Figure 1

Figure 1. Bat mammarenavirus detection and host distribution in study of highly diverse arenaviruses in neotropical bats, Brazil. A) Geographic ranges of arenavirus-positive bat species indicated by blue (Artibeus planirostris),...

We investigated diverse specimens from 1,047 adult bats belonging to 32 species collected from southeastern Brazil (Appendix). We analyzed a total of 3,670 different tissue specimens, including spleens (n = 893), lungs (n = 889), intestines (n = 973), and livers (n = 915), for arenavirus RNA by using reverse transcription PCR (RT-PCR) (9) modified to promote NWA amplification (Appendix Table 1, Figure 1).

We detected arenavirus RNA in 4 Artibeus lituratus, 1 A. planirostris, and 12 Carollia perspicillata bats; the overall detection rate was 1.62% (95% CI 0.95%–2.59%). Arenavirus-positive bats were collected during 2007–2011 from 3 sampling sites located in both forest and urban areas within a 60-km radius (Figure 1), suggesting arenavirus maintenance in bat populations in this region. All 3 arenavirus-positive bat species are abundant in tropical environments and well-adapted to urban landscapes, indicating potential for dispersion and spillover to humans and other animals.

Most arenavirus-positive bats were collected in 2 forest fragments in 2007 (Tables 1, 2; Figure 1), where most bat species positive for arenavirus RNA were sampled. Whether high detection rates at those sites correspond to epizootics or sampling bias remains unknown.

All arenavirus-positive animals appeared healthy, suggesting limited negative effects of arenavirus infection on bat hosts. This observation was similar in rodent arenavirus hosts (10) and consistent with high TCRV seroprevalence in a serologic survey (4) but different from experimental TCRV infections (5), likely because of different routes and high doses used for infecting bats in laboratory settings. High seroprevalence and low arenavirus detection rates suggest that arenaviruses do not infect bats persistently, which is distinct from results for rodent arenavirus infections (11). Lack of persistence is important for public health because it indicates potential limitations of arenavirus shedding by bat hosts whose lifespan is <8–12 years (12).

We detected arenavirus RNA in multiple organs at similar concentrations, including spleens (mean, 1.2 × 10⁷ RNA copies/mg) and lungs (mean, 6.4 × 10⁶ RNA copies/mg) (p = 0.53 by Mann-Whitney U test) (Table 2), suggesting systemic infection similar to that observed in experimentally infected bats (5). We observed the highest arenavirus RNA concentration in the single arenavirus-positive intestine specimen, followed by the spleen, lung, liver, and kidney in that animal (Table 2). High arenavirus RNA concentrations in intestines are consistent with virus shedding through the enteric route, which has been observed during experimental infections with TCRV (5). Although rodents shed arenaviruses primarily through urine and saliva, shedding also occurs in feces (2). Determining differences in arenavirus transmission routes between bats and rodents will require further investigation. We were unsuccessful isolating bat arenaviruses from organ homogenates despite repeated attempts (Appendix), likely because of tissue degradation under tropical conditions.

Figure 2

Figure 2. Phylogenetic analyses of highly diverse arenaviruses in neotropical bats, Brazil. Maximum-likelihood consensus trees compare partial RNA-dependent RNA polymerase genes (A), complete large (L) segment genes (B), and complete small (S)...

We performed phylogenetic analysis of the partial sequence for the arenavirus RNA-dependent RNA polymerase gene obtained from RT-PCR screening. We found 2 NWA clades in bats from Brazil: 1 clade for both Artibeus spp. and 1 clade for C. perspicillata bats (GenBank accession nos. ON648806–16) (Figure 2, panel A). We obtained complete arenavirus coding sequences from 1 A. planirostris and 3 C. perspicillata bats (GenBank accession nos. ON648817–24) by using Illumina-based deep sequencing (Illumina, https://www.illumina.com); genome organization was identical to other mammarenaviruses. Both arenaviruses formed a well-supported monophyletic clade with TCRV in sister relationship to Junin and Machupo viruses (Figure 2, panel B) and Ocozocoautla de Espinosa virus that was possibly responsible for a hemorrhagic fever outbreak in Mexico (Figure 2, panel C) (13). These results highlight the genetic relationship of those bat-associated arenaviruses with highly pathogenic NWAs (Appendix Table 2). Identical topology in phylogenetic reconstructions argued against potential reassortment (Figure 2, panels B, C), and homogeneous sequence distances and recombination analyses along the genome did not indicate recombination events (Appendix Figure 2).

The A. planirostris bat was infected with a previously unknown TCRV strain (Appendix Table 2) that had an amino acid identity of 93.8%–95.5% with other TCRV sequences, depending on the protein analyzed. The arenaviruses from C. perspicillata bats formed a separate species in clade B of the TCRV serogroup (Figure 2, panels B, C). Species assignment relied on taxonomic criteria (14) that included exclusive detection in a distinct host, nucleotide sequence identity of <80% in the small segment, and 88.6%–90% amino acid identity in the nucleocapsid protein compared with TCRV and pairwise sequence comparison (https://www.ncbi.nlm.nih.gov/sutils/pasc/viridty.cgi?textpage=overview) results for large and small segments (Appendix Figure 3). The 5′ and 3′ ends of large and small genomic segments obtained from the newly identified arenavirus from C. perspicillata bats were nearly identical to TCRV, consistent with a close genetic relationship between those NWAs (Appendix Table 3, Figure 4). We propose that the arenavirus sequenced from C. perspicillata bats should be named Tietê virus (species Tiête mammarenavirus) and abbreviated as TETV; the name comes from the main river located <4 km from the capture site (Figure 1).

Arenavirus genetic diversity is hypothesized to result from a complex macro-evolutionary pattern that includes both co-evolution and host switching in the Muridae family of rodents. In South America, arenaviruses might have co-evolved with rodents in the Sigmodontinae subfamily, with the exception of TCRV (10). Further investigation will be required to determine whether bat arenaviruses evolved from an ancestral host switch involving rodents, which would be consistent with the genetic relationship between TCRV or Tietê virus and rodent-derived Ocozocoautla de Espinosa virus, or whether bats and arenaviruses co-evolved. Of note, bats play an essential role in ecosystems, and stigmatization of bats as sources of zoonotic viruses is unwarranted.

In summary, the epidemiology, genealogy, and zoonotic potential of bat arenaviruses deserve further investigation. Our results suggest that bats are an underrecognized arenavirus reservoir.