Porcine reproductive and respiratory syndrome (PRRS) was first recognized in 1987 in North America (
21) and shortly thereafter in Europe (
35). It has since become one of the most common and economically significant infectious diseases in the swine industry worldwide (
1,
15). It is characterized by mild to severe reproductive failure in sows and gilts and respiratory problems in piglets (
9,
10,
17). The PRRS virus (PRRSV) was first isolated almost simultaneously in Europe and North America; the strains are designated Lelystad (
53) and VR-2332 (
6,
10), respectively. Although these strains induce phenotypically indistinguishable disease symptoms (
19), they are genetically (
2,
25,
32) and serologically (
33,
54,
58) distinct. They currently constitute the two distinct genotypes of known PRRSV species.
PRRSV belongs to the family
Arteriviridae in the order
Nidovirales together with equine arteritis virus (EAV), simian hemorrhagic fever virus, and the lactate dehydrogenase-elevating virus of mice (
8,
45). Like other arteriviruses, PRRSV is a small-envelope virus with a positive-sense, single-stranded RNA genome of ∼15 kb in length. The genome has a cap structure at its 5′ end and a poly(A) tail at its 3′ end. The genome contains at least nine open reading frames (ORFs) flanked by 5′ and 3′ noncoding regions (NCRs) (
13,
29,
45,
59). Two overlapping ORFs, ORF1a and ORF1b, are expressed from the genomic RNA. These two ORFs are predicted to be processed into 13 mature nonstructural proteins that are believed to be involved in viral replication (
5,
43,
44,
45,
49,
57). ORF2a, ORF2b, and ORF3 to ORF7 are translated from the 5′ ends of a coterminal nested set of subgenomic mRNAs. The small ORF2b is completely embedded within the larger ORF2a (
59). These ORFs encode the viral structural proteins (
14,
27,
29,
59).
To analyze positive-sense RNA viruses such as PRRSV at molecular and genetic levels, a number of reverse genetics systems that allow the determination of the functions of the genes and gene products of these RNA viruses through their manipulation and genetic analysis have been developed (
7). With regard to arteriviruses, two “RNA-launched” reverse genetics systems for EAV have been independently developed by two research groups (
16,
50). Although both functional EAV cDNAs were constructed using the same virus isolate, different bacteriophage promoters, either the SP6 or T7 RNA polymerase promoter, were placed immediately upstream of the viral genome for runoff transcription in vitro. The first infectious EAV cDNA was constructed in a high-copy-number plasmid, pUC18, and did not show any genetic instability during its construction and propagation in
Escherichia coli, in contrast to the infectious cDNAs of several other viruses (
50).
For PRRSV, the first RNA-launched reverse genetics system was constructed for the European Lelystad strain by assembling its full-length cDNA under the T7 promoter in the low-copy-number plasmid pOK12 (
28). Identical strategies were also previously used in the construction of an infectious cDNA for the North American strain VR-2332 (
34). Recently, two additional T7 promoter-driven RNA-launched reverse genetics systems were developed independently for the highly virulent American isolate NVSL 97-7895 (
47) and for the highly virulent “atypical” North American isolate P129 (
23) by using the low-copy-number plasmid pBR322 and the high-copy-number plasmid pCR2.1, respectively. An additional “DNA-launched” system has also been reported for the P129 isolate and employed full-length cDNA under the human cytomegalovirus immediate early promoter (
23).
The present report describes the construction of not only a full-length infectious PRRSV cDNA in a bacterial artificial chromosome (BAC) designed to produce synthetic RNAs bearing authentic 5′ and 3′ ends of the viral genome but also a panel of self-replicating viral replicons. We have used our system to define 5′- and 3′-terminal nucleotide sequences that are essential for RNA replication. In addition, we have also demonstrated the potential of our system to generate vectors that express heterologous genes in eukaryotic cells in a variety of forms, including recombinant infectious PRRSV cDNA, viral replicons, and synthetic infectious viruses. Thus, the system we have constructed not only provides an important platform from which the basic biology of PRRSV can be investigated but also is a useful tool for designing new heterologous gene expression vectors and for generating genetically defined antiviral vaccines.
MATERIALS AND METHODS
Cells and viruses.
MARC-145 (
20) and BHK-21 cells (
61) were maintained as described previously. All reagents used in cell culture were purchased from Life Technologies, Inc., Gaithersburg, MD. The parental PRRSV used in this study is the first Korean PRRSV strain, PL97-1, which was isolated in 1997 from the serum of an infected pig (
20). High-titer virus stocks were obtained by cultivation in MARC-145 cells at a low multiplicity of infection (MOI) of 0.1 for 72 h. The viruses were then clarified by centrifugation (2,000 rpm for 10 min), aliquoted, and stored at −80°C until use.
Virus titration.
Virus titers were determined by plaque assay using MARC-145 cells. For this, cells were preseeded in a six-well plate at a density of 3 × 105 per well for 12 to 18 h and then infected with serial 10-fold dilutions of virus for 1 h at 37°C with frequent agitation. The cell monolayers were then overlaid with minimal essential medium containing 0.5% SeaKem LE agarose (FMC BioProducts, Rockland, Maine) and 5% fetal bovine serum and incubated for 4 days at 37°C with 5% CO2. The resulting plaques were visualized by fixation with 7% formaldehyde followed by staining with crystal violet (1% [wt/vol] in 5% ethanol).
Oligonucleotides.
All oligonucleotides used for cDNA synthesis, PCR amplification, and mutagenesis in this study are listed in the supplemental material (see Table S1 in the supplemental material). Their sequences were designed according to the complete nucleotide sequence of PL97-1 (GenBank accession number AY585241 ) (
20).
Construction and mutagenesis of the full-length PRRSV cDNA as a BAC.
Because of space limitations, only the salient features of plasmids are described here. A detailed description of plasmid construction is provided in the supplemental material, and computer-readable sequence files are also available upon request. All plasmids were constructed using standard molecular biology procedures (
38).
pBAC/PRRSV/FL contained the full-length PRRSV PL97-1/LP1 cDNA flanked by the SP6 RNA polymerase promoter and three unique restriction sites (AclI, NotI, and SdaI) in a row for in vitro runoff transcription. Eight constructs (pBAC/PRRSV/FL/Δnt1 to -15) harbored deletions of 1, 3, 5, 7, 9, 11, 13, and 15 nucleotides (nt) from the utmost 5′ end of the viral genome, respectively. pBAC/PRRSV/FL/Δnt3/Rev2 and pBAC/PRRSV/FL/Δnt3/Rev3 were reconstructed by the addition of 4 (TATG) and 3 (AAG) nucleotides at the beginning of the viral genome in pBAC/PRRSV/FL/Δnt3, respectively. Similarly, six additional constructs of pBAC/PRRSV/FL/Δnt7/Rev1 to -6 were reconstructed by the addition of 7 (ATTATAT), 8 (TATTATAT), 8 (TATCATAT), 10 (ATATATATAT), 12 (ATATATATATAT), or 8 (ATTTATAT) nucleotides, respectively, at the beginning of the viral genome in pBAC/PRRSV/FL/Δnt7. pBAC/PRRSV/FL/IRES-EGFP was constructed by the insertion of the encephalomyocarditis virus (EMCV) internal ribosome entry site (IRES)-driven enhanced green fluorescent protein (EGFP) expression cassette immediately downstream of the first 33 nucleotides of the ORF7 coding region, followed by the 3′ 911 nucleotides of the viral genome. pBAC/PRRSV/FL/Npro-EGFP is identical to pBAC/PRRSV/FL/IRES-EGFP, except that the EMCV IRES-driven EGFP expression cassette was replaced by the coding sequence of the Npro-EGFP fusion protein in such a way that expression of the fused Npro-EGFP was driven by its own subgenomic promoter. The autoprotease Npro gene of bovine viral diarrhea virus was fused adjacent to the N terminus of the EGFP gene so that the correct N terminus of the EGFP protein was created by Npro cleavage.
We constructed a panel of 11 PRRSV viral replicon vectors that express the luciferase (LUC) gene as a reporter. One set of three constructs, namely, pBAC/PRRSV/RepLuc MB, pBAC/PRRSV/RepLuc ME, and pBAC/PRRSV/RepLuc DI, were first constructed to contain internal deletions of nt 12714 to 14194, nt 12163 to 14194, and nt 12163 to 15252 in the viral genome, respectively, and also the insertion of the EMCV IRES-driven LUC expression cassette at the site of each deletion to facilitate the monitoring of viral replication. The other set of eight constructs, namely, pBAC/PRRSV/RepLuc S1 to S8, are identical to pBAC/PRRSV/RepLuc ME, except that the internal deletion was further extended to nt 15200, nt 15150, nt 15100, nt 15050, nt 15000, nt 14950, nt 14900, and nt 14500, respectively.
In vitro synthesis of RNA transcripts.
The full-length PRRSV cDNA (2 μg) was digested with the appropriate restriction enzyme (AclI, NotI, or SdaI) and/or treated with mung bean nuclease (MBN). With regard to the PRRSV viral replicons, all the recombinant cDNAs were linearized by NotI digestion and extracted with phenol-chloroform and precipitated with ethanol. In vitro transcription from these template cDNAs, subsequent removal of the template cDNAs from transcription reaction, and extraction and quantitation of synthetic RNA transcripts were performed as previously described (
61).
Transfection of cells with RNA transcripts.
MARC-145 or BHK-21 cells were preseeded at a density of 2 × 106 or 3 × 106 cells per p150 culture dish for 24 h at 37°C with 5% CO2, respectively. The subconfluent cells were then trypsinized and washed three times with ice-cold RNase-free phosphate-buffered saline. After resuspension at a density of 2 × 107 cells/ml in phosphate-buffered saline, 400 μl of the cells was mixed with 2 μg synthetic RNA in a cuvette with a gap width of 0.2 cm. The MARC-145 and BHK-21 cells were electroporated with 10 or 5 pulses of current, respectively, by using an ECM 830 electroporator (BTX Inc., San Diego, CA) set at 900 V and a 99-μs pulse length. The cells were then transferred to 10 ml of fresh medium. Infectious-center assays were used to quantitate the specific infectivity of synthetic RNA transcripts. The electroporated cells were then serially diluted 10-fold and plated onto monolayers of nontransfected MARC-145 cells (3 × 105) in a six-well plate. After 6 h, they were overlaid with agarose-containing minimal essential medium, incubated for 4 or 5 days, and stained with crystal violet.
Real-time quantitative reverse transcription-PCR.
Total RNA was extracted from duplicate wells with TRIzol reagent (Invitrogen Co.); 100 ng of total cellular RNA was used for the reverse transcription reaction with primers specific for the PRRSV ORF1a region as well as for the BHK-21 β-actin RNA to normalize total RNA levels. PRRSV and BHK-21 β-actin cDNAs were generated by reverse transcription at 45°C for 30 min, followed by inactivation of the reverse transcriptase at 95°C for 10 min. PRRSV-specific and BHK-21 β-actin-specific cDNAs were amplified with an iQ Supermix quantitative PCR system (Bio-Rad Laboratories, Hercules, CA) and detected with an iCycler iQ multicolor real-time PCR detection system (Bio-Rad Laboratories). One-tenth of the reverse transcription reaction mixture was used for PCR amplification with 45 cycles of 95°C for 15 s and 60°C for 1 min. The PRRSV forward and reverse primers were 5′-CATGTGAGTGATAAACCTTTCCCG and 5′-TCATAGACAGTAGCCATAGCACAC, respectively. The probe sequence (nt 659 to 682) was 5′-6FAM-ACGTGTTGACCAACCTGCCGCTCC-BHQ1 (where 6FAM is 6-carboxyfluorescein and BHQ is black hole quencher; Integrated DNA Technologies Inc., Coralville, IA). The forward and reverse primers for BHK-21 β-actin were 5′-ACTGGCATTGTGATGGACTC and 5′-CATGAGGTAGTCTGTCAGGTC, respectively. The probe sequence was 5′-HEX-CCAGCCAGGTCCAGACGCAGG-BHQ2 (where HEX is hexachlorofluorescein; Integrated DNA Technologies Inc.). The 2
−ΔΔCT method was used to analyze relative changes in PRRSV RNA levels from real-time, quantitative PCR experiments (
41,
56).
Direct immunofluorescence and analysis of reporter gene expression.
PRRSV ORF7 proteins were visualized on the surface of formaldehyde-fixed cells by incubation with mouse anti-ORF7 monoclonal antibody (MAb; 6D7/D2) followed by secondary labeling with either fluorescein isothiocyanate-conjugated or Cy3-conjugated goat anti-mouse immunoglobulin G (IgG; Jackson ImmunoResearch Labs Inc., West Grove, PA), using a procedure described elsewhere (
61). LUC activity was estimated in cell lysates, as previously described (
61). EGFP expression was visualized under a confocal microscope fitted with a fluorescein filter, as described previously (
61).
DISCUSSION
In this report, we have described the construction and characterization of a reverse genetics system for PRRSV that involves generating a functional PRRSV cDNA as a BAC which is based on
E. coli and its single-copy plasmid F factor (
42,
52). BACs have been successfully used to clone and maintain large fragments of DNA from a variety of complex genomic sources (including humans and herpesvirus) in bacteria (
26,
42). Using this functional cDNA, we not only demonstrated the importance of the PRRSV 5′-end nucleotide sequences for replication but also discovered novel AU-rich sequences of various sizes that can functionally replace the authentic 5′-proximal 7 nucleotides of the viral genome for RNA replication. We also showed the utility of this system by generating a panel of self-replicating, self-limiting PRRSV viral replicons that aided the identification of a 3′
cis-acting element required for PRRSV replication. This reverse genetics system will thus permit direct molecular genetic studies of the PRRSV pathogen that will greatly enhance our understanding of the molecular mechanisms behind its replication, transcription, translation, virulence, and pathogenesis. In addition, the results we obtained present the possibility that this system can be used to produce novel heterologous gene expression vectors that can express a foreign gene of interest. Thus, this system will also be useful in developing new, safe, and genetically defined antiviral vaccines.
With the aim of generating an infectious cDNA that has an unaltered native sequence, we also sought to ensure that the in vitro transcription of the infectious PRRSV cDNA would generate RNA transcripts possessing the authentic 5′- and 3′-terminal nucleotide sequences of the viral genome. These sequences are usually essential for the replication of viral RNA. For example, it has been shown that the 5′- and 3′-terminal regions of flavivirus RNA are required for the initiation of replication in vitro and in vivo (
24). To ensure the generation of authentic 5′- and 3′-terminal sequences, we adapted the approaches that were used previously to design flavivirus infectious cDNAs (
37,
61). The cap structure at the 5′ end of the PRRSV viral genome was followed by the nucleotide sequence
1AUG ACG
6, which is highly conserved in most PRRSV strains, including PL97-1/LP1 (
20). We achieved the authentic 5′ end of the viral genome by engineering the SP6 promoter transcription start immediately upstream of the viral genome. The in vitro runoff transcription reaction in the presence of the m
7G(5′)ppp(5′)A cap structure analog then allowed us to synthesize capped RNA transcripts with the authentic 5′ ends (
11). These transcripts were highly infectious upon transfection into susceptible MARC-145 cells and nonpermissive BHK-21 cells. Previous studies with PRRSV cDNAs used the m
7G(5′)ppp(5′)G cap structure analog instead in the in vitro T7 polymerase-driven transcription reaction (
23,
28,
34,
47). This would result in capped RNA transcripts containing one (
23,
28,
34) or two (
47) extra nonviral G nucleotides upstream of the
1AUG ACG
6 sequence. When we used the m
7G(5′)ppp(5′)G cap structure analog in our own transcription reactions, we found that the extra G did not affect the infectivities of the RNA transcripts (data not shown). We also found that the uncapped RNA transcripts were not infectious, which indicates that the cap structure is essential for viral replication (data not shown), in agreement with a previous report (
28).
The 3′ end of the PRRSV viral genome terminates with the poly(A) tail (
8,
29,
45). The poly(A) tail in PL97-1/LP1 is 54 nucleotides long. To ensure that the infectious PPRSV cDNA that we constructed would produce RNAs bearing the 54 nucleotides of the poly(A) tail, we followed the tail sequence with a unique restriction recognition site that would linearize the cDNA for runoff transcription. The resulting RNA transcripts contained the 54-nucleotide poly(A) tail and gave a specific infectivity of >10
5 PFU/μg of RNA that produces synthetic viruses of >5 × 10
4 PFU/ml upon transfection. That the 54-nucleotide poly(A) tail is important was then clearly demonstrated by the fact that unpolyadenylated RNA transcripts were not infectious (data not shown). With regard to the other PRRSV cDNAs published previously, poly(A) tails of various lengths, such as 109 (
28), 43 (
47), 40 (
34), or 21 (
23) A's, have been included during construction of each cDNA. It is not clear whether poly(A) tails of such variable lengths may affect viral replication.
Several studies have shown that it is important that the synthetic RNA transcripts that are generated from functional cDNAs for RNA viruses carry the authentic 3′ end, as this promotes their infectivity and the establishment of a productive infection (
60,
61). To ensure this for the PRRSV RNAs, we designed the infectious PRRSV cDNA template so that AclI digestion followed by MBN treatment would generate a linearized template lacking the 5′ CG overhang left by the AclI digestion. This then served as the template in SP6 polymerase runoff transcription and was found to produce highly infectious RNA transcripts bearing the authentic 5′ and 3′ termini of the viral genome. We also observed similar levels of infectivity with RNA transcripts that terminated with 2 and 10 extra nonviral nucleotides at their 3′ ends. However, 14 extra nonviral 3′ nucleotides slightly decreased the specific infectivity and delayed virus production. Interestingly, the 2 to 14 extra nonviral nucleotides were not found in the genomic RNAs of the synthetic viruses produced from the cDNA in cells. This could be because these extra nonviral nucleotides are lost during propagation in the transfected cells. Alternatively, it is also possible that the SP6 RNA polymerase in the in vitro transcription reaction may not efficiently synthesize the entire poly(A) tail each time, which would sometimes result in synthetic RNA transcripts with poly(A) tails of different sizes that lack these extra 3′ nonviral nucleotides. Thus, replication-competent RNA transcripts could be selected over replication-incompetent RNAs containing these extra nonviral nucleotides. Further investigation will be needed to distinguish between these two possibilities. With regard to the other PRRSV cDNAs published previously, linearization was induced by a unique restriction site, such as PvuI (
28), AclI (
34,
47), or SwaI (
23), which was introduced downstream of the poly(A) tail. The digestion of either PvuI or AclI would produce synthetic RNA transcripts bearing two extra nonviral nucleotides (CG) downstream of the poly(A) tail, whereas SwaI digestion would make synthetic RNAs with nine nonviral nucleotides. Our own observations suggest that these extra 2 or 9 nucleotides should not affect the infectivities of their resulting RNA transcripts.
Our work presented here elucidates the importance of the PRRSV-conserved 5′-end nucleotide sequence
1AUGACGU
7 in RNA replication. Several novel AU-rich PRRSV 5′ sequences detected in pseudorevertants were able to functionally replace the deleted
1AUGACGU
7. According to the predicted RNA secondary structure of PRRSV 5′ NCRs (
48), the presence of these novel AU-rich PRRSV 5′ sequences did not seem to significantly alter their stem-loop structures, which might be critical for viral replication. Although the functional role of these novel sequences is not understood at all, the complementary sequence of each of these novel 5′ sequences at the utmost 3′ end of negative-sense RNA is predicted to be involved in the initiation of positive-sense RNA synthesis. For some positive-sense RNA viruses, only minimal
cis-acting sequences at the 3′ ends of negative-sense RNAs are essential for positive-sense RNA synthesis (
4,
18,
36). It is interesting to speculate on the origin of these novel 5′ AU-rich sequences and the molecular mechanism of their insertion at the very beginning of the viral genome. Based on the heterogeneity and size of these novel 5′ sequences, it is likely that they are acquired from cellular RNAs in the process of recombination involving template switching, during either negative- or positive-sense RNA synthesis, as has been described for pestivirus (
30) and poliovirus (
22). It is less likely, although not impossible, that these sequences are derived from the viral genome. Understanding this issue may provide new information on RNA replication of arteriviruses.
Several studies have shown that
cis-acting elements within the coding region of the viral genome are essential for RNA virus replication (
24). By analyzing a panel of self-replicating, self-limiting, LUC-expressing PRRSV viral replicons with internal deletions starting from the beginning of ORF2a and extending towards the 3′ NCR, we found that there was a
cis-acting element within the coding region of PRRSV as well. These analyses revealed that the three viral replicons that contained at least 911 nucleotides from the 3′ end of the genome were replication competent. In contrast, eight viral replicons that did not contain 911 nucleotides at the 3′ end of the genome were all replication incompetent. The LUC-expressing PRRSV viral replicon system is advantageous for monitoring viral replication in a highly quantitative and sensitive manner. On the other hand, the sequence requirements for RNA replication of the LUC-containing PRRSV viral replicon might not necessarily be the same as those for full-length genome replication. Especially, the insertion of a highly structured RNA sequence of EMCV IRES and the 1,653-nucleotide LUC gene could have a deleterious effect on the overall level of RNA structure when this insertion is combined with a specific deletion within the viral genome. Investigations on how the
cis-acting element mapped in this study works in RNA replication in the context of the full-length viral genome might be an important step towards a better understanding of PRRSV RNA synthesis. A study using the infectious Lelystad strain cDNA has also shown that a 34-nucleotide stretch within the PRRSV ORF7 is essential for RNA replication (
51). In addition to this stretch of 34 nucleotides, the presence of an additional 3′
cis-acting element which is essential for viral replication was revealed in our own data. Investigations using our replicon system that aim to finely map this minimal 3′
cis-acting element and elucidate the molecular mechanism by which it controls viral replication are currently under way.
Several RNA viruses have been successfully exploited as eukaryotic expression vectors (
12,
40). We showed here that the infectious PRRSV cDNA BAC that we constructed could also serve as a novel PRRSV-based virus/vector system in which foreign genes of interest could be expressed. We showed the possibility that this system can be used to express foreign genes in two ways: first, by generating infectious recombinant vector RNAs/viruses that carry a foreign gene, and second, by producing viral replication-competent but propagation-deficient PRRSV viral replicon vector RNA that carries a foreign gene. As a transient-expression system, PRRSV offers several advantages: (i) the recombinant virus is rapidly produced, (ii) PRRSV can replicate in a variety of eukaryotic cells upon transfection of synthetic RNAs, (iii) PRRSV is unable to infect humans, (iv) the genetically stable infectious cDNA is available and readily manipulated, and (v) the cytoplasmic replication of the RNA genome minimizes the possibility of integration and unwanted mutagenic consequences. Recently, a human coronavirus with a positive-strand RNA genome has been developed as a multigene expression vector; the transcription strategy of this vector involves producing a nested set of multiple subgenomic mRNAs (
46). PRRSV employs a similar strategy in its gene expression. That the genome of PRRSV is smaller (∼15 kb) than that of coronaviruses (∼30 kb) suggests that our PRRSV-based expression system can be even more useful as a multigene expression vector.
In conclusion, our reverse genetics system for PRRSV, which involves an infectious cDNA BAC and viral replicons, may be useful in a variety of research fields. First, this system enables us not only to investigate the molecular mechanisms of viral replication, transcription, and translation but also to identify the viral genetic elements involved in virulence and pathogenesis. Second, it can serve as an invaluable genetic tool for heterologous gene expression in a wide variety of eukaryotic cells; this has many applications in biological research. Finally, targeted manipulation of the infectious PRRSV cDNA may permit the production of effective genetically modified antiviral vaccines against this pathogen that may abrogate the ongoing worldwide spread of PRRSV.