STAT1-Dependent Innate Immunity to a Norwalk-Like Virus

See allHide authors and affiliations

Science  07 Mar 2003:
Vol. 299, Issue 5612, pp. 1575-1578
DOI: 10.1126/science.1077905


Norwalk-like caliciviruses (Noroviruses) cause over 90% of nonbacterial epidemic gastroenteritis worldwide, but the pathogenesis of norovirus infection is poorly understood because these viruses do not grow in cultured cells and there is no small animal model. Here, we report a previously unknown murine norovirus. Analysis of Murine Norovirus 1 infection revealed that signal transducer and activator of transcription 1–dependent innate immunity, but not T and B cell–dependent adaptive immunity, is essential for norovirus resistance. The identification of host molecules essential for murine norovirus resistance may provide targets for prevention or control of an important human disease.

The understanding of human infectious diseases has often depended on identification of animal viruses that provide experimental models to define mechanisms of host resistance. We therefore pursued the observation that mice deficient in signal transducer and activator of transcription 1 (STAT1) and recombination-activating gene 2 (RAG2) (RAG/STAT1–/–mice) (1) sporadically succumbed to a pathogen that could be serially passed by intracerebral (i.c.) inoculation (fig. S1). Lethal infection was associated with encephalitis, vasculitis of the cerebral vessels, meningitis, hepatitis, and pneumonia (fig. S1). The pathogen was more virulent in mice lacking both the interferon αβ and the interferon γ receptors (IFNαβγR–/–) (2) than in wild-type mice and passed through a 0.2-μm filter (fig. S1A). Evaluation of diseased brain for known human and murine pathogens was negative (fig. S1) (3). These data suggested that a previously uncharacterized IFN-sensitive virus was present in diseased RAG/STAT1–/–mice.

To identify the pathogen, we used representational difference analysis (RDA) (4) and obtained sequences homologous to regions of many calicivirus genomes (Fig. 1A) (3), demonstrating that we had identified a previously unknown calicivirus. We determined a consensus sequence of the 7382–base pair polyadenylated genome, which contained the expected three calicivirus open reading frames (ORFs) (Fig. 1B). ORF1 encoded a 182.5-kD polyprotein containing amino acid motifs shared by caliciviruses and picornaviruses. We determined that ORF2 encoded the 58.9-kD capsid protein, because virus-like particles (VLPs) were found in the supernatant of cells infected with a recombinant baculovirus expressing ORF2 (Fig. 2A). ORF3 encoded a 22.1-kD basic protein.

Figure 1

Sequencing and phylogenetic analysis of the MNV-1 genome. (A) Sequencing of MNV-1. RDA was used to identify clones (red lines) homologous to calicivirus sequences that spanned the Norwalk virus genome (black line). Sequences within RDA clones (indicated by asterisks) were used to clone and sequence five fragments (blue lines) of the MNV-1 genome (3). The MNV-1 sequence has been submitted to GenBank and assigned accession numberAY228235. bp, base pairs. (B) Schematic of the MNV-1 genome with its predicted ORFs. The nucleotide positions of the putative NH2 and COOH termini of each ORF are indicated above each box, whereas the numbers for each reading frame relative to ORF1 are indicated below. The locations of amino acid motifs in ORF1 (red box) are indicated: 2C helicase, Gly-X-X-Gly-X-Gly-Lys-Thr (X indicates any amino acid residue); 3C protease, Gly-Asp-Cys-Gly; 3D polymerase, Lys-Asp-Glu-Leu, Gly-Leu-Pro-Ser, Tyr-Gly-Asp-Asp. The putative S and P domains of the ORF2 (yellow box)-encoded capsid protein were identified on the basis of sequence alignments with Norwalk virus. Blue box, ORF3; AAA, 3′ poly-A tail. (C) Alignment of the capsid protein sequence of MNV-1 with capsid protein sequences of representative members of the four genera ofCaliciviridae and several picornaviruses (3).

Figure 2

Visualization and pathogenicity of MNV-1 particles. (A) MNV-1 VLPs and (B) purified infectious virions visualized by negative staining electron microscopy (3). Scale bars indicate 100 nm (left) or 50 nm (right). (C) Survival of RAG/STAT1–/– mice infected i.c. with unpurified or purified MNV-1 (densities of 1.365 to 1.337 g/cm3) or with gradient fractions from mock-infected brain. The P values for mock compared to infected mice are indicated. All statistical analyses were performed with the use of the Mantel-Haenszel test (GraphPad Prism, Version 3.03).

The Caliciviridae are divided into four genera, with the Vesivirus and Lagovirus genera including animal viruses. The Norovirus and Sapovirusgenera contain the human caliciviruses (5). Noroviruses are important human pathogens worldwide and cause more than 90% of all cases of nonbacterial epidemic gastroenteritis (5–7). Phylogenetic analysis of the capsid protein of the murine virus (using Clustal W or PAUP) revealed that the virus is in the genus Norovirus but does not cluster within previously identified genogroups (Fig. 1C). Clustal W alignments using the complete genomic sequence confirmed this conclusion (8). Therefore, we propose the name Murine Norovirus 1 (MNV-1) and the placement of the virus into a new Norovirus genogroup.

We purified MNV-1 from the brain of an IFNαβγR–/–mouse on CsCl gradients (fig. S2). MNV-1 had a buoyant density (1.36 ± 0.04 g/cm3) and size (diameter of 28 to 35 nm) (Fig. 2B) similar to that of other noroviruses (5,9). CsCl-purified MNV-1 was pathogenic; 18 of 18 RAG/STAT1–/– mice inoculated with MNV-1–containing gradient fractions died, whereas mice inoculated with control gradient fractions did not (Fig. 2C). Mice inoculated with gradient-purified virions showed encephalitis, meningitis, cerebral vasculitis, pneumonia, and hepatitis (8). Induction of disease with CsCl-purified MNV-1 strongly argues that MNV-1 is the causative agent of the disease we initially detected and passed (fig. S1).

The availability of a norovirus that infects mice provided us an opportunity to identify components of the immune system required for norovirus resistance. Human norovirus disease is characterized by a short (about 24 hours) incubation period followed by 24 to 48 hours of vomiting and/or diarrhea (5,10). The immune mechanisms involved in resistance to human norovirus disease are not well defined. Nonimmune host factors, such as blood groups, have been associated with susceptibility to infection (11, 12). Infected individuals can develop short-term immunity to homologous viruses, but the development of long-term immunity is questionable, and it is unclear whether seropositive persons are more or less susceptible to disease (13).

We studied whether lack of T and B cell–dependent adaptive immunity in RAG–/– mice predisposed the animals to lethal infection. Wild-type and RAG1–/– (14) or RAG2–/– (15) mice were infected by the i.c., intranasal (i.n.), and peroral (p.o.) routes and followed for 90 days (Fig. 3A). The p.o. and i.n. routes were tested because the physiologic routes of infection for human caliciviruses are oral and respiratory (5). Unexpectedly, MNV-1 infection did not kill RAG–/– mice even after direct i.c. inoculation (Fig. 3A), although these mice are typically highly susceptible to infection with a range of different viruses. Similarly, only a statistically insignificant proportion of RAG–/–animals died within 90 days of p.o. or i.n. infection (Fig. 3A). RAG–/– mice were persistently infected, because they contained high levels of viral RNA in feces and multiple visceral organs (Fig. 4A). Survival of persistently infected RAG–/– mice demonstrated that B and T cell–dependent adaptive immune responses are not required for protection from lethal murine norovirus disease. However, it may be that adaptive immunity, for example after vaccination or natural infection, does influence the outcome of secondary MNV-1 infection.

Figure 3

IFNαβ or IFNγ receptors and STAT1 are required to protect from lethal MNV-1 challenge. Mice were inoculated with MNV-1 (3) or mock-inoculated with the use of 10 μl intracerebrally (ic), 25 μl intranasally (in), or 25 μl perorally (po). (A) Mouse strains that did not show increased mortality. The P value for RAG–/– mice mortality compared with that of wild-type mice inoculated i.n. is 0.2185 and for RAG–/– mice mortality compared with that of wild-type mice inoculated p.o. is 0.3980. dpi, days post-infection. The survival after inoculation with MNV-1 or mock is shown for (B) IFNαβγR–/–, (C) STAT1–/–, (D) RAG/STAT1–/–, and (E) STAT1/PKR–/– mice. For (B) to (D), all Pvalues for mock compared to infected groups were ≤0.0001 except IFNαβγR–/– i.n., P= 0.0021; STAT1–/– i.n., P = 0.0005; and STAT1–/– p.o., P = 0.0026.

Figure 4

MNV-1 infection in wild-type and immunodeficient mice. Viral RNA levels were determined with the use of quantitative real-time reverse transcriptase polymerase chain reaction and a standard curve (fig. S3), and the number of viral RNA copies per μg total RNA are reported. All samples from mock-infected mice were negative in this assay (8). (A) Three RAG-deficient animals (129RAG2–/– and B6RAG1–/–) per group were infected with MNV-1 for 90 days either p.o. or i.n., and viral RNA was quantified in the indicated tissues. (B) Antibody titers in 129 mice after p.o. infection. Antibody to MNV-1 was detected by enzyme-linked immunosorbent assay (3). Data shown is a representative of two experiments. (C to E) Seronegative STAT1–/– and 129 mice were inoculated p.o. with either 25 μl MNV-1 stock or mock-infected and analyzed 1, 3, or 7 days postinfection (five mice per time point from two experiments). (C) and (D) Viral RNA in tissues of 129 and STAT1–/– mice. (E) Representative tissue pathology in infected STAT1–/–mice.

The studies in RAG–/– mice suggested that the innate immune system might be sufficient for resistance to MNV-1 infection. This is an attractive hypothesis, because the usual 24-hour incubation period and 24- to 48-hour time course of human norovirus disease may offer too short a time for induction of adaptive immune responses (5). We therefore tested a variety of mouse strains lacking components of the innate immune system with MNV-1. Mice lacking IFNαβ receptor (2), IFNγ receptor (2), protein kinase RNA-activated (PKR) (16), or inducible nitric oxide synthase (17) were no more susceptible to lethal infection than wild-type controls (Fig. 3A). More subtle contributions of these genes to MNV-1 resistance cannot be ruled out by these studies. However, mice lacking both IFNαβ and IFNγ receptors were at least 10,000-fold more susceptible to lethal infection than controls after either i.c. or i.n. inoculation (Fig. 3B, fig. S4). These data show that IFNs are essential for resistance to MNV-1 infection and that the IFNαβ and IFNγ receptors can compensate for one another.

Because RAG/STAT1–/– mice die (fig. S1A) whereas RAG–/– mice survive after MNV-1 infection (Fig. 3A) and because STAT1 is involved in signaling through both the IFNαβ and IFNγ receptors (18), we studied whether STAT1 is required for resistance to MNV-1 infection. STAT1 deficiency resulted in lethal MNV-1 infection in mice (i) with intact B and T cell compartments (19) (STAT1–/–, Fig. 3C), (ii) lacking T and B cells (RAG/STAT1–/–, Fig. 3D), and (iii) lacking PKR (STAT1/PKR–/–, Fig. 3E). Therefore, STAT1 is required for survival after MNV-1 infection. Together with the survival of RAG–/– mice, this result strongly argues that a STAT1-dependent innate immune response is sufficient to prevent lethal MNV-1 infection.

To better understand the role of STAT1-dependent innate responses in controlling MNV-1 infection, we quantified MNV-1 RNA in tissues after p.o. infection of STAT1–/– or 129 wild-type mice. We measured RNA because we have been unable to measure viral infectivity by plaque assay (3). Wild-type mice were infected after p.o. inoculation as suggested by the presence of viral RNA in intestine, liver, and spleen one day after inoculation (Fig. 4C). Furthermore, these mice seroconverted to MNV-1 capsid protein after p.o. (Fig. 4B) or i.c. inoculation (fig. S5). However, wild-type mice did not develop symptoms or tissue pathology and cleared viral RNA from visceral and mucosal tissues by 3 days after infection (Fig. 4C), a time course consistent with clearance via innate immunity. Wild-type mice, therefore, resemble asymptomatically infected humans that have been detected during norovirus epidemics (5, 20).

STAT1–/– mice, like wild-type mice, contained MNV-1 genome in intestine, liver, and spleen one day after inoculation (Fig. 4D). In contrast to wild-type mice, STAT1–/– mice had high levels of MNV-1 RNA in multiple organs, as well as substantial tissue pathology, 3 and 7 days after infection. (Fig. 4, D and E). Decreases in viral RNA between 3 and 7 days of infection suggest that STAT1-independent immune responses can have some effect on MNV-1 infection. These data confirm a critical role for STAT1-dependent innate immunity in control of acute norovirus infection and show that a norovirus can cause systemic disease in immunocompromised hosts after mucosal infection.

Our finding that MNV-1 causes disease in immunocompromised mice suggests that human caliciviruses may be important systemic pathogens in patients with hereditary or acquired immunodeficiency. It is interesting to speculate, on the basis of our demonstration of persistent infection in RAG–/–mice and prolonged viral infection in STAT1–/– mice, that persons with immune deficiencies in either adaptive or innate immunity could be chronic norovirus carriers, thereby providing a source for epidemic outbreaks. Importantly, researchers using mouse models need to consider that MNV-1 can cause tissue pathology in immunocompromised mice. Acute or chronic MNV-1 infection might, therefore, contribute to phenotypes originally attributed directly to alterations in immune function. Lastly, we speculate that immunocompromised mice may harbor additional unidentified pathogens that could be studied to elucidate mechanisms of human disease and immunity.

Supporting Online Material

Materials and Methods

Figs. S1 to S5

References and Notes

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: Virgin{at}


View Abstract

Navigate This Article