Widespread recombination, reassortment, and transmission of unbalanced compound viral genotypes in natural arenavirus infections

Abstract

Arenaviruses are one of the largest families of human hemorrhagic fever viruses and are known to infect both mammals and snakes. Arenaviruses package a large (L) and small (S) genome segment in their virions. For segmented RNA viruses like these, novel genotypes can be generated through mutation, recombination, and reassortment. Although it is believed that an ancient recombination event led to the emergence of a new lineage of mammalian arenaviruses, neither recombination nor reassortment has been definitively documented in natural arenavirus infections. Here, we used metagenomic sequencing to survey the viral diversity present in captive arenavirus-infected snakes. From 48 infected animals, we determined the complete or near complete sequence of 210 genome segments that grouped into 23 L and 11 S genotypes. The majority of snakes were multiply infected, with up to 4 distinct S and 11 distinct L segment genotypes in individual animals. This S/L imbalance was typical: in all cases intrahost L segment genotypes outnumbered S genotypes, and a particular S segment genotype dominated in individual animals and at a population level. We corroborated sequencing results by qRT-PCR and virus isolation, and isolates replicated as ensembles in culture. Numerous instances of recombination and reassortment were detected, including recombinant segments with unusual organizations featuring 2 intergenic regions and superfluous content, which were capable of stable replication and transmission despite their atypical structures. Overall, this represents intrahost diversity of an extent and form that goes well beyond what has been observed for arenaviruses or for viruses in general. This diversity can be plausibly attributed to the captive intermingling of sub-clinically infected wild-caught snakes. Thus, beyond providing a unique opportunity to study arenavirus evolution and adaptation, these findings allow the investigation of unintended anthropogenic impacts on viral ecology, diversity, and disease potential.

Fig 1. Numerous viral S segment genotypes were present in infected animals, including recombinant genotypes.

Multiple sequence alignments of NP and GPC CDS were used to create a Bayesian phylogeny, which is depicted as a paired co-phylogeny. Lines connect CDS from individual genome segments. Lines are colored according to genotype. Red lines indicate recombinant genome segments listed in Table 2. Taxa are labeled by snake # and genotype #. For example “3–1” corresponds to snake #3 genotype S1. Sequences from snakes in Europe are labeled BAv_N (Boa AV NL) and UHV-1.

Fig 2. Numerous viral L segment genotypes were present in infected animals, including recombinant genotypes.

Multiple sequence alignments of Z and L CDS were used to create a Bayesian phylogeny, which is depicted as a paired co-phylogeny. Lines connect CDS from individual genome segments. Lines are colored according to genotype. Red lines indicate recombinant genome segments listed in Table 2. Taxa are labeled by snake # and genotype #. For example “35–18” indicates snake #35 L18. Sequences from snakes in Europe are labeled BAv_N (Boa AV NL) and UHV-1.

Fig 3. Phylogeny of representative snake and mammalian arenaviruses provides an overview of arenavirus diversity.

Representative snake and mammalian arenavirus sequences were collected and used to create a multiple sequence alignment of L CDS, which was used to create a Bayesian phylogeny. Red lines indicate Old World mammalian arenaviruses and blue lines New World viruses. Sequences accession numbers are indicated.

Fig 4. Individual snakes are infected by complex unbalanced sets of viral genotypes.

These tables depict the fractional abundance of S and L genotypes detected in individual animals. Each row corresponds to an individual animal. Each column corresponds to a particular S or L segment genotype. Phylogenies on top of the tables were created using representative sequences from each genotype and a neighbor joining clustering method. Shading of cells indicates the fractional abundance of that genotype in the indicated animal, which was calculated as the proportion of sequencing reads mapping to that genotype divided by the total number of arenavirus-mapping reads from that animal. Recombinant segments are depicted with a triangle. All shaded boxes correspond to coding-complete assemblies, except for those indicated with a circle. Groups of animals harboring similar virus genotype combinations that were housed together are indicated with brackets. Neg snake is a sample from an uninfected snake and HeLa is a sample from total HeLa cell RNA.

Fig 5. Examples of recombinant genotypes.

(A) An example recombinant S segment: S8 from snake #35. Plots of pairwise nucleotide identity in 100 nt sliding window between this segment and two other 2 segments with sequences similar to presumed parental genotypes: S6 from snake #35 and S11 from snake #34. A cartoon of the recombinant segment is depicted, as is the approximate location of the recombination junction. (B) An example recombinant L segment (snake #24 L10) is depicted as in (A).

Fig 6. Recombinant genome segments with unusual organizations.

Recombinant S and L genome segments with unusual double-intergenic regions (2xIGR) and/or partial coding sequences are depicted as cartoons. Plotted below each cartoon are coverage levels (the number of sequencing reads supporting each base in the assembly) and predicted free energy of folding (i.e. predicted RNA secondary structure; -∆G) of 140 nt sliding windows. Approximate locations of recombinant breakpoints are indicate with triangles and dotted lines. Partial coding sequences are indicated. Some of these segments, or very closely related versions thereof, were detected in multiple animals, as indicated. In these cases, cartoons and plots are based on the segment from the snake listed in bold font.

Fig 7. Transmission of multiple genotypes in a natural infection: (A) A cartoon and timeline depicting the circumstances surrounding the cohabitation of snakes #35 and #36.

(B) Viral genotypes detected in snakes #35 and #36 at indicated time points. Fractional abundance of genotypes depicted as in Fig 4.

Fig 8. Virus populations replicate as stable ensembles in culture: (A) Liver homogenate from snake #38 was applied to cultures of JK cells and replication was monitored by measuring supernatant viral RNA levels using qRT-PCR and genotype-specific primers.

Levels of distinct S and L genotypes detected are indicated and are normalized to the amount of S RNA detected in the first time point. Points and error bars represent mean and standard deviation of two independent experiments. (B) As in (A), but a liver homogenate from snake #47 was used as inoculum. (C) The 2xIGR L4 segment detected in snake #47 replicates stably in culture. Same experiment as (B), but qRT-PCR used primers that targeted two different regions of the L4 segment as depicted in the inset cartoon.

Fig 9. Most or all virus particles contain one L genotype.

End-point dilution experiments were performed to determine the viral genotype of individual virus particles. Serial dilutions were prepared and applied to JK cells in 96-well plates. After 7–10 days, supernatant was reserved and wells were stained with anti-NP antibody to identify infected wells. qRT-PCR using discriminating primers was used to genotype the virus in individual well supernatant. Each row corresponds to one well and each column to an S or L genotype. The fractional abundance of S and L genotypes detected in individual wells are indicated as in Fig 2 as are the dilution used to inoculate the genotyped well. Negative wells (“neg”), not staining with anti-NP antibody, from the highest dilutions served as negative controls. The amount of each genotype detected in the inoculum is also indicated. Inoculum from snake #37 liver was used in (A) and from snake #47 liver in (B).