Cell Host & Microbe Commentary

Sociovirology: Conflict, Cooperation, and Communication among

Samuel L. Dı´az-Mun˜ oz,1,* Rafael Sanjua´ n,2,3 and Stuart West4 1Department of and Molecular Genetics, University of California, Davis, One Shields Avenue, Davis, CA 95616, USA 2Institute for Integrative Systems Biology (I2SysBio), Universitat de Vale` ncia, C/Catedra´ tico Agustı´n Escardino 9, 46980 Paterna Valencia, Spain 3Department of Genetics, Universitat de Vale` ncia, C/Dr. Moliner 50, Burjassot, Vale` ncia 46100, Spain 4Department of Zoology, University of Oxford, South Parks Road, Oxford OX1 3PS, UK *Correspondence: [email protected] https://doi.org/10.1016/j.chom.2017.09.012

Viruses are involved in various interactions both within and between infected cells. Social evolution theory offers a conceptual framework for how -virus interactions, ranging from conflict to cooperation, have evolved. A critical examination of these interactions could expand our understanding of viruses and be exploited for epidemiological and medical interventions.

Virus-virus interactions are pervasive and Potential misgivings with this social- of another individual. As such, evolu- highly diverse (DaPalma et al., 2010; evolution approach include the idea that tionary analysis of social interactions re- Figure 1). Some viruses need another, viruses are too simple to interact socially, lies fundamentally on natural selection. ‘‘helper’’ virus to complete their infection or that to infer traits like cooperation The only major difference with non-social cycle, and other viruses are commonly would be anthropomorphic. On the flip models is that two or more interacting activated or suppressed by the presence side, social interactions have driven the individuals are involved. For example, of secondary viral infections. Viral pro- evolution of life at all levels of complexity cooperative traits, which benefit others, teins can mix and produce mosaic-like and thus could have a role in viral will only be selected for by natural selec- viral particles (pseudotypes) when a cell evolution as well: genes cooperate to tion if they provide a fitness benefit to the is coinfected with two different viruses. form genomes, cells cooperate to form individual performing them (Figure 2A). Viral coinfection of microbes is wide- multicellular organisms, and multicellular Cooperative traits cannot evolve merely spread (Dı´az-Mun˜ oz, 2017), and viruses organisms form complex cooperative so- because they provide a benefit to a have mechanisms enabling multiple cieties. Conflict is also ubiquitous: ‘‘self- population. Furthermore, the evolution of viral genomes to be cotransmitted in the ish’’ genes drive their transmission at a cooperation does not require the ability same infectious unit (reviewed in Sanjua´ n, cost to the genome, cells exploit collec- of individuals to foresee the conse- 2017). Coinfecting viral genomes can be tively produced goods like nutrient- quences of their actions. In other words, distinct, variants of the same virus, or scavenging molecules, and organisms cooperation and social evolution can be even genetically identical, suggesting compete with each other for resources. explained by, and are not at odds with, different types of functional interplay. Since no complex phenotypes are neces- individual-based natural selection. Furthermore, use a form sary for the evolution of social traits, the One way cooperation can be evolution- of communication to regulate lysis of the appearance of anthropomorphism or the arily favored is when it is directed toward infected cell (Erez et al., 2017). Finally, ‘‘simple’’ organization of viruses may not genetically identical individuals or rela- virus-virus interactions in the absence of pose problems for the application of tives. From an evolutionary perspective, cellular coinfection can also be mediated sociobiology models in . helping a genetically identical individual by changes at the host level, such as We argue for a social evolution reproduce is the same as reproducing immune responses. approach to understand and predict yourself. By extension, helping relatives Despite this growing body of empirical virus-virus interactions. This framework reproduce provides a fitness benefit to evidence suggesting virus-virus interac- can clarify unexplained phenomena in the actor, as relatives share a fraction of tions, we lack a well-founded conceptual virus-virus interactions by identifying the actor’s genes. This process by which framework that provides an understand- evolved viral social traits, their genetic individuals increase their fitness by ing of how these interactions have and mechanistic basis, and the selective helping relatives reproduce is termed kin evolved and how they could shape viral pressures underlying these traits. We selection. pathogenesis. Social evolution theory outline how this approach could lead to An important potential difficulty in was originally developed to explain animal new breakthroughs in both fundamental applying social evolution to virology re- behavior, but has since been extended to and applied virology. sides in the definition of an individual. microorganisms, including bacteria and One may consider the virion as an individ- unicellular eukaryotes. Yet this social Cooperation and the Individual ual. Yet in some cases, including para- perspective has not been embraced in Social interactions take place when the myxoviruses, birnaviruses, filoviruses, the study of viruses. traits of one individual influence the fitness retroviruses, and inoviruses, a virion can

Cell Host & Microbe 22, October 11, 2017 ª 2017 The Authors. Published by Elsevier Inc. 437 This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Cell Host & Microbe Commentary

Figure 1. Virus-Virus Interactions Are Diverse and Provide Multiple Opportunities for Social Evolution (A–F) Indicated by gray shadows and depicting social interactions, where the action of a viral genome changes the fitness of other viral genomes. Cells are ovals with black borders. (A) A viral genome (thick curved segment) enters the cell and performs transcription (mRNAs, thin curved segments +) and translation, leading to generation of shared intracellular viral (open hexagons). Transcription-translation is a social and potentially cooperative trait because it benefits the other viral genomes in the cell. (B–F) Viral genomes can prevent or promote reproduction of other genomes by changing the probability that they can infect a cell or host by: (B) blocking the entry of other viral genomes into the cell; (C) producing host-level immune changes (dark red outline) that favor the transmission of all infecting viral genomes; (D) inducing the cell to produce molecules essential for transmission to neighboring cells, benefiting all viral genomes in the cell; (E) producing viral proteins that communicate cell infection status, signaling to other viral genomes the abundance of cells available for reproduction; and (F) manipulating host immune signals to induce distant cells to differentially expose receptors favoring entry of some viral genomes over others.

carry multiple genome copies (n-ploidy). If we think of a genome as an individ- Conflict and Cheating Viruses Functionally, these n-ploid virions should ual, then the natural history of viruses is Evidence in favor of viruses acting as be similar to multiple virions entering the filled with opportunities for social inter- cooperative social agents is, paradoxi- same cell, because they would both lead actions (Figure 1), even though experi- cally, provided by the occasional spread to infections with multiple viral genomes. mental demonstration is still lacking in of uncooperative ‘‘cheats.’’ If individuals As such, we suggest that the definition of many cases. A single viral genome are not genetically identical, they will not an individual should be set at the level of entering a cell needs to accomplish both necessarily have the same evolutionary the single infectious viral genome. replication and gene expression (tran- interests, creating the potential for con- Another potential complication stems scription and translation). The mRNAs flict. Social evolution theory predicts from the fact that some viruses, notably and proteins resulting from viral gene that if multiple genomes infect a cell, one RNA viruses, show extremely high muta- expression can provide a collective that invests less in cooperative traits tion rates, leading to the suggestion that benefit to genomes in the cell, such as such as transcription and invests more in the minimal level at which an individual generating and proteins that its own replication will be favored, RNA virus can be defined is a sequence block host immunity. The fact that these because it will benefit from cooperation cloud, or quasispecies (Andino and Dom- factors act as public goods permits, but without paying as much of the cost ingo, 2015). In this case, the individual does not guarantee, cooperation (Chao (Chao and Elena, 2017). This is analogous could be redefined as the consensus (or and Elena, 2017). Cooperation is more to the tragedy of the commons in humans, predominant) sequence of the quasispe- likely to evolve if most interacting ge- where cooperation breaks down due to cies. If, however, functional social interac- nomes are identical, since such high selfish interests, even though everyone tions are established among genetic relatedness allows kin selection to oper- could do better in the long run by variants within such clouds, then it re- ate (Turner and Chao, 1999). Studied cooperating. mains useful to keep the definition of the with this social evolution lens, even the Two important variables determine the individual at the single viral genome level, infection and replication of a single viral extent to which multiple genomes infect because this allows identifying and genome within a single cell is an inher- a cell: the multiplicity of infection (MOI) analyzing such interactions. ently social process (Figure 1A). and the rate of spontaneous mutation.

438 Cell Host & Microbe 22, October 11, 2017 Cell Host & Microbe Commentary

Figure 2. The Problem of Cooperation and Testing Virus-Virus Interactions Using Social Evolution Theory (A) Imagine a population of viruses that perform a costly cooperative action such as transcription (indicated by C). A selfish cheater that does not perform the cooperative action (indicated by S) arises in this population through mutation or coinfection. This selfish cheater is able to benefit from the cooperative behavior of the cooperators without paying the cost. Consequently, the selfish cheater will increase in frequency, even though this leads to a reduction in mean fitness. (B) Testing for evolved social interactions, such as cheating and cooperation. Left: Strains are competed in isolation and in combination to quantify fitness and confirm social traits. Middle: When grown as single infections, a population of cooperators (blue) achieves higher growth than a cheater population (red). Right: In coinfection (purple), the population has lower growth than a population composed solely of cooperators. The social composition of the population reveals why: cheaters (dotted red) initially increase population growth by exploiting cooperators (dotted blue), but growth stalls as the population becomes dominated by cheaters.

High mutation rates such as those ex- fitness. In the laboratory, artificially high Diversity and Heterotypic hibited by RNA viruses result in extremely MOIs are well-known to promote the Coinfection diverse populations, whereas a high MOI emergence of DIPs (Marriott and Dim- As detailed above, kin selection theory can bring genomes from different line- mock, 2010). However, a pending chal- predicts that high coinfection rates select ages into the same host cells. These vari- lenge is to measure MOIs in vivo and to against cooperation, because they re- ables can enable genomes to interact in determine the ensuing levels of intra-host duce genetic relatedness. However, coinfection, as long as there is sharing of genetic relatedness in viruses. mutually beneficial cooperation involving viral products within cells. In turn, the association between muta- genetically distinct individuals (‘‘hetero- A clear experimental test of the associ- tion rate and cheating has been shown typic cooperation’’) can also evolve. All ation between MOI and cheater evolution in experimental populations of RNA that is required is that the individuals was generated with phi6 viruses treated with base analog muta- have some shared interest, such that (Turner and Chao, 1999). Serial passage gens. This selected for a fraction of viral they have higher fitness if they cooperate. at high MOI led to the evolution of a genomes with DIP-like behavior, greatly Cooperation between different genetic cheater virus optimized to increase its reducing population fitness (reviewed in variants is a familiar concept in virology, fitness in coinfection by favoring its Andino and Domingo, 2015). but rigorous evidence supporting it re- replication over other coinfecting viruses, Finally, evidence that viral proteins can mains scarce. The finding that population reducing average population fitness. function as public goods inside the cell genetic diversity correlated with the ability In contrast, passage at low MOI selected is supported by well-known processes of poliovirus to cause disease in mice led a virus optimized to grow efficiently in virology, such as pseudotyping, in to the suggestion of cooperation, with in monoinfection and to high popula- which a viral particle contains the enve- similar processes postulated for hepatitis tion fitness. Thus, coinfection generates lope proteins of a different virus. How- B virus, among others (reviewed in Andino conditions ripe for the emergence of ever, many plus-strand RNA viruses and Domingo, 2015). Measuring genetic cheats. replicate in well-defined subcellular diversity (e.g., using quasispecies theory Defective interfering particles (DIPs) structures. This situation suggests that, or mutation-selection balance) is a first appear to be ‘‘cheats.’’ DIPs usually have in some cases, intracellular viral product step to establish that interactions poten- a large portion of their genome deleted, sharing might be restricted, which would tially exist, but does not provide infor- including essential proteins, and hence tend to prevent the spread of cheaters. mation on the nature of the interactions can only reproduce in the presence of Thus, the details of viral population and their fitness outcomes. completely functional ‘‘helper’’ viruses. structure at the intracellular level and Genetic complementation, whereby However, their smaller genomes provide the identification and mechanistic char- the genetic defects of the interacting a replication advantage, outcompeting acterization of public goods will be viruses are mutually compensated, is not helper viruses in mixed infections of cells crucial milestones to reach in the socio- intrinsically cooperation, but represents and severely reducing viral population virology field. a possible mechanism for heterotypic

Cell Host & Microbe 22, October 11, 2017 439 Cell Host & Microbe Commentary

cooperation. A potential example comes evolved defecation to help dung beetles, dental interactions. Finally, if the condi- from influenza viruses, when one strain and so although defecation is mutually tions favoring a cooperative or cheating has a more efficient hemagglutinin, which beneficial, it is not cooperation. interaction are known or hypothesized, mediates virus attachment to host cells, Thus, the challenge resides in explain- evolutionary experiments in which these and another has a more efficient neur- ing cooperative traits that evolved to conditions are manipulated can provide aminidase that facilitates the release of benefit others, which becomes difficult further conclusive evidence for an evolved virions from infected cells. In cell culture, with complex interactions. Intercellular vi- social interaction. these viruses reproduce better together rus-virus interactions mediated through The recently discovered arbitrium sys- than they do independently, because changes in the host, such as breakage tem, whereby bacteriophages produce one is advantaged in cell entry and the of physical barriers or immune suppres- virus-encoded small peptides that are de- other is advantaged in cell exit (Xue sion, could easily fall into the elephant- tected by specific virus-encoded recep- et al., 2016). Multiple genome infections dung beetle category. Similarly, some of tors in neighboring cells, provides a good also lead to higher viral growth in measles these synergistic interactions reported test case (Erez et al., 2017). This system virus and even gain of functions, such as heterotypic cooperation between virus could be a cooperative trait, because it as extended cell tropism (Shirogane variants may be transient. specifically senses phage population den- et al., 2012). A possible reason why inter- The designation of byproduct rather sity and suppresses lysis. Thus, viruses acting genomes may surpass wild-type than cooperation does not imply that the forgo the chance to produce more prog- fitness is that some beneficial mutations observed phenomena are not important. eny to prevent the exhaustion of available show strong negative epistasis when Rather, they raise different questions host cells. If tests, as described above, combined in the same viral genome. about why they are favored by natural se- determine arbitrium is an evolved Adaptive immunity evasion in hepatitis lection, which can be crucial for under- social trait, this knowledge can be ex- C virus provides another suggested standing and manipulating these viral so- ploited, for instance, to develop phage instance of cooperation. Some variants cial interactions. The utility of identifying therapies that disrupt the arbitrium sys- may encode dominant antigens to monop- cooperation or cheating, as opposed to tem, achieving better bacterial clearance. olize immune responses and favor other a transient interaction, is that they repre- viruses, leading to a complex network of sent evolved social interactions. This Virulence and Social Interactions cross-immunoreactivity that mitigates the identification has two consequences. Virulence can be a social trait. In the immune pressure acting on certain vari- First, we can uncover the genetic under- simplest scenario, greater pathogen ants (Skums et al., 2015). As opposed to pinnings of the trait. For instance, the spe- growth will increase transmission, but the above examples, this may not be a cific mechanism underlying cheating by also increase host mortality (virulence). mutually beneficial interaction, because DIPs has led to DIP-derived treatments Consequently, overall viral transmission the variants encoding the flag antigen that depend only on the gene, rather can be maximized at an intermediate would play an altruistic role (Figure 1F). than the entire viral particle. Second, we growth rate, where the host is prudently To what extent this outcome is evolution- can establish the environmental and so- exploited, with the associated virulence arily stable is unclear because, as dis- cial pressures that led to selection of an unavoidable consequence. Myxoma cussed above, altruistic cooperative that trait. This knowledge of the selective virus provides a potential example of pru- behavior is critically dependent on genetic regime can lead to generalizable princi- dent exploitation in the field. Myxoma vi- relatedness and kin selection. ples, such as low-MOI vaccine propaga- rus causes a lethal disease known as tion to avoid the appearance of DIPs. myxomatosis in the European rabbit and Cooperation or Not Cooperation? In practice, the key first step will be to was released as a means to control rabbit As a note of caution, many of the above ex- test whether evolved social interactions populations in Australia and France. How- amples of distinct viral genomes engaging are really occurring between strains, and ever, after years of natural infection, the in positive or mutually beneficial interac- what form they take (Figure 2B). First, virus evolved toward slightly reduced tions are not necessarily cooperation. one can measure the fitness of each part- fatality rates and prolonged host survival Whereas cooperation is colloquially used ner alone and in combination (Borderı´a times, increasing opportunities for both as a synonym for ‘‘helping,’’ in evolutionary et al., 2015). This will show whether a vector-borne and direct transmission terms it requires that a trait provides a given interaction is beneficial or detri- (Kerr et al., 2015). benefit to another individual (mutual mental, and whether it can be exploited Virus or host dispersal can also influ- benefit or altruism) and that the trait has by potential cheats. Second, to show an ence the social evolution of virulence. evolved at least partially due to this benefit evolved social interaction, as opposed to Consider a spatially structured population (West et al., 2007). This latter clause is a transient interaction, the genetic mech- with limited dispersal of both virus and required because we are interested in anisms underlying the interaction can be hosts. In this case, if a viral strain grows whether the benefit to others is an adapta- uncovered and exploited in experimental quickly, then it can deplete the local sup- tion (Figure 2), and not just a byproduct of manipulations. For instance, one can en- ply of susceptible hosts and hence slow an otherwise selfish trait. For example, gineer strains with and without the trait transmission of that strain in the long when an elephant produces dung, this is to test whether a beneficial trait is cooper- run. Theoretical models predict that beneficial to the elephant, but also benefi- ation or a detrimental trait is cheating. This such spatially structured populations will cial to a dung beetle that uses that dung. experimentation can falsify alternative, select for lower virulence and hence However, elephants have clearly not non-adaptive explanations such as inci- more prudent host exploitation, as first

440 Cell Host & Microbe 22, October 11, 2017 Cell Host & Microbe Commentary

experimentally verified in an insect virus evolution theory can inform which types REFERENCES (Boots and Mealor, 2007). of viruses are more likely to be affected 479- In contrast to the above example, low by this phenomenon and its impact on Andino, R., and Domingo, E. (2015). Virology 480, 46–51. virulence can also evolve in response to routine laboratory studies. decreased relatedness. When host ex- The social evolution perspective also Boots, M., and Mealor, M. (2007). Science 315, ploitation relies on intra- and intercellular opens the door to engineering of new, bet- 1284–1286. cooperation between individual genomes, ter viral interventions by deploying our Borderı´a, A.V., Isakov, O., Moratorio, G., Hen- coinfection can select for cheats, leading knowledge of social evolution. New atten- ningsson, R., Aguera-Gonza€ ´ lez, S., Organtini, L., € to the breakdown of cooperation and, uated vaccines using DIPs could be Gnadig, N.F., Blanc, H., Alcover, A., Hafenstein, S., et al. (2015). PLoS Pathog. 11, e1004838–e20. consequently, lower parasite growth and generated by using social principles reduced virulence (Turner and Chao, 1999). underlying the emergence of cheats Chao, L., and Elena, S.F. (2017). Proc. Biol. Sci. 284 (Marriott and Dimmock, 2010). Theory on , 20170228–20170229. Implications cheating and how it impacts population DaPalma, T., Doonan, B.P., Trager, N.M., and A social perspective on virus evolution fitness could be used in epidemiology to Kasman, L.M. (2010). Virus Res. 149, 1–9. can open new research and treatment predict the best strategies to limit viral Dı´az-Mun˜ oz, S.L. (2017). Virus Evol 3, vex011. possibilities. Social thinking can cast spread. On the other hand, knowledge of viruses in a new light and question funda- kin-selected clonal reproduction can be Erez, Z., Steinberger-Levy, I., Shamir, M., Doron, S., Stokar-Avihail, A., Peleg, Y., Melamed, S., mental assumptions to advance basic used to increase viral yield in vaccine pro- Leavitt, A., Savidor, A., Albeck, S., et al. (2017). Na- virology. Similarly, by focusing on virus- duction or to increase gene de- ture 541, 488–493. virus antagonism and alliances, we may livery. Likewise, knowledge of evolved Kerr, P.J., Liu, J., Cattadori, I., Ghedin, E., Read, be able to uncover new weapons in our interactions between bacteriophages A.F., and Holmes, E.C. (2015). Viruses 7, own virus-host battle. could lead to improved phage therapy. 1020–1061. The central role of genetic relatedness In sum, there is growing evidence that Marriott, A.C., and Dimmock, N.J. (2010). Rev. and kin selection in the evolution of coop- virus-virus interactions are pervasive and Med. Virol. 20, 51–62. eration can help us better understand key that viruses have mechanisms to mediate processes in virology, such as superinfec- these interactions. We suggest that these Sanjua´ n, R. (2017). Trends Microbiol. 25, 402–412. tion exclusion, which makes the cell traits are best understood and predicted Shirogane, Y., Watanabe, S., and Yanagi, Y. refractory to entry of other, potentially using a social evolution approach. (2012). Nat. Commun. 3, 1235–1237. non-identical genomes, and why some vi- As such, sociovirology holds promise Skums, P., Bunimovich, L., and Khudyakov, Y. ruses tend to be transmitted as groups for breakthroughs in fundamental and (2015). Proc. Natl. Acad. Sci. USA 112, 6653–6658. (collective infectious units). The social applied virology. evolution perspective also draws atten- Turner, P.E., and Chao, L. (1999). Nature 398, 441–443. tion to certain long-held assumptions ACKNOWLEDGMENTS regarding standard virological practices. West, S.A., Griffin, A.S., and Gardner, A. (2007). For example, studies have found that pla- J. Evol. Biol. 20, 415–432. R.S. was funded by ERC Consolidator Grant que assays may not yield clonal popula- 724519-Vis-aVis. S.L.D.-M. was funded by NIH Xue, K.S., Hooper, K.A., Ollodart, A.R., Dingens, tions (reviewed in Sanjua´ n, 2017). Social grant 4R00AI119401-02. A.S., and Bloom, J.D. (2016). Elife 5, e13974.

Cell Host & Microbe 22, October 11, 2017 441