key: cord-290290-wyx9ib7s
authors: Sinegubova, Maria V.; Orlova, Nadezhda A.; Kovnir, Sergey V.; Dayanova, Lutsia K.; Vorobiev, Ivan I
title: High-level expression of the monomeric SARS-CoV-2 S protein RBD 320-537 in stably transfected CHO cells by the EEF1A1-based plasmid vector
date: 2020-11-05
journal: bioRxiv
DOI: 10.1101/2020.11.04.368092
sha: 
doc_id: 290290
cord_uid: wyx9ib7s

The spike (S) protein is one of the three proteins forming the coronaviruses’ viral envelope. The S protein of the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) has a spatial structure similar to the S proteins of other mammalian coronaviruses, except for a unique receptor-binding domain (RBD), which is a significant inducer of host immune response. Recombinant SARS-CoV-2 RBD is widely used as a highly specific minimal antigen for serological tests. Correct exposure of antigenic determinants has a significant impact on the accuracy of such tests – the antigen has to be correctly folded, contain no potentially antigenic non-vertebrate glycans, and, preferably, should have a glycosylation pattern similar to the native S protein. Based on the previously developed p1.1 vector, containing the regulatory sequences of the Eukaryotic translation elongation factor 1 alpha gene (EEF1A1) from Chinese hamster, we created two expression constructs encoding SARS-CoV-2 RBD with C-terminal c-myc and polyhistidine tags. RBDv1 contained a native viral signal peptide, RBDv2 – human tPA signal peptide. We transfected a CHO DG44 cell line, selected stably transfected cells, and performed a few rounds of methotrexate-driven amplification of the genetic cassette in the genome. For the RBDv2 variant, a high-yield clonal producer cell line was obtained. We developed a simple purification scheme that consistently yielded up to 30 mg of RBD protein per liter of the simple shake flask cell culture. Purified proteins were analyzed by polyacrylamide gel electrophoresis in reducing and non-reducing conditions and gel filtration; for RBDv2 protein, the monomeric form content exceeded 90% for several series. Deglycosylation with PNGase F and mass spectrometry confirmed the presence of N-glycosylation. The antigen produced by the described technique is suitable for serological tests and similar applications.

Humanity is faced with an unprecedented challenge -the Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which causes a severe respiratory illness -coronavirus disease 2019 (COVID- 19) pandemic.

Countries were sent to lockdown; people could not make informed decisions about the possibility of social contacts; the need for diagnostic tests is very high. Existing tests for SARS-CoV2 are reviewed in [1] .

At the beginning of the pandemic, PCR testing methods dominated since such test systems can be developed urgently, soon after the emergence of a new virus in the population. Among the disadvantages of PCR-tests is a high sensitivity to contamination and dependence on sampling's correctness, a high proportion of falsepositive signals. Unlike PCR diagnostics, serological testing gives positive results long after the event of infection, at least for several months. This testing method makes it possible to reliably determine whether a person is infected with the SARS-CoV-2, even in the absence of disease symptoms. We need serological tests, both in express format and screening tests based on ELISA. Serologic tests are also needed to detect convalescent plasma of therapeutic interest and assess emerging vaccines' effectiveness.

In order for serological testing to have a more significant predictive value, mapping of the epitopes to which neutralizing antibodies appear should be carried out, as was done for SARS-CoV1 [2] , аnd convalescent or postvaccinal sera should be massively tested for the presence of neutralizing antibodies, for example, with a surrogate virus neutralization test based on antibody-mediated blockage of ACE2-spike protein-protein interaction [3] or another that can be carried out on a relatively large scale.

The use of highly specific and high-affinity viral antigens is already a big step towards improving diagnostic accuracy.

The immunodominant antigen of SARS-CoV2 is the RBD domain of the spike protein [4] . Another antigen widely used for diagnostics -the nucleocapsid (N) protein -combines high sensitivity and low specificity; therefore, it needs accurate antigen mutagenesis to remove highly conserved areas without compromising affinity. Cases are described for SARS-CoV when the results of testing with N-protein were clarified using two subunits of spike protein [5] .

The coronaviruses' spike (S) protein forms large coronal-like protrusions on the virions surface, hence the name of the family Coronaviridae. The S protein plays a crucial role in receptor recognition, cell membrane fusion, internalization of viruses, and their exit from the endosomes. It is described in detail in the review [6] .

It consists of S1 and S2 subunits and, in the case of the SARS-CoV-2 virus, has 1260 amino acids [7] .

The S protein is co-translationally incorporated into the rough endoplasmic reticulum (ER) and is glycosylated by N-linked glycans. Glycosylation is essential for proper folding and transport of the S protein.

The S protein trimer is transported from the ER. Interacting with the M and E proteins S protein trimer is transported to the virus's assembly site. S protein is required for cell entry but not necessary for virus assembly [8] .

During their intracellular processing, S proteins of many types of coronaviruses, including SARS-CoV-2 and MERS-CoV, but not SARS-CoV, undergo partial proteolytic degradation at the furin signal protease recognition site with the formation of two subunits S1 and S2. Apparently, most of the S protein copies on the membrane of SARS-CoV-2 viral particles are trimers of S2 subunits that are incapable of interacting with the receptor. The full-length S protein trimer on the viral particle's surface also undergoes complex conformational rearrangements during the formation of the RBD-receptor complex and the virus's penetration into the cell.

The S protein homotrimer binds to the ACE2 dimer, detailed study of this interaction is available here [9] .

As part of the trimer, the spike protein's monomer "moves its head" -the S1 subunit can form the open or closed conformation; that is, it can have a raised fragment or a lowered RBD domain, this can influence the affinity of antibodies targeted to it.

The S protein of SARS-CoV-2 amino acid sequence is variable, with more than 200 and 18 relatively frequent S protein amino acid variations. A glycan shield is formed by N-linked glycans on the S protein surface, which is likely to help viral immune escape. In a comparative study of genome-wide sequencing data of natural isolates of SARS-CoV-2 [10] for the detected 228 variants of the S protein, all 22 potential Nglycosylation sites within the S protein's ectodomain were completely conserved, which confirms the importance of each of these sites for maintaining the integrity of the S protein oligosaccharide envelope.

It should be noted that not all of the 22 potential N-glycosylation sites are occupied, for S1 and S2 subunits, obtained from transiently transfected HEK293 cells [11] N-glycosylation events were experimentally confirmed only for 17 out of 22 sites, also at least one O-glycosylation site was experimentally found inside the RBD-domain area of the S1 subunit with the mucin-like structures. Non-vertebrate cells may be used to produce the S protein or its fragments; in this case, N-glycans are present mostly in the form of bulky high mannose or paucimannose structures, possibly blocking the interaction of antibodies with the folded S protein [12] . Computational modeling of the glycan shield, performed for the HEK293-derived S protein, revealed that in the case of human cells, around 40% of the protein's surface is effectively shielded from IgG antibodies [13] .

The use of full-length S protein for practical serological testing is nearly impossible due to its insolubility, caused by the presence of transmembrane domain. An artificial trimer of its ectodomain has been successfully used as an antigen in serological tests; however, such complex protein cannot be obtained in large quantities in mammalian cells, apparently due to the limitation on the folding of the trimerized abundantly glycosylated protein and subsequent difficulties in its isolation and purification. It is generally believed that the SARS-CoV-2 S protein receptor-binding domain is a minimal proteinaceous antigen, adequately resembling the immunogenicity of the whole spike protein. This domain contains only two occupied N-glycosylation sites [11] and 1-2 occupied O-glycosylation sites. It does not contribute to the trimer formation, and its surface is mostly unshielded. Isolated RBD's of the S proteins of beta-coronaviruses were produced in various expression systems.

Bacterial expression of the RBD from MERS-CoV produced no soluble target protein, refolding attempts also were unsuccessful [14] . Budding yeasts Pichia pastoris were the suitable host for the secretion of MERS-CoV RBD with at least two (from three) N-linked glycosylation sites present. Similar data were obtained for the RBD from SARS-CoV virus -removal of all N-glycosylation sites resulted in the sharp drop of protein secretion rate in the P. pastoris yeast, in the case of full RBD domain (residues 318-536), secretion of the unglycosylated target protein was stopped completely [15] . It may be proposed that the addition of N-glycans in these sites is needed for correct folding of the RBD in the ER of eukaryotic cells.

The SARS-CoV-2 S protein RBD, expressed in E.coli, also was detected only as inclusion bodies and was found to be unreactive even on blotting [16] . Hyperglycosylated yeast-derived SARS-CoV-2 RBD was obtained in reasonable quantities (50 mg/L in bioreactor culture) by the P. pastoris expression system and successfully used for mice immunization [17] . Unfortunately, yeast-derived glycosylated proteins contain immunogenic glycans and cannot be used for immune assays with human antibodies. Similarly, SARS-CoV-2 RBD may be produced in the Nicotiana benthamiana plant, resulting in non-vertebrate N-glycans addition, potentially reactive with human antibodies [18] .

Most early preprints and peer-reviewed articles describing the SARS-CoV-2 S protein and its RBD domain production methods were focused on transient transfection of HEK293 cells [11] [19] and purification of small protein lots in a very short time. For example, D. Stadlbauer [20] reports more than 20 mg/L target protein titer in transiently transfected HEK-293 cells. Simultaneously, the scalability of transiently transfected cell lines cultivation is still questionable, and gram quantities of RBD, needed for large scale in vitro diagnostic activity, may be produced only by stably transfected cell lines.

Previously we have developed the plasmid vector p1.1, containing large fragments of non-coding DNA from the EEF1A1 gene of the Chinese hamster and fragment of the Epstein-Barr virus long terminal repeat concatemer [21] and employed it for unusually high-level expression of various proteins in CHO cells, including blood clotting factors VIII [22] , IX [23] , and heterodimeric follicle-stimulating hormone [24] . CHO cells were successfully used for transient SARS-CoV RBD expression at 10 mg/L secretion level [25] . We have proposed that SARS-CoV-2 RBD, suitable for in vitro diagnostics use, may be expressed in large quantities by stably transfected CHO cells, bearing the EEF1A1-based plasmid.

p1.1-Tr2-RBDv1 construction. The RBD coding sequence was synthesized according to [26] . The DNA fragment encoding the RBDv1 ORF with Kozak consensus sequence and C-terminal c-myc and 6xHis tags were obtained by PCR using primers AD-COV-AbsF and AD-RBD-myc6HNheR (listed in Table 1 The Resulting pTM vector was sequenced as described above, available from Addgene, plasmid # 162783.

pTM-RBDv2 construction. RBD ORF was amplified using adaptor primers AD-SFR2-NheF and AD-SFR2-XmaR restricted by NheI and XmaI (Sibenzyme, Novosibirsk, Russia) and cloned into pTM vector, restricted by NheI and AsiGI (Sibenzyme, Novosibirsk, Russia). The resulting construct was sequenced using SQ-5CH6-F and SQ-MycH-R primers. pTM-RBDv2 is available from Addgene, plasmid # 162785

Plasmids for cell transfections were purified by the Plasmid Midiprep kit (Evrogen, Moscow, Russia) and concentrated by ethanol precipitation in sterile conditions.

The transgene copy number in the CHO genome was determined by the quantitative real-time-PCR (qPCR) as described in [22, 24] . Serial dilutions of p1.1-eGFP [21] or pGem-Rab1 plasmids were used for calibration curves generation. The weight of one CHO haploid genome was taken as 3 pg, according to [27] . Genomic 

Chinese hamster ovary DG-44 cells (Thermo Fischer Scientific) were cultured in the ProCHO 5 medium (Lonza, Switzerland), supplemented by 4 mM glutamine, 4 mM alanyl-glutamine and hypoxanthinethymidine supplement (HT) (PanEco, Moscow, Russia). Cells were grown as a suspension culture in sterile 125 ml Erlenmeyer flasks with vented caps, routinely passaged 3 to 4 days with centrifugation (300 g, 5 min) and seeding density 3-4*10 5 cells/ml.

The 50-80 µg of each plasmid were precipitated by the addition of 96% ethanol and 3M sodium acetate, washed with 70% ethanol, dried, and resuspended in 100 µl of sterile R-buffer, Neon transfection kit (Thermo 

Seeding cell culture was grown in 125 ml Erlenmeyer shake flasks with 30 ml of Lonza ProCHO 5 medium, supplemented with 4 mM glutamine, 4 mM alanyl-glutamine and 2-8 µM MTX until cell concentration exceeds 1-1.5 mln cells/ml. Cell suspension was transferred to four 250 ml Erlenmeyer flasks, each containing 60 ml of culture medium, and grown to the same cell density. The entire cell suspension was transferred to a single 2 L Erlenmeyer flask with 1 L culture medium, final seeding density 3-4*10 5 cells/ml. Cells were cultured for three days, on the fourth day of culture, daily glucose measurements were started.

Glucose concentration in the cell supernatant was measured by the Accutrend Plus system (Roche, Switzerland); if glucose level was below 20 mM, it was added up to 50 mM as the sterile 45% solution. The culture in 2 L flask was grown for 6 to 8 days until the cell viability, measured by trypan blue exclusion, dropped below 50%.

The clonal cell line was obtained by the limiting dilution method from the cell population, cultured in 8 µM MTX. Methotrexate was omitted in the culture medium for two 3 d passage before cloning. Cells were additionally split by 1:1 dilution 24 hours before the cloning procedure. Cells were diluted in EXCELL-CHO (Merck, Germany) culture medium supplemented with 4 mM glutamine, 4 mM alanyl-glutamine, HT and 10% of untransfected CHO DG 44 conditioned medium resulting in seeding density 0.5 cell/well, and the suspension was seeded into 96-well plates (200 μl/well). Plates were left undisturbed for 14 days at 37°C, 5% CO2 atmosphere. Wells with single colonies were screened by microscopy; well grown colonies were detached by pipetting and transferred to the wells of 12-well plate, containing 4 ml of the EXCELL-CHO, supplemented as described above and grown for 7 days undisturbed. Product titer was measured by ELISA, as described below, 6 wells with highest RBDv2 titer were used for further cultivation. Best-producing clonal cell lines were transferred to 125 ml Erlenmeyer flasks with the ProCHO 5 culture medium supplemented with 4 mM glutamine, 4 mM alanyl-glutamine and 8 µM MTX and after 5 days in suspension culture, the best producing clone was determined by measuring the product titer and cell concentration. 

SDS-PAGE was performed with the 12.5% acrylamide in the separating gel, in reducing conditions, if not stated otherwise, with the PageRuler prestained marker, 5 µl/lane (ThermoFisher Scientific). Gels were stained by the colloidal Coomassie blue according to [28] , scanned by the conventional flatbed scanner in the transparent mode as 16-bit grayscale images and analyzed by the TotalLab TL120 gel densitometry software (Nonlinear Dynamics, UK).

SDS-PAGE was performed as described above, protein transfer, blocking, hybridization and color development were done according to [29] using nitrocellulose transfer membrane (GVS Group, Bologna, Italy) and Towbin buffer with methanol. Primary anti-c-myc antibody (SCI store, Moscow, Russia #PSM103-100) was used at the 1:2000 dilution, anti-mouse-HRP conjugate (Abcam, Cambridge, UK, ab6789) was used at 1:2000 dilution; membrane was developed by the DAB-metal substrate and scanned by the flatbed scanner in the reflection mode.

Multimeric forms of the RBD were quantified by size exclusion chromatography, utilizing Waters Extracts were vacuum-dried and redissolved in the 0.5% trifluoroacetic acid (TFA), 3% ACN solution.

Prepared solutions were mixed at 3:1 ratio with 20% α-cyano-4-hydroxycinnamic acid (Merck) solution in 20% ACN, 0.5% TFA on the target plate.

Solutions of intact and deglycosylated proteins were passed through the ZipTip C18 microcolumns (Millipore), washed and eluted according to manufacturer protocol. One and a half µl of protein solutions were mixed on the target plate with 0.5 µl of the 20% 2,5-dihydroxybenzoic acid (Merck) solution in 20% ACN, 0.5% TFA. Mass spectra were obtained by the MALDI-TOF mass spectrometer Ultraflextreme peptides identification was performed by the GPMAW 4.0 software (Lighthouse data, Denmark) and by the Mascot server (Matrix Science, Boston, USA). Glycopeptides mass assignment was performed by the GlycoMod online software tool [30] .

Sandwich ELISA with anti-S protein antibodies was performed using a prototype of the SARS-CoV-2 antigen detection kit (Xema Co., Ltd., Moscow, Russia, a generous gift of Dr. Yuri Lebedin). Pre-COVID-19 normal human plasma sample (Renam, Moscow, Russia) was used for preparation of the SARS-CoV-2 negative serum sample. Serum samples of five patients with the PCR-confirmed SARS-CoV-2 infection were pooled for testing and one serum sample with the borderline IgG titer level was tested separately. The blood sampling protocol conformed to the local hospital human ethics committee guidelines.

Antibody capture ELISA with human serum samples was performed according to [29] at the 100 ng per well antigens load. Antigens were applied on ELISA 96-well plates (Corning, USA) overnight at + 4oC, in PBS, 

The t-test was performed using the GraphPad QuickCalcs Web site: https://www.graphpad.com/quickcalcs/ttest1.cfm (accessed November 2020).

The native N-terminal signal peptide of SARS-CoV-2 S protein (amino acid sequence MFVFLVLLPLVSSQ) was fused to the RBD sequence (319 -541, according to YP_009724390.1) and joined with a C-terminal c-myc epitope (EQKLISEEDL), short linker sequence, and hexahistidine tag. N-terminal part of the RBDv1 gene was constructed according to [26] , utilizing the optimized codon usage gene structure. C-terminal tags were not optimized for codon usage frequencies. The resulting synthetic gene was cloned into the p1.1-Tr2 vector plasmid, a shortened derivative of the p1.1 plasmid [21] , and used for transfection of DHFR-deficient CHO DG44 cells. The resulting expression plasmid p1.1-Tr2-RBDv1 [GenBank: MW187858] is shown on Fig 1A. The stably transfected cell population was obtained by selection in the presence of 200 nM of DHFR inhibitor methotrexate, RBD titer 0.33 mg/L was detected for 3-days culture (Fig. S1 ). One-step target gene amplification was performed by increasing the MTX concentration tenfold and maintaining the cell culture for 17 days until cell viability restored to more than 85%; the resulting polyclonal cell population could secrete up to 3,0 mg/L RBD in the 3-days culture. The target protein was purified by a single IMAC chromatography step, utilizing the IDA-based resin Chelating Sepharose Fast Flow (Cytiva), Ni2+ ions, and step elution by increasing imidazole concentrations (Fig 1B, Fig 1C) . The resulting protein production method was found to be sub-optimal due to unexpectedly low secretion rate, signs of cellular toxicity of the target gene -33 h cell duplication time, maximal cell density in shake flask of 2.3 mln сells/ml (Fig S2) , and unacceptable level of contaminant proteins co-eluting with the RBDv1. At the same time, the RBDv1 protein was stable in the culture medium during the extended batch cultivation of cells for at least 7 days (Fig 1D) , making the long-term feed batch cultivations a viable option for its production in large quantities.

We proposed that target protein secretion rate and its purity after one-step purification could be significantly improved by a simultaneous shift of the RBD domain boundaries, exchange of the SARS-CoV-2 S protein native signal peptide to the signal peptide of more abundantly expressed protein, two-step genome amplification and switch from IDA-based resin to the NTA-based one (Fig 1E) .

Human tissue plasminogen activator signal peptide (hTPA SP, amino acid sequence MDAMKRGLCCVLLLCGAVFVSAS) is commonly used for heterologous protein expression in mammalian cells. It was successfully used for the expression of SARS-CoV S protein in the form of DNA vaccine [31] and envelope viral protein gp120 [32] .

In the case of MERS-CoV S protein RBD -Fc fusion protein, various heterologous signal peptides modulate target protein secretion rate by the factor of two [14] .

Corrected boundaries of the SARS-CoV-2 RBD were determined according to the cryo-EM data [PDB ID: 6VXX] [33] obtained for the trimeric SARS-CoV-2 S protein ectodomain. Initially used 319 -541 coordinates, described in the [26] include one unpaired Cys residue originated from the N-terminal part of the next domain SD1 (structural domain 1), so we excluded Lys319 from the N-terminus of the mature RBD protein, aiming at the maximization of signal peptide processing, and removed C-terminal aminoacids C 538 VNF 541 , which form the structure of the SD1 domain. Both linker areas surrounding the folded RBD domain core remain present in the RBDv2 protein (320 -537, according to YP_009724390.1). Additionally, we redesigned C-terminal tags by introducing the Pro residue immediately upstream of the c-myc tag, adding the short linker sequence SAGG between the c-myc tag and polyhistidine tag, and extending the polyhistidine tag up to 10 residues. We expected this structure to expose the c-myc tag properly on the protein globule's surface and move the decahistidine tag away from possible masking negatively charged protein surface areas.

We constructed an expression vector pTM [GenBank: MW187855], where consensus Kozak sequence, hTPA SP and c-myc and 10-histidine tags are coded in the polylinker. RBD coding fragment was cloned in-frame, resulting pTM-RBDv2 expression plasmid [GenBank: MW187856] is shown on Fig.2A .

CHO DG44 cells were transfected by the pTM-RBDv2 plasmid, stably transfected cell population was established at the 200 nM MTX selection pressure. Target protein titer was similar to the previous plasmid design -0.9 mg/L for 3-days culture, but after one step of the MTX-driven genome amplification, it increased eleven-fold to 9.7 mg/L at 2 µM MTX ( Fig 2B) and then increased by a factor of 2.5 after second amplification step at 8 µM MTX, resulting titer was 24.6 mg/L for 3-days culture (Fig 3A) . A steady increase of the target protein titer was detected for the extended batch cultivation of polyclonal cell population obtained at 8 µM MTX, peaking at 50 mg/L at 8 days of cultivation in the 2 L shake flask (Fig 3D, 3E) . A similar ratio of product titer increase after multi-step MTX-driven genome amplification was described for the MERS-CoV RBD -40-fold increase after 9 steps of consecutive increments of MTX concentration, overall amplification period length was 60 days [14] . Vector plasmid pTM, used in this study, allowed a much more rapid amplification course -a 27-fold titer increase in two steps, 33 days total. This All cell populations, secreting RBD proteins, were analyzed by the quantitative PCR and it was found, that increased productivity of populations, adapted to higher concentrations of MTX corresponds to higher copy numbers of target gene (Fig 3C) . Higher cell productivity in the case of RBDv2 protein was not due to higher target gene copy numbers, then in the case of RBDv1.

Cell culture medium Pro CHO5 (Lonza), utilized in this study, contains unknown components, blocking Histagged RBD protein's interaction with the Ni-NTA chromatography resin. Clarified conditioned medium, used for protein purification, was concentrated approximately tenfold by tangential flow ultrafiltration on the 5 kDa MWCO cassettes and completely desalted by diafiltration, 20 diafiltration volumes of the 10 mM imidazole-HCl, pH 8.0 solution. RBDv1 and RBDv2 proteins were purified by IMAC utilizing Ni-NTA Agarose (Thermo Fischer Scientific, USA) in the same conditions. Desalted conditioned medium was applied onto the column in the presence of 10 mM imidazole; the column was washed by the solution containing Elution was performed by the 300 mM imidazole solution; further column strip by the 50 mM EDTA-Na solution revealed no detectable target protein RBDv2 in the eluate (Fig 2C) . Purified proteins were desalted by another round of ultrafiltration/diafiltration on the centrifugal concentrators with 5 kDa MWCO membranes; diafiltration solution was PBS; final concentration 3-7 mg/ml. Purified proteins were flashfrozen in liquid nitrogen and stored frozen in aliquots. Overall protein yield for RBDv2 was 64%, 32 mg of purified RBDv2 were obtained from 1 L shake flask culture.

The apparent molecular weight of intact RBDv1 was determined as 35.3 kDa, deglycosylated RBDv1 -26.1 kDa, theoretical molecular weight -27647 Da. RBDv2 molecular weight was determined as 35.7 kDa for the intact protein, deglycosylated protein -28.5 kDa, theoretical molecular mass -27459 Da (Fig 4A) .

Both protein variants possess two distinct forms of intramolecular disulfide bonds sets, visible as two closely adjacent bands in non-reducing conditions and complete absence of such band pattern in reducing conditions.

Previously it was reported that SARS-CoV-2 RBD 319-541, expressed transiently in HEK-293 cells, tends to form a covalent dimer, around 30% from the total, visible as the 60 kDa band on the denaturing gel in nonreducing conditions [19] . We confirmed this observation; in the case of stably transfected CHO cells, covalent dimerization was also 31% according to gel densitometry data. At the same time, it should be noted that the RBDv2 protein, redesigned explicitly for mitigation of this unwanted dimerization and containing an even number of Cys residues, still forms 6 % of the covalent dimer.

Purified RBDv2 was tested by size exclusion chromatography. The major monomer form's apparent molecular weight was determined as 32.4 kDa (Fig S3) , admixtures peaks apparent molecular masses corresponded well to RBD dimer, tetramer, and two high molecular mass oligomers accounting for 6% of all peak areas (Fig 4B) .

Mass-spectrometry analysis of RBDv1 and RBDv2 revealed that both proteins' molecular masses diminished ( Fig S5, S6 ). This long peptide was completely absent in both spectra of de-glycosylated proteins (Table S1 -S4) . A more detailed analysis of this area of the RBD protein may be of some interest for the S protein structure-function investigation but is out of scope for the present study.

Purified RBD variants were used as antigens for microplates coating and subsequent direct ELISA with pooled sera obtained from patients with the RT-PCR-confirmed COVID-19 diagnosis, 1 weakly positive serum sample from the RT-PCR-confirmed COVID-19 patient, and serum sample obtained from a healthy volunteer before December 2019 (Fig 4E) . Both RBD variants perform equally -all serum samples produce highly similar OD readings for all dilutions tested with both antigens.

Here we describe a method of generating stably transfected CHO cell lines, secreting large quantities of monomeric SARS-CoV-2 RBD, suitable for serological assays. At present, serological assays for detection of seroconversion upon SARS-CoV-2 infection are mostly based on two viral antigens -nucleoprotein (NP) and S protein or fragments of the S protein, including the RBD. There are various reports on the specificity and sensitivity of assays based on these two antigens. In some cases, the sensitivity of clinically approved NPbased assays was challenged by direct re-testing of NP-negative serum samples by the RBD-based assays [34] . Other studies question the specificity of NP-based ELISA tests, demonstrating a significant level of false-positive results for the full-length SARS-CoV-2 NP [35] . It may be proposed that testing of serum samples with both SARS-CoV-2 antigens will produce the most accurate results, as was done, for example, in the South-East England population study [36] ; this conclusion was made in the microarray study of a limited number of patients serum samples [37] .

It is unclear yet, which part of the S protein is the optimal antigen for serological assays; microarray analysis revealed that S2 fragment generates more false-positive results than S1 or RBD antigen variants [37] in the case of IgG detection, at the same time the RBD protein generated much lower signals on COVID-19 patients serum samples then S1 or S1+S2 antigens. In another microarray study it was found that IgG response toward the RBD domain in the convalescent plasma samples correlates well with the response toward full-length soluble S protein [38] . In the conventional ELISA test format, RBD demonstrated nearly 100% specificity and sensitivity on a limited number of SARS-CoV-2 patients and control serum samples [4] . As of 26.10.20, at least 104 various immunoassays for SARS-CoV-2 antibodies were authorized for in vitro diagnostic use in the EU [39], many of them use RBD as the antigen. A simple ELISA screening test with the 96-well microplate will consume around 10 µg of the RBD antigen for 40 test samples, so even one million tests will require 250 mg of the purified RBD protein, making the antigen supply a critical step in the production of such tests. Method of the generation of highly productive stably transfected CHO cell line, secreting the RBD protein, may be important for IVD test manufacturers in securing the sources of RBD antigen with highly predictable properties.

Although the RBD fragment of the S protein from SARS-CoV-2 is not the most popular antigen variant in the current efforts of anti-SARS-CoV-2 vaccine development [40] , it may be considered as the viable candidate for a simple subunit vaccine. It demonstrated the significant protective immune response development in rodents, without signs of ADE effect [41] and some RBD-based protein subunit vaccine have advanced to Phase II clinical trials. Cultured CHO cells are the reliable source of RBD protein for this kind of vaccines; at the productivity level achieved in our study, only 30 m 3 of cell culture supernatant will provide enough antigen material for 100 mln of typical 10 µg/vial vaccine doses. С, D -protein sequence coverage by tryptic peptides, MALDI-TOF analysis. Glycosylated peptides found are not pictured, signal peptides are yellow, detected tryptic peptides -violet, experimentally obtained masses, [M+H]+, are stated in the boxes. E -Immunoreactivity of RBDv1 and RBDv2 by ELISA with pooled serum samples from PCR-positive patients -(+)pooled, single serum sample from PCR-positive patient (+) and pre-COVID-19 pooled sera (-). All sera samples were analyzed in duplicates, data are mean.

Supporting Figure S1 . Cell growth and viability dynamics of initial selection and MTX-driven target gene amplification.

Supporting Figure S2 . Cell growth curve for the extended batch cultivation of RBDv1 and RBDv2producing cell populations, 2 uM MTX selection pressure.

Supporting Figure S3 . Size exclusion chromatography trace of molecular mass calibrators and molecular mass calibration curve.

Supporting Figure S4 . MALDI-TOF spectra traces of intact proteins in glycosylated and deglycosylated forms.

Supporting Figure S5 . MALDI-TOF spectra traces of tryptic peptides mxtures from intact and deglycosylated RBDv1.

Supporting Figure S6 . MALDI-TOF spectra traces of tryptic peptides mxtures from intact and deglycosylated RBDv2.

Supporting Table S1 . Peptides mass list of the RBDv1 intact protein, in-gel digestion, reduced protein.

Supporting Table S2 . Peptides mass list of the RBDv2 intact protein, in-gel digestion, reduced protein.

Supporting 

Molecular and Immunological Diagnostic Tests of COVID-19: Current Status and Challenges. iScience

Antigenic and immunogenic characterization of recombinant baculovirus-expressed severe acute respiratory syndrome coronavirus spike protein: implication for vaccine design

A SARS-CoV-2 surrogate virus neutralization test based on antibody-mediated blockage of ACE2-spike protein-protein interaction

The receptor binding domain of the viral spike protein is an immunodominant and highly specific target of antibodies in SARS-CoV-2 patients

False-positive results in a recombinant severe acute respiratory syndromeassociated coronavirus (SARS-CoV) nucleocapsid-based western blot assay were rectified by the use of two subunits (S1 and S2) of spike for detection of antibody to SARS-CoV

Structure, Function, and Evolution of Coronavirus Spike Proteins

Structural and functional properties of SARS-CoV-2 spike protein: potential antivirus drug development for COVID-19

Fenner and White's medical virology

Molecular interaction and inhibition of SARS-CoV-2 binding to the ACE2 receptor

Variations in SARS-CoV-2 Spike Protein Cell Epitopes and Glycosylation Profiles During Global Transmission Course of COVID-19. Front Immunol

Deducing the N-and O-glycosylation profile of the spike protein of novel coronavirus SARS-CoV-2

Site-specific N-glycosylation Characterization of Recombinant SARS-CoV-2 Spike Proteins

Analysis of the SARS-CoV-2 spike protein glycan shield reveals implications for immune recognition

Engineering a stable CHO cell line for the expression of a MERS-coronavirus vaccine antigen. Vaccine

Yeast-expressed recombinant protein of the receptor-binding domain in SARS-CoV spike protein with deglycosylated forms as a SARS vaccine candidate

Recombinant SARS-CoV-2 spike proteins for sero-surveillance and epitope mapping. bioRxiv

Structural and Functional Comparison of SARS-CoV-2-Spike Receptor Binding Domain Produced in <em>Pichia pastoris</em> and Mammalian Cells

Rapid production of SARS-CoV-2 receptor binding domain (RBD) and spike specific monoclonal antibody CR3022 in Nicotiana benthamiana

Purification of recombinant SARS-CoV-2 spike, its receptor binding domain, and CR3022 mAb for serological assay

SARS-CoV-2 Seroconversion in Humans: A Detailed Protocol for a Serological Assay, Antigen Production, and Test Setup

Improved elongation factor-1 alpha-based vectors for stable high-level expression of heterologous proteins in Chinese hamster ovary cells

Stable high-level expression of factor VIII in Chinese hamster ovary cells in improved elongation factor-1 alpha-based system

A Highly Productive CHO Cell Line Secreting Human Blood Clotting Factor IX

High-level expression of biologically active human follicle stimulating hormone in the Chinese hamster ovary cell line by a pair of tricistronic and monocistronic vectors

A 219-mer CHO-expressing receptor-binding domain of SARS-CoV S protein induces potent immune responses and protective immunity

A serological assay to detect SARS-CoV-2 seroconversion in humans

Eukaryotic genome size databases

Highly Sensitive and Fast Protein Detection with Coomassie Brilliant Blue in Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis

Antibodies : a laboratory manual

GlycoMod--a software tool for determining glycosylation compositions from mass spectrometric data

Identification of two neutralizing regions on the severe acute respiratory syndrome coronavirus spike glycoprotein produced from the mammalian expression system

Extracellular Matrix Proteins Mediate HIV-1 gp120 Interactions with alpha4beta7

Structure, Function, and Antigenicity of the SARS-CoV-2 Spike Glycoprotein

Testing for responses to the wrong SARS-CoV-2 antigen?

Whole Nucleocapsid Protein of Severe Acute Respiratory Syndrome Coronavirus 2 May Cause False-Positive Results in Serological Assays

Estimates of the rate of infection and asymptomatic COVID-19 disease in a population sample from SE England

Analysis of SARS-CoV-2 Antibodies in COVID-19 Convalescent Blood using a

): p. 3581. 39. database. Foundation for Innovative New Diagnostics. SARS-CoV-2 diagnostic pipeline

A systematic review of SARS-CoV-2 vaccine candidates

The SARS-CoV-2 receptor-binding domain elicits a potent neutralizing response without antibody-dependent enhancement

We thank Mr. Arthur Isaev (Genetico, Moscow, Russia) and Dr. Alexander Ivanov (Institute of Molecular biology Russian Academy of Sciences, Moscow, Russia) for valuable comments and early access to the SARS-CoV-2 S protein sequence data, Dr. Yuri Lebedin, Eugenia Kostrikina and Xema Co., Ltd., for providing anti-RBD mAbs and conjugates.The measurements were carried out on the equipment of the Shared-Access Equipment Centre "Industrial Biotechnology" the Research Center of Biotechnology of the Russian Academy of Sciences. DNA sequencing was carried out in the inter-institutional Center for collective use "GENOME" IMB RAS, organized with the support of the Russian Foundation of Basic Research.The authors would like to acknowledge all the doctors who diagnose and treat patients during the COVID-19 pandemic.

Primers for RBDv1 cloning, restriction sites are underlined AD-COV-AbsF AACCTCGAGGCCGCCACCATGTTCATGCCTTCTT AD-RBD-myc6HNheR GCTAGCCTAATGGTGATGGTGATGATGACCGGTATGCATAT TCAGATCCTCTTCTGAGATGAGTTTTTGTTCGAAGTTCACGC ATTTGTT Primers for pTM construction, sticky ends of annealed pairs are underlinedCTAGTGATGGTGATGGTGATGGTGATGGTGATGACCGCCTG CAGACAGATCCTCTTCGCTGATCAGTTTTTGTTCACCGGTA Primers for RBDv2 cloning, restriction sites are underlined AD-SFR2-NheF GCTAGCGTGCAGCCCACCGAATCC AD-SFR2-XmaR CCCGGGTTTGTTCTTCACGAGATTGGT Sequencing primers SQ-5CH6-F GCCGCTGCTTCCTGTGAC IRESA rev AGGTTTCCGGGCCCTCACATTG SQ-MycH-R GATGACCGCCTGCAGAC