key: cord-306901-uuwgpuhw
authors: Roy, Sylvie; Ghani, Karim; de Campos-Lima, Pedro O.; Caruso, Manuel
title: Efficient production of Moloney murine leukemia virus-like particles pseudotyped with the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) spike protein
date: 2020-09-16
journal: bioRxiv
DOI: 10.1101/2020.09.16.298992
sha: 
doc_id: 306901
cord_uid: uuwgpuhw

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) outbreak that started in China at the end of 2019 has rapidly spread to become pandemic. Several investigational vaccines that have already been tested in animals and humans were able to induce neutralizing antibodies against the SARS-CoV-2 spike (S) protein, however protection and long-term efficacy in humans remain to be demonstrated. We have investigated if a virus-like particle (VLP) derived from Moloney murine leukemia virus (MLV) could be engineered to become a candidate SARS-CoV-2 vaccine amenable to mass production. First, we showed that a codon optimized version of the S protein could migrate efficiently to the cell membrane. However, efficient production of infectious viral particles was only achieved with stable expression of a shorter version of S in its C-terminal domain (ΔS) in 293 cells that express MLV Gag-Pol (293GP). The incorporation of ΔS was 15-times more efficient into VLPs as compared to the full-length version, and that was not due to steric interference between the S cytoplasmic tail and the MLV capsid. Indeed, a similar result was also observed with extracellular vesicles released from parental 293 and 293GP cells. The amount of ΔS incorporated into VLPs released from producer cells was robust, with an estimated 1.25 μg/ml S2 equivalent (S is comprised of S1 and S2). Thus, a scalable platform that has the potential for production of pan-coronavirus VLP vaccines has been established. The resulting nanoparticles could potentially be used alone or as a boost for other immunization strategies for COVID-19. IMPORTANCE Several candidate COVID-19 vaccines have already been tested in humans, but their protective effect and long-term efficacy are uncertain. Therefore, it is necessary to continue developing new vaccine strategies that could be more potent and/or that would be easier to manufacture in large-scale. Virus-like particle (VLP) vaccines are considered highly immunogenic and have been successfully developed for human papilloma virus as well as hepatitis and influenza viruses. In this study, we report the generation of a robust Moloney murine leukemia virus platform that produces VLPs containing the spike of SARS-CoV-2. This vaccine platform that is compatible with lyophilization could simplify storage and distribution logistics immensely.

(MLV) could be engineered to become a candidate SARS-CoV-2 vaccine amenable to mass 23 production. First, we showed that a codon optimized version of the S protein could migrate 24 efficiently to the cell membrane. However, efficient production of infectious viral particles was only 25 achieved with stable expression of a shorter version of S in its C-terminal domain (DS) in 293 cells 26 that express MLV Gag-Pol (293GP). The incorporation of DS was 15-times more efficient into VLPs 27 as compared to the full-length version, and that was not due to steric interference between the S 28 cytoplasmic tail and the MLV capsid. Indeed, a similar result was also observed with extracellular 29 vesicles released from parental 293 and 293GP cells. The amount of DS incorporated into VLPs 30 released from producer cells was robust, with an estimated 1.25 µg/ml S2 equivalent (S is comprised 31 of S1 and S2). Thus, a scalable platform that has the potential for production of pan-coronavirus VLP 32 vaccines has been established. The resulting nanoparticles could potentially be used alone or as a 33 boost for other immunization strategies for IMPORTANCE Several candidate COVID-19 vaccines have already been tested in humans, 35 but their protective effect and long-term efficacy are uncertain. Therefore, it is necessary to continue 36 developing new vaccine strategies that could be more potent and/or that would be easier to 37 manufacture in large-scale. Virus-like particle (VLP) vaccines are considered highly immunogenic 38 and have been successfully developed for human papilloma virus as well as hepatitis and influenza 39 viruses. In this study, we report the generation of a robust Moloney murine leukemia virus platform 40 that produces VLPs containing the spike of SARS-CoV-2. This vaccine platform that is compatible 41 with lyophilization could simplify storage and distribution logistics immensely. 42 FACS analysis on 293-ACE2 cells, a cell line generated by stable transfection that is 61% positive 110 for ACE2 (Fig. 2) . Titers of 3.2 x 10 7 infectious units (IU)/ml and 1.5 x 10 6 IU/ml were obtained for 111 VSV-G-and Galv-pseudotyped viruses, although titers of S and DS-pseudotyped viruses were below 112 the detection limit of 10 4 IU/ml (Fig. 3A ). Only few GFP cells could be observed by fluorescence 113 microscopy after infection with the DS-pseudotyped virus and there were none when the S-114 pseudotyped vector was used (Fig. 3B) . Thus, these results indicated that the transient production was 115 extremely inefficient for generating VLP-S, even with DS. 116 DS-pseudotyped MLV recombinant viral particles are efficiently released from stable 117 producer cells. We have shown that stable retrovirus packaging cell lines can generate pseudotyped vectors with at least 10-fold higher titers as compared to transient transfection 119 productions (47). We then hypothesized that S or DS stably expressed in 293GP cells (293 cells that 120 express MLV Gag-Pol) could be a better system to produce VLP-S. Stable populations of 293GP 121 cells expressing S and DS were then generated by transfection. In these cells, S and DS were able to 122 localize at the cell surface at even higher levels than what we found in transient transfection (Fig.  123 4A). A GFP retroviral vector was then introduced in these cells by infection (Fig 4B) , and titers of 124 GFP viruses released by these new producers were measured after infecting 293-ACE2 cells. Only 125 few GFP positive cells could be detected by fluorescence microscopy after infection of 293-ACE2 126 cells with the S-pseudotyped vector, but a very high percentage of fluorescent cells was observed 127 after infection with the DS virus. A high number of GFP positive cells was seen with the Galv virus 128 diluted 10-times as compared to the two other vectors (Fig. 5A) . Titers of 1.6 x 10 7 IU/ml and 10 5 129 IU/ml were measured for the Galv and DS-pseudotyped viruses, respectively, and the S-pseudotyped 130 vector titer was below the detection limit of 10 4 IU/ml, as expected ( Fig 5B) . We could conclude that 131 the production of recombinant viral particles was robust from stable producers expressing DS and 132 inefficient with the full-length version of SARS CoV-2 S. 133

The deletion of the 19 amino acid cytoplasmic tail of S does not enhance its fusogenicity. 134

As producer cells express the same amount of S and DS at the cell surface, one possible explanation 135 for the high transduction efficiency of DS-pseudotyped vectors could be increased fusogenicity. The 136 fusion capacity of S and DS was then assessed in a syncytia formation assay by mixing 293GP cells 137 expressing S or DS with 293-ACE2 cells. The number and the size of syncytia evaluated one day 138 after mixing were very similar between S and DS mixtures, and there were none with the control 293 139 cells (Fig. 6) . Thus, the deletion of the 19 amino acids in the S cytoplasmic tail does not have a 140 significant effect on its fusogenicity. 141

High amounts of SARS-CoV-2 DS protein are incorporated into MLV VLPs released 142 from stable producer cells. A VLP-derived SARS CoV-2 vaccine will be a viable option if 143 sufficient amounts of S protein are incorporated at the surface of the released particles. Western blots 144

were performed with an anti-S2 antibody to evaluate the quantity of S protein into VLPs produced in 145 transient transfections and from stable producers. Two bands were detected around 90 KDa that are 146 most likely two glycosylated forms of S2. The uncleaved S protein migrated around 180 kDa, and 147 two other bands above 250 kDa were also detected in the DS samples that had more intense signals. 148

These bands could be dimeric and trimeric forms of S as it has been suggested (19). The amount of 149 S2 detected at the surface of VLPs produced in transient transfections or released from stable 150 producers was much higher with the truncated version of S than with the full-length molecule (Fig.  151 7A). MLV viral particles produced in transient transfection or from stable producers were detected 152 with an antibody against p30. A 4-and a 15-fold difference was found with the transient and the 153 stable production systems, respectively (Fig. 7B ), although there was less than a 1.5-fold difference 154 between S and DS in cellular extracts (Fig. 7C ). More DS was also released as compared to the full-155 length protein in the supernatants of stably transfected 293 cells, however the amount of DS detected 156 was 4-to-5 times lower than the one released from the 293GP-DS. The amount of S2 equivalent 157 present in the supernatant of 293GP-DS cells was high and evaluated at 1.25 µg/ml using the IgG-S2 158 standard (Fig. 7A ). Our results indicated that the incorporation of S into MLV VLPs is very efficient 159 in stable producers but only with the truncated version of S. Immunization will be the best preventive strategy to address the current COVID-19 pandemic, 175

although therapeutic alternatives cannot be neglected as an efficient vaccine is not a certainty (23, 25, 176 50 ). Yet preliminary results from preclinical and clinical studies are encouraging as several types of 177 vaccines are able to trigger the production of Nabs against SARS-CoV-2 S (25, 27-31, 33-39, 51, 52) . 178

How efficient and how long these Nabs will be present in vaccinated people remains an open 179 question that will only be answered with time (25). Also, antibody-dependent enhancement will have 180 to be carefully monitored in these trials as it is a side effect that cannot be underestimated with 181 coronaviruses (23, 25, 50). One other major challenge ahead will be the capacity to mass produce 182 COVID-19 vaccines. In this study, we have established and characterized a new MLV-derived VLP 183 platform that could be used for the production of a COVID-19 vaccine. 184

The efficient pseudotyping of MLV particles with S is a prerequisite to establish a robust VLP CoV-2 S at the cell surface was not improved after disrupting the ER retention signal by missense 191 mutations (56). In this study, we showed that S could be detected at the cell surface at a similar level 192 to that achieved by DS in transiently transfected cells as well as in stable producers ( Fig. 1 and Fig.  193 4A), a finding that has also been reported for SARS-CoV S expressed in transient transfections (40, 194 54) . These results indicate that S can bypass its natural localization and efficiently migrates to the cell 195 surface when it is overexpressed. 196

Despite similar amounts of S and DS at the cellular membrane, the truncated version was more 197 efficiently incorporated into MLV viral particles. Four-and 15-fold differences were obtained with 198

VLPs produced in transient transfection experiments and from stable producers, respectively (Fig. 199 7B) . The hypothesis that has been proposed for SARS-COV and SARS-CoV-2 is that the 19 amino-200 acid deletion in the S cytoplasmic tail facilitates the pseudotyping by decreasing the steric 201 interference with the retroviral matrix proteins (54, 55, 57). Our results invalidate this hypothesis as 202 more DS was also found in the supernatant of 293 transfected cells that did not express MLV Gag-Pol 203 (Fig. 7A) . Parental 293 and 293GP cells release EVs that can incorporate DS more efficiently than S 204 ( Fig. 7A and Fig. 8 ). VLPs and EVs are very similar in composition, and it has been postulated that 205 they use similar pathways for vesicle trafficking (58, 59). So, unlike S, DS was efficiently 206 incorporated into VLPs or EVs like for example tetraspanins or endosomal markers that are equally 207 found in both particle types (58, 59). 208

EVs released from 293GP-DS contain less than 10% of the total DS protein, and they would not 209 need to be removed from vaccine preparations as they could be as good immunogens as VLPs. It was 210 even reported that EVs containing the S protein of SARS-CoV could induce high levels of Nabs (60). 211

Titers of recombinant GFP retroviruses released from stable producers were at least a 1000-fold 212 higher with DS versus S despite a 15-fold difference in the amount of the two proteins incorporated at 213 the surface of VLPs ( Fig. 5A and Fig 7B) . As we did not find major differences in fusogenicity 214 between S and DS in a syncytia formation assay (Fig. 6) , our results suggest that recombinant viruses 215 become fully infectious when a certain threshold of S protein is incorporated at their surface. 216

Recombinant GFP or luciferase pseudotyped retroviruses are commonly used to measure the 217 activity of Nabs present in serum of infected or vaccinated people (55-57). These reagents are 218 convenient, as unlike SARS-CoV-2 they can be manipulated in a BSL-2 laboratory. The robust 219 production system with the 293GP-DS cell line could be highly valuable to evaluate the presence of 220 Nabs in large cohorts. 221

Mass production will be a major challenge with all types of SARS-CoV-2 vaccine that are 222 being developed as the entire worldwide population will have to be vaccinated. Based on the results 223 of a nanoparticle vaccine containing S, whose 5 and 25 µg doses triggered a high level of Nabs in 224 people (28), we assume that a vaccine derived from the VLP platform described in this study could 225 be efficient with similar or lower amounts of S per dose. The yield of VLPs produced from the 226 293GP-DS cells could be increased if a high producer clone is selected instead of a bulk population, 227

and if cells are cultured in bioreactors in fed-batch or perfusion modes. The average titer of gene 228 therapy vectors produced with a derivative of the 293GP cell line was increased by 5.6-fold in 229 bioreactor versus a 10-layer cell factory, and the total vector yield was increased by 13.1-fold (61). 230

Mutations of the furin cleavage site located between S1 and S2 and the D614G variant that is now 231 more prevalent in the infected population could increase the amount of S incorporated into VLPs (57, 232

A very concise review that compared the first results of different COVID-19 vaccines 234 concluded that the most immunogenic ones were made with recombinant proteins (25). These results 235 emphasize the importance of the platform developed in this study because VLPs present the antigen 236 in a protein format that seems more potent for vaccination than the protein alone. Indeed, MLV VLPs 237 displaying the human cytomegalovirus glycoprotein B antigen could trigger 10-times more Nabs in 238 mice than the protein alone using the same amount of antigen (63). Finally, VLP-S could be used as a 239 boost for other types of vaccine like measle virus-and adenovirus-based recombinant vectors. These 240 combinations were highly potent for triggering Nabs against hepatitis C proteins in mice and 241 macaques (64). 242

In conclusion, we have developed and characterized a new MLV VLP platform that can 243 efficiently incorporate the S protein from SARS-CoV-2, and that has the potential to produce a pan-244 coronavirus vaccine. The next logical step is to validate this vaccine in experimental animals and in 245 humans thereafter. 246

Plasmids. The expression plasmid pMD2ACE2iPuro r containing the human angiotensin-248 converting enzyme (ACE2) cDNA used to generate ACE2 positive cells was constructed as follows: 249 the ACE2 PmeI cDNA fragment obtained from the plasmid hACE2 (Addgene; #1786) was cloned in 250 pMD2iPuro r opened in EcoRV. 251

The SARS-CoV-2 S gene from the Wuhan-Hu-1 isolate (GenBank: MN908947.3) was codon 252 optimized (Genscript, Township, NJ) and cloned in pMD2iPuro r in EcoRI/XhoI. A shorter version with 253 a 19-codon deletion in C-terminal (DS) was also constructed in a similar way. 254

The pMD2.GalviPuro r and pMD2.G plasmids that encode the Galv and VSV-G envelopes, and 255 the retroviral vector plasmid containing the GFP gene under the control of the 5' long terminal repeat 256 sequence have been described elsewhere (65). human chimeric anti-S1 antibody (Genscript; 1:200 dilution) followed by an Alexa647-conjugated goat 298 anti-human IgG (Jackson Laboratories; 1:400) were successively incubated with cells for labelling. 299

The fixable viability stain 450 (BD Biosciences, San Jose, CA, USA) was used to exclude dead cells. 300

The presence of S was then analyzed by flow cytometry with a BD FACSAria II (BD Biosciences). 301

Cells transfected with a Galv expression plasmid were used as control. The presence of stably 302 expressed S at the cell surface of 293GP-S and 293GP-DS was similarly analyzed by flow cytometry. 303

The presence of ACE2 at the surface of 293-ACE2 cells was also checked by FACS. Detached 304 cells were labelled with a mouse anti-ACE2 antibody (R&D Systems, Minneapolis, MN1/200) 305 followed by an Alexa488 goat anti-mouse (1:1,000; Invitrogen, Carlsbad, CA). 306

The presence of S released in the supernatant of transiently transfected 293GP cells was 307 analyzed by Western blot. Subconfluent cells plated in 60 mm were transfected for 4 h with 5 µg of 308 envelope expression plasmids and 5 µg of the GFP retroviral plasmid. One day later, the media was 309 replaced with 2.5 ml of SFM that was then harvested the following day. Supernatants were 310 concentrated 10-fold with a 30 kDa Amicon centrifugal unit (Millipore Sigma, Oakville, Canada) and 311 were stored at -80 o C until use. The GFP fluorescence evaluated under a microscope at the time of 312 harvest was very similar among the different transfected plates. 313

Supernatants from confluent 293GP-S, 293GP-DS, 293-S and 293-DS cells were also harvested 314 and concentrated from 60-mm dishes. 315

Cell pellets of 1 x 10 6 cells were resuspended in 100 µl RIPA lysis buffer containing a protease 316 inhibitor cocktail (Roche). Samples were centrifuged for 5 min to remove cell debris and stored at -317 20 o C until use for Western blot analysis. 318

Samples of 20 µl were incubated 5 min at 95°C in loading buffer containing 1% SDS and 2.5% 319 b-mercaptoethanol, and run on a 10% SDS-polyacrylamide gel (4% stacking), followed by transfer 320 onto nitrocellulose membranes (GE Healthcare). Immunoblotting was performed with a rabbit 321 polyclonal antibody anti-S2 (1:400 dilution, SinoBiological, Beijing, China) and a rat monoclonal 322 antibody anti-MLV p30 produced from the hybridoma R187 (1:2,000 dilution; American Type 323 Culture Collection, Manassas, VA). Blots were then incubated with secondary antibodies 324

IRDylight680 goat anti-rat IgG (1:10,000; Invitrogen) and IRDye 800CW anti-rabbit IgG (1:10,000; 325

Li-Cor Biosciences, Lincoln, NE), and analyzed with the Odyssey Infrared Imaging System Biosciences). Serial dilutions of known amounts of C-terminally Fc-tagged S2 (BioVendor, Brno, 327 Czech Republic) were used for quantification. 

China Novel Coronavirus I, Research T. 2020. A 348 Novel Coronavirus from Patients with Pneumonia in China

A pneumonia outbreak 352 associated with a new coronavirus of probable bat origin

The 354 socio-economic implications of the coronavirus pandemic (COVID-19): A review

Human Coronavirus: Host-Pathogen Interaction

Origin and evolution of pathogenic coronaviruses

A new threat from an old enemy: Reemergence of 363 coronavirus (Review)

Possible Bat Origin of Severe Acute Respiratory Syndrome Coronavirus 2

Identifying SARS-CoV-2 related coronaviruses in 369 Malayan pangolins

Cross host transmission in the 371 emergence of MERS coronavirus

Bats are natural 374 reservoirs of SARS-like coronaviruses

376 Rooting the phylogenetic tree of middle East respiratory syndrome coronavirus by 377 characterization of a conspecific virus from an African bat

Severe acute respiratory syndrome coronavirus-like virus in Chinese horseshoe bats

Dipeptidyl peptidase 4 is a functional receptor for the emerging human coronavirus-EMC. 384

Functional assessment of cell entry and receptor usage 386 for SARS-CoV-2 and other lineage B betacoronaviruses

Probable Pangolin Origin of SARS-CoV-2 Associated with the 388 COVID-19 Outbreak

SARS-CoV-2 Cell 391 Entry Depends on ACE2 and TMPRSS2 and Is Blocked by a Clinically Proven Protease 392 Inhibitor

Angiotensin-converting enzyme 2 is a 395 functional receptor for the SARS coronavirus

Structure, Function, 397 and Antigenicity of the SARS-CoV-2 Spike Glycoprotein

Characterization of spike glycoprotein of SARS-CoV-2 on 400 virus entry and its immune cross-reactivity with SARS-CoV

Structural Basis for Potent Neutralization of Betacoronaviruses by Single-404

Monoclonal Antibodies for Prevention and 406 Treatment of COVID-19

Fruitful Neutralizing Antibody Pipeline Brings 408

Hope To Defeat SARS-Cov-2

SARS-CoV-2 Vaccines: Status Report

Development of SARS-CoV-2 411 vaccines: should we focus on mucosal immunity?

SARS-CoV-2 vaccines: 'Warp Speed' needs mind melds not 414 warped minds

Major findings and recent 416 advances in virus-like particle (VLP)-based vaccines

Safety, tolerability, and immunogenicity of a recombinant adenovirus type-5 vectored 420 COVID-19 vaccine: a dose-escalation, open-label, non-randomised, first-in-human trial

First-in-Human Trial of a SARS CoV 2 Recombinant 426 Spike Protein Nanoparticle Vaccine

An mRNA Vaccine against 433 SARS-CoV-2 -Preliminary Report

An alphavirus-derived replicon RNA vaccine induces SARS-CoV-2 neutralizing antibody 438 and T cell responses in mice and nonhuman primates

Oxford 444 CVTG. 2020. Safety and immunogenicity of the ChAdOx1 nCoV-19 vaccine against SARS-CoV-445 2: a preliminary report of a phase 1/2, single-blind, randomised controlled trial

Primary exposure to SARS-CoV-2 protects against reinfection in rhesus macaques

Immunogenicity of a DNA 457 vaccine candidate for COVID-19

CoV-2 spike S1-Fc fusion protein induced high levels of neutralizing 461 responses in nonhuman primates

Development of an Inactivated 465 Vaccine Candidate, BBIBP-CorV, with Potent Protection against SARS-CoV-2

Development of an inactivated vaccine 470 candidate for SARS-CoV-2

Phase 1/2 study of COVID-19 RNA vaccine BNT162b1 in adults

Immunogenicity and safety of 479 a recombinant adenovirus type-5-vectored COVID-19 vaccine in healthy adults aged 18 years 480 or older: a randomised, double-blind, placebo-controlled, phase 2 trial

Intradermal-486 delivered DNA vaccine provides anamnestic protection in a rhesus macaque SARS-CoV-2 487 challenge model

Retroviral vectors pseudotyped with severe acute respiratory syndrome coronavirus S 490 protein

The contribution of the cytoplasmic 492 retrieval signal of severe acute respiratory syndrome coronavirus to intracellular 493 accumulation of S proteins and incorporation of S protein into virus-like particles

Intracellular targeting signals contribute to 496 localization of coronavirus spike proteins near the virus assembly site

Cytoplasmic tail of coronavirus spike protein has 498 intracellular targeting signals

Genetic analysis of the SARS-coronavirus spike glycoprotein functional domains 501 involved in cell-surface expression and cell-to-cell fusion

Aromatic amino acids in the juxtamembrane domain of severe acute respiratory syndrome 504 coronavirus spike glycoprotein are important for receptor-dependent virus entry and cell-505 cell fusion

Vesicular stomatitis virus pseudotyped with severe acute respiratory 508 syndrome coronavirus spike protein

510 Efficient human hematopoietic cell transduction using RD114-and GALV-pseudotyped 511 retroviral vectors produced in suspension and serum-free media

Discrimination between 513 exosomes and HIV-1: purification of both vesicles from cell-free supernatants

Highly purified human immunodeficiency virus type 1 reveals a 516 virtual absence of Vif in virions

The Improbability of the Rapid Development of a Vaccine for SARS-CoV-2

DNA vaccine protection against SARS-CoV-2 in rhesus macaques

Single-shot Ad26 vaccine protects against SARS-CoV-2 in rhesus 534 macaques

Human 536 immunodeficiency viral vector pseudotyped with the spike envelope of severe acute 537 respiratory syndrome coronavirus transduces human airway epithelial cells and dendritic 538 cells

Retroviruses pseudotyped with the severe acute 541 respiratory syndrome coronavirus spike protein efficiently infect cells expressing 542 angiotensin-converting enzyme 2

Measuring SARS-CoV-2 neutralizing antibody activity using pseudotyped and chimeric 547 viruses

Protocol and Reagents 550 for Pseudotyping Lentiviral Particles with SARS-CoV-2 Spike Protein for Neutralization 551 Assays

Optimized pseudotyping conditions for the SARS COV-2 Spike glycoprotein

Extracellular vesicles and viruses: Are 556 they close relatives?

The Trojan exosome hypothesis

Exosomal vaccines containing the S protein of 560 the SARS coronavirus induce high levels of neutralizing antibodies

Large-scale clinical-grade retroviral vector production in a fixed-bed bioreactor

The 565 D614G mutation in the SARS-CoV-2 spike protein reduces S1 shedding and increases 566 infectivity

Enveloped virus-like particle 569 expression of human cytomegalovirus glycoprotein B antigen induces antibodies with potent 570 and broad neutralizing activity

A prime-boost strategy using virus-like particles pseudotyped for HCV 574 proteins triggers broadly neutralizing antibodies in macaques

Generation of a high-titer packaging cell line for 576 the production of retroviral vectors in suspension and serum-free media

Fiji: an open-source platform for biological-image analysis