Rapid emergence of virus–host mutualism under stress

Broad Spectrum of Outcomes in Virus–Host Interactions

The article by González et al. (1) in PNAS illuminates the flexibility of the virus–host relationships by demonstrating a rapid shift to mutualism under stress in an RNA plant virus model. A common public perception of a virus, particularly relevant in the time of the COVID-19, is that of a pathogenic, often deadly parasite. The obligate intracellular parasitism of viruses rightly made it into the textbook virus definitions. However, such a human health-centric perspective on the vast and extremely diverse world of viruses turns out to be short-sighted due to the dramatic acceleration in understanding the broader picture of virus–host interactions over the past decade or so. A major driver of this process is inexpensive, rapid sequencing technologies that spawned metaviromics—discovery of unsuspected multitudes of viruses in organisms’ holobionts and various environments. In particular, the number of known RNA viruses identified in protists, animals, or complex aquatic systems has increased severalfold (2–4). The greatly expanded knowledge of RNA virus diversity enabled the development of a comprehensive evolution-based taxonomy system that culminated in the unification of RNA viruses into the kingdom Orthornavirae (5, 6).

The big wave of new virus discovery in plants brought at least two remarkable outcomes. First, the evolutionary connections between the plant virome and the vast virome of invertebrates that apparently gave rise to most of the plant RNA virus lineages via horizontal virus transfer have transpired (7). A second outcome was a major reassessment of virus–plant relationships. It was found that the virome of wild plants is dominated by cryptic, persistent viruses that do not cause any apparent disease (8, 9).

Another major beneficiary of the facile nucleotide sequencing is the field of experimental RNA virus evolution. Due to the small size of RNA virus genomes and the remarkably low fidelity of RNA-dependent RNA polymerases (RdRPs) that replicate these genomes, it became possible to trace pathways of virus evolution by sequencing entire genome populations within an experimentally amenable time frame (10). In plant models of RNA virus microevolution, an important current direction is investigation of the roles of environment and stress in virus–plant interactions (11, 12). This research was not in small part stimulated by the discovery of an amusing virus–fungus–plant system in which a virus reduces fungal virulence and as a consequence endows host plants with salt stress tolerance (13). González et al. provide an impactful insight into this field by using the experimental evolution approach to show that drought stress modulates the path of virus–plant coadaptation, rendering virus from a disease-causing parasite to a mutualist that increases the plant tolerance to this abiotic stress.

Adaptability of Virus Genotype and Infection Phenotype

An imaginative experimental setting of this work involved the double whammy of switching virus host under strong selection pressure due to abiotic stress. More specifically, four distinct ecotypes of the model plant Arabidopsis thaliana that responded to turnip mosaic virus (TuMV) (Potyviridae) either with severe disease symptoms and strong induction of pathogen defense genes (dubbed group 1) or with mild symptoms accompanied by overexpression of abiotic stress response genes (group 2) were used as virus hosts. The A. thaliana-naive TuMV was passaged five times on each ecotype of the host plants either under normal watering or under severe drought conditions, and the resulting virus lineages were genotyped and analyzed for their infection phenotypes. The extent of TuMV adaptation was evaluated as an integrated measure of virus infectivity and speed of symptom development relative to the ancestral virus. This design allowed asking key questions on the general routes of virus–host coevolution under standard or abiotic stress conditions.

The answers to these questions appear to be rather nontrivial. Although all passaged virus lineages showed substantial adaptation to A. thaliana, the extent of adaptation was dramatically different between the group 1 and group 2 ecotypes, as well as between standard and drought conditions. In group 1 lines, virus adaptation levels were higher under standard conditions, whereas in group 2 lines, the opposite was observed: The virus adapted to the new host better under drought conditions, attesting to crucial contributions of the host genotype to virus–host coevolution.

Comparisons of the genotypes of the ancestral viruses to those evolved under standard or stress conditions revealed similar numbers of mutations (∼30), most of which happen to be nonsynonymous substitutions. Thus, as could be expected, adaptation to the new host involved positive selection on the evolving virus. Strikingly, the distribution of these substitutions along the ∼10-kb TuMV genome was highly nonrandom, with most concentrated within a short stretch of ∼40 nucleotides within the sequence encoding VPg (virus protein, genome-linked). In many picornavirus-like, positive-strand RNA viruses (orders Picornavirales and Patatavirales within phylum Pisuviricota), VPg acts as a protein primer for the initiation of RNA synthesis by the virus RdRP (14). In addition, VPg contributes to virus transport within plans and suppression of antiviral RNA interference response (15). However, multifunctionality is common for all potyvirus proteins leaving the question “why VPg?” to hang in midair. Potential answers might be that VPg indeed forms the most relevant hub in virus–plant–environment interaction networks or, perhaps, more likely, that other potyvirus proteins could be targeted under distinct experimental evolution designs.

Plant Response to Double Challenge by Drought and Virus

To get insight into the plant responses accompanying virus adaptation with or without abiotic stress, the authors profiled the whole-genome transcriptomes and identified the differentially expressed genes (DEGs). The sheer number of DEGs was significantly higher in group 2 than in group 1 accessions suggestive of a greater plasticity of the gene regulation network in plants less severely affected by virus infection.

Functional profiling of the DEGs in the group 1 accession showed significant down-regulation of genes involved in circadian clock although the profiles of the two group 1 accessions shared only a few such genes. Given a broad array of group 1 DEGs that are not known to contribute to the circadian clock, this overall functional signal appears to be relatively weak but nevertheless is biologically relevant because circadian clock is known to contribute to plant stress and infection responses (16). Another potentially interesting DIG category in group 1 accessions were down-regulated genes involved in the nucleocytoplasmic transport, also a category previously associated with drought response. However, because nucleocytoplasmic transport is also involved in potyvirus infection (two virus proteins, NIa, the main protease, and NIb, the RdRP, are transported to the nucleus where they form inclusions) (15), it cannot be ruled out that reduction of this transport could modulate the virus–host interaction network.

Analysis of the two group 2 accessions showed no similarity in the functional profiles of enriched or depleted DEG categories, implying, unsurprisingly, that there are multiple pathways across the plant fitness adaptation landscape. As would be expected, investigation of the plant hormonal profiles showed an increase in salicylic acid levels in virus-infected plants, in agreement with the critical role of this hormone in systemic plant defense against viruses (17). Several other plant hormone profiles showed varying degrees of differential accumulation among the used lines and across experimental conditions. Given the well-known complexity of the hormonal regulation networks that shape plant development and responses to biotic and abiotic stresses, the observed hormonal profile changes were difficult to reduce to a clearcut functional interpretation.

The article by González et al. in PNAS illuminates the flexibility of the virus–host relationships by demonstrating a rapid shift to mutualism under stress in an RNA plant virus model.

Some Virus–Plant Combinations Like It Dry Better than Others

The final and most striking outcome of this expansive study is that infection caused by drought-evolved viruses substantially enhanced plant survival under severe drought conditions, clearly attesting to a shift from parasitic to mutualistic virus–host relationships. Indeed, it was demonstrated that mean drought survival probabilities of the plants infected with drought-evolved TuMV isolates were significantly higher than those of uninfected plants. This result held true for three out of four tested A. thaliana ecotypes. The only standout ecotype was doomed under drought conditions independent of its infection status, obviously, due to a genotype that is unfit to face water stress. Further analysis of virus adaptation specificity demonstrated that drought stress strongly favored evolution of generalist virus lineages that increased drought tolerance of distinct plant ecotypes compared to more specialist lineages that evolved under normal conditions.

Not entirely unexpectedly, the extensive investigation of the changes in virus genotypes and host plant gene expression profiles fails to provide a mechanistic explanation of how the viruses that evolved under drought conditions help plants to survive severe abiotic stress. On the virus side, preferential accumulation of mutations within VPg coding sequence suggests a potential role of this multifunctional protein in changing the outcomes of virus–plant interactions. However, experimental validation of such a role, for example, comprehensive functional characterization of the interactions between VPg and host proteins for the ancestral and drought-evolved VPg variants, seems to be prohibitively challenging.

On the host side, interpretations of the whole-genome transcriptome analyses are inherently difficult due to the sheer numbers of DEGs that belong to a variety of either known or yet-unknown functional categories. Such analyses often end up in preferential discussion of the DEG functionalities that seem to be the most relevant, such as circadian clock or nuclear transport genes singled out in this work. The solution to such bias would require the ability to model and interpret the entire system of the organisms’ gene regulation networks under varying environmental and biotic stress conditions, a remote possibility even with the arrival of the current machine learning and artificial intelligence capabilities. However, what once again transpired from this study is the uncanny phenotypic plasticity of the plant networks of gene expression regulation and hormone signaling in a double-stress (infection and drought) selection pressure system.

Despite the understandable limitations in uncovering mechanisms of virus–plant coadaptation, the paper by González et al. provides a captivating biological story of a remarkable opportunism in virus microevolution pathways in response to changing host plant genotype and drought stress. The take-home message of virus rapidly switching from parasitic to mutualistic mode of virus–host interaction is strong and clear.

An extraordinary flexibility in virus–host interactions described in this work is by no means limited to plant viruses, but rather resonates with the recent progress in human virology (18, 19). Deeper insight into the virome of healthy humans reveals a continuum of outcomes of virus presence in the human body (20). Many human viruses, such as the ubiquitous anneloviruses, seem to be commensals causing no apparent disease (21). Other viruses (e.g., herpes simplex viruses 1 and 2) alternate between latency and disease, in response to conditions that vary from cold or heat to UV exposure to emotional stress. Related herpesviruses, such as Epstein–Barr virus and human cytomegalovirus, cause human population-wide, lifelong, often unapparent infections that help in shaping human immune profiles and protect from heterologous virus infections (22). Furthermore, slow infections by commensal papillomaviruses provide broad protection from skin cancers (23). It can be expected that the ever deeper understanding of the intricate virus–host coevolution routes in diverse model systems, from plants to humans, could eventually lead to a better, less chaotic management of the future virus pandemics.