INTRODUCTION
Posttranslational modifications (PTMs) of proteins regulate many cellular processes. Common PTMs include phosphorylation, acetylation, ubiquitination, and ADP-ribosylation. Viruses are well known for their ability to manipulate PTMs for their advantage, often encoding proteins that are able to add modifications (protein kinases, E3 ligases) or remove them (phosphatases, deubiquitinases) from proteins. ADP-ribosylation is a common but little-studied PTM whereby diphtheria toxin-like ADP-ribosyl transferases (ARTDs) transfer ADP-ribose from NAD
+ onto target proteins. ADP-ribose can be attached either as a single unit via mono-ADP-ribosylation (MAR) or as polymers via poly-ADP-ribosylation (PAR). ARTDs have both antiviral and proviral effects on replication (reviewed in reference
1). In support of their antiviral roles, several ARTDs are induced by interferon (IFN) (
2,
3); others enhance interferon-stimulated gene (ISG) expression (
4,
5); and some are under positive selection, a hallmark of proteins involved in virus-host conflict (
6,
7). Specific examples of antiviral ARTD1s include ARTD10, ARTD12, and ARTD14, which block alphavirus replication (
2). Also, ARTD14 and ARTD12 have been demonstrated to modulate IFN and NF-κB pathways, respectively (
8,
9). Finally, ARTD13, known as zinc-antiviral protein (ZAP), is an inactive ARTD that has strong activity against a number of viruses (
10).
Several proteins have been identified that regulate ADP-ribosylation by removing ADP-ribose from target proteins. These include PAR-glycohydrolases (PARG), ADP-ribosyl hydrolases (ARH), and macrodomain-containing proteins (reviewed in reference
11). A macrodomain is an evolutionarily conserved domain consisting of approximately 170 amino acids with a well-defined structure of central β-sheets flanked by α-helices (
12,
13). They are present in all domains of life, and humans have 9 genes that encode macrodomain proteins. Many studies have shown that these domains bind to mono- or poly-ADP-ribose (MAR or PAR) and that several are able to hydrolyze ADP-ribose-1′′ phosphate, a by-product of tRNA splicing, to ADP-ribose (
14,
15). Recently, these domains were shown to remove ADP-ribose from proteins and thus may play a key role in the regulation of protein ADP-ribosylation (
16–18). Furthermore, as several ARTDs have been shown to have antiviral activity, macrodomains may inhibit these functions; however, this has not been experimentally demonstrated. Consistent with this idea, viruses from the
Hepeviridae,
Togaviridae, and
Coronaviridae families all encode macrodomains. Originally, it was thought that these enzymes primarily act as ADP-ribose-1′′-phosphatases (ADRPs) or bind to PAR (
14,
19–22), as these activities had been convincingly demonstrated
in vitro. However, it was recently shown that macrodomains from hepatitis E virus (HEV) and SARS-CoV can deMARylate and dePARylate protein substrates
in vitro (
23), although no specific protein targets for viral macrodomains have been identified. No clear role for any potential macrodomain activity during infection has been identified, and no connection to innate immunity has been shown for ADR-1′′-phosphate or free ADP-ribose.
Coronaviruses (CoVs) are large, positive-sense RNA viruses that cause a variety of veterinary and human diseases. In humans, CoVs were originally thought to cause only mild respiratory or gastrointestinal disease. Then, in 2002 to 2003, severe acute respiratory syndrome (SARS)-CoV emerged, spreading across multiple countries and causing a severe respiratory disease with a mortality rate of ~10% (
24). In 2012, the Middle East respiratory syndrome coronavirus (MERS-CoV) emerged as another highly pathogenic CoV with a high mortality rate (
25). In addition, a number of SARS-like CoVs are currently circulating in bats and are able to infect human cells, suggesting that a single zoonotic transmission event could set off another epidemic (
26). Currently, there are no approved therapeutics or vaccines to treat human CoVs, and as such, increased research into further understanding CoV biology to identify therapeutic targets and novel vaccine strategies is needed to control these viruses.
All known CoVs encode a macrodomain within the large transmembrane protein nonstructural protein 3 (nsp3). Using mutation of the highly conserved asparagine residue, which was shown to be essential for ADP-ribose-1′-phosphatase activity
in vitro (
14), this domain was shown to be required for murine hepatitis virus (MHV) virulence in mice, despite only minor replication defects being associated with mutant viruses
in vitro (
19,
27). In the context of infection with the HKU-39849 strain of SARS-CoV, macrodomain mutation rendered virus highly sensitive to IFN, although this same phenotype was not shown in the MHV studies (
19,
20,
27). That study also found increased levels of IFN and CXCL-10 in macrodomain mutant-infected 293 cells (
20). The macrodomain has also been analyzed in other viruses. The Sindbis virus macrodomain was required for replication in neuronal cells and virulence in 2-week-old mice (
28). Also, while the role of the macrodomain in HEV infection has not been specifically analyzed, it was shown to block interferon production when expressed in isolation (
29). Taken together, those studies suggest that the viral macrodomains impact host innate immune pathways to promote disease. While it has been clearly demonstrated that the CoV macrodomain is essential for virulence in MHV, it remains unclear whether this domain is also critical for virulence in pathogenic human respiratory CoV infection. Here we used a mouse-adapted SARS-CoV (MA15) that causes a lethal infection in BALB/c mice (
30). We developed a series of recombinant viruses with mutations in the ADP-ribose binding pocket of the macrodomain, which we predicted would reduce or eliminate enzymatic activity, and confirmed this prediction in an
in vitro deMARylation assay. We found that all of these mutant viruses were highly attenuated and unable to cause lung disease in mice, while
in vitro replication was unaffected. In addition, we found that the catalytic activity of the macrodomain was required to repress cytokine expression both
in vivo and
in vitro. Our results suggest that the SARS-CoV macrodomain functions to suppress the expression of innate immune genes and promote virulence.
DISCUSSION
Results presented here demonstrate that the SARS-CoV nsp3 macrodomain is critical for virulence. It was also required for optimal virus replication
in vivo and cytokine repression both
in vivo and
in vitro. These results are consistent with reports demonstrating an important
in vivo role for CoV macrodomains. Specifically, catalytic mutants with mutations of the macrodomain in murine MHV and in Sindbis virus were unable to cause severe hepatitis and encephalitis, respectively (
19,
27,
28). However, in addition to demonstrating reduced replication
in vivo, our results indicate a role for the conserved SARS-CoV macrodomain in suppressing the early IFN and proinflammatory cytokine response and promoting pathological changes, such as edema, in the lung and lethality in infected mice. Specifically, SARS-CoV N1040A, a virus devoid of macrodomain catalytic activity, induced significantly elevated expression levels of IFN, ISGs, and other cytokines at 16 to 24 hpi. Importantly, mice coinfected with both wild-type and N1040A had better outcomes and increased IFN and proinflammatory cytokine expression than mice infected with wild-type virus, despite the presence of similar virus titers at 24 hpi. This suggests that an early innate immune response plays an important role in protecting mice from lethal disease. However, dually infected mice were not completely protected from infection, suggesting that the ability of the macrodomain to augment virus load
in vivo also plays a significant role in its ability to promote disease.
It has long been known that SARS-CoV is able to repress the cellular IFN response, and a number of SARS-CoV proteins have been shown to block the IFN response in isolation or
in vitro (
43–47). This report presents the first example of a SARS-CoV mutant virus whose activity leads to an elevated level of antiviral cytokines
in vivo and
in vitro, thus clearly demonstrating a prominent role for the macrodomain in inhibiting the early innate immune response.
Notably, we were unable to definitively address the role of IFN-I in protection using IFNAR
−/− mice, due to IFN having dual roles during infection. We recently showed that a dysregulated IFN response enhanced virulence of SARS-CoV (
37). SARS-CoV strongly represses IFN production initially following infection, and mice are completely protected from disease if exogenous IFN is given before peak replication, demonstrating an important role for early IFN-I production in protection. In the absence of exogenous IFN, IFN-I is rapidly produced at later times and recruits inflammatory monocytes to the lung. There, these monocytes produce additional proinflammatory cytokines, cause vascular leakage, and impair virus specific T-cell responses, ultimately leading to lethality (
37). Consequently, SARS-CoV is unable to cause disease in IFNAR
−/− mice. It is currently unknown how an early IFN/cytokine response protects mice from SARS-CoV-mediated disease. Identifying the precise mechanism of cytokine-mediated protection from a lethal SARS-CoV infection will help us understand how the innate immune system responds to and protects animals from CoV infection.
Another compelling issue is that of exactly how the SARS-CoV macrodomain suppresses innate immune gene expression. Until recently, it was thought that the primary role for the CoV macrodomain was to dephosphorylate ADR-1′′-phosphate, a by-product of tRNA splicing, to ADR (
14,
19–22,
31). Hence, it was primarily referred to as an ADRP (ADPR-1′′-phosphatase). However, this intermediate has never been detected during a CoV infection, and there is no known connection between ADR-1′′-phosphate or ADR and the innate immune system. Therefore, we believe it is much more likely that the CoV macrodomain acts as a de-MAR/PARylating enzyme than as an ADRP, and future efforts will be devoted to identifying its target protein(s).
Identifying protein targets of ARTDs is a priority for researchers, and such studies will likely identify many points of cell biology that may be regulated by macrodomains (
48,
49). At this time, a few ADP-ribosylated proteins involved in innate immunity have been identified. It was recently shown that ARTD14, also known as Ti-PARP, ADP-ribosylated TBK-1, which led to inhibition of the IFN response (
9). ARTD12 has been shown to localize to stress granules and interact with TRIF. Furthermore, its catalytic activity enhances NF-kB-dependent gene expression and blocks both host and virus translation during alphavirus infection (
8,
50). Also, ARTDs located in stress granules are able to ADP-ribosylate Argonaute proteins, which leads to blocking of RNA interference (RNAi), leading to the increased translation of ISGs (
4,
5). Finally, ARTD15 was shown to ADP-ribosylate PERK and IRE1α, 2 endoplasmic reticulum (ER) stress sensors required for the induction of the unfolded-protein response (UPR) (
51). ARTDs are also auto-ADP-ribosylated, which may impact their specific interactions with proteins in these pathways, making them potential macrodomain targets as well. As ARTDs are often found in stress granules, it will be of interest to determine if nsp3 localizes to these sites, in addition to replication compartments.
In addition, viral proteins may be ADP-ribosylated to modify their functions. Recently, two influenza A virus proteins, PB2 and PA, were shown to be PARylated and targeted for ubiquitin-dependent degradation when overexpressed. This degradation was countered by the coexpression of the PB1 protein of influenza A virus, explaining why this degradation does not occur during infection (
52). As the SARS-CoV macrodomain was unable to block cytokine expression in isolation, it is possible that it removes antiviral ADP-ribosylation from SARS-CoV proteins. Furthermore, nsp3 is a transmembrane protein mostly localized within virus replication complexes, which may limit its access to ADP-ribosylated cellular proteins. While several viral proteins could be targeted by the macrodomain, likely targets include the PL
pro domain of nsp3, which is located in close proximity to the macrodomain and contains deubiquitinase activity that can repress innate immune signaling (reviewed in reference
53), and the N protein, which binds to nsp3 and is also able to block innate immune signaling in overexpression (
54,
55). Experiments are under way to identify specific protein targets of the SARS-CoV macrodomain.
The nsp3 macrodomain is completely conserved across the
Coronavirinae subfamily, suggesting that it plays a key role in the CoV lifecycle. Recent work, including this study, has clearly shown that this domain is critical for the virulence of CoVs and their ability to suppress the innate immune response (
19,
20,
27). Identifying the molecular targets of the CoV macrodomains will further enhance our understanding of how CoVs evade the innate immune response and address the issue of why this domain has been conserved throughout coronavirus evolution. Resolution of such issues will improve our ability to identify new strategies and targets for antiviral therapies.
MATERIALS AND METHODS
Cell culture, plasmids, and reagents.
Vero E6, Huh-7, 293T, and HeLa cells were grown in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS), and Calu-3 2B4 cells (Kent Tseng, University of Texas Medical Branch) were grown in MEM supplemented with 20% FBS as previously described (
41). Codon-optimized sMacro (nucleotides 3262 to 3783 of SARS-CoV MA15) and MERS-CoV ORF4a were synthesized and cloned into pUC57 (GenScript). The sMacro and GFP sequences were PCR amplified and ligated into a linearized pcDNA3 plasmid using In-Fusion (Invitrogen) cloning. The ORF4a sequence was PCR amplified and then restriction digested and ligated into the pLKO plasmid. The resulting constructs were confirmed by restriction digestion, PCR, and direct sequencing. Human IFN-α (B/D) and IFN-β were purchased from PBL (Piscataway, NJ). High-molecular-weight poly(I-C) and poly(dA-dT) were purchased from InvivoGen (San Diego, CA). Cells were transfected with either Polyjet (Amgen, Thousand Oaks, CA) or Lipofectamine 2000 (Fisher Scientific, Waltham, MA) per the instructions of the manufacturers.
Mice.
Pathogen-free BALB/c mice were purchased from the National Cancer Institute (Frederick, MD) or Jackson Laboratories (Bar Harbor, ME). IFNAR−/− mice on a BALB/c background were obtained from Joan Durbin (Rutgers-New Jersey Medical School). Mice were bred and maintained in the animal care facility at the University of Iowa. Animal studies were approved by the University of Iowa Institutional Animal Care and Use Committee (IACUC) and met stipulations of the Guide for the Care and Use of Laboratory Animals.
Generation of recombinant pBAC-MA15 constructs.
All recombinant pBAC-MA15 constructs were created using Red recombination (see primers in
Table S1 in the supplemental material). The recombinant BAC with the N1040A mutation (AA3382-3383GC) was engineered using a GalK/kanamycin dual marker cassette that was previously described (
34,
56). Additional BACs with point mutations in the nsp3 macrodomain were engineered using the Kan
r-I-SceI marker cassette for dual positive and negative selection as previously described (
27,
33). Final BAC DNA constructs were analyzed by restriction enzyme digestion, PCR, and direct sequencing for isolation of correct clones.
Reconstitution of recombinant pBAC-MA15-derived virus.
All work with MA15 virus was conducted in the University of Iowa biosafety level 3 (BSL3) Laboratory Core Facility. Approximately 10
6 Vero E6 cells were transfected with 1 μg of pBAC-MA15 DNA using Lipofectamine 2000 (Fisher Scientific) as a transfection reagent. Two separate bacterial clones were used for the N1040A mutation. Viral plaques were evident by 72 to 96 h after transfection. Then, recombinant virus underwent 2 rounds of plaque purification followed by 2 amplification steps prior to its use. The resulting BAC-derived recombinant viruses used in this study are listed in
Table S1.
Virus infection.
Vero-E6 or Calu-3 2B4 cells were infected at the indicated MOIs. Infected cells and supernatants were collected, and titers were determined on Vero E6 cells. Mice were lightly anesthetized using isoflurane and were intranasally infected with 3 × 104 PFU in 50 μl DMEM. To obtain tissue for virus titers, mice were euthanized at different days postchallenge, lungs were removed and homogenized in phosphate-buffered saline (PBS), and titers were determined on Vero E6 cells. Virus titers are represented as numbers of PFU/lung. Two separate clones of the N1040A virus were used in independent experiments.
Immunoblotting.
Total cell extracts were lysed in sample buffer containing SDS, protease and phosphatase inhibitors (Roche, Basel, Switzerland), β-mercaptoethanol, and a universal nuclease (Fisher Scientific). Proteins were resolved on an SDS polyacrylamide gel, transferred to a polyvinylidene difluoride (PVDF) membrane, hybridized with a primary antibody, reacted with an infrared (IR) dye-conjugated secondary antibody, visualized using a Li-COR Odyssey Imager (Li-COR, Lincoln, NE), and analyzed using Image Studio software. Primary antibodies used for immunoblotting included anti-N polyclonal antibody (IMG548; Imgenex, San Diego, CA); anti-SARS nsp3 polyclonal antibody (kindly provided by Mark Denison, Vanderbilt University, Nashville, TN); and anti-actin monoclonal antibody (clone AC15; Abcam, Inc., Cambridge, MA). Secondary IR antibodies were purchased from Li-COR.
Immunofluorescence.
To analyze intracellular protein localization by immunofluorescence, cells grown on glass coverslips were fixed in 4% paraformaldehyde, permeabilized with 0.1% Triton X-100 for 15 min, blocked with 1% goat serum–PBS, incubated with anti-SARS nsp3 polyclonal antibody in blocking buffer, and subsequently labeled with secondary antibody. Labeled cells were counterstained with TO-PRO-3 (Fisher Scientific) to visualize the nuclei and then mounted on slides with Vectashield antifade reagent (Vector Laboratories, Burlingame, CA). Images were captured using a Leica STED SP8 confocal laser scanning microscope, and images were analyzed using LAS X software.
Lung cell preparation.
Lung cells were prepared as previously described (
37).
Flow cytometry.
For surface staining, cells derived from the lungs were treated with Fc block (CD16/32, 2.4G2) and then incubated with specific MAbs or isotype controls. The monoclonal antibodies used for these studies, CD45-PECy7/FITC (30-F11), CD11b-Percp-Cy5.5 (M1/70), Ly6C-APC (AL-21), Ly6G-FITC (1A8), F4/80-PE (BM8), CD11c-eFluor 450 (N418), PerCp-Cy5.5-anti-IA/IE (M5/114.15.2), CD3-APC (145-2C11), CD4-e450 (RM4-5), CD8-FITC (53-6.7), and NKP46-PECy7 (29A1.4), were purchased from BD Biosciences or eBiosciences. Cells were analyzed using a fluorescence-activated cell sorter (FACS) Verse flow cytometer (BD Biosciences, Mountain View, CA). All flow cytometry data were analyzed using FlowJo software (Tree Star, Inc., Ashland, OR).
Real-time quantitative PCR (RT-qPCR) analysis.
RNA was isolated from tissue culture cells or from perfused lung homogenates using Trizol (Fisher Scientific) and treated with RNase-free DNase (Promega, Madison, WI), and cDNA was prepared using Moloney murine leukemia virus (MMLV) reverse transcriptase per the manufacturer’s instructions (Invitrogen). RT-qPCR was performed using an Applied Biosystems 7300 real-time PCR system (Applied Biosystems, Foster City, CA) and RT
2 2× SYBR green qPCR master mix (Qiagen). Primers used for qPCR are listed in
Table S2. Cycle threshold (
CT) values were normalized to those of the housekeeping gene encoding hypoxanthine phosphoribosyltransferase (HPRT) by the following equation: Δ
CT =
CT(gene of interest) −
CT(HPRT). All results are shown as a ratio to HPRT calculated as −2
ΔCT.
Lung histology and immunohistochemistry.
Lungs were removed, fixed in zinc formalin, and paraffin embedded. Sections were stained with hematoxylin and eosin and examined by light microscopy. For immunohistochemical staining of tissues, 10% formalin-fixed, paraffin-embedded lung sections (6 to 7 μm in thickness) were microwaved in 10 mM citrate buffer (pH 6.0) for 5 min. Endogenous peroxidase was inactivated with 3% hydrogen peroxide (H2O2) at room temperature for 10 min. Sections were then incubated (overnight at 4°C) with rabbit anti-N protein (IMG548; Imgenex, San Diego, CA) (1:1,000). Secondary labeling with biotinylated goat anti-rabbit IgG (1:200) was performed at room temperature for 1 h, followed by color development with 3,3′-diaminobenzidine for 3 min.
Protein expression and purification.
SARS-CoV macrodomains were expressed and purified as described previously, with minor adjustments (
57). Briefly, C41 (DE3) cells carrying pET vector encoding SARS-CoV macrodomain or point mutants were grown in TB media to an optical density at 600 nm (OD
600) of ~0.6, and the cultures were cooled to 16°C and induced with 50 µM IPTG (isopropyl-β-
d-thiogalactopyranoside) for 20 h. Harvested pellets were stored at −80°C until purification. The pellets were lysed with 1× BugBuster reagent–50 mM Tris (pH 7.6)–150 mM NaCl–15 mM imidazole supplemented with lysozyme, Benzonase, and phenylmethylsulfonyl fluoride (PMSF). The lysates were clarified by centrifugation, and the supernatants were subjected to nickel-nitrilotriacetic acid (Ni-NTA) resin. Proteins were eluted using 50 mM Tris (pH 7.6)–150 mM NaCl–300 mM imidazole, dialyzed to 25 mM Tris (pH 7.6)–150 mM NaCl–1 mM dithiothreitol (DTT), frozen in liquid nitrogen, and stored at −80°C.
ARTD10 de-ADP-ribosylation assay.
ARTD10 de-ADP-ribosylation assays were performed as described previously (
16,
23). Briefly, glutathione
S-transferase (GST)-ARTD10 catalytic domain (cd) was bound to glutathione Sepharose beads in reaction buffer (50 mM HEPES [pH 7.2], 150 mM NaCl, 0.2 mM DTT, 0.02% NP-40). A 15-μl volume of bead slurry per reaction was used with 7.5 µl of ~0.5 mg/ml GST-ARTD10cd. The beads were washed with reaction buffer, and an automodification reaction was performed at 37°C for 20 min with 5 to 10 kBq
32P-labeled NAD
+ per reaction, at a final NAD
+ concentration of 1 µM. The beads were washed with reaction buffer and incubated with 0.5 µM SARS-CoV macrodomains at 37°C for 30 min. Reactions were separated by SDS-PAGE, subjected to Coomassie staining, dried, and exposed to autoradiography films. The films were quantified using ImageJ.
Statistics.
A Student’s t test was used to analyze differences in mean values between groups. All results are expressed as means ± standard errors of the means (SEM). Differences in survival were calculated using a Kaplan-Meier log-rank test. P values of ≤0.05 were considered statistically significant (*, P < 0.05; ** P < 0.01; *** P < 0.001; n.s., not significant).