Bats host major mammalian paramyxoviruses

Abstract

The large virus family Paramyxoviridae includes some of the most significant human and livestock viruses, such as measles-, distemper-, mumps-, parainfluenza-, Newcastle disease-, respiratory syncytial virus and metapneumoviruses. Here we identify an estimated 66 new paramyxoviruses in a worldwide sample of 119 bat and rodent species (9,278 individuals). Major discoveries include evidence of an origin of Hendra- and Nipah virus in Africa, identification of a bat virus conspecific with the human mumps virus, detection of close relatives of respiratory syncytial virus, mouse pneumonia- and canine distemper virus in bats, as well as direct evidence of Sendai virus in rodents. Phylogenetic reconstruction of host associations suggests a predominance of host switches from bats to other mammals and birds. Hypothesis tests in a maximum likelihood framework permit the phylogenetic placement of bats as tentative hosts at ancestral nodes to both the major Paramyxoviridae subfamilies (Paramyxovirinae and Pneumovirinae). Future attempts to predict the emergence of novel paramyxoviruses in humans and livestock will have to rely fundamentally on these data.

Figure 1. Sampling sites and specimens.

The bat samples used in this study and their countries of origin are depicted in red, rodent samples in blue. Sampled animals are listed by species in Table 1. Numbers indicate paramyxovirus-positive individuals and individuals tested in total. DRC=Democratic Republic of Congo, RCA=Central African Republic. Sampling years and results of individual RT–PCR assays are given in Supplementary Table S1.

Figure 2. Partial L-gene phylogenies including novel paramyxoviruses (PVs) from small mammals.

The genera Rubulavirus (a), Morbillivirus (b), Henipavirus (c) and Respirovirus (e), as well as the subfamily Pneumovirinae (d) are shown with bat viruses coloured in red and rodent viruses in blue. Major PVs are indicated by a pictogram of typical ordinal host and designations of virus species (SV, simian virus; PorPV, porcine PV; PIV, parainfluenzavirus; CDV, canine distemper virus; PPRV, peste des petits ruminants virus; RSV, respiratory syncytial virus; hMPV, human metapneumovirus). Bat and rodent viruses marked by an asterisk have been described previously. Values at node points indicate Bayesian posterior probabilities of grouping (only values above 0.6 are shown). The scale bar indicates substitutions per site. For selected henipaviruses (c), the four amino acid GDNE/GDNQ motif at the catalytic site of the RNA-dependent RNA polymerase is shown. Abbreviations used in virus designations are detailed in Supplementary Table S1.

Figure 3. Bat paramyxovirus (PV) dispersion among bat host families.

(a) Maximum likelihood phylogeny reconstructed from 186 amino acids of the L-gene corresponding to positions 367–552 in NiV strain UM-0128 RNA-dependent RNA polymerase (GenBank, AJ564623) with Paramyxoviridae genera shown in blue, red, green and yellow colour. Rabies virus, Atlantic salmon PV (ASPV), Fer-de-lance PV (FdlPV) and genera not containing bat viruses are shaded in grey. The scale bar represents evolutionary distance in substitutions per site. (b) Chiropteran phylogeny adapted from Simmons. Those bat families sampled in this study are shown in bold type. Detected PV genera per bat family are symbolised by squares coloured as in (a). Black dots in rectangles indicate virus-host pairs (bat families versus bat genera) identified for the first time in this study. The approximate numbers of bat species per family were as follows: Vespertilionidae (more than 350), Pteropodidae (about 200), Phyllostomidae (about 190), Molossidae (about 100), Hipposideridae (about 90, including Rhinopomatidae), Rhinolophidae (about 80), Emballonuridae (about 50), Megadermatidae (about 5), Nycteridae (about 20), Natalidae (about 5), Thyropteridae (about 5), Noctilionidae (2), Myzopodidae (2), Mystacinidae (1), Furipteridae (1).

Figure 4. Serology and virus concentrations for selected bat PV.

(a) Reactivity of different sera (diluted 1/40) from African bats with human mumps virus in Vero cells by indirect immunofluorescence assay, a) Rhinolophidae (R.s., Rhinolophus sp.), b) Hipposideridae (H.c., Hipposideros ruber/caffer), c) Vespertilionidae (M.i., Miniopterus inflatus), d) Pteropodidae (M.t., Myonycteris torquata), e) Emballonuridae (C.a., Coleura afra), f) Pteropodidae (M.p., Micropteropus pusillus), g) and h) non-reactive sera from C.a and M.i. (b) Prototype PVs mumps, measles, RSV, parainfluenza 1 (PIV1) and PIV2 infected in Vero cells, stained with bat sera R.s., H.c., M.p. as above. Signal intensities were rated from negative to +++. (c) Reactivity of Eidolon helvum (E.h., sample GH64) serum from Ghana with NiV antigen using an indirect immunofluorescence assay: a) and b) represent 1/10 and 1/40 dilution of E.h. serum, c) positive control using a guinea pig anti-NiV serum, including a large NiV-induced syncytia stained in red, and d) E.h. serum GH64 diluted 1/10 applied on non-infected cells. Bars in a, b and c represent 50 μm. All panels used the same microscope and camera settings to demonstrate the strength of cross-reactivity between NiV and African viruses. (d) Henipavirus RNA concentrations in 22 E. helvum solid organs and serum, of which 21 tested positive in spleens, 5 in livers, 3 in kidneys, 2 in lungs, 1 in intestine and 1 in serum. Virus concentrations (log₁₀ RNA copies per millilitre of serum or per gram of tissue) are given on the y axis for each bat organ tested. Organ types are identified on the x axis. Bars represent mean virus concentrations per organ type, whiskers show s.d.

Figure 5. Phylogeny of henipaviruses.

(a) Bayesian phylogeny (WAG model) of the genus Henipavirus and sister genera Morbillivirus and Jeilong (J-virus) showing novel bat viruses in red colour. Human parainfluenzavirus 1 (hPIV1) was used as an outgroup. Blue numbers represent viruses also represented in (b). The asterisk indicates the virus from which the full genomic sequence was analysed. For selected henipavirus clades, the four amino acid GDNE/GDNQ motif at the catalytic site of the RNA-dependant RNA polymerase is shown. The scale shows amino acid substitutions per site in a translated 558 nt L-gene fragment, corresponding to positions 367–557 in NiV strain UM-0128L-gene (GenBank, AJ564623). (b) Bayesian nt-based phylogenies (GTR+G+I model) using a 642 nt N-gene fragment (left) corresponding to positions 401–1,042 in NiV strain UM-0128 and the 558 nt L-gene fragment (right). GB1535, virus from Eidolon helvum, Gabon; M74a, E. helvum (full genome available), Ghana; KCR245H, Pteronotus parnellii, Costa Rica. Numbers at nodes indicate Bayesian posterior probabilities. The scale bar represents nt substitutions per site.

Figure 6. Paramyxovirus (PV) faecal shedding over three reproductive seasons.

PV faecal shedding over three reproductive seasons from 2008 to 2010 in a Myotis myotis bat maternity colony. Samples were obtained approximately every 3 weeks during the reproductive season from a Myotis myotis maternity roost located in the attic of a private house in Rhineland-Palatinate, Germany, throughout three sampling years (2008 to 2010). Sampling dates are shown on x axes. Each sample was tested by specific real-time RT–PCR for morbilli-related PV. RNA concentrations per gram of faeces are given on the y axis. Numbers on the x axis represent individual faecal pools tested, consisting of five single faecal pellets each. Gaps between columns represent pools testing negative.

Figure 7. Temporal and geographically linked PV detection in African fruit bat species.

(a) Bat rubulaviruses, henipaviruses and pneumoviruses detected in more than one sampling site or year in Rousettus aegyptiacus (Rou aeg), Eidolon helvum (Eid hel) and Hypsignathus monstrosus (Hyp mon) are represented by different colours and symbols. (b) Detection of virus lineages in different sites and years. GHA=Ghana, CON=Republic of the Congo, DRC=Democratic Republic of Congo, RCA=Central African Republic. (c) Distribution of bat hosts as indicated by the IUCN red list (Benda et al., 2008. Rousettus aegyptiacus. Mickleburgh et al., 2008. Eidolon helvum. Mickleburgh et al., 2008. Hypsignathus monstrosus. In: IUCN 2011. IUCN Red List of Threatened Species. Version 2011.2. (www.iucnredlist.org). Downloaded on 23 November 2011).

Figure 8. Ancestral state reconstruction.

(a) Patristic distances within PV genera detected per ordinal host categories (amino acid-based). Blue and orange colours represent diversification before and after inclusion of novel viruses from this study. (b) Summary of average host switches over ca. 10,000 trees as modelled by parsimony ancestral state reconstruction in Mesquite. Symbols on the x axis depict donor (large symbols) and recipient hosts in the level of ordinal mammalian categories (bats, rodents, primates, carnivores, cetartiodactyls) and birds. Before the analysis, the rabies virus outgroup was removed, because this lyssavirus would have to be mapped to a bat host, biasing host state reconstruction for the root of the PV tree toward bats. Four single PV were removed because they were the only available PV in their particular host species (salmon, snake, dolphin, tupaia). (c) Summary of results from hypothesis testing using maximum likelihood (ML) ancestral state reconstruction in Bayestraits. After analysis of 1,000 ML bootstrap trees, the trait change model was restricted by one of the ten criteria listed, and the resulting average likelihood of the restricted model was compared with that of the unrestricted model. The list shows the factor (n-fold) by which each restricted model was less likely than the unrestricted model (averaged over 1,000 bootstrap replicates of the underlying ML tree).