|Home | About | Journals | Submit | Contact Us | Français|
Noroviruses (NoVs) are the main agents of gastroenteritis in humans and the primary pathogens of shellfish-related outbreaks. Some NoV strains bind to shellfish tissues by using carbohydrate structures similar to their human ligands, leading to the hypothesis that such ligands may influence bioaccumulation. This study compares the bioaccumulation efficiencies and tissue distributions in oysters (Crassostrea gigas) of three strains from the two principal human norovirus genogroups. Clear differences between strains were observed. The GI.1 strain was the most efficiently concentrated strain. Bioaccumulation specifically occurred in digestive tissues in a dose-dependent manner, and its efficiency paralleled ligand expression, which was highest during the cold months. In comparison, the GII.4 strain was very poorly bioaccumulated and was recovered in almost all tissues without seasonal influence. The GII.3 strain presented an intermediate behavior, without seasonal effect and with less bioaccumulation efficiency than that of the GI.1 strain during the cold months. In addition, the GII.3 strain was transiently concentrated in gills and mantle before being almost specifically accumulated in digestive tissues. Carbohydrate ligand specificities of the strains at least partly explain the strain-dependent bioaccumulation characteristics. In particular, binding to the digestive-tube-specific ligand should contribute to bioaccumulation, whereas we hypothesize that binding to the sialic acid-containing ligand present in all tissues would contribute to retain virus particles in the gills or mantle and lead to rapid destruction.
Noroviruses (NoVs) are recognized as the leading cause of epidemics of gastroenteritis and an important cause of sporadic cases in both children and adults (13). Members of the Caliciviridae family, these small round nonenveloped RNA viruses are highly genetically and antigenically diverse. The genetic classification system, based on relatedness of the complete VP1 capsid protein, currently separates the strains into five recognized genogroups (GI to GV), with GI, GII, and GIV infecting humans, while GIII and GV strains infect cows and mice, respectively (41, 57). In addition, a number of new genogroups have been proposed recently (10, 34, 37). Histo-blood group antigen (HBGA) expression is a factor in the genetic resistance of humans to norovirus infection and has been proposed to affect the transmission and epidemiology of noroviruses in human populations (17, 33). HBGAs are complex glycans present on many cell types, including red blood cells and vascular endothelial cells, as well as some epithelial cells (intestinal, urogenital, and respiratory). Volunteer studies and outbreak analyses indicate that binding to these carbohydrates is required for infection, with many strains infecting only a subset of the population based on their HBGA expression (15, 20, 25, 27, 28, 48, 49).
NoV ligands have also been identified in oyster tissues, and as observed in humans, variations between NoV strains have been described (30). Oysters are usually grown in coastal waters and thus exposed to waters potentially contaminated by human activities. A better understanding of the role played by oysters in terms of manner of contamination, pathogen persistence, and pathogen selection may help improve the sanitary quality of shellfish by promoting new methods of prevention or depuration to increase the consumer's safety. We recently demonstrated that the lack of GIII NoV ligand in oysters induced a lower accumulation compared to human NoVs, in terms of both frequency and concentrations, as determined by quantitative real-time reverse transcription-PCR (rRT-PCR) during a field study (56). Likewise, mutant GI virus-like particles (VLPs) unable to recognize the shared human and oyster ligand had a greatly decreased ability to accumulate in oysters (30). Such observations clearly suggested that specific glycan ligands impact bioaccumulation efficiency and supported our earlier hypothesis based on in vitro VLP binding and bioaccumulation (23).
In order to get a more complete picture of the strain-specific potential for contamination of oysters, the aim of the present study was to obtain quantitative data on viral bioaccumulation in oyster tissues by one GI strain and two GII strains by rRT-PCR quantification of virus recovery from in vitro bioaccumulation experiments performed at several time points during the year. In parallel, measurements of ligand expression were made during the same periods.
Fecal samples containing GI.1 (Norwalk virus 8FIIa strain; GenBank accession no. M87661.1), GII.3 (94% identity in the polymerase and beginning of the capsid coding regions to the GII.3 used for the VLPs, Toronto CAN strain; GenBank accession no. AF247431), and GII.4 (Houston strain; GenBank accession no. EU310927) NoVs were used for bioaccumulation experiments. Viral RNAs were extracted from 10% suspensions of stool by using the Nuclisens kit (bioMérieux) as recommended by the manufacturer and were eluted in 100 μl of RNase-free water. Mengovirus strain vMC0 (kindly provided by A. Bosch, University of Barcelona) was propagated in HeLa cells, and the virus titer was determined as described previously (35).
GI.1 and GII.4 VLPs, produced using plasmids containing open reading frames 2 and 3 (ORF2 and -3, respectively) of the Norwalk virus strain 8FIIa and the Houston strain, were used for the enzyme-linked immunosorbent assay (ELISA), as well as their corresponding antibodies produced in rabbits (2, 21). A construct containing ORF2 of a GII.3 strain (accession no. AY247431) and the corresponding antibody were kindly provided by L. Svensson (44).
The following neoglycoconjugates were examined in this study: H type 1-polyacrylamide (PAA) (Fucα2Galβ3GlcNAcβ-R), H type 3-PAA (Fucα2Galβ3GalNAcα-R), and Ley-PAA (Fucα2Galβ4[Fucα3]GlcNAcβ-R), kindly provided by N. Bovin (Moscow, Russia), as well as sialyl-Lex-human serum albumin (HAS) (NeuAcα2,3Galβ4[Fucα3]GlcNAcβ3Galβ4Glcβ-R) and sialyl-LNnT-HSA (NeuAcα2,3Galβ4GlcNAcβ3Galβ4Glcβ-R), purchased from Isosep AB (Uppsala, Sweden).
Fecal samples containing the GI.1, GII.3, and GII.4 strains were diluted in seawater in a volume of 3 ml and kept at 12°C. Sampling was performed after 1 and 24 h. A volume of 1 ml was removed, and nucleic acids were extracted directly.
For all experiments, live oysters (Crassostrea gigas) were purchased directly from the same producer (the same batch kept over the 6-month study period on special request), and environmental data such as water temperature and salinity were monitored on a daily basis with a Marel Smatch TPS (NKE, Hennebont, France). The temperatures measured were 15.3 ± 0.92, 12.2 ± 1.2, 7.7 ± 0.6, and 8.8 ± 1°C in October, November, January, and March, respectively. Oysters were immersed the day after in large tanks of seawater at the laboratory. For all bioaccumulations, control tests performed on oysters assayed before bioaccumulation showed no preexisting GI or GII NoV contamination. After 24 h of immersion at the designated temperature, oysters were individually checked, and only living oysters showing filtration activity were kept.
Natural seawater freshly collected from a single clean area was used for bioaccumulation experiments. The bioaccumulation experiments for all strains and virus concentrations were conducted on the same day for each season. Ten aquariums were filled with 3 liters of seawater each, 9 were artificially contaminated with fecal samples containing GI.1, GII.3, or GII.4 NoV at three different concentrations, and the last aquarium was kept as a control. Twelve oysters were added to each aquarium and incubated for 24 h at 12 ± 1°C for the first two bioaccumulation experiments and 8 ± 1°C for the last two under oxygenation. Six oysters were harvested after 1 h and the remaining six after 24 h. Only six oysters were introduced into the control aquarium, and harvesting occurred at 24 h.
Harvested oysters were shucked just after collection, and the total weights of the six oyster bodies were determined. Dissections were performed immediately after harvesting and were performed at the same time for each strain by different laboratory members to avoid any delays or differences from subsequent assays. Digestive tissues (DT), gills, and mantle were collected from the six oysters; cut into small pieces; mixed carefully to lower individual variation; and frozen immediately in 1.5-g separate portions. All recovered tissue weights were recorded.
Samples were extracted based on tissue series (i.e., all gills together) on the same day for all experiments. Mengovirus (106 50% tissue culture infectious doses [TCID50]) was added to each dissected tissue (1.5 g) before homogenization. The laboratory procedure included tissue homogenization, extraction by vortexing with an equal volume of chloroform-butanol for 30 s and Cat-Floc T (173 μl per tube) (Calgon, Ellwood City, PA) treatment for 5 min on the bench before centrifugation for 15 min at 13,500 × g. The resulting suspension was precipitated with polyethylene glycol 6000 (PEG 6000) (Sigma, St. Quentin, France) for 1 h at 4°C and centrifuged for 20 min at 11,000 × g at 4°C (1).
The Nuclisens extraction kit (bioMérieux, Lyon, France) was used following the manufacturer's instructions with minor modifications (26). The PEG pellet was suspended in 1 ml of RNase-free H2O, mixed with the lysis buffer (2 ml), and incubated for 30 min at 56°C. After a brief centrifugation to eliminate particles (if needed), 50 μl of paramagnetic silica was added, and the mixture was incubated for 10 min at room temperature. All washes were performed with the magnetic ramp, and nucleic acids were recovered in 100 μl of elution buffer (bioMérieux, Lyon, France). All extractions were conducted in a short time to avoid freezing before rRT-PCR. The remaining nucleic acids were kept frozen (−80°C).
For NoV, rRT-PCRs were conducted with previously described primers and probes targeting the beginning of ORF2, using different sets for GI and GII in separate reactions (26). For mengovirus, primers and probe were previously described (42). The rRT-PCR was carried out with the Ultrasens quantitative RT-PCR (qRT-PCR) kit (Invitrogen, France), using adjusted concentrations of primers and TaqMan probes (26).
Five microliters of nucleic acid extracts or controls was added per well, for a final total well volume of 25 μl. All samples were analyzed in duplicate undiluted and after 10-fold dilution.
The cycle threshold (CT) was defined as the cycle at which a significant increase in fluorescence occurred (i.e., when fluorescence became distinguishable from background). To be included in the quantitative analysis, all wells had to yield a CT value of ≤41, which was considered the quantification threshold (QT).
To estimate loss of viral nucleic acids during extraction, a defined amount of mengovirus was spiked into each sample, and the recovery was determined to allow calculation of an extraction efficiency. After extraction of samples seeded with the mengovirus, undiluted and 10-fold-diluted extracts were subjected to rRT-PCR for mengovirus. The CT value of the sample was compared to the CT value of the positive control used in the extraction series and to a standard curve made by endpoint dilution. This difference (ΔCT) was used to determine the extraction efficiency, using the equation 100e−0.6978ΔCT and expressed as a percentage for each tissue (26).
The absence of inhibitors was verified for each sample by comparing undiluted and 10-fold-diluted extracts' CT values. The mean CT value was calculated for each sample, and as variations of <1 CT were observed, standard variation was not considered. No adjustment was made for rRT-PCR efficiency as no significant inhibition was observed. The number of RNA copies present in each positive sample was estimated by comparing the CT value to GI or GII standard curves derived from in vitro transcription plasmids containing nucleotides 146 to 6935 of the Norwalk virus (GenBank accession no. M87661) or nucleotides 4191 to 5863 of the Houston virus (GenBank accession no. EU310927). The final concentration was then adjusted based on the volume of nucleic acids analyzed and extraction efficiency and was reported per gram of tissue (26). Thus, virus concentration (copies per gram) was determined as follows: concentration = (no. of copies/5 μl analyzed) × (100 μl extract/1.5 g tissue) × (1/extraction efficiency).
Oysters from the same batch (before bioaccumulation experiments) were prepared as previously described (30). Briefly, digestive tissues, gills, and mantle were dissected, homogenized in phosphate-buffered saline (PBS [pH 7.4]) in a one-third dilution, heated for 10 min at 95°C, and centrifuged, and the supernatant was recovered. After measurement of the protein concentration with a BC assay kit (Uptima, San Diego, CA), Nunc Maxisorp immunoplates (ThermoFischer Scientific, Roskilde, Denmark) were coated in duplicate with tissue extracts at 40 μg/ml in carbonate buffer (pH 9.6) and were blocked with 10% nonfat dried cow's milk in PBS for 1 h. VLPs (1 μg/ml) were added and incubated for 1 h at 37°C. The respective rabbit anti-VLP antibodies were added, the mixture was incubated for 1 h at 37°C, and anti-rabbit IgG conjugated to horseradish peroxidase (Uptima) was added. Between each step, plates were washed three times with PBS–5% Tween 20 (Sigma-Aldrich, France). The enzyme signals were detected with 3,3′,5,5′-tetramethylbenzidine (TMB; BD Bioscience, San Jose, CA) and read at 450 nm with a spectrophotometer (Safire; Tecan). For each tissue, negative controls (without VLPs or antibodies) and positive controls (human secretor type A, B, and O saliva samples) were included. After validation of the positive and negative controls, optical density (OD) values obtained for each sample were read, and a test ratio sample was determined (OD values of the test sample divided by the OD values of the negative control). A sample was considered positive if the ratio was ≥2.
Nunc Maxisorp Immunoplates were coated with oligosaccharides conjugated to either polyacrylamide (PAA) or human serum albumin (HSA) as previously described (32). After blocking with 5% defatted dry cow's milk, VLPs at 4 μg/ml (2 × 1010 particles per well) were added for 2 h at 37°C. VLP binding was detected by incubation with the respective anti-GI.1 or anti-GII.4 rabbit antisera diluted at 1/1,000 followed by incubation with peroxidase-conjugated goat anti-rabbit immunoglobulins (Uptima, Montluçon, France). The peroxidase substrate TMB (BD Bioscience, San Jose, CA) was used, and OD values at 450 nm (OD450) were determined. OD values twice above background were considered positive.
All concentrations obtained were log transformed, and geometric mean titers (GMT) and standard deviations (SDs) were calculated. Data were analyzed with StatGraphic software (Sigma Plus, Levallois-Perret, France). Analysis of variance (ANOVA) tests were used to compare NoV concentrations between samples, and significance was declared at P ≤ 0.05.
The stability of the three viral strains was verified on three separate experiments conducted at the seawater temperature used for bioaccumulation. For each strain, the mean CT values calculated based on all CT values obtained (at last 12 values for each) and standard deviations showed no difference during the time period considered (Table 1). The stability of the VLPs' reactivity with rabbit hyperimmune sera corresponding to the GI.1 and GII.4 strains was verified over 1 day in seawater. For GI.1 VLPs, the OD values after 1 h and 24 h were 111% and 85%, respectively, of the OD value at baseline (for an OD value of 1.3 at time zero). Similarly, for GII.4 VLPs, the OD values were 115% and 83% after 1 h and 24 h, respectively, compared to baseline (for an OD value of 0.9 at time zero). The lack of change in the CT values and the small decrease of immune reactivity suggest that the viral particles, used for the bioaccumulation, are equally stable in seawater for at least 24 h.
Attachment to five neoglycoconjugates of VLPs from the GI.1, GII.3, and GII.4 strains used for the bioaccumulation experiments were compared. The GI.1 VLPs attached preferentially to H type 1, as previously reported (16, 32), whereas the GII.3 and GII.4 attached preferentially to sialyl-Lex and Ley and, to a smaller extent, to the sialylated type 2 precursor, similar to previous observations for the GII.3 strain and for another GII.4 strain (44) (Fig. 1). Thus, the binding patterns were quite different between the GI.1 and the GII strains, the latter being characterized by an ability to recognize sialylated structures.
Fresh oysters collected from a clean environment were bioaccumulated at a water salinity of 30.5 ± 1.2 g/liter. Three concentrations were assayed for each strain, the highest one being 8.52 ± 0.2, followed by 8.38 ± 0.3 and 8.48 ± 0.33 expressed as log10 RNA copies/liter for the GI.1, GII.3, and GII.4 strains, respectively. The middle concentration corresponded to a 10-fold dilution and the lowest concentration to a 100-fold dilution. Thus, approximately 108, 107, and 106 RNA copies/liter were used for the high, middle, and low doses and were introduced into the respective aquariums for the three strains.
The GI.1 virus was bioaccumulated efficiently at all three doses (Fig. 1). A dose-dependent response was observed for the three concentrations tested (106, 107, and 108 copies) with final amounts observed in DT of 4.5 ± 0.9, 5.5 ± 0.8, and 6.4 ± 0.7 log10 RNA copies/g, respectively (P = 0.0001, ANOVA). Concentrations detected in DT were quite similar during the two first experiments conducted in October and November, whereas increased concentrations in DT were observed in January and, to a lesser extent, in March (Fig. 2 A). To assess this difference, relative amounts of virus recovered in DT after 1 or 24 h were calculated. After 1 h, less than 1% of virus seeded into seawater was detected for three experiments (October, November, and March), but 41% was already concentrated in the DT in January (Fig. 2B). After 24 h, 5.5 and 1.2% of the seeded-virus amounts were detected in the DT in October and November, respectively, and 88 and 27% were detected in the experiments conducted in January and March, respectively. The difference between October/November and January/March was statistically significant (P = 0.0004). Concentrations in gills and mantle were quite stable and at least 100-fold lower than the concentration in DT. For example, for the lowest dose, 1.9 ± 0.8 and 2.0 ± 1 log10 RNA copies/g of gills or mantle, respectively, were detected after 1 h and 2.0 ± 0.6 and 2.0 ± 0.7 log10 RNA copies/g of gills and mantle, respectively, were detected after 24 h. For the highest concentration, the difference after 24 h of bioaccumulation was even larger with 3.2 ± 1 log10 RNA copies/g of gills, compared to 6.4 ± 0.7 log10 RNA copies/g detected in DT.
The GI.1 VLPs' capacities to bind to the different tissues were then compared for the 4 months considered (October, November, January, and March) (Fig. 2C). These VLPs bound readily to DT, with a clear increase in January and March. No binding above the sensitivity threshold was observed for the two other tissues. The distributions of bioaccumulated virus within the different tissues analyzed for the three strains were compared after 1 h and 24 h. Concentrations measured in all three tissues after 1 h and in gills and mantle after 24 h were expressed as a percentage of concentrations measured in DT at 24 h (Fig. 2D). The virus was directly accumulated in DT (13%), with only 0.5 and 0.3% of viruses detected after 1 h in gills and mantle, respectively. After 24 h, the relative concentrations detected in gills and mantle remained very low (0.15 and 0.12%, respectively), showing that the virus specifically accumulated in the DT.
The GII.3 strain was also efficiently bioaccumulated by oysters (Fig. 3). The concentrations detected after 24 h in DT were quite similar to those observed for GI.1 (except in January) (Fig. 3A). Unlike with the GI.1 strain, no difference between the 4 months was observed in terms of bioaccumulation efficiency (P = 0.18). After 1 h, only 0.1 to 0.5% of the inoculum was detectable in DT (Fig. 3B). After 24 h, concentrations in DT increased for all three inoculum dosages but never reached more than 4% (0.1 to 4.1%) of the virus dose seeded into seawater (Fig. 3B). However, as for the GI.1 strain, a clear dose impact was observed in recovered DT concentrations (4.2 ± 0.4, 5.1 ± 0.4, and 6.4 ± 0.2 log10 RNA copies/g for the three seeded doses assayed, respectively) (P = 0.0002).
GII.3 VLPs bound to all three tissues assayed without clear variations between the 4 months (Fig. 3C). This is reflected by the observed tissue distributions (Fig. 3D). If up to 41% of the virus was already detected in DT after 1 h, nonnegligible proportions were also present in gills (11%) and in the mantle (5%) (Fig. 3D). However, after 24 h, the concentrations in these two tissues represented only approximately 0.5% for each of the concentrations detected in DT. Thus, this strain accumulated in all organs in a transient manner before being concentrated in DT. After 1 h, the concentrations observed in gills and mantle were approximately 10 times lower than those in DT. In the former organs, relative concentrations then decreased for the next 23 h to reach a difference in concentration compared to DT between 100- and 1,000-fold lower. For example, for the lowest dose, 2.3 ± 1.3 and 2.1 ± 1.6 log10 RNA copies/g were detected in gills and mantle, respectively, after 1 h, and 1.9 ± 0.7 and 2.1 ± 0.7 log10 RNA copies/g of gills and mantle, respectively, were detected after 24 h. As observed for the GI.1 strain, the difference between virus concentration in gills and DT reached ~3 log10 after 24 h for the highest dose, with detection of 3.5 ± 0.5 log10 RNA copies/g in gills compared to 6.4 ± 0.2 log10 RNA copies/g in DT.
Surprisingly, the GII.4 strain showed very poor bioaccumulation, with less than 0.01% of the inocula being concentrated by the oysters (Fig. 4). Even at the higher dose (108 viral particles diluted into seawater), the concentrations detected in DT were low, with no difference between the four experiments. At variance with the GI.1 strain, the lower efficiency was observed in January. (The absence of inhibitors was verified, the extraction efficiency control was identical to the one observed for the GI.1 or GII.3 experiments, and rRT-PCR was repeated several times.) The mean concentrations at 24 h for the four experiments were 2.2 ± 0.3, 2.7 ± 0.6, and 3.5 ± 0.5 log10 RNA copies/g of DT for the three concentrations assayed, respectively, showing the absence of a clear relationship with the seeded doses (P = 0.57). These concentrations were quite similar to those detected after 1 h (1.7 ± 0.3, 2.0 ± 0.1, and 2.3 ± 0.2 log10 RNA copies/g of DT), showing almost no increase over time, unlike for the two other strains assayed. After 1 h, concentrations in gills and mantle were similar to those in DT, unlike for the two other strains. For example, for the lowest dose, 2.3 ± 1.1 and 2.4 ± 0.2 log10 RNA copies were detected per g of gills and mantle, respectively, after 1 h. After 24 h, the concentrations detected were 1.7 and 2.0 ± 0.4 log10 RNA copies/g in gills and mantle, respectively. Finally, similar to DT, recovered concentrations in gills or mantle showed no clear dose relationship.
As observed for the GII.3 VLPs, the GII.4 VLPs attached to all three tissues, except that in October lower binding to the gills was observed. However, considering the 4 experiments conducted, this showed a clearly different pattern, with about 45% of relative viral concentration detected in gills after 1 h compared to 5.5% detected in DT and 9% in mantle. After 24 h, gills and mantle tissues still represented 31 and 40%, respectively, of the virus accumulated in DT, indicating that the accumulation of this strain was not organ specific in addition to being very low, as described above.
Shellfish contamination by infectious agents is classically monitored based on detection of Escherichia coli in shellfish tissues (European regulation 91/492/EC) or fecal coliforms in growing waters (United States National Shellfish Sanitation Program). However, shellfish meeting regulation criteria have been implicated in outbreaks, and depuration, efficient at eliminating some bacteria, does not eliminate viruses (7, 29, 43). The observed differences in clearance of bioaccumulated bacteria and viruses raise questions about potential interactions between oysters and viral human pathogens. This is of particular interest for NoVs, the pathogens most frequently involved in shellfish-borne outbreaks (4, 7, 40). The finding that NoV-specific ligands exist in shellfish led to the hypothesis that expression of these ligands may influence bioaccumulation and behavior of these viruses in oysters (23, 30, 51, 56). The ligand for Norwalk virus, the prototype GI strain, was characterized as an A-like carbohydrate structure indistinguishable from human blood group A antigen and whose expression is restricted to the digestive tissues of these animals and shows a clear seasonal variation (23, 30, 50). For the Houston GII.4 strain, we recently demonstrated that the interaction in digestive tissues involved both a sialic acid in α2,3 linkage and an A-like carbohydrate ligand and that the virus binds to gills and mantle tissue involving the sialic acid-containing ligand exclusively (30). To evaluate the impact of these ligands on NoV bioaccumulation in oysters, three representative strains of NoV GI and GII were compared in terms of efficiency of bioaccumulation, tissue distribution, and seasonal influence. Selection of the tissues analyzed was based not only on the VLPs' binding ability but also on oyster physiology. For feeding activity, oysters pump water over their gills. Suspended particles are captured and passed on to the alimentary tract, with some sorting of particles occurring prior to ingestion to help regulate what is presented to the digestive tract. The organs involved in the ingestion and digestion of food and the elimination of feces include the mouth, a short esophagus, stomach, a crystalline-style sac, digestive diverticula, midgut, rectum, and anus. (All of these tissues were dissected and called “digestive tissues” [DT] in this study.) With the exception of a short section of the rectum, the entire alimentary canal lies within the visceral mass and is completely immobilized by the surrounding connective tissue (included here in mantle tissues). Food is moved from the mouth toward the anus by the strong ciliary activity from epithelial cells that line the alimentary tract. The digestive gland surrounds the stomach entirely and also part of the intestine. It comprises a series of branched ducts that open into the stomach, and the duct branches serially to terminate in blind-ending tubules, the location of the digestion activity (14). Based on these physiological parameters, we chose to apply a quantitative approach to three groups of tissues: i.e., gills, digestive tissues, and mantle.
Environmental conditions have an impact on oyster growth, respiration, and nutrient assimilation (8, 31). As these aspects were not considered here, to avoid as much as possible variability due to environmental conditions and to follow the seasonal cycle of oysters, all experiments were performed with the same batch of oysters kept in a clean area during the 6 months of the study. The seawater was collected from the same area, and the aquarium temperature was adapted to the in situ measured temperature. The bioaccumulation experiments (the three strains at three concentrations and the control batch) were conducted on the same day for each season, and oysters were dissected at the same time, by different members of the laboratory, to avoid any delay and differences within assays. Similarly, all tissues were then extracted by organs rather than by strains or level of contamination, and rRT-PCRs were performed in a single experiment, including all negative controls and the standard curve. These precautions were important to avoid artifactual experimental variations and to allow safe comparisons of the four experiments.
The first striking observation was that oysters concentrated the three strains with very different efficiencies and tissue distributions. The GI.1 strain was previously shown to bind specifically through an A-like carbohydrate structure to DT but not to other tissues (23, 50). We observed here that it was readily bioaccumulated in DT, with less than 1% of the virus detected in other tissues after 1 h and a 1,000-fold difference between gills/mantle and DT after 24 h, consistent with the lack of a ligand in gills and mantle. The high concentration of GI.1 recovered in DT is also consistent with earlier observations (1, 45). The efficiency of this DT-specific bioaccumulation paralleled the season-dependent expression level of the carbohydrate ligand, strongly arguing in favor of its involvement in the bioaccumulation process. Moreover, we previously observed that GI.1 VLPs bioaccumulated in a manner dependent on this carbohydrate recognition since mutant VLPs that had lost the carbohydrate ligand-binding property were less well accumulated (30). We also previously observed that in the environment, the ratio between genome copies in oysters and in water was much higher (50 times) for GI strains than for GIII strains that have no carbohydrate ligand in oyster tissues (56). Collectively, the previously reported observations and those presented here represent compelling evidence for a major role of the TD-specific carbohydrate ligand in the highly efficient GI.1 strain bioaccumulation in oysters. In contrast, the GII.4 strain was poorly bioaccumulated, regardless of the month considered, and showed a different distribution within the shellfish body. Even though a preferential accumulation in DT occurred after 24 h, after 1 h, a large percentage of virus was detected in gills and mantle, consistent with the binding to sialic acid in the α2,3 linkage previously detected by ELISA and histochemistry (30). Bioaccumulation of GII.4 strains in gills or labial palps of Pacific oysters (Crassostrea gigas) (36) or in gills but not in DT of Crassostrea ariakensis (53) was reported, suggesting that the behaviors of various NoV GII.4 strains may be similar in distinct oyster species, even if the results obtained here are to be considered for the GII.4 Houston strain. Over the last 15 years, strains of the GII.4 genotype became predominant across human populations (up to 80 to 90% of clinical cases) and have been responsible for several large outbreaks (3, 6, 47). Despite this high prevalence in human infections and thus large amounts of GII.4 particles discharged in sewage (46), strains of this cluster are not predominant in oyster-related outbreaks (11, 19, 22, 24, 39). Different factors such as viral load in human feces, resistance to sewage treatment, and adsorption to different particles may influence the behavior of viral particles (12, 52). However, our observation of very poor accumulation by oysters compared to other NoV strains is concordant with this epidemiological observation.
What is the reason for the weak DT-specific bioaccumulation of GII.4? Assays of water in the bioaccumulation tanks at the end of the last two experiments showed negligible numbers of virus genome copies leftover for all three strains compared to the inocula (data not shown), suggesting that the three strains have almost entirely been captured by the oysters. Although the low GII.4 bioaccumulation might be explained by a lower stability compared to that of other strains, as previously suggested (9), control experiments showed no decrease in genome copy numbers or capsid immune reactivities in seawater during the short period of time considered (24 h). However, it should be noted that detection of RNA by RT-PCR may not correlate with infectivity. Furthermore, the very low bioaccumulation of GII.4 compared to GI.1 was already clearly visible after 1 h. It is therefore unlikely that the inefficient bioaccumulation of the GII.4 strain can be explained by a lower stability in seawater in such short periods. The binding of the GII.4 strain to gills (and the mantle) may prevent the passage of viral particles to the mouth and thus to DT. Even after 1 day and irrespective of the season or concentrations, GII.4 virus persists on the gills (and the mantle) in accordance with the binding with VLPs detected by ELISA (30). Because we were unable to detect residual virus in the seawater after a 24-h bioaccumulation period (as observed also for the two other strains tested), we hypothesize that binding to gills and mantle through a sialic acid-containing ligand prevents the passage into the digestive tract and is followed by rapid destruction of the virus by unknown mechanisms that need to be further analyzed and identified.
GII.3 NoVs have an ELISA binding pattern to oyster tissues similar to that observed for the GII.4 strain, with VLPs binding to DT, gills, and mantle. However, the bioaccumulation efficiency was much higher than that observed for the GII.4 strain. After 1 h, as observed for the GII.4 strain, NoV GII.3 was detected in gills and mantle but also in DT. After 24 h, gills and mantle tissues displayed concentrations 1,000-fold lower than in DT, suggesting that after being transiently retained in the gills, probably due to binding to sialic acid, they are either destroyed, as observed for the GII.4 strain, or they are released to enter the mouth, as observed for the GI.1 strain. The release from the gills or mantle might occur if the GII.3 strain has a lower binding affinity for the sialic acid-containing ligand, an aspect that will require further investigation.
GI NoVs represent about 30% of NoV-contaminated shellfish (field or market studies) (5, 12, 54), and this genogroup is also frequently detected in shellfish outbreaks (18, 19, 22, 24, 39, 55). Similarly, NoV GII.3 strains are frequently reported in shellfish-related outbreaks (11, 19, 38). These data fit with our observation that even after 1 h, mimicking an accidental contamination, there is already considerable uptake of GI.1 and GII.3 viruses in oysters. The efficiency of oysters at bioaccumulating GI strains such as Norwalk virus, particularly during the winter months (January through March) when oyster consumption is highest in France, may explain why this genogroup of NoVs are so often implicated in shellfish-related outbreaks despite their relatively low frequency of detection in the community. Virus contamination of oysters can occur in the absence of a specific ligand, as observed through a field study with GIII NoV strains (56). However, it is far less efficient than when a digestive-tissue-specific ligand is present, as in the case of the GI.1 strain.
From a public health perspective, identification of a correlation between ligands in shellfish and bioaccumulation efficiency may help to predict periods of high risk, guide the development of testing protocols that will help to increase the sanitary quality of shellfish put on the market, or even lead to the selection of oyster species that may be less sensitive to NoV contamination.
This work was supported in part by grant 2006 SEST 08 01 (Coquenpath) from the Agence Nationale pour la Recherche (ANR, no. 538), by a grant (CIMATH) from the Région des Pays de la Loire, and by DGAL (Direction Générale de l'Alimentation). H.M. was supported by a fellowship from IFREMER and Conseil Régional des Pays de la Loire.
We thank J.-C. Le Saux (Laboratoire de Microbiologie-LNR, IFREMER) for supplying oysters and environmental data and C. Jonsson and L. Svensson (University of Linkoping, Sweden) for GII.3 VLPs and antibody.
Published ahead of print on 25 March 2011.