Human Antibody Responses to Bovine (Newbury-2) Norovirus (GIII.2) and Association to Histo-Blood Group Antigens

had not only higherfrequency of antibodies(31.3%) compared with Le(a þ b (cid:2) ) (11%) ( P ¼ 0.068) but also higher antibody titer ( P ¼ 0.085) although this was not signiﬁcant statistically. No detectable cross-reaction between bovine GIII.2 and human GII.3 NoV VLP was found with human and animal sera. The results of this study suggest that bovine norovirus infections occur in Sweden and that secretor status but not ABO blood groups is a possible risk factor for infection. J. Med. Virol. 82:1241– 1246, 2010. (cid:1) 2010 Wiley-Liss, Inc.

Serum antibodies to bovine norovirus have been found recently in about 22% of humans. Whether this prevalence reflects limited virulence properties of the virus or that inherited host factors provide protection against bovine norovirus infection in humans remains to be established. To investigate whether histo-blood group antigens correlate with the presence of bovine norovirus (GIII.2) antibody, plasma (n ¼ 105) from Swedish blood donors, genotyped and phenotyped for secretor, Lewis and ABO, were tested and compared for the frequency of IgG antibody and antibody titer to Bo/Newbury2/ 76/UK. In total, 26.7% (28/105) of Swedish blood donors were antibody-positive. Two nonsecretors (2/21, 9.5%) were antibody-positive compared with 26/84 (31%) secretors (P ¼ 0.047). While no statistically significant correlation was found between the frequency of antibodies to bovine norovirus and different ABO blood groups, individuals with blood type B presented the highest frequency of antibodies (37.5%) compared with 0-30% among other blood groups. Individuals with Le(aÀbþ) had not only higher frequency of antibodies (31.3%) compared with Le(aþbÀ) (11%) (P ¼ 0.068) but also higher antibody titer (P ¼ 0.085) although this was not significant statistically. No detectable cross-reaction between bovine GIII.2 and human GII.3 NoV VLP was found with human and animal sera. The results of this study suggest that bovine norovirus infections occur in Sweden and that secretor status but not ABO blood groups is a possible risk factor for infection. J. Med. Virol. 82:1241-1246, 2010. ß 2010 Wiley-Liss, Inc.

INTRODUCTION
Bovine norovirus (NoV) forms a distinct third genogroup in the NoV genus [Oliver et al., 2003] and can be further divided into two genotypes. Genotype 1 is represented by the Jena virus [Liu et al., 1999] and genotype 2 by the Newbury agent 2 NoV [Dastjerdi et al., 1999;Oliver et al., 2007]. At present, there is limited information about how common bovine NoV infections are in humans [Oliver et al., 2003]. It has been reported recently that bovine NoV recognize an aGal epitope not present in human tissues suggesting that bovine NoV may not infect humans [Zakhour et al., 2009]. In contrast to this observation, Widdowson et al. [2005] found that 22% of the population of the Netherlands had antibodies to bovine NoV with veterinarians having a slightly higher frequency of antibodies.
Whether host genetic markers are associated with this susceptibility remains to be identified, but Lewis phenotype and secretor status determined by the FUT2 gene have been found to be associated with susceptibility to human NoV infections [Lindesmith et al., 2003[Lindesmith et al., , 2005Hutson et al., 2005;Thorven et al., 2005;Larsson et al., 2006;Kindberg et al., 2007;Tan et al., 2008;Bucardo et al., 2009;Kindberg and Svensson, 2009]. The secretor status of an individual describes the presence or absence of histo-blood group antigens on the gastrointestinal epithelia and in body fluids such as saliva. A Caucasian secretor individual has at least one functional allele of the FUT2 gene being either of the SeSe or Sese 428 genotype, whereas a nonsecretor is homozygous for non-functional alleles, the se 428 se 428 genotype. Lewis status is determined by the FUT3 gene and can become non-functional due to several single nucleotide polymorphisms.

Human Plasma
Plasma samples from 105 Swedish blood donors (26-68 years) were genotyped for secretor (FUT2) and Lewis (FUT3) and phenotyped for ABO and Lewis antigens [Larson et al., 1999;Larsson et al., 2006]. Genotyping on DNA, extracted from buffy coats, was performed using PCR-RFLP specific for the FUT2 428G > A and FUT3 59T > G, 202T > C, 314C > T, 508G > A and 1067T > A mutations. Phenotyping for ABO and Lewis was performed on red blood cells using hemagglutination assays with Novaclone TM -monoclonal ABO, anti-Le a and anti-Le b typing antisera (Dominion Biologicals Ltd, Dartmouth, Canada) and with Gamma-clone TM anti-Le a and anti-Le b typing antisera (Gamma Biological, Inc., Houston, TX). Ninety-three samples were further characterized by flow cytometry using Seraclone TM anti-Le a and anti-Le b typing antibodies (Biotest AG, Dreieich, Germany) [Larson et al., 1999;Larsson et al., 2006].
Production of Bovine Newbury-2 GIII.2 VLP and Human GII.3 VLP Production of Bo/Newbury2/76/UK VLP has been previously described [Oliver et al., 2007]. Briefly, Sf9 cell monolayers were infected with recombinant baculovirus at a multiplicity of infection of five and incubated in Hinks medium supplemented with 1% fetal calf serum for 5 days at 268C. VLPs were extracted from the supernatant of freeze-thawed, infected Sf9 cells. VLPs were purified by isopicnic centrifugation in cesium chloride gradients.

Detection of Antibodies to Bovine Norovirus
Detection of Bo/Newbury2/76/UK NoV antibodies was carried out as described previously for human NoV [Hinkula et al., 1995;Larsson et al., 2006]. Briefly, purified VLPs (Newbury) were coated (0.5 mg/ml) (0.05 M sodium carbonate, pH 9.5-9.7) (50 ml/well) in 96-well ELISA plates (Nunc Maxisorp, Thermo Fisher Scientific, Roskilde, Denmark) for 2 hr at 378C followed by overnight at 48C. After five washes with 0.9% NaCl/ 0.05% Tween 20 (Sigma-Aldrich, St. Louis, MO) plates were blocked with 3% bovine serum albumin (BSA) (Sigma-Aldrich, St. Louis, MO) in PBS for 1 hr at 378C. For determination of frequency of antibodies and antibody titers, plasma samples were serially twofold diluted starting at 1/100 in dilution buffer (0.5% BSA/ 0.05% Tween 20/PBS) and incubated for 2 hr at 378C. After five additional washes a horseradish peroxidelabeled goat-anti-human IgG antibody (Bio-Rad Laboratories, Hercules, CA) diluted 1/10,000 in dilution buffer was used as conjugate and incubated at 378C for 2 hr. Following five washes, tetramethylbenzidine (TMB) (DAKO, St. Louis, MO) was added as substrate. After stopping the reaction with sulfuric acid, the absorbance of the wells was measured at 450 nm. The mean value of negative controls (without plasma) plus three standard deviations was used as the cutoff value (0.1 OD). All samples were tested in duplicates.

Western Blot Analysis
To determine the degree of antibody cross-reactivity between bovine (GIII.2) and human (GII.3) NoV a Western blot was designed. Following separation in a 10% SDS-polyacrylamide gel (PAGE) proteins were transferred to a PVDF (0.45 mm) membrane (Millipore, Bedford, MA) for 1 hr at 375 mA and the remaining protein binding sites blocked by 3% BSA/PBS for 1 hr. After washing 2 Â 5 min (0.05% Tween 20/PBS), a mouse anti-bovine (GIII.2) VLP antibody (1/5,000 in dilution buffer) or a rabbit-anti-human GII.3 NoV (1/1,000 in dilution buffer) antibody was added and incubated with the corresponding membrane at 48C overnight. After washing 3 Â 5 min, a horseradish peroxidaseconjugated goat-anti-mouse IgG (1/20,000 in dilution buffer) (Bio-Rad) or peroxidase-conjugated goat-antirabbit IgG (1/20,000 in dilution buffer) (Bio-Rad) were added and incubated for 90 min at room temperature. Washing 3 Â 5 min was followed by 5 min incubation with ECL kit solution (Bio-Rad) and exposure to X-ray film (CEA AB, Strangnas, Sweden) or development by AEC staining kit (Sigma, Germany).
In addition, human sera selected randomly were used as primary antibody at different dilutions together with peroxidase-conjugated goat-anti-human IgG diluted 1/ 10,000. Incubation times and temperatures were as described above.

Statistical Calculations
To compare antibody titers the non-parametric Mann-Whitney two-sided test for significance was used (SPSS software 13.0); P < 0.05 was considered significant.

Human Antibody Prevalence to Bovine Newbury-2 Norovirus and Secretor Status
Of the 105 individuals investigated for antibody prevalence to the bovine NoV (Bo/Newbury2/76/UK) strain, 28 (26.7%) were antibody-positive (titer !200). The geometric mean antibody titers (GMT) were generally low (Fig. 1) compared with antibody titers to human NoV [Larsson et al., 2006] and high antibody titers (!800) were only found in 4/105 (3.8%) individuals. Two of these were of the type B blood group.
Of the 28 antibody-positive individuals two were nonsecretors compared with 26 among secretors (Table I) (P ¼ 0.047). Neither of the two antibody-positive non-secretors had antibody titers >400 compared with secretors where the highest titer was 3,200 ( Fig. 2A). Antibody titers were generally higher among secretors compared with non-secretors (P ¼ 0.082) although not statistically significant.
No significant correlation between frequency of antibodies and different Lewis phenotypes could be found (P ¼ 0.068). However, Le(aÀbþ) individuals had higher antibody titers than Le(aþbÀ) individuals (Fig. 2B) but this difference did not reach significance (P ¼ 0.085). Furthermore, no significant difference was found among different Lewis genogroups and antibody titer.
The number of antibody-positives within different ABO phenotypes spanned 0-37.5% with the highest antibody frequency among blood group B individuals (37.5%) ( Table I). No significant correlation between frequency of antibodies and ABO phenotypes was found. Worth noting, however, is that none (0/7) of the AB individuals were antibody-positive. The highest antibody titer (3,200) was found among individuals with blood groups A and B (Fig. 2C) compared with blood group 0 (800) and AB (negative). Among secretor individuals, a significant difference in antibody titers (P ¼ 0.02) were seen with individuals of blood group B showing higher antibody titers than individuals of blood group 0.
With blood donors spanning 26-68 years of age the possible correlation between frequency of antibodies and antibody titers to Newbury-2 NoV and age was investigated. The results showed that individuals born in the 1930s and 1940s (n ¼ 21) had higher frequency of antibodies (43%) and antibody titers (GMT 126) than individuals born in the 1970s (n ¼ 17) (12%, GMT 64; P ¼ 0.048, P ¼ 0.056).

No Detectable Cross-Reactivity Between
Bovine GIII.2 and Human GII.3 Limited cross-reactivity between human GI and bovine GIII.2 NoV has been previously reported Oliver et al., 2006]. To investigate if crossreactivity between human GII.3 and the bovine GIII.2 Bo/Newbury/76/UK strain would interfere with the interpretation, a Western blot was established to complement the ELISA results. To investigate if mouse anti-bovine NoV and rabbit anti-human NoV crossreacted with the heterologous strains, bovine (GIII.2) and human (GII.3) VLP proteins were separated by SDS-PAGE (Fig. 3A) and processed for Western blotting. As illustrated, mouse anti-bovine GIII.2 did not recognize GII human VLPs (Fig. 3B), and rabbit-anti-human GII.3 did not recognize the bovine VLP (Fig. 3C) all suggesting undetectable cross-reactivity between the VLPs in these assays. Further support for the absence of cross-reactivity was found with the human sera, some of which had high antibody titers (6,400) to human GII.3 NoV but with no detectable antibodies ( 100) to bovine NoV. While it have been shown previously that all seven AB individuals were antibody-positive for human GII.4 NoV [Larsson et al., 2006] they were all antibody-negative for the bovine strain. However, to further investigate cross-reactivity between GII.3 and GIII.2, a Western blot was designed using human serum as primary antibody. Figure 3D  shows a Western blot of a human serum with an ELISA titer of 3,200 for GII and no titer ( 100) for GIII. To further confirm that human sera recognize bovine VLP, Figure 3E illustrates binding of human sera to bovine VLP.

DISCUSSION
The present study showed similar antibody prevalence to bovine GIII.2 (26.7%) as the study by Widdowson et al. [2005] where 20% of sera from the general population and 28% of sera from veterinarians in the Netherlands were positive for bovine NoV. However, the results of this study are in contrast to Zakhour et al. [2009] who have suggested that the cellular receptor for bovine NoV is a aGal epitope, not present in humans, all suggesting that bovine NoV does not infect humans. Thus, the presence of antibodies to bovine NoV among humans may indicate the possibility of another receptor for bovine NoV that is present in humans or the possibility of a yet not discovered GIII human NoV giving rise to antibodies that cross react with Bovine NoV GIII.2.
Susceptibility for human NoV infections have been demonstrated to be associated with secretor status, with non-secretors being almost completely resistant to symptomatic infection [Lindesmith et al., 2003[Lindesmith et al., , 2005Hutson et al., 2005;Thorven et al., 2005;Larsson et al., 2006;Kindberg et al., 2007;Tan et al., 2008;Bucardo et al., 2009;Kindberg and Svensson, 2009]. However, secretor-independent infections have been documented [Lindesmith et al., 2005;Rockx et al., 2005;Carlsson et al., 2009;Nordgren et al., 2010] and antibody prevalence studies have shown that nonsecretors can be infected [Larsson et al., 2006]. Since the same blood donors have been used in an earlier study [Larsson et al., 2006] it was possible to investigate not only similarities and differences in frequency of antibodies and antibody titer between bovine NoV (GIII.2) and human NoV (GII.4) but also a possible association with histo-blood group antigens. A significant difference was found in frequency of antibodies to bovine GIII.2 between non-secretors and secretors (P < 0.047) with secretors having higher frequency of antibodies indicating that they might be more susceptible to infections. While all high antibody titer (!800) individuals were secretors, no non-secretors had antibody titers higher than 400. Non-secretors generally had lower antibody titers than secretors and these high antibody titer secretors could be the result of recent or repeated exposure to bovine GIII.2 NoV.
While no significant difference between frequency of antibodies and antibody titer to different ABO blood groups were found, individuals of blood group B had the highest frequency of antibodies to bovine NoV, which is in contrast to previous observations investigating seroprevalence to human GII.4 NoV [Larsson et al., 2006]. However, when looking at secretors only, blood group B individuals had significantly higher antibody titers compared with blood group 0 (P ¼ 0.02). No individuals of blood group AB (0/7) with antibodies against the bovine strain was found, which is in contrast to an earlier study, where the same seven AB individuals were all sero-positive for human GII.4 NoV [Larsson et al., 2006]. While this suggests that AB individuals may be less prone to infection with bovine NoV compared with individuals of other blood groups, the restricted number of AB individuals makes the conclusion uncertain.
Similar to a previous study from the Netherlands [Widdowson et al., 2005] an association between age and frequency of antibodies to bovine NoV was found, demonstrating that individuals born in the 1930s and 1940s had higher frequency of antibodies than individuals born in the 1970s (P ¼ 0.048). A similar correlation was also seen for antibody titers between age groups (P ¼ 0.056). A possible explanation for these observations is that individuals born in the 1930s and 1940s have been exposed repeatedly to bovine NoV during their lifetime.
Limited cross-reactivity between human and bovine NoV strains has been documented previously [Han et al., 2005;Widdowson et al., 2005;Batten et al., 2006;Oliver et al., 2006]. While Han and co-workers did not find cross-reactivity between bovine NoV and human GI and GII using an antigen ELISA [Han et al., 2005] a cross-reactive epitope between human GI and bovine GIII.2 NoV Oliver et al., 2006] has been reported. Limited cross reactivity has also been found between human GII.3 and bovine GIII.1 . In this study, no cross-reactivity by ELISA could be detected between bovine GIII.2 NoV and human NoV GII.3 using human and hyper-immune sera. Indeed, in several cases individuals with high antibody titers to human GII.4 NoV [Larsson et al., 2006] were antibody-negative to Newbury NoV, which strongly suggests that no or limited cross-reactivity exists between human GII.4 and bovine GIII.2 NoV. Furthermore, while all seven AB individuals are seropositive for human GII.4 NoV [Larsson et al., 2006] none of them were antibody-positive for bovine GIII.2 NoV. It has been suggested in earlier studies that bovine noroviruses are more closely related to GI NoV than GII NoV [Dastjerdi et al., 1999;Liu et al., 1999;van Der Poel et al., 2000]. This further supports the conclusion of no or limited cross-reactivity between bovine GIII.2 and human GII.3 NoV. However, the possibility of limited cross-reactivity between GIII.2 and certain human noroviruses other than GII.3 cannot be completely ruled out.
Oliver and co-workers have not previously found evidence of bovine NoV strains circulating in human; however, it does not exclude the possibility of a rare zoonotic transfer [Oliver et al., 2003] nor does it rule out the possibility of recombination between bovine and human strains. Even though bovine antibody titers were generally low in this study, a few secretors presented antibody titers !800, suggesting repeated exposure to bovine GIII.2 NoV or related viruses.