Preclinical evaluation of manufacturable SARS-CoV-2 spike virus-like particles produced in Chinese Hamster Ovary cells

Abstract

Background: As the COVID-19 pandemic continues to evolve, novel vaccines need to be developed that are readily manufacturable and provide clinical efficacy against emerging SARS-CoV-2 variants. Virus-like particles (VLPs) presenting the spike antigen at their surface offer remarkable benefits over other vaccine antigen formats; however, current SARS-CoV-2 VLP vaccines candidates in clinical development suffer from challenges including low volumetric productivity, poor spike antigen density, expression platform-driven divergent protein glycosylation and complex upstream/downstream processing requirements. Despite their extensive use for therapeutic protein manufacturing and proven ability to produce enveloped VLPs, Chinese Hamster Ovary (CHO) cells are rarely used for the commercial production of VLP-based vaccines.

Methods: Using CHO cells, we aimed to produce VLPs displaying the full-length SARS-CoV-2 spike. Affinity chromatography was used to capture VLPs released in the culture medium from engineered CHO cells expressing spike. The structure, protein content, and glycosylation of spikes in VLPs were characterized by several biochemical and biophysical methods. In vivo, the generation of neutralizing antibodies and protection against SARS-CoV-2 infection was tested in mouse and hamster models.

Results: We demonstrate that spike overexpression in CHO cells is sufficient by itself to generate high VLP titers. These VLPs are evocative of the native virus but with at least three-fold higher spike density. In vivo, purified VLPs elicit strong humoral and cellular immunity at nanogram dose levels which grant protection against SARS-CoV-2 infection.

Conclusions: Our results show that CHO cells are amenable to efficient manufacturing of high titers of a potently immunogenic spike protein-based VLP vaccine antigen.

Fig. 1. S expression in CHO cells is sufficient to generate S-coated SARS-CoV-2 VLPs.

a Sedimented supernatants with 15% iodixanol cushion from CHO²³⁵³ cells transfected with S or M-E/S plasmids. On the left, total protein staining (Coomassie blue) of S or M-E/S. On the right, immunoblot with anti-Spike (S1) detecting S (~180 kDa). The SARS-CoV-2 membrane (M) protein is detected with anti-M at ~17 kDa. b Representative TEM image of sedimented M-E/S VLPs. White arrows indicate potential VLPs. A scale bar of 100 nm is shown at the bottom of the image. c Representative TEM image of sedimented VLPs generated by the expression of S protein (S-VLPs). A scale bar of 200 nm is shown at the bottom of the image. White arrows indicate VLPs. d Representative immunoblot of different S-VLPs variants detected with anti-S1. From left to right, mock, beta (B.1.351), delta (B.1.617.2), and omicron (B.1.1.529) S-VLPs. e Representative TEM images of sedimented S-VLPs variants. From left to right, beta (B.1.351), delta (B.1.617.2), and omicron (B.1.1.529). A 50 nm scale bar is shown at the bottom of each image.

Fig. 2. S protein is the main component in purified S-VLPs derived from CHO-C2 cells.

a CHO-C2 (ERVLP-KO) cells. On the left, total protein staining (Coomassie blue) showing supernatants (7.5 µL) and sedimented (0.35 µg of total protein) VLPs produced in CHO²³⁵³ (parental) or CHO-C2. On the right, Western blotting of VLPs produced by CHO²³⁵³ or CHO-C2. The upper part of the blot shows the full-length S protein detected by anti-Spike (S1) in clarified harvest or sedimented VLPs. The lower blot is stained with anti-Gag p30 and shows the absence of Gag in CHO-C2-derived VLPs. b Scatter plot of total S protein (mg/L) from different samples at 5 days of transfection. Data are presented as mean ± SEM of six independent samples. c Total protein staining comparison of sedimented and affinity-purified S-VLPs. On the left, 1 µg of recombinant, soluble S protein (S STD) was loaded. For VLPs, 0.7 µg of total protein was loaded per well. d Representative TEM image of affinity-purified S-VLPs. A 200 nm scale bar is shown at the bottom. e Representative high-resolution TEM image of individual affinity-purified S-VLPs acquired with a HITACHI HT7700 120. A 50 nm scale bar is shown at the bottom. f Scatter plot of total S protein (mg/L) from different purified samples at 5 days of transfection. Data are presented as mean ± SEM of four independent samples. g ELISA assay. Dose-response curve of purified S-VLP (starting concentration [S] = 0.081 mg/mL) showing 11 serial dilutions. Red dots represent S-VLPs on a hACE2-coated plate in an S-shaped sigmoidal curve in red. Blue dots represent S-VLPs on an uncoated (mock) plate connected by a blue line.

Fig. 3. S-VLPs characterization shows a high number of S proteins on the particle surface with a glycan pattern similar to soluble S trimmers.

a Individual S-VLP image (positive contrast) was acquired by Cryo-EM. The portrayed VLP contains 20 peripheral S proteins. Individual S proteins are evident in the magnified image, protruding from the envelope with an average size of 20 nm. A 25 nm scale bar is displayed at the bottom of each image. b Schematic of calculation workflow from the measured arc length between two peripheral spikes to real Euclidian distance between spikes on the VLP core surface. c Reconstruction of S distribution on VLPs using the real Euclidian distance. VLP models correspond to core diameters of 40 nm/35 spikes, 50 nm/50 spikes, 60 nm/67 spikes, and 70 nm/86 spikes. d Compositional glycan identification of the S-VLP and recombinant soluble (SmT2v3) S proteins. The identity and proportion of 20/22 N-glycans were determined by LC–MS and presented in pie charts. See also Supplementary Fig. 3 and Supplementary Data file.

Fig. 4. Immunization with adjuvanted S-VLPs induces SARS-CoV-2-neutralizing, protective responses in rodents.

a* Humoral immune response titers (total anti-spike IgG) in vaccinated mice with different S-VLP formulations. Anti-S geometric mean titers (GMT) at day 28 from naïve and treated groups were determined by ELISA. b* Quantification of spike-specific T cells. Isolated splenocytes from vaccinated mice with S-VLP formulations were co-incubated with a spike peptide library. Quantification of IFN-γ+ secreting cells was performed by ELISpot and results were expressed as the mean of IFN-γ+/10⁶ splenocytes. c** Cell-based (hACE2-HEK293T) surrogate neutralization assay. The percentage neutralization means for SARS-CoV-2 was determined from dilutions of 1:25 in unadjuvanted and Adju-phos or 1:75 in AS01b conditions. a–c Data are presented as geometric mean ± SEM (10 mice per group). Black dots represent naïve animals, blue for unadjuvanted, red for Adju-phos, and green for AS01b S-VLPs. d Body weight change in immunized hamsters infected with SARS-CoV-2 (ancestral strain). The line graph represents daily body weight measurements of naïve and vaccinated animals post-challenge. e** Virus quantification by plaque assay. Plaque forming units (PFUs) were quantified in Vero E6 cells infected with hamster lung homogenates. Virus load (log10) is displayed in PFU per gram of lung. f** Viral RNA quantification in hamster lungs. RNA samples from naïve and vaccinated animals at five days post-challenge were subjected to viral RNA detection by RT-qPCR. SARS-CoV-2 RNA concentration is displayed by gRNA copies/lung (log10). d–f Data are presented as mean ± SEM (6 animals per group). Black dots (gray bars) represent naïve animals, blue is unadjuvanted S-VLPs, orange AS01b and green represents S-VLPs+AS01b. *A one-way ANOVA with Tukey’s or **Dunnette’s multiple comparison tests was performed to assess significance. p ≤ 0.0001, p ≤ 0.001, p ≤ 0.01 and p ≤ 0.05 are represented by four, three, two, and one asterisks, respectively, and p > 0.05 as not significant (NS).