Guanylate-binding protein 2 orchestrates innate immune responses against murine norovirus and is antagonized by the viral protein NS7

Noroviruses are the main causative agents of acute viral gastroenteritis, but the host factors that restrict their replication remain poorly identified. Guanylate-binding proteins (GBPs) are interferon (IFN)-inducible GTPases that exert broad antiviral activity and are important mediators of host defenses against viral infections. Here, we show that both IFN-γ stimulation and murine norovirus (MNV) infection induce GBP2 expression in murine macrophages. Results from loss- and gain-of-function assays indicated that GBP2 is important for IFN-γ–dependent anti-MNV activity in murine macrophages. Ectopic expression of MNV receptor (CD300lf) in human HEK293T epithelial cells conferred susceptibility to MNV infection. Importantly, GBP2 potently inhibited MNV in these human epithelial cells. Results from mechanistic dissection experiments revealed that the N-terminal G domain of GBP2 mediates these anti-MNV effects. R48A and K51A substitutions in GBP2, associated with loss of GBP2 GTPase activity, attenuated the anti-MNV effects of GBP2. Finally, we found that nonstructural protein 7 (NS7) of MNV co-localizes with GBP2 and antagonizes the anti-MNV activity of GBP2. These findings reveal that GBP2 is an important mediator of host defenses against murine norovirus.

Human norovirus (HuNV) infection is the major cause of epidemic nonbacterial gastroenteritis worldwide (1,2). Currently, there is no vaccination or specific antiviral treatment available. Clinical management is limited to supportive care and oral rehydration, and HuNV imposes a heavy global health burden (3). Thus, defining improved anti-HuNV therapy represents an urgent clinical need. Research into HuNV infection, however, has been hampered by the lack of robust experimental models. Murine norovirus (MNV), capable of replicating in both cell culture and small-animal models, shares similar traits with HuNV in structural and genetic features and has thus been widely used as a surrogate model (4,5). The recent discovery of the MNV receptor (CD300lf) has now enabled MNV infection in human cells by ectopically expressing this receptor. This has resulted in improved understanding of the mechanisms underlying viral replication and the identification of cellular factors as potential antiviral targets (6,7).
Noroviruses are nonenveloped, positive single-stranded RNA viruses belonging to the Caliciviridae family. The genome is about 7.5 kb in length and encodes three or four ORFs (3,8). The 5Ј-proximal ORF1 encodes a polyprotein that is posttranslationally cleaved into six nonstructural proteins (NS1/2 to NS7). ORF2 and ORF3, encoding the major and minor structural proteins, are referred as VP1 and VP2, respectively, which are translated from a subgenomic RNA. VP2 has been reported to possess important functions in viral replication and virion stability but may also corrupt host immune response (9). Specific for MNV, ORF4 overlaps with ORF2 and produces an additional protein called virulence factor (VF1). VF1 has been reported to antagonize innate immune response to MNV infection (8,9). MNV NS1/2 protein is associated with cell tropism and mediates resistance to interferon-(IFN-)-mediated clearance used for treating persistent viral infection (10). NS7 is the viral RNA-dependent RNA polymerase that can also modulate innate immune response (11,12). However, the exact interactions of these viral proteins with host innate antiviral immunity remain poorly understood.
IFN-mediated innate immune responses provide the first line of host defense against viral infections. Specific viral components sensed by pathogen recognition receptors including Toll-like receptors and RIG-I-like receptors lead to IFN production (13,14). The released IFNs bind to their cognate receptors to activate the Janus kinase (JAK)/signal transducer and activator of transcription (STAT) signaling pathway, resulting in the transcription of hundreds of interferon-stimulated genes (ISGs). A subset of ISGs are considered as the ultimate antiviral effectors limiting viral replication and spread (15). Currently, only a few ISGs have been identified to inhibit MNV infection, including interferon regulatory factor 1 (IRF1) and interferonstimulated gene 15 (ISG15) (16,17). Thus, it is largely unknown which factors are important for effective cell-autonomous defense against MNV infection.
Interesting candidate molecules to act in the defense against MNV infection are the guanylate-binding proteins (GBPs). They are members of the superfamily of IFN-inducible GTPases with many typical characteristics of ISGs. These proteins are composed of three distinct domains, including the N-terminal globular GTPase domain containing all motifs responsible for nucleotide binding and hydrolysis (G domain), the following helical part presenting as the middle domain (M domain), and the C-terminal GTPase effector domain (E domain) (18,19). To date, seven human GBPs and 11 murine GBPs as well as two mouse pseudogenes encoding GBPs have been identified (20,21 cro ARTICLE terial and protozoan pathogens (18,22,23), GBPs (e.g. GBP1, GBP2, and GBP5) have been reported to exert broad antiviral activity against HIV, Zika virus, hepatitis C virus (HCV), classical swine fever virus (CSFV), and influenza virus (24 -27). They are capable of regulating inflammasome activation (28,29), and this plays a role in controlling replication of rotavirus and influenza virus (30,31). A recent study has indicated that IFN-inducible GTPases (including GBP2) are important for IFN-␥-mediated inhibition of MNV replication and help to block the formation of viral replication complexes (32). In this study, we investigated in detail the regulation and function of GBP2 in MNV infection, revealing an important role in orchestrating host response to MNV infection.

IFN-␥ stimulation up-regulates GBP2 expression in mouse macrophages
IFN-␥ signaling can potently inhibit MNV replication and induce expression of GBPs. In this study, we observed up-reg-ulation of GBP2 following IFN-␥ treatment in RAW264.7 and J774A.1 cells (Fig. 1, A and B), two murine macrophage cell lines susceptible to MNV infection. This effect appears to be mediated by canonical JAK/STAT-mediated IFN signaling as the pharmacological compound JAK inhibitor 1 blocked IFN-␥induced GBP2 expression at both mRNA ( Fig. 1, C and D) and protein levels (Fig. 1E) in RAW264.7 and J774A.1 cells. Furthermore, IFN-␥ failed to stimulate GBP2 expression in STAT1deficient mouse embryonic fibroblasts (MEFs) (Fig. 1, F and G). We thus concluded that up-regulation of GBP2 is part of the canonical IFN-␥ response.

GBP2 enhances IFN-␥-mediated inhibition of MNV-1 replication in mouse macrophages
We next examined whether MNV infection could induce GBP2 expression. Interestingly, inoculation of both in RAW264.7 cells (Fig. 2, A and B) and J774A.1 cells (Fig. 2C) with MNV-1 up-regulated GBP2. To investigate the functional implication, we silenced GBP2 expression in RAW264.7 cells by h, respectively. mRNA (C, n ϭ 6; D, n ϭ 6) and protein (E) levels of STAT1 and GBP2 were analyzed by qRT-PCR and Western blotting, respectively. F, WT and STAT1 Ϫ/Ϫ MEFs were untreated or treated with IFN-␥ (100 units/ml) for 6 h. The mRNA level of GBP2 was analyzed by qRT-PCR (n ϭ 6). G, expression of STAT1 and GBP2 in WT and STAT1 Ϫ/Ϫ MEFs treated with IFN-␥ (100 units/ml) for 6 h were analyzed by Western blotting. Data in C and D were normalized to the untreated and JAK inhibitor-treated cells (both set as 1). Data in F were normalized to untreated WT and STAT1 Ϫ/Ϫ MEFs, respectively (both set as 1). **, p Ͻ 0.01. ␤-Actin was used as a loading control. Error bars, S.D.

GBP2 defends against MNV infection
lentiviral shRNAs. The efficacy of this approach was confirmed both with and without IFN-␥ treatment (Fig. 3, A-C). Although GBP2 silence per se has no major effect on MNV-1 replication, this strategy decreased anti-MNV activity of IFN-␥ (Fig. 3D). This is in accordance with a study demonstrating that MNV replicates to a high level in GBP2-deficient murine bone marrow-derived macrophages (BMDMs) in the presence of IFN-␥ (33). Thus, GBP2 is necessary for the anti-MNV activity of IFN-␥. Furthermore, we succeeded with stable GBP2 overexpression employing a lentiviral vector (Fig. 4, A and B). This did not affect MNV-1 RNA level (Fig. 4C), but it augmented IFN-␥-mediated inhibition of MNV-1 RNA transcription and NS1/2 protein expression (Fig. 4, D-G) as well as the viral titers (Fig. 4H). Thus, GBP2 is an important mediator of IFN-␥-triggered inhibition of MNV-1 replication in murine macrophages.

GBP2 overexpression restricts MNV-1 replication in human epithelial cells
Discovery of the MNV receptor (CD300lf) has enabled MNV infection in human cells by ectopic expression of this receptor (6). We found that transfection with FLAG-tagged CD300lf allowed MNV-1 replication in human HEK293T cells (Fig. 5A). Studies have reported that the anti-MNV ability of IFN-␥ is impaired in human HAP1 cells with complete loss of GBPs (33). To further examine whether GBP2 exerts anti-norovirus activity in human cells, HEK293T cells were co-transfected with FLAG-tagged CD300lf and different concentrations of GBP2 vectors for 24 h and subsequently infected with MNV-1 for another 24 h. Mirroring the results in murine macrophages, GBP2 overexpression alone can inhibit MNV-1, shown at both viral RNA (Fig. 5B) and protein levels (Fig. 5C). Furthermore, transfection with Myc-tagged GBP2 into CD300lf-expressing HEK293T cells confirmed the inhibitory effects on MNV-1 (Fig. 5, D-F). Thus, a role for GBP2 in constraining MNV rep-lication is not restricted to murine macrophages but also extends to human epithelial cells.
Of note, several ISGs with broad antiviral activity, including IRF1, RIG-I, and MDA5, have been shown to activate the transcription of many other ISGs (15,(33)(34)(35). To address whether GBP2 also exerts its action in a similar manner, we measured the transcriptional level of several selected ISGs in RAW264.7 cells in the absence or presence of stable GBP2 overexpression. Overexpression of GBP2 did not affect the expression of these tested ISGs (Fig. S1A). This was further confirmed in HEK293T cells (Fig. S1B). These results suggest that the antiviral activity of GBP2 is independent of ISG induction.

The N terminus of GBP2 is essential for the anti-MNV activity
Structural analysis of human GBP1 has revealed three distinct domains, including the N-terminal globular GTPase domain (G domain), the following two-helix part presenting as the middle domain (M domain), and the C-terminal GTPase effector domain (E domain) (19). Based on the human GBP1 structure, we predicted and modeled the mouse GBP2 protein structure (Fig. S2) by using the online software SWISS-MODEL (RRID:SCR_018123) and mapped GBP2 into three corresponding domains (Fig. 6A). To investigate which domain is responsible for the antiviral actions, we constructed FLAG-tagged truncated mutants of GBP2 and verified their expression in HEK293T cells (Fig. 6B). Their anti-MNV activities were examined in HEK293T cells transfected with FLAG-tagged CD300lf. We found that the G and GM domains of GBP2 inhibited viral replication shown at both viral RNA and protein levels ( Fig. 6, C and D). In contrast, the M, ME, and E domains did not exert anti-MNV activity (Fig. 6, C and D). To further confirm, we established stable expression of GBP2 truncated mutants in RAW264.7 cells (Fig. 6E). These cells were infected with MNV-1 and then treated with IFN-␥ for 24 h. We found that the G and GM domains of GBP2 enhanced IFN-␥-mediated

GBP2 defends against MNV infection
anti-MNV activity, whereas the remaining domains appeared not to influence viral replication (Fig. 6, F-H). Taken together, the N terminus of GBP2 is indispensable for restricting MNV-1 replication in human cells and augmenting IFN-␥-mediated anti-MNV ability in murine macrophages.
To determine the antiviral effects of these two GBP2 mutants, we transfected the constructed mutants and viral receptor vectors into HEK293T cells and then infected with MNV-1 for 24 h. Compared with the WT GBP2, these two mutants failed to restrict MNV-1 at both viral RNA (Fig. 7C) and NS1/2 protein levels (Fig. 7D). Measuring viral titers in the supernatants of infected cells further confirmed these results (Fig. 7E). Thus, the Arg-48 and Lys-51 residues of GBP2 are essential for its anti-MNV activity.

MNV NS7 antagonizes the antiviral activity of GBP2
Studies have reported that the viral replicase has a negative regulation on porcine GBP1-mediated inhibition of CSFV (27) and human GBP1-mediated restriction of HCV (36). Thus, we next determined whether the MNV replicase NS7 has a role on GBP2-mediated antiviral activity. We co-transfected pFLAG-GBP2 and pFLAG-NS7 into HEK293T cells. We found that GBP2 expression was not affected by NS7 expression (Fig.  8A), and NS7 did not affect viral NS1/2 protein expression (Fig.  8B). Moreover, confocal microscopy indicated that besides localization in the nucleus, NS7 could co-localize with GBP2 in the cytoplasm of the cells (Fig. 8, C and D). To further investigate NS7 on GBP2-mediated antiviral activity, we cotransfected pFLAG-CD300lf, pFLAG-GBP2, and pMyc-NS7 into HEK293T cells and then infected with MNV-1 for 24 h. We found that NS7 attenuated GBP2-mediated anti-MNV effects shown at both viral RNA and NS1/2 protein levels (Fig. 8, E and  F). These results indicated a potential antagonistic effects of NS7 on GBP2-mediated anti-MNV activity.

Discussion
GBPs as a group of IFN-induced proteins are essential for innate immune response against intracellular bacterial, viral, and protozoan pathogens (18,22,25). With respect to viruses, it has been shown that human GBP1 can restrict replication of vesicular stomatitis virus and encephalomyocarditis virus (39), whereas several GBPs have been linked to host defense against HIV, CSFV, and influenza virus (25,27,40). A recent study suggested potential interactions between this family of GTPases and norovirus replication (32). Here, we further show that GBP2 effectively responds to and defends murine norovirus infection. Our findings fit well with the increasing momentum and support for the notion that IFN-inducible GTPases are crucial for cell-autonomous host defense against viral infection in general and norovirus in particular (7). We first demonstrated that GBP2 mediates IFN-␥-triggered anti-MNV activity in murine macrophages, whereas GBP2 alone is not sufficient to inhibit MNV. By exploiting ectopic expression of the viral receptors (6), we conferred susceptibility of human HEK293T cells to MNV infection. We further demonstrated that GBP2 alone is sufficient to potently inhibit MNV in human epithelial cells, without requiring the presence of IFN-␥. We speculate that the disparity in requiring IFN-␥ may be attributed to the differences in species, cell types, and the expression patterns of GBP2.
Although ISGs are known as antiviral effectors, their modes of actions are diverse, including direct and indirect antiviral actions. Some ISGs have been linked to immunity against noro-

GBP2 defends against MNV infection
virus infection. IRF1 has been reported to contribute to IFN-␥mediated inhibition of MNV replication in macrophages (16). This may be an indirect effect, as shown in the setting of hepatitis E virus (HEV) infection. IRF1 activates STAT1 to induce the expression of a wide range of ISGs that eventually inhibit HEV replication (33). ISG15 inhibits an early step of the MNV life cycle upstream of viral genome transcription (17). GBP1 has been reported to restrict DENV replication by modulating NF-B activity, leading to the production of antiviral and proinflammatory cytokines (41). As seen in this study, the inhibition of MNV replication by GBP2 appears to be independent of ISG induction. Interestingly, recent studies have shown an association between GBPs and inflammasome activation. The inflammasome machinery is essential for host defense against viral pathogens (28,29). MNV infection triggers NLRP3 inflammasome activation in primary BMDMs with STAT1 deficiency or in TLR2-primed BMDMs (42). MNV infection persists much longer in NLRP6-deficient compared with WT mice (43). It is thus tempting to suggest that GBP2 functions through inflammasome activation, and this scenario should be investigated in future experimentation.
Structurally, hGBP1 can be mapped into three domains with distinct functionality (19). The N-terminal domain of GBP1 is responsible for the antiviral activity against influenza A virus, HCV, and CSFV infections (27,36,37). The C-terminal domain of mGBP2 dictates the recruitment to the Toxoplasma gondii parasitophorous vacuole and contributes to control of its replication (18). Based on the hGBP1 structure (Protein Data Bank entry 1F5N) (19), we modeled the different domains of mouse GBP2, including the G, M, and E domains. By constructing the truncated mutants of GBP2, we found that the N-terminal G-domain is important for the anti-MNV activity in human epithelial cells and for augmenting IFN-␥-mediated anti-MNV response in murine macrophages.
The Arg-48 and Lys-51 residues within the N-terminal domain are essential for the GTPase activity and antiviral function of GBP1 (27,36). Arg-48 in the P-loop is highly conserved across different GBPs and functions as a GTPase-activating "arginine finger" involved in the multimerization process. The R48A mutant has much weaker GTPase activity (38). Lys-51 (K51A) mutation in mouse GBP2 leads to a nearly complete loss of function, including hydrolysis, dimerization, and nucleotide binding (38). In this study, we found that the R48A and K51A mutants attenuate the anti-MNV effects of GBP2, suggesting the potential requirement of GTPase activity. Viruses have developed sophisticated strategies to evade host defense (44).

GBP2 defends against MNV infection
MNV NS1/2 interacts with host protein VAPA to enhance viral replication (45), and NS3 interacts with microtubule-associated protein GEF-H1, which plays a role in immune detection of viral replication (46). MNV NS7 is the viral replicase and catalyzes replication of the viral genome. NS7 presents a diffused pattern both in cell cytoplasm and nucleus (47,48). In this study, we revealed that NS7 co-localizes with GBP2 in the cytoplasm by transient expression and antagonizes GBP2-mediated anti-MNV activity. Viral replicases including NS5B of HCV and NS5A of CSFV, and NS1 of influenza A virus have been reported to interact with GBP1 (36, 37) and attenuate GBP1mediated antiviral activity (27,36). Thus, the potential interaction of NS7 with GBP2 and the possible inhibitory effect on GTPase activity of GBP2 will be interesting subjects for further study.
In summary, MNV-1 infection activates the expression of GBP2, an IFN-inducible GTPase. GBP2 orchestrates innate immune defense against MNV independent of its N terminus.
However, MNV NS7 can co-localize with GBP2 in the cytoplasm and antagonize GBP2-mediated anti-MNV activity. These findings shed new light on norovirus-host interactions and shall be helpful for better understanding the pathogenesis and developing new antiviral strategies.

Reagents
Mouse IFN-␥ (ab9922, Abcam) was dissolved in PBS. Stocks of JAK inhibitor 1 (SC-204021, Santa Cruz Biotechnology, Inc.) were dissolved in DMSO (Sigma) with a final concentration of 5 mg/ml. Puromycin (P8833, Sigma) was dissolved in PBS with a final concentration of 10 mg/ml. The Q5 site-directed mutagenesis kit (New England Biolabs) was used. GBP2 antibody (11854-1-AP) was purchased from Proteintech. STAT1 (catalog no. 9172) antibody was purchased from Cell Signaling Technology. Rabbit polyclonal antisera to MNV NS1/2 was Figure 7. The Arg-48 and Lys-51 residues of GBP2 are critical for its anti-MNV activity. A, schematic representation of the GBP2 mutations. B, expression of GBP2 mutations. The indicated expression plasmids were transfected into HEK293T cells for 24 h, and the cell lysates were analyzed by Western blotting (WB) using antibodies against FLAG tag and ␤-actin. HEK293T cells were transfected with pFLAG-CD300lf (0.5 g) and pFLAG-GBP2 (WT), pFLAG-GBP2 (R48A), pFLAG-GBP2 (K51A), or empty vectors for 24 h and then infected with MNV-1 at an MOI of 1 for 24 h. The MNV RNA (C) (n ϭ 4 -6) and NS1/2 protein (D) levels were analyzed by qRT-PCR and Western blotting, respectively. E, the viral titers in the supernatants were examined by a TCID 50 assay (n ϭ 6). Data in C and E were normalized to the empty virus (EV) control (set as 1). *, p Ͻ 0.05; **, p Ͻ 0.01; ns, not significant. ␤-Actin was used as a loading control. Error bars, S.D.

GBP2 defends against MNV infection
of Medicine) into RAW264.7 cells. The MNV-1 cultures were purified, aliquoted, and stored at Ϫ80°C for all subsequent experiments. The MNV-1 stock was quantified three independent times by the 50% tissue culture infective dose (TCID 50 ).

TCID 50
MNV-1 was quantified by a TCID 50 assay. Briefly, 10-fold dilutions of MNV-1 were inoculated into RAW264.7 cells grown in a 96-well tissue culture plate at 1,000 cells/well. The plate was incubated at 37°C for another 5 days, followed by observing the cytopathic effect of each well under a light scope. TCID 50 was calculated by using the Reed-Muench method.
HEK293T cells were transfected with various plasmids at the indicated concentrations using FuGENE HD transfection reagent (catalogue no. E2311; Promega) according to the manufacturer's instructions. Where necessary, the appropriate empty vector was used to maintain a constant amount of plasmid DNA per transfection. At 6 h post-transfection, fresh Dulbecco's modified Eagle's medium containing 10% FCS replaced the transfection mixture, and the cells were incubated at 37°C.

Silencing or overexpressing mouse GBP2 by lentiviral vectors
Lentiviral pLKO.1 knockdown vectors (Sigma-Aldrich) targeting mouse GBP2 were obtained from the Erasmus Biomics Center. The lentiviral pseudoparticles were produced in HEK293T cells as described previously (50). After a pilot study, the shRNA vectors exerting optimal gene knockdown were selected. These shRNA sequences are listed in Table S2. Stable gene knockdown cells were generated after lentiviral vector transduction and puromycin (5 g/ml; Sigma) selection. For stable expression, GBP2 WT and truncated overexpression lentiviral vectors were used to generate the GBP2 stable expression RAW264.7 cell lines. Meanwhile, control shRNA and the lentiviral empty vectors were also used as control, respectively.

qRT-PCR
Total RNA was isolated with a Macherey NucleoSpin RNA II Kit (Bioke, Leiden, The Netherlands) and quantified with a Nanodrop ND-1000 (Wilmington, DE). cDNA was synthesized from 500 ng of RNA using a cDNA synthesis kit (TaKaRa Bio, Inc., Shiga, Japan). The cDNA of all targeted gene transcripts were quantified by SYBR Green-based (Applied Biosystems) real-time PCR on the StepOnePlus TM system (Thermo Fisher Scientific) according to the manufacturer's instructions. Human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and murine GAPDH genes were used as reference genes to normalize gene expression. The relative expression of targeted gene was calculated as 2 Ϫ⌬⌬CT , where ⌬⌬C T ϭ ⌬C T sample Ϫ ⌬C T control (⌬C T ϭ C T (targeted gene) Ϫ C T (GAPDH)). All primer sequences are listed in Table S3.

Western blotting
Cultured cells were lysed in Laemmli sample buffer containing 0.1 M DTT and heated 5 min at 95°C and then loaded onto a 10% SDS-polyacrylamide gel. Proteins were electrophoretically transferred onto a polyvinylidene difluoride membrane (pore size, 0.45 M; Invitrogen) for 2 h with an electric current of 250 mA. Subsequently, the membrane was blocked with a mixture of 2.5 ml of blocking buffer (Odyssey) and 2.5 ml of PBS containing 0.05% Tween 20 for 1 h, followed by overnight incubation with primary antibodies (1:1000) at 4°C. The membrane was washed three times and then incubated with IRDye-conjugated secondary antibody (1:5000) for 1 h. After washing three times, protein bands were detected with the Odyssey 3.0 IR Imaging System (LI-COR Biosciences).

Statistical analysis
Data are presented as the mean Ϯ S.D. Comparisons between groups were performed with the Mann-Whitney test using GraphPad Prism 5.0 (GraphPad Software Inc., La Jolla, CA, USA). Differences were considered significant at a p value of Ͻ0.05.

Data availability
All of the data are contained within the article and supporting materials.