Attachment of Enteric Viruses to Bottles - Applied and Environmental

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Aug. 2007, p. 5104–5110 0099-2240/07/$08.00⫹0 doi:10.1128/AEM.00450-07 Copyright © 2007, American Society for Microbiology. All Rights Reserved.

Vol. 73, No. 16

Attachment of Enteric Viruses to Bottles䌤 S. Butot,1 T. Putallaz,1 C. Croquet,3 G. Lamothe,3 R. Meyer,2 H. Joosten,1 and G. Sa´nchez1* Quality & Safety Assurance Department, Nestle´ Research Center, Lausanne, Switzerland1; Quality Management Department, Nestle´ Product Technology Center, Orbe, Switzerland2; and Nestle´ Waters, Vittel, France3 Received 27 February 2007/Accepted 14 June 2007

Storage of water that was deliberately contaminated with enteric viruses in polyethylene terephthalate (PET) bottles led to a rapid decrease of the apparent viral load, thereby hampering the development of samples for a collaborative evaluation of viral detection methods for bottled water. To determine if this decrease was due to spontaneous inactivation or to adhesion, an elution protocol was developed and combined with a rapid and sensitive real-time reverse transcription-PCR-based method to quantify adsorbed norovirus (NV), hepatitis A virus (HAV), and rotavirus (RV) on bottle walls. The NV retention on PET bottle walls after 20 and 62 days reached an average level of 85% and 95% of the recovered inoculum, respectively. HAV and RV also showed adsorption onto PET bottles, reaching 90% and 80%, respectively, after 20 days of storage. NV and RV attachment was demonstrated to be dependent on the presence of autochthonous flora, whereas HAV adsorption was independent of it. Application of the elution and viral detection protocol to 294 commercially available water bottles obtained from 25 different countries did not give any positive result, thereby providing further evidence that the sources used for this product are free from enteric viruses and support for the theory that bottled water is not a vehicle for viral diseases. virus strain, the chemical composition of the water, and the presence of autochthonous microorganisms. After developing an efficient elution protocol, we also undertook a survey of 294 commercially available water bottles obtained from 25 different countries.

Food-borne and waterborne outbreaks of human enteric viruses, such as norovirus (NV), hepatitis A virus (HAV) and rotavirus (RV), are a matter of serious concern for public health. Although there is no epidemiological evidence that bottled water serves as a vehicle for viral diseases, some doubts were raised concerning its safety due to the reported finding of NV sequences in 33% of commercially available water samples sold in Switzerland (2, 3). However, attempts by other research groups to reproduce these results were always in vain (6, 15, 16, 19, 21, 27), even though much larger numbers of samples and more-sensitive detection methods were used. Further investigations strongly suggested that the original findings were due to artifacts and systematic mistakes (21). Since these unfounded allegations could have severe economic consequences for the bottled water industry, it is essential to develop internationally accepted virus detection methods for this matrix. An essential step in the validation of such methods is the organization of a collaborative trial to demonstrate its reproducibility. When preparing artificially contaminated samples for the validation of our NV detection method in bottled water based on membrane filtration and real-time reverse transcriptionPCR (RT-PCR) (16), we observed a substantial decrease of the viral load after a few days of storage. Since adsorption of human enteric viruses to the walls of different container materials had been reported previously (7, 8, 20, 25, 28), we suspected that a similar phenomenon was occurring in our samples. We therefore investigated whether enteric viruses may also adsorb to polyethylene terephthalate (PET) and glass bottles and to what extent such adsorption depends on the

MATERIALS AND METHODS Cells, viruses, and infections. A clinical stool sample positive for NV (genogroup I [GI], Valetta strain; kindly provided by RIVM, Bilthoven, The Netherlands) was used as NV reference material. The cytopathogenic HM-175 strain of HAV (courtesy of A. Bosch, Enteric Virus Group, University of Barcelona, Spain) and the simian RV strain SA-11 were propagated and assayed in FRhK-4 and MA-104 cell monolayers, respectively. Semipurified stocks were produced with the same cells by low-speed centrifugations of infected cell lysates. Infectious virus enumerations were performed by determining the 50% tissue culture infectious dose (TCID50) with eight wells per dilution and 20 ␮l of inoculum per well. Bottled waters. Locally purchased bottled waters with different mineral compositions were used during this study (Table 1). Viral genome quantification. Viral RNAs from water and walls were quantified using real-time RT-PCR. NV GI RNA was quantified using a specific assay for the Valetta strain, as described elsewhere (16), with an ABI Prism 7700 sequence detection system (PE Applied Biosystems, Foster City, CA). The NV RT reaction was performed at 41°C for 60 min using a Sensiscript RT kit (QIAGEN GmbH, Hilden, Germany) consisting of 1⫻ RT buffer, 500 ␮M (each) nucleotides, 1 ␮M 9.4 reverse primer, 1 ␮l of Sensiscript reverse transcriptase, 10 U of RNase inhibitor (Promega, Madison, WI), and 10 ␮l of NV RNA in a final volume of 25 ␮l. NV real-time PCR was performed using a TaqMan Universal PCR Master Mix (Applied Biosystems) consisting of 1⫻ TaqMan buffer, 0.2 ␮M TaqMan probe 9.4, and 0.3 ␮M 9.4 reverse and forward primers in a final volume of 50 ␮l containing 10 ␮l of cDNA. Amplification was performed for 1 cycle of 50°C for 2 min, 1 cycle of 95°C for 10 min, and 48 cycles of 95°C for 15 s and 58°C for 1 min. A NV standard curve was generated by amplifying 10-fold dilutions of the stool extract by using real-time RT-PCR. The cycle threshold value obtained from the assay of each dilution was used to plot a standard curve by assigning a value of 1 RT-PCR unit (PCRU) to the highest dilution showing a positive cycle threshold and values of 10, 100, and 1,000 PCRU sequentially to the lower dilutions. HAV RNA was quantified by using a LightCycler HAV quantification kit (Roche Diagnostics, Mannheim, Germany) with a LightCycler instrument as described previously (5, 22). Briefly, a total of 2.5 ␮l of extracted RNA was transferred into a capillary containing 7.5 ␮l of Master Mix. RT was performed

* Corresponding author. Mailing address: Nestle´ Research Center, Vers-chez-les-Blanc, CH-1000 Lausanne 26, Switzerland. Phone: 41 21 785 8692. Fax: 41 21 785 8553. E-mail: gloria.sanchez-moragas@rdls .nestle.com. 䌤 Published ahead of print on 22 June 2007. 5104

VOL. 73, 2007

VIRUS ADSORPTION IN BOTTLES

TABLE 1. Mineral composition of bottled water brands used in this study Amt (mg/liter) Brand no. Mg

1 2 3 4 5

2⫹

43.0 110.0 20.0 84.0 2.1

Ca2⫹

Na⫹

202 555 91 486 70

4.7 14.0 7.3 9.1 2.0

pH

7.3 7.1 7.1 7.3 7.7

at 55°C for 10 min, and amplification was performed for 1 cycle of 95°C for 30 s and 45 cycles of 95°C for 5 s, 55°C for 15 s, and 72°C for 12 s. The LightCycler HAV quantification kit contains internal HAV RNA standards that allow the number of RNA copies per sample to be estimated. RV real-time RT-PCR was based on a previous publication (23) with some modifications and was adapted to the LightCycler instrument (Roche Diagnostics). Briefly, real-time RT-PCR was performed using a QuantiTect SYBR Green RT-PCR kit (QIAGEN) consisting of 10 ␮l QuantiTect SYBR Green, 0.2 ␮l QuantiTect RT mix, 0.625 ␮M (each) P1 and P2 primers, and 0.8 U of RNase inhibitor (Promega). Five microliters of RNA was denaturated by heating for 5 min at 95°C and transferred into a capillary containing 15 ␮l of Master Mix. RT was performed at 50°C for 30 min; amplification was performed for 1 cycle of 95°C for 15 min and 50 cycles of 94°C for 30 s, 58°C for 60 s, 72°C for 1 min, and 75°C for 5 s. The data were collected in the single mode during the 75°C segment. A melting curve analysis was performed (segment 1 [95°C for 5 s; slope, 20°C/s; acquisition mode, none], segment 2 [65°C for 5 s; slope, 20°C/s; acquisition mode, none] and segment 3 [95°C for 0 s; slope, 0.1°C/s; acquisition mode, continuous]). A RV standard curve was generated by performing real-time RT-PCR with 10-fold dilutions of strain SA-11-extracted RNA. The crossing points obtained from the assay of each dilution were used to plot a standard curve by assigning the corresponding TCID50 values. Analysis of inoculated water samples. Three liters of bottled and deionized water were inoculated with ca. 2.4 ⫻ 106 PCRU of NV, 1.5 ⫻ 105 TCID50 of HAV, or 3.3 ⫻ 103 TCID50 of RV. Inoculated samples were gently mixed and distributed over nine bottles. Batches of three bottles were analyzed at days 0 (immediately after inoculation), 20, and 62. Samples were stored undisturbed in the dark at room temperature. Water samples were analyzed as described elsewhere (9, 16). Briefly, after the bottles were shaken vigorously, the water was filtered through a 0.45-␮m-poresize positively charged membrane (Zetapor filter membrane; CUNO, Inc., Meriden, CT). Virus particles were further concentrated by ultrafiltration (Amicon centrifugal filter device, 100K NMWL; Millipore, Molsheim, France) followed by RNA isolation using a commercially available kit (QIAamp viral RNA mini kit; QIAGEN). RNA extracts were either analyzed immediately by real-time RT-

5105

PCR or stored at ⫺80°C until use. Nucleic acid suspensions were analyzed in duplicate by the specific real-time RT-PCR method, which also allowed us to estimate the number of recovered viruses or RNA copies. Empty bottles were filled with 10 ml of elution buffer (50 mM glycine, 1% [wt/vol] beef extract [BD Difco beef extract paste; Becton Dickinson AG]; pH 9.5). Attached viruses were released from the bottle by vigorous shaking (Turbula T10B shaker mixer; Willy A. Bachofen AG) for 20 min at room temperature. The elution buffer was then transferred into a tube, and empty bottles were rinsed with 2 ml of elution buffer. The recovered elution buffer was then adjusted to pH 8.0 ⫾ 0.2 with 1 M HCl, transferred to an Amicon Ultra-15 centrifugal filter device (100K NMWL; Millipore), and centrifuged at 4,000 rpm to concentrate the virus particles. The eluate was adjusted to 140 ␮l with phosphatebuffered saline, and RNA extraction was performed using a QIAamp viral RNA mini kit (QIAGEN) according to the manufacturer’s instructions (Fig. 1A). Evaluation of alternative elution procedures. In an attempt to improve the recovery of attached viruses, the suitability of five different elution buffers was compared: (i) 50 mM glycine, 1% (wt/vol) BD Difco beef extract paste (pH 9.5); (ii) 50 mM glycine, 3% (wt/vol) BD Difco beef extract paste (pH 9.5); (iii) 50 mM glycine, 1% (wt/vol) BD Difco beef extract paste, 5 mM sodium dodecyl sulfate (SDS) (pH 9.5); (iv) 50 mM glycine, 1% (wt/vol) BD Difco beef extract paste, 10 mM SDS (pH 9.5); and (v) 50 mM glycine, 1% (wt/vol) BD Difco beef extract paste, 20 mM SDS (pH 9.5). An increased recovery was also attempted by treating the inoculated bottles by submersion for 4 min in an ultrasound bath (Bransonic ultrasonic cleaner, 47 kHz; Branson Ultrasonics Corporation, Danbury, CT). These experiments were all carried out with 330-ml PET bottles containing mineral water brand 3, which was inoculated with ca. 8.1 ⫻ 105 PCRU of NV. The bottles were stored undisturbed in the dark at room temperature and analyzed at day 20. Each alternative procedure was analyzed in triplicate and processed on the same day. Effect of autochthonous flora on adsorption. The autochthonous flora of bottled water was not characterized, but the influence of this flora was studied by comparing virus adsorptions for filter-sterilized water with those for non-filtersterilized water. Three liters of bottled water brand 3 and deionized water were sterilized through a 0.22-␮m-pore-size filter (Millipore) and then inoculated with ca. 2.4 ⫻ 106 PCRU of NV, 1.5 ⫻ 105 TCID50 of HAV, or 3. 3 ⫻ 103 TCID50 of RV. Inoculated samples were gently mixed and distributed over nine bottles (330 ml PET). Batches of three bottles were analyzed at days 0 (immediately after inoculation), 20, and 62. Samples were stored undisturbed in the dark at room temperature and analyzed as described above. Analysis of commercially bottled water samples. A simplified protocol for analyzing simultaneously the water phase and the packaging material was developed (Fig. 1B and C). The initial steps for water and bottle wall analysis were identical to those described above. Once viruses were released from positively charged membrane and bottle walls, elution buffers were pooled and further concentrated using a centrifugal filter device (100K NMWL) adapted to the volume required. The eluate was adjusted to 200 ␮l with phosphate-buffered saline, and then RNA extraction was performed by using a Magna Pure Compact Robot (Roche Diagnostics), applying the “DNA_Blood_100_400” protocol ac-

FIG. 1. Flow chart of the method. Method A uses independent analysis of water and wall phases, whereas methods B and C analyze them as a single sample. l, liter; EB, elution buffer.

5106

BUTOT ET AL.

APPL. ENVIRON. MICROBIOL.

FIG. 2. NV adsorption on PET bottles at days 0, 20, and 62. Percentages of virus recovery from the water and the wall (the amount of viruses recovered from the wall or the water divided by the total amount of recovered viruses) are depicted in white and gray, respectively. Error bars represent standard deviations (n ⫽ 3). Numbers depicted on the x axis correspond to the bottled water brands. H2Od, deionized water.

cording to the manufacturer’s instructions. Viral RNAs were detected using real-time RT-PCR for HAV and RV and by RT-PCR for generic NV detection as described elsewhere (16). The presence of inhibitors was monitored by using the internal control supplied with the HAV kit. For NV detection, RNA samples were analyzed twice, once without and once with the addition of a positive NV RNA sequence.

RESULTS Development of a protocol to release viruses from bottle walls. Storage for 3 weeks of water that was inoculated with NV in PET bottles led to a ca. 90% decrease of the viral load in the water phase compared with that of water analyzed immediately after inoculation. To determine whether adsorption played a role in this loss, a procedure was developed to release viruses from bottle walls. NV attached to bottle walls after 20 days of storage was tentatively released using different elution buffers. As each of the five tested elution buffers gave approximately 30% recovery (data not shown), the buffer that was also used for releasing viruses from the positively charged filter (50 mM glycine, 1% beef extract; pH 9.5) was selected for the standard protocol to elute the adsorbed viruses from the bottle wall. Exposure to ultrasound reduced the NV recovery to less than 0.1% and was therefore not included in the standard protocol. NV adsorption in different types of bottled waters. NV adsorption onto the bottle wall (PET and glass bottles) was evaluated after 0, 20, and 62 days using five different bottled water brands and deionized water. To investigate viral adhesion to

PET or glass bottles, the percentages of the recovered genome present in the aqueous phase and adhering to walls were monitored at days 0, 20, and 62. In PET bottles, NV retention on the bottle walls after 20 days reached an average level of approximately 75% of the total recovered inoculum versus only 2.5% at day 0. Viral adsorption rose to 91% after 62 days of storage. No significant differences between the five types of bottled waters were observed, whereas in deionized water, virus adsorption to the bottle wall was limited to only 35% after 62 days of storage (Fig. 2). Attachment on glass bottles ranged from 18% to 73% after 20 days of storage, depending on the bottled water composition, whereas at day 0, only 4.5% of adsorption, as an average for the five types of bottled water, was observed. Percentages after 62 days of storage rose to 75%, similar to those obtained after 20 days for PET bottles (Fig. 3). NV inoculated in bottled water brand 5 showed the lowest adsorption in glass bottles, where the virus particles quantity initially found in water at day 0 was more than 130 times greater than that found attached to the bottle. Between days 0 and 20, a small fraction of virus particles moved from the aqueous phase to the bottle walls, with more than four times more virus particles in the water phase than eluted from the bottle walls. Only after 62 days was the proportion of NV genome adsorbed onto the bottle walls higher (1.3 times) than in the water phase. HAV and RV adsorption onto PET bottles. When added to bottled water brand 3, HAV also showed adsorption onto the bottles, reaching 85% of virus retention after only 20 days

FIG. 3. NV adsorption on glass bottles at days 0, 20, and 62. Virus recovery (the amount of viruses recovered from the wall or the water divided by the total amount of recovered viruses) from the water and the wall are depicted in white and gray, respectively. Error bars represent standard deviations (n ⫽ 3). Numbers depicted in the x axis correspond to the bottled water brands. H2Od, deionized water.

VOL. 73, 2007


5107

FIG. 4. HAV adsorption onto PET bottles at days 0, 20, and 62. Comparison between bottled water and filter-sterilized water. Percentages of virus recovery (the amount of viruses recovered from the wall or the water divided by the total amount of recovered viruses) from the water and the wall are depicted in white and gray, respectively. Lines represent the total viral titer recovered (䡺), viral titer recovered in water (E), and viral titer recovered in walls (Œ) expressed as RNA copies. (A) Bottled water. (B) Deionized water. Error bars represent standard deviations (n ⫽ 3).

(Fig. 4A), whereas 46% virus adsorption was detected at day 0. At day 62, the quantity of virus particles in the aqueous phase was small (ca. 9.5%) compared with the quantity adhering to the bottle. Adsorption in other brands of bottled water was not investigated. RV retention onto the bottle walls after 20 days reached an average level of approximately 80% of the total virus recovered. This percentage rose to 95% after 62 days of storage. As observed with NV and HAV, only 40% of RV was retained on the walls of bottles filled with deionized water after 62 days (Fig. 5). Autochthonous flora effect on viral adsorption. To determine whether autochthonous flora plays a role in virus adsorption, 0.22-␮m-pore-size-filtered bottled water (filter sterilized) was inoculated separately with the three viruses. The use of this water in PET bottles strongly reduced the adsorption of NV, as more than 95% of the recovered inoculum was still detectable in the water phase after 62 days of storage. In the deionized water, where 38% of NVs were adsorbed to walls after 62 days of storage, more than 92% were still present in the water phase after 62 days if the water was filtered prior to inoculation (Fig. 6). RV attachment in filter-sterilized bottled water reached a level of 20% and 50% after 20 and 62 days, respectively, which

is four and two times lower than that observed with unfiltered water (Fig. 5). HAV adsorption, in contrast, was hardly affected by the presence of the autochthonous flora (Fig. 4). Kinetics of virus titer recovery. The total recovered virus titer from the aqueous phase and bottle walls was estimated using the corresponding standard curve at days 0, 20, and 62. NV (PCRU) and RV (TCID50) titers were quite stable in bottled water (18% and 38% titer loss for NV and RV, respectively), whereas ca. 95% and 83% decreases were observed with deionized water after 62 days of storage (Fig. 5 and 6). In contrast, the total HAV titer, surprisingly, increased with the storage time, 3 and 15 times in nonsterilized and filter-sterilized bottled water, respectively, after 62 days of storage (Fig. 4A). To elucidate this trend, the HAV titer was plotted separately in water and wall phases (Fig. 4), showing that the virus titer increase was observed only for the walls. At day 0, HAV was distributed equally on the bottle walls and in the water phase, but on day 20, the majority of genome was found adsorbed onto the bottle walls, with more than two times greater amounts than those found at day 0 in the aqueous phase. At day 62, the genome adhering to walls was more than 4 and 20 times that initially found in the aqueous phase for nonsterilized and filter-sterilized bottled water, respectively. To ascertain whether this titer increase was due to a technical problem, the

5108

BUTOT ET AL.


FIG. 5. RV adsorption on PET bottles and titer at days 0, 20, and 62. Comparison between bottled water and filter-sterilized water. Percentages of virus recovery (the amount of viruses recovered from the wall or the water divided by the total amount of recovered viruses) from the water and the wall are depicted in white and gray, respectively. Lines represent the total viral titer recovered, expressed as the TCID50. (A) Bottled water. (B) Deionized water. Error bars represent standard deviations (n ⫽ 3).

efficiency of the water filtration protocol (inoculated viruses/ recovered viruses in the water phase at day 0) was determined for the three viruses and was determined to be only 1% for HAV (instead of the ca. 30% and 10% obtained for NV and RV, respectively), suggesting that the HAV titer in water was underestimated; as soon as HAV was adsorbed to bottle walls, the efficiency of recovery was much better. Analysis of bottled water samples. In the framework of verifying the safety of bottled water, a worldwide survey of 294 commercial bottled water samples of different origins was carried out (Table 2). All tested bottled water samples were found to be negative for the presence of NV, HAV, and RV genomes in the aqueous phase and on bottle walls. The presence of RT-PCR inhibitors was evaluated for all brands of bottled waters included in this survey for NV and HAV. The HAV internal control was detected in all the tested samples, as was the NV RNA added to the RNA extracts. These results suggest that inhibitors did not play a role in this study. DISCUSSION We have shown that enteric viruses (NV, HAV, and RV) in bottled water can become adsorbed to the bottle wall. This adsorption depends on the type of virus, chemical composition

of the water, bottle material, duration of storage, and presence of autochtonous bacteria and may reach 99.7% of the total recovered virus after 62 days of storage, which is comparable to the attachment reported by Gassilloud and Gantzer (7), where 1.5 log10 of poliovirus genome was transferred from the aqueous phase onto the polypropylene walls during the first 20 days of incubation. Very recently, Gassilloud and coauthors (8) also reported a nearly 3-log10 reduction of infectious poliovirus levels in bottled water stored in PET bottles. From a method standardization point of view, virus adsorption onto bottles is undesirable, because it makes it more difficult to prepare stable reference samples that can be used for collaborative trials. However, our results also showed that viruses can also be desorbed from the bottle wall by rinsing the bottles with a small amount of elution buffer. By combining this elution procedure with the normal concentration procedure for virus detection, the total amounts of NV and RV recovered from the bottles were fairly stable (less than 38% loss over 62 days). It is therefore recommended that this combined protocol be used for collaborative studies to facilitate comparisons of the results. NV adsorption onto glass bottles was also detected, but more time was needed to reach the same adsorption level as

VOL. 73, 2007


5109

FIG. 6. NV adsorption on PET bottles and titer at days 0, 20, and 62. Comparison between bottled water and filter-sterilized water. Percentages of virus recovery (the amount of viruses recovered from the wall or the water divided by the total amount of recovered viruses) from the water and the wall are depicted in white and gray, respectively. Lines represent the total viral titer recovered, expressed in PCRU. (A) Bottled water. (B) Deionized water. Error bars represent standard deviations (n ⫽ 3).

that measured for PET bottles. NV adsorption may be influenced by the mineral composition of the water, since water brand 5, with the lowest content of Mg2⫹, Ca2⫹, and Na⫹ ions, and deionized water showed the lowest degrees of adsorption onto glass and PET bottles. It is noteworthy that virus attachment using 0.22-␮m-poresize-filtered bottled water showed different patterns depending on the virus. NV was detected mainly in the water phase of filter-sterilized bottled water after 62 days of storage, whereas

TABLE 2. Results for finished bottled water samples investigated during the virus survey No. positive/no. of samples tested Origin

Europe Middle East Asia South America

Norovirus GI

GII

0/180 0/43 0/41 0/30

0/180 0/43 0/41 0/30

HAV

RV

0/180 0/43 0/41 0/30

0/50 0/43 0/29 0/30

only 1% was found in the water phase if the bottled water contained its own autochthonous flora. For RV, these percentages were 50% and 3.5% for nonsterilized and filter-sterilized bottled water, respectively, whereas for HAV, 2.7% and 9.5% were recovered. In a recent publication on poliovirus adsorption to hydrophobic tubes (7), it was reported that the autochthonous flora did not appear to have any influence. Differences in adsorption depending on the viral species have also been described for adsorption to soils (ranging from 0 to 99%) and lettuce and have been attributed to the variability of the capsid, which influences the net charge on the virus (11, 26). Assuming no role for the autochthonous flora, the HAV attachment may be explained by the isoelectric point of the viral capsid, affecting the electrostatic potential between virus and bottle and enhancing the affinity of viruses for the walls. This assumption is supported by the fact that even at day 0, half of the HAV titer, in nonsterilized and filter-sterilized bottled water, was found attached to the bottle walls. It is well known that many bacteria are able to establish a biofilm on the bottle walls (13, 14), and it may well be that NV and RV adsorb more readily to the biofilm than to the bottle

5110

BUTOT ET AL.


itself. A second hypothesis, suggested also for virus interaction with clay (18), may involve some bacterial components that may reduce the electrostatic repulsive forces between the viruses and the PET bottle, thereby permitting the viruses to approach the PET surface sufficiently for adsorption to occur by cation exchange, hydrogen bonding, van der Waals forces, or hydrophobic interactions (17, 24). Many studies conducted on the survival of enteric viruses in water indicated that the loss of virus titer occurs at a variety of rates (1, 4, 7, 10, 12, 29). However, in several of these studies, loss of viruses due to adherence to the support was not considered. Under our experimental conditions, the total viral genome remained quite stable over the analyzed period of time in bottled water (nonsterilized and filter-sterilized), with the exception of HAV, where the titer increase was due to the better performance of the procedure to release virus adsorbed to walls than the water filtration method. It is debatable whether bottle analysis should also be used for monitoring purposes. Once the viruses are adsorbed, they are no longer of concern for the consumer, and it therefore seems pointless to try to detect them. However, one can argue that if an adsorbed virus were found, it would be an indication that the water source is contaminated and that viruses were present in the water phase at some point. From this perspective, it would be better to utilize the combined protocol, because a negative result with this method would be more meaningful than a negative result obtained with a method that does not detect adsorbed viruses. For this reason, we performed a survey of 294 samples of bottled water which were obtained from 25 different countries using the combined protocol. Since all samples tested negative for NV GI, NV GII, HAV, and RV, it is concluded that the corresponding sources were not contaminated with these viruses. These results, together with the absence of epidemiological data linking the consumption of bottled water to viral outbreaks, strongly support the theory that bottled water is not a vehicle for viral infections. ACKNOWLEDGMENT We are grateful to Thomas Kappeler for a critical reading of the manuscript. REFERENCES 1. Arnal, C., J. M. Crance, C. Gantzer, L. Schwartzbrod, R. Deloince, and S. Billaudel. 1998. Persistence of infectious hepatitis A virus and its genome in artificial seawater. Zentralbl. Hyg. Umweltmed. 201:279–284. 2. Beuret, C., D. Kohler, A. Baumgartner, and T. M. Lu ¨thi. 2002. Norwalk-like virus sequences in mineral waters: one-year monitoring of three brands. Appl. Environ. Microbiol. 68:1925–1931. 3. Beuret, C., D. Kohler, and T. Lu ¨thi. 2000. Norwalk-like virus sequences detected by reverse transcription-polymerase chain reaction in mineral waters imported into or bottled in Switzerland. J. Food Prot. 63:1576–1582. 4. Biziagos, E., J. Passagot, J. M. Crance, and R. Deloince. 1988. Long-term survival of hepatitis A virus and poliovirus type 1 in mineral water. Appl. Environ. Microbiol. 54:2705–2710. 5. Butot, S., T. Putallaz, and G. Sanchez. 2006. Procedure for rapid concentration and detection of enteric viruses from berries and vegetables. Appl. Environ. Microbiol. 73:186–192.

6. Ehlers, M. M., W. B. Van Zyl, D. N. Pavlov, and E. E. Muller. 2004. Random survey of the microbial quality of bottled water in South Africa. Water SA 30:203–210. 7. Gassilloud, B., and C. Gantzer. 2005. Adhesion-aggregation and inactivation of poliovirus 1 in groundwater stored in a hydrophobic container. Appl. Environ. Microbiol. 71:912–920. 8. Gassilloud, B., L. Huguet, A. Maul, and C. Gantzer. 2007. Development of a viral concentration method for bottled water stored in hydrophobic support. J. Virol. Methods 142:98–104. 9. Gilgen, M., D. Germann, J. Lu ¨thy, and P. Hu ¨bner. 1997. Three-step isolation method for sensitive detection of enterovirus, rotavirus, hepatitis A virus, and small round structured viruses in water samples. Int. J. Food Microbiol. 37:189–199. 10. Gordon, C., and S. Toze. 2003. Influence of groundwater characteristics on the survival of enteric viruses. J. Appl. Microbiol. 95:536–544. 11. Goyal, S. M., and C. P. Gerba. 1979. Comparative adsorption of human enteroviruses, simian rotavirus, and selected bacteriophages to soils. Appl. Environ. Microbiol. 38:241–247. 12. Hurst, C. J., and C. P. Gerba. 1980. Stability of simian rotavirus in fresh and estuarine water. Appl. Environ. Microbiol. 39:1–5. 13. Jayasekara, N. Y., G. M. Heard, J. M. Cox, and G. H. Fleet. 1999. Association of micro-organisms with the inner surfaces of bottles of non-carbonated mineral waters. Food Microbiol. 16:115–128. 14. Jones, C. R., M. R. Adams, P. A. Zhdan, and A. H. Chamberlain. 1999. The role of surface physicochemical properties in determining the distribution of the autochthonous microflora in mineral water bottles. J. Appl. Microbiol. 86:917–927. 15. Khanna, N., D. Goldenberger, P. Graber, M. Battegay, and A. F. Widmer. 2003. Gastroenteritis outbreak with norovirus in a Swiss university hospital with a newly identified virus strain. J. Hosp. Infect. 55:131–136. 16. Lamothe, G. T., T. Putallaz, H. Joosten, and J. D. Marugg. 2003. Reverse transcription-PCR analysis of bottled and natural mineral waters for the presence of noroviruses. Appl. Environ. Microbiol. 69:6541–6549. 17. Lipson, S. M., and G. Stotzky. 1983. Adsorption of reovirus to clay minerals: effects of cation-exchange capacity, cation saturation, and surface area. Appl. Environ. Microbiol. 46:673–682. 18. Lipson, S. M., and G. Stotzky. 1985. Effect of bacteria on the inactivation and adsorption on clay minerals of reovirus. Can. J. Microbiol. 31:730–735. 19. Liu, J., Q. Wu, and X. Kou. 2007. Development of a virus concentration method and its application for the detection of noroviruses in drinking water in China. J. Microbiol. 45:48–52. 20. Moore, R. S., D. H. Taylor, L. S. Sturman, M. M. Reddy, and G. W. Fuhs. 1981. Poliovirus adsorption by 34 minerals and soils. Appl. Environ. Microbiol. 42:963–975. 21. Sanchez, G., H. Joosten, and R. Meyer. 2005. Presence of norovirus sequences in bottled waters is questionable. Appl. Environ. Microbiol. 71: 2203–2205. 22. Sanchez, G., S. Populaire, S. Butot, T. Putallaz, and H. Joosten. 2006. Detection and differentiation of human hepatitis A strains by commercial quantitative real-time RT-PCR tests. J. Virol. Methods 132:160–165. 23. Schwarz, B. A., R. Bange, T. W. Vahlenkamp, R. Johne, and H. Muller. 2002. Detection and quantitation of group A rotaviruses by competitive and realtime reverse transcription-polymerase chain reaction. J. Virol. Methods 105: 277–285. 24. Shields, P. A., and S. R. Farrah. 1983. Influence of salts on electrostatic interactions between poliovirus and membrane filters. Appl. Environ. Microbiol. 45:526–531. 25. Taylor, D. H., R. S. Moore, and L. S. Sturman. 1981. Influence of pH and electrolyte composition on adsorption of poliovirus by soils and minerals. Appl. Environ. Microbiol. 42:976–984. 26. Vega, E., J. Smith, J. Garland, A. Matos, and S. D. Pillaii. 2005. Variability of virus attachment patterns to butterhead lettuce. J. Food Prot. 68:2112– 2117. 27. Villar, L. M., V. S. De Paula, L. Diniz-Mendes, E. Lampe, and A. M. Gaspar. 2006. Evaluation of methods used to concentrate and detect hepatitis A virus in water samples. J. Virol. Methods 137:169–176. 28. Ward, R. L., and P. E. Winston. 1985. Development of methods to measure virus inactivation in fresh waters. Appl. Environ. Microbiol. 50:1144–1148. 29. Yates, M. V., C. P. Gerba, and L. M. Kelley. 1985. Virus persistence in groundwater. Appl. Environ. Microbiol. 49:778–781.