|Home | About | Journals | Submit | Contact Us | Français|
The objective of this study was to determine the prevalence of anti-norovirus (NoV), -sapovirus (SaV) and -Tulane virus (TV) antibodies in rhesus macaques of the Tulane National Primate Research Center and to evaluate the antigenic relationship between these viruses. A high prevalence of NoV-binding (51–61%) and SaV-binding (50–56%) antibodies and TV-neutralizing (69%) antibodies were detected. Serum samples obtained during a human NoV outbreak and a multivalent anti-NoV hyperimmune serum were not able to neutralize TV infectivity. Conversely, low levels of cross-reactivity between the prototype TV and NoVs, but not between the TV and SaVs were detected by ELISA. These data indicate the preservation of some cross-reactive B-cell epitopes between the rhesus and human caliciviruses (CVs). The high prevalence of human and rhesus CV-specific serum antibodies suggests the frequent exposure of colony macaques to enteric CVs including the possibility of CV transmission between human and non-human primate hosts.
Caliciviruses (CVs) are small, non-enveloped, icosahedral viruses with an approximately 7.5–8.5 kb positive-sense, single-stranded, polyadenylated RNA genome. The family Caliciviridae consists of four established genera, Norovirus (NoV), Sapovirus (SaV), Lagovirus and Vesivirus (Green et al., 2000, 2001). The bovine enteropathogenic CVs (Newbury agent-1 and Nebraska), the rhesus enteric CVs (Tulane virus) and the St-Valerian-like CVs represent three currently unassigned new CV genera (Farkas et al., 2008; L'Homme et al., 2009; Oliver et al., 2003; Smiley et al., 2002). NoVs and SaVs are important aetiological agents of acute gastroenteritis in humans with NoVs accounting for 80–90% of non-bacterial gastroenteritis outbreaks, including >50% of all food-related gastroenteritis outbreaks (Fankhauser et al., 1998, 2002; Lopman et al., 2003). Based on phylogenetic analysis, both Norovirus and Sapovirus are further divided into five genogroups and several genetic clusters or genotypes (Farkas et al., 2004; Zheng et al., 2006). Viruses that are genetically and antigenically closely related to human NoVs and SaVs have also been isolated from animals (Dastjerdi et al., 1999; Liu et al., 1999; Martella et al., 2007; Sugieda et al., 1998), raising the concern about CV gastroenteritis as a zoonotic disease.
Studies to develop a primate animal model for human CVs suggest that non-human primates (NHPs) are susceptible to NoV infection but probably without the manifestation of clinical disease (Rockx et al., 2005; Subekti et al., 2002; Wyatt et al., 1978). Direct evidence for natural NoV infection in NHPs is so far missing. Nevertheless, a high prevalence of anti-NoV antibodies (both GI and GII) in serum samples collected from NHP species, including rhesus macaques, was reported (Jiang et al., 2004), and electron microscopy of faecal specimens from NHPs with diarrhoea revealed the presence of NoV-like particles in some specimens (Wang et al., 2007). Our group recently reported the isolation and characterization of the Tulane enteric CV (TV) from stool samples of an asymptomatic juvenile rhesus macaque (Farkas et al., 2008). Whether TV infection is associated with diarrhoea or other clinical symptoms and the prevalence of CV infections in NHPs is currently unknown.
Here, we report the prevalence of anti-CV antibodies in rhesus macaques of the Tulane National Primate Research Center (TNPRC) colony. In addition, we also evaluated the antigenic relationship between TV and human CVs.
A total of 556 serum samples were obtained between 2002 and 2008 from 515 rhesus macaques (Macaca mulatta) of Indian subspecies. While the majority of animals consisted of juvenile (<3 years old) macaques, the sample set contained also sera from 89 young adults (>3 years old). Five hundred serum samples were collected during the 2008 breeding colony inventory – each sample corresponded to a single animal. Fifty-six serum samples were collected from 15 nursery macaques that were each bled at least three times during the period from April 2002 to August 2003. These 56 samples and 132 randomly selected serum samples collected in 2008 were tested for the presence of anti-NoV-binding [Norwalk (GI.1) and MOH (GII.5)] and anti-SaV-binding [Mex14917 (GI.3) and Mex340 (GII.2)] antibodies. In the NoV ELISAs and the SaV ELISAs, virus-like particles (VLP) and SaV capsid fusion proteins, respectively, were used as antigens, as described previously (Farkas et al., 2005, 2006).
All rhesus macaque serum samples (n=556) were tested for the presence of anti-TV virus-neutralizing (VN) antibodies in a cytopathic effect (CPE)-based neutralization assay at 1:20 dilution. Each sample was tested in triplicate wells of a 96-well tissue culture plate (Corning Life Sciences) to neutralize 100 TCID50 of the prototype TV (GenBank accession no. EU391643). To avoid non-specific neutralization by complement, serum samples were heat-inactivated. Briefly, virus/serum mix was incubated at 37 °C for 1 h and transferred onto wells that were seeded with 1×104 LLC-MK2 cells per well 1 day prior to this. Plates were stained with crystal violet at 72 h post-inoculation (Sigma-Aldrich). At this time point, all cells in the virus control exhibited CPE characterized by cellular degeneration and detachment. Serum samples for which the cell monolayer was at least 50% intact in all of the three wells were considered to contain TV-neutralizing antibodies. A subset of rhesus samples (n=82) was end-titrated by twofold dilutions (Supplementary Fig. S1, available in JGV Online).
In addition, 65 human serum samples collected during a gastroenteritis outbreak investigation in 1999 (Farkas et al., 2003) were also tested for anti-TV VN antibodies at 1:10 dilution.
To determine whether cross-reacting epitopes exist among the TV and NoVs or SaVs, a solid phase ELISA similar to that described above for antibody detection was utilized. Anti-TV hyperimmune sera were generated by immunization of BALB/c mice with CsCl gradient-purified inactivated TV. Mouse hyperimmune sera were also generated against affinity column-purified glutathione S-transferase (GST)-TV capsid fusion protein. The production of multivalent anti-NoV hyperimmune rabbit sera obtained by cross-immunization with three GI [C59 (GI.2), Norwalk (GI.1) and VA115 (GI.3)] and five GII [Grimsby (GII.4), Hawaii (GII.1), Mexico (GII.3), VA207 (GII.9) and VA387 (GII.4)] human NoV capsid proteins was performed as described previously (Huang et al., 2001). The mouse anti-TV hyperimmune serum was titrated by twofold dilutions starting at 1:500 on plates coated with NoV VLPs, GST–SaV fusion proteins and GST. The rabbit anti-NoV hyperimmune serum was similarly titrated on plates coated with purified TV.
These anti-TV and anti-NoV hyperimmune mouse or rabbit serum samples were also evaluated for their ability to neutralize the prototype TV in tissue culture-based plaque reduction assay.
High prevalence of NoV- and SaV-specific antibodies was detected in TNPRC macaques (Table 1). Of the 188 total serum samples tested, 114 (61%) and 97 (52%) contained antibodies against Norwalk and MOH viruses, respectively. Moreover, a similarly high seroprevalence was observed against SaVs, with 94 (50%) and 106 (56%) of the 188 samples giving positive results in the Mex14917 (GI.3) and Mex340 (GII.2) ELISAs, respectively. When the serum samples collected from the 15 nursery macaques between 2002 and 2003 were analysed separately, 26 (46%), 25 (45%), 36 (64%) and 28 (50%) contained antibodies against Norwalk, MOH, Mex14917 and Mex340, respectively. Five, six, one and three of these 15 nursery animals were seronegative against Norwalk, MOH, Mex14917 and Mex340, respectively, at all collection time points. The prevalence of antibodies against Norwalk, MOH, Mex14917 and Mex340 among animals sampled in 2008 was 66, 54, 44 and 59%, respectively. Comparison of NoV and SaV seroconversion rates between these two groups of macaques sampled in 2002–2003 and 2008 revealed no substantial differences, although seroconversion against NoVs and Mex340 SaV was higher and against Mex14917 SaV was lower in the 2008 samples (Table 1). Statistical evaluation of these differences was not performed since different categories of animals were sampled in 2002–2003 (nursery) and 2008 (conventional juveniles).
There was a clear decrease of NoV but not SaV seropositivity among the 1–2- and 2–3-year-old juveniles, with the lowest seropositivity seen among 2–3-year-old animals followed by an increase in adult (>3 years old) animals (Table 2).
The distribution of SaV-specific antibodies in the 15 nursery macaques that were sequentially sampled during a 1 year period revealed that some animals seroconverted only to one of the SaV antigens. Whether this was due to natural resistance to infection against the particular SaV strain or due to other reasons is unknown. These serum samples specifically recognized only the corresponding antigen bands in Western blot immunostaining, confirming the specificity of the ELISA (Supplementary Fig. S2).
Of the 556 total rhesus macaque serum samples collected and tested in this study, 382 (69%) neutralized the prototype TV at ≥1:20 dilution. The prevalence of VN antibodies increased with age from 43% in the animals less than 1 year old to 94% in the animals more than 3 years old (Table 2). Among the 85 samples that were end-titrated, titres varied between 1:20 (25.8%) and ≥1:1280 (2.3%) (Supplementary Table S1).
Neither the human NoV outbreak sera nor the anti-NoV hyperimmune rabbit sera neutralized the TV even at a low (1:10) dilution. Similarly, the mouse sera raised against GST–TV capsid fusion protein had no neutralization activity. However, when the possibility of non-neutralizing (binding) cross-reacting epitopes between the TV and NoVs and/or SaVs was tested by solid phase ELISAs, the multivalent rabbit anti-NoV hyperimmne serum cross-reacted with TV up to a 1:2000 dilution (OD450>0.2). Similarly, the mouse anti-TV hyperimmune serum cross-reacted with all three NoV VLPs tested, but did not recognize the control GST or GST–SaV fusion proteins (Fig. 1).
In this study, consistent with a previous report (Jiang et al., 2004), the presence of anti-NoV antibodies was detected in captive rhesus macaques. In addition, for the first time, we demonstrated a high prevalence of anti-SaV antibodies in juvenile macaques, suggesting that captive NHPs are frequently exposed to human CVs or antigenically related agents. This finding by itself raises important questions regarding the epidemiology, epizootology and pathogenesis of these viruses that need to be thoroughly addressed in the near future.
By the age of 2, humans have high antibody prevalence (>80%) against NoVs and SaVs. The NoV antibody prevalence in humans usually drops around 4–9-months-old due to a decrease in maternal antibodies (transplacental and milk), after which it gradually increases as a result of primary and subsequent NoV infections (Jiang et al., 1995; Peasey et al., 2004). On the other hand, SaV antibody prevalence increases from birth without the drop associated with maternal antibody decrease and reaches a plateau by 2 years of age (Farkas et al., 2006).
Interestingly, our results are consistent with age-related serum NoV antibody drops seen in human infants, however, with a delay. Most of the captive juvenile rhesus macaques at TNPRC are raised in family groups consisting of several females, a male and a group of infants. In these groups, infants and juveniles have access to their mother's milk and most of the conventional animals breast feed up to 2 years of age. Breast milk provides protection (antibodies and milk glycans as decoy receptors) against NoV infection (Newburg et al., 2005). Thus, the delayed drop of NoV-specific antibodies in macaques compared with humans could be due to the delayed weaning of juvenile macaques.
It has been described previously that the major histocompatibility complex class I-related Fc receptor, FcRn, which facilitates the transport of immunoglobulins across epithelium, is expressed on the surface of various cell types in primates, including the intestinal epithelial cells of adults (Dickinson et al., 1999; Roopenian & Akilesh, 2007). The gradual decline of NoV-specific, presumably maternal, antibodies in 2–3-year-old macaques detected in our study is consistent with the above mechanism that contributes to maternal immunoglobulin passage in breast-fed infants even beyond the age of 6 months, when complete gut closure normally occurs (Kramer & Kakuma, 2004).
While the distribution of different NoVs circulating in the colony and the susceptibility of rhesus macaques to NoV infection based on their histo blood group antigen types is not yet known, we hypothesize that differences in the circulating NoV strains may also contribute to differences in age-stratified antibody prevalence between the human and macaque populations.
The fact that we observed drops of presumably maternal antibodies to NoVs but not to SaVs in this study is also consistent with the situation that has been described in children (Farkas et al., 2006). Since the majority of animals in this study were older than 8 months, we were not able to address whether the SaV antibody prevalence was at a lower level in infant macaques than in human infants.
Hyperimmune sera generated against purified TV but not against NoV VLPs neutralized the prototype TV, suggesting that TV and NoVs might represent different serotypes without sharing the virus-neutralizing epitopes. Nevertheless, these hyperimmune sera revealed low but consistent cross-reactivity between the prototype TV and NoVs, although not between the TV and SaVs (Fig. 1). These results also confirmed the specificity of our ELISAs while selectively pointing to the existence of cross-reactive B-cell epitopes among rhesus TVs and human NoVs.
In this study, NHP serum samples were screened at a single dilution of 1:100, while in a previous study by Jiang et al. (2004), this was done at a dilution of 1:200. Despite the relatively low level of cross-reactivity between the TV and NoVs observed here, it is still possible that anti-NoV antibodies seen in rhesus sera at low dilutions may reflect the presence of cross-reactive antibodies generated by TV infection. To eliminate and examine such a scenario, more detailed (epitope mapping) studies could be conducted in the future.
None of the 65 human serum samples neutralized the prototype TV even at a 1:10 dilution. This further corroborates that NoVs and TVs represent different CV serotypes. Whether the TV infections take place in humans and/or whether the NoV/SaV infections take place in NHPs remains to be elucidated. Thus, a larger study involving analysis of serum samples and faecal specimens from different populations and age categories of NHPs and humans, including animal keepers that are in daily contact with NHPs, is warranted.
The inability of the mouse GST–TV capsid fusion protein-specific hyperimmune sera to neutralize the TV indicates that VN epitopes are, in the case of TV, conformation dependent. This could be important for NoV vaccine development assuming that proper expression of the CV capsid protein with preservation of conformationally dependent immunogens (VLP or P particle) is necessary for the induction of protective immunity.
The significance of the TV as a pathogen is currently unknown. In order to assess the pathogenicity of TV, it will be necessary to perform a challenge study with seronegative macaques in order to reproduce the TV infection in NHPs. In this study, 382 (69%) of the 556 serum samples collected from TNPRC rhesus macaques neutralized the prototype TV at ≥1:20 dilution. Moreover, over 10% of the animals shed a wide diversity of TVs in their stool (T. Farkas and others, unpublished data). Taken together, our results indicate that in addition to TV infections, both human NoV and SaV infections are common in colony macaques. Therefore, investigations of possible transmission of these viruses from human to NHPs and vice versa are needed. The fact that we also detected partial cross-reactivity (presence of binding but not VN antibodies) between the TV and NoV further suggests that TV is related to NoV. This characteristic could indicate the suitability of TV as model for NoV.
We thank Drs Rudolf Bohm and Marion Ratterree for their support. Excellent technical assistance from Mrs Amanda Tardo and Nadia Slisarenko is greatly appreciated. We also thank Miss Christine T. Bulot for help with manuscript editing. This study was supported by NCRR/NIAID grant R21RR024871 to K.S. Partial support was also provided by NCRR grants U24 RR018111 to Dr Rudolf Bohn, Chairman of the Division of Veterinary Medicine, Tulane National Primate Research Center, and P51RR000164-47 to K.S.
Two supplementary figures and a supplementary table are available with the online version of this paper.