Comparative Study of Ten Thogotovirus Isolates and Their Distinct In Vivo Characteristics

Over the last years, several historical specimen and new isolates were classified to the genus Thogotovirus within the Orthomyxoviridae. For the present comparative study we collected 10 different isolates of thogotoviruses from four continents that were kindly provided by colleagues all over the world (Table 1). All these virus isolates were propagated in our lab under comparable conditions in Vero or BHK cells to titers of 10⁶ to 10⁷ PFU/ml.

For many thogotovirus isolates only partial or no sequence information is available in public databases. To confirm the authenticity of our virus stocks, we analyzed their genomes by next-generation sequencing (NGS) using ultracentrifugation concentrated preparations of the 10 isolates. We determined the full-length genome sequences by ultradeep Illumina sequencing of the extracted RNAs generating 10 to 15 million reads per virus resulting in a 30.000–80.000-fold coverage depending on the individual isolate. The full-length genomes were de novo assembled and deposited in GenBank (accession numbers see Table 1). The genome of Oz virus (OzV) was not included in this NGS analysis and for further analyses the sequences from a reference genome (8) were used. Based on the deduced amino acid sequences we constructed maximum likelihood trees for each segment. The phylogenetic relationships between the isolates remained largely consistent in all six segments and showed a clear separation into two groups: The THOV-like and the DHOV-like thogotoviruses. This was furthermore, reflected by the amino acid sequence identities of the individual open reading frames (orf) that are high within (40–100%) but low (20–60%) between both clades (Fig. S1). Moreover, the phylogenetic analysis showed that JOSV (THOV-like) as well as OzV and BRBV (DHOV-like) are more distantly related members of the two clusters. Furthermore, we observed a distant relationship between OzV and BRBV in segment 4 and 6 (Fig. 1D and F) as previously suggested (8).

To investigate the morphology of thogotovirus virions, we analyzed virus-infected cells and supernatants of infected cell cultures by transmission electron microscopy. Representative images of ultrathin-sections show spherical particles with electron-dense walls budding from the surface of infected Vero cells (Fig. 2A and B). Extracellular and protruding particles contained electron dense material representing densely packaged viral proteins beneath the lipid envelope (Fig. 2A and B, insets). A detailed analysis of the clarified supernatants of virus-infected cells showed spherical particles with a diameter of around 100 nm (Fig. 2C to F). Interestingly, a substantial number of filamentous virions were observed as described previously for BRBV (9). The analyses of the four different isolates suggest that the frequency and size of these filamentous particles is higher for DHOV-like viruses (Fig. 2C to F left panel).

So far, the capacity of different thogotoviruses to replicate and cause disease in mammals has not been comparatively analyzed. To address this, we studied the replication of the different thogotovirus isolates in mice. C57BL/6 mice were intraperitoneally infected with 100 PFU of the THOV-like (Fig. 3) and DHOV-like isolates (Fig. 4). Only for the infection with the Ken-IIA and the SiAr126 isolates, we chose a 100-fold reduced infectious dose of one PFU to delay their excessive pathogenic effects until day 6. The amount of viral progeny was analyzed in liver, lung, spleen, kidney, serum and brain after two, four and 6 days postinfection. Interestingly, upon infection with the THOV-like isolates we detected viral titers in almost all organs, except the brain, and already observed infectious particles in liver and spleen 2 days postinfection. Overall, the highest titers were present in the livers of Ken-IIA and the SiAr126 infected animals, whereas JOSV, Kami-25 and PoTi503 replicated to 100 to 1000-fold reduced titers (Fig. 3). The DHOV-like isolates India/61, BKNV and PoTi461 also showed efficient replication in liver, lung and spleen (Fig. 4C to E). However, OzV and BRBV infections did not lead to measurable viral titers in all analyzed organs (Fig. 4A and B).

Overall, the spleen is one of the first organs affected by all thogotoviruses upon intraperitoneal infection. Remarkably, there was a clear difference in the organ tropism of both clusters as in contrast to the more hepatotropic THOV-like isolates (Fig. 3) the viral replication of DHOV-like isolates peaked in the lung (Fig. 4).

To further compare the virulence of the different thogotovirus isolates, C57BL/6 mice were intraperitoneally infected with serially diluted virus inoculums and daily monitored. The isolates Ken-IIA, SiAr126, India/61, BKNV and PoTi461 which grew to high titers caused severe lethal disease at a low viral dose of one PFU within five to 7 days (Fig. 5C, E, H to J). Interestingly, mice infected with the THOV-like viruses Ken-IIA and SiAr126 dramatically dropped in body weight within five to 6 days (Fig. 5C and E) and displayed severe symptoms, including lethargy, ruffed fur and hunched posture similar to the previously published clinical score (28). In contrast, mice infected with the DHOV-like isolates India/61, BKNV and PoTi461 got terminally ill in the absence of a dramatic weight loss (Fig. 5H to J). JOSV, Kami-25 and PoTi503 that did not grow to high titers (Fig. 3) and BRBV and OzV that were strongly attenuated in vivo (Fig. 4) did not cause severe disease even at the highest infection doses of 100,000 PFU per animal (Fig. 5A, B, D, F and G). Notably, mice infected with JOSV and PoTi503 (Fig. 5A and D) transiently dropped in weight at high infection inoculums, indicating a mild self-limiting disease.

In summary, these data indicate that the course of disease caused by thogotovirus infections correlates with their capacity to replicate in vivo (Table 2).

In order to evaluate the humoral immune response against thogotoviruses, we isolated and pooled postinfectious antisera from at least four C57BL/6 mice. In case of the highly pathogenic viruses, we infected congenic Mx1^+/+ mice that can potently restrict thogotovirus replication (17, 29). We subsequently investigated whether these sera are able to recognize viral antigens present in the lysates of infected cells by Western blotting analyses (Fig. 6). The antibody responses of all viruses were primarily directed against GP (58 kDa) and NP (52 kDa) with NP showing the strongest signal in all cases (Fig. 6). Importantly, the GP was reported to have a high molecular weight of 75 kDa due to its glycosylation (4, 30). Antibodies against the M protein (31 kDa) were only detected in sera of OzV infected animals (Fig. 6F). Interestingly, antisera from mice infected with closely related virus isolates showed a high degree of cross-reactivity. This was observed between the isolates within the THOV-like (Fig. 6A to E) or within the DHOV-like clusters (Fig. 6F to J) but not between the two groups. However, postinfectious sera of JOSV and BRBV, both distantly related members of the two clusters (Fig. 1), were not cross-reactive (Fig. 6A and G). Curiously, postinfectious sera of the other THOV-like isolates, except SiAr126, recognized JOSV NP, suggesting that during a JOSV infection the humoral response is exclusively directed against strain specific epitopes. In contrast, the antiserum from OzV infected animals recognized antigens of all DHOV-like viruses (Fig. 6F) but sera from India/61, BKNV and PoTi461 showed comparable cross-reaction with each other but not with antigens of OzV and BRBV (Fig. 6H to J). The variations in the specificity of the immune sera against the different viruses that sometimes recognize only NP, sometimes additionally GP or M might be due to differences in the intensity of viral protein production in the infected cells as well as the presentation of the viral antigens in the infected animals. To confirm the specificity of the antigen recognition, we performed Western blot analysis of lysates of transfected cells expressing recombinant GP, NP and M proteins of THOV SiAr126 and DHOV India/61, respectively. The recombinant proteins of SiAr126 were best recognized by the antisera directed against Ken-IIA and to a lesser extent against SiAr126 but both sera did not recognize the India/61 proteins (Fig. S2B, C). Accordingly, the anti-India/61 and anti-BKNV antisera strongly recognized the India/61 NP but there was no cross-recognition of the SiAr126 proteins (Fig. S2E, F), confirming the clade specificity shown with the lysates of infected cells (Fig. 6). Only faint bands of SiAr126 GP and NP were recognized by the anti-JOSV antiserum (Fig. S2A), confirming the more distant position of JOSV in the THOV-like clade. The cross-reactivity of the anti-OzV antiserum recognizing GP and M but not NP in the lysates of DHOV-like viruses infected cells (Fig. 6F) could be confirmed for the recombinant India/61 proteins (Fig. S2D).

Furthermore, the postinfectious antisera were tested for virus neutralization using a set of selected thogotoviruses, THOV-like SiAr126 and JOSV as well as DHOV-like India/61 and OzV. These analyses confirmed the clear separation between the THOV-like and the DHOV-like clade. Group specific antisera were only able to neutralize the respective prototypic isolate of each group (SiAr126 or India/61) (Fig. 7A and C and Fig. S3A, C). However, these two viruses were not neutralized by the anti-JOSV and anti-OzV antisera. Accordingly, JOSV and OzV were selectively neutralized by anti-JOSV and anti-OzV antisera, respectively (Fig. 7B and D and Fig. S3B, D). Unexpectedly, SiAr126 was not well neutralized by the anti-SiAr126 serum (Fig. 7A, Fig. S3A). However, the Western blot analysis of the recombinant proteins suggests that this antiserum is much less potent than, e.g., the anti-Ken-IIA antiserum in recognizing the viral SiAr126 GP and NP (Fig. S2B, C). Therefore, we tested other batches of anti-SiAr126 antisera and also the neutralization capacity of these sera batches against THOV-like PoTi503 and Ken-IIA with similar results (data not shown). Because of the extremely high virulence of the SiAr126 isolate in C57BL/6 mice these sera were obtained from Mx1^+/+-positive mice that dampen the replication and pathogenic effects of the SiAr126 infection (17). We speculate that under these restricting conditions, the cell tropism of the virus is modified (19) and the expression of viral antigens is hampered and therefore production of neutralizing antibodies attenuated. In conclusion, the main humoral response during a thogotovirus infection is directed against the viral NP and GP and with a few exceptions the produced antibodies are only cross-reactive with closely related viruses.

To evaluate tick-independent routes of virus transmission and the possibility that the infection route might influence the disease progression of thogotoviruses, C57BL/6 mice were infected with the THOV-like SiAr126 or the DHOV-like India/61 isolate by either the intraperitoneal, subcutaneous, intranasal or intravenous route. For both viruses all infection routes with 100 PFU led to a severe and fatal disease (Fig. 8A and B). However, intranasal and subcutaneous application delayed the time of death by up to 3 days for the SiAr126 isolate (Fig. 8A) and by 1 day for the India/61 isolate (Fig. 8B) in comparison to the most potent intravenous route of infection.

The successful infection via the intranasal route and the fact that both THOV and DHOV-like viruses efficiently replicated in the lung (Fig. 3 and 4) suggested the possibility of a contact or aerosol driven horizontal transmission between mice. To test for such transmission events, SiAr126 or India/61 virus infected C57BL/6 index mice were cohoused for 4 days with IFNAR^−/−/IL28R^−/− contact animals. Mice lacking these receptors are nonresponsive to type I and III interferon (IFN) and are therefore highly susceptible to viral infections and ideal sentinels to detect viral transmission (31). Virus progeny was detected in the bladder and snout of the index animals 4 days after infection, indicating a possible release of infectious material (Fig. 8C and E). The contact animals were monitored for weight and seroconversion for 19 days after onset of cohousing. During that time no weight loss greater than 10% was observed and none of the animals were symptomatic or seroconverted. In conclusion, no thogotovirus transmissions were observed, rendering such events rather unlikely.

The low replication capacity of some isolates suggested the possibility that these viruses might not be well adapted to the murine host. Influenza viruses are frequently adapted to mice by consecutive passages of lung or liver homogenates to increase their pathogenicity for mice (32–35). Following this rationale, we chose the THOV-like isolates JOSV and Kami-25 for mouse adaptations because they showed a limited replication and a subclinical disease progression in vivo (Fig. 3A and B). For each isolate two independent series of 10 liver passages with consecutive intraperitoneal infections of 1000 PFU were performed in C57BL/6 mice. We chose an infection time of 4 days to avoid any effect of a possible up-coming adaptive immune response directed against viral antigens. Viral titers in the liver homogenates were determined and used for the next round of infection with 1000 PFU.

The viral titers of JOSV remained almost unchanged over the 10 passages (Fig. 9A). The titers of Kami-25 increased slightly about 10-fold over the 10 liver passages (Fig. 9B). Next, the viruses from the 10th mouse passages were amplified by a single passage in Vero cells and concentrated by ultracentrifugation to reach sufficient virus yield for deep sequencing and low variant frequency detection in the viral genomes. We mostly observed low frequency genome variants (<1%) apart from some variants that were present in a frequency above 10% (Fig. 9C and Fig. S4 and S5). Of these only a few resulted in amino acid substitutions (Fig. 9C). In comparison to the JOSV adaptations the Kami-25 adaptations showed more as well as higher frequency variants (Fig. 9C). The 10th passage of JOSV resulted in a relative uniform accumulation of variants per nucleotide (variants/nt) over all segments. However, segment 6 of both JOSV adaptations showed an increased variants/nt ratio (Fig. 9D). In contrast, the analysis of the 10th passage of Kami-25 indicated that segments 1 and 2 were highly preserved and variants were more frequently found in the other segments, arguing for a high structural conservation of PB2 and PB1 (Fig. 9E). For JOSV we only observed two low-frequency but nonsynonymous variations (∼1%) at position 40 and 49 of segment 3 that were present in both adaptations (Fig. S4). Interestingly, in segment 6 of Kami-25 we found two variants present in both adaptations at position 494 and 573 (Fig. S5). The variant at position 573 had a frequency of 15 and 3% but does not translate into an amino acid change. However, variant position 494 causes a proline to leucine exchange in codon 159 and was present at a high frequency of 20% and 46%, respectively. We also detected a synonymous mutation G855A in segment 4 in a frequency of 16% and 7%, respectively. This suggests a convergent selection for these variants during the adaptation to the mouse host. However, overall, the genomes of both viruses remained largely stable over the 10 mouse passages, implicating the absence of new selection pressure possibly due to a long history of coevolution of these viruses with rodents.

All work with thogotoviruses was performed under biosafety level (BSL) 2 conditions, except for the human isolate of BRBV that was handled under BSL3 conditions. The animals were handled in accordance with guidelines of the Federation for Laboratory Animal Science Associations and the national animal welfare body. Animal experiments were performed in compliance with the German animal protection law and approved by the local animal welfare committee (Regierungspraesidium Freiburg, permit 35–9185.81/G-15/127).

The 10 different thogotovirus isolates (Table 1) were kindly provided by our colleagues: THOV/SiAr/126/72 (isolate number 113.3) by Robert E. Shope, Emerging Infectious Diseases, University of Texas, Galveston, TX (60); the African THOV/Ken-IIA (isolate number 82.1) by Patricia A. Nuttall, Institute of Virology, University of Oxford, UK (5); THOV/PoTi503 (isolate number 106.1) by Armindo R. Filipe, Laboratory of Virology, NIH, Lisbon, Portugal (61); JOSV (TVP10564, isolate number 105.1) by Robert B. Tesh, Emerging Infectious Diseases, University of Texas, Galveston, TX (7); HI-Kamigamo-25 (strain KSU-25, isolate number 69.1) by Kentaro Yoshii, Laboratory of Public Health, Hokkaido University, Sapporo, Japan (36); DHOV/India/1313/61 (isolate number 77.1) by Fred J. Fuller, Department of Microbiology, North Carolina (10); BKNV (strain LEIV306K, isolate number 95.3p.1) by Robert E. Shope (6, 11); DHOV/PoTi461 (isolate number 100.1) by Armindo R. Filipe (62); BRBV-KS (strain NR-50132, ATCC VR-1842, Kansas-KS isolate number 87.1) by Amy J. Lambert and Brandy Russell, Centers for Disease Control and Prevention, Fort Collins, Colorado (24); and Oz virus (strain OzV, isolate number 264.1) by Kyoko Sawabe, National Institute of Infectious Diseases, Tokyo, Japan (8). All THOV-like viruses code for a functional orf encoding the ML IFN antagonist in their segment 6 (7, 37).

Most strains were initially isolated from tick lysates by intracranial infection of new borne mice or by inoculating Vero cells, followed by several passages on BHK-21 or Vero cells. Viral titers were determined by plaque assay on Vero cells.

African green monkey kidney Vero cells (ATCC CCL-81) and Syrian golden hamster kidney BHK-21 cells (ATCC CCL-10) were cultivated in Dulbecco's Modified Eagle Medium (DMEM, Gibco) supplemented with 10% fetal calf serum (FCS) at 37°C and 5% CO₂. Cell infections were performed using DMEM with 0.1% FCS. For viral replication, DMEM with 2% FCS and 20 mM HEPES was used.

Wild-type C57BL/6 mice were purchased from Janvier Labs (France). Congenic B6.A2G-Mx1 mice with a functional Mx1 allele (Mx1^+/+) (63) (for the generation of postinfectious sera) and IFNAR^−/−/IL28R^−/− mice (31) (index animals for transmission experiments) were bred in-house. Animal experiments were performed in compliance with the German animal protection law and approved by the local animal welfare committee (Regierungspraesidium Freiburg, permit G-15/127). All experiments were performed with age and sex matched 7 to 10 weeks old animals. Depending on the experiment the animals were infected intraperitoneally (i.p.), subcutaneously (s.c.) or intravenously (i.v.) with a total volume of 100 μl PBS. In case of the intranasal (i.n.) infection the animals were mildly sedated with 5% isofluran and then 40 μl of PBS diluted virus was applied directly onto the snout. Animals were daily monitored for weight and clinical signs and killed by cervical dislocation at the indicated time points. Organs, liver, lung, spleen and kidney were harvested at day 4 postinfection and homogenized (FastPrep Homogenizer, MP Biomedicals) in PBS in a final volume of 1 ml. After centrifugation at 5,000 × g for 10 min and 4°C the supernatants were analyzed by plaque assay on Vero cells.

Post-infectious sera were generated by challenging animals that survived the infection with 10,000 PFU/animal. The second infection did not result in any weight loss or disease symptoms. At 10 days after the second infection the blood was harvested by heart punctuation of anesthetized animals (Ketamine 100 mg/kg body weight and Rompun 5 mg/kg body weight) prior to cervical dislocation. The serum was prepared from the blood by incubation at 37°C for 10 min and centrifugation at 5,000g for 10 min. Due to the high lethality of the virus strains SiAr126, India/61, BKNV, and PoTi461 B6-Mx1^+/+ mice were used to elicit specific, postinfectious antisera as described previously (29).

For the determination of the lethal dose (LD) weight and disease symptoms were monitored daily. The mice were killed, if the weight loss reached 25% or the mice showed severe disease symptoms.

For transmission experiments C57BL/6 index mice (n = 9) were infected i.p. with 1.000 PFU/animal. At 6 h postinfection the animals were cohoused for 4 days in a fresh cage with naive IFNAR^−/−/IL28R^−/− contact mice (n = 12) that are highly susceptible to virus infection. After 4 days the C57BL/6 index animals were sacrificed. Liver, lung, spleen, and kidney were harvested to confirm the successful infection. Bladder, ileum, snout, and feces (3 pellets/animal) were obtained to analyze transmission relevant organs for viral propagation. The weight and disease symptoms of the contact animals were monitored for additional 16 days. Afterwards sera were obtained from the infected animals and checked for seroconversion in a neutralization assay as described (28) (data not shown).

For in vivo passaging C57BL/6 mice (n = 2) were i.p. infected with 1.000 PFU for 4 d. The livers were homogenized and the viral load determined. Diluted liver homogenates were used for the next passage by i.p. infection with 1.000 PFU for 4 d. 10 consecutive liver passages were performed for each virus in two independent series. The 10th liver passages were used to grow virus stocks on Vero cells for next-generation sequencing.

Vero cells were infected with different thogotovirus isolates at an MOI of 5 and fixed 16 h postinfection in 4% paraformaldehyde (PFA) and 0.1% glutaraldehyde in a 100 mM PHEM buffer for 30 min at room temperature (64). The method of in situ prefixation allowed us to preserve the original shape of cells and virus particles in the process of budding. Then, cells were scrapped, pelleted, and incubated overnight with 4% PFA in the 100 mM PHEM buffer at 4°C. After washing with a 100 mM sodium cacodylate buffer, cells were post-fixed for 60 min with 1% osmium tetroxide and 50 mM potassium ferricyanide in a 100 mM sodium cacodylate buffer (pH 7.0), washed and treated with 0.1% tannic acid in HEPES buffer for 30 min. After washing, the cell pellets were incubated in 2% uranyl acetate. Then, the cells were dehydrated in grade ethanol, embedded in a mixture of Epon and Araldite, and polymerized at 60°C for 24 h. Ultrathin sections (60 to 90 nm) of the cells were cut with a Leica EM UC6 microtome. The sections were contrasted with uranyl acetate and lead citrate and analyzed with a JEM1400 transmission electron microscope at 120 kV. The images were acquired using a TVIPS TemCam F416 camera.

Viral supernatants from Vero-infected cells were fixed with 4% PFA. Formvar-coated 400-mesh grids were pretreated with 1% alcian blue to optimize adhesion of viral particles. The viral suspensions were applied onto EM grids using airfuge sedimentation of 80 μl viral suspension. All preparations were negatively stained with 2% phosphotungstic acid and analyzed with either a Zeiss 109 or a JEM1400 electron microscope. For quantification, particles in several independent fields of view of the EM preparations were analyzed. About 70 to 100 virions were counted for each isolate to determine the percentage of spherical versus filamentous particles as well as their diameter and length. Particles were classified as filamentous when their long axis was at least two times their diameter.

Vero cells were infected with the indicated viruses (MOI of 1) for 24 h and lysed in T-PER tissue protein reagent (Thermo scientific). Proteins were denaturated in Laemmli buffer, separated by 12% SDS-polyacrylamid gel electrophoresis and transferred onto PVDF membrane (Merck Millipore). Viral proteins were detected using the postinfectious polyclonal mouse antisera. β-actin specific rabbit antiserum (Abcam) was used as an internal control. Primary antibodies were detected using fluorescent-labeled anti-mouse secondary antibodies (LI-COR). The fluorescent signals were detected using the ODYSSEYFc (Licor).

For the expression of recombinant viral proteins, pCAGGS-based expression plasmids (65) encoding the viral GP, NP, and M of SiAr126 and India/61 were transfected into 293T cells (2 μg GP, 0.2 μg NP, 2 μg M plasmid per six well). At 48 h posttransfection the cells were lysed and analyzed by Western blotting as described above.

To test the postinfectious mouse sera for virus neutralization, serial 2-fold dilutions (1:32 to 1:16,384) were prepared and incubated with a fixed amount of 100 PFU of the appropriate virus strain for 1 h at room temperature (28). As a control virus was incubated with PBS. The virus-serum mixture was transferred onto Vero cells and a plaque assay was performed. The PFU of the antibody-treated viruses were normalized to the PBS/virus control. For each experiment and sera a four parameter logistic regression with an upper (y = 100) and lower limit (y = 0) was calculated and based on the regression the neutralization titer NT₅₀ determined.

About 60 ml of virus stock (about 1 × 10⁷ PFU/ml) were subjected to ultracentrifugation through a 28% glycerin cushion in PBS for 2 h at 100,000 × g using a SW32 rotor (Beckman Coulter). The virus pellets were resuspended in 1 ml of PBS, virus titers were determined and about 2–5 × 10⁸ PFU were subjected to RNA extraction. Subsequently, the viral RNAs were pooled for the library preparation (TruSeq RNA Library Prep kit v2) and sequenced with an Illumina HiSeq2500 (50 cycles, single end).

Quality of raw sequencing data were assessed with FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/, accessed on 13.07.2021), no adapter or low-quality sequences were observed. Assembly of the different samples was done with SOAPdenovo-Trans (version 1.03, default parameter) (66), Trinity (version 2.8.4, default parameter) (67) and rnaSPAdes (version 3.13.1, default parameter) (68). Final assemblies were achieved by clustering the contigs of all individual assemblies using cd-hit-est (version 4.6.6, sequence identity threshold set to 0.95) (69) and BLAST (version 2.7.1+, E-value threshold set to 1e-10) (70). Visualization of the assemblies and their comparison was performed with QUAST (version 5.0.2, default parameter) (71).

Raw sequencing data of the host adaptation experiments were mapped to the final assemblies with HISAT2 (version 2.0.4, default parameter) (72). Mappings were processed with samtools (version 1.3) (73). For SNP calling, loFreq (version 2.1.3.1) (74) was used with default parameters. In-house Python3 scripts were used to determine changes on amino acid level. The scripts are available at https://github.com/klamkiew/thogotovirus_host_adaptation.

Circular plots, as shown in Fig. S1 and S2, were created using circos (10.1101/gr.092759.109). Preparation of the data for circos configuration files was done with an in-house Python script, deposited in the mentioned github repository.

Multiple sequence alignments of the deduced amino acid sequences of the different THOV and DHOV-like isolates were generated using MAFFT (75). Based on these alignments, maximum likelihood trees were constructed using PhyML 3.1 (76). The Smart Model Selection (SMS) was utilized to determine the best fitting substitution model (77). The analysis was performed under the LG substitution model with 1,000 bootstrap replicates and the tree visualized using Geneious 10.

Data were analyzed and statistically evaluated with GraphPad Prism 7. Viral titers of mouse organs were displayed on a log-scale (scatterplot, geometric mean) Statistics were computed by a one-way ANOVA with a Tukey’s multiple-comparison test on log-transformed values.

Full genome nucleotide sequences of the thogotoviruses were submitted to GenBank. For GenBank accession numbers see Table 1.

Raw reads of the virus stocks sequencing are available under https://osf.io/zd2mx/ (10.17605/OSF.IO/ZD2MX).