Molecular epidemiology of human astroviruses

Stephan S. Monroe*, Jennifer L. Holmes*f and Gael M. Belliot*{

* Viral Gastroenteritis Section, 'Division of Viral and Rickettsial Diseases, National Center for Infectious Diseases, Centers for Disease Control and Prevention, A tlanta, GA 30333, USA, f Oak Ridge Institute for Science and Education, Oak Ridge, TN, USA, and %A tlanta Research and Education Foundation, A tlanta, GA 30033, USA

Abstract. Human astroviruses (HAstVs) are associated with 5—9 percent of cases of gastroenteritis in young children. Seven serotypes (HAstV-1 to -7), which correlate with genotypes, have been defined by using immune typing methods. We have used partial nucleotide sequence information from the capsid protein gene for molecular typing of 29 unique human astrovirus strains obtained from prospective studies of children with gastroenteritis in Egypt and Malawi. HAstV-1 was the most commonly detected strain, consistent with previous studies, but a surprising variety of strains were identified in both collections. An eighth astrovirus type, HAstV-8, has been defined on the basis of the complete capsid protein gene sequence and was detected in both collections analysed in this study. Although HAstV-8 and HAstV-4 strains segregate into well resolved clades by analysis of sequences from the region encoding protein P2 (VP32), the pair-wise distances between these types are less than those between strains of the other serotypes. In contrast, analysis of sequences from the region encoding protein P3 unambiguously resolve HAstV-4 and HAstV-8 strains, consistent with their classification as distinct serotypes. Overall, strains representing six of the eight serotypes were detected in two collections of samples from prospective studies of gastroenteritis in young children indicating that multiple astrovirus types are frequently co-circulating within communities.

2001 Gastroenteritis viruses. Wiley, Chichester (Novartis Foundation Symposium 238) p 237-249

Astroviruses were first detected in 1975 during examination of the faeces of infants with diarrhoea by using electron microscopy (Madeley & Cosgrove 1975, Appleton et al 1977). The viruses were described as 28—30 nm particles with a smooth outer edge and a surface star appearance and were given the name astrovirus. Particles with similar structural features were subsequently detected in the faeces of a variety of animal species (reviewed in Matsui & Greenberg 1996). Cultivation of human astroviruses was originally achieved in primary human embryonic kidney cells, with a stringent requirement for trypsin in the growth medium (Lee & Kurtz 1981). Direct isolation of human astroviruses from clinical specimens was later reported using a continuous human colon carcinoma cell line, Caco-2 (Willcocks et al 1990), allowing for routine isolation of strains from well characterized collections.

Our awareness of the importance of astroviruses as a cause of infantile diarrhoea has changed as more sensitive detection methods have been developed in recent years. The concentration of astrovirus shed during infection is lower than that observed for many other enteric viruses (e.g. rotavirus and adenovirus) and initial studies, which relied on detection by relatively insensitive electron microscopy, reported astrovirus in less than 3% of cases of childhood diarrhoea (Lew et al 1990, Monroe et al 1991, reviewed in Glass et al 1996). The generation of a group-reactive monoclonal antibody (Herrmann et al 1988) allowed for the development of more sensitive enzyme immunoassays (EIAs) for detection of astrovirus in clinical specimens (Herrmann et al 1990). Using these more sensitive methods, astrovirus was found to be associated with 3—9% of cases in several prospective studies of childhood diarrhoea (Herrmann et al 1991, Kotloff et al 1992, reviewed in Glass et al 1996). The determination of the complete nucleotide sequence of two human astrovirus isolates (Jiang et al 1993, Willcocks et al 1994) allowed for the development of highly sensitive reverse transcriptase (RT)-PCR detection assays (Major et al 1992, Jonassen et al 1993, 1995, Mitchell et al 1995). A direct comparison of the detection efficiency of EIA and RT-PCR using samples from an outbreak in a day care setting demonstrated the increased sensitivity of RT-PCR and documented asymptomatic shedding of astrovirus in older children (Mitchell et al 1995).

Serotypes of human astrovirus have historically been defined on the basis of immunoelectron microscopy, immunofluorescence, neutralization assays and type-specific EIA (Lee & Kurtz 1994, Noel et al 1995, Koopmans et al 1998). Type-specific rabbit antisera, generated by Dr John Kurtz and Terry Lee at the Public Health Laboratory in Oxford, UK, have been used as the reference reagents for most of the assay formats (Kurtz & Lee 1984). Seven serotypes of human astrovirus (HAstV1—7) have been fully characterized (Lee & Kurtz 1994), with an eighth serotype suggested on the basis of nucleotide sequence information available in GenBank (Z66541). The independence of HAstV-8 as a distinct serotype was questioned, however, by the observation of partial one-way reactivity with HAstV-4 by using standard immunoelectron microscopy assays (J. Kurtz, personal communication). Studies on the relative prevalence of astrovirus serotypes in various populations have demonstrated that HAstV-1 is the type most commonly detected, while HAstV types 6 and 7 are rarely detected (Lee & Kurtz 1994, Noel & Cubitt 1994, Noel et al 1995).

The astrovirus genome consists of a single-stranded RNA of positive polarity encoding three open reading frames (ORFs) (Jiang et al 1993, Willcocks et al 1994, Matsui & Greenberg 1996, Matsui etal 2001, this volume). ORFs 1a and 1b encode proteins involved in virus replication and their expression involves generation of an ORFla—lb fusion protein via a ribosomal frameshift (Marczinke et al 1994, Lewis & Matsui 1996). ORF2 encodes the capsid protein precursor (Monroe et al 1993) which is processed to the mature capsid proteins by a pathway that remains to be rigorously established (Sanchez-Fauquier et al 1994, Bass & Qui 2000). All serotypes contain at least three capsid proteins, Pl, P2 and P3, with the P2 protein encoding group-reactive epitopes, and the P3 protein encoding serotype-specific epitopes (Belliot et al 1997a).

With the development of RT-PCR assays for detecting astroviruses came the potential for typing strains by comparison of nucleotide sequence information. The usefulness of this approach was demonstrated when genetic types inferred by phylogenetic analysis of sequences from a region within ORF2 were shown to correlate precisely with antigenic types determined by type-specific EIA (Noel et al 1995). Surprisingly, a similar analysis of sequences from within ORF1a resulted in a different grouping of strains (Belliot et al 1997b). Strains from types 1—5 clustered in one distinct group, termed genogroup A, while those from types 6 and 7 clustered in genogroup B. While it was possible to clearly separate strains into these two well resolved genogroups by using either analysis of nucleotide sequence information or the results of probe hybridization experiments, it was not possible to separate strains within a genogroup into individual serotypes by using these techniques (Belliot et al 1997b).

We have recently participated in characterizing astrovirus strains from two studies of gastroenteritis in young children, one in Egypt (Naficy et al 1999) and another in Malawi (Cunliffe et al 1998). The details of the study designs and the methods for astrovirus detection and characterization will be presented elsewhere (Holmes et al 2001). In brief, astrovirus positive samples identified by group-reactive EIA were genetically typed by analysis of RT-PCR products generated from the P2 region of ORF2 (Noel et al 1995). On the basis of unique sequences of the amplification products, 17 distinct astrovirus strains were identified in the samples from Egypt and 12 in the samples from Malawi (Fig. 1). Interestingly, each collection contained samples clustering with six of the eight serotypes, including the recently described HAstV-8. These results confirm our previous work, and that of more recent studies, which indicated that multiple astrovirus serotypes co-circulate in communities within a few years (Noel et al 1995, Mustafa et al 2000).

We next examined the clustering of strains based upon analysis of sequences of RT-PCR products generated from ORF1a (Belliot et al 1997b). As previously reported, the strains segregated into two well resolved genogroups, A and B (Fig. 2). Consistent with the higher overall similarity in this region of the genome compared to ORF2, several of the strains that were distinguishable on the basis of sequences in ORF2 contained identical sequences in the region of ORF1a analysed. As noted previously, it was not possible to unambiguously

FIG. 1. Relationship of astrovirus strains based upon partial sequences from ORF2. Distances were calculated based on the 348 nucleotide amplification products from ORF2 by using the distances program of the GCG package. The country of origin of each strain is indicated by three letter IS03166 code. All strains of an individual serotype fall within a single lineage, indicated Tl—'T8.

FIG. 1. Relationship of astrovirus strains based upon partial sequences from ORF2. Distances were calculated based on the 348 nucleotide amplification products from ORF2 by using the distances program of the GCG package. The country of origin of each strain is indicated by three letter IS03166 code. All strains of an individual serotype fall within a single lineage, indicated Tl—'T8.

0 0

Post a comment