Phytochemical Diversity: the Sounds of Silent Metabolism Plant Science

Plant Science 176 (2009) 161–169

Contents lists available at ScienceDirect

Plant Science

journal homepage: www.elsevier.com/locate/plantsci

Review Phytochemical diversity: The sounds of silent metabolism

Efraim Lewinsohn a,*, Mark Gijzen b a Department of Vegetable Crops, Newe Ya’ar Research Center, Agricultural Research Organization, P.O. Box 1021, Ramat Yishay 30095, Israel b Agriculture and Agri-Food Canada, 1391 Sandford Street, London, Ontario N5 V 4T3, Canada

ARTICLE INFO ABSTRACT

Article history: Plants produce tens of thousands of different natural products also referred to as secondary metabolites. Received 10 June 2008 These metabolites were once thought to be the result of aberrant metabolism, or a form of transient Received in revised form 1 September 2008 storage of byproducts and intermediates thereof. Although the true role of such metabolites in plants Accepted 27 September 2008 remains mostly unknown, it is evident that plants invest a great deal of resources in synthesizing, Available online 21 October 2008 accumulating and sorting such metabolites, often produced through complex and highly regulated biosynthetic pathways operating in multiple cellular and sub-cellular compartments. There is also Keywords: growing evidence indicating that many biosynthetic pathways leading to the accumulation of plant Silent metabolism Occult metabolic capacity natural products are not fully active. Thus, occult enzymes exist, sometimes without any apparent Plant secondary metabolism endogenous substrate or function, suggesting that plants have a reservoir of metabolic capabilities that Natural product biosynthesis normally remains hidden or unused. It is often difﬁcult to accurately guess what are the actual biological Metabolic engineering roles of such enzymes solely based on bioinformatics, due to promiscuity towards substrates and the Metabolic networks relatively ease to change substrate or product speciﬁcity by introducing minor changes in sequence of the enzymes. It could be that such orphan enzyme activities are relics of a recent past that have not been fully eliminated through selection and evolution. Additionally, it could be that such occult activities possess unknown biochemical roles and coincidentally are able to accept novel substrates. We have coined the term ‘‘silent metabolism’’ to describe occult metabolic capacities present or induced in plants. A few examples illustrating silent metabolism in the terpenoid and phenylpropanoid pathways, as well as their repercussion in the metabolic engineering of plant secondary metabolism are discussed in this review. ß 2008 Elsevier Ireland Ltd. All rights reserved.

Contents

1. Introduction ...... 162 2. Occult terpenoid pathways ...... 162 2.1. Silent metabolism in carotenoid biosynthesis and degradation reactions...... 163 3. Silent metabolism in the biosynthesis of volatile ethers ...... 165 4. Silent metabolism in the phenylpropanoid pathways ...... 166 4.1. Flavonoid biosynthetic pathways provide numerous examples of silent metabolism ...... 166 4.2. Silent ﬂavonoid metabolism in soybean seed coats ...... 166 4.3. Soybean seed coat anthocyanin pigments are substrates for the seed coat peroxidase enzyme ...... 167 5. Mechanism of action of silent metabolism...... 167 5.1. Silent metabolism and relaxed selection...... 167 6. Concluding remarks...... 167 Acknowledgements...... 168 References...... 168

* Corresponding author. E-mail addresses: [email protected] (E. Lewinsohn), [email protected] (M. Gijzen).

Hello darkness, my old friend The much greater multitude of metabolites in living organisms as I’ve come to talk with you again compared to their unexpected low calculated number by the small Because a vision softly creeping number of relevant genes identified during genome sequencing projects has puzzled biologists [14]. It has been postulated that Left its seeds while I was sleeping promiscuity of enzymes, including multiple product and substrate And the vision that was planted in my brain enzyme specificity as well of changes in compartmentation patterns Still remains may also contribute to plant metabolome diversity [15]. Moreover, Within the sound of silence the more knowledge we mine on specific biosynthetic pathways and the accumulation of information on the complexity of biosynthetic (P. Simon, 1964) networks have revealed that the transcription patterns as a whole are rarely directly reflected in the proteome of an organism [16] and 1. Introduction in turn, the proteome in many cases only vaguely reflects the metabolome [17]. The more knowledge we extract on the regulation Plants produce and accumulate a vast number of different of plant natural product biosynthesis, the more puzzling it gets. natural products, also called secondary metabolites. Although tens Perhaps the old-fashioned unpopular and forgotten theory of being of thousands of secondary metabolites have been chemically natural products part of an ‘‘aberrant metabolism’’ was not so off- identified [1–3], the biological roles of most of these compounds line. It thus seems that the combination of enzymes and substrates remain obscure. Many natural compounds are of commercial and generated by serendipity may be a platform to generate the much industrial importance, imparting colors and scents to flowers, needed variation to attain effectiveness in plant defenses, and this fruits and vegetables, and are also key ingredients in medicinals process might be favored by evolution [7]. In this scenario, it is and nutraceuticals. Many studies have indicated that natural evolutionary advantageous to keep parts of the enzymatic products accumulated in plants have clear ecological roles such as machinery active through generations, and this in turn is the basis protection against predation, protection against fungal and of what we have defined as ‘‘silent metabolism’’: Occult biosynthetic bacterial diseases or against adverse climatic conditions [1,2,4– capacities that is not easily detected but readily active when 6]. Additionally, many natural products serve as signal molecules challenged. ‘‘Silent metabolism’’ can be either constitutive or to attract pollinators and seed-dispersers, or mediate pathogenic, induced by development, the environment and genetic manipula- parasitic, or symbiotic interactions [1]. Still, the biological roles of tion, and might seem to the naı¨ve spectator as if active enzymes and most specialized compounds in the plants producing them are biosynthetic pathways do not have any apparent role in the tissue unknown [3]. According to the so called ‘‘Screening Hypothesis’’ studied, but are uncovered when a change in metabolism is effected. theory [7], it is a rarity for a natural product to possess biological In this review we will discuss a few of many examples of ‘‘silent activity. Therefore, evolution favors means to increase biodiversity metabolism’’ based on our own results and on other published to provide novel compounds to be challenged by evolution. In this work. They all indicate that plants have practically a network of context and according to the ‘‘screening hypothesis’’ many of the occult biosynthetic capacities that confers them an unlimited compounds manufactured and stored by plants (and retained potential to produce a large array of different compounds that can through evolution) are part of the ‘‘screening’’ arsenal that might be activated when novel substrates become available. become important in due evolutionary time. It is therefore advantageous for occult and partial metabolic pathways to be 2. Occult terpenoid pathways retained through evolution as means to readily provide the much needed biochemical diversity. Many monoterpenes are key constituents of essential oils and Plant secondary metabolites were once thought to be the result impart unique aromas to flowers and fruits. The metabolic of aberrant or accidental metabolism or an odd form of storage of engineering of the monoterpene pathway has been the focus of nutrients. Still, numerous studies have demonstrated that plants many studies [18]. The Clarkia breweri floral gene linalool synthase have developed intricate mechanisms to modulate the production (LIS) encodes an enzyme that catalyzes the formation of (S)-linalool and accumulation of such metabolites. In most cases studied, it has from the monoterpene precursor geranyl diphosphate. Over- been found that natural products are usually formed by elaborate expression of Clarkia’s LIS in tomato fruit caused the accumulation arrays of enzymes, concertedly controlled by the expression of of (S)-linalool [19] as expected, but also the unexpected formation their respective genes. Although the chemical structures of more of 8-hydroxylinalool, a compound absent in control fruits (Fig. 1). than 40,000 different terpenes, 20,000 phenolics and 5000 The formation of 8-hydroxylinalool is thought to be mediated by alkaloids are known [1], plants utilize a relatively small number an endogenous ‘‘occult’’ hydroxylase activity able to act on (S)- of biosynthetic pathways for the production of such an impressive linalool. The actual biological role of this hydroxylase in tomato number of different compounds [3]. Thus, the vast phytochemical fruit metabolism is presently unknown, but the novel availability diversity is paradoxically attained by relatively minor changes in of substrate due to expression of a foreign gene enabled us to existing and ubiquitous metabolic pathways [8]. The reasons and uncover this activity. The Clarkia LIS gene was also over-expressed origins of such diversity in plant phytochemistry are still largely in petunia flowers. In this case, (S)-linalool was formed but was unknown. Are the huge numbers of structures present merely rapidly glycosylated (Fig. 1) [20]. In turn, when the Clarkia LIS gene there for serendipitous reasons? Or is their presence directed and was over-expressed in carnation flowers, linalool was also favored by evolution. Relatively small chemical changes in a detected in the transgenic flowers but it was apparently further molecule, can have a major impact in its biological function [9–11] metabolized to linalool oxides (Fig. 1) [21]. Thus, the over- and minor changes in protein structure might bring about to expression of an identical gene in different target tissues and profound changes in enzymatic activity or substrate specificity organisms gave rise to distinct phenotypes, according to the silent ([12,13]). metabolism present or induced in the target plant. Therefore, in Thus, very minor changes in the DNA of an organism are enough metabolic engineering experiments, one has to take into account to confer a change in a gene that will be reflected in the chemistry not only the potential functions of the genes manipulated, but also of the organism that might provide adaptive advantages to the the patterns of the silent metabolism in the target organism that variant organism, and thus favored by selection. might interact with the novel products generated. E. Lewinsohn, M. Gijzen / Plant Science 176 (2009) 161–169 163

Fig. 1. Silent linalool metabolism revealed by metabolic engineering. Overexpression of the Clarkia breweri (S)-linalool synthase (LIS) gene in different target tissues and organisms gives rise to different phenotypes. Linalool and 8-hydroxylinalool were detected in transgenic tomato fruit (top arrows [19]. Linalool glycoside was detected in petunia flowers [20] and linalool oxides were accumulated in carnation flowers (bottom arrows [21]). The phenotypes reflect gene overexpression as modulated by the existing or induced silent metabolism of the target organism.

Subsequent metabolic engineering of the terpenoid pathway in 2.1. Silent metabolism in carotenoid biosynthesis and degradation tomato fruit further demonstrated the existence of a very active reactions grid of seemingly ‘‘occult’’ enzymes. The lemon basil (Ocimum basilicum L. cv. Sweet Dani) geraniol synthase (GES) gene was over- Carotenoids are nutritionally important tetraterpenoid pig- expressed in tomato fruits in attempts to modify their aroma and ments that that are also part of the photosynthetic apparatus flavor [22]. The GES gene encodes a protein able to synthesize and are needed in human nutrition as pro-vitamin A. Attempts geraniol, a rose-scented monoterpene alcohol from geranyl to improve the nutritional value of rice have focused on diphosphate. Diversion of the plastidial terpenoid pathway of increasing the levels of b-carotene in the endosperm yielding tomato to the formation of geraniol was achieved as evidenced by what we term ‘‘Golden Rice’’. This was attained by the ectopic the accumulation of relatively high levels of geraniol in transgenic expression of PSY (phytoene synthase and CTRL1 abacterialgene tomato fruit [22]. Nevertheless, transgenic fruit accumulated at coding for carotenoid desaturase [24,25]. In the initial experi- least eleven novel additional metabolites that share a common ments, a construct directing the expression of lycopene b- chemical backbone and most likely are derived from geraniol. cyclase was also employed, to further direct carotenoid These volatiles included the monoterpene alcohols nerol and biosynthesis to b-carotene but shown to be unnecessary, as citronellol, the monoterpene aldehydes geranial, neral and transgenes overexpressing PSY and CTRL1 only displayed a citronellal, the monoterpenol esters geranyl, neryl and citronellyl similar carotenoid pattern as transgenes overexpressing the acetate, geranic and neric acid and rose oxide (Fig. 2). The three genes [24]. The product expected to be formed by the conversion of geraniol to geranial and neral is catalyzed by alcohol action of the two carotenoid biosynthesis transgenes PSY and dehydrogenase, an enzyme long known to be active in tomato CTRL1 is lycopene, a red pigment. Still, Golden Rice accumulates fruits and characterized by a broad substrate specificity able to b-carotene and xanthophyll derivatives. The absence of lyco- accept ethanol (and also geraniol) as substrates [23]. Geraniol pene in Golden Rice shows that the pathway proceeds beyond reductase activity, able to generate citronellol from geraniol and the transgenic end point and thus that an endogenous pathway NADP+, was readily detected in nontransgenic tomato fruit [22]. was also acting. Moreover, by using realtime PCR, it was Additionally, alcohol acetyltransferase activity able to act on elegantly shown that in wild-type rice endosperm the mRNA geraniol, nerol or citronellol and acetyl CoA to generate geranyl expression of the relevant carotenoid biosynthetic enzymes acetate, neryl acetate, and citronellyl acetate, respectively, was encoding phytoene desaturase, z-carotene desaturase, carotene also readily detected in ripe control tomatoes [22]. The actual role cis-trans-isomerase, b-lycopene cyclase, and b-carotene hydro- of these enzymes can only be speculated, but the above evidence xylase were all active. Only the mRNA corresponding to PSY was suggests that tomato fruits possess active but occult enzymatic virtually absent in wild type rice indicating that this was the activities that can readily accept novel substrates if the substrates limiting step for carotenoid accumulation [26]. Thus, an occult, become available. To conclude, metabolic engineering of the silent, but potentially active biosynthetic pathway was revealed plastidial terpenoid pathway revealed that a ‘‘silent’’ array of in wild-type rice. This suggests that the wild ancestor of rice enzymes is present or can be readily induced in tomato fruit. The contained a pigmented endosperm. This trait was probably metabolic flow through this occult, ‘‘silent’’ (but not silenced) grid selected against early during rice domestication and this was can be rapidly triggered and detected when a novel substrate is attained by incorporating mutants deficient in the expression of present. PSY in cultivated rice. 164 E. Lewinsohn, M. Gijzen / Plant Science 176 (2009) 161–169

Fig. 2. Silent metabolism of geraniol in tomato fruit. The geraniol produced by the overexpression of the Ocimum basilicum geraniol synthase (GES) gene was isomerized, oxidized, reduced and/or esteriﬁed to the monoterpene alcohols nerol and citronellol, the monoterpene aldehydes geranial, neral and citronellal, the monoterpenol esters geranyl, neryl and citronellyl acetate, the monoterpene acids geranic and neric acid and to rose oxide. Some of the enzymes able to catalyze these reactions can readily be measured in nontransformed fruit, including geraniol dehydrogenase, geraniol reductase, and an alcohol acetyltransferase activity readlily acting on geraniol, nerol or citronellol and acetyl CoA to generate geranyl acetate, neryl acetate and citronellyl acetate, respectively [22].

Carotenoid content not only influences fruit color, but indirectly levels of both b-carotene and its degradation product b-ionone affects the volatile composition of fruits. Pleiotropic effects on [29]. To assess the biochemical functionality of CmCCD1, the clone aroma and flavor caused by genes affecting carotenogenesis have was overexpressed in E. coli strains previously engineered to been detected by utilizing defined mutants and near-isogenic lines produce b-carotene. The CmCCD1 gene product efficiently cleaved of tomatoes that solely differ in carotenoid patterns [27,28]. b-carotene at the 9, 10 and 90,100 positions and b-ionone was Similar effects of carotenoid content and composition on the aroma released from the bacterial cultures. Further examination revealed apocarotenoid volatiles of watermelon, melon, bell peppers and that the CmCCD1 gene product cleaves many other carotenoids, carrots [27–29] and [Azulay, Y., Tadmor, Y., and Lewinsohn E., but only at positions 9, 10 and 90,100, generating geranylacetone unpublished results] have been noted. Carotenoids give rise to from phytoene; pseudoionone from lycopene; as well as a-ionone apocarotenoid (norisoprene) aroma volatiles upon oxidative and pseudoionone from d-carotene. These acyclic and monocyclic cleavage [30], and thus carotenoid composition dictates the carotenoids are normally absent in melons, still the CCD enzyme composition of apocarotenoid volatiles. This relationship is efficiently accepted them as substrates generating novel apocar- probably mediated by the oxidative cleavage of carotenoids into otenoid volatiles. Thus, it seems that the broad substrate specificity apocarotenoid volatiles [19,27]. A group of carotenoid cleavage of the CmCCD1 allows the formation of novel volatile metabolites dioxygenase enzymes (CCD’s) catalyze such cleavages and accept a when the proper carotenoid substrate is available. A similar CCD variety of carotenoids as substrates [19,27,29,31–33]. b-Carotene enzyme with a broad substrate-specificity was also discovered in is the predominant pigment in orange-fleshed melon (Cucumis tomatoes [33]. These findings easily explain the pleiotropic effects melo) varieties, while pale-green and white cultivars have much on aroma previously observed in tomato carotenoid mutants and lower b-carotene levels. In parallel, the potent odorant b-ionone, the general correlation between carotenoid composition and the 9, 10 (90,100) cleavage product of b-carotene, is present in apocarotenoid aroma chemicals observed [27,28]. However, the orange-fleshed melons and in much lower levels in pale green and question of why is an active CCD enzyme is needed in the fruits of a white fleshed varieties. A search for a gene putatively responsible melon genotype that lacks b-carotene, the apparent substrate of for the cleavage of b-carotene into b-ionone was carried out in this enzyme. annotated melon fruit EST databases yielding a sequence Carotenoids are biosynthesized and accumulated within (CmCCD1) similar to other plant carotenoid cleavage dioxygenase plastids, yet the carotenoid cleavage dioxygenases are apparently genes [29]. Interestingly, the sequence originated from ‘Tam Dew’ cytosolic. Thus, it is likely that subcellular compartmentation melons, a pale green fleshed variety that lacks either b-carotene limits the access of the CCD enzymes to carotenoid substrates and and its oxidative cleavage product b-ionone. Still, the CmCCD1 limits carotenoid cleavage. This might explain the vast difference gene is up-regulated upon maturation, similarly to the corre- (about a thousand fold) between the carotenoid levels (measured sponding CCD genes in orange-fleshed varieties that contain high in mg/g FW) and apocarotenoid aroma levels (measured in mg/g E. Lewinsohn, M. Gijzen / Plant Science 176 (2009) 161–169 165

FW) in many fruits [19,27]. However, this leaves the question open orcinol dimethyl ether, still express OOMT2 and possess enzymatic for speculation if carotenoids are the bona ﬁda substrate for CCD’s capability in vitro to produce orcinol methyl ether from orcinol, or whether these genes actually have a different role in the cell indicating that another biosynthetic step must be preventing metabolism and serendipitously accept carotenoids as substrates. orcinol dimethyl ether accumulation. Thus, if OOMT2 is not One could speculate that this activity was important in the melon operational in European roses, why is it active in cell-free extracts? wild progenitors, and the gene was retained without any apparent It could therefore be that the biological function of OOMT2 is not function during domestication. related to orcinol dimethyl ether formation, providing another example of occult metabolic capabilities driven by ‘‘silent’’ 3. Silent metabolism in the biosynthesis of volatile ethers metabolism. Chemical diversity with respect to volatile monoterpene and The scent of roses is a very complex trait that is a result of the phenylpropene contents has been documented and rationalized in presence of dozens of aroma chemicals that have distinct sweet basil (Ocimum basilicum), using genomic and metabolomic biosynthetic origins; among those are the mono- and sesquiter- tools [12,16,37]. A sweet basil type that contains predominantly penes, phenylalanine derived compounds, fatty acid derivatives the phenylpropene derivative estragole, possess OMT activities and phenolic methyl ether compounds. The modern hybrid tea able to methylate chavicol to estragole, while a type of basil that rose cultivars emit phenolic methyl ethers, a group compounds accumulates only methyl eugenol (a 3 methoxylated estragole that are only present in the Chinese ancestral species and not in the derivative) possesses an activity able to methylate eugenol to European ancestral species [34,35]. The last steps in the formation methyl eugenol (Fig. 3). However, if chavicol is offered as a of these compounds are the sequential methylation of orcinol (3,5- substrate it will be methylated to estragole in cell-free extracts, dihydroxytoluene) and 3-hydroxy, 5-methoxytoluene to give rise although the plant itself does not contain estragole. Moreover, to orcinol dimethyl ether (3,5-dimethoxytoluene). The reactions some of the lines able to O-methylate chavicol in vitro, do not are catalyzed by O-methyltransferases (OMT’s) that utilize S- accumulate estragole in vivo, and some of the lines are able to O- adenosyl methionine as a methyl donor and an alcoholic substrate methylate isoeugenol in vitro to generate methylisoeugenol, a as an acceptor. Hybrid roses possess two types of orcinol compound that has not been detected in sweet basil. Interestingly, methyltransferase genes with similar but not identical substrate the ‘‘lemon-scented’’ basil line 197 contains citral (a mixture of the preferences OOMT1 and OOMT2 [34,36]. Both gene products monoterpene aldehydes geranial and neral) but lacks volatile perform both of the methylations involved, but OOMT1 is much phenylpropenes in its essential oil (Fig. 3). Still, cell-free extracts more active towards orcinol than towards 3-hydroxy, 5-methox- derived from this lemon-scented line can readily O-methylate ytoluene. Interestingly, OOMT2-like genes seem to be present in all chavicol to estragole, eugenol to methyl eugenol or isoeugenol to rose types, while OOMT1 types are only present in Chinese types methyl-isoeugenol (Fig. 3) although none of these compounds are and their progenies [35]. Still, only those roses that possess OOMT1 present in the plant. This further indicates the presence of ‘‘silent’’ emit orcinol dimethyl ether. European roses, that do not emit O-methyltransferase activities in basil lines that may readily

Fig. 3. Silent metabolism in phenylpropene metabolism in basil (Ocimum basilicum). Essential oil compositions (left panels) were compared to phenylpropene O- methyltransferase towards various substrates (right panels). ‘‘Exotic’’ basil contains almost exclusively methyl chavicol, but displays enzymatic activities able to act on chavicol, eugenol and isoeugenol Line R3 contains almost exclusively estragole and linalool, and is able to methylate chavicol at a high rate as expected; but unable to methylate eugenol or isoeugenol. Line 145 possesses almost exclusively estragole, and displays a specific chavicol O-methyltransferase activity generating estragole, but unable to act efficiently on either eugenol or isoeugenol. Line 149 possesses methyl eugenol and some methyl chavicol in its essential oil and displays O-methyltransferase activity towards chavicol, eugenol, and isoeugenol. Line ‘‘4’’ Contains eugenol and linalool (lacks p-methylated volatile phenylpropenes, and possess no detectable O- methyltransferase activity as expected. Line 197 is a lemon-scented basil and contains almost exclusively the monoterpene aldehydes citral, still it displays high chavicol, eugenol and isoeugenol O-methyltransferase activity. Adapted from [55]. 166 E. Lewinsohn, M. Gijzen / Plant Science 176 (2009) 161–169 accept novel substrates if the ability to produce those substrates is acquired by breeding or mutation, easily yielding to novel chemotypes. Two types of genes coding for phenylpropene specific OMT’s are found in sweet basil. Both gene product types O-methylate phenylpropene alcohols at the 4 position. CVOMT1 readily accepts chavicol to generate methylchavicol (estragole), while EOMT1 accepts eugenol to generate methyleugenol, but still retains the ability to methylate chavicol to estragole at a significant rate. A single T-to-C mutation, causing a phenylalanine residue at position 260 changed to a serine controls this substrate specificity [12]. Nevertheless, plants possessing EOMT activity still have the ‘‘silent’’ ability to O-methylate chavicol to estragole if chavicol is offered as a substrate. Breeders can predict that if the trait to generate the substrate chavicol was introduced by crosses or metabolic engineering the resulting progenies will contain estragole by the action of silent metabolism.

4. Silent metabolism in the phenylpropanoid pathways

The ﬂavonoid phenylpropanoid biosynthetic pathways of plants have been studied from many angles due to the importance, ubiquity, and the visibility of the products.

4.1. Flavonoid biosynthetic pathways provide numerous examples of silent metabolism

Phenotypes that are controlled by flavonoid pigments are powerful tools for genetic investigations because of they can easily be measured. The coloration of flowers, seeds, and other plant Fig. 4. Wild and cultivated soybean seed coat. (A) Schematic illustration of the organs captures our attention, even if one is not a biochemist, soybean seed coat, in cross section. Anthocyanin pigments accumulate in the geneticist, or plant biologist. It also seems that flavonoid pathways palisade epidermis while peroxidase is concentrated in the hourglass cells. Hourglass cells are approximately 75 mm in length. (B) A photograph of commonly determine traits that are subject to heavy selection pressures; at grown yellow-seeded soybeans, lacking seed coat pigmentation due to the least this is clearly so for domesticated crop and ornamental plants. dominant I locus (Glycine max cv. Lindarin). (C) A photograph of wild-type The flavonoid pathway provides a rich source of examples to soybeans with a black-seeded phenotype due to accumulation of anthocyanin illustrate silent metabolism because it is such a large and well- pigments (Glycine max PI437654). studied pathway that has evolved a variety of functions that serve plants. With regard to seed crops, for most species the selection has been for lack of color. Wild wheat, rice, barley, soybean and many likely because flavonoids play essential roles in other plant tissues, other crops have colored seed while most domesticated strains such as in reproductive tissues where they are essential for lack color. fertility. Even though most commonly grown soybeans are yellow- 4.2. Silent flavonoid metabolism in soybean seed coats seeded due to a lack of chalcone synthase in the seed coat tissues, genes corresponding to steps in the downstream pathways Seed coat pigmentation in soybean is determined by the continue to be expressed at high levels. For example, dihydro- accumulation of anthocyanin and proanthocyanidin molecules in flavonol-4-reductase (DFR) and flavonoid 30 hydroxylase (F30H) the palisade cells of epidermis [38]. Nearly all commonly grown transcripts are highly expressed in seed coats where they could soybean cultivars are yellow-seeded but may vary with regard to have participated in the biosynthesis of anthocyanin pigments pigmentation of the hilum (point of connection of the seed to the during the late stages of seed development. Moreover, these genes ovary wall). Conversely, wild-type soybeans are deeply pigmented are also expressed in yellow-seeded phenotypes that do not and black-seeded (Fig. 4). Over the years, through genetic crossing accumulate any pigments [42,43]. The isoflavonoid branch also and inheritance studies, a bewildering array of seed coat appears to be constitutively active in seed coats, regardless of pigmentation phenotypes have been uncovered and described chalcone synthase levels, as evidenced by the expression of [39]. There are many genes that can influence seed coat isoflavone synthase [44]. But in this case, isoflavonoid products can pigmentation, but three major ones, I, R, and T, have the greatest be detected in seed coat tissues, whereas anthocyanins clearly are effects [40]. Among these, the I locus is the most important. This not present. How does this occur? There are several possibilities, gene controls the expression and distribution of chalcone synthase but it seems most likely that the isoflavonoids present in in the seed coat tissues, and thus it effectively operates as a developing seed coats were synthesized in other plant organs bottleneck that can restrict entry of substrate into various and simply transported to the seed coat. In any case, it is apparent branches of the flavonoid pathway [41]. The I locus exerts that the seed coat is programmed to make anthocyanins, powerful epistatic effects that can mask phenotypes conditioned isoflavonoids, and other phenylpropanoids that are dependent by R and T, because these other genes operate in pathways that are upon the key entry point enzyme, CHS, regardless of whether there downstream from chalcone synthase. The I locus is also tissue is a sufficient amount of this enzyme to ensure flux through the specific, only altering flavonoid metabolism in the seed coat. This is pathway. E. Lewinsohn, M. Gijzen / Plant Science 176 (2009) 161–169 167

4.3. Soybean seed coat anthocyanin pigments are substrates for the pigments, and that this has become obscured as a consequence of seed coat peroxidase enzyme domestication and selection of yellow-seeded types.

Another example of silent metabolism in soybean seed coats 5. Mechanism of action of silent metabolism that is apparently linked to anthocyanin pigmentation relates to the accumulation of peroxidase. The soybean peroxidase is a class 5.1. Silent metabolism and relaxed selection III plant peroxidase and is the most abundant protein in mature seed coats, where it accounts for approximately 5% of the total It is instructive to consider the idea of silent metabolism soluble protein [45]. The occurrence of soybean peroxidase in the together with the well-known genetic concepts of epistasis and seed coat is controlled by a single dominant gene, Ep. Homozygous relaxed selection, as well as the related topics of sub- and neo- recessive epep genotypes do not accumulate soybean peroxidase functionalization of gene products. In cases where genes interact to due to a deletion mutation in the structural gene encoding the produce particular phenotypes, epistatic genes may either hide or enzyme [46]. Despite this difference in the amount of soybean reveal the phenotypic expression of other genes. This differs from peroxidase enzyme in Ep- versus ep ep genotypes, there is no dominant and recessive interactions because the genes are not known phenotypic effect such as change in seed color, perme- allelic. For example, epistatic interactions are commonly observed ability, lignification, longevity, or hardness that can be associated among genes that encode enzymes within a linear biochemical with the Ep locus. Thus, the function of SBP in the seed coat remains pathway. The catalytic efficiency, substrate specificity, or other enigmatic. Perhaps this is because its role is dependent on the characteristics of a particular enzyme in the pathway can alter the presence of anthocyanin pigments that are absent from virtually all final products or the metabolic flux through the entire path. A modern soybean cultivars. mutation in a gene that results in a non-functional enzyme would The participation of anthocyanin pigments and other flavonoid effectively block the pathway at that step, possibly leading to molecules in oxidative reactions is gaining more attention, since accumulation of substrate upstream and a depletion of substrate this may constitute a significant portion of their functionality. downstream. For individuals that carry such an epistatic mutation, With regards to soybean seed coats, wild-type (black) soybeans the enzymes that operate downstream are biochemical orphans accumulate large amounts of anthocyanin pigments and soybean and become part of the silent metabolism of the plant cell. Thus, peroxidase in separate but adjacent cell layers of the seed coat. The epistatic genetic effects may provide a context for silent anthocyanins are concentrated in the palisade cells of the metabolism that is otherwise cryptic or puzzling. epidermis of the seed coat, whereas soybean peroxidase accumu- Epistatic genetic effects that become fixed in a population may lates within the hourglass cells of the subepidermis [45,47]. The result in the degeneration of a biochemical pathway [49] or supply vacuolar targeting signal of the soybean peroxidase is somewhat opportunities for enzymes that have become biochemical orphans unusual among class III plant peroxidases, as most other enzymes to evolve new capabilities. Genetically, this is considered a form of of this type are secreted into the apoplast. Nonetheless, maceration relaxed selection, a reduction or removal of selective pressure that of the seed or imbibition of water simultaneously releases these operates on a gene. Gene duplication is another genetic mechanism two soluble seed coat constituents. In fact, seed coat anthocyanin that results in relaxed selection and offers similar prospects. The pigments are substrates for soybean peroxidase-catalyzed reac- sub-functionalization of an enzyme, so that it acquires character- tions, as shown in Fig. 5. The polymerization of seed coat istics that are somewhat different from its ancestral type, is an anthocyanin pigments by soybean peroxidase into insoluble adaptation that is facilitated by removing any conservative selective products may provide defense or protection for the seed. It is pressure. Likewise, neo-functionalization is the acquisition of a reminiscent of binary systems where enzymes and substrates are completely new property that is distinct from its ancestral type. This spatially separated in different cell types or layers, such as occurs more rarely but potentially offers the organism novel traits glucosinolates and cyanogenic glycosides [48]. Thus, it is plausible that may drastically alter its fitness and success. Therefore, the silent that function of SBP is related to its activity towards anthocyanin metabolism of a plant cell represents a biochemical manifestation of the genetic notion called relaxed selection. The impact of gene duplication on primary and secondary metabolic pathways has been estimated using the model plant Arabidopsis [50]. The relatively high levels of intraspecific variation observed in pathways of secondary metabolism contrast with the general conservation of function in primary metabolism. The flexibility of secondary metabolism is supposed to be enabled by a greater genetic redundancy of these pathways compared to primary metabolic routes [50]. It is a compelling hypothesis that plant populations maintain a diverse reservoir of secondary metabolic capability to meet the ever changing selective pressures in their environment.

6. Concluding remarks

It has become evident that the initial Darwinian view of

Fig. 5. Polymerization of anthocyanin pigments by seed coat peroxidase. biological systems that have been optimized by evolution is much Anthocyanins purified from soybean seed coats were dissolved in 250 mM more complex that we initially thought. The large phytochemical NaH2PO4 buffer (pH 7.0), and H2O2 was added to 0.01% (v/v), alone or in diversity that we find in nature is a result of the different selection combination with purified soybean seed coat peroxidase (SBP) enzyme. Photos pressures that plants have successfully coped with during were taken 5 min and 24 h after additions. This figure illustrates that SBP catalyzes a evolution. The ubiquity of plant silent metabolism may be a great rapid polymerization of anthocyanin pigments into insoluble products. The addition of H2O2 alone, without SBP, can de-colorize solutions of anthocyanin advantage in a changing and evolving world. It is likely that plants pigments over long periods, but no precipitate forms under these circumstances. have evolved means of preserving the biochemical information to 168 E. Lewinsohn, M. Gijzen / Plant Science 176 (2009) 161–169 generate chemicals without actually generating them. Moreover, References the availability of plant complete genomes and the advancement in [1] R. Croteau, T.M. Kutchan, N.G. Lewis, Natural products (secondary metabolites), our understanding of gene evolution, has led to the discovery that in: B. Buchanan, W. Gruissem, R. Jones (Eds.), Biochemistry and Molecular Biology plants that do not normally accumulate a certain natural product of Plants, American Society of Plant Physiologists, Rockville, MD, 2000, pp. 1250– still posses homologous genes to those present in plants that 1318. [2] J. Gershenzon, N. Dudareva, The function of terpene natural products in the actively synthesize them [51]. natural world, Nat. Chem. Biol. 3 (2007) 408–414. It has long been known by geneticists that through hybridization [3] W. Schwab, R. Davidovich-Rikanati, E. Lewinsohn, Biosynthesis of plant-derived it is possible to augment the generation of qualitative and flavor compounds, The Plant J. 54 (2008) 712–732. [4] J.B. Harborne, Recent advances in chemical ecology, Nat. Prod. Rep. 6 (1989) 85– quantitative variation in secondary chemistry. Qualitatively, hybrids 109. may express all of the secondary chemicals of the parental taxa, may [5] J.B. Harborne, Recent advances in the ecological chemistry of plant terpenoids, in: fail to express certain parental chemicals, or may express novel J.B. Harborne, F.A. Tomas-Barberan (Eds.), Ecological Chemistry and Biochemistry chemicals that are absent in each parent. Quantitatively, concentra- of Plant Terpenoids, 6, Clarendon Press, Oxford, 1991, pp. 399–426. [6] J.H. Langenheim, Higher plant terpenoids: a phytocentric overview of their tions of parental chemicals may vary markedly among hybrids. This ecological rules, J. Chem. Ecol. 20 (1994) 1223–1280. phenomenon has been termed ‘‘additive inheritance’’ [52],andit [7] R.D. Firn, C.G. Jones, Natural products—a simple model to explain chemical might be a reflection or operate in a similar or related way as the diversity, Nat. Prod. Rep. 20 (2003) 382–391. [8] R. Croteau, F. Karp, The origin of natural odorants, in: P.M. Muller, D. Lamparsky phenomenonlongknownas‘‘heterosis’’or‘‘hybridvigor’’.Inessence, (Eds.), Perfumes. Art, Science and Technology, Elsevier Applied Science, London, each parent may contribute previously silent but complementary 1991, pp. 101–126. enzymes to generate complementary networks, and thus the [9] C.L. Steele, J. Crock, J. Bohlmann, R. Croteau, Sesquiterpene synthases from grand fir (Abies grandis). Comparison of constitutive and wound-induced activities, and hybrid’s phytochemistry might be very different to that of their cDNA isolation, characterization, and bacterial expression of delta-selinene parents. Recently, a combined metabolomic–genomic set of tools synthase and gamma-humulene synthase, J. Biol. Chem. 273 (1998) 2078–2089. have been employed to rationalize flavonoid biosynthesis in poplar [10] D. Tholl, F. Chen, J. Petri, J. Gershenzon, E. Pichersky, Two sesquiterpene synthases are responsible for the complex mixture of sesquiterpenes emitted from Arabi- and has provided interesting insights about this phenomenon [53]. dopsis flowers, Plant J. 42 (2005) 757–771. Most of the biochemical, molecular and genetic studies on [11] V. Portnoy, Y. Benyamini, E. Bar, R. Harel-Beja, S. Gepstein, J.J. Giovannoni, A.A. secondary metabolism have been conducted using domesticated Schaffer, J. Burger, Y. Tadmor, E. Lewinsohn, N. Katzir, The molecular and biochemical basis for varietal variation in sesquiterpene content in melon (Cucu- plants. All cultivated crops originated from wild relatives and mis melo L.) rinds, Plant Mol. Biol. 66 (2008) 647–661. many of the original traits present in the progenitors have been lost [12] D.R. Gang, N. Lavid, C. Zubieta, F. Chen, T. Beuerle, E. Lewinsohn, J.P. Noel, E. through crop domestication [54]. Cultivated plants have been Pichersky, Characterization of phenylpropene O-methyltransferases from sweet subjected to directed hybridization and selected taking into basil: facile change of substrate specificity and convergent evolution within a plant O-methyltransferase family, Plant Cell 14 (2002) 505–519. account commercially important or practical traits. Thus, breeding [13] D.S. Lee, P. Nioche, M. Hamberg, C.S. Raman, Structural insights into the evolu- and crop improvement has allowed for a limited set of important tionary paths of oxylipin biosynthetic enzymes, Nature 455 (2008) 363–368. traits to prevail and be overrepresented in our modern crops. The [14] W. Schwab, Metabolome diversity: too few genes, too many metabolites? Phy- tochemistry 62 (2003) 837–849. number of generations in which selection pressure has been [15] E. Pichersky, D.R. Gang, Genetics and biochemistry of secondary metabolites in attained is very small as compared to the number of generations plants: an evolutionary perspective, Trends Plant Sci. 5 (2000) 439–445. that wild plants have experienced selection pressures throughout [16] Z. Xie, J. Kapteyn, D.R. Gang, A systems biology investigation of the MEP/terpenoid and shikimate/phenylpropanoid pathways points to multiple levels of metabolic natural evolution. Silent metabolism could be a relic of important control in sweet basil glandular trichomes, Plant J. 54 (2008) 349–361. and valuable traits or functions that the ancestral wild plants [17] M. Dafny-Yelin, I. Guterman, N. Menda, M. Ovadis, M. Shalit, E. Pichersky, D. acquired and preserved during their evolution, but their cultivated Zamir, E. Lewinsohn, Z. Adam, D. Weiss, A. Vainstein, Flower proteome: changes in protein spectrum during the advanced stages of rose petal development, Planta offspring have lost. If the expression of a silent partial pathway or 222 (2005) 37–46. biochemical orphan do not have harmful effects and do not [18] N. Dudareva, E. Pichersky, Metabolic engineering of plant volatiles, Curr. Opin. significantly affect the overall fitness of the plant, only very little Biotechnol. 19 (2008) 181–189. [19] E. Lewinsohn, F. Schalechet, J. Wilkinson, K. Matsui, Y. Tadmor, K.H. Nam, O. Amar, selection pressure is applied towards eliminating those traits, and E. Lastochkin, O. Larkov, U. Ravid, W. Hiatt, S. Gepstein, E. Pichersky, Enhanced therefore they will be conserved during the domestication process. levels of the aroma and flavor compound S-linalool by metabolic engineering of Another possibility of the widespread occurrence of silent the terpenoid pathway in tomato fruits, Plant Physiol. 127 (2001) 1256–1265. metabolism is that many of the enzymes we detect as ‘‘silent’’ do [20] J. Lucker, H.J. Bouwmeester, W. Schwab, J. Blaas, L.H. van der Plas, H.A. Verhoe- ven, Expression of Clarkia S-linalool synthase in transgenic petunia plants have an important yet unknown metabolic function. Due to the results in the accumulation of S-linalyl-beta-D-glucopyranoside, Plant J. 27 broad substrate specificity displayed by many of these enzymes, (2001) 315–324. novel substrates can be readily utilized by serendipitously [21] M. Lavy, A. Zuker, E. Lewinsohn, O. Larkov, U. Ravid, A. Vainstein, D. Weiss, Linalool and linalool oxide production in transgenic carnation flowers expressing recruited enzymes and novel phenotypes can be generated that the Clarkia breweri linalool synthase gene, Mol. Breeding 9 (2002) 103–111. are subjected to natural or human selection. Thus, if the new traits [22] R. Davidovich-Rikanati, Y. Sitrit, Y. Tadmor, Y. Iijima, N. Bilenko, E. Bar, B. generated provide an evolutionary value, they will be retained. Carmona, E. Fallik, N. Dudai, J.E. Simon, E. Pichersky, E. Lewinsohn, Enrichment of tomato flavor by diversion of the early plastidial terpenoid pathway, Nat. Increasing and retaining phytochemical diversity is key to the Biotechnol. 25 (2007) 899–901. survival and adaptation of plants, and ‘‘silent metabolism’’ might [23] T.A. Bicsak, L.R. Kann, A. Reiter, T. Chase Jr., Tomato alcohol dehydrogenase: be one of the key factors by which plants cost-effectively retain purification and substrate specificity, Arch. Biochem. Biophys. 216 (1982) 605– 615. their flexibility to defend themselves and communicate with other [24] X. Ye, S. Al-Babili, A. Kloti, J. Zhang, P. Lucca, P. Beyer, I. Potrykus, Engineering the organisms. In this context, silent metabolism is not at all silenced, provitamin A (beta-carotene) biosynthetic pathway into (carotenoid-free) rice but is a loud reservoir of metabolic pathways that can easily be endosperm, Science 287 (2000) 303–305. [25] J.A. Paine, C.A. Shipton, S. Chaggar, R.M. Howells, M.J. Kennedy, G. Vernon, S.Y. activated, conferring fitness benefits to the bearing organism. Wright, E. Hinchliffe, J.L. Adams, A.L. Silverstone, R. Drake, Improving the nutri- Silent metabolism constitutes an evolutionary asset and thus tional value of Golden Rice through increased pro-vitamin A content, Nat. widespread the plant kingdom, contributing to the impressive Biotechnol. 23 (2005) 482–487. phytochemical diversity that we presently experience. [26] P. Schaub, S. Al-Babili, R. Drake, P. Beyer, Why is Golden Rice golden (yellow) instead of red? Plant Physiol. 138 (2005) 441–450. [27] E. Lewinsohn, Y. Sitrit, E. Bar, Y. Azulay, A. Meir, D. Zamir, Y. Tadmor, Carotenoid Acknowledgements pigmentation affects the volatile composition of tomato and watermelon fruits, as revealed by comparative genetic analyses, J. Agric. Food Chem. 53 (2005) 3142– 3148. We thank Yaron Sitrit, Kobi Tadmor and Ilan Levin for useful [28] E. Lewinsohn, Y. Sitrit, E. Bar, Y. Azulay, M. Ibdah, A. Meir, E. Yosef, D. Zamir, Y. comments. Tadmor, Not just color—carotenoid degradation as a link between color and E. Lewinsohn, M. Gijzen / Plant Science 176 (2009) 161–169 169

aroma in tomato and watermelon fruit, Trends Food Sci.Technol. 16 (2005) [41] J.H. Tuteja, S.J. Clough, W.C. Chan, L.O. Vodkin, Tissue-specific gene silencing 407–415. mediated by a naturally occurring chalcone synthase gene cluster in Glycine max, [29] M. Ibdah, Y. Azulay, V. Portnoy, B. Wasserman, E. Bar, A. Meir, Y. Burger, J. Plant Cell 16 (2004) 819–835. Hirschberg, A.A. Schaffer, N. Katzir, Y. Tadmor, E. Lewinsohn, Functional char- [42] C.S. Wang, J.J. Todd, L.O. Vodkin, Chalcone synthase mRNA and activity are acterization of CmCCD1, a carotenoid cleavage dioxygenase from melon, Phyto- reduced in yellow soybean seed coats with dominant I alleles, Plant Physiol. chemistry 67 (2006) 1579–1589. 105 (1994) 739–748. [30] P. Winterhalter, R. Rouseff, Carotenoid-derived aroma compounds: an introduc- [43] G. Zabala, L. Vodkin, Cloning of the pleiotropic T locus in soybean and two tion, in: P. Winterhalter, R. Rouseff (Eds.), Carotenoid-derived aroma compounds, recessive alleles that differentially affect structure and expression of the encoded vol. 802, American Chemical Society, Washington, DC, 2002, pp. 1–17. flavonoid 30 hydroxylase, Genetics 163 (2003) 295–309. [31] F. Bouvier, J.C. Isner, O. Dogbo, B. Camara, Oxidative tailoring of carotenoids: a [44] S. Dhaubhadel, B.D. McGarvey, R. Williams, M. Gijzen, Isoflavonoid biosynthesis prospect towards novel functions in plants, Trends Plant Sci. 10 (2005) 187– and accumulation in developing soybean seeds, Plant Mol. Biol. 53 (2003) 733– 194. 743. [32] G. Giuliano, S. Al-Babili, J. von Lintig, Carotenoid oxygenases: cleave it or leave it, [45] M. Gijzen, R. Van Huystee, R.I. Buzzell, Soybean seed coat peroxidase (a compar- Trends Plant Sci. 8 (2003) 145–149. ison of high-activity and low-activity genotypes), Plant Physiol. 103 (1993) 1061– [33] A.J. Simkin, S.H. Schwartz, M. Auldridge, M.G. Taylor, H.J. Klee, The tomato 1066. carotenoid cleavage dioxygenase 1 genes contribute to the formation of the [46] M. Gijzen, A deletion mutation at the ep locus causes low seed coat peroxidase flavor volatiles beta-ionone, pseudoionone, and geranylacetone, Plant J. 40 activity in soybean, Plant J. 12 (1997) 991–998. (2004) 882–892. [47] J.B. Carlson, N.R. Lersten, Reproductive morphology, in: H.R. Boerma, J.E. Specht [34] G. Scalliet, N. Journot, F. Jullien, S. Baudino, J.L. Magnard, S. Channeliere, P. Vergne, (Eds.), Soybeans: Improvement, Production, and Uses, American Society of Agr- C. Dumas, M. Bendahmane, J.M. Cock, P. Hugueney, Biosynthesis of the major onomy, Madison, WI, 2004. scent components 3,5-dimethoxytoluene and 1,3,5-trimethoxybenzene by novel [48] B.A. Halkier, J. Gershenzon, Biology and biochemistry of glucosinolates, Annu. Rev. rose O-methyltransferases, FEBS Lett. 523 (2002) 113–118. Plant Biol. 57 (2006) 303–333. [35] G. Scalliet, F. Piola, C.J. Douady, S. Rety, O. Raymond, S. Baudino, K. Bordji, M. [49] R.A. Zufall, M.D. Rausher, Genetic changes associated with floral adaptation Bendahmane, C. Dumas, J.M. Cock, P. Hugueney, Scent evolution in Chinese roses, restrict future evolutionary potential, Nature 428 (2004) 847–850. Proc. Natl. Acad. Sci. USA 105 (2008) 5927–5932. [50] D.J. Kliebenstein, A role for gene duplication and natural variation of gene [36] N. Lavid, J. Wang, M. Shalit, I. Guterman, E. Bar, T. Beuerle, N. Menda, S. Shafir, D. expression in the evolution of metabolism, PLoS ONE 3 (2008) e1838. Zamir, Z. Adam, A. Vainstein, D. Weiss, E. Pichersky, E. Lewinsohn, O-Methyl- [51] P.J. Facchini, D.A. Bird, B. St-Pierre, Can Arabidopsis make complex alkaloids? transferases involved in the biosynthesis of volatile phenolic derivatives in rose Trends Plant Sci. 9 (2004) 116–122. petals, Plant Physiol. 129 (2002) 1899–1907. [52] C.M. Orians, The effects of hybridization in plants on secondary chemistry: [37] D.R. Gang, J. Wang, N. Dudareva, K.H. Nam, J.E. Simon, E. Lewinsohn, E. Pichersky, implications for the ecology and evolution of plant–herbivore interactions, An investigation of the storage and biosynthesis of phenylpropenes in sweet basil, Am. J. Bot. 87 (2000) 1749–1756. Plant Physiol. 125 (2001) 539–555. [53] K. Morreel, G. Goeminne, V. Storme, L. Sterck, J. Ralph, W. Coppieters, P. Breyne, M. [38] J.J. Todd, L.O. Vodkin, Pigmented soybean (Glycine max) seed coats accumulate Steenackers, M. Georges, E. Messens, W. Boerjan, Genetical metabolomics of proanthocyanidins during development, Plant Physiol. 102 (1993) 663–670. flavonoid biosynthesis in Populus: a case study, Plant J. 47 (2006) 224–237. [39] R.G. Palmer, T.W. Pfeiffer, G.R. Buss, T.C. Kilen, Qualitative genetics, in: H.R. [54] D. Zamir, Improving plant breeding with exotic genetic libraries, Nat. Rev. Genet. Boerma, J.E. Specht (Eds.), Soybeans: Improvement, Production, and Uses, Amer- 2 (2001) 983–989. ican Society of Agronomy, Madison, WI, 2004. [55] I. Ziv-Raz, Methyltransferases involved in the formation of methyl chavicol and [40] D. Qutob, F. Ma, C.A. Peterson, M.A. Bernards, M. Gijzen, Structural and perme- methyl eugenol in sweet basil (Ocimum basilicum L.). M.Sc. Thesis, The Technion – ability properties of the soybean seed coat, Botany 86 (2008) 219–227. Israel Institute of Technology, Haifa (1999) pp. 75.