key: cord-278438-bnjkmegh
authors: Yuan, Lijuan; Saif, Linda J
title: Induction of mucosal immune responses and protection against enteric viruses: rotavirus infection of gnotobiotic pigs as a model
date: 2002-09-10
journal: Vet Immunol Immunopathol
DOI: 10.1016/s0165-2427(02)00046-6
sha: 
doc_id: 278438
cord_uid: bnjkmegh

Enteric viruses are a major cause of diarrhea in animals and humans. Among them, rotaviruses are one of the most important causes of diarrhea in young animals and human infants. A lack of understanding of mechanisms to induce intestinal immunity and the correlates of protective immunity in neonates has impaired development of safe and effective vaccines against enteric viruses. Studies of candidate vaccines using an adult mouse model of subclinical enteric viral infections often do not predict vaccine efficacy against disease evaluated in neonatal large animals. A series of studies have been conducted using a neonatal gnotobiotic pig model of rotavirus infection and diarrhea to identify correlates of protective immunity and to evaluate traditional and novel vaccine approaches for the induction of mucosal immune responses and protection to enteric viruses. Gnotobiotic pigs recovered from infection with virulent Wa human rotavirus (HRV) (mimic natural infection) had high numbers of intestinal IgA rotavirus-specific primary antibody-secreting cells (ASCs) and memory B-cells (to recall antigen) measured by ELISPOT assay, which correlated with complete protection against rotavirus challenge. Most short-term IgA memory B-cells were resident in the ileum, the major site of rotavirus replication. Spleen, not the bone marrow, was the major resident site for longer-term IgG memory B-cells. Candidate rotavirus vaccines evaluated in pigs for their ability to induce intestinal or systemic ASC and protection against rotavirus infection and diarrhea included attenuated live virus, inactivated virus, and baculovirus-expressed double-layered rotavirus-like particles (2/6-VLPs). In combination with those candidate vaccines, various adjuvants, delivery systems, and immunization routes were tested, including incomplete Freund’s adjuvant for i.m. immunization, and a mutant Escherichia coli heat labile enterotoxin R192G (mLT) for i.n. immunization. It was shown that orally administered replicating vaccines were most effective for priming for intestinal IgA ASC and memory B-cell responses, but i.n. administered non-replicating 2/6-VLPs plus mLT were effective as booster vaccines. We conclude that protective immunity depends on the magnitude, location, viral protein-specificity, and isotype of the antibody responses induced by vaccination. Therefore highly effective enteric viral vaccines should: (i) induce sufficient levels of intestinal IgA antibodies; (ii) include viral antigens that induce neutralizing antibodies; and (iii) require the use of effective mucosal adjuvants or antigen delivery systems for non-replicating oral or i.n. vaccines.

Enteric viruses infect and damage intestinal epithelial cells leading to diarrheal disease (Saif, 1999b) . Enteric viral infections remain a major cause of morbidity and mortality for neonatal farm animals and children (Saif and JackWood, 1990) . A lack of understanding of mechanisms to induce intestinal immunity and the correlates of protective immunity in neonates has impaired development of safe and effective vaccines against enteric viruses for farm animals and children. Protection induced by candidate vaccines observed in studies using lab animals, such as mouse and rabbit models of rotavirus subclinical infections often do not predict the protective ef®cacy of these vaccines against clinical infections in studies of neonatal large animals O'Neal et al., 1998; McNeal et al., 1999a,b; Yuan et al., 1998 Yuan et al., , 2000 Welter and Welter, 1990) . Efforts to identify immunologic correlates of protective immunity to rotavirus infection using knockout mice for selected immune-related genes, passive lymphocyte transfer experiments, and immunodepletion of selected populations of lymphocytes by the injection of speci®c monoclonal antibodies have shown that none of the known effectors for acquired immune responses (CD4 or CD8 T-cells, antibodies, interferon-g, perforin, etc.) are essential for protective immunity in adult mice (Franco and Greenberg, 1999; O'Neal et al., 2000) . The outcome of vaccine ef®cacy studies using the adult mouse model of rotavirus infection also largely depends on the genetic background of the mouse strain (Franco and Greenberg, 2000) . Uncharacterized innate immune mechanisms were suggested to play a major role in protective immunity against rotavirus infection in some mouse strains (Choi et al., 1999; Franco and Greenberg, 1999; McNeal et al., 1999b) . The uncertainty of the roles of known immunologic effectors in protective immune responses to rotavirus in adult mice raises questions of the relevance of studies using adult mice as models for rotavirus immunity in neonatal farm animals and human infants.

Studies of mucosal immune responses and protective immunity induced by experimental rotavirus infections and candidate vaccines using a gnotobiotic pig model of rotavirus diarrhea (Chen et al., 1995; Yuan et al., 1996 Yuan et al., , 1998 Yuan et al., , 2000 Yuan et al., , 2001a or trans-missible gastroenteritis coronavirus (TGEV) infection in conventional piglets (VanCott et al., 1993 (VanCott et al., , 1994 have consistently demonstrated that protective immunity against diarrhea from enteric viral infections depends upon the production of intestinal IgA antibody-secreting cell (ASC) and memory B-cell responses at the sites of virus replication (Saif, 1996 (Saif, , 1999a . Studies of natural rotavirus infections in children showed a similar positive correlation of IgA antibodies and protective immunity against rotavirus infection and diarrhea (Matson et al., 1993; Velazquez et al., 2000) . Strategies to enhance production of virus-speci®c IgA antibodies to the neutralizing viral antigens in the small intestine (intestinal lamina propria) may be the most effective approach to induce protection against viral diarrheas. This paper will: (i) review vaccination strategies for induction of mucosal immune responses and protection against enteric viral infections and disease with emphasis on candidate rotavirus vaccines evaluated in a gnotobiotic pig model of rotavirus disease; (ii) discuss comparative vaccine protective ef®cacies evaluated using the neonatal gnotobiotic pig model of rotavirus disease versus the adult mouse model of rotavirus infection; (iii) compare the distribution pattern of memory B-cells in pigs to that of T-and B-memory cells in mice and their role in the protective immunity; and (iv) brie¯y summarize the features of three novel mucosal vaccine delivery systems (coronavirus derived expression systems, microencapsulation and immune stimulating complexes (ISCOMs)) and their potential for improving the ef®cacy of enteric virus vaccines.

The characteristics of the mucosal immune system and the pathogenesis of enteric viruses, including rotavirus have been reviewed (Mowat and Viney, 1997; Saif, 1999b) . Using neonatal gnotobiotic pigs, a unique model to study rotavirus immunity , a series of studies have been conducted to identify the immune correlates of protective immunity to rotavirus infection and disease (Chen et al., 1995; Ward et al., 1996b; Yuan et al., 1996 Yuan et al., , 1998 Yuan et al., , 2000 Yuan et al., , 2001a and to evaluate various vaccination approaches (vaccine types, immunization routes and adjuvants) for their ability to induce intestinal and systemic immune responses and protection against rotavirus challenge (Ward et al., 1996b; Yuan et al., 1996 Yuan et al., , 1998 Yuan et al., , 2000 Yuan et al., , 2001b . The questions addressed were: What are the basic parameters of the host mucosal and systemic antibody responses (isotype and tissue distribution) to infectious rotavirus (virulent or attenuated) or non-replicating (inactivated or VLP) vaccines in naõ Ève pigs? How do these responses correlate with protection? How does the type of vaccine and route (oral, i.m., i.n., or sequential combinations of oral and i.n.) of administration in¯uence the mucosal and systemic antibody responses to rotavirus and the degree of protection?

We ®rst investigated the ASC and memory B-cell responses to a virulent human rotavirus (HRV) Wa strain (to mimic natural infections) and an attenuated Wa HRV strain (to mimic live oral vaccines), the degree of protection induced against virulent Wa HRV challenge, and potential correlates of protection (Yuan et al., 1996 (Yuan et al., , 2001a . Four groups of gnotobiotic pigs were orally inoculated with one dose of virulent Wa HRV at 3±5 days of age, or 2 or 3 doses of attenuated Wa HRV (AttHRV 2Â, 3Â) at 10-day intervals, or mock-infected with cell-culture¯uids, respectively, and challenged at 21±28 postinoculation days (PIDs) with virulent Wa HRV. The virus-speci®c ASC and memory B-cell responses in the intestinal (duodenum, ileum, and mesenteric lymph nodes) and systemic (spleen, peripheral blood, and bone marrow) lymphoid tissues were enumerated using an isotype-speci®c ELISPOT assay at PID 21±28 and postchallenge days (PCDs) 7 (PID 28±35). To quantitate virus-speci®c ASC, an ELISPOT assay was performed on the day of mononuclear cells (MNCs) extraction (Yuan et al., 1996) . To detect virus-speci®c memory B-cells, an ELISPOT assay was performed after the MNC from each tissue had been stimulated in vitro with semipuri®ed Wa HRV antigen in cell culture for 5 days (Yuan et al., 2001a) . The ASC and memory B-cell responses in three representative tissues at PCD 0 and protection rates against challenge are summarized in Table 1 . Pigs recovered from virulent rotavirus infection were completely protected from virus shedding upon challenge. These pigs developed signi®cantly higher numbers of virus-speci®c IgA and IgG ASC and memory B-cells in the intestinal lamina propria compared to the attenuated Wa HRV inoculation groups. Oral inoculation of pigs with 2 or 3 doses of attenuated Wa HRV induced lower numbers of IgA and IgG ASC and memory B-cells in the intestinal lamina propria, but higher numbers of memory B-cells in the spleen compared to the virulent Wa HRV inoculated pigs. The attenuated Wa HRV inoculated pigs were partially protected against virus shedding and diarrhea after challenge ( Table 1 ). The protection rates from the virulent, AttHRV2Â, and AttHRV3Â inoculation groups were positively correlated with the magnitude of IgA ASC and memory B-cell responses in the intestinal lamina propria. A positive correlation between intestinal IgA antibody responses and protection was also shown in a study of mice comparing the differences in the mucosal and systemic immune responses generated by oral inoculation of mice with heterologous (non-murine) and homologous (murine) rotaviruses and the ability of these infections to produce protective immunity against subsequent reinfection upon challenge (Feng et al., 1994) . Sucking mice were inoculated orally with live rhesus (RRV), bovine (NCDV), or murine (EHP) rotavirus and challenged at PID 42 with virulent murine rotavirus (ECW). Conclusions from this study using the adult mouse model of rotavirus infection agreed with the ®ndings from studies of pigs infected with virulent or attenuated Wa HRV and challenged with virulent Wa HRV; i.e., the ability of a heterologous rotavirus to stimulate a detectable intestinal IgA antibody response correlated with its ability to generate protective immunity.

Due to the nature of enteric virus infections (super-®cial infections of the intestinal epithelial cells), IgA antibodies present in the intestine at the time of exposure provide the most important ®rst line of protection by neutralizing virus infectivity in the lumen or intracellularly (Burns et al., 1996) , preventing the viruses from attaching to and penetrating the Yuan et al. (1996 Yuan et al. ( , 2001a . b Not determined. epithelium, and excreting bound antigens (immune complexes) through the epithelium back into the lumen (Saif, 1999a) . To stimulate intestinal IgA antibody responses, oral immunization is preferable because it mimics the route of natural infection. However, oral vaccines have shown insuf®cient or inconsistent ef®cacy in the ®eld and clinical trials (Bresee et al., 1999; Saif and JackWood, 1990) . The reduced immunogenicity of the vaccine strain compared to the virulent strain of the virus (Yuan et al., 1996; Ward et al., 1996b) , the potential risk of adverse effects of high vaccine doses or live vaccines in infants (intussusception) (Anonymous, 1999) , and the interference of preexisting antibodies in the intestine of neonates (mainly maternal antibodies derived from colostrum and milk) with replicating virus vaccines Parreno et al., 1999) are major obstacles to improving the ef®cacy of live oral rotavirus vaccines.

To circumvent the obstacles for oral live vaccines, non-replicating vaccines have been evaluated for their immunogenicity and protective ef®cacy. Studies of adult mice and rabbits have shown that non-replicating rotavirus vaccines (inactivated virus or VLPs) administered via extraintestinal immunization routes (i.m., i.n. or i.p.) induced complete or signi®cant partial protection against infection with or without adjuvants (Conner et al., 1993; McNeal et al., 1992 McNeal et al., , 1998 McNeal et al., , 1999a Of®t and Dudzik, 1989; O'Neal et al., 1997 O'Neal et al., , 1998 . We evaluated the immunogenicity and protective ef®cacy of non-replicating rotavirus vaccines in combination with different adjuvants, including inactivated Wa HRV given by the oral or i.m. route with incomplete Freund's adjuvant (IFA) (Yuan et al., 1998) and 2/6-VLP vaccines given by the i.n. route with or without a mutant Escherichia coli heat labile enterotoxin R192G (mLT) as a mucosal adjuvant in gnotobiotic pigs (Yuan et al., 2000) . The mLT and other LT or Cholera toxin (CT) derivatives have been shown to retain the adjuvant effects of the native LT or CT to enhance immune responses to unrelated antigens when coadministered orally or i.n. through induction of both Th1-and Th2-type immune responses (LT) or by promoting Th2-type immune responses (CT) (Freytag and Clements, 1999; Pizza et al., 2001) . The ASC and memory B-cell responses and the protection rates induced by the nonreplicating rotavirus vaccines are summarized in Table 2 . Binary-ethyleneimine inactivated Wa HRV administered orally alone or by the i.m. route with IFA twice to gnotobiotic pigs induced few IgA ASC and memory B-cells in the intestinal lamina propria and conferred minimal protection (0±6%) against challenge. Substantially higher numbers of IgG ASC and memory B-cells were induced in the systemic lymphoid tissues (spleen and peripheral blood) by i.m. inoculation of pigs with inactivated Wa HRV; however, IgG ASC or memory B-cells in the systemic lymphoid tissues were not associated with protective immunity. Thus, inactivated rotavirus vaccines administered via ether the oral or i.m. route were signi®cantly less ef®cacious in inducing intestinal IgA antibody responses and conferring protective immunity than the oral live rotavirus vaccines. The segregation of the mucosal and systemic immune responses appears to be the cause of the failure of i.m. vaccination to induce intestinal immune responses. It was shown in a study of humans that T-and B-cells activated by the oral and parenteral routes differ markedly with regard to their respective homing potentials (Kantele et al., 1997) . The majority of lymphocytes activated by parenteral inoculation expressed L-selectin, the principal peripheral lymph node homing receptor; and the majority of this population of B-cells were committed to IgG production (Quiding-Jarbrink et al., 1995) , which explains the high numbers of IgG ASC and memory B-cells detected in the spleen of pigs inoculated i.m. with inactivated rotavirus ( Table 2 ). The observations on the isotype and tissue distribution of antibody responses and protection in pigs inoculated i.m. with inactivated rotavirus concur with the ®ndings from studies of mice showing a relationship between the homing receptor expression pattern of memory B-cells and protective immunity Rott et al., 1997; Williams et al., 1998) whereby protective immunity for an intestinal pathogen, such as rotavirus, resides in the subsets of memory B-and T-cells expressing the intestinal homing receptor a4b7, and this phenotype is not shared with memory cells elicited by i.m. immunization.

Lymphocytes sensitized in the nasal-associated lymphoid tissues (NALTs) can relocate to distant effector sites through the common mucosal immune system (Brandtzaeg et al., 1999) . The advantages of NALT as an inductive site for vaccine antigens, particularly Yuan et al. (1998 Yuan et al. ( , 2000 .

for non-replicating antigens, include the presence of antigen-retaining crypts and the absence of antigendegrading digestive enzymes. Although the i.n. route has been shown to be more ef®cient for the induction of IgA antibody responses in the upper respiratory and genitourinary tract, and less ef®cient for the intestinal lamina propria in rodents and pigs (Brandtzaeg et al., 1999; McDermott and Bienenstock, 1979; VanCott et al., 1994) , the i.n. vaccination route has been shown to be effective for induction of protective immunity against rotavirus infection in adult mice (O'Neal et al., 1997 (O'Neal et al., , 1998 McNeal et al., 1999b) . In gnotobiotic pigs, double-layered 2/6-VLP vaccines derived from co-expression of the bovine RF core protein VP2 gene and the inner capsid protein VP6 gene of simian SA11 or human Wa rotavirus strains were evaluated using the i.n. route (Yuan et al., 2000) . The VP2 core and VP6 inner capsid rotavirus proteins did not induce virus neutralizing antibodies which are induced only by the rotavirus outer capsid proteins VP4 and VP7. Four groups of gnotobiotic pigs received 250 ug per dose SA11 2/6-VLPs in 2 or 3 i.n. dose regimens (SA11 2/6-VLP mLT IN 2Â and 3Â), two doses of SA11 2/6-VLPs alone (SA11 2/6-VLPs IN 2Â), or mLT alone as controls, respectively. Intranasal inoculation of pigs with SA 11 2/6-VLPs plus mLT induced higher numbers of intestinal IgA and IgG ASC and memory B-cell responses than three oral doses of live attenuated Wa HRV, but lower IgA and IgG ASC and memory B-cells in the spleen (Tables 1 and 2 ). The mLT adjuvant enhanced intestinal ASC and memory B-cell responses to the VLP vaccines, as evident by the greatly increased numbers of intestinal IgA and IgG ASC and memory B-cells in the pigs vaccinated with SA11 2/6-VLPs mLT compared to the SA11 2/ 6-VLPs alone at PID21 (PCD 0). However, SA11 2/6-VLPs administered i.n. in two or three dose regimens with or without mLT did not confer protection to pigs against diarrhea upon challenge ( Table 2 ). The failure of 2/6-VLP vaccines suggests that protective immunity to rotavirus diarrhea in pigs depends not only on the location, the magnitude, and the isotypes of antibodies, but also on the protein-speci®city (VP4 and VP7 neutralizing antibodies) of the B-cell responses.

Although a similar SA11 2/6-VLP vaccine conferred complete protection against infection when adult mice were inoculated i.n. (O'Neal et al., 1998) , protection against diarrhea in neonatal pigs probably requires the presence of IgA neutralizing antibodies in the intestine to the outer capsid rotavirus proteins, VP4 and VP7. Most non-replicating antigens (e.g. inactivated virus, recombinant proteins, and VLPs) are usually poor immunogens for inducing mucosal immune responses when administered without adjuvants, and may even result in immunological tolerance instead of immunity (Czerkinsky et al., 1999) . To induce suf®cient levels of immune responses, an effective mucosal adjuvant or delivery system that can greatly enhance the immunogenicity of the vaccine antigens is of critical importance. However, in a study of mice inoculated i.n. with puri®ed inactivated double-or triple-layered rotavirus particles, similar rates of protection were induced by the vaccines with or without adjuvant; although mLT signi®cantly increased antibody responses and reduced antigen concentrations needed for full protection (McNeal et al., 1999b) . The discrepancies in rotavirus vaccine ef®cacies in studies of adult mice and neonatal pigs will be discussed further in a later section.

Gut-associated lymphoid tissues (GALTs) and NALT are two major inductive sites for mucosal immune responses. The use of combined oral and i.n. vaccination routes stimulates multiple mucosal inductive sites, i.e., both GALT and NALT in contrast to the individual oral or i.n. vaccination route alone. We postulated that by exploiting the advantages of multiple mucosal immunization routes (oral and i.n.), and vaccine types (replicating virus and non-replicating VLPs), a combined vaccination regimen may optimally stimulate the mucosal immune system and increase the protective ef®cacy of rotavirus vaccines. Two combined vaccine regimens were evaluated in gnotobiotic pigs. In the ®rst combined vaccine regimen group, gnotobiotic pigs were inoculated with one oral dose of attenuated Wa HRV at 3±5 days of age to prime the GALT inductive sites by live replicating virus. This was then followed by two i.n. boosting doses of Wa 2/6-VLP plus mLT (AttHRV/VLP2Â) at 10-day intervals. In the second combined vaccine regimen group, pigs were inoculated with two i.n. doses of Wa 2/6-VLP plus mLT followed by one oral dose of live attenuated Wa HRV (VLP2Â/AttHRV1Â). A single oral dose of attenuated Wa HRV (AttHRV1Â) and three i.n. doses of Wa 2/6-VLP plus mLT (Wa 2/6-VLP mLT 3Â) groups were included as controls. The ASC responses and protective ef®cacy of the combined regimens are summarized in Table 3 . The AttHRV/VLP2Â regimen induced similar mean numbers of intestinal IgA ASC as induced after virulent Wa HRV infection (135 versus 171 ASC per 5 Â 10 5 MNC, respectively, Tables 1 and 3) and the IgA ASC numbers were signi®cantly higher than those induced by a single oral dose of attenuated Wa HRVor three i.n. doses of Wa 2/6-VLP plus mLT at PID 28 (PCD 0). The AttHRV/VLP2Â regimen induced the highest mean numbers of intestinal IgA ASC at challenge among all rotavirus vaccines tested to date in gnotobiotic pigs including one to three oral doses of live attenuated Wa HRV, two or three i.m. doses of inactivated Wa HRV with IFA, and two or three i.n. doses of SA11 or Wa 2/6-VLPs with or without mLT (Yuan et al., 1996 (Yuan et al., , 1998 (Yuan et al., , 2000 (Yuan et al., , 2001a . These results suggest that the combined immunization routes, oral priming followed by intranasal boosting, and vaccine types, live attenuated Wa HRV for priming followed by 2/6-VLPs plus mLT for boosting, was more effective in stimulating intestinal IgA ASC responses than each individual vaccine. The protective ef®cacy of the AttHRV/VLP2Â regimen was slightly lower than that of three doses of attenuated Wa HRV (Table 1) with a 58% protection rate against virus shedding and a 44% protection rate against diarrhea upon challenge of pigs with virulent Wa HRV. Although the AttHRV/VLP2Â regimen did not confer a higher rate of protection compared to the AttHRV3Â vaccine, the potential advantage of the combined AttHRV/VLP2Â vaccine was indicated by the signi®cantly higher ASC responses induced in the intestinal lymphoid tissues and the potential greater safety of using a non-replicating vaccine as boosters to reduce the risk of intussusception associated with a live oral reassortant HRV vaccine in infants (Anonymous, 1999) . The absence of the outer capsid VP4 and VP7 rotavirus neutralizing antigens from the i.n. boosting doses of 2/6-VLPs in the AttHRV/VLP2Â regimen was again presumably a determinant in its failure to confer higher protection rates as it was for the complete failure of i.n. 2/6-VLP vaccines tested in pigs to confer any protection. If a triple-layered 2/4/6/ 7-VLP vaccine were used for boosting in the combined regimen, a much higher protection rate might have been achieved based on the high numbers of IgA ASC in the intestinal lymphoid tissues induced by this vaccine approach and based on the positive correlation between the numbers of IgA ASC in the intestinal lymphoid tissues induced by live rotavirus and the degree of protection (Yuan et al., 1996 (Yuan et al., , 2001a . The VLP2Â/AttHRV1Â regimen was less ef®cacious than the AttHRV/VLP2Â regimens in inducing intestinal ASC responses and protection against virus shedding or diarrhea, but more ef®cacious than the i.n. Wa 2/6-VLP mLT alone (Table 3 ). The ineffectiveness in inducing intestinal ASC responses by the VLP2Â/ AttHRV1Â vaccine regimen was possibly partially due to the functional compartmentalization of the primary mucosal inductive site (NALT) involved in this vaccination approach, suggesting that oral priming (at the portal of infection and the site of virus replication) with live replicating vaccine is more effective for inducing intestinal IgA responses. Taken together, the use of a replicating vaccine to prime lymphocytes in the major inductive site (GALT) followed by boosting with a non-replicating vaccine at a second mucosal inductive site (NALT) may be a highly effective approach to stimulate the mucosal immune system and induce protective immunity against enteric infections.

There are numerous reports of protective immunity against rotavirus infection in adult mice induced by different types of rotavirus antigens (e.g. live or inactivated, homologous or heterologous rotavirus (Feng et al., 1994; McNeal et al., 1992 McNeal et al., , 1998 McNeal et al., , 1999a Of®t and Dudzik, 1989) ; recombinant rotavirus proteins or VLPs Choi et al., 1999; O'Neal et al., 1997 O'Neal et al., , 1998 ; and DNA plasmids (Chen et al., 1997 (Chen et al., , 1998 Herrmann et al., 1996) ], using various routes of inoculation (oral, i.n., i.m., i.p., or i.d.) . When comparing studies of mice, there are discrepancies in the protective ef®cacies of rotavirus vaccines possibly due to the use of different strains of in-or out-bred adult mice or murine rotaviruses. The genetic background and virulence of murine rotavirus strains used for challenge plays a critical role in the outcome of an ef®cacy study in mice (Franco and Greenberg, 2000) . The results often differ even more dramatically between studies of rotavirus vaccines in adult mice and neonatal pigs. It is dif®cult to compare data derived from studies of adult mice with neonatal pig studies due to species genetic differences, the body size, age, and in-bred (mice) or out-bred (pigs) animals, in addition to their different susceptibilities to enteric viruses or the different pathogenesis of the viruses in mice versus pigs (Conner and Ramig, 1997; Saif and Greenberg, 1997; Yuan et al., 2000) . Inoculation of mice i.m. with inactivated rotavirus or i.n. with 2/6-VLPs mLT or puri®ed inactivated native rotavirus particles induced complete protection against infection in adult mice. However, similar vaccines did not confer protection against rotavirus disease in neonatal pigs. There are many possible reasons for the discrepancies; however, the difference in the pathogenesis of rotaviruses between pigs and mice is likely one of the most critical factors. The pathogenesis of rotavirus disease in pigs, calves and probably in human infants is similar; but it differs from that in mice (Conner and Ramig, 1997; Saif et al., 1994; Saif and Greenberg, 1997) . Rotavirus infections of young pigs, calves, and human infants induces watery diarrhea, anorexia, depression, and dehydration (Conner and Ramig, 1997; Saif et al., 1994) . Pronounced histologic changes (e.g. villous atrophy as a result of virus replication and epithelial cell lysis) occur in gnotobiotic pigs infected with virulent Wa HRV (Ward et al., 1996a) and similar intestinal villous atrophy has been reported for human infants (Conner and Ramig, 1997; Davidson and Barnes, 1979; Saif and Greenberg, 1997) . In mice under 15 days of age, homologous rotavirus infection causes diarrhea and lethargy, but only mild intestinal lesions consisting of vacuolization of villous tip epithelial cells with little or no villous atrophy (Little and Shadduck, 1982) . In adult mice, rotavirus infection occurs without causing disease or histopathologic changes (Ward et al., 1990) . High levels of intestinal IgA antibodies to the neutralizing rotavirus antigens are required to prevent rotavirus disease in neonatal pigs. In contrast, protection against infection or clearance of an established infection in mice apparently can be mediated by any one of the immune effectors (IgA or IgG; CD4 or CD8 T-cells) or some unknown innate immune effectors. The redundancy of immune mechanisms (innate or acquired) in mice may be better developed than that in pigs or humans. Such differences are very important to consider when comparing the protective ef®cacy of a vaccine evaluated in the adult mouse model of infection with the neonatal pig model of rotavirus infection and disease.

Vaccination induced protective immunity depends on the induction of memory B-and T-cells. To investigate the distribution and sites of residence of memory B-cells, and their role in protective immunity to reinfection against an enteric virus, we assessed the memory B-cells responses and protection in pigs inoculated with virulent and attenuated Wa HRV (Yuan et al., 2001a ). Short-term rotavirus-speci®c IgA and IgG memory B-cells were detected in the intestinal lamina propria (duodenum and ileum), MLN, and spleen, but with much higher numbers in the ileum and slightly higher numbers in MLN than in the other tissues at PID 28 and 35. Complete protection against HRV infection was associated with sig-ni®cantly higher numbers of IgA and IgG effector and memory B-cells in the intestinal lamina propria (especially in the ileum) of virulent HRV-inoculated pigs at challenge (Table 1) . Our ®ndings clearly demonstrate that the short-term memory B-cells induced by virulent Wa HRV reside primarily in the ileum, the major site of rotavirus replication, up to PID 35. Our ®ndings parallel the recent ®ndings for both memory CD4 and CD8 T-cells (Mackay and von Andrian, 2001; Masopust et al., 2001; Reinhardt et al., 2001 ) that a substantial subset of memory T-cells reside in extralymphoid tissues (e.g. skin, lung, and intestinal lamina propria) and this subset of memory cells acts rapidly to execute effector activity when exposed to recall antigen. It is logical that a similar distribution pattern of memory T-and B-cells would facilitate the cognate interaction between T-and B-cells required for the successful establishment of the T-cell-dependent immune response. Based on the study of CD4 and CD8 memory T-cells (Masopust et al., 2001; Reinhardt et al., 2001) , Mackay and von Andrian (2001) proposed that immunological memory could be divided into two phasesÐa short-term phase followed by a long-term phase. The short-term memory depends on the survival and activation of the extralymphoid tissue-origin memory T-cells, whereas long-term memory depends on secondary lymphoid tissue-origin memory T-cells. In gnotobiotic pigs infected with virulent Wa HRV (Yuan et al., 2001a) , the numbers of IgA and IgG memory B-cells in all the tissues, but especially IgA in the ileum declined substantially from PID 28 to PID 83. Signi®cant waning of IgA memory B-cell responses in the intestine from PID 28 to PID 83 indicates a short IgA B-cell memory to rotavirus infection in neonatal pigs. Our ®ndings agree with previous observations that IgA memory to mucosal pathogens is short-lived and requires periodic boosting to be maintained, as often occurs after repeated reexposure to endemic enteric pathogens in the environment (Bernstein et al., 1989; Saif and Fernandez, 1996) . At PID 83, the spleen had the highest numbers of IgG memory B-cells compared to the other tissues, including the bone marrow, suggesting that the spleen, and not the bone marrow was the major resident site for longer-term IgG memory B-cells after an intestinal viral infection (Yuan et al., 2001a) . Our ®ndings for the resident sites for the shortterm (up to PID 35) and longer-term (at PID 83) memory B-cells agree with the two-phase theory for memory T-cells (Mackay and von Andrian, 2001) . However the time scale of the two phases observed for memory B-cells in pigs orally infected with virulent or attenuated Wa HRV was shorter compared to that investigated for memory CD8 T-cells in mice infected with vesicular stomatitis virus (Masopust et al., 2001) , and in the latter studies the short-term phase was proposed to last for weeks to months (up to PID 296) and the long-term phase supposedly lasts for years (Mackay and von Andrian, 2001) . The difference in the observed persistence of memory cells is consistent with the different lengths of protective immunity induced by local (short-term) versus systemic infections (longer-term) with viral or bacterial pathogens.

According to the conventional immunological dogma based on observations in mice, bone marrow was considered to be the major site for long-term antibody production (Benedetti et al., 1998) . In contrast to this concept, our study demonstrated that bone marrow is not a major site for IgA and IgG rotavirus-speci®c antibody production or for memory B-cells in pigs from PID 28 up to at least PID 83 after oral inoculation with HRV, as indicated by the lowest mean numbers of IgA and IgG memory B-cells and ASC in bone marrow compared to the other tissues tested (Yuan et al., 2001a) . Studies of mice also showed that no IgA and IgG ASC or memory B-cells were detected in bone marrow after adult mice were orally inoculated with the heterologous rhesus rotavirus strain RRV from PID 21 to PID 126 (Moser and Of®t, 2001) . Williams et al. (1998) reported that rotavirus-speci®c, FACS-sorted a4b7 IgA memory B-cells adoptively transferred from donor mice seven months after oral murine EC strain rotavirus inoculation into RAG-2 mice preferentially resided in the lamina propria with ASC numbers at least 250-fold higher than that in bone marrow. It was shown that the memory B-cell subset responsible for the secretory IgA response and protective humoral immunity to rotavirus expressed the intestinal homing receptor, a4b7. In conclusion, after rotavirus infection of pigs or mice, a substantial subset of memory B-cells migrate to and reside in the lamina propria of the small intestine and another major subset of memory B-cells reside in the spleen. Memory B-cells resident in the lamina propria of the small intestine likely mediate protective immunity upon reinfection with rotavirus. The memory B-cells resident in the spleen may be a source for long-term antibody production (mainly IgG antibody) in the circulation, but were not associated with protective immunity to reinfection with rotavirus. Bone marrow was not a major site for rotavirus-speci®c antibody production or for memory B-cells in pigs and mice orally infected with rotavirus.

Novel delivery systems, such as use of recombinant attenuated bacteria as carriers of heterologous antigens (Gentschev et al., 2001) , encapsulation of antigens into microspheres, or ISCOMs have been investigated for enhancing the immunogenicity of mucosal vaccines. The goal of these novel delivery systems is to deliver the antigens more ef®ciently to mucosal lymphocyte populations (i.e., Peyer's patch) after inoculation, or to increase the persistence of antigen in lymphoid tissues at mucosal inductive sites. A newly developed coronavirus expression system (Enjuanes et al., 2001) showed attractive features and great potential for use as a virus vector for delivery of mucosal vaccines. Since coronaviruses infect the mucosal surfaces of the respiratory and/or enteric tracts, they can target the vaccines to the mucosal inductive sites like the natural infections. Non-pathogenic coronavirus strains infecting most species of interest, including humans, swine, and cattle are available to develop expression systems. In addition, the tissue and species tropism of coronavirus may be manipulated through engineering the S gene which is the tropism determinant; thus the expressed vaccine antigens may target to speci®c tissues in speci®c animal species (Enjuanes et al., 2001) . The use of the coronavirus expression system as a potential delivery vehicle for rotavirus vaccines is of interest to evaluate in the gnotobiotic pig model of rotavirus diarrhea.

Microencapsulation has been extensively studied for mucosal vaccination because of its attractive features (O'Hagan et al., 1993; O'Hagan, 1998) , including delivery of encapsulated antigens to provide protection from stomach acidity and the myriad proteases present in the gastrointestinal tract. Microencapsulation has the potential to disseminate antigen to various lymphoid tissues dependent upon the size of the microsphere; therefore, manipulation of the size of the microcapsules allows targeting of vaccine antigens to speci®c sites. Varying the ratio of lactide to glycolide allows antigen release over a long or short period of time. Thus it may be possible to deliver a primary immunization plus a booster immunization with a single dose of vaccine. Inoculation of mice orally with microencapsulated rotavirus enhanced the frequencies of virus-speci®c IgA ASC and the quantities of virus-speci®c IgA antibodies produced in the intestinal lymphoid tissues (Moser et al., 1998) . Use of microencapsulated Wa HRV in the combined AttHR/ VLP2Â vaccine regimen may further improve vaccine ef®cacy by providing the prolonged presence of rotavirus antigen in the inductive sites of the small intestine and enhancing the homing of antigen-speci®c T-and B-lymphocytes after i.n. boosting.

ISCOMs are another attractive approach for delivery of mucosal vaccines. The ISCOMs are 30±40 nm highly structured particles composed of antigen, cholesterol, phospholopids and adjuvant Quilaja saponins (Rimmelzwaan and Osterhaus, 1995) . The advantages of ISCOM vaccines include their potential to induce high local and systemic antibody responses and CD4 T-cell responses, even in the presence of pre-existing antibodies; and their capacity to induce MHC class I restricted CD8 CTL responses, which are important in the clearance of established virus infections (Osterhaus and Rimmelzwaan, 1998) . ISCOMs have been shown to induce antibody responses and activate Th cells and cytolytic T lymphocytes in a number of animal species, including non-human primates and humans (Sjolander et al., 2001) . The immune modulating effects of ISCOMs for an oral non-replicating rotavirus vaccine was evaluated in gnotobiotic lambs (van Pinxteren et al., 1999) . ISCOMs and ISCOMmatrix preparations enhanced virus-speci®c IgG ASC responses after vaccination and signi®cantly increased IgA and IgG ASC responses in the blood after challenge compared to lambs given inactivated virus alone. This study showed that different oral ISCOM vaccines could induce different types of immune responses, either Th1-like (recombinant UK VP6 ISCOMs) or Th2-like (inactivated rotavirus mixed with ISCOM-matrix). Besides the oral route, ISCOMs have been also evaluated for i.m. and i.p. immunization routes. Intraperitoneal injection of ISCOMs recruited and activated many components of the innate immune system, including neutrophils, macrophages, and dendritic cells (Smith et al., 1999) . It was suggested that ISCOMs act by targeting antigen and adjuvant to macrophages and/or dendritic cells (Smith et al., 1999) . This pathway may be important to exploit for vaccine development, especially if combined with another vaccine type or vaccination route, plus a different mucosal adjuvant, such as mLT. We are currently evaluating the ef®cacy of oral or i.n. VLP ISCOM vaccines in the neonatal gnotobiotic pig model, alone or combined with priming by oral live attenuated Wa HRV. Further exploring the combination of multiple vaccination routes, vaccine types, adjuvants, and new mucosal delivery systems may lead to the optimal stimulation of the intestinal mucosal immune system and the maximal ef®cacy of mucosal vaccines against enteric viruses.

Withdrawal of rotavirus vaccine recommendation

Long-term antibodies after an oral immunization with cholera toxin are synthesized in the bone marrow and may play a role in the regulation of memory B-cell maintenance at systemic and mucosal sites

Induction and persistence of local rotavirus antibodies in relation to serum antibodies

Regional specialization in the mucosal immune system: primed cells do not always home along the same track

Current status and future priorities for rotavirus vaccine development, evaluation and implementation in developing countries

Protective effect of rotavirus VP6-speci®c IgA monoclonal antibodies that lack neutralizing activity

Enumeration of isotype-speci®c antibody-secreting cells derived from gnotobiotic piglets inoculated with porcine rotaviruses

Protective immunity induced by rotavirus DNA vaccines

Protective immunity induced by oral immunization with a rotavirus DNA vaccine encapsulated in microparticles

Antibody-independent protection against rotavirus infection of mice is stimulated by intranasal immunization with chimeric VP4 or VP6 proteins

Viral enteric diseases

Rotavirus vaccine administered parenterally induces protective immunity

Virus-like particles as a rotavirus subunit vaccine

Oral tolerance and antipathological vaccines

Structural and functional abnormalities of the small intestine in infants and young children with rotaviral enteritis

Coronavirus derived expression systems

Comparison of mucosal and systemic humoral immune responses and subsequent protection in mice orally inoculated with a homologous or a heterologous rotavirus

Isotype-speci®c antibody responses to rotavirus and virus proteins in cows inoculated with subunit vaccines composed of recombinant SA11 rotavirus core-like particles (CLP) or virus-like particles (VLP)

Immunity to rotavirus infection in mice

Immunity to homologous rotavirus infection in adult mice

Bacterial toxin as mucosal adjuvants

Recombinant attenuated bacteria for the delivery of subunit vaccines

Protection against rotavirus infections by DNA vaccination

Effects of maternal antibodies on protection and development of antibody responses to human rotavirus in gnotobiotic pigs

Homing potentials of circulating lymphocytes in humans depend on the site of activation

Pathogenesis of rotavirus infection in mice

Memory T cellsÐlocal heroes in the struggle for immunity

Preferential localization of effector memory cells in nonlymphoid tissue

Fecal antibody responses to symptomatic and asymptomatic rotavirus infections

Evidence for a common mucosal immunologic system. I. Migration of B immunoblasts into intestinal, respiratory, and genital tissues

Active protection against rotavirus infection of mice following intraperitoneal immunization

Stimulation of local immunity and protection in mice by intramuscular immunization with triple-or double-layered rotavirus particles and QS-21

Effects of different adjuvants on rotavirus antibody responses and protection in mice following intramuscular immunization with inactivated rotavirus

Antibody-dependent and -independent protection following intranasal immunization of mice with rotavirus particles

Distribution of rotavirus-speci®c memory B cells in gut-associated lymphoid tissue after primary immunization

Effect of water-based microencapsulation on protection against EDIM rotavirus challenge in mice

The anatomical basis of intestinal immunity

Noninfectious rotavirus (strain RRV) induces an immune response in mice which protects against rotavirus challenge

Microparticles and polymers for the mucosal delivery of vaccines

Biodegradable microcapsules for oral immunization

Rotavirus virus-like particles administered mucosally induce protective immunity

Rotavirus 2/6 virus-like particles administered intranasally with cholera toxin, Escherichia coli heat-lablie toxin (LT), and LT-R192G induce protection from rotavirus challenge

Protection of the villus epithelial cells of the small intestine from rotavirus infection does not require immunoglobulin A

Induction of virus-speci®c immunity by iscoms

Serum and intestinal isotype antibody responses to Wa human rotavirus in gnotobiotic pigs are modulated by maternal antibodies

Human circulating speci®c antibody-forming cells after systemic and mucosal immunizations: differential homing commitments and cell surface differentiation markers

Visualizing the generation of memory CD4 T cells in the whole body

A novel generation of viral vaccines based on the ISCOM matrix

Expression of the mucosal homing receptor a4b7 correlates with the ability of CD8 memory T cells to clear rotavirus infection

Expression of mucosal homing receptor a4b7 by circulating CD4 cells with memory for intestinal rotavirus

Mucosal immunity: an overview and studies of enteric and respiratory coronavirus infections in a swine model of enteric disease

Enteric viral infections of pigs and strategies for induction of mucosal immunity

Group A rotavirus veterinary vaccines

Rotavirus gastroenteritis

Enteric virus vaccines: theoretical considerations, current status, and future apporaches

Animal rotaviruses

Comparative studies of the pathogenesis, antibody immune responses, and homologous protection to porcine and human rotaviruses in gnotobiotic piglets

Immune responses to ISCOM(R) formulations in animal and primate models

Immune-stimulating complexes induce an IL-12-dependent cascade of innate immune responses

Isotype-speci®c antibody-secreting cells to transmissible gastroenteritis virus and porcine respiratory coronavirus in gut-and bronchusassociated lymphoid tissues of suckling pigs

Contribution of antibody-secreting cells induced in mucosal lymphoid tissues of pigs inoculated with respiratory or enteric strains of coronavirus to immunity against enteric coronavirus challenge

Serum antibody as a marker of protection against natural rotavirus infection and disease

Development of an adult mouse model for studies on protection against rotavirus

Pathogenesis of an attenuated and a virulent strain of group A human rotavirus in neonatal gnotobiotic pigs

Development of mucosal and systemic lymphoproliferative responses and protective immunity to human group A rotavirus in a gnotobiotic pig model

Evaluation of killed and modi®ed live porcine rotavirus vaccines in cesarean derived colostrum deprived pigs

The memory B-cell subset responsible for the secretory IgA response and protective humoral immunity to rotavirus expresses the intestinal homing receptor, a4b7

Systemic and intestinal antibody-secreting cell responses and correlates of protective immunity to human rotavirus in a gnotobiotic pig model of disease

Antibody-secreting cell responses and protective immunity assessed in gnotobiotic pigs inoculated orally or intramuscularly with inactivated human rotavirus

Intranasal administration of 2/6-rotavirus-like particles with mutant Escherichia coli heat-labile toxin (LT-R192G) induces antibody-secreting cell responses but not protective immunity in gnotobiotic pigs

Short-term immunoglobulin A B cell memory resides in intestinal lymphoid tissues but not in bone marrow of gnotobiotic pigs inoculated with Wa human rotavirus

Protective immunity and antibody-secreting cell responses elicited by combined oral attenuated Wa human rotavirus and intranasal Wa 2/6-VLPs with mutant Escherichia coli heat-labile toxin in gnotobiotic pigs