In total 1,147 blood samples were collected from rodents: 1,060 were live-trapped, wild animals from <20 sites (Table 1), and 87 blood samples were collected from a captive colony of wild house mice (9) and tested serologically. All animal research was conducted under license, according to UK regulations.

Serum samples were separated by centrifugation (10,000 rpm, 10 min) and tested for LCMV antibody by using the manufacturer’s protocol for commercial indirect fluorescent antibody assay slides (Charles River Laboratories, Wilmington, MA, USA). A 1:40 dilution of anti-rat or anti-mouse immunoglobulin G fluorescein isothiocyanate (Sigma-Aldrich, Gillingham, UK) or a combination of both were used as secondary antibody. Ninety-three serum samples (from the original serum samples tested for antibody) that were either antibody positive or from sites with high seroprevalence were tested for arenavirus RNA by PCR. Another 379 blood samples from the captive colony of house mice, which had not been previously tested for antibody, were also tested. The PCR targeted a fragment of the glycoprotein precursor gene (GPC) (10). A selection of samples found negative by the GPC PCR were subsequently retested by PCR targeted at a fragment of the nucleoprotein (N) gene (8), by using primers to sequences common to the Old World arenaviruses. Total RNA was extracted by using QIAamp viral RNA mini-kit (QIAGEN, Crawley, UK), converted to cDNA, and amplified by using a single-step kit (Superscript III one-step RT-PCR with Platinum Taq polymerase system; Invitrogen, Paisley, UK) in conjunction with oligonucleotides arena1⁺ and LCMV322^– (10) or 1010C and either OW1696R or NW1696R (8). Products were separated and visualized by agarose gel electrophoresis, and amplicons were purified with the QIAquick PCR purification kit (QIAGEN). Bidirectional sequencing was performed off-site (MWG Biotech AG, Ebersberg, Germany). The 97-nt sequences generated here were deposited with GenBank (accession nos. DQ275199–DQ275295).

The software package MEGA version 4.0 (11) was used to construct an alignment of a 283-nt fragment of the GPC gene nucleotide sequences and predicted amino acid sequences, and for phylogenetic analysis with the neighbor-joining method (p distance model), with bootstrap support based on 1,000 pseudoreplicates. Other GenBank sequences included for comparison are listed in Table 2. Pairwise genetic distances were calculated by using the p distance model; percentage sequence identities were calculated by subtracting the genetic distances from 1.0 and multiplying by 100.

Overall, 66 of 1,147 serum samples and 7 of 9 rodent species had antibodies to arenaviruses. Sciurus vulgaris had the highest prevalence, 26%, although only 15 squirrels were tested. M. musculus had the second highest prevalence, 17.5%. Antibodies were also detected in Apodemus sylvaticus, Microtus agrestis, Micromys minutis, captive-housed Cynomys ludovicianus, and Rattus norvegicus. Seroprevalence varied between species (1.4%–26%) and between sites (0%–50%) (Table 1).

GPC PCR amplicons were obtained from 127 of 472 tested samples, and sequences were determined for 97 samples (Table 1). All positive samples were from Mus musculus except 1 from A. sylvaticus. Twenty samples negative in the GPC PCR, but seropositive or from high prevalence sites, were tested by N gene PCR, and 2 were weakly positive: 1 S. vulgaris and 1 A. sylvaticus. In neither case, however, could a sequence be obtained from the amplicon.

Appendix Figure

Appendix Figure. Unrooted neighbor-joining radial tree that used the p-distance model (1,000 replicates) for a section of the glycoprotein precursor gene gene, rooted to Lassa virus strain LP. A total of 24 representative...

Nucleotide and amino acid GPC sequence identities for all the samples in this study ranged from 93.6%–100% (Table 2), and 94.5%–100%, respectively (data not shown). When compared with other arenaviruses, the nucleotide sequences exhibited 78.7%–85.8% identity with LCMV reference sequences and only 58.1%–59.9% identity with Lassa virus (Appendix Figure).

Although antibodies to arenaviruses have been reported in a range of European rodent species, our study provided evidence of arenaviruses infecting red squirrels (S. vulgaris) and European harvest mice (M. minutis). Antibodies to arenaviruses have been reported in introduced S. carolensis in Great Britain (12) but were not detected in this study. We also reported antibodies to arenaviruses in black-tailed prairie dogs (Cynomys ludovicianus): those tested in this study were part of a colony in a zoo, however, and had contact with wild mice, some of which were seropositive. As found in previous studies, Mus musculus was more likely to be infected with LCMV than other rodent species.

The nucleotide sequences of most PCR amplicons clearly identified LCMV as the most frequent cause of the antibody detected. However, the detection of arenaviral RNA in 2 animals by the N gene PCR, but not by the LCMV-specific GPC PCR, may suggest the presence of another species of arenavirus. Further studies are needed to determine if other arenaviruses species are present in European rodent populations (8).

Figure

Figure. Unrooted neighbor-joining tree using the p-distance model (1,000 replicates) for a section of the glycoprotein precursor gene gene, showing bootstrap values of >60 for all sequences identified in this study (283...

Genetic heterogeneity was present within and between sites (Figure), as seen in previous studies of arenaviruses (13,14). Sequences from animals in the captive colony and a nearby farm (MF) clustered and were different from those from a more distant farm (BHF). Furthermore, all of the British sequences clearly clustered separately from the reference strain sequences (from the United States, France, Germany, or Slovakia). These findings suggest spatial heterogeneity in sequence may be reflected in host range and pathogenicity. Sequencing might be useful in tracing sources of future human outbreaks.

This study has increased the list of European (and North American) rodents that may be infected with LCMV and that might therefore pose a risk to humans. The genetic variation observed and potential variations in pathogenicity may indicate that some wildlife populations pose more of a public health risk than others. Further studies are needed to assess which mutations cause increased pathogenicity and to establish whether or not LCMV represents the only arenavirus present in European rodent populations.