Norovirus Capture with Histo-Blood Group Antigens Reveals Novel Virus-Ligand Interactions

Norovirus specimens.

Norovirus clinical stool specimens were obtained from individuals challenged with live NV or Snow Mountain virus (SMV) or from various norovirus outbreak samples. The norovirus outbreaks occurred over at least a 5-year period, and the viruses were archived and previously demonstrated to represent various norovirus genotypes (52; C. Moe, unpublished data; J. Vinjé, unpublished data). Various strains from the outbreak specimens were identified and retyped in a blind fashion by reverse transcription-PCR (RT-PCR) followed by DNA sequencing of the RT-PCR products. When necessary to produce a liquid suspension, ∼20% stool/phosphate-buffered saline (PBS) sample solutions were made and clarified by low-speed centrifugation. Viral RNA was extracted from stool samples using Qiamp viral RNA minikits (Qiagen, Valencia, Calif.). A broadly reactive RT-PCR amplification targeting a 327-bp region of the polymerase gene was performed using a One-Step RT-PCR kit (Qiagen) and modified JV12/JV13 primers (52). RT-PCR-positive products were sequenced and aligned to previously published sequences to identify specimens that represent distinct norovirus genotypes. Human challenge specimens and outbreak samples representing different norovirus genotypes were then subjected to titer determination by end-point dilution RT-PCR using the same primers described above, except for Norwalk virus, where strain-specific primers were used (15).

Viral capture and purification method.

A total of 50 μl of streptavidin-coated magnetic bead suspensions (M270; Dynal Biotech, Oslo, Norway) were rinsed three times with 1 ml of sterile PBS in a 1.5-ml microcentrifuge tube. The beads were then preblocked with 5% milk in PBS (5% BLOTTO) for 1 h at ambient temperature on an end-over-end mixer. Synthetic, biotinylated histo-blood group antigens were obtained from Glycotech (Rockville, Md.) and resuspended in PBS at 1 mg/ml. A 50-μl volume of the synthetic carbohydrates was added to the preblocked magnetic beads, and the mixture was incubated at ambient temperature for 2 h on an end-over-end mixer. The beads were then rinsed three times with PBS-Tween (0.05% Tween) and blocked for 1 h at ambient temperature and then overnight at 4°C in 5% BLOTTO containing 0.25% Tween, while continually kept on an end-over-end mixer. The beads were then rinsed three times with PBS-Tween and resuspended in 5% BLOTTO. A norovirus-containing stool specimen (1 to 10 μl) with 10⁵ PCR-detectable units of virus was then added to the bead suspensions, and the mixture was incubated at ambient temperature for 4 h on an end-over-end mixer. The beads were then washed five times with PBS-Tween and three times with sterile PBS and subsequently resuspended in 50 μl of sterile, RNase-free PBS. The washed beads were then 10-fold serially diluted three times and stored at −80°C until tested by RT-PCR.

RT-PCR detection of norovirus attachment.

RT-PCR detection of viral attachment was conducted in a blind fashion, with no prior knowledge of the norovirus strains or histo- blood group carbohydrates being tested for attachment. First, encapsidated viral RNA was heat released from the bead suspensions by incubating 10 μl of each bead dilution at 99°C for 5 min. Then 2.5 μl of the heat-released RNA, as well as RNA extracted and purified from the original stool sample as a positive control, was subjected to RT-PCR using the same primers and conditions for determining the titers of the samples. RT-PCR products were examined on an ethidium bromide-stained 2% agarose gel.

VLP attachment assays.

Two microwell-based assays were designed to detect and quantify norovirus VLP attachment to histo-blood group antigens. In the first assay, synthetic biotinylated carbohydrates (2.5 μg/ml in Tris-buffered saline, 100 μl per well) were added to streptavidin-precoated microwells (Thermo Labsystems, Basingstoke, United Kingdom) and the mixtures were incubated for 2 h at 37°C. The microwells were then blocked with 300 μl of 5% BLOTTO overnight at 4°C. Sucrose gradient-purified norovirus VLPs (5, 27) were then added to the carbohydrate-coated microwells at various concentrations and in the presence or absence of various stool extracts (as indicated in Results). The wells were incubated at 37°C for 4 h and then washed five times with PBS-Tween. To detect attachment, 100 μl of pooled polyclonal antisera from SMV, NV, and Hawaii virus (HV)-infected human volunteers was added per well at a 1:100 dilution in 5% BLOTTO and the wells were incubated at 37°C for 2h. The plates were then washed five times with PBS-Tween, a goat anti-human immunoglobulin G (IgG)-alkaline phosphatase conjugate (1:1,000 in 5% BLOTTO, at 100 μl/well) (Sigma, St. Louis, Mo.) was added, and the mixture was incubated for 1 h at 37°C. After a final washing step, p-nitrophenyl phosphatase substrate (pNPP) (Sigma-FAST tablets; Sigma) was added at 150 μl/well and the plates were incubated for 25 to 30 min at ambient temperature. Developed plates were then read at an optical density of 405 nm (OD₄₀₅) using a microplate reader (Molecular Devices, Sunnyvale, Calif.). Wells with no VLPs added were used as plate blanks.

A second VLP attachment assay was used to detect the attachment of carbohydrates to VLP-coated microwells, similar to previously described methods (27). Briefly, 2.5 μg of the appropriate VLP reagent per ml (100 μl/well in Tris-buffered saline) was added to high-binding microwells (Costar, Corning, N.Y.) and the plates were incubated for 4 h at ambient temperature and then blocked with 5% BLOTTO overnight at 4°C. Synthetic biotinylated histo-blood group antigens were then added at various concentrations in the presence or absence of stool extracts (as indicated in Results) in 5% BLOTTO, and the plates were incubated at 37°C for 4 h. The plates were then washed five times with PBS-Tween, streptavidin-alkaline phosphatase conjugate (1:500 dilution in 5% BLOTTO, at 100 μl/well) (Sigma) was added, and the mixture was incubated at 37°C for 1 h. Finally, the plates were washed and developed with pNPP and the OD₄₀₅ was determined. Wells with no carbohydrate added were used as plate blanks. Student's t test (unpaired, two-tailed, unequal variance) was used to determine the statistical significance of stool extract-enhanced attachment of SMV VLPs.

Identification and genotyping of clinical norovirus specimens.

To identify high-titer clinical specimens to be used for this study, we first screened and determined the titers of norovirus stool specimens obtained from human challenge studies or outbreak samples for the presence of noroviruses by RT-PCR using both broadly reactive and NV-specific primer pairs. We then sequenced RT-PCR products from norovirus-positive samples and aligned them with previously published norovirus sequences to determine the genotype of the viruses. Finally, we determined the titers of norovirus stool specimens representing various genotypes by end-point RT-PCR using the same primer sets used to identify the norovirus strains. Stool samples with norovirus titers of 10⁷ to 10⁸ RT-PCR-detectable units were identified and used for attachment characterization assays. For a genetic classification of the strains used in this study and a general summary of attachment data, see Fig. 10.

Norovirus capture and histo-blood group antigen attachment characterization.

We designed and optimized a magnetic bead-based method to capture and characterize the histo- blood group antigen attachment properties of noroviruses in clinical stool specimens (Fig. 1A). Using this assay, we evaluated the ability of noroviruses representing five different genotypes to attach to synthetic carbohydrates representative of the various histo-blood group antigens that would be present on gut mucosal cell surfaces of individuals of different secretor and Lewis phenotypes (Fig. 1B). The histo-blood group antigen attachment properties of NV, a genogroup I (GI) strain and the prototype norovirus (GI.1 genotype) have been previously characterized using VLP reagents and a variety of biochemical and histological methods (27, 33, 42). Consistent with these studies, NV obtained from a human challenge stool specimen (39) bound specifically to H type 1, Le^b, and, to a relatively lesser extent, H type 3 (Fig. 2), suggesting that NV VLPs and wild-type virus are biologically indistinguishable in the case of attachment to these histo-blood group antigens.

Also consistent with our previous studies using VLP reagents (27), different norovirus strains exhibited a variety of histo-blood group antigen attachment phenotypes (Fig. 2). A second GI strain, M23, representative of the Desert Shield/Stavanger-virus genotype (GI.3) did not attach to any of the histo-blood group antigens tested. This observation was confirmed by repeating the experiment with a 10-fold-higher concentration of virus (data not shown). Two GII strains, M15 and M116, bound specifically only to Le^b. The M15 virus is representative of the Toronto (GII.3) virus genotype, while M116 represents a novel norovirus genotype (52).

Finally, SMV, which is a prototype GII strain (GII.2), was obtained from a human challenge stool specimen and tested for attachment to the various histo-blood group antigens. Previous studies from our laboratory demonstrated that SMV VLPs did not attach to any of the synthetic carbohydrates used in this study (27). However, we observed that SMV from the clinical stool specimen did in fact attach specifically to H type 3 (Fig. 2).

SMV attachment characterization.

We further investigated the SMV-H type 3 attachment properties to identify the cause of the discrepancy between the SMV VLP and clinical virus attachment observations. First, we tested if SMV VLPs will compete with the virus for attachment to H type 3. Using the magnetic bead method described above, the SMV stool specimen was added to H type 3-coated magnetic beads in either the absence or presence of excess SMV VLPs (10 μg, sucrose pure) or to beads pretreated with the same amount of SMV VLPs and then washed with PBS before addition of the SMV stool specimen. Interestingly, SMV VLPs outcompeted the virus to attach to H type 3 (Fig. 3). However, the SMV VLPs could be washed from the H type 3-coated beads before addition of the SMV stool specimen, restoring the ability of the virus to attach to H type 3 (Fig. 3).

We then used a series of microwell-based assays to further characterize SMV VLP attachment to H type 3. Using a biochemical microwell method, which previously established that SMV VLPs do not attach to H type 3 (27), we first characterized the attachment of three additional batches of sucrose-purified SMV VLPs. Consistent with the previous results, synthetic, biotinylated H type 3 did not attach to microwells coated with any of the three different batches of SMV VLPs (data not shown).

Using another method to detect VLP attachment to carbohydrate-coated microwells, similar to the method previously described by Hutson et. al. (33), SMV VLPs again did not attach to H type 3-coated microwells (Fig. 4). However, when the SMV stool specimen was added to sample buffer along with the VLPs, attachment to H type 3 was detected (Fig. 4). The binding signal appeared in a linear fashion in relation to VLP concentration, suggesting that the SMV VLPs and not the SMV virions in stool attached to H type 3.

Stool contents promote SMV attachment to H type 3.

To determine if SMV VLP attachment is dependent on the presence of SMV virions or on another component of the extracellular environment in the clinical sample, we investigated the ability of other stool extracts to enhance SMV VLP attachment to H type 3. Forty-two archived clinical stool specimens from various norovirus outbreaks, representing at least 10 different norovirus genotypes, were screened for their ability to enhance SMV VLP attachment to H type 3. Remarkably, nearly all of the extracts promoted SMV VLP attachment to some extent (Fig. 5). The mean attachment (net OD₄₀₅ ± standard deviation) of SMV VLPs to H type 3 in the presence of the stool samples was 1.57 ± 0.84, whereas attachment of three different batches of SMV VLPs to H type 3 in the absence of stool samples was 0.08 ± 0.01. The difference in the ability of SMV VLPs to attach to H type 3 in the presence and absence of stool extracts is significant (P < 0.001; Student's t test). No correlation was found between attachment enhancement levels and the presence, absence, genotype, or titer of noroviruses in the outbreak specimens (data not shown). In fact, even prechallenge, virus-free stool extracts from human volunteers promoted SMV attachment regardless of the individuals' secretor phenotype (Fig. 5). Therefore, stool contents alone are most likely promoting SMV attachment to H type 3, suggesting that a component of the extracellular environment in normal human feces might play a role in SMV attachment in vivo.

We further investigated the attachment enhancement mechanism to characterize its specificity for SMV and H type 3. The ability of the SMV stool sample to enhance SMV VLP attachment was specific for H type 3, since attachment to H type 1, the H type 3 precursor, or mock-coated microwells was not detected even in the presence of the stool specimen (Fig. 6). In fact, neither wild-type SMV nor SMV VLPs attached to any of the carbohydrates used in this study other than H type 3, regardless of the presence or absence of stool extracts (Fig. 2 and data not shown).

The attachment enhancement mechanism was not necessarily specific for SMV. The ability of NV VLPs to attach to H type 1, H type 3, or Le^b was not enhanced by the presence of stool contents (Fig. 7A). In addition, no enhancement effect on NV VLP attachment with other histo-blood group antigens was evident (data not shown). However, we previously demonstrated that HV VLPs do not attach to any of the histo-blood group antigens used in this study (27). Although the effect was not as robust as for the SMV VLPs, addition of stool extracts modestly promoted HV VLP attachment specifically to Le^b (Fig. 7B).

We further characterized the attachment enhancement mechanism to begin to dissect the components of stool that might play a role. Enzyme-linked immunosorbent assays of the stool extracts were performed to determine if antibody might play a role in SMV attachment enhancement. Of the 42 different outbreak specimens tested, only a small fraction had detectable levels of norovirus-specific IgA, IgM, or IgG, with no correlation between the presence of detectable norovirus-specific antibody and the ability of the stool extracts to enhance SMV attachment (data not shown). In fact, denaturing the stool samples by boiling did not impair their ability to enhance SMV attachment to H type 3 (Fig. 8), suggesting that neither antibody, enzymes, nor other protein components in stool are responsible for SMV attachment enhancement.

We then determined if pretreatment of the SMV VLPs or H type 3 with the stool extracts was sufficient to detect the attachment of SMV VLPs to H type 3. Before the addition of SMV VLPs, H type 3-coated microwells were pretreated with stool extracts for 2 h and washed to remove unbound stool components. Similarly, SMV VLP-coated microwells were pretreated with stool components and washed before addition of synthetic H type 3. In both cases, SMV VLP attachment to H type 3 was evident only when all three components—stool extract, SMV VLPs, and carbohydrate—were present at the same time (Fig. 9). Pretreatment of the SMV VLPs or H type 3 with stool extracts was not sufficient to promote attachment.