Introduction

Hantaviruses are tri-segmented negative-stranded RNA viruses within the family Hantaviridae that can cause two severe diseases in human populations, namely hemorrhagic fever with renal syndrome (HFRS) and hantavirus cardiopulmonary syndrome (HCPS). Hantaviruses in the Americas are more closely associated with HCPS, which is characterized by fever, headache, myalgia, hypotension, and thrombocytopenia that can progress to cardiopulmonary failure. The mortality rate for HCPS is estimated between 15–50% and varies among virus species and across countries [1–5].

Despite high mortality rates in humans, hantaviruses are maintained naturally in rodent populations and can persist for months to the lifetime of the animal [6]. Infected rodents shed virus through saliva, urine, and feces which can form aerosols that can be inhaled by other rodents or humans [7]. In Panama, the Costa Rican pygmy rice rat (Oligoryzomys costaricensis) serves as a rodent reservoir for Choclo virus (CHOV, Orthohantavirus chocloense). O. costaricensis is distributed across areas which are experiencing high levels of habitat conversion from natural to agricultural lands. This transformation is hypothesized to increase population densities of commensal rodents and subsequent human contact with infected rodents, ultimately increasing zoonotic pathogen transmission [8]. CHOV was first identified by RT-PCR after an outbreak of HCPS in the agroecosystems of western Panama from December 1999 to February 2000 [9,10]. Monitoring of human and rodent populations in Panama over the past two decades have discovered multiple hantaviruses in Panama (e.g., Calabazo virus and Rio Segundo virus); however, CHOV is responsible for almost all human cases [11–13]. From 2001 to 2007 multiple community-wide surveys of western Panamanians without reported HCPS symptomology found 16–60% of the individuals were hantavirus seropositive depending on region [14], documenting that many mild or asymptomatic exposures were not accounted for in clinical case count data. CHOV-associated disease generally has a lower case fatality rate (mean 7.9%, range from 3% to 50.0%) than other New World Hantaviruses, but unfortunately predominantly affects young people (ages 20–49) [12].

Despite being first reported more than 40 years ago, hantaviruses are often described as ‘emerging’ pathogens due to their increasing number of infections, global distribution, and great breadth of pathogen diversity [15]. Current proactive approaches aimed at pathogen prediction are utilizing hantaviruses as a model for understanding spillover events [16]. Diverse specimens of wild mammals archived in museum biorepositories over temporal and spatial scales are increasingly being utilized for surveillance and characterization of emerging diseases [17–20]. For instance, the first complete CHOV sequence was obtained from archival Oligoryzomys costaricensis (= fulvescens, [21]) splenic tissue (MSB:Mamm:96073) using Sanger sequencing [13]. Of the 69 CHOV sequences available in NCBI GenBank, only nine are considered complete segments, eight of which are derived from the same voucher specimen (accessed August 10, 2023). Here, we deep sequenced archived mammalian tissue specimens to generate a metatranscriptome-assembled CHOV genome doubling the total available CHOV genomes.

Methods

Results and Discussion

An outbreak from late 1999 to early 2000 of hantavirus cardiopulmonary syndrome (HCPS) in western Panama [9,10] has spurred over two decades of epidemiological and wildlife surveillance [12,36]. During those 20 years 712 clinical cases of HCPS were reported in Panama [12] and >11,000 specimens of non-volant mammals with archived biological materials were contributed to museum repositories (https://arctos.database.museum/). Of these, 883 have been identified as O. costaricensis, and 778 (88%) have been screened for prior hantavirus infection using an IgG strip immunoblot assay [39] with an average seropositivity of 16% [36]. We requested samples designated hantavirus positive at the University of New Mexico MSB for metatranscriptomic sequencing. Initial taxonomic assignment by Kraken2 classified < 15 reads Orthohantavirus in all samples except MSB:Mamm:131232, for which it identified 658 reads. Hantavirus-specific IgG antibodies can be detected up to six months after initial infection [40]. Given that 90% of the hantavirus-positive samples had 0 to 14 reads, there was likely no active infection.

We then took a reference-based mapping approach to maximize recovery of CHOV reads. We generated 113,237,341 reads from deep sequencing the metatranscriptome of MSB:Mamm:131232, of which 2,072 reads (0.00002%) mapped to the CHOV reference. The average depth of coverage from the mapping-based alignment was 16X; however, there was variation between segments with the greatest coverage of the small (S) segment (25X) followed by the medium (M) segment (19X) and the large (L) segment (12X) (Fig 1). Due to the low abundance of sequencing reads mapping to CHOV, we implemented a 3X coverage threshold of base-calling in the consensus genome. If the coverage at a site was below 5X, we only called a base when the allele frequency was 100% at that site. Using these parameters, we recovered a 91% complete CHOV assembly. Similar coverage thresholds have aided in the recovery of more complete assemblies [41,42]. This was an improvement upon our initial de novo assembly which was only 64% complete. Under these thresholds, there were 1,199 ambiguous base calls of which 93% (1,114 bp) were in the L segment. Therefore, completeness of the L segment (83%) was less than the M (99%) and the S (98%) segments. While our L segment is not complete, this marks a major improvement compared to many of the hantaviruses deposited in GenBank which are missing the L segment (ELMCV, LANV, and NYV) or only sequenced small portions of the L segment (BCCV). The International Committee on Taxonomy of Viruses Hantaviridae Study Group is revisiting requirements for proper hantavirid classification including standards for minimum sequence quality and genome completeness [43] as well as minimum information standards for uncultured viral genomes [44]. Limited biological material precluded additional sequencing necessary for genome completion and verification of low coverage base-calls. However, our efforts are still highly valuable as they provide the second CHOV, which will aid in our ability to design PCR amplicon tiling array and hybridization target capture sequencing strategies that can lead to affordable and scalable targeted sequencing efforts [45]. Such protocols that enable high throughput and cost-effective means to generate robust genetic information are needed for future genomic comparative studies and systematic exploration of understudied viruses, such as CHOV.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Coverage plot of sequencing reads mapped to the S, M, and L segments of the CHOV reference genome.

https://doi.org/10.1371/journal.pntd.0011672.g001

To place our CHOV genome into the phylogenetic context of other hantaviruses, we performed single and concatenated segment alignments and generated maximum likelihood phylogenetic trees. Our sequence fell into a well-supported clade (≥99% bootstrap values) with the CHOV reference genome from voucher MSB:Mamm:96073 (Figs 2A and S1). Both the reference and our sequence were isolated from O. costaricensis specimens from Los Santos, Panama, and were captured 23 months apart (Fig 3). Excluding ambiguous base calls (Ns), there are 125 single nucleotide polymorphisms (SNPs) differing from the CHOV reference of which 40 SNPs were in the L segment, 58 SNPs were in the M segment, and 27 SNPs were in the S segment. This corresponds to a divergence of 1.1% from the CHOV reference genome. This variation resulted in eight non-synonymous mutations with five located in the M segment (p.D246G, p.S247F, p.K510R, p.A658T, and p.V1133G) and three in the L segment (p.L342S, p.V412I, and p.N778S) (S3 Table). Overall, the limited number of amino acid differences discovered across two samples collected nearly a year apart, but within the same geographic region is interesting. However, to understand the true extent of genomic heterogeneity of CHOV regionally and temporally, generating additional complete genomes from both rodents and humans are necessary. It should be noted, that while the CHOV reference genome was generated with conventional Sanger sequencing (average accuracies of ~99.4%), Illumina next-generation sequencing (NGS) produces comparable sequence accuracies for reads with phred-like quality scores above 30, and overall accuracy increased with increasing depth of coverage [46]. The M segment encodes for two glycoproteins (G1 and G2), involved in signal transduction and mitigation of host antiviral defenses, positions 22–545 and 658–1136, respectively. While amino acid substitutions from lysine (K) to arginine (R), or alanine (A) to threonine (T) are less likely to affect protein structure because of similar side chain polarity and charge, modifications from aspartic acid (D) or valine (V) to glycine (G) are more likely to cause structural changes because of differing side chain biochemistry. Structural and functional studies of the CHOV L segment, similar to those in SNV [47], should be investigated. Non-synonymous mutations can have effects on protein stability and structure or alter protein-protein interactions [48]. In addition, variation in the non-coding regions at the 5′ and 3′ termini could disrupt complementary sequences which aid in the panhandle structure formation essential for transcription and replication [49]. These regions are generally highly conserved, and we observed only one SNP in the S segment (g.T29C), which is not predicted to interfere with panhandle formation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2.

Maximum likelihood phylogenies of the S and M segments and a concatenated alignment (A) and a tanglegram used to visualize possible reassortment histories (B). Phylogenies were built with the GTR+GAMMA model with 10,000 ultrafast bootstraps and 10,000 bootstraps for the SH-aLRT on single segment and concatenated alignments. In the tanglegram, lines connect the same taxa/tip in each tree to one another such that crossing lines suggest topological discordance.

https://doi.org/10.1371/journal.pntd.0011672.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Geographic distribution of Panamanian CHOV sequences from 32 capture sites of the Costa Rican pygmy rice rat (O. costaricensis) and approximate residence of 21 clinical cases georeferenced using the Datum UTM, WGS 1984, with ArcMap software from ArcGIS (ESRI2019).

Red dots equating to the overlap in the locality of human and rodent sequences represent six rodents and one human in Los Santos, two rodents and two humans in Veraguas, and two rodents and one human in Coclé.

https://doi.org/10.1371/journal.pntd.0011672.g003

Hantavirus evolution has likely been shaped by their co-evolutionary history with their rodent reservoirs [50]. The phylogenetic relationships between South American and North American hantaviruses from sigmodontine rodents (Fig 2B) are most likely derived from a complex history of co-speciation events and the biogeographic constraints that influenced rodent expansion into South America [51]. However, intra- and inter- lineage reassortment between closely related variants have been reported for several hantaviruses [52–56]. Although assumptions of reassortment events are often based on conflicting phylogenetic tree topologies, reassortment has also been demonstrated in in vitro experiments [57–59]. The high degree of genomic similarity in the S and L segments suggests the exchange of the M segment is more common and potentially beneficial [60]. We did not find evidence of reassortment between the two CHOV genomes (Fig 2B). With additional genomic variants, this question should be revisited with more robust analyses [61].

One outstanding question is the relatedness of hantaviruses isolated from rodents to human clinical cases. A phylogenetic analysis of all publicly available partial S segment CHOV sequences clearly demonstrates viruses isolated from O. costaricensis, including the strain described here, are intimately related to all clinical strains from Panama (Fig 4). Both clinical and rodent CHOV strains were captured from Los Santos, Veraguas, and Coclé provinces; however, only rodent-acquired sequences were found in Panamá (Fig 3). This is reflective of the clinical disease burden with the greatest number of cases in Los Santos (77%) followed by Veraguas (12%) and Coclé (7%) [12]. However, it should be noted that locations of clinical cases are by residence, and it is possible patients were exposed while traveling to another endemic area of the country. The strains from the province of Panamá form a well-supported clade demonstrating potential geographic substructure (Fig 4) which is congruent with previous findings [36]. More sequencing is needed to determine how closely these are related to clinical strains; however, only seven Panamá residents have reported HCPS in the last 20 years (1% of all cases) [12]. Five other sequences from Panamanian hantaviruses were isolated from the short-tailed cane mouse (Zygodontomys brevicauda) and the Chiriqui harvest mouse (Reithrodontomys creper), but these are likely representative of other unclassified hantaviruses (e.g., Calabazo virus and Rio Segundo virus) that have yet to be fully sequenced [43] or associated with human disease (S2 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. A maximum likelihood phylogeny of the partial S segment of 51 CHOV genomes demonstrates sequences isolated from O. costaricensis are associated with human disease.

The phylogeny was built with the GTR+GAMMA model with 10,000 ultrafast bootstraps and 10,000 bootstraps for the SH-aLRT. Bootstrap values ≥ 65% are shown. The sequence in bold was acquired in this study.

https://doi.org/10.1371/journal.pntd.0011672.g004

Virus discovery has been dramatically accelerated with the advances in sequencing [62]. Here, we provide a proof-of-concept for using deep metatranscriptomics to recover a hantavirus genome from archived museum tissue samples. Viral enrichment protocols prior to sequencing, including those that utilize fraction and filtration methods, may even increase recovery of target organisms [63,64]. Other methods, such as PCR or target enrichment by capture [45], that seek to enrich virus sequences during library prep, require a priori knowledge of genomic sequences to preform optimally, and will be assisted by the metatranscriptomic sequencing described here. Furthermore, we demonstrate that tissues from voucher MSB:Mamm:131232, which were ultrafrozen for 20 years before our current sequencing efforts, were still viable for NGS, highlighting the longevity and utility of such preserved specimens. Our understanding of hantavirus systematics, evolution, and ecology have been profoundly influenced by integrating collaborative efforts between public health agencies, virologists, field biologists, and museum scientists. Museum specimen collections have been leveraged to document the distribution of SNV [65] and the discovery of multiple novel hantaviruses [66]. Utilizing these invaluable archives is essential to understanding pathogens circulating past and present. Given that habitat conversion [8] and climate change [67] could influence reservoir host densities and therefore increase the burden of CHOV and other hantaviruses, it is necessary to continue capturing sequence data to inform diagnostics, detection, and vaccine design.

Supporting information

Acknowledgments

The authors thank the field crews and museum scientists at the Instituto Conmemorativo Gorgas de Estudios de la Salud (Panama City) and Museum of Southwestern Biology (Albuquerque) for collecting and preserving archival tissues of the rodents. Specifically, we thank the Panamanian Ministry of Environmental Affairs, the Panamanian Ministry of Health, the Panamanian Institute of Livestock and Agricultural Research, the Gorgas Committee for Animal Care and Use, and the International Center for Infectious Disease Research Program of the National Institutes of Health for their support. We also thank the UNM Center for Advanced Research Computing for providing high performance computing resources.