Design of SARS-CoV-2-Wuhan-Hu1 clone and rescuing of icDNA-derived SARS-CoV-2

Because coronavirus genomes are substantially longer than those of most RNA viruses, many coronavirus RG systems involve techniques not normally used in RNA virus rescue. Specifically, the in vitro transcription of multiple ligated cDNAs or the propagation of YACs are not necessary for rescuing most RNA viruses. Thus, we sought to construct a simple and accessible single-plasmid rescue system to accelerate COVID-19 research (particularly in laboratories unaccustomed to working with coronaviruses).

We designed our SARS-CoV-2 infectious cDNA clone based on the Wuhan-Hu-1 isolate and the viral cDNA was cloned into pCC1-4K, a derivative of the pCC1BAC (EPICENTRE). Importantly, a substantial fraction of the regulatory sequences was removed to make the pCC1-4K plasmid (which lacks an intact F factor). In the correct strains, the pCC1-4K system is maintained at a single copy per cell. Because coronavirus-derived cDNA sequences can be unstable when propagated in bacteria, the low copy number facilitates the simple genetic manipulation of SARS-CoV-2 icDNA. Importantly, following successful genetic manipulation, the presence of an arabinose-inducible origin in pCC1 means that supplementing bacterial cultures with arabinose can greatly enhance DNA yield, and therefore large quantities of transfection-quality DNA can be prepared using standard reagents.

We assembled the SARS-CoV-2 icDNA genome from 5 synthetic fragments, each flanked by unique restriction sites SanDI (1524), PacI (8586), MluI (13956), Bsu36I (18176), and BamHI (25313) (Fig 1A). A human cytomegalovirus (CMV) promoter was inserted at the position corresponding to the 5′ end of the genome, and a dual hepatitis delta virus (HDV) ribozyme and Simian Virus 40 terminator (SV40) were added downstream of the poly-A tail at the 3′ end of the virus genome. These elements ensure efficient transcription and homogenous 3′ end processing during the rescue of infectious virus. To enable direct visualisation and quantification of replicating virus in cells, we also generated variants containing sequences encoding for fluorescent (mCherry and ZsGreen) and bioluminescent (Nanoluciferase, or NLuc) markers inserted into the plasmid backbone. To avoid deletion of viral sequences, the markers were cloned in-frame after the C-terminus of the ORF7a protein using a foot-and-mouth disease virus (FMDV) 2A “ribosomal skipping” linker (2A) to liberate the nascent reporter protein from the ORF7a protein (Fig 1A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. An openly available plasmid launched SARS-CoV-2 RG system.

(A) Schematic of the design and construction of a pCC1-4K-SARS-CoV-2-Wuhan-Hu-1 icDNA clone. Synthetic DNA fragments 1, 2, and 3 based on the SARS-CoV-2-Wuhan-Hu1 sequence were cloned into the pCC1 plasmids, and fragments 4 and 5 were cloned into high copy plasmid pUC57Kan by the gene synthesis company (Genscript). Fragments were designed to contain specific restriction cloning sites SanDl, PacI, Mlul, Bsu36I, and BamHI for cloning purposes. Sequences encoding for mCherry, ZsGreen, and NLuc markers were cloned in-frame to the C-terminus of the ORF7a protein via an FMDV 2A linker. (B) A summary of the passage (P) history of the extensively propagated wt 1 and wt 2 plasmid preps (expanded 6 times on solid agar and 5 times in liquid culture prior to being “grown up” in liquid culture for DNA extraction). (C) Summary plots of the number of reads mapping to the wt rescue plasmid and to the rescued viral genome. (D) Detection of SARS-CoV-2 N antigen in Vero E6 cells infected with RG-rescued SARS-CoV-2-Wuhan-Hu-1. Cells were not infected (Mock) or infected at MOI 1.0 for 48 h and then fixed and stained with mouse monoclonal anti-N antibody and counterstained with DAPI. Cells were imaged using a confocal microscope. Scale bar = 50 μm. (E) Titre of the mCherry virus (plaque forming units on Vero E6 cells) following propagation in Vero E6 cells (P3). (F) Detection of infected Vero E6 cells using the RG-rescued SARS-CoV-2-mCherry. Cells were infected for 48 h at MOI 1.0, then fixed, stained with anti-N antibody, and imaged as in (D). Scale bar = 50 μm. (G) Quantification of mCherry and N expression in Vero E6 cells infected with SARS-CoV-2-mCherry at an MOI of 0.1 for 48 h. Fixed and permeabilised cells were stained for N protein and with Hoechst 33342 (2 μg/ml). Cells were imaged using the Celigo Imaging Cytometer to identify the proportion of infected cells positive for mCherry and/or N protein. (H) Fixed plaque assays were scanned using a Celigo Imaging Cytometer (Nexcelom Bioscience) using the red channel to visualise mCherry. The cells were then subsequently stained with Coomassie staining solution and imaged again. (I) Detection of viral replication in infected Vero E6 cells using the RG-rescued SARS-CoV-2-NLuc. Cells were not infected (Mock) or infected at MOI 1.0 for 24 h, then lysed. NLuc activity was measured in the lysate using a luminometer. (J) Fluorescent plaques of the SARS-CoV-2-ZsGreen virus were visualised using the green channel as in panel H. (K) A schematic of the passage (P) history of the SARS-CoV-2 mCherry virus used to assess reporter stability. Each filled circle represents 1 day of propagation. (L) Plaque assays of passages 3 to 5 were scanned (Celigo) as in (H). The percentage of plaques visible following Coomassie staining that were mCherry-positive in the linear range of the dilution series (total plaque number in the range of 22 to 55 for each replicate) are plotted for each lineage at each passage. (M) Typical images of fluorescent plaques used for the quantification in (L) are shown. (N) A summary of the variation generated during the passage of the mCherry virus summarised in panel K. The variation is detailed in S1 Table. (O) The percentage of the viral swarm displaying all variants that exceed 25% of the swarm at any time point is shown. The low-level deletion of the furin cleavage site is also highlighted for interest. The data underlying Fig 1E, 1G, 1I and 1L may be found in S1 Data. FMDV, foot-and-mouth disease virus; icDNA, infectious cDNA; MOI, multiplicity of infection; NLuc, Nanoluciferase; RG, reverse genetics; SARS-CoV-2, Severe Acute Respiratory Syndrome Coronavirus 2; wt, wild-type.

https://doi.org/10.1371/journal.pbio.3001091.g001

Because plasmid clones harbouring the full-length SARS-CoV-2 genome could be unstable, we transformed the wild-type (wt) rescue plasmid, picked 2 colonies, and passaged each lineage through a series of 5 liquid cultures (summarised in Fig 1B). Bacterial cells were plated in between each passage, and a single colony was used to initiate each subsequent liquid passage. To validate plasmid integrity, a single colony derived from each lineage of the fifth liquid passage was chosen at random and “grown up” (1 litre culture induced with arabinose), and the culture used to extract plasmid DNA. The plasmids were then sequenced with an average coverage of 649 reads (wt 1) and 829 reads (wt 2) (S1 Fig). Using a conservative estimate of 1 duplication per hour of culture, the 2 sequenced colonies underwent over 100 divisions and both colonies (out of over 1 × 10³⁰ progeny) yielded DNA with the correct consensus sequence. There were no subconsensus substitutions detected at a frequency greater than 5% and in this approximately 35 kb construct, only 2 substitutions were detected in each lineage at a frequency >1%. Thus, this plasmid system is stable for functional use. To rescue icDNA-derived infectious viruses, we transfected BHK-21 cells with icDNA clones of SARS-CoV-2, alongside derivatives harbouring mCherry, ZsGreen, and NLuc markers. The following day, culture supernatant was used to inoculate subconfluent monolayers of Vero E6 cells. Following 4 days of culture, the virus-containing supernatant was harvested and the viral RNA was extracted and sequenced (available at GenBank BioProject PRJNA658321). This confirmed that “full-length” viruses had been rescued and that no nucleotide substitutions had accumulated during the rescue and amplification steps. As expected, the frequency of subconsensus substitutions was higher in the rescued viruses with 4 to 15 substitutions detected at a frequency >5%. Crucially, the sequencing reads of the rescued viruses overwhelmingly mapped to the viral genome (and not the flanking plasmid) indicating that we sequenced rescued viral RNA and not contaminating plasmid DNA (Fig 1C and S1 Fig).

The resulting RG-rescued SARS-CoV-2 viruses replicated efficiently and could readily be detected by IF microscopy using a commercial anti-SARS-CoV-2 nucleocapsid (N) antibody (Fig 1D). Moreover, the rescued viruses induced cytopathic effect (CPE) and can be used in conventional plaque assays (Fig 1E). When reporter expression and antigen expression were compared, approximately 80% of infected Vero E6 cells expressed the reporter and N (Fig 1F and 1G). The apparent 10% to 20% of infected cells that expressed N but not mCherry (Fig 1G) likely reflected asynchrony between N and mCherry expression as >95% of plaques from the same passage were mCherry positive (Fig 1H). The NLuc and ZsGreen reporter systems also facilitated the detection and quantification of recombinant SARS-CoV-2 replication in infected cells. SARS-CoV-2-NLuc provided a robust readout of over 4 orders of magnitude of bioluminescence intensity (Fig 1I), and SARS-CoV-2-ZsGreen plaques were all ZsGreen positive at early passages (Fig 1J).

Reporter viruses are usually less replicatively fit than their unmodified counterparts, which can lead to reporter-negative variants emerging during extended propagation. To assess the stability of the ORF7a-2A-Reporter cassette, we passaged the SARS-CoV-2-mCherry virus 5 times in Vero E6 cells (Fig 1K). After 5 passages (P5), >95% of plaques were mCherry positive in 2 out of 3 replicates (Fig 1L and 1M). In one of the replicates, however, an mCherry-negative population emerged by P4 and appeared to outcompete the parental mCherry virus (Fig 1L and 1M). To better characterise this process, we sequenced P2, P3, and P5 viral swarms (Fig 1N and 1O). By P3, a deletion mutant was detected at low frequency (4%) that had lost 1,636 nucleotides (Δ1,636) encompassing the last 2 codons of ORF6 and all of ORF7a, all of mCherry, all of ORF7b, and the majority of ORF8 (the last 2 codons remain). By P5, this variant accounted for the majority of the viral swarm (approximately 60% to 90%). Thus, although the majority of the replicates of the reporter viruses appear stable over many passages, deletion variants can rapidly overrun the culture once they arise.

Generation of coronavirus antibodies

There is an urgent need to develop antibodies specifically targeting SARS-CoV-2 proteins to accelerate COVID-19 research. We therefore generated a near-comprehensive panel of sheep polyclonal antibodies against the structural, nonstructural, and accessory proteins of SARS-CoV-2. In addition, we raised antibodies to the structural proteins of SARS-CoV and MERS-CoV (S, E, M, and N proteins), as well as the N proteins of HCoV 229E and HCoV OC43 (Fig 2A). To maximise the usefulness of these tools to the global scientific community, we sought to experimentally validate these antibodies through established assays (although at the time of submission, SARS-CoV-2 nsp4 and nsp6, and SARS-CoV and MERS-CoV S and M antibodies were not yet available for validation).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Production and validation of a near-comprehensive panel of openly available SARS-CoV-2 and coronavirus antibodies.

(A) Schematic representation of the SARS-CoV-2 and other coronavirus-processed proteins for which antibody synthesis and validation is described in this study. All antibodies (and corresponding proteins and cDNAs) listed are available upon request at https://mrcppu-covid.bio/. Accession numbers for the sequences utilised can be found in S2 Table and at https://mrcppu-covid.bio/. (B) A schematic illustrating the method of production for the antibodies shown in (A), utilising N (nucleocapsid) as an example. Each cDNA encoding a coronavirus protein was cloned into both pGex (carrying a GST tag) and pMex (MBP tag) plasmids, in order to yield corresponding GST- and MBP-tagged viral proteins. Sheep were subsequently immunised by utilising the GST-tagged proteins as antigens. At day 7 postimmunisation, serum was harvested and antibodies were affinity purified using the corresponding MBP-tagged protein. Additional inoculations (up to 5 in total) occurred 28 days apart. Exceptions to this method occurred for SARS-CoV-2 ORF7a, S, and S-RBD antibodies, where the MBP-tagged proteins were used as antigens, and after harvest, serum was again purified against the MBP-tagged protein, followed by depletion of the antibody against MBP. The SARS-CoV-2 S antibody was purified using protein G Sepharose chromatography. (C) IF validation for three of the antibodies (against nonstructural nsp2, structural N, and accessory ORF3a proteins) shown in (A) was conducted in Vero E6 cells that were uninfected (mock) or infected with the SARS-CoV-2 England-02 virus at an MOI of 0.1 for 48 h prior to fixation, permeabilization, and staining. IF validation for the remaining antibodies in (A) is shown in S2 Fig. (D) WB validation for the 3 antibodies shown in (C) was conducted using Vero E6 cells that were uninfected (mock) or infected with England-02 virus at an MOI of 0.1 or 1 (as indicated) for 72 h followed by WB analysis of whole cell lysates. WB validation for the remaining antibodies in (A) is shown in S3 and S4 Figs. (E) IP validation for the 3 antibodies shown in (C) was conducted using Vero E6 cells that were uninfected (mock) or infected with England-02 virus at an MOI of 0.1 for 3 days. IP validation for the remaining antibodies in (A) is shown in S4 and S5 Figs. (F) Immunostaining of cells either uninfected (mock) or infected with the RG-rescued SARS-CoV-2 and its mCherry and NLuc derivatives. Cells were stained using the N antibody utilised in (C–E) and imaged using a Celigo imaging cytometer. (G) WB validation of Vero E6 cells infected for 48 h with the RG viruses shown from Fig 1A for 48 h, using the 3 antibodies shown in (C–E). GST, glutathione S-transferase; IB, immunoblotting; IF, immunofluorescence; IP, immunoprecipitation; MBP, maltose-binding protein; MERS-CoV, Middle East Respiratory Syndrome Coronavirus; MOI, multiplicity of infection; NLuc, Nanoluciferase; RG, reverse genetics; SARS-CoV-2, Severe Acute Respiratory Syndrome Coronavirus 2; WB, western blotting; wt, wild-type.

https://doi.org/10.1371/journal.pbio.3001091.g002

The generation of the antibodies is illustrated in Fig 2B. Briefly, individual glutathione S-transferase (GST)-tagged SARS-CoV-2 proteins were used to inoculate sheep, and immunoglobulin G (IgG) was subsequently isolated and affinity purified using maltose-binding protein (MBP)-tagged variants of each viral protein. To validate our panel of antibodies, we initially examined their utility in indirect IF confocal microscopy. Vero E6 cells were mock infected or infected with SARS-CoV-2 England-02 (from Public Health England) at multiplicity of infection (MOI) 0.1 for 48 h, and then fixed and stained with each antibody. Antigen staining was highly specific to SARS-CoV-2-infected cells, exemplified by the staining of ORF3a, nsp2, and N proteins (Fig 2C). Antibodies raised against the other SARS-CoV-2 proteins were also highly specific, and more importantly, revealed the distinct staining and spatial distribution profiles of most SARS-CoV-2 proteins (S2 Fig).

In a similar fashion, we validated our panel of antibodies targeting SARS-CoV-2 proteins using WB. Vero E6 cells were infected with the England-02 isolate of SARS-CoV-2 at MOI 0.1 or 1 (as indicated) for 72 h (to ensure that most of the culture was infected using both multiplicities). Importantly, we were able to detect all available SARS-CoV-2 proteins using our antibodies, again exemplified by ORF3a, N, and nsp2 (Fig 2D) with the data for all remaining antibodies shown in S3 Fig. Interestingly, previous work has proposed that the ORF10 protein is unlikely to be expressed and may be a misannotation [17]. Remarkably, ORF10 was specifically detected in infected cells by WB and IF using our antibodies (S2 and S3 Figs).

Furthermore, in the small number of instances examined, substantial cross-reactivity of the antibodies was observed. For example, the antibodies raised against SARS-CoV N, MERS-CoV N, and HCoV OC43 N efficiently recognised SARS-CoV-2 N, whereas cross-reactivity was not observed for antibodies raised against N from the alphacoronavirus HCoV 229E (S4 Fig).

In addition to IF and WB, we examined the utility of our panel of antibodies in immunoprecipitation (IP) assays. Here, Vero E6 cells were infected with England-02 for 3 days at MOI 0.1. Interestingly, the antibodies generally appeared more sensitive in IP assays than WB, and anti-HCoV 229E N cross-reactivity with SARS-CoV-2 N that was not observed by WB was evident by IP (S4 Fig). Although cross-reactivity was present for SARS-CoV E, no cross-reactivity was observed by IP or WB using anti-MERS-CoV E and SARS-CoV-2 antigens (S4 Fig). Importantly, the antibodies resulted in efficient and specific IP of all but one (envelope [E]) of the SARS-CoV-2 proteins we raised antibodies to (Fig 2E and S5 Fig).

Next, we validated our SARS-CoV-2 N antibody in cells infected with the icDNA-derived viruses. Vero E6 cells were infected with the RG-rescued SARS-CoV-2-Wuhan-Hu-1 (wt), SARS-CoV-2-NLuc, or SARS-CoV-2-mCherry for 48 h, and N protein was detected by IF (Fig 2F). Similarly, ORF3a, nsp2, and N proteins were detected in Vero E6 cell lysates after infection with all the RG viruses from Fig 1A, confirming the utility of our antibody panel (Fig 2G). As nearly all antibodies are compatible with IF, WB, and IP, they represent an invaluable resource to researchers investigating SARS-CoV-2 and COVID-19.

We next investigated whether our antibody panels could be a useful tool in the study of protein–protein interactions, for which co-IP is a common technique. We performed co-IPs to look at interactions between different viral proteins based on studies performed with other CoVs [18–21]. Using lysates from infected Vero E6 cells (England-02, MOI 0.1), we first immunoprecipitated ORF3a protein using our ORF3a antibody and probed for the presence of spike (S) protein and ORF3a (Fig 3A) [18–21]. This revealed a predicted interaction between SARS-CoV-2 S and the accessory protein ORF3a, that had been observed in SARS-CoV. Likewise, we performed co-IPs and confirmed interactions between the SARS-CoV-2 nsp12 and nsp13 proteins (Fig 3B) [22]. In addition, we revealed a predicted interaction between SARS-CoV-2 N and M proteins (Fig 3C). Altogether, our coronavirus toolkit’s near-comprehensive SARS-CoV-2 antibody panel will greatly facilitate the study of coronavirus protein–protein interactions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Usage of our toolkit antibodies to demonstrate or confirm SARS-CoV-2 protein interactions.

(A) A co-IP was performed using lysates from Vero E6 cells infected with England-02 at MOI 0.1 for 3 days. Using the specific anti-ORF3a antibody described herein, SARS-CoV-2 ORF3a was immunoprecipitated (alongside a preimmune IgG control), and the immune complexes were western blotted for the presence of SARS-CoV-2 spike (S) and ORF3a. (B) As in (A), SARS-CoV-2 nsp13 (or IgG control) was immunoprecipitated and the immune complexes were probed for nsp13 and nsp11/12 by WB. (C) As in (A, B), SARS-CoV-2 N (or IgG control) was immunoprecipitated and the immune complexes were probed for matrix (M) and N by WB. co-IP, co-immunoprecipitation; IB, immunoblotting; IgG, immunoglobulin G; MOI, multiplicity of infection; SARS-CoV-2, Severe Acute Respiratory Syndrome Coronavirus 2; WB, western blotting.

https://doi.org/10.1371/journal.pbio.3001091.g003

Generation of SARS-CoV-2 permissive cell lines

SARS-CoV-2 has been shown to infect a variety of cell types, in particular nasal, epithelial, and lung (bronchoalveolar, bronchial, and epithelial) cells [13]. In addition, multiple reports indicate that gut epithelial and brain cells may also be natural targets of infection [23,24]. In humans, angiotensin-converting enzyme 2 (ACE2) is the main receptor and the TMPRSS serine proteases cleave the viral S protein, priming it for infection. Importantly, blockade/inhibition of these proteins has been shown to inhibit viral infection [25]. Because many cell lines derived from human lung are not susceptible to SARS-CoV-2 infection, such as A549 cells, we sought to engineer susceptibility in cell lines that can easily be adapted for many assays (including medium- to high-throughput formats). We transduced Vero E6 and A549 epithelial cell lines using LV lentiviral vectors to generate ACE2 and TMPRSS2-expressing cell lines, as well as double-expressing, ACE2-TMPRSS2 lines (Fig 4A). We visualised ACE2 and TMPRSS2 expression by WB (Fig 4B) and also assessed ACE2 expression using IF (Fig 4C–4F). The modified cell lines expressed ACE2 at higher levels than Calu-3 and Vero E6 cells (Fig 4B) and exogenous TMPRSS2 expression reduced ACE2 expression as has been reported previously [26]. Notably, in both cell backgrounds, exogenous ACE2 increased the susceptibility to infection, increasing the clarity of plaques in Vero E6 cells and converting A549s from nonsusceptible to permissive target cells (Fig 4G–4I). The exogenous expression of TMPRSS2 alone, in Vero E6 cells, increased the size of the plaques. Interestingly, Vero E6-ACE2-TMPRSS2 (VAT cells) yielded larger and clearer plaques (Fig 4G), which were more susceptible to infection as evidenced by the >10-fold increase in pfu/ml relative to unmodified cells (Fig 4H). Similar results were observed in AAT (A549-ACE2-TMPRSS2) cells, where exogenous TMPRSS2 expression greatly increased plaque size and clarity, and modestly increased the number of plaques (Fig 4I).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Conventional cell lines modified to express ACE2 and TMPRSS2 lentiviruses have utility in SARS-CoV-2 phenotypic assays.

(A) A schematic illustrating the LV-ACE2 and LV-TMPRSS2 lentiviruses that have been used to transduce multiple cell lines. The corresponding abbreviations of the modified lines are also shown. (B) The abundance of ACE2 and TMPRSS2 in Vero E6 cells, A549 cells and derivatives modified to express exogenous ACE2 and/or TMPRSS2 was assessed by WB. Permissive Calu-3 and Caco-2 samples were included for reference. Two actin blots are presented as different samples of equivalent cells were used to stain for ACE2 and TMPRSS2. (C) ACE2 intensity values from confocal microscopy images of ACE2 and Hoechst-stained Vero E6 cells and VA, VT, and VAT derivatives were measured using Cell Profiler (cellprofiler.org). (D) Randomly selected confocal microscopy images used for the quantification in (C) are shown. (E) As in panel C, using stained A549 cells, as well as AA and AAT cell derivatives. (F) Randomly selected confocal microscopy images used for the quantification in (E) are shown. (G) The plaque phenotype produced by the same preparation of SARS-CoV-2 England-02 on Vero E6 cells and VA, VT, and VAT derivatives. (H) As in (G), indicating the titre (in pfu/ml) of the same preparation of SARS-CoV-2 England-02 in cells with or without exogenous ACE2 and/or TMPRSS2. (I) As in panel G, AA and AAT derivatives of A549 cells were infected with SARS-CoV-2 England-02 to observe the plaque phenotype. (J) An example of a phenotypic well clearance/monolayer integrity assay used to assess the anti-SARS-CoV-2 activity of various compounds (nafamostat, apilimod, EIDD_2801, or remdesivir). The cells were treated with 2-fold serially diluted compound (10 μM to 20 nM) before being mock infected or infected with SARS-CoV-2. At 72 h postinfection, the monolayers were fixed and Coomassie-stained before scanning using a Celigo imaging cytometer. (K) As in panel J, well clearance assays in AA and AAT cells using SARS-CoV-2 England-02 are shown. (L) Quantification of the anti-SARS-CoV-2 activity of apilimod in AA and AAT cells. The mean and standard error of 4 replicates are plotted. (M) Dose response curve of remdesivir using the well-clearance assay in AA cells (as in panel K) multiplexed with a dead cell protease toxicity assay. The mean and standard error of 4 replicates are plotted. (N) as in panel M using AAT cells. The data underlying Fig 4C, 4E, 4H, 4L, 4M and 4N may be found in S1 Data. AA, A549-ACE2; AAT, A549-ACE2-TMPRSS2; ACE2, angiotensin-converting enzyme 2; SARS-CoV-2, Severe Acute Respiratory Syndrome Coronavirus 2; VA, Vero E6-ACE2; VAT, Vero E6-ACE2-TMPRSS2; VT, Vero E6-TMPRSS2; WB, western blotting.

https://doi.org/10.1371/journal.pbio.3001091.g004

Crucially, these modified cell lines are suitable for cytopathic effect (CPE)-based assays. CPE assays can be useful tools in compound screening and compound characterisation, and these cell lines provide a clear output for medium-throughput screening. For illustration, using VA (Vero E6-ACE2) cells, we examined potential inhibitory compounds using conditions where a fixed dose of SARS-CoV-2 would cause substantial CPE, resulting in the cell monolayer being “cleared” in the absence of any inhibitory compound. Virus-induced CPE was effectively abrogated following treatment with apilimod, a phosphatidylinositol-3-phosphate 5-kinase (PIKfyve) inhibitor (Fig 4J). In contrast, ribonucleoside analogues remdesivir and EIDD-2801 were less effective, but showed dose-dependent inhibition, while nafamostat, a serine protease inhibitor, displayed no inhibitory effect against SARS-CoV-2. Importantly, the inhibitory effect of apilimod was lost in the presence of TMPRSS2 (VAT cells), whereas the inhibitory effects displayed by remdesivir and EIDD-2801 were unaffected (Fig 4J), suggesting that apilimod’s mode of action may be ineffective when TMPRSS2-dependent infection occurs. These observations were recapitulated in AA (A549-ACE2) and AAT cells (Fig 4K). The CPE-based assays are also quantitative, allowing dose response curves to be generated and EC₅₀ values to be calculated. For example, dose response curves of apilimod inhibition of SARS-CoV-2 in AA and AAT cells are shown in Fig 4L. Moreover, the CPE assays can be multiplexed with other assays to monitor compound toxicity in parallel (Fig 4M and 4N), and the utility of the AA and AAT cells is indicated in S6 Fig, where these cells were used to consider the ability of a panel of compounds to inhibit SARS-CoV-2.

Establishment and analysis of clinical isolates

The modified cell lines forming part of this resource are also useful for isolating and characterising SARS-CoV-2 from clinical samples [27]. We have isolated 3 SARS-CoV-2 viruses, CVR-GLA-1, CVR-GLA-2, and CVR-GLA-3, from samples collected from patients hospitalised in Scotland (2 from sputum samples, one from a bronchoalveolar lavage (BAL) sample). Using our modified cell lines to characterise the plaque phenotypes of these isolates, we found that the CVR-GLA virus isolates all caused larger plaques than England-02 (Fig 5A and 5C), consistent with the enhanced particle infectivity proposed for the D614G substitution [28]. In addition, while titres across isolates were generally similar to those of England-02, they were increased in the presence of exogenous ACE2 and further increased by the presence of exogenous ACE2 and TMPRSS2, in line with our earlier observations (Fig 5B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Toolkit of SARS-CoV-2 clinical isolates.

(A) A comparison of plaque phenotypes in VAT cells from the 3 SARS-CoV-2 viruses isolated in this study and the England-02 SARS-CoV-2 virus. (B) A comparison of viral titers (in pfu/ml) of the 3 SARS-CoV-2 viruses isolated in this study and the England-02 SARS-CoV-2 virus in AA, AAT, Vero E6 (E6), VA, and VAT cells. All cells with exogenous ACE2 or TMPRSS2 were transduced with the lentiviruses described in Fig 4. (C) Quantification of plaque area (in mm²) was calculated in ImageJ (plaques not overlapping other plaques or the side of the well) from 4 replicates of the data presented in (A) and (B). Significance was determined using a Mann–Whitney U test. (D) A table showing the amino acid substitutions present in England-02 and 3 SARS-CoV-2 viruses isolated from clinical samples in this study, relative to Wuhan-Hu-1 (NC_045512). CVR-GLA-1 (MT882022) was isolated from sputum (CVR837 [EPI_ISL_461705]), CVR-GLA-2 (MT906650) was isolated from sputum (CVR2224 [EPI_ISL_448167], and CVR-GLA-3 (MT906649) was isolated from bronchoalveolar lavage (CVR3899_BAL [EPI_ISL_490695]). (E) IF of the 3 SARS-CoV-2 viruses isolated in this study and the England-02 SARS-CoV-2 virus in AAT cells (at MOI 0.01 for 48 h), using ORF3a, N, and nsp2 antibodies, as in Fig 2. The data underlying Fig 5B and 5C may be found in S1 Data. AA, A549-ACE2; AAT, A549-ACE2-TMPRSS2; ACE2, angiotensin-converting enzyme 2; IF, immunofluorescence; MOI, multiplicity of infection; SARS-CoV-2, Severe Acute Respiratory Syndrome Coronavirus 2; VA, Vero E6-ACE2; VAT, Vero E6-ACE2-TMPRSS2.

https://doi.org/10.1371/journal.pbio.3001091.g005

Clinical isolates can be particularly useful given the interest in phenotypic variation as the pandemic unfolds, and for this reason, these viruses were sequenced to identify amino acid substitutions relative to Wuhan-1 and England-02 (Fig 5D). For example, variants such as D614G in spike (S), which occurs in all of our isolates, have been the subject of intense investigation [28]. Moreover, CVR-GLA-2 exhibits the N439K substitution in S that has also been linked to immune escape from neutralising antibodies [29].

We used our antibody panel to validate the 3 CVR-GLA isolates by IF. AAT cells were uninfected or infected at MOI 0.01 for 48 h with England-02, CVR-GLA-1, CVR-GLA-2, or CVR-GLA-3. The cells were then probed with antibodies against ORF3a, nsp2, and N. As observed in Fig 2C, each of these antibodies displayed strong specificity against their target proteins expressed by all 3 viruses (Fig 5E), further validating the reagents generated in this study.

Because most viruses become tissue culture-adapted when they are extensively passaged in vitro, we sought to evaluate whether the human AAT cells might be an alternative substrate to African green monkey Vero E6 cells for amplifying SARS-CoV-2. We passaged the CVR-GLA-1 isolate on Vero E6 cells and AAT cells (Fig 6A) and sequenced the P2 and P6 swarms (Fig 6B). Strikingly, there was little overlap between the variants observed during passage of the mCherry (Fig 1N and 1O and S1 Table) and GLA-1 (Fig 6B and S1 Table) viruses in Vero E6 cells, making it difficult to draw generalisable conclusions. We specifically examined the furin cleavage site in our sequencing data as variation in this region has received a lot of attention. The multibasic S1/S2 furin cleavage site in SARS-CoV-2 S is essential for the infection of human lung cells [30], and multiple studies have reported that this site can be rapidly lost during passage in Vero E6 cells [31,32], likely driven by the enhanced growth of furin deletion mutants in Vero E6 cells [33]. In the 6 lineages of SARS-CoV-2-mCherry and CVR-GLA-1 that were propagated in Vero E6 cells, only 1 lineage out of 6 exhibited a deletion of the furin site at >0.5% abundance. Notably, in lineages that did contain a furin cleavage site deletion (2/6), the deletion variant did not overrun the culture (and even decreased slightly in frequency in 1 lineage), suggesting that these mutants did not have a substantial growth advantage in Vero E6 cells. Surprisingly, while CVR-GLA-1 passaged in AAT cells only lost the furin cleavage site in 1 of 3 replicates (at 1.46%), 1 replicate did contain a deletion mutant near this site at a frequency of approximately 10%. Thus, the AAT cells appear worse than the Vero E6 cells at preserving the integrity of the SARS-CoV-2 furin cleavage site during passage.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Sequence changes observed in SARS-CoV-2 CVR-GLA-1 following in vitro propagation.

(A) A schematic of the passage history of the SARS-CoV-2 CVR-GLA-1 virus in Vero E6 and AAT cells. Each filled circle represents 1 day of propagation. (B) A summary of the variation generated during the passage of the CVR-GLA-1 virus indicated in panel A. The variation is detailed in S1 Table. The percentage of the viral swarm displaying all variants that exceed 5% or 25% of the swarm at any time point is shown. The percentage occurrence of several deletion mutants is highlighted for interest. AAT, A549-ACE2-TMPRSS2; SARS-CoV-2, Severe Acute Respiratory Syndrome Coronavirus 2.

https://doi.org/10.1371/journal.pbio.3001091.g006

These data highlight the heterogeneous nature of SARS-CoV-2 propagation, where virus strain, cell culture conditions, and passaging protocol all likely influence lab-to-lab variation of virus stocks. Moreover, these data underscore the importance of facile, robust, high fidelity rescue systems (such as the plasmid system described herein) for generating uniform SARS-CoV-2 virus stocks that avoid the variation introduced when viral stocks are extensively passaged.

DNA-launched SARS-CoV-2 plasmid construction

pCC1-4K-SARS-CoV-2-Wuhan-Hu1 (MT926410) and its derivatives, pCC1-4K-SARS-CoV-2-mCherry (MT926411), pCC1-4K-SARS-CoV-2-NLuc (MT926412), and pCC1-4K-SARS-CoV-2-ZsGreen (MW289908) were assembled from synthetic DNA fragments (Genscript). The clones were assembled from 5 synthetic DNAs assembled in a stepwise fashion using conventional restriction enzyme cloning.

Rescue of infectious virus

A total of 3 μg of plasmid DNA (containing icDNA of SARS-CoV-2) and 3 μl of Lipofectamine LTX with 3 μl of PLUS reagent were used to transfect BHK-21 cells (ECACC) in 6-well plates. On the next 3 days posttransfection, supernatant was transferred to Vero E6 cultures in T25 flasks. Virus was allowed to propagate in Vero E6 cells for a further 4 days before harvesting the P0 stock. Infectious titres were enumerated by counting infected foci or via plaque assay. Rescued virus was used to infect Vero E6 cells in 96-well plates. For the Nanoluc assay, cells were lysed using a passive lysis buffer (Promega, cat#E1941) at 24 hpi. NanoLuc activity was determined using the Nano-Glo Luciferase Assay kit (Promega, Cat# N1130) and measured using a GloMax Luminometer (Promega).

Modified cell lines

The pLV-EF1a-IRES-Hygro (Addgene plasmid #85134) and pLV-EF1a-IRES-Neo (Addgene, plasmid #85139) lentiviral vectors were modified to encode synthesised (Integrated DNA Technologies) ACE2 (GenBank NM_001371415.1) and TMPRSS2 (GenBank NM_005656). Lentiviral vectors were produced through transient transfection as described previously [42] and used to modify A549 cells (ATCC#CCL-185; generous gift from Prof. Ben Hale, validated by STR analysis (Eurofins)) and Vero E6 cells (ATCC#CRL-1586; generous gift of Prof. Michele Bouloy) that were cultured and transduced under standard culture conditions.

Isolation of SARS-CoV-2 viruses from clinical samples

Sputum and BAL residual clinical samples from SARS-CoV-2-infected individuals were obtained for culture and used to inoculate Vero E6 or VAT cells in T25 flasks. Samples were harvested between 48 and 96 h postinfection, depending on the severity of CPE. Viral titers were determined by plaque assay.

Sequencing and metagenomics of isolated SARS-CoV-2 viruses

To determine the sequence of CVR-GLA-1, CVR-GLA-2, and CVR-GLA-3 isolated from clinical samples (CVR837, CVR2224, CVR3899_BAL) and the RG-rescued SARS-CoV-2 viruses, RNA was extracted from culture supernatant using a hybrid Trizol-RNeasy protocol. Library was made from cDNA using Kapa LTP Library Preparation Kit for Illumina Platforms (Kapa Biosystems, cat# KK8232). The sequencing of the libraries was carried out on Illumina’s NextSeq 550 System (Illumina, cat# SY-415-1002).

Plaque assays

Monolayers of A549 or Vero E6 cells and their ACE2/TMPRSS2 derivatives were prepared by seeding 2.5 × 10⁵ cells/well in 12-well plates, and plaque assays were executed using standard conditions. Briefly, a 1-ml overlay of MEM/0.6% Avicel RC-591/2% FCS was used, and fixed plates were stained with Coomassie Blue staining solution (0.1% Coomassie Brilliant Blue R-250/45% methanol/10% acetic acid) before scanning using a photo scanner (Epson Expression 1680 Pro).

Well-clearance/monolayer integrity CPE assay

Cells were seeded in optical 96-well plates (Perkin Elmer CellCarrier-96 Ultra Microplates Cat# 6055302) The following day, the cells were incubated with 10 μM to 20 nM of the various compounds. The cells were then mock infected or infected with a predetermined dose of SARS-CoV-2 that would cause substantial CPE by 72 h postinfection. Fixed plates were stained (0.1% Coomassie Brilliant Blue R-250/45% methanol/10% acetic acid) and scanned using a Celigo imaging cytometer (Nexcelom), allowing monolayer integrity to be quantified (to facilitate quantification of well-clearance to assess virus growth and toxicity). Dead cell protease activity was measured in the same plates using the CytoTox-Glo Cytotoxicity Assay (Promega) in accordance with the manufacturer’s instructions.

Generation of coronavirus antibodies

N-terminal GST fusion proteins for each coronavirus protein were used as antigens, and sheep were immunised with each antigen, followed by up to 5 further injections 28 days apart, with bleeds performed 7 days after each inoculation. Antibodies were subsequently affinity purified from serum using N-terminal MBP-tagged recombinant protein. All antibodies (and associated proteins and cDNA clones) generated in this paper are openly available at: https://mrcppu-covid.bio/.

Ethics statement

This project falls under the scope and authority of project licence P134D890E, issued by the UK Home Office in accordance with the Animals (Scientific Procedures) Act 1986. This licence will stay in force, unless revoked earlier, until 2 April 2024. Prior to submission to the Home Office, the application for the licence was scrutinised by the University of Dundee Ethical Review Committee, which approved the proposal at its meeting in March 2019. The Committee (now termed the Welfare and Ethical Use of Animals Committee) is a subcommittee of the University Court, its highest governing body. All projects undertaken under the licence are reviewed in terms of scientific rigor and possible alternative methods to avoid animal use, as well as AWERB in terms of welfare.

Immunofluorescence

Cells infected with SARS-CoV-2 were fixed and permeabilised in 8% formaldehyde/1% Triton X-100 at designated time points postinfection. DAPI (Thermo Fisher Scientific, cat# P36395) was included in the mounting medium, and slides were stained with primary sheep anti-coronavirus antibodies (1:500) and a rabbit anti-sheep secondary antibody (Abcam cat# ab150182) at a 1:1000 dilution before imaging (Zeiss LSM 880 confocal microscope). Alternatively, fixed and permeabilised cells were labelled using a mouse anti-N (nucleocapsid) IgG antibody (SinoBiologicals; cat# 40143-MM08), which was detected using an AlexaFluor 488-conjugated goat anti-mouse secondary antibody (Life Technologies; cat# A28175). Cells were subsequently counterstained with DAPI (1 μg/ml) and mounted using ProLong Gold antifade (Life Technologies; cat# P36930). Sequential channels were acquired as z-stacks (1.2 μm step) using an Olympus FV3000 confocal microscope and processed using Imaris software (v 9.0.2; Bitplane).

For the enumeration of infected foci, the sheep anti-SARS-CoV-2 N antibody was used (at a 1:500 dilution) and visualised using donkey anti-sheep antibody (A11015, Thermo Fisher Scientific, 1:1000 dilution) prior to scanning (Celigo, Nexcelom).

Immunoprecipitation

At 3 days postinfection (at MOI 0.1), cells were trypsinised and rinsed in ice cold PBS before being lysed in ice-cold lysis buffer [43]. IPs were set up overnight at 4°C with the antibody and protein lysate in a ratio of 3 μg of antibody per mg of protein lysate. The immunoprecipitated proteins were subjected to SDS-PAGE and then transferred to PVDF membranes (Merck Millipore, IPFL00010), which were then incubated with their respective biotinylated primary antibodies (0.5 to 1.0 mg/ml). Bands were visualised (LI-COR Odyssey CLx imaging system) using labelled streptavidin (DyLight 680/800, Thermo Fisher Scientific, 21848/21851).

Western blotting

Briefly, sonicated cell lysates were resolved using Novex 4% to 12% acrylamide gels (Life Technologies) before wet-transfer onto nitrocellulose membranes (Sigma Aldrich, cat# GE10600003), after which membranes were probed with the indicated primary antibody (full list of antibodies and their availability found in S2 Table) in SEA BLOCK (Thermo Scientific, cat# 37527), unless otherwise specified. Bands were visualised using DyLight-labelled secondary antibodies (Thermo Fisher Scientific) prior to scanning with a Li-Cor Odyssey CLx imaging system.

Materials availability

All sheep CoV antibodies (and corresponding proteins and cDNAs) listed, lentiviral constructs (and stable Vero E6 and A549 cell lines), and DNA-launched SARS-CoV-2 RG constructs are openly available through a not-for-profit web interface (antibodies, proteins, rescue plasmids at https://mrcppu-covid.bio/; cell lines and virus strains have been deposited at the UK National Institute for Biological Standards and Controls; www.nibsc.org).