Hantaviridae: Current Classification and Future Perspectives

In recent years, negative-sense RNA virus classification and taxon nomenclature have undergone considerable transformation. In 2016, the new order Bunyavirales was established, elevating the previous genus Hantavirus to family rank, thereby creating Hantaviridae. Here we summarize affirmed taxonomic modifications of this family from 2016 to 2019. Changes involve the admission of >30 new hantavirid species and the establishment of subfamilies and novel genera based on DivErsity pArtitioning by hieRarchical Clustering (DEmARC) analysis of genomic sequencing data. We outline an objective framework that can be used in future classification schemes when more hantavirids sequences will be available. Finally, we summarize current taxonomic proposals and problems in hantavirid taxonomy that will have to be addressed shortly.


Introduction
Recent environmental, animal, and plant metagenomic studies have resulted in an avalanche of viral genomic sequencing data, vastly expanding the known virus biodiversity [1][2][3][4][5][6][7][8]. These advancements in the field of virus discovery led to a striking discrepancy between the number of potential new viral taxa described in literature and the number of officially recognized taxa by the International Committee on Taxonomy of Viruses (ICTV) [9]. Reasons for the backlog in official classification were not only the sheer number of novel viruses but also the absence of described biological properties of these viruses beyond genomic sequencing data and sequence-inferred characteristics. In the past, most ICTV Study Groups were reluctant to create new taxa in the absence of additional information on phenotypic virus properties, such as host range, antigenic relatedness, and virion morphology [9,10]. A consensus statement endorsed by the ICTV Executive Committee, explicitly permitting classification based on genomic sequence data alone (while still encouraging the acquisition of additional data) has opened the door to a much-needed reformation of the taxonomy of many virus families [9] and, therefore, an improved official depiction of the evolutionary relationships in the virosphere [11].
These criteria indirectly imply that hantavirid classification into a species requires knowledge of its natural host, significant coverage of the viral genome sequence, and virus isolation in culture. In addition, cross-neutralization experiments, typically requiring biosafety level 3 containment, should be performed. Given stringent criteria, not all hantavirid species listed in the ninth ICTV report actually meet the these criteria [17]. For a minority of hantavirids, isolates were not available. For three hantavirids, the M segment sequence was incomplete or unavailable. Furthermore, certain hantavirids can cross host species barriers in opposition to the first criterion that suggests that a distinct hantavirus should be associated with a unique ecological niche [43,44]. Moreover, not all hantavirids listed in the ninth ICTV report meet the second criterion that denotes a minimum amino acid difference of 7% in nucleocapsid (encoded by the small (S) genomic segment) and glycoprotein (encoded by the medium (M) genomic segment) amino acid sequences. Consequently, the second criterion was proposed to be changed to a difference of >10% amino acid differences of the nucleoprotein and >12% amino acid difference of the glycoprotein [45].
Taxonomy is a continuous process that needs to keep pace with virus discovery and novel methodologies. The taxonomy of Hantaviridae clearly requires a comprehensive overhaul. The rationale and methodology for the beginning of this overhaul, formulated in official ICTV taxonomic proposals (TaxoProps) 2016.023a-cM, 2017.006M, 2017.012M, and 2018.010M (https://talk.ictvonline.org/), is outlined in the next sections of this manuscript.

DEmARC Analysis for Hantaviridae
The analysis was limited to hantavirids for which coding-complete S and M segment sequences were available. The deduced amino acid sequences of the proteins encoded by these segments (nucleoprotein and glycoprotein, respectively) of all available tentative hantavirid sequences were downloaded from NCBI's GenBank. A concatenated multiple sequence alignment was generated with MAFFT v7 [46]. Bayesian phylogenetic inference was conducted in BEAST 1.8.4 [47] using 20 independent runs that continued until adequate effective sample sizes (ESS > 200) were obtained. Independent runs were combined using LogCombiner 1.8.4 (BEAST) [47], employing a burn-in of 10%. A consensus tree was built using TreeAnnotator 1.8.4 (BEAST) [47] with the maximum clade credibility method and visualized in FigTree v1.4 [48]. This consensus tree was used as a guide tree for the DivErsity pArtitioning by hieRarchical Clustering (DEmARC) analysis [49,50]. Pairwise evolutionary distance (PED) values were calculated using a maximum-likelihood approach with a WAG substitution model in Tree-Puzzle. A PED cut-off value of 0.1 was used for species demarcation within Hantaviridae.

Phylogenetic Inference for the Bunyavirales
The polymerase amino acid sequences of significant representative members of Bunyavirales were extracted from NCBI's GenBank. In addition, new sequences stemming from viruses likely to be related to order members, including Jiāngxià mosquito virus 2 (JMV-2) [4], were considered in the analysis. Multiple sequence alignment was performed with MAFFT v7 after which a Bayesian phylogenetic reconstruction was conducted with BEAST 1.8.4. Two independent Markov Chain Monte Carlo analyses were run until adequate ESS were obtained. A consensus tree was built employing a burn-in of 10% in TreeAnnotator 1.8.4.

Change of Demarcation Criteria
To establish an impartial hantavirid classification that is easily reproducible and adheres to the consensus about the exclusive use of sequencing data, we abandoned the ninth ICTV report's species demarcation criteria and instead applied a classification approach based solely upon genetic data. DEmARC analysis was used to objectively define classification ranks based upon PED [49] and to establish taxonomic revisions of Hantaviridae in consecutive years since 2017.
Ideally, sequence-based classification relies on complete or at least coding-complete genome sequences [51], which, in the case of hantavirids, would be sequences of the three genomic segments S, M, and large (L). Unfortunately, for a large number of hantaviruses, availability of coding-complete sequences is limited, and, in particular, L segment sequences are frequently missing. Because the coding sequence of a single genomic segment does not contain sufficient information to achieve meaningful classification, we used a multiple sequence alignment of concatenated amino acid sequences of the S and M segments. DEmARC analysis gave a frequency distribution of PED values of which a threshold of 0.1 gave an optimal clustering cost of zero. Consequently, this threshold was selected as a hantavirid demarcation criterium at the species rank. Genera are demarked by a PED-value threshold of 0.95. Subfamilies are demarked based on their distinct clustering in the maximum clade credibility tree (see Figure 3) and a PED-value threshold of 3.5.

Addition of New Taxa to Hantaviridae
Numerous new hantavirid species were incorporated into the ICTV-official taxonomy based on DEmARC analysis in 2017. Of these, 8 hantavirids have rodents as their natural hosts, whereas 3 newly discovered hantaviruses infect bats, 5 infect moles, and 8 infect shrews. Today, these viruses are distributed among the four mammantavirin genera Loanvirus, Mobatvirus, Orthohantavirus, and Thottimvirus (Table 1) [52].

Addition of New Taxa to Hantaviridae
Numerous new hantavirid species were incorporated into the ICTV-official taxonomy based on DEmARC analysis in 2017. Of these, 8 hantavirids have rodents as their natural hosts, whereas 3 newly discovered hantaviruses infect bats, 5 infect moles, and 8 infect shrews. Today, these viruses are distributed among the four mammantavirin genera Loanvirus, Mobatvirus, Orthohantavirus and Thottimvirus (Table 1) [52].
In 2018, viral metagenomics led to the discovery of new hantavirids in reptiles, jawless fishes (Agnatha), and ray-finned fishes (Actinopterygii) [7]. In line with the DEmARC analysis results, five additional hantavirid species were created in three genera: Actinovirus (Actantavirinae), Agnathovirus (Agantavirinae) and Reptillovirus (Repantavirinae). In addition, the complete genome sequences of 2 additional orthohantaviruses became available ( Using metagenomics, Jiāngxià mosquito virus 2 (JMV-2) was discovered. This virus is a highly divergent virus most closely (albeit very distantly) related to hantavirids [4] that has subsequently been described as the first mosquito-borne hantavirid [40]. However, phylogenetic analysis of the amino acid sequence of the coding-complete sequence of the JMV-2 L segment demonstrates that JMV-2 is divergent from all hantavirids and likely represents a novel family in Bunyavirales (Figure 2).

Abolishment of Hantavirid Taxa and Declassification of Hantavirids
From the introduction of new objective classification criteria based on sequence data, 8 previously recognized hantavirid species were abolished because they did not fulfill all criteria used for DEmARC analysis-based classification: •

Creation of Subfamilies and Genera within Hantaviridae
The recent discoveries of hantaviruses in a wide spectrum of host species have significantly increased the known hantavirus diversity. Phylogenetic inference divides Hantaviridae in well-supported subclades ( Figure 3). These taxonomic sub-groups are now better defined by the introduction of genera and subfamilies (Table 3).

Creation of Subfamilies and Genera within Hantaviridae
The recent discoveries of hantaviruses in a wide spectrum of host species have significantly increased the known hantavirus diversity. Phylogenetic inference divides Hantaviridae in wellsupported subclades (Figure 3). These taxonomic sub-groups are now better defined by the introduction of genera and subfamilies (Table 3).

Megataxonomy of Hantaviridae
A recent global phylogenetic analysis confirmed the monophyly of negative-sense RNA viruses [11]. A top taxonomic rank was introduced by the ICTV for all RNA viruses [80] including a phylum, 2 subphyla, and several classes for negative-sense RNA viruses [81,82]. The current megataxonomy of Hantaviridae is outlined in Table 4.

Future Taxonomic Perspectives
In 2020, hantavirid taxonomy will likely only change minimally, because only a single TaxoProp has been submitted by the 2019 submission deadline. This TaxoProp outlines the addition of one loanvirus species for the recently discovered Brno virus (BRNV) [83]. The megataxonomic placement of Hantaviridae will likely remain steady, but phylum Negarnaviricota will likely be included in the newly proposed kingdom "Orthornavira".
Novel TaxoProps are already expected to be submitted by the next submission deadline in 2020 for the 2021 taxonomy cycle to accommodate several recently described putative mobatviruses [84,85]. Furthermore, the ICTV Hantaviridae Study Group is currently discussing whether hantavirids, for which coding-complete S+M+L genomic segment sequences are not available, ought to be declassified and whether hantavirid name abbreviations ought to be unique (and be changed if they are not). The ICTV Hantaviridae Study Group is also discussing how species "complexes" (species that harbor more than one member virus) could be resolved, and how hantavirid species names could be changed to Linnaean binomial names [86] should this become an ICTV requirement.

Discussion
The current hantavirid taxonomy (Table 5) is based upon concatenated amino acid sequences of S and M genomic segment-encoded proteins. To provide a more robust framework, ideally only coding-complete sequences of hantavirids should be used for any classification efforts, with various methods analyzing all segments. Unfortunately, very few hantavirus genomes have been sequenced fully, precluding such a robust taxonomic classification for now. Increased sequencing efforts of partially characterized hantavirids, some of them discovered decades ago, could substantially improve future taxonomic efforts. In many cases, obtaining missing sequence information is not challenging scientifically, as most historic hantavirids have been isolated in culture. However, owing to the high sequence diversity and saturation of informative sites, classification with inclusion of the M segment might become increasingly difficult as hantavirid diversity may be enormous. Such diversity is indicated by detection of more divergent hantavirids in metagenomic samples and in fish and reptiles. Although hantavirid interspecies segment reassortment is thought to be fairly limited, reassortment events have shaped hantavirid evolution [43,44,87] and may further complicate classification efforts. We are calling on the hantavirid research community to weigh in on these issues and to contribute to taxonomic efforts, including TaxoProp writing and submission, to achieve a taxonomy that best reflects hantavirid evolutionary relationships.