Species-specific RT-PCR amplification of human enteroviruses: a tool for rapid species identification of uncharacterized enteroviruses

The 65 human enterovirus serotypes (family Picornaviridae) are classified into four species, Human enterovirus (HEV)-A, HEV-B, HEV-C and HEV-D, largely on the basis of phylogenetic analysis of multiple genome regions (King et al., 2000). Analysis of the complete genome sequences of all the prototype strains has shown that the species remain phylogenetically coherent in all genome regions except in the 5'-non-translated region (NTR) (Brown et al., 2003; Oberste et al., 2004a, d). Seven of the 13 simian enterovirus prototype strains are provisional members of two HEV species, with six strains in HEV-A and one in HEV-B (Oberste et al., 2002). The remaining six strains are proposed to define three new species in the genus Enterovirus (Oberste et al., 2002), while the bovine and porcine enteroviruses comprise two additional species (King et al., 2000). Recombination, a major mechanism of enterovirus evolution, has been observed to occur in nature only among members of the same species, except in the 5'-NTR (Brown et al., 2003; Lindberg et al., 2003; Lukashev et al., 2003, 2004; Oberste et al., 2004a, d, e; Oprisan et al., 2002). Species identity therefore appears to have some biological significance integral to virus survival.

Genus-specific primers are widely used for the amplification of enteroviruses from clinical specimens by RT-PCR (Rotbart & Romero, 1995). Serotype-specific primers have also been designed for the identification of epidemiologically related enterovirus isolates during an outbreak (Huang et al., 2003; Kilpatrick et al., 1998, 2001; Mullins et al., 2004; Oberste et al., 2003a). In certain cases, it may be desirable to screen a large collection of enterovirus isolates for viruses of a given species to, for example, examine serotypes related to polioviruses within HEV-C. We present here a set of four RT-PCR primer pairs that may be used to rapidly identify the species of HEV isolates, allowing the screening of relatively large collections of isolates without the need for more detailed initial characterization of the isolates.

Primer design.
To design species-specific RT-PCR primers, 3'-NTR sequences of the HEV prototype strains were aligned using PILEUP (Wisconsin sequence analysis package, version 10.3; Accelrys). Primers were chosen in regions that are highly conserved among viruses of the same species but divergent among viruses of different species as described in Results (see Fig. 1).

(42K): Fig. 1. Conservation of 3'-NTR sequences and structures within an enterovirus species and location of species-specific PCR primers. (a) Intraspecies consensus sequences. For each species, the majority residue at a given site is shown on the top line with variant residues shown below. The absence of a residue (alignment gap) is indicated by a hyphen and termination codons are underlined. (b) Predicted secondary and tertiary structures, based on the published structures of CVA2 (HEV-A), CVA9 (HEV-B), PV1 (HEV-C) and EV70 (HEV-D) (Mirmomeni et al., 1997; Pilipenko et al., 1992a). Stemloops X, Y and Z are indicated and termination codons are underlined. Sites of sequence variation are indicated by arrows: upper case, base pairing preserved; lower case, base pairing disrupted; Δ, base deleted. Boxed regions indicate alternative structures within a species. The portion of the poly(A) tract that is not part of stem X is denoted as An. (c) Alignment of 3'-NTR consensus sequences for HEV-A to HEV-D, from panel (a). The termination codons are boxed; dots indicate alignment gaps. The location and orientation of primers 480483 and 423426 are shown above the sequences.

Viruses.
To test the species-specific RT-PCR primers, three panels of enterovirus isolates were assembled. The first panel consisted of reference strains of each of the 65 recognized human and 13 simian enterovirus serotypes (Oberste et al., 2002). The proposed prototype strains for enteroviruses numbered beginning at EV73 were not included in this panel because they have not yet been fully characterized. The second panel contained 115 human clinical isolates of known serotype, which have been previously characterized in our laboratory, and it included 32 enterovirus serotypes of all four species and three human rhinovirus serotypes (Oberste et al., 1999a, 2000, 2004b). The third panel was composed of viruses isolated from patients with acute flaccid paralysis that were identified during virological surveillance activities in Bangladesh in support of the global polio eradication initiative. These isolates were randomly selected from the Bangladesh non-polio enteroviruses isolated between August 1999 and January 2002. The isolates were chosen from among those that were positive for enterovirus, as shown by RT-PCR using pan-enterovirus primers (Yang et al., 1992), and negative for poliovirus, as shown by RT-PCR using pan-poliovirus primers (Kilpatrick et al., 1996), in the initial screening.

RT-PCR.
For RT-PCR analysis, RNA was extracted from 140 µl of virus-infected cell culture supernatant by using the QIAamp viral RNA mini kit (Qiagen) according to the manufacturer's recommended procedure, and eluted with 60 µl sterile water. Each RT-PCR reaction contained 1 µl RNA, 67 mM Tris/HCl (pH 8·8), 17 mM ammonium sulfate, 6 µM EDTA, 2 mM MgCl₂, 200 µM each dNTP, 1 mM dithiothreitol, 1 µM each primer, 10 U placental RNase inhibitor (Roche Applied Science), 3 U avian myeloblastosis virus (AMV) reverse transcriptase (Roche Applied Science) and 2·5 U Taq DNA polymerase (Roche Applied Science) in a total volume of 50 µl. The reactions were incubated at 50 °C for 30 min followed by 94 °C for 3 min. Thermocycling was performed for 35 cycles at 94 °C for 30 s, 42 °C for 30 s and 72 °C for 30 s in a 9700 model thermocycler (Applied Biosystems). Thermocycling was followed by incubation at 72 °C for 5 min. The reaction products were analysed by electrophoresis in a 10 % polyacrylamide/TBE gel and stained with 0·5 µg ethidium bromide ml¹.

Virus identification.
Virus serotypes were identified by analysis of the VP1 gene by sequencing the amplicon product using primers 292/222 (Table 1) as previously described (Oberste et al., 2003b). In a few cases where the 292/222 sequence was of poor quality, or to resolve the components of mixtures, additional primers (486497) were used (Table 1). The serotype of each isolate was determined by pairwise comparison of the VP1 amplicon sequence to a database of complete VP1 sequences of all enterovirus serotypes (Oberste et al., 1999b). We have previously shown that strains which were at least 75 % identical in VP1 sequence belong to the same serotype, whereas those that were less than 70 % identical to one another belong to different serotypes (Oberste et al., 1999a, b, 2000, 2001). When all recognized serotypes were included in the database search, a high score of less than 70 % indicates that the sequence of the isolate does not match the serotype of any sequence in the database; therefore it is potentially a new serotype (Oberste et al., 2000, 2001). Such viruses are classified as untyped, pending further characterization. In this study, the complete VP1 sequence was determined using primers 486497 (Table 1) for all isolates whose partial VP1 sequence was less than 70 % identical to those of all recognized serotypes. For these isolates, VP1 was amplified in two fragments using primer pairs 486/488 and 487/489 (HEV-A), 490/492 and 491/493 (HEV-B), or 494/496 and 495/497 (HEV-C). Sequences of the two fragments were assembled, and the complete VP1 sequences were compared with sequences in the VP1 sequence database as described above to confirm that they were less than 70 % identical to those of known serotypes.

Table 1. Primers used to screen enterovirus isolates by species and for identification of enterovirus isolates by VP1 sequencing

Plaque-purification of components of mixed isolates.
For mixed isolates whose components could not be resolved using primers 486497, the components were separated by plaque-purification. Briefly, human rhabdomyosarcoma (RD) cell monolayers in six-well plates were infected with virus by incubation at 37 °C for 1 h with rocking. The supernatant was removed and replaced with 0·6 % SeaPlaque agarose (Cambrex Bioscience) in minimal essential medium and the plates were incubated at 37 °C with 5 % CO₂. After 48 h, the plaques were visualized by adding 1 ml of 2 % neutral red to each well. Following incubation at 37 °C for 2 h, 10 plaques were harvested and the individual virus stocks were prepared by inoculation of RD cells in 24-well plates. The serotypes of the plaque-purified viruses were determined by PCR amplification and sequencing with primers 292/222 as described above.

3'-NTR sequence analysis.
For isolates that were not amplified by any of the species-specific primers (pairs 480/423, 481/424, 482/425 and 483/426; Table 1), the 3'-NTR was sequenced to determine the degree of primertemplate mismatch. The 3'-NTR sequences were determined using primer 784 (5'-ATYCAYCCRACCATGCCAATG-3', EV76 nt 71207140) and the 3'-RACE method (Roche Applied Science). To provide a standard of comparison, the 3'-NTR sequences of the HEV-A clinical isolates were determined using primer 233 (5'-TTGAYTACWCWGGNTATGATGC-3', PV1 nt 66816702) and 3'-RACE. The newly derived sequences were aligned with the homologous sequences of the HEV-A prototype strains by using PILEUP.

Nucleotide sequence accession numbers.
The sequences described here have been deposited in the GenBank sequence database, accession nos AY919400AY919592 and AY923167AY923169 (Supplementary Table S1 available in JGV Online).

3'-NTR conservation
The 3'-NTR was targeted for design of species-specific RT-PCR primers because it was observed from the analysis of complete genome sequences of all serotypes of HEVs that the 3'-NTR sequences are well conserved within an enterovirus species (7099 % identity) but highly divergent between species (<62 % identity) (Brown et al., 2003; Oberste et al., 2004a, b, d). The 3'-NTR sequences of the prototype strains of each of the four HEV species, HEV-A to HEV-D, were aligned by species, and conserved regions were identified within each species (Fig. 1a). To determine whether the predicted structures were phylogenetically conserved when all members of a species are considered, sequence variation was mapped onto the 3'-NTR predicted secondary and tertiary structures of each species on the basis of structures proposed by Mirmomeni et al. (1997). The predicted RNA secondary and tertiary structures are highly conserved within each species with three predicted stems (X, Y and Z) in all members of HEV-A and HEV-B, and only two (X and Y) in HEV-C and HEV-D viruses. In HEV-A, there are two alternative sequences/structures in the Z stemloop structure: one is fully conserved among CVA4, CVA14 and CVA16 and the other is fully conserved among all other members of HEV-A (Fig. 1b). Only three paired bases (two in stem Y and one in stem Z) are disrupted by sequence variation in the HEV-A 3'-NTR. The HEV-B 3'-NTR is the most highly variable, probably because of the greater number of sequences, but sequence variation disrupts only three paired bases (two in stem Z and one in stem Y) (Fig. 1b). The HEV-C 3'-NTR structure is fully conserved, as all sequence variations maintain the predicted structure or are compensated by changes in the complementary base (Fig. 1b). The predicted 3'-NTR structure is also conserved in HEV-D except that EV70 contains four additional stacked base pairs at the top of stemloop Y that are absent from EV68 (Fig. 1b).

Primer design and validation
Primers were chosen in regions that are highly conserved among members in a given species and yet unique relative to the other species (Fig. 1c and Table 1). The reverse-strand primers anneal at homologous sites in each species alignment, immediately upstream of the poly(A) tract, and the forward-strand primers anneal at upstream sites in the 3'-NTR. The resulting amplicons have predicted sizes of 61 bp (HEV-A), 64 bp (HEV-B), 59 bp (HEV-C) and 71 or 82 bp (HEV-D) (Fig. 1c). The difference in product sizes for members of HEV-D is due to an 11 nt deletion in the EV68 3'-NTR relative to that of EV70 (Fig. 1).

For initial validation of the primers, RNA extracted from each of the 65 HEV prototype strains was used as template in RT-PCRs with each of the four primer pairs (Supplementary Table S2). As expected from analysis of the prototype strain 3'-NTR sequences (Fig. 1), the species-specific primers amplified a product from all members of the homologous species but no product was amplified using template RNA from viruses of any other HEV species (Supplementary Table S2). Thirteen simian enterovirus strains were also included in the initial primer characterization to determine whether the primers could discriminate between the human and simian enteroviruses. None of the simian enteroviruses were amplified with any of the species-specific primers (Supplementary Table S2) despite their close relationship to the HEVs when their VP1 capsid-coding sequences were compared (Oberste et al., 2002).

To ensure that the results in Supplementary Table S2 were not simply artefacts of testing the isolates from which the primers were designed, a panel of 115 viruses with unknown 3'-NTR sequences was assembled from isolates that have been previously characterized by partial sequencing of the VP1 gene (Oberste et al., 1999a, 2000, 2004b). This panel represents 32 enterovirus serotypes of all four species plus three human rhinovirus serotypes. Each species-specific PCR primer pair amplified a product of the correct size from all viruses of the cognate species but failed to produce a product from template RNA derived from viruses of other species (Supplementary Table S3). No product was amplified from the rhinovirus templates with any of the four species-specific PCR primer pairs (Supplementary Table S3).

Screening of a collection of isolates
To validate further the specificity of the species-specific primer pairs, a panel of unknowns consisting of 186 enterovirus isolates, which have been pre-screened to exclude polioviruses, was screened with each of the four species-specific primer sets (Table 2 and Supplementary Table S1). For 166 isolates (89·2 %), an amplification product was produced with only one of the four primer sets. Twelve isolates (6·5 %) were positive with more than one set of primers and eight isolates (4·3 %) were negative with all primers. For each sample, the enterovirus serotype was identified by amplification and sequencing of a portion of the VP1 capsid gene, followed by comparison of the amplicon sequence to a database of VP1 sequences for all enterovirus serotypes. In all cases, the species of the identified virus(es) corresponded to the species of the amplification-positive species-specific primer pair(s). Of the 166 singly positive viruses, 127 (76 %) were members of HEV-B, while HEV-A and HEV-C accounted for 11 (18/166) and 13 % (21/166) of the single-species isolates, respectively (Table 2 and Supplementary Table S1). No members of HEV-D were identified in the panel of unknowns. Within each species, a wide range of serotypes was amplified, confirming that each species-specific primer pair has broad intraspecies specificity. For example, 31 of 37 recognized HEV-B serotypes were among those amplified by the HEV-B primers, and the HEV-A primers amplified 9 of 12 HEV-A serotypes. In addition, 28 isolates contained at least one member of a new serotype candidate (EV74 to EV101) from species HEV-A, HEV-B and HEV-C (Table 2, Table 3 and Supplementary Table S1).

Table 2. Amplification of a panel of 186 enterovirus clinical isolates of unknown serotype using species-specific RT-PCR primers Viruses were subsequently identified by partial sequencing of the gene encoding the VP1 capsid protein.

Table 3. Comparison of enterovirus serotypes numbered above EV73 identified in this study to those of existing serotypes (range of VP1 nucleotide sequence percentage identity) Sequence identities to members of the homologous species are indicated in bold.

Each of the isolates that were positive with more than one set of primers was shown by partial VP1 gene sequencing to contain viruses of two (11 isolates) or three (one isolate) different species (Table 2 and Supplementary Table S1). Five isolates contained members of both HEV-A and HEV-B: CVA4 plus E9, CVA4 plus E14, CVA6 plus E3, CVA7 plus E13 and CVA14 plus EV86 (Table 2 and Supplementary Table S1). Six isolates contained mixtures of HEV-B and HEV-C viruses: E9 plus CVA13, E14 plus CVA20, E17 plus CVA11, E25 plus EV96, EV97 plus CVA13 and EV97 plus CVA20 (Table 2 and Supplementary Table S1). One isolate was a mixture of CVA16 (HEV-A), CVB2 (HEV-B) and CVA13 (HEV-C). Three homospecies mixtures were also identified: CVA6 plus EV76 (HEV-A), EV76 plus EV90 (HEV-A) and CVB4 plus E12 (HEV-B) (Table 2 and Supplementary Table S1).

The eight isolates that were not amplified with any of the species-specific primer pairs all belonged to one of four newly described types that have been provisionally classified in HEV-A: EV76, EV89, EV90 and EV91 (Oberste et al., 2005); one of the isolates was a mixture of EV76 and EV90. To determine whether the failure of primers 480/423 to amplify these isolates was due to a mismatch of one or a few bases near the end of the primer(s) or to more extensive sequence divergence, the 3'-NTR sequences were determined for each of the isolates, except the mixture, as well as for 14 previously described isolates of these four types (Oberste et al., 2005). The sequences were compared to one another, to those of the enterovirus prototype strains and to those of the 16 isolates of conventional HEV-A serotypes that were amplified with primers 480/423 (Supplementary Table S1). The 3'-NTR sequences of all of the EV76, EV89, EV90 and EV91 isolates were closely related to one another (>83 % identity) but were not monophyletic with respect to type (Fig. 2a and data not shown). For all isolates of these four types, the 3'-NTR sequence is 94 nt long, not including the termination codon. Alignment of these sequences with those of the enterovirus prototype strains showed that they were highly diverged from members of all enterovirus species (<35 % identity), including the simian enteroviruses (Fig. 2b), whereas those of the conventional Bangladesh HEV-A isolates were 7094 % identical to those of the HEV-A prototype strains. The EV76, EV89, EV90 and EV91 3'-NTR sequences are highly divergent from those of the other HEV-A viruses (<63 % sequence identity), so that the species-specific PCR primers (480483 and 423426) are predicted to anneal poorly or not at all, due to the large number of mismatches (Fig. 2b).

(23K): Fig. 2. 3'-NTR sequences of isolates that were not amplified by any of the four species primer sets. (a) Alignment of the 3'-NTR sequences of isolates of EV76, EV89, EV90 and EV91, including isolates that have been previously described (Oberste et al., 2005). The majority consensus line (top) indicates the majority residue at each position. For individual isolates, only the sites that differ from the majority consensus are shown. The strict consensus line (bottom) indicates the range of diversity at each site and uses standard IUB nucleotide ambiguity codes. Termination codons are underlined. (b) Comparison of the EV76, EV89, EV90 and EV91 3'-NTR consensus sequence and 3'-NTR sequences of the simian enteroviruses baboon enterovirus (SV-BaEV), SV19, SV43 and SV46 to the consensus of the human viruses in HEV-A (from Fig. 1a). Alignment gaps are indicated by a hyphen. Residues that are identical to the HEV-A consensus, taking into account nucleotide ambiguity characters, are indicated by a dot. Termination codons are boxed. The positions of primers 480 and 423 are also indicated.

To enable detection of EV76, EV89, EV90 and EV91 by 3'-NTR RT-PCR, an additional primer pair was designed (primer 1133, 5'-GCWAHAYCTTTGGTGYACCCTG-3', EV76 nt 73717392, and primer 1134, 5'-GGGTTAGGGTAATTRAYCTTTGG-3', EV76 nt 74307408), based on the sequences in Fig. 2(a). When used under the same conditions as the other species-specific primer pairs, primers 1133/1134 amplified the expected 60 bp product from all 21 isolates listed in Fig. 2(a), but did not amplify a product when RNAs from the other HEV-A serotypes were used as template (data not shown). The enterovirus 3'-NTR, the site of initiation of negative-strand RNA synthesis, is required for efficient genome replication (Brown et al., 2004; Mirmomeni et al., 1997; Rohll et al., 1995). While 3'-NTR sequences vary widely among the various enterovirus species (Brown et al., 2003; Oberste et al., 2004a, b, d) the existence of highly conserved secondary structures suggests that these structures are the functional unit(s) involved in replication (Mirmomeni et al., 1997; Pilipenko et al., 1992b, 1996). The predicted structures consist of three stemloops termed X, Y and Z in HEV-A and HEV-B and two stemloops (X and Y) in HEV-C and HEV-D (Mirmomeni et al., 1997). X and Y form a tertiary structure through a so-called kissing interaction of their loop residues (Melchers et al., 1997; Mirmomeni et al., 1997; Pilipenko et al., 1992b). The Z domain is apparently dispensable for replication in culture but may play a role in viral pathogenesis in vivo (Merkle et al., 2002). This grouping of species with regard to the number of 3'-NTR structural domains parallels the species groupings observed in the sequences of the 5'-NTR (Pöyry et al., 1996). The 3'-NTR structures are thought to mediate the binding to viral RNA of viral and cellular proteins that are essential for replication (Harris et al., 1994; Mellits et al., 1998). Mapping of the observed intraspecies sequence variation onto the 3'-NTR structures of the four HEV species shows that most of the sequence differences should maintain the predicted base-pairing structures (Fig. 1b). Many of the potentially disruptive sequence differences are compensated by a change in the complementary base. Phylogenetic conservation of the proposed structures lends further support to their importance in the virus life cycle.

Conservation of the 3'-NTR, resulting from its essential role in virus replication, has allowed us to design species-specific PCR primers to screen collections of isolates. In our example, the five most frequently identified serotypes in the collection of 186 enterovirus isolates from Bangladesh were CVB4 (12 isolates, including a CVB4-E12 mixture), E12 (11 isolates, including the CVB4-E12 mixture), E11 (eight isolates), E26 (seven isolates) and E13 (five isolates). CVB4 and E11 are often among the most commonly reported enteroviruses in the USA, but E12 and E26 are not commonly reported (Centers for Disease Control & Prevention, 1997, 2000, 2002; Strikas et al., 1986). These two serotypes are also uncommon elsewhere: only 16 E12 and no E26 isolates were reported in Japan from 1982 to 1999, out of over 60 000 total enterovirus isolates (Infectious Disease Surveillance Center, 2003). Throughout the world, E13 was rarely reported until it emerged as a major cause of aseptic meningitis outbreaks in 20002001 in Europe, the USA and Japan (Centers for Disease Control & Prevention, 2001; Communicable Disease Surveillance Centre, 2000; Keino et al., 2001; Twisselmann, 2000). To the best of our knowledge, the EV69 isolates in this report represent the only noted description since the isolation of the prototype strain in 1959 (Melnick et al., 1974). The high frequency of HEV-B serotypes, relative to those of HEV-A and HEV-C, is also consistent with published enterovirus surveillance reports. It is significant that no members of HEV-D were identified in the screening of the 186 isolates of unknown serotype. All of the viruses in this panel were isolated from stool specimens. We have recently shown that EV68 is associated exclusively with respiratory disease and that it has been isolated only from respiratory tract specimens (Oberste et al., 2004b). It has also been shown that human rhinovirus 87 (HRV-87) and EV68 are independent isolates of the same serotype (Blomqvist et al., 2002; Ishiko et al., 2002). Similarly, EV70, the only other known member of HEV-D, has only rarely been isolated from faecal specimens (Nakazono & Kondo, 1989). Therefore, it may be that the failure to detect HEV-D is related more to the specimen type tested in this study rather than a reflection of the prevalence of infection in the population.

Mixed enterovirus infections, and the resulting mixed virus cultures containing enteroviruses of two or more serotypes, have long presented a challenge to enterovirus identification (Hara et al., 1968). In the traditional neutralization assay, mixed cultures present as breakthrough cytopathic effect in the presence of antibody to one of the mixture components. Such cytopathology is usually indistinguishable from virus growth in the presence of irrelevant antisera (i.e. antibodies specific for an enterovirus serotype that is not present in the mixture). With the recently developed molecular serotyping methods (Oberste et al., 1999a, 2000, 2003b) mixtures may appear to be single isolates if one serotype is present in much higher titre than the other(s). If the different serotypes are of approximately equal titre, the sequence is usually of poor quality and uninterpretable, providing only a non-specific clue that the culture may contain a mixture of serotypes. Only after laborious purification by the production of plaques or limiting dilution is it possible to resolve the mixture and identify its components. The species-specific RT-PCR provides a simple method to identify hetero-specific mixed cultures, which comprised approximately 6 % of the samples screened in this study (Table 2). Components of these mixtures were easily resolved by sequencing using primers that are specific for sequence motifs that are highly conserved within species but divergent between species (Table 1). Homo-specific mixtures still pose a potential problem, but the use of multiple VP1 primer sets may be helpful in these cases. In this study, for example, three homo-specific mixtures were resolved using additional primers after 292/222 produced poor quality sequence (Table 2). Because of the difficulty in resolving homospecies mixtures, particularly when one component is present at much higher titre than the other, the number of such mixtures will tend to be underestimated by this approach.

Molecular approaches to enterovirus typing using VP1 sequence as a surrogate for neutralizing epitopes have recently identified numerous new enterovirus types (Norder et al., 2002, 2003; Oberste et al., 2000, 2001, 2004c, 2005) (M. S. Oberste, unpublished data). This study identified 29 isolates (16 % of isolates, including components of mixtures) of 15 new types beyond those serotypes that were originally identified by antigenic analysis alone. Several of the new types found in this screening (EV7376 and EV8991) have recently been described in the literature (Norder et al., 2003; Oberste et al., 2000, 2001, 2004c, 2005). The new types are distributed among three species, HEV-A (four types), HEV-B (10 types) and HEV-C (one type), demonstrating further the wide diversity of enteroviruses that may be isolated in a given geographical area.

Four of the new types (EV76, EV89, EV90 and EV91) appear to be members of HEV-A based on the phylogenetic analysis of VP1 or complete capsid sequences, but they constitute a distinct subgroup from other members of HEV-A, including the simian enteroviruses, when other genome regions (e.g. the 3D gene) are analysed (Oberste et al., 2005). While their 3D sequences are clearly related to those of members of HEV-A (Oberste et al., 2005), the region immediately downstream (the 3'-NTR) is unique (Fig. 2), suggesting that EV76, EV89, EV90 and EV91 comprise a novel group within HEV-A. This classification is analogous to that of six simian enteroviruses (baboon enterovirus A13, SV19, SV26, SV35, SV43 and SV46; SV19/26/35 represent a single serotype) in that these four simian serotypes are clearly related to members of HEV-A in VP1 sequence (Oberste et al., 2002) but distinct in 3D and in the 3'-NTR (Fig. 2). The biological basis for these distinctions remains unknown.

All four 3'-NTR primer sets are highly specific since they amplified only viruses of the correct species. The primers that are specific for HEV-B, HEV-C and HEV-D are also highly sensitive, amplifying all isolates of their respective species. While only included explicitly in this study as part of the testing of prototype strains, the polioviruses have always amplified with the HEV-C-specific primers in our laboratory, consistent with the proposal that they could be reclassified as members of HEV-C (Brown et al., 2003). While the HEV-A-specific primers amplified all recognized members of HEV-A, they failed to amplify EV76, EV89, EV90 and EV91 isolates and the simian enteroviruses that appear to be members of HEV-A. Additional primer sets will be developed and evaluated for these isolates as well as the other species in the genus Enterovirus as sequence data on multiple isolates become available. Further studies are needed to determine whether these types should be considered as divergent members of HEV-A, possibly as a distinct subspecies, or whether they represent two additional enterovirus species. In either case, the primers described here provide a convenient reagent to distinguish between conventional HEV-A serotypes and those of these two unique groups.

While it may be possible to adapt this method to the detection of enteroviruses from clinical specimens to narrow their identification, we anticipate that the primary application will be to use this method for the screening of large collections of enterovirus isolates like those that may exist in state or national virology laboratories, which isolate a large number of viruses from faecal specimens or from aseptic meningitis patients. Such screening could identify the enterovirus species present, thereby directing the choice of primers to be used in the subsequent characterization.

We are grateful to the Bangladesh staff of the Expanded Programme on Immunization, WHO, and the Bangladesh national polio eradication programme for investigating cases of acute flaccid paralysis and obtaining stool specimens. We also thank Dr Nalini Withana, WHO/SEARO, for helping to facilitate our collaboration.