Plants activate defence responses to protect themselves against microbial pathogens and herbivorous insects. However, induction of defences comes at a price, as the associated allocation costs, autotoxicity costs and ecological costs form fitness penalties. Upon pathogen or insect attack, resources are allocated to defences instead of to plant growth and reproduction, while above- and below-ground interactions with beneficial organisms may also be disturbed. The phytohormones salicylic acid and jasmonic acid are major players in the regulation of induced defences and their associated fitness costs. Hormone-controlled signalling pathways cross-communicate, providing the plant with a finely tuned defence regulatory system that can contribute to a reduction of fitness costs by repressing ineffective defences. However, this sophisticated regulatory system causes ecological costs, because activated resistance to one organism can suppress resistance to another. Moreover, the system can be hijacked by invading organisms that manipulate it for their own benefit. Priming for enhanced defence emerged as a defence mechanism with limited fitness costs. Because priming results in a faster and stronger activation of defence only after pathogen or insect attack, the limited costs of the primed state are often outweighed by the benefits in environments with pathogen or herbivore pressure. The balance between protection and fitness is crucial for a plant's success and is therefore of great interest for plant breeders and farmers. By combining molecular knowledge and ecological relevance of defence mechanisms, one can gain fundamental insight into how and why plants integrate different immune signals to cope with their natural multitrophic environment in a cost-effective manner.
During their lifetime, plants encounter innumerable attackers, including microbial pathogens and herbivorous insects that try to retrieve nutrients from the plant. Plants can ward off the majority of these attackers for which they rely on preformed defences and activation of their innate immune system. Preformed plant defences include physical barriers such as thick cuticles, rigid cell walls, thorns, needles and trichomes, and chemical weapons such as toxic or repellent compounds (Osbourn, 1996). In a second line of defence, inducible defences can be activated when pattern recognition receptors of plants recognize general features of microbial pathogens, such as flagellin, lipopolysaccharides, peptidoglycan, β-glucans and chitin, referred to as pathogen- or microbe-associated molecular patterns (PAMPs or MAMPs; Jones & Dangl, 2006; Pel & Pieterse, 2013). Similarly, recognition of so-called damage-associated molecular patterns (DAMPs), such as galacturonides, systemins and AtPep1, which are endogenous elicitors that accumulate as a result of enzymatic degradation of plant cell walls or proteins upon attack by pathogens or insects, leads to activation of defences (Fig. 1; Lotze et al., 2007; Boller & Felix, 2009; Heil, 2009; Ferrari et al., 2013). Other defence-inducing compounds are the herbivore-associated molecular patterns (HAMPs), such as fatty acid–amino acid conjugates (FACs) from oral secretions (Felton & Tumlinson, 2008; Mithöfer & Boland, 2008) and effectors of pathogens that are produced to suppress immune responses but that the plants, under evolutionary pressure, have learned to recognize (Jones & Dangl, 2006).
The immune response that is activated upon pathogen or insect attack is modulated by the induced production of a phytohormonal blend in the plant. The phytohormones salicylic acid (SA), jasmonic acid (JA), ethylene (ET) and abscisic acid (ABA) are important regulators of induced defence mechanisms (Robert-Seilaniantz et al., 2011; Pieterse et al., 2012). Whereas SA and JA are the main players, ET and ABA have more modulating roles (van Loon et al., 2006a; Ton et al., 2009). The SA pathway is primarily induced by and effective against biotrophic pathogens, whereas the JA pathway is primarily induced by and effective against necrotrophic pathogens and herbivorous insects (Fig. 1; Glazebrook, 2005; Howe & Jander, 2008). The quantity, composition and timing of the hormonal signal signature tailors the defence response specifically to the attacker at hand, thereby prioritizing effective over ineffective defences, which minimizes fitness costs (De Vos et al., 2005; Pieterse & Dicke, 2007).
The benefits of plant defences are obvious; they help the plant to survive in the presence of harmful organisms (Fig. 1). However, the inducible character of plant defences leaves a time slot between attack and the expression of defences in which the plant is vulnerable to the invading organism. Constitutive expression of defence traits does not have this drawback, making it probable that inducible defences have other selective advantages over constitutive defences (Heil & Baldwin, 2002). Fitness costs that are associated with defences have been postulated to be a driving force behind the evolution of inducible defences (Simms & Fritz, 1990). There are also costs associated with the genetic maintenance of inducibility, such as receptors and defence signal transduction routes, which all constitutively require energy and resources (Purrington, 2000; Cipollini et al., 2003). These maintenance costs may be minimal, because many inducible pathway components have been co-opted from other processes, such as growth and development (Pieterse et al., 2009).
The actual induced resistance status entails direct and indirect fitness costs. Direct resistance costs include allocation and autotoxicity costs (Heil & Baldwin, 2002; Strauss et al., 2002). The latter is inflicted on plants by induced secondary chemicals that are toxic to the plant itself as well (Baldwin & Callahan, 1993; Heil & Baldwin, 2002; Strauss et al., 2002). Allocation costs occur when valuable resources are allocated to resistance instead of to growth and reproduction (Herms & Mattson, 1992; Heil & Baldwin, 2002; Strauss et al., 2002; Walters & Heil, 2007). Re-allocation of plant resources has also been postulated as a means by which the plant starves the pathogen in order to halt the infection (Canet et al., 2010). Once induced, the enhanced resistance status needs to be maintained, but this is less costly (Van Hulten et al., 2006).
Indirect resistance costs, also known as ecological costs (Heil & Baldwin, 2002; Strauss et al., 2002), occur as a result of the changed physiology of the plant that in turn affects interactions with other biotic and abiotic environmental factors, such as beneficial or harmful organisms, competing plants and resource availability (Heil, 2002; Cipollini et al., 2003; Kessler & Halitschke, 2007; Poelman et al., 2008; Traw & Bergelson, 2010). In this review, an overview of current knowledge on benefits and costs associated with inducible defences that are controlled by phytohormones is provided and how this knowledge can be applied for improved crop protection is discussed.
SA is a phenolic compound that plays a key role in disease resistance signalling in plants (Mishina & Zeier, 2007; Vlot et al., 2009). Besides its role in plant defence, SA also influences seed germination, vegetative growth, photosynthesis, respiration, thermogenesis, flower formation, seed production, senescence and responses to abiotic stress (reviewed in Rivas-San Vicente & Plasencia, 2011). Plants rapidly synthesize SA upon pathogen infection (Malamy et al., 1990; Métraux et al., 1990). SA can be synthesized via two distinct enzymatic pathways that both require chorismate. The bulk of pathogen-induced SA is produced from isochorismate via ISOCHORISMATE SYNTHASE 1 (ICS1), but chorismate can also be converted into SA via a series of enzymatic reactions initially catalysed by PHENYLALANINE AMMONIA LYASE (PAL; Vlot et al., 2009). Defence signalling downstream of SA is largely regulated by the regulatory protein NONEXPRESSOR OF PR GENES1 (NPR1; Dong, 2004). Activation of the SA signalling pathway leads to a change in the cellular redox state, which reduces NPR1 from its inactive oligomeric form to its active monomeric form. Monomeric NPR1 is then translocated to the nucleus where it interacts with TGA transcription factors (Mou et al., 2003; Dong, 2004; Moore et al., 2011), resulting in the activation of a large set of defence-related genes, amongst which are genes coding for PATHOGENESIS-RELATED (PR) proteins and WRKY transcription factors (van Loon et al., 2006b; Rushton et al., 2010; van Verk et al., 2011).
Benefits: SA triggers disease resistance
The first indication for a role of SA in disease resistance signalling came from White (1979), who showed that exogenous application of SA to tobacco plants enhanced resistance against Tobacco mosaic virus (TMV). Nowadays, numerous examples exist that demonstrate the resistance-inducing capacity of SA in a wide variety of plants against (hemi-)biotrophic pathogens and some phloem-feeding insects (Klessig & Malamy, 1994; Walling, 2008; Vlot et al., 2009). The significance of SA was further shown by the use of mutant or transgenic plants (mostly in Arabidopsis thaliana, tobacco and tomato) that are affected in the production or the perception of SA. For example, transgenic NahG plants, which are incapable of accumulating SA, and mutant npr1 plants, which are impaired in SA signalling, are more susceptible to oomycete, fungal, bacterial and viral pathogens (reviewed in Glazebrook, 2005).
Trade-offs: allocation costs of SA-inducible defences
Several studies investigated the costs of SA-inducible defences. In general, exogenous application of SA or its chemical analogue benzothiadiazole (BTH) has been shown to reduce plant growth and seed production of different plant species (Heil et al., 2000; Cipollini, 2002; Canet et al., 2010). However, environmental conditions such as growing period, competition with neighbouring plants, and nitrogen supply, can influence these fitness effects and sometimes avert the growth costs associated with SA-inducible defences (Heidel et al., 2004; Dietrich et al., 2005). Arabidopsis mutants constitutively expressing SA-inducible defences, such as constitutive expressor of PR genes 1, 5 and 6 (cpr1, cpr5 and cpr6), were shown to be dwarfed and severely affected in seed production (Bowling et al., 1994; Heil & Baldwin, 2002; Heidel et al., 2004; Van Hulten et al., 2006). Conversely, SA-deficient NahG and SA induction deficient2 (sid2, mutated in the SA biosynthesis gene ICS1) Arabidopsis plants had higher growth rates and seed production compared to wildtype plants under pathogen-free conditions (Cipollini, 2002; Abreu & Munné-Bosch, 2009), confirming the negative effects of SA on growth and reproduction. The decrease in growth that was observed after treatment with BTH was reduced in the SA signalling mutant npr1, implying a pivotal role of NPR1 in inhibiting plant growth when SA-dependent resistance mechanisms are activated (Van Hulten et al., 2006; Canet et al., 2010). However, after infection with the SA-inducing downy mildew pathogen Hyaloperonospora arabidopsidis, npr1 mutant plants displayed a lower fitness than wildtype plants (Heidel & Dong, 2006). This demonstrates that, although costly, SA-inducible defences are beneficial when plants grow under pathogen pressure. The beneficial effects of SA-regulated defences was particularly apparent under low-nutrient conditions (Heidel & Dong, 2006), which supports the theory on allocation costs as a driver of the evolution of inducible defences. Mutants cpr1 and cpr5 that constitutively express SA-regulated defences failed to show a fitness benefit under pathogen pressure, supporting the hypothesis that the inducible character of SA-dependent resistance prevents excessive fitness costs (Heidel & Dong, 2006; Van Hulten et al., 2006).
Although negative effects of SA on fitness have mostly been ascribed to allocation costs (Heil et al., 2000; Walters & Heil, 2007), toxic effects of SA may also contribute to reduced fitness (Bi et al., 2010; Asaduzzaman & Asao, 2012). However, most studies focusing on autotoxicity costs of SA have not included plant genotypes that rule out effects of allocation costs, e.g. SA signalling mutants such as npr1, which makes claims on a role for SA in autotoxicity costs obscure. Moreover, most studies on allocation costs of SA signalling that have tested npr1 made use of BTH as inducer of the SA pathway, which induces SA signalling and resistance without the toxic side effects of SA (Lawton et al., 1996), thereby omitting autotoxicity effects of SA in their studies. One of the few studies that applied SA to npr1 (Cipollini, 2002) found no decrease in seed set in comparison to non-treated npr1 plants, whereas SA treatment did decrease seed set in wildtype plants, indicating that the costs incurred by SA are (mostly) allocation costs and not autotoxicity costs. Besides direct allocation costs, SA-inducible defences also inflict ecological costs, which include crosstalk effects between the SA and JA signalling pathways, which are described in the section on crosstalk.
JA is a key regulator in the defence response against herbivorous insects and necrotrophic pathogens (Glazebrook, 2005; Howe & Jander, 2008). Besides its essential role in regulating disease and pest resistance, JA has also been implicated in senescence, root growth, fruit ripening, tendril coiling, pollen development, tuberization and responsiveness to abiotic stress (Wasternack & Hause, 2013).
JA is an oxylipin that accumulates rapidly in plants in response to wounding, herbivory and infection by necrotrophs (Creelman et al., 1992; Penninckx et al., 1996). The initial phase of JA formation takes place in the chloroplasts, where fatty acids of membrane lipids (e.g. linoleic acid) are metabolized by lipoxygenases to generate oxylipins including the JA precursor 12-oxo-phytodienoic acid (OPDA). Subsequently, OPDA is transported to the peroxisomes where it undergoes three steps of β-oxidation to generate JA (reviewed in Wasternack & Hause, 2013). JA can be conjugated to amino acids, such as l-isoleucine, resulting in JA-Ile, the most biologically active member of the JAs (Staswick & Tiryaki, 2004; Fonseca et al., 2009). The F-box protein CORONATINE INSENSITIVE1 (COI1) is a key regulator of the JA signalling pathway (Xie et al., 1998), as it is part of the JA receptor complex (Yan et al., 2009; Sheard et al., 2010). Binding of JA to COI1 targets JASMONATE ZIM-DOMAIN (JAZ) proteins for degradation via the 26S proteasome pathway (Chini et al., 2007; Thines et al., 2007). In the uninduced state, JAZ proteins repress JA-responsive gene expression by binding to transcriptional activators, such as MYC2, 3 and 4 (Fernández-Calvo et al., 2011). Accumulation of JA triggers the degradation of JAZ proteins, resulting in derepression of JA-regulated genes.
In Arabidopsis, there are two distinct branches of the JA signalling pathway that antagonize each other. The ERF-branch is activated upon infection with necrotrophic pathogens and is regulated by the APETALA2/ETHYLENE RESPONSE FACTOR (AP2/ERF)-domain containing transcription factors ERF1 and ORA59 (Anderson et al., 2004; Pré et al., 2008). The ERF-branch of the JA response pathway also requires ET and results in the activation of the marker gene PLANT DEFENSIN1.2 (PDF1.2; Penninckx et al., 1998; Lorenzo et al., 2003). The MYC-branch is activated upon wounding and feeding by herbivorous insects and is regulated by the MYC transcription factors MYC2, 3 and 4 in a synergistic action with ABA (Anderson et al., 2004; Fernández-Calvo et al., 2011; Niu et al., 2011). Activation of the MYC-branch leads to induced expression of the marker gene VEGETATIVE STORAGE PROTEIN2 (VSP2; Lorenzo et al., 2004).
Benefits: JA triggers disease resistance
Many JA-inducible defence responses and their effectiveness in plant resistance against diseases and pests were identified by exogenous application of JA and by the analysis of mutants with defects in JA signalling compounds such as COI1, MYCs and ERFs. This demonstrated that JA signalling is indispensable for resistance to a wide range of necrotrophic pathogens and herbivorous insects, whereby in general the ERF-branch is associated with resistance against necrotrophic pathogens (Berrocal-Lobo et al., 2002; Lorenzo et al., 2003) and the MYC-branch with resistance against herbivorous insects (Lorenzo et al., 2004; Howe & Jander, 2008; Kazan & Manners, 2012).
Trade-offs: allocation costs of JA-inducible defences
Studies on the costs of JA-inducible defences have mostly been executed with plants that were infested with insects. In contrast, cost studies on JA-inducible defences that are associated with infection by necrotrophic pathogens are scarce. Infestation by insects and exogenous application of JA comes with costs, which is apparent from a decreased seed set and delayed flowering and fruit ripening (Agrawal et al., 1999; Redman et al., 2001; van Dam & Baldwin, 2001). In addition, the Arabidopsis mutant cev1 and the transgenic line overexpressing JA carboxyl methyltransferase (JMT), both constitutively expressing JA-dependent defences, showed reduced growth phenotypes (Ellis & Turner, 2001; Cipollini, 2010). The effect on delayed flowering of JMT-overexpressing lines was especially apparent under low nutrient conditions (Cipollini, 2010). Furthermore, competition with neighbouring plants increased JA-induced fitness costs in tobacco (van Dam & Baldwin, 2001). The JA-associated trade-offs in reduced plant performance have therefore been mostly explained by allocation costs. However, JA is also known to directly regulate several plant developmental processes such as growth and seed production (Creelman & Mullet, 1997; Yang et al., 2012), which complicates the assignment of the origin of the fitness decrease detected in plants expressing JA responses. Despite the fact that JA-induced responses are costly, they benefit plants when under attack, even in field situations (Baldwin, 1998). To our knowledge no studies on autotoxicity costs of JA have been described so far. Ecological costs of JA-inducible defences include crosstalk effects between the ERF- and MYC-branches of the JA signalling pathway, which are described in the section below on crosstalk.
Hormonal crosstalk in defence signalling
Plant hormones are integral to plant immune responses and are differentially effective against different types of attackers. During plant–attacker interactions, multiple hormones are induced that together steer the immune response of the plant (De Vos et al., 2005). Hormone homeostasis is vital for a successful immune response upon attack, as extensive cross-communication between defence signalling pathways allows the plant to fine-tune the defence response to the attacker at hand (Reymond & Farmer, 1998). Hormonal crosstalk has often been interpreted as a cost-saving strategy and may have evolved as a means of the plant to reduce allocation costs by repression of unnecessary defences that are ineffective against the attacker that is encountered (Pieterse & Dicke, 2007; Thaler et al., 2012). However, proof for this hypothesis has not been demonstrated yet, as to the authors' knowledge there has been no study that measured the fitness levels of plants exhibiting hormonal crosstalk in comparison to that of crosstalk mutant plants. In this review crosstalk between SA and JA signalling, and between the ERF- and MYC-branches of the JA signalling pathway is covered, but other hormones have also been reported to modulate hormone-controlled immune signalling (Robert-Seilaniantz et al., 2011; Pieterse et al., 2012).
Crosstalk between the SA and JA pathways
The first indication of crosstalk between SA and JA signalling came from Doherty et al. (1988), who showed that SA and its acetylated derivative aspirin are strong antagonists of the JA pathway. Many cases of crosstalk between the SA and JA pathways have been reported since (Bostock, 2005; Stout et al., 2006; Pieterse et al., 2012). Pharmacological experiments with Arabidopsis revealed that the JA-responsive genes PDF1.2 and VSP2 are highly sensitive to suppression by SA. The antagonistic effect of SA on JA signalling was observed in a large number of Arabidopsis accessions (Koornneef et al., 2008) and was even reported to remain active in the next generation (Luna et al., 2012), highlighting the potential significance of this phenomenon in the regulation of induced plant defences in nature.
Several important regulatory proteins of SA–JA pathway crosstalk have been identified, including NPR1 (Spoel et al., 2003; Luna et al., 2012; Pieterse et al., 2012). Timing and concentration of ET was shown to modulate the strength of SA–JA crosstalk and its NPR1-dependency. Recent data showed that, in Arabidopsis, SA does not affect JA biosynthesis but it affects JA signalling downstream of COI1, at the level of transcriptional regulation of JA-responsive genes (Leon-Reyes et al., 2010b; van der Does et al., 2013). The ERF transcriptional activator ORA59, which regulates many JA-responsive genes, was identified as an important target for SA (van der Does et al., 2013).
Many studies found an antagonistic effect of SA signalling on the JA pathway. However, in several cases JA signalling could suppress the SA pathway as well. A few molecular players have been reported to play a role in this JA–SA crosstalk, such as COI1 and MYC2 (Zheng et al., 2012). Furthermore, neutral and synergistic interactions between SA and JA have also been reported (Schenk et al., 2000; van Wees et al., 2000; Mur et al., 2006). Timing, sequence of initiation and the relative concentration of each hormone are important for the outcome of the SA–JA crosstalk (Mur et al., 2006; Koornneef et al., 2008; Leon-Reyes et al., 2010a). Besides the aforementioned ET, other hormones can also modulate SA–JA crosstalk (Pieterse et al., 2012). Under different plant–attacker conditions, simultaneous inductions of SA and JA signalling do not always lead to predicted outcomes, highlighting the complexity of the hormonal interactions.
Crosstalk between the ERF- and MYC-branches of the JA pathway
As described above, defence responses to necrotrophic pathogens and herbivorous insects are regulated by distinct branches of the JA signalling pathway: the ERF- and the MYC-branch, respectively (Lorenzo & Solano, 2005; Kazan & Manners, 2008; Pieterse et al., 2012). Transcriptional changes in response to these diverse types of attackers show limited overlap, suggesting that the context in which the induced JA signal is perceived is crucial in tuning the JA response (De Vos et al., 2005). The hormones ET and ABA determine the direction in which the JA pathway is steered, towards the ERF- or the MYC-branch, respectively (Verhage et al., 2011; Pieterse et al., 2012). Activation of the ERF-branch resulted in reduced expression of the MYC-branch marker gene VSP2, whilst silencing of ORA59 or mutating the ET pathway caused enhanced levels of VSP2 expression (Lorenzo et al., 2004; Verhage et al., 2011). Reciprocally, activation of the MYC-branch suppressed transcription of the ERF-branch marker gene PDF1.2, whereas mutation of MYC transcription factors genes or ABA signalling components led to enhanced expression of PDF1.2 and the ERF transcription factor gene ORA59 (Verhage et al., 2011). These data clearly indicate a mutually antagonistic interaction between the different branches of the JA pathway.
Ecological costs of defence signalling
Fitness costs associated with induced defence arise from allocation and ecological costs (Fig. 1). Allocation costs are incurred when resources are allocated to resistance instead of to growth and reproduction (Heil & Baldwin, 2002) and have been described in the previous sections on SA- and JA-inducible defences. Ecological costs arise when defence-induced plants have altered abilities to interact with their biotic and abiotic environment (adjusted from Heil, 2002). For example, induction of the JA defence pathway resulted in reduced numbers of visitations by beneficial pollinators (Strauss et al., 2002) and JA-regulated herbivore-induced plant volatiles (HIPVs) alter the interaction of a plant with herbivores, carnivores and competing plants (Dicke & van Loon, 2000). Also above- and below-ground interactions with beneficial microbes can be affected by the activation of defence in foliar tissue. Exogenous application of SA to the soil inhibited the growth and formation of root nodules in the Rhizobium–legume symbiosis, whereas growth of Rhizobium cells itself was not affected by SA (Sato et al., 2002; Mabood & Smith, 2007). Furthermore, de Román et al. (2011) found that foliar application of acibenzolar-S-methyl (ASM), a functional analogue of SA, to soybean led to a transient reduction in arbuscular mycorrhizal colonization of roots. Negative effects of foliar herbivory on colonization of the roots by mycorrhizal fungi have also been reported (Barber et al., 2012). In contrast, recruitment of soilborne beneficial microbes upon stress induction in the leaves has also been shown (Fig. 1). Beneficial Bacillus subtilis bacteria were recruited to the rhizosphere upon foliar infection of Arabidopsis with the bacterial pathogen Pseudomonas syringae (Rudrappa et al., 2008). Moreover, foliar application of JA and wounding of the leaves of Medicago truncatula resulted in enhanced JA signalling and enhanced mycorrhization by Rhizophagus irregularis (formerly known as Glomus intraradices; Landgraf et al., 2012). These results show the importance of testing effects of altered defences in plants under realistic environmental conditions, because otherwise relevant ecological costs might be missed. The net effect of induced defence signalling on plant fitness strongly depends on the community context.
Ecological costs as a result of hormonal crosstalk are becoming increasingly recognized. Whereas hormonal crosstalk may be advantageous for the plant to keep allocation costs in check, evidence is accumulating that crosstalk at the level of gene expression is translated into crosstalk at the level of resistance. When plants encounter multiple attackers simultaneously or successively, the induction of a hormone signalling route might elevate the resistance to one attacker, but at the same time hormonal crosstalk can decrease the resistance to another attacker (Pieterse et al., 2012). Furthermore, as a common virulence strategy, successful pathogens and insects can hijack crosstalk mechanisms by targeting plant hormone biosynthesis and perception to rewire immune signalling, rendering the plants more susceptible (Fig. 1; Grant & Jones, 2009; Verhage et al., 2010; Pieterse et al., 2012). Ecological costs of crosstalk between different hormones are described in the next sections, focusing on crosstalk between SA and JA signalling, and between the ERF- and MYC-branches of JA signalling.
Ecological costs of SA–JA crosstalk
Many examples of ecological costs of SA–JA crosstalk have been described. For instance, in Arabidopsis infection by the hemibiotrophic pathogen P. syringae leads to induction of the SA pathway, resulting in an effective resistance response against this pathogen. However, through SA–JA crosstalk mechanisms JA signalling is suppressed, which renders the infected leaves more susceptible to the necrotrophic fungus Alternaria brassicicola (Spoel et al., 2007). Similarly, induction of SA signalling in Arabidopsis by exogenous application of SA inhibited JA-induced resistance to the generalist herbivores Spodoptera exigua and Trichoplusia ni (Cipollini et al., 2004; Cui et al., 2005). In tobacco, Manduca sexta larvae consumed up to 2·5-times more leaf tissue from plants exhibiting increased SA signalling after inoculation with TMV than from mock-treated plants (Preston et al., 1999). Furthermore, reduced SA signalling in Arabidopsis genotypes NahG and npr1 was correlated with reduced feeding by T. ni in comparison to wildtype plants (Cui et al., 2002). Crosstalk the other way around, namely of JA on SA, was also effective on the level of disease resistance: the JA-insensitive mutant coi1 showed enhanced expression of SA-dependent defences and enhanced resistance to P. syringae (Kloek et al., 2001).
Several pathogens have evolved ways to hijack host crosstalk mechanisms as a virulence strategy. One of the best-studied examples is the production of coronatine by P. syringae. Coronatine is a pathogen-derived functional and structural mimic of JA-Ile that suppresses SA signalling, thereby promoting susceptibility to this pathogen (Kloek et al., 2001; Brooks et al., 2005; Cui et al., 2005; Zheng et al., 2012). Furthermore, the necrotrophic pathogen Botrytis cinerea was shown to produce an exopolysaccharide that acts as an elicitor of the SA pathway and causes suppression of the JA pathway and consequently promotes pathogen growth (El Oirdi et al., 2011). Likewise, nymphs of the phloem-feeding silverleaf whitefly Bemisia tabaci activated SA-responsive gene expression in Arabidopsis, thereby suppressing the JA signalling pathway. This was shown to be associated with accelerated nymphal development, suggesting that the nymphs of B. tabaci can rewire the plant's immune signalling network to their own benefit (Zarate et al., 2007). Additionally, eggs of Pieris brassicae butterflies have been reported to induce SA signalling upon oviposition, which suppresses JA signalling and provides an advantage for the freshly hatched larvae (Bruessow et al., 2010).
Ecological costs of crosstalk between the ERF- and MYC-branches
Indications of effects of crosstalk between the ERF- and MYC-branches on the level of resistance to insects and necrotrophic pathogens have come mainly from studies with Arabidopsis mutants affected in one of the branches. In a two-choice setup, larvae of the specialist insect herbivore Pieris rapae preferred to feed from Arabidopsis genotypes that highly expressed the ERF-branch of the JA pathway, such as MYC2 mutant jin1 plants and ORA59-overexpressing plants, over wildtype plants that highly expressed the MYC-branch upon feeding by the caterpillars (Verhage et al., 2011). Furthermore, the jin1 mutant and the ABA biosynthesis mutant aba2 were more resistant to the necrotrophic pathogens B. cinerea, Plectosphaerella cucumerina and Fusarium oxysporum due to a potentiated expression of the ERF-branch in these mutants (Anderson et al., 2004; Lorenzo et al., 2004; Nickstadt et al., 2004; Adie et al., 2007). A comparable mechanism may underlie findings from the early 1990s with black bean aphids, Aphis fabae, that displayed a higher growth rate and fecundity on bean leaves infected with the necrotrophic pathogen Botrytis fabae, compared to uninfected leaves (Zebitz & Kehlenbeck, 1991).
Indications of hijacking of hormone crosstalk mechanisms by attackers have been found for ERF–MYC crosstalk as well. For example, application of the oral secretion of P. rapae caterpillars into wounded leaf tissue stimulated expression of the ERF-branch, suggesting that insecticidal compounds have the potential to manipulate the plant response towards the caterpillar-preferred ERF-branch (Verhage et al., 2011; Fig. 1). Together, these data support the existence of high ecological costs of crosstalk on the level of resistance to pathogens and pests.
Priming for enhanced defence
Besides the fact that inducible defences involve fitness costs, the inducibility of defences comes with an unsafe time slot between attack and the expression of defences. This might have been a driving force for the development of a sophisticated, cost-effective way to activate inducible defence responses, namely by priming (Fig. 1). Plants that are primed for enhanced defence do not express defences in the absence of an attacker, but show a faster and stronger activation of cellular defence responses upon attack compared to non-primed control plants (Conrath et al., 2002, 2006; Frost et al., 2008). Prior activation of defences is not a prerequisite for the primed state, which makes priming a cost-efficient form of induced immunity. Another benefit of priming is that it offers enhanced resistance against a broad spectrum of attackers. Multiple inducers of priming for defence have been identified, including beneficial microbes, pathogens and herbivorous insects, but also chemical elicitors and wounding (Conrath et al., 2002, 2006). The diverse forms of priming are described in the sections below.
Various mechanisms underlying priming have been reported. Inactive cellular proteins that play a role in cellular signal amplification have been shown to accumulate in primed plants where they remain dormant until activation by stressors, resulting in an accelerated response. Examples of such dormant signal transducers implicated in priming are transcription factors and mitogen-activated protein kinases (MPKs; Pozo et al., 2008; Beckers et al., 2009; Van der Ent et al., 2009). Chromatin modifications at the promoters of priming-associated genes have also been implicated in the regulation of the primed state (Jaskiewicz et al., 2011; Luna et al., 2012; Rasmann et al., 2012). Priming has in several cases been demonstrated to be transferred to the plant's offspring, which in some cases was associated with epigenetic changes, allowing plants to retain memory of a threatening situation into one or more successive plant generations (Luna et al., 2012; Pieterse, 2012; Rasmann et al., 2012; Slaughter et al., 2012).
SA-dependent systemic acquired resistance
Systemic acquired resistance (SAR) is a well-studied form of induced resistance in which priming is thought to play an important role (Conrath et al., 2002, 2006; Durrant & Dong, 2004; Vlot et al., 2009). SAR is activated locally and systemically upon infection by a (hemi-)biotrophic pathogen and enhances resistance of uninfected plant parts to subsequent infection by the same or a broad range of other pathogens. SAR was first described by Ross (1961), who demonstrated that uninfected leaves of TMV-infected tobacco plants became more resistant to subsequent infection by TMV. SAR is associated with endogenous accumulation of SA, both at the site of infection and in healthy systemic tissues. Mutant plants that are impaired in SA signalling, including the npr1 mutant, are incapable of developing SAR, indicating that SAR requires SA signalling (Durrant & Dong, 2004). Recently, several long-distance signals involved in the communication between SAR-induced tissue and systemic SAR-expressing tissue have been identified (Vlot et al., 2009; Dempsey & Klessig, 2012; Shah & Zeier, 2013). SAR is accompanied by priming of SA-dependent defences, resulting in potentiated expression of SA-responsive genes, such as PR1 (Mur et al., 1996; van Wees et al., 1999). Additionally, SA-independent callose deposition is primed during SAR, resulting in accelerated strengthening of the cell wall at the site of pathogen penetration (Ton & Mauch-Mani, 2004). Exogenous application of low concentrations of SA or BTH does not directly activate defences, but primes plants for enhanced expression of cellular defences after pathogen attack (Conrath et al., 2002, 2006). This indicates that SA-mediated priming is an intrinsic part of pathogen-activated SAR.
JA/ET-dependent induced systemic resistance
Induced systemic resistance (ISR) triggered by nonpathogenic microbes is another well-studied form of induced resistance in which priming plays an important role. Plant roots contain a large number of rhizosphere-associated microbes, called the root microbiome, that aid in plant growth and reproduction (Berendsen et al., 2012). Beneficial ISR-inducing microbes include soilborne plant growth-promoting rhizobacteria (PGPR) and fungi (PGPF) as well as mycorrhizal fungi (Van Hulten et al., 2010; Zamioudis & Pieterse, 2012). Like SAR, ISR offers a broad-spectrum resistance to foliar and root pathogens, but on top of that it is also effective against certain herbivores (van Wees et al., 2008; Pineda et al., 2010). In contrast to SAR, rhizobacteria-mediated ISR was shown to be independent of SA and instead requires JA, ET and ABA signalling (van Wees et al., 2008; Van der Ent et al., 2009; Pieterse et al., 2012). Most SA-signalling components in Arabidopsis proved to be dispensable for ISR, except the NPR1 protein that acts downstream of JA and ET in ISR (Pieterse et al., 1998; van Wees et al., 2000).
Large-scale gene expression analysis revealed that induction of ISR can occur with only minor changes in gene expression in the leaves (Verhagen et al., 2004; van Wees et al., 2008; Van der Ent et al., 2009). However, upon encounter with an invader, ISR-expressing plants show enhanced and accelerated expression of JA- and ET-regulated genes and accumulation of callose-rich papillae at the site of infection (Verhagen et al., 2004; Pozo & Azcón-Aguilar, 2007; Pozo et al., 2008). Concordantly, ISR is predominantly effective against necrotrophic pathogens and herbivorous insects, but also offers protection against biotrophs that are sensitive to cell wall defences (Ton et al., 2002; van Oosten et al., 2008). In roots, the induction of ISR results in altered expression of several genes, including the transcription factor gene MYB72 (Verhagen et al., 2004) that emerged as an important component of ISR, as myb72 mutants were abolished in their ability to express ISR (Van der Ent et al., 2008; Segarra et al., 2009).
Priming has most often been studied in the context of plant–pathogen interactions, but plants can also be primed by signals associated with herbivore feeding. Tissue damage can lead to wound-induced resistance (WIR), which can be induced in neighbouring plants via HIPVs or in systemic leaves of the same plant via internal signals or externally via HIPVs (Heil & Silva Bueno, 2007; Frost et al., 2008). These HIPVs prime the plant for JA-inducible defence mechanisms, or act as signals to attract parasitic and predatory insects to combat attacking herbivores (Baldwin et al., 2006; Ton et al., 2007). In addition to the effect on insect performance, WIR can also prime the plant for enhanced resistance against microbial pathogens (De Vos et al., 2006).
Crosstalk during priming
Despite the shared dependency on the NPR1 protein, distinct signalling cascades are important for SAR and ISR, requiring SA or JA and ET, respectively (Pieterse et al., 1998; van Wees et al., 2000). Crosstalk between the SA and JA pathways could entail high ecological costs, as described in a previous section, but is this also true for SA- and JA-dependent priming? Simultaneous induction of SAR and ISR was shown to result in an additive effect on the level of resistance against P. syringae (van Wees et al., 2000). In plants mutated in either the SA or JA signalling pathway, this additive effect was not found. Furthermore, induction of ISR did not affect expression of SAR-induced PR1. Thus, there is no evidence for SA–JA crosstalk during simultaneous activation of SAR and ISR. The Arabidopsis accession Bur-0 is constitutively primed for both PR1 and PDF1.2 expression upon exogenous application of SA and JA, respectively. Consequently, Bur-0 is more resistant to the hemibiotrophic pathogen P. syringae and also to the necrotrophic pathogen P. cucumerina (Ahmad et al., 2011). Together, these results suggest that there is no SA–JA crosstalk during priming. However, when SAR was inherited in the next generation, the progeny showed a weaker induction of the JA-inducible gene PDF1.2, which was accompanied by increased susceptibility to the necrotrophic pathogen B. cinerea. This was associated with a chromatin modification at the PDF1.2 promoter that is associated with transcriptional silencing (Luna et al., 2012). Mycorrhizal fungi and PGPR have been reported to induce resistance to leaf chewing insects but also to increase susceptibility to phloem feeders (Koricheva et al., 2009; Pineda et al., 2010). These findings were recently expanded with studies on the effects of the ISR-inducing Pseudomonas fluorescens on the attraction of parasitoids by volatiles of aphid-infested plants. Pineda et al. (2013) reported that ISR-expressing plants attracted less parasitoids. It is unknown whether there is a role for SA–JA crosstalk in this ecological cost of ISR.
Benefits: limited allocation costs of priming
Priming of Arabidopsis with low concentrations of β-amino-butyric acid (BABA) was shown to have only marginal effects on plant growth and seed production in the absence of pathogens, suggesting that there are no or only limited allocation costs associated with priming (Van Hulten et al., 2006). In the presence of pathogens, a clear fitness advantage was observed for primed plants over non-primed plants and plants expressing constitutive defences. Walters et al. (2008) also found that priming in barley by saccharin did not incur fitness costs, both in greenhouse and field conditions. Furthermore, there are several studies that show that PGPR not only prime for defence but also increase plant growth and seed production, although these traits are not causally related (Raupach & Kloepper, 1998; Zehnder et al., 2001; Zamioudis et al., 2013). These results indicate that there are fitness benefits for plants that interact with PGPR whilst no allocation costs are associated with this. Ahmad et al. (2011) found that the Arabidopsis accession Bur-0 is constitutively primed for enhanced defence against pathogens and insects, without growth restraints. Together these results show that the benefits of priming outweigh the marginal costs of it in environments in which disease occurs. Therefore, priming for enhanced defence seems to be a very useful tool for application in crop protection.
Inducible defences and trade-offs with crop protection
In complex natural environments, plants encounter a multitude of pathogens and pests. In agriculture this leads to tremendous annual crop losses, representing a total value of over €450 billion worldwide. Allocation and ecological costs of (induced) plant defences are a major problem for the implementation of induced resistance in agriculture (Walters & Heil, 2007). To successfully use inducible defences in crop protection, a functional understanding of the physiological and ecological consequences of the induced state is indispensable and demands more research (Bostock, 2005; Koornneef & Pieterse, 2008). Hormonal pathway crosstalk presents a challenge for translating fundamental knowledge into crop disease resistance traits. Plants often have to deal with simultaneous or subsequent attack by very different attackers. Genetic traits that are associated with contrasting resistance mechanisms to different attackers, for example SA signalling that causes elevated resistance to biotrophs, but reduced resistance to necrotrophs, can greatly impact plant fitness and thus crop yield. The extensive interactions between different hormone signalling routes that are activated upon encounter of a plant with an attacker, and the concentration-, space- and time-dependent context in which this occurs, need to be dissected. However, to fully comprehend the plant's immune system, so that this knowledge can be applied to sustainable agriculture, plants need to be studied in agricultural and natural environments as well, because predictions on hormonal interactions and fitness effects during the encounter of plants with their biotic and abiotic environment do not always lead to the predicted outcomes (Clarke et al., 2009; Ritsema et al., 2010; Ballaré, 2011; Cerrudo et al., 2012).
Breeders usually select for plant traits, such as yield and quality, while disease resistance is rarely in the top three of selected traits (Brown, 2002). Elevated resistance is usually correlated with detrimental effects on yield, but genetically and physiologically it is possible to heighten disease resistance while conserving plant fitness (Bechtold et al., 2010). Research into the mechanisms of how plants successfully combine high disease resistance and high yield could open up new possibilities for the development of valuable crop species. Furthermore, priming for enhanced defence provides also an opportunity to protect plant species while minimizing the costs of resistance (Van Hulten et al., 2006). Simultaneous activation of ISR and SAR provides an attractive tool for the improvement of crop species (van Wees et al., 2000). Overall, understanding of the functioning of the complex defence signalling network and the fitness costs involved is necessary for successful application of defence traits in crops. Therefore, molecular biologists and ecologists should join forces to place molecular mechanisms of inducible plant defences in an ecological perspective.
The authors of this review are supported by VIDI grant no. 11281 of the Dutch Technology Foundation STW, which is part of the Netherlands Organization of Scientific Research, and ERC Advanced Investigator Grant no. 269072 of the European Research Council.
Conflicts of interest
The authors have no conflicts of interest to declare.