Matryoshka Dolls of Virus Evolution: The Internal Forces That Shape Viral Fate

Synopsis

Viral evolution unfolds across nested layers of adaptation, much like a set of Matryoshka dolls. The outermost, well-studied layer involves interactions between viruses and their hosts—where immune evasion, cross-species transmission, and long-term coevolution drive viral diversification. Yet, hidden within this framework is an often-overlooked inner layer: the coevolution of viruses with their own molecular parasites, defective interfering (DI) particles, and defective viral genomes (DVGs). These molecular parasites exploit viral replication machinery, reshaping infection dynamics and imposing selective pressures that influence viral fitness, transmission, and persistence. This perspective synthesizes evidence from experimental evolution, mathematical modeling, and molecular virology to propose a more integrated view of viral evolution. By framing host–virus interactions and virus-DI particle dynamics within a unified evolutionary framework, we highlight the underappreciated role of DI particles as evolutionary players, not just aberrant byproducts. Recognizing these internal layers of viral evolution may inform the development of antiviral strategies and broader questions in host–pathogen coevolution.

Introduction

Viruses continuously adapt to dynamic host environments, evolving across multiple timescales. Within individual hosts, high mutation rates and replication efficiencies enable rapid adaptation to immune pressures (Steinhauer and Holland 1987; Drake and Holland 1999; Marques et al. 2024). Over months to years, cross-species transmission events create evolutionary bottlenecks that select for variants capable of infecting new hosts (Taubenberger and Kash 2010; Singh and Yi 2021). On even longer timescales, a continuous arms race unfolds: viral adaptations drive host immune countermeasures, which in turn shape further viral diversification (Hoffmann et al. 2015; Tenthorey et al. 2022).

This perspective frames these processes as nested “Matryoshka dolls,” highlighting both established and hidden layers of viral evolution (Fig. 1). The more conspicuous “outer doll” encompasses host immune evasion, drug resistance, and host switching, often inferred from genetic changes and selection signatures. Nestled inside lies an often-overlooked “inner doll”: viruses co-evolving with their own molecular parasites, classically known as defective interfering (DI) particles (Huang and Baltimore 1977) and more recently as defective viral genomes (DVGs) (Brennan and Sun 2024). We view such particles and genomes as parasites that prey on productive virus infection. From this perspective, theories that have been developed to understand the coevolution of hosts and their parasites may be both challenged and enriched.

Fig. 1.

Matryoshka dolls of virus evolution

The external doll: viral evolution over time and across hosts

The “external doll” represents the most readily observed and better-studied layer of viral evolution. It encompasses mutations, recombinations, and selective sweeps identifiable through genome sequencing, functional assays, and comparative analyses. These events lead to shifts in host receptor binding, immune evasion, and species adaptation—features central to our current grasp of how viruses establish infections, spread, and persist.

Short-term evolution: rapid adaptation and immune evasion

Within a single infected host, viral populations are in constant flux, shaped by selection pressures that favor mutants capable of evading immune recognition. RNA viruses, in particular, exhibit astonishing adaptability due to their high mutation rates and error-prone replication mechanisms (Domingo and Perales 2019). Influenza viruses exemplify antigenic drift, where mutations in the hemagglutinin (HA) protein accumulate under immune selection, allowing evasion of pre-existing immunity (Bush et al. 1999). These antigenic changes fuel seasonal influenza epidemics and necessitate frequent vaccine updates (Hensley et al. 2009).

A similar evolutionary trajectory is observed in HIV-1, where continuous within-host selection drives viral escape from both neutralizing antibodies and cytotoxic T lymphocyte responses. As soon as the immune system mounts a targeted attack against a particular viral epitope, new mutants emerge that weaken or entirely evade immune recognition (Borrow et al. 1997; Richman et al. 2003). This relentless cycle of immune pressure and viral adaptation explains why HIV-1 is so difficult to control and why, even within a single patient, the virus can diversify into a complex quasi-species (Mansky and Temin 1995).

SARS-CoV-2, the virus responsible for the COVID-19 pandemic, further illustrates these principles on a population scale. As global immunity increased through infection and vaccination, new variants of concern emerged, each carrying mutations that confer enhanced transmissibility and immune evasion (Harvey et al. 2021). The rapid rise of Omicron, which carried multiple spike protein mutations that diminished neutralization by existing antibodies, exemplifies the strong selection pressures shaping viral evolution in real time (Cele et al. 2022).

The forces driving short-term viral evolution are relatively well understood: high mutation rates generate genetic diversity, while immune-driven selection favors variants that can escape recognition. However, viral evolution is not confined to single hosts or isolated epidemics. In some cases, a virus may face an even greater evolutionary challenge—the transition to an entirely new host species.

Host switching: evolutionary bottlenecks and adaptation to new hosts

Few events in viral evolution are as consequential as a successful host switch. When a virus jumps from one species to another, it must overcome multiple biological barriers, including differences in cell entry receptors, host immune defenses, and replication machinery (Parrish et al. 2008; Longdon et al. 2014). Most spillover events fail, but when adaptation occurs, the consequences can be profound, leading to the emergence of viral diseases in new hosts and reshaping viral evolution in the long term.

Influenza viruses provide a textbook example of host switching and viral emergence. The segmented nature of the influenza genome allows for reassortment when multiple strains infect the same host, a process that can generate entirely new viral subtypes with pandemic potential (Koçer et al. 2013). The relatively recent 2009 H1N1 pandemic strain, known as swine flu, originated from reassortment events that facilitated viral adaptation to human transmission (Nelson et al. 2008). In contrast, the H1N1 variant responsible for the 1918 pandemic was likely caused by an avian flu variant that adapted to humans (Taubenberger et al. 2005).

Similarly, SARS-CoV-2 emerged after a likely zoonotic jump from animal reservoirs, possibly bats or intermediate hosts, requiring genetic changes in its spike protein to efficiently bind human ACE2 receptors (Starr et al. 2020). Early in the pandemic, rapid evolutionary adaptations fine-tuned the virus for efficient human-to-human transmission, highlighting the powerful selection pressures at play following host switching events (Carabelli et al. 2023).

HIV-1, too, exemplifies the impact of host switching on viral evolution. The virus arose through multiple independent cross-species transmission events from nonhuman primates, each introducing distinct viral lineages into the human population (Sharp and Hahn 2011). These transmission events required substantial genetic adaptation, particularly in viral accessory proteins that counteract human immune restriction factors (Sheehy et al. 2003; Neil et al. 2008). The ability of HIV-1 to adapt to human-specific antiviral defenses, such as tetherin and APOBEC3, underscores the intense evolutionary bottlenecks viruses face when adapting to new hosts.

While vertebrate host switching has historically received the most attention, horizontal virus transmission (HVT) events are evolutionarily broader. Many plant viruses have emerged via cross-kingdom HVT involving invertebrate vectors or fungal hosts, as evidenced by phylogenomic studies of tobamoviruses and potyviruses (Dolja et al. 2020). Even more remarkably, recent work has identified prokaryote-infecting viruses with close evolutionary ties to eukaryotic Partitiviridae, suggesting rare but profound HVT across the prokaryote–eukaryote divide (Neri et al. 2022). These examples provide evidence that host switching is not merely a vertebrate phenomenon but may be a central driver of viral emergence across the entire tree of life.

Long-term coevolution: shaping viral genomes and host immune defenses

At the broadest evolutionary scale, viruses and their hosts are engaged in an ancient arms race, with each side continuously adapting in response to the other. This long-term coevolution has left indelible marks on both viral and host genomes, revealing the deep evolutionary imprint of past infections (Emerman and Malik 2010; Kaján et al. 2020).

The long-running evolutionary conflict between primates and retroviruses is well illustrated by the host gene TRIM5α, which restricts retroviral infection by recognizing and disrupting viral capsids. This gene has undergone repeated positive selection in primate lineages, reflecting past battles with ancient retroviruses—but in humans, these changes have left TRIM5α largely ineffective against HIV-1 (Sawyer et al. 2005; Goldschmidt et al. 2008). Viruses have also evolved countermeasures against other host defenses: the Vif protein in HIV-1 disables the host antiviral enzyme APOBEC3G, which would otherwise introduce mutations into the viral genome. This interaction is finely tuned—HIV-1 Vif can neutralize human APOBEC3G, but not versions from other primates, while SIV Vif proteins have adapted to target the APOBEC3G variants in their specific hosts (Bogerd et al. 2004). These examples reveal a clear pattern of co-evolution, with host and virus continually adapting to one another’s defenses and counter-defenses.

Beyond individual immune genes, the legacy of ancient viral infections is etched into host genomes. Endogenous retroviruses (ERVs), the genetic traces of long-past retroviral infections, now constitute a significant portion of mammalian DNA (Feschotte and Gilbert 2012). While many of these sequences are defunct, some have been repurposed for essential host functions, including immune regulation and development of the placenta (Dunlap et al. 2006; Katzourakis and Gifford 2010).

While external selection pressures—immune evasion, host switching, and long-term coevolution—have shaped viral evolution in profound ways, they do not tell the whole story. Within viral populations, an obscured layer of evolution is unfolding, one that does not fit neatly into the traditional framework of adaptive selection.

The internal doll: molecular parasites as evolutionary players

Long regarded as infection byproducts, DI particles and their associated DVGs may play a significant role in virus evolution. First described by von Magnus in the early 1950s, DI particles were shown to accumulate during serial-passage cultures of influenza virus, leading to periodic collapses in viral yield (von Magnus 1951)—a phenomenon now recognized as the von Magnus effect. These molecular parasites hijack the replication machinery of co-infected cells, outcompeting wild-type viruses and altering infection dynamics (Huang and Baltimore 1970; Bhat et al. 2022). DI particles carry and deliver truncated genomes that lack the functions needed for independent replication yet retain their ability to replicate and package into virus-like particles in the presence of functional viruses. Moreover, their defective virus genomes or DVGs are prevalent in natural and clinical virus isolates, they can trigger protective host immune responses, and the absence or presence of DVGs has been associated with more or less severe disease (Vignuzzi and López 2019; Bhat et al. 2022; Brennan and Sun 2024). To frame how DVGs may contribute to viral evolution, we focus on a single feature of their multifaceted nature: their ability as molecular parasites to prey on productively infected hosts.

Pathogen-host coevolution

More than 50 years ago, Van Valen introduced the Red Queen Hypothesis, arguing that species must evolve continuously just to maintain their relative fitness in response to coevolving adversaries (Van Valen 1973). Unlike traditional views of evolution as progress toward a static optimum, Van Valen’s model suggests that the biotic environment, shaped by interactions with competitors, predators, and pathogens, creates ever-changing selection pressures. So the survival of a species depends on how well it adapts compared to its co-evolving counterparts. As one species gains an advantage, others must evolve to counter it. The hypothesis predicts that evolutionary rates remain relatively constant over time because adaptation never stops; molecular evolution in pathogens and hosts reflects this continuous struggle.

While the Red Queen Hypothesis explains why evolution is an ongoing race, May and Anderson explored the specific trade-offs that shape pathogen evolution, integrating epidemiological dynamics with population genetics (May and Anderson 1983). Their study employed mathematical modeling to challenge the widely held assumption that pathogens inevitably evolve toward reduced virulence (Burnet and White 1972; Holmes 1982). They found that pathogen virulence does not necessarily decline over time; instead, the evolutionary success of a pathogen depends on the balance between virulence (how much harm it causes) and transmissibility (how efficiently it spreads). If a pathogen spreads before killing its host, high virulence can be advantageous. Further, host–pathogen interactions drive genetic diversity; pathogens apply strong selection pressure, so host populations must maintain diverse resistance alleles, preventing their complete extinction. Finally, multiple evolutionary trajectories exist. Some pathogens evolve toward lower virulence, allowing hosts to survive longer and spread the infection. But others evolve greater virulence, maximizing short-term transmission before the host dies. More broadly, the mathematical models of May and Anderson showed that the evolution of pathogens depends on their basic reproductive number (R₀), which depends on host susceptibility, pathogen adaptation, and environmental factors. This work laid the foundation for modern models of infectious disease evolution, helping predict how diseases like influenza or COVID-19 might change over time.

Microbial arms races

Bacteria and their viral predators, bacteriophages, provide a well-characterized model of pathogen-host coevolution (Chao et al. 1977; Bohannan and Lenski 1997). Mathematical models demonstrate that as bacteria evolve resistance, phages counter-adapt, fueling an ongoing arms race that drives genetic diversity and ecosystem complexity. Importantly, this coevolutionary diversification occurs even when external environmental conditions are stable, such as in a laboratory chemostat with constant resource supply (Weitz et al. 2005) . Modeling shows that the continual evolutionary battle between bacteria and phage, rather than environmental variability, leads to the emergence and coexistence of multiple bacterial and phage strains, thereby increasing ecosystem complexity. Moreover, bacteria–phage interactions involve gradual evolutionary changes, where bacteria modify their surface receptors, and phages adapt their tail fibers to regain infectivity. However, this coevolutionary process is not a simple matter of one mutation cancelling another. Instead, gradual adaptations accumulate and diverge over time, promoting the coexistence of multiple bacterial and viral strains. Rather than a single dominant genotype emerging, coevolution fosters a diverse ecosystem in which different bacteria and virus strains adapt to distinct infection and resistance strategies.

Viruses and their DI variants

While host–pathogen interactions are often framed in terms of the coevolutionary arms race between viruses and their hosts, an equally intricate evolutionary struggle may be unfolding at the molecular level within viral populations themselves. Many viruses do not evolve in isolation but instead interact dynamically with their DI particles or DVGs (defective viral genomes), molecular parasites that arise spontaneously through replication errors. This layer of coevolution challenges more established views of viral adaptation, revealing that evolutionary success for a virus is not simply about outpacing host defenses but also about managing internal parasites that can divert resources and disrupt replication.

While DIs and DVGs are among the most studied examples of molecular parasites within viral populations, they are far from unique. Diverse satellite viruses, satellite RNAs, and subviral agents such as viroids also act as viral hyperparasites, engaging in complex parasitic, commensal, or mutualistic relationships with their helper viruses and hosts (Roossinck 2011). These elements can influence replication efficiency, symptom severity, and transmission dynamics in mixed infections. For example, satellite RNAs associated with cucumber mosaic virus can attenuate or exacerbate disease symptoms depending on host genotype and environmental conditions. Similarly, hepatitis delta virus, which requires hepatitis B virus for replication and packaging, can alter disease outcomes and immune interactions in human infections (Taylor 2020). These examples suggest that molecular hyperparasitism may be a general and evolutionarily significant phenomenon across diverse viral systems.

Phage Φ6 and its DI particles

Experimental evolution studies with RNA bacteriophage Φ6 demonstrated that viral populations can be shaped by game-theoretic principles akin to the “prisoner’s dilemma,” where viral variants competing within the same host cell evolve either cooperative or selfish replication strategies (Turner and Chao 1999). At high multiplicities of infection (MOI), where multiple phages infect the same cell, defective phage variants—functionally akin to defectors emerge. These defectors or DI particles exploit the replication machinery of fully functional “cooperator” phages, gaining a fitness advantage while contributing minimally to collective replication efforts. This dynamic mirrors desertion or betrayal within a group, where individual defectors prosper short-term but ultimately weaken the overall viral population. As their frequency rises, the overall fitness of the viral population declines, illustrating an evolutionary tension between cooperation and defection (Turner and Chao 2003).

Vesicular stomatitis virus and its DI particles

DI particles also play a significant role in the evolution of vesicular stomatitis virus (VSV), an RNA virus widely used as a model for studying viral replication and evolution. DI particles of VSV are truncated viral genomes that retain sequences necessary for replication but lack essential coding regions, rendering them incapable of independent propagation. These particles emerge most frequently in high-MOI conditions, where they compete with full-length VSV genomes for replication machinery, often suppressing wild-type virus production (DePolo et al. 1987). A striking feature of VSV evolution is its ability to co-evolve with DI particles over extended passages. Experimental studies have shown that as DI particles increase in frequency, the wild-type virus responds by selecting for mutants resistant to DI interference. Resistance, conferred by mutations in the nucleocapsid and polymerase genes, accumulates in a stepwise manner, leading to virus variants that can withstand interference by hundreds of thousands of DI particles per cell. This resistance stems from altered specificity in the viral replicase complex, which can no longer efficiently replicate DI genomes—an adaptation that allows viral replication to continue despite high DI loads (Giachetti and Holland 1988). However, each wave of viral adaptation is countered by the emergence of novel DI particles, generating a perpetual, dynamic interplay. The VSV–DI system exemplifies both “Escalatory Red Queen” dynamics—wherein virus and DI particles engage in an arms race of increasingly sophisticated adaptations—and aspects of “Fluctuating Red Queen” co-evolution, as the dominant DI and viral genotypes shift over time (Brockhurst et al. 2014). This work not only illuminates the mechanics of virus evolution but also underscores how the virus’s own processes—specifically, its error-prone replication—generate DI particles that act as internal antagonists, creating a self-imposed, dynamic conflict that drives genetic diversification and perpetual evolutionary innovation.

Therapeutic interfering particles

Beyond natural examples of virus-DI particle coevolution, efforts have long sought to harness DVGs for therapeutic applications (Doyle and Holland 1973; Dimmock and Easton 2014; Baltes et al. 2017; Chaturvedi et al. 2021; Xiao et al. 2021; Pitchai et al. 2024). Inspired by the principles of DI particle interference, researchers have explored the possibility of engineering therapeutic interfering particles (TIPs) that can selectively suppress pathogenic viruses while maintaining long-term stability within a host population (Metzger et al. 2011).

HIV presents an especially compelling case for TIP-based interventions. Given the high mutation rate and immune evasion capabilities of HIV, conventional antiviral strategies struggle with the emergence of resistant strains. TIPs, however, exploit the same growth mechanisms as HIV, competing for replication and packaging resources while lacking the ability to cause disease. Recent work in nonhuman primates has demonstrated that engineered TIPs can establish stable, long-term presence in infected hosts, suppressing HIV replication and delaying disease progression (Pitchai et al. 2024). By maintaining an evolutionary arms race with HIV, TIPs may offer a novel therapeutic approach that resists viral escape and reduces overall transmission potential. Mathematical modeling suggests that TIPs could exert strong selective pressure on HIV populations, favoring variants with reduced fitness and lowering overall viral loads in infected individuals. Unlike conventional antivirals, which necessitate ongoing treatment and risk resistance, TIPs may persist as stable molecular parasites, co-evolving with HIV to sustain long-term suppression (Pitchai et al. 2024).

Pathogen–host interactions: the external and internal layers

As highlighted in Table 1, the relationships between DI particles and viruses mirror, but also significantly diverge from, the classic virus–host interaction paradigm. A defining feature of DI particles is their ability to replicate at rates comparable to, or even exceeding, their parental virus (Thompson and Yin 2010; Timm et al. 2014). This is a striking contrast to the virus-host relationship, where viral replication is orders of magnitude more productive than host cell division (Jin and Yin 2021). The comparable generation times of DI particles and viruses lead to a dynamic interplay where DI particles can rapidly amplify within an infection cycle, often outcompeting the virus for essential replication and packaging resources (Akpinar et al. 2016; Baltes et al. 2017).

Table 1.

Comparative dynamics of viruses, DI particles, and host cells

Feature	Virus vs. Host cells	DI particle vs. virus
Generation time	virus << host cell (viruses reproduce much faster than their host cells)	DI particle ≅ virus (generation times for DI particles and their viruses are similar)
Emergence	virus ≠ byproduct of cell growth (viruses are not a byproduct of host cell growth)	DI particle = byproduct of virus growth (DI particles are byproducts of virus growth)
Mutation rate	virus >> cell (mutation rates are higher for viruses than host cells)	DI particle ≅ virus (mutation rates are similar for DI particles and their viruses)
Evolutionary potential	viruses (large) (viruses readily adapt their gene functions)	DI particles (Astronomical) (DI particles readily innovate and adapt their gene functions)

A critical difference between viruses and DI particles lies in their origins. Viruses can propagate independently of host replication, while DI particles are obligate byproducts of viral replication—emerging from replication errors, internal deletions, or recombination events within the viral genome (Lazzarini et al. 1981; Aguilar Rangel et al. 2023). Though they depend on co-infection with a functional virus, DI particles can proliferate aggressively under favorable conditions.

Because DI particles share the same error-prone replication machinery as their parental virus, their mutation rates are similarly high—approaching the upper limit for RNA genome stability ​ (Drake 1993; Drake and Holland 1999). Yet DI particles possess an additional advantage: freedom from the evolutionary constraints of maintaining complete coding sequences. This freedom allows them to explore a vastly expanded mutational space. For instance, a 10,000-nucleotide viral genome might support around 30,000 distinct single-nucleotide substitutions, but it could harbor over 50 million possible single deletion mutations—a mutational landscape that is more than 1000-fold larger (Aguilar Rangel et al. 2023).

This mutational breadth can enable remarkable functional innovation. DI genomes frequently evolve enhanced replication signals, more efficient packaging elements, and novel gene products that modulate host responses (Patel and Elliott 1992; Boergeling et al. 2015; Imamichi et al. 2016; Girgis et al. 2022; Ghorbani et al. 2023; Ranum et al. 2024). These adaptations reveal DI particles not merely as defective byproducts but as potent evolutionary players that can actively reshape viral population dynamics.

That said, viruses themselves are no strangers to innovation. Through point mutation, recombination, and horizontal gene transfer, viruses can acquire and repurpose genetic material in ways that yield new molecular functions—sometimes co-opting host genes (Irwin et al. 2022). The contrast, then, is not about which entity can innovate, but how they do so: DI particles exploit structural rearrangement and relaxed constraints, while viruses operate through gene acquisition and functional refinement. Both internal and external layers of viral evolution are rich with creativity—driven by different rules but converging in their capacity to invent.

Conclusion

The “Matryoshka dolls” framework unifies external viral adaptations with the underappreciated role of DI particles in shaping evolution. Recognizing these nested layers may inspire next-generation antiviral strategies and deepen our understanding of viral evolution in dynamic host environments.

Funding

I am grateful for support by the US National Science Foundation (DMS-2151959, MCB-2029281, and CBET-2030750), the National Institutes of Health (R01DK133605), and the University of Wisconsin-Madison: Chemical and Biological Engineering, Wisconsin Institute for Discovery, the Office of the Vice-Chancellor for Research, and the Wisconsin Alumni Research Foundation.