Molecular typing of enteroviruses associated with viral meningitis in Cyprus, 2000-2002

Non-polio human enteroviruses (HEVs) belonging to the Picornaviridae family represent the most common cause of aseptic meningitis, which, conversely, is the most commonly identified illness in association with enterovirus infections (Rotbart, 1995). HEVs are ubiquitous faecal-orally transmitted small RNA viruses with a seasonal peak of infection in summer and autumn. Outbreaks of disease caused by a single serotype strain are frequently reported and represent a major public health problem (Ozkaya et al., 2003; Avellon et al., 2003; Thoelen et al., 2003).

The typing of enteroviruses is based conventionally on virus isolation in cell culture, followed by neutralization with intersecting pools of type-specific antisera and confirmation with monospecific antisera to identify the serotype. Due to the large number of antigenically distinct serotypes, serotyping procedures are time-consuming, labour-intensive and costly. Moreover, the limited supply of reference sera and their inability to detect new antigenic variants or emerging serotypes are major drawbacks of neutralization typing (el Sageyer et al., 1998).

Because VP1 is the major surface-accessible protein in the picorna virion that exhibits relevant serotype-specific neutralization epitopes, the sequence of the VP1-encoding gene has been shown to correlate well with the classical serotype classification (Oberste et al., 1999, 2003). However, assays utilizing highly degenerated primers to amplify a portion of the VP1 gene in all HEV serotypes usually require cell culture prior to extraction due to a lack of sensitivity (Oberste et al., 1999, 2003; Caro et al., 2001).

Since sample material is often used up completely for RNA and DNA extractions, for diagnostic purposes the typing of HEVs utilizing cell culture is often not possible. For this reason a typing strategy based on the sequencing of a large part of the 5' untranslated region and amplification of a part of the VP1 region was developed.

In this study, samples from 218 patients with viral meningitis, who were diagnosed enterovirus positive between January 2000 and December 2002, have been analysed.

Data obtained by sequencing a 399 bp fragment in the 5'NCR of the enterovirus genome, together with information about currently circulating HEV strains, were used to design serotype-specific primers amplifying a portion of the VP1 gene directly from clinical specimens in order to unambiguously identify the serotype.

Patients and clinical samples. Between February 2000 and December 2002 cerebrospinal fluid (CSF) and/or stool samples from 298 patients with suspected viral meningitis from all parts of Cyprus were sent to our laboratory for analysis. The median patient age was 6 years, ranging from one month to 77 years, with a male : female ratio of 1.5 (Fig. 1). For routine laboratory diagnosis, an in-house nested RT-PCR assay targeting a sequence in the highly conserved 5'NCR was employed. One hundred and sixty-eight of 228 CSF samples (74 %) and 50 of 70 stool samples (71 %) were found to be enterovirus positive. Subsequently, these 218 positive samples were subjected to molecular typing for epidemiological surveillance.

(14K): Fig. 1. Age distribution of patients diagnosed with enteroviral meningitis in Cyprus between January 2000 and December 2002.

Extraction and 5'NCR nested RT-PCR. Stool samples were treated as previously described (World Health Organization, 2004). Viral RNA was extracted from 140 µl of clinical specimen (CSF or stool) by the QIAamp Viral RNA Mini kit (Qiagen), according to the manufacturer's instructions. For reverse transcription, 10 pmol primer HEV 7 (see Table 1) and 8 µl 2.5 mM dNTP mix were added to 5 µl RNA. The mixture was heated to 70 °C for 5 min and subsequently snap-cooled on ice for 1 min. To each sample were added 5 µl of a reaction mixture containing 1x Taq buffer, 20 U ribonuclease inhibitor and 50 U M-MuLV reverse transcriptase (New England BioLabs). cDNA synthesis was performed for 30 min at 42 °C. cDNA product (2.5 µl) was added to 22.5 µl of a PCR mixture containing 1x RT buffer, 5 nmol dNTPs, 5 pmol primers HEV 2 and HEV 7, and 2.5 U AmpliTaq Gold DNA polymerase (Applied Biosystems). Amplification involved 30 cycles of denaturation at 95 °C for 30 s, annealing at 42 °C for 45 s and elongation at 72 °C for 1 min. For nested PCR, 2.5 µl of first PCR product was used in an equivalent assay with primers HEV 5 and HEV 8. The 399 bp products were visualized in a 1.5 % agarose gel containing 1 µg ethidium bromide ml¹.

Table 1. Oligonucleotide primers used for 5'NCR nested RT-PCR and VP1 RT-PCR The positions of the primers in the 5'NCR are relative to the genome of Poliovirus 1 Mahoney. The positions of the primers in the VP1 region are relative to the genomes of Human echovirus 30 strain Bastianni (AF162711), Human echovirus 13 strain Del Carmen (AY302539), Human echovirus 9 isolate DM (AF524867), Human echovirus 6 lytic strain (U16283), Human echovirus 4 (AY302557), HEV71 polyprotein gene (U22522), CVA6 strain Gdula (AY421764), CVB1 (M16560), CVB2 strain Ohio (AF081485) and CVB5 (X67706), respectively.

Sequencing. Amplicons were purified using the Montage DNA Gel Extraction kit (Millipore) and sequenced on a CEQ 8000 sequencer (Beckman Coulter) in the forward and reverse directions using the respective PCR primers and the CEQ Dye Terminator Cycle Sequencing with Quick Start kit (Beckman Coulter).

Sensitivity. The sensitivity and specificity of the 5'NCR nested RT-PCR assay was verified using the 2004 Quality Control for Molecular Diagnostics (QCMD) HEV proficiency panel. The panel consisted of 12 lyophilized samples, of which nine were HEV positive, one was Human rhinovirus 16 (RV16) positive and two were negative. The positive samples contained various dilutions of tissue-culture-grown heat-inactivated viruses in a virus transport medium or a PBS/fetal calf serum matrix, the lowest concentration being 0.03 tissue-culture infective dose 50 (TCID₅₀) ml¹. All HEV-positive samples were successfully amplified, while the rhinovirus and negative samples did not yield a product.

Amplification of VP1-encoding sequences. Primers in the VP1 region of relevant enterovirus serotypes were developed based on prototype strain sequences available from GenBank. The selected serotype-specific primers were tested for heterologous priming with BLAST and are listed in Table 1. A VP1 one-tube two-step RT-PCR was carried out as follows: 10 pmol antisense primer and 4 µl 2.5 mM dNTP mix were added to 2.5 µl viral RNA template. The mixture was heated to 70 °C for 5 min and subsequently snap-cooled on ice for 1 min. To each sample were added 2.5 µl of a reaction mixture containing 1x Taq buffer, 20 U ribonuclease inhibitor and 25 U M-MuLV reverse transcriptase (New England BioLabs). cDNA synthesis was performed for 30 min at 42 °C. After completing cDNA synthesis, to each tube were added 40 µl of a PCR mixture containing 1x Taq buffer, 10 pmol of the respective sense primer and 2.5 U AmpliTaq Gold DNA polymerase (Applied Biosystems). Amplification involved 40 cycles of denaturation at 95 °C for 30 s, annealing at 53 °C for 45 s and elongation at 72 °C for 1 min. The PCR products were visualized in a 1.5 % agarose gel containing 1 µg ethidium bromide ml¹. Purification and sequencing of the VP1 amplicons were carried out as described above.

Phylogenetic analysis. Multiple sequence alignments were built with CLUSTALW. Phylogenetic analyses were conducted using MEGA version 2.1 (Kumar et al., 2001). Genetic distances were calculated using the Kimura two-parameter model with a transition : transversion ratio of 2. The phylogenetic tree was constructed using the neighbour-joining method. Statistical significance of phylogenies was estimated by bootstrap analysis with 500 pseudoreplicates. For reasons of clarity, interior branches representing distinct clusters were collapsed into elongated triangles, whose thickness was proportional to the number of taxa condensed and whose width was proportional to the maximum distance between the taxa.

Two hundred and eighteen CSF and/or stool samples collected over a period of 3 years from patients diagnosed with enteroviral meningitis were analysed for epidemiological surveillance. A nested RT-PCR was employed to amplify a 399 bp fragment from the highly conserved 5'NCR. A phylogenetic tree was constructed using the neighbour-joining method to group the samples for further characterization (see Fig. 2a). Representative prototype sequences were included in the alignment. The 5'NCR tree revealed 12 clusters. Since it had been reported previously that clustering in the 5'NCR tree is concordant with clustering in the VP1 phylogenetic tree (Thoelen et al., 2003, 2004), it was assumed that samples belonging to one cluster represented the same enterovirus type. Consensus sequences for each cluster were determined and subjected to a BLAST search against GenBank.

(12K): Fig. 2. Phylogenetic trees based on (a) the partial 5'NCR sequences and (b) the partial VP1 sequences of the 218 clinical isolates. Trees were constructed using the neighbour-joining method and were evaluated with 500 bootstrap pseudoreplicates; only bootstrap values over 75 % are shown. Poliovirus 1 Mahoney was used as outgroup. Interior branches representing distinct clusters are shown as triangles. Scale bars show the number of substitutions per nucleotide position, i.e. 0.1 represents 10 % nucleotide differences.

Based on these results, primers amplifying a part of the VP1 region of potential enterovirus serotypes were designed using prototype enterovirus strain sequences available from GenBank. One-step RT-PCR was then performed in the VP1 region to verify the serotype of each cluster, starting with the highest match of the 5'NCR sequencing result. For six clusters, the serotype could already be determined in this manner after one RT-PCR [Human echovirus 4, Human echovirus 9, Human echovirus 6 strain A, Human echovirus 30 strain A, HEV71 and Human coxsackievirus (CV)A6]; three further clusters yielded a positive result after testing them for the second-highest identity score (CVB1, CVB2 and CVB5). However, the differences between the highest and second-highest matches were in all three cases only minor, i.e. 1 or 2 nucleotides, thus making a clear decision difficult. For the remaining three clusters, however, a positive result was not obtained at this stage. Since two of these clusters contained a significantly large number of samples, it was decided to probe them for the predominant serotypes found in Europe during 2000 and 2001. Human echovirus 30 was one of the enteroviruses isolated most frequently, causing outbreaks of aseptic meningitis in France, Iceland, Kosovo and the Netherlands in 2000, and in England and Wales, Scotland, Ireland and Germany in 2001 (Noah & Reid, 2002; Communicable Disease Surveillance Centre, 2001; Bernit et al., 2004). Human echovirus 13, previously regarded a rare enterovirus serotype, caused outbreaks in England and Wales, Scotland, Ireland, Germany, France and the Netherlands in 2000 (Chomel et al., 2003; Noah & Reid, 2002). In an epidemiological study of an outbreak of aseptic meningitis in the summer of 2000 in Belgium, human echoviruses 30, 13 and 6 were identified most frequently during the epidemic (Thoelen et al., 2003).

Using primers in the VP1 region for human echoviruses 30 and 13, two more clusters could be identified (Human echovirus 13 and Human echovirus 30 strain B). The last remaining cluster was then tried for the next highest matches of the 5'NCR BLAST result. All VP1 amplicons generated were sequenced and BLAST analysis was carried out. Each cluster was assigned the serotype that yielded the highest identity score, as proposed by Oberste et al. (1999). A phylogenetic tree was constructed on the basis of a nucleotide alignment of all VP1 fragments using the neighbour-joining method, including relevant reference strains from GenBank (Fig. 2b). Comparing the clustering of samples in the 5'NCR and VP1 trees, it is noticeable that some serotypes in the 5'NCR tree (i.e. Human echovirus 6 strain A and Human echovirus 30 strain B) did not cluster monophyletically as they did in the VP1 tree. However, samples forming a distinct subcluster in the 5'NCR tree were found to be of the same serotype by VP1 typing.

Table 2 summarizes the distribution of serotypes identified in Cyprus from 2000 to 2002. Ten different enterovirus serotypes were identified, with human echoviruses 30, 13, 6 and 9 being the most prevalent enteroviruses encountered. CVB2, B1 and B5, Human echovirus 4, HEV71 and CVA6 represented rather rare serotypes isolated in this study. In Fig. 3 is depicted the temporal distribution of the four most prevalent serotypes mentioned above. In spring 2000, Human echovirus 30 was identified most frequently as the cause of aseptic meningitis, but its incidence declined rapidly in summer as it was superseded by Human echovirus 13, which dominated until the end of November 2000. In 2001, another outbreak of Human echovirus 30 was observed, which lasted from April to September with some sporadic cases of Human echovirus 6 in May and June. Human echovirus 6 then became the predominant enterovirus serotype from October 2001 to February 2002, when it was superseded by Human echovirus 9, which was isolated until July 2002. From Fig. 3 it is notable that the different virus strains caused temporally separated outbreaks that succeeded each other, having a mean duration of about 5 months.

Table 2. Enterovirus types identified from 218 samples in Cyprus between 2000 and 2002

(22K): Fig. 3. Temporal distribution of the four most prevalent serotypes isolated in Cyprus between 2000 and 2002.

Recently, many RT-PCR assays have been developed for the molecular typing of enteroviruses (Ishiko et al., 2002; Caro et al., 2001; Santti et al., 1999a; Oberste et al., 2003). In the past, different regions of the enteroviral genome have been evaluated for the correlation of genotype with serotype, and it has been shown that only the VP1 region fulfils the criterion of absolute correlation (Oberste et al., 1999; Casas et al., 2001; Caro et al., 2001). The VP1 protein is immunodominant among the enterovirus capsid proteins (Mateu, 1995; Minor, 1990). In particular, the BC loop, which covers amino acids 8293 of the 292 amino acid VP1 protein, has been shown to be important for the reactivity of type-specific antibodies (Norder et al., 2001; Reimann et al., 1991). In this study, data obtained by sequencing a 399 bp fragment in the 5'NCR, together with information about currently circulating HEV strains, were used to design serotype-specific primers for the VP1 region of the most probable serotypes. The partial VP1 sequences amplified encompassed entirely the hypervariable BC loop, making an unambiguous identification of the serotype possible. In most instances, knowledge of the 5'NCR clustering was very useful to guide the choice of primers for the VP1 RT-PCR. With additional information about currently circulating HEV strains it was also possible to identify the few remaining clusters. A major benefit of this approach is that it can be carried out directly on clinical specimens without the requirement for virus propagatation in cell culture. In addition, the sample material is often used up for RNA and DNA extractions for diagnostic purposes, leaving this approach as the only opportunity for possible typing. Furthermore, results can be obtained in as little as 2 days.

Our results are in agreement with the findings of Thoelen et al. (2003), who observed that clustering in the 5'NCR is concordant with clustering in the VP1 region, but seem to contradict studies by Oberste et al. (2004a, b), who suggest that there is only minimal correlation between the 5'NCR and serotype due to recombination between different enteroviruses. Recombination, intratypic as well as intertypic, has been observed previously, and is thought to play an important role in enterovirus evolution (Santti et al., 1999b, Oprisan et al., 2002). However, the predictive power of the 5'NCR sequencing depends largely on the existence in GenBank of recent 5'NCR sequence data for currently circulating strains. The close relationship of the Cypriot isolates to previously identified European strains whose sequences have already been published may account for the success of this approach.

In the summer of 1996, an outbreak of viral meningitis occurred in Cyprus (Ramsay et al., 1996; Communicable Disease Surveillance Centre, 1996a, b). More than 400 cases were admitted to hospital, the majority of whom were infants and young children. Most cases were mild and were discharged from hospital within a few days. No deaths or serious complications were observed. To our laboratory were sent 217 samples from the Limassol district, 88 from the Nicosia district, 84 from the Larnaca district and 16 from the Paphos district for analysis. Cell culture and RT-PCR confirmed the presence of enteroviruses in 292 samples (72 %). Sequencing of a small fragment (135 bp) of the 5'NCR of 38 samples suggested CVB5 as the causative agent (unpublished data). Several stool samples were sent to the Public Health Laboratory Service (PHLS), Virus Reference Division, Colindale, UK, which isolated CVB5 from 14 of 21 specimens. CVB5 was also the most common cause of viral meningitis in England and Wales in 1996 (Communicable Disease Surveillance Centre, 1996c).

In this study, samples from 218 patients diagnosed with enteroviral meningitis between January 2000 and December 2002 were analysed. The most frequent enterovirus serotypes identified were human echoviruses 30 (55.5 %), 13 (15.1 %), 6 (13.8 %) and 9 (8.3 %) (see Table 2). Human echovirus 30 is one of the most commonly isolated enteroviruses, causing outbreaks of aseptic meningitis around the world every few years, and outbreaks in Europe in 2000 and 2001. Human echovirus 13, rarely reported previously, emerged in Cyprus as the second most predominant serotype in 2000. Many reports about Human echovirus 13 outbreaks in 2000 and 2001 across Europe and the USA have confirmed the emergence of this enterovirus as a human pathogen (Mullins et al., 2004; Kirschke et al., 2002; Avellon et al., 2003; Noah & Reid, 2002; Chomel et al., 2003). In general, the pattern of the most frequently isolated enterovirus serotypes in Cyprus seems to follow those in European countries. The large number of tourists arriving in Cyprus each year (∼2 400 000), mainly from European countries, compared to the size of the population (∼730 000), is probably a major factor in the epidemiology of enteroviral infections in Cyprus.

In conclusion, this study is the first molecular epidemiological analysis of enterovirus meningitis cases in Cyprus. A serotype distribution that basically corresponded with observations in other European countries during the same period was observed. Our approach showed that sequencing of a large part of the 5'NCR is very helpful in guiding the choice of primers for the VP1 RT-PCR, thus allowing a reliable typing of enteroviruses from clinical samples. This approach is especially useful in cases in which sample volume is limited or insufficient for cell culture, or the virus is uncultivable.