Surrogate Virus Neutralisation Test Based on Nanoluciferase-Tagged Antigens to Quantify Inhibitory Antibodies against SARS-CoV-2 and Characterise Omicron-Specific Reactivity in a Vaccination Cohort

Virus-specific antibodies are crucial for protective immunity against SARS-CoV-2. Assessing functional antibodies through conventional or pseudotyped virus neutralisation tests (pVNT) requires high biosafety levels. Alternatively, the virus-free surrogate virus neutralisation test (sVNT) quantifies antibodies interfering with spike binding to angiotensin-converting enzyme 2. We evaluated secreted nanoluciferase-tagged spike protein fragments as diagnostic antigens in the sVNT in a vaccination cohort. Initially, spike fragments were tested in a capture enzyme immunoassay (EIA), identifying the receptor binding domain (RBD) as the optimal diagnostic antigen. The sensitivity of the in-house sVNT applying the nanoluciferase-labelled RBD equalled or surpassed that of a commercial sVNT (cPass, GenScript Diagnostics) and an in-house pVNT four weeks after the first vaccination (98% vs. 94% and 72%, respectively), reaching 100% in all assays four weeks after the second and third vaccinations. When testing serum reactivity with Omicron BA.1 spike, the sVNT and pVNT displayed superior discrimination between wild-type- and variant-specific serum reactivity compared to a capture EIA. This was most pronounced after the first and second vaccinations, with the third vaccination resulting in robust, cross-reactive BA.1 construct detection. In conclusion, utilising nanoluciferase-labelled antigens permits the quantification of SARS-CoV-2-specific inhibitory antibodies. Designed as flexible modular systems, the assays can be readily adjusted for monitoring vaccine efficacy.


Introduction
Since the WHO declared coronavirus disease 2019 (COVID-19) a pandemic in 2020, vaccines based on the original Wuhan-Hu-1 strain of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and its descendants, specifically Omicron BA.1 and BA.4/5, have proven to be effective in controlling the disease [1].These vaccines mainly target the Vaccines 2023, 11, 1832 2 of 16 viral spike (S) surface glycoprotein as the predominant mediator of immunity [2,3].Immune responses following infection, vaccination, or both (known as hybrid immunity) are under thorough investigation to determine the extent and duration of protection against emerging virus variants.Studies on pre-Omicron variants have shown that hybrid immunity provides optimal and long-lasting protection against severe disease, an important finding that is already influencing current vaccination schemes [4].Furthermore, enhanced immune responses have been observed following vaccination post-infection, such as an increase in antibodies directed against the more conserved domain 2 of the SARS-CoV-2 spike protein (S2), which have been identified and verified in various variants, including Omicron B.1.1.529and BA.5 [5].While immunity mediated by B and T cells appears to persist, with an even greater effect noted in hybrid immunity, antibody levels gradually decrease over time.However, booster vaccinations have shown their efficacy in inducing broad humoral immunity against Omicron variants BA.1, BA.2, and BA.5 [6].Conversely, prior infection or vaccination may impede neutralising immune responses to subsequent viral variants such as B.1.1.529,BA.4, or BA.5 [7,8].In addition to monitoring population-level developments in (hybrid) immunity and immune responses in high-risk patient groups, it will be crucial to project the efficacy of vaccines developed to counter evolving immune escape variants to guide future vaccination strategies.
Hence, a serologic characterisation of neutralising antibodies has increasingly gained importance, particularly regarding emerging escape variants.To this end, standard serologic tests that detect SARS-CoV-2-specific antibodies irrespective of their neutralising capacity, such as enzyme-linked or chemiluminescent immunoassays (ELISA, CLIA), have to be complemented by functional virus neutralisation assays.The majority of neutralising antibodies in convalescent sera target epitopes within the receptor binding domain (RBD) of the viral spike protein domain 1 (S1), which mediates attachment to its cellular receptor, human angiotensin-converting enzyme 2 (hACE2), and, thus, host cell entry [9][10][11][12][13].As the RBD displays sufficient variability to discriminate between different human coronaviruses and simultaneously high specificity when used as a diagnostic antigen, it has become the target of choice in serologic assays [14,15].In addition, the N-terminal domain (NTD) of S1 has been shown to harbour neutralising epitopes recognised by convalescent and vaccinee sera [16][17][18].Viral variants frequently evade neutralising immune responses or therapeutic monoclonal antibodies through mutations within immunodominant epitopes such as E484 of the RBD, with Omicron and its subvariants accumulating more than 30 mutations within spike [11,[19][20][21][22][23].Whereas minimal protective antibody titres have not yet been defined, neutralising antibodies are regarded to correlate with protection from severe disease, rendering them suitable surrogate markers for immunity and vaccine efficacy studies [24].
While conventional virus neutralisation tests (cVNTs) such as plaque reduction neutralisation tests (PRNTs) are considered the gold standard, they require successful and time-consuming virus isolation and biosafety level (BSL) 3 containment, which impedes many diagnostic laboratories.As an alternative, vector-based pseudotyped virus neutralisation tests (pVNTs) have been developed, enabling faster testing within standard BSL-2 facilities; however, as in-house tests, they still lack standardisation.To facilitate the assessment of neutralising antibodies, virus-free assays commonly termed surrogate virus neutralisation tests (sVNTs) have been introduced to the market and can be conducted in a standardised format under clinical routine BSL-2 laboratory conditions within only a few hours [25][26][27].These assays rely on the inhibition of SARS-CoV-2 RBD binding to hACE2 and have been shown to correlate well with cVNTs and various SARS-CoV-2 S protein-based pVNT assays, rendering them valuable tools for high-throughput functional antibody screening [25][26][27][28][29][30][31][32][33].In the wake of the first primary Omicron infections, commercial antibody assays targeting the S protein based on the ancestral Wuhan-Hu-1 strain of SARS-CoV-2, including sVNTs, have revealed a decline in sensitivity, warranting the adaptation of diagnostic antigens to the currently prevailing variants [34].
To investigate the influence of variant-specific mutations on the reactivity of vaccinee sera and to differentiate between different immunoglobulin classes without handling in-fectious virus particles, we developed virus-free assays applying S protein fragments of varied lengths, including the RBD and NTD (Supplementary Figure S1) linked to nanoluciferase (NanoLuc, NLuc, cf.Materials and Methods) as bioluminescent reporter-tagged antigens.The nanoluciferase reporter system has been demonstrated to be appropriate for immunoassays such as the fluid-phase luciferase immune precipitation (LIPS) assay [35].In our assessment of a cohort of vaccinated individuals, our in-house sVNT and immunoglobulin capture enzyme immunoassay (EIA) complemented routine serologic tests convincingly, displaying high levels of sensitivity, specificity, and reproducibility.The accuracy of the findings was confirmed through comparison with commercial serologic assays, as well as an in-house VSV-based pVNT (VSV-pVNT) employing a vesicular stomatitis virus (VSV) bearing the SARS-CoV-2 S protein.The sVNT and capture EIA are highly adaptable and simple assays that allow for a rapid analysis of variant-specific humoral immune responses, thus characterising vaccine efficacy.

Human Serum Samples
Serum samples were collected within the framework of a study approved by the ethics committee of the Aerztekammer Westfalen-Lippe, Muenster, Germany, and the University of Muenster aiming at the "Characterisation of the Humoral Immune Response to SARS-CoV-2 in Vaccinees".Participants received either Comirnaty (BNT162b2, BioNTech SE/Pfizer, Mainz, Germany) or Spikevax (mRNA-1273, Moderna Inc., Cambridge, MA, USA) mRNA vaccines based on wild-type Wuhan-derived S protein sequences.Samples were obtained at baseline before immunisation (t0) and 4 weeks post first vaccine dose (t1); 4 weeks (t2), 3 months (t3), and 6 months after the second vaccine dose (t4); and 4 weeks after (t5) the third vaccine dose, respectively.The sampling period ranged from January 2021 to January 2022.

Routine Serology
Systematically, IgG antibodies specifically targeting the receptor binding domain (RBD) of the SARS-CoV-2 S protein (anti-RBD IgG) were quantified using the commercial, CE/IVD certified chemiluminescence microparticle immunoassay (CMIA) SARS-CoV-2 IgG II Quant on an Architect platform (Abbott Diagnostics, Wiesbaden, Germany).Seropositivity was indicated by values greater than or equal to 50.0 arbitrary units (AU)/mL, where 0.142 Abbott AU/mL corresponds to 1 binding antibody unit (BAU) as defined by the WHO.Abbott claims a sensitivity of up to 99.37% depending on the day post-symptom onset and a specificity of 99.55%.Accordingly, the presence of IgG antibodies against the nucleocapsid (N) protein (anti-N IgG) of SARS-CoV-2 was qualitatively determined using the CMIA Abbott SARS-CoV-2 IgG in vaccinees to detect any potential infection (sensitivity 100%, specificity 99.63%).In addition, the recomLine SARS-CoV-2 IgG CE/IVD (Mikrogen, Neuried, Germany) commercial line blotting assay was applied to assess sera with equivocal anti-N IgG levels at t0.The test detects IgG directed towards the SARS-CoV-2 RBD, S1, and N antigens semi-quantitatively using a line blot format.All tests were conducted according to the manufacturer's instructions.
The commercial, CE/IVD-certified cPass SARS-CoV-2 Neutralization Antibody Detection Kit (GenScript Biotech, Leiden, The Netherlands) was employed to semiquantitatively evaluate the neutralising activity of sera using sequences derived from the Wuhan strain.This assay measures the degree to which antibodies inhibit the binding of the RBD to the hACE2 receptor using RBD-HRP (horse-radish peroxidase) conjugates in a blocking ELISA format and is considered one of the most accurate sVNTs available [25,27,31].Briefly, serum samples are incubated with RBD-HRP to allow for the binding of inhibitory antibodies to the RBD and subsequently transferred to hACE2-coated ELISA plates.RBD-HRP not blocked by serum antibodies will bind to hACE2, which is visualised by a colour reaction and quantitatively determined in a microtitre plate reader at 450 nm.Following the manufacturer's manual, serum samples were tested in technical duplicates with a final dilution of 1:20.The percentage of inhibition was calculated as ( The in-house VSV-pVNT utilised a vesicular stomatitis virus (VSV) carrying the SARS-CoV-2 S protein.To enhance trans-complementation through the transiently expressed S protein in the VSV-pVNT, we employed site-directed mutagenesis to introduce a 21amino-acids (aa)-long C-terminal deletion into the expression vector pCG1-SARS-2-S [10] that comprises the wildtype (wt) S protein aa sequence (GenBank ID NC_045512.2) [1].The plasmid pCG1-SARS-2-S was cleaved with SalI and BsaBI.Subsequently, the product of primers S BsaB1 forward (fwd) and S Delta1253 backward (bwd) amplified from plasmid pCG1-SARS-2-S was inserted into the vector by In-Fusion cloning (Clontech/Takara Bio Inc., Mountain View, CA, USA).The resulting construct was designated pCG1-SARS-2-S-Delta1253.The S protein with the Omicron BA.1-specific amino acid sequence and the C-terminal 21 aa deletion was expressed from the vector pcDNA3.1 SARS-2 Omicron containing a synthetic S-insert derived from BA.1 (Thermo Fisher Scientific GmbH, Schwerte, Germany).
SARS-CoV-2 neutralisation was analysed using the VSV-pVNT system (VSV ∆G/GFP-Luc + SARS-CoV-2 S protein variants) as described by [36].The pseudotyped virus was generated according to the method outlined by [37].The sera were diluted (1:20) and preincubated with the pseudotyped virus at 37 • C/5% CO 2 for 1 h (hour).Subsequently, Vero E6 cells were infected with the pseudotyped virus-sera mixture for 1 h at 37 • C/5% CO 2 (with a final multiplicity of infection (MOI) of 0.01).At 16 h post infection (h.p.i.), the number of GFP-positive cells was quantified with the Celigo Image Cytometer (Nexcelom/Perkin Elmer Inc., Waltham, MA, USA).Four technical replicates were measured, from which the average was calculated.A pool of 17 sera containing high levels of neutralising antibodies against S, collected at t2, was used as a positive control (PSP).Additionally, a corresponding pool of sera collected from the same patients at t0 served as a negative control (NSP).The degree of neutralisation was calculated as the reduction of the GFP signal (%) = (1 − GFP signal of the treated sample/GFP signal of the untreated sample) × 100%.

Surrogate Virus Neutralisation Test
The in-house surrogate virus neutralisation test (sVNT) was designed to measure the inhibition of the binding of secreted RBD-containing S protein fragments, which are Nterminally labelled with NLuc to the hACE2 receptor in an ELISA format (Supplementary Figure S2A).Black high-binding 96-well plates (Greiner Bio-One GmbH, Frickenhausen, Germany) were coated with 200 ng/well of recombinant hACE2 (Sino Biological Inc., Beijing, China, #10108-H08H) diluted in 100 µL carbonate buffer at 4 • C for 16 h.The coated plates were washed four times using 300 µL/well TRIS-buffered saline (TBS)/0.1% Tween (TBS-T), blocked with 300 µL/well TBS-T/2% BSA (w/v) for 1.5 h at 37 • C, and washed four times using TBS-T.
Subsequently, 110 µL of serum, which was diluted 1:20 in TBS-T and contained 5.5 × 10 6 relative light units (rlu) secNLuc-RBD, underwent incubation for 30 min at 37 • C. Following this, 100 µL/well was transferred to hACE2-coated 96-well microtiter plates and kept at 37 • C for 1 h.The plates were washed with 300 µL/well TBS-T four times before the addition of 100 µL Nano-Glo reagent/well.Next, the resulting NLuc activity was quantified.Luciferase activity in the absence of human serum and SARS-CoV-2 antibody negative (NSP) and positive (PSP) human serum pools served as controls.The inhibition of the binding of the RBD to hACE2 was calculated as a reduction in the NLuc signal (%) = (1 − NLuc signal of the sample/NLuc signal of the untreated sample) × 100%.

Immunoglobulin Capture Enzyme Immunoassay
Immunoglobulin class-specific detection of antibodies was performed by in-house enzyme immunoassays (EIA) (Supplementary Figure S2B).To this end, black high-binding strip plates (Greiner Bio-One) were coated with 400 ng/well of anti-human IgG (GtxHu-004-J), anti-human IgA (GtxHu-001-G), and anti-human IgM (GtxHu-008-G) goat F(ab') 2 fragments (Dianova/Biozol GmbH, Hamburg, Germany), respectively.The procedure was carried out at 4 • C for 16 h in 100 µL of bicarbonate buffer (pH 9.6).After washing four times with 300 µL of wash buffer (TBS pH 7.4 containing 0.05% Tween 20, TBS-T), 300 µL of blocking buffer (TBS-T/10% foetal calf serum (FCS)) was added at 37 • C for 90 min.After washing again four times, 100 µL of each serum sample was incubated on the sealed plate at a dilution of 1:100 in the sample buffer (TBS-T/10% FCS) for 1 h at 37 • C. Controls consisting of 100 µL of TBS-T/10% FCS without serum as well as NSP and PSP were included.Next, plates were washed again four times with wash buffer and incubated with the NLuc-labelled antigens.The antigens were added at a concentration of 5 × 10 6 rlu/100 µL in a total volume of 100 µL of the sample buffer for 1 h at 37 • C.After washing four times, the luciferase activity of NLuc-RBD proteins bound to the plate was quantified by adding 100 µL of NanoGlo reagent.Results were expressed as activity (rlu) after subtracting the blank without serum addition.Cut-off values for all in-house assays were determined through receiver operating characteristic (ROC) curves using GraphPad Prism 10.0 (San Diego, CA, USA).

Routine Serologic Assessment of Serum Samples
All 124 participants of the vaccination study underwent routine serology to test for SARS-CoV-2-specific antibodies immediately before their first vaccination.At t0, before the first vaccination, sera from 53 participants were non-reactive in the commercial anti-RBD IgG CMIA, anti-N IgG CMIA, and line blot.These participants were also available for antibody testing 4 weeks after the first vaccination (t1) and 4 weeks after the second vaccination (t2), and their sera were included in the study.Additional sera from a subset of these vaccinated individuals were gathered 3 months (t3, 42 sera) and 6 months (t4, 26 sera) post second vaccination, and 4 weeks after the third vaccination (t5, 12 sera).The quantitative anti-RBD-IgG CMIA revealed that all serum samples collected from t1 to t5 were reactive (Figure 1).

Recognition of NanoLuc-Tagged S-Fragments by Serum Antibodies
As an initial experimental step, the suitability of NLuc-tagged S fragments containing the RBD derived from the original Wuhan-Hu-1 sequence as diagnostic antigens (Table 1) was evaluated.The gamma-chain (human immunoglobulin heavy chain gamma) capture EIA measuring IgG was used to assess the reactivity of constructs comprising the RBD of S (aa 319 to 541) and fragments of the external domain of S spanning from aa 15 to 541 (NTD plus RBD), aa 15 to 680 (S1 fragment), and aa 15 to 1213 (ectodomain of spike), respectively, with sera collected at time points t0, t1, and t2 (Supplementary Figure S3).For comparison, an NLuc-labelled NTD fragment (aa 15 to 307) was included.Cell culture supernatants adjusted to an NLuc activity of 5 × 10 6 rlu per well were used as antigen preparations without any further purification steps (cf.Materials and Methods).The reactivity of sera in the quantitative anti-RBD-IgG CMIA (Abbott) served as the gold standard.ROC curves were applied to characterise the diagnostic performance of constructs and to determine cut-off values that allowed for the discrimination of sera obtained before (t0) and after the first vaccination (t1) with 100% specificity.The signal intensity of IgG reactivity measured in the gamma-capture EIA at time points t1 and t2 served as a second criterion to identify the most appropriate NLuc-tagged S protein fragments.

Recognition of NanoLuc-Tagged S-Fragments by Serum Antibodies
As an initial experimental step, the suitability of NLuc-tagged S fragments containing the RBD derived from the original Wuhan-Hu-1 sequence as diagnostic antigens (Table 1) was evaluated.The gamma-chain (human immunoglobulin heavy chain gamma) capture EIA measuring IgG was used to assess the reactivity of constructs comprising the RBD of S (aa 319 to 541) and fragments of the external domain of S spanning from aa 15 to 541 (NTD plus RBD), aa 15 to 680 (S1 fragment), and aa 15 to 1213 (ectodomain of spike), This approach revealed that, amongst the constructs tested as described in Section 2.3.2, the RBD-containing construct secNLuc-RBD exhibited the highest sensitivity at time point t1 with 92.5% (49/53 sera reactive) (Supplementary Figures S3 and S4; Table 2).In comparison, construct secNLuc-S15-541 comprising the NTD and RBD showed a similar, slightly lower sensitivity of 86.8%.In contrast, the sensitivity achieved with constructs secNLuc-S15-680 containing the S1 domain and secNLuc-spike-CSdel-PPmut including the spike ectodomain at t1 was lower, i.e., at 73.6% and 71.7%, respectively.The secNLuc-NTD construct, which was tested for comparison, displayed a sensitivity comparable to secNLuc-S15-541, i.e., 86.8%.At time point t2, all constructs mentioned above exhibited a sensitivity of 100% in the gamma-chain capture EIA.In terms of signal strength at time point t2, constructs secNLuc-RBD and secNLuc-S15-541 showed the strongest reactivity, while the reactivity of secNLuc-NTD, secNLuc-S15-680, and secNLuc-spike-CSdel-PPmut in the gamma-chain capture EIA was lower (Supplementary Figures S3 and S4).These experiments demonstrated that the secNLuc-RBD antigen was most sensitively recognized by human sera after the first vaccination, allowed for the best differentiation between sera obtained before and after the first vaccination, and yielded the strongest signals after the second vaccination.Therefore, the construct was chosen for further experimentation.
As the sVNT does not distinguish between antibody classes, all serum samples collected at time points t0 to t5 were analysed using EIA to determine their antibody classspecific reactivity.Cut-off values for IgA-and IgM-EIA were identified through a ROC curve analysis (Supplementary Figure S4).This approach revealed that secNLuc-RBD reacted most strongly with IgG antibodies from t1 to t5 (Figure 1).The IgG reactivity pattern in the gamma-chain capture EIA was almost indistinguishable from that found by the commercial anti-RBD CMIA.A notable increase was evident from t1 to t2, with the highest levels reached at time points t2 and t5.In contrast, IgG levels decreased from t2 to t3 to t4.
The reactivity pattern of IgA with secNLuc-RBD highly resembled that of IgG but tended to be elevated after the third vaccination (t5) and sustained a minor descent from t2 to t4.Overall, the signal strength of IgA was approximately 10-fold lower as compared to that of IgG.Dissimilar to IgG and IgA, the maximum IgM levels were detected at t1 and t5, while no increase occurred from t1 to t2. Between t2 and t4, RBD-specific IgM antibodies decreased more markedly than IgG and IgA antibodies.Additionally, IgM antibody levels were significantly lower than those of IgG.

Detection of Inhibitory Antibodies by sVNT
Employing the secNLuc-RBD construct as a diagnostic antigen in our in-house sVNT showed the effectiveness of this experimental approach to detecting antibodies that impede RBD binding to hACE2 (Figure 2).To ensure 100% specificity when discriminating between sera obtained at time points t0 and t1, the cut-off was set at 25% inhibition (Supplementary Figure S4), resulting in a sensitivity of 98% at t1 (Table 3).RBD-specific antibody detection by the in-house sVNT at t2 was achieved with a sensitivity and specificity of 100%.At time points t3, t4, and t5, 42 out of 42 sera, 24 out of 26 sera, and 12 out of 12 sera were reactive in the in-house sVNT.showed the effectiveness of this experimental approach to detecting antibodies that im-pede RBD binding to hACE2 (Figure 2).To ensure 100% specificity when discriminating between sera obtained at time points t0 and t1, the cut-off was set at 25% inhibition (Supplementary Figure S4), resulting in a sensitivity of 98% at t1 (Table 3).RBD-specific antibody detection by the in-house sVNT at t2 was achieved with a sensitivity and specificity of 100%.At time points t3, t4, and t5, 42 out of 42 sera, 24 out of 26 sera, and 12 out of 12 sera were reactive in the in-house sVNT.Compared to the commercial cPass sVNT (Figure 2), the in-house sVNT displayed comparable or slightly improved sensitivity at time point t1 (52 out of 53 sera were reactive, with 98% sensitivity vs. 94%) (Table 3).One serum sample non-reactive at t1 in the in-house sVNT was reactive in the cPass sVNT and in the heavy-chain capture EIA; however, it exhibited low IgG levels and moderate IgM levels and was IgA negative.All three serum samples that did not react in the cPass assay at t1 were also nonreactive in IgG, IgA, and IgM capture EIA.Compared to the commercial cPass sVNT (Figure 2), the in-house sVNT displayed comparable or slightly improved sensitivity at time point t1 (52 out of 53 sera were reactive, with 98% sensitivity vs. 94%) (Table 3).One serum sample non-reactive at t1 in the in-house sVNT was reactive in the cPass sVNT and in the heavy-chain capture EIA; however, it exhibited low IgG levels and moderate IgM levels and was IgA negative.All three serum samples that did not react in the cPass assay at t1 were also nonreactive in IgG, IgA, and IgM capture EIA.
At time point t4, the sensitivity of the in-house sVNT was slightly lower as compared to cPass.Both serum samples that were non-reactive in the in-house sVNT at t4 exhibited reactivity in cPass and the IgG EIA, had low levels in the IgA EIA, and were negative in the IgM EIA.At time points t2 and t5, all sera displayed reactivity in both sVNTs.A ROC curve analysis of the cPass sVNT results revealed a 98% sensitivity and specificity in discriminating sera at t0 and t1 using a cut-off of 20% inhibition, which differs from the specifications of the CE/IVD certified product by the manufacturer (Table 3, Supplementary Figure S4).
While the Abbott CMIA and the capture EIAs demonstrated a considerable decrease in antibody levels following the second vaccination, the cPass and in-house sVNT analysis indicated a much lower significant decline.Both sVNTs produced consistent results at time points t0 and t5 but varied in signal strength more widely at t1 to t4.The greatest disparities were observed at time points t3 and t4 (Figure 2).Overall, the correlation between both sVNTs was strong, as indicated by the Spearman r value of 0.91 (95% confidence interval: 0.88 to 0.93).To evaluate the functional significance of inhibitory antibodies by sVNTs, the formation of neutralizing antibodies was examined via VSV-pVNT (Figure 3).Employing ROC curve analysis and vaccinated vs. non-vaccinated differentiation as criteria, we established a cut-off of 65% inhibition for detection of neutralising antibodies at time point t1, which ensured 100% specificity (Supplementary Figure S4).This resulted in a 72% sensitivity of the pVNT at time point t1 (Table 3).At t2, all human sera were reactive in the pVNT.Differences in outcomes between the pVNT, the in-house sVNT, and the cPass sVNT were mainly noticeable in sera collected at t1.Taken together, a strong correlation between the results of the in-house sVNT and pVNT was observed with a Spearman r of 0.9 (95% confidence interval: 0.86 to 0.92) (Figure 3).
in antibody levels following the second vaccination, the cPass and in-house sVNT analysis indicated a much lower significant decline.Both sVNTs produced consistent results at time points t0 and t5 but varied in signal strength more widely at t1 to t4.The greatest disparities were observed at time points t3 and t4 (Figure 2).Overall, the correlation between both sVNTs was strong, as indicated by the Spearman r value of 0.91 (95% confidence interval: 0.88 to 0.93).
To evaluate the functional significance of inhibitory antibodies by sVNTs, the formation of neutralizing antibodies was examined via VSV-pVNT (Figure 3).Employing ROC curve analysis and vaccinated versus non-vaccinated differentiation as criteria, we established a cut-off of 65% inhibition for detection of neutralising antibodies at time point t1, which ensured 100% specificity (Supplementary Figure S4).This resulted in a 72% sensitivity of the pVNT at time point t1 (Table 3).At t2, all human sera were reactive in the pVNT.Differences in outcomes between the pVNT, the in-house sVNT, and the cPass sVNT were mainly noticeable in sera collected at t1.Taken together, a strong correlation between the results of the in-house sVNT and pVNT was observed with a Spearman r of 0.9 (95% confidence interval: 0.86 to 0.92) (Figure 3).Comparison of results obtained in in-house sVNT and pVNT in sera obtained on t0, t1, and t2.Values < 0 were set to 0, and the correlation coefficient was calculated according to Spearman.
Using the pVNT as the gold standard, the effectiveness of sVNTs, Anti-RBD CMIA, and Anti-RBD IgG EIA in predicting neutralising antibodies among the 159 sera collected from t0 to t2 was analysed.By raising the cut-off of the in-house sVNT to over 73% inhibition, the detection of neutralising antibodies by pVNT could be predicted with a 98.6% Values < 0 were set to 0, and the correlation coefficient was calculated according to Spearman.
Using the pVNT as the gold standard, the effectiveness of sVNTs, Anti-RBD CMIA, and Anti-RBD IgG EIA in predicting neutralising antibodies among the 159 sera collected from t0 to t2 was analysed.By raising the cut-off of the in-house sVNT to over 73% inhibition, the detection of neutralising antibodies by pVNT could be predicted with a 98.6% specificity and 87.8% sensitivity.At a cut-off of over 76.5% inhibition, the commercial cPass sVNT showed 100% specificity and 95.6% sensitivity (Table 4).By increasing the cutoff to over 1706 AU/mL in the Anti-RBD CMIA, 95.65% specificity and 96.67% sensitivity were achieved relative to the pVNT benchmark.For the in-house RBD IgG EIA, a cut-off of over 1103 rlu resulted in 100% specificity and 75.6% sensitivity.Adjusting the cut-off of the in-house RBD IgG EIA to over 617.5 rlu enabled the prediction of neutralising antibodies with 97.1% specificity and 92.2% sensitivity (Table 4).As expected, raising the cut-off values caused a significant decline in the sensitivity of all assays regarding differentiation between vaccinated and unvaccinated individuals.At time points t0 and t1, the in-house sVNT, cPass sVNT, Anti-RBD CMIA, and Anti-RBD IgG EIA showed sensitivity reductions of 51.9%, 60.4%, 69.8%, and 60.4%, respectively. 1Optimal threshold values were defined by ROC curve analysis of data and calculation of Youden s index. 2 As compared to pVNT as the gold standard in sera obtained from t0 to t2.

Detection of Inhibitory Antibodies Reactive with the Omicron BA.1 Variant
Finally, we investigated whether the RBD-based in-house sVNT would detect the induction of cross-inhibitory antibodies against Omicron BA.1 upon vaccination with wt S. To this end, inhibitory antibodies against the wt (in-house sVNT) and Omicron BA.1 RBD (in-house BA.1 sVNT), respectively, were measured in nine vaccinees whose serum samples were taken at all time points of the vaccination study.The determination of neutralising antibodies against wt S and BA.1 S via in-house pVNT served as a control.
The pVNT results indicated that all vaccinees tested required three vaccinations (t5) for the induction of a robust cross-neutralizing antibody response against Omicron BA.1 S (Figure 4).At earlier time points, none (t1), four (t2), three (t3), and a single (t4) out of nine serum samples contained neutralising antibodies against Omicron BA.1 S with an inhibitory activity at least 65% at a serum dilution of 1:20.In contrast, neutralizing antibodies against wt S were detected in all sera obtained from t2 to t5, as well as in six out of nine sera gathered at t1 (Figure 4).Significantly higher neutralising antibody levels against wt S were observed from t1 to t4 (Figure 4).At t5, all sera neutralized wt S and BA.1 S with nearly 100% activity at a dilution of 1:20.Overall, the pattern of antibody reactivity in the in-house sVNTs correlated with the outcome of the pVNT in the detection of BA.1-reactive neutralizing antibodies (Figure 5).At time point t0, a higher level of background reactivity of sera with the Omicron BA.1 RBD as compared to the wt RBD was observed.Raising the cut-off of the in-house BA.1 sVNT to 50% inhibition to avoid false-positive results at t0 led to equivalent differentiation between Omicron BA.1 reactive and non-reactive samples in the sVNT and pVNT.As observed by the pVNT, none of the sera reacted with Omicron BA.1 at t1.Three wt S-based vaccinations were necessary to elicit a robust reactivity in the in-house BA.1 sVNT in all sera.Levels of inhibitory antibodies against the wt RBD were significantly higher in all samples from t1 to t5 (Figure 5).In contrast, the determination of variant-reactive antibodies via a heavy-chain capture EIA for comparison revealed that this method less reliably distinguished between wt and Omicron BA.1-reactive antibodies despite using the same antigens as in the in-house sVNTs (Supplementary Figure S5).Significant differences in reactivity were observed in the IgG EIA at time points t3 and t4, in IgA EIA at t2 and t3, and in IgM EIA at t3 and t4, respectively.The relative differences in reactivity with the wt RBD and Omicron BA.1 RBD, however, were less pronounced as compared to sVNT and pVNT.

Discussion
In summary, the utilisation of NLuc-labelled S protein fragments in in-house sVNT and EIA permits a comprehensive characterisation of antibodies induced by SARS-CoV-2 infection and/or vaccination.Both assay systems are characterised by high specificity, sensitivity, and reproducibility and can be conveniently performed in a routine laboratory environment.As the in-house assays employ unpurified cell-culture supernatants as the source of diagnostic antigens, they are easy to conduct.
a capture EIA to investigate isotype-specific responses to viral variants.Implementing the NLuc-tagged wt RBD, the in-house EIA discriminated between IgM, IgA, and IgG reactivity.However, when testing vaccinee sera using an NLuc-tagged Omicron BA.1 RBD as the diagnostic antigen, the EIA exhibited inferior performance in discriminating between variant-specific IgG antibodies compared to the sVNT and pVNT.
In contrast to EIA-based serologic assays, the sVNT detects antibodies of all isotypes, which may result in higher sensitivity and better negative predictive values compared to isotype-specific tests, as shown for the cPass sVNT vs. commercial IgG enzyme immunoassays [27].We verified this observation when comparing the in-house capture EIA and sVNT using the NLuc-tagged RBD as the antigen.However, it has been reported that the cPass sVNT may overestimate low neutralising antibody levels when using the cut-off value of 20% recommended by its manufacturer for the RUO (research use only) version upon release, requiring a threshold of up to 50% when analysing high-titre sera [28].Subsequently, this issue was resolved by raising the cut-off value to 30% when launching the CE/IVD certified cPass sVNT on the market [25], at a slight expense of sensitivity while increasing specificity [30,32].Alternatively, a threshold area between 15% and 35% indicating equivocal results could be defined, which would warrant confirmation by other VNTs, especially when testing low-titre sera with a focus on individual results [32].
Neutralising antibody titres exceeding the quantifiable range of the assay, which are often found in vaccinees, may require an adjustment of the cut-off values.It has been recommended that ROC curves be calculated [31] or end-point dilution of sera be performed.We set the cut-off value for the in-house sVNT to 25% to avoid false-positive results in sera diluted 1:20.As the range of NLuc signals surpasses that of HRP, the reporter enzyme deployed in the cPass sVNT, displaying the remaining hACE2 binding capacity instead of inhibition could improve resolution when testing high-titre sera.In the case of investigating variant-specific antibodies, cut-off values may have to be adapted accordingly.A cut-off value of 65% was established for optimal discrimination between Omicron BA.1 reactivity vs. non-reactivity in the in-house sVNT.
Using the reactivity of sera in the pVNT as the benchmark, the cut-off of the sVNT needs adjusting to improve agreement between the assays.Our data show that a cut-off of over 75% inhibition in the sVNT enables the prediction of the presence of neutralising antibodies in sera with high specificity and reasonable sensitivity.Hence, the same method can determine the appropriate cut-off titres for the anti-RBD CMIA and anti-RBD IgG EIA, respectively.With an extended interval from vaccination or infection, the determination of serum antibody levels with non-functional tests such as CMIA or EIA could pose problems due to antibody waning.Contrary to the commercial CMIA and the in-house EIA, which both revealed a decline in antibody levels in sera collected from t2 to t4, the inhibitory capacities of antibodies determined by in-house pVNT and sVNT persisted at high levels, suggesting affinity maturation.This reinforces the requirement to complement routine antibody quantification with a functional characterisation of the humoral immune response.Unlike the in-house pVNT that involves the entire ectodomain of the S protein, the in-house sVNT solely measures antibodies that interfere with the RBD-hACE2 interaction, which pertains to most neutralising antibodies; however, antibodies that bind to different epitopes will escape detection [29].Since the sample size available for our study was limited, further studies should be performed to corroborate our findings and to investigate the influence of different sample types on the test performance.

Conclusions
To conclude, our data show that the in-house sVNT and, to a lesser degree, the EIA utilising the NLuc-tagged RBD, enable rapid quantification and a functional evaluation of variant-specific antibodies in a hybrid-immune population that is prevalent worldwide these days.Conceivable applications comprise monitoring the efficacy of vaccines against virus variants, screening drug candidates that interfere with the RBD-hACE2 binding and following up on prolonged COVID-19 courses in high-risk groups in clinical settings.As secreted RBD fragment of S protein, aa 319 to aa 541, wt sequence, N-terminal NLuc tag pEN-secNL-RBD Omicron secreted RBD fragment of S protein, aa 319 to aa 541, Omicron BA.1 sequence, N-terminal NLuc tag pEN-secNL-S15-307 secreted NTD fragment of S protein, aa 15 to aa 307, wt sequence, N-terminal NLuc tag pEN-secNL-S15-541 secreted NTD and RBD fragment of S protein, aa 15 to aa 541, wt sequence, N-terminal NLuc tag pEN-secNL-S15-680 secreted S1 fragment of S protein, aa 15 to aa 680, wt sequence, N-terminal NLuc tag pEN-secNL-spike-CSdel-PPmut secreted stabilised trimeric, histidine-tagged ectodomain of S protein, aa15 to aa 1213, lacking the basic, furin-sensitive cleavage site, N-terminal NLuc tag pCG1-SARS-2-S-Delta1253 wt S protein with C-terminal 21aa deletion pcDNA3.1 SARS-2 Omicron Omicron BA.1 S protein with C-terminal 21aa deletion CAG ATC CGC TAG C ATG AAC TCC TTC TCC ACA AGC secNL-BamHI bwd GGA TCC CGC CAG AAT GCG TTC GCA CAG RBD-BamHI fwd ATT CTG GCG GGA TCC CGG GTG CAG CCC ACC GAA TCC ATC RBD-NotI bwd TC TAG AGT CGC GGC CGC TTA TCA GAA GTT CAC GCA TTT GTT CTT CAC GAG RBD-BamHI_2 fwd C ATT CTG GCG GGA TCC CGG GTG CAG CCC ACC GAA TCC ATC S15-BamH1 fwd C ATT CTG GCG GGA TCC TGT GTG AAC CTG ACC ACA AGA ACC S307-Not1 bwd TC TAG AGT CGC GGC CGC TTA TCA GGT GAA GGA CTT CAG GGT GCA C S680-Not1 bwd TC TAG AGT CGC GGC CGC TTA TCA GCT GTT TGT CTG TGT CTG GTA GCT S CSdel-PPmut bwd T CTA GAG TCG CGG CCG CGA GCT CGA ATT TCA TTA GTG S BsaB1 fwd TC GCC GGA CTG ATT GCC ATC GTG S Delta1253 bwd G GTT TAA ACA GTC GAC ACT AGT TTA GCT GCC ACA GCT ACA ACA GCC CTT

Figure 2 .
Figure 2. Detection of inhibitory antibodies by sVNT.(A) In-house sVNT employing the secNLuc-RBD construct was used to determine antibodies inhibiting binding of the RBD to hACE2 in vaccinee sera at time points t0-t5.The cut-off determined by ROC curve analysis (25% inhibition) is indicated by the dashed line.Horizontal bars delineate the variable median.(B) Analysis of vaccinee sera by the cPass sVNT at time points t0-t5.The cut-off value according to the manufacturer's manual is indicated by the dashed line (30% inhibition), and horizontal bars delineate the variable median.(C) Correlation of in-house sVNT and cPass sVNT values < 0% were set to 0%, and the correlation coefficient was calculated according to Spearman.Ih: in-house.

Figure 2 .
Figure 2. Detection of inhibitory antibodies by sVNT.(A)In-house sVNT employing the secNLuc-RBD construct was used to determine antibodies inhibiting binding of the RBD to hACE2 in vaccinee sera at time points t0-t5.The cut-off determined by ROC curve analysis (25% inhibition) is indicated by the dashed line.Horizontal bars delineate the variable median.(B) Analysis of vaccinee sera by the cPass sVNT at time points t0-t5.The cut-off value according to the manufacturer's manual is indicated by the dashed line (30% inhibition), and horizontal bars delineate the variable median.(C) Correlation of in-house sVNT and cPass sVNT values < 0% were set to 0%, and the correlation coefficient was calculated according to Spearman.Ih: in-house.

Figure 3 .
Figure 3. VSV-pseudotyped pVNT, (A) neutralising activity of sera obtained from 53 vaccinees on t0, t1, and t2 in pVNT, dashed line depicts cut-off, horizontal lines show the variable median.(B)Comparison of results obtained in in-house sVNT and pVNT in sera obtained on t0, t1, and t2.Values < 0 were set to 0, and the correlation coefficient was calculated according to Spearman.

Figure 3 .
Figure 3. VSV-pseudotyped pVNT, (A) neutralising activity of sera obtained from 53 vaccinees on t0, t1, and t2 in pVNT, dashed line depicts cut-off, horizontal lines show the variable median.(B)Comparison of results obtained in in-house sVNT and pVNT in sera obtained on t0, t1, and t2.Values < 0 were set to 0, and the correlation coefficient was calculated according to Spearman.

Figure 4 .
Figure 4. Detection of antibodies reactive with the Omicron BA.1 variant in pVNT.(A,B) Consecutive sera of nine vaccinees obtained at time points t0 to t5 were tested for neutralising antibodies against wt S (A) and BA.1 S (B) by sVNT.(C) Differences in reactivity with the wt S and BA.1 S at time points t0 to t5.Horizontal lines represent the variable median.The significance of differences in reactivity at individual time points was determined by the Mann-Whitney test (p > 0.05: n.s., p ≤ 0.01: **, p ≤ 0.001: ***, p ≤ 0.0001: ****).

Figure 4 .
Figure 4. Detection of antibodies reactive with the Omicron BA.1 variant in pVNT.(A,B) Consecutive sera of nine vaccinees obtained at time points t0 to t5 were tested for neutralising antibodies against wt S (A) and BA.1 S (B) by sVNT.(C) Differences in reactivity with the wt S and BA.1 S at time points t0 to t5.Horizontal lines represent the variable median.The significance of differences in reactivity at individual time points was determined by the Mann-Whitney test (p ≤ 0.01: **, p ≤ 0.001: ***, p ≤ 0.0001: ****).

Figure 4 .
Figure 4. Detection of antibodies reactive with the Omicron BA.1 variant in pVNT.(A,B) Consecutive sera of nine vaccinees obtained at time points t0 to t5 were tested for neutralising antibodies against wt S (A) and BA.1 S (B) by sVNT.(C) Differences in reactivity with the wt S and BA.1 S at time points t0 to t5.Horizontal lines represent the variable median.The significance of differences in reactivity at individual time points was determined by the Mann-Whitney test (p > 0.05: n.s., p ≤ 0.01: **, p ≤ 0.001: ***, p ≤ 0.0001: ****).

Figure 5 .
Figure 5. Detection of antibodies reactive with the Omicron BA.1 variant in in-house sVNT.(A,B) Consecutive sera of nine vaccinees obtained at time points t0 to t5 were tested for antibodies reactive with wt RBD (A) and BA.1 RBD (B) by sVNT.(C) Differences in reactivity with the wt RBD and BA.1 RBD at time points t0 to t5.Horizontal lines represent the variable median.The significance of differences in reactivity at individual time points was determined by the Mann-Whitney test (p ≤ 0.01: **, p ≤ 0.001: ***, p ≤ 0.0001: ****).
1 − OD value of the sample/OD value of the negative control) × 100%.Values below the cut-off threshold of 30% are indicative of a negative result; values equal to or exceeding the cut-off indicate the presence of SARS-CoV-2 neutralising antibodies.

Table 1 .
(A) List of expression plasmids.(B) List of primers.

Table 2 .
Reactivity of secreted NLuc-tagged S antigens in heavy-chain capture EIA.
1,2The cut-off to achieve 100% specificity at time point t0 and the ability to distinguish between time points t0 (immediately before vaccination) and t1 (four weeks after the first vaccination) was defined by ROC curve analysis and calculation of AUC values.

Table 3 .
Sensitivity of sVNT and pVNT in sera obtained at t0 to t2.

Table 3 .
Sensitivity of sVNT and pVNT in sera obtained at t0 to t2.