Zoonotic potential of a novel bat morbillivirus

Bats are significant reservoir hosts for many viruses with zoonotic potential1. SARS-CoV-2, Ebola virus, and Nipah virus are examples of such viruses that have caused deadly epidemics and pandemics when spilled over from bats into human and animal populations2,3. Careful surveillance of viruses in bats is critical for identifying potential zoonotic pathogens. However, metagenomic surveys in bats often do not result in full-length viral sequences that can be used to regenerate such viruses for targeted characterization4. Here, we identify and characterize a novel morbillivirus from a vespertilionid bat species (Myotis riparius) in Brazil, which we term myotis bat morbillivirus (MBaMV). There are 7 species of morbilliviruses including measles virus (MeV), canine distemper virus (CDV) and rinderpest virus (RPV)5. All morbilliviruses cause severe disease in their natural hosts6–10, and pathogenicity is largely determined by species specific expression of canonical morbillivirus receptors, CD150/SLAMF111 and NECTIN412. MBaMV used Myotis spp CD150 much better than human and dog CD150 in fusion assays. We confirmed this using live MBaMV that was rescued by reverse genetics. Surprisingly, MBaMV replicated efficiently in primary human myeloid but not lymphoid cells. Furthermore, MBaMV replicated in human epithelial cells and used human NECTIN4 almost as well as MeV. Our results demonstrate the unusual ability of MBaMV to infect and replicate in some human cells that are critical for MeV pathogenesis and transmission. This raises the specter of zoonotic transmission of a bat morbillivirus.


Introduction
Bats are signi cant reservoir hosts for many viruses with zoonotic potential 1 . SARS-CoV-2, Ebola virus, and Nipah virus are examples of such viruses that have caused deadly epidemics and pandemics when spilled over from bats into human and animal populations 2,3 . Careful surveillance of viruses in bats is critical for identifying potential zoonotic pathogens. However, metagenomic surveys in bats often do not result in full-length viral sequences that can be used to regenerate such viruses for targeted characterization 4 . Here, we identify and characterize a novel morbillivirus from a vespertilionid bat species (Myotis riparius) in Brazil, which we term myotis bat morbillivirus (MBaMV). There are 7 species of morbilliviruses including measles virus (MeV), canine distemper virus (CDV) and rinderpest virus (RPV) 5 . All morbilliviruses cause severe disease in their natural hosts [6][7][8][9][10] , and pathogenicity is largely determined by species speci c expression of canonical morbillivirus receptors, CD150/SLAMF1 11 and NECTIN4 12 . MBaMV used Myotis spp CD150 much better than human and dog CD150 in fusion assays. We con rmed this using live MBaMV that was rescued by reverse genetics. Surprisingly, MBaMV replicated e ciently in primary human myeloid but not lymphoid cells. Furthermore, MBaMV replicated in human epithelial cells and used human NECTIN4 almost as well as MeV. Our results demonstrate the unusual ability of MBaMV to infect and replicate in some human cells that are critical for MeV pathogenesis and transmission. This raises the specter of zoonotic transmission of a bat morbillivirus.

Results
Isolation of MBaMV sequence. During a metagenomic genomic survey of viruses in bats, we identi ed a full-length morbillivirus sequence from a riparian myotis bat (Myotis riparius) in Brazil. This myotis bat morbillivirus (MBaMV) had a genome length of 15,804 nucleotides consistent with the rule of six and comprise of six transcriptional units encoding the canonical open reading frames (ORFs) of nucleo (N) protein, phospho (P) protein, matrix (M) protein, fusion (F) protein, receptor binding protein (RBP), and large (L) protein (Extended Data Fig. 1a). The sizes of these ORFs are comparable to their counterparts in the other morbilliviruses (Extended Data Fig. 1b). Phylogenetic analysis using the full-length L protein sequence indicated that MBaMV is most closely related to canine distemper virus (CDV) and phocine distemper virus (PDV) (Extended Data Fig. 1c, Extended Data Table 1).
Paramyxovirus proteins with the most frequent and direct interactions with host proteins, such as P and its accessory gene products (V and C) as well as the RBP, tend to exhibit the greatest diversity 13 .
Morbillivirus P, V and C antagonize host-speci c innate immune responses while its RBP interacts with host-speci c receptors. That these proteins are under evolutionary pressure to interact with different host proteins is re ected in the lower conservation of MBaMV P/V/C (31-43%) and RBP (27-32%) with other morbillivirus homologs. This is in contrast to the relatively high conservation (52-76%) of MBaMV N, M, F, and L proteins with their respective morbillivirus counterparts (Extended Data Fig. 2).
Species speci c receptor usage. The use of CD150/SLAMF1 to enter myeloid and lymphoid cells is a hallmark of morbilliviruses, and also a major determinant of pathogenicity. CD150 is highly divergent across species, and accounts for the species restricted tropism of most morbilliviruses 14 . Thus, we rst characterized the species-speci c receptor tropism of MBaMV. We performed a quantitative image-based fusion assay (QIFA) by co-transfecting expression vectors encoding MBaMV-F and -RBP, along with CD150 from the indicated species into receptor-negative CHO cells. MeV-RBP and F formed more syncytia in CHO cells upon human-CD150 (hCD150) co-transfection compared to dog-CD150 (dCD150) or bat-CD150 (bCD150) (Fig. 1a, top row). In contrast, MBaMV-RBP and F formed bigger and more numerous syncytia upon bCD150 overexpression than hCD150 or dCD150 (Fig. 1a, middle row). CDV-RBP and F formed extensive syncytia with both dCD150 and bCD150, and moderate syncytia with hCD150 and even mock-transfected cells (Fig. 1a, bottom row), suggesting a degree of promiscuity. We quanti ed these differential syncytia formation results on an image cytometer as described 15 (Fig. 1b).
We also evaluated the receptor usage of MBaMV in a VSV-pseudotype entry assay. VSV-DG[Rluc] bearing MeV-RBP and F entered hCD150-transfected CHO cells better than dCD150-, bCD150-, or mocktransfected cells (Fig. 1c) as expected. MBaMV-pseudotypes entered only bCD150-transfected CHO cells. CDV-pseudotypes showed good entry into dCD150-and bCD150-transfected, but not hCD150-transfected CHO cells. These results are generally consistent with our fusion assay results and support the species speci city of morbilliviruses. The promiscuity of CDV RBP for bCD150 suggest potential for epizoonotic transmissions from carnivores into some chiropteran species.
Generation of MBaMV by reverse genetics. Next, we attempted to generate a genomic cDNA clone of MBaMV that we could rescue by reverse genetics. We synthesized and assembled the putative MBaMV genome in increasingly larger fragments. Two silent mutations were introduced in the N-terminal 1.5 kb of the L gene to disrupt a cryptic open reading frame (Extended Data Fig. 3) that initially prevented cloning of the entire MBaMV genome. We introduced an additional EGFP transcription unit at the 3' terminus and rescued this MBaMV-GFP genome using the N, P, and L accessory plasmid from MeV (Extended Data Fig. 1a). MBaMV-GFP was initially rescued in BSR-T7 cells but passaged, ampli ed, and titered on Vero-bCD150 cells (Extended Data Fig. 4a). MBaMV formed GFP-positive syncytia containing hundreds of nuclei at 3 days post-infection (dpi) (Fig. 2a) and relatively homogenous plaques by 7 dpi (Fig. 2b). Transmission electron microscopy (TEM) (Fig. 2c)  Evaluation of receptor usage by MBaMV. To understand how well CD150 from various hosts supports MBaMV replication, we tested MBaMV growth in parental Vero-CCL81 cells and isogenic derivatives constitutively expressing CD150 of human, dog, or bat. MBaMV formed huge syncytia (Fig 3a) at 2 dpi in Vero-bCD150 cells and reached peak titers of ~10 5 PFU/ml at 3 dpi (Fig 3b). MBaMV showed moderate syncytia spread and growth in Vero-dCD150 cells but peak titers at 5 dpi was ~100-fold lower. No signi cant virus growth was detected in Vero or Vero-hCD150 cells. These results con rm that MBaMV can use bCD150 but not hCD150 for e cient cell entry and replication. MBaMV appears to use dCD150, albeit to a much lesser extent than bCD150.
MeV uses human nectin-4 as the epithelial cell receptor 17,18 which mediates e cient virus shedding from the affected host 12,19 . CDV also uses human nectin-4 e ciently for entry and growth 20 . To test if MBaMV can use human nectin-4 in an epithelial cell context, we evaluated the replication kinetics of MBaMV in human lung epithelial cells that express high (H441) or low (A549) levels of nectin-4 12,21 (Extended Data Fig. 4b). Surprisingly, MBaMV showed e cient virus spread (Fig. 3c) in H441 cells and reached 10 4 PFU/ml by 6 dpi (Fig. 3d). In contrast, MBaMV showed small GFP foci and 10 times lower titer in A549 cells. Comparing the Area Under Curve (AUC) revealed signi cant differences in this growth curve metric ( Fig. 3e). However, MeV still replicated to higher titers than MBaMV in H441 cells (Fig. 3d-e). This could be due to species speci c host factors or differences in interferon antagonism between human and bat morbilliviruses. Thus, we tested MBaMV versus MeV growth in interferon-defective Vero-human nectin-4 cells (Vero-hN4). MBaMV and MeV replicated and spread equally well on Vero-hN4 cells (Fig 3f-g), validating the ability of MBaMV to use human nectin-4, and suggesting that MBaMV may not have fully adapted to counteracting human innate immune responses.
Molecular characterization of MBaMV. To better understand the transcriptional pro le of MBaMV, we used Nanopore long-read direct RNA sequencing to sequence the mRNAs of MBaMV-infected Vero-bCD150 cells at 2 dpi (MOI=0.01). We found a characteristic 3'-5' transcriptional gradient where GFP>N>P>M>F>RBP>L (Extended Data Fig. 5a).
Morbilliviruses have a conserved intergenic motif (CUU) between the gene end and gene start of adjacent genes 'AAAA-CUU-AGG'. This intergenic motif was not immediately apparent in the long complex M-F intergenic region of the assembled MBaMV genome. However, the high coverage of this M-F intergenic region (M read-through transcripts) identi ed the M-F intergenic motif as 'CGU' instead of 'CUU' (Extended Data Fig. 5b).
The P gene of morbilliviruses is known to generate the V or W genes through the insertion of one or two guanines, respectively, at the conserved editing motif (AAAAGGG) 22 , which is present in MBaMV.
Amplicon sequencing of the P gene editing motif-from the same mRNA pool used above-revealed the frequency of P, V, and W mRNA is 42.1%, 51.2%, and 2.6%, respectively (Extended Data Fig. 5c), suggesting that the major interferon antagonist (V) is produced even in the absence of interferon.
We next evaluated the expression and cleavage of two surface glycoproteins (RBP and F). C-terminal AU-1 tagged F construct showed uncleaved F0 and cleaved F1 (Extended Data Fig. 5d). C-terminal HA tagged RBP construct showed monomer in addition to oligomers (Extended Data Fig. 5e). MBaMV-RBP showed smear above 110 kDa which is suggestive of oligomerization even in the reducing condition.
Species tropism of MBaMV. The two suborders of chiropterans (bats), Pteropodiformes (Yinpterochioptera) and Vespertilioniformes (Yangochiroptera), include more than 1,400 species grouped into 6 and 14 families, respectively 23 . Myotis bats belong to the prototypical Vespertilionidae family that is the namesake of its suborder. Jamaican fruit bats (Artibeus jamaicensis) belong to the same suborder as myotis bats, albeit from a different family (Phyllostomidae). We inoculated 6 Jamaican fruit bats available in a captive colony via two different routes with MBaMV to assess its pathogenicity in vivo. All bats remained asymptomatic and showed no evidence of developing systemic disease up to 3 weeks post-infection. Nor could we detect any molecular or serological evidence of productive infection (Extended Data Fig. 6). Inspection of Jamaican fruit bat and myotis CD150 sequences revealed key differences in the predicted contact surfaces with RBP (discussed below), which we speculate are responsible for the species-speci c restriction seen in our experimental challenge of Jamaican fruit bats with MBaMV.
To identify RBP-CD150 interactions likely involved in determining host species tropism, we compared the amino acid sequences at the putative contact surfaces of morbillivirus RBPs and their cognate CD150

Discussion
Metagenomic viral surveillance studies aided by next-generation sequencing have allowed scientists to monitor viruses circulating in animal species and identify potential zoonotic threats 31 . Surveillance of bat species has been particularly critical. For instance, >60 novel paramyxovirus sequences were identi ed in a 2012 bat surveillance study, several of which mapped to the Morbillivirus genus 4 . While comparing novel virus sequences to known pathogens may help inform the risks associated with future spillover events, this type of in silico modeling based on incomplete viral sequences needs to be complemented by functional characterization of such viruses. In this study, we identi ed a full-length morbillivirus genomic sequence from Myotis riparius bats in Brazil and generated an infectious virus clone using reverse genetics. With this approach, we circumvented the arduous process of isolating and culturing live virus directly from animals and instead produced MBaMV in the lab.

MBaMV characterized as a morbillivirus
Prior to this study, there were only 7 ICTV recognized morbilliviruses species, none of which were isolated from bats. While the annotated MBaMV genome aligned with the classic morbillivirus genome organization (N, P/V/C, M, F, RBP, and L), it was important to verify that virus generated by reverse genetics successfully recapitulated morbillivirus biology. Fusion assays and entry experiments con rmed that MBaMV preferentially used myotis CD150 over human or dog CD150 to enter transgenic Vero cells (Fig. 3), which ts the paradigm that CD150 is the major determinant of host speci city for morbilliviruses. We also assessed P-editing-a hallmark of paramyxoviruses-and found RNA editing of P-mRNA, creating V-mRNA (single G insertion) or W-mRNA (double G insertion) of MBaMV. Interestingly, the proportion of V-mRNA at 51.2% of total P transcripts is unusually high for orthoparamyxoviruses, resembling the now extinct rinderpest virus (RPV) more than extant morbilliviruses 32 .
In their natural hosts, morbillivirus are highly pathogenic and can cause deadly acute infections 33 . Thus, a reasonable prediction is that MBaMV would cause visible disease in the bat host. However, when we challenged Jamaican fruit bats with MBaMV, we found the virus was not able to cause systemic disease in the bats (Extended Data Fig. 6) and there was no evidence that MBaMV productively infected these bats. This lack of infection is likely due to the CD150 differences between the species-CD150 of Jamaican fruit bats and Myotis species is only 70% conserved on the amino acid level (Extended Fig. 8).
We predict that MBaMV infection is more likely to cause serious disease in the Myotis riparius species.
Potential zoonotic threat of MBaMV based on receptor usage While non-human morbilliviruses are not currently known to jump the species barrier and infect humans, we did nd that MBaMV was able to utilize human receptors in vitro to a certain extent. Notably, MBaMV replicated well in H441 cells and in Vero cells expressing human nectin-4 (Fig. 3). CDV is also reported to use human nectin-4 20 and can replicate in H358 cells 34 . Alarmingly, there have been several outbreaks of CDV in non-human primates, resulting in acute disease or death in the animals 35 . In one outbreak, mutations were found in the RBP which rendered CDV-RBP capable of e ciently using primate-CD150 20 . However, CDV is unlikely to adapt to humans in the presence of cross-reactive MeV immunity. Whether such cross-reactivity extends to MBaMV remains to be seen. Traditionally, morbilliviruses use CD150 to enter myeloid and lymphoid cells. However, unlike MeV which infects human macrophages via CD150, MBaMV infects human macrophages in a CD150-independent manner (Fig. 4c) Table S1.

Plasmids
We cloned the open reading frame of hCD150, dCD150, and bCD150 (from Myostis brandtii since the CD150 sequence from M. riparius is unknown) into the pCAGGS vector cut by EcoRI (NEB) and NheI-HF (NEB). We introduced HA tag-linker-Igk signal peptides (amino acids corresponding to; MVLQTQVFISLLLWISGAYG-YPYDVPDYA-GAQPARSP) at the N-terminus of CD150s as previously For MeV RBP and F expressing plasmid, we ampli ed RBP and F sequence from p(+) MV323-AcGFP with the addition of HA-tag and AU1-tag same as MBaMV-RBP and -F, creating pCAGGS-MeV-RBP-HA, pCAGGS-MeV-F-AU1. For CDV RBP and F cloning, we ampli ed RBP and F sequence from pCDV-5804P plasmid with the addition of HA-tag and AU1-tag, creating pCAGGS-CDV-RBP-HA, pCAGGS-CDV-F-AU1.
Genome coding plasmids for MeV; (p(+) MV323-AcGFP) and CDV; pCDV-5804P were kindly gifted from Dr. Makoto Takeda 42 and Dr. Veronica von Messling respectively 43 . We transferred the MeV genome sequence into pEMC vector, adding an optimal T7 promotor, a hammer head ribozyme, and we introduced an eGFP transcriptional unit at the head of the genome (pEMC-IC323-eGFP), which is reported in the previous study 15 . The cells were trypsinized and passed onto Vero-bCD150 cells (2.0x10 6 cells / ask in one 75cm 2 ask.).
We collected supernatant 2 days after overlay and reampli ed MBaMV in fresh Vero-bCD150 cells.
The rescue of MeV was done exactly same way as MBaMV rescue except that 5 mg of pEMC-IC323eGFP was used for transfection and Vero-hCD150 cells were used for coculturing.

Titration of viruses and plaque assay
For MBaMV, a monolayer of Vero-bCD150 cells in 12 well was infected by 500 ml of serially diluted samples for 1 hour, followed by medium replacement with methylcellulose containing DMEM. 5 dpi, the number of GFP positive plaque was counted to determine titer. For the plaque assay, infected Vero-bCD150 cells were incubated under methylcellulose containing DMEM for 7 days. Cells were then stained with 1% crystal violet and 1% neutral red sequentially. For MeV, we used Vero-hCD150 cells and xed the plates at 4dpi. Soluble CD150 production and puri cation Production and puri cation of soluble CD150 is as previously reported 46 . Soluble CD150 is a chimera comprising the human V (T25 to Y138) and mouse C2 domains (E140 to E239) + His6-tag, which was cloned into pCA7 vector. The expression plasmid was transfected by using polyethyleneimine, together with the plasmid encoding the SV40 large T antigen, into 90% con uent HEK293S cells lacking Nacetylglucosaminyltransferase I (GnTI) activity. The cells were cultured in DMEM (MP Biomedicals), Analysis was completed using FCSExpress-7. A total of 2 donors were utilized for this analysis Western blot for RBP and F protein 1 x 10 6 of 293T cells were seeded on to collagen coated 6 well plate. 293T cells were transfected by 2 mg of pCAGGS, pCAGGS-MBaMV-RBP-HA, or pCAGGS-MBaMV-F-AU1 using polyethylenimine max (polysciences Bat challenge experiment and evaluation of infection. Six Jamaican fruit bats (Artibeus jamaicensis) were inoculated with 2x10 5 PFU MBaMV-eGFP; three bats were intranasally (I.N.) and 3 bats were intraperitoneally (I.P.). At 1 week post virus inoculation, bats were subjected to blood and serum collection, visually inspected for GFP expression around the nares, oral cavity, and eyes by LED camera in each group (I.N. and I.P.). At 2 weeks post virus infection, blood, serum, and tissues (lung, spleen, and liver) were collected from one bat in each group. At 3 weeks post virus infection, blood, serum, and tissues (lung, spleen, and liver) were collected from one bat in each group. Transmission electron microscopy (TEM) Routine transmission electron microscopy processing was done as described 50 . The Vero-bCD150 cells infected by MBaMV for 3 days were washed with phosphate-buffered saline and then xed with 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.4) on ice for 1 hour. The cells were scraped off the 100 mm tissue culture treated petri dish and pelleted by low-speed centrifugation (400g for 5 minutes). The pellet was xed for 30 minutes with the same xative before secondary xation with 2% osmium tetraoxide on ice for 1 hour. The cells were then stained with 2% uranyl aqueous solution en bloc for 1 hour at room temperature, dehydrated with a series of increasing ethanol gradients followed by propylene oxide treatment, and embedded in Embed 812 Resin mixture (Electron Microscopy Sciences). Blocks were cured for 48 h at 65°C and then trimmed into 70 nm ultrathin sections using a diamond knife on a Leica Ultracut 6 and transferred onto 200 mesh copper grids. Sections were counterstained with 2% uranyl acetate in 70% ethanol for 3 min at room temperature and in lead citrate for 3 minutes at room temperature, and then examined with a JEOL JSM 1400 transmission electron microscope equipped with two CCD camera for digital image acquisition: Veleta 2K x 2K and Quemesa 11 megapixel (EMSIS, Germany) operated at 100 kV.
Animal study was performed following the Guide for the Care and Use of Laboratory Animals. Animal experiment was approved by the Institutional Animal Care and Use Committee of Colorado State University (protocol number 1090) in advance and conducted in compliance with the Association for the Assessment and Accreditation of Laboratory Animal Care guidelines, National Institutes of Health regulations, Colorado State University policy, and local, state and federal laws.

Human subjects research
Normal primary dendritic cells and macrophages used in this project were sourced from 'human peripheral blood Leukopack, fresh' which is provided by the commercial provider New York Blood center, inc. Leukapheresis was performed on normal donors using Institutional Review Board (IRB)-approved consent forms and protocols by the vendor. The vendor holds the donor consents and the legal authorization that should give permission for all research use. The vendor is not involved in the study design and has no role in this project. Samples were deidenti ed by the vender and provided to us. To protect the privacy of donors, the vendor doesn't disclose any donor records. If used for research purposes only, the donor consent applies.
Data and materials availability: The raw next generation sequencing results of bat surveillance, P gene editing, and transcriptome by MinION are uploaded at NCBI GEO: GSE166170, GSE166158, and GSE166172, respectively. values are from ordinary one-way ANOVA with Dunnett's multiple comparisons test. c, VSV-pseudo particle (pp) entry assay showed similar trends. Adjusted p values obtained as in (b) but only for comparing groups at the highest viral inoculum used (10-1 reciprocal dilution).