Previous Article | Next Article ![]()
Journal of Virology, March 2009, p. 2109-2118, Vol. 83, No. 5
0022-538X/09/$08.00+0 doi:10.1128/JVI.02109-08
Copyright © 2009, American Society for Microbiology. All Rights Reserved.

Centre for Infectious Diseases, University of Edinburgh, Summerhall, Edinburgh EH9 1QH, United Kingdom,1 PHLS Coxsackievirus Reference Unit, Department of Medical Microbiology, West Park Hospital, Epsom, Surrey, United Kingdom,2 Enterovirus Laboratory, National Centre for Microbiology, Carlos III Institute of Health, Majadahonda, Madrid, Spain,3 Institute of Health and Community Medicine, University Sarawak Malaysia, Sarawak, Malaysia,4 Department of Virology, University of Turku, Finland,5 M. P. Chumakov Institute of Poliomyelitis and Viral Encephalitides, Moscow, Russia,6 National Virus Reference Laboratory, Dublin, Ireland,7 Department of Medical Microbiology, Leiden University Medical Centre, Leiden, The Netherlands,8 National Institute for Biological Standards and Controls, London, United Kingdom,9 Discipline of Immunology and Infectious Diseases, University of Sydney, Australia,10 Enterovirus Laboratory, National Public Health Institute, Helsinki, Finland,11 Department of Biological Sciences, University of Warwick, United Kingdom,12
Received 7 October 2008/ Accepted 5 December 2008
|
|
|---|
|
|
|---|
Enteroviruses were originally classified as polioviruses, coxsackie virus type A or B viruses, or echoviruses (enteric cytopathic human orphan viruses), depending upon the infectious properties of the virus such as pathogenicity in mice (reviewed in reference 22). From the 1960s onwards, enteroviruses within these groups were further differentiated into serotypes originally by using panels of specific neutralizing antisera and, more recently, by sequence comparisons of structural gene regions such as VP1 (9, 34, 38, 43). There are currently over 100 recognized human enterovirus serotypes that fall into four main species (designated A to D) using phylogenetic analysis (54). The Enterovirus genus additionally contains several other species infecting primates, cattle, and pigs and has recently been expanded to include the genetically related human rhinovirus A and B (54).
The species B serotype, echovirus 30 (E30), is a major cause of meningitis in both children and adults. Among the many serotypes associated with this disease presentation, E30 is generally the most commonly isolated in Europe (8, 31, 49), the United States (10, 37), Asia (1, 60), and South America (33). E30 infections typically occur as a series of outbreaks every 3 to 5 years, frequently over large geographical areas. For example, high frequencies of E30 detection in meningitis cases and surveillance programs were reported for 2000 to 2001 throughout Europe, including Denmark (58), Belgium (57), Cyprus (45), Germany (46), and France (3, 5), and again in 2005 to 2006 (8). Similarly, in the United States, long-term surveillance by the Centers for Disease Control and Prevention revealed peaks of E30 isolation in 1981, 1991 to 1993, 1997, and 2003 (10, 37). The underlying basis for this periodicity in E30 infections and the possible association of different genetic variants of E30 with outbreaks are currently poorly understood.
At any one time point, a range of different species B enterovirus serotypes circulate in human populations. The evolution of enteroviruses occurs through genetic drift and, over much longer periods, antigenic diversification in the structural gene region encoding the virus capsid (7, 14, 25, 30, 51, 55); it may also occur by recombination between the capsid and nonstructural coding parts of the genome and the 5' untranslated region (2, 13, 16, 20, 26, 28, 29, 35, 39, 41, 47, 48, 53). To date, almost all documented examples of recombination have been limited to members of the same species (e.g., between species B serotypes), with the exception of the 5' untranslated region, where only a single genetic group can be identified within human species A and B and a second with species C and D (48).
In this study, we have carried out an extensive investigation of VP1 sequence divergence and recombination through sequencing the 3Dpol region of E30 isolates and samples collected from several European countries, Southeast Asia, and Australia over a combined 8-year observation period. Using this geographically diverse sample collection, our aims were to document the time span and geographical extent of different E30 variants as they emerged and spread during the observation period. The identification of individual recombinants of E30 provides the means to document in detail the dynamics of E30 population turnover, geographical ranges of enterovirus transmission networks, and, ultimately, the relationship between the emergence of new variants of E30 and longer-term changes in disease associations and pathogenicity.
|
|
|---|
|
View this table: [in a new window] |
TABLE 1. Sources of survey specimens
|
Nucleotide sequencing. Amplified DNA from RT-PCR-positive pools was directly sequenced using BigDye (ABI) with the inner sense or antisense primer used for amplification, followed by annotation and alignment of E30 nucleotide sequences using the Simmonic sequence package, version 1.6 (52; http://www.virus-evolution.org).
Phylogenetic analysis.
Phylogenetic trees were constructed by neighbor joining from 1,000 samplings of maximum-composite-likelihood (MCL) distances using the MEGA4.0 software package (56) with pairwise deletion for missing data. Regression analysis used MCL distances (calculated by MEGA [56] with a gamma distribution value,
, of 0.8) between each sequence and the early isolate, 14916net87. A Markov chain Monte Carlo (MCMC) method implemented in the BEAST package, version 1.4 (17), was used to estimate the rate of sequence change (18). Dated sequence sets were run three times with chain lengths of 50 million under the SRD06 model of substitution (50) and assumptions of a strict clock and constant population size. All other parameters were optimized during the burn-in period. Output from BEAST was analyzed using the program TRACER (http://beast.bio.ed.ac.uk/Tracer).
Nucleotide accession numbers. Sequences obtained in this study have been deposited in the GenBank database under accession numbers FJ525954 to FJ526200 and FJ538597 to FJ538865.
|
|
|---|
VP1 sequences from the main E30 group in the sequenced region of VP1 showed approximately 9% nucleotide sequence divergence (mean pairwise P distance), with variability almost entirely restricted to synonymous sites (nonsynonymous to synonymous substitution [dN/dS] ratio based on uncorrected synonymous and nonsynonymous distances of 0.04). The evidence for neutral drift rather than positive (Darwinian selection) in the evolution of this region justifies measurement of its rate of sequence change to estimate the time of the most recent common ancestor (MRCA) of this subgroup of E30 isolates. There was a significant correlation between sequence distance from the earliest main group E30 isolate (14916net87) and distances of sequences from more recently collected isolates (317 sequences; R = 0.341; P = 4.3 x 10–10), providing evidence for measurable sequence drift of E30 over the observation period.
The same data set was analyzed by the Bayesian MCMC method (18). A relaxed (log normal) molecular clock fitted the data better than a strict clock (log likelihoods of –4,895 and –4,966, respectively; log10 Bayes factor of 30.6). Using the relaxed clock, the nucleotide substitution rate was calculated as 8.8 x 10–3 (high-probability distribution [HPD] range, 7.4 x 10–3 to 1.04 x 10–3) substitutions per site per year (Table 2). MCMC predicted a date for the MRCA of the main group of E30 variants of 1977 (HPD range, 1968 to 1985) (Table 2). A similar analysis of a data set that included the older Bastianni prototype isolate (collected in 1958) produced similar substitution rates (8.5 x 10–3 [HPD range, 7.0 x 10–3 to 9.9 x 10–3] substitutions per site per year using a relaxed clock), but an earlier MRCA (1948 [HPD range, 1933 to 1958]), consistent with the earlier isolation date of the oldest collected sequence.
|
View this table: [in a new window] |
TABLE 2. Estimates of rates of sequence change of E30 by regression and BEAST analysis
|
![]() View larger version (40K): [in a new window] |
FIG. 1. Neighbor-joining tree of MCL-corrected pairwise distances of VP1 sequences (A) and 3Dpol sequences (B) from the study subjects. Because of the number of sequences (>300), only those showing greater than 1.0% or 1.5% divergence, respectively, from each other are shown. Dot colors indicate the 3Dpol clade of the displayed sequences (see text), not the H, E, F, and G VP1 subgroups assigned in previous analyses (8, 42). 3Dpol region sequences of study subjects were interspersed with sequences from other species B serotypes and have been included in panel B, labeled with the following abbreviations: E, echoviruses; CVA, coxsackie virus type A; CVB, coxsackie virus type B; EV, new enteroviruses. Accession numbers are given. Clades containing three or more members are colored. The scale bar depicts an evolutionary distance of 0.02. Bootstrap resampling was used to determine robustness of the groupings; values of 70% are shown.
|
![]() View larger version (16K): [in a new window] |
FIG. 2. Distributions of pairwise evolutionary distances between sequences from study subjects and published E30 sequences in the VP1 region (A) and 3Dpol (B). In the latter, the dotted line represents the threshold between intra- and interclade distances.
|
![]() View larger version (28K): [in a new window] |
FIG. 3. Association between VP1 sequence divergence (x axis shows maximum distance in each category) (A) and geographical/temporal separation with recombination frequency (y axis) (B). This records the proportion of pairwise comparisons between sequences that show different 3Dpol clades in each category (total numbers of pairwise comparisons are shown above bars). In panel B, for comparison of sequences of variants >10,000 km apart, separate proportions have been calculated for data sets including (light gray) and excluding (dark gray) sequences from Australia.
|
The relationship between recombination and geographical separation was separately analyzed. Removing the time element by comparing variants collected in the same year (Fig. 3B) indicated that variants collected in the same city or within 1,000 km (generally within Europe for this data set) were mostly nonrecombinant, in contrast to >80% for variants 1,000 to 3,000 km apart and >95% of those collected more 3,000 km apart. The only exception were E30 variants collected in Australia; they showed the same 3Dpol sequences as those in Western Europe in their collection year (2000) (see next section). Combined, these data are consistent with relatively large although geographically restricted transmission networks for each E30 RF and with relatively rapid temporal turnover of variants within individual geographical zones.
Turnover of E30 in a single geographical area. Western Europe was the most intensively sampled region in the current study and provided the opportunity to investigate the turnover of individual RFs with time (Fig. 4). Turnover of E30 variants occurred throughout Western Europe approximately synchronously, as indicated by the low frequency of comparisons between different RFs in the same calendar year for sequences collected <1,000 km apart (40% compared with 22% for variants from the same city).
![]() View larger version (42K): [in a new window] |
FIG. 4. Frequencies of different RFs in Europe over the observation period (240 isolates), color coded as for phylogenetic trees (see the legend of Fig. 1). RFs represented by two or fewer occurrences have been individually labeled in unfilled boxes.
|
We have insufficient sampling outside of Europe to clearly define the extent of the transmission networks within which this turnover of E30 occurred. However, on the boundaries of Europe, different RFs were observed; RF-AA (not found in Europe) was the most commonly detected RF in southern Russia, Georgia, and Azerbaijan in 2000, and RF-A was detected in southern Russia in 2002 despite its virtual disappearance from Europe by that time. Further from Europe, in Southeast Asia, E30 variants in Malaysia (this study), Taiwan (11), and China (60) were entirely distinct. RF-T accounted for 16 of 18 E30 variants found in Malaysia and Taiwan in 2001, to be replaced by RF-F and RF-W in 2003 and 2004. (None of these RFs was found in Europe.) Despite this evidence for the circulation of a different set of E30 variants in Southeast Asia, variants collected in 2000 from Perth, Australia, were RF-P and RF-Z, the two most common RFs in Europe in that year. What these findings imply for transmission routes of E30 and potentially other enteroviruses is discussed below.
Minor RF populations. Circulating populations of E30 are dominated by a series of major RFs that show time-related turnover. In Europe, just 10 of the 26 RFs observed (38%; RF-AX, -Q, -P, -R, -A, -V, -L, -Z, -J, and -O) accounted for 92% (222/240) of isolates characterized over this period, whereas 12 RFs were found only once. These 12 "sporadic" RFs frequently showed different patterns of recombination from those of the main RFs (Fig. 5; Table 3), typically falling into VP1 lineages normally associated with a different 3Dpol group. For example, a well-defined, bootstrap-supported lineage in the VP1 region contained 17 RF-T isolates from Malaysia and Taiwan and a single, sporadic isolate (TW01/Tp/AY/01) with an AY 3Dpol group (Fig. 5). Similarly, a large lineage of European RF-Z isolates contained several sporadic RFs (RF-U, -S, -J, and -AV) interspersed with RF-Z variants. In total, we observed 12 sporadic RFs, each with a 3Dpol group unique to that isolate. A further two isolates clustered with RF-Z isolates in VP1, but 3Dpol sequences fell in the J clade, which included a third E30 isolate with an entirely different VP1 sequence. Collectively, these isolates break the previously observed association between VP1 divergence and linkage to 3Dpol (Fig. 2A), and it is these 14 isolates (Table 3) that account for the low but observable frequency of recombination between variants showing VP1 pairwise distances of <0.05 (6% to 8% of pairwise comparisons in the distance categories of <0.005, 0.005 to 0.1, and 0.01 to 0.025).
![]() View larger version (29K): [in a new window] |
FIG. 5. Parts of VP1 phylogenetic tree showing lineages containing sporadic RFs (filled circles) with different 3Dpol groups from others in the clade (RF-Z, RF-T, and RF-R in the examples shown; isolates shown as unfilled circles). Trees are sections from a complete phylogenetic tree of all 318 study isolates constructed by neighbor joining using Jukes-Cantor corrected pairwise distances. The scale bar depicts an evolutionary distance of 0.1 used for the three trees shown and the complete tree of all isolates (not shown). Bootstrap resampling was used to determine robustness of the groupings; values of 70% are shown.
|
|
View this table: [in a new window] |
TABLE 3. Characteristics of sporadic RFs
|
|
|
|---|
Sequence diversity in the VP1 region occurred predominantly at synonymous sites, consistent with previous analyses of variability of VP1 (4) and the whole P1 region (51). As discussed previously (4, 37, 49), the highly conservative nature of the substitutions that have accumulated in the diversification of E30 genotype II over the past 30 to 40 years does not immediately suggest that sequence change in structural gene regions is primarily driven by immune-mediated selection for antigenic change. The high rate of sequence change of the VP1 region characterized in this study (around 8 x 10–3 substitutions per site per year) (Table 2) lies within the range of substitution rates of other picornaviruses; among enteroviruses, rates of 13.5 x 10–3 for EV71 (species A), 9 x 10–3 for poliovirus serotype 1 (species C), and 5 x 10–3 for EV70 (species D) are described (7, 24, 55), similar to foot-and-mouth disease viruses (e.g., 4 x 10– to 14 x 10–3) (reference 15 and references therein).
The pattern of sequence diversity in the nonstructural region (as exemplified by the phylogeny and sets of pairwise distances in the 3Dpol region analyzed in the current study) differed substantially from VP1. Differences include the existence of discrete clusters rather than the continuum of sequence distances in VP1 (Fig. 2) and the interspersed pattern of 3Dpol sequence groups of E30 with those of other species B serotypes that contrasts markedly with the monophyletic grouping of E30 sequences in VP1 (including the more divergent genotype I sequences) (37, 42). These observations are consistent with the occurrence of frequent recombination and separate "modular" evolution of structural and nonstructural gene regions in species B and other enteroviruses (2, 13, 16, 20, 28, 29, 32, 35, 39-41, 47, 48, 51, 53). Our analysis of a large and geographically diverse set of E30 isolates in the current study provides evidence for an extremely large number of distinct 3Dpol lineages associated with E30 over the past 10 years and beyond; the finding that none was shared with other species B serotypes hints at the enormous diversity of these viruses circulating worldwide at any one time. What remains unclear is the time depth behind the current diversity of 3Dpol lineages and, by extension, this circulating pool of nonstructural region sequences with which E30 recombines. Regression analysis in the 3Dpol region analyzed (Table 2) failed to find a temporal relationship between the sample date and the sequence divergence from the tree root (in contrast to similar analysis of VP1), preventing calculation by BEAST of substitution rates in this region, or a date for the MRCA of the species B nonstructural region population. The latter date is important in further understanding timescales and sources of diversity in this region. It will also perhaps enable a dating of the timescale for the genesis of more than 40 different serotypes of this species since they seem biologically restricted to this nonstructural region population as recombination partners. In the future we propose to analyze much longer sequences in the P3 region on a wider range of sequences in each 3Dpol group to achieve this.
Recombinant forms of E30. In the current study we have used the 3Dpol group as a marker to identify different circulating populations of E30 recombinant forms. The data we have obtained add to previous investigations of E30 population turnover and geographical distributions based on the identification of E30 subpopulations by phylogenetic groupings in VP1 (1, 3, 4, 8, 37, 49, 59). The use of RF- and VP1-based classifications of E30 for these epidemiological investigations has its own advantages and disadvantages. Assignment of E30 into a series of RFs for population studies has the benefit of being unambiguous: recombination events are essentially irreversible, the assignment of 3Dpol sequences into different groups is relatively straightforward (e.g., by pairwise distances) (Fig. 2B), and there is no conceivable process by which 3Dpol sequences might resemble each other through evolutionary convergence except by multiple recombination events. In contrast, E30 classifications based on VP1 sequences have to impose divisions into what, ultimately, is a continuum of variability (Fig. 2A), reflected in part by the wide variety of different classifications with different resolutions of VP1 lineages that have been proposed to date (4, 37, 42, 49). On the other hand, the evolutionary process underlying the differentiation of VP1 can be more easily understood than possibly random or evolutionarily selected recombination events that lead to the appearance of new RFs; VP1 phylogenies represent a much better approximation of the evolutionary history of the E30 serotype than 3Dpol groups and allow time of origin and differentiation of structural gene regions to be more easily inferred. For these reasons, we advocate the use of both methods for investigation of the molecular epidemiology of E30 and other enteroviruses.
Our finding of a 3- to 5-year turnover and replacement cycle for individual RFs within Europe (Fig. 5) is consistent with previous VP1-based (3, 4, 8, 27, 37, 49) and combined VP1/RF-based (27, 32, 53) investigations of E30 population change within individual countries, such as Spain, France, the United Kingdom, and the United States. In one of the earliest studies (37), successive emergence and extinction of VP1 groups 1 to 4 was observed between 1956 and 1998 although in all likelihood (based on the diversity within each group and the time spans involved) they must each have contained a number of separate RFs. A more recent Spanish study identified emergence and extinction of five more closely similar VP1 lineages (all falling within genotype IIB, lineage F) over a period of 8 years (1992 to 2000), very similar to the pattern, shown in Fig. 4, which was observed throughout Europe in the subsequent 7 years. The use of similar time-correlated recombination analysis to that presented in Fig. 3 to a range of other species B serotypes found in blood from donors in the United Kingdom indicates that the rapid turnover observed for E30 is a general attribute of viruses within this species (53). This commonality, indeed, suggests the intriguing possibility of interserotype exchange of nonstructural gene region sequences as a source of 3Dpol diversity we have documented (see next section).
E30 transmission networks. The availability of samples from several European countries, Western Asia, Southeast Asia, and Australia allowed a detailed analysis of the geographical extent of the spread of individual RFs. Although numbers of samples from some areas were restricted, our comparative study provided evidence for a substantial commonality of E30 strains circulating at any one time in Europe, and additionally, at least in the year sampled (2000), in Australia. In contrast, all variants from Southeast Asia were entirely distinct throughout the study period. In terms of their geographical extents, these observations are consistent with previous observations of large-scale emergence and turnover of E30 variants identified as separate lineages in VP1 in samples from the United States (37) and Europe (49) and with detection of related viruses in isolates from widely separated countries. These include China and Japan (1), Western Russia and surrounding countries with Far Eastern Russia and Taiwanese variants (27, 61), and perhaps Spain and Argentina (42). Despite the intervening geographical bloc of Asia, which harbors different E30 populations from those circulating in Europe, the similarity in RF distributions between Australia and Europe in 2000 provides further evidence for very long-distance, rapid transmission of enteroviruses, in this case most likely through intercontinental air travel. These findings heighten concern about the potential rapid global spread of highly pathogenic enteroviruses that might emerge in the future, such as hemorrhagic conjunctivitis EV70 and neurovirulent EV71.
Despite the strong linkage between VP1 lineage (or sequence similarity) and 3Dpol grouping (Fig. 3A), a surprising new finding from the study was the frequent detection of sporadic recombinants of E30. These were embedded within VP1 lineages of E30 variants with a different 3Dpol group, such as the occurrence of an RF-AO isolate from Iceland deeply nested within a VP1 lineage of other European RF-R isolates. Of the 240 isolates analyzed from Europe, a total of 11 were sporadic, and 9 were found uniquely (Table 3); each of these was reamplified and resequenced to verify the presence of these unexpected, discrepant 3Dpol groupings. Their frequent occurrence among circulating strains in both Europe and Southeast Asia (e.g., RF-T) suggests that there is continuing ongoing recombination during even extremely short-term evolution of E30, but in these instances, the variants did not expand sufficiently to replace the virus population (i.e., VP1 lineage) from which they originated. This hypothesis is supported by the pattern of replacement of the larger RF populations in which a new recombinant frequently emerges from within a VP1 clade of the RF it replaces. For example, the RF-R variants that first appeared in Europe in 2002 and continued to circulate until 2007 are direct evolutionary descendants of the older RF-P population (2000 to 2001). Similar emergence of RF-Z from within RF-L is also observed.
The underlying biological or epidemiological factors that govern which recombinants might be short lived or sporadic and which ones emerge and replace their ancestors are entirely unknown. As discussed previously (4, 49), it is difficult to imagine that the turnover of E30 variants results from a process of antigenic shift and drift that drive the evolution of structural proteins of influenza A virus. Although not formally tested, the low level of amino acid sequence variability in the VP1 region and other capsid proteins of E30 genotype II variants would be most unlikely to alter neutralization susceptibility and would therefore not allow a new RF to reinfect previously exposed and immune hosts. Likewise, the primary determinants of virus tropism are the receptor-binding capsid sequences, predicting that RFs probably exhibit a similar cell and tissue tropism. More difficult to discount is the possibility that successful RFs might have enhanced transmissibility, perhaps through achieving longer or more intense virus shedding after infection or greater infectivity on exposure. As discussed (21, 49), genetic determinants for such fitness differences may locate to the nonstructural region and thus vary between RFs. Finally, the process of population emergence, replacement, and extinction might be simply stochastic. Seasonality in E30 incidence alongside temporal and geographical differences in the proportion of human population susceptible to E30 infection potentially combine to create population bottlenecks that determine the chance survival and extinction of competing E30 populations year on year. As with other viruses that show unexplained rapid, large-scale genotype replacements, such as parvovirus B19 genotype 2 displaced by genotype 1 in the 1960s throughout the Northern hemisphere (36), our real understanding of virus population structures, transmission dynamics, host susceptibility changes, and possible evolutionary selection on the virus is currently so poor that these alternative explanations cannot be resolved.
Sources of viral diversity. A final unresolved issue is the immediate source of the wide range of 3Dpol lineages in our collection of E30 isolates. Emerging clearly from the finding of frequent sporadic RFs, the number of 3Dpol groups is much larger than the number of separately identifiable VP1 lineages, suggesting a source outside the E30 population. Although one isolate (ES27) showed evidence for recombination between two European cocirculating lineages (RF-R and RF-Z) (Table 3), it was remarkable that none of the others matched those of earlier E30 isolates circulating in the 1980s or 1990s or any other species B serotype characterized to date. However, this may be because there has been much less large-scale investigation of other commonly isolated enterovirus serotypes, such as echovirus 9 and 11, CAV-9, and some of the coxsackie B viruses that have circulated contemporaneously with E30 over the past 10 years. The available data set for these viruses does not rule out the possibility of frequent interserotype recombination with E30, leading to extensive sharing of 3Dpol lineages with similar timescales for emergence and replacement of the major European 3Dpol lineages (A, P, Z, Q, and R). We are currently investigating this possibility through a second-phase collaborative investigation of 3Dpol sequence comparisons of E9 and E11 isolates with those of E30 over the same observation period of 2000 to 2007. From this, we may understand more about the genetic and epidemiological interrelationships of species B enteroviruses and the nature of their possibly interlinked evolution.
This study was funded by the Wellcome Trust.
Published ahead of print on 17 December 2008. ![]()
|
|
|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2009 by the American Society for Microbiology. For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»