In a survey of rodentborne hemorrhagic fever viruses, 1,591 small mammals were trapped in Guinea from 2002 to 2005. Total RNA was extracted from rodent blood preserved in liquid nitrogen by using the Blood RNA kit (Peqlab, Erlangen, Germany). A 1-step RT-PCR targeting a highly conserved region of the RNA polymerase (L) gene was performed by using primers LVL3359A-plus (5´-AGAATTAGTGAAAGGGAGAGCAATTC), LVL3359D-plus (5´-AGAATCAGTGAAAGGGAAAGCAATTC), LVL3359G-plus (5´-AGAATTAGTGAAAGGGAGAGTAACTC), LVL3754A-minus (5´-CACATCATTGGTCCCCATTTACTATGATC), and LVL3754D-minus (5´-CACATCATTGGTCCCCATTTACTGTGATC) (Note: Underlined letters represent differences in nucleotides among plus and minus primers). Because Lassa virus antigen should be frequently detectable in the natural host, and specific antibodies are known to be negatively correlated with its presence, they were not investigated in this study (9).

Figure 1

Figure 1. Map of Guinea showing the location of the 18 trapping sites (small circles). Sites where only Mastomys erythroleucus or M. natalensis were trapped are shaded in gray and black, respectively. Sites...

The animals were caught in 18 different study sites representative of the principal geographic regions of Guinea (10) (Figure 1). All sites were rural villages with a population <1,000. Human Lassa fever seroprevalence was previously reported to be low (0%–11%) in 6 trapping areas and high (25%–55%) in 12 trapping areas (11–13, Figure 1). Of 1,482 murid rodents belonging to 13 genera and at least 20 species, we typed 847 as M. natalensis and 202 as M. erythroleucus but none as M. huberti, by using DNA from liver biopsies. In addition, we karyotyped 12 members of the genus Mastomys in the field by using standard procedures (14).

We obtained positive RT-PCR results from 98 (1.2%) of 1,482 murid rodents. Sequence analysis showed 96 Lassa virus strains with 96%–100% amino acid homology with the prototypic strain Josiah. Lassa virus–positive rodents were only captured in the prefectures of Faranah (villages of Gbetaya, Bantou, Tanganya) and Guéckédou (Denguédou), both situated along the border with Sierra Leone (Figure 1). We PCR-typed all Lassa virus–positive rodents unequivocally as M. natalensis, with 1 male (no. BA686) additionally confirmed by karyotyping (2n = 32, autosomal fundamental number = 53). Overall, 11.3% of M. natalensis were infected with Lassa virus (Table), with 0% in the low seroprevalence area and 5.4%–32.1% in the Lassa fever high seroprevalence area. In the coastal region, where the lowest human Lassa virus seroprevalence (0%–6%) has been reported, only M. erythroleucus was captured. In contrast, in the forest region, where the highest seroprevalence of up to 55% in selected villages has been found, only M. natalensis was trapped (11). Both species were captured in the savannah regions, where seroprevalence was 2%–42%, but M. natalensis was captured more frequently (Figure 1).

Figure 2

Figure 2. Phylogenetic relationships of Lassa virus strains based on a nucleoprotein gene fragment (631 bp) determined by using the neighbor-joining method. The numbers above branches are bootstrap values >50% (1,000 replicates). Scale...

Lassa virus was isolated in cell culture from 32 rodents, and a 631-bp fragment of the nucleoprotein gene previously used for phylogenetic analysis was sequenced (GenBank accession nos. DQ832667–DQ832699) (15). The phylogenetic tree shows that all isolates belong to the lineage IV of Lassa virus and that strains from the prefecture of Faranah cluster with strains isolated previously from human patients of the same region (15) (Figure 2). We detected 2 novel L-gene sequences that shared 68%–74% homology with lymphocytic choriomeningitis virus in 2 rodents of the Mus subgenus Nannomys.

Our results indicate that M. natalensis is very likely the only reservoir host of Lassa virus in Guinea. Previous studies that identified M. erythroleucus and M. huberti as reservoirs of Lassa virus in Sierra Leone may have confused the species, especially because M. natalensis and M. huberti have an identical number of chromosomes (5,7). Despite a massive trapping effort, M. huberti was not detected in the regions of Guinea bordering Sierra Leone that had high human Lassa virus seroprevalence. Therefore, M. natalensis is likely also the only reservoir in Sierra Leone. Whether this is also the case for the genetically more remote lineage I–III strains of Lassa virus remains to be evaluated because M. kollmannspergeri is also present in Nigeria (7,16). While it is reassuring that the reservoir for Lassa virus is M. natalensis, as reported in the literature since the 1970s, our study demonstrates that proper taxonomic identification of the host is necessary before drawing inferences about the ecology of Lassa virus infection.

The relative abundance of Mastomys correlated with human Lassa virus seroprevalence in Guinea: M. natalensis was absent from the region with the lowest seroprevalence and was the only Mastomys species caught in the highest seroprevalence region. However, if both M. natalensis and M. erythroleucus were present, human Lassa virus prevalence was either low, intermediate, or high (Figure 1). This is a novel finding, which confirms and expands on 2 previous studies conducted in Guinea that reported a correlation between the absence of Mastomys spp. and a low human Lassa virus seroprevalence (9,12).

The results of this study have multiple implications for explaining the patchy occurrence of Lassa virus in Guinea and neighboring countries, as well as for Lassa fever control. First, assuming that M. natalensis is the only host of Lassa virus, natural nidality may occur in a similar fashion as that described for Bolivian hemorrhagic fever caused by Machupo virus. The Machupo virus reservoir host, Callomys callosus, has been shown by genetic analysis to be paraphyletic. The populations of rodents responsible for the maintenance and transmission of Machupo virus were monophyletic lineages different from C. callosus in non-disease–endemic regions of Bolivian hemorrhagic fever and coevolving with the virus (17). M. natalensis specimens from regions with high and low Lassa virus activity are being genetically investigated to determine lineages and population structure. Second, the geographic region of potential rodent-to-human transmission of lineage IV strains of Lassa virus is most likely defined by the occurrence of M. natalensis. Mastomys collections from Sierra Leone and Liberia could be molecularly retyped to reevaluate areas of potential Lassa fever reemergence. Third, our mass screening for arenaviruses found only Lassa virus in Mastomys and a novel lymphocytic choriomeningitislike virus in Mus spp. of the subgenus Nannomys, which are closely related to Mus musculus, the Eurasian host of lymphocytic choriomeningitis virus. These findings support the hypothesis of a species-specific association of arenaviruses with their rodent hosts, resulting in cophylogeny.