Experiments involving animals were performed in strict accordance with EU (2010/63/UE) and Spanish (RD 53/2013 and 32/2007) guidelines. All the protocols were approved by the on-site Ethical Committee (permit n°. PROEX 146.6/20). Infected mice were housed in a self-contained ventilated rack (Allentown, NJ).

Monkey Vero E6 (ATCC CRL-1586), hamster kidney (BHK-21, CCL-10) and human normal lung fibroblasts (MRC-5, CCL-171) cells were obtained from American Type Culture Collection. Vero E6 cells expressing human TMPRSS2 protease (VeroE6-TMPRSS2, JCRB1819) were obtained from Japanese Collection of Research Bioresources. Human liver-derived Huh-7 cells were kindly provided by R. Bartenschlager (University of Heidelberg, Germany). The bronchial epithelial cell line Calu-3 2B4 (Yoshikawa et al., 2010) was kindly provided by C.T. Tseng (University of Texas Medical Branch, USA). Human lung-derived A549 cells expressing human ACE2 (A549-ACE2) cells were kindly provided by R. Andino (University of California San Francisco, USA) and A549 cells expressing both human ACE2 and TMPRSS2 (A549-ACE2-TMPRSS2) were obtained from InvivoGen (San Diego, CA, USA). Human kidney-derived cells expressing human ACE2 (HEK293-ACE2) were obtained from Innoprot (Derio, Spain). Cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM, Lonza) supplemented with 25 mM HEPES, 10% fetal bovine serum (FBS, HyClone), 2% glutamine, 1% non-essential amino acids (Sigma) and maintained at 37°C in a humidified atmosphere of 5% CO₂. Calu-3 2B4 cells were grown in the medium described above supplemented with 20% FBS. Several cell lines required antibiotics for maintenance: 1 mg/ml of G418 for Huh7-ACE2, A549-ACE2 and VeroE6-TMPRSS2 cells; 80 µg/ml of hygromycin for HEK293-ACE2 cells; 0.5 µg/ml of puromycin and 300 µg/ml of hygromycin for A549-ACE2-TMPRSS2.

Recombinant SARS-CoV-2 (rSARS-CoV-2) viruses were rescued from the corresponding infectious cDNAs. To obtain a SARS-CoV-2 clinical isolate as a control, six nasopharyngeal swabs from COVID-19 patients were kindly provided by R. Delgado (Hospital 12 de Octubre, Madrid, Spain). The swab samples were diluted in 500 µl of cell culture medium. Confluent Vero E6 cells, cultured as indicated above in a F12.5 flask, with medium supplemented with 100 U/ml penicillin, 100 mg/ml streptomycin, 50 mg/ml gentamicin and 0.5 mg/ml amphotericin B, were infected with half of the clinical sample. At 48 hpi, supernatants were collected and filtered through a 0.22 µm Millex-GS filter (Millipore) to eliminate potential bacterial and fungi contamination. Total RNA was extracted from half of the supernatant, using QIAamp Viral Mini kit (Qiagen) following the manufacturer’s instructions, and sent for next generation sequencing at Hospital 12 de Octubre (Madrid, Spain). Attending to the observed cytopathic effect and the obtained sequencing data, SARS-CoV-2-MAD6 was selected, which derived from a 69-year-old male patient clinical sample. The virus was plaque purified three times and viral stock was prepared following standard procedures set up in our laboratory (DeDiego et al., 2007). Virus titration was performed in Vero E6 cells, following standard procedures (DeDiego et al., 2007). All experiments with SARS-CoV-2 infectious viruses were performed in BSL-3 facilities at CNB-CSIC according to the guidelines set forth by the institution.

Immunofluorescence analysis of human ACE2 expression in Huh-7 cells showed heterogeneous expression of this protein (data not shown). Then, two strategies were followed. First, Huh-7 cells were reverse transfected with 2 µg of plasmid pcDNA-hACE2, encoding human ACE2, kindly provided by M. Farzan (Harvard Medical School, USA) (Li et al., 2004), using LipofectamineTM 2000 (ThermoFisher Scientific) following the manufacturer’s recommendations. At 48 h post transfection (hpt) G418 (1 mg/ml) selection was applied for 10-12 days. The G418-containing growth medium was replaced every two days and cell toxicity was visually examined. Huh-7-ACE2 cells were then infected for subsequent analysis. The second strategy was to isolate independent cell clones from Huh-7 cells and select for high ACE2 expression levels. After several rounds of cell cloning by limiting dilution, seven independent ACE2-positive clones were selected by immunofluorescence. Huh-7 1D11 cell clone was finally selected based on high ACE2 expression levels and other cell characteristics, such as duplication time, morphology, etc.

To engineer a SARS-CoV-2 infectious cDNA (pBAC-SARSCoV-2^FL), six DNA fragments were chemically synthesized and purchased from GenScript (New Jersey, USA) ( Table 1 ). The full-length cDNA clone (pBAC-SARSCoV2^FL) was assembled by sequential cloning of overlapping DNA fragments (F1-F6), covering the entire SARS-CoV-2 Wuhan-Hu-1 isolate (SARS-CoV-2-WH1) genome (GenBank MN908947), into the plasmid pBAC using the unique restriction sites selected shown in Table 1 .

The plasmid pBAC-SARSCoV2^FL-D614G, containing the prevalent A23403>G mutation, was generated by overlapping PCR, using plasmid pUC57-F5 ( Table 1 ) as a template. First, a 655 bp PCR product was amplified with oligonucleotides WH-22763-VS (5’- GTAATTAGAGGTGATGAAG-3’) and D614G-RS (5’-CTGTGCAGTTAACACCCTGATAAAGAACAGC-3’, A23403>G mutation in bold type), and a 1,788 bp PCR product was amplified using oligonucleotides D614G-VS (5’-GCTGTTCTTTATCAGGGTGTTAACTGCACAG-3’, A23403>G mutation in bold type) and WH-25174-RS (5’- TCCAAGTTCTTGGAGATCG-3’). Overlapping PCR product was then amplified using intermediate PCR products as templates and oligonucleotides WH-22763-VS and WH-25174-RS. The final 2,412 bp PCR product was digested with BstEII and EcoRV and cloned into the same sites of pUC57-F5 ( Table 1 ) to generate pUC57-F5-D614G. Finally, pUC57-F5-D614G was digested with SanDI and BamHI and cloned into the same sites of pBAC-SARSCoV2^FL, to generate the pBAC-SARSCoV2^FL-D614G. The genetic integrity of the cloned DNAs was verified throughout the subcloning and assembly process by extensive restriction analysis and sequencing.

To engineer a SARS-CoV-2-derived RNA replicon, the full-length infectious cDNA clone pBAC-SARSCoV2^FL, as described above, was used as a backbone. A DNA fragment of 4,028 bp was chemically synthesized and purchased from GeneArt (Thermo Fisher Scientific, Germany). This fragment contained SARS-CoV-2 genome nt 20,085 to 21,556, fused to M gene transcription regulating sequence (TRS-M, nt 26,431 to 26,523) preceding mNeonGreen reporter gene (Shaner et al., 2013), and nt 28,177 to 29,870, including N gene and its TRS sequence. The synthetic DNA fragment was digested with SanDI and RsrII, and cloned into the same restriction sites of pBAC-SARSCoV2^FL, leading to plasmid pBAC-SARSCoV2-REP-mNG.

To generate a non-replicative SARS-CoV-2 replicon as a control, plasmid pUC57-F2 ( Table 1 ) was digested with NheI. The insert, containing nt 1,545 to 4,201, was cloned in pUC57-F2 in the reverse orientation, generating plasmid (pUC57-F2-NR). Finally, pUC57-F2-NR was digested with BsiWI and PmeI, and cloned into the same restriction sites of pBAC-SARSCoV2-REP-mNG, leading to plasmid pBAC-SARSCoV2-REP-mNG-NR. All constructs were verified by extensive restriction analysis and sequencing.

VeroE6-TMPRSS2 cells were grown to 95% confluence in F12.5 flasks. Cells were transfected with 4 μg of SARS-CoV-2 infectious cDNAs and 12 μl of Lipofectamine™ 2000 transfection reagent (Thermo Fisher Scientific) according to the manufacturer’s instructions and incubated at 37°C. At 72 hours post-transfection (hpt), the supernatant was collected (P0, passage 0) and stored at -80°C. Viruses were cloned from P0 by three rounds of plaque purification following standard procedures. Two selected viral clones were amplified to obtain viral stocks.

Cells were infected with either SARS-CoV-2-MAD6 virus or recombinant rSARS-CoV-2 viruses at an moi of 0.1. Up to nine blind passages were performed each 48 hpi. Supernatants were collected for titration and cells for RNA extraction. The loss of furin cleavage site was suggested by the appearance of a large plaque phenotype during plaque titration assay. In addition, total RNA from infected cells at different passages was isolated using RNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions. Total cDNA was synthesized with random hexamers from 100 ng of total RNA as a template using High-Capacity cDNA reverse transcription kit (Applied Biosystems), following the manufacturer’s recommendations. A set of overlapping PCRs covering the whole SARS-CoV-2 genome was performed using oligonucleotides in Supplementary Table 1 . PCR products were sequenced and sequences were assembled using SeqMan Ultra software from DNA Lasergene 17.3 package (Madison, WI).

Transgenic K18-hACE2 mice (Zheng et al., 2021) were maintained at the CNB-CSIC animal core facility. Six-week-old Syrian golden hamsters were provided by Envigo. All work with infected animals was performed in a BSL3+ laboratory at the Center for Animal Health Research (CISA, CSIC-INIA). Animals were anesthetized with 5% inhaled isoflurane for induction and 1.5-2% for maintenance, and intranasally inoculated with different doses of SARS-CoV-2 in DMEM. Weight loss and mortality were evaluated daily. Animals approaching 25% of weight loss were euthanized by 5% overdose of inhaled isoflurane until respiratory arrest for 90 seconds (Douglas et al., 2018). Three animals per group were euthanized as indicated above and necropsied at days 3 and 6 after inoculation, depending on the experiment. Lungs were collected and examined for macroscopic lesions. Lungs were then divided and were either frozen for subsequent virus titration, placed in RNAlater stabilization reagent (Life Technologies) for RNA extraction, or fixed in ready-to-use zinc formalin fixative (Z2902, Sigma-Aldrich) for histopathology. To determine SARS-CoV-2 titers, lungs were homogenized in phosphate-buffered saline (PBS) containing 100 IU/ml penicillin, 0.1mg/ml streptomycin, 50 µg/ml gentamicin, and 0.5 µg/ml amphotericin B (Fungizone), using a BeadBug 6 microtube homogenizer (Merck) and Lysing matrix S tubes (MPbio).

Total intracellular RNA was purified from cells or tissue samples with RNeasy Mini kit (Qiagen) according to the manufacturer’s instructions. Total cDNA was synthesized with random hexamers from 100 ng of total RNA as a template using High-Capacity cDNA reverse transcription kit (Applied Biosystems), following the manufacturer’s recommendations. SARS-CoV-2 viral RNA synthesis was analyzed using two custom TaqMan assays specific for gRNA (forward primer 5’- GTGARATGGTCATGTGTGGCGG-3’, reverse primer 5’- CARATGTTAAASACACTATTAGCATA-3’, and MGB probe 5’- CAGGTGGAACCTCATCAGGAGATGC-3’) and sgmRNA-N (forward primer 5’- CCAACCAACTTTCGATCTCTTGT-3’, reverse primer 5’- GGGTGCATTTCGCTGATTTT-3’, and MGB probe 5’- TTCTCTAAACGAACAAACTA-3’). Cellular gene expression was analyzed using specific TaqMan gene expression assays ( Table 2 ) and hydroxymethylbilane synthase (HMBS) as a reference housekeeping gene. Data were acquired with a QuantStudio 5 real-time PCR system (Applied Biosystems) and analyzed with QuantStudio Desing & Analysis software v1.5.1. Relative quantifications were performed using the 2^-ΔΔCt method (Livak and Schmittgen, 2001). All experiments and data analysis were MIQE compliant (Bustin et al., 2009).

Cells grown on 13 mm glass coverslips were fixed by incubation with 4% paraformaldehyde in PBS for 40 min at room temperature. Cells were permeabilized with ice-cold methanol for 10 min and blocked with 10% FBS in PBS for 1h at room temperature. Primary antibodies were diluted in PBS with 5% FBS as follows: rabbit anti-ACE2 [EPR4436] (1:300, Abcam); rabbit anti-TMPRSS2 [EPR3862] (1:300, Abcam). Cells were then washed five times for 5 min each time with PBS and incubated for 30 min at room temperature with species-specific secondary antibodies conjugated to Alexa Fluor 488 or Alexa Fluor 594, diluted 1:500 in PBS with 5% FBS. Cell nuclei were stained with DAPI (1:200, Sigma). Finally, coverslips were mounted in Prolong Gold antifade reagent (Invitrogen). Confocal microscopy analysis was performed using a Leica SP5 laser scanning microscope, and images were collected and processed with LAS AF software (Leica, Wetzlar, Germany).

Analysis of viral RNA synthesis by SARS-CoV-2 replicons was performed as previously described (Becares et al., 2016). Briefly, RNA was isolated from transfected cells at 48 hpt using RNeasy kit (Qiagen). DNA was removed from samples with DNaseI and cleaned RNA was subjected to RT-qPCR for the detection of viral RNAs as described above.

Reporter gene expression was detected by fluorescence microscopy using a Leica DMI6000B fluorescence microscope and OrcaR2 digital camera for image capture. Fiji software (Schindelin et al., 2012) was used for minimal adjustment to enhance the contrast for bright-field images and to merge them with the fluorescence images.

Two-tailed, unpaired Student’s t tests were used to analyze the difference in mean values between groups. All results were expressed as mean ± the standard deviation (SD) or standard error of the mean (SEM). P values <0.05 were considered significant.

A SARS-CoV-2 infectious cDNA clone (pBAC-SARS-CoV-2^FL) was engineered by sequential assembly of synthetic DNA fragments, covering the entire viral genome, using selected unique restriction sites ( Figure 1 ). The cDNA clone sequence was identical to the reported Wuhan-Hu-1 isolate sequence (GenBank accession MN908947), with the exception of two silent point mutations: A20085>G, creating a unique SanDI restriction site, and G26840>C, eliminating MluI and BsiWI restriction sites. These point mutations were also used as genetic markers to distinguish between the virus recovered from the cDNA clone and a wild-type virus. By February 11^th 2023, there were more than five million SARS-CoV-2 genome sequences in the databases, but according to NextStrain analysis (Hadfield et al., 2018), the two engineered silent mutations were not present in the sequenced genomes, highlighting its usefulness as genetic markers.

The recombinant rSARS-CoV-2 virus was efficiently recovered from the infectious cDNA with a titer of 6×10⁶ pfu/ml. However, a mutated virus with a large plaque phenotype that emerged in passage 1 (rSARS-CoV-2 *), grew to up to 15-fold higher titers and outcompeted the wild-type virus within three passages on VeroE6 cells. Sequencing results revealed that the mutant virus contained a 15-nt deletion from nt 23583 to 23597 in the S gene. This deletion eliminated five amino acids (QTQTN) upstream to the furin cleavage site. Similar deletions, leading to an attenuated virus, were previously described (Lau et al., 2020). Using the best available animal model at that time, the hamster model, we observed that, in fact, the mutant virus adapted to VeroE6 cells was attenuated, causing no weight loss ( Figure 2A ) or macroscopic damage measured as lung weight increase ( Figure 2B ), in contrast with the patient-isolated SARS-CoV-2-MAD6 virus. These data were in agreement with published observations indicating that mutations close to or eliminating the furin cleavage site attenuate SARS-CoV-2 (Johnson et al., 2021; Vu et al., 2022).

To avoid virus instability, rSARS-CoV-2 virus was recovered using human Calu3 2B4 cells (Yoshikawa et al., 2010) and monkey VeroE6 cells overexpressing human TMPRSS2 protease (VeroE6-TMPRSS2). Recovered virus reached titers of 2×10⁶ pfu/ml and 1×10⁷ pfu/ml in Calu3 2B4 and VeroE6-TMPRSS2 cells, respectively. The rSARS-CoV-2 virus genome was fully sequenced and no genetic changes were identified. Moreover, in both cell lines the virus was fully stable for at least nine passages, maintaining an intact furin cleavage site.

The transgenic K18-hACE2 mice model was then available, being a better lethal model for SARS-CoV-2 infection than hamster (Zheng et al., 2021). To analyze the virulence of the engineered rSARS-CoV-2 virus, 16-week-old K18-hACE2 mice were infected with two different doses of isolate SARS-CoV-2-MAD6 or of recombinant rSARS-CoV-2 virus recovered from Calu3 2B4 cells. The mice infected with SARS-CoV-2 MAD6 and rSARS-CoV-2 viruses significantly lost weight compared with mock-infected mice ( Figure 2C ). In addition, all the mice infected with both isolated SARS-CoV-2-MAD6 and recombinant rSARS-CoV-2 viruses died between days 6 and 9 post-infection, depending on the inoculated dose ( Figure 2C ). These data indicated that recombinant rSARS-CoV-2 virus, recovered from the infectious cDNA, was virulent in K18-hACE2 mice.

As indicated above, genomic instability in SARS-CoV-2 was observed when the virus was passaged in VeroE6 cells, with deletions appearing in the S protein region close to the furin cleavage site. Calu3 2B4 cells are difficult to culture in monolayer and other authors have extensively used alternative cells lines for SARS-CoV-2 infection, although virus stability was not systematically tested in these cells. To improve the cell culture system, eleven human and non-human cell lines were tested. Seven of them were susceptible to SARS-CoV-2 infection ( Figure 3A ). Among these susceptible cell lines, SARS-CoV-2 growth was more robust in VeroE6-TMPRSS2, Huh-7-ACE2 and Calu3 2B4 cells with titers of 1×10⁷ pfu/ml, 3×10⁶ pfu/ml and 2×10⁶ pfu/ml, respectively. Moreover, virus sequencing indicated that the furin cleavage site was maintained at least for six passages ( Table 3 ). Significant virus production was also obtained in A549-ACE2, VeroE6 and Huh-7.5.1-ACE2 cells, with virus titers of 1×10⁷ pfu/ml, 6×10⁶ pfu/ml and 1×10⁶ pfu/ml, respectively. However, the furin cleavage site was not maintained during the passages in these cell lines ( Table 3 ). Huh-7.5.1 cells are derived from Huh-7.5 cells, both containing more than 50 mutations compared with parental Huh-7 cells, all of the changes being adaptive mutations for better hepatitis C virus (HCV) growth (Appel et al., 2005; Zhong et al., 2005; Kawamoto et al., 2020). This may explain the differences in virus stability observed between the two cell lines. Surprisingly, virus replicated efficiently in A549-ACE2-TMPRSS2 cells as determined by RT-qPCR ( Supplementary Figure 1 ), but no infectious virus was detected by plaque assays ( Figure 3A ).

It has been reported that SARS-CoV-2 infection and genetic stability required the expression of both human ACE2 receptor and TMPRSS2 protease (Hoffmann et al., 2020b; Peacock et al., 2021). The levels of these proteins were analyzed by immunofluorescence in VeroE6-TMPRSS2, Calu3 2B4 and Huh-7-ACE2 cells. As expected, human ACE2 was not detected in VeroE6-TMPRSS2 cells, while it was expressed in human cells ( Figure 3B , upper panels). TMPRSS2 protease was expressed in all cell lines, although nuclear and cytoplasmic location in Huh-7-ACE2 cells differed from the exclusive cytoplasmic location observed in the other two cell lines ( Figure 3B , lower panels). Similar cell-type dependent TMPRSS2 subcellular pattern was previously reported for other cell types and may have influence in cell susceptibility to the infection (Bertram et al., 2013; Thul et al., 2017).

A slight delay in weight loss and lethality was observed in susceptible transgenic mice infected with the rSARS-CoV-2-WH1 virus compared with those infected with isolate SARS-CoV-2-MAD6 virus ( Figure 2C ). The genomes of these two viruses were identical except for the two genetic markers and three-point mutations: the silent mutation C3037>T and two missense mutations C14408>T (affecting nsp12) and A23403>G (leading to D614G in S protein). Interestingly, the S protein D614G mutation became prevalent in SARS-CoV-2 isolates early during the global pandemic and it has been maintained in all SARS-CoV-2 variants since then (Antoneli et al., 2021; Chakraborty et al., 2021). To further characterize the cell culture system for SARS-CoV-2, based on our previous observation and given its rapid rise in human isolates and enhanced transmission, we engineered rSARS-CoV-2-D614G mutant virus as a more representative virus compared with the parental one. The genome of the rSARS-CoV-2-D614G mutant virus was identical to that of the parental rSARS-CoV-2 virus except for the point mutation A23403>G. Viral growth kinetics of rSARS-CoV-2-D614G mutant virus was determined in VeroE6-TMPRSS2 ( Figure 4A ), Calu3 2B4 ( Figure 4B ), and Huh-7-ACE2 ( Figure 4C ). No significant differences were observed in VeroE6-TMPRSS2 cells between the mutant and parental viruses. In contrast, the viral titers of D614G variant were significantly higher (up to 10-fold) than those of the parental virus in human cells, both Calu3 2B4 and Huh-7-ACE2 cells. These data were in agreement with previous observations (Hou et al., 2020; Yurkovetskiy et al., 2020; Zhang et al., 2020; Plante et al., 2021; Zhou et al., 2021). The increased viral titer ( Figure 5A ) was not due to differences in viral RNA synthesis, as the accumulation of gRNA and sgmRNA-N was similar in rSARS-CoV-2-D614G and parental virus infected Calu3 2B4 cells ( Figure 5B ). The rSARS-CoV-2-D614G mutant virus was virulent in K18-hACE2 mice with a slightly accelerated weight loss compared with the rSARS-CoV-2 virus, as expected ( Figure 5C ). Both viruses had similar virulence measured as mice survival, in line with published data (Korber et al., 2020; Stauft et al., 2021).

Based on previous data from in vivo models mimicking SARS-CoV-2 infection in humans (Zheng et al., 2021; Wong et al., 2022), a panel of cytokines was selected to analyze their mRNA accumulation in cells infected with rSARS-CoV-2-D614G virus. From the previously selected cell lines, only the cytokine expression in human Calu3 2B4 and Huh-7-ACE2 was evaluated, since VeroE6-TMPRSS2 cells are defective in the innate immune response. The levels of genes related with IFN response, such as IFN-β, IFN-λ, or IFN stimulated genes (ISGs) such as ISG15 and MX1, and the levels of pro-inflammatory cytokines, such as TNF, CCL2, CXCL10 or IL-6 were evaluated by RT-qPCR ( Figure 5 ). mRNAs for these genes were significantly increased in both infected Calu3 2B4 and Huh-7-ACE2 cells as compared with mock-infected cells. Nevertheless, the cytokine accumulation pattern observed in infected Calu3 2B4 mimicked the one observed in vivo, both in mouse models (Zheng et al., 2021; Wong et al., 2022) or in patients (Mudd et al., 2020; Galani et al., 2021; Tang et al., 2021), while cytokine increases were lower in infected Huh-7-ACE2 cells ( Figure 6 ).

Altogether, the data indicated that both VeroE6-TMPRSS2 and Calu3 2B4 cells were an adequate cell culture system to amplify SARS-CoV-2 virus, maintaining viral genetic stability and virulence. In addition, innate immune responses in Calu3 2B4 cell line resembled those observed in animal models and humans, making this cells line optimal for virus-host interaction studies.

The availability of a SARS-CoV-2 replicon provides an important tool for the study of fundamental viral processes and the screening of antiviral drugs in the absence of infectious virus, not requiring BSL-3 conditions. Therefore, a SARS-CoV-2-derived replicon was constructed (pBAC-SARSCoV2-REP-mNG) derived from the full-length infectious cDNA clone as described above ( Figure 7 ). This plasmid retained the untranslated 5’ and 3’ ends of the viral genome, the replicase genes, and the N gene, which is required for efficient CoV RNA synthesis (Almazan et al., 2004). A mNeonGreen (mNG) reporter gene was also introduced in the replicon plasmid, under the control of M gene TRS (Shaner et al., 2013). Since first amplification of replicon RNA comes from CMV promoter, a non-replicative RNA replicon was also engineered as negative control of viral RNA polymerase-dependent RNA synthesis ( Figure 7A ).

To evaluate the SARS-CoV-2 replicon functionality, the replicon was transfected into BHK (easily transfected cell line often used for CoV replicons analysis) or VeroE6-TMPRSS2 cells. Reporter protein expression was tested by fluorescence microscopy at 48 hpt. The mNG protein was expressed to high levels, compared with the non-transfected cells and the cells transfected with the non-replicating replicon ( Figure 7B ). Fluorescence was observed in isolated BHK cells, and in isolated foci of VeroE6-TMPRSS2 fused cells, demonstrating the inability of the replicon to propagate ( Figure 7B ).

Viral RNA synthesis was analyzed by RT-qPCR. Both gRNA and sgmRNA-N accumulated significantly in cells transfected with the SARS-CoV-2 replicon, compared with the non-replicative replicon control ( Figure 8 ). Altogether, the results indicated that SARS-CoV-2 replication-transcription complex was assembled, and viral RNAs and proteins were efficiently synthetized.