An update on eukaryotic viruses revived from ancient permafrost

One quarter of the Northern hemisphere is underlain by permanently frozen ground, referred to as permafrost. Due to climate warming, irreversibly thawing permafrost is releasing organic matter frozen for up to a million years, most of which decomposes into carbon dioxide and methane, further enhancing the greenhouse effect. Part of this organic matter also consists of revived cellular microbes (prokaryotes, unicellular eukaryotes) as well as viruses that remained dormant since prehistorical times. While the literature abounds on descriptions of the rich and diverse prokaryotic microbiomes found in permafrost, no additional report about “live” viruses have been published since the two original studies describing pithovirus (in 2014) and mollivirus (in 2015). This wrongly suggests that such occurrences are rare and that “zombie viruses” are not a public health threat. To restore an appreciation closer to reality, we report the preliminary characterizations of 13 new viruses isolated from 7 different ancient Siberian permafrost samples, 1 from the Lena river and 1 from Kamchatka cryosol. As expected from the host specificity imposed by our protocol, these viruses belong to 5 different clades infecting Acanthamoeba spp. but not previously revived from permafrost: Pandoravirus, Cedratvirus, Megavirus, and Pacmanvirus, in addition to a new Pithovirus strain.


Introduction
Ongoing international modeling and monitoring studies keep confirming that the 40 continuous release of greenhouse gas (mostly CO2) due to human activities since the in-41 dustrial revolution is causing significant climate change through global warming. It is 42 now widely acknowledged that an average temperature increase of 1.5°C relative to 1850- 43 1900 would be exceeded during the 21st century, under all realistic circumstances [1] even 44 though the adequacy of present climate models to predict regional changes remains in 45 debate [2]. For instance, climate warming is particularly noticeable in the Arctic where average temperatures appear to increase more than twice as fast as in temperate regions [3]. One of the most visible 47 consequences is the global thawing of permafrost at increasing depths [4,5], the rapid erosion of permafrost bluffs [6,7], 48 as well as erosion of deep and old permafrost by thaw slumping in hillslopes [8,9]. This rapid permafrost thaw causes 49 mobilization of ancient organic matter previously preserved for millennia in permafrost deep layers, a phenomenon 50 most visible in Siberia where deep continuous permafrost underlays most of the North Eastern territories. 51 The thawing of permafrost has significant microbiological consequences. First, above freezing temperatures, the return 52 of liquid water triggers the metabolic reactivation of numerous soil microorganisms (bacteria, archaea, protists, 53 fungi) [10][11][12][13][14], exposing the organic material previously trapped in permafrost to decomposition, releasing additional 54 CO2 and methane further contributing greenhouse gas to the atmosphere [5,15,16]. Yet, a more immediate public health 55 concern is the physical release and reactivation of bacteria (or archaea) that have remained in cryptobiosis trapped in 56 deep permafrost, isolated from the Earth's surface for up to 2 million years [10, 17] (although a more consensual limit 57 would be half a million years [18]). On a shorter time scale, the periodical return of anthrax epidemics devastating 58 reindeer populations has been linked to the deeper thawing of the permafrost active layer at the soil surface during 59 exceptionally hot summers, allowing century-old Bacillus anthracis spores from old animals burial grounds or carcasses 60 to resurface [19-21] 61 One could imagine that very deep permafrost layers (i.e. million-year-old), such as those extracted by open-pit 62 mining, could release totally unknown pathogens [22]. Finally, the abrupt thawing vertically operating along the whole 63 wall of permafrost bluffs (consisting of specific ice-rich deposits called "yedoma") such as seen in the Kolyma lowland 64 or around the Yukon river Alaska, causes the simultaneous release of ancient microorganisms from frozen soils dating 65 from the whole Holocene to the late Pleistocene (i.e. up to 120.000 years ago) [23]. Many culture-based and culture- 66 independent studies (i.e. barcoding and/or metagenomics) have documented the presence of a large diversity of bacteria 67 in ancient permafrost [10-12, 17, [24][25][26][27][28], a significant proportion of which are thought to be alive, although estimates 68 vary greatly with the depth (age) and soil properties [17,29,30]. These bacterial populations include relatives of com-69 mon contemporary pathogens (Acinetobacter, Bacillus anthracis, Brucella, Campylobacter, Clostridia, Mycoplasma, various 70 Enterobacteria, Mycobacteria, Streptococci, Staphylococci, Rickettsia) [11,12,24,29,31]. Fortunately, we can reasonably hope 71 that an epidemic caused by a revived prehistoric pathogenic bacterium could be quickly controlled by the modern 72 antibiotics at our disposal, as they target cellular structures (e.g. ribosomes) and metabolic pathways (transcription, 73 translation or cell wall synthesis) conserved during the evolution of all bacterial phyla [32], even though bacteria carry-74 ing antibiotic-resistance genes appear to be surprisingly prevalent in permafrost [26,31,33]. 75 The situation would be much more disastrous in the case of plant, animal, or human diseases caused by the revival 76 of an ancient unknown virus. As unfortunately well documented by recent (and ongoing) pandemics [34,35], each new 77 virus, even related to known families, almost always requires the development of highly specific medical responses, 78 such as new antivirals or vaccines. There is no equivalent to "broad spectrum antibiotics" against viruses, because of 79 the lack of universally conserved druggable processes across the different viral families [36,37]. It is therefore legitimate 80 to ponder the risk of ancient viral particles remaining infectious and getting back into circulation by the thawing of 81 ancient permafrost layers. Focusing on eukaryote-infecting viruses should also be a priority, as bacteriophages are no 82 direct threat to plants, animals, or humans, even though they might shape the microbial ecology of thawing permafrost 83 [38]. 84 Our review of the literature shows that very few studies have been published on this subject. To our knowledge, 85 the first one was the isolation of Influenza RNA from one frozen biopsy of the lung of a victim buried in permafrost 86 since 1918 [39] from which the complete coding sequence of the hemagglutinin gene was obtained. Another one was 87 the detection of smallpox virus DNA in a 300-year-old Siberian mummy buried in permafrost [40]. Probably for 88 safety/regulatory reasons, there was not follow up studies attempting to "revive" these viruses (fortunately). The first 89 isolation of two fully infectious eukaryotic viruses from 30,000-y old permafrost was thus performed in our laboratory 90 and published in 2014 and 2015 [41,42]. A decisive advantage of our approach was to choose Acanthamoeba spp. as a 91 host, to act as a specific bait to potentially infectious viruses, thus eliminating any risk for crops, animals or humans. 92 However, no other isolation of a permafrost virus has been published since, which might suggest that these were lucky 93 shots and that the abundance of viruses remaining infectious in permafrost is very low. This in fact is wrong, as numer-94 ous other Acanthamoeba-infecting viruses have been isolated in our laboratory, but not yet published pending their com-95 plete genome assembly, annotation, or detailed analysis. In the present article we provide an update on thirteen of them, 96 most of which remain at a preliminary stage of characterization. These isolates will be available for collaborative studies 97 upon formal request through a material transfer agreement. The ease with which these new viruses were isolated sug-98 gests that infectious particles of viruses specific to many other untested eukaryotic hosts (protozoans or animals) remain 99 probably abundant in ancient permafrost.

101
Permafrost sampling 102 The various on-site sampling protocols have been previously described in [31,43] for samples #3 and #5 (collected in 103 the spring 2015), in [13,44] for sample #4, in [45] for sample #6, and [46, 47] for samples #7-9 (see Table 1). 104 Liquid sample #2 and #4 were collected in pre-sterilized 50 ml Falcon tube in august 2019, as well as sample #1 consisting 105 of surface soil without vegetation from the Shapina river bank collected on 07/15/2017 and since maintained frozen at -106 20°C in the laboratory. 107 108 Sample preparation for culturing 109 About 1g of sample is resuspended in 40 mM Tris pH 7.5, from 2-10% V/V depending on its nature (liquid, mud, solid 110 soil) and vortexed at room temperature. After decanting for 10 minutes, the supernatant is taken up, then centrifugated 111 at 10,000 g for one hour. The pellet is then resuspended in 40 mM Tris pH 7.5 with a cocktail of antibiotics (Ampicillin 112 100µg/mL, Chloramphenicol 34µg/mL, Kanamycin 20µg/mL). This preparation is then deposited one drop at a time 113 onto two 15 cm-diameter Petri dishes (Sarsted 82.1184.500) one previously seeded with Acanthamoeba castellanii (Doug-114 las) Neff (ATCC 30010TM) at 500 cells/cm 2 , the other with A. castellanii cells previously adapted to Fungizone (Ampho-115 tericin B, Gibco, Pasley, UK) by serial passages in presence of increasing concentration of the drug up to 2.5 µg/ml. 116 Fungizone is used to inhibit the proliferation of viable microfungi known to be present in permafrost. Changes in the usual appearance of A. castellanii cells (rounding up, non-adherent cells, encystment, change in vacuoli-120 zation and/or refractivity) might eventually become visible after 72 h, but might be due to a variety of irrelevant causes 121 such as overconfluency, the presence of a toxin, or the proliferation of bacteria or microfungi. Under a light microscope, 122 the areas exhibiting the most visible changes are spotted using a p1000 pipetman. This 1 mL volume is then centrifu-123 gated (13,000 g for 30 minutes), the pellet resuspended in 100 µL and scrutinized under a light microscope. This sub-124 sample is also used to seed further T25 cell culture flasks of fresh A. castellanii cells.  (Table 2), a PCR test is performed using the Terra PCR Direct Polymerase Mix (Takara Bio Europe SAS, Saint-130 Germain-en-Laye, France). Amplicons are then sequenced (Eurofins Genomics, Ebersberg, Germany) to confirm the 131 presence of new isolates of a given acanthamoeba-infecting virus family (Table 2) suggested by their particle morphol-132 ogy and ultrastructural features. Nomenclature of new isolates 137 We used the binomial format for the naming of virus species, where the genus name and a species epithet together form 138 a unique species name. The genus name (e.g. "Pithovirus") was attributed on the basis of concordant similarities with 139 previously characterized amoeba-infecting viruses: genome sequences (PCR amplification using specific probes, partial 140 or complete genome sequences), cell-infection patterns, and virion morphological features. The species epithet was cho-141 sen to reflect the location or nature of the source sample (e.g. "duvanny"). A strain name (e.g. "Tums1") was added to 142 further specify the precise sample (there might be several from the same location/source) from which the isolation was 143 performed. Strain names can thus be shared by different species. Virus cloning, virus particles purification using a cesium chloride gradient, and DNA extraction from approximately 5 158 x 10⁹ purified particles (using the Purelink Genomic extraction mini kit, ThermoFisher) have been previously described 159 [49]. Sequence data was generated from the Illumina HiSeq X platform provided by Novogene Europe (Cambridge, 160 UK). Genome data assembly was performed in-house as previously described [49]. The draft genome sequences listed 161 in Table 3-6 are provided as supplementary material (S1-S8).
162 163 Design of virus-specific PCR primers 164 Clusters of protein-coding genes common to all known members of a viral family or clade were identified using Or-165 thofinder [50]. The protein sequence alignments of these clusters were converted into nucleotide alignments using 166 Pal2nal [51]. Statistics on the multiple alignments where then computed using Alistat [52] and sorted using the "most 167 unrelated pair criteria". The corresponding alignments were thus visually inspected to select the variable regions 168 flanked by strictly conserved sequences suitable as PCR primers. The primers and their genes of origin are listed in 169 Table 2.    dense tegument with a lamellar structure parallel to the particle surface and interrupted by an ostiole-like apex 220 (Figure 1.A). In complement of these morphological features unique to the Pandoraviridae [53], PCR tests were 221 performed to confirm the identification of the new isolates using family-specific sets of primers (Table 2) and the 222 amplicons sequenced to evaluate their genetic divergence with other members of the family (Table 3). All new isolates 223 were found to be significantly distinct from each others and from contemporary strains, albeit within the range of 224 divergence (93%-86% nucleotide identity) previously observed (Table 3)  recalling that the prototype of this family was previously isolated from an ancient permafrost layer of more than 286 30,000-y BP [58]. Other members of this family are the most abundant in a recent metagenomic study of various 287 Siberian permafrost samples focusing on eukaryotic viruses [59]. 288 We In complement of these visual clues, PCR tests were performed to confirm the identification of the new isolates using 295 two different clade-specific sets of primers (Table 2) and the amplicons sequenced to evaluate their genetic divergence 296 with known members of the family (Table 4). All new isolates were found to be significantly distinct from each others 297 and from contemporary strains, but within a range of divergence (94%-87%) consistent with that of previously 298 characterized members of these clades (Table 4). In addition, we sequenced the genomes of the 3 new isolates. These   In complement of the above unambiguous observations, a PCR tests was performed to confirm the identification of 324 the new isolate using a Megavirinae specific sets of primers (Table 2) and the amplicon sequenced to evaluate its 325 genetic divergence with known members of the family. M. mammoth was found to be a very close relative of the 326 modern prototype M. chilensis (Table 5). Such very low levels of divergence are actually customary within the 327 Megavirus genus (also referred to as the C-clade Megavirinae) [64].

Pacmanvirus lupus 344
Pacmanvirus is a clade of recently discovered Acanthamoeba-infecting viruses distantly related to the African swine fe-345 ver virus, until then the only known members of the Asfarviridae family that infects pigs [65]. We now report the isola-346 tion of a third member of this newly defined group from the frozen intestinal remains of a Siberian wolf (Canis lupus) 347 preserved in a permafrost layer dated >27,000-y BP. At variance with the other truly giant viruses (i.e. exhibiting unu-348 sually large particles), their icosahedral virions are about 220 nm in diameter (Figure 1.F Table 7). The tree (rooted at midpoint) was built using IQ-TREE (version 1.6.2) [54] (best fit model: " LG+F+I+G4" ). 437 The two closest Mimiviridae RPB1 sequences are used as an outgroup.

444
Following initial reports published more than 5 years ago [41,42], this study confirms the capacity of large DNA 445 viruses infecting Acanthamoeba to remain infectious after more than 48,500 years spent in deep permafrost. Moreover, 446 our results extend our previous findings to 3 additional virus families or groups: 4 new members of the Pandoraviridae, 447 one member of the Mimiviridae, and one pacmanvirus (Table 1). One additional pithovirus was also revived from a 448 particularly productive sample dated 27,000-y BP (sample#7, Table 1) exhibiting mammoth wool. Given these viruses' 449 diversity both in their particle structure and replication mode, one can reasonably infer that many other eukaryotic 450 viruses infecting a variety of hosts much beyond Acanthamoeba spp. may also remain infectious in similar conditions. 451 Genomic traces of such viruses were detected in a recent large-scale metagenomic study of ancient permafrost [59] as 452 well as in Arctic lake sediments [66]. They include well documented human and vertebrate pathogens such as pox-453 viruses, herpesviruses, and asfarviruses, although in lower proportions than protozoan infecting viruses. 454 In our recent metagenomic study [59], pandoraviruses are notably absent while they constitute the large majority 455 of the viruses revived from permafrost and cryosols. Such a discrepancy might originate from the fact that the extraction 456 of genomic DNA from their sturdy particles requires a much harsher treatment than for most other viruses. Their abun-457 dance in environmental viromes might thus be much larger than the small fraction they contribute to the DNA pool. 458 Such DNA extraction bias may apply to many other microbes, and is a serious limitation to the validity of metagenomic 459 approaches for quantitative population studies. 460 The types of viruses revived in our study are indeed the results of even stronger biases. First, the only viruses we 461 can expect to detect are those infecting species of Acanthamoeba. Second, because we rely on "sick" amoeba to point out 462 potentially virus-replicating cultures, we strongly limit ourselves to the detection of lytic viruses. Third, we favor the 463 identification of "giant" viruses, given the important role given to light microscopy in the early detection of positive 464 viral cultures. It is thus likely that many small, non-lytic viruses do escape our scrutiny, as well as those infecting many 465 other protozoa that can survive in ancient permafrost [10]. 466 However, we believe that the use of Acanthamoeba cells as a virus bait is nevertheless a good choice for several 467 reasons. First, Acanthamoeba spp. are free-living amoebae that are ubiquitous in natural environments, such as soils and 468 fresh, brackish, and marine waters, but are also in dust particles, pools, water taps, sink drains, flowerpots, aquariums, 469 sewage, as well as medical settings hydrotherapy baths, dental irrigation equipment, humidifiers, cooling systems, ven-470 tilators, and intensive care units [67]. The detection of their virus may thus provide a useful test for the presence of any 471 other live viruses in a given setting. Second, if many Acanthamoeba species can be conveniently propagated in axenic 472 culture conditions, they remain "self-cleaning" thanks to phagocytosis, and are capable of tolerating heavy contamina-473 tion by bacteria (that they eat) as well as high doses of antibiotics and antifungals. The third, but not the least, advantage 474 is that of biological security. When we use Acanthamoeba spp. cultures to investigate the presence of infectious unknown 475 viruses in prehistorical permafrost (in particular from paleontological sites, such as RHS [46, 47]), we are using its billion 476 years of evolutionary distance with human and other mammals as the best possible protection against an accidental 477 infection of laboratory workers or the spread of a dreadful virus once infecting Pleistocene mammals to their contem-478 porary relatives. The biohazard associated with reviving prehistorical amoeba-infecting viruses is thus totally negligi-479 ble, compared to the search for "paleoviruses" directly from permafrost-preserved remains of mammoths, woolly rhi-480 noceros, or prehistoric horses, as it is now pursued in the Vector laboratory (Novosibirsk, Russia) [68], fortunately a 481 BSL4 facility. Without the need of embarking on such a risky project, we believe our results with Acanthamoeba-infecting 482 viruses can be extrapolated to many other DNA viruses capable of infecting humans or animals. It is thus likely that 483 ancient permafrost (eventually much older than 50,000 years, our limit solely dictated by the validity range of radiocar-484 bon dating) will release these unknown viruses upon thawing. How long these viruses could remain infectious once 485 exposed to outdoor conditions (UV light, oxygen, heat), and how likely they will be to encounter and infect a suitable 486 host in the interval, is yet impossible to estimate. But the risk is bound to increase in the context of global warming 487 when permafrost thawing will keep accelerating, and more people will be populating the Arctic in the wake of industrial 488 ventures.