Isolation, characterization, and circulation sphere of a filovirus in fruit bats

Significance

Filoviruses (FiVs) often trigger devastating infectious diseases in Africa, but their natural origin and distribution remain largely unclear, though bats are deeply associated with their circulation. This study reports the identification of a FiV (DEHV) in Chinese fruit bats following a decade of active tracking, further adding evidence that bats are natural hosts for certain FiV lineages. DEHV is significantly distinct from other FiVs in genomic features and phylogeny, expanding the genetic diversity of FiVs in the world. Especially, the virus was repetitively detected at an orchard next to a human settlement, but no associated diseases have ever been reported. The isolation of DEHV provides an invaluable resource to develop control measures for FiV-associated diseases.

Abstract

Bats are associated with the circulation of most mammalian filoviruses (FiVs), with pathogenic ones frequently causing deadly hemorrhagic fevers in Africa. Divergent FiVs have been uncovered in Chinese bats, raising concerns about their threat to public health. Here, we describe a long-term surveillance to track bat FiVs at orchards, eventually resulting in the identification and isolation of a FiV, Dehong virus (DEHV), from Rousettus leschenaultii bats. DEHV has a typical filovirus-like morphology with a wide spectrum of cell tropism. Its entry into cells depends on the engagement of Niemann-Pick C1, and its replication is inhibited by remdesivir. DEHV has the largest genome size of filoviruses, with phylogenetic analysis placing it between the genera Dianlovirus and Orthomarburgvirus, suggesting its classification as the prototype of a new genus within the family Filoviridae. The continuous detection of viral RNA in the serological survey, together with the wide host distribution, has revealed that the region covering southern Yunnan, China, and bordering areas is a natural circulation sphere for bat FiVs. These emphasize the need for a better understanding of the pathogenicity and potential risk of FiVs in the region.

Mammalian filoviruses are among the pathogens of greatest concern worldwide with two most prominent members, Ebola and Marburg viruses, causing highly contagious viral hemorrhagic fevers (1). Two genera are well established within the Filoviridae, Orthoebolavirus, and Orthomarburgvirus (1). The former consists of genetically very diverse members including four African species, Orthoebolavirus zairense (Ebola virus, EBOV), Orthoebolavirus sudanense (Sudan virus, SUDV), Orthoebolavirus bundibugyoense (Bundibugyo virus, BDBV), and Orthoebolavirus taiense (Taï Forest virus, TAFV), and an Asian species, Orthoebolavirus restonense (Reston virus, RESTV) (1, 2). Recently, a new ebolavirus species, Orthoebolavirus bombaliense (Bombali virus, BOMV), has been identified in Sierra Leone bats (3). In contrast, the genus Orthomarburgvirus has less genetic diversity and is comprised of a unique species, Orthomarburgvirus marburgense, with two members, Marburg virus (MARV) and Ravn virus (RAVV) (1, 2). Two more mammalian FiV genera have been established with the discovery of Lloviu virus (LLOV) in Spanish bats in 2011, which forms the genus Cuevavirus (4), and Měnglà virus (MLAV) in Rousettus bats in China, which forms another new genus, Dianlovirus (2, 5). Some FiVs show distinct pathogenicity to humans or other animals. EBOV, SUDV, and MARV frequently trigger deadly human hemorrhagic fevers in Africa (6, 7), while diseases caused by BDBV and TAFV are much more rare (6). RESTV is nonlethal to humans (1) but fatal to nonhuman primates, with several outbreaks in cynomolgus monkeys in the Philippines (1, 8). LLOV is a possibly deadly agent for bats and presumably responsible for mass mortalities in Miniopterus schrebrisii bats in Spain, Portugal, France, and Hungary (4, 9–11). In contrast, the pathogenicity of MLAV and BOMV is still unknown.

Bats participate extensively in the circulation of mammalian FiVs, but cell isolation of them from bats has been very difficult. So far, only two bat FiVs have been isolated in cell culture, MARV from R. aegyptiacus bats in Africa (12, 13) and LLOV from an M. schreibersii bat in Europe (10). Although the most mammalian FiVs have not been isolated from bats, increasing numbers of serological and/or viral RNA investigations show that a broad range of bat species are involved in the ecological circles of EBOV, SUDV, and BOMV in Africa (3, 14–18), and RESTV, EBOV-like, and multiple new FiVs in China (19, 20), Bangladesh (21), the Philippines (22, 23), Singapore (24), and India (25).

The southwest region of Yunnan province in China and bordering regions are tropical or subtropical mountainous areas with high human population densities that also provide highly suitable habitats for various bat species. These vast areas are also recognized hotspots of (re)emerging infectious diseases (EIDs) (26, 27) and, indeed, have frequently suffered from outbreaks of such bats-associated EIDs as Nipah and SARS (28, 29). Notably, many divergent bat FiV genomic fragments have been uncovered in southwest Yunnan during the past decade (5, 20, 30, 31), and further serological surveys of bats across a broad area of South China have revealed wide and complex FiV infections (19, 20), indicating the existence of yet-to-be-identified new FiV(s) circulating in the area.

Results and Discussion

Identification of a Bat FiV.

Dehong Autonomous Prefecture is located in southwest Yunnan bordering with Myanmar with abundant forests and wildlife resources. Large plantations of tropical fruits, such as lychee, longan, and jujube, are important economic sources in the region but have also attracted frequent visits of fruit bats during harvest seasons. To prevent bats from stealing their fruits, orchard owners usually set up nets (Fig. 1 A and B). Since the discovery of FiV DH04 from the lung of an Rousettus leschenaultii bat trapped in an orchard in this area in 2013 (30), we have been conducting a long-term surveillance of bat FiVs in the area. A total of 476 fruit bats trapped in orchard-protecting nets have been collected during seven sampling visits between December 2014 and December 2022 (Fig. 1C), consisting of R. leschenaultii (n = 396), Cynopterus sphinx (n = 68), Megaerops niphanae (n = 3), and Eonycteris spelaea (n = 9). Using a pan-FiV nested RT-PCR (30) to screen all samples, we found only a single positive, in the lung and liver of an R. leschenaultii bat (sample code Rl133-16) collected in December 2016. The 353 nucleotide (nt)-long amplicons shared the highest (79.4% nt) similarity with the corresponding region of the RNA-dependent RNA polymerase gene of MARV, followed by 77.5% with Chinese MLAV, indicating the identification of a new FiV. We tentatively named it Dehong virus (DEHV) based on its identification location.

Fig. 1.

Long-term tracking of FiVs and bat sampling. (A) R. leschenaultii bats trapped in a net at one of the sampled jujube (Ziziphus muaritiana) orchards. (B) A jujube partially eaten by a bat, identified by the orchard owner. He witnessed bats hanging upside down on branches to eat fruits from the bottom in the evenings. The darkened bottom of the jujube indicates that bat consumption had occurred sometime previously. (C) Chronological sampling and testing of bats to track FiVs. Red boxes: DEHV seroreactivities tested by IIFA at each sampling time, expressed as numbered bars, with four colors identifying the four bat species collected and tested at each sampling time. Black boxes: Samples positive for DEHV RNA, with viral isolation in cell culture indicated by a red asterisk. The discovery of DH04 in our previous study (June 2013 triangle) is included on the time axis.

Cell Isolation, In Vitro Cell Tropism, and Morphological Characterization.

We incubated Vero cells with the clarified supernatant of the homogenized positive lung sample. At passage 2, cytopathic effects (CPE) were observed on day 3, becoming consistent following three passages (Fig. 2 A and B). The culture reached a titer of 1 × 10^7.7 TCID₅₀/0.1 mL at passage 5, and this was used in follow-up experiments. The in vitro cell tropism of DEHV was assessed in nine human and animal cell lines. All were susceptible to the virus with six showing morphologically different CPEs and with the remaining three not showing visible CPE during the 7-d culture period (SI Appendix, Fig. S1). Growth curves in these lines showed similar curves with peak titers of 1.1 × 10⁷ to 6.2 × 10⁸ genomic copies/μL or 1 × 10^6.7 TCID₅₀/0.1 mL in human liver tumor (Huh7) cells and 1 × 10^5.5 TCID₅₀/0.1 mL in African green monkey (Vero) cells at 4 or 5 d.p.i. (Fig. 2C). These results show that DEHV has a wide cell species tropism and can replicate efficiently in liver, lung, kidney, or ovary cell lines of the human, monkey, hamster, Myotis bat, bovine, canine, or pig. Transmission electron microscopy (TEM) revealed multiple morphologies of DEHV with a size of 900 to 1,300 nm in length by ~80 nm in width (Fig. 2D and SI Appendix, Fig. S2). The virions were covered with ~10-nm membrane-anchored peplomers. Ultra-thin sections showed inclusion bodies of nucleocapsids (NCs) within the cytoplasm, with some virus particles budding from the cell surface.

Fig. 2.

Isolation and morphological observation of DEHV. (A) Mock-infected Vero cells showing a healthy appearance, with negative staining (Inset) by IIFA using DEHV-specific anti-NP hyperimmune mouse serum at 4 d.p.i. (B) DEHV-infected Vero cells showing clear CPE and bright green fluorescence staining in the cytoplasm (Inset) at 4 d.p.i. (C) Growth curves of DEHV in nine cell lines of different animal origin, with titers peaking mostly at 4 to 5 d.p.i. Virus titers were determined by qRT-PCR (viral RNA) for all cell lines and IIFA (infectivity) for Vero and Huh-7 cell lines. (D) Transmission electron microscopy of DEHV-infected Vero cells, with a virion shown in Inset. The ultrathin section of infected Vero cells at 3 d.p.i. shows a virus particle (arrowed) attaching to the cell surface and numerous inclusion bodies.

Genomic Characteristics and Phylogeny.

High-throughput sequencing of infected Vero cells at passage 5 generated the full genome of the virus, and no sequences of other viruses were found. The result showed that DEHV has the largest genome size (20,943 nt) so far found within the family Filoviridae, with 13 nt-long reverse complementary 3′ and 5′ ends (Fig. 3A). Extragenic sequences include a leader at the 3′ end (49 nt) and a much longer trailer at the 5′ end (774 nt). Its genomic structure resembles those of MARV, RAVV, and MLAV more closely than those of other FiVs, encoding seven proteins: 3′-nucleoprotein (NP) - virion protein (VP) 35 - VP40-glycoprotein (GP) - VP30 - VP24-polymerase (L)-5′. The genes are separated by four intergenic regions, with VP35 and VP30 partially overlapped by VP40 and VP24, respectively. In contrast to the GPs of LLOV and ebolaviruses, which are encoded by two open reading frames (ORFs) via transcriptional editing, DEHV GP is encoded by a single ORF, similar to orthomarburgviruses and MLAV.

Fig. 3.

Genomic and phylogenetic characterization of DEHV. (A) Genomic structure comparison of DEHV with other FiVs. The phylogenetic tree is based on complete genomic sequences using the maximum-likelihood (ML) method. Filled circles indicate that the divergence is supported by bootstrap values of ≥80%. (B) Similarity profiles of DEHV with consensus sequences of the genera Orthomarburgvirus, Orthoebolavirus, Cuevavirus, and Dianlovirus. (C) Taxonomy assignment of DEHV based on ICTV-proposed BLAST-based alignment comparison. (D) ML phylogenies of DEHV with other representatives of mammalian FiVs based on NP, VP35, and partial L aa sequences. The phylogenetic trees of NP and VP35 are almost identical and are therefore represented by a consensus tree. Filled circles indicate nodes supported by ≥ 80% bootstrap replicates.

Global alignment based on complete genomes revealed that DEHV shares low sequence identities with all other FiVs. It has the highest (~60.0% nt) identity with MARV and RAVV, followed by 53.9% with MLAV, and is only distantly related to ebolaviruses (39.9 to 41.0%) and LLOV (37.2%). DEHV shows a similar sequence similarity profile across the genome to currently known mammalian FiVs (Fig. 3B), without the recombinational events found in some FiVs, suggesting a long-term independent evolution of its lineage. The International Committee on Taxonomy of Viruses (ICTV) has proposed an algorithm for filovirus classification using pairwise sequence comparison (PASC)–derived sequence demarcation criteria (32). Genomic sequences of different filovirus genera should differ from each other by ≤45% identities based on BLAST alignment (32). On this basis. DEHV showed the highest (46.3 to 47.2%) identity with MARVs (Fig. 3C). Further analyses of the heterogeneity of nucleotide frequencies among different FiVs revealed that DEHV had a different nucleotide frequency pattern (chi-square test, χ² = 1,577.3, P < 0.0001), with the highest AU content (68.2%) as compared with those of other FiVs (54.0 to 63.5%) (SI Appendix, Fig. S3).

Phylogenetic analyses based on complete genomes and seven ORF amino acid (aa) sequences of mammalian FiVs showed highly similar phylogenetic tree topologies (Fig. 3 A and D and SI Appendix, Fig. S4). These results demonstrate that mammalian FiVs are polyphyletic and are clearly segregated into two major phylogroups, EBOV-like and MARV-like phylogroups. It is interesting to note that this classification corresponds to whether or not they have a GP editing site (Fig. 3A). DEHV formed a separate branch between MLAV and orthomarburgviruses in these phylogenies. Considering the low genome sequence identity, along with the largest genome size and the significantly different base frequency, we propose the classification of DEHV as a member of a new genus within the family Filoviridae. As per Dianlovirus (5), we coined “Delovirus” for the name of the genus, reflecting a new filovirus identified in Dehong prefecture, and accordingly, DEHV is the prototype of deloviruses within the genus.

Cell Receptor Determination and Compound Assay.

Cell receptors are key determinants for tissue tropisms and host range of a virus. Niemann-Pick C1 (NPC1) is an essential receptor for GP binding to mediate entry of EBOV, MARV, LLOV, and MLAV (31, 33–35). The NPC1 domain C displays a helical core structure with two protruding loops that engage a hydrophobic cavity in the head of GP1 (36). The two loops in the protein of humans and many animals are highly conserved with almost identical key residues responsible for interactions with FiVs (Fig. 4A), thereby partially explaining the wide cell tropism of DEHV. The receptor-binding domains (RBDs) of FiVs are much more variable between the two phylogroups but are relatively conserved within them (Fig. 4B). Three-D structure modeling showed that, although the RBDs of DEHV and MARV shared a more similar structure than with EBOV, they all formed a hydrophobic cavity to bind the two NPC1 loops (Fig. 4C). We used three specific interfering RNA (siRNA) fragments to suppress the NPC1 expression in Huh-7 cells (Fig. 4 D and E). In their presence, DEHV replication decreased by 27.6 to 49.5% (t test, P < 0.01) (Fig. 4F), indicating that DEHV also uses NPC1 as its entry receptor. Remdesivir is an adenosine analog with significant antiviral activity against the polymerases of FiVs, including recently discovered LLOV and BOMV (37, 38). It inhibited DEHV replication in Vero cells with a half-maximum effective concentration (EC₅₀) value of 0.10 μM (95% CI: 0.095 to 0.103 μM) (Fig. 4G), consistent with EC₅₀ values for EBOV (0.06 to 0.14 μM) in multiple human endothelial cells (37).

Fig. 4.

Entry of DEHV is mediated by the binding of NPC1, with replication inhibited by remdesivir. (A) Alignments of core aa sequences of loops 1 and 2 of human and animal NCP1s. Shadowed regions are predicted to directly interact with FiV GP1, with five key residues marked by asterisks. These species were selected since have either been reported as natural hosts of FiVs or their cell lines have been determined susceptible to DEHV in this study. Residue numbering refers to the sequence of human NPC1. (B) Alignments of three GP1 regions forming a hydrophobic cavity. The aa motifs predicted to interact directly with the two loops of NPC1 are shadowed with five key aa marked with red asterisks. Residue numbering is based on the GP1 of EBOV. (C) Structure modeling of the binding between GP1 of three FiVs and NPC1 domain C of humans and two bats. The binding interface is magnified to show key residues. (D) qRT-PCR showing that the three siRNAs effectively interfere with the expression of NPC1 in Huh-7 cells. (E) WB analysis confirms the effectiveness of siRNA interference 3 of the expression of NPC1 in Huh-7 cells. (F) Reduction in titer of DEHV in Huh-7 cells due to interference in the expression of NPC1 by the siRNAs. (G) Anti-DEHV activity of remdesivir in Vero cells. **P < 0.01; ***P < 0.001; ****P < 0.0001.

Intense Cross-Reactivity between DEHV, MLAV, and MARV.

Among the FiV structural proteins that elicit Abs, NP is a major antigen with its central region showing a certain similarity with the counterpart of paramyxoviruses and rhabdoviruses (1). A previous study from our laboratory showed cross-reactivity of NPs between MARV and MLAV as well as between EBOV and RESTV, but no cross-reactivity of NPs between FiVs and two other mononegaviruses, Tuhoko pararubulavirus 1 (TUHV) and rabies virus (RABV) (20). Comparison of C terminal aa sequences of these NPs showed 43.8 to 52.1% similarities among DEHV, MLAV, and MARV (Fig. 5A). To investigate the cross-antigenicity of DEHV with other FiVs, six mouse anti-NP hyperimmune serum samples against prokaryotically expressed NPs of different FiVs were used to separately analyze their cross-reactivity with eukaryotically expressed NPs of seven FiVs, TUHV, and RABV. As shown in Fig. 5B, WB did not detect NP cross-reactivity between MARV-like and EBOV-like viruses except for a weak reaction between anti-DEHV-NP serum and the NP of LLOV. As well, there was no reactivity detected between FiVs and the two mononegaviruses, but intense cross-reactivity of both anti-DEHV-NP and anti-MARV-NP sera was found with NPs of DEHV, MARV, and MLAV.

Fig. 5.

Cross-antigenicity and serological survey of DEHV. (A) Phylogenic and pairwise comparison of the 320-aa NP C-terminal sequences (referring to FiV DH04). Brown boxes identify viruses showing significant cross-reactivity with the depth of color representing the levels of cross-reactivity. (B) WB analyses showing reactivity of nine eukaryotically expressed FiV NPs with six anti-NP sera. The same quantity of NPs were loaded in each lane, which were taken from different gels. (C) WB analyses of partial bat sera, showing the reactivity of IIFA-reactive (DEHV+) and nonreactive (DEHV−) samples with nine FiV NPs and EGFP control. Boxed area: Bat serum Rl15-14 strongly reacting with NPs of DEHV, MLAV, and MARV, similarly to that shown with anti-MARV-NP and anti-DEHV-NP mouse sera.

A Natural Circulation Sphere of DEHV.

To better understand how DEHV circulates in the region, we established specific RT-PCR and qRT-PCR methods to re-screen the 476 bat samples. These not only confirmed the positive bat Rl133-16 but also, surprisingly, identified four additional positive R. leschenaultii bats collected in the Decembers of 2014 (n = 1), 2021 (n = 2), and 2022 (n = 1) (Fig. 1C), with virus loads of 4.62 × 10³ to 9.04 × 10⁴ copies/0.1g. Furthermore, sequence comparison showed that the 249 nt-long FiV contig, previously identified by meta-transcriptomic sequencing of the pooled samples of 57 R. leschenaultii bats collected in January 2019 in the same region (20), is 100% identical to the L gene of DEHV. But the two specific RT-PCR methods failed to detect any positives in these bats probably due to the very low viral loads in the samples. Altogether, the ongoing detection of DEHV in R. leschenaultii bats in 2014, 2016, 2019, 2021, and 2022 in the region’s orchards suggests a natural circulation of DEHV there.

We further conducted a serological survey using the indirect immunofluorescent assay (IIFA), western blotting (WB), and a neutralization assay (NA) to determine DEHV-reactive antibody (Ab) levels of bats. By IIFA, 30.8% (28/91) R. leschenaultii bat serum samples collected from the 479 bats reacted with DEHV-infected cells (SI Appendix, Table S1). Most reactive samples (21/28) had a titer of 1:100, but two from 2015 and 2016 had titers of 1:800. Six IIFA-reactive and four IIFA-nonreactive serum samples for which sufficient volumes were available were further validated by WB analysis (Fig. 5C). All six IIFA-reactive samples showed varying levels of reactivity with NPs of DEHV and some other FiVs, but none with the NPs of TUHV and RABV. Of note is that serum Rl15-14 from 2015 had an IIFA titer of 1:800 and showed very strong reactivity with the NPs of MLAV, DEHV, and MARV, but not with those of the other viruses. Its reaction pattern was identical to that of anti-DEHV-NP and anti-MARV-NP mouse sera (Fig. 5 B and C), indicating that bat Rl15-14 had been infected by DEHV or a close relative. Of the four IIFA-nonreactive sera, except for serum Rl16-13 that showed moderate reactivity with the NPs of DEHV and some other FiVs, the remaining three showed no reactivity with DEHV, TUKV, and RABV, but weak reactivity with DH04, MLAV, and LLOV, suggesting possible infection of the bats with other FiV(s). We then tested anti-DEHV neutralizing Ab levels of the 27 bat sera for which there was sufficient volume and found that 14 had varying levels of neutralizing activity with titers between 5 and 1,280 ND₅₀ (50% neutralizing doses) (SI Appendix, Table S1). Notably, two serum samples, Rl17-13 from 2017 and Rl15-14 from 2015 (with IIFA titers of 1:200 and 1:800, respectively) had remarkably high ND₅₀ levels of 640 and 1,280 respectively, indicating a recent exposure of the two bats to DEHV. These serological results provide more robust evidence supporting the claim of long-term circulation of DEHV or closely related one(s) in R. leschenaultii bats in the region.

Previous investigations have revealed high FiV seroprevalences in several bat species in the Philippines (22, 23), Singapore (24), Bangladesh (21), and India (25), but no virus has been detected except for RESTV mainly from M. schreibersi bats in the Philippines (22). Most recently, Makenov et al. reported the detection of FiVs in R. leschenaultii bats in China-bordering Lai Chau Province of Vietnam (SI Appendix, Fig. S5), where they found five 310 nt-long L gene fragments (39). Three of these formed tight clades with previously identified Chinese bat FiV sequences, while the remaining two joined the clade of DEHV with up to 99.0% nt identities (SI Appendix, Fig. S4), indicating the potential distribution of DEHV in northern Vietnam (SI Appendix, Fig. S5). In China, while some other frugivorous and insectivorous bat species are prevalent in Yunnan and other southern provinces (19, 20), FiV RNAs have been detected there only in Rousettus spp. and E. spelaea (30, 31), species widely distributed throughout southern China, southeast Asia, and the entire Indian subcontinent (40). FiVs in R. leschenaultii have been detected six times on six occasions between 2013 and 2022 in our previous (20, 30) and present studies (Fig. 1C). In addition, other diverse FiVs have also been identified in Rousettus spp. and E. spelaea bats from Jinghong and Mengla of Xishuangbanna prefecture (31), leading to the identification of MLAV in a Rousettus bat (5) (SI Appendix, Fig. S5). Our previous serological survey in this district showed 60.8% (87/143) anti-FiV seroprevalence in R. leschenaultii bats (20). A serological study of bats collected between 2006 and 2009 in this region by Yuan et al. reported that the highest seroprevalence against ebolavirus-like viruses was among R. leschenaultii, Pipistrellus pipistrellus, and Myotis sp. (19). All these results clearly demonstrate that R. leschenaultii is a dominant reservoir harboring multiple bat FiVs in Yunnan province, including DEHV, FiV DH04 (30), MLAV (5), and possibly even other yet-to-be-identified FiVs (31). Considering the distribution of FiV-carrying bats along border areas and the long distances that R. leschenaultii can migrate (40), the circulation range of multiple FiVs is undoubtedly wider than Dehong prefecture. Consequently, we propose a natural circulation sphere of diverse FiVs in southern Yunnan province and neighboring southeast Asia (SI Appendix, Fig. S5), within which diverse MARV-like FiVs are harbored by R. leschenaultii or even other bat species. This is a FiV hotspot proposed in Asia, where R. leschenaultii bats are a major host for FiVs, and follows the identification of such in Central Africa, West Africa, and South-Central Europe.

Spillover of these viruses into human society is a major public health concern. Particularly, in these orchards, bat-eaten jujubes are frequently found (Fig. 1B), and the long-term circulation, wide distribution, and intimate interface between bats and humans provide ample opportunities for DEHV spillover. Fortunately, no indigenous FiV-associated diseases have ever been reported in the region even across Asia till now; hence, like RESTV, DEHV is a likely nonlethal agent for humans. Nevertheless, further systematic serology investigations and experimental infections of animals with the isolated virus will be required to fully understand its virulence, pathogenicity, and potential spillover risk to human and domestic animals, as well as to determine its risk classification for biosafety control.

Materials and Methods

Ethics and Biosafety Statement.

The procedures for sampling and processing bats were reviewed and approved by the Administrative Committee on Animal Welfare of Changchun Veterinary Research Institute (Institutional Animal Care and Use Committee Authorization, permit numbers JSY-DW-2011-02 and JSY-DW-2015-01). All animals were handled according to the principles and Guidelines for Laboratory Animal Medicine (41).

The detection and cell culture–based assays of DEHV were performed in a biosafety level 3 (BSL-3) laboratory by well-trained and certified experimental operators, who wore personal protective equipment during the experiments, including protective clothing, safety gloves, goggles, and N95 masks and strictly followed the standard operation protocol of a high-level biosafety laboratory. Containment of DEHV-related activity in BSL-3 is based on the following. i) Among all FiVs, only EBOV and MARV can widely cause highly lethal human diseases; therefore, their biosafety level is classified as BSL-4 in the world, while the biosafety level of members within other FiV genera has not been officially defined. ii) As described above, long-term circulations of divergent bat FiVs have been identified across broad areas of south China and bordering countries by virological and serological investigations, but no FiV-associated diseases have been identified and reported in history in the vast cross-country areas, indicating that they are likely less pathogenic or even apathogenic. iii) Till now, FiVs that have caused human diseases belong to Orthoebolavirus and Orthomarburgvirus genera, while the evidence to show significant human pathogenicity of other newly identified FiV genera has not been found. DEHV is a FiV likely belonging to a new genus, genetically far from orthoebolaviruses and orthomarburgviruses. It has a long-term circulation in the area and has been frequently detected in the last 10 y at the orchard, a very intimate human/bat interface, where cross-species transmission of DEHV from bats to cause human disease should have occurred if it were pathogenic to humans. However, no diseases associated with DEHV in the area have been reported so far. iv) Given the facts above, and according to the WHO standard for risk classification of pathogens (42), the classification of DEHV as risk group 2 is more precise than as risk group 3 or higher. In case potential laboratory exposure might occur, BSL-3 containment of DEHV experiments was approved by the Biosafety Committee of Changchun Veterinary Research Institute (Institutional biosafety authorization: JSYABSL3-D-4[0]-JL025-2022).

Sample Collection.

Between December 2014 and December 2022, we collected bats trapped in protective nets that were set up at several adjacent orchards planting lychees, longans, and jujubes, which were very close to human residences (Fig. 1A). These collections were conducted during seven fruit harvest seasons with each spanning 6 to 20 d (Fig. 1C). We inspected these nets every morning and picked trapped bats from nets. Most were adult, but their exact ages and genders were not recorded at sampling. The majority were dead at collection, and they were immediately subjected to necropsy at the local laboratory to sample their brains, lungs, livers, kidneys, spleens, and recta, and sera when possible. All samples were cryotransported to our laboratory and stored at −80 °C. The species was morphologically identified and confirmed by sequencing the mitochondrial cytochrome b (Cyt b) gene (43).

Virus Detection.

All samples were subjected to the pan-FiV detection as per our published method (30), which uses nested degenerate primer pairs focusing on the most conserved region of the L gene of all known mammalian filoviruses and had been successfully employed to detect diverse bat FiVs (31). Briefly, about 0.1 g of each tissue was subjected to total RNA extraction using a QIAamp RNeasy Mini Kit (Qiagen). The cDNA was synthesized for PCR using a PCR master mix (Tiangen). The specific amplicons were cloned into pMD18T vectors (TaKaRa), and 10 clones of each amplicon were randomly picked for sequencing by the Sanger method on an ABI 3730xl DNA sequencer (ComateBio).

To specifically detect DEHV, the degenerate primers used above were replaced by specific ones based on the genome sequence of DEHV. DEHV-specific qRT-PCR was established to quantify the virus load of positive samples with the following primer and probe sequences: DEHV-GP-QF1 (5′-ACTGAACCAGGCGAGAAGAAA-3′), DEHV-GP-QR1 (5′-ACCAGGACCAAAAAAGGGAAT-3′), and DEHV-GP-probe (HEX5′-CCATCTCGTCCTCTTGAATGCTCCATA-3′MGB). The plasmid bearing the amplicon was constructed to generate the standard curve following serial 10-fold dilutions. Positive tissues were accurately weighed and fully homogenized in PBS for total RNA extraction. The qRT-PCR program was performed using the MonAmp Taqman qPCR mix (Monad, Wuhan, China) with the following program: 15 s at 95 °C then 40 cycles of 5 s at 95 °C, 30 s at 60 °C.

Virus Isolation, Electron Microscopy, and In Vitro Cell Tropism Assessment.

Vero cells (ATCC CCL-81) maintained in our laboratory were grown in 24-well plates. About 0.2 g of the positive lung sample of Rl133-16 was homogenized using 1 mL serum-free Dulbecco’s Modified Eagle Medium (DMEM) (HyClone) with 100,000 U/mL penicillin and 100 μg/mL sodium streptomycin sulfate. The homogenates were clarified by centrifugation at 6,000 × g for 10 min at 4 °C, and 500 μL of the resulting supernatants were incubated with Vero cells at ~80% confluency at 37 °C for 1 h following confirmation as DEHV-positive by RT-PCR. The cells were then maintained in DMEM supplemented with 2% fetal bovine serum (FBS) (Cell-Box) and antibiotics, and inspected daily for CPE. At 7 d, cells were freeze-thawed 3×, and 500 μL culture supernatants were fourfold diluted in DMEM and added to fresh cells plated in a T-25 flask. Cells were harvested for total RNA extraction and assayed for the growth of DEHV by RT-PCR. IIFA was also employed to determine viral growth using specific mouse anti-DEHV NP hyperimmune serum and fluorescein isothiocyanate (FITC)–conjugated donkey anti-mouse IgG H&L (Thermo Fisher Scientific) as primary and secondary antibodies, respectively (see Serological survey). Fifty percent tissue culture infectious doses (TCID₅₀) were determined using IIFA by the Spearman–Kärber method. For electron microscopy, the procedures to prepare negatively stained samples and ultrathin sections of infected Vero cells were as published elsewhere (44). Virus-cell tropisms were determined using A549 (ATCC CCL-185), HeLa (ATCC CCL-2), Huh-7, BHK-21 (ATCC CCL-10), MDBK (ATCC CCL-22), MDCK (ATCC CCL-34), PK-15 (ATCC CCL-33) cells, and an immortalized embryonic Myotis petax bat kidney cell line (BFK) established by our laboratory (45). These cells were cultured in 24-well plates and infected at a multiplicity (MOI) of 0.1 with DEHV harvested from Vero cells at passage 5. Cells were observed daily for CPE for 7 d, and the growth kinetics were determined using qRT-PCR (genomic copies) for all cell lines and IIFA (TCID₅₀) for Huh-7 and Vero cells.

High-Throughput Sequencing.

To obtain the complete genomic sequence of DEHV, infected cells at passage 5 were lysed using TRIzol reagent (TaKaRa). Complete RNA was extracted using phenol–chloroform and ribosomal RNA was removed using an NEBNext^® rRNA Depletion Kit (Human/Mouse/Rat) (NEB, Ipswich, MA). An RNAseq library was prepared by using an NEBNext^® Ultra directional RNA library prep kit (NEB) and then sequenced on an Illumina NovaSeq 6000 sequencer. About six gigabases of raw data were generated. The reads were quality checked using fastp version 0.20.0, and the resultants were de novo assembled using SPAdes genome assembler version 3.14.1 in meta mode. Contigs of ≥ 1 kb were used for ORF prediction using Prodigal version 2.6.3 with translation table 1 and meta mode. Derived amino acid sequences of ≥50 aa were queried against the Eukaryotic Viral Reference Database (EVRD)-aa version 1.0 (46) using DIAMOND blastp version 0.9.35 with e-value cutoff 1e−5. This search found a 21,937 nt-long contig corresponding to a new FiV. We performed self-to-self alignment using reverse complementary sequences to identify the terminal 3′ and 5′ ends and further validated them using RT-PCR and Sanger sequencing. To check the quality of assembly, we mapped these reads back to the complete sequence using bowtie2 version 2.4.1 and calculated the sequencing coverage using samtools version 1.10.

Sequence Comparisons and Phylogenetic Analyses.

For sequence and phylogenetic analysis of DEHV, representative sequences of mammalian FiVs were retrieved from RefSeq as follows: EBOV (NC_002549), BDBV (NC_014373), TAFV (NC_014372), BOMV (NC_039345), SUDV (NC_006432), RESTV (NC_004161), LLOV (NC_016144), MLAV (NC_055510), RAVV (NC_024781), and MARV (NC_001608). Partial cds of BtFiV DH04 were retrieved from GenBank. ORFs of DEHV were predicted de novo using ORF Finder (https://www.ncbi.nlm.nih.gov/orffinder) and further compared with those of other known representative sequences. The locations of transcriptional start and stop signals of DEHV were determined by searching the consensus motifs of the other FiVs.

Base composition bias refers to uneven distribution and composition of the four genomic bases. Usually, base composition is stationary among genomes or genic homologs of close taxa. The base compositional heterogeneity of DEHV with other filoviruses was tested using DAMBE version 7.3.11 (47). A pooled sequence set (i.e., all complete genomes of FiVs) was first subjected to the chi-square test, and if the set did not show homogeneity of base frequencies across taxa (P < 0.05), the sequences were then pairwise-tested. To assess the global sequence identity of DEHV with other FiVs, their complete genomes were aligned using mafft version 7.470 by the e-ins-i method, and pairwise identities were calculated using the MegAlign module within the Lasergene suite version 7.1.0. PASC, the BLAST-based alignment web tool proposed by ICTV, was used to classify a sequence within the family Filoviridae (32). Accordingly, we used the strategy in the taxonomic assignment of DEHV.

Possibilities of historical recombination among filoviruses and the extent of these were assessed using PhiTest, a robust statistical method for detecting the presence of recombination (48) with a window size of 100. No significant evidence for recombination was found (P = 1.0) and confirmed by a series of predictions implemented in RDP4 (49), in which nine methods (RDP, GENECONV, BootScan, MaxChi, Chimaera, SiSican, 3Seq, LARD, and PhyIPro) were employed. If a recombinational prediction was simultaneously supported by at least three methods with MC-corrected probability ≤1e-3, it was considered significant.

The phylogenies of the seven ORFs in the aa and complete genomic nucleotide sequence were inferred using MEGA version X. The alignments were built using mafft, and their hypervariable regions were automatically trimmed by trimAL version 1.2 using default settings. The best substitution models were assessed using ModelFinder version 1.6.8 with the Akaike information criterion. Maximum likelihood trees were constructed using the Subtree-Pruning-Regrafting fast method. The phylogenies were tested by the bootstrap method with 1,000 replicates.

Structural Analysis of the Binding Interface of GP1 and NPC1.

The receptor and GP1 binding interfaces of EBOV, MARV, and DEHV were structurally compared using a previously determined crystal structure (PDB: 5F1B) of EBOV GP bound to human NPC1 domain C (NPC1-C) as template (36). NPC1-C sequences of R. leschenaultii and R. aegyptiacus were retrieved from GenBank, and their counterparts to the human NPC1-C in the structure were modeled using the Swiss-Model server and visualized with PyMOL version 2.4.0. The GP1 structures of MARV and DEHV were also modeled in the same way. Since the GP1 sequences of MARV and DEHV were divergent from that of EBOV, their conformations were evaluated by calculating the global root mean square deviation (RMSD) using the DALI server (50). Pairwise comparison showed that the RMSDs ranged from 0.3 to 1.3 with identities between 42 and 80%, suggesting identical structures and convincing modeling.

RNAi Assay.

Three siRNAs were designed targeting the human NPC1 gene (accession number NM_000271.5) and chemically synthesized (JTS Scientific, Wuhan, China). Huh-7 cells were grown in 24-well plates to 70 to 80% confluence and then transfected with these siRNAs at concentrations of 20 pM using Lipofectamine 2000 reagent (Thermo Fisher Scientific). To validate the interference, cells at 48 h post transfection were harvested for qRT-PCR and WB analyses. Genomic copies were quantified using SYBR PrimeScript Ex Taq II polymerase (TaKaRa) and an Agilent Stratagene Mx3000P Q-PCR system. Fold variations between RNA samples were calculated by the 2^−δδCt threshold cycle (Ct) method after normalization to the amount of GAPDH mRNA. For WB, cells were thoroughly lysed in RIPA Lysis Buffer (Millipore) and about 20 µg aliquots of protein were subjected to SDS-PAGE before being transferred onto polyvinylidene difluoride membranes. After blocking with skimmed milk for 1 h, the membranes were incubated with primary anti-NPC1 (Abcam) and anti-α-tubulin antibodies (Beyotime), and HRP-conjugated secondary antibodies (Beyotime). Specific protein bands were visualized by automatic chemiluminescence (Tanon, Shanghai, China) and quantified by ImageJ software version 1.36b. Cells were infected with DEHV at a MOI of 0.01 at 48 h post-transfection with the three siRNAs and maintained for 96 h, after which viral loads and infectivity were determined by qRT-PCR and IIFA, respectively.

Inhibition Assays of Remdesivir.

Previously published methods (37, 38) were modified to test the in vitro inhibitory effect of remdesivir on DEHV. Its cytotoxicity in Vero cells was determined using a Cell Counting Kit-8 (APExBIO, Houston, TX), and a half-maximum cytotoxic concentration (CC₅₀) of 24.3 μM was found. The compound was prepared in fourfold dilutions from 0.1 to 10.24 μM. Cells grown in 24-well plates were infected with DEHV at an MOI of 0.1 for 1 h and then maintained in DMEM/2% FBS containing dilutions of remdesivir. Inhibition at each dilution was tested in triplicate. Cells were harvested by freeze-thawing 3× at 4 d.p.i., and virus loads and infectivity were determined by IIFA and by qRT-PCR after calibration with the reference GAPDH gene.

Cross-Antigenicity Analysis of NPs by WB.

Prior to use, commercial proteins A and G were tested to validate their reactivity and specificity using IIFA or WB analyses, and their optimal concentrations were determined by serial dilutions. To characterize the cross-antigenicity of FiV NPs, six specific mouse anti-NP hyperimmune sera against Escherichia coli-expressed C terminal NPs (~320 aa) of DEHV, DH04, MARV, RESTV, RABV, and TUHV were western-blotted against BHK-21 cell-expressed EGFP-tagged NPs (the same C terminus) of EBOV, RESTV, LLOV, DH04, MLAV, DEHV, MARV, and two other mononegavirus controls (RABV and TUHV). The hyperimmune sera against NPs of RESTV, DH04, MARV, RABV, and TUHV and the EGFP-tagged NPs of EBOV, RESTV, LLOV, DH04, MLAV, MARV, RABV, and TUHV had been prepared for our previous study (20). Using the same protocol, we prepared transiently expressed EGFP-tagged C-terminal NP of DEHV, and specific mouse anti-NP hyperimmune sera against E. coli-expressed C-terminal NP of DEHV in the present study. EGFP expressed by blank vector pEGFP-C1 was also used in the WB analysis as a control. After quantification by a Bradford Protein Assay kit (Tiangen), the 10 eukaryotically expressed proteins were separated by SDS-PAGE, transferred onto polyvinylidene difluoride membranes, blocked with 5% skimmed milk, and then incubated with the above six hyperimmune sera at 1:1,000 dilution for 2 h. After washing, the membranes were incubated with 1:1,000 diluted Alexa Fluor 680-conjugated donkey anti-mouse IgG H&L (Thermo Fisher Scientific) for 50 min, then visualized using an Odyssey imager (LI-COR Bioscience). To ensure the same amount of these proteins, the loading volume was adjusted according to assessment by ImageJ, which were 50.5 μg, 51.9 μg, 82.4 μg, 63.1 μg, 54.9 μg, 54.6 μg, 119.1 μg, 118.7 μg, 88.8 μg, and 10.4 μg for EBOV, RESTV, DH04, DEHV, MARV, MLAV, TUHV, RABV, LLOV, and empty EGFP blank control, respectively.

Serological Survey.

To determine the seroprevalence of DEHV in the regional bats, we conducted a serological survey of DEHV in bats using IIFA. Prior to IIFA, samples had been inactivated at 56 °C for 30 min. At 4 d.p.i., Vero cells infected with 100 TCID₅₀ of DEHV in 96-well plates were washed 3× with sterile PBS, fixed in cold acetone, and incubated with the samples at a dilution of 1:100 at 37 °C for 1 h, followed by three washes with PBS-Tween (PBST). Equal volumes of FITC-labeled proteins A and G (Boster, Wuhan, China) mixed at a 1:2,000 dilution were used to detect the primary Abs by immunofluorescence microscopy (Olympus). Controls of virus (DEHV-infected Vero cells without addition of serum samples) and reagent (proteins A/G directly added to mock-infected Vero cells) were also included. Antibody titers of samples for which there was sufficient volume were determined by endpoint titration using serial fourfold dilutions (1:100 to 1:3,200). All IIFA-reactive sera were further analyzed by WB with BHK-21-expressed NPs, together with several IIFA-nonreactive sera randomly selected for the analysis. The WB procedure was the same as described above, except using 1:300 diluted bat serum samples as primary Ab and 1:5,000 diluted HRP-conjugated proteins A and G (Boster) as secondary Ab.

The IIFA results were further confirmed by a neutralization assay against DEHV. The IIFA-reactive and some IIFA-nonreactive sera were inactivated as described above. After an initial fivefold dilution, the samples were twofold serially diluted up to 1:5,120, incubated with equal volumes of 100 TCID₅₀ DEHV at 37 °C for 1 h, and then added to Vero cells in 96-well plates for 1-h absorption. After draining, cells were washed 3× and incubated at 37 °C for 4 d. DEHV was detected by IIFA with mouse anti-DEHV NP-specific hyperimmune serum. Fifty percent infectious titers were determined by the Spearman–Kärber formula. Appropriate controls were included.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

The complete sequence for DEHV has been deposited in GenBank under accession number OP924273 (51) and amplicon sequences under accession numbers OP903520–OP903530 (52) and OP941450–OP941459 (53). The RNAseq raw data have been made available in the NCBI Sequence Read Archive (SRA) under accession number PRJNA905542 (54). All other data are included in the manuscript and/or SI Appendix.