Absence of human astrovirus RNA in sewage and environmental samples


Dr JM Darville, Public Health Laboratory, Myrtle Road, Kingsdown, Bristol, BS2 8EL, UK (e-mail: j.mervyn.darville@bristol.ac.uk).


Over a period of several weeks during the summer of 1996, samples of sewage, sea water, river water, sand and silt were collected from a sewage works at Weston-super-Mare, England and from coastal areas nearby. A sensitive reverse-transcriptase polymerase chain reaction (RT-PCR) was used to search for human astrovirus (HAstV) RNA in concentrates of the samples. No evidence of astrovirus was found in any sample, which suggests that contamination with these viruses is not a problem in this area during the summer holiday season. Furthermore, the single case of astrovirus diarrhoea diagnosed in this laboratory in the summer occurred at the end of the sampling period, and not in the survey area. The primers used sometimes yielded a product two-thirds the expected size but bearing no sequence homology with HAstV. The confirmation that poliovirus adsorbs to sand and silt shows that these materials might be able to concentrate other enteric viruses in water to a level which could be a threat to the health of people coming into contact with it.

Astroviruses are small (28–30 nm), non-enveloped, positive-sense, single-stranded RNA viruses (Matsui & Greenberg 1996), which are transmitted by the faeco-oral route and are a worldwide cause of enteritis, particularly in children (Ashley et al. 1978; Caul 1996; Matsui 1997), although diagnosis is less frequent than that of rotaviruses and enteric adenoviruses. To date, seven serotypes of human astrovirus (HAstV) have been reported (Lee & Kurtz 1994), although recent papers refer to a human astrovirus type 8. Astroviruses distinct from HAstV have also been reported from domestic animals, both mammalian and avian (Kurtz & Lee 1987), and it is probable that they also infect many species of wild animals. Astrovirus enteritis is most common in children of one to three years and is usually seen in winter or early spring in temperate zones such as the UK (Kurtz & Lee 1987; Caul 1996). For example, during 1996, the number of astrovirus reports each month from this laboratory were 6, 6, 1, 1, 3, 0, 0, 0, 1, 0, 5 and 23. Infections are usually sporadic within the community, although outbreaks of infection have occurred in institutions (Matsui & Greenberg 1996). Despite a few anecdotal reports, water has not yet been shown to be a significant factor in outbreaks (Cubitt 1991). In 1995, there was publicity in the UK media following reports of astroviruses in coastal waters (Myint et al. 1994). This preliminary finding, which has not been confirmed, gave rise to considerable concern in the water industry about the origin of these viruses and the potential impact on leisure activities.

Traditional methods for the examination of water for human faecal contamination may not be entirely satisfactory as a means of assessing contamination with human enteric viruses (Gerba et al. 1978; Cook & Myint 1995; Metcalf et al. 1995WMetcalf et al. 1995 has been changed to Metcalf et al. 1995a. For example, coliform bacteria may be derived from animals (Wolf 1972) or be of non-faecal origin (Scarpino 1971). Furthermore, levels of human enteric bacteria do not correlate with the presence of human enteric viruses (Shuval & Katzenelson 1972; Kopecka et al. 1993), which are not removed as efficiently by treatment (Goddard et al. 1981; Morris 1984) and are more persistent in water (Havelaar et al. 1993), sediment (Liew & Gerba 1980) and sewage sludge (Goddard et al. 1981). Certain human enteric viruses are known to cause outbreaks of water-borne disease (Hung et al. 1984; Kukkula et al. 1997; Rab et al. 1997). Molecular techniques such as the polymerase chain reaction offer the possibility of more sensitive detection methods (Cook & Myint 1995).

In this paper, the results are reported of a search at Weston-super-Mare during the summer of 1996 for HAstV RNA in untreated and treated sewage samples, and in samples of environmental waters, potentially contaminated by that sewage, which were collected at the same time.

Materials and methods


Samples were collected on six occasions from the beginning of August to the middle of September. Samples of raw (filtered through 0·2 mm mesh) and chlorine-treated sewage were collected from the sewage works at Weston-super-Mare, England Fig. 1 near here(Fig. 1a). Water samples were collected from the River Axe and from Uphill beach approximately 1 h before high tide (Fig. 1b). At this time, sewage discharge from the outfall flows back into the river and may also be carried towards the beach. Silt (river) and sand (beach) samples were collected at the same time (Fig. 1b). Control water samples were taken from sites (Brean Sands or Sand Point) remote from the outfall at high tide (Fig. 1b). Sodium thiosulphate was added to treated sewage and water samples to inactivate any chlorine. Duplicate sewage and water samples were collected in 4·5 litre plastic containers which were cleaned and steam-sterilized before re-use. Solid samples were collected in duplicate in sterile 250 ml plastic containers.

Figure 1.

(a) Map of England and Wales with the Weston-super-Mare study area arrowed. (b) The Weston-super-Mare coastal area showing sampling sites (arrowed)


The samples were concentrated by standard methods (Standing Committee of Analysts 1995), modified with respect to the volume processed. One litre of the water sample was acidified to pH 3·5 with molar HCl. Using a peristaltic pump, the sample was passed through a 0·45 μm cellulose acetate filter (Whatman) which was fitted, with a pre-filter, to a stainless steel membrane (Sartorius, Epsom, UK) filtration apparatus. The filtrate was discarded. The pH of a sterile 0·5% skimmed milk solution was adjusted to 9·5 using NaOH, and approximately 350 ml of the solution was passed through the filter in the same direction as the water sample. The filtrate was adjusted to pH 3·5 using HCl to induce flocculation and was centrifuged at 3500 g for 20 min at 10 °C in order to pellet the floc. The pellet was resuspended in 1 ml 0·15 mol l−1 Na2HPO4 solution and either processed immediately for RNA extraction, or stored at – 20 °C.

Sand and silt samples were thoroughly mixed with skimmed milk solution at pH 9·5 in order to elute any adsorbed virus. After the solid particles had settled, the supernatant fluid was then treated as above.

In order to determine the recovery rate, 2 ml of a faecal suspension previously shown to contain at least 106 HAstV particles ml−1 was divided into two; 1 ml was added to 1 litre of sea water and concentrated as detailed above. Serial dilutions of the resulting floc suspension and of the remaining faecal suspension were processed by RT-PCR as detailed below. A comparison of the PCR end-points indicated a recovery rate of about 10%, which represents a concentration factor of 100.

RNA extraction

Usually, the method of Boom et al. (1990) was used unmodified. Briefly, 0·1 ml of the floc pellet was mixed with 40 μl silica suspension (Sigma) and 1 ml L6 buffer. After centrifugation, the supernatant fluid was discarded and the silica pellet washed successively in L2 buffer, ethanol twice, and then acetone. After drying the silica pellet, the RNA was eluted in 55 μl sterile water and separated from the silica by centrifugation. When samples were found to be inhibitory to the reactions, 0·1 ml of the redissolved floc pellet was mixed with 0·5 ml TRI reagent (Sigma). Chloroform (0·1 ml) was added and after mixing and centrifugation, the upper aqueous phase (approximately 300 μl) was mixed with the silica and L6 buffer. This extra step reduced the yield of RNA from positive controls.

cDNA synthesis

MMLV Reverse Transcriptase (Gibco BRL), Random Hexamers (Pharmacia), dNTPs and 5 × buffer were added to 35 μl of the RNA solution, as recommended by the supplier, in a final reaction volume of 50 μl. Care was taken to avoid carry-over of silica which may inhibit the RT reaction. The mixture was incubated at 37 °C for 1 h, then used immediately for PCR or stored at – 20 °C.

Polymerase chain reaction

Taq DNA polymerase (Promega), 25 mmol l−1 MgCl2 (Promega), dNTPs and buffer were prepared as a master mix. The primers (1) MON 244 TTA GTG AGC CAC CAG CCA TC and (2) MON 245 GGT GTC ACA GGA CCA AAA CC were as described by Noel et al. (1995) and were synthesized by the oligonucleotide synthesis facility, Department of Biochemistry, University of Bristol. They yield a specific HAstV product of 413 bp from the capsid protein precursor region of the genome. To 18 μl of the mix, 2 μl of cDNA were added. Reaction tubes were placed in the block of a thermal cycler (Perkin Elmer, Norwalk, CT, USAWP kerningPage: 2Please supply company details/address of ) and amplified in the following protocol: 35 cycles of 20 s at 94 °C (first cycle 3 min), 20 s at 50 °C, 20 s at 72 °C. Astrovirus positive controls both for cDNA synthesis and for DNA amplification were included with every batch of samples processed. Diluted astrovirus was added to replicate concentrates of environmental samples in order to look for inhibition of the reverse transcriptase and/or the Taq DNA polymerase.

At the end of the cycle, 2 μl gel loading buffer was added to each reaction and the products were loaded onto a 1% agarose gel with ethidium bromide. After electrophoresis at 120 V for 0·6 h, products were visualized on a u.v. transilluminator (Ultra-Violet Products Inc., San Gabriel, CA, USAWP kerningPage: 3Please supply company details/address of ) and recorded by photography (Polaroid). It was shown that this method could detect HAstV 1–5 and by comparison with electron microscopy, could detect as few as four astrovirus particles.


Weak bands were cut out of the gel and re-amplified in order to obtain sufficient product for sequencing. PCR products were mixed with an equal volume of 40% polyethylene glycol (PEG) 6000 (BDH) in 2·5 mol l−1 NaCl (BDH). After 15 min at room temperature, the mixture was centrifuged at 13 000 g for 30 min. The pellet was resuspended in water and the solution treated with chloroform to remove residual PEG. The products in the aqueous phase were sequenced in a PCR-based automated sequencer (Perkin Elmer) in the Department of Pathology and Microbiology at the University of Bristol. The sequences obtained were compared by computer with sequences lodged in the databases GenBank (release 98·0) and EMBL (release 49·0).


No test samples yielded amplicons comparable in size or intensity with those observed in the controls. About 25% of samples, mainly but not wholly sewage, yielded PCR products. The bands seen, however, were always faint and were approximately two-thirds of the size of those seen with control material WP kerningPage: 3Typesetter: insert Fig. 2 near here(Fig. 2). The smaller, faint bands were considered to be non-specific but nonetheless, some of these products were sequenced along with those from known astroviruses found in stool samples. The sequences were compared with each other and with published HAstV sequences.

Figure 2&.

emsp;PCR products with HAst primers. Lane 1: molecular weight markers (123 bp ladder); lanes 2 and 3: products from water samples (reamplified); lanes 4 and 5: products from astrovirus controls (reamplified). Approximate product sizes in base pairs shown

The astrovirus control products were shown to match closely published sequences from a range of HAst serotypes, and 100% identity was found with some sequences of HAst-1. A comparison between the control virus and a published HAst-1 sequence (accession number Z25771) is shown in WP kerningFig. 3. However, the small products obtained from the environmental samples bore no sequence relationship to these astroviruses. In contrast, weak homology (50–60%) with a range of non-viral sequences was demonstrated. The sources of these sequences include bacteriophage, bacteria, fungi, protozoa, man and other animals. A region of partial homology with bacteriophage P22 is shown in Fig. 3.

Figure 3&.

emsp;Published astrovirus sequence (Z25771) compared with that from the control virus (HAstV C), and phage P22 sequence compared with that of nucleic acid amplified from the environmental samples. The dashes indicate idential nucleotides

Although astrovirus was not detected in sand samples, control experiments showed that sand and silt from the area and elsewhere in the southwest of England were able to bind a vaccine strain of poliovirus (data not shown).


It is unlikely, for two reasons, that HAstV is present in environmental waters in considerable amounts in summer. First, HAstV is relatively uncommon, contributing to only 1–2% of virologically-positive childhood diarrhoeas although, as 80% of the population is seropositive for astrovirus antibody by the age of 10 years, there is probably much asymptomatic infection. Secondly, astrovirus prevalence peaks in winter/spring rather than in summer. Nevertheless, notwithstanding the epidemiology of astroviruses, many children are present in holiday areas such as Weston-super-Mare in summer, and the impact on leisure activities of an outbreak of gastrointestinal infection would be high. Although 1 litre samples were analysed, these included sewage as well as environmental waters. Astrovirus would have been present in far higher numbers in the sewage than in the environmental water into which it is discharged. Had any HAstV been detected in the sewage, then greater volumes of water would have been analysed.

Samples from sporadic cases of gastroenteritis in the community were examined in the laboratory during the survey period. Only one was attributable to astrovirus and this was found at the end of the period in an area over 20 miles from the survey area. This contrasts markedly with the numbers detected during the winter months. It is perhaps not unexpected that astroviruses were not detected in environmental samples collected for the survey, and therefore, the results indicate that human astroviruses are not a major cause of environmental contamination during the summer months. It is, however, worth noting a study from Spain (Pintóet al. 1996) in which astrovirus was detected in environmental water. In that study, water concentrates which were negative for astrovirus RNA by RT-PCR were inoculated into and serially passaged blind in CaCo-2 cells. After the third such passage, astrovirus RNA was detected by RT-PCR in the supernatant fluid of the cells. Most significantly, the water samples had come from a reservoir into which flowed sewage effluent from patients known to be shedding astroviruses. This technique has also been used to demonstrate the persistence of HAstV in dechlorinated drinking water in the laboratory for as long as 90 d (Abad et al. 1997).

It was particularly interesting to note that the astrovirus-specific primers used in this study could amplify reproducibly a small product from a significant number of environmental samples. Although the bands were weak, it might have been tempting to consider them to be derived from astroviruses had they been of the appropriate size. When sequenced, however, these products proved to be totally unrelated to those from human astroviruses, and bore some homology with a wide range of sequences likely to be present in sewage and environmental samples such as those listed above. This finding highlights one of the drawbacks of PCR, namely that the reaction is not always specific. It is therefore essential to improve specificity, for example by hybridization or by restriction endonuclease analysis of amplicons. Both specificity and sensitivity may be improved by using a nested PCR, although this introduces potential problems of cross-contamination with amplicons. Ultimately, the best evidence of specificity is achieved by sequencing the product, as was done in this survey. Another drawback is that PCR does not differentiate infectious from inactivated virus.

Although no evidence of astrovirus was found, this does not rule out the possibility WP kerningthat it would be present when causing disease in the population. Furthermore, the more common agents of rotavirus and the SRSVs (Norwalk-like) probably pose a greater threat. Although sea bathing is unusual in the winter and spring, other activities such as sailing and surfing are practised all year round.

Viruses have been reported to bind to sand and silt particles (Smith et al. 1978; Liew & Gerba 1980) and thus, to survive for prolonged periods. This may be an effective means of concentrating low titres of virus. The confirmation that sand and silt can bind, and perhaps concentrate, viruses indicates a risk with even low concentrations of viable human enteric viruses.


The authors wish to thank Tabytha Coleman, Simon Kershaw, Val Player, Alan Renton, Katya Skerbec, Paul Walker and Polly Weston for their help with this survey. They also thank Wessex Water plc who funded the survey.