3. Results 3.1 Illness and clinical features of sick pangolins Between July and August in 2018, four sick wild pangolins were sent to the Jinhua Wildlife Rescue Station of Zhejiang province, China (Fig. 1). For simplicity, they were referred to here as ‘1-Dongyang’, ‘2-Lishui’, ‘3-Ruian’, and ‘4-Wucheng’, according to the location found. Morphological examination and molecular identification revealed that these pangolins belong to the Sunda pangolin, Manis javanica (1-Dongyang, 2-Lishui, and 4-Wucheng) and the Chinese pangolin, Manis pentadactyla (3-Ruian). The details and clinical signs exhibited by these pangolins are described in Table 1. All four pangolins exhibited anorexia when sent to the rescue station, and twitching and slobbering behavior were observed in 1-Dongyang and 2-Lishui. Hemorrhage and skin lesions were obvious in 1-Dongyang, whereas edema on the front legs was found in 2-Lishui. Finally, pangolin 4-Wucheng appeared febrile and contained skin lesions, whereas pangolin 3-Ruian exhibited relatively mild clinical signs. Table 1. Background information, clinical features, and examination of the rescued pangolins. Pangolin ID Species Gender/age State Clinical features Clinical examination X-ray Hematological analysis 1-Dongyang M. javanica Male/adult Dead Anorexia, twitch, slobber, hemorrhage,a and skin lesionsb Normal PLT ↓, GLU ↓, cholesterol ↓, and amylase ↑ 2-Lishui M. javanica Female/adult Dead Anorexia, twitch, slobber, and edemac Large area of shadow on the left lung ALT ↑, AST ↑, amylase ↑, and total bilirubin ↓ 3-Ruian M. pentadactyla Female/juvenile Alive Anorexia, anxiety, and edemac – – 4-Wucheng M. javanica Female/adult Dead Anorexia, fever, and skin lesionsb Stones in the stomach PLT ↓, ALP ↑, BUN ↑ a On the left ear. b On the face and paws. c On the front legs. PLT, platelet count; GLU, glucose; AST, aspartate aminotransferase; ALT, alanine aminotransferase; BUN, blood urea nitrogen; ALP, alkaline phosphatase. X-ray tests revealed a large area of shadow in the left lung of pangolin 2-Lishui, suggestive of pneumonia (Supplementary Fig. S1). Due to the lack of healthy pangolin as control in this study, we used the blood biochemical values of healthy Formosan pangolins (Chin et al. 2015) as a reference. 1-Dongyang showed a significant decrease in platelet count, glucose, and cholesterol, whereas 2-Lishui exhibited a significant increase in alanine aminotransferase, aspartate aminotransferase, and amylase levels, but a decrease in total bilirubin. In addition, a lower PLT count and higher levels of alkaline phosphatases and blood urea nitrogen were observed in 4-Wucheng (Table 1 and Supplementary Table S4). Although these pangolins received careful resuscitation in the rescue station, only the pangolin 3-Ruian recovered from illness. Pangolins 1-Dongyang, 2-Lishui, and 4-Wucheng died on 3, 16, and 5 days after the admission, respectively. 3.2 Pathologic changes in dead pangolins Congestion was observed in the liver and lung of pangolin 1-Dongyang following autopsy, although no obvious pathological changes were observed in other inner organs. Consistent with the obvious congestion, histological tests revealed hemorrhage in the liver and lung (Supplementary Fig. S2). In addition, pyknotic nucleus of the hepatic cells, lymphocyte pyknosis in spleen, necrosis of the hepatic plate, and glomerular necrosis were observed. Finally, collagen fiber degeneration was observed in the liver, spleen, and lung (Supplementary Fig. S2). For pangolin 2-Lishui, obvious necrosis was observed in the lung, kidney, and spleen, and congestion was observed in the liver (Supplementary Fig. S2) in addition to mesenteric lymphadenopathy. Notably, many milky white lesions were present in both lungs, especially in the left lower lobe (1 × 2 cm). Histological tests revealed necrosis in the liver, spleen, lung, kidney, trachea, and small intestinal, as well as hemorrhage in the lung (Supplementary Fig. S2). 3.3 Identification of viral agents by meta-transcriptomics and PCR To identify the possible etiologic agents of disease in the four pangolins, eight meta-transcriptomic libraries from blood, liver, spleen, lung, kidney, and fecal samples were constructed, generating a total of 306,908,179 paired-end sequence reads. De novo assembly revealed the high abundance of a pestivirus- and coltivirus-like virus in all the meta-transcriptomic libraries of the pangolin 1-Dongyang and 2-Lishui, representing 6–80 and 1–29 per cent of total viral contigs, respectively (Supplementary Table S2). Notably, despite the presence of other putative viruses (Supplementary Table S2), only the pestivirus- and coltivirus-like virus could be successfully identified by PCR. As those putative viral sequences could not be confirmed by PCR and probably were from contamination; hence, they were not possible etiologic agents of disease in four pangolins. Additional assays revealed that the novel pestivirus was only present in 1-Dongyang, whereas the novel coltivirus was only present in 2-Lishui. No potential bacterial and fungal pathogens were found in the libraries generated from 1-Dongyang and 2-Lishui. Finally, no abundant viral, bacterial and fungal sequences were found in the libraries generated from blood and tissue samples of 3-Ruian and 4-Wucheng (Supplementary Table S2). Genetic analysis revealed that the novel pestivirus shared <57 per cent nt similarity to known pestiviruses, while the novel coltivirus showed <68 per cent nt similarity to known coltiviruses. Considering that they are related, yet clearly genetically distinct, from known members of Pestivirus and Coltivirus (see below), we designated these two newly identified viruses as Dongyang pangolin virus (DYPV) and Lishui pangolin virus (LSPV), respectively, reflecting their hosts species and the geographic location of sampling. Attempts at virus isolation by cell culture (BHK-21, Vero-E6, and DH82) and inoculation of suckling mice proved unsuccessful. To determine the distribution of the newly identified viruses, each organ was tested by PCR. Consequently, DYPV was identified in the heart, liver, spleen, lung, kidney, brain, blood, throat swab, and fecal sampled from pangolin 1-Dongyang, as well as the nymph ticks (Amblyomma javanense) collected from this animal. Interestingly, the virus was not identified in the adult ticks also sampled from 1-Dongyang. Similarly, LSPV was also identified in a broad range of tissue organs including heart, liver, spleen, lung, kidney, blood, throat swab, and fecal matter of 2-Lishui. 3.4 Genetic features of DYPV and LSPV To further characterize both viruses, primers were designed based on the sequences obtained here and those of related viruses described previously. In this manner we were able to recover the complete genomes of DYPV from 1-Dongyang and the ticks (A. javanense) collected from this animal. The genetic features of DYPV are described in Fig. 2a. Notably, viruses obtained from the pangolin and ticks were closely related to each other, although still exhibited almost 2 per cent nt sequence difference. In addition, both viruses were genetically and phylogenetically distinct from known viruses of the genus Pestivirus, with >43 per cent nt and >48 per cent aa differences (Supplementary Table S5). As with classical pestiviruses, the genome of DYPV encodes twelve proteins. All known cleavage sites of the NS3 protease were observed in DYPV (Supplementary Fig. S3). In addition, the NS4A-NS4B cleavage site, located at Position L2426/A2427 in BVDV1 and L2336/A2337 in CSFV, appeared at Position L2315/K2316 site in DYPV (Supplementary Fig.S3). Figure 2. Schematic of the annotated DYPV and LSPV genomes. (a) Genome comparison between DYPV and known pestiviruses. (b) Genome comparison between LSPV and known coltiviruses. Notably, despite the use of both meta-transcriptomic and PCR methods, only nine genome segments (1–5 and 8–11) were obtained from the novel coltivirus—in animal 2-Lishui—and our attempt to recover segments 6, 7, and 12 failed. Details of the genetic features of LSPV are described in Fig. 2b. All recovered segments had the consensus sequences (GAG/AUU/A) at the 5ʹ-terminus and (G/CAGUC) and the 3ʹ-terminus (Fig. 3a), respectively. However, the genome sequences of LSPV were distinct from those of recognized coltiviruses, with <75 per cent aa similarity (Supplementary Table S6). Like other coltiviruses, LSPV Segment 9 also contained two ORFs. However, the sequences adjacent to the UGA stop codon, which are the signal of read-through, were different from known coltiviruses. Notably, however, a similar pattern was observed in SHLV (Fig. 3b), the closest relative of LSPV, that was sampled from ticks in Australia. Figure 3. Characteristics of the LSPV genome. (a) Sequence conservation of the 5ʹ- and 3ʹ-terminal 10 nts in genomic segments of the LSPV genome were analyzed and visualized using Weblogo. (b) Sequence information around the stop codon of segment VP9 of LSPV and known coltivirus. 3.4 Phylogenetic relationships of the newly identified viruses To determine the evolutionary relationships among the newly identified and previously recognized viruses, we estimated ML phylogenetic trees based on aa sequences (Figs 4 and 5). Figure 4. ML trees based on aa sequences of the entire coding sequences (polyprotein) and NS3 genes of DYPV and other known pestiviruses. The numbers at nodes indicate bootstrap support values after 1,000 replications. Bootstrap values higher than 70 per cent were considered significant and shown on the trees. Figure 5. ML trees based on aa sequences of the RdRp genes and putative RNA methyltransferase (VP2) genes of LSPV and other known coltiviruses. The numbers at nodes indicate bootstrap support values after 1,000 replications. Bootstrap values higher than 70 per cent were considered significant and shown on the trees. In phylogenies based on the aa sequences of the entire coding sequences and the NS3 genes, DYPV formed a highly distinct lineage, confirming that it represents a novel pestivirus. Notably, the DYPV lineage occupied the basal position within the Artiodactylous lineage, with those pestiviruses present in rodents and bats forming more divergent lineages. In the phylogenies based on the aa sequences of the RdRp gene and putative RNA methyltransferase gene (VP2; Fig. 5), LSPV also formed a distinct lineage within the genus Coltivirus. Interestingly, however, the virus was most closely related to Shelly headland identified in ticks (I. holocyclus) sampled from the long-nosed bandicoot (a marsupial) in Australia (Harvey et al. 2018). 3.5 Phylogenetic analysis of pangolin sequences The Sunda pangolin (M. javanica) is geographically distributed widely in South-East Asia, including Indonesia (Java, Sumatra, Borneo, and the Lesser Sunda Islands), Malaysia, Singapore, The Philippines, Thailand, Vietnam, Laos, and Cambodia (Zhang et al. 2015), whereas Chinese pangolins (M. pentadactyla) are relatively commonplace in Southern China (Choo et al. 2016). To determine the likely geographic origin of these sick pangolins, sequences of mt-cyt b gene were amplified from their tissue. Genetic analysis revealed that all the sequences obtained from Sunda pangolins in this study fell into the M. javanica group, while the Chinese pangolins clustered with those of M. pentadactyla. Notably, three Sunda pangolins sampled here were very closely related to those from Indonesia, Malaysia, and Thailand and Singapore, with 99.9, 99.7, and 99.9 per cent nt sequence identity, respectively, indicating that they were most likely (illegally) imported into China from abroad. In contrast, the pangolin 3-Ruian (M. pentadactyla) clustered together with those from Taiwan/China, with 99.1 per cent nt sequence identity, suggesting that this animal was not imported. As no sequences related to the A. javanense mitochondrial 16S rDNA were available, we could not determine the origin of the A. javanense ticks collected from sick Sunda pangolins. Hence, a systemic effort should be considered to establish comprehensive databases for the speciation of arthropod vectors and as a tool for determining the geographic origin of the collected arthropods. 3.6 Molecular investigation of DYPV and LSPV in local ticks As DYPV was identified in ticks (A. javanense) collected from pangolin 1-Dongyang, and LSPV was closely related to SHLV also identified in ticks (I. holocyclus) from Australia, we collected ticks at the locations from where the sick pangolins were found. Consequently, 452 ticks representing 7 species were collected, including 220 Haemaphysalis hystricis, 147 Rhipicephalus microplus, 49 Haemaphysalis longicornis, 11 Ixodes granulatus, 11 Rhipicephalus haemaphysaloides, 9 Rhipicephalus sanguineus, and 5 Haemaphysalis mageshimaensis. Unfortunately, neither DYPV nor LSPV were identified in these ticks by meta-transcriptomics and nested RT-PCR.