Assembly and Entry of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV2): Evaluation Using Virus-Like Particles

Research on infectious severe acute respiratory syndrome coronavirus 2 (SARS-CoV2) is currently restricted to BSL-3 laboratories. SARS-CoV2 virus-like particles (VLPs) offer a BSL-1, replication-incompetent system that can be used to evaluate virus assembly and virus-cell entry processes in tractable cell culture conditions. Here, we describe a SARS-CoV2 VLP system that utilizes nanoluciferase (Nluc) fragment complementation to track assembly and entry. We utilized the system in two ways. Firstly, we investigated the requirements for VLP assembly. VLPs were produced by concomitant synthesis of three viral membrane proteins, spike (S), envelope (E), and matrix (M), along with the cytoplasmic nucleocapsid (N). We discovered that VLP production and secretion were highly dependent on N proteins. N proteins from related betacoronaviruses variably substituted for the homologous SARS-CoV2 N, and chimeric betacoronavirus N proteins effectively supported VLP production if they contained SARS-CoV2 N carboxy-terminal domains (CTD). This established the CTDs as critical features of virus particle assembly. Secondly, we utilized the system by investigating virus-cell entry. VLPs were produced with Nluc peptide fragments appended to E, M, or N proteins, with each subsequently inoculated into target cells expressing complementary Nluc fragments. Complementation into functional Nluc was used to assess virus-cell entry. We discovered that each of the VLPs were effective at monitoring virus-cell entry, to various extents, in ways that depended on host cell susceptibility factors. Overall, we have developed and utilized a VLP system that has proven useful in identifying SARS-CoV2 assembly and entry features.


Introduction
Coronaviruses (CoVs) include a remarkable diversity of animal and human-infecting species. There are seven human-infecting species, each causing disease of varying severity, ranging from mild or asymptomatic infections resembling the common cold to severe wheezing, bronchitis, or pneumonia [1,2]. These human CoVs are descendants of animal viruses that have invaded zoonotically [3]. There are a remarkable diversity of animalinfecting CoV species, and while most appear limited to their natural animal reservoirs, the emergence of severe acute respiratory syndrome coronavirus (SARS-CoV) in 2002 [4] and Middle East respiratory syndrome coronavirus (MERS-CoV) in 2012 [5], and SARS-CoV2 in 2019 [6], clearly demonstrated zoonotic CoV capabilities [3,7]. During December 2019, SARS-CoV2 was first reported in China, and within months, spread across the globe, causing pandemic coronavirus disease 2019 (COVID- 19), with more than 2 million fatalities as of February 2021 [8].
Enveloped CoV particles enclose a single positive-sense RNA genome. The limiting envelopes contain three major proteins, matrix (M), envelope (E), and spike (S). M

VLP Production and Purification
Adherent HEK293T cells at 80% confluence were co-transfected with equimolar amounts of plasmids encoding the CoV-S, E, M, and HiBiT-N proteins using LipoD transfection reagent (SignaGen, Frederick, MD, USA). 1:3 DNA:LipoD mixtures were mixed in serum free-DMEM for 10 min at room temperature followed by dropwise addition onto cells. Four h later, media were replenished with DMEM-1% FBS. Media containing VLPs were collected at 24 h post-transfection, clarified by sequential centrifugation, first at 300× g, 4 • C, 10 min, then at 3000× g, 4 • C, 10 min. VLPs in media were then concentrated 20-fold by ultrafiltration using 100 K Amicon ultra filters (MilliporeSigma, St. Louis, MO; USA), and 0.5-mL volumes were then applied to size-exclusion chromatography (SEC) columns. Fractions were collected according to manufacturer's instructions (Izon Science Ltd., Medford, MA, USA), and VLPs were identified in eluted fractions using Nano-Glo luciferase assay reagents (Promega, Inc.). Briefly, aliqouts of each fraction were mixed with LgBiT protein in the presence of Nano-Glo luciferase assay substrate (Promega), and Nluc enzyme levels were measured in terms of relative light units (RLU) using a Veritas microplate luminometer. Fractions containing HiBiT-VLPs were aliquoted and stored at −80 • C.

Binding and Entry Assay of SARS-CoV2 VLPs
For the binding assay, HeLa and HeLa-ACE2 cells were incubated with the normalized Nluc input multiplicities of EMN (E+M+N) and SEMN (S+E+M+N) containing VLPs for 1.5 h at 4 • C. The cells were washed and lysed to quantify the associated Nluc activity using the Promega's Nano-Glo luciferase assay system. The EMN VLPs were used to calculate the background.
For the VLP-cell entry assay, HeLa cells were transfected with pcDNA-ACE2-LgBiT. Twenty-four h later, cells were inoculated with EMN and SEMN, containing VLPs for 1 h at 4 • C. Thereafter, cells were incubated with or without trypsin (20 ng/µL) in the presence of Nluc live cell substrate vivazine and transferred to 37 • C to initiate VLP-cell entry. Nluc activity was measured at various time points to assess kinetics of VLP entry. After 2 h, cells were dissolved in Nano Glo lysis buffer (Promega Corporation, Madison, WI, USA) to allow for maximal HiBiT:LgBiT complementation. The proportions of VLPs that entered cells in the 2 h time period were estimated relative to the maximal complementation after detergent cell lysis.

Statistical Analysis
Results are expressed as means ± SE of the mean of at least three independent experiments. Unpaired t-test and ANOVA with Dunnette post hoc test were applied to statistical differences amongst the groups. In all the experiments, p < 0.05 was considered statistically significant.

Determinants of SARS-CoV2 VLP Assembly
CoV VLPs can be produced upon co-expression of the E, M, and N structural proteins in a mammalian expression system, and the S glycoprotein can efficiently incorporate onto these secreted VLPs [30]. Recent publications have also shown that SARS-CoV2 VLPs can be produced similarly upon co-expression of structural proteins. These studies have either used tagged M protein [31] or tagged E protein [32], both of which have been previously shown to be the minimal requirement for production of VLPs [33]. To determine optimal conditions for VLP production, we co-expressed the E, M, and N structural proteins in HEK-293T cells and collected the VLP-containing supernatant at early 24 h post-transfection times ( Figure 1). We tagged the N protein with an 11 amino-acid peptide (HiBiT) for sensitive detection of secreted VLPs ( Figure 2). We specifically chose to only tag the N protein because of our previous observation that revealed that the N protein of coronaviruses can tolerate amino-terminal extensions without affecting VLP production [34]. Secreted HiBiT-N VLPs were identified using Binary NanoLuc Technology (NanoBiT) ( Figure 2B, left panel). Purified VLPs eluted from SEC columns at or near void volumes (fractions 7-9; Figure 2B, right panel). These SEC-purified VLPs contained E, M, and HiBiT-N, as evaluated by Western blotting ( Figure 2C). S proteins were incorporated into VLPs when co-synthesized with E, M, and N ( Figure 2D, right panel). Variant S proteins were equally incorporated when present ( Figure 2D, right panel, lanes 5-6), making it clear that VLPs are suitable to evaluate S variants of concern. S proteins synthesized alone did not secrete ( Figure 2D, right panel, lane 2). VLPs are suitable to evaluate S variants of concern. S proteins synthesized alone did not secrete ( Figure 2D, right panel, lane 2).

Nucleocapsid Drives Efficient SARS-CoV2 VLP Production
To delineate the role of individual structural proteins in VLP assembly, we performed a series of transfections with E, M, and N, alone or together, purified secreted VLPs using SEC, and analyzed VLP proteins by Western blotting (Figure 3). Unlike previous reports showing that E and M were sufficient to produce VLPs, we identified dependence on E, M, and N proteins ( Figure 3, right panel, lane 6). This finding fits more closely with a recent report demonstrating some dependence on N proteins for M secretion [32]. Of note, previous studies evaluated VLP secretion at relatively late times after structural gene expression (2 to 3 days post transfection), while we focused on earlier time frames, when structural proteins were less abundant and less likely to undergo unconventional secretion. Apparently, it is these conditions in which N proteins are central to VLP formation and secretion.  VLPs using SEC, and analyzed VLP proteins by Western blotting (Figure 3). Unlike previous reports showing that E and M were sufficient to produce VLPs, we identified dependence on E, M, and N proteins ( Figure 3, right panel, lane 6). This finding fits more closely with a recent report demonstrating some dependence on N proteins for M secretion [32]. Of note, previous studies evaluated VLP secretion at relatively late times after structural gene expression (2 to 3 days post transfection), while we focused on earlier time frames, when structural proteins were less abundant and less likely to undergo unconventional secretion. Apparently, it is these conditions in which N proteins are central to VLP formation and secretion.

Nucleocapsid Carboxyl-Terminal Domains (CTDs) Drive Efficient VLP Production
The central role for N proteins in VLP production ( Figure 3) prompted us to ask whether individual N protein fragments might facilitate assembly. Towards this end, we constructed HiBiT-N NTD and CTD fragments to assess their support of VLP production. Unfortunately, these fragments accumulated at levels nearly 10 times lower than the fulllength N protein (data not shown) and hence could not be utilized to obtain conclusive data. Therefore, we adopted an alternative approach in which we asked whether N proteins from related betacoronaviruses could replace SARS-CoV2 N. Betacoronavirus N proteins exhibit considerable variability; SARS-CoV2 N is 49% and 37% similar to MERS-CoV and MHV-CoV N, respectively. We expressed the MERS or MHV N proteins in conjunction with SARS-2 membrane proteins S, E, and M, and found that MERS N proteins were largely incapable of supporting assembly, providing only ~10% as many VLPs as the ho-

Nucleocapsid Carboxyl-Terminal Domains (CTDs) Drive Efficient VLP Production
The central role for N proteins in VLP production ( Figure 3) prompted us to ask whether individual N protein fragments might facilitate assembly. Towards this end, we constructed HiBiT-N NTD and CTD fragments to assess their support of VLP production. Unfortunately, these fragments accumulated at levels nearly 10 times lower than the fulllength N protein (data not shown) and hence could not be utilized to obtain conclusive data. Therefore, we adopted an alternative approach in which we asked whether N proteins from related betacoronaviruses could replace SARS-CoV2 N. Betacoronavirus N proteins exhibit considerable variability; SARS-CoV2 N is 49% and 37% similar to MERS-CoV and MHV-CoV N, respectively. We expressed the MERS or MHV N proteins in conjunction with SARS-2 membrane proteins S, E, and M, and found that MERS N proteins were largely incapable of supporting assembly, providing only~10% as many VLPs as the homologous SARS-2 N, as measured by tracking HiBiT-N secretion (Figure 4, panel B, rightmost lane). MHV N proteins were more capable in supporting SARS-2 VLPs, providing up to 40% VLP production relative to SARS-2 N (data not shown).
These differences between SARS-2 and MERS N proteins provided opportunities to delineate the roles of individual domains of N protein. We made a series of MERS-SARS-2 chimeric N constructs ( Figure 4A). The chimeras divided the N protein into three major domains: (1) NTD; including the N1a + N1b domains; (2) linker domain, consisting of N2a; and (3) CTD; consisting of N2b + N3 domains. The chimeric HiBiT-tagged N proteins were co-expressed with SARS-CoV2 membrane proteins, and secreted HiBiT levels were measured as reflections of VLP production and secretion ( Figure 4B).   The results demonstrated a primary role for the nucleocapsid CTDs in VLP production and secretion ( Figure 4B). This was most evident from results of two chimeras, designated N MS 3 and N SMS 5, both of which are~80% as effective as complete SARS-2 N and both of which retain the SARS-2 CTDs ( Figure 4B). However, there was also a minor role for the nucleocapsid NTDs, as evidenced by N chimeras with SARS-2 NTD or NTD + linker having 40% the activity of complete SARS-2 N ( Figure 4B). Linker domains did not contribute specifically to VLP production, as evidenced by the properties of N MSM 6, which was equivalent to MERS N in support of VLPs. These results were independently confirmed by detecting VLP-associated S and M proteins in purified VLP preparations ( Figure 4C, top VLP panels). Also of note, the diminished support of several VLP production processes were not due to reduced synthesis of chimeric N proteins, as measured by detection of viral proteins in cell lysates ( Figure 4C, bottom panels). Overall, the findings demonstrate complex domain-specific contributions to VLP assembly that are largely dominated by the CTDs. These results are concordant with some previous reports demonstrating the role of coronavirus N-CTD in protein oligomerization and assembly [35,36].

A SARS-CoV2 Nucleocapsid Fragment Interferes with VLP Production
Further evaluation of specific nucleocapsid domains in VLP assembly were approached by asking whether SARS-2 nucleocapsid fragments might interfere with the natural VLP production process. To this end, a deleted form of N (∆N) was constructed in which a portion of the CTD was eliminated ( Figure 5). This ∆N construct, which lacks HiBiT, was co-transfected with the standard SARS-2 VLP assembly components, S, E, M, and HiBiT-N. The presence of ∆N significantly hindered in the assembly of VLPs and reduced VLP production by~30% ( Figure 5B), a marginal amount, consistent with the CTD of complete SARS-2 N protein having a dominant role in VLP assembly (Figure 4). Western blot detection of VLP proteins ( Figure 5C) confirmed the finding that the ∆N fragment interfered weakly. These findings form the basis for continued tests of assembly interference by short nucleocapsid fragments or nucleocapsid peptidomimetics.

HiBiT-Tagged SARS-CoV2 VLPs Assess Virus Binding and Entry Events
CoV VLPs are tools to evaluate particle assembly (Figures 1-5) and can also be used to assess particle entry into target cells. Indeed, HiBiT tagged flavivirus VLPs have been previously shown to be a useful quantitative method for the analysis of viral entry and release [37]. Here, we considered the potential of HiBiT-N VLPs in target-cell events, first in measuring virus-cell binding ( Figure 6A). SEC-purified VLPs ( Figure 6C) were applied to target cells in a 4 • C binding process. Definitive ACE2 receptor-dependent binding was observed ( Figure 6D). Second, HiBiT-N VLPs were applied to target cells expressing a recombinant ACE2 in which Nluc "LgBiT" fragments were appended to ACE2 cytoplasmic termini. Here, VLPs were incubated at 37 • C, in the presence of S protein-activating trypsin proteases [38]. Nluc, assembled only after successful fusion of VLP and cell membranes, was measured at specific time points after 37 • C incubation. This was found to be a highly sensitive, accurate, and quantitative approach to measure SARS-CoV2 VLP entry, with signals up to 70-fold over background ( Figure 6E). We quantified the efficiency of the HiBiT-N tagged VLP entry process by determining the maximal HiBiT-LgBiT complementation achieved after detergent-mediated lysis of virus-cell cultures, and by using this maximal value as a denominator. From these calculations, we estimated~10% of VLPs successfully entered target cells in a 2 h incubation period.

HiBiT-Tagged SARS-CoV2 VLPs Assess Virus Binding and Entry Events
CoV VLPs are tools to evaluate particle assembly (Figures 1-5) and can also to assess particle entry into target cells. Indeed, HiBiT tagged flavivirus VLPs h previously shown to be a useful quantitative method for the analysis of viral e release [37]. Here, we considered the potential of HiBiT-N VLPs in target-cell eve in measuring virus-cell binding ( Figure 6A). SEC-purified VLPs ( Figure 6C) were to target cells in a 4 °C binding process. Definitive ACE2 receptor-dependent bin We considered whether HiBiT tags could be placed on CoV structural proteins other than nucleocapsid, so that the membrane proteins might also be monitored using Nanobit approaches. To this end, we produced two more types of SARS-CoV2 VLPs having the HiBiT tags on C terminus of M and the E proteins. Production and SEC purification of the Eand M-HiBiT VLPs revealed that they could be obtained in yields similar to HiBiT-N VLPs (compare areas of HiBiT-containing fractions 7-9 in Figure 6C,F,H). In VLP-cell entry assays, the E-HiBiT and M-HiBiT VLPs were clearly effective at monitoring cell entry, however, they were~10 and~30-fold less robust than HiBiT-N VLPs in generating Nluc entry signals (compare Figure 6E,G,I). The relatively lower entry signals may be attributable to the limited mobilities of M and E proteins after VLP-cell fusion, and the resultant failure of many HiBiT moieties to reach LgBiT tags extending into cytosols from ACE2 transmembrane proteins. Nonetheless, these findings clearly demonstrate the potential for Nluc fragment-tagging and CoV VLP reagents in the study of CoV-cell entry processes.

Discussion and Conclusions
Here, we demonstrated that steps in SARS-CoV2 assembly and entry can be dissected using VLPs. We used the VLP system in two ways; first, to identify an underappreciated role for N proteins in SARS-CoV2 assembly, and second, to establish a platform for evaluating SARS-CoV2 cell entry with reductionist, quantitative assays that isolate the initial entry process from all other infection stages.
CoV N proteins carry out several functions in the context of natural infection. They suppress host cellular response to viral infection [39][40][41], and they arrest host cell cycling, thereby promoting viral replication [42,43]. N proteins interact with genomic RNA, and with virion E and M proteins, to integrate the coronavirus genome into virus particles [44]. Early reports focusing on CoV particle assembly did not reveal a role for N proteins [45], yet here in this study, N proteins were critical for SARS-CoV2 VLP production ( Figure 3). This provided opportunities to dissect N domains for their roles in VLP assembly and secretion.
CoV N proteins are comprised of NTDs, central Ser/Arg (SR)-rich containing linker regions, and CTDs [24]. Our results demonstrated that SARS-CoV2 particle assembly and secretion was maximized by N proteins containing the SARS-CoV2 CTD (Figure 4). Conversely, N proteins lacking the CTD only modestly interfered with particle assembly (Figure 5), further implying that that the CTDs have a prominent role in assembly. The findings are consistent with a previous report suggesting a role for the CTD in N protein dimerization and oligomerization, both of which crucial in forming virus particles [36,46]. Other domains in addition to the CTD appeared to contribute more modestly to particle production. The linker (N2a) domain, when present with the CTD, moderately enhanced VLP production. This may arise because the SR-rich linker domain regulates N phosphorylation, oligomerization, and resulting higher order ribonucleoprotein assembly [47]. The strong CTD requirement could also come from the N3 domains that are present in CTD chimeras. N3 domains interact with M [48,49] at discontinuous regions of M endodomains [23,50,51], and may be central in promoting particle assembly.
A remaining question is whether N protein: RNA binding has a role in particle morphogenesis. Both NTDs and CTDs were associated with RNA binding in SARS-CoV [46,52], and N-RNA complexes may form prior to assembly into enveloped particles. At present, the VLP systems do not reveal the N: RNA interactions that take place in natural infections, and we have not yet investigated the relationships between N domain: RNA interactions and N domain functions in virus particle assembly.
The VLPs can, however, provide assay platforms for particle assembly and its inhibition. SARS-CoV2 assembly is a target for pharmacologic inhibition, and in silico approaches have identified several phytocompounds [53,54] and anti-viral and anti-microbial drugs [55,56], which may prove effective against SARS-CoV2. These computational methods help in designing of new drug candidates more rapidly, however, their actual validation in natural viral infection takes time and resources due to limitations of required biosafety facilities to handle highly pathogenic viruses. The VLP system provides a rapid and safe platform to validate computationally predicted antiviral compounds, and dissect antiviral operating mechanisms, as a step toward clinical application.
The VLP systems described here also provide platforms for evaluating virus-cell entry. Here, we found that the 11-amino acid HiBiT tags used to monitor VLP production and entry can be appended to different structural proteins of SARS-CoV2, as demonstrated by the VLP-cell entry assays ( Figure 6). This advances the NanoBiT technology to SARS-CoV2, building from a recent report that has shown the utility of HiBiT tags on spike or membrane protein of infectious bronchitis coronavirus [57]. Using sensitive NanoBiT technology, the system isolates virus-cell entry and can thereby recognize the receptors and proteases operating as coronavirus-cell susceptibility factors ( Figure 6). Notably, various sarbecovirus S proteins can be incorporated into VLPs, offering ways to determine whether newly-evolving SARS-CoV-2 variants of concern depend variably on host factors or exhibit distinctive cell entry pathways into target cells.
Overall, we report a HiBiT-VLP system that can be utilized in BSL-1 laboratory settings to quantify features of CoV assembly and entry. The VLPs were used here to illuminate roles for N protein domains in SARS-CoV-2 assembly and to demonstrate utility in viruscell entry assays. We anticipate the VLP systems will be adaptable to screens for inhibitors targeting SARS-CoV2 entry and assembly, thereby accelerating antiviral drug discovery for COVID19 and other CoV diseases.