Previous Article | Next Article ![]()
Journal of Clinical Microbiology, April 2004, p. 1396-1401, Vol. 42, No. 4
0095-1137/04/$08.00+0 DOI: 10.1128/JCM.42.4.1396-1401.2004
Copyright © 2004, American Society for Microbiology. All Rights Reserved.
David Lewis,2 Alison F. Richards,1 Benjamin A. Lopman,3 Antony D. Hale,2 Roger Eglin,2,
Jim J. Gray,1 and David W. G. Brown1
Enteric, Respiratory and Neurological Virus Laboratory, Central Public Health Laboratory, Specialist and Reference Microbiology Division,1 Gastrointestinal Infections Division, Communicable Disease Surveillance Centre, Health Protection Agency, London,3 Leeds Laboratory, Health Protection Agency Yorkshire and Humber, Leeds, United Kingdom2
Received 22 August 2003/ Returned for modification 5 November 2003/ Accepted 22 December 2003
|
|
|---|
|
|
|---|
Recently, detailed studies on the molecular epidemiology of NVs have been published by various groups (3, 28, 37). The genomic diversity of NVs includes two genogroups (I and II) (2, 37), possibly a third, and a number of genotypes, which have yet to be formally agreed upon. The genogroup I strains include Norwalk virus (GI-1), Southampton virus (SOV; GI-2), Desert Shield virus (DSV; GI-3), and Valetta virus (VLV; GI-4), and the genogroup II strains include Hawaii virus (GII-1), Melksham virus (GII-2), Mexico virus (GII-3), and Grimsby virus (GRV; GII-4) (see Table 1) (8).
|
View this table: [in a new window] |
TABLE 1. NV strains associated with outbreaks and sporadic cases of gastroenteritis
|
Recombinant NV particles have been produced for several strains representing a number of NV genotypes, including Norwalk virus (20), Snow Mountain agent (18), Hawaii virus (10), and GRV (13), and have enabled the development of genotype-specific enzyme-linked immunosorbent assays (ELISAs), allowing the rapid screening of large numbers of fecal samples.
This paper describes the diversity of a representative population of NVs circulating in a defined region of the United Kingdom between 1998 and 2001 and compares the diversity of NVs in two particular geographical locations within that region and in different outbreak settings.
|
|
|---|
The outbreaks occurred between September 1998 and August 2001, and the sporadic cases were identified between November 1999 and August 2001, which together represent the majority of NV incidents identified in this health care region.
A 20% fecal suspension was prepared for EM (undertaken at the Leeds Laboratory), and an aliquot of this suspension was sent to the Enteric, Respiratory, and Neurological Virus Laboratory at the Central Public Health Laboratory, Colindale, London, for testing by an ELISA specific for GRV antigen (GRV-ELISA), reverse transcriptase (RT)-PCR, and the heteroduplex mobility assay (HMA) and for cloning and DNA sequencing. All samples were stored at 4°C prior to testing.
Testing algorithm. The GRV-ELISA was used to screen the EM-positive samples for GRV antigen. The samples negative by ELISA were then tested by RT-PCR. PCR amplicons of suitable quality were characterized by HMA as Grimsby/1995/UK-like, Mexico/1989/MX-like, Southampton/1995/UK-like, Valetta/1995/UK-like, or Desert Shield/1990/SA-like strains. HMA was not performed on weak PCR amplicons or when PCR products were generated with the SG1-D1 and SG2-D1 primer sets (11). DNA sequencing was used to characterize PCR amplicons not identified or tested by HMA and to confirm representative HMA profiles.
EM. Fecal suspensions were made by mixing 20% (wt/vol) feces in 0.1 M phosphate-buffered saline (pH 7.2) and vortexing the mixture for 20 s. Suspensions were then centrifuged at 2,500 x g for 30 min. Virus pellets were produced by centrifuging 1 ml of supernatant at 45,000 x g for 1 h (Optima TL; Beckman Coulter, High Wycombe, United Kingdom). A 100-µl cushion of 30% (wt/vol) sucrose was injected into the bottom of each tube before centrifugation. The pellet was resuspended in 50 µl of phosphate-buffered saline, and 25 µl of 1,1,2-trichlorotrifluoroethane was added before the mixture was vortexed for 10 s. The suspension was clarified by centrifugation at 2,500 x g for 10 min. To prepare grids, a copper grid with a hydrophilic carbon-Formvar support film was floated for 1 h on 25 µl of the final supernatant and then stained with 2% sodium phosphotungstate, pH 6.5. The grids were examined for 5 min with a JEOL 1200EX electron microscope at a magnification of x50,000.
Antigen ELISA. The GRV-specific ELISA was carried out as described previously (13).
Nucleic acid extraction and RT-PCR. Fecal samples for RT-PCR were processed as 10% suspensions and extracted by the guanidinium isothiocyanate-silica method (6). The primers used in this study were broadly reactive primers amplifying a region of open reading frame 1 (ORF1) and have been previously described; they included primer sets Ni-E3 (5), Ando-E3 (28), and SG1-D1 and SG2-D1 (11). These primers amplify regions of 113 to 150 bases of the RNA polymerase gene of NVs. Samples that were NV negative by RT-PCR were further screened for sapoviruses (SVs) by use of primer set SR80-JV33 (32, 36).
HMA. PCR amplicons generated from fecal samples were analyzed by the HMA technique (30). Briefly, for an HMA of Ni-E3 amplicons, two reaction volumes were prepared. The first contained 4.5 µl of PCR amplicon with 4.5 µl of reference amplicon A (Grimsby-like virus DNA), and the second contained 4.5 µl of PCR amplicon with 4.5 µl of reference amplicon B (Mexico-like virus DNA). Each reaction volume was mixed with 1 µl of annealing buffer, denatured at 95°C for 5 min (Perkin-Elmer 9600 thermocycler), and allowed to cool to 4°C. For Ando-E3 amplicons, three reference strains were used: Southampton-like virus, Desert Shield-like virus, and Valetta-like virus (VLLV). The assay used VLLV and Southampton-like virus or VLLV and Desert Shield-like virus, depending on the prevalence of the two non-VLLV strains at the time of testing. The HMA reaction mixtures were then electrophoresed with a mutation detection enhancement polyacrylamide vertical gel (Biowhittaker Molecular Applications, Wokingham, United Kingdom) (30). The gel was stained with Gelstar (Biowhittaker Molecular Applications) and photographed with a Polaroid camera.
The presence of a homoduplex profile for one reference strain and a heteroduplex profile for the other reference strain in an HMA was taken to indicate the characterization of that strain. PCR products giving heteroduplex profiles for both reference strains were further characterized by DNA sequencing. Pattern matching of strains, which resulted in heteroduplex profiles identical to either reference strain in a single gel, negated the requirement to sequence all strains not characterized at the HMA stage. For genogroup II characterization, the strains used in the HMA were GRV-like (Carousel/1998/UK; GenBank accession no. AF439539), with 100% identity to Grimsby/1995/UK, and Mexico virus-like (Driffield/1999/UK; GenBank accession no. AF439540), with 95% identity to Mexico/1989/MX. For genogroup I characterization, the following were used: SOV-like (Ramridge/1998/UK; GenBank accession no. AF439543), with 93% identity to Southampton/1995/UK; DSV-like (Windlesham/1998/UK; GenBank accession no. AF439542), with 96% identity to Desert Shield/1990/SA; and VLV-like (Luton/1999/UK; GenBank accession no. AF439541), with 99% identity to Valetta/1995/MA. Reference strains were determined with primer sets Ni-E3 and Ando-E3 to have 74 and 72% sequence identity, respectively, over the region of the polymerase amplified by these primers. Strains not typed by HMA were characterized by DNA sequencing of a cloned amplicon.
Cloning and DNA sequencing. PCR amplicons of NVs were cloned with a TOPO TA cloning system (Invitrogen, Paisley, United Kingdom) prior to sequencing as previously described (27). Contiguous sequences and pairwise alignments of the 76-bp interprimer region (for Ni-E3 and Ando-E3) and the 109-bp interprimer region (for SG1-D1 and SG2-D1) of the NV ORF1 sequences were generated with the Genebuilder and Clustal programs of Bionumerics version 2.5 (Applied Maths, Kortrij, Belgium).
Virus nomenclature. As NV genotypes are defined by their capsid sequences (8), the viruses identified in this study, through the amplification of a region of the RNA-dependent RNA polymerase gene, have been designated strains. The strains are then related to an NV genotype based on published polymerase and capsid sequences of prototype NV strains (37).
Statistical analysis.
The odds of an outbreak being caused by a predominant strain (Grimsby/1995/UK or Girlington/1993/UK) versus any other strain were also stratified by outbreak setting. Odds ratios were compared by the
2 test with 1 df.
|
|
|---|
HMA was able to type 39 of 235 (16%) NVs, which included the GRV, SOV, DSV, and VLV genotypes. Of 235 NV strains, 196 (83%) were not tested by HMA but rather were characterized by DNA sequencing.
The GRV-ELISA was used to examine fecal samples from patients with sporadic cases of NV gastroenteritis. GRV was detected by ELISA in 57 of 141 (40%) sporadic-case samples. Of the ELISA-negative samples, six were RT-PCR positive for GRV (63 of 141 GRV positives or 45%). The remaining 55 (39%) sporadic-case samples were non-GRV NVs, and in 19 (13%), no virus was detected by RT-PCR. Seven samples were retested by EM and found to be negative. The remaining 12 were not tested. Four SVs were detected by RT-PCR in the remaining samples from patients with sporadic cases.
Characterization of caliciviruses. Enteric viruses from 496 outbreaks and 122 sporadic cases of gastroenteritis were characterized. NV was found in 494 outbreaks and 118 sporadic cases. SV was found in two outbreaks and four sporadic cases. The genotypes found and their frequencies are shown in Table 1.
Two outbreak strains originally identified as NV by EM were confirmed by RT-PCR as containing SVs (32, 36). The strain from one outbreak was characterized as Houston/1990/US-like (GenBank accession no. U67859), and the other was Lyon/1998/FR-like (GenBank accession no. AJ251991). Also, four sporadic-case strains were shown to be associated with SVs, including one Lyon/1998/FR-like strain, one strain designated Leeds/2000/UK (which had only 76% identity to Lyon/1998/FR), a third strain designated Leeds/2001/UK (which had only 78% identity to London/1992/UK; GenBank accession no. U67858), and a fourth strain which was London/1992/UK-like.
Distribution of NV strains by geographical location within the North region. A comparative analysis of strains circulating in two metropolitan areas within the North region is shown in Fig. 1. Overall, in the period of September 1998 to August 2001 in the West Yorkshire area, 53% of the NV outbreaks were due to Grimsby/1995/UK, 15% were due to Girlington/1993/UK, 17% were due to other genogroup II strains, and 15% were due to genogroup I strains. In Tyne and Wear, 40% of the NV outbreaks were due to Grimsby/1995/UK, 30% were due to Girlington/1993/UK, 17% were due to other genogroup II strains, and 13% were due to genogroup I strains. In the remaining areas, 55% of the NV outbreaks were due to Grimsby/1995/UK, 22% were due to Girlington/1993/UK, 15% were due to other genogroup II strains, and 9% were due to genogroup I strains.
![]() View larger version (26K): [in a new window] |
FIG. 1. Temporal distribution of NV strains in two metropolitan areas of the North of England from 1998 to 2001. A, Girlington/1993/UK; B, Grimsby/1995/UK; C, Halle445/1999/DE; D, Berlin385/2000/DE; E, Hillingdon/1994/UK; F, Harrow/2001/UK; G, 12C/92/UK; H, Stepping Hill/2001/UK; I, H/Calv/94E/NO; J, Mexico/1989/MX; K, Lancaster/2000/UK; L, Desert Shield/1990/SA; M, Southampton/1991/UK; N, Hak1d/2000/JP (Valetta/1995/MA); O, 12121/1989/UK; P, Fort Lauderdale/1998/US. Genogroup II strains are indicated by dark gray shading; genogroup I strains are indicated by light gray shading. Untyped strains are not shown but are included in the numbers of outbreaks.
|
![]() View larger version (27K): [in a new window] |
FIG. 2. Distribution of NV strains circulating in outbreaks in hospitals, nursing homes, and other community settings. Percentages of NVs for each group are shown for a three-year period, from 1998 to 2001 (see Table 1 for details); untyped strains are not included. Black, Grimsby virus; white, Girlington virus; dark gray, other genogroup II strains; light gray, genogroup I strains.
|
|
|
|---|
The GRV-ELISA proved successful in detecting NV genotypes antigenically related to GRV (12), with 243 of 529 (43%) outbreaks being characterized at this stage. However, there were 16 (3%) outbreak strains that were negative by GRV-ELISA but positive by RT-PCR, demonstrating the increased sensitivity of PCR compared to ELISA. Of 529 RT-PCR-positive NV outbreaks, 251 (48%) were not detected by the GRV-ELISA (including the 16 GRVs). The GII-1 strain was responsible for 104 outbreaks, predominantly between August 1998 and October 1999, with 93 of 104 (89%) occurring between these dates. Only 11 GII-1 outbreaks have occurred in this region since November 1999.
Only small regions of the polymerase gene were used for phylogenetic analysis, as the primers used had to be broadly reactive in order to amplify the cDNA of a wide range of genogroup I and II NV strain types. Although the sequence data generated from a small region of the RNA polymerase should be interpreted with caution, previous phylogenetic studies using related primers have revealed a robust relationship between this small amplified region and more extensive regions of the polymerase gene (6). The use of other primer pairs, which amplify a larger region of the polymerase gene, compromises the sensitivity of the PCR and would have led to far fewer strains being detected and characterized.
A novel NV strain was defined as having 90% or less identity at the nucleotide level with published sequences. This is a figure based on the analysis of other RNA genomes, as there are no clear guidelines defining what constitutes a novel strain of NV based on a partial sequence of the RNA-dependent RNA polymerase-encoding region. Several of the NV strains detected appear to be novel strains that have not previously been reported in the United Kingdom.
One outbreak was associated with Fort Lauderdale/1998/US (Alphatron/1998/NL-like; GenBank accession no. AF195847), a putative genogroup III NV (8) with less than 70% homology at the nucleotide level to all genogroup I and II strains.
Several outbreaks and sporadic cases were associated with the Harrow/2001/UK strain not seen previously in our laboratory, although an identical strain, Bad Berleberg/477/01 (GenBank accession no. AF409066), was detected in Germany also in 2001. It has been shown to be a recombinant NV associated with at least two different capsids (GII-1 Girlington and GII-3 Mexico) (unpublished data).
The Stepping Hill/2001/UK strain, determined by sequence analysis of the ORF1-ORF2 region to be a variant of Hawaii/1971/US (unpublished data); the Rossendale/2000/UK strain, a variant of Norwalk/1968/US, its nearest neighbor at 89% identity; and the Lancaster/2000/UK strain, a variant of Snow Mountain/1976/US, its nearest neighbor at 90% identity, were detected in outbreaks and/or sporadic cases.
The distributions of NV strains of both genogroups were very similar in sporadic cases of gastroenteritis in adults and children, although a KY/89/JP-like strain, a variant of Norwalk/1968/US, was not seen in any outbreaks during this study.
In West Yorkshire, the distribution of NVs was similar to the overall pattern in the North region. However, in the Tyne and Wear area, there was an absence of GII-4 in the 1998-1999 season, with GII-1 strains being predominant. In the West Yorkshire area, GII-4 and GII-1 strains were evenly distributed in the 1998-1999 season; however, GII-1 strains were predominant in 1998 to 1999 in the North region as a whole.
The cocirculation of diverse NV strains in a particular location can lead to the genesis of naturally occurring recombinant strains through recombination events during dual infection (16, 37).
The temporal distribution of NV strains (Fig. 1) shows that in a particular geographical area of the North region in a single month (December 2000), three genogroup II strains were cocirculating, i.e., Grimsby/1995/UK, 12C/92/UK, and Stepping Hill/2001/UK. In September 1999, three genogroup I strains, Southampton/1991/UK, Desert Shield/1990/SA, and Hak1d/2000/JP, were cocirculating.
Genogroup II strains were responsible for the majority (83%) of NV outbreaks. GII-4, the predominant genogroup and -type in the United Kingdom, was the predominant NV genogroup and -type worldwide during this time (3, 12). NV strains capable of causing epidemics have previously been identified. In the 1993-1994 season in the United Kingdom, a GII-3 strain, Mexico/1989/MX, was associated with epidemic spread (12), and this phenomenon was repeated in the 1998-1999 season with the GII-1 strain Girlington/1993/UK. This occurrence demonstrates that the GII-4 strain, although predominant in the United Kingdom and worldwide, can be displaced, even if for only one or two seasons. This finding has implications for virus detection, particularly with the increased use of commercial antigen enzyme immunoassays (34).
The results of this study strengthen the reported association between the Grimsby/Lordsdale cluster of NV strains and outbreaks of gastroenteritis in semiclosed institutions (3, 12). The Grimsby/1995/UK strain was associated with 49% of the outbreaks that occurred in 27 of the 36 and 30 of the 36 months of the study period in nursing homes and hospitals, respectively.
The data presented describe in detail the diversity and temporal distribution of NV strains cocirculating in this region of the United Kingdom between 1998 and 2001, the establishment or presence of a predominant endemic type, and the introduction and subsequent disappearance of an epidemic type.
Although NV diversity due to the accumulation of point mutations and genetic recombination has been described, the differing properties of endemic and epidemic NV strains and the role played by immunity in the selection of NV strains are poorly understood.
Several factors may play a role in determining the prevalent strains circulating in the population, including viral factors such as infectious dose, environmental stability, and virulence, which may have an important role in determining transmissibility in this setting.
The possibilities for investigating biological differences between NV genotypes in the absence of an animal model and tissue culture system are limited, but the findings presented here are consistent with the Grimsby/Lordsdale virus cluster having distinct characteristics. Several other factors may have contributed to the pattern seen, including herd immunity, population mobility and density, and the level of hygiene in institutional settings.
Future approaches to the control and prevention of NV-associated outbreaks, particularly in the hospital setting, will require a better understanding of the importance and interactions of these different factors.
Present address: Genomics, Proteomics and Bioinformatics Laboratory, Central Public Health Laboratory, Specialist and Reference Microbiology Division, Health Protection Agency, London, United Kingdom. ![]()
Present address: National Microbiology Transfusion Laboratories, National Blood Service, London, United Kingdom. ![]()
|
|
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2009 by the American Society for Microbiology. For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»