A bat MERS-like coronavirus circulates in pangolins and utilizes human DPP4 and host proteases for cell entry

Abstract

It is unknown whether pangolins, the most trafficked mammals, play a role in the zoonotic transmission of bat coronaviruses. We report the circulation of a novel MERS-like coronavirus in Malayan pangolins, named Manis javanica HKU4-related coronavirus (MjHKU4r-CoV). Among 86 animals, four tested positive by pan-CoV PCR, and seven tested seropositive (11 and 12.8%). Four nearly identical (99.9%) genome sequences were obtained, and one virus was isolated (MjHKU4r-CoV-1). This virus utilizes human dipeptidyl peptidase-4 (hDPP4) as a receptor and host proteases for cell infection, which is enhanced by a furin cleavage site that is absent in all known bat HKU4r-CoVs. The MjHKU4r-CoV-1 spike shows higher binding affinity for hDPP4, and MjHKU4r-CoV-1 has a wider host range than bat HKU4-CoV. MjHKU4r-CoV-1 is infectious and pathogenic in human airways and intestinal organs and in hDPP4-transgenic mice. Our study highlights the importance of pangolins as reservoir hosts of coronaviruses poised for human disease emergence.

Keywords: bat MERS-like coronavirus; dipeptidyl peptidase-4; furin cleavage site; pangolin.

Graphical abstract

Figure S1

Genome characterization and phylogenetic analysis of MjHKU4r-CoV, related to Figure 1 (A) MjHKU4r-CoV-2, -3, and -4 genome sequences mapped to the reference sequence, MjHKU4r-CoV-1. SNPs are displayed on the genomes. (B and C) Phylogenetic trees based on the RdRp nucleotide sequences (B) or complete S gene nucleotide sequences (C) of representative Alpha and Beta coronaviruses. MjHKU4r-CoVs are shown in red. Different color shades represent different subgenera of coronaviruses. The scale bars represent 0.2 or 0.3 substitutions per nucleotide position. The software and setting used are described in the STAR Methods.

Figure 1

Genome characteristics of pangolin MERS-like CoV, Manis javanica HKU4-related CoV (MjHKU4r-CoV) (A) Genome structure of MjHKU4r-CoV-1. (B) Similarity plot based on the full-length genome sequences. MjHKU4r-CoV-1 was used as a query sequence, Tylonycteris bat CoV HKU4-1, Pipistrellus bat CoV HKU5-1, bat MERS-related NeoCoV, and human MERS-CoV were used as reference sequences. (C) Phylogenetic tree based on the nucleotide sequences of the complete genomes of representative Alpha coronavirus (Alpha-CoV) and Beta coronavirus (Beta-CoV). MjHKU4r-CoV-1–4 are shown in red. Shaded colors represent different subgenera of coronaviruses. The scale bars represent 0.3 substitutions per nucleotide position. The software and setting used are described in the STAR Methods. (D) Schematic representation of the S protein. S1 and S2 subunits are indicated, as well as four domains within S1, including the N-terminal domain (NTD), RBD, subdomain 1 (SD1), and subdomain 2 (SD2). RBM alignment is shown at the bottom left, and the 16 key residues situated at the surface between MERS-CoV RBM and human DPP4 are pink shaded. Predicted furin cleavage sites of MjHKU4r-CoV-1 and the corresponding sites in other merbecoviruses are shown at the bottom right. The key arginine sites are shown in red, and the predicted furin cleavage sites are gray shaded. See also Figures S1, S2, and S3 and Table S1.

Figure S2

Serological investigation through LIPS and microneutralization test (MNT) assays, related to Figure 1 (A) MjHKU4r-CoV NP antibodies in 80 pangolin serum samples were detected using LIPS (the only positive sample in the MNT assay is shown in red). (B and C) Pangolin sera were tested for neutralizing antibodies against MjHKU4r-CoV-1 by the MNT assay at 1:10 and 1:20 dilutions. IF staining of a pangolin serum sample positive for neutralization antibody at 1:10 and 1:20 dilutions is shown. Mock-infected and no-serum infection were included as controls (B). Neutralizing activity was quantified using ImageJ (C).

Figure S3

Prediction and alignment of putative furin cleave sites in pangolin MjHKU4r-CoV and bat HKU4r-CoVs, related to Figure 1 (A) Prediction of furin cleavage sites in the S proteins of MjHKU4r-CoV-1 and bat HKU4 was carried out using ProP-1.0 Server (https://services.healthtech.dtu.dk/service.php?ProP-1.0), using the furin-specific prediction as the default. (B) Predicted furin cleavage site of MjHKU4r-CoV and the corresponding sites in all publicly available bat HKU4r-CoVs. The arginine sites are shown in red, and the predicted furin cleavage sites are gray shaded.

Figure 2

Isolation of MjHKU4r-CoV-1 and elucidation of its cell-entry mechanism (A) Transmission electron microscopy analysis of MjHKU4r-CoV-1 virions. Scale bars, 50 nm. (B) Cytopathic effect in Caco-2 cells at 48 h.p.i. with MjHKU4r-CoV-1 or mock infected (left, scale bars, 400 μm), and immunofluorescence assay (IFA) with an antibody against the Tylonycteris bat coronavirus HKU4 NP (right, scale bars, 200 μm). Red, NP; blue, nuclei. (C) Caco-2 cells were infected with MjHKU4r-CoV-1 at a multiplicity of infection (MOI) of 1 or 0.01. Viral titers and viral RNA copies at indicated time points were determined by TCID₅₀ assay and qRT-PCR, respectively. (D–G) Viral infection in Huh-7 cells with or without hDPP4 expression. Wild-type (WT) or DPP4-knockout Huh-7 (KO) cells were infected with MjHKU4r-CoV-1 or MERS-CoV for 24 h. IFA was performed (D, scale bars, 200 μm) and quantified by high content screening (E), and viral titer (F) and RNA copy number (G) were determined in the supernatant. (H–K) Viral infection in KO cells after hDPP4 re-introduction. IF staining in Huh-7 KO cells transfected with hDPP4 expression plasmid or empty vector (H, scale bars, 200 μm) quantified by high content screening (I), and viral titer (J) and RNA copies (K) were determined in the supernatant at the indicated times. Data are presented as means and standard errors of the means (SEMs) of at least triplicate measurements in (E–G and I–K). Statistical significance was assessed using a two-tailed Student’s t-test in (E and I) and two-way ANOVA in (F, G, J, and K). See also Figure S4.

Figure S4

Mapping analysis of NGS raw reads and metagenomics analysis of MjHKU4r-CoV-1 cell culture, related to Figure 2 (A) High-throughput sequencing was performed using RNA extracted from supernatants of MjHKU4r-CoV-1-infected Caco-2 cell cultures. Reads randomly extracted from NGS raw data of isolated MjHKU4r-CoV-1 were mapped to the reference sequence obtained from an anal swab sample. (B) Metagenomics analysis of sequencing data from MjHKU4r-CoV-1 viral supernatant.

Figure 3

Effect of host protease on viral entry and S protein cleavage of MjHKU4r-CoV-1 (A and B) Proteolytic cleavage dependency in human cells. Huh-7 cells were pretreated with the indicated protease inhibitors or exogenous trypsin for 2 h before infection with MjHKU4r-CoV-1 at a MOI of 10. Cells were stained by IFA at 8 h.p.i. The cells were pretreated with 10 μM furin inhibitor decRVKRcmk or water as a control; 50 μM serine-class protease inhibitor camostat or 0.5% dimethylsulfoxide (DMSO) as control; 50 μM cysteine-class protease inhibitor E64D or 0.1% DMSO as a control; or 0.5 μg/mL trypsin or Dulbecco’s modified Eagle’s medium (DMEM) as control. Data are shown as IF images (A, scale bars, 400 μm) or IF quantification (B). (C) Mutations introduced into the potential furin cleavage site in MjHKU4r-CoV-1 S protein. (D) Structure model for MjHKU4r-CoV-1 S protein predicted based on the PDB: 5X5C structure as a template using the SWISS-MODEL online tool (swissmodel.expasy.org). The predicted furin cleavage site is shown in red. (E) Western blot analysis of S protein cleavage products during protein biosynthesis. HEK293 cells were transfected with MERS-CoV, wild-type MjHKU4r-CoV-1, mutant MjHKU4r-CoV-1, or bat HKU4-CoV S protein expression plasmid containing a C-terminal S-tag. Empty vector transfection served as a negative control. At 48 h.p.t., the cells were harvested, lysed, and subjected to western blotting using the anti-S-tag antibody. S protein and cleaved S are indicated with point markers. (F) Huh-7 and Caco-2 cells were infected with pseudoviruses harboring MERS-CoV, wild-type MjHKU4r-CoV-1, mutant MjHKU4r-CoV-1, or HKU4-CoV S protein carrying a detectable luciferase. Entry efficiency was compared by measuring luciferase activity. (G) Analysis of pseudovirus package. Medium containing pseudovirions harboring MERS-CoV, wild-type MjHKU4r-CoV-1, mutant MjHKU4r-CoV-1, or HKU4-CoV S protein was harvested and ultra-centrifuged, followed by western blot analysis. Detection of vesicular stomatitis virus matrix protein (VSV-M) served as control. (H) Analysis of furin-mediated S protein processing. MERS-CoV, wild-type MjHKU4r-CoV-1, mutant MjHKU4r-CoV-1, or HKU4-CoV S protein expression plasmid was co-transfected with empty vector or furin expression plasmid into HEK293 cells in the presence or absence of a furin inhibitor (10 μM decRVKRcmk). At 48 h.p.t., the cells were lysed, and S protein cleavage was analyzed by western blotting using the anti-S-tag antibody. S protein and cleaved S are indicated with point markers. (I) Effect of furin cleavage on viral entry mediated by S protein. MERS-CoV, wild-type MjHKU4r-CoV-1, mutant MjHKU4r-CoV-1, or bat HKU4 pseudovirus was produced in the presence or absence of a furin inhibitor (10 μM decRVKRcmk) and used to infect Huh-7 cells or Caco-2 cells. Viral entry was quantified by measuring luciferase at 48 h.p.i. Data are presented as means and SEMs of at least triplicate measurements in (B, F, and I). Statistical significance was assessed using a two-tailed Student’s t test.

Figure 4

MjHKU4r-CoV-1 binds to human, bat, and pangolin DPP4 proteins (A) HeLa cells overexpressing the following DPP4 orthologous: hDPP4, Tylonycteris pachypus bat DPP4 (TpDPP4), Pipistrellus bat DPP4 (PpDPP4), or Manis javanica DPP4 (MjDPP4) were infected with MjHKU4r-CoV-1 at a MOI of 1 and subjected to IF staining at 24 h.p.i. Shown as MjHKU4r-CoV-1 NP or DPP4 protein expression (left, scale bars, 10 μm), or the replication dynamics of MjHKU4r-CoV-1 (viral titer, middle panel; genomic RNA, right panel). For IF staining, green, DPP4; red, NP; blue, nuclei. (B) Binding affinity of MjHKU4r-CoV-1, MERS-CoV, and HKU4-CoV RBD proteins to human, bat, and pangolin DPP4 proteins as measured by Bio-layer interferometry assay. The quantification of the binding affinity is shown in Table S2. See also Figure S5 and Table S2.

Figure S5

Functional assessment of different DPP4 orthologs mediating MjHKU4r-CoV-1 infection, related to Figure 4 HeLa cells overexpressing DPP4 orthologs from various mammals or empty vector were infected with MjHKU4r-CoV-1 (MOI = 1). At 24 h.p.i., the cells were fixed and subjected to IF staining. MjHKU4r-CoV-1 NP (red), DPP4 protein expression (green), and nuclei (blue) are shown (scale bars, 10 μm).

Figure S6

Human cell tropism and human colon ex vivo infection of MjHKU4r-CoV-1, related to Figure 5 (A) NP staining (red) of the indicated human cell lines at 24 h and 72 h.p.i. with MjHKU4r-CoV-1. Blue, nuclei; scale bars, 200 μm. Replication of MjHKU4r-CoV-1 in cell culture supernatant was analyzed by TCID₅₀ assay. (B–D) qRT-PCR detection of the replication of MjHKU4r-CoV-1 in human ex vivo colon tissue (B) or culture supernatant (C). IF staining of human ex vivo colon at 24 h.p.i. with MjHKU4r-CoV-1 (D). Red, viral NP; green, human DPP4; blue, nuclei; scale bars, 100 μm.

Figure 5

MjHKU4r-CoV-1 infection in human colon organoids and airway organoids (A–C) Replication of MjHKU4r-CoV-1 in human colon organoids (hCO), shown as viral copy numbers in cells (A) and supernatants (B) and viral titers in supernatants (C) as analyzed by qRT-PCR and TCID₅₀ assay, respectively. (D) Co-staining of MjHKU4r-CoV-1 NP (red) and hDPP4 (green) in human colon organoids at 72 h.p.i. (scale bars, 100 or 30 μm). Blue, nuclei. (E–G) Replication of MjHKU4r-CoV-1 in human airway organoids (hAWOs), shown as viral copy numbers in cells (E) and supernatants (F) and viral titers in supernatant (G) as analyzed by qRT-PCR and TCID₅₀ assay, respectively. (H) Co-staining of MjHKU4r-CoV-1 NP (red) and hDPP4 (green) in human airway organoids at 48 h.p.i. (scale bars, 50 or 10 μm). Blue, nuclei. See also Figure S6.

Figure 6

MjHKU4r-CoV-1 is infectious in hDPP4-Tg mice (A and B) hDPP4-Tg mice were intranasally inoculated with 1 × 10⁶ TCID₅₀ MjHKU4r-CoV-1. (A) Mean body weights (n = 7 for MjHKU4r-CoV-1 infected mice and n = 3 for mock-infected mice), (B) body weights of individual mice. (C and D) qRT-PCR detection of viral replication in the lung (C) and brain (D) of MjHKU4r-CoV-1-infected hDPP4-Tg mice at the indicated time points (n = 4 for 1, 3, and 5 d.p.i. and n = 7 for 8 d.p.i.). (E) Lung and brain sections from MjHKU4r-CoV-1-infected or mock-infected mice stained for MjHKU4r-CoV-1 NP (red) at 5 d.p.i. (lung) and 8 d.p.i. (brain) (scale bars, 100 μm). Red, viral NP; blue, nuclei. (F) Pathological changes in the lungs of MjHKU4r-CoV-1-infected hDPP4-Tg mice at 5 d.p.i. Yellow arrows indicate severely affected areas (scale bars, 1,000 or 100 μm). See also Figure S7.

Figure S7

Evasion of the type I IFN response by MjHKU4r-CoV-1, related to Figure 6 (A and B) Caco-2 cells were infected with MjHKU4r-CoV-1 (at a MOI of 2). At 6, 12, 24, and 48 h.p.i., viral RNA copies in cell lysates were determined by qRT-PCR (A), and viral titers in the supernatant were determined by TCID₅₀ assay (B). (C and D) Caco-2 cells infected with MjHKU4r-CoV-1 (at a MOI of 2), Sendai virus (100 hemagglutinating units/mL), or mock infected were collected at indicated times. Total RNA extracted from cells was used to evaluate IFNB-β (C) and ISG15 (D) mRNA levels, shown as fold changes relative to the GAPDH level. (E) HEK293 cells were co-transfected with viral protein expression plasmids or empty vector, simulator plasmid RIG-I (2CARD), firefly luciferase reporter plasmid pIFN-β-luc, and Renilla luciferase plasmid pRL-TK. At 24 h.p.t., luciferase activity was measured. Empty vector was set to 100%. NS1 protein of influenza A virus served as a positive control. (F) HEK293 cells were co-transfected with viral protein expression plasmids or empty vector, firefly luciferase reporter plasmid pISRE-luc, and Renilla luciferase plasmid pRL-TK. At 24 h.p.t., the cells were treated with 500 U/mL universal IFN-α for another 8 h and luciferase activity was measured. Empty vector was set to 100%. NS1 protein of influenza A virus served as a positive control. Data are presented as means and SEMs of at least triplicate measurements in (E) and (F). Statistical significance was assessed using a two-tailed Student’s t test.

Figure 7

Therapeutic treatments against MjHKU4r-CoV-1 infection (A) Nsp5 and nsp12 amino acid similarity between MjHKU4r-CoV-1, MERS-CoV, and SARS-CoV-2. (B) Partial amino acid sequence alignment of MjHKU4r-CoV-1, MERS-CoV, and SARS-CoV-2 for nsp5 and nsp12 proteins. GC376 and remdesivir target sites are shaded pink. (C and D) Antiviral activities of small molecules against MjHKU4r-CoV-1 (C) and MERS-CoV (D), as determined by qRT-PCR. (E and F) Neutralization efficacy of human monoclonal antibodies (targeting the MERS-CoV RBD) against MjHKU4r-CoV-1 (E) and MERS-CoV (F), as quantified by high content screening. Details can be found in STAR Methods. (G) Partial S protein amino acid sequence alignment of MjHKU4r-CoV-1, MERS-CoV England 1, and MERS-CoV EMC/2012. The binding sites in RBD of MERS-CoV England 1 with m336 and the corresponding sites in MERS-CoV EMC/2012 and MjHKU4r-CoV-1 are shaded pink. Data are presented as means and SEMs of at least triplicate measurements in (C) and (D). Statistical significance was assessed using a two-tailed Student’s t test.