Construction of a Severe Acute Respiratory Syndrome Coronavirus Infectious cDNA Clone and a Replicon To Study Coronavirus RNA Synthesis

In this article, we use the BAC approach to assemble the SARS-CoV full-length cDNA under the control of the cytomegalovirus (CMV) immediate-early promoter to allow the expression of the viral RNA in the nucleus by cellular RNA polymerase II. This approach was used in our laboratory for the construction of the first CoV infectious cDNA clone (3) and, more recently, for the successful construction of other coronavirus (26) and arterivirus (7) cDNA clones. This system permits the stable maintenance in bacteria of large DNA fragments from a variety of complex genomic sources (1, 24). Furthermore, the manipulation of the cDNA clone is similar to that of a conventional plasmid and directly allows the recovery of infectious virus from the cDNA clone without the need for in vitro ligation and transcription steps.

The BAC clone carrying an infectious genome of the SARS-CoV Urbani strain was generated in three steps. The first was selection of appropriate restriction sites in the viral genome (Fig. 1A). The second was construction of the intermediate plasmid pBAC-SARS-CoV 5′-3′ as the backbone for assembling the full-length cDNA clone (Fig. 1B). This plasmid includes the first 681 nucleotides (nt) of the genome under the control of the CMV promoter and the last 975 nt of the genome, followed by a 25-nt synthetic poly(A), the hepatitis delta virus ribozyme, and the bovine growth hormone termination and polyadenylation sequences to produce synthetic RNAs bearing authentic 5′ and 3′ ends of the viral genome. In addition, a multiple-cloning site containing the restriction sites ClaI, MluI, PmeI, BamHI, and NheI, selected in the first step, was cloned between the viral sequences to allow the assembly of the full-length clone. After transfection of baby hamster kidney (BHK) cells with the intermediate plasmid, the sequence of the 5′ and 3′ ends of the synthetic RNA generated was confirmed by 5′ and 3′ rapid amplification of cDNA ends analysis (data not shown). Finally, the third step was assembly of the full-length cDNA clone (pBAC-SARS-CoV^FL) by sequential cloning of five overlapping cDNA fragments into the plasmid pBAC-SARS-CoV 5′-3′ (Fig. 1C). These cDNAs were generated by standard reverse transcriptase PCR (RT-PCR) with specific oligonucleotides using total RNA from cells infected with the SARS-CoV Urbani strain, kindly provided by the Centers for Diseases Control and Prevention (Atlanta, Ga.), as the template. The cDNA clone sequence was identical to the reported Urbani sequence (21), with the exception of two silent point mutations at positions 10338 (C to T) and 11163 (T to A) that were introduced as genetic markers to distinguish between the virus recovered from the cDNA and the wild-type virus. The assembled SARS-CoV infectious cDNA clone was fully stable during its propagation in E. coli DH10B cells for more than 200 generations, considerably facilitating the genetic manipulation of the viral genome (data not shown). The detailed cloning strategy, plasmid maps, and sequences are available upon request.

To recover infectious virus, BHK cells were grown to 95% confluence in a 25-cm² flask and transfected with the cDNA clone by using Lipofectamine 2000 (Invitrogen) according to the manufacturer's specifications. At 6 h posttransfection (hpt), cells were trypsinized, plated over a confluent monolayer of VeroE6 cells grown in a 25-cm² flask, and incubated at 37°C for 72 h. Virus titers quickly increased to around 2 × 10⁶ PFU/ml at 72 hpt (Fig. 2A). After two passages in VeroE6 cells, the recovered virus was cloned by three rounds of plaque purification, and the genotypic and phenotypic properties were determined. The rescued virus (rSARS-CoV) induced a clear cytopathic effect in VeroE6 cells, and its identity was confirmed by indirect immunofluorescence using SARS-CoV-specific antibodies (Fig. 2B). Furthermore, it conserved the genetic markers introduced and was identical to the parental virus in terms of plaque morphology, growth kinetics, and mRNA and protein patterns (data not shown).

All work involving infectious SARS-CoV was performed in a biosafety level 3 laboratory at the National Health Institute Carlos III (Madrid, Spain), following the guidelines of the European Commission and the National Institutes of Health. Personnel were double-gloved and wore powered air-purifying respirators (HEPA AirMate; 3M, Saint Paul, MN) to provide a positive-pressure environment within the hoods.

The availability of a SARS-CoV replicon provides an important tool for the study of fundamental viral processes and for the selection of antiviral drugs against this human pathogen, without the need for growing infectious virus.

To engineer the SARS-CoV-derived replicon, the BAC system was used because it presents several advantages over other approaches, such as easy manipulation, unlimited production of the cDNA, high efficiency of transfection, and intracellular expression of the replicon RNA. The generation of the SARS-CoV replicon was based on the full-length cDNA clone described above and on published data showing that the N protein is required for efficient CoV replication (2, 22). Accordingly, a cDNA containing the untranslated 5′ and 3′ ends of the Urbani strain genome, the replicase gene, and the N gene preceded by the upstream 143 nt including its natural transcription-regulating sequence was cloned as a BAC under the control of the CMV promoter by following the same approach described for the generation of the full-length cDNA clone (Fig. 3A). In addition, a multiple-cloning site containing the unique restriction sites PacI, AscI, and BamHI was cloned downstream of the replicase gene to allow the cloning and expression of heterologous genes.

To study SARS-CoV replicon activity, the synthesis of N gene mRNA was analyzed in human 293T and BHK cells by RT-PCR (Fig. 3B). Cells were grown to 95% confluence on 35-mm-diameter plates and transfected with 4 μg of SARS-CoV replicon by using 12 μg of Lipofectamine 2000 (Invitrogen) according to the manufacturer's specifications. Total intracellular RNA was extracted at 24 hpt by using the RNeasy mini kit (QIAGEN) and was used as a template for RT-PCR analysis of N gene mRNA synthesis. Amplifications were performed with the reverse primer URB-28630RS (5′-TGCTTCCCTCTGCGTAGAAGCC-3′), complementary to nucleotides 511 to 532 of the N gene, and the forward primer URB-29VS (5′-GCCAACCAACCTCGATCTCTTG-3′), spanning nucleotides 29 to 50 of the Urbani leader sequence. High levels of transcript were detected in both 293T and BHK cells transfected with the replicon (Fig. 3B), indicating that the SARS-CoV replicon was functional, at least at the transcriptional level. To confirm whether the SARS-CoV replicon was also functional at the replication level, the synthesis of genomic positive-strand RNA was analyzed on DNase I-treated samples by real-time RT-PCR using the reverse primer URB-1995RS (5′-ATGGCGTCGACAAGACGTAAT-3′), complementary to nucleotides 1995 to 2015 of the genome, and the forward primer URB-1931VS (5′-ACCACTCAATTCCTGATTTGCA-3′), spanning nucleotides 1931 to 1952 of the genome. In our system, the amount of replicon RNA in the cytoplasm of transfected cells is the sum of the RNA transcribed from the CMV promoter in the nucleus plus the amount of RNA self amplified by the viral replicase in the cytoplasm. To determine the amplification exclusively due to the viral replicase, the synthesis of genomic RNA was analyzed in cells transfected with either the wild-type replicon or a nonreplicative construct in which most of the polymerase domain has been deleted. An increase in RNA synthesis of at least 10-fold over the RNA background level due to the CMV promoter was detected in cells transfected with the wild-type replicon, indicating RNA self-amplification of the SARS-CoV replicon (data not shown).

Similar to the full-length cDNA clone, the replicon was fully stable during its propagation in bacteria for more than 200 generations, as determined by restriction endonuclease analysis. To confirm this result, the replicon activities of cDNAs purified after amplification in bacteria during 20, 100, and 200 generations were analyzed in 293T cells following the procedure described above. No differences in the replicon activities after 200 generations were detected (Fig. 3C), demonstrating that the generated replicon was fully stable in bacteria.

These RNA-processing enzymes are encoded at the 3′ end of ORF 1b, downstream of the helicase domain (25). The identification of a stable processing intermediate comprising these three domains in infectious bronchitis virus-infected cells (31) suggests that their activities might cooperate in the same metabolic pathway.

To study the role of the enzymes ExoN, NendoU, and 2′-O-MT in CoV RNA synthesis, in-frame single-deletion mutants in which the conserved domain of each enzyme (25) was deleted and a mutant replicon lacking the three conserved enzymes were generated by using the SARS-CoV replicon (Fig. 4A). For the generation of these mutants, PmeI-PacI fragments with the desired in-frame-deletions were generated by PCR mutagenesis and cloned into the SARS-CoV replicon, replacing the wild-type PmeI-PacI fragment. The functionality of these constructs was analyzed in comparison with that of the wild-type replicon in 293T and BHK cells by RT-PCR as described above. Interestingly, high levels of N gene mRNA were detected only for cells transfected with the parental replicon (Fig. 4B), suggesting that the RNA-processing enzymes ExoN, NendoU, and 2′-O-MT are essential for SARS-CoV RNA synthesis. Formally, it is not possible to exclude the possibility of a lethal effect of these deletions at the level of RNA structure or pp1ab processing, even though their N- and C-terminal borders were relatively distant from the proteolytic cleavage sites. To overcome this potential limitation, three point mutants, in which putative catalytic and highly conserved residues of ExoN (Asp-6145), NendoU (His-6678), and 2′-O-MT (Asp-6905) (14, 19, 25) were replaced by Ala (D6145A, H6678A, and D6905A, respectively), were generated. In addition, since the Ser-6744 residue has been described to be nonessential for NendoU activity (14), an additional mutant containing an Ala instead of a Ser at this position (S6744A) was generated as a control to confirm that single amino acid substitutions could be tolerated in this domain of the polyprotein (Fig. 4C). The activity of these replicons was compared to that of the deletion mutants and the wild-type replicon. To this end, a quantitative analysis of N gene mRNA expression was performed by real-time RT-PCR using the reverse primer URB-28163RS (5′-TGGGTCCACCAAATGTAATGC-3′), complementary to nucleotides 44 to 64 of the N gene, and the forward primer URB-27VS (5′-AAGCCAACCAACCTCGATCTC-3′), spanning nucleotides 27 to 47 of the Urbani leader sequence. As expected, high levels of N gene mRNA were detected in cells transfected with the wild-type replicon or with the control S6744A mutant (Fig. 4D). However, a reduction of more than 90% was observed in the case of both the deletion and point mutants, confirming that the effect on CoV RNA synthesis is most probably due to the abrogation of ExoN, NendoU, and 2′-O-MT activities and not to major changes affecting the RNA structure or polyprotein processing (Fig. 4D). In addition, a 10-fold reduction in the amount of genomic positive-strand RNA was detected for all mutant replicons (data not shown), suggesting that ExoN, NendoU, and 2′-O-MT are essential for efficient CoV replication and transcription. Although it was previously reported that NendoU and ExoN were essential for CoV RNA synthesis (14, 19), this is the first evidence that 2′-O-MT activity is also essential. Further genetic and biochemical studies will be required to address the specific role of these enzymes in CoV transcription and replication.

Within the Nidovirales order, these three RNA-processing enzymes are highly conserved in the families Coronaviridae (coronavirus and torovirus) and Roniviridae. It has been postulated that these enzymes are involved in RNA processing, possibly in the synthesis of the 5′- and 3′-coterminal subgenomic RNAs (11, 14, 25, 36). Interestingly, only the NendoU domain is conserved in the family Arteviridae (20, 25), suggesting that the RNA synthesis mechanisms of coronavirus and arterivirus might present some differences at the molecular level or that the extra RNA-processing activities (ExoN and 2′-O-MT) present in coronaviruses and roniviruses may be required to synthesize and maintain the large RNA genome (∼30 kb) of these virus families. In this context, the ExoN domain, which is predicted to contain a 3′→5′ exonuclease activity, may have been acquired to operate in proof-reading mechanisms to improve the fidelity of the RNA-dependent RNA polymerase.

The SARS-CoV replicon assembled as a BAC provides a useful tool for basic studies on CoV transcription and replication, for the safe screening of new antiviral drugs, and for the development of SARS-CoV vaccines.

We thank P. Pérez-Breña from the Human Health Department, Instituto de Salud Carlos III, for facilitating the development of the project and D. Dorado and M. Gonzalez for technical assistance.

This work was supported by grants from the Ministerio de Educación y Ciencia of Spain (BIO2004-00636), the European Community (Frame VI, Projects DISSECT SP22-CT-2004-511060 and RIVIGENE SSPE-CT-2005-022639), and Fort Dodge Veterinaria.