Virome profiling of long-tailed marmots reveals tissue tropism and cross-species transmission of a diversity of uncharacterized mammalian viruses Authors: Le Yi, Yan Liang, Guangzhi Yan, Chunli Cai, Yang Li, Zihan Zhao, Yonggang Qu, Changchun Tu, Biao He Published: 2025 ABSTRACT AbstractBackgroundAnthropogenic land conversion and population encroachment into wildlife habitats are amplifying zoonotic risks through intensified human-livestock-wildlife contacts. As ecologically resilient rodents exhibiting dynamic population fluxes, long-tailed marmots (Marmota caudata) are crucial viral reservoirs in Central Asian ecosystems. Their sympatric coexistence with domestic herds creates sustained spillover risks, yet comprehensive virome characterization remains limited. This study employs a whole-virome profiling strategy to comprehensively characterize the viral diversity of 101 wild M. caudata across 5 habitats in Xinjiang, aiming to address key knowledge gaps in rodent-borne zoonoses under anthropogenic pressures.ResultsOur study identified 3314 viral contigs spanning 21 families, with intestinal and lung samples exhibiting the highest viral diversity. Organ-specific tropism was characterized by preferential niches, notably adenoviral predominance in intestinal ecosystems and herpesviral selectivity for lymph node microenvironments. Phylogenetic reconstruction suggested potential cross-species transmission of rotavirus and ephemerovirus from bovines to marmots, along with the identification of 72 novel viruses expanding taxonomic diversity across 17 species and 2 genera. Particularly, two novel coronaviruses showing lung tropism are sufficiently divergent to be new subgenus candidates within the genus Betacoronavirus.ConclusionsThis comprehensive virome profiling delineates the ecological niches of marmot-associated viruses and identifies a diversity of uncharacterized mammalian viruses, including taxonomically novel lineages and multiple potential zoonotic pathogens requiring prioritized attention. These findings expand our understanding of the circulation dynamics and diverse background of marmot viruses, providing preliminary data for developing surveillance frameworks and prevention strategies targeting zoonotic risks associated with this ecologically pivotal species. Video AbstractSupplementary InformationThe online version contains supplementary material available at 10.1186/s40168-025-02304-6.Keywords: Marmots, Tissue tropism, Cross-host transmission, Virome, Coronavirus Introduction Emerging infectious diseases (EIDs), predominantly caused by wildlife-borne zoonotic pathogens, have been occurring with growing frequency worldwide, posing severe threats to public health as well as massive burdens to socio-economic development [1–3]. Rodents are the most abundant, diverse, and widely distributed mammals, representing > 43% of the total number of mammalian species [3–6]. Their habitats overlap extensively with human societies, providing ample opportunities for them to spill over pathogens to humans and other animals, especially those viral ones often resulting in such deadly human infections as Lassa fever, hemorrhagic fever with renal syndrome, hantavirus cardiopulmonary syndrome, etc. Indeed, rodents harbor the most abundant zoonotic viruses [7], covering at least 31 virus families like Hantaviridae, Arenaviridae, and Coronaviridae [8–11], hence are of special concern as reservoirs of emerging viruses. Long-tailed marmots (Marmota caudata) are large, gregarious burrowing rodents within the family Sciuridae, which primarily live in alpine meadows and typical steppes of China, Tajikistan, Kyrgyzstan, Pakistan, Afghanistan, and India [12–15]. They prefer ecological niches that are rich in green plant biomass and, hence, share habitats with a variety of wildlife, such as argali, Himalayan ibex, tolai hare, and pika, and also intimately interact with local livestock like goats, cows, yaks, and horses by competing for food [12]. Marmots are known to carry and transmit Yersinia pestis, the causative agent of plague [16]. Besides, Brucella spp., Anaplasma phagocytophilum, and other important infectious pathogens have also been found in them [17, 18]. They also harbor a variety of viruses, e.g., Powassan virus, a tick-borne encephalitis flavivirus circulating between Ixodes cookei and M. monax [19–21], and woodchuck hepatitis virus, belonging to the genus Orthohepadnavirus within the family Hepadnaviridae and causing marmot liver disease similar to hepatitis of humans [22]. Particularly, with the application of the high-throughput sequencing (HTS)-based viromic technique, an increasing number of viruses with potential public health or veterinary concern have been uncovered in these animals [23–27], exemplified by canine distemper virus, orthopoxvirus, hepatovirus, bocaparvovirus, astrovirus, tick-borne encephalitis virus, etc., highlighting the necessity to continuously investigate the virus diversity they harbor and to systematically understand the virus circulation at the interface between them and humans. Situated in the northwest of China and the hinterland of the Eurasian continent, Xinjiang Uygur Autonomous Region (XUAR) connects China with Central/Western Asia and Europe by serving as an important terrestrial corridor of the Silk Road. It has diverse landscapes, with the mountain grasslands providing ideal habitats for marmots. XUAR is known for its free-ranging husbandry-raising mode, as well as the epidemic focus of several notorious zoonoses, such as plague and hemorrhagic fever [28–31]. The overlapping spheres of marmots and humans/livestock make pathogens readily transmitted between them, particularly exemplified by Yersinia pestis [32]. Unfortunately, the knowledge of marmot-borne viruses in the vast area has been limited, posing an obstacle to controlling and preventing marmot-related EIDs. In the present work, we report the total viromic profiling of long-tailed marmot tissue samples using a combination of the DNA-specific multiple displacement amplification (MDA) and the RNA-specific meta-transcriptomics (MTT) viromic approaches. Results not only identify diverse novel viruses, particularly the discovery of coronaviruses divergent enough to be candidates of a new subgenus within the family but also reveal the tropism specialty of several viruses and the viral circulation cases between marmots and domestic animals. These findings contribute to our understanding of viral ecology at critical human-wildlife interfaces and provide valuable data for proactive surveillance of marmot-associated zoonoses. Results Expanded marmot virome reveals organ tropism of viruses As a part of the Y. pestis surveillance project, we trapped 101 long-tailed marmots inhabiting 5 cattle- and sheep-roaming pastoral areas in Kashi Prefecture of XUAR in 2016, which are adjacent to Tajikistan and Pakistan (Fig. 1A). Among the five sampling locations (HQ, KD, KT, MZ, WM), we collected 26, 31, 14, 14, and 16 long-tailed marmot individuals, respectively. These animals were detected negative for Y. pestis at the local laboratory using the national surveillance standard for plague epizootic [33] and then were dissected to collect their lungs, lymph nodes, kidneys, and intestines for viromic examination. In order to capture the complete spectrum of virome, we used a combination of MDA and MTT techniques to profile the whole eukaryotic virome of each organ type [34]. The same organs of each location were mixed for sample preprocessing, which resulted in 20 MDA and 20 MTT viromic libraries. The clean datasets were first subjected to host genome removal and then to a rapid metagenomic classification. After de novo assembly of the remaining reads, contigs were queried against the eukaryotic viral reference databases (EVRD) version 2.0 [35] using BlastX. As a result, 3314 contigs of ≥ 1000 bp were annotated to 21 eukaryotic viral families (Fig. 1B). We collapsed these viral contigs at 90% average nucleotide identity (ANI90) over 90% coverage of the short contig, i.e., at the approximate species level, which led to 1231 viral clusters (vcANI90s). Overview of sampling and viral diversity. A Map of China with the study region highlighted. The color shading indicates the sampling area in XUAR. The pie chart displays the percentage distribution of samples across sampling sites. Detailed geographical distribution of marmot sampling sites: KT (Kalasu and Taheman), WM (Waqia and Maeryangdaban), HQ (Khunjerab), KD (Kalasudaban), and MZ (Mazhaer). B Heatmap of RPM (reads per million mapped reads) values for 21 eukaryotic viral families. Viral genotypes and families are color-coded on the left, with corresponding color keys on the right. Group labels (classified by organs and geographic locations) are listed in a text column at the bottom. Relative viral abundance in each group is visualized as a heatmap ranging from low (blue) to high (red) based on RPM values in each sample pool The vcANI90 heatmap based on the read abundance showed clear tropism dynamics of these viruses for specific organs. The intestine samples revealed the richest and the most abundant viral composition attributed to their direct connection with the external environment. Viruses within Astroviridae, Picornaviridae, Sedoreoviridae, Parvoviridae, and Adenoviridae were mainly concentrated in the intestine. Second only to the intestine, substantial viral reads are presented in the lung samples, typically represented by the members of Coronaviridae. To further characterize the viral distribution across tissues, we compared the relative abundance of these viral reads (Fig. 2). The analysis demonstrated statistically significant enrichment of astrovirus (AstV), norovirus (NoV), adenovirus (AdV), rotavirus (RV), and bocaparvovirus (BoV) in the intestines, with their relative abundance exceeding that in other tissues by 2–3 orders of magnitude. Meanwhile, we found that polyomavirus (PyV), bovine ephemeral fever virus (BEFV), herpesvirus (HV), and papillomavirus (PaV) revealed preferential concentration in lymph nodes, with quantifiable presence observed in other organs. Notably, Betacoronavirus (βCoV) exhibited exclusive pulmonary localization, with viral reads being strictly confined to the lung specimens. Distribution patterns of different viral genera in tissues and organs New subgenus of coronaviruses Viruses within the family Coronaviridae are a group of enveloped, single-stranded positive-sense RNA (+ssRNA) viruses with a genome size of ~30 kb [36]. CoVs are classified into four genera, with members within the genus Betacoronavirus being of particular concern, since some of them have caused major epidemics and a global pandemic over the past two decades, i.e., the SARS, MERS, and the COVID-19 pandemic [37–39]. Currently, βCoVs are very diverse and are divided into 14 species within 5 subgenera with most being associated with bats, but all rodent-associated βCoVs are relatively constrained in genetic diversity and are only classified into 3 species within the subgenus Embecovirus [40]. The MTT viromics detected two distinct βCoVs in the lung samples collected from the Khunjerab (HQ) area, and their complete genomes were successfully validated by RT-PCR and rapid amplification of cDNA ends (RACE) assays. We named them M. caudata coronavirus (McCoV) XJ72F and McCoV XJ87F, respectively. The full genomes of McCoV XJ72F and McCoV XJ87F were 27,432 nucleotides (nt) with a 30 nt-long poly-A tail and 27,537 nt with a 29 nt-long poly-A tail, respectively, both much shorter than other βCoVs (Fig. 3A). They showed 82.3% nt identity between each other and the highest 70.0% identity with the human coronavirus HKU1 at the full genome level. The genomic structures of the two viruses were similar, containing seven sequentially arranged genes in the order 5′-UTR-ORF1ab-S-hypothetical protein-NS5-E-M-N-UTR-poly(A)−3′ (Fig. 3A). Compared to other βCoVs, they both have an additional 225 nt-long open reading frame (ORF) with unknown function between the S and NS4. Their transcriptional regulatory sequences (TRSs) had the same core 5′-CUAAAC-3′, which conformed to the consensus motif of all embecoviruses but differed from any other subgenus of βCoV [41]. Genome structures (A) of representative betacoronaviruses and the phylogenetic analyses (B) based on RdRp, S, and N proteins of coronaviruses The phylogenetic trees were reconstructed based on the full-length amino acid (aa) sequences of RdRp, S, and N proteins (Fig. 3B). The family-level phylogeny based on RdRp showed that McCoV XJ87F and McCoV XJ72F fell into an independent cluster within the clade of βCoVs. The genus-level phylogenetic analyses based on S and N clearly demonstrated that the two viruses were sisters next to the subgenus Embecovirus, suggesting new members within the genus Betacoronavirus. The International Committee on Taxonomy of Viruses (ICTV) has proposed a demarcation criterion for coronavirus classification that uses the divergence of the 3CLpro, NiRAN, RdRp, ZBD, and HEL1 domains of the replicase protein, in which the ranks of genus, subgenus, and species are divided by the aa divergences of 36.0%, 14.7%, and 7.6%, respectively [42, 43]. The comparison of these domains showed that McCoV XJ87F and McCoV XJ72F were 8.7–11.8% divergent from each other, and the minimum divergence from other known βCoVs was 23.0%, fulfilling the criterion to be two species within a new subgenus. Following the nomenclature of established subgenera within the genus Betacoronavirus and based on phylogenetic analyses that support a distinct evolutionary lineage, we propose the candidate subgenus Mabecovirus to encompass the new betacoronaviruses identified in marmots. Within this proposed taxonomic framework, McCoV XJ87F and McCoV XJ72F are tentatively classified as member species. RT-PCR detection showed that the novel coronaviruses only existed in the lungs of the marmots from HQ, with a total positive rate of 2.0% (2/101) (Table S1). Cross-host transmission of rotavirus and Ephemerovirus The interface of contact between human/domestic animals and wildlife is a key factor mediating cross-host transmission of zoonotic viruses. Pastoral areas represent such interfaces, where free-ranging livestock frequently interact with diverse wildlife due to sharing food resources. Our study found phylogenetic evidence suggestive of cross-host transmission events involving group A rotavirus (RVA) and Ephemerovirus between marmots and cattle, largely from cattle to marmots. RVAs, which contain 11 double-stranded RNA segments [44], are the most notorious diarrheal pathogens with a wide host range spanning humans, mammals, and avian species [45]. Ephemerovirus, a single-stranded negative-sense RNA (-ssRNA) virus within the family Rhabdoviridae, has a genome of ~15 kb and circulates between ruminants and arthropod vectors, causing ephemeral fever in ruminants [46]. Through the MTT viromics, we recovered 4 RVA contigs in the intestines of KT marmots, covering partial or near-complete segments of VP1, VP2, VP3, and NSP3, alongside a 14,905 nucleotide (nt) Ephemerovirus contig in the lymph nodes of HQ marmots. These RVA contigs, derived from the same sample library but annotated to distinct rotavirus segments, were provisionally designated RVA/Marmot-wt/CHN/XJC1 to C4/2016. Phylogenetic analyses demonstrated that the VP3 segment was closely next to ruminant rotaviruses, whereas the VP1, VP2, and NSP3 segments all grouped specifically within the ruminant RVA lineages, exhibiting maximum nt identities of 93.0–95.3% (Fig. 4A). The Ephemerovirus contig, spanning almost the entire genome (3′-N-P-M-G-GNS-α1-α2-β-γ-δ-L-5′), was designated marmot Ephemerovirus McEFV/XJ82F/CHN/2016. Phylogenetic analysis of the L gene sequences placed it within a bovine-derived cluster (Fig. 4B), sharing 98.2% nt identity with bovine ephemeral fever virus (BEFV) strain JT02L (KY315724.1). Molecular detection confirmed the presence of RVA XJC1 in 4 samples and McEFV/XJ82F/CHN/2016 in 1 sample with positive rates of 4.0% (4/101) and 1.0% (1/101), respectively. These findings suggest that the RVAs and McEFV/XJ82F/CHN/2016 likely originated from cattle and underwent cross-species transmission to marmots within this sympatric ecosystem. Genomic characterization and phylogenetic profiling of rotaviruses and ephemeroviruses. A Comparison of the genomic structures between RVA/DQ-75 and RVA/XJC1-4 and the phylogenetic tree based on the VP1, VP2, VP3, and NSP4 gene nt sequences of rotaviruses. B Genomic structure of MCEFV XJ82F and the phylogenetic tree based on the L gene nt sequences of ephemeroviruses New species of adenoviruses and herpesviruses Adenoviruses (AdVs) and herpesviruses (HVs) are two large groups of double-stranded DNA (dsDNA) viruses. AdVs possess a 25–48 kb-sized genome and are associated with various clinical manifestations, including respiratory infections, keratoconjunctivitis, cystitis, and gastroenteritis in humans [46]. In contrast, HVs exhibit substantial genomic complexity with larger genomes ranging from 125 to 241 kb, capable of causing diverse pathologies in humans and other animals, such as cutaneous vesicles, encephalitis, meningitis, and chicken Marek’s disease [47]. The MDA viromics revealed 67 adenoviral contigs that were predominantly identified in intestinal samples. Notably, we recovered a near-complete adenoviral genome contig (34,824 bp) from the intestinal library of HQ with a positive rate of 3.0% in the sampled populations (Table S1). We named the virus M. caudata adenovirus (McAdV) XJ96C. Furthermore, two additional adenoviral variants (McAdV XJ11C and McAdV XJ72C) encoding complete DNA polymerase sequences were identified within the same library. McAdV XJ96C encodes 24 genes, with the majority annotated as canonical adenoviral proteins, while a 1092-bp long ORF cannot be functionally annotated. Phylogenetic reconstruction based on the pol gene revealed distinct evolutionary relationships: XJ72C and XJ96C formed a distinct cluster adjacent to the red squirrel adenovirus [48], while XJ11C occupied an intermediate phylogenetic position within the genus Mastadenovirus, showing substantial divergence from both marmot adenoviruses and other established adenoviruses. XJ72C and XJ96C were 77.5% identical to each other and 66.5–69.5% to the red squirrel AdV, while XJ11C showed 49.9–50.1% identity to XJ72C and XJ96C and only the highest 49.5% identity to porcine M. caudata. Crucially, all three McAdVs exhibited pairwise genetic distances exceeding 15% both between themselves and relative to previously characterized adenoviruses. In accordance with the ICTV species demarcation of adenoviruses [46, 49], the three McAdVs meet the established criteria for novel species classification within the genus Mastadenovirus. The MDA virome analysis identified 349 HV-like contigs ranging from 1009 to 69,933 bp in length, predominantly derived from lymph node libraries. Specific PCR tests demonstrated differential detection rates of HVs across tissues, with lymph nodes showing 8.9% (9/101) positivity versus 1.0% (1/101) in the intestine. To resolve the phylogenetic classification of these herpesviruses, we designed targeted primers based on viral contigs to amplify conserved pol gene fragments. This approach yielded three distinct herpesvirus sequences designated as M. caudata herpesvirus (McHV) XJ5L, XJ16L, and XJ36L. Phylogenetic reconstruction using the pol gene revealed that all three viruses clustered within the subfamily Betaherpesvirinae (Fig. 5). XJ5L and XJ16L clustered distinctly (86.5% aa identity) and occupied a separate phylogenetic position within the murid herpesvirus group, sharing 53.8–60.5% sequence identity with other members. In contrast, XJ36L exhibited substantial genetic divergence, forming a sister clade with Suid betaherpesvirus 2 (52.9% aa identity) while showing markedly lower similarity with XJ5L and XJ16L (47.6–48.7% aa identities). The phylogenetic evidence demonstrates that the three sequences form two distinct clades divergent from all currently characterized herpesviruses, and in accordance with the ICTV species demarcation criteria for the family Orthoherpesviridae [47], McHV XJ36L, XJ5L, and XJ16L should be considered as three new species within Betaherpesvirinae. Genomic organization and phylogenetic analysis of the marmot adenovirus and herpesvirus. A Schematic representation of the genomic structure of the marmot adenovirus McADV/XJ96C identified in this study. Key coding regions are indicated. B Linear comparison of the marmot herpesvirus McHV/XJ16L/CHN/2016 genomic contig obtained in this study with the reference sequence Murid herpesvirus 8. Open reading frames (ORFs) are indicated by shaded areas, and the genomic backbone is depicted. C Phylogenetic trees of adenoviruses (left) and herpesviruses (right) based on the amino acid sequences of the DNA polymerase (Pol) protein. The Pol protein coding region is highlighted in red in both genomic maps (the coding region is named pol for adenovirus and M54 for herpesvirus). Sequences from this study are indicated by solid red circles Abundant and diverse DNA viruses with circular genomes The MDA viromes revealed a substantial diversity of circular DNA viruses, comprising four major groups: circular replicase-encoding single-stranded (CRESS) DNA viruses, polyomaviruses (PyVs), papillomaviruses (PaVs), and torque teno viruses (TTVs) (Fig. 6A). Among 11 CRESS DNA viral genomes from intestinal libraries, 6 taxonomic lineages were resolved, including the recognized families Circoviridae and Naryaviridae alongside four unclassified clusters (Fig. 6B). Most of these genomes shared remarkable sequence similarities with other rodent-associated viruses, sharing aa identities ranging from 52.3% to 76.7%. Strikingly, two exceptionally large circular genomes, XJ21C (6719 nt) and XJ24C (4788 nt) (Fig. 6A), were recovered and independently validated through PCR amplification. XJ21C encoded five unidirectionally oriented proteins, including a capsid (Cap) protein (1587 nt) and a replicase (Rep) protein (789 nt), while the three remaining ORFs lacked functional annotation. Phylogenetic positioning demonstrated that XJ21C forms an independent branch at the family level (Fig. 6B), sharing only 25.9% aa identity with its closest relative, turbot circovirus (WYD62157.1). We propose the establishment of a new family, tentatively named “Megacoviridae,” within the clan of CRESS DNA viruses to accommodate XJ21C, based on its distinct phylogenetic positioning and significant genetic divergence. In contrast, XJ24C harbored two unidirectionally oriented ORFs of unknown function and an inversely oriented Rep gene. It clustered within the ICTV non-reference virus clade, displaying 34.1% aa identity to the closest relative, canine circovirus (WDW25929.1). Additionally, XJ1C, classified within the Circoviridae clade, exhibited 78.3% genome-wide nt identity to Calfel virus LSF45_cir359, below the Circoviridae species threshold of 80% [50], representing a putative new species. Specific PCR tests revealed positive rates of 22.8% (23/101) and 5.9% (6/101) for XJ21C and XJ24C, respectively, with all positives derived from intestinal specimens. Genomic and phylogenetic characterizations of viruses with a circular DNA genome. A Genomic structure of the representative circular DNA viruses recovered in this study. B Phylogenetic tree of CRESS DNA viruses based on the Rep proteins. C Phylogenetic tree of PyVs based on the LTAg proteins. D Phylogenetic tree of PaVs based on L1 proteins. E Phylogenetic tree of TTVs based on the ORF1 proteins PyVs are known to infect mammals and birds, with some species associated with progressive multifocal leukoencephalopathy, respiratory infections, or cancers [51]. These viruses possess ~5.2-kb circular dsDNA genomes and are divided into nine genera [52]. Three complete PyV genomes (XJ7L, XJ10L, and XJ34L: 5136–5297 bp) with the canonical PyV genomic architecture were successfully recovered. Phylogenetic analysis revealed three distinct evolutionary clusters: XJ34L clustered within the Alphapolyomavirus branch, while XJ7L and XJ10L formed two novel lineages distinct from established polyomaviruses, both remaining taxonomically unassigned. Comparative sequence analysis revealed 61.3–79.9% intra-cluster aa identity, compared to 40.5–53.1% inter-cluster identity. Analysis of all three large T antigen (LTAg) sequences showed ≥ 24% nt divergence to recognized species, exceeding the ICTV 15% species threshold [52], confirming their classification as novel distinct viral species. Specific PCR for XJ7L showed a 16.8% (17/101) positivity rate among studied samples, with the WM region demonstrating the highest prevalence of 56.3% (9/16). PaVs are vertebrate-infecting agents associated with warts, papillomas, and tumors, characterized by ~8-kb circular double-stranded DNA genomes [53]. In this study, PaV-like contigs were identified exclusively in lymph node specimens derived from the KD group. PCR-based genome assembly enabled the reconstruction of a complete 7718-bp viral genome (designated Papillomavirus McPaV XJ32L), encoding six canonical PaV genes (L1, L2, E1, E2, E6, and E7) arranged in unidirectional orientation. Comparative genomic analysis revealed 55.2% aa identity and 59.6% nt identity of L1 to the closest known PaV, which is below the ICTV 60% nucleotide threshold for genus demarcation [54, 55]. Phylogenetic reconstruction positioned XJ32L within the subfamily Firstpapillomavirinae, where it formed a distinct monophyletic clade (Fig. 6D). Combined with comparative genomic analysis, these findings could support the proposal of a candidate new genus, which we have provisionally designated Treislambdapapillomavirus, within this subfamily. The positive rate of McPaV XJ32L was 5.0% (5/101) by PCR test, with infection confirmed in both the intestine and lymph nodes of the same marmot, demonstrating systemic viral dissemination in this rodent species. TTVs, classified within the family Anelloviridae, are circular single-stranded DNA (ssDNA) viruses characterized by genome sizes ranging from 2 to 4 kb in length [56]. In the current study, 29 contigs were annotated as TTVs, exhibiting extensive genetic divergence with aa identities of 34.5–85.2% compared to known TTVs. Phylogenetic analysis based on ORF1 sequences revealed two major clusters (provisionally designated Marmot TTV1 and Marmot TTV2) with intra-cluster nt identities of 55.5–95.1% and inter-cluster nt identities of 41.2–49.1% (Fig. 6E). Notably, members within the Marmot TTV1 cluster shared 67.8–72.2% nt identities with the yellow-bellied marmot TTV in the USA. According to the taxonomic standard for anelloviruses, a nucleotide identity threshold of 69% across the complete ORF1 coding region serves as the species demarcation criterion [57]. These findings suggest the identification of two novel anellovirus species belonging to a currently unclassified genus within the family Anelloviridae. Extensive discoveries of other mammal viruses In addition to the viruses described above, our combined viromic approaches identified numerous other diverse viruses, most of which are variants of known marmot viruses. Here, we focused on astroviruses (AstVs), hepatoviruses (HAVs), noroviruses (NoVs), phenuiviruses (PhVs), adeno-associated dependoparvoviruses (AAVs), and bocaparvoviruses (BoVs) to summarize their phylogenetic relationships and molecular detections. The viromics successfully recovered the complete or nearly complete genomes for 7 AstVs, 1 HAV, and 2 BoVs, while only partial sequences were obtained for 1 NoV (4451 nt), 2 AAVs (3515 nt and 3976 nt), and 2 PhV L segments. All these viruses exhibited conserved genomic architectures characteristic of their respective viral families. For instance, McAstV XJ84C demonstrated the overlap of ORF1a and ORF1b with a highly conserved ribosome coding frameshift sequence (5′-AAAAAAAC-3′), while McHAV XJ96x encoded a polyprotein predicted to undergo proteolytic cleavage into four structural and seven nonstructural proteins. Phylogenetic analyses demonstrated that the seven diverse McAstVs, McHAV XJ96x, and McNoV XJ16C and two McBoVs were all clustered closely with their counterparts found in other marmots (Fig. 7) and were identified as variants of established virus species. However, the two PhVs and two McAAVs formed distinct phylogenetic subclades, and their incomplete genome assemblies preclude definitive classification as novel virus species under current ICTV guidelines. Given the unique phylogenetic lineage, McAbuV XJ15C was singled out for targeted RT-PCR screening, revealing a positivity rate of 2.0% (2/101) with exclusive detection in intestine specimens. Genomic structures and phylogenetic analyses of astroviruses (A), hepatoviruses (B), noroviruses (C), phenuiviruses (D), and parvoviruses (E). The fragments used for phylogenetic reconstruction were highlighted in red Discussion Proactive detection of emerging viruses could contribute to outbreak prevention and mitigation of the scale of potential epidemics and pandemics. Cross-species transmission events often occur at the human-animal interface, constituting major pathways for zoonotic disease emergence. Therefore, researchers are focusing on the diversity of viruses carried by animals. Notably, however, most of the current studies employed a single viromic technique to profile viruses of specific genome types by pooling different tissue samples [34], which made it impossible to capture the complete spectrum of viruses or understand the tropism dynamics of viruses in certain tissues. To overcome the limitations of single-modality viromic analyses, we employed an integrated framework combining MDA and MTT sequencing to comprehensively characterize the viromes across different tissues of long-tailed marmots in Xinjiang. Our findings reveal an unexpectedly rich viral diversity encompassing vertebrate-associated families alongside insect- and plant-derived viruses. The latter likely represent dietary components from grassland foraging, though intrinsic viral carriage within the marmots cannot be excluded. Notably, while the overall viral family composition corresponds with previous findings in Marmota himalayana from the Qinghai-Tibet Plateau [25, 26], there were still some viral genera and families that had not been reported in marmots before, such as Phenuiviridae and Betacoronavirus. These differences were potentially attributable to host speciation and geographic isolation. Comparative virome profiling of long-tailed marmots revealed distinct tissue specialization of viral communities. RNA viral families exhibited particularly low abundance in both kidneys and lymph nodes, with +ssRNA viruses almost absent from the kidneys. Furthermore, herpesviruses showed predominant lymph node distribution, while adenoviruses, astroviruses, and noroviruses displayed preferential intestinal localizations. Notably, coronaviruses were uniquely restricted to pulmonary tissues. These distinct distribution patterns not only reflect intrinsic viral tropisms but also provide epidemiological insights into potential transmission routes and zoonotic risks. DNA viruses dominated the whole virome profile, particularly the circular genomes. This representation bias likely originates from the rolling circle replication mechanism inherent to MDA, which preferentially amplifies circular genomes. Despite the inherent technical bias, MDA proved superior to conventional metagenomics (MTG) in virome characterization, enabling the recovery of a greater diversity of viral species sequences [34]. Phylogenetic analysis revealed substantial genetic divergence between the recovered contigs and existing GenBank references, including several unclassified genera within recognized families, suggesting the presence of novel rodent-associated viruses. These findings highlight gaps in the current understanding of marmot virome diversity while highlighting that the risk of emerging rodent-borne infectious diseases should be considered. Of particular concern were the vertebrate-infecting viruses exhibiting cross-species transmission capabilities. Phylogenetic analysis revealed that marmot rotaviruses clustered with the ruminant rotavirus clade, while ephemeroviruses exhibited high homology with bovine strains. Given the sympatric distribution of marmots and grazing herbivores within shared grassland ecosystems, these findings indicate potential viral transmission via the fecal–oral route. Beyond these specific cross-species transmission events, our findings suggest that long-tailed marmots may function as intermediate hosts for viruses originating from domestic animals. Their ecological overlap with livestock, social population structure, and the phylogenetic evidence of viral sharing create conditions conducive to viral circulation and maintenance within marmot colonies. Additionally, the rich viral diversity observed in specific tissues, such as the simultaneous presence of multiple virus families in intestinal samples, indicates potential coinfection scenarios that could facilitate viral evolution or synergistic interactions. However, due to our sample pooling strategy, further studies with individual-level analysis are needed to confirm coinfection in single animals and quantify amplification dynamics. Phylogenetic reconstruction of Anelloviridae contigs revealed characteristic clustering between long-tailed marmot and yellow-bellied marmot strains within an unclassified genus-level lineage, demonstrating viral similarity between closely related hosts and validating the reliability of virome profiling. The marmot-derived coronaviruses McCoV-XJ87F and McCoV-XJ72F, identified in separate regions of Xinjiang, represent distinct species within a novel subgenus, implicating long-tailed marmots as probable maintenance hosts. There are currently no reports of these coronaviruses in other rodents, potentially reflecting host specificity or geographical limitation. RT-PCR screening of the sampled marmots revealed that extrapulmonary tissues were coronavirus-negative, demonstrating the lung-specific tropism while suggesting respiratory system association for both the viruses. Herpesviruses identified in rodents primarily cluster within the subfamilies Betaherpesvirinae and Gammaherpesvirinae [58]. In this study, three novel species from marmots were phylogenetically positioned within the Betaherpesvirinae clade. Virome analysis revealed that herpesvirus sequences originated predominantly from marmot lymph nodes, a finding corroborated by PCR detection, indicating specific tissue tropism in marmots. Metagenomic analysis identified three novel adenovirus species phylogenetically clustering with a squirrel adenovirus found in a dead red squirrel exhibiting acute diffuse catarrhal enteritis [48]. As the marmot adenoviruses are also derived from rodent intestines, the potential pathogenic risks warrant further attention. Polyomaviruses infect a wide range of animals, including birds, rodents, and fish [59]. Three novel polyomavirus species were also identified in marmots, highlighting the significant genetic diversity within marmot polyomaviruses. Owing to the incomplete taxonomic characterization of certain rodent-associated viruses, phylogenetic analysis was constrained to species and genus levels. In conclusion, we employed viral metagenomic analysis to profile the whole viromic overview of long-tailed marmots in XUAR. The identification of novel viral species and genera could expand the virological taxonomic framework of certain viral families. The results also suggest that long-tailed marmots are potential reservoirs of emerging viral variants, warranting continued surveillance to assess spillover risks at the human-livestock-wildlife interface. Limitations This study has several limitations. The sample size and geographic focus within XUAR may limit the generalizability of our findings to other marmot populations. Additionally, while our viromic approach provides a broad overview, it may miss low-abundance viruses. It is important to acknowledge that while MDA proved superior to conventional metagenomics (MTG) in virome characterization for viral discovery, enabling the recovery of a greater diversity of viral species sequences [34], this whole-genome amplification technique is known to potentially introduce stochastic errors and alter the relative abundance of taxa compared to amplification-free approaches, which may impact nucleotide-level comparisons and fine-scale genetic analyses. Consequently, the abundance estimates and single-nucleotide variants reported herein should be interpreted with caution, recognizing this as a common trade-off in richness-focused viromic studies. Furthermore, this viromics approach, which includes filtration and nuclease treatment to enrich for viral particles, inherently depletes bacterial nucleic acids. Therefore, our sequencing data are not suitable for evaluating bacterial co-infections or for confirming the absence of specific bacterial pathogens like Y. pestis. Finally, while phylogenetic analysis suggests cross-species transmission, this evidence requires further ecological and experimental studies for confirmation, and the proposed new viral taxa (e.g., Mabecovirus) are contingent upon formal ratification by the ICTV. Materials and methods Sample collection XUAR is the natural epidemic focus of the plague, and plague surveillance is a regular assignment in local areas. During the surveillance conducted in Kashgar from May to July 2016, traps were placed at marmot holes and checked every 2 h. A total of 101 apparently healthy long-tailed marmots were captured. The captured animals were humanely euthanized through excessive muscle injection of pentobarbital sodium at the local Center for Animal Disease Control and Prevention, followed by dissections to extract the lungs, kidneys, lymph nodes, and intestinal tissues, respectively. All the animals were tested for plague, and those that were negative were transported to the laboratory on dry ice and stored at −80 °C for viromic analysis. Sample pretreatment and next-generation sequencing The marmot samples were divided into 20 groups according to the 5 sampling areas and 4 types of organs. About 0.1 g of each sample was mixed in a 1.5-mL centrifuge tube according to grouping. The sterile magnesium salt buffer (50 mM Tris, 10 mM MgSO4, 0.1 M NaCl, pH = 7.5) was added at a v/w ratio of 5:1 and then homogenized using a ground-glass tissue grinder; homogenates were centrifuged at 10,000 × g for 5 min at 4 °C to remove debris. The centrifuged supernatants were filtered through 0.45-μm pore size membranes and digested with DNase I and RNase A (TaKaRa, Dalian, China) to eliminate the free nucleic acids. For the MDA processing, viral DNA was extracted using the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany) and isothermally amplified using the illustra GenomiPhi V2 DNA Amplification Kit (GE Healthcare, Chicago, IL, USA) according to the manufacturer’s protocol. The products were purified using the QIAquick PCR Purification Kit (Qiagen, Hilden, Germany) with 1 µg used for Illumina pair-end (150 bp) sequencing on an Illumina NovaSeq 6000 sequencer (Illumina Inc., San Diego, CA, USA). For the MTT processing, total RNA was extracted using the TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and quantified by a Qubit 4 Fluorometer (Thermo Fisher Scientific). The extracted RNA was checked for integrity by an Agilent 2100 Bioanalyzer (Agilent Technologies), and rRNA was removed using the Ribo-Zero™ Magnetic Gold Kit (Epicentra Biotechnologies, Madison, WI, USA). The final libraries were prepared using the NEBNext Ultra Directional RNA Library Prep Kit (NEB, Ipswich, MA, USA) and then processed for next-generation sequencing (NGS) on an Illumina NovaSeq 6000 sequencer. Viromic analyses The data processing and bioinformatics analyses were similar to our previous procedure [34]. First, quality control was conducted using fastp (v0.20.0) to exclude the poor-quality reads [60], and the clean reads were obtained from 40 pools. Host genomes were removed from clean data using Bowtie2 (v2.4.1) with the very sensitive end-to-end mode by mapping against the reference genome of M. himalayana (a close relative to M. caudata) in the NCBI database (GenBank accession no. RAQP00000000.1) [61]. Kraken 2 (v2.0.9) was employed to taxonomically classify the bacteria, archaea, and fungi genomes among the remaining reads with a custom RefSeq-based database [62]. All unaligned reads of MDA were de novo assembled using MEGAHIT (v1.2.9) [63], while those of MTT were de novo assembled using Trinity (v2.10.0) [64]. Contigs were clustered to remove duplicate sequences using Cd-hit (v4.5.4) [65]. In order to obtain high-quality virus sequences, the contigs with lengths above 1 kb were retained. Then, the resulting contigs were submitted to BLASTn (v2.10.0) and DIAMOND (v0.9.35) BLASTx searches with e-values cutoff of 1e-5 against EVRD [66]. Previously cleaned reads were mapped back to the viral-like contigs using Bowtie2 and counted using Samtools (v1.10). PCR validation Specific primers were designed based on the viral contigs. PCR was performed with the Phanta Flash Master Mix (Vazyme, Nanjing, China) under the following cycling conditions: pre-denaturation at 98 °C for 30 s; 35 cycles of denaturation at 98 °C for 10 s, annealing at 58 °C for 5 s, and extension at 72 °C for 5 s/kb; and a final extension at 72 °C for 1 min. RACE (rapid amplification of cDNA ends) was used to obtain the complete terminal of viral genomes using the HiScript-TS 5′/3′ RACE Kit (Vazyme, Nanjing, China). Two pairs of specific primers, 5′GSP1 and 5′GSP2 and 3′GSP1 and 3′GSP2, were designed according to the known region at the 5′/3′ end of the viral genome. Reverse transcription and PCR were carried out according to the manufacturer’s protocol. The expected products were directly sequenced on an ABI 3730xl DNA analyzer (ComateBio, Changchun, China). Genomic annotation and phylogenetic analyses The ORFs in the viral genomes were predicted with the NCBI ORF Finder tool (https://www.ncbi.nlm.nih.gov/orffinder/). The nt and aa sequences of the putative viral proteins were further verified by performing online BLASTn or BLASTx searches against the nonredundant database of GenBank. The conserved domains within the nt and aa sequences were also searched by the CD-search (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi) in NCBI. The structural prediction of the coronavirus genomes was performed using ZCURVE_CoV2.1 [67]. All references were retrieved from GenBank (Table S4). The alignments of the nt and deduced aa sequences were conducted by MAFFT (v7.313) [68] with default parameters. TrimAl (v1.2) [69] was employed to clip ambiguous alignments. The best-fit substitution model was automatically determined by the ModelFinder module of IQ-TREE [70]. The phylogenetic analyses were conducted using the maximum likelihood method with the bootstrap set to 1000 replicates in IQ-TREE (v1.6.8) [71] of Phylosuit (v1.2.2) [72]. Phylogenetic trees were visualized and manipulated using the Interactive Tree of Life (iTOL) (https://itol.embl.de) [73]. Supplementary Information Supplementary Material 1: Table S1. Molecular detection of the representative viruses recovered in this studya. a Numbers in each cell are positive tissue number. Supplementary Material 2: Table S2. Comparison of the MCCoV-XJ87 genome with other representative betacoronavirus strains. Supplementary Material 3: Table S3. Comparison of amino acid homology in five demarcated domains between MCCoV-XJ87, MCCoV-XJ72, and the current coronaviruses in GenBank. Supplementary Material 4: Table S4. Accession list for phylogenetic analyses. Acknowledgements Not applicable. Authors’ contributions All authors made substantial contributions to this work. B.H. and C.T. conceived and designed the study. Y.L. and Y.Q. participated in the sample collection. L.Y., G.Y., C.C., Y.L., and Z.Z. performed the experiments. L.Y., Y.L., G.Y., and B.H. analyzed the data. L.Y., B.H., and G.Y. wrote the paper. B.H., C.T., and Y.Q. revised the paper. All authors approved the manuscript. Funding This research was supported by the National Natural Science Foundation of China (32300424) and the National Key Research and Development Program of China (2023YFF1305401). The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. Data availability All genome sequences were submitted to GenBank. Accession numbers for the representatives of assembled contigs are PP098962-PP098971, PP728723-PP728766, PP531426-PP531443, and OR882008. Illumina Novaseq sequence data were deposited into NCBI BioProject under accession number PRJNA1055180. Declarations Ethics approval and consent to participate All animal experiments were reviewed and approved by the Administrative Committee on Animal Welfare of Changchun Veterinary Research Institute (Laboratory Animal Care and Use Committee Authorization, permit number: JSY-DW-2016-02). All methods were carried out in accordance with relevant guidelines and regulations. Consent to participate is not applicable as this study did not involve human participants. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Le Yi, Yan Liang and Guangzhi Yan contributed equally to this work. Contributor Information Changchun Tu, Email: changchun_tu@hotmail.com. Biao He, Email: heb-001001@163.com. References Associated Data Supplementary Materials Supplementary Material 1: Table S1. Molecular detection of the representative viruses recovered in this studya. a Numbers in each cell are positive tissue number. Supplementary Material 2: Table S2. Comparison of the MCCoV-XJ87 genome with other representative betacoronavirus strains. Supplementary Material 3: Table S3. Comparison of amino acid homology in five demarcated domains between MCCoV-XJ87, MCCoV-XJ72, and the current coronaviruses in GenBank. Supplementary Material 4: Table S4. Accession list for phylogenetic analyses. Data Availability Statement All genome sequences were submitted to GenBank. Accession numbers for the representatives of assembled contigs are PP098962-PP098971, PP728723-PP728766, PP531426-PP531443, and OR882008. Illumina Novaseq sequence data were deposited into NCBI BioProject under accession number PRJNA1055180.