Expression of the SARS-CoV-2 receptor-binding domain by live attenuated influenza vaccine virus as a strategy for designing a bivalent vaccine against COVID-19 and influenza

Two African green monkey cell lines (Vero ATCC CCL-81 and Vero E6 ATCC C1008), and canine kidney cell culture (MDCK, ATCC CCL-34) were maintained according standard protocol in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1×antibiotic-antimycotic solution (all from Capricorn, Germany) at 37°C and 5% CO₂.

A previously rescued live attenuated influenza vaccine (LAIV) virus served as a viral vector for generation of recombinant influenza viruses expressing immunogenic fragments of SARS-CoV-2 [17]. This virus carries hemagglutinin (HA) and neuraminidase (NA) genes of A/17/Anhui/2013/61 (A/Anhui/1/2013-based LAIV strain) and the remaining six genes from A/Leningrad/134/17/57 (H2N2), a master donor virus for LAIV. An H7N9 LAIV virus expressing truncated NS1 protein was used here as an additional vector control; generation and main characteristics of this vaccine virus were previously reported [17].

A reassortant influenza virus PR8-IDCDC-RG32A (Sh/PR8) carrying HA and NA genes of A/Shanghai/2/2013 (H7N9) virus and the remaining genes of A/PR/8/34 (H1N1) strain was obtained from the Centers for Disease Control and Prevention (CDC, Atlanta, GA, USA).

Influenza viruses were propagated in 10–11-day-old embryonated chicken eggs at 33°C (for attenuated viruses) or 37°C (for Sh/PR8 virus), clarified by low-speed centrifugation and stored at -70°C. For virus concentration, the clarified allantoic fluid was subjected to ultracentrifugation on a 30%/60% sucrose gradient, as previously described [18].

The SARS-CoV-2 virus, HCoV-19/Russia/StPetersburg-3524/2020 (Wuhan lineage, D614G, GISAID ID EPI_ISL_415710) was obtained from the Smorodintsev Research Institute of Influenza (Saint Petersburg, Russia). This virus was cultured in Vero (ATCC CCL-81) cells grown in DMEM supplemented with 2% FBS (DMEM/2% FBS) with addition of 1× antibiotic-antimycotic solution (Capricorn, Germany) and 10 mM HEPES at 37°C and 5% CO₂. Cell supernatants were harvested 72-96 h after inoculation and aliquoted into single-use stock vials after centrifugation at 3500 rpm for 15 min. The sucrose-gradient purified viruses were obtained using a previously described method of the high-speed centrifugation in sucrose gradient [18].

The recombinant RBD of spike protein of SARS-CoV-2 (Wuhan lineage) was stably expressed in HEK293 cells and provided by JSC «BIOCAD» [19]. The recombinant full-length HA protein of A/Shanghai/2/2013 (H7N9) virus was kindly provided by Professor Florian Krammer (Mount Sinai School of Medicine, New York, USA).

For cell stimulation in ELISPOT analyses, the pools of SARS-CoV-2 peptides – PepTivator® SARS-CoV-2 Prot_S, Prot_N (Miltenyi Biotec, Germany) – were used. All proteins were stored at −70°C in aliquots.

The infectious virus titers were assessed in 10-11 days old developing chicken embryos at 33°C for 72h, and the titers were calculated according to the Reed and Muench method [24] and expressed as log₁₀EID₅₀/mL. The genetic stability of the recombinant influenza viruses was assessed after 3, 5 and 10 sequential passages in eggs. The identity of the chimeric influenza virus genes containing SARS-CoV-2 inserts was evaluated by Sanger sequencing.

The RBD expression in MDCK cells infected with the recombinant viruses was assessed by sandwich ELISA of lysed cells with anti-RBD antibodies. MDCK monolayers were infected with test viruses at 0.002 MOI and incubated at 33°C, 5% CO2. At 60 hours post infection, cells were lysed with lysis buffer (250 mM sucrose, 50 mM Tris-HCl, 25 mM NaCl, 2 mM EDTA, 0.1% Triton X-100, 1 µg/mL trypsin inhibitor, 1 mM PMSF) on ice for 5 minutes with periodic mixing. Cell debris was pelleted by centrifugation at 13000 g and 4°C for 10 minutes. The supernatant was tested by ELISA in high-sorbent 96-well plates coated with rabbit polyclonal anti-RBD antibody (BIOCAD, Russia), 100 ng/well. Fourfold dilutions (from 20 µg/ml to 1.2 ng/mL) of a recombinant RBD protein expressed in mammalian cells [19] was used for standard curve generation. After blocking with 5% skim milk, the cell lysate supernatant was added in triplicates and incubated for 1h, followed by 2× washing with PBST. Then, a mouse monoclonal anti-RBD antibody (kindly provided by Dr Alexey Sokolov, FSBSI “IEM”, St. Petersburg) was added to the plates at a concentration of 1 mg/mL. After 1-hour incubation and washing, HRP-conjugated goat anti-mouse antibody (BioRad, USA) was added to the wells and incubated for 1 hour at 37°C. Plates were washed with PBS-T and the color was developed with 1-Step TMB Substrate Solution (HEMA, Russia). The reaction was stopped with 1 M H₂SO₄, and the resulting absorbance was measured at wavelength 450 nm (OD₄₅₀) using xMark Microplate Spectrophotometer (BioRad, USA). The RBD expression level was calculated by approximation of the mean OD₄₅₀ value to the standard curve generated from the recombinant RBD protein.

The SDS-PAGE and Western blot analysies were performed with sucrose-purified viruses taken at an equal protein concentration of 0.5 mg/mL of each. The discontinuous SDS-PAGE was performed under non-reducing conditions according to the previously described method [25] in 5% stacking and 10% resulting polyacrylamide gels loaded with the samples, which were prepared by mixing of sucrose gradient-purified viruses with a loading Laemmli buffer. The resulting gels were blotted as previously described [26] on the nitrocellulose membranes with a pore diameter of 0.45 μm. The membranes, blocked with 5% skimmed milk on PBST for 1 h at 37°C, were overnight-incubated with rabbit polyclonal anti-RBD antibodies (5 µg/ml in blocking buffer; Bio-Rad, USA) or mouse polyclonal anti-H7 serum raised to the recombinant H7 HA protein (1:200) at 4°C. The next day, the membranes were washed three times and immunodetected with anti-mouse or anti-rabbit secondary antibodies conjugated with horseradish peroxidase (1:3000 in PBS-T, Bio-Rad, USA) for 1 h at 37°C. Finally, 0.05% solution of diaminobenzidine (Sigma, USA) in PBS containing 1% hydrogen peroxide was used to stain the treated membranes.

Female BALB/c mice were purchased from the Stolbovaya farm (Moskow region, Russia). Mice were immunized intranasally with 10⁶ EID₅₀ of each experimental vaccine strain, in a volume of 50 μL. Viral titers in respiratory tissues were assessed at 3 dpi. For this, mice were euthanized with isoflurane, and the nasal turbinates and lungs were aseptically collected and stored at -70°C. Organs were homogenized using a small bead mill (TissueLyser LT, QIAGEN, Germany) in 1 mL of sterile PBS. Titers were determined in eggs as described above and expressed as log₁₀EID₅₀/mL.

For immunogenicity studies, animals were immunized twice with the same vaccine dose, at a 3-week interval. Serum samples and spleens were collected 21 days after the second dose. Immune responses to influenza were studied in enzyme-linked immunosorbent assay (ELISA) against a sucrose-gradient purified whole influenza virus; antibody responses to the SARS-CoV-2 were assessed in ELISA against RBD recombinant protein and, as well as in a microneutralization (MN) test of live SARS-CoV-2.

Six- to eight-week-old female Syrian hamsters were intranasally immunized with 5×10⁶ EID₅₀ of the studied influenza viruses, or with PBS. To determine the ability of the vaccine prototypes to replicate in the upper (URT) and lower (LRT) respiratory tracts of immunized hamsters, nasal and lung tissue samples were collected on day 3 after the first vaccine dose (4 animals from each group). Viral titers in tissue homogenates were determined by titration in eggs as described above. To study antibody responses to influenza and SARS-CoV-2, animals were immunized twice with a 3-week interval with the 5×10⁶ EID₅₀ and the sera were collected at day 21 after the second immunization. To assess protective potential against influenza virus and SARS-CoV-2, groups of hamsters immunized twice with the recombinant virus, as well as with the H7N9 LAIV vector, and mock-immunized animals were i.n. challenged 3 weeks after the second dose with Sh/PR8 influenza virus at a dose of 10⁶ EID₅₀ or with SARS-CoV-2 Wuhan (D614G) at a dose of 10⁵ TCID₅₀. Influenza virus-infected animals were euthanized 4 days post challenge and viral titers in tissue homogenized were determined by titration in eggs as described above. SARS-CoV-2-challenged hamsters were monitored for weight loss and clinical symptoms for 5 days after infection. The scoring of behavior and clinical signs was assessed as the sum of declining of following signs: behavior in the cage (normal=0; depressed=1); behavior in the open area (normal=0; sluggish=1); reaction to taking in hands (normal=0; sluggish=1); fur condition (normal=0; lack of grooming=1); interest to food (normal=0, decreased=1). On day 5 after challenge, animals were euthanized and lungs, nasal turbinates and spleens were collected for virological, immunological and/or histopathological evaluations. Viral titers in the URT and LRT were determined by titration tissue homogenates in Vero cells and expressed as lg TCID₅₀/gram tissue. Spleens were used to measure recall T-cell immunity to SARS-CoV-2 antigens in the ELISPOT assay.

The levels of serum antigen-specific antibodies were assessed by ELISA. Briefly, 96-well high-binding polystyrene plates (Thermo Scientific, USA) were coated with sucrose-gradient purified H7N9 LAIV virus (16 HA units per well) or with recombinant RBD protein (100 ng per well) in carbonate-bicarbonate buffer (pH 7.4) for 20 hours at +4°C. The coated plates were 3 times washed with PBS-T, and 2-fold serum dilutions were added in duplicates for 1 h at 37°C. After intensive washing with PBS-T a solution of secondary anti-mouse (Bio-Rad, USA) or anti-hamster (Thermo Scientific, USA) IgG HRP-conjugated antibodies were added to the plates and incubated for 1 h at 37°C. Thoroughly washed plates were finally developed with a 1-Step TMB Substrate Solution (HEMA, Russia). The reaction was stopped with 1 M H₂SO₄, and the resulting absorbance was measured at 450 nm (OD₄₅₀) using xMark Microplate Spectrophotometer (BioRad, USA). Antibody titers were determined as the last serum dilution with the OD₄₅₀ value exceeding twice the mean OD₄₅₀ values of the control wells (no serum added).

The microneutralization test was performed as previously described [18]. In brief, 300 TCID₅₀ of SARS-CoV-2 virus were mixed with 2-fold dilutions of serum samples and incubated for 1 h, followed by mixture transfer to 96-well plates with confluent monolayers of Vero cells. After 1 h of incubation at 37°C, 5% CO₂, the inoculum was removed, culture medium containing corresponding serum dilutions was added to appropriate wells, and the plates were incubated for 48 h at 37°C, 5% CO_2. After incubation, medium was removed and cells were fixed with 2% formaldehyde in PBS solution, and virus replication was detected using ELISA with rabbit anti-RBD antibodies (BIOCAD, Russia) and secondary HRP-conjugated anti-rabbit IgG antibodies (BioRad, USA). The color was developed with a 1-Step TMB Substrate Solution (HEMA, Russia) and optical density was measured at wavelength 450 nm using xMark Microplate Spectrophotometer. The 50% inhibitory concentration (IC₅₀) was calculated with a four-parametric nonlinear regression method.

The IFN-γ response of isolated splenocytes and lung cells of Syrian hamsters on the 5th day after challenge was measured using an IFN-γ ELISPOT Plus kit (Mabtech, Sweden) according to the manufacturer's protocol and as described in [21]. Briefly, a pre-coated with monoclonal antibody ELISPOT plate was washed 4 times with sterile PBS (200 µl/well) and then incubated with the CR-10 medium (RPMI supplemented with 10% FBS, 5 mM HEPES, 1× antibiotic-antimycotic, and 50 μM β-mercaptoethanol) for 30 minutes at room temperature. Then, CR-10 media was removed and “cells + stimuli” mixtures were added to each well, followed by 18h incubation at 37°C, 5% CO2. 500,000 cells were stimulated either with 0.1 MOI of purified SARS-CoV-2 or with 1 MOI of purified influenza virus, or with PepTivator S + N mixture (30 pmol per peptide) (Miltenyi Biotec Bergisch Gladbach, Germany). The detection of spots was performed according to the manufacturer’s protocol using detection antibody and substrate solution. Color development was stopped by extensively washing in tap water. Before counting, the plate was left to dry overnight, then spots were counted in an AID vSpot Spectrum reader (Advanced Imaging Devices, Germany).

We designed SARS-CoV-2 RBD-based cassettes for incorporation into influenza HA molecules because this strategy was successfully used in our previous studies [20, 29]. We inserted the RBD-encoding fragment into the H7N9 influenza virus HA gene between the signal peptide-encoding sequence and the HA1 subunit of the molecule using the GGGGSGGGGS flexible linker. It was shown in previous experiments that cassettes in such constructs are expressed as a part of HA protein and exposed at the surface of the virion [20, 30]. According to our previous studies, the size of the cassette may have a significant impact on the virus growth characteristics and immunogenicity [31]. Therefore, we designed two variants of the SARS-CoV-2 RBD-based immunogenic cassettes of different lengths. The first variant, HA+RBD 194, contained the insertion of the SARS-CoV-2 spike protein’s 333-526 amino acid residues, as described in [32]. Another construct was designed based on the full-length RBD protein of the Wuhan strain which comprises 223 amino acid residues (319-541) of the spike protein [33] (Fig. 1).

We rescued two recombinant influenza viruses expressing chimeric HA proteins as shown in Fig. 1, and carrying intact N9 gene, as well as six remaining genes of A/Leningrad/17 LAIV master donor virus. The chimeric viruses encoding the inserts RBD 194 and RBD 223 were named FluCoVac-19 and FluCoVac-20, respectively (Table 1).

The rescued LAIV-RBD viruses were amplified in eggs, and their titers ranged from 8.3 to 8.4 lgEID₅₀/mL (Table 1). Although they were significantly lower than that of the H7N9 LAIV vector, such infectious activity of the recombinant strains is suitable for further manufacturing processes. Ten sequential passages in eggs of the rescued viruses revealed high genetic stability of the RBD 194 (FluCoVac-19) and RBD 223 (FluCoVac-20) inserts, since only a single substitution was found in each virus, both times within the flexible linker: GGGGSGRGGS in FluCoVac-19 virus and GGGGNGGGGS in FluCoVac-20 variant. It is very unlikely that these minor changes will affect the antigenicity of the chimeric HA molecule, since the RBD sequence remained unchanged.

We also designed a panel of RBD-based constructs for their expression in the target cells as part of the modified NS gene of the LAIV virus. We previously developed recombinant LAIV strains with modified NSs that successfully stimulated T-cell immunity to other respiratory pathogens, such as RSV [35] and human adenovirus [22]. Here, we explored different variants of influenza NS gene modifications, targeting strategies and transgene cassette processing pathways that were found to be prospective for other recombinant vaccines.

We tested two modifications of the NS gene sequence. In the first type, the NS1 coding region was truncated up to 126 residues, followed by the linker and the RBD-based insert. The noncoding fragment of the NS1 ORF was removed from the sequence, except for the regions necessary for splicing for NEP. In another type of construct, we removed all noncoding sequences of NS1 and added necessary sequence after the cassette insertion for full-length NEP ORF recovery, through the T2A self-cleavage site (Fig. 2A). This strategy was previously studied by DiPiazza et al. [36]. In addition to the intact RBD fragment, we designed four variants of the RBD inserts with different targeting sequences that were supposed to enhance the humoral immune response to the transgene.

Two strategies were used for SARS-CoV-2 cassette targeting to lysosomal compartment. A human LAMP-1 transmembrane C-end peptide sequence (35 amino acid residues) was added to the C-end of the RBD cassette (Fig. 2B). Early studies on DNA antiviral vaccines demonstrated that the addition of this peptide to an antigen leads to enhanced immunogenicity [37‐39]. The mechanism is based on enhanced antigen presentation through the MHC class II pathway. In the second variant, the HLA-II invariant chain with transmembrane anchor domain was added to the N-end of the RBD cassette to serve as a signal of lysosomal targeting [40]. The role of the HLA-II invariant chain in the regulation of Th immune responses [41] suggested using it as a part of the transgene to modulate immune responses to DNA vaccination. In the mouse model, incorporation of CD4+ peptide into the HLA-II invariant chain led to high level of Th immune response after the peptide stimulation [42].

The other two designs included the tissue plasminogen activator (tPA) signal peptide at the N-terminus of the RBD cassette, and one of them also included the HER-2 transmembrane domain (TMD) and the IL2Ra cytoplasmic domain (CPD) fused to the C-end of RBD cassette (Fig. 2B). The tPA signal peptide is a signal sequence often used to transport the attached proteins towards the secretion pathway in mammalian cells. This signal peptide was used in several studies and its addition enhanced the immunogenicity of different constructs (MVA-based vaccine against tuberculosis, DNA vaccine with HIV-1 antigen, DNA vaccine against influenza) [43‐45]. The idea is based on direct presentation of the target protein to immune cells in the bloodstream. In our case, the RBD structure was further stabilized by additional 55 residues of the spike protein and a single mutation (see below, [19]).

The usage of membrane targeting is a common strategy for enveloped viral vector presentation of foreign antigens. This strategy is widely used in recombinant baculoviruses [46]. The membrane targeting of the antigen in vector vaccines can improve immunogenicity even if the antigen is not originally membrane-bound [47]. The effect of different cytoplasmic domains on the protein’s effective surface presentation and overall vaccine potential was also studied [48]. Here, we tested the erbB-2 (HER-2) transmembrane domain fused with the alpha-subunit of the IL-2 receptor (IL2Ra) cytoplasmic domain, which was described earlier [49] (Fig. 2B). This combination was proven to be useful in previous experiments on cell lines development, e.g., for overexpression of human Fc receptors on the surface of CHO and HEK293 cells (data not shown). Also, there are no tyrosine residues in IL2Ra cytoplasmic domain, and therefore there is no chance of interference with complex signaling pathways in target cells, unlike the biologically active HER2 cytoplasmic domain.

Therefore, we tested several different approaches to trigger and enhance the immune response toward the key receptor-binding domain of the SARS-CoV-2 spike protein, along with the response to the influenza virus.

We also tested three ways to insert the RBD-based cassettes into truncated NS1 ORF (Fig. 2A): (i) using the P2A self-cleavage site previously described for designing T-cell-based vaccines and which ensured independent processing of influenza virus antigens and the inserted transgene [21, 35]; (ii) via the pentanucleotide Stop-Start codon (TAATG) used by influenza B virus to terminate and reinitiate translation, which was tested earlier to express reporter GFP or functional IL-2 [50] or RSV epitopes [51, 52] from the truncated NS1 ORF; and (iii) by fusing the NS₁₂₆ protein with the RBD fragment through a flexible linker, so that the both proteins are expressed without disintegration. This strategy was shown to be promising in several studies of influenza viruses as viral vectors [53‐55].

Of note, most of our RBD inserts were designed to have a prolonged region of the spike protein, comprising of residues 319-596 (278 amino acids), since this modification was proven to be highly immunogenic when expressed by an adeno-associated virus (AAV) vector [19].

Overall, we rescued seven recombinant LAIV viruses expressing various RBD-based cassettes from the modified NS1 protein ORF, as listed in Table 2. FluCoVac-35, FluCoVac-41, FluCoVac-59 and FluCoVac-72 encoded different RBD cassettes following the P2A autocleavage site, whereas FluCoVac-78 and FluCoVac-79 were connected to the NS1 protein fragment via the Stop-Start pentanucleotide. The last variant, FluCoVac-83, encoded the RBD antigen in-frame with the NS1 and was separated from the influenza protein by the flexible linker.

The replicative activity of the recombinant viruses in chicken embryos was compared to that of the modified H7N9 LAIV virus encoding truncated to 126 residues NS1 protein, which was characterized earlier [17]. Most of the chimeric influenza viruses replicated efficiently in eggs, except FluCoVac-79 variant (Table 2). It is likely that this combination of the RBD cassette and the TAATG linking region interfered with the infectious activity of the chimeric virus. Interestingly, the same RBD cassette with SP, TMD and CPD inserted via the P2A self-cleavage site had no negative effect on the growth properties of the recombinant virus; in fact, this variant, FluCoVac-72, replicated better in eggs than the control LAIV NS₁₂₆ vector virus (Table 2).

Serial passaging of the rescued LAIV/RBD variants revealed high level of genetic stability of all but one recombinant virus (Table 2). Strikingly, the only variant that encoded the full-length NEP following the RBD cassette and the T2A autocleavage site (FluCoVac-41) was unstable and the insert was not detected in the virus after six passages in eggs. However, more research is needed to elucidate the exact genetic mechanisms underlying this phenomenon.

The expression of correctly folded RBD protein in MDCK cells infected with experimental vaccine strains was evaluated by sandwich ELISA of cell lysates. The productive infection of the cell with each recombinant virus was confirmed by hemagglutination assay of cell supernatants, as well by the detection of cytopathic effect in each virus-infected well. Unexpectedly, the expression of high levels of RBD protein was detected only in cells infected with two variants with RBD insertions into the influenza HA molecule – FluCoVac-19 and FluCoVac-20 (Fig. 3). No significant expression of the RBD was detected in MDCK cells infected with any of the recombinant viruses with insertions into the NS1 ORF, indicating that synthesis of the RBD protein from the NS1 open reading frame does not result in proper folding of the target antigen within the infected cell (Fig. 3).

Since RBD fragments inserted into HA molecule of influenza virus are supposed to be exposed on the surface of the virion, we conducted Western blot analyses of the sucrose gradient-purified viruses FluCoVac-19 and FluCoVac-20, using a polyclonal anti-RBD rabbit antibody and a mouse hyperimmune sera raised to the recombinant H7 HA protein expressed in insect cells. The H7N9 LAIV vector, as well as recombinant RBD protein, were used as control antigens in this assay. As shown in Fig. 4A, three apparent bands reacting with anti-RBD antibodies were observed in the FluCoVac-19 virus, suggesting the presence of RBD antigen in complex with the monomeric, dimeric and trimeric influenza HA molecules. The recombinant RBD protein used as a positive control in this study was also detected by anti-RBD antibodies in monomeric and multimeric forms, each monomer with expected size about 35 kDA (rhombus at Fig. 4A). Unexpectedly, no anti-RBD antibody binding was detected in the case of the FluCoVac-20 variant (Fig. 4A), whereas clear RBD expression was noted when MDCK cells were infected with this virus (Fig. 3). The absence of the RBD fragment within the HA molecule of FluCoVac-20 was confirmed by Western blot with anti-H7 antibody: the HA bands in various forms in this variant were identical to the H7N9 LAIV control virus, whereas corresponding bands of the FluCoVac-19 virus appeared at higher molecular weight, confirming the presence of an additional fragment within this antigen (Fig. 4B). Since FluCoVac-20 encoded the RBD fragment within the chimeric HA gene, which was confirmed by Sanger sequencing of the purified virus material, and expressed significant quantities of RBD within infected MDCK cells, most likely that the RBD fragment in this virus is subjected to proteolytic cleavage post-translationally and is not exposed on the surface of the virion.