Bats Worldwide Carry Hepatitis E Virus-Related Viruses That Form a Putative Novel Genus within the Family Hepeviridae

INTRODUCTION

A third of the world's human population may have been infected with hepatitis E virus (HEV), the prototype of the family Hepeviridae (59). HEVs are small, nonenveloped viruses with an approximately 7,200-nucleotide (nt) positive-sense, single-stranded RNA genome. Human HEV is classified into four genotypes (41). Following food-borne and waterborne fecal-oral infection, clinical symptoms range from asymptomatic to severe hepatitis (57). Hepatitis B virus and HEV constitute the most common causes of acute hepatitis in developing countries (57). In industrialized countries, hepatitis E has long been considered rare, but there is now a growing body of evidence suggesting that particularly HEV genotypes 3 and 4 constitute major etiologies of acute viral hepatitis (13). Moreover, HEV was found to be present in blood donors at rates between 0.07% in Chinese blood donors and 19.5% in Japanese donors with elevated liver enzyme levels (27, 63) as well as in plasma fractionation pools at rates of up to 10% (5). HEV infection can become chronic in immunocompromised patients and has been associated with high rates of mortality in pregnant women (6, 14, 36). Along with a growing awareness of the clinical relevance of HEV, candidate vaccines have been developed and tested in phase II and III clinical trials (64, 72, 75).

HEV differs from all other human hepatitis viruses in that it has been linked to animal reservoirs, beginning with its isolation from swine in 1997 (43). In particular, HEV genotypes 3 and 4 are associated foremost with food-borne zoonotic transmission from deer, domestic pigs, and wild boar (55, 58, 68). Antibodies against human HEV have been detected in several other animal species, including cattle, sheep, and horses (55). Zoonotic infection seems to contribute to a high seroprevalence in swine handlers and veterinarians (7, 22, 42, 44). Critically, those viruses detected in livestock animals such as swine, deer, farmed rabbits, and mongooses all cluster closely with human viruses (12, 23, 50, 73). Beyond this monophyletic clade, more divergent animal hepeviruses have been described recently. Rat sera were long known to cross-react with human HEV antigens serologically (16, 55), and a distinct hepevirus lineage associated with mild hepatitis in infected animals was detected in wild Norway rats (31, 38, 60). Avian hepeviruses have been identified in farmed chickens but in no other birds so far (28). These viruses are globally widespread in poultry farms, causing a range of symptoms from asymptomatic infection to severe hepatitis and splenomegaly (40, 74). The genetically most distant animal hepevirus was already isolated from apparently healthy trout in 1988 but was identified as a member of the family Hepeviridae only in 2011 (3).

Because of the zoonotic nature of HEV infection, it is highly relevant to learn more about its genetic diversity and association with mammalian hosts. Current data suggest the existence of specific virus clades in specific mammalian hosts, but the restriction of previous studies to farmed and peridomestic animals provided little opportunity to investigate the origin of human viruses. The phylogenetic placement of viruses of several unrelated animal taxa with human viruses suggests an acquisition from humans, but data are insufficient to reject alternative hypotheses. Critically, the very recent detection of rodent viruses in a sister relationship with human HEV suggests the existence of a wider, undiscovered diversity of Hepeviridae in mammals. In several studies by us and other groups, bats have proven to be highly efficient indicators of mammalian virus diversity, which may be due to their exceptionally large social group sizes, which promote the acquisition and maintenance of viruses (8, 15, 19, 20, 69, 70). Bats have been linked to a growing number of emerging viruses, including lyssaviruses, coronaviruses, henipaviruses, and filoviruses (8). For all of these viruses, bats carry larger viral diversities than other mammals, supporting the notion that bats might act as viral reservoirs. To examine if bats may also play a specific role in the ecology and evolution of mammalian Hepeviridae, we investigated a globally representative biological sample from 85 different bat species, including over 3,000 specimens. Our results suggest the existence of a genetic clade of viruses whose patristic distance resembles that encountered in rodents and humans (including the associated livestock viruses), altogether yielding genetic criteria to define putative genera within the family Hepeviridae. We conclude that three putative genera of mammalian hepeviruses may exist.

MATERIALS AND METHODS

RESULTS

Heminested RT-PCR was used to screen a total of 3,869 fecal, liver, and blood specimens from 85 different bat species from five continents (Table 2). Hepevirus RNA was found in seven (0.18%) specimens, including five fecal samples, one blood sample, and one liver specimen. RNA-positive samples stemmed from Europe, Central America, and Western Africa (Fig. 1A). Detected sequences originated from five bat species of three different families (Hipposideridae, Vespertilionidae, and Phyllostomidae), representing all three major stem lineages in the phylogenetic tree of bats (Fig. 1B). The detection rate in feces (5 of 2,624) was not significantly higher than those in blood (1 of 1,173) and liver (1 of 72) (P = 0.40 and 0.15, respectively, by two-tailed Fisher's exact test). Very high virus RNA concentrations were found in all fecal samples (median, 8.5 × 10⁸ RNA copies per gram of feces; range, 2.5 × 10⁸ to 6.1 × 10¹⁰ RNA copies per gram of feces). These concentrations were several orders of magnitude higher than that in the single positive blood specimen (1.8 × 10⁵ RNA copies per ml). Hepevirus RNA was also detected in a liver sample at a high concentration comparable to those detected in feces (5.4 × 10⁹ RNA copies per gram of tissue). Virus concentrations were analyzed in all solid organs of the animal whose liver tested positive. As shown in Fig. 2, RNA concentrations in liver tissue exceeded those in any other organ by at least 1,000-fold. The comparable virus concentrations in solid organs other than liver were likely the result of a high level of viremia. Unfortunately, no blood sample was available from this animal.

For phylogenetic analysis, a putative RNA-dependent RNA polymerase (RdRp) gene fragment corresponding to 108 amino acids (aa) was amplified from all seven positive specimens. Distance-based and probabilistic phylogenetic analyses provided an acceptable robustness of major nodes, placing all chiropteran viruses in a separate monophyletic clade within the family Hepeviridae (Fig. 3). Long phylogenetic branches linking the bat viruses to their common ancestor suggested the existence of a high level of diversity of related viruses. The clade of bat viruses was about as diversified as all human HEVs, including related HEVs found in nonhuman hosts such as rabbits, deer, and wild boar (Table 3; see also Table S1 in the supplemental material). The rodent-specific hepeviruses described more recently in wild Norway rats (Rattus norvegicus) (31, 32, 38) had a common ancestor with human HEV and related viruses but not with bat or avian HEV. The patristic distance within the clade of bat viruses exceeded that in the rat-associated clade more than 2-fold (21.1% versus 9.3% amino acid divergence) (Table 3). The bat viruses were also considerably more diversified than all known avian viruses.

Several viruses from livestock and peridomestic animals are placed phylogenetically within and between the four human HEV genotypes. To look for signs of human HEV-related viruses in bats, a subset of 49 bat sera representing major bat lineages, including 20 Myotis dasycneme, 5 Hipposideros gigas, 5 Hipposideros cf. caffer-ruber (Hipposideros sp. that looks like Hipposideros caffer or H. ruber), 4 Rhinolophus alcyone, 5 Rousettus aegyptiacus, 5 Miniopterus inflatus, 5 Coleura afra, and 5 Vampyrodes caraccioli sera, was analyzed. These sera were tested for antibodies to human HEV in an immunofluorescence assay using HEK293T cells transiently expressing the HEV genotype 1 capsid gene. No reactivity was observed for any sample (representative reactions are shown in Fig. 4).

To determine, on the contrary, if descendants of bat viruses are encountered in humans, pooled plasma samples from 93,146 blood donors were tested. As a control, all samples were first screened for human HEV. HEV genotype 3 viruses were detected in 11 pools. This was in agreement with previously determined prevalences of HEV RNA in blood products (2, 27, 63), confirming the suitability of the pooled samples for HEV detection in principle. Notably, the method of blood donor pooling involved the concentration of viral particles by high-speed centrifugation, allowing a nearly quantitative recovery of viruses from pools. The sensitivity limit of the assay for individual blood donations contained in the pools could therefore be projected to be approximately 4.4 log₁₀ copies/ml. This was compatible with previously reported viral loads in individual plasma donations ranging between 3.2 and 5.7 log₁₀ IU/ml (4). None of the blood donor pools yielded positive results in single and heminested RT-PCR assays capable of detecting all bat-associated HEVs, ruling out the existence of human-specific viruses related to the bat-specific clade in these samples. To investigate humans with potentially closer exposures to bats and a propensity for persistent infection (11, 14, 45), an additional test of 453 anonymized sera from Cameroonian HIV-positive patients was done. No evidence of bat or human hepevirus RNA was obtained, whereas the suitability of these samples for the detection of viral RNA was proven previously (51). This may indicate that the level of immunosuppression might not have been as severe in these patients as in other cohorts of HIV-infected individuals tested for HEV (11, 14).

To compare the genome properties of the novel bat viruses with those of other members of the family Hepeviridae, the full genome of one virus, termed BS7, from a German Eptesicus serotinus bat was sequenced. As shown in Fig. 5A, the bat virus represented an independent branch among the full-length hepevirus genomes. The complete genome comprised 6,767 nucleotides (nt), excluding the poly(A) tail at the 3′ end, constituting the shortest mammalian hepevirus genome, with a size comparable to that of the avian viruses (6,654 nt in the U.S. prototype strain [GenBank accession number AY535004]). The coding regions were flanked by a 33-nt 5′-untranslated region (UTR) and a 77-nt 3′-UTR. In the coding region, at least three open reading frames (ORFs) typical of all hepeviruses were identified (Table 4). An internal putative 660-nt ORF overlapping ORF1 in the −2 reading frame was identified at positions 326 to 985 and provisionally termed ORF NX (for N-terminal unknown). Whether ORF NX corresponded to the putative ORF4 or ORF5 described for rodent hepeviruses remains unclear, since the sequence identity with these ORFs was very low (33.3% and 14.4% nucleotide identities and 30.5% and 17.5% amino acid identities with ORF4 and ORF5, respectively), and no significant similarity of the putative gene product of the internal reading frame to any described protein domain could be detected by BLAST. Similar to bat hepevirus ORF NX, rat hepevirus ORF4 is located at the N terminus of ORF1, while ORF5 is located approximately 2,000 nucleotides downstream of ORF1 (31).

In the first ORF of BS7 (ORF1) (nt 34 to 4776), several domains could be predicted (Table 5). A putative methyltransferase domain was found to contain all four conserved amino acid residues identified previously in plus-stranded RNA viruses belonging to the alphavirus-like supergroup (62) at ORF1 amino acid positions 65 (H), 115 (D), 118 (R), and 229 (Y). The beginning of a putative Y domain, as typical for hepeviruses, was found at around aa 216 (VVTY). The level of conservation decreased downstream, and the end of the domain could not be identified. The genome region encoding the papain-like protease in HEV was highly divergent in hepeviruses from different hosts, including bats, and the amino acid residues around a putative catalytic cysteine (TCFL) defined previously by Koonin et al. (35) were found in neither the bat hepevirus nor the previously described rat, bird, or fish hepeviruses.

DISCUSSION

In this study, we have described novel hepeviruses from a globally representative sample of bat specimens. The genomic characterization of a bat hepevirus clearly supported their classification as members of the recently established family Hepeviridae and indicated that they may be the most divergent mammalian hepeviruses described so far.

While a full genomic characterization is generally required for exact conclusions on taxonomy, the RdRp-based criterion established here may represent a useful tool for the typing of partially sequenced field specimens, similar to, e.g., the VP1 gene in enteroviruses and the RdRp gene in coronaviruses (20, 53). Our approach might help to resolve the classification of rodent hepeviruses, for which a fifth genotype has been proposed (38), as well as that of the rabbit hepeviruses, which seem to belong to HEV genotype 3 (23, 24). On the other hand, classification based on the RdRp gene may not always be sufficient to assign novel genera, and criteria based on full-ORF data may be required. The sequence of ORF3, which was found to be highly distinct between hepeviruses from different hosts, might be useful as an additional marker for classification. Our cumulative classification efforts suggested the existence of at least four putative genera in the family Hepeviridae. One genus would comprise human HEV genotypes and closely related animal viruses, while the other three would include viruses from chiropteran, rodent (rat), and avian (chicken) hosts. The trout hepevirus might correspond to a separate taxonomic unit of a higher rank, e.g., a subfamily (3).

The rate of hepevirus detection in bats was similarly low compared to that in other mammals, including humans (1, 9, 16, 24, 38, 58, 60). The highest anti-HEV seroprevalence and RNA detection rates described so far were for farmed animals, e.g., rabbits, chickens, and piglets (23, 33, 40). The accumulation of very large groups of susceptible individuals is uncommon in wild animals, with bats probably constituting the only mammals beside humans that form social groups exceeding several hundred thousand individuals in one place (52). Our study was performed on samples from very large social groups, including a breeding roost of Eidolon helvum bats in Ghana with more than 300,000 individuals (19) as well as vespertilionid bats hibernating in different roosts in Germany and Bulgaria, all exceeding 10,000 members (20, 25). While we have not observed any particularly high prevalence of viruses in these samples, the diversity of bat viruses found in our whole sample was comparable to that of the extensively studied human hepeviruses and much higher than that of the known avian or rodent viruses. We can thus be confident that our study provides a fair representation of the wider bat hepevirus diversity. Despite this, all chiropteran hepeviruses were monophyletic. In contrast to other viruses for which bats are assumed to constitute relevant mammalian reservoirs, such as mammalian coronaviruses and paramyxoviruses (19, 56), bat hepeviruses were not interspersed with human or other mammalian viruses. Bats therefore do not seem to constitute reservoirs of mammalian hepeviruses in general. The clear distinction of the four proposed hepevirus genera implicated rather that hepevirus evolution might have involved a nonrecent invasion of ancestral mammalian hosts. The marked exception to this idea is human hepeviruses, which can be found in both humans and peridomestic animals. It may be plausible that the intensification of farming activities during human history led to transmissions of some but not all human viruses to farmed animals, which would then have maintained these viruses and formed a potential source of zoonotic reinfection. This hypothesis is coherent with the observation that HEV genotypes 1 and 2 have so far been detected only in humans and not in any other animals.

Based on our data, it would now be possible to search for hepeviruses in more distant relatives of farmed or peridomestic animals. This search should include lagomorphs other than rabbits, wild (nonlivestock) ungulates, as well as carnivores other than mongooses. If, indeed, hepeviruses very distinct from human HEV could be found in these taxa, the idea of an acquisition of human hepeviruses by farmed animals would be strongly supported. HEV in wild primates should also be studied to confirm a possibly more general association of human-related viruses with primates.

As with other newly described viruses in animals, we have to address potential zoonotic risks. Human infection has been associated with the consumption of pig, wild boar, and deer meat in industrialized countries (10, 37, 58, 68). Even though bats are consumed by humans in parts of the world (45), we consider it unlikely that bat hepeviruses would easily transmit to humans. As discussed above, the viruses carried by pig, boar, and deer are closely related to human viruses. Neither avian nor rodent hepeviruses were transmissible to primates experimentally (29, 60), and the latter are even more closely related to human viruses than the bat viruses. Despite reports on the serologic reactivity of human and rodent sera with hepevirus antigens from the opposed species (16, 17), proof of zoonotic transmission would require direct virus detection. In our study, we did not find any bat hepevirus RNA in a very broad range of human specimens.

Our data underline the importance of investigating targeted and balanced samples when studying viral host associations. While the surveillance of pathogens in livestock and peridomestic animals is important for epidemiology, ecologically valid conclusions regarding host associations can be very hard to reach using such samples. It is essential to cover large geographic and phylogenetic samples from the spectrum of potential viral hosts (19).

Supplementary Material