Noroviruses are highly prevalent, positive-sense RNA viruses that are transmitted by the fecal–oral route and infect the gastrointestinal tract of their host. Human noroviruses cause the majority of acute nonbacterial gastroenteritis infections worldwide, resulting in significant morbidity, mortality, and economic losses. The study of human viruses in their native host has inherent limitations. Thus, studies in animal models are critical for gaining a greater understanding of viral pathogenesis as well as the development of vaccines and therapeutics. Here, we describe multiple models that exist to study human noroviruses in nonhuman hosts and related viruses in their native hosts. The strengths and limitations of each of these models will also be discussed. Together, knowledge gained from these different models has significantly increased our understanding of these important human pathogens.
Chapter 3.4
Animal Models of Norovirus Infection
C.E. Wobus*
J.B. Cunha*
M.D. Elftman**
Abstract
Keywords
animal models of infection
positive sense RNA virus
calicivirus
norovirus
enteric virus pathogenesis
1. Introduction
Advances in the understanding of human noroviruses (HuNoVs) are challenging because, despite being such ubiquitous pathogens, these viruses cause species-specific infections and, until very recently (Jones et al., 2014), have not been successfully cultured in vitro. The use of animal models to address questions that are difficult or impossible to answer in human studies is a common approach to complement in vitro and epidemiological data for human viruses in general. Animal models are based either on studies of human viruses in animals, often requiring the use of immunocompromised hosts and/or host-adapted viruses, or the use of related viruses in their natural hosts. As outlined in this chapter, both approaches have been used to study HuNoVs.
Human volunteer-based studies and epidemiologic analyses have been invaluable in determining features of HuNoV infection and disease (Vashist et al., 2009; Kaufman et al., 2014). These are: (1) short incubation time (average 1 day) (Lee et al., 2013); (2) quick (1–4 days) resolution of disease symptoms, characterized mainly by diarrhea and/or vomiting (Kaufman et al., 2014; Atmar et al., 2008); (3) viral shedding for weeks to months with or without disease symptoms (Kaufman et al., 2014; Teunis et al., 2014); (4) viral spread mainly by the fecal–oral route and person-to-person transmission (Bitler et al., 2013), (5) histo-blood group antigen (HBGA) expression and secretor status as genetic susceptibility factors (Lindesmith et al., 2003; Hutson et al., 2002, 2005; Tan and Jiang, 2014). However, studies relying on human subjects have ethical limitations and require vast resources, therefore preventing them from being routinely conducted (Vashist et al., 2009). Therefore, animal models for HuNoV infection, as well as HuNoV surrogates, are valuable and necessary tools that can elucidate basic aspects of HuNoV infection and pathogenesis (Karst et al., 2014; Kniel, 2014).
Ideally, the best HuNoV animal model would use HuNoV or a virus with high genetic similarity that closely mirrors the biology and clinical features of HuNoV infection in an animal host. However, HuNoVs are members of the genetically diverse Norovirus genus within the Caliciviridae family that exhibit very narrow species specificity. Thus, HuNoV infection in a nonnative host is limited to some closely related primates (chimpanzees, newborn pigtail macaques), immunocompromised animals (ie, mice), and gnotobiotic (Gn) pigs and calves, albeit with varying levels of clinical signs (Bok et al., 2011; Cheetham et al., 2006; Souza et al., 2008; Taube et al., 2013). Alternatively, many studies have been carried out using related caliciviruses as HuNoV surrogate models. Historically, feline calicivirus (FCV) was the first surrogate for human caliciviruses due to the availability of an established cell culture system, and its study provided an early understanding of calicivirus biology (Radford et al., 2004). However, FCV causes upper respiratory tract infections in cats (Radford et al., 2007), and therefore has limited utility as a model for gastroenteritis and will not be discussed further. In contrast, porcine sapovirus (PoSaV) causes diarrhea in swine and can be propagated in culture in the presence of bile acids (Flynn and Saif, 1988), providing an animal model of calicivirus gastroenteritis. However, it was not until the discovery of murine norovirus 1 (MNV-1), a norovirus that infects a small animal species (Karst et al., 2003) and the subsequent development of a cell culture model (Wobus et al., 2004), that a HuNoV surrogate became widely available for both in vitro and in vivo studies (Wobus et al., 2006). MNV has been instrumental in elucidating many basic aspects of NoV infection and pathogenesis (Karst et al., 2003; Gonzalez-Hernandez et al., 2013, 2014; Taube et al., 2012, 2009; Chachu et al., 2008). More recently, Tulane virus (TV) was discovered as a causative agent of diarrhea in rhesus macaques (Sestak et al., 2012). Combined with the ability to readily grow in cell culture (Farkas et al., 2008) and availability of an infectious clone (Wei et al., 2008), TV is another important HuNoV surrogate (Farkas, 2015). In this chapter, the characteristics of HuNoV animal and surrogate models under investigation in the field will be discussed, including their advantages and disadvantages. We will also point out contributions of each model to our collective understanding of NoV biology and therapeutic approaches for HuNoVs.
2. Animal models of HuNoV infection
To date, a limited number of HuNoV strains was shown to infect mice deficient in the recombination activating gene (Rag) and the common gamma chain (γc) (Taube et al., 2013), gnotobiotic pigs (Souza et al., 2007a), gnotobiotic calves (Souza et al., 2008), macaques (Subekti et al., 2002), and chimpanzees (Bok et al., 2011). So far, no single animal model recapitulates all aspects of HuNoV infection. As outlined later, each model has strengths to allow the study of specific aspects of HuNoV biology (Table 3.4.1).
Table 3.4.1
Summary of the Discussed HuNoV Animal Models With Their Strengths and Limitations
| Host | Virus | Strengths | Limitations |
| Balb/c Rag2/γc−/− mouse (Taube et al., 2013) | HuNoV (GII.4) | Genetically tractable host Relatively low cost | Immune defects in host Transient infection Not orally susceptible Asymptomatic infection Lack of fecal shedding Not natural host |
| Chimpanzee (Bok et al., 2011) | HuNoV (GI.1, untyped strain) | Genetically similar to human Orally susceptible Transmission between animals Similar shedding and antibody responses to humans | Asymptomatic infection High cost Not natural host Moratorium on use |
| Newborn pigtail macaques (Subekti et al., 2002) | HuNoV (GII.3) | Orally susceptible Symptomatic infection Fecal shedding Similar serum IgG responses to humans Transmission between animals | High cost Unknown IgA responses Not natural host |
| Gnotobiotic pigs (Cheetham et al., 2006; Souza et al., 2007a; Bui et al., 2013; Takanashi et al., 2011; Cheetham et al., 2007) | HuNoV (GII.4, GII.12) | Symptomatic and asymptomatic infections Orally susceptible Fecal shedding Susceptibility partially mediated by HBGAs Similar incubation period, duration of symptoms to humans Mild villous atrophy Similar cytokine, antibody responses to humans | High cost Limited ability to passage Lack of microbiota Underdeveloped immune system Not natural host |
| Gnotobiotic calves (Souza et al., 2008) | HuNoV (GII.4) | Symptomatic infection Orally susceptible Fecal shedding Villous atrophy | High cost Lack of microbiota Underdeveloped immune system Not natural host |
2.1. Mice as an Animal Model
The first small animal model to study HuNoV infection was described in 2013 (Taube et al., 2013). Balb/c mice deficient in recombination activating genes (Rag−/−) 1 or 2 and common gamma chain (γc−/−) (Rag−/−/γc−/−) engrafted with human CD34+ hematopoietic stem cells (humanized mice), nonengrafted siblings, and immunocompetent wild type controls were challenged with a pool of HuNoV strains (Fig. 3.4.1). Viral genome titers were measured by quantitative polymerase chain reaction preceded by reverse transcription (RT-qPCR) in intestinal and nonintestinal tissues. Infection was measured by detecting increased genome titers over input in both humanized and nonhumanized Balb/c Rag−/−/γc−/− mice. Viral genomes were present in all tissues analyzed, except the brain. However, infection was asymptomatic and transient (lasting 2 days). Immunocompetent wild type Balb/c or B6/B10 Rag−/−/γc−/− mice were not infected, suggesting both the immune deficiency and genetic background of the mouse as determinants of susceptibility in this model. Unlike natural HuNoV infection, mice were susceptible only via the intraperitoneal, not the oral route, and no shedding of virus in the stool was detected. The former is likely a result of the genetic defects of this mouse strain, which lacks Peyer’s patches and associated M cells, a known gateway for NoV entry into the host (Gonzalez-Hernandez et al., 2014). Expression of structural and/or nonstructural proteins in cells with macrophage-like morphology in the spleen, small intestinal lamina propria, and Kupffer cells in the liver of Balb/c Rag−/−/γc−/−, but not of mock-infected mice, verified replication of HuNoV.
Figure 3.4.1 Human Norovirus Mouse Model.
Balb/c mice lacking the recombination activation genes (Rag) 1 or 2 and common gamma chain (γc) (Rag−/−γc−/−) were humanized with CD34+ hematopoietic stem cells. Humanized and nonhumanized Balb/c Rag−/−γc−/−, wild-type (WT) Balb/c controls and B6/B10 Rag−/−γc−/− mice were challenged with a pool of HuNoV strains by intraperitoneal (i.p.) and/or oral (p.o.) routes. Viral infection was determined by detecting increased viral genome titers over input and viral protein expression by immunohistochemistry. Productive infection routes are indicated by green arrows and presence of a virion, while nonproductive infections are indicated by black arrows and crossed-out virions. (Source: Summary of data from Taube et al., 2013.)
Balb/c mice lacking the recombination activation genes (Rag) 1 or 2 and common gamma chain (γc) (Rag−/−γc−/−) were humanized with CD34+ hematopoietic stem cells. Humanized and nonhumanized Balb/c Rag−/−γc−/−, wild-type (WT) Balb/c controls and B6/B10 Rag−/−γc−/− mice were challenged with a pool of HuNoV strains by intraperitoneal (i.p.) and/or oral (p.o.) routes. Viral infection was determined by detecting increased viral genome titers over input and viral protein expression by immunohistochemistry. Productive infection routes are indicated by green arrows and presence of a virion, while nonproductive infections are indicated by black arrows and crossed-out virions. (Source: Summary of data from Taube et al., 2013.)
A mouse model of HuNoV infection can facilitate translational and basic science applications, such as drug development and identification of viral and host factors mediating infection. The advantages of a mouse model as opposed to large animal models are manifold, including reduced cost, shorter gestation length, and the ability to genetically manipulate the host. However, the necessity for deleting components of the immune system to permit HuNoV to cross the species barrier represents an inherent caveat that limits the use of this model for certain applications. For example, vaccine testing is not possible due to the absence of T and B cells in Rag-deficient animals. Therefore, the identification of mouse strains with other immune deficiencies that permit HuNoV infection could expand the utility of the mouse model. Furthermore, the discovery of different permissive mouse strains that can be orally infected, shed virus in feces and have a longer duration of infection, which would improve this model. Some of the open questions to address are the minimum number of particles required to initiate an infection, the ability of different HuNoV genotypes to cause an infection, and the ability for serial passaging and adaptation of HuNoV in the murine host.
2.2. Chimpanzees as an Animal Model
The close genetic relatedness of chimpanzees to humans led to the use of chimpanzees as animal model for human-specific pathogens. HuNoV infections of chimpanzees were first performed in 1978 (Wyatt et al., 1978). Although none of the chimpanzees challenged orally with filtered HuNoV-containing fecal samples developed symptoms associated with HuNoV infection, infected animals developed NoV-specific antibody responses, and viral antigen was detected in feces. Importantly, HuNoV was serially passaged via feces to additional uninfected animals, demonstrating that chimpanzees support asymptomatic HuNoV infection. This finding was confirmed a few decades later when Bok et al. (2011) intravenously inoculated seronegative chimpanzees with HuNoV. No clinical signs of gastroenteritis were observed but the onset and duration of virus shedding in stool and serum antibody responses were similar to those observed in humans (Fig. 3.4.2). HuNoV capsid protein was detected in intestinal lamina propria cells from the B cell and dendritic cell lineages, and NoV RNA was found in intestinal and liver biopsies. Two infected chimpanzees were protected against re-infection of homologous HuNoV when challenged either at 4 and 24 months, or at 10 months after the first infection. In addition, chimpanzees vaccinated intramuscularly with Norwalk virus (GI) virus-like particles (VLPs), but not those derived from GII VLPs, were protected from homologous Norwalk virus infection 18 months after vaccination and developed antibodies that blocked VLP binding to HBGAs in a surrogate neutralization assay. Although this study did not use the natural route of infection, it provided important information regarding the ability of VLP-based vaccination to protect from homologous but not heterologous, cross-genogroup challenge. Protection from homologous challenge is also observed in human clinical trials (Atmar et al., 2011), while vaccine efficacy after heterologous challenge remains to be studied in humans. These data demonstrate the utility of chimpanzees as a valuable infection model for HuNoV but the NIH moratorium on the use of chimpanzees for biomedical research has halted any future NIH-funded studies in this HuNoV animal model.
Figure 3.4.2 Human Norovirus Chimpanzee Model.
Seronegative chimpanzees were intravenously (i.v.) inoculated with genogroup I (GI) HuNoV (Norwalk virus). Virus replication was measured by RT-qPCR, virus protein by IHC and serum antibody by EIA. Infected chimpanzees were protected against re-infection with homologous HuNoV when rechallenged after the first infection or vaccinated intramuscularly (i.m.) with GI (Norwalk virus) VLPs. GII VLPs and placebo did not protect from challenge with GI HuNoV, thus providing important information regarding the ability of VLP-based vaccination to protect from homologous but not heterologous, cross-genotype challenge. (Source: Schematic summary of data from Bok et al., 2011.)
Seronegative chimpanzees were intravenously (i.v.) inoculated with genogroup I (GI) HuNoV (Norwalk virus). Virus replication was measured by RT-qPCR, virus protein by IHC and serum antibody by EIA. Infected chimpanzees were protected against re-infection with homologous HuNoV when rechallenged after the first infection or vaccinated intramuscularly (i.m.) with GI (Norwalk virus) VLPs. GII VLPs and placebo did not protect from challenge with GI HuNoV, thus providing important information regarding the ability of VLP-based vaccination to protect from homologous but not heterologous, cross-genotype challenge. (Source: Schematic summary of data from Bok et al., 2011.)
2.3. Macaques as an Animal Model
Nonhuman primates other than chimpanzees may provide an alternative HuNoV animal model but only limited infections in a few species have been performed to date. The susceptibility of common marmosets, cotton top tamarins, cynomolgus, and rhesus macaques to oral HuNoV infection was tested in one study (Rockx et al., 2005). Only rhesus macaques were found susceptible to Norwalk virus infection, albeit asymptomatically, with virus shedding in some animals and development of NoV-specific IgG and IgM, but not IgA, antibodies. While the asymptomatic infection status and duration of shedding resembles that found in humans (Graham et al., 1994; Rockx et al., 2002), the lack of IgA responses limits the use of this species as a vaccine model.
More promising results were observed in a study which showed newborn pigtail macaques (Macaca nemestrina) to be symptomatically infected with Toronto virus P2-A (a GII.3 HuNoV strain) (Subekti et al., 2002). Two of three monkeys developed diarrhea and dehydration. All animals developed IgG antibody responses and shed virus in stool for at least 3 weeks. Similar development of disease and shedding results were obtained following passage into additional newborn macaques. In addition, the mothers of the infected infant monkeys became symptomatically infected via natural infection, demonstrating the ability of HuNoV to be transmitted and passaged in this species of monkeys. This promising first study warrants further development of Macaca nemestrina as a HuNoV disease and transmission model.
2.4. Gnotobiotic Pigs and Calves as Animal Models
Gnotobiotic (Gn) pigs are models for enteric diseases because their gastrointestinal anatomy and physiology are similar to that of humans (Saif et al., 1996). Gn pigs support symptomatic HuNoV infection characterized by mild diarrhea, transient viremia, viral shedding in feces, and the presence of HuNoV structural and nonstructural proteins in enterocytes (Cheetham et al., 2006) (Fig. 3.4.3). In one study (Cheetham et al., 2006), Gn pigs were orally inoculated with fecal filtrates of the NoV/GII/4/HS66/2001/US strain, or with pig-passaged intestinal contents. Two thirds of the inoculated animals developed mild diarrhea, 57% seroconverted, but only 44% of pigs shed low levels of virus in their feces, suggesting disease symptoms can develop without fecal shedding or seroconversion. Using different virus strains [a GII.4 2006b variant (Bui et al., 2013) or the GII.g/GII.12 HS206 strain (Takanashi et al., 2011)], more infected animals shed virus than developed diarrhea, demonstrating asymptomatic infections can occur in Gn pigs. Increases in genome titers over input were detected in most (18/22) pigs (Bui et al., 2013), confirming productive infection. The short incubation period (24–48 h) and duration of disease symptoms (1–3 days) or asymptomatic infection resemble clinical features of HuNoV infection in humans (Lee et al., 2013; Lindesmith et al., 2003; Lopman et al., 2004). To date, only GII.4 and GII.12 strains have been reported to infect pigs, but further studies are needed to test susceptibility of Gn pigs to additional genogroups/genotypes.
Figure 3.4.3 Human Norovirus Gnotobiotic Pig Model.
Gnotobiotic pigs support symptomatic HuNoV infection characterized by mild diarrhea, transient viremia, viral shedding in feces, and detection of HuNoV structural and nonstructural proteins in enterocytes by immunofluorescence microscopy. (Source: Schematic representation of findings from Cheetham et al., 2006 and Souza et al., 2007a.)
Gnotobiotic pigs support symptomatic HuNoV infection characterized by mild diarrhea, transient viremia, viral shedding in feces, and detection of HuNoV structural and nonstructural proteins in enterocytes by immunofluorescence microscopy. (Source: Schematic representation of findings from Cheetham et al., 2006 and Souza et al., 2007a.)
Susceptibility factors of HuNoV infection in the human population are secretor status and HBGA type (Lindesmith et al., 2003; Hutson et al., 2002). A similar correlation between infection outcome and HBGA expression was seen in Gn pigs, wherein increased incidence of diarrhea, fecal shedding, and seroconversion rate were observed in pigs with the A+/H+ genotype compared to the A−/H− genotype (Cheetham et al., 2007).
Histopathologic examination of HuNoV infected Gn pigs detected mild villous atrophy in the duodenum of one out of seven animals and an increased number of apoptotic cells in intestinal villi of all infected animals (Cheetham et al., 2006), features also reported during HuNoV infection (Troeger et al., 2009). Structural or nonstructural HuNoV antigens were detected in enterocytes of the proximal small intestine, suggesting HuNoV replication occurs in these cells, but whether additional cell types also support infection in this model remains unknown. The tropism for enterocytes differs from the recently described tropism of HuNoV for B cells in vitro (Jones et al., 2014), but the cellular tropism of HuNoV in infected human intestines remains undefined (Lay et al., 2010).
The power of animals models for pathogenesis studies is highlighted by a study where naïve Gn pigs were infected with the HuNoV HS66 GII.4 strain, and serum and intestinal contents were analyzed for intestinal and systemic humoral and cellular immune responses (Souza et al., 2007a). Such studies are difficult to perform in human volunteers with unknown and variable NoV-exposure history. Infection induced low antibody titers and antibody-secreting cell numbers, which may explain the incomplete seroconversion rate observed in pigs. However, the kinetics of IgA and IgG antibody-secreting cell responses and the greater induction of a Th1 compared to Th2 and proinflammatory IL-6 responses are also observed in HuNoV-infected volunteers (Lindesmith et al., 2003; Ramani et al., 2015).
Animal models of disease are important for the development of therapeutics and Gn pigs have been evaluated as test models for both vaccines and antivirals. Similar to vaccine trials in humans (Atmar et al., 2011; Bernstein et al., 2015; Treanor et al., 2014), HuNoV VLPs are immunogenic in Gn pigs and induce homologous protection (Souza et al., 2007b). In an additional study, Gn pigs were vaccinated with VLPs and P particles (globular structures formed by multimerization of 12–36 copies of the P domain (Bereszczak et al., 2012)) and challenged with a closely related GII HuNoV. Primary NoV infection protected 83% of challenged animals against diarrhea and 57% against virus shedding. In contrast, VLP or P particles provided protection against diarrhea in 60% or 47%, respectively, and against virus shedding in 0% or 11%, respectively (Kocher et al., 2014). These studies suggest that a replicating vaccine (eg, attenuated virus) should be pursued as an alternative vaccination approach as it will likely confer superior protection to the recombinant subunit vaccine of VLPs or P particles.
The use of Gn pigs as an antiviral efficacy test model for HuNoV infection is only just beginning but will increase as more compounds move through the drug development pipeline (Arias et al., 2013; Rocha-Pereira et al., 2014). One study (Jung et al., 2012) looked at interferon (IFN) α an important innate immune mediator with proven antiviral activity in humans (Feld and Hoofnagle, 2005). Treatment of Gn pigs with IFNα decreased viral shedding with a delayed onset and shorter duration of shedding, but viral shedding increased after treatment was discontinued (Jung et al., 2012). The inability of type I IFNs (IFNα/β) to clear NoV infection has also been seen in mice (Nice et al., 2014). Taking into account the known side effects of IFNα treatment in humans (Gota and Calabrese, 2003), IFNα will unlikely become a widely used treatment option for HuNoV infection in humans, and more specific anti-NoV drugs will need to be developed. Another area of future development will be studies regarding the utility of Gn pigs as a transmission model. Although the HS66 strain was serially passaged three times through Gn pigs with evidence of virus shedding at each passage and evidence of HuNoV-infected cells, viral titers and duration of fecal shedding decreased with each passage and no evidence of mutations was observed in the region analyzed (ie, ORF1/2 junction) (Cheetham et al., 2006). Thus, whether adaptive mutations occur in the HuNoV genome during infection in pigs remains to be addressed.
Additional Gn pig studies demonstrated that susceptibility to HuNoV infection changes with age, with lower median infectious dose (ID50) in neonates (4–5 days old) compared to older (33–34 days old) pigs (Bui et al., 2013). Whether susceptibility similarly changes with age in humans remains unknown. However, Gn pig studies (Bui et al., 2013; Jung et al., 2012) did confirm earlier observations in humans that the use of statins (cholesterol lowering drugs) exacerbates HuNoV disease severity (Rondy et al., 2011). Simvastatin use in Gn pigs increased the incidence of diarrhea, and led to earlier onset and longer virus shedding (Bui et al., 2013; Jung et al., 2012). Treated animals showed impaired Toll-like receptor (TLR) 3-dependent innate immune responses and lowered IFNα expression by alveolar macrophages and intestinal dendritic cells (Jung et al., 2012), but future studies are needed to determine whether similar mechanisms operate during NoV infection of humans.
In addition to Gn pigs, Gn calves were also investigated as an alternative large animal model (Souza et al., 2008), although to a limited extent. HuNoV infection of Gn calves with the GII.4 HS66 strain resulted in intestinal lesions (villous atrophy and loss of epithelial cells) in the small intestine, diarrhea and virus shedding in all analyzed animals. One of five calves also developed viremia. Viral capsid antigen was detected in small intestinal enterocytes and in lamina propria macrophage-like cells. HuNoV infection induced low levels of antibodies with only 67% (2/3) seroconversion but 100% (3/3) coproconversion (ie, antibodies in feces) in calves. Furthermore, early proinflammatory TNFα responses during diarrhea, high antiinflammatory IL-10 responses, and low to moderate Th1 (IL-12, IFNγ) and Th2 (IL-4) responses were observed. While more pronounced intestinal lesions were induced in Gn calves compared to Gn pigs, coinciding with increased numbers of IgA and IgG antibody secreting cells, most features of HuNoV infection in Gn calves are similar to those in Gn pigs.
In summary, both species of Gn animals are suitable models for the study of disease mechanisms, and thus represent important large animal models for preclinical, IND (investigational new drug)-enabling studies. However, the high cost of maintenance and required specialized care limit the broader use of this model in the scientific community. In addition, the lack of commensal bacteria in Gn animals, which promote development of the immune system (Sommer and Backhed, 2013) and of enteric viral infections (Kuss et al., 2011; Uchiyama et al., 2014; Kane et al., 2011), including NoV (Jones et al., 2014; Miura et al., 2013), represent caveats for studies of viral pathogenesis and immune responses. However, recent studies of Gn animals colonized with human commensal bacteria (Kandasamy et al., 2014; Vlasova et al., 2013) promise to overcome this limitation in future studies of HuNoV infection.
3. Human norovirus surrogates as animal models
The alternative approach to studying human viruses in a nonnative animal host is the study of surrogate viruses in their native hosts. The most widely used surrogate viruses for modeling HuNoV infection in vivo were PoSaV and are now MNV and recoviruses. All these viruses share the ability to readily grow in cell culture while sharing varying degrees of genetic relatedness and biological features with HuNoVs. Although no surrogate animal model recapitulates all aspects of HuNoV infection, each model has contributed important knowledge to our understanding of calicivirus biology as detailed later.
3.1. The Porcine Sapovirus Model
Sapoviruses (SaVs) are gastrointestinal pathogens transmitted by the fecal–oral route (Oka et al., 2015). Like HuNoVs, human SaVs (HuSaVs) are enteric viruses that cause outbreaks and sporadic cases of acute gastroenteritis in people of all ages around the globe (Oka et al., 2015). The clinical features of HuSaV disease are indistinguishable from HuNoV infections and laboratory diagnosis is required for accurate identification of these viruses (Oka et al., 2015). Unlike HuNoVs, HuSaVs do not bind HBGAs (Shirato-Horikoshi et al., 2007) but other cell attachment factors like sialic acid might be shared (Kim et al., 2014; Rydell et al., 2009). No host genetic factor(s) of susceptibility has been identified for HuSaVs to date (Oka et al., 2015). PoSaVs are genetically closely related to HuSaVs (Scheuer et al., 2013), and the PoSaV Cowden strain is currently the only member of the Sapovirus genus with a tissue culture system that requires the addition of intestinal contents or bile acids during infection (Flynn and Saif, 1988; Chang et al., 2004). This combined with the ability to genetically manipulate the viral genome (Chang et al., 2005) and development of acute gastroenteritis in PoSaV-infected pigs (Guo et al., 2001; Flynn et al., 1988) led to the use of PoSaV as a surrogate for HuNoV and HuSaV infections.
Gnotobiotic pigs orally infected with the PoSaV prototype strain Cowden develop mild to severe diarrhea, shed virus in the feces, and some develop viremia (Guo et al., 2001). Histological analysis of small intestinal segments from PoSaV-infected pigs showed villous atrophy, villous shortening and blunting, villous fusion, crypt hyperplasia, and cytosolic vacuolization of enterocytes. PoSaV antigen detection indicates that the major site of PoSaV replication is in villous enterocytes in the proximal small intestine. Therefore, the clinical and pathological features of PoSaV infection resemble those found in HuSaV (Oka et al., 2015) and HuNoV (Green, 2013) infections in humans. However, whether the cellular tropism of PoSaV mirrors the in vivo tropism of HuNoVs and HuSaVs remains to be determined. Importantly, this model demonstrated that limited amino acid changes in a virus strain following serial passaging in cell culture resulted in an attenuated PoSaV phenotype in vivo (Guo et al., 2001). Given the large genetic variability of HuNoVs and HuSaVs, this suggests a wide range of phenotypes for these human viruses.
In summary, the PoSaV experimental infection model using Gn pigs is a valuable tool that has and will undoubtedly contribute to a greater understanding of SaV pathogenesis and disease. Future studies using the PoSaV model also have the potential to advance translational studies towards vaccine and antiviral development for HuSaVs. However, the use of PoSaV as a surrogate for HuNoV is less ideal given the availability of MNV and TV, which are more closely related to HuNoV and for which the molecular tools and infection or disease models are available that are needed to perform detailed mechanistic and translational studies.
3.2. The Mouse Norovirus Model
Murine norovirus (MNV) is a natural pathogen of wild rodents and laboratory mice (Henderson, 2008; Farkas et al., 2012; Tsunesumi et al., 2012; Smith et al., 2012). To this date, MNV is the only NoV that has a small animal model and can be efficiently cultivated in tissue culture (Wobus et al., 2006). Since its discovery in 2003 (Karst et al., 2003), the MNV model has enabled significant advances in the understanding of norovirus-host interactions on the molecular, cellular, and organismal levels and has provided valuable perspectives on host–pathogen interactions in general (reviewed, eg, in Karst et al., 2014; Wobus et al., 2006). The utility of MNV as a model system for other NoVs is based on the following main features: (1) MNV can readily be cultivated in primary cells and stable cell lines derived from its natural host (Wobus et al., 2004), enabling a thorough analysis of the entire virus life cycle from attachment to release at the cellular and molecular levels. (2) The genetic tractability of mice has allowed researchers to identify diverse host factors that are important for survival and clearance of MNV, providing insights into host defenses that are likely to be important for protection during HuNoV infection (reviewed in Karst et al., 2014; 2015; Karst, 2010). (3) The availability of several reverse-genetics systems (Ward et al., 2007; Yunus et al., 2010; Chaudhry et al., 2007; Sandoval-Jaime et al., 2015) permits a detailed understanding of viral determinants of NoV infections, including receptor utilization, tissue tropism and viral persistence (eg, Taube et al., 2012; Nice et al., 2013; Strong et al., 2012). While MNV does not recapitulate clinical symptoms of HuNoV infection in immunocompetent hosts, it provides novel perspectives toward understanding interactions between intestinal viruses and their natural, coevolved host (Jones et al., 2014; Cadwell, 2015; Kernbauer et al., 2014; Cadwell et al., 2010).
Sequence analysis of the MNV genome shows that these viruses are homologous to other viruses in the NoV genus, including HuNoV, but form a separate genogroup (GV) (Karst et al., 2003; Smith et al., 2012). Three of MNV’s four open reading frames (ORFs) encoding the nonstructural proteins (ORF1), major and minor capsid proteins (ORF2 and ORF3, respectively) are shared with all other NoVs, while an additional fourth ORF is only present within GV (McFadden et al., 2011; Thackray et al., 2007). Studies of NoV proteins have suggested that products of the ORF1 polyprotein of MNV and HuNoV have common functional properties, such as the rearrangement of cellular membranes during virus replication or during translation initiation (reviewed in Thorne and Goodfellow, 2014). In addition, MNV shares with HuNoVs the use of carbohydrates for cell attachment, although the specific carbohydrates only partially overlap (Tan and Jiang, 2014; Taube et al., 2009). Many strains of HuNoV bind to HBGA (reviewed in Tan and Jiang, 2014) but some HuNoV strains also bind sialic acids (Rydell et al., 2009; Maalouf et al., 2010; Esseili et al., 2012). MNV binds to terminal sialic acids on gangliosides and N- or O-linked glycoproteins in a strain-dependent manner (Taube et al., 2009, 2012), although no strains of MNV are known to date that use HBGA (Taube et al., 2010). Thus, while the MNV model does not allow studies into the function of HBGA during NoV pathogenesis, it does provide a system to understand the role of carbohydrates during NoV infection in general.
MNVs and HuNoVs also share the physical properties of the capsid and route of transmission. Like HuNoV, MNV virions are highly stable in the environment and are relatively resistant to heat, dessication, and low pH (Cannon et al., 2006; Arthur and Gibson, 2015; Kotwal and Cannon, 2014; Tuladhar et al., 2012). In addition, viral particles are also highly infectious. The estimated 50% HuNoV infectious dose is between 18 and 2800 particles depending on the study (Teunis et al., 2008; Atmar et al., 2014), and MNV infection can be initiated with as few as 10 particles (Liu et al., 2009). These similarities are consistent with their common fecal–oral route of transmission making MNV a widely used surrogate for food safety studies (reviewed in Kniel, 2014).
Given the shared route of transmission, viral shedding in the stool is characteristic for all NoVs. While the infectious potential of shed HuNoV virions remains unknown, genomes are detected in the stool for weeks to months in immunocompetent individuals, long after symptomatic disease has resolved, and for years in immunocompromised individuals (Kaufman et al., 2014; Teunis et al., 2014; Green, 2014). Infectious MNV is shed for months as most strains cause a persistent infection in wild-type mice (Thackray et al., 2007; Hsu et al., 2006). In addition, the MNV-1 strain, which causes a self-resolving, acute infection in wild-type mice with transient shedding, causes long-term shedding in mice lacking components of the adaptive immune system (ie, Rag-deficient mice) (Karst et al., 2003). Thus, different durations of viral shedding can be modeled depending on the combinations of MNV and mouse strains. Multiple factors influencing shedding have been identified; viral gene products (ie, NS1/2 [N-terminal] and VP1 [viral capsid]) (Nice et al., 2013; Strong et al., 2012), host factors (eg, adaptive immune responses, interferon λ) (Karst et al., 2003; Nice et al., 2014), and the microbiota (Baldridge et al., 2015). While transmission occurs readily between mice in a cage (Manuel et al., 2008) (Fig. 3.4.4), the parameters that influence transmission remain largely unexplored. Nevertheless, a recent study successfully demonstrated the ability of an antiviral compound to disrupt transmission of MNV (Rocha-Pereira et al., 2015), highlighting the utility of the MNV system as a transmission model.
Figure 3.4.4 Transmission of MNV between mice in a cage.
One Balb/c mouse each was orally infected with 1 × 106 pfu of MNV.CR3, the same dose of UV-inactivated MNV.CR3, or mock-infected. Two days later, each infected mouse was cohoused with 2 naïve mice for 28 days. Serum was collected from all animals on days 0, 7, 14, 21, and 28 of cohousing. ELISA was performed on a 1:100 dilution of serum as described (Wobus et al., 2004) and absorbance values were graphed. Naïve mice housed with a MNV.CR3-inoculated, but not mock or UV inactivated virus-inoculated cage mate seroconverted, indicating viral infection.
One Balb/c mouse each was orally infected with 1 × 106 pfu of MNV.CR3, the same dose of UV-inactivated MNV.CR3, or mock-infected. Two days later, each infected mouse was cohoused with 2 naïve mice for 28 days. Serum was collected from all animals on days 0, 7, 14, 21, and 28 of cohousing. ELISA was performed on a 1:100 dilution of serum as described (Wobus et al., 2004) and absorbance values were graphed. Naïve mice housed with a MNV.CR3-inoculated, but not mock or UV inactivated virus-inoculated cage mate seroconverted, indicating viral infection.
Despite many similarities, important differences exist between MNVs and HuNoVs. One of the greatest differences is that MNV does not cause overt clinical disease in an immunocompetent host. HuNoV infections frequently result in vomiting and diarrhea but mice are unable to vomit and MNV-infected wild-type mice do not develop overt diarrhea. Nevertheless, MNV-infected mice show histopathologic changes in the intestine (mild inflammation, limited villous atrophy) similar to those observed in HuNoV-infected volunteers (Karst et al., 2015; Mumphrey et al., 2007). It is also important to note that MNV causes diarrhea in mouse strains with impaired innate immunity, including strains lacking signal transducer and activator of transcription (Stat) 1 or the type I and II interferon receptors (Karst et al., 2003; Mumphrey et al., 2007; Rocha-Pereira et al., 2013). While the immunodeficient status of the host limits the use of those strains for studies of disease mechanisms, it does provide a small animal disease model for antiviral efficacy testing to develop compounds that not only reduce viral loads but also disease scores, as one recent study elegantly demonstrated (Rocha-Pereira et al., 2013).
Differences exist also at the level of genetic diversity. HuNoV strains identified to date exhibit a broader genetic diversity (up to 45% divergence at the nucleotide level), spanning 3 genogroups (GI, GII, and GIV) with at least 29 genotypes (Zheng et al., 2006) than known MNV strains. MNVs identified to date separate into two genotypes with ∼23% nucleotide divergence (Smith et al., 2012). However, a recently identified rat NoV that clusters with MNVs in GV diverges at the nucleotide level by 31% (Tsunesumi et al., 2012). Thus, it is likely that the diversity of rodent NoVs will increase in the future as more strains are identified. While the broad genetic diversity of HuNoV requires specific considerations during vaccine development, vaccination studies of HuNoV VLPs in mice demonstrated the development of heterologous, cross-genogroup immunity that reduces MNV viral loads upon challenge (LoBue et al., 2009). Thus, the MNV model can also be used as one tool for the development of vaccines with cross-genogroup protective efficacy.
Many features of HuNoV infection and replication remain poorly understood, but studies with MNV may offer clues to these processes. One open question in the field is the cell tropism of NoV. The identity of HuNoV-infected cell types in vivo remains unknown. However, viral antigen from both MNV and HuNoV has been detected within the intestinal lamina propria of infected mice and humans, respectively (Lay et al., 2010; Mumphrey et al., 2007), suggesting a tropism for immune cells (Karst and Wobus, 2015). This is consistent with the cell types that support HuNoV and MNV replication in vitro, specifically B cells for HuNoV (Jones et al., 2014) and macrophages, dendritic cells, microglia, and B cells (Jones et al., 2014; Wobus et al., 2004; Cox et al., 2009) for MNV. Thus, to date a partial overlap in cellular tropism exists between HuNoV and MNV. Future studies are needed to determine whether human and murine NoVs share the tropism for all these different immune cells. Another open question is how the broad genetic diversity of HuNoV influences viral phenotypes. Genetic analysis of MNV strains has already revealed viral determinants of NoV biology, including virulence and glycan utilization (Karst et al., 2014; Taube et al., 2012; Nice et al., 2013; Strong et al., 2012). Similarly, genetic variability in HuNoV may eventually be linked to specific phenotypes. Furthermore, studies with MNV have illuminated many aspects of NoV biology, for example, identifying roles for microfold (M) cells in intestinal entry into the host (Gonzalez-Hernandez et al., 2014), for dendritic cells in dissemination (Elftman et al., 2013), and for components of the innate immune system in controlling infections (McCartney et al., 2008; Thackray et al., 2012; Changotra et al., 2009). Future studies will need to determine the translatability of these findings to HuNoV infections.
One of the most promising aspects of the MNV model is the opportunity to study an enteric pathogen in its natural host and elucidate how the host’s genetic background and microbial environment influence an encounter with a pathogen. For example, studies using MNV demonstrate that the interaction between enteric viruses, the microbiome, and the host’s genetic background can influence the development of complex conditions such as inflammatory bowel disease (Cadwell et al., 2010). More recent studies have shown how concurrent helminth infections create a type 2 cytokine environment resulting in poor control of NoV replication (Osborne et al., 2014), and that the microbiome as a whole or specific members promote MNV infection and persistence (Jones et al., 2014; Baldridge et al., 2015), at least in the latter case via modulating the host immune response.
Overall, MNV provides a unique opportunity to mechanistically dissect, not only the effects of host factors and environment on a naturally coevolved host–pathogen relationship, but also many aspects of NoV biology. Despite the limitations of MNV as a model for clinical disease (ie, need for an immunodeficient host), MNV infection of wild-type mice represents a genetically tractable infection and transmission model.
3.3. The Recovirus Model
Rhesus enteric caliciviruses represent the most recently established HuNoV model. Recoviruses were first described in 2008, after the detection and isolation of a novel calicivirus from fecal samples of rhesus macaques (Farkas et al., 2008). These viruses are phylogenetically distinct from the genus Norovirus and from the proposed genus Recovirus. However, they share many properties with HuNoV, including their genomic organization, large genetic diversity, the ability to bind to HBGA, transmission by the fecal–oral route, and clinical disease presentations. Recovirus genetic diversity encompasses three genogroups and five genotypes (Farkas, 2015). GI and GII strains are found in rhesus macaques (Farkas et al., 2014), while GIII strains were discovered in samples collected from six humans with gastroenteritis in Bangladesh (Smits et al., 2012). The prototypic Tulane virus (TV), a GI.1 strain, replicates efficiently in cell culture in the rhesus macaque kidney epithelial cell line LLC-MK2 (Farkas et al., 2008), has a reverse genetics system available (Wei et al., 2008), and is associated with diarrheal disease in rhesus macaques after experimental infection (Sestak et al., 2012), providing crucial tools for mechanistic studies.
One of the most important advantages of the recovirus model is that the clinical disease presentation, histopathology, and transient serological response to TV are reminiscent of volunteer studies of HuNoV infection (Sestak et al., 2012). The similarities are multi-fold. First, intragastric infection of juvenile macaques with TV causes fever, a semiliquid diarrhea, and viral shedding in feces, although neither vomiting nor viremia were observed (Sestak et al., 2012). Second, intestinal biopsies from the duodenum of these monkeys revealed slightly blunted villi and mild mononuclear infiltration of the lamina propria. Third, experimentally infected macaques developed a robust but transient neutralizing antibody response against TV. However, future studies are needed to determine the ability of these antibodies to protect from infection or disease. Fourth, TV has a tropism for B cells. TV antigen-positive B cells were observed in the intestinal lamina propria, and infection of peripheral blood mononuclear cells obtained from healthy macaques led to an increase in viral RNA and antigen expression within B cells (Sestak et al., 2012). This tropism is similar to at least one HuNoV GII.4 strain (Jones et al., 2014).
Another important similarity of HuNoV and recoviruses is their ability to bind HBGAs. Diverse HuNoV strains have different patterns of HBGA-utilization and dependency (Huang et al., 2005) (see Chapter 3.3). Three HBGA binding patterns have been identified to date for recoviruses (Farkas, 2015). In case of TV, binding occurs with the type 3 form of the A antigen and all forms of the B antigen (Zhang et al., 2015). HBGAs are a known susceptibility factor for HuNoV infection (reviewed in Tan and Jiang, 2014) but the primate colonies used for recovirus studies exhibit a fairly homogeneous HBGA phenotype with greater than 95% of monkeys producing B antigen (Farkas et al., 2010). Thus, future experiments in a more phenotypically diverse population of macaques are needed to investigate whether specific HBGAs determine susceptibility of macaques to recovirus infection. Multiple roles have been proposed for HBGAs and its different forms (ie, free, bacterially associated, cell associated) during NoV infection (Karst and Wobus, 2015). Thus, the ability of recoviruses to bind HBGA and cause disease in a conventionally housed (ie, bacterially colonized) immunocompetent host provides a unique opportunity to understand the role of HBGA during enteric calicivirus infection and disease in the future.
The relationship between genetic and serologic diversity of HuNoVs remains unknown. Study of recoviruses with a similarly broad genetic diversity and available cell culture system has yielded initial insights; namely that genotypic classifications correlate generally with serotypes, although exceptions exist (Farkas, 2015; Farkas et al., 2014). These data suggest that the great HuNoV genetic diversity is likely accompanied by broad serological diversity. While this may present challenges in vaccine design, the recently described HuNoV culture system (Jones et al., 2014) will now enable future serotyping studies.
In summary, the recovirus model combines many beneficial features of an ideal HuNoV model. Most importantly, recoviruses can be genetically manipulated in vitro (Wei et al., 2008), replicate efficiently in tissue culture (Farkas et al., 2008), use the same HBGA attachment receptors as HuNoV (Sestak, 2014), and cause similar disease in vivo (Sestak et al., 2012). While the financial cost of in vivo studies and the inability to genetically manipulate the host are drawbacks of this model, the limitations are outweighed by many benefits, foremost the availability of an immunologically and microbially competent disease model. This model might be particularly useful in elucidating various roles of HBGAs in calicivirus pathogenesis, including how changing HBGA binding affinities and escape from neutralizing immune responses influence individual and herd immunity, an important driver of HuNoV evolution (Debbink et al., 2013).
4. Conclusions
Animal models are an integral part of our studies of human viruses. Large animal models are particularly valuable for translational studies and preclinical efficacy testing of antivirals and vaccines, while small animal models, especially knock-out and transgenic mouse models, are invaluable to gain a greater mechanistic understanding of all aspects of virus—host interactions at the organismal and cellular levels.
Although the past two decades have seen much progress in our understanding of HuNoVs, our understanding of these viruses lags behind that of other human viruses. The study of HuNoV has been particularly difficult due to the long-term lack of a cell culture system (Duizer et al., 2004; Herbst-Kralovetz et al., 2013; Papafragkou et al., 2013; Takanashi et al., 2014) and thus independent validation of the recently described B-cell culture model (Jones et al., 2014) is critical. The field also struggles to identify an animal model that closely reproduces HuNoV disease and is widely accepted. No “perfect” model has been developed to date but multiple different animal models for the study of HuNoV are available either by overcoming the species barrier of HuNoV or using HuNoV surrogates. Each of these models has its strengths and limitations (Tables 3.4.1 and 3.4.2) and overcoming some of these limitations remains an important goal for future research. One potentially fruitful avenue could be the generation of host-adapated HuNoVs, which has not been reported for any HuNoV animal model. Improvements might also come from the study of little explored (eg, newborn pigtail macaques) or new animal hosts, the identification of new surrogate viruses, or discovery of additional strains of HuNoV that more readily adapt or transmit to other species. While data obtained from any of the existing or future models are critical for our improved understanding of HuNoVs, validation of findings in the human host will ultimately determine how appropriate each model is to study a given aspect of HuNoV biology.
Table 3.4.2
Summary of the Discussed HuNoV Surrogates With Their Strengths and Limitations
| Host | Virus | Strengths | Limitations |
| Wild-type mouse (Wobus et al., 2006; Chang et al., 2005; Farkas et al., 2012) | MNV (all strains) | Natural host Fecal shedding Orally susceptible Genetically tractable host Relatively low cost Similar histopathology to humans (villous blunting, mild intestinal inflammation) Available cell culture system Available reverse genetics system | Asymptomatic infection Does not bind HBGAs Relatively limited genetic diversity of MNV strains |
| STAT1-/-AG129 mouse (Karst et al., 2003; Hsu et al., 2006) | MNV (eg, MNV-1, MNV-3, CR3) | Orally susceptible Symptomatic infection Fecal shedding Relatively low cost Available cell culture system Available reverse genetics system | Immune deficiency Systemic infection Exacerbated histopathology Does not bind HBGAs Relatively limited genetic diversity |
| Rhesus macaques (Sestak et al., 2012; Farkas, 2015) | Recoviruses/ Tulane virus | Natural host Orally susceptible Symptomatic infection Virus binds HBGAs Large genetic diversity Genomic organization similar to HuNoVs Fecal shedding Similar histopathology to humans (villous blunting, mild intestinal inflammation) Transient neutralizing antibody response like humans Available cell culture system Available reverse genetics system | Not a norovirus High cost |
| Gnotobiotic pigs (Rydell et al., 2009; Scheuer et al., 2013) | PoSaV | Natural host Orally susceptible Symptomatic infection Fecal shedding Villous atrophy and blunting Available cell culture system Available reverse genetics system | Not a norovirus High cost Lack of microbiota Underdeveloped immune system Does not bind HBGAs |
Acknowledgments
We apologize to all colleagues whose work could not be cited due to lengths restrictions. C.E.W. is a recipient of the Friedrich Wilhelm Bessel award from the Alexander von Humboldt Foundation. Work in her laboratory was supported by NIH grants R21/R33 AI102106, R01 AI080611, R21 AI103961, and DARPA grant HR0011-11-C-0093. J.B.C. was funded by Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Brasília, Brazil.
Full access? Get Clinical Tree
