The Evolution of Antiviral Defense Systems

lab-led

A synthesis arguing that antiviral defenses across the three domains of life reuse a small set of designs, antisense recognition joined to nuclease activity and later to transcriptional and secreted responses, and proposing that chordates lost RNA interference through incompatibility with interferon.

Benjamin R. tenOever

2016 · Cell Host & Microbe · perspective

Senior authors

Benjamin R. tenOever

Correspondence

Benjamin R. tenOever

Research areas & themes

Citation

tenOever BR. The Evolution of Antiviral Defense Systems. Cell Host & Microbe. 2016. Volume 19, issue 2, pages 142-149.

DOI 10.1016/j.chom.2016.01.006. PMID 26867173.

Note on provenance. The source file is named tenOever_CHM2015 but the article was published in February 2016, and the inventory metadata is the authority. This is a single-author Perspective, so the great majority of the science discussed was produced by other laboratories and is attributed to them throughout this record.

One-sentence contribution

A synthesis arguing that antiviral defenses across the three domains of life reuse a small set of designs, antisense recognition joined to nuclease activity and later to transcriptional and secreted responses, and proposing that chordates lost RNA interference through incompatibility with interferon.

Executive summary

This Perspective traces antiviral defense across bacteria, archaea and eukaryotes and asks how the changing nature of genetic parasites shaped the systems that oppose them. The argument is organized by evolutionary parsimony, the principle that a component present in all three domains probably existed in their last universal ancestor, which the author notes is probabilistic and not universally accepted. Drawing chiefly on comparative genomic work by Koonin and colleagues, the article sets out a trajectory in which self-replicating RNA gave rise to polymerases and then to DNA-based parasites, so that the earliest defenses faced mobile DNA and could do little better than transcribe complementary antisense RNA. Sequence-specific endonucleases, Argonaute proteins guided by small nucleic acids, CRISPR-Cas, and the eukaryotic piRNA and RNA interference systems are then presented as successive combinations of the same two elements, a guide and a nuclease. In chordates, pattern recognition receptors derived from Toll-like proteins, antiviral cytokines including tumor necrosis factor and the interferons, and recombinatorial adaptive immunity replaced small RNA defense. The article's own proposal, labeled an attractive hypothesis, is that this replacement was forced by incompatibility. Chordate size and receptor-mediated virus spread would require an RNA-dependent RNA polymerase for systemic small RNA amplification, and expressing such a polymerase in mammalian somatic cells triggers innate immunity. Work from the tenOever laboratory supplies two supporting observations, that engineered mammalian RNA interference can protect against virus without interferon, and that microRNA function in acute infection appears limited to cytokine control.

Scientific context

Whether chordates retain an antiviral function for RNA interference was, at the time of writing, actively contested, and the article says so. Reports from the laboratories of Ding and of Voinnet argued for antiviral RNA interference in mammalian cells, while a joint opinion piece by Cullen, Cherry and tenOever questioned its physiological relevance, and work from Bogerd, Cullen and colleagues reported that replication of many human viruses is refractory to inhibition by endogenous cellular microRNAs. Separately, comparative genomics had begun to resolve the deep history of both viruses and defense systems. Koonin and colleagues had argued that viruses have no single common ancestor, that their diversity arose through genetic exchange, that Polintons may be the ancestors of most eukaryotic DNA viruses, and that most eukaryotic RNA viruses may descend from DNA phages that acquired an RNA-dependent RNA polymerase. Koonin and Krupovic had proposed that both known adaptive immune systems, CRISPR-Cas and the immunoglobulin system, were built when transposable elements conferred recombination capacity on pre-existing innate responses. The Perspective assembles these separate literatures into one narrative and adds a specific evolutionary argument about the loss of RNA interference in chordates.

Central question

Given that the composition of the virome changed over evolutionary time, from mobile DNA elements in early cells to the RNA viruses that dominate eukaryotes, how did that change shape the succession of antiviral defense systems observed across the tree of life, and why did chordates abandon RNA interference in favor of the interferon system.

Experimental strategy

No experiments are reported. The method is inference from conservation and phylogeny under evolutionary parsimony, applied to defense systems and, in parallel, to the evolution of the parasites themselves, since the argument is that the two histories constrain each other. The article states explicitly that the trajectory it proposes for each defense system is necessarily speculative and rests on observed conservation rather than on demonstration. Two figures organize the synthesis. One presents cartoon models of antisense RNA, restriction nucleases, Argonaute-based targeting, CRISPR-Cas, piRNAs, RNA interference, interferons and immunoglobulins side by side. The other places the appearance of each system on a timeline against the three domains of life.

Key findings

This is a Perspective, so the entries below are the article's principal arguments and the evidence it marshals, not results generated by the author.

1. Applying parsimony, a component found in bacteria, archaea and eukaryotes likely existed in the last universal ancestor, and the best route to reconstructing the first antiparasite defense is comparative analysis of genes that process nucleic acid. The article credits this framework to Woese and colleagues, to Koonin, and to Anantharaman, Koonin and Aravind, and notes that the approach is probabilistic and contested.
2. Virus diversity does not trace to a single ancestor and arose through genetic exchange, with DNA viruses dominating prokaryotes and RNA viruses abounding in eukaryotes. From this the article infers that the first genuine genetic parasite of unicellular life was DNA based, and that the scarcity of RNA phages suggests early cellular environments were poorly suited to RNA spread. The underlying phylogenetics is attributed to Koonin and colleagues, Edwards and Rohwer, and Krupovic and colleagues.
3. The earliest defense available to a protocell would have been transcription of antisense RNA producing steric interference with a mobile element, a strategy still widely evident in bacteria and archaea, where small RNAs under about 150 nucleotides control plasmid, transposon and phage amplification. Attributed to Wagner and colleagues, Gottesman and Storz, and Thomason and Storz.
4. The emergence of DNA phages is proposed as the driver for sequence-specific endonucleases, with self protection arising by modification of host DNA, creating the two-component restriction and modification design. Attributed to Kobayashi, Ishikawa and colleagues, and Mochizuki and colleagues, with the causal reading offered by the author as speculative.
5. Three later systems are presented as fusions of antisense specificity with nuclease activity. Prokaryotic Argonaute proteins use small DNA or RNA guides against mobile elements, work attributed to Makarova and colleagues, Olovnikov and colleagues, and Swarts and colleagues. CRISPR-Cas, present in about 90 percent of sequenced archaeal and 40 percent of bacterial genomes, incorporates phage or plasmid fragments into genomic loci and transcribes them as guides, attributed to Marraffini, Grissa and colleagues, and van der Oost and colleagues. The eukaryotic piRNA pathway uses clustered transposon-derived small RNAs of 25 to 30 nucleotides with a Piwi nuclease, largely in germline cells, attributed to Ishizu and colleagues, Iwasaki and colleagues, and Gunawardane and colleagues. The article states that piRNA and CRISPR share no common ancestry and are analogous rather than homologous.
6. RNA interference is presented as an elaboration on the prokaryotic Argonaute theme, with Dicer generating 19 to 21 nucleotide fragments loaded into RISC. A key distinction drawn is amplification. Plants and worms encode RNA-dependent RNA polymerases acquired from phage, while arthropods do not, so their defense depends on transport of small interfering RNAs and weakens with distance from infection, a point attributed to Saleh and colleagues and to Tomoyasu and colleagues.
7. Whether chordates retain antiviral RNA interference is described as unresolved and controversial, with conflicting positions cited from Cullen, Cherry and tenOever, from Li and Ding and colleagues, and from Maillard, Voinnet and colleagues. The article's own position is that eukaryotic microRNAs descend from the ancestral antiviral system but no longer act in an antiviral capacity, supported by work from the tenOever laboratory reported by Aguado and colleagues and by Bogerd, Cullen and colleagues, and consistent with the finding of Obbard and colleagues that in Drosophila only Dicer-2, the antiviral one, shows the selective signature of immune function.
8. In place of RNA interference, chordates use pattern recognition receptors evolved from Toll-like proteins that originated roughly 700 million years ago in development and were independently co-opted as sentinels, attributed to Leulier and Lemaitre and to Putnam and colleagues, coupled to secreted cytokines. Tumor necrosis factor is cited as an early antiviral cytokine with about 550 million years of functional conservation, attributed to Quistad and colleagues, with interferons and interleukins following.
9. The article's own hypothesis is that loss of RNA interference in chordates reflects incompatibility with the interferon system rather than simple redundancy. The reasoning is that large chordates would need polymerase-driven amplification and circulation of small interfering RNAs, that receptor-mediated cell entry prevents the kind of passive spread plants exploit, and that expression of an RNA-dependent RNA polymerase in mammalian somatic cells itself triggers innate immunity, shown independently by Painter and colleagues and by Yu and colleagues. Supporting evidence for mutual exclusivity is drawn from Billy and colleagues, Paddison and colleagues and Maillard and colleagues on the divergent responses of stem and somatic cells to long double-stranded RNA, and from Girardi and colleagues and Seo and colleagues on reciprocal inhibition between RISC and interferon signaling. The author notes that work from his own laboratory, reported by Benitez and colleagues, showed that engineered mammalian RNA interference can protect against virus when the requirement for an RNA-dependent RNA polymerase is removed.
10. Both known adaptive immune systems are argued to derive from the very elements they defend against, with CRISPR-Cas built largely from casposon-derived genes and the RAG recombinase derived from a Transib family transposon. Attributed to Koonin and Krupovic, Krupovic and colleagues, Kapitonov and Jurka, and Kapitonov and Koonin. The article also notes that comparative analysis of gene duplication shows defense systems diverging faster than any other process, citing Daugherty and Malik and Siddle and Quintana-Murci.

Mechanistic model

The Perspective does not establish a mechanism and says of its central evolutionary claims that they are necessarily speculative. What it offers is a structural account and one testable hypothesis.

The structural account is that antiviral defense has repeatedly been assembled from a small parts list. A guide that recognizes foreign sequence, first as an antisense transcript and later as a small RNA or DNA, combined with an effector, first steric hindrance and later a nuclease, then extended in multicellular organisms by a transcriptional response and finally by a secreted signal that protects cells that are not yet infected. Adaptive systems in this account are innate systems that acquired recombination from a transposable element.

The specific hypothesis, about why chordates replaced RNA interference with interferon, is presented explicitly as an attractive hypothesis whose evolutionary cause cannot be determined. It links three premises, that systemic small RNA defense in a large organism requires amplification by an RNA-dependent RNA polymerase, that chordate virus spread by receptor-mediated entry does not carry small RNAs along with it as plant cell-to-cell movement does, and that such a polymerase cannot be expressed in mammalian somatic cells without triggering innate immunity. The correlative evidence for incompatibility, the divergent handling of long double-stranded RNA in stem versus somatic cells and the reciprocal inhibition between RISC and interferon signaling, is consistent with the hypothesis but does not establish that incompatibility caused the evolutionary loss. The article is careful to frame what would follow if the hypothesis holds, namely that the loss of RNA interference in chordates may have everything to do with polymerase biology.

Conceptual or technical advance

The article's contribution is framing. By reading defense systems against the changing composition of the virome rather than in isolation, it makes the sequence of systems look like responses to a sequence of problems, and it identifies the guide plus nuclease pairing as the recurring solution across domains that share no common defense ancestry.

The sharper contribution is the incompatibility hypothesis, which converts a descriptive observation, that chordates have interferon and appear not to use antiviral RNA interference, into a mechanistic proposal with an identified molecular obstacle. Because that obstacle is the innate immune consequence of RNA-dependent RNA polymerase expression, the proposal is addressable experimentally, and the article points to a test already performed in the author's own laboratory, that engineering around the polymerase requirement allows RNA interference to substitute for the interferon response in mammalian cells.

Relationship to the broader research program

This Perspective sets the conceptual frame for a line of work in the tenOever laboratory on whether mammals use, or could be made to use, RNA-based antiviral defense. Two of the laboratory's own papers carry weight in the argument. Benitez and colleagues in 2015 reported that engineered mammalian RNA interference can provide antiviral protection that negates the requirement for the interferon response, and Aguado and colleagues in 2015 reported that microRNA function in the acute response to virus infection is limited to cytokine control. The 2013 opinion piece co-written with Cullen and Cherry is the laboratory's earlier position on the same controversy.

Category 3 synthesis. Placed beside the laboratory's engineering papers, in which RNA viruses were built to produce microRNAs and then used to deliver silencing libraries, this Perspective supplies the reason those experiments are informative rather than merely technical. The recurring question across those papers, why RNA viruses do not naturally exploit the host small RNA machinery and why mammals do not use it against them, is here given an evolutionary answer in terms of incompatibility with interferon. That connection is visible when the papers are read together and is not asserted by any one of them.

Related publications

• Benitez and colleagues, 2015, Engineered Mammalian RNAi Can Elicit Antiviral Protection that Negates the Requirement for the Interferon Response, from the tenOever laboratory and cited here. Predecessor supplying evidence for the central hypothesis.
• Aguado and colleagues, 2015, microRNA Function Is Limited to Cytokine Control in the Acute Response to Virus Infection, from the tenOever laboratory and cited here. Predecessor.
• Cullen, Cherry and tenOever, 2013, Is RNA interference a physiologically relevant innate antiviral immune response in mammals, cited here. Review or synthesis, the earlier statement of the position developed in this Perspective.
• Varble and colleagues, 2010, Engineered RNA viral synthesis of microRNAs, and Langlois and colleagues, 2012, In Vivo Delivery of Cytoplasmic RNA Virus-derived miRNAs. Conceptual extension in the reverse direction, since this Perspective supplies the evolutionary reading of why RNA viruses do not naturally make microRNAs.
• Varble and colleagues, 2013, An In Vivo RNAi Screening Approach to Identify Host Determinants of Virus Replication. Application, using engineered small RNA delivery as a tool in the interferon-competent host discussed here.
• Koonin and Krupovic, 2015, on the evolution of adaptive immunity from transposable elements combined with innate immune systems, from another group and a principal source for the closing argument. Conceptual foundation.

Limitations and boundaries

The article is a Perspective by a single author, and it is not evidence that the tenOever laboratory performed the work it describes. Almost all of the primary findings summarized here belong to other groups, above all the comparative genomics of virus and defense system evolution developed by Koonin and colleagues, the CRISPR-Cas literature, the prokaryotic Argonaute and piRNA literatures, and the mammalian RNA interference work from the Ding, Voinnet, Cullen, Pfeffer and Sullivan laboratories. Only the Benitez, Aguado and Cullen, Cherry and tenOever citations are from the author's own group.

The evolutionary reconstructions are inferences from conservation under parsimony, which the article itself states is probabilistic and not universally agreed upon, and it describes the proposed trajectory for each defense system as necessarily speculative. No experiment, dataset or analysis is presented, so nothing here is independently verifiable from the article. The exact phylogeny of virus evolution is stated to be unachievable, so the ordering of parasite types that motivates the ordering of defenses is itself uncertain.

The central hypothesis about chordates is explicitly labeled as one attractive hypothesis among possible explanations, and the article says the evolutionary cause of the transformation cannot be determined. The evidence assembled for incompatibility between RNA interference and interferon is correlative, drawn from cell culture comparisons of stem and somatic cells and from reciprocal inhibition experiments, and the question of whether chordates retain antiviral RNA interference is presented as unresolved rather than settled. The article allows that any residual use may be confined to specialized pluripotent cells. Coverage is also acknowledged to be incomplete, since the author apologizes for original work that could not be cited for reasons of space, and whole areas including plant defense, programmed cell death and dormancy, and viral counter-defense are treated only briefly.

Audience summaries

25 words

Defenses against viruses, from bacterial antisense RNA to antibodies, reuse a few designs. This Perspective asks why vertebrates traded RNA silencing for interferon, and proposes incompatibility.

75 words

Across bacteria, archaea and eukaryotes, defenses against genetic parasites repeatedly pair a sequence-specific guide with a nuclease, from antisense RNA and restriction enzymes to Argonaute, CRISPR-Cas, piRNAs and RNA interference. Vertebrates instead use pattern recognition receptors, interferons and antibodies. This single-author Perspective, synthesizing work from many laboratories, argues that vertebrates could not keep RNA interference because systemic small RNA defense needs a viral-type polymerase, and expressing one triggers innate immunity.

150 words

This Perspective reads the succession of antiviral defense systems against the changing composition of the virome, arguing that early cells faced mobile DNA and could answer only with antisense transcription, that DNA phages drove sequence-specific endonucleases and then restriction and modification, and that Argonaute, CRISPR-Cas, piRNAs and RNA interference are successive fusions of a nucleic acid guide with a nuclease. Chordates instead built pattern recognition receptors from Toll-like proteins and coupled them to secreted cytokines, with tumor necrosis factor among the earliest and interferons and recombinatorial adaptive immunity following. The article proposes that RNA interference was not merely made redundant but was incompatible, since protecting a large organism would require an RNA-dependent RNA polymerase to amplify and circulate small RNAs, and expressing such a polymerase in mammalian somatic cells triggers innate immunity. Supporting observations come from many groups, including two studies from the author's own laboratory. The evolutionary argument is presented as hypothesis.

Discoveries supported by this paper

Documented publication relationships

• microRNA Function Is Limited to Cytokine Control in the Acute Response to Virus Infection — predecessor. Direction: this paper → linked paper. stated in the Related publications section of 2016-tenoever-the-evolution-of-antiviral-defense
• Is RNA Interference a Physiologically Relevant Innate Antiviral Immune Response in Mammals? — review or synthesis. Direction: this paper → linked paper. stated in the Related publications section of 2016-tenoever-the-evolution-of-antiviral-defense
• Engineered RNA viral synthesis of microRNAs — conceptual extension. Direction: this paper → linked paper. stated in the Related publications section of 2016-tenoever-the-evolution-of-antiviral-defense

Pathogens

Technologies