ABSTRACT
Noroviruses (NoV) are responsible for many shellfish outbreaks. Purification processes may be applied to oysters before marketing to decrease potential fecal pollution. This step is rapidly highly effective in reducing Escherichia coli; nevertheless, the elimination of virus genomes has been described to be much slower. It is therefore important to identify (i) the purification conditions that optimize virus removal and (ii) the mechanism involved. To this end, the effects of oyster stress, nutrients, and the presence of a potential competitor to NoV adhesion during purification were investigated using naturally contaminated oysters. Concentrations of NoV (genomes) and of the viral indicator F-specific RNA bacteriophage (FRNAPH; genomes and infectious particles) were regularly monitored. No significant differences were observed under the test conditions. The decrease kinetics of both virus genomes were similar, again showing the potential of FRNAPH as an indicator of NoV behavior during purification. The T90 (time to reduce 90% of the initial titer) values were 47.8 days for the genogroup I NoV genome, 26.7 days for the genogroup II NoV genome, and 43.9 days for the FRNAPH-II genome. Conversely, monitoring of the viral genomes could not be used to determine the behavior of infectious viruses because the T90 values were more than two times lower for infectious FRNAPH (20.6 days) compared to their genomes (43.9 days). Finally, this study highlighted that viruses are primarily inactivated in oysters rather than released in the water during purification processes.
IMPORTANCE This study provides new data about the behavior of viruses in oysters under purification processes and about their elimination mechanism. First, a high correlation has been observed between F-specific RNA bacteriophages of subgroup II (FRNAPH-II) and norovirus (NoV) in oysters impacted by fecal contamination when both are detected using molecular approaches. Second, when using reverse transcription-quantitative PCR and culture to detect FRNAPH-II genomes and infectious FRNAPH in oysters, respectively, it appears that genome detection provides limited information about the presence of infectious particles. The comparison of both genomes and infectious particles highlights that the main mechanism of virus elimination in oysters is inactivation. Finally, this study shows that none of the conditions tested modify virus removal.
INTRODUCTION
Shellfish are recognized to be vulnerable toward human-pathogenic viruses such as human norovirus (NoV) (1, 2). Among them, oysters are particularly involved in virus transmission because of (i) their capacity to concentrate microorganisms while filtering large volumes of seawater (3); (ii) the presence of A-like histo-blood group antigens (A-like HBGAs), specific ligands that may favor NoV retention in digestive tissues of oysters (4); and (iii) the fact that oysters are generally consumed raw.
According to EU regulations (5), the microbiological quality of shellfish is governed by the concentration of Escherichia coli. Based on the concentration of E. coli detected in 100 g of flesh and intervalvular fluid (FIL), purification processes (relaying [i.e., shellfish storage in unpolluted water] or depuration [i.e., shellfish storage in treated seawater]) are implemented to achieve acceptable levels of E. coli in mollusks before putting them on the market (6). However, NoV outbreaks associated with the consumption of oysters compliant with EU regulations have been reported (7, 8). To date, it is well acknowledged that this bacterial criterion has numerous limitations (9) and does not reflect the behavior of pathogenic viruses in shellfish (10–12). Thus, identification of a reliable viral indicator is required for better assessment of the NoV hazard in shellfish, especially during purification processes. Despite recent advances in cellular models allowing the in vitro replication of NoV using human intestinal enteroids (13), this approach would not be suitable for routine food testing owing to its complexity, high cost, and time-consuming nature. To date, the ISO 15216-1 standard, which allows the quantification of NoV genomes by reverse transcription-quantitative PCR (RT-qPCR) (14), is therefore the only approach to specifically detect NoV in vulnerable foodstuffs. However, this method is unable to discriminate infectious from noninfectious NoV when its genomes are detected. As has already been demonstrated on NoV surrogates, viral genomes have better persistence than infectious viruses in the environment (15, 16). As a result, surveillance studies using the ISO 15216-1 standard may overestimate NoV infectious particles. This observation is probably applicable to NoV, but nothing has been formally demonstrated due to the lack of culture methods. This is why no regulations exist in Europe for the assessment of NoV risk in foodstuffs.
Alternative approaches to NoV genome detection have been proposed, including the use of human fecal contamination indicators, especially bacteriophages infecting enteric bacteria (17). Among them, F-specific RNA bacteriophages (FRNAPH) have been extensively studied because of their structural similarity to waterborne viruses (18). Both genomes and infectious particles of FRNAPH can be easily detected by RT-qPCR (19, 20) and culture (21, 22), respectively. Four genogroups of FRNAPH have been described and commonly circulate in the environment (23). Among them, genogroups II (FRNAPH-II) and III (FRNAPH-III) are mainly associated with human fecal pollution (24), while genogroups I (FRNAPH-I) and IV (FRNAPH-IV) are mainly related to animal fecal pollution. A positive correlation has recently been demonstrated by Hartard et al. between the genomes of FRNAPH-II and NoV in oysters during purification processes (22), showing that FRNAPH-II could be proposed as a potential indicator to evaluate NoV risk in shellfish (25). On the other hand, it has been demonstrated that FRNAPH are excreted by only a part of the population (26%) (26). Furthermore, Furuse et al. showed in 1983 that FRNAPH are more frequently shed by hospitalized patients (14%) than by healthy individuals (1.6%) (27). Therefore, FRNAPH may have some limits in describing fecal pollution coming from small populations (i.e., boat discharge or a small septic system). However, it is possible to consider the presence of FRNAPH as the proof of a fecal pollution especially coming from urban areas. Their behavior can easily be followed during a purification process.
McMenemy et al. modeled the depuration efficacy of different systems for the removal of E. coli, NoV, and FRNAPH (28). Several physical-chemistry criteria may impact the effectiveness of depuration processes in reducing the microbiological load of oysters (i.e., water temperature, salinity, oxygenation, and flow rates) (28). For example, it has already been demonstrated that water temperature plays a major role in the persistence of RNA and DNA F-specific bacteriophages in oysters during depuration (29, 30). In addition, biological criteria may also have high impact (the presence of nutrients, the expression of HBGAs as in oysters, and biological activity).
Studies focusing on virus removal during oyster depuration have reported variable reduction in NoV genomes for conventional systems (31). In their review, McLeod et al. suggest that depuration would be less effective than relaying because more outbreaks have been described. These researchers noted that relaying was able to decrease NoV genome significantly (by nearly 1 log10) after at least 17 days and that there have been no illnesses reported with 28-day relayed products (31). However, relaying is a less common practice than depuration, which may have potentially contributed to the relative absence of relay-associated outbreaks. Conversely, another study found no NoV genome reduction differences between relaying and depuration after 30 days; indeed, a similar and slight decrease in NoV genome levels was observed in both cases (25). However, because of the limited number of studies on this topic, there are discrepancies in the effectiveness of purification in NoV genome removal.
This study first aimed to evaluate the effects of different conditions on the elimination of viruses in oysters during purification processes (i.e., relaying and depuration) and then to define the main mechanisms involved. We tested the effects of purification conditions on NoV and FRNAPH elimination as a function of (i) accessibility to nutrients; (ii) the presence of citric acid, a potential competitor of NoV-specific ligands (A-like HBGAs) in oysters; and (iii) stress conditions taking oysters out of water for a period of time. Concerning citric acid, it was demonstrated that it mimics the pyranoside ring of fucose and therefore inhibits the recognition of genogroup II (GII) NoV to HBGAs (32). Concerning the stress condition, a period outside water may be the reason why the virus content of shellfish was modified during the first day of purification. For example, we previously (32) observed a breaking point in the elimination kinetics of NoV and FRNAPH-II genomes from oysters after the first day of depuration, which was also observed with murine norovirus in mussels (33). The behaviors of both viruses were compared by following their genomes, whereas infectivity was only followed for FRNAPH. Comparison of FRNAPH genomes and infectious particles was done to help determine whether viruses were inactivated in oysters or released in water. To be as close as possible to environmental conditions, the study was performed using a set of oysters coming from a commercial harvesting area naturally affected by fecal contamination.
RESULTS
Microbiological quality of the initial set of oysters.The initial samples were found to be significantly affected by fecal pollution with an initial concentration of E. coli of approximatively 1,200 most probable number (MPN)/100 g of FIL. On day 1 after depuration and relaying, the E. coli concentration fell below the limit of detection (LOD; <67 MPN/100 g of FIL). The initial genome concentrations of genogroup I (GI) NoV, GII NoV, and FRNAPH-II were 293, 802, and 463 genome copies (gc)/g of digestive tissues (DT), respectively. Infectious FRNAPH were detected by two methods: (i) total infectious particle enumeration in digestive tissues (about 140 PFU/g of DT at day 0) and (ii) a culture-based approach on the whole body (qualitative integrated cell culture [ICC]-RT-qPCR), where FRNAPH-I and FRNAPH-II were detected at quantification cycles (Cq) of 10.9 and 30.5 at day 0, respectively. The Cq of FRNAPH-II was higher than that of FRNAPH-I because of the better growth rate of the latter. Therefore, for the same initial concentration, the Cq mean values of FRNAPH-I were always lower than those of FRNAPH-II (e.g., for an initial inoculation of 50 PFU/analyzed fraction, FRNAPH-ICq = 18.8 and FRNAPH-IICq = 24.8 [22]). No FRNAPH-III was detected using this method.
Influence of purification and stress conditions during the purification process of oysters. (i) Purification conditions.Shellfish batches were subjected to different conditions, four in UV-treated seawater tanks (i.e., depuration) and one in natural seawater (i.e., relaying) (Table 1). Oysters were stored for 43 days, except for the depuration condition with citrate buffer (batch B2), because the oyster mortality was observed after 22 days due to the acidic condition of the medium (i.e., pH 6.2). No oyster mortality was observed for the other test conditions. In order to identify a purification condition that influences viral load removal, we compared FRNAPH and NoV concentrations over time under each testing condition.
Experimental conditions of the study
During purification, FRNAPH-II, GI NoV, and GII NoV were all monitored by targeting their genomes, while infectivity was also quantified by a plaque assay for FRNAPH. Values for recovery rates for genome extraction (i.e., difference between enteric cytopathic bovine orphan [ECBO] initial genome quantities and ECBO extracted genome quantities; average, 19.4% ± 12.4%) and RT-qPCR inhibition (11% ± 13.2%) were minor and agreed with the performance criteria defined by the NF EN ISO 15216 standard (i.e., >1 and < 75%, respectively). Infectious FRNAPH were detected throughout the whole purification period (i.e., 43 days), except for B2 (i.e., 22 days) because of oyster mortality. Less than 10 PFU/g of DT was detected after 22 days of purification under all purification conditions. All of the GI NoV, GII NoV, and FRNAPH-II genomes were detected throughout the purification period, except under the Bat and D conditions for NoV GI and under the B3 condition for NoV GII, whose genomes were not detected after days 29 and 36 and after day 36, respectively (Fig. 1).
Example of viral target behaviors during MBBATD purification processes. (A) GII NoV genome. (B) Infectious FRNAPH. (C) GI NoV genome. Black circle/black line, B1; gray circle/gray line, B2; white circle/dotted line, B3; black cross/black dashed line, D; gray cross/gray dash line, Bat. The initial batch (day 0) is indicated by a black square.
A Wilcoxon signed-rank test was used to compare the distributions of viral targets corresponding to the different conditions. The results of these pairwise comparisons are shown in Table 2. Among the 40 tests thus performed, only three showed P values of <0.05: B2/B3 for infectious FRNAPH (P = 0.0143), B1/Bat for GI NoV (P = 0.0423), and B1/B3 for FRNAPH-II (P = 0.0323). Five tests yielded a P value of <0.10. These last two proportions are in full accordance with the allocation which is expected under the null hypothesis (i.e., the storage conditions are equivalent). Because no statistical differences were highlighted, we suggest that the storage conditions do not seem to have influenced the purification of the tested viral targets (FRNAPH and NoV), and no difference was highlighted between depuration and relaying. This result led us to consider the use of the viral target’s concentration values obtained under each condition as repetitions, making up a unique variable called “MBBATD,” for use in subsequent statistical analyses. MBBATD refers to the grouped means of B1, B2, B3, Bat, and D for each viral target under depuration and relaying time point conditions.
Pairwise comparisons of viral target abundance corresponding to different storage conditionsa
(ii) Stress conditions.In order to test the influence of stress conditions on virus removal, infectious FRNAPH, as well as GI NoV, GII NoV, and FRNAPH-II, genomes were monitored as previously described for the MBBATD group. A Wilcoxon signed-rank test was also used for distribution comparisons (Table 2). In the 12 tests performed, no significant differences between conditions were shown for each viral target. As previously mentioned, this result led us to consider the values obtained as repetitions, making up a unique variable called “MSB.” MSB refers to the grouped means of SB1, SB2, and SB3 for each viral target under depuration time point conditions.
More generally, the Wilcoxon signed-rank test was used to compare the storage (i.e., the “MBBATD” group) with the stress (i.e., the “MSB” group) conditions for each viral target. None of these tests showed any significant differences (P = 0.3125, 0.8438, 0.2188, and 1 for the FRNAPH II genome, the GI NoV GI genome, the GII NoV genome, and infectious FRNAPH, respectively). This was followed by a unique test performed on the four viral targets taken as a whole (P = 0.2776). Thus, there seemed to be no difference in virus elimination between storage and stress conditions.
Similarly, the Friedman test was used to compare the results of the eight conditions. Blocks were formed on the virus-time combinations, provided that they showed complete data for all the conditions, to be usable. Thus, the number of blocks was limited to twelve (i.e., 4 viral targets × 3 time points). The null hypothesis (i.e., there are no differences between the eight conditions) could not be rejected (Friedman's test statistic [TFR] = 10.72; P = 0.1512). In other words, this result confirms once more, by following a different approach, that there is no significant differences between the eight conditions B1/B2/B3/BAT/D and SB1/SB2/SB3.
Viral target behaviors during purification processes.Decrease kinetics of the four monitored viral targets were compared in order to evaluate whether FRNAPH-II was a good indicator of NoV (GI and GII) and whether viruses were inactivated (i.e., genomes remained stable while infectious particles decreased) or released (i.e., genomes were released in the same way as infectious particles) during the purification processes of oysters. As a first step, the concentrations of viral targets over time were analyzed in graph form for each purification condition. Two representative examples are shown in Fig. 1.
A regression analysis using a log-linear model to express the viral concentration as a function of time was performed for each of the four viral targets. Because of the observed variability of these repetitions, which may be associated with experimental conditions, raw data were logarithmically transformed for statistical processing (log-linear model). SB1, SB2, and SB3 data were excluded here because they did not represent all of the sampled days. Moreover, analysis was performed only until day 22 because B2 data were missing beyond this purification time point (because of oyster mortality).
Examination of the experimental data indicated a temporary slight increase in viral target concentrations between day 0 and 1 under the various conditions. This phenomenon is observed for each viral target/condition variable, except for infectious FRNAPH under Bat and D and for the GII NoV genome under D. After this breaking point, concentrations decrease gradually over time following an exponential decay pattern. Analyses of the results were then performed only from days 1 to 22 in order to better respect the relevance of the log-linear survival model. The results of the regression analysis by fitting a log-linear model to the experimental data are presented in Table 3 for each of the four viral targets. From these results it can be pointed out that the slopes of all the fitted lines represented in Fig. 2 are significantly different from zero (t test; P < 0.01), which seems to establish the existence of a decreasing link between each of the four viral target concentration and the time (Table 3). Decay coefficients range from 0.8941 for infectious FRNAPH to 0.9530 for GI NoV. This means that the decay of infectious FRNAPH was more rapid than that of the other target genomes (NoV and FRNAPH-II). More specifically, T90 (time to reduce 90% of the initial titer) values are 20.57 days for infectious FRNAPH, 47.83 days for GI NoV genome, 26.67 days for GII NoV genome, and 43.90 days for FRNAPH-II genome. However, the rather low values of r2 for GI NoV and FRNAPH-II genomes (i.e., 0.3361 and 0.2397, respectively) indicate that this link is not very strong. This is related to the variability of the experimental results observed for the same time point. Indeed, the degrees of dispersion of the results, expressed by the coefficient of variation of the residual variability, are approximately 52.23, 52.24, 49.01, and 76.59% for infectious FRNAPH, GI NoV, GII Nov, and FRNAPH-II, respectively. These values point out the high levels of variability in the experimental results that were obtained for each of the considered viral targets.
Regression analysis for each viral target during the storage of shellfish after day 1a
Behavior of infectious FRNAPH and of FRNAPH-II, GI NoV, and GII NoV genomes during all oyster purification processes.
The Friedman test was also used in order to compare the viral target concentrations under storage conditions, with blocks formed on the time points. The outcome of the test shows that the slopes characterizing the kinetics of the four viral targets may not be considered equal (TFR = 16.20; P = 0.0010). The Wilcoxon signed-rank test was then used to compare the distributions of the microorganisms taken two by two. Simultaneously, the Student t test was used to perform a pairwise comparison of the corresponding slopes and thereby the removal of microorganisms over the time. The results of these tests are shown in Table 4. Statistically significant differences (P < 0.01) were observed for each comparison concerning infectious FRNAPH, except for the slope comparison against GII NoV (P = 0.1176). The distribution of infectious FRNAPH concentrations is always located below that of the genome concentrations, as can be seen in Fig. 2. On the other hand, the genome concentrations of FRNAPH-II, GI NoV, and GII NoV do not show any significant differences. Considering the slope comparison, infectious FRNAPH decay appears to be much more rapid than its genomes, as well as GI NoV. Furthermore, note that the slope corresponding to GI NoV genome is significantly (P = 0.0207) smoother than that of GII NoV genome.
Comparisons of the distributions and slopes between each pair of microorganisms
Finally, this study compared the decreases in FRNAPH genomes and total infectious FRNAPH between days 0 and 43 to identify the removal mechanism occurring in oysters during purification. For this, we excluded batch B2 to avoid missing data after day 22. Totals of 463 gc/g of DT for FRNAPH-II and 140 PFU/g of DT for total infectious FRNAPH were detected at day 0. At day 43, the FRNAPH-II genome concentration was approximately 196 gc/g of DT, whereas total infectious FRNAPH decreased to 1 PFU/g of DT. Since FRNAPH-II genomes decreased much slower than the corresponding infectious particles during purification, we can suggest that phage inactivation is thought to prevail over release in water.
Infectious FRNAPH genotyping: use of two complementary methods.Infectious FRNAPH were subjected to genotyping by two methods: a plaque assay followed by RT-qPCR of plaques and a sensitive qualitative approach of ICC-RT-qPCR (22).
When using ICC-RT-qPCR, FRNAPH-I and -II were detected up to the end of the study for each condition tested, whereas FRNAPH-III was not detected at all, even at day 0. It is well established that the FRNAPH-III subgroup is rapidly inactivated under natural conditions (34). The mean Cq values of RT-qPCR controls were approximately 10.87 ± 0.629 for FRNAPH-I, 9.93 ± 0.68 for FRNAPH-II, and 11.84 ± 1.30 for FRNAPH-III. FRNAPH-IV was used as an internal control (i.e., the culture-positive control of FRNAPH) for each ICC-RT-qPCR because this genogroup displays very low resistance/prevalence in the natural environment. The mean Cq values of the internal control were approximately 22.87 ± 3.04 with oyster samples and 11.87 ± 4.24 without oyster samples in the growth medium. Even if no inhibition control was used for this qualitative approach, the observed difference shows that oysters inhibited FRNAPH multiplication or contained RT-qPCR inhibitors. Sixty-four ICC-RT-qPCR assays were performed on the samples in this study: 93.7% of them detected FRNAPH-I with Cq values below 20 (++) against 55% for FRNAPH-II with Cq values below 32 (++) and 17.2% for FRNAPH-I with Cq values between 32 and 37 (+). Again, this difference could be explained by a higher growth rate for FRNAPH-I compared to FRNAPH-II, as previously reported (22). For example, for the same initial concentration of 0.5 PFU for both genogroups, Hartard et al. (25) demonstrated that 57% of ICC-RT-qPCR detected FRNAPH-I against only 14% for FRNAPH-II. Because of the high rate of bacteriophage multiplication induced by this method, we recommend performing the analysis on four samples at a time to limit contamination risk.
For both FRNAPH genogroups, Cq values were homogeneous over time. It is also worth noting that ICC-RT-qPCR signals could still be detected when no PFU were observed by plaque assay, which indicates greater sensitivity of the ICC-RT-qPCR method. However, both methods provided complementary results for the global detection of infectious FRNAPH. When the strains detected by plaque assay methods were subjected to genotyping assays, 41.8% of them (i.e., from days 0 to 43) were found to belong to FRNAPH-I, 36% to FRNAPH-II, and 22.2% to unidentified strains. Both genogroups were detected up to day 43 (60% of FRNAPH-I and 40% of FRNAPH-II). Neither of them showed a specific selection during depuration or relaying (Fig. 3).
FRNAPH-I and -II genome proportions corresponding to infectious FRNAPH detected by plaque assay over time. Dark gray, GI; light gray, GII.
Both of the two genogroups of FRNAPH detected by ICC-RT-qPCR were always detected above 100 PFU/g of DT of total infectious FRNAPH detected by plaque assay. Below this limit, only one genogroup was detected in 28.5 and 62.5% of ICC-RT-qPCR assays when titration was between 50 and 10 PFU/g and below 10 PFU/g, respectively. The two methods could therefore be viewed as complementary, variability being only detectable below 100 PFU/g of DT.
A signal was still observed for at least one genogroup, even when no PFU of total infectious FRNAPH was detected by the plaque assay method. No correlation was found between plaque assay and ICC-RT-qPCR methods for genogroup identification (e.g., on day 15, sample B1 showed 100% of infectious FRNAPH-I using genotyping of PFU, whereas only infectious FRNAPH-II was detected using ICC-RT-qPCR).
DISCUSSION
In this study, the decreased kinetics of GI and GII NoV (genomes) and FRNAPH (infectious particles and genomes) in oysters under different purification conditions were established. When focusing only on purification conditions applied to oysters coming from polluted areas, many studies have shown not only the incidence of physicochemical conditions (i.e., water temperature, salinity, and dissolved oxygen) on shellfish physiology but also the effect of the initial microorganism load (i.e., natural or artificial contamination) (10, 35–37). For these reasons, we decided to work with naturally NoV-contaminated oysters that were also positive for FRNAPH, because of the high correlation previously demonstrated (25). The initial batch had been collected in winter and the high E. coli count observed (1,200 MPN/100 g of FIL) indicate a fecal contamination. The lack of seasonal trend in E. coli accumulation in shellfish has been well described (18). Conversely, NoV genomes showed a greater accumulation in shellfish during winter. This seasonal effect can be partially explained by (i) the potential overexpression of A-like HBGAs during winter (38, 39), (ii) the seasonality of gastroenteritis epidemics leading to massive excretion of NoV in the environment (40, 41), and (iii) low temperatures and low UV radiation levels, providing greater stability to viral particles in the environment. Therefore, the simultaneous presence of GI and GII NoV is not surprising. FRNAPH, which are also known to exhibit increased accumulation in oysters during the winter period (3, 42), were detected in high concentrations for both infectious particles and genomes. The presence of FRNAPH-II genome was indicative of potential human pollution as previously shown (24) and therefore of a significant correlation with the presence of NoV genomes in oysters (8, 22). The presence of E. coli was suggestive of a recent fecal pollution event considering that they have a low persistence in oysters. On the whole, the initial oyster batch was identified as a potential high risk for NoV infection. In such a situation, depuration or relaying was clearly needed.
While depuration has been reported to be highly effective in removing E. coli generally after 48 h (10, 43, 44), several studies have suggested low NoV genome reduction rates (45, 46), providing that NoV prevalence and concentration in shellfish are similar whether the latter are subjected or not to purification processes (12, 47). Difficulties in removing NoV are usually explained by their specific interactions with A-like HBGAs especially in the digestive tissues of mollusks, but also by their sequestration in zones out of the digestive tract lumen, such as in hemocytes (i.e., phagocytic blood cells of oysters) (4, 48). Hemocytes contain acidic vesicles, which contribute to food digestion. Some authors underlined that NoV persistence in shellfish could be related to their resistance to acidic digestion (48). These authors’ assumption, further supported by their demonstration of the high acid resistance of NoV, could explain the long persistence of the virus in oysters. Indeed, a 1-log-unit reduction (T90) in NoV genome concentration in shellfish has been reported to take up to 28 days (10, 49). Our study confirmed these results as the GII NoV genome displayed a mean T90 of 26.7 days. GI NoV and FRNAPH-II genomes seemed to have longer persistence under our test conditions. This difference can easily be explained by the variability of the experimental results observed at a same time point. Nevertheless, it appears highly difficult to eliminate viral genomes (both NoV and FRNAPH) by depuration or relaying when initial pollution is detected. The overall conclusion to be drawn is that depuration of NoV and FRNAPH genomes may only be achieved over an extended depuration period, which is clearly much longer than those classically used by shellfish producers to mitigate the viral risk. Moreover, NoV removal is not effective enough because some NoV foodborne outbreaks related to the consumption of oysters have been reported (50, 51). Therefore, optimization of the purification conditions may be a key challenge to identifying effective parameters and ensuring the removal of infectious NoV.
In this study, three hypotheses were tested in order to identify a purification condition that could promote virus removal. Four conditions of depuration and one of relaying were investigated, but none of them yielded better virus removal. Nevertheless, the results raised several interesting discussion points.
Our depuration system provided highly reproducible results because the Bat condition was used for a previous study (25). These results appeared to be very similar when comparing T90 values and decay coefficients; the T90 values of infectious FRNAPH were 20.6 days here and 21.3 days in the previous study, and the decay coefficients were approximatively 0.90 for GII NoV in both studies. This shows that even if we use only one naturally polluted oyster batch for all experiments, the results seem reproducible for other oyster batches. In addition, the absence of any significant differences between the industrial (Bat) and pilot (B1) tanks also showed that this study was highly representative of industrial practices.
In depuration processes, reactivation of the filtration activity is a crucial point, as is adaptation to suboptimal conditions such as low food availability (52). We have demonstrated here that the presence of nutrients does not provide added value to virus removal, as already shown in other studies comparing viral depuration rates between fed and starved oysters (49, 53). As a consequence, the digestive metabolism of oysters seems to have no influence on virus release (seen for both viruses by following their genomes) or inactivation (seen for FRNAPH followed by culture). Moreover, it can therefore be assumed that no differences exist between relaying and depuration conditions.
Maalouf et al. (38) demonstrated that GII.4 NoV binding to oyster digestive tissues involves sialic acid and A-like HBGAs. Little is known about the presence and distribution of sialic acid in mollusks, even though it has been found in the hemolymph of some marine bivalve species (54). Such a specific interaction has been described to partially explain the different depuration rates observed for different kinds of viruses or mollusks (52). In particular, it has been argued that FRNAPH may not be able to mimic NoV behavior because they cannot recognize these ligands (4). Numerous studies have reported depuration differences between NoV and FRNAPH (17, 44), but most of them compared NoV genomes quantified by RT-qPCR and FRNAPH infectious particles determined by plaque assay. As previously demonstrated (22), we confirm that (i) there was no relationship between the genomes and infectious particles of a same virus (i.e., FRNAPH) and that (ii) the kinetic removal of NoV and FRNAPH genomes were similar during the purification process of oysters. Our results show that there is no difference between virus able to recognize such ligands (NoV) or not (FRNAPH) and that citric acid had no effect on virus depuration. The acidic digestion within hemocytes may than explain the behavior difference discussed above between FRNAPH infectious particles and genomes. Overall, we may conclude that the possible presence of some of HBGA-like ligands is not a reliable argument that could explain limited depuration of some NoV strains in oysters.
Finally, oyster stress was investigated by taking the oysters out of the water for 72 h. Previous studies have reported a two-phase elimination pattern with more rapid elimination on the first day than on subsequent days (25, 52). The first phase (sharp decline) was described to be mainly due to the extracellular digestion and purging of the digestive tract (i.e., defecation) (52). The second phase (stabilization) was suspected to result from specific interactions between viruses and oysters, as already discussed above. In our study, a typical elimination rate was observed in the second phase, as had been expected. Conversely, the sharp decline trend of the first day was reversed, with a slight increase in virus (NoV and FRNAPH) count being detected. It is difficult to explain the temporary slight increase in viral target concentrations, which was observed 17 times out of 20 cases (i.e., four organisms and five conditions) between days 0 and 1, by chance alone (two-sided sign test; P = 0.0013). However, it must be pointed out that the data from day 0 (initial batch) corresponded to a unique value and, as such, had less statistical significance and therefore less reliability than the data from the other days. Comparison between the initial and the stress-subjected batches showed no significant differences, including under nutrient availability conditions (i.e., B3). As a whole, our results suggest that taking oysters out of water for 72 h before reimmersion did not improve virus removal.
All in all, parameters indicative of infectious NoV risk are still lacking since in vitro replication remain complex and do not allow routine analysis. Our results regarding FRNAPH are therefore of interest because the bacteriophages can be detected simultaneously by culture and RT-qPCR. First of all, our findings are in line with many other studies showing that viral genomes are more persistent than infectious particles (15, 16). Indeed, FRNAPH-II genome and the corresponding infectious particles have T90 values of approximately 43.9 and 20.6 days, respectively. Such a difference provides information on the mechanisms involved in virus elimination during oyster purification. In our study, genome levels remained constant, but infectivity levels decreased for the same viral particle (i.e., FRNAPH), suggesting that viruses were inactivated in the oysters rather than released. When viral inactivation is the main process to occur, it is better to focus on infectious particles rather than viral genomes to avoid overestimation of the risk. To sum up, all our data led to the same conclusion: FRNAPH and NoV genomes may display similar behaviors with low kinetic removal from the oysters under all the purification conditions tested. In order to determine whether one of the four FRNAPH genogroups was specifically persistent in oysters, genotyping of PFU obtained by plaque assay methods was performed. This showed no variation frequency of FRNAPH-I and -II during purification, indicating that no specific FRNAPH genotype selection occurred.
Conclusive evidence of a relationship between the presence of infectious FRNAPH in shellfish and NoV risk has not yet been demonstrated. Nevertheless, it is now recognized that the two particles have similar behaviors (followed using the genome quantification) (17). A recent study successfully identified infectious FRNAPH in three of eight NoV foodborne outbreak-related samples (55). In that study, although contamination of the positive batches was a priori linked to wastewater discharges from a treatment plant and a boat, the authors pointed out that the five negative samples had probably been affected by individual contamination. More recently, when targeting the presence of FRNAPH and NoV in oysters, Lowther et al. (8) demonstrated that both viruses were in higher concentrations in outbreak-related samples and that infectious FRNAPH were detected in all outbreak samples (n = 9). The same authors suggested that combining RT-qPCR testing with a test for infectious FRNAPH detection could improve NoV risk assessment in shellfish. Our study is very much in keeping with these two studies and proposes the use of ICC-RT-qPCR for qualitative detection of infectious FRNAPH in shellfish, which may be very useful to better estimate the actual health risk of NoV infection. Furthermore, it would be interesting to validate FRNAPH as an indicator when a routine cell culture approach is available for NoV.
In conclusion, despite some variability in data, the results were consistent and reproducible, as previously demonstrated by Hartard et al. (25). Genome detection has been shown here and elsewhere to provide limited information about the presence of infectious particles in oysters. Next, inactivation has been demonstrated to be the probable primary mechanism of virus elimination from oysters. In addition, FRNAPH-II has been confirmed as a reliable tool for NoV monitoring during purification processes and should be used to better understand NoV behavior and risk in shellfish. Finally, this study has revealed that none of the conditions tested modify virus removal. Therefore, further tests should be performed to identify new strategies allowing the effective elimination of infectious NoV.
MATERIALS AND METHODS
Oyster samples and purification conditions.Oysters (Crassostrea gigas) were collected in January 2018 in class B production areas. Investigated specimens were then positive for E. coli (i.e., >230 MPN/100 g of FIL), according to EC regulation 854/2004 (5). Oysters were split into five different batches and subjected to different purification conditions over a 43-day period. Experimental conditions are described in Table 1. Bat utilized a 200-m3 industrial tank containing 20 tons of seawater and 250 kg of oysters (i.e., approximately 2,500 oysters), whereas B1, B2, and B3 utilized 0.6-m3 pilot tanks of 600 liters containing 60 kg of oysters (i.e., approximately 600 oysters). These tanks contained oxygenated seawater disinfected by continued UV treatment (80 to 90 mJ/cm2) and continuous running. Batch D was stored under relaying conditions by placement in a natural class A site, according to European regulations (5).
Each stress period was induced by taking 20 oysters from batches B1, B2, and B3 out of the water for 72 h at 10°C ± 2°C (named SB1, SB2, and SB3, respectively). These oysters were then reimmersed in water during 24 h before resampling. As a first step, oysters for B1, B2, and B3 were taken out of the water on day 3 and then put again in water on day 6. They were resampling at 24 h (day 7), 48 h (day 8), and 72 h (day 9) after reimmersion. This experiment was repeated according to the same protocol, except for the resampling 48 and 72 h later (Table 1). Samples were analyzed as described for the other experimental conditions (Table 1).
After sampling, all specimens were kept at 4°C for transport and analyses were performed within 3 days.
NoV, FRNAPH, and E. coli analyses. (i) NoV and FRNAPH genome detection.Genomes of FRNAPH subgroup II (FRNAPH-II) and of human NoV (GI and GII) were detected from oysters using a method previously described (25) and the recommendations of the NF EN ISO 15216-1 standard (14), respectively. Briefly, the digestive tissues (DT) of 10 live specimens were recovered by dissection. Each sample was homogenized, and 2 g of the mixture was used for RNA extraction with a NucliSENS kit for easyMAG (bioMérieux, Marcy l’Etoile, France) according to the manufacturer’s recommendations. Using the same RNA extracts, quantification of FRNAPH and NoV genomes was performed in duplicate with an RNA ultrasense one-step quantitative RT-qPCR system (Life Technologies, Carlsbad, CA).
FRNAPH-II genomes were detected using the probes designed by Wolf et al. (20). Quantification was carried out using a standard curve of diluted viral genome with a concentration range of 2.5 to 2.5 × 105 gc/reaction. GI NoV and GII NoV genomes were then detected according to the recommendations of the NF EN ISO 15216-1 standard (14). Quantification was carried out using a standard curve of plasmids with a concentration range of 5.0 to 5 × 105 gc/reaction.
According to the procedures used, the theorical limit of detection (LOD) for FRNAPH and NoV, corresponding to the presence of 1 gc in a PCR well, was close to 40 gc/g of DT. Recovery rates for genome extraction and RT-qPCR inhibition were also determined for each sample using enteric cytopathic bovine orphan (ECBO; ATCC VR-248) virus to determine the efficiency of extraction procedure according to the recommendations of the ISO 15216 standard. ECBO virus was propagated in Madin-Darby bovine kidney cells (MDBK; ATCC CCL-22) as described previously (33). The genome detection of ECBO virus was performed with the primers and probe described previously (56). The inhibition of the RT-PCR for the quantification of NoV genome was controlled using specific external control RNA as described in the ISO 15216-1 standard (14).
(ii) E. coli detection.E. coli was detected by direct impedance measurement in shellfish FIL, according to NF V08-106 (57). Results were expressed as the most probable number (MPN).
(iii) Infectious FRNAPH detection and genotyping.Infectious FRNAPH were detected by two different methods using Salmonella enterica serovar Typhimurium WG49 (NCTC 12484) as the host strain (58). The first method was derived from the ISO 10705-1 standard (21). For each oyster sample, DT tissues from 10 specimens were dissected and mixed with 2 volumes of phosphate-buffered saline–0.15% peptone for 3 min in a DT-20 tube with Ultra-Turrax tube drive (IKA-Werke GmbH & Co. KG, Staufen, Germany) and kept in ice for 4 h. After centrifugation (2,500 × g for 5 min), the supernatant was collected. Culture was performed from 1.5 ml of supernatant, four times, in 150-mm-diameter petri dishes, allowing the analysis of 6 ml of supernatant (corresponding to 2 g of DT). Infectious FRNAPH concentration was expressed in PFU per g of DT after an 18-h incubation period. The theoretical LOD was approximately 1 PFU/2 g.
The second method used for infectious FRNAPH detection was a qualitative ICC-RTqPCR approach performed on the whole shellfish (i.e., FIL) without any dissection step, as described by Hartard et al. (22), with slight modifications. Briefly, 10 oyster specimens were mixed for 3 min in a DT-50 tube with Ultra-Turrax tube drive. Culture of infectious FRNAPH was then performed in 250-ml Erlenmeyer flasks by adding 20 ml of the oyster mixture, 25 ml of 2× tryptone yeast-extract glucose broth, 500 μl of a calcium-glucose solution, 100 μl of a 25 mg/ml kanamycin and nalidixic acid solution, and 30 PFU of FRNAPH-IV used as a culture-positive control. Finally, 5 ml of a S. enterica serovar Typhimurium WG49 suspension prepared as described in the ISO 10705-01 standard (21) was added. Biological amplification was performed at 37°C for 4 h, under agitation (110 rpm). Infectious FRNAPH genome extraction was then performed from 1 ml of the total suspension. After centrifugation (18,000 × g for 3 min), 500 μl of the supernatant was collected, and extraction was performed using NucliSens EasyMag (bioMérieux) in 100 μl of elution buffer. The genomes of each FRNAPH subgroup were detected using the primers and probes developed by Wolf et al. (20) under the conditions described in previous studies (25). The LOD of the ICC-RT-qPCR method was 1 PFU/20 g of whole shellfish flesh.
Statistical analyses.In order to rank the abundance of each viral target (i.e., FRNAPH-II, GI NoV, and GII NoV genomes, and infectious FRNAPH) in shellfish, a three-level classification was applied. Thus, concerning FRNAPH and NoV genomes, specimens were considered negative (–) if no genomes were detected (<40 gc/g of DT), positive (+) if the concentrations were between 40 and 400 gc/g of DT (i.e., 10 times the LOD), and quantifiable (++) if the concentrations were >400 gc/g of DT. Concerning infectious FRNAPH, samples were considered negative (–) when the ICC-RT-qPCR approach gave a quantification cycle (Cq) value greater than 37. Because FRNAPH-I have a better growth rate, the threshold used to differentiate positive (+) from strongly positive (++) samples was 20 for this subgroup, whereas it was 32 for FRNAPH-II and -III (22).
A Wilcoxon signed-rank test was used to perform pairwise comparisons of (i) the storage conditions for each of the four viral targets and (ii) the viral targets across the entire set of storage conditions. The Friedman test was used to carry out a global comparison of (i) the eight storage conditions considered (blocks are formed on the virus-time combinations) and (ii) the four viral targets (blocks are formed on the time points).
A parametric approach was used to examine the kinetics of each of the four viral targets. To this end, the raw data were subjected to a prior logarithmic transformation. A regression analysis was then performed to model the variation of the concentration of each viral target as a log-linear function of the time. This allowed us to estimate, with a confidence interval, the slope and decay coefficient associated with each viral target considered. Finally, a t test was used to test and compare the slopes taken two by two. All of the statistical analyses were generated using R statistical software (v.3.6.1).
ACKNOWLEDGMENTS
The results of this study were obtained within the scope of OxyVir, a project funded by the Fonds Européen pour les Affaires Maritimes et la Pêche (FEAMP). This study was supported by the Joint Technological Unit ACTIA VIROcontrol.
FOOTNOTES
- Received 2 March 2020.
- Accepted 8 April 2020.
- Accepted manuscript posted online 17 April 2020.
- Copyright © 2020 American Society for Microbiology.
