Evolution and ecology of Jeilongvirus among wild rodents and shrews in Singapore

Abstract

Background

Jeilongvirus was proposed as a new genus within the Paramyxoviridae in 2018. The advancement in metagenomic approaches has encouraged multiple reports of Jeilongvirus detection following the initial species discovery, enriching species diversity and host range within the genus. However, Jeilongvirus remains understudied in Singapore, where interfaces between humans and small mammals are plentiful.

Methods

Here, we utilized metagenomic sequencing for the exploration of viral diversity in small mammal tissues. Upon discovery of Jeilongvirus, molecular screening and full genome sequencing was conducted, with the data used to conduct statistical modelling and phylogenetic analysis.

Results

We report the presence of Jeilongvirus in four species of Singapore wild small mammals, detected in their spleen and kidney. We show that full genomes of three Singapore Jeilongvirus encode for eight ORFs including the small hydrophobic and transmembrane proteins. All generated genomes cluster phylogenetically within the small mammal subclade, but share low genetic similarity with representative Jeilongvirus species. Statistical modelling showed no spatial or temporal patterns and differences among species, life history traits and habitat types.

Conclusions

This study serves as a basis for understanding dynamics between Jeilongvirus and small mammal hosts in Singapore by displaying the virus generalist nature. In addition, the initial detection can help to invoke improved routine surveillance and detection of circulating pathogens in synanthropic hosts.

Supplementary Information

The online version contains supplementary material available at 10.1186/s42522-023-00094-1.

Background

The family Paramyxoviridae consists of enveloped, negative-sense RNA viruses with genomes ranging from 14 kb to 20 kb in length. The non-segmented genomes encode for six open reading frames (ORFs), namely the nucleocapsid (N), phosphoprotein (P), matrix (M), fusion (F), receptor binding protein (RBP) and large polymerase (L). The RBP is given alternate designations, including hemagglutinin (H), hemagglutinin-neuraminidase (HN) and attachment (G) protein for different taxa [1]. This family of viruses include several medically important genera such as Morbillivirus, Respirovirus, Rubulavirus, and Henipavirus [2]. Within these genera, there are several instances of zoonotic spillover, with notable examples being Hendra virus (HeV) and Nipah virus (NiV) that originated from Pteropodidae fruit bats [3].

Jeilongvirus was proposed as a new genus within the subfamily Orthoparamyxovirinae in 2018 following the discovery of its two initial members, Beilong virus (BeiV) and J-virus [4]. The Jeilongvirus genus is segregated from other members of the Paramyxoviridae family due to the presence of two additional proteins that encode a transmembrane (TM) and/or small hydrophobic (SH) protein, in addition to a larger than average G protein. However, this unique characteristic (TM and SH proteins) has also been found absent in two Jeilongvirus species, Mount Mabu Lophuromys virus 1 (MMLV-1) and Mount Mabu Lophuromys virus 2 (MMLV-2) from Rungwe brush-furred rats from Mozambique [5]. BeiV and J-virus were believed to have rodent origins due to their initial discovery in their host animals [6, 7]. Subsequent surveillance with pan-paramyxovirus PCR and metagenomics approaches uncovered a wider host range of Jeilongvirus [5, 8–11], with 15 established species [12] from bats, small mammals, and feline origin. Despite extensive reporting of Jeilongvirus diversity, their host specificity, mode of transmission and pathogenic effects remain understudied and largely unknown.

Small mammals such as rodents play an important role in maintaining virus genetic diversity but little is known about the viral diversity in Singapore. There are around 16 small mammal species in Singapore, and many of them inhabit peri-urban areas. Common and widely distributed species include Mus castaneus, Rattus norvegicus, and R. tanezumi [13]. The synanthropic lifestyle of small mammals, especially in a highly urbanized city like Singapore, creates opportunities for high contact interfaces with humans. This, in combination with small mammals’ competency as reservoirs for zoonotic diseases [14, 15], increases the risk for pathogen spillover and the spread of infectious diseases.

Although the zoonotic potential of Jeilongvirus is unknown, the diversity of infected species indicates potential past spillover events. Here, we investigated the prevalence of Jeilongvirus from small mammals in Singapore and elucidated the evolutionary relationships of three full length Jeilongvirus genomes retrieved in this study using a metagenomic approach. We also examined the ecology and epidemiology of this virus in Singapore by studying the relationships between host factors, such as species variety, life history traits and habitat types, and the likelihood of Jeilongvirus infection.

Methods

Sample collection

All animal work was conducted with approvals from the National University of Singapore Institutional Animal Care and Use Committee (IACUC B01/12), National Parks Board (Permit no NP/RP12-004) and the Agri- Food and Veterinary Authority of Singapore (Permit no AV16.01.004.0004). Small mammals (n = 100), including rodents and shrews, were captured using Sherman and Tomahawk traps from April 2012 to March 2016 from 17 locations across Singapore (Fig. 1). Trapped animals were identified to species using morphological features in addition to a COI barcoding PCR [16] when animals were unable to be morphologically identified. Individuals classified as pest species by Animal and Veterinary Science were euthanized using isoflurane. Both euthanized animals, as well as animals that died during handling were transported back to the laboratory for dissection in a biosafety level two cabinet. Lung, spleen and kidney tissues were collected and stored individually in viral transport media (VTM) or PBS and kept at -80 ºC until processed. Dissection instruments were disinfected with 1% virkon and 70% ethanol between harvesting each tissue and between individual animals.

Fig. 1.

Map of Singapore showing the location of 17 sampling sites of captured small mammals. Red sites denotes location of Jeilongvirus RT-qPCR positive sampling sites, and black denotes Jeilongvirus negative sites. Inset map shows the location of Singapore within Southeast Asia

Sample processing and RNA extraction

For initial exploratory pathogen discovery, the lung, spleen, and kidney homogenates from 30 Rattus tanezumi individuals were pooled for metagenomic sequencing. For subsequent screening and next generation sequencing of individual small mammals samples, a 1–3 mm³ piece of individual lung, spleen and kidney from each animal was collected and separately weighed (2.0 to 32.2 mg). All tissues were homogenized in 500 µl of AVL buffer (Qiagen, Hilden, Germany) using silicon carbide shards and a Mini-Beadbeater-96 (Biospec Products, Bartlesville, USA) at 2000 rpm for 2.5 min. The homogenates were centrifuged at 12,000 rpm for 1 min, and 140 µl was used for RNA extraction with the Direct-zol RNA miniprep kit (Zymo Research Corporation, Irvine, USA) as per the manufacturer’s instructions.

Metagenomics and virus full genome sequencing

Library preparation, sequencing and metagenomics analysis of the pooled R. tanezumi tissues were performed as previously described [17]. Briefly, the metagenomic library was prepared with the TruSeq RNA library preparation kit (Illumina, San Diego, USA) and sequenced on an Illumina MiSeq platform with 150 bp single-end reads. The subsequent FASTQ file was trimmed and taxonomically classified using DIAMOND v0.9 [18]. Jeilongvirus reads were extracted and mapped to a closely related genome sequence (MT085297) identified through BLASTn, with a full-length genome obtained and annotated for reference. This dataset is available at the NCBI Sequence Read Archive under BioProject accession number PRJNA990547. For whole genome sequencing of individual small mammal samples (n = 3), RNA was treated with the RiboZero Plus rRNA depletion kit (Illumina, San Diego, USA), prior to library preparation with the TruSeq RNA library preparation kit (Illumina, San Diego, USA) following the manufacturer’s instructions. The library was quality checked and sequenced on an Illumina MiSeq platform with 250 bp paired-end reads. FASTQ files were quality checked and trimmed with BBDuk2, before mapping to a reference genome created from the metagenomic dataset (PRJNA990547) to obtain the full-length genomes.

Jeilongvirus detection and quantification

To screen for Jeilongvirus in the individual lung, spleen and kidney tissues of small mammals (n = 100), real-time qPCR (RT-qPCR) was performed using in-house designed primers, probe and positive control (179 bp minigene plasmid, pUCIDT, AMP) that targeted the Jeilongvirus L gene. The qPCR primers and probe were designed using Geneious Prime 2022.1.1. Unbiased metagenomic reads extracted from the R. tanezumi metagenomic dataset (PRJNA990547) were mapped to the previously obtained Jeilongvirus genome. Within the L gene region a 99% threshold was set for generating a consensus sequence, which did not include any ambiguities, that was used for oligo design.

The qPCR assay was optimized prior to screening with the AgPath-ID One Step RT-PCR kit (Applied Biosystems, Waltham, USA). A dilution range with triplicates of the positive control plasmid was used to generate a standard curve. The 25 µl reaction contained 0.5 µm of each forward (5’ - TTACAGGTCATGACGTCGGA − 3’) and reverse (5’ - TGGAATCTCGGGCTAGAGGG − 3’) primers, 0.25 μm probe (FAM - ACAATGAAAGGTTGACAAGCCA - BHQ1), 12.5 µl of 2 ´ buffer, 1 µl of 25 ´ reverse transcriptase enzyme mix and 5 µl of RNA template. The reaction profile was as follows: Reverse transcription at 50 °C for 15 min, followed by an initial denaturation step at 95 °C for 15 min and 45 cycles of 95 °C for 15 s and 54 °C for 30 s. Positive controls at an appropriate dilution were used in all qPCR runs. The optimized RT-qPCR assay generated a R² of 0.995, an amplification efficiency (E) of 1.92, a limit of detection (LOD) of 144 copies per reaction and a limit of quantification (LOQ) of 425 copies per reaction. Viral copy numbers were measured in copies/mg based on the standard curve equation.

Phylogenetic analysis

Phylogenetic analysis was conducted with Jeilongvirus genomes (n = 3) from this study and available Paramyxovirus sequences (n = 92) from GenBank. Multiple sequence alignments were performed using MAFFT v7 [19] in Geneious Prime 2022.1.1., and the L gene phylogenies were constructed with maximum likelihood method in RAxML v8 under the general time reversible (GTR + G) nucleotide substitution model. Branch support was assessed using 1000 bootstrap replicates. Phylogenetic analysis of the generated whole genomes was also constructed as mentioned above.

Statistical analysis

We fit hierarchical Bayesian models with a Bernoulli distribution to reflect the temporal and spatial structure in our data and the binomial response variable of negative or positive for Jeilongvirus. To accommodate this hierarchical structure, we used multilevel regression models with a three-level structure, nesting individual positives within districts and years. The use of a varying effects model allows us to use partial pooling which provide better inferences on the estimates of prevalence, especially, in groups where sample size is low. Non-centered parameterization was utilized to navigate the posterior more efficiently. Priors were chosen to be weakly regularizing to control for both under- and overfitting of the model to the data. Convergence criteria, such as effective sample sizes and R-hat values, were used to check for appropriate model convergence throughout, and trace plots were inspected for signs of incomplete mixing when necessary. All models were fit with the ‘rethinking’ [20] package as a high-level interface to the Stan platform for statistical computing [21], model results and visualizations were conducted using the ‘tidybayes-rethinking’ [22] and ‘ggplot2’ [23]. All computing was conducted in ‘R’ [24]. To estimate the probability of RT-qPCR positivity among species, a three-level hierarchical Bernoulli distribution model was fit with district and year as the primary two levels and species, age, and sex as the tertiary level. To estimate the probability of RT-qPCR positivity among habitat types, a three-level hierarchical Bernoulli distribution model was fit with the interaction between district and year as the primary two levels and habitat as the tertiary level. For our numerically generated posterior samples, an estimated median probability with two-tailed 89% Bayesian credible intervals (BCI) for each parameter was reported (essentially, Bayesian confidence intervals).

Results

A total of 100 small animals from eight rodent and shrew species were collected in Singapore from 2012 to 2016 (Fig. 1). The species were identified as Mus castaneus (n = 21), Rattus exulans (n = 2), R. norvegicus (n = 20), R. sp. (n = 1), R. tanezumi (n = 32), R. tiomanicus (n = 3), Sundamys annandalei (n = 15), and Tupaia glis (n = 6). Animals were captured from four habitat types and 17 locations across Singapore (Fig. 1). There were 53 males, 45 females and two animals of undetermined sex. Seventy-nine small mammals captured were identified as adults and the remaining were juveniles (Additional Table 1).

Metagenomic sequencing of the R. tanezumi pool yielded a total of 4,849,707 reads. Approximately 0.32% (n = 15,741) reads were classified as viral, of which 34.7% (n = 5,457) were classified as Jeilongvirus. Other viral reads detected in the R. tanezumi pool were bacteriophages and retroviruses. This dataset is available at the NCBI Sequence Read Archive under BioProject accession number PRJNA990547.

Twelve of 100 (12%) animals captured were RT-qPCR positive for Jeilongvirus (Table 1). Positive animals were from four species (M. castaneus, R. exulans, R. tanezumi and R. tiomanicus) that were trapped from May to August 2012, March 2013, September 2013, and December 2015, across seven sites (Fig. 1). Jeilongvirus was detected in spleen and kidney samples, but not in the lung tissues of all 100 animals. The kidney and spleen samples had an average viral copy number of 8.1 × 10⁶ and 3.4 × 10⁴ copies/mg respectively.

Table 1.

Amplification cycle threshold (ct) values and viral copies per mg of tissues calculated from the RT-qPCR standard curve

Sample ID	Species	District	Collection Date	Spleen		Kidney
				Ct valuea	Copies per mg spleen	Ct valuea	Copies per mg kidney
RTIO-2	R. tiomanicus	West	2012-06-08	29.6	40,986.8	25.7	179,632.8
RTIO-3	R. tiomanicus	West	2012-06-08	28.5	36,504.3	21.4	2,686,558.5
RE-1	R. exulans	Central	2012-08-28	-	-	23.4	1,501,978.1
RE-2	R. exulans	East	2015-12-11	-	-	21.0	4,253,155.6
RT-2	R. tanezumi	Central	2012-05-29	30.4	25,957.4	28.6	157,470.9
RT-5	R. tanezumi	Central	2012-05-26	-	-	21.5	4,297,980.4
RT-7	R. tanezumi	West	2012-06-05	-	-	18.8	16,334,042.9
RT-8	R. tanezumi	Central	2012-05-27	-	-	20.4	14,867,601.5
RT-16	R. tanezumi	North	2013-03-17	-	-	30.6	12,654.9
RT-19	R. tanezumi	North	2013-09-24	-	-	20.3	12,812,335.6
RT-21	R. tanezumi	North-East	2013-09-06	-	-	26.2	234,851.6
MU-21	M. castaneus	East	2015-12-06	-	-	30.4	4,842.4

- Ct value not detected; CWC Central Water Catchment, CCK Choa Chu Kang

^aLimit of Quantification cut off at 32.08 (425.64 copies per reaction)

Unbiased metagenomic sequencing was conducted on the 12 Jeilongvirus positive kidneys, with complete Jeilongvirus genomes recovered from three R. tanezumi samples (RT-05, RT-07 and RT-08). The total number of reads post-trimming and reads mapped to the reference genome are shown in Additional Table 2. The Jeilongvirus genomes from this study displayed a nucleotide similarity ranging from 97.1–100%, and an amino acid similarity of 97.7–100%. The annotated genomes have eight open reading frames (ORF) in the order: 5’- N/PVC/M/F/SH/TM/G/L -3’ (Fig. 2), consistent with the typical Jeilongvirus genome organization. The sequences generated in this study are available at NCBI Genbank (accession numbers OR233792 to OR233794) and as BioProject PRJNA990547.

Fig. 2.

Genome organization of Singapore Jeilongvirus generated in this study and representative members of Jeilongvirus (AY900001, DQ100461, JN689227), Henipavirus (KF278639) and Narmovirus (AY286409). The Singapore Jeilongvirus genomes have eight open reading frames, which encode nucleocapsid (N), phosphoprotein (P), matrix protein (M), fusion protein (F), small hydrophobic protein (SH), transmembrane protein (TM), glycoprotein (G), and RNA polymerase (L). The figure is drawn to scale and the scale bar is shown at the bottom

Maximum likelihood phylogeny of the L gene reveals that the three Jeilongvirus species collected from R. tanezumi in Singapore clustered in a monophyletic clade with other unclassified paramyxoviruses identified in small mammals from Southeast Asia, Madagascar, Reunion, and Tunisia (Fig. 3), that is a sister clade to the Wenzhou rattus norvegicus jeilongvirus 1. Phylogenetic analysis of the full genomes (Additional Fig. 1) reflects similar phylogenetic relationships as the L gene phylogeny. The genomes of all three Jeilongvirus species identified from R. tanezumi also shared low genetic similarity with other representative virus species within the same genogroup, ranging from 55.0 to 70.1% nucleotide and 51.0–74.3% amino acid identities, respectively (Table 2).

Fig. 3.

Maximum likelihood phylogeny of Jeilongvirus L gene sequences generated using the GTR + GAMMA model in RAxML. Coloured boxes indicate exemplar Jeilongvirus species. Red fonts indicate novel sequences generated from R. tanezumi in this study. Bootstrap values greater than 70.0% are indicated at branch nodes. The scale bar denotes nucleotide substitutions per site

Table 2.

Genetic similarities between Jeilongvirus genomes from Singapore (n = 3) and related Jeilongvirus genomes

Sample ID	RT-05	RT-07	RT-08	RtBi-ParaV/Tt2013	TailamV	Wenzhou Rn jeilongV 1	BeilongV	J-virus
	Nucleotide Similarity (%)
RT-05		97.09	99.97	95.82	70.50	70.14	68.05	55.80
RT-07	97.76		97.09	95.67	70.43	70.01	67.98	55.83
RT-08	99.97	97.73		95.81	70.45	70.11	67.95	55.70
RtBi-ParaV/Tt2013	96.63	96.64	96.63		70.49	70.08	67.83	55.77
TailamV	75.27	72.58	72.59	72.31		68.92	67.94	55.98
Wenzhou Rn jeilongV 1	73.69	73.58	73.82	73.50	71.77		68.00	55.23
BeilongV	71.75	71.71	71.68	71.44	71.88	71.17		55.32
J-virus	51.58	51.39	51.60	51.45	51.00	51.17	51.30
	Amino Acid Similarity (%)

Amongst the Rattus species in Singapore, R. exulans and R. tiomanicus showed higher probability of positivity for Jeilongvirus, although there was overlap of credible intervals (Fig. 4a). There were no differences in the likelihood of RT-qPCR positives across sexes and age classes (Fig. 4b and c). We also compared the likelihood of RT-qPCR positives among the four habitat types. Our data suggests that young forest and scrub habitats showed higher probability of Jeilongvirus positivity than urban and old forest habitats (Fig. 4d). There were no spatial or temporal patterns in the probability of positives (Fig. 4e), with overlapping credible intervals for the interaction term of district and year.

Fig. 4.

Results from Bayesian hierarchical models showing probability of RT-qPCR positive with 89.0% credible intervals for: (a) species, (b) sex, (c) age class (d) habitat, and (e) interaction term of year and district

Discussion

Several novel murine Jeilongviruses have been reported and sequenced in the last few decades, including J-virus, Beilong virus (BeiV) and Tailam virus (TaiV) [6, 7, 11, 25]. Continued discovery and characterization of viruses within Jeilongvirus have been augmented by global and national zoonotic surveillance programs, along with the increasing use of metagenomic sequencing. In this study, we detected Jeilongvirus in four small mammal species in Singapore, M. castaneus, R. exulans, R. tanezumi, and R. tiomanicus, suggesting a wide rodent host range. This is consistent with previous studies where Jeilongvirus detection occurred across a variety of small mammal species in countries such as Thailand, China, and Madagascar [26–28].

Phylogenetic analysis shows that the Jeilongvirus species divide into two major clusters, with the bats and felines falling in one cluster and small mammals in the other. The Singapore Jeilongvirus falls within the small mammal subclade, grouping with unclassified Paramyxoviruses from small mammals sampled in islands of the Indian Ocean and in Africa. Previous studies indicate a geographical and host correlation of J-virus and BeiV in mainland China, specifying that well-defined lineages circulated widely within their taxonomical host groups [26, 29, 30].

The results from the hierarchical model analysis showed no differences among species, life history traits, or habitat types. The first and second hierarchies in the model consisted of district and year which correspond respectively to spatial and temporal patterns in Jeilongvirus positivity, that were consistent in Singapore. The estimated posterior distribution from a Bayesian analysis is structured by the sample size of the data. For M. castaneus, where there was one positive individual, we can still estimate the probability of positivity, based on partial pooling, however the large variance is a result of the small sample size. We realize that it may be difficult to generalize this finding to the entire extent of M. castaneus distribution or among the species that we captured as part of this study. A larger sample size of M. castaneus may have given us more precise insight into these relationships. However, our initial findings can act as prior knowledge for future investigations into M. castaneus as a reservoir for Jeilongvirus.

There were no differences of positivity among small mammal life history traits such as sex and age, or among habitat types, although young forest and scrub habitats exhibited slightly higher positivity estimates than other habitats. Bayesian models perform well with unequal sample sizes since sample size is a component of the posterior distribution and is reflected in the variance components of the habitat positivity estimates. Larger sample sizes may have reduced the size of the credible intervals. By using a hierarchical Bayesian approach, we maximized the information in the data to arrive at the results presented. Our model results illustrate the generalist nature of this virus and can serve as a foundation for further study on the dynamics of Jeilongvirus.

At present, the pathology and clinical representation of Jeilongvirus remain unclear. A member species, J-virus, was first isolated in mice with hemorrhagic lung lesions [6] and its pathogenicity was investigated in mice experiments [31, 32]. While newly established member species have been identified in a variety of organs, including the heart, kidney, and liver [5, 11, 26], Jeilongvirus was detected mainly in kidney and spleen and was absent in the lung in this study, supporting the previous observations that Jeilongvirus potentially results in a systemic infection [26].

Conclusions

There is reported evidence of human seroconversion to J-virus [25], however the zoonotic potential of Jeilongvirus remains unknown. As a compact and highly urbanized island, Singapore holds many human and animal interfaces. Robust routine surveillance including increased number of animals sampled and increased spatio-temporal extent of sampling across both populations, combined with virus isolation and serological studies is important to further improve our understanding towards this understudied genus and help instigate the management of a potential spillover.

Abbreviations

BeiV

Beilong virus

F

Fusion

G

Attachment

L

Large polymerase

LOD

Limit of detection

LOQ

Limit of quantification

M

Matrix

MMLV-1

Mount Mabu Lophuromys virus 1

MMLV-2

Mount Mabu Lophuromys virus 2

N

Nucleocapsid

NGS

Next generation sequencing

ORFs

Open Reading Frames

P

Phosphoprotein

PBS

Phosphate buffered saline

RT-qPCR

Real time quantitative polymerase chain reaction

rRNA

Ribosomal RNA

SH

Small hydrophobic

TaiV

Tailam virus

TM

Transmembrane

VTM

Viral transport medium

Authors’ contributions

Sample collection and processing: DHWL, SAB.  Laboratory work: LC, ZAO, DHWL.  Phylogenetic analysis: LC, DHWL, RZ, YCFS, IHM.  Statistical modelling: ATH.  Manuscript writing and editing: All.  Conceptualization and design: GJDS, IHM.  Final approved manuscript: All.

Funding

This study was supported by the Duke-NUS Signature Research Programme funded by the Ministry of Health, Singapore.

Availability of data and materials

All metagenomic data and sequences generated from this study are available in NCBI Genbank in BioProject PRJNA990547 and accession numbers OR233792 to OR233794.

Declarations

Ethics approval and consent participate

All small mammal research was approved by the National University of Singapore Institutional Animal Care and Use Committee (IACUC protocol approval B01/12), National Parks (Permit no NP/RP12-004) and Agri-Food and Veterinary Authority of Singapore (Permit no AV16.01.004.0004).

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.