Emodin – a secondary metabolite with multiple ecological functions in higher plants
Department of Biology, Faculty of Science and Science Education, University of Haifa at Oranim, Tivon 36006, Israel; Present address: Department of Zoology, University of Florida, Gainesville, FL 32611–8525, USA
The anthraquinone emodin, identified in 17 plant families distributed worldwide, has numerous biological activities, some of which exhibit a wide spectrum of ecological impacts by mediating biotic or abiotic interactions of plants with their environment. Here the evidence for direct and indirect effects of emodin on plant survival and reproduction is reviewed. Emodin in vegetative organs may help protect plants against herbivores, pathogens, competitors and extrinsic abiotic factors (e.g. high light intensities). In unripe fruit pulp, emodin may facilitate seed dispersal by protecting the immature fruit against predispersal seed predation whereas in ripe pulp it may deter frugivores and thus reduce the chances that seeds will be defecated beneath the parent plant. It also accelerates the passage of seeds through the digestive tract, potentially reducing dispersal distance and increasing seed viability upon dispersal. In certain circumstances both of the last two effects could also have negative fitness consequences for plants. Natural selection should favor secondary metabolites with multiple functions because they protect the plants against a variety of unpredictable biotic and abiotic environments. Such metabolites also enhance plant defenses by using different molecular targets of specific enemies through a variety of mechanisms of action. Emodin illustrates the wide and often overlooked potential for chemical multifunctionality in plant secondary metabolites.
The original definition of chemical compounds as ‘secondary’ emerged from the notion that they are not absolutely essential to the survival and reproduction of plants (Bentley, 1999). However, the current concept is that the wide variety of secondary metabolites that is produced by higher plants plays important roles in many complex biotic and abiotic interactions (Waterman, 1992; Waterman & Mole, 1994). In several instances a group of related secondary compounds has exhibited more than one biological activity, sometimes in contradictory modes (Rhoades, 1977; Swain, 1986; Wink, 1999; Cipollini, 2000). The same anthocyanins can act as attractants to the flower, antioxidants, phytoalexins and as antibacterial agents (Cooper-Driver & Bhattacharya, 1998). Many of the same trichome constituents, which are toxic to insects, are antimicrobial and also have deleterious effects on mammalian herbivores (Harborne, 1993). The leaf resin of Larrea spp. deters herbivores, protects against desiccation and screens UV radiation (Rhoades, 1977). Ripe fruits of Solanum often contain glycoalkaloids that possess multiple biological activities, such as germination inhibition, laxative effects on mammals, constipating effects on birds, deterrents to herbivores and more (Cipollini, 2000). Capsaicin in chili (Capsicum spp.) fruits exhibits multiple functions that promote seed dispersal (Tewksbury & Nabhan, 2001). Many alkaloids simultaneously exhibit allelopathic, antibacterial and animal toxicity effects (Wink & Schimmer, 1999). Thus, the same secondary metabolites may act to deter different groups of organisms by different modes of action. Rhoades (1977) argued that any chemical substances must be integrated into the total metabolic scheme of the plant and by nature are expected to be multifunctional. Wink (1999) claims that multiple functions of secondary metabolites are common and do not contradict their main role for chemical defense and signaling. Furthermore, he argues that natural selection will favor those metabolites that possess multiple functions.
Here, I illustrate the broad diversity of biological activities exhibited by one secondary metabolite – emodin. Although this compound was first described > 75 yr ago (reported as ‘frangula-emodin’, Kögl & Postowsky, 1925), many of its diverse biological properties have been discovered in the last decade. I focus on the variety of ecological interactions mediated by emodin and on its role in facilitating plant survival under the risks induced by biotic and abiotic hazards and also on the potential cost of its production to plant fitness. I then discuss several of the ecological and evolutionary aspects of secondary metabolite multifunctionality.
Emodin: chemical structure and distribution
Emodin belongs to the anthraquinones, a group of > 170 natural compounds that make up the largest group of natural quinones (Thomson, 1987; Thomson, 1997; Harborne et al., 1999). More than half of the natural anthraquinones are found in lower fungi, particularly in Penicillium and Aspergillus species, and in lichens. Others are found in higher plants, and, in isolated instances, in insects (Thomson, 1987, 1997; Evans, 1996). The Rubiaceae, Rhamnaceae, Fabaceae, Polygonaceae, Bignoniaceae, Verbenaceae, Scrophulariaceae and Liliaceae are particularly rich in anthraquinones (Van den Berg & Labadie, 1989).
Anthraquinone may either be formed via the acetate-malonate pathway in Polygonaceae and Rhamnaceae or O-succinylbenzoic acid in the Bignoniaceae and Verbenaceae (Evans, 1996; Dewick, 1998; Harborne et al., 1999). The basic chemical structure of anthraquinone is an anthracene ring (tricyclic aromatic) with two ketone groups in position C9 and C10 (Fig. 1). In plants, anthraquinones are mostly present as sugar derivatives – the glycosides – but the free form – the aglycones – are widely distributed as well (Thomson, 1987, 1997; Harborne et al., 1999). Among the most common naturally occurring anthraquinone aglycones in higher plants are emodin, rhein, chrysophanol, aloe-emodin, and physcion (Fig. 1) (Evans, 1996; Harborne et al., 1999). Several biochemical pathways that transform one anthraquinone to another have been discovered. For example, chrysophanol is synthesized in plants by dehydroxylation of emodin, an enzymatic conversion that is mediated by nicotinamide adenine dinucleotide phosphate (NADPH) as cofactor (Anderson et al., 1988). It was also suggested that physion is derived from emodin (Thomson, 1997). The anthraquinone glycosides formed when one or more sugar molecules, mostly glucose or rhamnose, are bound to the aglycone by a β-glycoside linkage to hydroxyl group at position C8 (in the case of glucose) or the one at C6 (in the case of rhamnose) (Fig. 1) (Dewick, 1998). Among the most common emodin-related glycosides are emodin-8-glucose, frangulin and glucofrangulin (Harborn et al., 1999) (Fig. 1).
Emodin (1,3,8-trihydroxy-6-methylanthraquinone, Fig. 1) is mainly reported in three plant families: Fabaceae (Cassia spp.), Polygonaceae (Rheum, Rumex and Polygonum spp.) and Rhamnaceae (Rhamnus and Ventilago spp.) (Table 1). It seems that this is not a function of the larger number of species assayed in these families relative to other families but that emodin is actually more common among species of these three families. However, a comprehensive literature survey revealed that emodin has already been identified in at least 17 plant families, 28 genera and 94 species (Table 1). Interesting records of emodin have been documented recently in Asteraceae, Poaceae and Simaroubaceae, among others (Table 1). The presence of emodin in Aloe (Liliaceae) is rather controversial (Dange, 1996), and it might be rare in this family. Emodin has a worldwide distribution, occurring in subtropical and tropical families (e.g. Bignoniaceae and Simaroubaceae), in families that mainly inhabit the temperate region (e.g. Polygonaceae and Saxifragaceae), and in families inhabiting both the tropics and the temperate regions (e.g. Rhamnaceae and Clusiaceae). Furthermore, emodin occurs in diverse life forms including trees, shrubs, lianas and herbs.
Table 1. Families, genera, number of species (in parentheses) and parts of plants containing emodin
Emodin was originally reported as being more common in bark and roots (Evans, 1996; Bruneton, 1999), but it is clear now that emodin is present in other vegetative organs (stem, foliage) as well as in reproductive organs (flower, fruit, seeds, pods, Table 1). According to the nonadaptive hypothesis, the distribution of secondary metabolites within organs may be roughly equivalent to the distribution of the primary metabolic pathways responsible for the creation of the secondary metabolite (as a byproduct) and thus they do not necessarily have an adaptive function in each organ (Eriksson & Ehrlén, 1998). However, concentrations of emodin in different plant organs are frequently unequal, a phenomenon found in many secondary metabolites (Zangerl & Bazzaz, 1992; Bernays & Chapman, 1994). For example, the flowers and roots of Rumex luminiastrum contain higher emodin concentrations than the leaves (Abd El-Fattah et al., 1994). The average (± sd) emodin aglycone concentration in R. alaternus leaves (0.04% ± 0.05% d. wt, n= 14) is 8 times higher than in unripe fruit (seeds and pulp) (Tsahar, 2001). From the ecological point of view it is expected that the use of secondary metabolites in ripe pulp and seeds would be radically different, as the pulp should reward the seed disperser with nutrition whereas the seed should remain unattractive to predators (Waterman, 1992). Indeed, emodin content in ripe pulp of R. alaternus (0.0017% ± 0.0006% d. wt, n= 12) is 14 times lower that in the seeds (Tsahar, 2001). The allocation of emodin in various plant organs and the unequal distribution of emodin concentration among them may in itself provide circumstantial evidence for multifunctionality of emodin in plants.
Nevertheless, a similar content of secondary metabolites among organs does not necessarily support the nonadaptive hypothesis. For example, the similar content of secondary metabolites (alkaloids, saponins, cyanogenic glycosides, terpenoids, and phenolics) found in unripe fruits and in leaves of several species (Joslyn & Goldstein, 1964; McKey, 1979) may indicate that in these cases a similar secondary metabolite content is required to fight against the specific potential enemies of each organ.
Abiotic factors that govern emodin levels in plants
Synthesis and accumulation of secondary metabolites in plants is regulated in space and time (Wink & Schimmer, 1999) and is affected by abiotic environmental factors, such as light intensity, soil minerals, osmotic stresses (drought and salinity), and seasonality (Waterman & Mole, 1989, 1994). Abiotic environmental factors that constrain the production of secondary metabolite in plants indirectly influence the interactions of plants with their biotic environment (Waterman & Mole, 1994). Therefore, in order to better understand the role of emodin as a mediator of biotic interactions it is essential to explore how its synthesis is influenced by abiotic factors.
Studies indicate that emodin levels in plants depend on season and light intensity. The highest amount of total anthraquinones (where approx. 50% was emodin) in the leaves of Rheum undulatum in Europe was observed in springtime (April) followed by a continuous decrease during the summer with the lowest anthraquinone amounts in late summer (September) (Paneitz & Westendorf, 1999). Such a seasonal pattern may reflect the trade-off between plant development and defense (Waterman & Mole, 1989). Hence, the level of emodin in R. undulatum may diminish in summer when high metabolic activities, such as growing, flowering and fruiting, utilize the limited plant resources. However, the temporal pattern of emodin concentration in Rhamnus frangula bark in Europe showed three peaks (April, July–August and November), and metabolic activities seem insufficient to explain it (Kubiak, 1977).
Emodin (and total anthraquinones) content in suspension cultures of Rhamnus purshiana was significantly raised by a daily photoperiod of 12 h (Van den Berg et al., 1988). Similar positive correlations between secondary metabolites and light intensity emerged among individuals of the same species and among organs of the same individual exposed to different light intensities (Waterman & Mole, 1989). This phenotypic response of the plant to increased light intensity is probably due to increased carbon availability as predicted by the carbon/nutrient hypothesis (Bryant et al., 1983).
Biological activities of emodin
Emodin (as well as physcion, chrysophanol, aloe-emodin and rehin) forms the basis of a range of purgative anthraquinone derivatives and from ancient times has been widely used as a laxative compound (Evans, 1996; Bruneton, 1999). It is believed that the presence of hydroxyl groups in position 1 and 8 of the aromatic ring system is essential for the purgative action of the compound (Paneitz & Westendorf, 1999). Because of its chemical structure, emodin glycosides (and other anthraquinones) in mammals are carried unabsorbed to the large intestine, where metabolism to the active aglycones takes place by intestinal bacterial flora. The aglycone exerts its laxative effect by damaging epithelial cells, which leads directly and indirectly to changes in absorption, secretion and motility (Mueller et al., 1999; Van den Gorkom et al., 1999). Emodin also inhibits the ion transport (Cl−-channels) across colon cells, contributing to the laxative effect (Rauwald, 1998).
Recent studies indicate that emodin exhibits numerous other biological activities. It affects the immune system, vasomotor system and metabolic processes (Table 2). Other biological activities are anti-inflammatory, antimicroorganism and antifeedant (Table 2). The molecular mechanisms underlying many of these biological effects remain either unknown or controversial. Nevertheless, emodin has diversified modes of action. For example, emodin increases the rate and force of heart contractions probably through release of endogenous catecholamines (Dwivedi et al., 1988), contributes to DNA damage by inhibiting the catalytic activity of topoisomerase II (Mueller et al., 1999), and induces muscle contractions by the release of internal Ca2+ as a result of oxidation of the ryanodine receptor (Cheng & Kang, 1998). Many of the biological activities of emodin are dose-dependent and different effective doses for a specific activity have been found in different organisms (Boik, 1995). Of most relevance here, emodin has many biological activities potentially mediating plant–plant, plant–animal, plant–microorganism and plant–abiotic environment interactions.
Table 2. Pharmacological and biological activities of emodin
Emodin–mediated interactions between plants and the biotic environment
The feeding-deterrent property is an important function of emodin in mediating plant–animal interactions. Emodin has a deterrent effect on a large spectrum of organisms from invertebrates to vertebrates. In relatively low concentrations of emodin, gypsy moth (Lymantria dispar) larvae reduced feeding and prolonged development and with high concentrations a pronounced mortality occurred in 2–3 d (Trial & Dimond, 1979). The mode of action of emodin as an insect antifeedant is yet to be studied.
The effective feeding deterrent property of emodin to phytophagous insects was suggested as an explanation for the little insect attack on Rhamnus alnifolia foliage (Trial & Dimond, 1979). The number of species of phytophagous insects on introduced Rhamnus cathartica plants in Canada is much lower than on native plants in Europe (Malicky et al., 1970). This may be indirect evidence of the importance of emodin (as well as other anthraquinones), which occurs in R. cathartica leaves (Maw, 1981), in the evolution of plant defense against phytophagous insects. A long coevolution period in Europe, but not in Canada, may enable the insects to penetrate the defensive properties of R. cathartica. This is based on the assumption that a long history of association between certain groups of insects and plants caused a progressive accumulation of adaptations enabling insects to tolerate feeding deterrents (Bernays & Chapman, 1994).
There are several indications that at certain levels emodin is toxic and deters vertebrates. For example, yellow-vented bulbuls (Pycnonotus xanthopygos) and house sparrows (Passer domesticus) always preferred to feed on emodin-free banana-mash diet than on diets that contained > 0.001% emodin aglycone (f. wt) (Tsahar, 2001). Acute toxicity tests with emodin that was given orally to redwing blackbirds (Agelaius phoeniceusin) and starlings (Sternus vulgaris) revealed that LD50 (which estimates the dose of a substance needed to kill half of a group) was > 100 mg kg−1 (Schafer et al., 1983). The oral administration of emodin to 1-d-old cockerels caused moderate diarrhoea and mortality within 5 d of ingestion (LD50 = 3.7 mg kg−1, Wells et al., 1975). The capacity of emodin to deter mammals is less well documented. In feeding trials, captive white-footed mice (Peromyscus leucopus) almost totally avoided emodin-containing food (Sherburne, 1972).
The mode of action of emodin as a feeding deterrent on birds and mammals is unknown. It is worth mentioning that a closely related but naturally uncommon anthraquinone (9,10 anthraquinone, Thomson, 1987) has been recognized for many years as an avian deterrent (Avery et al., 1997). The mode of action of this compound on birds is controversial. Aversive taste was suggested by Thomson (1988) but refuted by Avery et al. (1997). Either way, a slight difference in the chemical structure of emodin may totally change its pharmacological activity, as is the case with many other secondary metabolites (Waterman, 1992; Wink, 1999). Thus, it is impossible to infer the mode of function of emodin from the mode of function of 9,10-anthraquinone.
The antifeedant property of emodin may also have negative effects on plant reproduction by deterring potential pollinators and seed dispersal agents. However, the antideterrent cost of emodin for plant fitness has not been studied yet.
Allelopathy is defined as the effect of one plant on another plant through the release of a chemical compound in the environment (Rice, 1995). Phenolics are among the major substances that cause allelopathy (Inderjit, 1996). Emodin at 10–100 mg l−1 inhibits the growth of roots and shoots of sunflower (Helianthus annuus, LD50 = 45 mg l−1) and corn (Zea mays var. everte, LD50 = 65 mg l−1) (Hasan, 1998). Inoue et al. (1992) reported that emodin extracted from Polygonum sachalinense exhibited inhibitory activities against the seedling growth of lettuce (Lactuca sativa), green amaranth (Amaranthus viridis) and timothy grass (Phleum pratense). The growth of lettuce seedlings was severely inhibited at relatively low concentrations of emodin (50 ppm). Emodin > 100 ppm inhibited root and hypocotyls growth or leaf sheath growth. It was also argued that allelochemicals present in R. cathartica hinder growth of many herbaceous woodland species surrounding the trees by suppressing their germination (Boudreau & Wilson, 1992). Preliminary studies suggest that the dense shade produced beneath the shrubs may contribute to this effect (Taft & Solecki, 1990) but also that the fruit pulp may contain allelopathic chemicals, including emodin, that leach from fallen fruits (and leaves) into the soil and retard the growth of competing plants (A. Krebsbach & C. Wilson, pers. comm.).
Inoue et al. (1992 ) speculated on the allelopathic mechanisms of emodin in S. sachalinense , based on data on emodin concentrations in rhizomes, aerial part and in the soil. The fresh rhizome with roots of P. sachalinense contained 158 mg kg −1 (f. wt) emodin and the aerial part contained 72 mg kg −1 (f. wt) emodin. Four months after defoliation, the leaves still contained 213 mg kg −1 (d. wt) emodin. Furthermore, emodin content of the soil was 55 mg kg −1 d. wt and was sufficient to inhibit seedling growth of lettuce, green amaranth and timothy grass. Inoue et al. (1992 ) suggested that emodin is released from the rhizome (as glycoside or as aglycone) into the soil and then the glycosides decompose to emodin aglycones, which is the active form that exhibits seedling growth inhibition. In addition, the litter of leaves is decomposed by physical or biological processes that release emodin, which may affect early seedling growth of nearby plants. Their results also demonstrated that emodin is very stable in the ecosystem.
Allelochemicals that are released by plants may also influence nutrient availability in the soil and hence indirectly may affect plant survival (Appel, 1993; Inderjit & Dakshini, 1999). Schlesinger (1991) has suggested that plant phenolics have the potential to influence soil nutrients and rates of nutrient cycling, which ultimately influence plant growth. Northup et al. (1999) have reviewed the different ways that plant polyphenolics affect nutrient cycling and the implications for community structure. Indeed, emodin in the soil directly decreased the Mn2+ availability, increased the Na+ and K+ availability and indirectly decreased the PO43– availability (Inderjit & Nishimura, 1999). Therefore it is likely that emodin that is released by plants adversely affects phosphate-requiring species by depleting phosphate from the soil (Inderjit & Nishimura, 1999). This is in accordance with the assertions of Black (1973) that allelochemicals may cause imbalance in the availability of nutrients in the soil and hence are likely to influence plant growth and distribution.
Emodin may also indirectly affect soil microbial ecology as a result of its effect on the composition of soil inorganic ions, which may have a further influence on the availability of soil nutrients (Inderjit & Dakshini, 1999). To date, the specific contribution of the ion change caused by emodin on microbial activity remains unknown.
Many plant secondary metabolites have an allelopathic effect on microorganisms (Rice, 1995). Surprisingly, it is unclear whether phenols and other secondary metabolites have a role in protecting plants from disease in vivo (Harborne, 1993). But it has been demonstrated that anthraquinones, extracted from different species of Aloe, exhibit antibacterial activity by inhibition of nucleic acid synthesis in Bacillus subtilis (Levin et al., 1988). For many years it has been known that emodin has direct antibacterial activities (Anke et al., 1980; Kitanaka & Takido, 1986). Emodin was found to inhibit nine soil bacterial species (Arthrobacter globiformis, Chlorella pyrenoidosa, Bacillus megaterium, 4 Rhizobium spp., and Azotobacter chroococcum) in vitro with minimal concentrations of 10–200 µg ml−1 (Le Van, 1984). In the presence of emodin, the cells of these bacteria developed aberrant morphological forms, especially enhanced length (Le Van, 1984).
Emodin was also found to be an efficient antifungal toxin. Emodin isolated from Rhamnus triquetra bark was highly effective against spore germination of 17 tested fungi species (including seven species of Alternaria and three species of Fusarium, Singh et al., 1992). Less than 50% inhibition of the spores of all fungi species was achieved at an emodin concentration of 500 µg ml−1 and maximum inhibition (up to 100% in some species) was observed at an emodin concentration of 2000 µg ml−1 (Singh et al., 1992). Emodin was also highly inhibitory against a pathogenic basidiomycete fungus, Fomes annosus (Donnelly & Sharidan, 1986).
From the plant perspective, several direct and indirect nonexclusive ecological consequences of the antimicrobial activity of emodin are possible. Because plant survival frequently depends upon its disease resistance (Rice, 1984), emodin may directly render the plant resistant to disease caused by microbial pathogens. It seems that emodin is a preinfection toxin because it is consistently present in the plant (Tsahar, 2001), and no study has indicated that it increases subsequent to infection as an inducible defense. Singh et al. (1992) recently suggested that the ability of some higher plants to resist fungal pathogens depends on the presence of emodin along with other chemicals in their tissues. In addition, because emodin inhibits soil microorganisms, which may compete with the plants for inorganic nutrients, emodin may indirectly improve the competitive advantage of plants under some circumstances (Rice, 1995; Schmidt & Ley, 1999). But, emodin may also inhibit some mutualistic soil microorganisms, which are beneficial for their host plants and therefore may reduce plant fitness. This possible trade-off has not yet been studied.
To conclude, direct evidence of the ecological role of the antimicrobial activity of emodin in plants in vivo is limited. However, the in vitro studies and other circumstantial evidence indicate that the allelopathic properties of emodin may directly and indirectly affect the survival and growth of plants.
Seed dispersal and germination
Recent studies has demonstrated that emodin might have an impact at different stages of seed dispersal and facilitate plant reproductive success (Tsahar, 2001; Tsahar et al., 2002). The scope here is limited to fleshy-fruited species like Rhamnus spp. as no study yet has investigated the effect of emodin on other types of seed dispersal.
Unripe fleshy fruits are usually well protected chemically against seed predators (Janzen, 1983). Indeed, as demonstrated in the Old World for R. alaternus and R. palestina (lycioides), in the New World for R. cathartica, most bird species do not consume the unripe fruits, probably because they contain unpalatable chemicals (Sherburne, 1972; Maw, 1981; Izhaki, 1986; Tsahar, 2001). Captive American robins (Turdus migratorius) and white-footed mice (Peromyscus leucopus) always preferred ripe over unripe R. cathartica fruits (Sherburne, 1972). They significantly reduced feeding on ripe Prunus virginiana and Rosa multiflora fruits that were coated with emodin but readily accepted ripe fruits without emodin. Furthermore, birds that were starved for 12 h consumed only a few of the green fruits of R. cathartica (Sherburne, 1972). American robins, catbirds (Dumetella carolinensis) and redwings (Agelaius phoeniceus), which were forced to feed on either Rhamnus green fruits or on gelatin capsules containing 50 µg−5 mg emodin, showed similar abnormal physiological effects, such as diarrhoea (Sherburne, 1972). Many mammals and birds have developed detoxification mechanisms that enable them to cope with certain levels of dietary phenols by conversion of the toxic compound in three metabolic reactions: glucuronide synthesis, ethereal sulphate synthesis and methylation (Williams, 1964). From the results described above and the results on bulbuls and house sparrows (Tsahar, 2001), however, it seems that birds and small mammals are unable to detoxify emodin efficiently, particularly when it is present in high concentrations. Hence, emodin in unripe fruit pulp protects against premature fruit consumption.
Secondary metabolites in fruit pulp are often broken down during ripening, a process that is thought to promote fruit consumption when the seeds are mature and ready for dispersal (Janzen, 1983; Garguillo & Stiles, 1993). A study on Rhamnus alaternus indicates that although emodin concentration in the pulp decreases during ripening, it does not fully disappear in ripe pulp (Tsahar et al., 2002). As emodin deters potential seed dispersal agents, its presence in ripe fruits may reduce the amount of fruits eaten and thus may have a negative effect on plant fitness. However, several nonexclusive hypotheses may explain the adaptive significance of secondary metabolites in ripe fruits (Cipollini & Levey, 1997a,b; Cipollini, 2000), some of which seem to be relevant in the case of R. alaternus. Nevertheless, in contrast to the ‘protein assimilation hypothesis’ holds that secondary metabolites reduce fruit digestion and thus force the dispersal agents to forage and deposit the seeds away from the parent tree (Izhaki & Safriel, 1989; Izhaki, 1992), emodin actually improves bird digestion (Tsahar, 2001).
Insects and microorganisms seem to avoid not only the foliage and green fruits of Rhamnus spp. but also the ripe fruits. R. alaternus plants with relatively high emodin concentration in the pulp suffered less seed damage by insects and microorganisms without reducing the total fruit removal by seed dispersal agents than trees with lower concentrations (Tsahar et al., 2002). These findings support the microbe/pest specificity model, which states that negative effects of secondary metabolites should be stronger on microbial and invertebrate pests than on vertebrates, including dispersers (Cipollini, 2000). Despite the presence of certain amounts of emodin in the ripe pulp, relatively large assemblages of avian frugivores consume the ripe fruits of many Rhamnus species and act as seed dispersers (Herrera, 1984; Izhaki & Safriel, 1985; Tsahar, 2001; Izhaki, 2002). Frugivores may avoid the accumulation of secondary metabolites by consuming only a limited number of fruits, thus shortening their foraging bouts on the fruiting plants (Izhaki & Safriel, 1989, 1990b; Barnea et al., 1993). This behavior ensures that the seeds would be deposited away from parent plant and in small groups, thus reducing plant-sibling and among-sibling competition (‘attraction/repulsion’ hypothesis, Cipollini & Levey, 1997a; Cipollini, 2000).
The laxative activity of emodin that was well documented in humans (Van den Gorkom, 1999) was also observed in potential dispersal agents like small-mammals and birds (Sherburne, 1972). Rapid passage of seeds through the gut may ensure their viability. However, the ‘gut retention hypothesis’ (Cipollini & Levey, 1997a) may not apply to Rhamnus spp. because the seeds are encased in a moisture-sensitive envelope that splits and ejects the seeds only after exposure to dry air (Izhaki & Safriel, 1990a). Thus, the seeds are fully protected during the passage through the gut and germination is totally independent of gut characteristics of the various dispersers (Izhaki & Safriel, 1990a; Barnea et al., 1991). Short dispersal distance is the potentially disadvantageous consequence of rapid seed passage through the gut and reflects the potential negative effect of emodin on plant fitness (Cipollini & Levey, 1997a).
The direct effect of emodin on seed germination was studied only for two plant species. The germination of control sunflower (Helianthus annuus) seeds was 98% but emodin treatments (50 and 100 mg l−1) inhibited the germination to 76% and 55%, respectively (Hasan, 1998). However, the same treatments did not affect the germination of corn (Zea mays var. everte) seeds (Hasan, 1998). Furthermore, there is some circumstantial evidence that emodin along with other secondary metabolites may inhibit seed germination of other plant species. Seeds of Rhamnus cathartica and R. alaternus apparently will not germinate unless the fruit pulp is removed either by hand or through seed passage in the digestive tract of dispersal agents (Gourley & Howell, 1984; Barnea et al., 1991). Eight other fleshy-fruited species, including R. palaestinus, germinated within the intact fruits (Barnea et al., 1991), but whether or not their pulp contain emodin is unknown. Nevertheless, emodin (and other compounds) in the pulp may inhibit germination of Rhamnus seeds, depending upon concentration, either via direct chemical inhibition as demonstrated for other seed species or by controlling microenvironment factors such as light. Such an autoallelopathy mechanism may prevent seed germination within the fruits while they are either still attached to the mother plant or on the ground beneath the mother plant. In both cases the germination of undispersed seeds would be detrimental to plant fitness and hence emodin may play an important role as a seed germination inhibitor that allowed germination only after the seeds are dispersed (Cipollini & Levey, 1997a).
To conclude, emodin in unripe and ripe pulp may contribute to reduction of predispersal seed predation by insects and microorganisms, prevention of dispersal of premature seeds by vertebrates, dispersal of seeds away from the parent plant and inhibition of seed germination within the fruit. Although these functions have an adaptive value as they increase plant fitness (Cipollini, 2000) some of them may induce negative effects on plant fitness as well. Yet, the adaptive approach to the presence of secondary metabolites in fruits is controversial (Eriksson & Ehrlén, 1998) and the possible trade-offs of their presence for plant reproduction were not studied yet.
Antioxidant activity of emodin
The UV region of the solar radiation (c. 290–400 nm) is potentially most damaging to higher plants (Lumsden, 1997). The studies in the last two decades on the effectiveness of naturally occurring compounds in plants as antioxidants revealed that phenols have superior antioxidant capacity, and it was argued that they may have evolved as a protection against the intense UV radiation endured by early angiosperms (Swain, 1986; Cheeke, 1998). Emodin and Cassia tora extracts that contain emodin do have a photo-protective function (Yen et al., 1998). Yuan and Gao (1997) also reported on the antioxidant activity of emodin that was shown to be a potent inhibitor of superoxide radicals. The antioxidant activity of emodin probably depends on scavenging hydroxyl radicals (Yen et al., 2000). Thus, emodin, especially when occurring on the leaf surface, may protect plants against harmful environmental effects of excessive incident UV light.
Emodin is not alone
Because plants contain numerous secondary metabolites it is important to recognize that organisms such as herbivores, frugivores, granivores and pathogens do not interact with chemicals in isolation. For example, herbivores that ingest Rhamnus leaves simultaneously consume several other anthraquinones, tannins and flavonoids (Paya et al., 1986; Coskun, 1989, 1992; Abegaz & Peter, 1995) in addition to emodin. Frugivores that feed on the fruit of Rhamnus prinoides actually ingest emodin, physcion, rhamnazin, prinoidin, and many other emodin-derived compounds (Abegaz et al., 1996). Likewise, the roots of Rheum franzenbachii contain five anthraquinones (emodin, chrysophanol, physcion, aloe-emodin and rhein (Ma et al., 1989)).
Because of the complexity of chemical interactions in tissues of living plants, a popular approach in chemical ecology has been to quantify the total amount of all compounds that belong to a single class (e.g. total phenolics, total alkaloids, etc.) and to associate them with a common biological property (Rosenthal & Berenbaum, 1991; Berenbaum & Zangerl, 1992). Several of the anthraquinones, such as emodin, rhein, chrysophanol and aloe emodin, indeed have common laxative activity (Van den Gorkom et al., 1999). Furthermore, the higher laxative effect of a combination of anthraquinones was explained by their synergistic effects (Rauwald, 1998). Yet, individual anthraquinones also exhibit different biological activities due to their different structures (Evans, 1996; Bruneton, 1999). Minor differences in chemical structure can lead to appreciable differences in biological activity because the addition or reduction of a simple element may open new molecular targets (Waterman, 1992; Waterman & Mole, 1994; Wink & Schimmer, 1999). Small differences among anthraquinones, for instance, may determine the perception and toxicity of different anthraquinones to different consumers, with potentially large effects for the plants (Bowers, 1988; Wink & Schimmer, 1999). Therefore, quantifying the total anthraquinones in plants may be sufficient to predict purgative effects but may not be sufficient to predict the full effects of these compounds on plant fitness.
When a single compound has a profound effect on organisms, the appropriate approach is to isolate and quantify this compound and to investigate its ecological attributes (Berenbaum & Zangerl, 1992). Given the large and varied effects of emodin, it seems valid to consider the distinct attributes of it as an isolated compound. However, because there is a potential for chemical interactions and synergism of emodin with other secondary compounds, we should be aware that its isolated ecological functions might not accurately reflect its function within the chemical milieu of a plant.
The evolution of multifunctionality of emodin
Emodin is almost inevitably multifunctional in its effects. It exhibits numerous biological activities; several of them are thought to play significant ecological roles in the life of many plant species by mediating their interactions with their biotic and abiotic environment. But the multifunctional attribute of emodin is probably an example of a much wider phenomenon of ‘multipurpose defense’ secondary metabolites (Wink & Schimmer, 1999). Although the many biological functions attributable to emodin may in part reflect more intensive study of this compound in comparison to other secondary metabolites, it no doubt contradicts the assumption that secondary metabolites are highly specific in their biological functions (see also Cipollini, 2000).
Two nonexclusive hypotheses may explain the evolution of multifunctionality of secondary metabolites. Firstly, plants have to cope with a wide range of unpredictable biotic and abiotic constraints, and selection should favor the production of a single compound that exhibits the ability to protect the plant against as many potential enemies and abiotic hazards as possible (Wink & Schimmer, 1999). The available data suggest that in some species emodin is present in certain plant parts all year long, and therefore it constantly protects these parts against unpredictable environmental risks like herbivory (both insects and vertebrates), pathogenic diseases, competition (allelopathy) and high light intensities. However, the data presented in Table 1 suggests that in many species emodin has not been found throughout all parts of the plant. This could simply be a result of lack of emodin analyses in different plant parts. But in case it represents the actual emodin distribution among plant parts, an alternative interpretation of these data would be that there are trade-offs in the production of emodin. Thus, emodin presence in different plant parts at different concentrations in different species reflects the diversity of evolutionary trajectories and ecological conditions in which emodin is found, and potentially, the diversity of different adaptive functions it may have. Secondly, the defense strategy of plants should be dynamic because the enemies usually evolve resistance. Selection should maintain a single compound as long as it is able to affect different molecular targets of the same enemy simultaneously. The probability that the enemy will evolve resistance against multitarget molecules is much lower than for a single-target molecule and thus provides an advantage for the plant in an arms race (Wink & Schimmer, 1999). The phenolic OH-groups enable anthraquinones to interact with proteins by forming hydrogen and ionic bonds (Wink & Schimmer, 1999). This explains the various interactions of emodin with enzymes, transporters, channels and receptors. As these interactions are possible with every protein, we would expect multiple targets and multifunctional activities (Wink & Schimmer, 1999). Thus, emodin in higher plants may interfere with the same enemy through a variety of molecular targets and mechanisms.
Although simultaneous multiple selection pressures may be responsible for the production of multifunctional chemical compounds, the conditions under which the chemical first evolved may be different than the conditions that the plants currently experience, such that the original selection pressures that gave adaptive advantage to the new chemical are either a subset of, or are even completely different from, the selection pressures that currently maintain the adaptive advantage of synthesis (Waterman, 1992; Cipollini, 2000). Furthermore, emodin may have evolved in fungi and lichen independent of its occurrence in higher plants. Although biotic and abiotic factors may play a marginal role in the evolution of the compound itself, the distribution and concentration of emodin within different plants could be shaped by multiple interactions of plants with their biotic and abiotic constraints. Cooper-Driver & Bhattacharya (1998) suggested that a compound, which arose early in the evolution of plants, might gradually take on new functions. However, plant families that currently contain emodin do not have monophyletic relationships and therefore we cannot preclude the possibility that emodin arose independently on a number of occasions (Wink & Waterman, 1999). In any case, selection would probably maintain emodin in higher plants as long as its benefit to fitness outweighs its production cost.
Multiple functions of a single secondary metabolite have fascinated many researchers. Cipollini (2000) called such chemical compounds ‘jacks of all trades’ and Wink & Schimmer, 1999) explained the evolution of such compounds in terms of nature's tendency ‘to catch as many flies with one clap as possible’. I suggest that emodin is a case study representing a much broader phenomenon among secondary metabolites that have evolved as an integral part of the diverse interactions between plants and their biotic and abiotic environment. Nevertheless, most of the studies reviewed here were not designed to examine the multifunctional approach. Furthermore, there are many gaps in our understanding of costs and benefits of emodin production and function. Future studies should focus on the differential distribution of emodin among plant parts and the biotic and abiotic factors that are associated with emodin production in these parts. The importance of basic abiotic factors such as seasonality and soil minerals and water on emodin production are mostly unknown. To advance our understanding of why plants do and do not contain secondary metabolites in various tissues we should also consider the trade-offs presented by their production. Although this review reflects the lack of studies that explored the potential cost of emodin production on plant fitness, only by considering both costs and benefits in the chemical multifunctionality puzzle will we be able to develop predictive models that explain emodin occurrence.
My aim here has been to increase the awareness of chemical multifunctionality in plant secondary metabolites and to stimulate studies focusing on multiple ecological functions of single secondary compounds in higher plants. Integrated multidisciplinary efforts from ecologists, biochemists and molecular biologists will be needed for this effort. Future ecological studies of plant-secondary metabolites will be well served by integrating multiple pathways by which metabolites may mediate interactions of plants with their biotic and abiotic environment.
I am grateful to William Setzer, Phyllis Coley, Michael Wink and three anonymous reviewers for critical comments on earlier versions of the manuscript, and Ella Tsahar, Jacob Friedman, Emma Kvitnitsky, Irena Paloy, Zvia Shapira, Moshik Inbar, Josef Katz, Doug Levey and Marty Cipollini for their help in developing my ideas about the adaptive roles of emodin. This research was supported by grants from the University of Haifa and Oranim. The manuscript was written while the author was on sabbatical leave at the Department of Zoology, University of Florida.