Animal Models
Although no animal model to date has been entirely satisfactory, it has been demonstrated that chimpanzees can be successfully infected with GI.1 norovirus (
269), while gnotobiotic pigs (
270) and gnotobiotic calves (
271) can be successfully infected with GII.4 norovirus. The development of a norovirus replicon that expresses nonstructural viral proteins in human hepatoma cells (
272) and a reverse-genetics system driven by human elongation factor 1 alpha that produces progeny virus containing infectious viral RNA (
273) may provide insights into virus-host interactions at the cellular level. Furthermore, the development of a humanized immunodeficient murine model of human norovirus infection (
274) may also prove useful although perhaps only to a limited extent. For example, Taube and colleagues reconstituted
Rag1 −/− yc −/− BALB/c mice (yc is the γ chain of receptors for interleukin-2 [IL-2], IL-4, IL-7, IL-9, and IL-15) with human CD34
+ human hematopoietic stem cells after neonatal irradiation (
275). Despite this manipulation, infection by the oral route was entirely unsuccessful, and intraperitoneal inoculation with a strain of human norovirus GII.4 was required but with no evidence of associated illness.
Some insights into norovirus infection may, however, be obtained from studies of murine models utilizing MNVs. MNVs are widespread in laboratory mice, and in contrast to human noroviruses, they can be propagated in tissue culture (
276). Although MNV produces little or no outward signs of infection in wild-type laboratory mice (
277), knockout studies indicate that both innate and adaptive immunity play critical roles in the control of MNV infection. Particularly important are type 1 and type 2 interferons (
278 – 282), dendritic cell functions (
283 – 285), and the production of neutralizing antibody (
286 – 288). As such, MNVs may provide some important understanding of human norovirus infection that would not otherwise be possible to obtain (
286).
Human Host Factors in Susceptibility and Resistance to Norovirus Infection
Almost 4 decades ago, Parrino and colleagues challenged and then rechallenged, after an interval of 27 to 42 months, 12 volunteers with stool filtrates containing the norovirus (GI.1) from the index outbreak that had been serially passaged through other human volunteers (
289). Six subjects developed symptoms of gastroenteritis after the initial challenge, and each of these subjects also became ill upon rechallenge. In contrast, none of the 6 subjects who remained asymptomatic upon initial challenge became ill after rechallenge. Four of the six patients who developed symptoms were challenged a third time 4 to 8 weeks after the second challenge, and only one patient became ill. The investigators concluded that they had demonstrated the presence of short-term but the absence of long-term immunity to Norwalk virus infection. This conclusion does not, however, address the 6 subjects who failed to become ill after either the first or second challenge, a finding that suggested the possibility that some individuals are intrinsically resistant to Norwalk virus infection. Five years later, the discovery of familial clustering of infection, with some related groups having apparent resistance, was also observed in a large community outbreak traced to a swimming pool (
111). These observations suggested the possibility that some individuals possess inherited protection against norovirus infection. Subsequent investigations determined that the reason for this inherited resistance lies in the natural variability of histo-blood group antigens (HBGAs) and their expression on the mucosal epithelial surface of the gastrointestinal tract (
290).
HBGAs are oligosaccharide epitopes whose diversity results from the sequential addition of monosaccharides to glycan precursors (
291,
292). The A and B red cell antigens are synthesized from a common intermediate molecule, the H substance, of which there are 3 types, but with the final fucosylated product having the same terminal disaccharide. There is no “O antigen”: group O erythrocytes express the H antigen; hence, the ABO system is better labeled the ABO(H) system.
The fucosyltransferases FUT1 and FUT2 each add a fucose to the precursor disaccharides of the H substance. While the
FUT1 gene is expressed in erythroid progenitor cells,
FUT2 is expressed mostly in mucosal epithelial cells from which the enzymatic gene product is secreted into surrounding secretions such as saliva and breast milk. While also produced by FUT2, Lewis (Le) antigens are predominantly the result of fucosylation by a third enzyme, FUT3. Approximately 5% of the white population are homozygous for inactive
FUT3 alleles and, as a consequence, lack Le
a and Le
b and are termed Lewis negative (
291). Lewis antigens are also present on mucosal surfaces and are secreted into the surrounding milieu.
The presence of H, Le
b, A, or B antigens in saliva and other secretions operationally defines the secretor phenotype and is evidence of an active
FUT2, or “secretor,” gene. Approximately 20% of individuals of European descent have, however, inherited two null
FUT2 alleles as a result, in 99% of cases, of a G428A mutation that introduces a stop codon (
293). The resultant absence of HBGAs at mucosal surfaces and in surrounding secretions that characterize the nonsecretor phenotype plays a key role in protection against infection by most noroviruses, as demonstrated in challenge studies (
294 – 296) and analyses of outbreaks, as reviewed by Rydell and colleagues (
297).
Norwalk virus-derived VLPs bind to H antigen
in vitro (
298) and also hemagglutinate type A, AB, O, and, much more weakly, B red blood cells (
295). They also bind to epithelial cells of the gastroduodenal mucosa of individuals who are secretors but not to those of nonsecretors (
290). Thus, it appears that norovirus is among organisms, such as
Helicobacter pylori (
299), that utilize HBGAs expressed in the gastrointestinal tract as cell surface receptors (
300).
Given the multiplicity of viral strains and of HBGAs, it is not surprising that there are diverse patterns of individual strain associations with individual HBGAs. In a study involving 14 norovirus strains, Huang et al. reported having identified 7 different patterns that could be classified into 2 groupings: one group of viruses recognized A and/or B and H antigens, while the second group reacted only with Lewis and/or H antigens (
301). While there was some clustering of binding patterns depending on the phylogenetic relatedness of the viruses, both patterns could be found among both GI and GII viruses. GII.4 VLPs bind strongly to saliva of secretor-positive individuals regardless of the ABO(H) blood group (
302), while strong binding of GI.1 VLPs is observed only with secretions of type A, AB, or O secretors (
22,
294,
301,
303 – 306).
While initial studies indicated that a nonsecretor status seemed to provide total resistance to norovirus-associated illness, the degree of protection has now been demonstrated to be less than absolute. Examples of infection in nonsecretors include ones due to norovirus GII.2 Snow Mountain (
307), GII.4 (
308,
309), and GI.3 (
310,
311). In fact, combining the data from two GI.3 outbreaks, it can be calculated that 11 of 22 (50%) nonsecretors and 46 of 90 (51%) secretors were infected (
310,
311). In addition, binding to human Caco-2 intestinal cells by GII.6 norovirus strain VLPs is independent of HBGAs and is instead associated with cell maturity (
312), as is that by the GII.4 Desert Shield strain (
313). The receptors for these viruses remain unknown. In addition, inherited factors other than HBGAs may affect susceptibility to norovirus infection. For example, travelers to Mexico who have the lactoferrin T/T genotype may be protected against norovirus infection despite being at an increased risk of all-cause traveler's diarrhea (
314).
In addition to resistance to infection based on secretor status, norovirus infection results in adaptive immunity to homotypic viral challenge, albeit of relatively short duration, as was first reported by Parrino and colleagues in 1977 (
289) and subsequently confirmed (
294). For instance, a multiple-challenge experiment with Norwalk virus found evidence of protection for ≥6 months (
315), and it is commonly stated that immunity lasts as long as 2 years (
316). However, the relevance of many challenge experiments has been questioned because the inocula used in this and many other studies were likely several orders of magnitude higher (
317) than that required for infection, and as a consequence, the duration of immunity in the natural setting may be significantly longer than that estimated from these experiments. Thus, Teunis and colleagues estimated the 50% infectious dose (ID
50) to be between 18 and 1,015 genomic equivalents (
317). Atmar and colleagues found that the Norwalk virus ID
50s were estimated to be ≈1,320 genomic equivalents (≈3.3 RT-PCR units) in subjects who proved to be secretor-positive blood group O or A persons and ≈2,800 (≈7.9 RT-PCR units) for all secretor-positive persons (
318,
319). On the other hand, the potential for natural exposure to much larger numbers of viral particles is significant. Thus, 1 ml of virus-containing vomitus from symptomatically infected subjects contained a median of 41,000 genomic equivalents (
319). The norovirus load in feces is much higher and appears to be genogroup dependent, with reported medians of 8.4 × 10
5 cDNA copies per g in individuals with GI infection and 3.0 × 10
8 cDNA copies per g in individuals with GII infection (
320). It is therefore possible that estimates from observational studies of natural infection may provide the best estimates of the duration of immunity. Thus, mathematical modeling taking into account observational data describing the age-specific incidence of norovirus disease, seasonality, and population immunity led to an estimated duration of immunity of 4.1 years (95% CI, 3.2 to 5.1 years) to 8.7 years (95% CI, 6.8 to 11.3 years) (
316).
Complicating the determination of immunity duration by examination of naturally acquired infections is the fact that immunity may be strain specific, and any analysis is potentially confounded by the multiple genogroups, genotypes, and strains of virus, together with the continuing evolution of the virus. Thus, administration of stool filtrates from two distinct outbreaks to prison volunteers led to a lack of cross-protection between the Norwalk and Hawaii strains (
321), now known to be prototypes of GI and GII noroviruses, respectively. A prospective study of diarrhea in infants in the first weeks of life (the median age at recruitment was 19 days) to the age of 2 years found evidence of protection that was genotype specific: 97% of repeat infections were due to a genotype that differed from that of a previous infection (
322).
Although IgG antibody directed against norovirus antigens is present in the serum of >90% of individuals by the time they reach adulthood, and they nonetheless remain at risk of repeat norovirus infection (
323), evidence indicates that blocking antibody plays a role in immunity to reinfection. While the inability to culture the virus
in vitro precludes the ability to detect neutralizing antibody by classical methods, the prevention of binding of norovirus-derived VLPs or P particles to HBGA (blocking antibody) is believed to be an accurate surrogate of neutralization (
324,
325).
In a volunteer study, preexisting antibody did not protect against disease resulting from an initial challenge, but a correlation with protection emerged after subsequent rechallenges (
315). However, others have reported that preexisting antibody provides at least partial protection against naturally acquired infection (
326,
327). The appearance of IgA antibody against norovirus in saliva by day 5 after challenge of genetically susceptible individuals correlated with protection (
294), as did an early Th1 lymphocyte response (
307). High prechallenge titers of blocking antibodies also correlate with protection against experimental challenge (
325). Volunteers challenged with a GII.2 strain (Snow Mountain) elicited serum IgG antibody responses that were cross-reactive with another GII.1 viral strain (Hawaii), but lesser cross-reactivity was noted with salivary IgA, and neither antibody type was cross-reactive with GI.1 (
307). A similar pattern of cross-reactivity was seen with T cells exposed to Snow Mountain virus, which elicited a predominantly Th1 response. Passive protection from a maternal source occurs, as demonstrated by the observation that exclusive breastfeeding of infants from the ages of 3 to 5 months was associated with protection, and infections that did occur in the first 6 months of life were more likely to be asymptomatic than were those that occurred at 6 to 11 months of age (55% versus 36%;
P < 0.001) (
322).
The IgG antibody response to challenge with different GI strains results in variable heterotypic responses that are likely determined by an individual's history of natural exposure to noroviruses and deceptive imprinting (e.g., by development of “decoy epitopes”) (
328).
Antibodies to antigens other than those of the viral capsid also result from viral challenge. Three-fourths of volunteers challenged with GI.1 virus had an antibody response to viral protease from either the homologous virus or GII.4 virus (Houston) (
329), but these antibodies are not believed to be protective.
Immune Selection and Development of Viral Diversity
The viral capsid protein VP1 has 3 structural domains, with the shell (S) being the core from which 2 other domains, P1 and P2, protrude (
330). P2 is the most exposed portion and is likely the major point of contact with HBGA ligands and with neutralizing antibody (
22,
88,
331 – 333). In addition, mutations and recombination events involving P2 can significantly affect antigen properties and interactions with HBGAs (
334). GII.4 strains, in particular, are undergoing rapid evolution that affects receptor binding by changes in surface-exposed P2 and antigenic expression, resulting in the emergence of new epidemic strains of the virus (
25,
335 – 340). In contrast, there has been only limited evolution of GI viruses over the last 4 decades (
334).
GII.4, although first recognized as a major epidemic strain in the middle of the last decade of the 20th century, has been circulating since at least 1974 (
25). GII.4 undergoes epochal evolution characterized by periods of stasis followed by the emergence of a new epidemic strain. There have been 7 different GII.4 variants associated with global epidemics since the 1990s, which occurred in 1996, 2002, 2004, 2007 to 2008 (2 variant strains), 2009 to 2012, and 2012 onward (the Sydney strain) (
341). Thus, on average, new variants of GII.4 have appeared every 2 to 3 years.
Evidence indicates that new variants emerge under positive selection (
25), likely as a result of the pressure exerted by the development of herd immunity as larger portions of the population experience infection, with resultant viral antigenic drift (
20,
342 – 345). GII.4 2012 Sydney, which has, to a large extent, replaced previously circulating GII.4 variants, had undergone changes in at least 2 epitopes recognized by blocking antibodies (
344).
While norovirus evolution is believed to result from the selective pressure of the immune system, Schorn and colleagues demonstrated amino acid changes in P2 and P1-2 during individual chronic infections due to GII.4 in renal transplant recipients and due to GII.7 and GII.17 in an additional patient each (
60). Others evaluating two hematopoietic stem cell transplant recipients and one small bowel transplant recipient chronically infected with GII.4 or GII.7 found evidence of positive selection, with accumulation of an average of 5 to 9 mostly nonsynonymous mutations per 100 days in the hypervariable region of the P2 domain of the VP1 capsid gene (
346).
The immune selection of norovirus variants has important implications for vaccine development.