A Rapid Method for Generating Infectious SARS-CoV-2 and Variants Using Mutagenesis and Circular Polymerase Extension Cloning

ABSTRACT The appearance of SARS-CoV-2 variants in late 2020 raised alarming global public health concerns. Despite continued scientific progress, the genetic profiles of these variants bring changes in viral properties that threaten vaccine efficacy. Thus, it is critically important to investigate the biologic profiles and significance of these evolving variants. In this study, we demonstrate the application of circular polymerase extension cloning (CPEC) to the generation of full-length clones of SARS-CoV-2. We report that, combined with a specific primer design scheme, this yields a simpler, uncomplicated, and versatile approach for engineering SARS-CoV-2 variants with high viral recovery efficiency. This new strategy for genomic engineering of SARS-CoV-2 variants was implemented and evaluated for its efficiency in generating point mutations (K417N, L452R, E484K, N501Y, D614G, P681H, P681R, Δ69-70, Δ157-158, E484K+N501Y, and Ins-38F) and multiple mutations (N501Y/D614G and E484K/N501Y/D614G), as well as a large truncation (ΔORF7A) and insertion (GFP). The application of CPEC to mutagenesis also allows the inclusion of a confirmatory step prior to assembly and transfection. This method could be of value in the molecular characterization of emerging SARS-CoV-2 variants as well as the development and testing of vaccines, therapeutic antibodies, and antivirals. IMPORTANCE Since the first emergence of the SARS-CoV-2 variant in late 2020, novel variants have been continuously introduced to the human population, causing severe public health threats. In general, because these variants acquire new genetic mutation/s, it is critical to analyze the biological function of viruses that such mutations can confer. Therefore, we devised a method that can construct SARS-CoV-2 infectious clones and their variants rapidly and efficiently. The method was developed based on a PCR-based circular polymerase extension cloning (CPEC) combined with a specific primer design scheme. The efficiency of the newly designed method was evaluated by generating SARS-CoV-2 variants with single point mutations, multiple point mutations, and a large truncation and insertion. This method could be of value for the molecular characterization of emerging SARS-CoV-2 variants and the development and testing of vaccines and antiviral agents.

IMPORTANCE Since the first emergence of the SARS-CoV-2 variant in late 2020, novel variants have been continuously introduced to the human population, causing severe public health threats. In general, because these variants acquire new genetic mutation/s, it is critical to analyze the biological function of viruses that such mutations can confer. Therefore, we devised a method that can construct SARS-CoV-2 infectious clones and their variants rapidly and efficiently. The method was developed based on a PCR-based circular polymerase extension cloning (CPEC) combined with a specific primer design scheme. The efficiency of the newly designed method was evaluated by generating SARS-CoV-2 variants with single point mutations, multiple point mutations, and a large truncation and insertion. This method could be of value for the molecular characterization of emerging SARS-CoV-2 variants and the development and testing of vaccines and antiviral agents.
KEYWORDS SARS-CoV-2, infectious clone, mutagenesis, CPEC, reverse genetics S evere acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is one of three recognized, significant respiratory disease-causing viruses of the Coronaviridae family, together with severe acute respiratory syndrome coronavirus (SARS-CoV) and the Middle East Respiratory Syndrome Coronavirus (MERS-CoV). Since its first emergence in December 2019, coronavirus variants for further application of in vitro and in vivo studies will aid in the genetic characterization of these variants and enhance our understanding of virus evolution.

RESULTS
CPEC approach for generating recombinant SARS-CoV-2 virus. To confirm the utility of the CPEC method for reconstructing SARS-CoV-2 infectious clone, we designed primers to synthesize three cDNA fragments of 8.7 to 11.8 kb in size (Fig. 1A, B). Viral RNA extracted from the first passage of virus grown in Vero E6 cells was used for generating cDNA fragments by RT-PCR (Table 1). Each fragment was in-fused and subcloned individually into a designed pUC19 plasmid vector (Fig. S1A). The SARS-CoV-2 subcloning vector was modified to maintain the genomic fragment of SARS-CoV-2 and facilitate mutagenesis of the viral genome (Fig. S1B). The subcloning vector is equipped with two multirestriction endonuclease regions containing EagI, AsiSI, and ApaI sites; a T7 promoter for in vitro RNA transcription; two self-cleaving ribozymes, hammerhead (Hh) and hepatitis delta virus (HdV) ribozymes; and a T7 terminator (Fig. S1). Before assembly, each subclone was amplified using primers designed to contain a 15-bp extension of the 59-or 39-end sequence of the linearized subcloning vector (Table 1) for PCR generation of the three target fragments (8.7 to 11.8 kb each). After purification, the three genomic fragments of SARS-CoV-2 were assembled with a pYES1L vector using the CPEC cloning method. The generated PCR products can be directly processed for transformation into competent bacterial cells by electroporation without ligation or purification.
After successful transformation, 10 colonies were randomly picked for positive screening by junction PCR using fragment-specific primers (Table S1). After confirmation of correct size, positive colonies were seeded. To successfully perform in vitro transcription of the whole genomic length of 59 cap-containing SARS-CoV-2 cDNA by T7 RNA polymerase, we linearized the assembled infectious clone plasmid by digestion at ApaI sites located upand downstream of the insert (Fig. 1A). The recombinant SARS-CoV-2 virus was recovered by direct electroporation of the in vitro-transcribed, whole-genome-length RNA in BHK-21 cells. To further optimize the infectivity of RNA transcripts (25, 26), we further co-electroporated the SARS-CoV-2 N gene together with full-length SARS-CoV-2 RNA transcripts. Transfected BHK-21 cells were then overlaid onto Vero-E6 cells, and following transcription, the medium was replaced with Dulbecco's modified Eagle's medium (DMEM) containing 2% fetal bovine serum (FBS). After 72 to 96 h, transfected cells with cytopathic effects were observed, indicative of successful rescue of recombinant virus (WT-Rg); the resulting titer was estimated to be 7.8 Â 10 5 PFU/mL, and cells showed a plaque morphology similar to that of wild-type (WT-PC) (Fig. 1D). Furthermore, we observed comparable time courses and peaks of viral titers between regenerated SARS-CoV-2 (WT-Rg) and WT-PC SARS-CoV-2 (Fig. 1D). Collectively, these observations demonstrate successful recovery of SARS-CoV-2 virus using our CPEC method.
Versatile primer design for point mutagenesis and the recovery of SARS-CoV-2 variants. To further optimize the utility of the three-subclone/CPEC method, we demonstrated the ease of applying mutagenesis using a specific scheme for designing primers for both site-directed mutagenesis (SDM) ( Fig. 2A) and in-fusion cloning methods for multiple mutagenesis (Fig. 3A). For this, a set of forward primers was designed containing 15 nt bases from the 59 end that bind complementarily to the 15 bases from the 59 end of the reverse primer and extend to 20 to 50 nt bases conferring the mutation of interest  Fig. 2A, dotted square). The SDM PCR product amplified using these primer sets was restriction digestion with DpnI and then used directly together with the respective CP-1 to -6 primer sets to amplify the mutated fragment for CPEC, obviating the need for bacterial transformation of the SDM product ( Fig. 2A and B). Compared to traditional cloning methods, our CPEC mutagenesis method optimizes the process and reduces the time spent by up to 3 days (Fig. 2B). Using the resulting clones, we directly applied mutagenesis to generate mutants containing single substitutions (K417N, L452R, E484K, N501Y, D614G, P681H, and P681R) in the spike protein. We also introduced adjacent double substitutions (E484K1N501Y), small deletions (D69-70 and D157-158), and a small insertion (ins 38F) in NSP6 using a single primer set ( Table 2). To verify the success of mutagenesis, we assessed the whole genome of all clones by Sanger sequencing using the customdesigned primers (Table S2). In addition, to verify the mutagenesis efficiency introduced by our specific primer design strategy, we confirmed positive mutant subclones by sequencing 10 colonies of each mutant subclone. This analysis revealed a mutagenesis efficiency of 30% to 100% (Table 3). To further determine the phenotypic characteristics of generated SARS-CoV-2 variants, we measured their replicative properties in vitro and compared them with the genetically reverse-generated WT-Rg. Each of the rescued recombinant SARS-CoV-2 variants exhibited cytopathic effect (CPE) and homogeneity of plaque size, as shown in Fig. 2C and Table 4. Notably, however, the plaque size of the combined E484K1N501Y variant was smaller than that of other variants. Most rescued mutagenic viruses possessed titers in the range of 2.0 Â 10 4 to 4.9 Â 10 5 PFU/mL compared with 7.2 Â 10 5 PFU/mL and  Table 4). The replication kinetics of single mutants revealed that a peak was reached within 48 h in Vero E6 cells for almost all tested mutants (Fig. 2C). Notably, all mutants showed an infectious titer similar to that of control WT-Rg (log 10 TCID 50 /mL = 5.134 6 0.236) after 48 h. Nonetheless, the efficiency and rapidness of three-subclone/CPEC combined with the introduced primer design scheme support the utility of this approach in performing SARS-CoV-2 point mutagenesis. The genetic correspondence of single-passaged WT recombinant viruses to their infectious clones was verified by deep sequencing, which showed no unwanted mutations ( Table 4). Versatility of SDM primers for the simultaneous editing of multiple mutations and generation of corresponding variants. To edit DNA plasmids containing multiple mutations that are relatively distinct from each other, we further determined whether the primers designed for point mutagenesis were versatile enough for use in simultaneous introduction of multiple point mutations (Fig. 3A). For substitution of N501Y and D614G in the spike-encoding gene, we applied the same scheme for primer designs. However, during generation of the insert containing the targeted mutation, the forward primers cross-pair with the reverse primers of the adjacent primer set; specifically, the forward primer of the N501Y cross-pairs with the reverse primer of D614G, whereas the forward primer used to generate the insert containing D614G cross-pairs with the reverse primer of N501Y. This cross-pairing primer amplification scheme generated fragments that shared 15 homologous overlapping bases on their ends. Each amplified PCR product contains overlapping ends homologous to those of the vector or adjacent fragments that serve to facilitate subsequent in-fusion reactions. Once the subcloned fragment is successfully transformed with multiple mutations, the plasmid can be assembled and cloned to produce a full-length genome by CPEC. Applying this same design scheme, we generated SARS-CoV-2 variants with a triple E484K/N501Y/D614G mutation. The genetic correspondence of each single-passaged recombinant mutant virus to its infectious clone was also verified by deep sequencing. Among the 16 recombinant viruses, 50%, including K417N-Rg, D614G-Rg, P681R-Rg, E484K1N501Y-Rg, D69-70-Rg, D157-158-Rg, DORF7A-Rg, and GFP-Rg, acquired one to three unwanted nucleotide mutations ( Table 4). Most of these mutations were synonymous, but in four cases-D614G-Rg, P681R-Rg, D157-158-Rg, and GFP-Rg-the virus acquired one amino acid substitution. The proportion of most mutations was in the 27% to 54% range, although two synonymous mutations acquired in 484K1N501Y-Rg and DORF7A-Rg viruses accounted for more than 97% of rescued and single-passaged isolates (Table 4). No nucleotide mutations were found in the in vitro-transcribed viral RNA genome before cell transfection (data not shown). Overall, more than 70% of mutations showed a proportion less than 50%, and some of the amino acid changes (i.e., R682W and L179F in the spike protein) are reported to be adaptive mutations in a cell or human infection (27,28). Therefore, these mutations were potentially acquired during virus propagation in cells.
Observations of successfully regenerated mutants showed that plaque sizes of recombinants containing double or triple mutations in the spike were smaller than those of WT-Rg ( Fig. 3B; Table 4). However, the growth kinetics of the E484K/N501Y/D614G triple mutant in Vero E6 cells (mean peak log 10 TCID 50 /mL value = 5.468 6 0.236) revealed that viral titers of this mutant were higher than those of the WT-Rg SARS-CoV-2 virus (mean peak log 10 TCID 50 /mL value = 5.134 6 0.236). Overall, the use of cross-pairing primers and a two-step amplification scheme for subclones reduces the complexity of the method for generating multiple point mutations in SARS-CoV-2 while increasing time efficiency.
Mutagenesis for introducing a large insertion and deletion in the SARS-CoV-2 genome. To determine whether our primer design scheme can also be applied to largescale editing of the SARS-CoV-2 genome, we recovered a gene-deleted SARS-CoV-2 Rg-virus after removal of the ORF7A gene (27,410 to 27,710 nt), a process that eliminated a total of 244 bases or 81 amino acids (Fig. 4A). For truncation of more than 30 nt, a two-step amplification is recommended, with an initial step in which a fragment containing the deletion is generated by PCR in a subclone followed by CPEC for whole-genome cloning. We successfully rescued a SARS-CoV-2 variant containing a large deletion (DORF7A) and noted that the recombinant virus exhibited a lower viral titer and reduced plaque formation compared with WT-Rg virus ( Fig. 4B; Table 4).
We also explored whether the same primer design scheme could be used to generate a large insertion in the SARS-CoV-2 genome. In this case, the ORF7A gene was deleted and replaced with Turbo GFP, a total of 694 nucleotides (nt), using two designed primer sets, one for the GFP insert and one for the vector component (Table 4). To subclone GFP-mutant F3, we in-fused and transformed the two amplified fragments (Fig. 4A) and used CPEC to process the GFP-mutant F3 with the other fragments. GFP-expressing SARS-CoV-2 formed smaller plaques relative to those of WT-Rg virus (Fig. 4B). Notably, cells containing GFP were readily observed from 48 to 72 h postinfection, indicating the successful recovery of infectious virus with GFP expression (Fig. 4C). Although the replication kinetics of the recovered GFP-expressing viruses were slightly lower than those of the WT-Rg strain (Fig. 4D), the SARS-CoV-2 clone expressing GFP will aid in future antiviral screening and quantification assays.

DISCUSSION
The public availability of complete genomes has enabled researchers to monitor changes in the viral genome of SARS-CoV-2 since the first emergence of its variants in 2020. Several mutations that have occurred in the genetic composition of SARS-CoV-2 have enhanced the fitness and transmissibility of the virus, making it more difficult to control (29)(30)(31). Principle concerns have been the vast array of point mutations in the S protein, which primarily aids in the entry of the virus into the host cell (11,13,25,32). The presence of these mutations may reduce the efficacy of available vaccines as they weaken binding of the antibody (32). Because the combination of these point mutations has given rise to current SARS-CoV-2 VOCs and VOIs, which in turn impact public health, there is an urgent need for the development of methods for determining the impacts of these mutations on viral infectivity and antigenicity.
Among RNA viruses, coronaviruses have the largest genome size, which reaches almost 30 kb (6). Hence, generating an infectious clone has always been a challenge. Compared to a previously reported reverse genetics method (14,16,19,30), the method utilized in the current study employed fewer fragments (three versus seven to 10) to facilitate various mutagenesis strategies in subclones and mitigate errors by reducing CPEC PCR cycles (24). Although previously utilized yeast cloning and other traditional cloning methods show proven cloning efficiency and are unhampered by the large genome size of SARS-CoV-2, they require considerable time for successful completion of the assembly process. Hence, in this study, we demonstrated how the application of CPEC simplified the process and reduced the cloning time required for the most available method for recovering SARS-CoV-2. The advantages of the CPEC method rely on the overlapping strands of the fragments and cloning vectors during the amplification and the quality of the PCR polymerase used. Here, we specifically designed primers that generate fragments and a vector with double-overlapping homologous termini. We also used GXL DNA polymerase, which allows successful amplification of products more than 40 kb in length, like SARS-CoV-2 genome, together with a 10-kb vector. Compared with transformation-associated recombination (TAR) cloning in yeast, CPEC can substantially reduce the number of steps as well as the time required for assembly. Thus, compared with previous reverse genetics systems developed, this customized CPEC approach shows improved efficiency, speed, and simplicity in regenerating large-size genome viruses like SARS-CoV-2.
We also demonstrate an easier and uncomplicated guide for performing mutagenesis in the SARS-CoV-2 genome for amplification of fragments. The versatility of our designed primers has expanded the utility of each primer, reducing the cost and time needed for redesigning additional primers when performing multiple-substitution mutagenesis. The primer design scheme for mutagenesis devised in this study can be readily applied for use in SDM and in-fusion cloning methods, facilitating simple and broad mutagenesis, as demonstrated here.
Notably, VOCs that have arisen contain the following key mutations in the spike protein: N501Y and D614G, common among all; E484K and L452R, common among Alpha VOC (B.1.1.7); K417N for Delta (B.1.617.2); and P681H for Gamma (P.1) (7,12). Additional contributions of VOCs include observed deletions in the spike (D69/70) (25) and a 382-nt deletion in ORF8 in an isolate from a patient in Taiwan (33). In the current study, we found that most single point mutations minimally impacted the growth properties of SARS-CoV-2, whereas a triple mutant significantly improved in vitro viral growth compared with that of WT-Rg. The results of our phenotypic assays potentially support the interpretation that combinations of each of these mutations in the spike have a greater impact on the replicability of the virus, and that the combination of these mutagenic events has allowed SARS-CoV-2 to evolve detrimental changes in pathogenicity and epidemiology (7). Recombinants generated using the approach described here can be utilized in studying emerging variants, especially studies designed to assess the impact of these variants on antibody therapies. In addition, SARS-CoV-2 viruses with large deletions of ORF7A and insertion of GFP, successfully and simply generated via this method, will contribute to the development of live-attenuated vaccines or strategies for screening antivirals. Overall, our study enables a simple, uncomplicated approach for performing mutagenesis and recovering SARS-CoV-2 variants for further study and diagnostic evaluation, providing a rapid preparedness capability in the face of continuously emerging VOCs.

MATERIALS AND METHODS
Cells and virus. Vero E6 (ATCC, CRL1586) and Baby Hamster Kidney-21 (BHK-21) cell lines utilized in this study were obtained from Korean Cell Line Bank (KCLB; Seoul, Republic of Korea). Vero E6 cells were maintained in DMEM (Gibco-Invitrogen, Carlsbad, CA, USA), whereas BHK-21 cells were maintained in minimum essential medium-alpha (MEM-a; Gibco-Invitrogen), both supplemented with 5% to 10% (vol/vol) heatinactivated FBS and 1% antibiotics (Gibco-Invitrogen) at 37°C and 5% CO 2 . The SARS-CoV-2 virus resource (NCCP43326) for this study was provided by the National Culture Collection for Pathogens, Korea National Institute of Health, and was propagated at 37°C and 5% CO 2 in Vero E6 cells in 2% FBS containing DMEM. After 72 to 96 h, viral supernatants were harvested and processed for storage at 280°C before use.
Primers and viral cDNA synthesis. The KDCA (BetaCoV/Korea/KCDC03/2020 [GISAID id: EPI-ISL: 407193]) strain, from the 59-untranslated region (UTR) to 3'-UTR (29,870 bp), was used as a reference strain. For the design of primer sets (Table 1), the KDCA sequence together with all available SARS-CoV-2 sequences in the GenBank database from January 2020 to November 2021 were aligned using CLC Genomics Workbench 7.0 (Qiagen, Hilden, Germany). SARS-CoV-2 genomic RNA was extracted using a RNeasy minikit (Qiagen, Hilden, Germany) following the manufacturer's instructions. Viral cDNA was synthesized using reverse transcription (RT) primers (RT-1, RT-2, and RT-3) ( Table 1) at 65°C for 5 min. After a 2-min incubation on ice, Superscript III Reverse Transcriptase (Invitrogen, Waltham, MA, USA) was added and the elongation process was allowed to proceed at 50°C for 90 min, followed by inactivation of the RT enzyme at 70°C for 15 min.
Subcloning of the SARS-CoV-2 genome. The SARS-CoV-2 genome was subcloned by first dividing it into three fragments corresponding to approximately 8.7 to 11.8 kb regions (Fig. 1A) by digesting the vector with AgeI, which serves to prevent back-transformation, followed by amplification of fragments using Phusion High Fidelity DNA polymerase (New England Biolabs, Ipswich, MA, USA) with specifically designed primer sets: Fragment 1, SP-1 and SP-2; Fragment 2, SP-3 and SP-4; and Fragment 3, SP-5 and SP-6; the final subcloning vector was amplified using primers SP-7 and SP-8 (Table 1; Fig. S2A). All fragments overlapped each other by at least 30 bp, with F1 and F2 overlapping by 450 bp. After amplification of the vector, back-transformation was further prevented by incubating the PCR product with DpnI to digest the methylated plasmids used as template. PCR was performed using an initial denaturation step of 98°C for 30 s, followed by 34 cycles of denaturation at 98°C for 10 s, annealing at 68 to 69°C for 30 s, and extension at 72°C for 5 min and 30 s for inserts and 2 min for the vector, with a final extension step of 72°C for 10 min. Electrophoretic analyses of PCR products for each fragment and vector showed bands of 8.7 to 11.8 kb for fragments F1, F2, and F3, and 3.16 kb for the vector. The amplified PCR products were purified using a NucleoSpin Gel and PCR Clean-up kit (Macherey-Nagel, Hoerdt, France), and then each fragment was cloned into the subcloning vector using an In-Fusion HD cloning kit (TaKaRa Bio Inc., Kusatsu, Shiga, Japan), following the manufacturer's protocol. The in-fused clone samples were transformed into OneShot TOP10 chemically competent cells (Invitrogen) using a heat shock protocol and incubated on ampicillin-containing Luria Bertani (LB) plates at 25°C for 48 h. Positive clones were selected by performing colony PCR with enzyme 2Â TOPsimple DyeMIXtenuto (Enzynomics Co., Yuseong-gu, Daejeon, Republic of Korea), which detects junctions of each fragment using custom-designed primers ( Fig. S2B; Table S1). The PCR conditions used were an initial denaturation step of 95°C for 30 s, followed by 34 cycles of denaturation at 95°C for 10 s, annealing at 58°C for 90 s, and extension at 72°C for 90 s, with a final extension step of 72°C for 2 min. Also, two silent mutations, A7486T and T7489A, were introduced in F1 subclone to confirm whether rescued viruses from transfected RNA or not.
CPEC using vector and three SARS-CoV-2 genome fragments. Each subclone was amplified individually using its respective CPEC primers (Primer code: CP-1 to CP-6) ( Table 1) under the following conditions: initial denaturation step of 98°C at 30 s, followed by 34 cycles of denaturation at 98°C for 10 s, annealing at 69°C for 30 s, and extension at 72°C for 5 min and 30 s, with a final extension step of 72°C for 10 min. The vector, pYES1L, was then also amplified under the same conditions using the designed primer set CP-7/CP-8 (see Table 1). All amplified products were treated with DpnI (New England Biolabs) for up to 8 h, and purified using a Macherey-Nagel NucleoSpin Gel and PCR Clean-up kit (Hoerdt). Inserts (F1, F2, and F3) and pYES1L vector were assembled through the CPEC method using 0.1 pmol of each fragment. CPEC processing employed PrimeSTAR GXL DNA polymerase (TaKaRa Bio Inc.) under the following conditions: an initial denaturation step of 98°C for 2 min, followed by 15 cycles of denaturation at 98°C for 10 s, annealing at 55°C for 15 s, and extension at 68°C for 15 min, with a final extension step of 68°C for 15 min. The CPEC product was electroporated into electrocompetent cells (Invitrogen) following the manufacturer's instructions and incubated on an LB agar plate supplemented with a 0.01% (wt/ vol) spectinomycin for 24 to 36 h at 37°C. For confirmation of recombinant clones, colony PCR targeting junctions of each fragment was performed using specifically designed CPEC junction primers ( Fig. S2C; Table S1) with 2Â TOPsimple DyeMIX Tenuto (Enzynomics Co.) and the following PCR conditions: an initial denaturation step at 95°C for 2 min, followed by 34 cycles of 95°C for 30 s, 61°C for 30 s, and 72°C for 1 min and 30 s, with a final extension step of 72°C for 5 min.
SARS-CoV-2 mutagenesis of subclones and direct construction of a full-length infectious clone variant. The SARS-CoV-2 genome was manipulated by applying mutagenesis strategies to the constructed subclone plasmids. Noted variations in the SARS-CoV-2 genome were selected and introduced into the corresponding target fragments using designed, versatile primer sets. Overlapping nucleotides were positioned 15 nt from the 59 end in forward primers, which contain a further 20 to 50 nt extension that confers the mutation of interest. The reverse primer contained 15 bases of the paired forward primer's 59 end in reverse complementary orientation and extended a further 20 to 30 bases ( Fig. 2A; Table 2). In this study, we designed a versatile primer scheme for generating single or multiple point mutations, a large insertion, and a large truncation in spike and ORF7A genes of SARS-CoV-2, as shown in Table 2. To introduce point mutations or adjacent double mutations, we utilized SDM PCR of target subclones employing PrimeSTAR GXL DNA polymerase (TaKaRa Bio Inc.). The cycling conditions for PCR mutagenesis were as follows: an initial denaturation step of 98°C for 2 min, followed by 15 cycles of denaturation at 98°C for 10 s, annealing at 55 to 60°C for 15 s, and extension at 68°C for 15 min, with a final extension step of 68°C for 15 min.
The resulting PCR products produced through mutagenesis can be generated as subclones by transformation or undergo PCR and CPEC for use in constructing a full-length clone. In the former case, PCR products can be used for transformation after DpnI treatment to generate subclones possessing target mutation(s); in the latter, the site-directed mutagenesis PCR product can be used as a template to directly generate a fragment for CPEC by using the appropriate primer sets (CP-1 to CP-6) ( Table 1). The cycling conditions were as follows: an initial denaturation step of 98°C for 2 min, followed by 15 cycles of denaturation at 98°C for 10 s, annealing at 55 to 60°C for 15 s, and extension at 68°C for 15 min, with a final extension step of 68°C for 15 min. The composition of the PCR mixture followed the standard protocol of the manufacturer.
In vitro transcription. The SARS-CoV-2 whole genome was first separated from clones using the restriction enzyme ApaI and then purified using a manual phenol-precipitation method (34). The linearized plasmid was processed for in vitro transcription using a mMessage mMachine T7 Transcription kit (Thermo Fisher Scientific, Waltham, MA, USA) following the manufacturer's protocol, with slight modifications. Briefly, a 40-mL reaction containing 6 mL GTP (GTP:cap ratio, 1:1) and 12 mg of DNA template was incubated for 8 to 12 h at 32°C, after which template DNA was removed by adding 2 mL of Turbo DNase and incubating for 15 min at 37°C. RNA was subsequently extracted by lithium chloride precipitation. The stability of viral RNA during replication in transfected cells was improved by also transcribing the N gene of SARS-CoV-2 in vitro and co-transfecting it into the cell. The N gene DNA template was prepared by PCR as previously described (14). Transcription of the SARS-CoV-2 N gene was performed by incubating with 1 mL of GTP (GTP: cap ratio, 1:1) in a 20-mL reaction at 37°C for 3 h.
Virus recovery. SARS-CoV-2 RNA transcripts were transfected into BHK-21 cells by electroporation as previously described (14). Briefly, 10 mg of whole-genome transcripts, 2 mg of N gene transcript, and 8 Â 10 6 BHK cells in 800 mL of Opti-MEM were mixed in a 4-mm gap cuvette and incubated for 10 min on ice. After incubation, cells were electroporated (voltage, 850V; pulse, 5 times with 0.01-ms pulse length and 1-s pulse interval) using a BTX Gemini X2 twin-wave electroporator and Opti-MEM as an electroporation reagent. After a 5-min recovery, electroporated cells were seeded on a monolayer of Vero E6 cells in a T-75 flask and incubated with 10 mL of 5% DMEM at 37°C and 5% CO 2 . After 24 h, the medium was replaced with 2% DMEM and plates were further incubated for a minimum of 72 h or until ;90% of cells exhibited virus-mediated cytopathic effects. The rescued virus-Passage 0 (P0)-was inoculated into another T-75 flask containing Vero E6 cells and passaged. After incubating for 72 to 96 h at 37°C and 5% CO 2 , cell supernatants (P1) were harvested. After virus recovery, viral RNA was sequenced to confirm it has silent mutations (A7486T and T7489A) by the Sanger sequencing method for ensuring the viruses were rescued from transfected RNA (Fig. S3). To confirm whether viruses have undesired mutations, viral RNAs were extracted from single-passaged viruses using RNeasy minikit (Qiagen, Hilden, Germany), sequenced by deep sequencing method (DNALINK, Republic of Korea) and analyzed by CLC genomics workbench (Qiagen, Hilden, Germany).
Virus quantification and plaque formation. Plaque formation and the concentration of recombinant viral stocks relative to their parental viruses were determined by preparing Vero E6 cells in 6-well plates and infecting with 1 mL of 10-fold serially diluted recombinant/parental viruses. Plates were incubated at 37°C and 5% CO 2 for 1 h to allow for virus-cell attachment. During this incubation period, a 0.7% DMEM-agarose overlay medium was prepared by mixing 1.4% agarose and 2Â DMEM media. After incubating, the viral inoculum in each well was removed and replaced with a 3-mL volume of overlay medium. The plate with overlay was then cooled at 4°C for 10 min before incubating in a 37°C and 5% CO 2 incubator. At 96 h postinfection, the cells were fixed with 10% formaldehyde and stained with 1% (wt/vol) crystal violet. Wells containing 20 to 100 plaques were selected and analyzed using ImageJ software (35) (National Institute of Health, Bethesda, Maryland), and the titer of virus stocks, in PFU, was calculated.
In vitro growth kinetics. For comparison of in vitro viral growth kinetics, cells were infected with wild-type or recombinant SARS-CoV-2, diluted to a multiplicity of infection (MOI) of 0.0001. All plates were incubated at 37°C and 5% CO 2 for 1 h. After the viral adsorption period, media were replaced with DMEM containing 2% (vol/vol) FBS. Each cell culture supernatant was harvested at 24, 48, 72, and 96 h postinfection, and viral titers at each time point were measured as TCID 50 (median tissue culture infectious dose), calculated based on the Reed-Muench method.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.4 MB.