Freezing viruses in time

The 2014 outbreak of Ebola in West Africa has generated understandable interest in how long viruses can survive outside of a host. Is it a matter of minutes, several hours, a few days? How about 700 years? Remarkably, a new study in PNAS by Ng et al. (1) shows that viruses can survive intact in ice patches for at least this period and retain their infectivity.

Ng et al. (1) based their study on a 700 ± 40-y-old caribou fecal sample cored from a permanent ice patch in the Northwest Territories of Canada. From this sample, the authors recovered two novel viruses: a single-stranded DNA virus that they termed Ancient caribou feces-associated virus (aCFV), and the partial sequence of a single-stranded RNA virus called the Ancient Northwest Territories cripavirus (aNCV). That aCFV is a distant relative of the geminiviruses, a well-known group of plant viruses, strongly suggests that the virus was present in the plant diet of the caribou. aNCV is more intriguing, in part because it is most closely related to a group of insect RNA viruses (dicistroviruses) not known to infect plants, such that an insect may have been consumed with the vegetable matter or died on the fecal sample. Studies of “ancient” RNA viruses are also few and far between, and many have question marks over their authenticity. Indeed, the pioneering work recovering the genome of the influenza virus associated with the notorious pandemic of 1918–1919 still serves as the benchmark for work in this area (2, 3). Although aCFV and aNCV are not the oldest viral genomes obtained, and earlier this year it was proposed that a replication-competent DNA virus (Pithovirus sibericum) could be recovered from Siberian permafrost dating to 30,000 y ago (4), they are undoubtedly some of the most believable.

Challenges in Studying Ancient Pathogens

The characterization of aCFV and aNCV looks robust because Ng et al. (1) took a meticulous approach to their work, following many of the guidelines set down for the study of ancient DNA (5). Although a number of purportedly ancient viruses have been examined, it is unclear how many will stand the test of time because the specter of contamination, a dominant factor in all ancient DNA studies, cannot be completely excluded. For example, one of the issues that has plagued studies of ancient viruses and bacteria is that the laboratories performing this work have often housed contemporary microbes, so that

The characterization of aCFV and aNCV looks robust because Ng et al. took a meticulous approach to their work, following many of the guidelines set down for the study of ancient DNA.

contamination can sometimes be difficult to exclude. It is therefore essential that all such work be independently replicated, which is also the case for P. sibericum. Satisfyingly, the study of Ng et al. (1) was confirmed in a specialist ancient DNA laboratory, so that the risk of inadvertent contamination was greatly reduced. Although definitive proof that these viruses come from the fecal material and not from more recently deposited ice encountered during the process of extraction is probably impossible to obtain, Ng et al. took as many precautions as are reasonably possible.

Despite all this careful work, aCFV and aNCV do lack two of the major signatures of sample authenticity: that their nucleic acids are damaged because of inevitable sample degradation and that the branch lengths leading to these ancient viruses are shorter (i.e., closer to the root of the tree) than those of contemporary viruses. Although at face value these may seem like major limitations, there are important mitigating circumstances. The lack of damage could reflect the fact that the viral genomes are protected by a highly durable capsid shell (see below) and were rapidly frozen. In the case of aCFV and aNCV, the rapid freeze happened because the fecal sample was deposited on a permanent ice patch, with the ice core in this case taken more than 1 m below the surface. Hence, it may be that frozen encapsidated viruses break the rule that damage is a necessary expectation in ancient DNA, although this clearly needs to be explored further. The absence of short branch lengths may simply reflect the fact that there are no contemporary viral relatives by which to make a meaningful comparison, and that 700 y is only a small proportion of the total depth of the trees in question.

The discovery of two 700-y-old viruses has a number of wider implications. First, that aCFV and aNCV are so divergent from contemporary viruses not only means that there has likely been a turnover (birth-and-death) of virus lineages through time, but it serves as a timely reminder of how little of the virosphere has been explored. Along with the in vogue metagenomic screening of contemporary host populations, a broader survey of archival samples may provide important insights into the diversity of the virus world. Fecal samples, including those frozen in ice, may be a particularly rich source of viral diversity, and provide a good representation of the viruses that are inevitably encountered with the ingestion of food. As with humans (6), the analysis of caribou fecal samples hints at a virome composed of plant viruses that are ingested and then excreted with no active host infection.

On a more practical level, the analysis of ancient viruses like aCFV and aNCV will be central to understanding the dynamics of viral evolution, which have to date largely been confined to viruses sampled over time-scales of decades or based on the assumption that hosts and viruses codiverge over millions of years. As a case in point, there have been long-running questions over whether evolutionary rates in viruses are “time-dependent,” such that they are highest among closely related sequences and then decline with time, and how quickly one of its major causes—site saturation—occurs (7). Ancient viruses could provide robust and valuable calibration points for these molecular clock-based studies. However, obtaining viruses from archival clinical samples, which will be central to understanding the emergence of human infectious diseases, still represents a major challenge because sample degradation is likely to set in far more quickly than in ice patches.

The Hardy Viral Capsid

Perhaps the most remarkable finding of the Ng et al. (1) study is that the viruses were intact and that aCFV was infectious in experiments using modern plant (Nicotiana benthamiana) populations. Indeed, it will come as a surprise to many that the usually highly mutable viral nucleic acids were better preserved than the caribou mitochondrial DNA taken from the same ice core. As noted above, a possible explanation for this remarkable preservation is that the genomic material of viruses is contained within a robust viral capsid: the protective protein coat that is a defining feature of all viruses. Structural studies have found deep evolutionary relationships between capsid structures that have no apparent sequence similarity, including between RNA and DNA viruses, suggesting that they share an ancient common ancestry (8). Within the diversity of capsid structures, those with a simple icosahedral structure are particularly commonplace, including both geminiviruses and dicistroviruses, and are seemingly produced by homologous proteins. It therefore seems likely that icosahedrons provide viruses with uniquely stable structural characteristics. The long-term preservation of viral capsids in ice patches raises the more general question of whether this protein coat evolved as a means of viral survival in environments where hosts were sparse. Although it remains to be seen how long viral capsids can remain intact, it is interesting that aNCV is a member of a large order of RNA viruses known as the Picornavirales, all of which contain icosahedral capsids, and which are commonly found in environmental metagenomic surveys, including in “harsh” localities like the sea (9). In the same way, it might be expected that viruses that possess an additional and more permeable envelope protein surrounding the capsid (such as influenza virus) would degrade much more rapidly, and hence would not be expected to be intact in ancient samples. It may even be possible to perform experimental studies to test the preservation ability of viral capsids by simulating the degradation process.

Another of the intriguing aspects of this study is that aCFV remained infectious after 700 y in ice, adding weight to suggestions that the release of once frozen viruses might represent a potential source of new emerging pathogens (4). The infectivity of aCFV in N. benthamiana is particularly impressive because this relative of tobacco is native to northern Australia and so is clearly not the natural host of aCFV. Although this is a theoretical possibility, the risk posed by the potential release of viruses trapped in ice is likely to be minimal, particularly compared with the “normal” horizontal spread of viruses among susceptible host populations that occurs on a regular basis. In addition, the warmer temperatures associated with ice melting would presumably lead to the eventual degradation of the viral nucleic acids. Indeed, previous claims that viruses have been released from ice have been debunked, with contamination again the culprit (10).

The work on aCFV and aNCV by Ng et al. (1) represents an important step forward in studies of viral evolution, and reminds us that the characterization of any ancient microbe is only as good as the care taken to process and analyze the samples in question. The rigorous study of ancient DNA is evidently not for the faint hearted. As a case in point, much of the ongoing debate over what strains of Yersinia pestis (plague) bacteria were responsible for past pandemics may simply revolve around contamination issues (11). However, although studies of viral diversity from frozen samples may eventually open up research into viral evolution on a scale of millennia, it is likely that the only way to reveal the nature of past viral populations on very deep time-scales, depicting millions of years of viral history, is through genomic “fossils” in the guise of the endogenous viral elements that are being increasingly documented in eukaryotic genomes (12), and which ushered in the new science of paleovirology (13).