Plant secondary metabolites as defenses, regulators, and primary metabolites: the blurred functional trichotomy

23 The plant kingdom produces hundreds of thousands of low molecular weight organic 24 compounds. Based on the assumed functions of these compounds, the research 25 community has classified them into three overarching groups: primary metabolites, 26 which are directly required for plant growth; secondary (or specialized) metabolites, 27 which mediate plant–environment interactions; and hormones, which regulate 28 organismal processesand metabolism. For decades, this functional trichotomy of 29 plant metabolism has shaped theory and experimentation in plant biology. However, 30 exact biochemical boundaries between these different metabolite classes were never 31 fully established. A new wave of genetic and chemical studies now further blurs these 32 boundaries by demonstrating that secondary metabolites are multifunctional; they 33 can function as potent regulators of plant growth and defense as well as primary 34 metabolites sensu lato. Several adaptive scenarios may have favored this functional 35 diversity for secondary metabolites, including signaling robustness and cost-effective 36 storage and recycling. Secondary metabolite multifunctionality can provide new 37 explanations for ontogenetic patterns of defense production and can refine our 38 understanding of plant–herbivore interactions, in particular by accounting for the 39 Plant Physiology Preview. Published on July 7, 2020, as DOI:10.1104/pp.20.00433 Copyright 2020 by the American Society of Plant Biologists https://plantphysiol.org Downloaded on April 17, 2021. Published by Copyright (c) 2020 American Society of Plant Biologists. All rights reserved.


Abstract 23
The plant kingdom produces hundreds of thousands of low molecular weight organic 24 compounds. Based on the assumed functions of these compounds, the research 25 community has classified them into three overarching groups: primary metabolites, 26 which are directly required for plant growth; secondary (or specialized) metabolites, 27 which mediate plant-environment interactions; and hormones, which regulate 28 organismal processesand metabolism. For decades, this functional trichotomy of 29 plant metabolism has shaped theory and experimentation in plant biology. However, 30 exact biochemical boundaries between these different metabolite classes were never 31 fully established. A new wave of genetic and chemical studies now further blurs these 32 boundaries by demonstrating that secondary metabolites are multifunctional; they 33 can function as potent regulators of plant growth and defense as well as primary 34 metabolites sensu lato. Several adaptive scenarios may have favored this functional 35 diversity for secondary metabolites, including signaling robustness and cost-effective 36 storage and recycling. Secondary metabolite multifunctionality can provide new 37 explanations for ontogenetic patterns of defense production and can refine our 38 understanding of plant-herbivore interactions, in particular by accounting for the 39

Introduction 47
Plants can use simple, inorganic precursors to synthesize a large diversity of low 48 molecular weight organic compounds. This synthetic capacity helps plants to colonize 49 diverse and challenging environments. Low molecular weight organic compounds are 50 commonly separated by perspective function into primary metabolites, secondary 51 metabolites (also called specialized metabolites or natural products), and plant 52 hormones ( Figure 1) (Taiz et al., 2015). Primary metabolites are highly conserved 53 and directly required for the growth and development of plants (Fernie and Pichersky, 54 2015). Secondary metabolites, including major groups such as phenolics, terpenes, 55 and nitrogen-containing compounds, are often lineage-specific and aid plants to 56 interact with the biotic and abiotic environment (Hartmann, 2007). Finally, plant 57 hormones are defined as small compounds that regulate organismal processes, 58 including the production of the other metabolites, by interacting with receptor proteins 59 (Davies, 2004). 60 Despite the fact that definitions of secondary metabolites are inherently diffuse 61 (Hartmann, 2007;Pichersky and Lewinsohn, 2011;Davies, 2013), the distinction 62 between primary metabolites, secondary metabolites, and plant hormones has found 63 its way into textbooks and shapes our thinking in plant biology to this day. An 64 illustrative example is the field of plant-herbivore interactions, where major efforts 65 have gone into disentangling how plants protect their primary metabolites (serving as 66 nutrients for herbivores) using secondary metabolites (serving as defenses for 67 plants), and how adapted herbivores manage to extract primary metabolites while 68 avoiding the negative effects of secondary metabolites (Awmack and Leather, 2002;69 Howe and Jander, 2008; Zhou et al., 2015;Erb and Reymond, 2019). In this context, 70 plant hormones are investigated as regulators of primary and secondary metabolism, 71 defense, and resistance that may be manipulated by adapted herbivores (Howe and 72 Jander, 2008;Schuman and Baldwin, 2016;Stahl et al., 2018), similar to pathogens 73 (Kazan and Lyons, 2014). The biochemical co-evolutionary arms-race theory (Ehrlich 74 and Raven, 1964), a key concept in plant-herbivore interactions (Berenbaum and 75 Zangerl, 2008;Jander, 2018), postulates that plant secondary metabolites evolve in 76 response to herbivore pressure, resulting in the evolution of resistance mechanisms 77 in herbivores. The resulting arms race is thought to drive the diversity of plant 78 secondary metabolites and insect herbivores (Futuyma and Agrawal, 2009). 79 Over the last decades, the distinction between primary metabolites, secondary 80 metabolites, and plant hormones has proven a useful approximation. However, the 81 emergence of a more detailed understanding of plant metabolism may require us to 82 revisit this functional partitioning (Neilson et al., 2013;Maag et al., 2015;83 Kliebenstein, 2018;Pichersky and Raguso, 2018;Zhou et al., 2018). In particular, an 84 increasing number of genetic and functional studies on plant secondary metabolites 85 are blurring the functional trichotomy by showing that plant secondary metabolites 86 can have regulatory functions and serve as precursors for primary metabolites. In this 87 review, we discuss this evidence, mostly focusing on examples that rely on the use of 88 natural knockout variants, mutants, and transgenic plants altered in their capacity to 89 produce certain secondary metabolites in combination with chemical 90 complementation assays to demonstrate activity of the metabolites. We illustrate that 91 for an increasing number of plant secondary metabolites, a strict functional 92 separation from regulators and primary metabolites may not do them justice and 93 possibly hinders our progress in understanding their roles for plant survival in hostile 94 environments. 95

Integration of plant secondary metabolites into regulation and metabolism 96
Early evidence for metabolic integration of secondary metabolites 97 In 1977, David Rhoades studied the properties of creosotebush (Larrea spp.) leaf 98 resin. He found that the resin, which contained high levels of phenylpropanoid 99 derivatives (lignans), absorbed UV radiation, reduced evaporative water loss across 100 cellulose membranes, and had the capacity to form complexes with proteins, thus 101 possibly reducing the digestibility of plant materials for herbivores (Rhoades, 1977). 102 Rhoades thus postulated that "…any chemical system possessed by a plant must 103 necessarily be integrated into the total metabolic scheme and multiple functions are 104 to be expected". In other words, Rhoades proposed that secondary metabolites are 105 not endpoints, but integrated components of plant metabolism, and may, by 106 consequence, take on any number of functions, similar to other plant metabolites. 107 Indeed, evidence was emerging at that time that secondary metabolites may regulate 108 growth and defense, as exogenously applied flavonoids could modulate polar auxin 109 transport and catabolism (Stenlid, 1963;Stenlid, 1976), glucosinolate breakdown 110 products could replace auxins in inducing hypocotyl bending (Hasegawa et al., 111 1986), and induced volatiles promoted resistance and defense regulation in 112 neighboring trees (Baldwin and Schultz, 1983;Rhoades, 1983 aphid-and chitosan-induced callose deposition and callose induction is rescued by 124 the addition of DIMBOA or DIMBOA-Glc (Ahmad et al., 2011;Meihls et al., 2013). In 125 both cases, the capacity to regulate callose is structurally specific and depends on 126 the modification of the indole-derived ring. In Arabidopsis, indol-3-127 ylmethylglucosinolate, which lacks a methylated hydroxy-group on the aromatic ring, 128 Interestingly, glucosinolates and benzoxazinoids also seem to regulate the 141 accumulation of other secondary metabolites (Hemm et al., 2003;Kim et al., 2015;Li 142 et al., 2018a). In Arabidopsis, mutants that are defective in the atypical myrosinase 143 PEN2 release lower amounts of Trp-derived metabolites such as camalexin upon 144 flg22 treatment (Frerigmann et al., 2016) and infection by Pseudomonas syringae 145 (Stahl et al., 2016). Furthermore, mutants defective in the CYP83B1 enzyme required 146 for indole glucosinolate production also show lower accumulation of the 147 phenylpropanoid sinapoylmalate (Kim et al., 2015). The phenylpropanoid phenotype 148 is rescued in mutants that no longer produce the substrate of CYP83B1, indole-3-149 acetaldoxime (Kim et al., 2015), suggesting that it may be the aldoxime 150 overaccumulation rather than the lack of downstream glucosinolates that suppresses 151 sinapoylmalate. Suppressor screens showed that the phenylpropanoid phenotype is 152 also absent in plants that have mutated MEDa/b genes, which encode key 153 components of a large multisubunit transcriptional complex that regulates 154 phenylpropanoid biosynthetic genes (Kim et al., 2015;Dolan et al., 2017). A recent 155 study demonstrates that a group of Kelch Domain F-Box (KFB) genes that are 156 involved in PAL inactivation (Zhang et al., 2013) defective in their capacity to  176  produce volatile indole are unable to prime their systemic tissues to rapidly release  177 terpenes upon herbivore attack (Erb et al., 2015). Adding indole to the headspace of 178 maize plants restores this priming phenotype (Erb et al., 2015). Rice (Oryza sativa) 179 plants also respond to indole through priming of early defense signaling elements 180 such as the map kinase OsMPK3 (Ye et al., 2019). Transgenic plants that are 181 deficient in OsMPK3 expression are no longer responsive to indole, suggesting that 182 indole acts via the priming of early defense signaling (Ye et al., 2019). In Arabidopsis, 183 geranylgeranyl reductase1 mutants are defective in systemic acquired resistance 184 against Pseudomonas syringae (Riedlmeier et al., 2017). Adding the pathogen-185 induced volatiles αand β-pinene to the headspace of the mutant restores resistance, 186 with the response depending on intact salicylic acid signaling and the AZELAIC ACID 187 INDUCED (AZI1) gene (Riedlmeier et al., 2017 LOX2 mutation leads to stronger expression of defense-related genes in neighbors 197 than wild-type plants, suggesting that volatiles can also suppress defenses (Paschold 198 et al., 2006). 199 In summary, at least five classes of secondary metabolites (glucosinolates, 200 benzoxazinoids, terpenes, aromatics, and green-leaf volatiles) are now confirmed to 201 act as potential regulators of in planta defense. It is exciting to speculate that there 202 are many other secondary metabolites that play similar regulatory roles. An important 203 gap of knowledge is the mechanism by which secondary metabolites regulate 204 defenses. As many of the secondary metabolites are chemically reactive (Farmer and 205 Davoine, 2007;Hadacek et al., 2010), it is possible that they act indirectly by 206 depleting detoxification enzymes, thus triggering the accumulation of known signaling 207 molecules such as reactive oxygen species (ROS) (Khokon et al., 2011). However, 208 as discussed below, secondary metabolites may also have hormone-like properties 209 by binding to specific receptor proteins (Katz et al., 2015). More work on the targets 210 of secondary metabolites in planta is clearly warranted and would help to clarify the 211 ecological and evolutionary context of their capacity to regulate defenses. 212

Secondary metabolites as regulators of growth and development 213
Plants regulate their growth dynamically and often reduce their investment into 214 growth and development upon herbivore-or pathogen attack. This reduction in 215 growth is thought to be largely due to the reconfiguration of a plant's signaling 216 network rather than a lack of resources (Kliebenstein, 2016;Machado et al., 2017;217 Guo et al., 2018). Strikingly, plant secondary metabolites and their breakdown 218 products are being (re)-discovered as plant growth modulators, thus adding another 219 layer of regulation to growth-defense patterns. Again, glucosinolates provide a 220  (Salehin et al., 2019). Iaa5,6,19 mutants fail to close their stomata upon drought 237 stress, a phenotype that can be reverted by adding 4-MSOB (Salehin et al., 2019). 238 Together with the finding that glucosinolate biosynthesis and activation mutants are 239 less tolerant to drought (Salehin et al., 2019), and that glucosinolate breakdown 240 products can trigger stomatal closure in Arabidopsis and Vicia faba (Khokon et al., 241 2011;Hossain et al., 2013), these results provide evidence that aliphatic 242 glucosinolates are involved in stomatal regulation. Interestingly, glucosinolate-243 mediated stomatal regulation requires a functional ROS receptor kinase (GHR1) 244 (Salehin et al., 2019). Given that the myrosinase TGG1 accumulates in guard cells 245 and is required for stomatal regulation (Zhao et al., 2008) with the substantial variation in glucosinolate biosynthesis within species, creates a 271 wealth of metabolic networks and phenotypes, which can be acted upon by natural 272 selection. It is tempting to speculate that this diversity is a reflection of the highly 273 diverse habitats and environments that a single species can inhabit and may provide 274 adaptive potential beyond conserved hormonal pathways. 275 In addition to glucosinolates, flavonoids are implicated in regulating plant growth, 276 development, and environmental responses. Exogenously applied flavonoids have 277 long been known to modulate auxin transport (Stenlid, 1976). Evidence that 278 flavonoids may also act as endogenous growth regulators came from an Arabidopsis 279 chalcone synthase mutant, transparent testa (tt4). However, the oxidation state of a cell can directly influence signaling by altering 302 disulfide bridge formation or other protein modifications. Thus, it is possible that 303 flavonols also function as signals and further work is needed to differentiate between 304 these hypotheses. 305 Other secondary metabolites may also regulate plant development. Diploid oat sad2 306 mutants that overproduce the triterpene β-amyrin produce shorter roots and 307 significantly more root hairs than wild-type plants, phenotypes which are absent in 308 other mutants of the pathway that do not overproduce β-amyrin (Kemen et al., 2014). 309 However, this phenotype cannot be phenocopied by adding β-amyrin to roots, 310 possibly because its activity requires specific spatiotemporal accumulation patterns 311 (Kemen et al., 2014) distinguishable from mechanisms normally assigned to plant hormones ( Figure 2). 324 Whereas some of these secondary metabolite regulators are ancient and highly 325 conserved (e.g. flavonoids, terpenes), others evolved more recently (e.g. 326 glucosinolates and benzoxazinoids) and are restricted to specific plant families. 327 Plants thus have both a conserved and a unique, variable, and flexible repertoire of 328 regulators at their disposition to adjust growth and development, which likely 329 contributes to their potential to colonize variable and challenging habitats. 330

Secondary metabolites as primary metabolites 331
If secondary metabolites can regulate growth, development, and defense, can they 332 also function as primary metabolites? Whereas primary metabolites are highly 333 conserved, secondary metabolites evolve dynamically and are inherently variable in 334 structure and production (Wink, 2008). This rapid evolution would seem to complicate 335 their integration into the most fundamental workings of plant metabolism because it 336 would require a rapid evolution of enzymes to connect these novel structures into the 337 more conserved metabolic pathways. However, evidence for secondary metabolites 338 that are not strictly essential, but nevertheless contribute to primary metabolism, is 339 emerging. In Arabidopsis, plants with mutations in the flavonoid pathway upstream of 340 the FLAVANONE-3-HYDROXYLASE (F3H) show a reduction in the respiratory 341 cofactor ubiquinone (coenzyme Q) (Soubeyrand et al., 2018). Ubiquinone levels can 342 be restored by adding dihydrokaempferol or kaempferol to the mutants. Labelling 343 experiments demonstrate that the aromatic ring of kaempferol is integrated into 344 ubiquinone, and that heme-dependent peroxidases likely use kaempferol to produce 345 4-hydroxybenzoate as a substrate for ubiquinone (Soubeyrand et al., 2018). 346 The integration of flavonoids into primary metabolism is perhaps not surpising, as 347 they represent one of the oldest and most conserved classes of secondary 348 metabolites (albeit with substantial inter-specific variation in glycosylation patterns). 349 Flavonoid evolution precedes the emergence of many innovations in plant primary 350 metabolism, such as C 4 photosynthesis. Whether younger, more specialized 351 secondary metabolites can act as primary metabolites is not well understood. This 352 lack of knowledge is closely related to a limited understanding of secondary 353 metabolite catabolism. Where do these compounds go when they are no longer 354 needed? One would assume that re-integrating secondary metabolites into primary 355 metabolism is beneficial for plants (Neilson et al., 2013). Such a re-integration 356 pathway has been proposed for cyanogenic glycosides (Selmar et al., 1988 deglycosylation, HCN may be assimilated into asparagine via the formation of β-358 cyanoalanine (Selmar et al., 1988 glucosinolates is at least partially controlled by conserved phytohormonal pathways 469 (Schweizer et al., 2013), plant enemies that are capable of overcoming these 470 conserved pathways may also suppress more specific regulators. Interestingly, an 471 opposite pattern has also been found for the tomato leaf spot fungus, which uses a 472 hydrolase to detoxify steroidal glycoalkaloids and benefits from the defense-473 suppressing properties of the resulting breakdown products (Bouarab et al., 2002). 474 This illustrates that specialized plant enemies may also misuse the regulatory 475 properties of secondary metabolites of their host plants. 476

Multifunctionality as a cost-saving strategy 477
Producing secondary metabolites has energetic and metabolic costs (Gershenzon, 478 1994 to explore the role of secondary metabolite reintegration as a cost-saving strategy. 496 Another way to minimize costs is to utilize the same secondary compound for 497 multiple purposes (Neilson et al., 2013). As many secondary compounds are 498 chemically reactive, they need to be managed by the plant through (potentially costly) 499 storage, inactivation, and/or resistance mechanisms, including specialized cells, 500 ducts, and glands (Sirikantaramas et al., 2008). By employing the same compound 501 class for multiple purposes, plants may spread these fixed costs across more fitness 502 components and increase their competitiveness. Metabolic costs may also be 503 lowered by using the same biosynthetic machinery to produce different compounds 504 for different purposes. Whereas the cost-saving aspects of multifunctionality are 505 difficult to quantify, multifunctionality seems to be a widespread property of 506 secondary metabolites, as discussed above, and it is difficult for this multifunctionality 507 to evolve without benefit. 508 improve if we take their full metabolic integration and potential multifunctionality into 537 account and do not limit their considered benefits to herbivore resistance. 538

Defense metabolites in plant-herbivore interactions 539
The functional trichotomy used to define plant metabolites has also shaped our 540 understanding of how these metabolites influence plant-herbivores interactions. 541 Herbivores are assumed to forage for primary metabolites while trying to avoid the 542 negative effects of secondary metabolites through behavioral and metabolic 543 adaptations (Behmer, 2009) (Stahl et al., 2018). If we accept that secondary 544 metabolites can also be regulators and precursors of primary metabolites, then it 545 becomes conceivable that they may have similar roles in herbivores. The root-546 feeding larvae of the western corn rootworm for instance forage for iron-547 benzoxazinoid complexes to acquire iron and improve their growth, thus effectively 548 using a plant secondary metabolite as a primary metabolite (Hu et al., 2018). Several 549 other herbivores also gain more weight in the presence of plant secondary 550 metabolites (Meldau et al., 2009;Richards et al., 2012;Marti et al., 2013;Veyrat et 551 al., 2016;Wetzel et al., 2016), and it is conceivable that some of these effects may 552 be due to the capacity of the herbivores to metabolize these compounds. Recent  553 examples also hint at the possibility that plant secondary metabolites may have 554 hormonal functions in herbivores. In rice, knocking down CYP71A1, a gene 555 responsible for the production of serotonin, a monoamine neurotransmitter, reduces 556 the performance of the rice brown planthopper (Nilaparvata lugens). Adding serotonin 557 to an artificial diet enhances its performance (Lu et al., 2018), suggesting that the 558 herbivore may benefit from the hormonal properties of this plant metabolite. Plants 559 may also benefit from producing secondary metabolites that act as (de)-regulators of 560 herbivore physiology. Spinach for instance produces the molting hormone 20-561 hydroxyecdysone (Bakrim et al., 2008), which can interfere with caterpillar 562 development (Kubo et al., 1983). 563 In general terms, a plant's metabolism is shaped by a dynamic landscape of 564 environmental selection pressure; conversely, the metabolic network of herbivores is 565 shaped by the functional and chemical potential of plant metabolites within the 566 herbivore's own selection landscape. One can thus expect that, similar to what 567 Rhoades postulated for plants (Rhoades, 1977), any chemical system taken up by a 568 herbivore must necessarily be integrated into its total metabolic scheme, and multiple 569 functions of plant secondary metabolites are to be expected, some of which likely 570 mirror their multiple functions in plants (Figure 3). Specialist herbivores are known to 571 use secondary metabolites as infochemicals (e.g. foraging cues), and some also 572 sequester defenses to protect themselves against herbivore natural enemies 573 (Nishida, 2002;Opitz and Müller, 2009), in analogy to the use of these chemicals as 574 defense regulators and resistance factors in plants (Figure 3). Cabbage aphids 575 (Brevicoryne brassicae) are an illustrative example in this context, as they can 576 activate glucosinolates by producing their own myrosinases (Bridges et al., 2002;577 Kazana et al., 2007). This allows them to use glucosinolates as two-component 578 defense system against predators (Kazana et al., 2007). As glucosinolate breakdown 579 products (isothiocyanates) also increase aphid responses to alarm pheromones 580 (Dawson et al., 1987), it was proposed that aphid-released isothiocyanates may also 581 https://plantphysiol.org Downloaded on April 17, 2021. -Published by Copyright (c) 2020 American Society of Plant Biologists. All rights reserved. act as danger signals (Bridges et al., 2002). Another example where herbivores use 582 secondary metabolites for several purposes that mirror their multiple uses by plants 583 are again benzoxazinoids, which are used as defense metabolites and siderophores 584 by a specialist root herbivore in maize (Box 1). Apart from mirroring plant functions, 585 adapted herbivores can also use plant secondary metabolites for herbivore-specific 586 functions. Cyanogenic glycosides for instance can be used by specialized lepidoptera 587 as defenses and nuptial gifts (Zagrobelny et al., 2018), 2018) are promising approaches to assess plant-herbivore interaction and to identify 602 metabolite functions and effects in herbivores without prior functional assumptions. 603

Concluding remarks 604
The functional separation of plant-derived, low molecular weight organic compounds 605 into primary metabolites, secondary metabolites, and hormones has proven to be a 606 useful approximation over the last decades. However, recent work has shown that 607 several classes of plant secondary metabolites are highly integrated into plant 608 metabolism and can serve as both regulators and primary metabolites. Thus, it is 609 likely that most secondary metabolites have additional functions for plants. Taking  610 into account these additional functions (see Outstanding Questions), we can refine 611 key concepts in plant-environment interactions and improve our understanding of the 612 chemical ecology of plants and their enemies. 613

Acknowledgements 614
We would like to thank Mike Blatt for the invitation to write this Inaugural Topical 615 Review, Pierre Mateo for drawing chemical structures, and Christelle A.M. Robert,616 Clint Chapple, Jonathan Gershenzon, and two anonymous reviewers as well as the 617 twitter community for helpful comments on an earlier version of this manuscript.   including resistance, regulation and primary metabolism (see Figure 2). Recent work 640 suggests that this multifunctionality is mirrored in adapted herbivores, which also 641 employ secondary metabolites for multiple purposes, including similar and novel 642 functions. Little is known about how adapted natural enemies use secondary 643 metabolites, but multifunctional integration across three trophic levels is likely (Box 644 2). Circles represent hypothetical individual secondary metabolites (for color code, 645 refer to Figs. 1 and 2). Solid lines indicate metabolic connections within an organism. 646 Dashed lines indicate similar functions of the same compounds in different 647 organisms. 648   (Glauser et al., 2011;Maag et al., 2016), the regulation of callose deposition as a defense against aphids (Ahmad et al., 2011;Meihls et al., 2013), and iron uptake through their capacity to chelate iron in the rhizosphere (Hu et al., 2018). Recent work shows that the western corn rootworm, a highly specialized maize pest, phenocopied these functions and is able to use benzoxazinoids for multiple purposes as well. The larvae of the herbivore use benzoxazinoids as foraging cues in the rhizosphere (Robert et al., 2012), are able to extract iron from benzoxazinoid-iron complexes to maximize their own growth (Hu et al., 2018), and are able to store benzoxazinoids to resist their own enemies, namely entomopathogenic nematodes (Robert et al., 2017). Strikingly, both the plant and the herbivore have evolved to modify the structure, localization, and stressinduced modification of benzoxazinoids to fit these different functions (Ahmad et al., 2011;Maag et al., 2016;Robert et al., 2017). This example illustrates how a detailed understanding of the functional integration of plant secondary metabolites can help to unravel the chemical underpinning of plant-herbivore interactions.

BOX 2. Multi-functionality of plant secondary metabolites in tritrophic interactions
Beyond plants and herbivores, the multifunctionality of plant secondary metabolites may also extend to the third trophic level. The current view on this topic is that i) plant primary metabolites influence tritrophic interactions by directly or indirectly feeding natural enemies of herbivores (Hunter, 2003;Sarfraz et al., 2009;Ugine et al., 2019), ii) volatile secondary metabolites can serve as foraging cues (Aartsma et al., 2017;Turlings and Erb, 2018), and iii) nonvolatile secondary metabolites can reduce herbivore host quality and thereby negatively affect natural enemies of herbivores, in particular when they are sequestered by specialized herbivores (Nishida, 2002;Opitz and Müller, 2009). To what extent natural enemies of herbivores may use plant secondary metabolites as primary metabolites for their own nutrition or as hormone-like regulators is currently unknown. Natural enemies of herbivores can metabolize secondary metabolites (Sloggett and Davis, 2010;Robert et al., 2017;Sun et al., 2019a), and can evolve resistance towards their toxic effects (Fink and Brower, 1981;Rafter et al., 2017;Zhang et al., 2019). Understanding if and how plant secondary metabolites are integrated into the metabolism of herbivore natural enemies represents an important current frontier in chemical ecology. This field can benefit from current insights into the blurred functional trichotomy of plant metabolism.  Figure 1. Low molecular weight compounds in plants are functionally classified as primary metabolites, secondary metabolites or hormones. Current work on plant secondary metabolites demonstrates that many of them also have regulatory roles, and some are demonstrated precursors of primary metabolites. Note that primary metabolites and hormones also show functional overlap with the other metabolite classes (not discussed here). These findings blur the functional trichotomy of plant metabolism and call for a reassessment of ecological and evolutionary frameworks that are based on this model.   Plants use secondary metabolites for multiple purposes, including resistance, regulation and primary metabolism (see Figure 2). Recent work suggests that this multifunctionality is mirrored in adapted herbivores, which also employ secondary metabolites for multiple purposes, including similar and novel functions. Little is known about how adapted natural enemies use secondary metabolites, but multifunctional integration across three trophic levels is likely (Box 2). Circles represent hypothetical individual secondary metabolites (for color code, refer to Figs. 1 and 2). Solid lines indicate metabolic connections within an organism.
Dashed lines indicate similar functions of the same compounds in different organisms.