New Virus Discovery May Aid in Disease Research

Naiavirus discovery, particle and cycle

We surveyed samples collected from the southern region of the Paraguay River in the state of Mato Grosso do Sul, in the Brazilian Pantanal biome. This region is widely recognized as one of the largest hotspots for biological diversity in the world. The prior isolation of the first Tupanvirus from a sample collected in this biome motivated our further exploration in the Pantanal to identify additional giant viruses7. To this end, water samples were collected at 44 locations along the river. In the laboratory, the samples were processed and then inoculated in Vermamoeba vermiformis.

After unsuccessful prospecting in 439 samples, we observed cytopathic effects in V. vermiformis (ATCC50237) inoculated with a water sample collected in the Porto Murtinho region. The amoebae exhibited rounding and lysis, and in the supernatant we observed regular elements, possibly giant virus particles or cells of an amoeba parasitic microorganism (Supplementary Fig. 1a, b). Once isolated, this agent was inoculated again simultaneously into V. vermiformis, A. castellanii (ATCC30010) and A. polyphaga (UFMG isolate), and after 48 h, we observed that all tested amoebae exhibited cytopathic effects. This lysate was subjected to crystal violet staining and then observed under an optical microscope. The cells of V. vermiformis were lysed, releasing a massive amount of a strongly stained entity that appeared to have a tail (Supplementary Fig. 1c). We then produced specific antibodies in mice against this entity and applied them in immunofluorescence (IF) assays. IF clearly confirmed the presence of a tail in the structure of the entity (Supplementary Fig. 1d).

Scanning electron microscopy (SEM) of the lysates revealed an unexpected and completely novel virion morphology (Fig. 1a–c). With a structure consisting of an enlarged region, here referred to as the head, and a narrowed region, here referred to as the tail, the virion was pleomorphic and asymmetric, reaching up to 1824 nm in length (mean 1350 nm) (Supplementary Fig. 2). Curiously, the entire external surface of the particles appeared to be made of the same membranous material, similar to the plasma membrane of amoebae. It is important to mention that what we refer to as a tail does not correspond to a viral tail structure sensu strictu, but rather to a tapered extension of the envelope structure originating from the region referred to as the head. After analyzing hundreds of particles, we identified two axes: one in the shape of a drop (head region with an average width of 540 nm) (Fig. 1a and Supplementary Fig. 2) and another in the shape of a triangle (head region with an average width of 570 nm) (Fig. 1c and Supplementary Fig. 2). At the upper end of the latter axis, at least one, and more frequently two, ostioles were present, asymmetrically arranged in an antipodal distribution. These ostioles were more clearly observed in the particle positioned along the drop-shaped axis and appeared as approximately circular protrusions on the particle surface, about 100–125 nm in size (Figs. 1d and 2a–c). In some particles, the ostioles resembled a flower of the Asteraceae family.

Transmission electron microscopy (TEM) revealed even more impressive details of this entity. In the head region, an oval structure, similar to the capsid of some giant viruses like Orpheovirus, occupied the upper region where the two ostioles were located (Fig. 1d and 2)8. A total of five structural layers were observed in the head region: 1) an external envelope covering the entire particle; 2) the outer layer of the oval capsid, thicker (averaging 25 nm) and electron-dense; 3) a layer below that, with internal undulations; 4) an internal membrane, as seen in nucleocytoviruses infecting amoebae; and 5) a broad electron-lucent region, where the genome is located (Fig. 2d and Supplementary Fig. 1d. In the ostiole region, there was a discontinuity in the external envelope and the layers composing the capsid wall (Fig. 2b, c). Interestingly, the internal content of the capsid is delimited in the ostiole region by the internal membrane and three additional layers of what appears to be membranous material, a structure never before described to our knowledge (Fig. 2b, c). Below the oval capsid, still within the head region, a roughly conical-shaped area (averaging 300 nm in length and 200 nm in width) was observed also covered by the external envelope (Fig. 1d). For a more detailed view of the outer membrane, new preparations were made for TEM, but using Spurr resin embedding with lower viscosity, which revealed more clearly the continuity of the envelope from the head region to the tip of the tail (Supplementary Fig. 3). In addition, longitudinal electron tomography analyses confirmed that the external envelope covers the entire viral particle from the apical region of the head to the tip of the tail (Fig. 2g, and Supplementary Movie 1). The oval-shaped capsid is clearly visible beneath the envelope, in the head region, with the aforementioned asymmetry (Fig. 2g). Particle analyses through tomography in the transverse plane revealed characteristics already described regarding the elements that make up the capsid and the tail region (Fig. 2h; and Supplementary Movies 1 and 2). At this point, we suspected that it was truly an unprecedented giant virus, with an average size of 1,350 nm, making it the largest tailed virus ever described in the virosphere and also the largest virus with an external envelope described to date. We named this entity Naiavirus, in honor of an indigenous Brazilian people, the Tupi-Guaranis. In their mythology, Naia was a beautiful indigenous woman who fell in love with and died from love for the Moon.

We then turned our attention to the tail region of Naiavirus. Unlike the head region, which showed less variation in size, the tail, with its membranous appearance, appeared quite flexible and variable in size in both SEM and TEM analyses (Fig. 3a–g and Supplementary Fig. 4). Notably, this structure appeared capable of folding, stretching, and shrinking in response to interactions with the adjacent substrate, a rare feature in the known virosphere (Fig. 3a–g). In TEM images, the tail appeared filled with an electron-lucent material and was delimited by a cylindrical membranous structure (Figs. 1d and 2c–f). Analyses of the replication cycle of Naiavirus in V. vermiformis and A. castellanii revealed that, most of the time, the first interaction between the amoebae and viral particles occurred via the tail. In several SEM images, the tail was observed attaching to the acanthapodia and pseudopodia of the amoebae, possibly stimulating their phagocytosis (Fig. 3a, b, d; Supplementary Fig. 5a and b). Once inside the phagosome, Naiavirus particles fused their internal membrane with the phagosomal membrane at the region of at least one of the ostioles, releasing the virion’s internal content into the amoeba’s cytoplasm (Supplementary Fig. 5c–g). TEM analyses revealed that the Naiavirus cycle occurs in the cytoplasm, with the cell’s nucleus remaining intact even at late stages (Fig. 3j and Supplementary Fig. 6g and h), raising questions about whether the nucleus might play a role in the late stages of the cycle, when morphogenesis takes place. An electron-lucent viral factory occupying about ¼ to ½ of the amoeba’s cytoplasm was observed (Supplementary Fig. 6g and h). The morphogenesis of Naiavirus particles occurs from crescent-like structures, a feature described for several giant viruses of amoebas (Fig. 3h; and Supplementary Fig. 6a). These crescents, measuring hundreds of nanometers, extend in various axes, encompassing the internal contents of the particle (Supplementary Fig. 6a–h). During their formation, at least two layers of membranous material are observed (Fig. 3h). Notably, the tail structure and the outer envelope appear to be incorporated into the particles within the viral factory (Fig. 1d; Fig. 3i, j; Supplementary Fig. 6d–h). At the viral factory’s periphery, the particles seem to undergo maturation stages, with the capsid appearing thicker and its outer layer more defined (Supplementary Fig. 6e–h). No evidence of budding or exocytosis was observed. SEM and TEM analyses strongly suggest that all morphogenesis occurs within the cell’s cytoplasm, and the particle release mechanism is via cell lysis induction (Fig. 3k, l; and Supplementary Fig. 7). It is important to mention that the complete replication cycle lasts ~24 h. In amoeba cultures infected synchronically, the viral factory becomes visible from 12 h; mature particles appear from 14 h; and cell lysis occurs 24 h onward.

As mentioned, Naiavirus has an external envelope that covers the entire particle. In the virosphere, envelopes are typically lipidic in nature, making them susceptible to detergents. Treatment of Naiavirus particles with Tween 20 at concentrations of 0.1% and 1% did not cause a significant reduction in viral titers. While the untreated particles induced infection with average titers of 10⁷.⁵, particles treated with 0.1% and 1% Tween induced infections with average titers of 10⁷.⁴ and 10⁷.², respectively. However, when treated with a 10% concentration, complete inactivation of the particles was observed. SEM analysis revealed damage to the viral envelope (Supplementary Fig. 8). These results suggest that the envelope of Naiavirus may have a composition different from that of regular lipids, possibly consisting of a lipid resistant to the detergent tested at lower concentrations. Although the nature of the envelope still requires further characterization, analyses using TEM, SEM, IF, and tomography indicate that the particle is completely covered by this mantle. Therefore, Naiavirus represents the largest enveloped virus in the known virosphere.

Naiavirus genome and evolutionary history

We then sequenced the Naiavirus genome using the Illumina MiSeq platform to provide insight into its evolution and genetic repertoire (see Methods). The DNA of purified virus was sequenced, and the assembly revealed a circular genome, 922 kbp in size with 33.6% G + C content and an average coverage of 791x. A total of 867 genes were predicted, ranging from 90 to 4323 bp. We performed homology searches using BLASTp against the NCBI nr database, revealing that a quarter of Naiavirus proteins had no detectable matches (i.e., ORFan genes). Furthermore, nearly 50% of the identified genes correspond to proteins with unidentified functions that were previously detected in other organisms, especially in nucleocytovirus metagenomic sequences. To improve the annotation of Naiavirus proteins, we additionally employed InterProScan, HHpred, and EggNOG-mapper searches9,10,11,12. Nevertheless, 127 proteins remained classified as ORFans.

Analysis of these annotation results revealed that the Naiavirus genome encodes a balanced distribution of predicted functional categories. Among other functional categories widely present in the Naiavirus genome, genes associated with protein translation are notable, including 5 aminoacyl tRNA synthetases (Glycine (ID_189), Methionine (ID_222), Tryptophan (ID_457), Tyrosine (ID_657), Aspartate/Asparagine (ID_694), a peptidyl-hydrolase (ID_318) and 14 translation factors, including initiation, elongation, and termination. Additionally, 6 tRNAs were predicted in Naiavirus genome [tRNA-Ile (tat), tRNA-Ile (aat), tRNA-Leu (taa), tRNA-Leu (caa), tRNA-Cys (aca), and tRNA-Cys (gca)]. There is no clear correlation between the presence of the viral tRNA genes and the codon usage observed for Naiavirus. Nevertheless, Naiavirus encodes tRNAs associated with the most commonly used codons for leucine and isoleucine, which may be advantageous for the viral translation process during infection (Supl. Table. 1). Genes involved in carbohydrate metabolism are also widely represented, particularly ten glycosyltransferase homologs, belonging to the families 2, 25, 45, 85 and others. Among the genes related to lipids, notable mentions include 7-dehydrocholesterol reductase (ID_7), three class-3 lipases (IDs_38, 176, 660), glycerol-3-phosphate cytidylyltransferase (ID_392), fatty acid hydroxylase (ID_407), cyclopropane-fatty-acyl-phospholipid synthase (ID_444) and two patatin phospholipases (IDs_783, 533). Host-virus and signal transduction genes were also found, including several kinases, phosphatases, and an endonuclease. A total of 47 genes related to transcription and RNA processing were found, including all those necessary for transcription, such as the canonical multimeric RNA polymerase subunits (IDs_5, 393, 396, 799), RNase III (IDs_61 and 469), RNA ligase (IDs_68, 353) Cap-specific mRNA (nucleoside-2-O-)-methyltransferase (IDs_102 and 103) and others. An impressive number of 52 genes related to DNA replication, recombination and repair were found, including DNA mismatch repair ATPase (ID_57), AP (apurinic) endonuclease family 2 (ID_84), NAD-dependent DNA ligase (ID_92), D5 helicase-primase (IDs_120), VV D6-like helicase (ID_159), DNA primase (ID_326), DNA topoisomerase IIA (ID_398), DNA repair exonuclease (ID_430) and DNA polymerase family B (ID_582).

In addition, a long list of proteins with unexpected functions was identified in Naiavirus. A phosphoglycerate mutase (ID_648), an enzyme involved in glycolysis and gluconeogenesis, was predicted and is essential for metabolizing glucose and 2,3-phosphoglycerate. Curiously, a staygreen protein (ID_776), typically found in plants and involved in regulating chlorophyll senescence, was also identified. Two paralogs of thaumatin family protein (IDs_365, 693) were predicted, which have also been previously identified in plants and are involved in defense functions. Bacterial-related proteins were also identified: an exosporium glycoprotein BclB-related protein (IDs_29 and 792), typically associated with the formation of the exosporium layer of bacterial spores; and two paralogs of hemolysin III (IDs_209 and 852), a virulence factor found in some bacteria that causes hemolysis. Additionally, an aquaporin (ID_689), a mitochondrial carrier protein (ID_365), and a tumor suppressor protein, the cyclin-dependent kinase inhibitor 3 (ID_181), were predicted. Finally, HHpred predicted that ID_802 bears remote homology to a 39S ribosomal protein L38 as well as a Phosphatidylethanolamine-binding protein.

Considering that the particles of Naiavirus do not resemble any virus previously described, a central question that has intrigued us since the discovery of this entity was its phylogenetic position in the virosphere. Thus, with the genome annotated, we searched for genes commonly shared among giant viruses. Orthologs of hallmark and core genes from nucleocytoviruses were consistently found in the Naiavirus genome, including the previously mentioned DNA polymerase family B (ID_582), DNA topoisomerase IIA (ID_398), VLTF3 late transcription factor (ID_540), D5 helicase-primase (ID_120), and a major capsid protein (ID_581) with a double-jelly roll domain. A phylogenetic tree of DNA polymerase family B was then constructed, as this gene has been used in the literature as a phylogenetic marker for nucleocytoviruses (Supplementary Fig. 9). The topology of the tree suggests that Naiavirus forms a deeply-branching lineage within the known representatives of the Pimascovirales order. Importantly, none of the pimascoviruses discovered through viral isolation, such as Pithovirus, Cedratvirus, and Orpheovirus, cluster near Naiavirus. Interestingly, one metagenome-derived genome that exhibits phylogenetic affinity for Naiavirus is the recently described Hydrivirus, which was assembled from a Siberian permafrost sample dated to 42,000 years ago13. Since it was detected through metagenomics, the morphological characteristics of the Hydrivirus particle, as well as data related to its host and life cycle, have remained mysterious.

To examine the prevalence of other viruses with phylogenetic affinity for Naiavirus, we surveyed > 15,000 publicly available metagenomes and identified several PolB sequences that fall within the new Naiavirus clade. We then performed metagenome binning to recover 11 novel metagenome-assembled genomes (MAGs) from this lineage. The other metagenomic sequences that cluster near Naiavirus were detected in various habitats, such as permafrost/cryo soil, acid mine sediment, bioreactor wastewater, water samples from lakes, etc., and several locations, including Russia, China, the USA, Canada, the Czech Republic, Sweden, Austria, and Greenland, with detections dating back to 2011 (Fig. 4c). This indicates that Naia-like viruses are present in different regions of the globe but, curiously, have never been isolated before. To assess the consistency of the results obtained from the previous phylogenetic analysis, a new tree was constructed, this time utilizing seven concatenated genes from both isolated viruses and MAGs (Fig. 4a). As previously observed, the concatenated tree revealed that Naiavirus falls within the order Pimascovirales, forming a distinct clade from the already known viruses in this group. Both Naiavirus, as well as Orpheovirus and Hydrivirus, are distributed among MAGs that also harbor several core genes of nucleocytoviruses. This finding indicates that this large branch of pimascoviruses remains poorly characterized, and that entirely new viruses belonging to this group are likely to be isolated in the coming years.

We also investigated the evolutionary history of Naiavirus through a global analysis of the sharing of predicted proteins among the pimascoviruses. In this analysis, not only Naiavirus but also all groups of pimascoviruses that have been isolated (cedrat-, pitho-, Orpheovirus) and metagenomic assembled genomes (MAGs) – including Hydrivirus – were considered. The result of this analysis is a network showing the protein sharing among 29 viruses (isolated or MAGs) (Fig. 4b). The largest cluster observed for Naiavirus consists of unique predicted proteins not shared with any other virus. However, several small clusters were observed between Naiavirus and various MAGs, with one large exclusive cluster between Naiavirus and MAG-Hydrivirus. This result suggests that Naiavirus has a distinct evolutionary history, with unique predicted proteins that are not shared with other known viruses, which may indicate a significant level of genomic differentiation. The observation of small clusters between Naiavirus and various MAGs, and a large exclusive cluster between Naiavirus and MAG-Hydrivirus, suggests a possible evolutionary relationship or exchange of genetic material between these viruses and other MAGs, which may indicate processes of coevolution or interaction in specific environments. Interestingly, the analysis of the contribution of paralogs in the genomes of pimascoviruses revealed that Naiavirus and Hydrivirus have the lowest percentage of gene duplications within this group (Supplementary Table 2).

Naiavirus particles’ proteome

Considering that the genome and virion organization of Naiavirus are unprecedented in the virosphere, we sought to examine the viral proteins that make up the particle. To address this, viral particles were produced by infecting 20 T175 flasks containing 107 Acanthamoeba castellanii and then purified twice, sequentially in a sucrose gradient. The ultrapure particles were subjected to proteomics and 254 viral proteins associated with the particle (Supplementary Table 3) were identified, a number that surpasses those reported for all giant viruses, except pandoraviruses. Overall, 17 proteins involved in transcription and RNA processing were detected, containing a variety of elements that could explain the occurrence of transcription of viral genes entirely in the cytoplasm, consistent with electron microscopy findings, where no phase of the cycle was observed in the cell’s nucleus. Additionally, 15 proteins involved in DNA replication, recombination, and repair were found in Naiavirus particles, suggesting that replication of its genome likely occurs in the early stages of the cycle. A homolog to a H2B histone was also found in the genome and proteomic results (ID_277), suggesting that this protein is potentially used for packaging the genome in the capsid. Finally, proteins involved in host interaction (e.g., kinases), lipid metabolism, protein metabolism, and others were also identified. Still, a total of 174 proteins were identified as ORFans or hypothetical proteins. To complement the analysis, we performed Foldseek searches on AlphaFold-predicted structures in the Naiavirus proteome (Supplementary data 1). As a result, we were able to successfully annotate only 229 proteins considering all of the approaches. A total of 19 proteins remained classified as hypothetical proteins, including the most abundant protein (ID_538), and six proteins were entirely novel, with no matches in any database (Supplementary Table 3), highlighting the vast untapped reservoir of novel proteins in giant viruses. Foldseek analysis also revealed that the viral particle packaging proteins potentially involved in host takeover, such as remote homologs to a interleukin-1 receptor-associated kinase 4 (ID_542), and a Ubx-domain-containing protein (ID_192). Notably, we identified some proteins with remote homology to phage structural proteins, including the baseplate tripod (ID_214), tail fibers (ID_206 and ID_561), and tailspike protein (ID_48). Interestingly, although the major capsid protein (MCP) is typically the most abundant protein in the capsids of various nucleocytoviruses, in Naiavirus particles, it ranks only as the 169th most abundant protein (Supplementary Table 3 and 4). Considering the completely unprecedented structure of Naiavirus particles, it is possible that the MCP’s primary structural function has been modified throughout the virus’s evolution, perhaps acting as a scaffolding protein. It is worth mentioning that such a phenomenon appears to have occurred in other pimascoviruses as well.

For viral isolation, 96-well plates containing 4 × 104 cells per well of Vermamoeba vermiformis (ATCC 50237) in Peptone Yeast Extract Glucose (PYG) medium supplemented with ciprofloxacin (4 μg/mL), vancomycin (4 μg/mL), doxycycline (20 μg/mL), penicillin (500 U/mL), streptomycin (100 mg/mL), and amphotericin B (25 mg/mL) were inoculated with 100 μL of pure or 1:10 diluted water samples collected from 44 different points along the Paraguay River in the state of Mato Grosso do Sul, Brazil. The collections were carried out within the context of the Navio project (Expanded Navigation for Intensive and Optimized Surveillance), with support from the Brazilian Navy. For the negative control, 100 μL of PBS was used instead. The plates were incubated at 29 °C and monitored daily under an optical microscope for cytopathic effects (CPE).

Upon CPE observation, the sample was collected, and viral stocks were produced using twenty T175 cm² cell culture flasks containing 1 × 107 V. vermiformis cells cultivated in PYG medium supplemented with penicillin (500 mg/mL), streptomycin (100 mg/mL), and amphotericin B (25 mg/mL). The cells were infected with Naiavirus at a multiplicity of infection (MOI) of 0.01 and incubated at 29 °C until typical CPE was observed.

For viral purification, the flask contents were collected and subjected to freezing and thawing to release viral particles that might still be trapped inside intact cells. The lysate was then centrifuged at 1200 × g for 10 min to remove cell debris. The supernatant was collected, layered over a 24% sucrose gradient, and centrifuged twice at 36,000 × g for 1 h. The resulting pellet was resuspended in phosphate-buffered saline (PBS), and the purified viruses were stored at −20 °C. The viral titer was determined using the endpoint method, calculated according to Reed-Muench, and expressed as the number of 50% tissue culture infective doses (TCID₅₀) per milliliter14. Host spectrum experiments were conducted by infecting the following amoebas with the Naiavirus seed pool: Acanthamoeba castellanii (ATCC30010), Acanthamoeba polyphaga (UFMG isolate), Vermamoebaa vermiformis (ATCC50237), and Naegleria gruberi (USP isolate). Considering that the virus did not replicate only in Naegleria, the remaining experiments were primarily conducted in Acanthamoeba castellanii and Vermamoeba vermiformis, as described in each section.

For the analysis of Naiavirus susceptibility to Tween 20, the purified viral particles were treated with this detergent at concentrations of 0.1%, 1%, and 10%. After 30 min of incubation, the particles were washed twice with PBS and then subjected to titration in Acanthamoeba castellanii. The detergent toxicity to the ameboid cells was previously calculated at different concentrations and dilutions, and these factors were considered in the interpretation of the results.