Coronavirus Non-Structural Protein 1 Is a Major Pathogenicity Factor: Implications for the Rational Design of Coronavirus Vaccines

Attenuated viral vaccines can be generated by targeting essential pathogenicity factors. We report here the rational design of an attenuated recombinant coronavirus vaccine based on a deletion in the coding sequence of the non-structural protein 1 (nsp1). In cell culture, nsp1 of mouse hepatitis virus (MHV), like its SARS-coronavirus homolog, strongly reduced cellular gene expression. The effect of nsp1 on MHV replication in vitro and in vivo was analyzed using a recombinant MHV encoding a deletion in the nsp1-coding sequence. The recombinant MHV nsp1 mutant grew normally in tissue culture, but was severely attenuated in vivo. Replication and spread of the nsp1 mutant virus was restored almost to wild-type levels in type I interferon (IFN) receptor-deficient mice, indicating that nsp1 interferes efficiently with the type I IFN system. Importantly, replication of nsp1 mutant virus in professional antigen-presenting cells such as conventional dendritic cells and macrophages, and induction of type I IFN in plasmacytoid dendritic cells, was not impaired. Furthermore, even low doses of nsp1 mutant MHV elicited potent cytotoxic T cell responses and protected mice against homologous and heterologous virus challenge. Taken together, the presented attenuation strategy provides a paradigm for the development of highly efficient coronavirus vaccines.


Introduction
Coronaviruses are vertebrate pathogens mainly associated with respiratory and enteric diseases [1]. They can cause severe diseases in livestock animals and lead thereby to high economic losses. In humans, coronavirus infections manifest usually as mild respiratory tract disease (common cold) that may cause more severe symptoms in elderly or immunecompromised individuals [2,3]. In 2002-2003, the appearance of severe acute respiratory syndrome (SARS), caused by a formerly unknown coronavirus (SARS-CoV), exemplified the potential of coronaviruses to seriously affect human health [4][5][6][7]. The frequent detection of SARS-like coronaviruses in horseshoe bats (Rhinolophus sp.) and the broad range of mammalian hosts that are susceptible to SARS-CoV infection may facilitate a potential reintroduction into the human population [8]. Therefore, the development of efficacious coronavirus vaccines is of high medical and veterinary importance.
Effective vaccines controlling virus spread and disease are available for a number of infections, such as smallpox, poliomyelitis, measles, mumps, rubella, influenza, hepatitis A, and hepatitis B [9,10]. Some of these vaccines consist of virus subunits or inactivated virus preparations that mainly induce the production of pathogen-specific antibodies. In contrast, live attenuated vaccines consist of replicationcompetent viruses that induce broad cellular and humoral immune responses without causing disease [10]. The most prominent live attenuated vaccines are vaccinia virus [11], poliovirus [12], and yellow fever virus (YF-17D) [13]. Despite their documented efficacy, it is still not fully understood why and how successful vaccines work [10,14]. However, recent concepts in immunology provide a link between innate and adaptive immune responses and suggest that the quality, quantity, and longevity of adaptive immune responses is determined very early after infection or vaccination [14]. Of major importance are professional antigen-presenting cells (pAPCs) such as dendritic cells (DCs) and macrophages, which play a major role in (i) sensing pathogen-associated molecular patterns, (ii) inducing innate immune responses, and (iii) shaping the upcoming adaptive immune response. Efficient live attenuated vaccines should therefore not only lack significant pathogenicity, but should also deliver antigens to pAPCs and activate the innate immune system.
Notably, the majority of currently available attenuated vaccines have been derived empirically. Given the recent proceedings in the areas of virus reverse genetics and virus-host interactions, the time should be ripe for more rational approaches in vaccine development. An attractive strategy is to target virally encoded pathogenicity factors, such as interferon (IFN) antagonists [15], to attenuate virulence while retaining immunogenicity. This concept has been proposed for the generation of live attenuated influenza virus vaccines encoding altered NS1 proteins [16,17].
Our rudimentary knowledge on coronavirus-encoded pathogenicity factors is reflected by the fact that only a few putative coronaviral pathogenicity factors have been identified and that functional analyses are still limited to the description of in vitro effects [18][19][20]. For a number of reasons, the non-structural protein 1 (nsp1) is of particular interest in this context. First, coronaviruses are positivestranded RNA viruses, and the replicase-encoded nsps are expressed from the viral genomic RNA immediately after virus entry by translation of two large polyproteins. nsp1 is encoded at the 59 end of the replicase gene and is therefore the first mature viral protein expressed in the host cell cytoplasm [21]. Second, a recent in vitro study suggests that SARS-CoV nsp1 may be associated with host cell mRNA degradation and may counteract innate immune responses [18]. Finally, nsp1 is encoded by all mammalian coronaviruses known to date (coronavirus groups 1, 2a, and 2b) [22], and recent structural data on SARS-CoV (group 2b) nsp1 suggest functional similarities to mouse hepatitis virus (MHV; group 2a) nsp1 [23].
Using a reverse genetics approach, we show here that nsp1 is a major pathogenicity factor. Recombinant MHV mutants encoding a deletion in nsp1 replicated as efficiently as wildtype virus in cell culture, but displayed an unprecedented degree of attenuation in mice. Interference with the type I IFN system appears to be the dominant mode of action of murine coronavirus nsp1. Vaccination with the nsp1 mutant virus elicited efficient memory cytotoxic T cell responses and protected against homologous and heterologous virus infections. Our study will pave the way for the generation of novel coronavirus vaccines based on modified coronavirus replicase genes.

Transient nsp1 Expression Affects Cellular Gene Expression
We assessed several replicase-encoded nsps of MHV (strain A59), SARS-CoV, and human coronavirus 229E (HCoV-229E) for their ability to interfere with host cell gene expression. Using transient gene expression studies, we found that MHV-A59, SARS-CoV, and HCoV-229E nsp1 significantly reduced luciferase reporter gene expression under the control of IFNb, IFN-stimulated response element (ISRE), and SV40 promoters ( Figure 1). This is consistent with a recent report suggesting that SARS-CoV nsp1 induces general host cell mRNA degradation [18]. Nevertheless, it should be noted that the nsp1-mediated reduction in reporter protein expression appeared more robust for ISRE and SV40 than for IFN-b promoter-driven expression. Our data also support the hypothesis that MHV encodes a SARS-CoV nsp1 homolog that displays the same function [23]. Although comparative sequence analyses suggest that nsp1 of group 2a/2b coronaviruses (e.g., MHV and SARS-CoV, respectively) and the nsp1 of group 1 coronaviruses (e.g., HCoV-229E) may belong to different protein families [22,23], we also observed reduced reporter gene expression in HCoV-229E nsp1-transfected cells ( Figure 1). Whether functional similarities may exist between nsp1 molecules encoded by coronaviruses of different phylogenetic lineages remains to be established in future studies. Importantly, our data also revealed that reporter gene expression from all tested promoters was not affected when C-terminally truncated MHV nsp1 molecules were tested ( Figure 1).

Generation of the MHV nsp1 Deletion Mutant
To assess the role of nsp1 in the context of virus replication, we constructed a recombinant MHV encoding a truncated nsp1 protein using our reverse genetic system [24]. Based on the results shown in Figure 1, we decided to delete MHV nucleotides (nts) 829-927 (99 nts). In the resulting mutant virus, MHV-nsp1D99, the replicase gene start codon, the translational reading frame, and the residues required for proteolytic release of nsp1 from the replicase polyprotein were maintained ( Figure 2A). As reported for a set of similar MHV mutants by Brockway et al. [25], viral growth and peak titers of MHV-nsp1D99 in murine 17Clone1 cells were indistinguishable from that of wild-type virus ( Figure 2B). To assess the stability of the recombinant MHV-nsp1D99, we analyzed the nsp1-coding region by RT-PCR sequencing after seven passages in tissue culture and no nucleotide changes were detected (unpublished data).
Infection of conventional DCs (cDCs) is an early and crucial event for the generation of protective antiviral immunity [26]. MHV productively infects cDCs and activates plasmacytoid DCs (pDCs) to generate a first wave of protective type I IFN [27]. To assess whether the mutant MHV-nsp1D99 has retained the ability to infect pAPCs, peritoneal macrophages ( Figure 2C), bone marrow-derived CD11c þ cDCs ( Figure 2D), and splenic, FACS-sorted CD11c þ cDCs ( Figure 2E) were exposed to MHV-nsp1D99 and wild-type control virus. Similar to replication kinetics in cell lines ( Figure 2B),

Author Summary
Prevention of viral diseases by vaccination aims for controlled induction of protective immune responses against viral pathogens. Live viral vaccines consist of attenuated, replication-competent viruses that are believed to be superior in the induction of broad immune responses, including cell-mediated immunity. The recent proceedings in the area of virus reverse genetics allows for the rational design of recombinant vaccines by targeting, i.e., inactivating, viral pathogenicity factors. For coronaviruses, a major pathogenicity factor has now been identified. The effect of coronavirus non-structural protein 1 on pathogenicity has been analyzed in a murine model of coronavirus infection. By deleting a part of this protein, a recombinant virus has been generated that is greatly attenuated in vivo, while retaining immunogenicity. In particular, the mutant virus retained the ability to replicate in professional antigen-presenting cells and fulfilled an important requirement of a promising vaccine candidate: the induction of a protective long-lasting, antigen-specific cellular immune response. This study has implications for the rational design of live attenuated coronavirus vaccines aimed at preventing coronavirusinduced diseases of veterinary and medical importance, including the potentially lethal severe acute respiratory syndrome.
MHV-nsp1D99 showed no significant growth defect in primary pAPCs ( Figure 2C-2E), indicating that the deletion of nsp1 did not alter the pronounced tropism of MHV for cDCs and macrophages.

Deletion in nsp1 Confers Strong Attenuation In Vivo
MHV-A59 is a hepatotropic and neurotropic virus that can cause acute hepatitis and encephalitis. Following intraperitoneal infection, virus replication is first detectable in spleen and liver, followed by virus spread to other organs, including the central nervous system. Hepatitis is the first clinical sign of disease, accompanied by elevated liver enzyme values in serum. Associated with the appearance of cytotoxic T cell responses approximately at day 5 post infection (p.i.), virus titers usually decline and are no longer detectable after day 7 p.i. Infections with a high dose (! 5 3 10 6 pfu, intraperitoneal) may, however, occasionally result in fatality. To evaluate the importance of nsp1 for virus replication and viral pathogenicity in vivo, C57BL/6 mice were infected intraperitoneally with different doses of wild-type MHV or MHV-nsp1D99. Both viruses replicated in the spleen, whereby MHV-nsp1D99 titers were consistently lower than wild-type virus titers ( Figure 3A). Furthermore, MHV-nsp1D99 was rapidly cleared and not detectable after day 2 p.i. ( Figure 3A). Wild-type, but not mutant virus, was detectable in the liver at low and intermediate dose (50 pfu and 5,000 pfu, respectively) ( Figure 3B). When high virus doses (5 3 10 6 pfu) were applied, MHV-nsp1D99 eventually reached the liver at day 2 p.i., but was not detectable at later time points ( Figure 3B). MHV-nsp1D99 was not detectable in other non-hematopoietic organs, such as lung and central nervous system (unpublished data). Mice infected with wild-type virus showed acute liver disease with elevated liver enzyme values in serum. Furthermore, after high dose infection with wild-type virus (5 3 10 6 pfu), a significant weight loss that peaked at approximately 10%-15% at day 4 was observed ( Figure 3E). In contrast, mice infected with the nsp1 mutant virus remained healthy after low, intermediate, or high dose infections. Even at the highest dose applied (5 3 10 6 pfu), MHV-nsp1D99-infected mice did not lose weight ( Figure 3E), and no elevated liver enzyme values were detected in the serum ( Figure 3C). This observation correlated well with the absence of hepatocyte necrosis and parenchymal inflammation following MHV-nsp1D99 infection ( Figure 3D). To further assess the attenuation of the MHV nsp1 mutant, mice were infected intracranially with 200 pfu and 20,000 pfu of MHV-nsp1D99  Figure 3F). Collectively, these data demonstrate that MHV-nsp1D99 is strongly attenuated in vivo, but has retained the ability to replicate in secondary lymphoid organs, such as the spleen.

Effect of nsp1 on Innate Immune Responses
We have previously shown that pDCs are the major source of IFN-a in the early stages of MHV infection and that type I IFN responses in CD11c þ cDCs are only weakly triggered by MHV [27]. To test whether nsp1 has an influence on the induction of IFN-a, we infected both cDCs and pDCs with MHV-nsp1D99 or wild-type MHV. Both viruses elicited rapid and high IFN-a production in Flt3-L-differentiated bone marrow-derived pDCs ( Figure 4A) and FACS-sorted primary pDCs ( Figure 4B). Furthermore, both wild-type and mutant MHV elicited only a late and weak IFN-a production in cDCs ( Figure 4A and 4B). These results suggest that nsp1 does not affect the induction of type I IFN. To assess a potential impact of nsp1 on type I IFN signaling and antiviral effector mechanisms in target cells that efficiently support MHV replication, cDCs and macrophages were pretreated with different dosages of IFN-a and infected with MHV-nsp1D99 or wild-type MHV. In cDCs, IFN-a treatment had a comparable effect on the replication of both MHV-nsp1D99 and the wild-type control virus ( Figure 4C). However, replication of MHV-nsp1D99 was, in a dose-dependent manner, more vulnerable to IFN-a treatment in macrophages ( Figure 4D), suggesting that nsp1 might counteract IFN signaling and/or the antiviral activities of IFN-induced effector proteins.
Indeed, in vivo experiments in type I IFN receptordeficient (IFNAR À/À ) mice [28] strongly support this interpretation. Infection of IFNAR À/À mice with wild-type MHV led to high titers in all tested organs ( Figure 5A-5D), indicating that signals transmitted via the IFNAR are crucial for preventing uncontrolled spread of the virus [27]. Surprisingly, the severe attenuation of MHV-nsp1D99 in wild-type 129Sv mice was not present in IFNAR À/À mice ( Figure 5A-5D). Replication of MHV-nsp1D99 in IFNAR À/À mice was largely restored and virus titers reached about 10 4 -10 5 pfu/g tissue in several organs after only 36 h (figures 5A-5D). These data strongly suggest that nsp1 has a pivotal role in counteracting type I IFN host responses and provide an explanation for the rapid clearance of MHV-nsp1D99 in wildtype mice. Interestingly, liver damage, measured as alanine 2oxoglutarate-aminotransferase (ALT) levels in serum, was not yet detectable in MHV-nsp1D99-infected IFNAR À/À mice at 36 h p.i. (Figure 5E). At 72 h p.i., MHV-nsp1D99 reached titers and ALT levels in IFNAR À/À mice comparable to those observed in MHV-A59-infected IFNAR À/À mice at 36 h p.i., demonstrating that MHV-nsp1D99 replication in IFNAR À/À is, although with slower kinetics, restored.

Immunization with the MHV nsp1 Deletion Mutant Protects against Homologous and Heterologous Virus Challenge
The phenotypic analysis of MHV-nsp1D99 revealed a number of features that are advantageous for live attenuated vaccines. MHV-nsp1D99 grows to high titers in cell culture, infects pAPCs, replicates almost exclusively in secondary lymphoid organs, and is strongly attenuated in vivo. To assess the potential of MHV-nsp1D99 as an attenuated live vaccine, we replaced accessory gene 4 of MHV-nsp1D99 and wild-type MHV-A59 by a gene encoding a fusion protein of the immunodominant cytotoxic T lymphocyte (CTL) epitope (KAVYNFATC) of the lymphocytic choriomeningitis virus (LCMV) and the enhanced green fluorescent protein (GP33-GFP) [29]. The resulting recombinant viruses, MHV-nsp1D99-GP33-GFP and MHV-GP33-GFP, were used to infect C57BL/6 mice with different doses (50 and 5,000 pfu, intraperitoneal), and CD8 þ T cell responses were assessed using flow cytometry-based detection of intracellular IFN-c following antigen-specific short-term in vitro restimulation. As shown in Figure 6A and 6B, infection with as few as 50 pfu of MHV-nsp1D99-GP33-GFP elicited strong CD8 þ T cell responses against both the H2-D b -restricted GP33 and the H2-K brestricted MHV S598 epitope.
To assess the level of protection against homologous MHV-A59 challenge, groups of C57BL/6 mice were immunized (5,000 pfu) with MHV-nsp1D99-GP33-GFP, MHV-GP33-GFP, or treated with PBS. Sixteen days p.i., mice were challenged with wild-type MHV (5,000 pfu) and viral titers were determined 5 d post challenge infection. Viral titers were below the limit of detection in MHV-nsp1D99-GP33-GFPand MHV-GP33-GFP-immunized mice ( Figure 6C). Together with the absence of elevated liver enzyme values in immunized mice ( Figure 6D), these data indicate that vaccination with the attenuated MHV nsp1 mutant provides complete protection against homologous virus challenge. The reverse genetic system facilitates incorporation of antigens derived from other infectious organisms. In order to determine whether the attenuated nsp1 mutant virus could confer protection against heterologous virus infection, MHV-nsp1D99-GP33-GFP-immunized C57BL/6 mice were challenged after 4 wk with LCMV (200 pfu, intravenous). LCMV titers in spleens were significantly reduced both in mice vaccinated with MHV-GP33-GFP and the attenuated MHV-nsp1D99-GP33-GFP virus ( Figure 6E). Remarkably, only 50 pfu of nsp1 mutant virus expressing the GP33 epitope were sufficient to achieve a reduction of LCMV titers by more than 4 orders of magnitude, indicating that nsp1 mutant viruses are well-suited to serve as attenuated recombinant virus vectors against heterologous viral infections.

Discussion
The rational design of live attenuated viral vaccines is greatly facilitated by the identification and targeting of pathogenicity factors. This study demonstrates an unprecedented level of attenuation of a murine coronavirus through a 99-nt deletion in nsp1. The nsp1 mutant virus was rapidly cleared in mice and did not induce clinical signs of disease in immunocompetent mice. These findings in the murine coronavirus model demonstrate that nsp1 is a major pathogenicity factor. In a stepwise approach, we made use of these observations to provide a blueprint for the construction and evaluation of live attenuated coronavirus vaccines encoding a truncated nsp1.
The presented results indicate that nsp1 plays a crucial role in the MHV life cycle by interfering with host innate immune responses. In accordance with the recent report by Kamitani et al. [18], we observed reduced reporter gene expression in transient nsp1 expression studies. The suggestion that SARS-CoV nsp1 may play a role in SARS-CoV pathogenesis by promoting host cell mRNA degradation [18] has now received support through the analysis of a coronavirus nsp1 mutant in a murine model. The MHV nsp1 mutant phenotype led us to conclude that nsp1 mainly affects IFN signaling pathways or other downstream events. The influence on IFN-a induction appears to be limited. These conclusions are based on several observations. First, the analysis of IFN-a production by pDCs and cDCs revealed no significant differences between wildtype and nsp1 mutant virus infections. Second, treatment of macrophages with IFN-a revealed a more efficient reduction of MHV-nsp1D99 replication compared with that of wild-type MHV. Finally, and most strikingly, IFNAR À/À mice were highly permissive for the mutant virus, and organ titers almost reached those of wild-type MHV-infected IFNAR À/À mice. Nevertheless, it should be noted that the nsp1 mutant virus replication was still slightly delayed in IFNAR À/À mice. Therefore, further studies are required to define molecular target(s) and the precise function(s) of coronavirus nsp1. Likewise, further studies are required to assess the impact of other coronaviral gene products on coronavirus pathogenicity. Recent reports indicate that coronaviruses most likely express a number of proteins, such as MHV and SARS-CoV nucleocapsid proteins, and SARS-CoV ORF3b and ORF6 proteins, that may interact with innate immune responses [19,20]. Also, the coronavirus replicase gene may harbor additional functions that play a role in virus-host interactions. It has been shown that the MHV and SARS-CoV nsp2 proteins [30], and the highly conserved ADP-ribose-199monophosphatase activity [31] encoded in nsp3, are both dispensable for virus replication in tissue culture, and that a single point mutation in the MHV nsp14 confers a strong attenuation of MHV in mice [32]. Clearly, the murine model, with MHV as a natural mouse pathogen, will be highly advantageous in the examination of this issue, because it allows the use of well-characterized inbred and transgenic mice in combination with well-established immunological techniques required to assess the full range of coronavirushost interactions.
The most remarkable finding of this study is the level of attenuation of the nsp1 mutant virus and its restricted replication in secondary lymphoid organs. It may well be that other coronaviral nsp1 molecules exert similar functions as the MHV nsp1. The coronavirus nsp1 has been suggested as a group-specific marker to differentiate group 1 coronaviruses from group 2a/2b coronaviruses [22]. Our transient nsp1 expression data indeed support the notion that SARS-CoV and MHV may encode evolutionarily conserved nsp1 homologs [22,23]. Nevertheless, further in vivo studies are required to determine whether the group 2b SARS-CoV nsp1 is indeed a functional equivalent to the structurally highly conserved group 2a nsp1 molecules encoded by MHV, bovine coronavirus, porcine hemagglutinating encephalomyocarditis virus, HCoV-OC43, and HCoV-HKU1. Likewise, it will be important to clarify in vivo, whether, despite the apparent lack of any sequence homology [22,23], the nsp1 of group 1 coronaviruses (e.g., HCoV-229E) may represent a functional correlate to the nsp1 of group 2a/2b coronaviruses. Recent progress in the establishment of suitable mouse models for SARS-CoV [33][34][35] and HCoV-229E [36] will enable researchers to address these questions in future studies.
The chosen attenuation strategy has resulted in the generation of a recombinant virus that fulfills important criteria of a live virus vaccine candidate: (i) growth to high titers in cell culture, which facilitates vaccine production, and (ii) generation of immunological memory that mediates efficient protection against viral challenge. One important aspect of protection against viral infections is the induction of specific cytotoxic T cells by pAPCs in secondary lymphoid organs [14]. A number of coronaviruses, such as MHV, HCoV-229E, feline infectious peritonitis virus, and SARS-CoV, have been shown to infect pAPCs and to replicate in the secondary lymphoid organs [27,[37][38][39][40][41][42]. Because of their pronounced tropism for pAPCs and the induction of strong CTL responses, we propose that coronaviruses represent promising vectors for the expression of heterologous antigens. The identification of nsp1 as a major pathogenicity factor will significantly increase the safety of coronavirus-based vectors  [43]. For example, the deletion of accessory genes (i.e., not replicase or structural genes) has been described for some coronaviruses to confer attenuation in the natural host [44][45][46], and the deletion of the structural envelope protein E has resulted in the development of replication-competent, but propagation-deficient, coronavirus vectors [47,48]. Now, with an accompanying deletion in the nsp1-coding sequence, such vectors would be considered ''recombination proof'', because the deletions are located at opposite genomic regions (i.e., within the replicase gene at the 59 end and within the structural gene region at the 39 end of the coronavirus genome), which make the reconstruction of virulent viruses by recombination unlikely. We therefore suggest that accessory gene, E gene, and partial nsp1 gene deletions will result in particular safe vectors with the potential to express multiple heterologous antigens [40,49].
Taken together, our results describe a novel type of coronavirus vaccines based on impaired function of a replicase gene product. We expect that our approach is applicable to most, if not all, mammalian coronaviruses and that it will enable the development of long-desired live attenuated vaccines for important coronavirus-induced diseases in humans and animals.

Materials and Methods
Mice and cells. C57BL/6 mice were obtained from Charles River Laboratories (http://www.criver.com/). 129Sv and type I IFN receptordeficient mice (IFNAR À/À ) [28] were obtained from the Institut fü r Labortierkunde (University of Zü rich) and bred in our facilities. All mice were maintained in individually ventilated cages and were used between 6 and 9 wk of age. All animal experiments were performed in accordance with the Swiss Federal legislation on animal protection.
MC57, BHK-21, L929, 293, and CV-1 cells were purchased from the European Collection of Cell Cultures (http://www.ecacc.org.uk/). D980R cells were a kind gift from G. L. Smith, Imperial College, London, United Kingdom. 17Clone1 cells were a kind gift from S. G. Sawicki, Medical University of Ohio, Toledo, Ohio, United States. BHK-MHV-N cells, expressing the MHV-A59 nucleocapsid protein under the control of the TET/ON system (Clontech, http://www. clontech.com/), have been described previously [24]. All cells were maintained in minimal essential medium supplemented with fetal bovine serum (5%-10%) and antibiotics.
Mutant vaccinia viruses are based on the recombinant vaccina virus vMHV-inf-1 (containing the full-length MHV-A59 cDNA) and were generated using our reverse genetic system as described previously [24]. Briefly, the gene to be mutated was replaced by the Escherichia coli guanine-phosphoribosyltransferase (GPT) gene through vaccinia virus-mediated homologous recombination, and GPT-positive clones were selected by three rounds of plaque purification on CV-1 cells in the presence of xanthine, hypoxanthine, and mycophenolic acid (GPT-positive selection) [50]. In a second round, the GPT gene was replaced by the mutated gene, and GPTnegative clones, containing the mutated gene, were selected by three rounds of plaque purification on D980R cells in the presence of 6thioguanine (GPT-negative seletion) [50]. To construct the recombinant vaccinia virus encoding the MHV-nsp1D99 cDNA, the 59 end of the MHV-A59 cDNA in vMHV-inf-1 was replaced by GPT using the plasmid pRec4. This plasmid is based on pGPT-1 [50], and the GPT gene is flanked to its left by 500 bp of vaccinia DNA and to its right by an internal ribosomal entry sequence (IRES) followed by MHV-A59 nts 952-1315. The GPT-negative selection was carried out using the plasmid pRec15. This plasmid contains 250 bp of vaccinia DNA followed by the bacteriophage T7 RNA polymerase promoter, one G nucleotide, and MHV nts 1-828 linked to MHV nts 928-1315. To replace the MHV-A59 accessory gene 4 in vMHV-inf-1 and vMHV-nsp1D99 by a gene encoding a fusion protein of EGFP and the LCMVderived CTL epitope KAVYNFATC (GP33-GFP) [29], the plasmid pRec8 was used for recombination with vaccinia viruses vMHV-inf-1 and vMHV-nsp1D99 . This plasmid contains the GPT gene flanked to its left by MHV nts 27500-27967 and to its right by MHV nts 28265-28700. GPT-negative selection was carried out using the plasmid pRec9. This plasmid contains MHV bp 27500-27967, the GP33-GFP gene, and MHV nts 28265-28700. Further cloning details, plasmid maps, and sequences are available from the authors upon request. Recombinant coronaviruses were rescued from cloned cDNA using purified vaccinia virus DNA as template for the in vitro transcription of recombinant MHV genomes as described [51].
Virus infections, determination of virus titers, and liver enzyme values. Mice were injected intraperitoneally or intracranially with indicated pfu of MHV A59 or intravenously with the indicated pfu of LCMV and sacrificed at the indicated time points. Organs were stored at À70 8C until further analysis. Blood was incubated at RT to coagulate, centrifuged, and serum was used for ALT measurements using a Hitachi 747 autoanalyzer (http://www.hitachi.com/). Peritoneal exudates cells (PECs) were isolated from the peritoneal cavity by flushing with 4 ml of ice-cold PBS. MHV titers were determined by standard plaque assay using L929 cells. LCMV titers in the spleens were determined 4 d after intravenous challenge in an LCMV infectious focus assay as previously described [55].
Histology, IFN-a ELISA, IFN-a treatment. Organs were fixed in 4% formalin and embedded in paraffin. Sections were stained with hematoxylin and eosin. Images were acquired using a Leica DMRA microscope (Leica, http://www.leica.com/) with a 253/0.65 NA objective (total magnification, 3162). Images were processed using Adobe Photoshop (Adobe Systems, http://www.adobe.com). Mouse IFN-a concentration in cell culture supernatants was measured by ELISA (PBL Biomedical Laboratories, http://www.interferonsource.com/) according to the manufacturer's instructions. IFN-a treatment of cells prior to MHV infection was performed using universal type I IFN (IFN-aA/D, Sigma).
Intracellular cytokine staining. Specific ex vivo production of IFNc was determined by intracellular cytokine staining. Organs were removed at the indicated time points following infection with recombinant MHV. For intracellular cytokine staining, single cell suspensions of 1 3 10 6 splenocytes were incubated for 5 h at 37 8C in 96-well round-bottom plates in 200 ll of culture medium containing 25 U/ml IL-2 and 5 lg/ml Brefeldin A (Sigma). Cells were stimulated with phorbolmyristateacetate (PMA, 50 ng/ml) and ionomycin (500 ng/ml) (both purchased from Sigma) as positive control or left untreated as a negative control. For analysis of peptide-specific responses, cells were stimulated with 10 À6 M GP33 peptide or 10 À4 M MHV S598 peptide. The percentage of CD8 þ T cells producing IFN-c was determined using a FACSCalibur flow cytometer (BD Biosciences). Both S598 (RCQIFANI) and GP33 (KAVYNFATC) peptides were purchased from Neosystem (http://www.neomps.com/). Statistical analysis. All statistical analyses were performed with Prism 4.0 (GraphPad Software, http://www.graphpad.com/). Data were analyzed with the paired Student's t-test assuming that the values followed a Gaussian distribution. A p-value of , 0.05 was considered significant.