Mammarenaviruses of Rodents, South Africa and Zimbabwe

We conducted a survey for group-specific indirect immunofluorescence antibody to mammarenaviruses by using Lassa fever and Mopeia virus antigens on serum specimens of 5,363 rodents of 33 species collected in South Africa and Zimbabwe during 1964–1994. Rodents were collected for unrelated purposes or for this study and stored at −70°C. We found antibody to be widely distributed in the 2 countries; antibody was detected in serum specimens of 1.2%–31.8% of 14 species of myomorph rodents, whereas 19 mammarenavirus isolates were obtained from serum specimens and viscera of 4 seropositive species. Phylogenetic analysis on the basis of partial nucleoprotein sequences indicates that 14 isolates from Mastomys natalensis, the Natal multimammate mouse, were Mopeia virus, whereas Merino Walk virus was characterized as a novel virus in a separate study. The remaining 4 isolates from 3 rodent species potentially constitute novel viruses pending full characterization.

We conducted a survey for group-specifi c indirect immunofl uorescence antibody to mammarenaviruses by using Lassa fever and Mopeia virus antigens on serum specimens of 5,363 rodents of 33 species collected in South Africa and Zimbabwe during 1964-1994. Rodents were collected for unrelated purposes or for this study and stored at −70°C. We found antibody to be widely distributed in the 2 countries; antibody was detected in serum specimens of 1.2%-31.8% of 14 species of myomorph rodents, whereas 19 mammarenavirus isolates were obtained from serum specimens and viscera of 4 seropositive species. Phylogenetic analysis on the basis of partial nucleoprotein sequences indicates that 14 isolates from Mastomys natalensis, the Natal multimammate mouse, were Mopeia virus, whereas Merino Walk virus was characterized as a novel virus in a separate study. The remaining 4 isolates from 3 rodent species potentially constitute novel viruses pending full characterization. USA) or prepared at NICD as described elsewhere for Crimean-Congo hemorrhagic fever virus (Table 2) (6).

Rodent Samples and Virus Isolation Studies
Most samples were opportunistically derived from material collected for unrelated surveys and stored at NICD. The initial 213 samples were collected at NICD during 1964-1981 for arbovirus surveys, 3,542 samples were collected and submitted by the Department of Health of South Africa during 1971-1988 for plague surveillance in the central part of the country, 831 rodents (with an emphasis on Mastomys natalensis mice) were collected in northeastern South Africa during 1984-1994 specifically for the investigation of mammarenaviruses, 764 rodent samples collected in Zimbabwe in 1974 were remnants of a study on Rift Valley fever virus (7), and 13 samples were collected in 1982 on a farm in south-central Zimbabwe where there had been a suspected but unconfirmed case of viral hemorrhagic fever in a patient admitted to a hospital in South Africa. Live-trapped rodents were euthanized and exsanguinated; serum samples and visceral organ (lung, heart, liver, spleen, and kidney) samples were conveyed to NICD with ice packs and stored at −70°C. Coordinates of sample collection sites were recorded as quarter-degree grid cells.
We confirmed identities of rodent species yielding virus isolates by determining partial cytochrome b gene sequences for 8 selected samples (8). Skull and skin preparations of rodents from plague surveillance were deposited in the Ditsong National Museum of Natural History (Pretoria, South Africa), and selected materials from other surveys were preserved at NICD.
We attempted isolation of mammarenaviruses for rodent species at locations where antibody was found. We inoculated serum and 10% clarified suspensions of pooled viscera onto Vero 76 mono-layer cultures in replicate Lab-Tek 8-chamber slides (ThermoFisher Scientific) and examined after incubation for 7-10 days at 37°C by IF with pooled mouse antiserum to MOPV and LASV. We passed samples 3 times before recording them as negative. The 5 original isolates of MOPV from M. natalensis rodents from Mozambique were taken to CDC in 1977 (9); we used duplicate organ samples stored at NICD to reisolate the viruses. We screened antigen cell spots prepared from cultures infected with selected known mammarenaviruses plus isolates from this study by IF against mammarenavirus monoclonal antibodies at doubling dilutions from 1:100. We tested all isolates for intracerebral pathogenicity for 1-day-old mice by inoculation of 2 litters (8 infant mice/litter) for each virus.

Results
We tested a total of 5,363 rodents of 33 species from collection sites throughout South Africa and Zimbabwe for antibody to mammarenaviruses (Table 3; Figures 1-4). Antibody was found to be widely distributed in the 2 countries (Figures 1-4) and was detected in serum samples of 1.2%-31.8% of 14 species of myomorph rodents; 19 mammarenavirus isolates were obtained from serum and viscera of 4 seropositive species ( Table 3).
Identities of the 4 myomorph species that yielded mammarenavirus isolates-M. natalensis mice and Aethomys chrysophilus, Micaelamys namaquensis, and Otomys unisulcatus rats-were confirmed from partial cytochrome b gene sequences (8) (GenBank accession nos. MK531528-35). However, the genus Micaelamys has subsequently proved to be polyphyletic and due for revision (16), whereas there is debate about inclusion of O. unisulcatus in the genus Myotomys (17). Furthermore, O. unisulcatus tissue remained available only for the Omdraaivlei isolates and not for the Merino Walk isolate. Most of the other myomorph rodents were identified from morphologic features and distribution patterns (18), but new species and subspecies with partially overlapping distributions have since been recognized in the genus Rhabdomys (19)(20)(21). No organs remained available, and serum specimens failed to yield DNA for phylogenetic studies; thus, the samples are recorded as R. pumilio sensu lato (Table 3).
No mammarenavirus antibody or virus was found in 14 of the myomorph rodent species (Table  3), and although these rodents were relatively poorly represented in the collection, they tend to be rare   (Table 3), but a single sample of O. irroratus rat produced an IF reaction on first pass in cell cultures that was lost during subculture and could not be repeated in further attempts to isolate virus. In addition, attempts to reisolate MOPV from 5 sets of M. natalensis organs from Mozambique (9) held in storage at NICD were successful ( Table 1). All isolates were pathogenic for day-old mice inoculated intracerebrally.
A total of 6 isolates from this study plus 1 reisolated MOVP from Mozambique demonstrated 4 patterns of reactivity in IF screening tests with monoclonal antibodies and selected mammarenavirus isolates ( Table 2). Deduced NP amino acid distances between selected isolates and closest relatives were calculated (Table 4). We determined the phylogenetic relationships of 48 mammarenavirus isolates, including 15/19 isolates from this study and the 5 reisolated MOPV isolates from Mozambique (Table 1), on the basis of neighbor-joining analysis of partial NP sequences (≈912 nt), together with host relationships ( Figure 5). The M. natalensis isolates from Mozambique and from this study grouped with 2 earlier isolates from Zimbabwe as Mopeia virus, whereas 5 isolates from this study fell into 4 groups; isolate Bobomene from South Africa grouped with more recent isolates Mariental from Namibia and isolate Witsand from South Africa grouped with Okahandja from Namibia and with isolate Bitu from Angola ( Figure 5). We determined phylogenetic relationships on the basis of neighborjoining analysis of a 136 bp cytochrome b barcode sequence for 8 selected rodents from which mammarenavirus isolates were obtained in this study and reference taxonomic voucher sequences from Gen-Bank ( Figure 6).

Discussion
The main impetus for this rodent survey came from the isolation of the mammarenavirus MOPV at NICD from M. natalensis rodents collected in a village in Mozambique during an arbovirus study in 1972; within months, the same rodent species was identified as the host of LASV in West Africa (9,22,23). As a consequence, work ceased on Mopeia virus at NICD and the isolates were transferred to CDC, where the relationship to LASV was confirmed (9). Although MOPV proved to be nonpathogenic for nonhuman primates (24), investigating the possible occurrence and role of mammarenaviruses as causes of human infection in South Africa was considered necessary.
Our survey detected widespread presence of antibody activity to mammarenaviruses in myomorph rodent serum specimens within the study area. Because M. natalensis mice have an eastern distribution in South Africa (18) distribution range of M. natalensis mice was found to be distinct from MOPV; 4 isolates obtained from 2 other rodent species further to the west also differed from MOPV (Tables 1-4; Figure 5).   Among routine diagnostic samples submitted to NICD, resting IF titers of 128 and 256 of IgG to MOPV antigen were detected in 2 patients from South Africa, but no etiologic significance could be attached to these findings. A single case of fatal LASV infection was diagnosed in a patient from Nigeria who was evacuated to a hospital in South Africa in 2007 (R. Swanepoel, unpub. data). The only other human arenavirus infections diagnosed within South Africa were in 2 patients referred successively from Zambia in 2008 who were infected with the novel Lujo virus and 3 local healthcare workers who acquired nosocomial infection from those patients (25). At the time of the Lujo virus outbreak, involvement of any of the mammarenaviruses isolated from rodents during the current study was ruled out; in the process, the Merino Walk isolate was characterized as a novel mammarenavirus (26).
The widely distributed M. natalensis mouse of sub-Saharan Africa consists of 6 matrilineages that fall into 2 clades, AI-III and BIV-VI, on the basis of the mitochondrial cytochrome b marker (27,28). Each lineage is associated with >1 mammarenavirus, ranging from LASV in lineage AI in West Africa to MOPV and Luna virus in lineage BVI in southern Africa (28,29). Our findings confirm the association of MOPV with M. natalensis mice in southern Africa, where this rodent is sympatric with M. coucha mice in northeastern South Africa and in Zimbabwe. However, the distribution of M. coucha mice extends westwards into the drier interior of South Africa; the low prevalence of MOPV antibody found in this species could represent spillover of infection from other rodents, rather than the harboring of a mammarenavirus (Table 3). Whereas M. natalensis mice in the mesic east are peridomestic, indigenous rodents tend to be sylvatic and less closely associated with human dwellings in the xeric west, where no evidence of infection was detected in humans.
The isolates from this study are provisionally named for their locations of origin (Table 1; Figure 5), but the isolates obtained from M. natalensis mice represent exemplar isolates of MOPV, and Merino Walk virus is clearly distinct. Although the apparent sharing of rodent hosts mitigates against species recognition within the mammarenaviruses (30), clarifying the interrelationships between the Bobomene, Witsand, and Omdraaivlei isolates and their relationship to the Mariental and Okahandja viruses from Namibia (31) and Bati virus from Angola (32) anticipates complete genomic characterization of the isolates.
The phylogenetic relationships between 8 rodents from which mammarenaviruses were isolated in this study and reference taxonomic voucher sequences from GenBank are compatible with the concept of cospeciation of arenaviruses and their rodent hosts (Figure 6), except that the interrelationships between Witsand, Okahandja, and Bitu isolates await clarification as previously noted. Moreover, the unavailability of rodent host tissue for Merino Walk virus precluded comparison with ostensibly the same host species, the O. unisulcatus rat, of the Omdraaivlei isolates. However, O. unisulcatus rats reportedly comprise a coastal lowland group that is located where the host of Merino Walk virus was collected and a central interior group that covers the area where the hosts of the Omdraaivlei isolates were obtained, although the low sequence divergences did not warrant recognition of subspecies (33). The observations on rodents from Zimbabwe were limited, and the single isolation of Mopeia virus obtained from M. natalensis mice from a farm near Masvingo was not related to the nonfatal illness of a former farm resident who was hospitalized in South Africa.
Further research on mammarenaviruses in rodents in South Africa should include attempts to isolate virus from O. irroratus rats and possibly Lemniscomys rosalia mice, which were underrepresented in this survey; the presence of Luna-related and Lunk-related viruses that were identified in Zambia in M. natalensis and M. minutoides rodents should also be investigated (34). Furthermore, the reservoir host and distribution range of Lujo virus in southern Africa have not been determined. A greater knowledge of the occurrence and diversity of mammarenaviruses in Africa is foundational to understanding the possible health risks associated with these viruses and preparedness for the emergence of such viruses in the future.