. Bovine EVs corresponded to 2% and 30% of total particle numbers in S⁺ VLP and S⁺ EV preparations, respectively.

To obtain quantitative data of our S⁺ VLPs at the level of single particles, we developed a nano flow technique to assess the fraction of S-positive particles among all particles released by HEK293T cells. Particles were purified from cell culture supernatants and incubated with the dye CellTraceViolet (CTV; Thermo Fisher Sci.), which exhibits fluorescence upon enzymatic ester hydrolysis in the lumen of intact vesicles after membrane penetration. Subsequently, particles were stained with the fluorescently labeled anti-S mAb 43A11 and analyzed by our nano flow technology using a cytometer (CytoFLEX, Beckman Coulter). To distinguish instrument noise from particles we pre-gated on CTV⁺ events and SSC-H (Fig. 4A) and measured the fraction of S-positive particles eventually. S⁺ VLPs, i.e., particles obtained after transient transfection of HEK293T cells with the plasmids encoding S, M, N and E, constituted 37.5% of all CTV⁺ particles. The inactivated SARS-CoV-2 stock contained about 10% S-positive particles (Fig. 4B). Thermal treatment of the SARS-CoV-2 sample might have lowered the esterase activities in both virions and EVs, which might explain the generally lower fraction of CTV⁺ particles (0.11% of all events) in this SARS-CoV-2 virus stock (top panels in Fig. 4B). We therefore estimated the fraction of SARS-CoV-2 virions and S⁺ VLPs to be in a comparable range but S⁺ VLPs were abundant by a factor of three or more.

• Download figure
• Open in new tab

Figure 4. Nano flow technology of S⁺ VLPs and heat-inactivated SARS-CoV-2 virus stock

(A) HEK293T cells transfected with S, M, N, E, CD63∼HiBiT or with M, N, E, CD63∼HiBiT but without S yield S⁺ VLPs or control Δ vFP EVs, respectively. After two rounds of low-speed centrifugation cell culture supernatants containing either S⁺ VLPs or control Δ vFP EVs were stained with the membrane permeable dye CellTraceViolet (CTV), which exhibits fluorescence only upon its uptake followed by esterase activation within the lumen of intact membranous vesicles. A heat-inactivated SARS-CoV-2 virus stock was also stained with CTV for comparison. Subsequently, samples were counter-stained for the presence of surface spike protein using the monoclonal anti-S antibody 43A11. The samples were diluted and analyzed using a CytoFLEX LX flow cytometer (Beckman Coulter Life Sci.).

(B) Panels in the top row show all recorded events analyzed for their sideward scatter (SCC-H) using a violet excitation laser (y-axis) and CTV staining (x-axis). CTV⁺ events were gated as shown to identify sub-cellular, intact particles (S⁺ VLPs, Δ vFP EVs, SARS-CoV-2 virus) to distinguish them from instrument noise seen in the PBS control. CTV⁺ events were analyzed for their staining with the anti-S antibody 43A11 coupled to AlexaFluor488 (bottom row of panels). 37.5% S⁺ particles were identified in the preparation of S⁺ VLPs, 10% S⁺ particles were identified in SARS-CoV-2 virus stock and fewer than 1% in control Δ vFP EVs. The low fraction of CTV-positive events in SARS-CoV-2 virus stock (0.11%) compared with preparations of S⁺ VLPs (0.61%) and Δ vFP EVs (0.37%) might be the consequence of a reduced esterase activity in virions (and extracellular vesicles) after heat inactivation at 56°C for 15 min to inactivate viral infectivity.

Our preparations of S⁺ VLPs were found to contain 626 ng mL^-1 S protein on average (Fig. 3C) and a particle concentration of 9.1×10¹⁰ mL^-1 (Fig. 3D), and consisted of 37.5% S⁺ particles (Fig. 4B). Assuming even distribution of S molecules per S⁺ particles and the given molecular weight of 134.4 kDa for the truncated recombinant S protein standard, we calculated a theoretical number of 82 S molecules, corresponding to 27 trimers per S⁺ VLP. This number is in the range described by others for SARS-CoV-2 (Ke et al., 2020; Turonova et al., 2020; Yao et al., 2020). Regarding the inactivated SARS-CoV-2 stock, we determined 149 ng mL^-1 S protein, 7.2×10⁹ mL^-1 overall particles and 10% S⁺ particles, which would correspond to 311 S trimers per particle. This surprisingly high number is probably attributable to technical limitations of staining heat inactivated particles along the protocol described above and thus underestimating of the number of S⁺ particles.

Virus-free neutralization test

Towards a virus-free neutralization test we analyzed the capacity of S⁺ VLPs to fuse with appropriate target cells. First, we transduced various cell lines (HEK293T, LN18, A549, Huh7, Vero and U251MG) to constitutively express human ACE2 and validated its expression by WB. Of these cell lines, Huh7 and Vero cells had previously been used for infection with spike-pseudotyped retrovirus vectors (Yang et al., 2020). Recently, we identified the human U251MG cell line to take up EVs equipped with VSV-G very efficiently suggesting that these cells might also be suitable to act as S⁺ VLP recipient cells (Albanese et al., 2021).

Next, we engineered EVs to contain SARS-CoV-2 spike together with CD63∼BlaM (a chimer of human CD63 and β-lactamase), concentrated them via ultracentrifugation and incubated them with the panel of recipient cells as described above for 4 h. Upon uptake and fusion of these EVs with recipient cells, BlaM translocates to the cytoplasmic compartment of the recipient cells. Only BlaM⁺ cells loaded with an appropriate substrate (CCF4-AM, Thermo Fisher Sci.) convert CCF4-AM to fluorescent CCF4, which is retained in the cytosol and can be quantified via flow cytometry on single cell level as described in the context of HIV-1 (Cavrois et al., 2002), and later by us (Albanese et al., 2021) and others (Carlon- Andres and Padilla-Parra, 2020; Cavrois et al., 2014; Desai et al., 2017; Feeley et al., 2011; Jones and Padilla-Parra, 2016).

ACE2⁺ Vero cells took up S⁺ EVs but not control EVs lacking a viral fusion protein (Δ vFP EVs) (Supplementary Fig. S2A). Yet only about 21% of Vero cells turned BlaM-positive even with a high dose of S⁺ EVs, while 75% of all cells became positive with control EVs equipped with CD63∼BlaM and VSV-G (vesicular stomatitis virus glycoprotein G) as a viral fusion protein (VSV-G⁺ EVs; Supplementary Fig. S2B). In contrast, up to 97% ACE2⁺ U251MG cells became BlaM-positive with S⁺ EVs, while cells incubated with Δ vFP EVs remained BlaM negative (Supplementary Fig. S2C, D). The fusion of ACE2⁺ U251MG cells with S⁺ or VSV-G⁺ EVs was equally efficient in this setting (Supplementary Fig. S2C, D). To demonstrate the endosomal uptake and fusion of EVs with the recipient cells, we pre-treated ACE2⁺ U251MG cells with chloroquine prior to incubation with S⁺ EVs (Supplementary Fig. S2E). Chloroquine deacidifies endosomes and thus also inactivates the endosomal-pH-dependent cysteine protease Cathepsin L (CTSL), which primes S for SARS-CoV-2 entry in certain cell lines in vitro (Hoffmann et al., 2020a; Hoffmann et al., 2020b). Thus, delivery of BlaM was found to be strictly dependent on endosomal processing of S in recipient cells.

Next, we used ACE2⁺ U251MG cells in preliminary neutralization experiments to test and quantify the reduction of S-mediated fusion of BlaM equipped S⁺ EVs by neutralizing serum antibodies. Sera from SARS-CoV-2 vaccinees and COVID-19 convalescent patients displayed dose-dependent neutralization, while sera from healthy and naïve donors barely showed any effect (Supplementary Fig. S2E). We concluded that all steps of the S-mediated and ACE2-dependent cellular uptake of S⁺ EVs are highly reminiscent of SARS-CoV-2 infection.

Optimization of the VLP fusion assay

The applicability of the assay described above is limited as its readout relies on flow cytometry, requires an overnight incubation step for the intracellular accumulation of CCF4 and depends on concentrated S⁺ EVs preparations. We therefore developed this assay further by replacing BlaM with nano-Luciferase (nLuc). To avoid known background problems due to protein leakage of intact luminescent nLuc, we adapted its split variant consisting of an incomplete and inactive nLuc polypeptide (LgBiT) and the self-associating activator peptide of 11 amino acids (HiBiT) for our purposes (Dixon et al., 2016; Miyakawa et al., 2020; Yamamoto et al., 2019).

Similar to BlaM, we fused HiBiT to the C terminus of CD63 (CD63∼HiBiT) (Somiya and Kuroda, 2021) to be incorporated into S⁺ particles. As recipient cells, we engineered ACE2⁺ U251MG cells to constitutively express N-myristoylated LgBiT (NM∼LgBiT) as membrane-associated reporter enzyme. This system proved to be almost free of leakage as HiBiT and LgBiT are tightly associated with the cellular and EV membranes, respectively, and are not secreted in detectable amounts. In the split nLuc system, the turnover of a suitable substrate is only catalyzed upon successful intracellular reconstitution of both parts of nLuc (Dixon et al., 2016). As a consequence, this assay has a very low background, which otherwise is a major problem when working with fully active enzymes as protein reporters.

We also turned from S⁺ EVs to SARS-CoV-2 VLPs, i.e., S⁺ VLPs encompassing all four structural SARS-CoV-2 proteins, because co-expression of M, N and E together with S and CD63∼HiBiT (Fig. 5A) led to higher particle numbers and enhanced fusogenicity of VLPs with ACE2⁺ U251MG cells (Fig. 3D). We determined the optimal stoichiometry of plasmid DNAs in this practical, readout-based approach to obtain S⁺ VLPs which resemble SARS-CoV-2 in many aspects as documented by cryo-EM, WB, ELISA, NTA and nano flow technology as shown in Fig. 1 to 4. The engineered S⁺ VLPs enter ACE2⁺ recipient cells in a receptor-dependent manner, require correct processing of S by proteases and escape from endosomes via the post-fusion conformation of S, very reminiscent of infectious SARS-CoV-2 virions. Upon endosomal fusion the membrane anchored HiBiT is delivered into the cytoplasm of the LgBiT⁺ recipient cell, where the functional enzyme reconstitutes in situ to support substrate turnover and emission of light as shown schematically in Fig. 5C. A scientific animation demonstrates this simple but very efficient principle (https://youtu.be/6wckXobT_bM).

• Download figure
• Open in new tab

Figure 5. VLP neutralization test (VLPNT)

(A) Engineered VLPs were generated in vitro by transient co-transfection of HEK293T cells with an optimized ratio of expression plasmids encoding the four SARS-CoV-2 structural proteins S, M, N, E and a chimeric membrane anchored activator peptide (CD63∼HiBiT). The resulting particles were termed S⁺ VLPs and obtained from conditioned cell culture medium three days after DNA transfection.

(B) Schematic view of a SARS-CoV-2 virion with the four structural proteins S, M, N and E and the viral genome of positive sense, single-stranded RNA complexed with N.

(C) Basic steps of VLP entry and reconstitution of nano Luciferase (nLuc). Similar to infection with SARS-CoV-2, spike, the trimeric viral fusion protein in the envelope of S⁺ VLPs (Fig. 1) mediates attachment to the host cell receptor ACE2, triggering endocytosis and proteolytic processing by TMPRSS2 or CTSL. The ensuing fusion of the S⁺ VLP envelope with cellular membranes exposes the HiBiT activator peptide to make contact with N-myristoylated LgBiT (NM∼LgBiT), which is stably expressed in the cytoplasm of the ACE2⁺ target cell. Upon in situ reconstitution of the functional nano Luciferase (nLuc*) reporter addition of substrate will induce bioluminescence, which can be quantified in a standard luminometer in 96-or 364-well cluster plates.

(D) To test body fluids for the content of neutralizing SARS-CoV-2 antibodies (NAbs), S⁺ VLPs are pre-incubated with serial dilutions (on the basis of 2) of the samples for 30 min. Suitable medical samples are sera of COVID-19 patients, vaccinated or naïve individuals or other body fluids such as saliva or nasal excretions. SARS-CoV-2 NAbs will interfere with all steps of S⁺ VLP attachment to ACE2, receptor-mediated intake, endosomal fusion of the VLP envelope with the endosome and escape to the cytoplasm. Target cells are U251MG cells engineered to express both ACE2 and NM∼LgBiT (LgBiT). Upon encounter with S⁺ VLP-borne CD63∼HiBiT, NM∼LgBiT is reconstituted into a fully functional nLuc reporter enzyme, which can be quantitated. Neutralizing SARS-CoV-2 antibodies reduce or even block the delivery of the CD63∼HiBiT activator entirely, which can be quantified in a standard clinical laboratory with aid of a luminometer and within 4.5 hours. A freely accessible scientific animation narrates the principle of the VLPNT (https://youtu.be/6wckXobT_bM).

Analogous to the EV system with BlaM as reporter protein, we assessed whether S⁺ VLPs fuse exclusively with susceptible ACE2⁺ cells via the viral entry factor S. We generated S⁺ VLPs (S, M, N, E, CD63∼HiBiT)⁺, VSV-G⁺ EVs (VSV-G, CD63∼HiBiT)⁺ and Δ vFP EVs (M, N, E, CD63∼HiBiT)⁺, incubated them with ACE2⁺ or ACE2^- NM∼LgBiT⁺ U251MG cells for 4 h and quantified their fusion with the different target cells. As expected, S⁺ VLPs fused exclusively with ACE2⁺ cells but not with ACE2^- cells (t-test, p<0.0001) (Fig. 6). Furthermore, fusion relied strictly on the presence of a viral fusion protein such as S because Δ vFP EVs were barely taken up (t-test, p<0.0001). In contrast, VSV-G⁺ EVs fused with ACE2⁺ and ACE2^- cells at similar levels due to VSV-G’s broad tropism (t-test, p value non-significant) (Fig. 6). We conclude that this system meets all requirements of a S⁺ VLP-based neutralization test.

• Download figure
• Open in new tab

Figure 6. Specificity and tropism of S⁺ VLPs

U251MG cells were transfected to express NM∼LgBiT constitutively. In addition, we established a derivative of these NM∼LgBiT cells with the human ACE2 receptor to obtain both ACE2-negative (ACE2^-) and positive (ACE2⁺) target cells. The two cell lines were incubated with S⁺ VLPs carrying CD63∼HiBiT, with EVs without a viral fusion protein (ΔvFP EVs) obtained from supernatants of HEK293T cells after transient transfection of expression plasmids encoding M, N, E and CD63∼HiBiT (but not S) or with EVs from HEK293T cells after transient transfection of expression plasmids encoding CD63∼HiBiT and protein G of the vesicular stomatitis virus (VSV-G⁺ EVs). The specificity of spike-mediated, ACE2-dependent fusion of all three particle classes was validated measuring luciferase activities upon reconstitution of the split nano-luciferase in the indicated cell types. The results document the high specificity of S⁺ VLPs to fuse exclusively with ACE2⁺ cells, while Δ vFP EVs yielded background signals, only. Both cell lines were equally positive when incubated with control VSV-G⁺ EVs indicating their ACE2 receptor independent uptake and fusion, reflecting the broad tropism of VSV. Data are based on at least four independent experiments. p-values of three independent t-tests are indicated (ns; not significant).

VLP neutralization test (VLPNT)

S-specific NAbs can prevent entry of SARS-CoV-2 virions into host cells and thus protect from viral infection via diverse modes of actions. Most antibodies block the attachment of S to ACE2 receptors by binding to the receptor binding motif (RBM) of S1 (Niu et al., 2021; Piccoli et al., 2020) or stall S in its closed, i.e., receptor binding domain (RBD) “down” pre-fusion conformation (Tortorici et al., 2020). Yet, certain S-specific NAbs also neutralize without disrupting the ACE2 interaction (Brouwer et al., 2020). Possible other mechanisms include the inhibition of proteolytic processing of S by, e.g., TMPRSS2 or CTSL or the interference with the heptad repeats or glycosylated surfaces in S2, which are required to promote the fusion of the viral envelope with the endosomal membrane as described in the context of SARS-CoV or MERS-CoV (Keng et al., 2005; Shanmugaraj et al., 2020). NAbs with their multiple mechanisms to disrupt S functions also reduce, interfere or even block S⁺ VLPs and the delivery of CD63∼HiBiT to susceptible target cells. As a result, reduction of luminescence from reconstituted nLuc might likely correlate with SARS-CoV-2 neutralization (Fig. 5D).

Towards a VLP-based virus neutralization test (VLPNT), we incubated a defined amount of S⁺ VLPs with serial dilutions (starting from 1:10 to 1:1,800) of sera from acute or convalescent COVID-19 patients, COVID-19 vaccinees or healthy, naïve donors and quantified the resulting dose-dependent neutralization as shown in Fig. 7A. Mean luminescence level of S⁺ VLPs, only, was set to 0% neutralization; while background luminescence obtained with Δ vFP EVs was set to 100% neutralization. The half maximal neutralization titer was determined by extrapolating sigmoidal curve values that correspond to 50% signal reduction after background correction. This value was termed VLPN₅₀. It is considered equivalent to VNT₅₀ and PRNT₅₀ of conventional VNTs (cVNTs) and plaque reduction neutralization tests (PRNTs), respectively.

• Download figure
• Open in new tab

Figure 7. Correlation of VLPNT and cVNT data and VOC cross-neutralization using sera from COVID-19 patients and vaccinees

(A) VLPNT neutralization data with dilutions of sera obtained from three individuals (a naïve healthy donor, the COVID-19 patient A12 and a BNT162b2 vaccinee) and the WHO reference serum (NIBSC 20/136) as international standard are shown. The graphs are examples and include mean neutralization results from three independent biological replicates. Serum dilution which resulted in half maximal signal reduction, equivalent to 50% neutralization, was termed VLPN₅₀ titer. A recombinant human mAb, REGN10987 with known neutralizing activity showed potent neutralization in the assay at low molar concentration (c), while an IgG isotype control mAb did not show any neutralizing activity.

(B) Validation of the VLPNT by correlating VLPN₅₀ titers versus VNT₁₀₀ titers obtained in a cVNT, the gold standard using infectious SARS-CoV-2 virus. Pearson correlation data (sample size n, coefficient r and p-value) of 23 sera from confirmed COVID-19 patients are shown and the linear relationship is indicated. The sample denoted A3 is a single serum sample referred to in the Result section. Results below the dotted horizontal line denote sera which scored negative in the VLPNT. Results left of the dotted vertical line denote sera which scored below the limit of detection (1:8) in the cVNT; these VNT₁₀₀ values were defined as 1:4 and indicated by square brackets.

(C) VLPNT with a SARS-CoV-2 variant of concern (VOC) compared with Wuhan-2019 D614G. S⁺ VLPs were harvested from supernatants of HEK293T cells transiently transfected with expression plasmids encoding either Wuhan-2019 D614G or Delta mutant B.1.617.2 S protein together with M, N, E, and CD63∼HiBiT. NAbs in sera of 13 COVID-19 vaccinees were cross-neutralizing, but less potent in neutralizing the δ-mutant B.1.617.2 compared with the Wuhan-2019 D614G mutant. Data were statistically analyzed using the paired t-test.

Sera of naïve donors resulted in no to little signal reduction, i.e., weak neutralization even at high serum concentrations. Of 12 sera from healthy, naïve donors obtained from mid 2019 and earlier, most did not reach 50% neutralization and a few showed VLPN₅₀ titers of 1:16 or lower (Supplementary Fig. S3A). Sera of 13 COVID-19 vaccinees after prime-boost immunization from 2021 showed high neutralization potential consistently up to 100% inhibition of S⁺ VLPs fusion and VLPN₅₀ titers of 1:27 to 1:352 (Supplementary Fig. S3A). Data on individual titers and types of vaccines which the donors received can be found in Supplementary Fig. S3D and Supplementary Tab. S1. Based on the maximal titers of naïve sera in our VLPNT, we added an additional margin and defined a preliminary minimal cutoff of 1: ≥25 to classify samples with neutralizing activities.

As there is considerable demand for global harmonization and standardization of NAb titers obtained from different laboratories using different versions of SARS-CoV-2 VNTs (Kristiansen et al., 2021), the World Health Organization (WHO) supplies laboratories with a standard plasma pooled from 11 British patients, who recovered from COVID-19 (Mattiuzzo et al., 2020, WHO/BS/2020.2403). We applied this standard (NIBSC 20/136) with a defined neutralization activity of 1000 International Units per mL (IU mL^-1). VLPN₅₀ values of multiple individual experiments with the WHO standard were obtained (Supplementary Tab. S1) and used to convert our serum VLPN₅₀ titers to harmonized titers expressed as IU mL^-1 as shown in Supplementary Fig. S3A, right panel. In six independent VLPNT experiment with the WHO standard, we calculated a coefficient of variation (CV) for log₁₀ transformed titers of CV = 0.04 and CV = 0.23 for untransformed titers, for the within-laboratory repeatability and the resulting imprecision of quantification, respectively. Clearly, additional experiments and standardizations have to be conducted to generate data for an assessment according to the Clinical Laboratory Standards Institute (CLSI) guidelines (Chesher, 2008).

We also used our VLPNT to test two potent neutralizing recombinant mAbs, REGN10987, imdevimab and REGN10933, casirivimab, which have been approved under emergency use authorization for the treatment of COVID-19 by the Food and Drug Administration (FDA) (2020; Baum et al., 2020; DeFrancesco, 2020). Their VLPN₅₀ were found to be 1.8 nM and 7.3 nM, respectively, while a human IgG isotype control mAb did not show any neutralization (Fig. 7A).

Based on these results we concluded that the VLPNT qualifies for the measurement of neutralizing activities in clinical samples.

VLPNT and cVNT compared

For the validation of our VLPNT we compared the VLPN₅₀ titers of a defined set of sera with results from a cVNT using infectious SARS-CoV-2, the ‘gold standard’ in the field (von Rhein et al., 2021). Towards this end, we tested 23 well documented serum samples from patients with confirmed SARS-CoV-2 infection under double-blinded, randomized conditions in our VLPNT. Relevant details about their clinical status, PCR results, SARS-CoV-2 strains and neutralizing serum titers data had been collected prior to our analyses (Supplementary Tab. S1).

When analyzed in our test, serum VLPN₅₀ titers of the 23 COVID-19 sera varied from 1:18 to 1: >2,000 and therefore scattered wider than serum titers from vaccinees, even though the medians were similar (Supplementary Fig. S3A, left panel). Data on the individual titers and the patients’ clinical symptoms can be found in Supplementary Fig. S3C and Supplementary Tab. S1. Due to the previously defined cutoff for the VLPN₅₀ of 1: ≥25 (Supplementary Fig. S3A, left panel), a single COVID-19 sample (sample A3) was classified as non-neutralizing even though the highest titer of a naïve, SARS-CoV-2 negative serum was found to be 1:16 (Supplementary Fig. S3A).

The 23 COVID-19 samples were also tested in a cVNT with replication competent SARS-CoV-2 virus. The titers were determined based on 100% reduction of CPE (VNT₁₀₀) and their values ranged from 1:8 to 1: >1024 (Fig. 7B). In the cVNT, 17 out of 23 samples were found to have neutralizing activity with VNT₁₀₀ 1: ≥8, while the 6 remaining sera performed below the limit of detection (LOD) and were set to 1:4 per definition. A comparison of the 23 COVID-19 samples in both tests found 16 sera to be concordant positive (CP), none was concordant negative (CN), whereas 7 sera were found to be discrepant (D) (Supplementary Fig. S3B, left panel). This discrepancy likely resulted from experimental differences between the two tests (Khoury et al., 2020). The cVNT ‘gold standard’ relies on 100% reduction of CPE (VNT₁₀₀) while the VLPNT scores at 50% reduction of S⁺ VLP fusion (VLPN₅₀) suggesting that the latter test might be more sensitive to identify serum samples with weakly neutralizing activities.

Excluding the 6 sera tested in the cVNT with VNT₁₀₀ values below LOD (samples in brackets in Fig. 7B), we analyzed the performance of the VLPNT using the 17 COVID-19 samples which scored positive in the cVNT along with 12 sera from healthy, naïve donors from mid 2019 and earlier as controls (Supplementary Fig. S3B, right panel). Of the 17 COVID-19 samples, the VLPNT classified one sample to be false negative, while all 12 healthy naïve sera were considered true negative. Therefore, the sensitivity of our VLPNT was calculated to be 94%, while its specificity was 100% (n = 29). Based on the small sample number the positive predictive value (PPV) was calculated to be 100% suggesting that our SARS-CoV-2 VLPNT has the potential to become a reliable diagnostic tool.

We also compared the VLPNT with the cVNT by correlating the VLPN₅₀ titer results of all 23 COVID-19 samples with the respective VNT₁₀₀ titers, regardless of whether they were considered CP, CN or D (Fig. 7B). The Pearson coefficient of log₁₀ transformed data (n = 23) revealed a highly significant positive correlation (r = 0.917 and p = 7.9×10⁻¹⁰). Given the very strong correlation, the VLPNT is not only a qualitative test for the in vitro diagnostics of NAbs but also yields reliable titers of NAbs faithfully reflecting titers obtained in a cVNT with infectious SARS-CoV-2.

Test adaptation to Delta-SARS-CoV-2 (B.1.617.2)

Finally, we adapted the VLPNT to test and compare the currently predominant variant of concern (VOC) of SARS-CoV-2 by mutating S (D614G) to S (B.1.617.2, δ-or Delta variant) and generated S⁺ (B.1.617.2) VLPs. We re-tested all 13 sera of SARS-CoV-2 vaccinees (Supplementary Fig. S3D) in the VLPNT with S⁺ (B.1.617.2) VLPs, determined their VLPN₅₀ titers and compared them to titers obtained with the D614G S variant (Fig. 7C). Pearson coefficient of log₁₀ transformed data (n = 13) was calculated and showed strong positive correlation (r = 0.911; p = 1.5×10⁻⁵). Apparently, NAbs induced by different SARS-CoV-2 vaccines based on Wuhan-2019 spike cross-neutralized VOC B.1.617.2, but neutralization titers were significantly lower as shown by a paired t-test of log₁₀ transformed data (p = 0.0002, n = 13). While the mean titer for D614G S⁺ VLPs was found to be 1:135, the mean titer of 1:80 obtained with B.1.617.2 S⁺ VLPs was reduced by a factor of 0.59. A reduced capacity of convalescent or vaccinee sera to cross-neutralize B.1.617.2 has been demonstrated previously in a pVNT (Liu et al., 2021a), but also in a cVNT (Planas et al., 2021). Thus, our finding appears to be consistent with published work. The successful adaptation of the VLPNT to a prevalent VOC indicates that our test is flexible and versatile to address and answer questions on cross-neutralizing antibodies between existing and future SARS-CoV-2 variants of concern including the currently spreading B.1.1.529 VOC, named Omicron.

Cell lines and cell culture

HEK293T (CID3915); HEK293, S⁺ (CID4618); U251MG, hACE2⁺ (CID4663); U251MG, hACE2^+, NM∼LgBiT⁺ (CID4697) and Vero, hACE2⁺ (CID4608) cells were maintained in DMEM medium (Gibco, Thermo Fisher Scientific), supplemented with 8% FBS (AC-SM-0143, Anprotec) and Pen/Strep (Gibco, Thermo Fisher Scientific). Cells were cultivated at 37°C in a water-saturated atmosphere with 5% CO₂. Cell viability was monitored by trypan blue exclusion and cultures with more than 95% viable cells were used in all experiments. Cell lines constitutively expressing S, hACE2 or NM∼LgBiT were generated by retroviral transduction or transient transfection of parental cells with expression plasmids followed by selection with the respective antibiotic and flow cytometric cell sorting.

SARS-CoV-2 VLPs and other engineered EVs

To generate S⁺ VLPs containing appropriate levels of SARS-CoV-2 spike (D614G or B.1.617.2) and control-EVs, HEK293T cells were transfected using TransIT-293 (MIR2700, Mirus) according to the manufacturers protocol with carefully adjusted ratios of codon-optimized expression vectors coding for S, VSV-G, or mock together with CD63∼HiBiT or CD63∼BlaM and, optionally, M, N and E. The expression plasmids are termed p7413.1 (S:D614G) or p7487.IA1 (S:B.1.617.2), p5451 (VSV-G), p5025 (mock), p7447.SA11 (CD63∼HiBiT), p7200.2 (CD63∼BlaM), p7395.LA3 (M), p7396.NA9 (E), p7391.MA5 (N), respectively.

After 72 hours, S⁺ VLPs, VSV-G⁺ EVs or control EVs were harvested from cell culture supernatants (DMEM, 8% FBS, supplemented with Pen/Strep, Gibco, Thermo Fisher Scientific) after low-speed centrifugation at 7°C for 10 min at 300×g and 20 min at 4200×g and generally used without further processing. If necessary and as indicated, S⁺ VLP and EV supernatants were concentrated by ultrafiltration with 100 kDa cutoff or sedimented by ultra-centrifugation at 100,000×g, 4°C, 2 h followed by floatation in a discontinuous iodixanol (Optiprep, Sigma-Aldrich) density gradient (bottom to top: 1.75 mL sample with 34% final Optiprep concentration; 1.65 mL 30% Optiprep in PBS; 0.6 mL PBS) and ultra-centrifugation at 160,000×g at 4°C for 4 h. Collected fractions were washed with PBS by ultra-centrifugation at 100,000×g, 4°C, 2 h, and resuspended. Particle preparations were flash frozen in liquid nitrogen for storage at −80°C and subsequently characterized by WB, NTA, nano flow technology, ELISA and cryo-EM.

Western blot

Concentrated S⁺ VLPs or EVs were lysed in non-reducing 5× Laemmli buffer and separated by SDS-PAGE. Proteins were transferred to nitrocellulose membranes (GE Healthcare Life Science) by semi-dry blotting (Bio-Rad) and membranes were blocked for 1 h in 5% (w/v) non-fat milk in ddH₂O at room temperature (RT). Membranes were incubated at 7°C overnight with primary anti-spike antibody (43A11, rat IgG, Helmholtz Zentrum Mu□nchen; 1A9, mouse IgG, GeneTex; PA5-81795, polyclonal rabbit antibody, Invitrogen) at 1:2000 dilution in 5% (w/v) non-fat milk (Carl Roth) in ddH₂O, washed 3 times in TBST (Tris-buffered saline with 0.1% Tween-20) and incubated at RT for 1 h with horseradish peroxidase (HRP)-conjugated secondary antibody (1112-035-062, goat anti-rat IgG, Jackson Immuno Research Europe; 7076S, horse anti-mouse IgG, Cell Signaling; 7074S, goat anti rabbit IgG, Cell Signaling) at 1:20,000, 1:2000 and 1:2000 dilution respectively, in 5% (w/v) non-fat milk in PBST (PBS, 0.05% Tween-20). After repeated (3×) washing in TBST, blots were incubated with ECL reagent (GE Healthcare) and imaged using a Fusion FX (Vilber).

ELISA

For the quantitation of spike in samples from various sources, a sandwich enzyme-linked immunosorbent assay (ELISA) was developed, using two anti-spike antibodies with non-overlapping epitopes. First, wells of a Nunc MaxiSorp plate (Thermo Fisher Scientific) were coated at RT for 5 h with 2 µg mL^-1 anti-spike capture antibody (55E10, rat IgG, Helmholtz Zentrum Mu□nchen) or isotype in PBS. After washing with PBST (PBS, 0.05% Tween-20), free binding sites were blocked at RT for 2 h in 5% (w/v) non-fat milk (Carl Roth) in PBS. Samples of recombinant spike protein (S1+S2 extracellular domain, 40589-V08B1, Sino Biological), S⁺ VLPs, or controls were diluted as indicated (Fig. 3A, B) in PBS and incubated at 7°C for 16 h on the antibody coated plate. After washing, HRP conjugated anti-spike detection antibody (43A11, rat IgG, Helmholtz Zentrum Mu□nchen) diluted 1:500 in 5% (w/v) non-fat milk in PBS) was added at RT for 2 h, washed again and incubated at RT with 100 µL TMB substrate reagent (BD555214, Becton Dickinson). The reaction was stopped by adding 50 µL 1 M H₂SO₄ and absorbance was measured at 450 nm in a CLARIOstar Plus (BMG Labtech). Data analysis was performed with GraphPad Prism 9.2.

NTA Analysis

Nanoparticle tracking analysis (NTA) was performed with the ZetaView PMX110 instrument (Particle Metrix) and the corresponding software (ZetaView 8.04.02) was used to measure the number and the size distribution of the S⁺ VLP, virus stock and EV preparations. Samples were diluted in filtered PBS to achieve a vesicle concentration of approximately 1×10⁷ mL^-1. Pre-acquisition parameters were set to a sensitivity of 75, a shutter speed of 50, a frame rate of 30 frames per second and a trace length of 15. The post-acquisition parameters were set to a minimum brightness of 20, a minimum size of 5 pixels and a maximum size of 1000 pixels.

Analysis of S⁺ VLP, SARS-CoV-2 and EV particles with nano flow technology

Single particles were analyzed using our nano flow technology and a CytoFLEX LX cytometer (Beckman Coulter Life Science) with samples of S⁺ VLPs, SARS-CoV-2 stock and EVs pre-stained with CellTraceViolet (CTV, Thermo Fisher Scientific) at 37°C for 20 min and quenched with PBS with 1% BSA. Upon washing via a 100 kDa cutoff Amicon ultra filter, the samples were stained with AlexaFluor488 conjugated anti-S antibody (43A11, rat IgG, Helmholtz Zentrum Mu□nchen), diluted and afterwards analyzed in the flow cytometer.

Cryo-electron microscopy

Aliquots of 4 μL from crude or purified S⁺ VLPs were deposited on EM grids coated with a perforated carbon film. After draining the excess liquid with a filter paper, grids were quickly plunged into liquid ethane cooled by liquid nitrogen, using a Leica EMCPC cryo-chamber, and stored in cryo-boxes under liquid nitrogen until use. For cryo-EM observation, grids were mounted onto a Gatan 626 cryoholder and transferred to a Tecnai F20 microscope (FEI, USA) operated at 200 kV. Images were recorded with an Eagle 2k CCD camera (Thermo Fisher, USA).

EV fusion assay with BlaM readout

For EV fusion or EV neutralization assays with cytometric readouts, 2×10⁴ Vero, hACE2⁺ or U251MG, hACE2⁺ cells (CID4608, CID4663, Helmholtz Zentrum München) in 100 µL DMEM (8% FBS, Pen/Strep supplemented) were seeded per well in a 96-well plate (353072, Falcon) and incubated o/n at 37°C, 5% CO₂. Where indicated, the cells were pre-incubated at 37°C for 1 h with chloroquine or were left without further treatment. The spent medium was replaced with up to 100 µL concentrated EVs (S⁺, mock or VSV-G⁺, CD63∼BlaM⁺) with or without prior pre-incubation with serum samples. After 4 h incubation at 37°C, 5% CO₂, the supernatant was removed, adherent cells were washed in PBS, trypsinized and moved to a V-bottom shaped 96-well plate and sedimented by centrifugation. After washing the cells once in CO₂ independent medium (8% FBS, Pen/Strep supplemented, Gibco, Thermo Fisher Scientific), 50 µL staining solution were added and the cells were incubated at RT in the dark o/n. The staining solution was composed of 2 µL CCF4-AM, 8 µL solution B (K1095, Thermo Fisher Scientific) and 10 µL 250 mM Probenecid (P8761, Sigma) per 1 mL supplemented CO₂ independent medium. After enzymatic turnover of the CCF4, cells were washed once with PBS (+3% FBS) and analyzed by flow cytometry using an LSR Fortessa instrument (BD) with a high throughput auto sampler. The 409-nm wavelength laser (violet) was used for excitation of the FRET substrate, resulting in emission of intact, non-cleaved CCF4 substrate at 520 nm (green), whereas emission of cleaved CCF4 substrate was detected at 447 nm (blue).

VLPNT with nanoLuciferase readout

For virus-like-particle neuralization tests (VLPNTs), 2×10⁴ U251MG (hACE2⁺, NM∼LgBiT⁺) cells (CID4697, Helmholtz Zentrum München) in 100 µL DMEM (8% FBS, Pen/Strep supplemented) were seeded per well in a 96-well Lumitrac200 (655075, Greiner Bio-One) plate and incubated at 37°C, 5% CO₂ overnight. The following day, serial dilutions of serum samples by a factor of 2 starting at 1:2 to 1:360 were prepared in PBS in a 96-well plate. 12 µL of the serial serum dilutions each were added to 48 µL S⁺ VLP (S, M, N, E, CD63∼HiBiT) to obtain a series of final dilutions ranging from 1:10 to 1:1800. S⁺ VLPs and serum dilutions were incubated at 37°C for 30 min in a 96-well plate. Thereafter, the mix was transferred to recipient cells after removal of their culture medium and the cells were incubated at 37°C, 5% CO₂ for 4 h. Prior to readout, the supernatant was removed and replaced with 25 µL substrate mix (20 µL OptiMEM, Gibco, Thermo Fisher Scientific +5 µL nano-Glo diluted 1:20 in LCS buffer, N2012, Promega). Bioluminescence was immediately quantified in a CLARIOstar Plus reader (BMG Labtech). Mean luminescence level of S⁺ VLPs, only, was set to 0% neutralization; while background luminescence obtained with Δ vFP EVs was set to 100% neutralization. The half maximal neutralization titer was determined by extrapolating sigmoidal curve values that correspond to 50% signal reduction after background correction. This value was termed VLPN₅₀. The WHO international standard (NIBSC 20/136) was used in between runs and as within-run reference.

Anti-spike antibody generation

To obtain spike antigen in its authentic conformation for the immunization of rats, a constitutive spike expressing HEK293 cell line (CID4618, Helmholtz Zentrum Mu□nchen) expressing SARS-CoV-2 spike^FL (D614G) was established. To boost expression levels further, the cells were transiently transfected with an S expression plasmid (p7413, Helmholtz Zentrum Mu□nchen) prior to the isolation of S⁺ EVs from conditioned media three days after transfection. EVs were purified and concentrated by ultrafiltration, density gradient ultra-centrifugation and subsequently characterized as described above. For the generation of anti-spike antibodies, S⁺ EVs were used to immunize rats in a prime-boost-boost scheme, before splenic B-cells were isolated to establish hybridoma cell clones. Specific clones were subsequently identified by screening of hybridoma supernatants for binding to the constitutive spike expressing HEK293 cell line (CID4618, Helmholtz Zentrum Mu□nchen) via flow cytometry. The resulting anti-spike antibodies (COVEV: 43A11 and 55E10) were then characterized thoroughly to validate their high affinity and specificity for spike^FL.

Antibody production

Antibodies 43A11 and 55E10 were purified from hybridoma supernatants (Helmholtz Zentrum Mu□nchen Core Facility Monoclonal Antibodies). Antibodies REGN10987 and REGN10933 were purified from CHO cells transiently transfected with expression plasmids encoding the heavy and light chains. Briefly, cell culture supernatant was loaded on a protein GammaBind Plus Sepharose (17088602, Cytiva) or HiTrap MabSelect SuRe (29049104, Cytiva) column, washed with PBS and antibodies were eluted at pH 2.7 or 3.5, respectively in citric acid buffer. Fractions were pooled, neutralized with 1 M Tris (pH 8.0) and separated on Superdex200 prep grade (17104301, Cytiva) in PBS for monomers or buffer exchanged to PBS via ultra-filtration, respectively. Directly coupled AlexaFluor488 labelled antibodies were generated using commercial kits (A10235, Thermo Fisher Scientific).

SARS-CoV-2 pseudotyped virus neutralization test (pVNT)

S-pseudotyped retrovirus stocks were generated by co-transfecting expression plasmids coding for S, gag and pol (from murine leukemia virus) and a myeloproliferative sarcoma virus (MPSV) based retroviral vector encoding eGFP (p7413, p4037, p6895, Helmholtz Zentrum Mu□nchen) into HEK293T cells. After 48 h, the supernatant was harvested and centrifuged at 7°C for 10 min at 300×g, 20 min at 4200×g and flash frozen for storage (−80°C). For pVNTs, 5×10³ Vero (hACE2⁺) cells (CID4608, Helmholtz Zentrum München) in 100 µL DMEM (8% FBS, Pen/Strep supplemented) were seeded per well in a 96-well plate (353072, Falcon) and incubated at 37°C, 5% CO₂ overnight. The cells were then pre-incubated for 1 h at 37°C in 50 µL, 8 µg mL^-1 protamine sulfate in DMEM with or without chloroquine and retroviral stocks were pre-incubated for 30 min at 37°C with 5× serum dilutions in PBS. 50 µL of this mix where then added to the pre-incubated cells and centrifuged for 90 min at 800×g, at 37°C. After 48 h at 37°C, GFP expression was quantified by flow cytometry using an LSR Fortessa instrument (BD).

SARS-CoV-2 neutralization test (VNT₁₀₀)

Human sera were heat-inactivated at 56°C for 30 min and diluted in a series of two-fold dilution steps (1:4 to 1:512). 100 plaque-forming units (PFU) of SARS-CoV-2 stock (German isolate BavPat1/2020; European Virus Archive Global # 026 V-03883, GenBank: MZ558051.1) contained in 50µL was added to an equal volume of diluted serum. The mixture was incubated at 37°C and approximately 20,000 Vero C1008 cells (ATCC, Cat#CRL-1586) were added after 1 hour. After a four-day incubation period, the cytopathic effect (CPE) was evaluated by light microscopy. Virus neutralization was defined as the complete absence of CPE in a given serum dilution (VNT₁₀₀). The reciprocal geometric mean titer (GMT) was calculated from the highest serum dilution without CPE based on three replicates. The lower detection limit of the assay was 1:8, corresponding to the first dilution of the serum tested. Two positive controls were used as inter-assay neutralization standards and quality control for each test. The WHO international standard (NIBSC 20/136) was tested seven times, resulting in a GMT of 377 for 100% absence of CPE in this test. Neutralization assays were performed in the BSL-4 laboratory of the Institute of Virology at Philipps University Marburg, Germany (Romero-Olmedo et al., 2021).

Patients and specimens

We included 23 serum specimens collected between April 6, 2020, and June 25, 2020, from 18 patients infected with SARS-CoV-2 at the LMU Klinikum, Munich, Germany. Patients are part of the COVID-19 Registry of the LMU Klinikum (CORKUM, WHO trial id DRKS00021225) and the study was approved on March 23, 2020 by the ethics committee (no. 20-245) of the Faculty of Medicine of the LMU (Ethik-Kommission bei der Medizinischen Fakultät der Ludwig-Maximilians-Universität München, Pettenkoferstr. 8a, 80336 München, Germany). Clinical data were obtained from health records and all patient data were anonymized for analysis. All patients were tested positive for SARS-CoV-2 by rRT-PCR in nasopharyngeal or oropharyngeal swabs. The median age of the 18 COVID-19 patients was 60 years (interquartile range 52 to 75 years), and 11.1% (2/18) of these individuals were female. We categorized the disease severity of the COVID-19 patients according to the WHO guideline “Clinical Management of COVID-19” (World Health Organization, 2020): asymptomatic (no clinical signs of infection), mild (symptomatic patients without evidence of viral pneumonia or hypoxia), moderate (clinical signs of pneumonia, including fever, cough, dyspnoea), severe (clinical signs of pneumonia, plus one of the following: respiratory rate >30 min^-1, severe respiratory distress, SpO₂ <90% on room air), critical (one of the following: acute respiratory distress syndrome, sepsis, septic shock). Three patients were categorized as asymptomatic, 2 patients as mild, 6 patients as moderate, 5 patients as severe, and 2 patients as critical.

Virus stock preparation

CaCo-2 cells (American Type Culture Collection, ATCC, Virginia, USA) in cell culture medium (Dulbecco’s Modified Eagle’s Medium containing 2% FBS) were challenged for 2 h with a clinical isolate of SARS-CoV-2 (GISAID EPI ISL 2967222) previously obtained from a nasopharyngeal swab of a COVID-19 patient. Subsequently, cell culture medium was exchanged, and three days post infection supernatants were passaged on Vero-E6 cells (ATCC, Virginia, USA). After three additional days, the cell culture supernatant was harvested and stored at −80°C. The virus stock was characterized by rRT-PCR. Heat-inactivation of the virus stock was performed by incubating the sample for 30 min at 56°C.