A systematic monthly collection of soil samples was undertaken over the course of approximately one year within a designated area of the riverbed (approximately 5 m × 10 m). Subsequent isolation experiments involving giant viruses and an amoeba (Acanthamoeba castellanii) consistently yielded positive results, confirming the monthly reproducibility of giant virus isolation from the same environmental context. However, in certain instances, viruses were not isolated concurrently from the samples obtained simultaneously, implying an uneven distribution of giant viruses within a local coastal region of approximately 5 × 10 m.

Soil and water specimens were systematically collected from May 2019 to March 2020 along the sandy expanse (35°42′39.2″N, 139°50′54.7″E) beneath the Hirai Bridge in Arakawa, Tokyo. Sampling was conducted during daylight hours, specifically avoiding rain, and at low tide. Each sampling event involved the random selection of three soil samples, which were then collected in 50 mL centrifuge tubes. River water was concomitantly procured in 50 mL centrifuge tubes using a 10 mL pipette, with sediment from the riverbed being perturbed during the collection process.

The procedure for seeding environmental samples into the amoebae commenced with the collection of approximately 1 g of soil in a 1.5 mL tube, followed by vigorous vortexing with 1 mL of sterile water. The supernatant was amalgamated with PYG medium and amoeba (Acanthamoeba castellanii), seeded into 96-well plates, and incubated at 26°C for subsequent observation. The PYG medium composition comprised 4 mM CaCl₂, 4 mM MgSO₄-7H₂O, 2.5 mM Na₂HPO₄-7H₂O, 2.5 mM KH₂PO₄, 0.1% [w/v] sodium citrate-2H₂O, 0.05 mM Fe(NH₄)₂(SO₄)₂-6H₂O, 100 mM glucose, pH 6.5. Wells displaying cytopathic effects (CPE) were chosen, and the culture medium was extracted and purified from the viral candidates by stepwise dilution in PYG medium containing amoeba. The virus was subsequently co-cultivated with the amoeba in a 25 cm² cell culture flask to facilitate growth.

Following growth, the culture medium was centrifuged at 500×g for 1 min and the supernatant was collected and subjected to further centrifugation at 8,000×g for 35 min to precipitate and gather the virus. The resultant virus solution was preserved at −80°C and subsequently utilized for virus identification through Polymerase Chain Reaction (PCR) analysis (Figure 1).

Giant virus typing was conducted by PCR using the Tks Gflex DNA Polymerase (Takara Bio, Inc., Tokyo, Japan), and capillary sequencing. For the detection of Marseilleviridae, primers targeting the MCP gene (For: 5′-STMYBDTKGAGAGTAATGACTTCTG-3′, Rev.: 5′-RRACTCGGAAGATRTGYTGG-3′) were utilized, as reported previously (Aoki et al., 2019). Mimivirus detection involved the use of custom-designed primers amplifying the MCP gene (For: 5′-ATGCTTGGTGAYGTWCCYGAACTWAC-3′, Rev.: 5′-RCTMATSATATTCTGAGAACATTGTAATTRAYMGC-3′). Additionally, primers were employed for the confirmation of the intein on polB of Mimivirus (For: 5´-GAGACGGATCATGTGGTTCCT-3′, Rev.: 5′-TGGCAGCCCTTTGACACTTC-3′). For the detection of Pandoravirus, primers amplifying a partial sequence of the polB gene (For: 5′-CCAAACCTCGCACCACTTTG-3′, Rev.: 5′-TCGAGCGTGTACGATTCGAG-3′), as previously reported (Akashi and Takemura, 2019), were used.

The PCR conditions for Marseillevirus and Mimivirus were as follows: 94°C for 1 min, {98°C for 10 s, 55°C for 15 s, 68°C for 1 min} repeated for 35 cycles, and a final extension at 68°C for 7 min, followed by an infinite hold at 8°C. Viral solutions were utilized as templates in these PCR procedures.

The PCR conditions for Pandoravirus were: 94°C for 1 min, {98°C for 10 s, 55°C for 15 s, 68°C for 30 s} repeated for 5 cycles, followed by {98°C for 10 s, 60°C for 15 s, 68°C for 30 s} repeated for 35 cycles, and a final extension at 68°C for 7 min, with an infinite hold at 8°C.

Template DNA was extracted using the standard NucleoSpin Tissue XS protocol (Macherey-Nagel GmbH and Co. KG, Germany). In instances of multibanded PCR products, gel purification was performed using NucleoSpin Gel and PCR Clean-up (Macherey-Nagel), followed by secondary PCR amplification. Only products with single bands were used for subsequent analyses.

PCR for Pithovirus and Cedratvirus employed the pandoravirus polB primers (For: 5´-CCAAACCTCGCACCACTTTG-3′, Rev.: 5´-CTCGGTCCACACCTCGATCG-3′), as previously described (Akashi and Takemura, 2019). PCR conditions for Pithovirus were as follows: 94°C for 1 min, {98°C for 10 s, 24.3°C for 15 s, 68°C for 15 s} repeated for 40 cycles, with an infinite hold at 8°C. For Cedratvirus, PCR conditions were 94°C for 1 min, {98°C for 10 s, 11.6°C for 15 s, 68°C for 15 s} repeated for 40 cycles, with an infinite hold at 8°C.

The primers used for PCR were also employed for capillary sequencing, and both the forward and reverse sequences were deciphered. Sequence analysis was conducted using FASMAC Co. The PCR cycling parameters and the FASTA file containing the deciphered sequences are available in Supplementary materials (Supplementary File S1).

Molecular phylogenetic analysis was performed using the following procedure. The sequences were aligned using MAFFT (v7.520) with specific options (—genafpair —maxiterate 1,000). Subsequently, Neighbor-Joining (NJ) trees were generated using rapidnj v2.3.3, with 10,000 bootstrap replicates, and the resulting trees were visualized using FigTree v1.4.4. Sequences derived from known viruses used for molecular phylogenetic analysis are documented in the Supplementary File S2. The classification of giant viruses adhered to the guidelines established by the International Committee on Taxonomy of Viruses (ICTV), as of January 2024, and is accessible at https://ictv.global/. In instances where specific classifications were not available at that time, commonly used names were employed.

Python 3.8 was used as the programming platform for visualizing both viral isolate counts and meteorological data. Cluster analysis was conducted using seaborn.clustermap, a function in the Python library. Cluster maps were generated using z-score normalized values for each virus and sample type (i.e., soil and water). Meteorological data pertinent to Tokyo, including temperature, vapor pressure, water level, precipitation, wind speed, cloud cover, and sunshine, were procured from the Japan Meteorological Agency.¹

A total of 55 strains of giant viruses were isolated from water samples, including river soil and surface sediments of the riverbed along the Arakawa River in Tokyo, through monthly sampling over approximately one year (Figure 2). These strains comprised 23 isolates of Marseilleviridae, 24 isolates of Mimiviridae, and six isolates of Pandoravirus (Figures 2A,B). Additionally, two previously unreported isolates, one each of Cedratvirus and Pithovirus, were obtained, marking their first isolation in Japan (Figures 2B,C).

PCR fragments of approximately 1.2 kb were successfully obtained from both the marseilleviruses and mimiviruses (Figure 2A). Molecular phylogenetic analysis utilizing the acquired sequence information revealed that among the 23 Marseilleviridae strains, three belonged to Lineage A whereas the remaining 20 were of Lineage B (Figure 3). Out of the 24 Mimiviridae strains, 13 were classified within the Mimivirus family, 10 were categorized as Megavirus strains, and one was designated as a Cotonvirus. A primer set targeting the intein-containing region polB of mimivirus yielded amplification in only three strains: Mi_13, Mi_15, and Mi_17 (Figure 4). Notably, both marseilleviruses and mimiviruses exhibited heterogeneity in the timing and source of isolation, with closely related strains spanning multiple months within both sub-phylogenetic groups (Figures 3, 4).

The multibanded pandoravirus strains were purified, and the cut PCR fragments were successfully reamplified using the same primers for subsequent sequencing analysis (Figure 2B, Lanes 3 and 4). Consequently, the polB gene region was decoded in five of the six isolates, whereas a portion of the BTB/POZ domain-containing protein was decoded in the remaining isolate. The breakdown of Pandoravirus isolates revealed three out of six isolates of Lineage A and three out of six isolates of Lineage B (Figure 5).

Remarkably, the genomic fragments of the two heterologous viral strains were successfully amplified using the polB primers of the pandoravirus (Figures 2B,C). One isolate, identified as Cedratvirus, was isolated from the soil in March 2020, showing amplification of a part of the ankyrin repeat-containing protein-coding gene. Phylogenetic analysis revealed its close relationship with the Brazilian Cedratvirus IHUMI (Figure 2B, Lane 7, Figure 6A). The other isolate, Pithovirus, exhibited amplification of a portion of the collagen triple helix repeat protein-coding gene and demonstrated a close relationship with Pithovirus sibericum P1084 based on phylogenetic relationships (Figures 2C, 6B). Both viruses were isolated from soil samples (Figure 6).

Comparing the sources of virus isolates over the course of the year, eight isolates of marseilleviruses originated from soil, whereas 15 were obtained from water and sediment. For mimiviruses, 15 isolates were sourced from soil and nine from water and sediment. Five Pandoravirus isolates were obtained from soil and one each from water and sediment (Figure 7A). Examination of the molecular phylogenetic data for pandoraviruses revealed that the isolation timing diverged from those of Marseilleviridae and Mimiviridae, with four of the six isolates obtained in May 2019. The other periods were July 2019 and January 2020, each with one isolate (Figures 5, 7A). Notably, only one of the six pandoravirus strains was isolated from water containing riverbed sediment, highlighting a bias in both timing and source compared with marseilleviruses and mimiviruses (Figure 7A). The top three-monthly isolates for marseilleviruses were eight in January 2020, seven in May 2019, and three in December 2019; for mimiviruses, five in May 2019; and four each in January 2020 and March 2020, showing a skewed distribution in the spring and winter months. Notably, no month throughout the year lacked a single giant virus isolation; however, in August, October, and November 2019, there was only one isolate each (August: Mimivirus from water and sediment samples; October: Cotonvirus from soil samples; November: Lineage B Marseillevirus from water and sediment samples) (Figures 4, 5, 7A).

Cluster mapping of the number of isolates of each virus based on the source and time of isolation categorized Pandoravirus, Pithovirus, and soil-derived Marseilleviridae and Mimiviridae as the May 2019-derived sample groups. Water- and sediment-derived Marseilleviridae and Mimiviridae were classified as the January 2020 group, whereas Cedratvirus (soil-derived), isolated in March 2020, formed an independent cluster. In summary, soil-derived viruses, excluding Cedratvirus, exhibited a bias toward May 2019, whereas the water sample-derived viruses Marseilleviridae and Mimiviridae demonstrated a bias toward the January 2020 isolation period (Figure 7B).

When the number of viral isolates from soil and water samples was quantified per gram of soil or per milliliter of water sample, the highest number of isolates from soil samples for a single sample occurred in July 2019, whereas the highest number of isolates from water samples was observed in January 2020. In May 2019, virus isolation was successful in all three soil samples collected each month. However, in the other months, some samples yielded no detectable viruses (Table 1).