Antagonistic within‐host interactions between plant viruses: molecular basis and impact on viral and host fitness

Summary

Double infections of related or unrelated viruses frequently occur in single plants, the viral agents being inoculated into the host plant simultaneously (co‐infection) or sequentially (super‐infection). Plants attacked by viruses activate sophisticated defence pathways which operate at different levels, often at significant fitness costs, resulting in yield reduction in crop plants. The occurrence and severity of the negative effects depend on the type of within‐host interaction between the infecting viruses. Unrelated viruses generally interact with each other in a synergistic manner, whereas interactions between related viruses are mostly antagonistic. These can incur substantial fitness costs to one or both of the competitors. A relatively well‐known antagonistic interaction is cross‐protection, also referred to as super‐infection exclusion. This type of interaction occurs when a previous infection with one virus prevents or interferes with subsequent infection by a homologous second virus. The current knowledge on why and how one virus variant excludes or restricts another is scant. Super‐infection exclusion between viruses has predominantly been attributed to the induction of RNA silencing, which is a major antiviral defence mechanism in plants. There are, however, presumptions that various mechanisms are involved in this phenomenon. This review outlines the current state of knowledge concerning the molecular mechanisms behind antagonistic interactions between plant viruses. Harmful or beneficial effects of these interactions on viral and host plant fitness are also characterized. Moreover, the review briefly outlines the past and present attempts to utilize antagonistic interactions among viruses to protect crop plants against destructive diseases.

Introduction

Plants, being sessile organisms, are continuously exposed to multiple simultaneous abiotic and biotic stresses that may directly impose restrictions on their growth or cause metabolic dysfunction (Atkinson and Urwin, 2012; Rejeb et al., 2014; Suzuki et al., 2014; Tanaka et al., 2014; Wasik and Turner, 2013). Abiotic stress factors, such as heat, cold, drought, salinity and nutrient stress, can substantially reduce the yields of major crop plants (Wang et al., 2003). Likewise, biotic stresses are known to considerably reduce crop yield and/or quality. Major biotic stresses include diseases induced by plant‐pathogenic bacteria, fungi, viruses and nematodes, and damage caused by herbivorous insects (Brown and Hovmoller, 2002; Mordecai, 2011; Reitz et al., 2015). Under natural conditions, plants are exposed to a combination of abiotic and biotic stresses (Eastburn et al., 2011; Suzuki et al., 2014). The expression of plant disease is the outcome of a three‐way interaction between a host, a virulent pathogen and the environment, known as the disease triangle (Eastburn et al., 2011; Scholthof, 2007).

Plants respond to abiotic and biotic stresses by employing versatile and intricate defence systems, often operating at significant defence costs (Bode and Kessler, 2012; Mine et al., 2014). The knowledge of the complex interactions between pathogen infection strategies and host plant defence mechanisms is poor (Dodds and Thrall, 2009) and derives from studies performed on a single pathogen clone infecting an isogenic host population, mostly under controlled experimental conditions (Pieterse et al., 2012). In nature, however, plants must defend themselves against simultaneous attacks of multiple pathogens and pests (Choisy and de Roode, 2010; Petek et al., 2014; Tack and Dicke, 2013). Among plant pathogens, RNA viruses have the highest evolutionary potential, combining large population sizes, fast and often error‐prone replication and high mutation rates (Belshaw et al., 2008; Duffy et al., 2008; Elena and Sanjuán, 2008; Koonin and Doolja, 2012, 2014; Makeyev and Bamford, 2004). As a result of this potential, many plant viruses are generalists and infect plants in both agricultural and wild populations (Alexander et al., 2014; Elena et al., 2009; Roossinck, 2010), quite often sharing the same hosts. Consequently, an individual plant can become infected with more than one virus.

Mixed infections of two related or unrelated viral pathogens frequently occur in single plants of cultivated or wild hosts (Choisy and de Roode, 2010; Malpica et al., 2006; Naidu et al., 2014, 2015; Tatineni et al., 2014; Valverde et al., 2007; Zhan and McDonald, 2013). They mostly occur following transmission by hemipteran vectors, many of which have evolved the ability to transmit more than one virus. Importantly, many invertebrate and even fungal vectors are able to simultaneously transmit two or more viruses (Syller, 2014; Syller and Grupa, 2014), thus directly generating mixed infections of new host plants (Syller, 2014). Subsequently, these plants become potential sources of the viruses, thereby increasing the chance of spread of diseases caused by these pathogens (Syller, 2014). Dual infection can be classified as either co‐infection or super‐infection (Miralles et al., 2001; Saldaña et al., 2003). Co‐infection is a term mostly used to refer to a situation in which two viruses invade the host simultaneously or in a short interval of time. In super‐infection, the host that has previously been infected by one virus is re‐infected with a different virus strain or isolate, or a distinct virus, at a later point in time (Carrillo et al., 2007; Kell et al., 2013; Saldaña et al., 2003).

Diseases caused by mixed viral infections result in a decline in plant vigour and reduced yields (Valverde et al., 2007; Wintermantel et al., 2008). Among them, grapevine leafroll disease (GLD) caused by Grapevine leafroll‐associated viruses (GLRaVs) has attracted the attention of virologists because of both its economical importance and intriguing and highly complex aetiology (Naidu et al., 2014, 2015). The severity of the negative effects depends on the type of within‐host interaction between the infecting viruses. Unrelated viruses generally interact with each other in a synergistic manner, which has a facilitative effect on one or both of the viral partners, manifested by higher accumulation of the beneficiary virus(es) in the host plant, and more severe symptoms than those induced by either virus alone (Syller, 2012; Tatineni et al., 2014). Numerous synergistic interactions have been described, the best characterized of which is that involving Potato virus Y (PVY) and Potato virus X (PVX) in tobacco plants (Rochow and Ross, 1955; Vance, 1991). Synergistic interactions are likely to result from suppression of the host defence mechanism based on RNA silencing by viral proteins (Carrington et al., 2001; Ratcliff et al., 1999). However, there are still some intriguing questions awaiting answers. For example, how does viral synergism alter host defence mechanisms and how does co‐infection by two unrelated viruses cause more severe symptoms than those produced by either virus alone (Mandadi and Scholthof, 2012; Tatineni et al., 2014)?

Unlike unrelated plant viruses, mutual relationships between related viruses are mostly antagonistic (competitive) (reviewed in Syller, 2012). These interactions have been less well documented than synergistic interactions, and insufficient research has been performed to elucidate their molecular basis. The relatively well‐known antagonistic interaction is cross‐protection, also referred to as homologous interference or super‐infection exclusion (SIE) (e.g. Bergua et al., 2014; Folimonova, 2012, 2013; Gutiérrez et al., 2012; Julve et al., 2013). This type of interaction occurs when a previous infection with one (primary/protecting) virus prevents or interferes with subsequent infection by a homologous secondary/challenge virus (DaPalma et al., 2010; Gal‐On and Shiboleth, 2006; González‐Jara et al., 2009; Ziebell and Carr, 2010). In this review, the term ‘super‐infection exclusion’ is used as it closely corresponds to the term ‘super‐infection’ and more accurately describes this phenomenon.

Not much is known about the effects of antagonistic interactions on viral and host plant fitness. In many cases, the performance of pathogens competing against one another cannot be predicted by their performance as single infections (Koskella et al., 2006). This article outlines the current state of knowledge concerning the molecular mechanisms underlying antagonistic interactions between plant viruses. Harmful or beneficial effects of these interactions on viral and host plant fitness are also characterized. Moreover, the review briefly outlines the past and current attempts to utilize antagonistic interactions among viruses to protect crop plants against destructive diseases.

General Outlook on Phenomenon of Competition Among Plant Viruses

Populations of RNA viruses are greatly diverse because of the low steadiness of their replication process (Ojosnegros et al., 2012). The intra‐host sets of viral strains, often referred to as quasispecies, consist of clouds of genetically related, but non‐identical, viral mutant types. The composition of a quasispecies is largely shaped by the competitive fitness of its individual viruses, and quasispecies diversity is maintained by mutation–selection balance (Domingo et al., 2012; Ojosnegros et al., 2012). Within‐host multiplication and between‐host transmission are considered to be the major components of viral fitness (Sacristán and García‐Arenal, 2008). It has been speculated that the former is linked to parasite virulence. However, high virulence resulting in a high cell‐killing rate, and subsequently in high host mortality and morbidity, will negatively affect between‐host transmission, at the same time decreasing the proportion of highly virulent parasite variants in the population. Paradoxically, it will negatively affect a virus fate because a healthier host is a better environment for a virus (Roossinck, 2005).

In mixed infections, the fitness of each of the viruses depends not only on its own ability to adapt to the host, but also on the activity (synergistic or antagonistic) of its counterpart(s) (Elena et al., 2014; Martin and Elena, 2009). Competitive interactions between parasite genomes co‐infecting cells are normally strong during parasite life stages that multiply rapidly and are highly capable of exploring host resources (Karvonen et al., 2012). Strong competition for host resources may result in a decreased rate of parasite reproduction, leading to decreased fitness, in comparison with that in a single infection (González‐Jara et al., 2009, 2013; Gutiérrez et al., 2010; Karvonen et al., 2012). An alternative effect of competition may be an enhanced rate of host exploitation and parasite multiplication (Karvonen et al., 2012), which is evidence of its increased fitness.

The basic parameter for the evaluation of the kinetics and progress of multiple infections is the multiplicity of infection (MOI), i.e. the number of virus genomes infecting a cell (González‐Jara et al., 2009, 2013; Gutiérrez et al., 2010, 2012; Zwart et al., 2013). MOI is of great importance from the standpoint of virus evolution because recombination, reassortment or complementation between different genotypes can only occur in mixed‐genotype‐infected cells, i.e. when the MOI is above unity (Gutiérrez et al., 2010; Zwart et al., 2013). However, despite its generally accepted importance, MOI has rarely been estimated for plant viruses, each time employing two engineered variants of the same virus under cell culture conditions (discussed in Zwart et al., 2013). As postulated, for the estimation of in vivo MOI in multicellular organisms, it is essential to replace the ‘cell culture’ conceptual model with one in which the effects of spatial processes occurring during viral expansion will be evaluated at higher levels. It should be taken into account that, during the colonization of a multicellular host by viruses, MOI may change considerably in different organs and at different infection stages (Gutiérrez et al., 2015). Indeed, during studies on the infection of turnip plants by the potyvirus Turnip mosaic virus (TuMV), (Gutiérrez et al., 2015) observed that MOI could depend largely on the route of cell infection in a systemically infected leaf. MOI is usually one genome per cell when cells are infected by virus particles moving long distances in the vasculature, whereas it is much higher during subsequent cell‐to‐cell transport in mesophyll. However, a rapidly established SIE prevents cell co‐infection by merging populations originating from different primary foci within a leaf. This complex process of colonization results in a situation in which within‐cell interactions occur almost exclusively among closely related viral variants. This can shed some light on the phenomenon of genotype spatial separation in infected plants (Gutiérrez et al., 2015).

Viral strains or closely related viruses invading the same host plant generally tend not to colonize cells already infected by their counterpart, consequently occupying separate niches. Spatial separation, also termed spatial exclusion, is a spectacular but poorly recognized consequence of a competitive interaction. This phenomenon is characteristic of both super‐infection and co‐infection, but different intracellular events seem to occur in each case. In the first instance, a virus entering cells of a so far uninfected host can freely replicate, move and finally colonize the host. It gains both a territorial and numerical advantage over a potential incoming competitor, thus becoming able to protect the host by excluding a super‐infecting virus. SIE is believed to occur even if the primary virus is weaker than its counterpart, which allows the first virus to maximize the production of progeny particles, once the cell has been infected (Beperet et al., 2014). The phenomenon has been studied in a number of human, animal and plant viruses (e.g. Bergua et al., 2014; Claus et al., 2007; Lee et al., 2005; Ramírez et al., 2010; Webster et al., 2013).

A more complex and thus unpredictable situation is likely to arise when two closely related viruses co‐infect the host, i.e. enter its cells at the same time. Here, various scenarios are possible because a cascade of molecular events associated with infections induced by the competing viruses can be triggered. However, all scenarios seem to lead to spatial separation between populations of the two viruses. In several studies, two related plant viruses or different variants of the same species have been reported to colonize separate cell clusters, thus exhibiting mutual exclusion (ME) (reviewed in Syller, 2012). It is worth recalling that spatial separation was observed in epidermal cells when Nicotiana benthamiana plants were doubly inoculated with cDNA clones of the potyviruses Plum pox virus (PPV), Tobacco vein mottling virus (TVMV) and Clover yellow vein virus (ClYVV) expressing green fluorescent protein (GFP) and red fluorescent protein (RFP), or with identical but differently labelled potyviruses (e.g. PPV‐GFP and PPV‐RFP). Both fluorescence signals were only detected in some cells at the border separating different coloured cell clusters. Spatial separation has also been reported for two populations of Apple latent spherical virus (ALSV) expressing yellow vs. cyan fluorescent proteins (YFP vs. CFP) in Chenopodium quinoa plants. Interestingly, differently labelled viral populations were separated from each other in both inoculated and young non‐inoculated leaves.

From an evolutionary standpoint, SIE can prove to be beneficial for newly emerging viral variants by favouring their entry into uninfected rather than already infected host cells, thereby promoting virus dissemination (Bergua et al., 2014; Syller, 2012). However, spatial separation between genetic variants sharing the same host reduces the opportunities for competition, thus reducing its beneficial effects on the overall fitness of a virus population (Elena et al., 2011). SIE has also been suggested to play a role in maintaining the stability of viral sequences by reducing the opportunity for recombination events between homologous viruses. If, hypothetically, homologous viruses were able to successfully super‐infect cells, recombination events would greatly increase virus variability (Bergua et al., 2014; Elena et al., 2011). This increased variability could, in turn, considerably hamper the development of efficient and stable strategies to protect humans, animals and plants from viral diseases, or to reduce their harmful effects (Bergua et al., 2014; Blackard et al., 2002; Syller, 2012).

Molecular Mechanisms of Antagonistic Interactions Between Viruses

Despite the importance of SIE and ME in viral pathogenesis, the knowledge on why and how one virus variant excludes another is scarce (Bergua et al., 2014; Folimonova, 2012; Natsuaki, 2011). For a better understanding of possible, although still hypothetical, mechanisms of these phenomena, we highlight briefly the molecular aspects of viral infection and the ensuing plant response. As it is beyond the scope of this article to characterize in detail multi‐layered plant defence strategies, readers are referred to comprehensive reviews on this topic (Coll et al., 2011; De Ronde et al., 2014; Dodds and Rathjen, 2010; Ghoshal and Sanfaçon, 2015; Hou et al., 2009; Huot et al., 2014; Mandadi and Scholthof, 2013; Molnar et al., 2011; Muthamilarasan and Prasad, 2013; Pallas and García, 2011; Pieterse and Van Wees, 2015).

When a virus infects a host cell, viral double‐stranded RNA (dsRNA) is synthesized by viral RNA‐dependent RNA polymerases (RdRps). From the initially infected epidermal or mesophyll cell(s), the virus moves to the neighbouring cells (short‐distance transport) through plasmodesmata (PD) (Niehl and Heinlein, 2010), and systemically (long‐distance transport) through vascular tissue, usually the phloem sieve tubes (Hull, 2014). The primary cell‐to‐cell transport stage does not essentially relate to phloem‐limited viruses, which are injected directly into the phloem by their vectors. The young leaves of the plant become infected following subsequent cell‐to‐cell movement. Virus movement is assisted by viral movement proteins (MPs) and other virus‐encoded proteins, which interact with one another and with plant components. Plants can counteract viral infection in its early stages. The recognition of viral dsRNA molecules by the plant defence system leads to the induction of RNA silencing, called RNA interference (RNAi), by targeting viral RNA for degradation (Muthamilarasan and Prasad, 2013; Nicaise, 2014) (Fig. 1). RNAi is a sequence‐specific gene regulatory pathway that provides a natural cellular response to viral infection in most eukaryotes. It involves the activities of a number of conserved proteins, such as the Dicer‐like protein (DCL), double‐stranded RNA‐binding proteins (DRBPs), Argonaute proteins (AGOs), which assemble in RNA‐induced silencing complexes (RISCs), RdRp and RNA helicase (Pallas and García, 2011; Peláez and Sanchez, 2013). In the first step of the RNA silencing pathway, viral dsRNA triggers the formation of virus‐derived small (or short) interfering RNAs (vsiRNA, 21–23‐nucleotide dsRNA) by DCLs. The vsiRNAs are loaded into AGOs and lead RISC complexes to degrade viral single‐stranded RNA (ssRNA) and/or to inhibit its translation (Ma et al., 2015; Pallas and García, 2011; Rana, 2007; Zhu and Guo, 2012). RNA silencing moves between cells over short and long distances (Molnar et al., 2011; Nicaise, 2014; Pyott and Molnar, 2015).

Figure 1.

A simplified model of the RNA silencing pathway (for abbreviations, see text and Table 1). When a virus, designated here as a primary virus (blue), enters a susceptible plant cell, its dsRNA molecules are recognized by specific cell receptors. The recognition leads to the induction of RNA silencing by targeting viral RNA for degradation. The silencing process involves the activities of different proteins, including DCLs, AGOs that assemble in RISCs, and RdRp. The RNA silencing pathway is initiated by the DCL‐mediated cleavage of viral dsRNA into vsiRNAs, which are loaded into AGOs to constitute the RISCs and to guide them to recognize complementary viral RNA sequences (secondary virus, red) for degradation. The silencing signal moves initially into adjacent cells through plasmodesmata, and from leaf to leaf through the plant vasculature.

Table 1.

List of viruses addressed and acronyms or abbreviations used in this article

Virus acronym	Species name	Genus	Family
ALSV	Apple latent spherical virus	Cheravirus	Secoviridae
BYMV	Bean yellow mosaic virus	Potyvirus	Potyviridae
ClYVV	Clover yellow vein virus	Potyvirus	Potyviridae
CMV	Cucumber mosaic virus	Cucumovirus	Bromoviridae
CTV	Citrus tristeza virus	Closterovirus	Closteroviridae
GLRaVs	Grapevine leafroll‐associated viruses (classified in three genera in the Closteroviridae family)	Ampelovirus Closterovirus Velarivirus	Closteroviridae
MCDV	Maize chlorotic dwarf virus	Waikavirus	Secoviridae
ORMV	Oilseed rape mosaic virus	Tobamovirus	Virgaviridae
PPV	Plum pox virus	Potyvirus	Potyviridae
PVM	Potato virus M	Carlavirus	Betaflexiviridae
PRSV	Papaya ringspot virus	Potyvirus	Potyviridae
PVX	Potato virus X	Potexvirus	Flexiviridae
PVY	Potato virus Y	Potyvirus	Potyviridae
SMV	Soybean mosaic virus	Potyvirus	Potyviridae
TBSV	Tomato bushy stunt virus	Tombusvirus	Tombusviridae
TICV	Tomato infectious chlorosis virus	Crinivirus	Closteroviridae
TMGMV	Tobacco mild green mosaic virus	Tobamovirus	Virgaviridae
TMV	Tobacco mosaic virus	Tobamovirus	Virgaviridae
ToCV	Tomato chlorosis virus	Crinivirus	Closteroviridae
ToMV	Tomato mosaic virus	Tobamovirus	Virgaviridae
ToRMV	Tomato rugose mosaic virus	Begomovirus	Geminiviridae
ToYSV	Tomato yellow spot virus	Begomovirus	Geminiviridae
TuMV	Turnip mosaic virus	Potyvirus	Potyviridae
TVMV	Tobacco vein mottling virus	Potyvirus	Potyviridae
ZYMV	Zucchini yellow mosaic virus	Potyvirus	Potyviridae

Abbreviations
AGO	Argonaute protein
Avr	Avirulence factor
CFP	Cyan fluorescent protein
CP	Coat protein
DCL	Dicer‐like protein
DRBP	Double‐stranded RNA‐binding protein
dsRNA	Double‐stranded RNA
ET	Ethylene
GFP	Green fluorescent protein
HR	Hypersensitive response
JA	Jasmonic acid
ME	Mutual exclusion
MOI	Multiplicity of infection
MP	Movement protein
PR (protein)	Pathogenesis‐related (protein)
RdRp	RNA‐dependent RNA polymerase
RFP	Red fluorescent protein
RISC	RNA‐induced silencing complex
RNAi	RNA interference
ssRNA	Single‐stranded RNA
SA	Salicylic acid
SAR	Systemic acquired resistance
SIE	Super‐infection exclusion
YFP	Yellow fluorescent protein
VIGS	Virus‐induced gene silencing
vsiRNA	Viral small interfering RNA

To counteract host RNAi‐based antiviral immunity, plant viruses encode RNA silencing suppressors, which interfere with a specific part of the RNA silencing pathway and thereby reduce its effectiveness against plant viruses (Burgyán and Havelda, 2011; De Ronde et al., 2014; Nicaise, 2014; Pumplin and Voinnet, 2013; Qu and Morris, 2005; Shimura and Pantaleo, 2011).

In specific cases, plants have evolved intracellular resistance (R) proteins recognizing a pathogen‐encoded avirulence factor (Avr), leading to effector‐triggered immunity (incompatible interaction), which results, among others, in the accumulation of pathogenesis‐related (PR) proteins, and often elicits programmed cell death, known as the hypersensitive response (HR) (Coll et al., 2011; De Ronde et al., 2014; Huot et al., 2014; Mandadi and Scholthof, 2012; Nicaise, 2014; Pumplin and Voinnet, 2013; Soosaar et al., 2005; Zhu et al., 2014; Zvereva and Pooggin, 2012). The HR confines the pathogen to the inoculated area, which results in the formation of necrotic lesions on the inoculated leaf, and thus potential propagation of the pathogen through the whole plant is restricted. However, unlike the resistant (incompatible) virus–host interactions, those in susceptible (compatible) hosts do not trigger HR, but produce a similar, but distinct, form of necrosis, called systemic necrosis (Mandadi and Scholthof, 2012). This form of response has been reported, for example, in young Nicotiana rustica plants inoculated with Tobacco mosaic virus (TMV), N. benthamiana and N. rustica plants infected with Tomato bushy stunt virus (TBSV), and tomato plants infected with Cucumber mosaic virus (CMV) and satellite RNA‐D (Mandadi and Scholthof, 2012; and references therein). It has also been demonstrated for young Datura metel plants inoculated with Potato virus M (PVM) (Grupa and Syller, 2015) (Fig. 2). Unlike HR‐associated necrosis, systemic necrosis primarily develops in the upper non‐inoculated leaves. Another difference is that systemic necrosis is a lethal response that often leads to premature plant death (Fig. 2). Moreover, systemic necrosis is considered not to impede virus multiplication or its systemic movement (Mandadi and Scholthof, 2012).

Figure 2.

Response of D atura metel plants to inoculation with a severe isolate of P otato virus  M (PVM) alone (right) and to sequential inoculation with mild and severe isolates (left). The plant singly infected with the severe isolate showed systemic stem and leaf necrosis, followed by detachment of the inoculated leaves, wilting and withering, resulting in plant death. Pre‐inoculation of the plant with the mild PVM isolate protected it against the destructive effects of infection with the severe isolate. A remarkable recovery of the plant is associated with RNA‐mediated exclusion of super‐infection with the severe isolate.

Subsequent to the HR, systemic acquired resistance (SAR) is induced in distant uninfected tissue (De Ronde et al., 2014; Mandadi and Scholthof, 2012; Nicaise, 2014; Oka et al., 2013; Soosaar et al., 2005; Vallad and Goodman, 2004). In contrast with HR, SAR is a long‐lasting immune response aimed to induce resistance against subsequent infections in distant tissue (Mandadi and Scholthof, 2012). The exact mechanisms of SAR are still obscure, but it is considered to be initiated through a local interaction among Avr and R proteins, and is accompanied by the accumulation of phytohormones, such as salicylic acid (SA), jasmonic acid (JA) and ethylene (ET) (Mandadi and Scholthof, 2012; Soosaar et al., 2005) (Fig. 3). A widely recognized feature of the SA and JA signalling pathways is their mutual antagonism (Alazem and Lin, 2015; Takahashi et al., 2004); it is likely that this regulatory cross‐talk has evolved to enable plants to fine‐tune the induction of their defences against different plant pathogens (Kunkel and Brooks, 2002). There are, however, exceptions to this general rule. It has been suggested recently that, in Nicotiana tabacum, which possesses the N resistance gene to TMV, JA signalling is not directly responsible for susceptibility to TMV, but is indirectly responsible for viral resistance through the partial inhibition of SA‐mediated resistance governed by the N gene, and that appropriate JA and SA levels are essential to establish the degree of resistance (Oka et al., 2013). More recently, the essential roles of SA and JA in systemic resistance against TMV have also been demonstrated in N. benthamiana (Zhu et al., 2014). Silencing of SA or JA biosynthetic and signalling genes increased the susceptibility of this plant to TMV, whereas foliar application of JA followed by SA enhanced systemic resistance against this virus (Zhu et al., 2014). Remarkably, although both SA and JA are essential for systemic resistance responses, these hormones seem to act sequentially, rather than in an additive or a synergistic manner. This suggests that a hitherto unknown interaction exists between these two hormones that needs to be studied further for a better understanding of the molecular mechanisms of plant defence against viral pathogens (Alazem and Lin, 2015; Zhu et al., 2014).

Figure 3.

A simplified schematic representation of hormone‐regulated defence responses of plants against viruses. The recognition of viral effectors by R proteins initiates the activation of small interfering RNA (siRNA) pathways and the induction of the hypersensitive response (HR), accompanied by the biosynthesis of salicylic acid (SA). These responses trigger siRNA antiviral mechanisms and restrict virus spread by the formation of necrotic lesions. SA activates systemic acquired resistance (SAR) and siRNA pathways in distant host tissues, thus contributing to the maintenance or restoration of plant fitness. Another phytohormone, ethylene (ET), has a negative effect on antiviral plant defence, because it interferes with the pathway activated by SA. Jasmonic acid (JA) has been recognized to have both positive and negative effects on the efficacy of plant defence against viruses (see text). The interaction mechanisms between SA and JA are unknown. PRs, pathogenesis‐related proteins.

Three essential scenarios to explain SIE in plants have been proposed (Gal‐On and Shiboleth, 2006). (i) The secondary virus enters a cell that is already infected with the primary virus. Here, different mechanisms regulating SIE are possible: over‐expressed coat protein (CP) may prevent uncoating of the secondary virus (Beachy et al., 1990; Lu et al., 1998) (see below for an additional explanation); uncoated RNA may be degraded by RISC (Fig. 1); and the minus‐strand RNA of the primary virus may hybridize to the secondary virus RNA. Subsequently, the dsRNA hybrids may be subjected to Dicer‐mediated cleavage, resulting in the generation of vsiRNAs. (ii) The secondary virus enters cells that have been primed by vsiRNA, but do not contain RNA of the primary virus. Here, RISC targets the RNA of the secondary virus for degradation. (iii) The secondary virus enters cells distant from the infected cell(s). In this case, a phloem‐mobile long‐distance signal is sent to amplify the silencing response by the endogenous RdRp, which is able to activate RISC and degrade RNA of the secondary virus. Thus, in this scenario, systemic silencing is involved, which can additionally explain the phenomena of ‘recovery’ and so‐called ‘green islands’ observed in some natural infections or in certain lines of transgenic plants expressing virus‐derived transgenes. In the ‘recovery’ phenomenon, young upper leaves of an infected plant are symptomless or show only mild symptoms (see Fig. 2). The viral RNA level in these leaves is either low or undetectable.

The latter scenario would be a reasonable explanation of why SIE is limited to closely related virus strains/isolates and requires an interval between inoculations (Ziebell and Carr, 2010). It would also explain the breakdown of plant protection in some cases. If the level of the secondary virus RNA infecting a primed cell exceeds the amount of activated RISC, the secondary virus may eventually succeed in infecting this plant. In addition, the occurrence of disparate virus strains/isolates might help the secondary virus to evade recognition and degradation by RISC.

It is worth mentioning that the hypothesis on CP‐mediated SIE was based on the finding that transgenic tobacco plants expressing the CP of TMV were resistant to infection with this virus, suggesting that CP interferes with the disassembly of TMV particles in inoculated cells (Beachy, 1999). However, the sole responsibility of this mechanism for SIE has been questioned, particularly because CP‐defective viruses and viroids can also confer SIE (Gal‐On and Shiboleth, 2006). It cannot be excluded that CP‐mediated resistance plays a role in SIE in some virus–host interactions, but is restricted to virus‐infected cells, because CP subunits do not move from cell to cell. More recently, CP‐independent SIE between deconstructed variants of TMV has been demonstrated to operate in N. benthamiana as an early induced, fast‐spreading and cell autonomous mechanism (Julve et al., 2013). Virus deconstruction means that virus genome functions which are limiting or undesired are removed and the virus is rebuilt by either transferring the missing obligatory functions to the host or replacing them with analogous functions that do not come from a virus. A viral MP appears to be dispensable for the phenomenon to occur, but its presence positively affects the process. The mechanism of SIE has been postulated to be based, at least in part, on competition for strain‐specific cell‐limiting factors. The competition between viruses for access to host cell resources can explain the early induction of SIE and the accelerating effect of MP, as this protein has previously been reported to mediate the fast cell‐to‐cell movement of TMV replicative complexes in the absence of CP. It has been proposed that virus replication complex (VRC)‐like structures are exclusion‐competent particles which, in the presence of an active MP, are transported from cell to cell at high rates, thus spreading immunization during early infection stages (Julve et al., 2013). In the authors' opinion, the identification of the factors limiting virus–host interactions will result in new specific weapons against pathogens, based on the immunization of plants by controlling the viral and host factors involved in SIE, rather than by using competitive infections.

The mechanism proposed for TMV (Julve et al., 2013) corresponds to the strain specificity demonstrated for Citrus tristeza virus (CTV), where SIE occurred between isolates of the same strain, but not between isolates of different strains (Folimonova, 2013; Folimonova et al., 2010). It has been shown that SIE by CTV requires a specific viral protein p33 (Folimonova, 2012). CTV p33 is one of the three proteins (p13, p18 and p33) playing a crucial role in extending the virus host range (Tatineni et al., 2011), but is dispensable for CTV infection (Tatineni et al., 2008). A lack of functional p33 has been demonstrated to completely abolish the exclusion ability of CTV (Folimonova, 2012). CTV mutants that are not able to encode p33 fail to exclude super‐infection by the parental wild‐type virus. Plants pre‐infected with mutants with a deleted p33 gene and then challenge inoculated with GFP‐marked CTV show GFP fluorescence distribution and intensity that are comparable with those observed in non‐pre‐infected trees. Moreover, hybrid viruses, in which the p33 protein is substituted with a cognate protein from a heterologous strain, fail to interfere with infection induced by the challenge wild‐type virus, which indicates that a heterologous p33 cannot mediate the exclusion. Most probably, the p33 protein functions in a homology‐dependent manner (Bergua et al., 2014; Folimonova, 2012). The diversification of strain‐specific interactions can be seen as a strategy employed by the virus to avoid plant defences (Julve et al., 2013). Alternatively, it may be used to favour heterologous rather than homologous co‐infections. Because large amounts of virus are not needed for SIE to be initiated, the systemic immunization of neighbouring cells can be achieved by occupying or depriving the limiting host factor. Such a model would intriguingly imply that the plant itself, rather than the virus, regulates homologous interference by controlling the availability of the host limiting factor (Julve et al., 2013). It is unclear whether SIE by CTV involves components of the RNA silencing pathway or operates through another novel mechanism (Folimonova, 2012). The data demonstrating that the protein p33 plays a substantial role in SIE suggest an existence of precise interactions between this protein and some other viral factor(s) involved in this phenomenon (Bergua et al., 2014; Folimonova, 2012).

Historically, SIE between plant viruses has predominantly been attributed to the induction of RNA silencing by the primary virus. However, it cannot be excluded that different mechanisms are involved in this phenomenon, depending on the virus or even on its strain or isolate (Folimonova, 2012; Zhou and Zhou, 2012; Ziebell and Carr, 2010). None of the mechanisms proposed so far can fully explain the intact process of SIE, and many molecular details remain obscure.

Effects of Antagonistic Interactions Among Viruses on Viral and Host Plant Fitness

Introductory remarks on virus and host fitness

Although a number of definitions of fitness, often inconsistent with each other, have been offered by biologists (Abrams, 2009, 2012; Krimbas, 2004; Ramsey, 2013), the generally accepted definition is that fitness involves the ability of organisms to survive and reproduce in the environment in which they find themselves, thereby contributing genes to the next generation (Laine and Barrès, 2013; Metcalf et al., 2015; Orr, 2009). Organisms differ in their fitness, and all such differences result in natural selection (Abrams, 2014; Metcalf et al., 2015).

For viruses, whose life cycle depends entirely on host cells, fitness is defined as the extent to which the virus has adapted to the host (Barbour and Grant, 2005) and is capable of producing infectious progeny, the ability being referred to more specifically as replicative fitness (Domingo, 2010; Domingo et al., 1997; Wargo and Kurath, 2012). Viral fitness includes a variety of components, such as genome unpacking, translation, replication, coating into new particles, and cell‐to‐cell and systemic movement in plants (Elena and Lalić, 2013). In all these steps, fitness depends on the ability of the virus to exploit host cellular resources. Moreover, it depends on the ability of the virus to impede or evade the defence mechanisms of the host plant. Because for survival and dissemination in a given environment a virus must be transmitted to new hosts (Andret‐Link and Fuchs, 2005; Gutiérrez et al., 2013; Syller, 2014), its fitness determining successful transmission depends on the stability of virion particles and on the direct interactions between the virus and the vector (Elena and Lalić, 2013; Jiu et al., 2007). Therefore, viral fitness during host‐to‐host transmission is considered to be an important component of overall viral fitness (Wargo and Kurath, 2012).

Replicative fitness of viral variants has been assessed in vivo or in vitro, typically by comparing the replication of two or more isolates representing the same species (Tripathi et al., 2015; Wargo and Kurath, 2012). In some studies, parallel hosts or cell cultures infected with single viral variants are used. However, a more sensitive and reliable measure of fitness differences between viral variants can be obtained using mixed infected hosts, in which the replication of each variant is determined in a competitive environment (Elena et al., 2014; Wargo and Kurath, 2012). The incorporation into assays of the effects of variables, such as whether infections are simultaneous or sequential, can be straightforward, but problems may be encountered in realistically describing interdependent processes, such as immunosuppression or cross‐reactive immunity (Metcalf et al., 2015).

Under natural conditions, plants are constantly confronted with a variety of biotrophic and necrotrophic pathogens. Hence, in order to survive and reproduce, they must grow and defend themselves at the same time, which are fundamentally conflicting processes (Herms and Mattson, 1992; Huot et al., 2014). The conflict between growth and defence imposes a trade‐off between these two systems, which has significant ecological and agricultural outcomes. The resistance mechanisms must be fine‐tuned because the use by plants of defence‐related secondary metabolites can come at a cost to other physiological processes, and thus have a negative impact on other plant traits, such as biomass and seed production (Denancé et al., 2013; Heil, 2002; Heil and Baldwin, 2002; Herms and Mattson, 1992; Reitz et al., 2015; Ruan et al., 2013; Walters and Heil, 2007). Seed mass and seed number are likely to have the greatest impact on the persistence of plant populations, as they determine the relative abilities of species to establish or disperse (Pierce et al., 2014).

Current knowledge on fitness costs and benefits of antagonistic interactions among plant viruses

The effects of within‐host competition between plant viruses on their fitness have so far been evaluated in few studies. In an Australian study, isolates of two tobamoviruses, TMV and Tobacco mild green mosaic virus (TMGMV) from New South Wales (NSW), found in both living wild Nicotiana glauca plants and NSW herbarium samples collected over a century ago, were analysed (Fraile et al., 1997). Early samples were found to be infected with both viruses, but, in later samples, TMGMV considerably prevailed over TMV. Interestingly, sequence analyses showed no increase in the genetic diversity among the TMGMV isolates over the century. By contrast, the genetic diversity of synonymous differences between TMV isolates varied and was correlated with their time of isolation. In experimental N. glauca plants, TMV accumulated to lower concentrations than did TMGMV and, in mixed infections, the latter strongly inhibited TMV accumulation, which indicates that TMGMV is fitter than its competitor. In another study, however, TMGMV appeared to be less fit than Oilseed rape mosaic virus (ORMV), another tobamovirus, in co‐inoculated N. tabacum cv. Samsun plants (Aguilar et al., 2000). As predicted by these authors, when two viruses belong to the same genus, but have a relatively distant genetic relationship, within‐host interactions between these viruses may be synergistic, antagonistic or neutral. In singly infected plants, TMGMV and ORMV were detected in both inoculated and non‐inoculated leaves (systemic infection), whereas, in co‐inoculated plants, ORMV inhibited TMGMV accumulation in both types of leaves. Interestingly, when the two viruses were sequentially inoculated on N. tabacum, ORMV protected the plants from challenge infection by TMGMV, and TMGMV protected the plants against challenge infection with ORMV. However, the interference between TMGMV and ORMV appeared to be host dependent, because, in Arabidopsis thaliana, protection only occurred when TMGMV acted as the protecting virus (Aguilar et al., 2000). TMGMV presumably does not prevent ORMV reproduction, as the latter is able to replicate and accumulate in inoculated leaves. Instead, TMGMV is likely to inhibit ORMV movement from the inoculated leaf to the phloem, delaying or impairing the transport of the secondary virus from inoculated to non‐inoculated leaves (Aguilar et al., 2000).

The plant response to double infections with TMGMV and ORMV was dependent on the host species (Aguilar et al., 2000). Nicotiana tabacum plants pre‐infected with ORMV developed severe symptoms, which were typical of infection with ORMV. Plants pre‐infected with TMGMV were protected from the severe symptoms caused by ORMV, but exhibited more severe symptoms than the plants infected with TMGMV alone. It can therefore be concluded that doubly infected N. tabacum plants paid no additional fitness costs as a result of the antagonistic interactions between TMGMV and ORMV. On the contrary, the plants first infected with TMGMV proved to benefit from such interactions. Arabidopsis plants protected by TMGMV did not show any symptoms, whereas those pre‐infected with ORMV displayed symptoms typical of this virus.

More recently, two criniviruses, Tomato chlorosis virus (ToCV) and Tomato infectious chlorosis virus (TICV), and two begomoviruses, Tomato rugose mosaic virus (ToRMV) and Tomato yellow spot virus (ToYSV), have been reported from the USA (Wintermantel et al., 2008) and Brazil (Alves‐Júnior et al., 2009), respectively, to coexist in tomato plants, indicating that infection by one virus does not prevent infection by a second virus from the same genus. In co‐infected Physalis wrightii, the accumulation of both criniviruses decreased, when compared with that in single infections, a pronounced decrease being recorded for TICV (Wintermantel et al., 2008). In N. benthamiana, TICV accumulation increased, whereas that of ToCV decreased. Seemingly, the presence of TICV reduced the replicative fitness of ToCV in this host. It is likely that the differences between TICV and ToCV in the adaptability to different hosts may eventually translate into competitiveness of each virus in dually infected hosts (Wintermantel et al., 2008). Like criniviruses, the accumulation of both begomoviruses decreased in doubly infected tomato plants (Alves‐Júnior et al., 2009). In N. benthamiana, DNA accumulation of ToYSV decreased, whereas that of ToRMV significantly increased between 14 and 28 days post‐inoculation, indicating the existence of both antagonistic (during the initial stages of infection) and synergistic interactions between ToYSV and ToRMV. In both hosts, ToYSV produced more severe symptoms than ToRMV. Interestingly, in dually infected tomato, symptoms were more severe than those produced by either virus alone, irrespective of the decreased titres of both viruses compared with single infections. Possibly, tomato plants paid fitness costs as a result of the negative interference between the invading begomoviruses. In co‐infected N. benthamiana plants, symptoms were comparable with those caused by single infection with more virulent ToYSV.

Reduced replicative fitness as the effect of antagonistic interactions has recently been reported for genetically different isolates of PVY in potato and N. tabacum plants (Syller and Grupa, 2014). PVY^NTN isolates proved to be fitter than PVY^O or PVY^NW isolates in potato plants, whereas they were either fitter or weaker than the competitors in tobacco. The results suggest that the competitive interference among PVY isolates is both isolate and host dependent. The differences in accumulation between PVY isolates were reflected in different rates of transmission by insect vectors. PVY^NTN isolates were transmitted by single aphids of Myzus persicae more readily than were the other isolates (Syller and Grupa, 2014).

The response of plants to dual vs. single infections with PVY isolates was dependent on their species (Syller and Grupa, 2014). Doubly infected potato plants did not seem to pay any fitness costs of the interactions among isolates. By contrast, the response of tobacco plants to double infections by certain isolates differed substantially from that of single infections. Interestingly, the severity of symptoms displayed by plants dually infected with a highly virulent and a low virulent isolate was similar to that in single infections with either the former or the latter virus, depending on the genetic background of the low virulent isolate. This differentiation in symptom severity corresponded with the effects of co‐infection on the accumulation of particular isolates in tobacco plants (Syller and Grupa, 2014). It may be hypothesized that, unlike potato, tobacco plants paid considerable fitness costs when the fitter of the competing PVY isolates achieved high accumulation levels.

A similar explanation may be valid for the outcome of co‐infection of a mild isolate I‐38 and a severe isolate Uran of PVM in D. metel plants (Grupa and Syller, 2015). Datura metel seedlings were mechanically inoculated with inoculum containing I‐38 or Uran alone, or a 1 : 1 (v/v) mixture of both isolates. In another experimental combination, seedlings were pre‐inoculated with I‐38 and, after 7 days, were challenge inoculated with Uran. Approximately 25% of the simultaneously inoculated plants exhibited no or slight symptoms, whereas the remaining 75% showed severe symptoms, typical of infection with the virulent isolate. The results indicated the occurrence of intraspecific competition between the isolates, the more virulent isolate Uran appearing to be fitter than its counterpart. Both isolates were detected in these plants by reverse‐transcription polymerase chain reaction. Remarkably, the plants pre‐infected with the mild isolate and challenge inoculated with the virulent isolate remained asymptomatic or exhibited only slight symptoms (Fig. 2). Recovery of the plants from disease symptoms suggested the occurrence of within‐host exclusion of the highly virulent isolate by the low virulent isolate. Indeed, only the isolate I‐38 was detected in these plants (Grupa and Syller, 2015).

Co‐infection of two viral strains differing in virulence can also result in intermediate symptoms, as has been reported for severe and mild strains of Maize chlorotic dwarf virus (MCDV) (Morales et al., 2014). These strains are quite distantly related to each other as they share only 57% identity of nucleotide sequence and 59% identity of amino acid sequence. Nonetheless, the symptoms of intermediate severity have been attributed to competitive rather than synergistic interaction of the strains (Morales et al., 2014). It may be concluded that doubly infected plants benefit from the competition between these strains when compared with plants infected with the severe isolate alone, but are at a disadvantage when compared with plants singly infected with the mild isolate.

Practical Exploitation of the Effects of Antagonistic Interactions Among Plant Viruses in Crop Protection—A Brief Overview

The effects of SIE among plant viruses were first reported by McKinney (1929), who found that, in tobacco plants infected with a ‘green’ strain of TMV, the appearance of yellow symptoms after re‐inoculation with a severe ‘yellow mosaic strain’ was restrained. Later, Salaman (1933) reported that an avirulent strain of PVX protected tobacco White Burley plants against superinfection with a virulent strain. Since then, purposeful infection with mild or attenuated strains has been used in attempts to protect crop plants against infection with more virulent strains, e.g. against Tomato mosaic virus (ToMV) in tomato, Zucchini yellow mosaic virus (ZYMV) in squash and melon, Papaya ringspot virus (PRSV) in papaya, Soybean mosaic virus (SMV) in soybean, CTV in citrus trees and others (reviewed in Gal‐On and Shiboleth, 2006; Lecoq and Raccah, 2001; Nishiguchi and Kobayashi, 2011; Ziebell and Carr, 2010). In agricultural practice, an attenuated or low virulent isolate must be artificially introduced into the plant. It acts as a vaccine applied in human and veterinary medicine (Nicaise, 2014; Syller, 2012), but no antibodies are produced by the plant in response to the infection. This method is applicable under glasshouse conditions, but it cannot be used in the field. For this and other reasons, this procedure is rarely used nowadays to protect annual crops. However, it might be effective in long‐term crops, in which the protecting virus (a naturally occurring mild strain, which can be inoculated into young trees in a nursery) is spread among stone‐fruit, pome‐fruit and citrus trees by insect vectors (Gal‐On and Shiboleth, 2006). For example, the disease caused by CTV has been reported recently to be controlled by inoculation of aphid‐transmitted mild CYV isolates into sour orange trees, which extends the productive life of the groves and enables a more gradual replanting of trees on CTV‐tolerant rootstocks (Lee and Keremane, 2013). CTV has a host range restricted to citrus and citrus‐related plants, in which it infects only phloem‐associated cells. Mild isolates of CTV were found to protect infected citrus trees against the harmful effects of virulent strains, but, in many varieties and growing areas, the protection completely failed (Folimonova et al., 2010). Further studies have revealed that SIE is only possible between CTV isolates of the same strain (Folimonova, 2013; Folimonova et al., 2010) (see the section on ‘Molecular mechanisms of antagonistic interactions between viruses’ in this review). This finding should be taken into account in future trials to create a mild CTV variant using an infectious clone based on the genetically characterized severe CTV isolate (Lee and Keremane, 2013). The mild variant could then be used as a ‘vaccine’ to protect trees against later infection by the virulent isolate (see below).

The development of engineering ‘vaccines’ based on viral vectors carrying a genomic fragment of the virulent virus is a promising alternative for labour‐ and time‐consuming identification of effective mild isolates for each economically important virus (Nicaise, 2014; Satoh et al., 2014). For example, vectors of ALSV, which is normally isolated from apples, but under experimental conditions causes latent infections in plants from the families Cucurbitaceae, Fabaceae and Solanaceae, from the Nicotiana genus, and in fruit trees (e.g. apple, cherry, peach and plum), may serve as effective ‘vaccines’ for the control of plant viral diseases by inducing virus‐induced gene silencing (VIGS) (Satoh et al., 2014). This method assumes that, if a plant is pre‐infected with an ALSV vector carrying the genomic sequence of the pathogenic target virus to induce VIGS, RNA silencing will be established to protect the plant against infection with the target virus. It has been reported that plants from the Cucurbitaceae family, which were pre‐inoculated with ALSV ‘vaccines’ containing a part of the genomic sequence of ZYMV and CMV, exhibited strong defence responses to challenge inoculation with the corresponding virulent viruses, and viral multiplication was largely inhibited (Tamura et al., 2013). Moreover, when ZYMV‐infected cucumber plants displaying mosaic symptoms were inoculated with an ALSV ‘vaccine’ containing part of the ZYMV genome, no mosaic symptoms developed in new upper leaves and normal plant growth was restored, indicating that the ALSV ‘vaccine’ also showed a protective effect. More recently, similar protective effects have been observed when pea, broad bean and Eustoma plants were inoculated with an ALSV‐based vector harbouring a segment of the Bean yellow mosaic virus (BYMV) genome (Satoh et al., 2014). In pea plants pre‐inoculated with the ALSV ‘vaccine’ and re‐inoculated with BYMV, the target virus multiplied in inoculated leaves, but its multiplication in upper leaves was strongly inhibited. No mosaic symptoms were developed by these plants. Simultaneous inoculation with the ALSV ‘vaccine’ and BYMV also prevented mosaic symptoms in broad bean and Eustoma plants. Remarkable effects of inoculation of pea and Eustoma plants with BYMV followed by inoculation with the ALSV ‘vaccine’ were observed. Plants of both species showed recovery from mosaic symptoms in upper leaves, associated with the inhibited BYMV accumulation, confirming the protective potential of the ALSV ‘vaccine’.

Concluding Remarks

In order to successfully establish infection, viruses modify the host cellular environment through diverse mechanisms aimed at impairing cellular function and optimizing virus replication (Villanueva et al., 2005). In a cell infected with two or more viruses, a conflict of interest is likely to arise because of the limited supply of resources essential for viral replication. The conflict leads to antagonistic interactions between closely related viruses, expressing similar basic requirements. In turn, antagonistic interactions result in fitness costs to both viruses or to the weaker of the two counterparts. In this conflict, neither of the viruses is the beneficiary, but one may become a winner (Roossinck, 2005) and thus bear relatively lower fitness costs. Which of the viruses can be expected to win? As mentioned, a virus entering a cell as the first virus gains both a territorial and numerical advantage over a potential counterpart, thus having a chance to exclude the intruder. However, the latter may eventually appear fitter and will defeat the primary virus, as observed in our research.

Without differences in fitness, natural selection cannot act and adaptation cannot occur (Orr, 2009). From the standpoint of evolutionary theory, within‐host competition between different pathogen genotypes will favour variants expressing a higher level of virulence (Choisy and de Roode, 2010). However, the phenotypic plasticity of parasites, including viruses, and an impaired immune response of the host can select for lower virulence. As a rule, in experimental studies, the overall virulence of multiple versus single infections is compared within one generation. Because RNA viruses are highly adaptable to new hosts, and host immunity response can be reduced by numerous factors, an erroneous conclusion may be drawn that pathology has increased. Hence, it is reasonable to avoid the conclusion of virulence evolution on the basis of single generation experiments (Choisy and de Roode, 2010). Moreover, although trials under experimental conditions are essential for the establishment of basic knowledge about individual signalling pathways, it is advisable to conduct future experiments under conditions that more accurately reflect natural environments, involving a wide array of biotic and abiotic stimuli. In such studies, interactions between plant viruses and their natural vectors should not be ignored, because of the diversity of strategies used by vector‐borne viruses. Vector transmission is a complex process, involving highly specific interactions between viral, host and vector determinants (e.g. Jeger et al., 2011; Mauck et al., 2012; Syller, 2014). These multi‐component interactions are likely to be even more complex if the host plant is simultaneously infected by two or more viruses. The effects of within‐host synergistic or antagonistic interactions among viruses on the rate of vector transmission should be given particular attention, as they may have serious ecological and epidemiological consequences (Syller, 2014). Therefore, a full understanding of plant virus epidemiology requires studies at scales of integration ranging from within‐plant cell processes to vector population dynamics, behaviour and broader ecological interactions (Jeger et al., 2011).

The plant response to a multiple viral infection is the outcome of a complex and dynamic host–virus(es) and virus–virus interplay. Plants have evolved mechanisms, such as hormone cross‐talk, to enable the fine‐tuning of their responses to environmental and developmental cues. There is a natural trade‐off in plants between the ability to resist a pathogen(s) and the inherent growth rate. In agriculture, crops have been bred for decades for growth‐promoting traits. This approach has resulted in a loss of genetic diversity that is known to compromise defence. However, prolonged induction of defence responses bears a fitness cost for the plant, because immune activation inhibits plant growth. The shift in the balance from growth to immunity will significantly reduce crop yield and will impair the efforts of breeders to develop next‐generation crops with improved yield. Therefore, a better understanding of the molecular mechanisms employed by plants to balance growth and defence should contribute to improving plant breeding and engineering strategies to select for desired genetic traits that will maximize plant fitness (Huot et al., 2014; Reitz et al., 2015).