Plant compounds that are perceived by humans to have color are generally referred to as ‘pigments’. Their varied structures and colors have long fascinated chemists and biologists, who have examined their chemical and physical properties, their mode of synthesis, and their physiological and ecological roles. Plant pigments also have a long history of use by humans. The major classes of plant pigments, with the exception of the chlorophylls, are reviewed here. Anthocyanins, a class of flavonoids derived ultimately from phenylalanine, are water-soluble, synthesized in the cytosol, and localized in vacuoles. They provide a wide range of colors ranging from orange/red to violet/blue. In addition to various modifications to their structures, their specific color also depends on co-pigments, metal ions and pH. They are widely distributed in the plant kingdom. The lipid-soluble, yellow-to-red carotenoids, a subclass of terpenoids, are also distributed ubiquitously in plants. They are synthesized in chloroplasts and are essential to the integrity of the photosynthetic apparatus. Betalains, also conferring yellow-to-red colors, are nitrogen-containing water-soluble compounds derived from tyrosine that are found only in a limited number of plant lineages. In contrast to anthocyanins and carotenoids, the biosynthetic pathway of betalains is only partially understood. All three classes of pigments act as visible signals to attract insects, birds and animals for pollination and seed dispersal. They also protect plants from damage caused by UV and visible light.
Plants produce more than 200 000 different types of compounds (Fiehn, 2002), including many colored (pigmented) ones. Humans recognize the color of a compound by perceiving reflected or transmitted light of wavelengths between 380 and 730 nm, while insects recognize light of shorter wavelengths (Davies, 2004). This review will focus on recent progress in understanding the biosynthesis of three main classes of pigments for coloration in plants: flavonoids/anthocyanins, betalains and carotenoids. Engineering of flavonoid and carotenoid biosynthetic pathways will be discussed separately (Tanaka and Ohmiya, 2008). The genetics and biochemistry of these pigments have recently been reviewed (Grotewold, 2006a).
Flavonoids, a group of secondary metabolites belonging to the class of phenylpropanoids, have the widest color range, from pale-yellow to blue. In particular, anthocyanins, a class of flavonoids, are responsible for the orange-to-blue colors found in many flowers, leaves, fruits, seeds and other tissues (Figure 1a–d). They are widely distributed in seed plants, are water-soluble, and are stored in vacuoles. Betalains, yellow-to-red nitrogen-containing compounds, are derived from tyrosine. They are also water-soluble and stored in vacuoles, but are found only in the order Caryophyllalles. Carotenoids, which are isoprenoids and are found ubiquitously in plants and micro-organisms, are essential components of photosystems and confer yellow-to-red coloration to flowers and fruits. Flavonoids/anthocyanins and carotenoids are often present in the same organs, and their combination increases color variety.
Flavonoids are among the best-characterized plant secondary metabolites in terms of chemistry, coloration mechanism, biochemistry, genetics and molecular biology (Grotewold, 2006b; Harborne, 1988, 1994; Stafford, 1990). The functions of flavonoids/anthocyanins, including their potential health benefits, have been reviewed (Harborne and Williams, 2000). Anthocyanins are widely used as natural food colorants, for example, the cyanidin acylglucosides from red cabbage (Brassica oleracea var. capitata) and Perilla frutescence (Figure 1e).
Structures and colors
Flavonoids, with a basic structure of C6–C3–C6, are widely distributed among land plants. Depending on their structures, flavonoids may be classified into about a dozen groups, such as chalcones, flavones, flavonols and anthocyanins. Modification of flavonoids with hydroxyl, methyl, glycosyl and acyl groups results in several thousand structures. An excellent flavonoid database is available (http://www.metabolome.jp/software/FlavonoidViewer/).
Orange to blue: anthocyanins
A total of 19 types of anthocyanidins, aglycons or chromophores of anthocyanins are known, but there are only six major ones: pelargonidin, cyanidin, peonidin, delphinidin, petunidin and malvidin (Figure 2a). Their color greatly depends on the number of hydroxyl groups on the B-ring; the larger the number of groups, the bluer the color. O-Methylation of anthocyanins has a slight reddening effect.
Anthocyanidins are modified by glycosyl moieties in versatile ways in a family- or species-specific manner. Anthocyanins are most frequently O-glycosylated (usually glucosylated) at the C3-position, followed by the C5-position. Glycosylation at C7′, C3′ or C5′ is often found. Glycosylation of anthocyanins results in slight reddening. The glycosyl moieties of anthocyanins are commonly modified by aromatic (hydroxycinnamic or hydroxybenzoic) and/or aliphatic (malonic, acetic, or succinic) acyl moieties. Aromatic acylation causes a blue shift and stabilizes anthocyanins. Anthocyanins modified with multiple aromatic acyl moieties (poly-acylated anthocyanins; Honda and Saito, 2002; Figure 1a) often show a stable blue color via intermolecular stacking. Aliphatic acylation does not change the color but increases the stability and solubility. 3-Deoxyanthocyanins, which are found in a limited number of plant species, such as maize (Zea mays), and 6-hydroxy anthocyanins, which are found in Alstroemeria, show a redder color than the corresponding anthocyanins.
The color of anthocyanins changes depending on the pH, co-existing colorless compounds (co-pigments, typically flavones and flavonols), and metal ions. In vitro, anthocyanidins are redder and more stable as the flavilium cation form at lower pH (pH <3), colorless under mildly acidic conditions (pH 3–6), and bluer and unstable as the quinonoidal base form at pH 6 and above. Intermolecular stacking by self-association or with flavones or flavonols (often called co-pigments) stabilizes anthocyanins and causes a bathochromic shift (blueing and intensifying of color). Metal ions, such as Al3+ and Fe3+, play a critical role in the generation of blue flowers in hydrangea (Hydrangea macrophylla; Yoshida et al., 2003) and tulip (Tulipa gesneriana; Shoji et al., 2007), respectively. A subtle stoichiometric balance of a cyanidin-based anthocyanin, a flavonol, Fe3+ and Mg2+ is critical to generation of the sky-blue petals of Meconopsis grandis (Himalayan poppy; Figure 1c; Yoshida et al., 2006). In cornflower (Centaurea cyanus), a super-molecular complex of six molecules each of cyanidin 3-(6-succinylglucoside)-5-glucoside and apigenin 7-glucuronide-4-(6-malonylglucoside), one ferric ion, one magnesium ion and two calcium ions is responsible for the blue flower color (Shiono et al., 2005). It is yet to be determined, however, how plants achieve such subtle balances of each component to generate appropriate colors.
Yellow: chalcones, aurones, flavonols and flavones
Many pale-yellow flowers, such as carnations (Dianthus caryophyllus) and cyclamens (Cyclamen persicum), contain 2′,4,4′,6′-tetrahydroxy chalcone (THC) 2′-glucoside (isosalipurposide). Safflower (Cathamus tinctorius) produces a rare yellow chalcone glucoside, safflomin A, which is used as a colorant. Some Asteraceae (the chrysanthemum family) produce pale-yellow 6′-deoxy chalcones. Aurones, a class of rare flavonoids, give brighter yellow flowers than chalcones, and are found in a limited number of species, such as snapdragon (Antirrhinum majus) cosmos (Cosmos bipinnatus) and Limonium. Flavonols and flavones are very pale-yellow and are mostly invisible to the human eye. As they absorb UV, which insects recognize, they give color and patterns to flowers to attract insects.
Biosynthesis of flavonoids
Main pathway. The flavonoid biosynthetic pathway is shown in Figure 2(b). The pathway is well understood and is conserved among seed plants. Flavonoids are synthesized in the cytosol. It has been proposed that the biosynthetic enzymes form a super-molecular complex (metabolon) via protein–protein interaction and are anchored in the endoplasmic reticulum (ER) membrane (Grotewold, 2006a; Winkel, 2004). The biosynthetic enzymes belong to various enzyme families, such as 2-oxoglutarate-dependent dioxygenases (OGD), cytochromes P450 (P450) and glucosyltransferases (GT), which suggests that plants recruited these enzymes from pre-existing metabolic pathways. Phylogenetic analysis indicates that genes encoding enzymes with the same or similar activity had diverged before the speciation of seed plants (Tanaka and Brugliera, 2006), with some exceptions (see below).
Chalcone synthase (CHS), a polyketide synthase, is the first committed enzyme in the pathway, and it catalyzes the synthesis of THC from one molecule of 4-coumaroyl CoA and three molecules of malonyl CoA. THC is rapidly and stereospecifically isomerized to the colorless (2S)-naringenin by chalcone isomerase (CHI). (2S)-Naringenin is hydroxylated at the 3-position by flavanone 3-hydroxylase (F3H) to yield (2R,3R)-dihydrokaempferol, a hydroflavonol. F3H belongs to the OGD family. F3H also catalyzes the hydroxylation of eryodictyol and pentahydroxyl flavanones to dihydroquercetin and dihydromyricetin, respectively.
Flavonoid 3′-hydroxylase (F3′H) and flavonoid 3′,5′-hydroxylase (F3′5′H), which are P450 enzymes, catalyze the hydroxylation of dihydrokaempferol (DHK) to form (2R,3R)-dihydroquercetin and dihydromyricetin, respectively. F3′H and F3′5′H determine the hydroxylation pattern of the B-ring of flavonoids and anthocyanins, and are necessary for cyanidin and delphinidin production, respectively. They are the key enzymes that determine the structures of anthocyanins and thus their color (Figure 2b; Tanaka, 2006). F3′H and F3′5′H catalyze the hydroxylation of flavanones, flavonols and flavones. Many important floricultural crops, such as roses (Rosa hybrida), chrysanthemums (Chrysanthemum morifolium) and carnations, do not produce delphinidin and thus lack violet/blue color varieties. This is attributed to the fact that they do not possess the F3 ′5 ′H gene, probably because they lost it during their evolution. Interestingly, some species of Asteraceae have re-acquired the F3′5′H function from their F3 ′H gene by convergent evolution (Seitz et al., 2006). Transgenic blue/violet carnations and roses have been developed by expressing a heterologous F3 ′5 ′H gene (Chandler and Tanaka, 2007; Katsumoto et al., 2007).
Dihydroflavonols are reduced to corresponding 3,4-cis-leucoanthocyanidins by the action of dihydroflavonol 4-reductase (DFR). In some plant species, such as petunia (Petunia hybrida) and cymbidium (Cymbidium hybrida), DFR has strict substrate specificity and cannot utilize dihydrokaempferol. This is the reason that these species lack pelargonidin-based anthocyanins and thus lack flowers of an orange/brick red color.
Anthocyanidin synthase (ANS, also called leucoanthocyanidin dioxygenase), which belongs to the OGD family, catalyzes the synthesis of corresponding colored anthocyanidins. The genes encoding the enzymes mentioned above have been isolated and characterized from flowers of many plants, including petunia, snapdragon (Antirrinum majus), gentian (Gentiana triflora), torenia (Torenia hybrida), morning glories, and other tissues of maize, Perilla and Arabidopsis.
Modification of anthocyanidins. In contrast to the well-conserved main pathway of flavonoid biosynthesis mentioned above, the modification of anthocyanidins is family- or species-dependent and full of diversity. Various UDP-glycose-dependent glycosyltransferases belonging to glycosyltransferase family 1 (http://www.cazy.org/fam/acc_GT.html), acyltransferases (ATs) mainly belonging to the BAHD family (D’Auria, 2006) and S-adenosylmethionine-dependent methyltransferases are responsible for such diversity. These enzymes whose genes have been isolated are summarized in Figures S1 and S2, and only recent interesting results are shown here. In general, these enzymes are specific to the position of modification on the anthocyanin and the donor substrates, but less specific to the anthocyanin structure, i.e. the enzymes catalyzing modification of A- and C-rings can modify anthocyanins irrespective of the hydroxylation pattern on the B-ring.
Anthocyanidins are initially 3-glucosylated by the action of UDP-glucose:flavonoid (or anthocyanidin) 3GT. However, in roses accumulating anthocyanidin 3,5-diglucoside, 5-glucosylation precedes 3-glucosylation, and both glucosylation reactions are catalyzed by a single enzyme, UDP-glucose:anthocyanidin 5,3GT (Ogata et al., 2005). Roses also have the conventional 3-glucosylation pathway catalyzed by 3GT, which functions in stressed cultured cells (Hennayake et al., 2006) and mature petals in some cultivars (Mizutani et al., 2007). While the phylogenetic analysis of GTs indicates that 3GT, 5GT and 3′/7GT form separate clusters (Figure S1), the amino acid sequence of 3′,5′GT from butterfly pea (Clitoria ternatea) is related to 3GTs rather than to gentian 3′GT (Figure S1), indicating that divergent evolution occurred after speciation of the pea (Noda et al., 2004).
Unlike other enzymes in the flavonoid biosynthetic pathway, BAHD-type anthocyanin AT genes diverted after speciation of seed plant families (Figure S2; Luo et al., 2007; Tanaka and Brugliera, 2006). Furthermore, the anthocyanin 3′,5′AT of butterfly pea is an acylglucose-dependent acyltransferase and a member of the serine carboxypeptidase-like (SCPL) family that localizes in the vacuole (Noda et al., 2006, 2007). Acyl glucose-dependent AT activity has also been reported in carrot (Daucus carota; Glassgen et al., 1998; Okuda et al., 2007). Elucidation of non-characterized biosynthetic enzymes and the mechanism for production of poly-acylated anthocyanins from plants such as cineraria (Senecio cruenta) and delphinium (Delphinium ajacis) may indicate more diversity.
The crystal structure of a chrysanthemum anthocyanin manolonytransferase and subsequent mutational study identified the binding sites of acyl CoA and the acyl acceptor (Unno et al., 2007). Salvia (Salvia splendens) anthocyanin 5-malonyltransferase was successfully modified to show hydroxycinnamoyltransferase activity towards the 3-glucosyl and 5-glucosyl moieties of anthocyanin in vitro by substituting only three amino acid residues (Suzuki et al., 2007). This study exemplifies the possible engineering of enzymes with desirable substrate specificity and kinetic properties.
Biosynthesis of aurones, flavonols and flavones. THC is unstable and requires stabilization by glucosylation or methylation in order to show its yellow color. Mutations of the CHI gene have been shown to be necessary for accumulation of isosalipurposide in carnation (Forkmann and Dangelmayr, 1980) and barley (Marinova et al., 2007a). Additional mutation of the DFR gene confers better yellow coloration in carnation (Itoh et al., 2002). The existence of several genes encoding THC 2′GT activity from carnation (Figure S1; Ogata et al., 2004; Okuhara et al., 2004) indicates that several non-specific enzymes may catalyze the reaction in vivo, unlike the case of anthocyanin GTs. In snapdragon, THC 4′-glucoside is synthesized by the action of THC 4′GT, presumably in the cytosol (Ono et al., 2006), transported to the vacuoles, and further converted to aureusidin 6-glucoside by aureusidin synthase [AS, a polyphenol oxidase (PPO)] in the vacuoles (Nakayama et al., 2000). As aurones are found in a limited number of unrelated species, several types of enzymes may be able to catalyze aurone biosynthesis.
Flavonols and flavones are synthesized from corresponding dihydroflavonols and flavanones by the action of the OGD flavonol synthase (FLS) and flavone synthase (FNS). Interestingly, there are two types of FNS: OGD (FNSI) and cytochrome P450 (FNSII). FNSI has been found in parsley (Petroselium crispum), and is suggested to have evolved from F3H (Martens et al., 2003). FNSII is more common, and is related to flavanone 2-hydroxylase (F2H) in legumes (Akashi et al., 1999), in which flavones are synthesized in two steps (F2H and dehydratase) from flavanones.
Transport to the vacuole and vacuolar pH regulation
Flavonoid glycosides, including anthocyanins, are usually transported into the vacuole. The transport mechanism is less well understood than the biosynthesis. Transport mechanisms may be redundant or depend on plant species and organs. The first and most established mechanism involves transport of anthocyanins via a glutathione S-transferase (GST)-like protein and a multi-drug resistance-like protein (a type of ABC transporter). Involvement of the former has been shown in maize (Marrs et al., 1995), petunia (Alfenito et al., 1998) and Arabidopsis (Kitamura et al., 2004), and the latter has been identified in maize (Goodman et al., 2004). The molecular mechanism whereby these proteins, especially GSTs, achieve the transport has not yet been clarified. A second mechanism involves vesicle-mediated mass transport of anthocyanins to vacuoles, as has been observed in, for example, lisianthus (Eustoma grandiflorum; Zhang et al., 2006). Anthocyanins are targeted directly to the protein storage vacuole via ER-derived vesicles in Arabidopsis seedlings, and this process does not depend on GST activity or an ATP-dependent transport mechanism (Poustka et al., 2007). A third mechanism may involve an Arabidopsis multi-drug and toxic compound extrusion (MATE) transporter (TT12), which is a vacuolar flavonoid/H+-antiporter that is necessary for vacuolar accumulation of proanthocyanidins and has been shown to mediate anthocyanin transport in vitro (Marinova et al., 2007b). Interestingly, intact flavonoid biosynthesis exerts control over the activity of the vacuolar flavonoid/H+-antiporter in barley (Marinova et al., 2007a). The molecular mechanism whereby a plant cell transports many kinds of flavonoids into a vacuole remains to be elucidated.
Regulation of the vacuolar pH, which greatly affects anthocyanin color, is partly understood. The only known structural gene that regulates vacuolar pH with relevance to color is the Japanese morning glory (Ipomea nil) Na+/H+-antiporter that is specifically expressed before flower opening and increases the vacuolar pH to generate blue flowers (Fukada-Tanaka et al., 2000). Petunia loci that regulate vacuolar pH have also been identified (Koes et al., 2005), and the petunia PH4 gene, which activates vacuolar acidification, has been shown to be R2R3 Myb (Quattrocchio et al., 2006).
The spatial and temporal expression of structural genes in anthocyanin biosynthesis is determined by a combination of R2R3 Myb, basic helix–loop–helix (bHLH) and WD40-type transcriptional factors and their interaction. This has been well established in maize, Arabidopsis, petunia and some other plants (reviewed by Koes et al., 2005), and in Japanese morning glory (Morita et al., 2006). The WD40 and bHLH proteins are pleiotropic and are involved in multiple processes in addition to anthocyanin synthesis, such as the control of vacuolar pH in petunia flowers and the formation of trichomes and root hairs in Arabidopsis. It is believed that they affect these processes via their interactions with specific MYB proteins, such as PH4 in petunia and GL1/Wer in Arabidopsis (Koes et al., 2005).
Intimate functional analysis of maize R (a bHLH) has provided new insight into its regulatory mechanism. R has a novel dimerization domain for its regulatory activity, and this domain was evolved from the ACT domain that is involved in the allosteric regulation of many amino acid metabolic enzymes (Feller et al., 2006). The bHLH region of R recruits an EMSY-like maize nuclear factor to flavonoid biosynthetic gene promoters, and this recruitment is associated with histone acetylation (Hernandez et al., 2007). R2R3 Myb contributes to speciation of some plant species. A small family of R2R3 Myb genes controls floral pigmentation intensity and patterning in the genus Antirrhinum, and their activity causes natural variation in anthocyanin pigmentation (Schwinn et al., 2006). Mutations in AN2 (an R2R3 Myb gene) result in acyanic corolla limbs in the wild petunia, P. axillaris, which has white moth-pollinated flowers, while P. integrifolia, which has a functional AN2 allele, has magenta bee- and butterfly-pollinated flowers (Hoballah et al., 2007).
Betalains show brilliant color in flowers (Figure 1f–i) or fruits of species belonging to the families of Caryophyllales, except for Caryophyllaceae and Molluginaceae. The mutual exclusiveness of anthocyanins and betalains in the Caryophyllales has given rise to considerable taxonomic debate (Clement and Mabry, 1996). Poor understanding of betalain biosynthesis in terms of biochemistry and molecular biology has prevented researchers from interpreting this taxonomic mystery. Betalains from red beet (Beta vulgaris) are used as a natural colorant. The advantage of betalain color is that the color does not depend on the pH and is more stable than that from anthocyanins.
Structures and colors
Betalains are classified into red (crimson) betacyanins and yellow betaxanthins. They are immonium conjugates of betalamic acid with cyclo-dihydroxyphenylalanine (cDOPA) glucoside and amino acids or amines, respectively (Figure 3; Strack et al., 2003). Only betacyanins are modified by glycosyl or acyl moieties. More than 50 molecular species of betacyanins and several betaxanthins have been isolated and identified, and novel betalain molecules are being reported in accordance with the progress in development of analytical equipment (Kugler et al., 2007; Wybraniec et al., 2007).
The earliest biological research on betalains was on genetic inheritance of the flower colors red (crimson), yellow and white in ‘four o’clocks’ (Mirabilis jalapa) by Mendel (Mendel, 1950). Later it was discovered that the C (Color) gene is required to synthesize both the yellow and red pigments, and that the R (Red) gene acts downstream of the C gene to yield the red pigment from the yellow pigment in the flowers of portulaca (Portulaca grandiflora; Trezzini and Zryd, 1990). A similar result was obtained in studies on flower coloring of M. jalapa (K. Kasahara, Yokohama, Japan, unpublished results).
Biosynthesis of betalains
The biosynthetic pathways of betalains and the enzymes and genes involved in the pathway are much less well understood than those of flavonoids and carotenoids. The biosynthetic steps were first proposed a quarter of a century ago, but mainly on the basis of phytochemical evidence, such as feeding experiments. Advances in molecular biological techniques will permit elucidation of the pathway in terms of biochemistry and molecular biology. Here, recent significant progress regarding the betalain biosynthetic pathway is reviewed, with particular focus on DOPA 4,5-dioxygenase (DOD) and cDOPA 5-O-GT (cDOPA5GT).
The biosynthetic pathways consist of several enzymatic reaction steps and spontaneous chemical reaction steps (Figure 3). DOPA formation is catalyzed by tyrosine hydroxylase (enzyme I in Figure 3), betalamic acid formation by DOPA 4,5-dioxygenase (DOD; enzyme II), cDOPA formation by plant PPO or DOPA oxidase (enzyme III), conjugation of betalamic acid and amino acid, amine or cDOPA by enzyme VIII, and modification with sugar molecules and aliphatic or aromatic compounds by enzymes IV–VII. Although DOPA is an important precursor not only of betalains but also of various secondary metabolites in plants, there have only been a few reports about the partial purification of tyrosine hydroxylase from P. grandiflora (Steiner et al., 1996; Yamamoto et al., 2001). cDOPA is presumably synthesized from DOPA by the action of PPO, but there is no direct evidence to prove it in vitro except correlation of the accumulation of PPO mRNA and betacyanin contents in tissues of pokeweed (Phytolacca americana; Joy et al., 1995). Recently, some betalains have been suggested to be synthesized by tyrosinase or PPO after the condensation of betalamic acid and amino acid (Gandia-Herrero et al., 2005a,b); however, the tyrosinase has not been successfully purified, nor has the cDNA encoding it been identified from betalain-producing plants. The condensation step of betalamic acid and amino acid, amine or cDOPA probably occurs as a spontaneous chemical reaction in vivo, as suggested by feeding experiments (Hempel and Boem, 1997; Schliemann et al., 1999).
As shown in Figure 3, betalains are synthesized through two independent pathways from DOPA, i.e. the betalamic acid biosynthetic pathway (enzyme II and onwards) and the cDOPA synthetic pathway (enzyme III and onwards). Betalamic acid is the chromophore molecule of both betacyanins and betaxanthins, and cDOPA and its derivatives are essential to produce betacyanin. DOD and cDOPA synthase are crucial enzymes for betacyanin synthesis.
DOPA 4,5-dioxygenase (DOD). Purification of the DOD enzyme and isolation of the gene encoding DOD were first achieved in fungi (Giord and Zryd, 1991). The fungal DOD converts DOPA to both betalamic acid and muscaflavin. As muscaflavin has not been detected in plants (Mueller et al., 1997), the biochemical properties of fungal DOD are likely to be different from those of plant DOD. Subsequent attempts to identify plant DOD protein homologs by immunoscreening using an antibody to the fungal DOD have not been successful, further indicating that fungal and plant DODs are structurally divergent (Christinet et al., 2004). However, Christinet et al. (2004) identified a candidate DOD cDNA from P. grandiflora using a cDNA subtraction method. The sequence of P. grandiflora DOD (PgDOD) is homologous to that of bacterial extradiol 4,5-dixoygenase. The activity encoded by the candidate cDNA was assayed in a P. grandiflora plant of the ccR− genotype, bearing whitish petals, using a particle bombardment method, and betacyanin was synthesized to generate red spots by transient expression. However, recombinant PgDOD expressed in Escherichia coli did not display DOD activity in vitro. DOD homologs have subsequently been found in many plant species irrespective of the production of betalains, even Arabidopsis (Christinet et al., 2004).
The DOD cDNA from M. jalapa has been successfully expressed in yeast and E. coli, and DOD activity has been detected in vitro. Some genes encoding DOD homologs derived from non-betalain-synthesizing plant species have also shown DOD activity in vitro, although the in vivo functions of the homologs are not known. DOD genes in betalain-synthesizing species may have evolved from these DOD homologs (Sasaki et al., 2005c, 2007).
Glucosylation steps. Like anthocyanins, many other plant secondary compounds are synthesized first as a basic skeleton and then modified with sugar moieties by the action of GTs and with aliphatic or aromatic acyl moieties by ATs. In the case of betalains, feeding experiments and early phytochemical studies suggested two routes for the biosynthetic pathways: one involves glucosylation at the betanidin step (enzyme IV, route A, blue arrow in Figure 3), which matches the biosynthetic processes of other plant secondary products, and the other involves glucosylation at the cDOPA step, in which modified cDOPA molecules with sugar and acyl moieties are condensed with betalamic acid (enzyme V, route B, magenta arrow in Figure 3 ). Which pathway is the main route for producing betacyanins in vivo remains controversial.
Evidence for route A was provided by experiments in which exogenously supplied radiolabeled-betanidin was incorporated into betanin in fruits of erect prickly pear (Opuntia dillenii; Sciuto et al., 1972). Biochemical data supporting route A include the identification of glucosyltransferase activity towards betanidin in crude enzyme prepared from cultured cells of livingstone daisy (Dorothenthus bellidiformis; Heuer and Strack, 1992). In addition, two region-specific enzymes that catalyze glucosylation of betanidin [UDP-glucose:betanidin 5-O-GT (B5GT) and UDP-glucose:betanidin 6-O-GT (B6GT)] were purified from livingstone daisy cell cultures, and their enzymatic properties were characterized in detail (Vogt et al., 1997). Isolation of their cDNAs and the subsequent enzymatic characterization of recombinant B5GT and B6GT revealed that they can utilize flavonoids as well as betanidin as substrates (Vogt, 2002; Vogt et al., 1999). B5GT and gentian 3′GT belong to the same cluster (Figure S1).
On the other hand, Sciuto et al. (1974) also reported that cDOPA 5-O-glucoside rather than betanidin or betanin was an efficient precursor of amaranthin (betanidin 5-O-β-glucuronosylglucoside) in feather cockscomb (Celosia cristata), which supports route B. The accumulation of cDOPA 5-O-glucoside in young beet plants (Wyler et al., 1984) and root peels of red beet (Kujala et al., 2001) also supports route B. In addition, cDOPA accumulation has been shown in the petals of mutant M. jalapa plants using phytochemical complementation (Figure S3). Crossing of a yellow flower line with a whitish flower line (W1, shown in Figure S3) gave progeny with yellow flowers. On the other hand, crossing of the same yellow line with another white flower line (W2, shown in Figure S3) produced progeny with red flower (K. Kasahara, Yokohama, Japan, unpublished results). Mixing betalamic acid with the petal extracts derived from W1 and W2 flowers produced yellow and red colors, respectively (Figure S3), which was consistent with the results of the cross. The red mixture contained mainly betanin and traces of betanidin. These results indicate that the extract from the W2 petal contained mainly cDOPA 5-O-glucoside not cDOPA, and that W2 contained cDOPA GT activity.
Activity of cDOPA GT has been detected in crude extracts prepared from red petals of M. jalapa and several betacyanin-producing plants, e.g. portulaca (P. grandiflora), pokeweed (Phytolacca americana), feather cockscomb (C. cristata), bougainvillea (Bougainvillea glabra) and Christmas cactus (Schlumbergera buckleyi), but not from anthocyanin-producing plants, e.g. carnation, chrysanthemum, tulip and anise-scented sage (Salvia guaranitica). In contrast to cDOPA GT activity, anthocyanidin 3GT activity was not detected in betacyanin-producing plants, but was detected in anthocyanin-producing plants (Sasaki et al., 2004). Furthermore, cDNAs encoding cDOPA5GT were isolated from the petals of M. jalapa and the inflorescences of C. cristata (Figure S1; Sasaki et al., 2005b). The profile of the transcripts of the cDOPA5GT gene agreed with the cDOPA5GT activity during development of red petals of M. jalapa, which closely paralleled the synthesis and accumulation of betacyanin. These results support the presence of route B via enzyme V shown in Figure 3, at least in M. jalapa.
Other modification of betalains with sugar and acyl moieties might occur at the cDOPA or glycosylated cDOPA stages rather than the aglycon (betanidin) stage. Recently, UDP-glucuronic acid:cDOPA 5-glucoside glucuronosyltransferase activity (enzyme VI) was detected in C. cristata (Sasaki et al., 2005a), indicating that modification with the glucuronic acid moiety occurs at cDOPA. The modification by acyl moieties on betanidin glucoside, such as a malonyl moiety of phyllocactin (malonylbetanin) and a 4-coumaroyl or ferulic moiety of lampranthins I (4-coumaroylbetanin) or II (feruloylbetanin), may be catalyzed by acyl CoA-dependent BAHD acyltransferase or acyl glucose-dependent acyltransferase belonging to the SCPL enzyme family, as in the case of anthocyanin acylation. As acyl-glucose-dependent acyltransferase activities towards betanidin glucosides (enzyme VII) have been reported (Bokern and Strack, 1988; Bokern et al., 1992), isolation of cDNAs of the SCPL protein family followed by their functional analysis may be a promising way to understand acylation in the betacyanin pathways. The step at which glucosylation and acylation occurs may depend on the species, and both routes may be possible. Obviously, further study is necessary for clarification.
Further characterization of the pathway
Mirabilis jalapa has many kinds of variegated flowers, and Mendel reported many of its intriguing patterns (Mendel, 1950). Such variegated patterns may be caused by transposons (Figure 1i; N. Sasaki and Yoshihiro Ozeki, Tokyo University of Agriculture and Technology, unpublished results). Transposon tagging, which has been successfully used to identify relevant genes in the flavonoid biosynthetic pathways in maize, petunia and morning glories, may be the most efficient way to identify the genes involved in betalain biosynthesis. Characterization of the genes encoding the flavonoid/anthocyanidin biosynthetic enzymes in betacyanin-producing plants has begun (Shimada et al., 2004, 2005, 2007). This may help clarify the taxonomic mystery of Caryophyllales.
Function of carotenoids
Carotenoids are isoprenoid compounds (mostly C40) with polyene chains that may contain up to 15 conjugated double bonds (Figure 4). More than 700 naturally occurring carotenoids have been identified (Britton et al., 1995, 2004). Carotenoids differ from anthocyanins and betalains in that they play essential roles in plant life, for example, photoprotective functions during photosynthesis (Green and Durnford, 1996; Niyogi, 2000) and provision of substrates for biosynthesis of the plant growth regulator abscisic acid (ABA; Nambara and Marion-Poll, 2005) and perhaps other hormones as well (Auldridge et al., 2006). Carotenoids also play an important role in human nutrition and health, providing provitamin A and having anti-cancer activities (Mayne, 1996). Some carotenoids are used as food colorants, cosmetics or pharmaceuticals.
Carotenoids show qualitative differences depending on the plant organs and species. For example, the green tissues of most plants show similar carotenoid profiles, accumulating both β,ε-carotenoids (with one β- and one ε-ring) and β,β-carotenoids (with two β-rings) (Goodwin and Britton, 1988). Carotenoids such as zeaxanthin, violaxanthin, antherxanthin and lutein are invariably found in leaves and stems. In contrast, carotenoids in non-green tissues show distinctive compositions that depend on the plant species. For example, tomato (Solanum lycopersicum) fruit accumulates a large amount of lycopene (Fraser et al., 1994). Capsanthin and capsorbin, ketocarotenoids that contain one and two acyl-cyclo-pentanol rings, respectively, are the typical carotenoids of red pepper (Capsicum annuum) (Hornero-Méndez et al., 2000). Bixa orellana is the only plant that accumulates bixin in its seeds (Bouvier et al., 2003a). Bixin is a dicarboxyl monomethyl ester apocarotenoid, also known as annatto, and is used in food and cosmetics as a red color additive. In general, plants do not accumulate carotenoids in their roots. The storage roots of carrot (Baranska et al., 2006) and sweet potato (Hagenimana et al., 1999) are exceptions. They accumulate a high concentration of β-carotene, and are an important source of vitamin A in the human diet. The majority of carotenoids in the petals of sandersonia (Sandersonia aurantiaca) are β,β-carotenoids, such as β-cryptoxanthin, zeaxanthin and β-carotene (Nielsen et al., 2003). On the other hand, more than 90% of the carotenoids in the petals of marigold (Tagetes sp.; Figure 1j; Moehs et al., 2001) and chrysanthemum (Kishimoto et al., 2004) are lutein and/or lutein derivatives (β,ε-carotenoids).
The main carotenoids of the flower petals of most plants are yellowish xanthophylls, which are pale to deep yellow in color (Table S1). The petals of some plants have a modified carotenoid biosynthetic capacity, accumulate unique carotenoids associated with their respective genus or even species, and are orange to red in color. Astaxanthin (3,3′-dihydroxy-4,4′-diketo-β,β-carotene) is a ketocarotenoid that is produced in a number of bacteria, fungi and algae; it furnishes an attractive orange–red color. Only a few plant species are known to produce astaxanthin. The petals of Adonis aestivalis and A. annua anomalously accumulate a large amount of astaxanthin, resulting in their blood-red color (Figure 1k; Cunningham and Gantt, 2005). The orange petals of calendula (Calendula officinalis) contain reddish carotenoids that are absent in yellow petals (Figure 1l). Some have a cis-structure at C5 or C5′, which is very rare in plants (Kishimoto et al., 2005). The red style branches of crocus (Crocus sativus), from which the spice saffron is derived, accumulate the unique apocarotenoids, crocetin glycosides, picrocrocin and safranal. They are produced by the cleavage of zeaxanthin and are responsible for the color, taste and aroma of saffron (Bouvier et al., 2003b).
Biosynthesis of carotenoids
In the past decade, the genes encoding nearly all the enzymes for carotenoid biosynthesis in plants have been identified, and their enzymatic activities have been characterized (see reviews by Cunningham and Gantt, 1998; Hirschberg, 2001; Howitt and Pogson, 2006). In plants, the entire pathway starting from isopentenyl pyrophosphate (IPP) occurs in the plastids, and it is there that the product accumulates. It is hypothesized that carotenogenic enzymes exist in complex with and associated with the plastid membranes (Cunningham and Gantt, 1998).
Figure 4 summarizes the carotenoid biosynthesis pathway in plants. Carotenoid biosynthesis starts from a C5 isoprene unit, IPP. Four IPPs are condensed to form C20 geranylgeranylpyrophosphate (GGPP). A head-to-head coupling of two GGPP molecules, catalyzed by phytoene synthase (PSY), yields the first C40 carotenoid, phytoene. In tomato, two different types of PSYs (Psy-1 and Psy-2) are expressed in an organ-specific manner (Fraser et al., 1999). Psy-1 encodes a fruit- and flower-specific isoform and is responsible for carotenogenesis in chromoplasts. In green tissues, Psy-2, which is homologous to Psy-1 but highly divergent from it, is predominantly expressed, and makes a major contribution to carotenogenesis in chloroplasts.
Conjugated double bonds are subsequently added by two structurally similar enzymes, phytoene desaturase (PDS) and ζ-carotene desaturase (ZDS). These desaturation reactions yield the intermediates phytofluene, ζ-carotene, neurosporene and lycopene, containing 5, 7, 9 and 11 conjugated double bonds, respectively. Increasing the number of conjugated double bonds shifts the absorption towards longer wavelengths, resulting in colorless phytoene and phytofluene, pale-yellow ζ-carotene, orange–yellow neurosporene and red lycopene. During the desaturation steps, several reaction intermediates with a cis-configuration are produced. Conversion of a cis- to a trans-configuration to form all-trans-lycopene is carried out by carotenoid isomerase (CRTISO), which has been identified in tomato (Isaacson et al., 2002) and Arabidopsis (Park et al., 2002). CRTISO is specific for adjacent double bonds at the 7,9 and 7′,9′-positions, and converts 7,9,9′-tri-cis-neurosporene and 7′,9′-di-cis-lycopene, the products of ZDS, to 9′-cis-neurosporene and all-trans-lycopene, respectively. Recently, Li et al. (2007) reported a second carotenoid isomerase (termed Z-ISO) in maize that converts the 15-cis-bond in 9,15,9′-tri-cis-ζ-carotene, the product of PDS, to 9,9′-di-cis-ζ-carotene, the substrate of ZDS.
The cyclization of lycopene is a branch point in the pathway, catalyzed by lycopene β-cyclase (LCYB) and lycopene ε-cyclase (LCYE). Because LCYE in most plants adds only one ε-ring to lycopene (Cunningham and Gantt, 2001; Cunningham et al., 1996), the pathway in plants typically proceeds only along branches, leading to carotenoids with one β- and one ε-ring (α-carotene and its derivatives) or two β-rings (β-carotene and its derivatives). Lycopene ε-cyclase in romaine lettuce (Lactuca sativa) has the ability to add two ε-rings to lycopene and yields a bicyclic ε-carotene, lactucaxanthin (Cunningham and Gantt, 2001). A single amino acid residue (457th Histidine) is important to form bicyclic ε-carotene.
β- and α-carotenes are further modified by hydroxylation or epoxidation, providing a variety of structural features. The oxygenated derivatives of carotene are called xanthophylls. Hydroxylation of the β- and ε-rings is catalyzed by β-hydroxylase (CHYB) and ε-hydroxylase (CHYE), respectively. CHYB is a non-heme di-iron mono-oxygenase, while CHYE is a P450, CYP97C1 (Tian et al., 2004). CHYB is a well-studied enzyme that has been cloned and characterized from many organisms, while CHYE has been identified only in Arabidopsis. A flower-specific CHYB (CrtR-b2) was recently identified in tomato (Galpaz et al., 2006). Taken together with the existence of flower- and fruit-specific PSY, GGPS and LCYB (tomato expression database, http://ted.bti.cornell.edu/), this finding supports the hypothesis that there is a chromoplast-specific carotenoid biosynthesis pathway.
Epoxidation at positions C5,6 and C5′,6′ of the β-ring of zeaxanthin, catalyzed by zeaxanthin epoxidase (ZEP), yields violaxanthin. Violaxanthin is converted to neoxanthin by neoxanthin synthase (NSY). Both 9-cis-violaxanthin and 9-cis-neoxanthin are cleaved to xanthoxin (C15) by 9-cis-epoxycarotenoid dioxygenase (NCED), and then converted to ABA via the ABA aldehyde intermediate (Nambara and Marion-Poll, 2005).
Regulation of carotenoid biosynthesis and accumulation of carotenoids
Plant tissues, in particular flower petals and fruits, have a wide variety of carotenoid contents, ranging from little or none to large amounts even within the same plant species. There is increasing evidence that carotenogenesis in plant tissues is predominantly regulated at the transcriptional level (see review by Sandmann et al., 2006). In marigold, the differences in petal color from pale-yellow to orange–red are caused by the different levels of accumulation of yellow carotenoid lutein (Figure 1j). Moehs et al. (2001) demonstrated that a higher level of PSY and 1-deoxy-d-xylulose-5-phosphate synthase might be responsible for the color development from pale-yellow to orange. It has also been demonstrated that PSY is a rate-limiting enzyme of carotenoid biosynthesis in canola (Brassica napus) seeds (Shewmaker et al., 1999) and tomato fruits (Fraser et al., 1994).
The successful isolation of genes for carotenoid biosynthesis will allow identification of the key regulatory steps of carotenoid biosynthesis. Nevertheless, knowledge on the molecular aspects that regulate the pathway is still limited. Recently, the genes responsible for hp1 and hp2 (mutations conferring a high level of carotenoids) have been shown to encode the proteins UV-DAMAGED DNA-BINDING PROTEIN 1 (DDB1) and DEETIOLATED 1 (DET1), components that are involved in the light-signal transduction pathway (Liu et al., 2004). In addition, other light-signaling components, such as HY5 and COP1, have been shown to antagonistically regulate the carotenoid level in tomato fruits (Davuluri et al., 2005; Liu et al., 2004). In petals of the Mimulus species, a single QTL at the YUP locus controls the presence and absence of carotenoids (Bradshaw and Schemske, 2003). It is interesting to note that the observed changes of pollinator preference associated with YUP alleles in Mimulus are comparable to those associated with AN2 alleles in petunia (Hoballah et al., 2007) as described above, although the identity of the YUP locus remains to be elucidated.
The amount of carotenoids in the tissues is not attributed solely to the ability to synthesize carotenoids. Some plant tissues have the capacity to synthesize carotenoids but contain only a trace amount of carotenoids. The mechanism that controls carotenoid accumulation is largely unknown. Recently, two different regulatory mechanisms were postulated. One is focused on carotenoid degradation, and the other is focused on sink capacity. In the case of chrysanthemum petals, there was no significant difference in the expression levels of the carotenogenic genes between the white and yellow petals of chrysanthemums (Kishimoto and Ohmiya, 2006). However, a gene encoding carotenoid cleavage dioxygenase (CmCCD4a) was specifically expressed in white petals (Ohmiya et al., 2006). Suppression of CmCCD4a expression resulted in a change in the petal color from white to yellow, indicating that normally white petals synthesize carotenoids but immediately degrade them into colorless compounds. The importance of sink capacity for carotenoid accumulation was first demonstrated in cauliflower (Brassica oleracea var. botrytis) Orange (Or) mutant. Or is a gain-of-function mutation, and single-locus Or mutation confers a high level of β-carotene accumulation in tissues where accumulation of carotenoid is normally repressed (Li et al., 2001). The Or gene encodes a plastid-associated protein with a cysteine-rich domain similar to that found in DnaJ-like molecular chaperones (Lu et al., 2006). This protein plays an important role in triggering differentiation of proplastids and/or other non-colored plastids into chromoplasts, which in turn act as a metabolic sink for carotenoids. Transformation of the Or gene into wild-type cauliflower (or) converts the white colour curd tissue into an orange color with increased levels of β-carotene.
Remaining questions and future challenges
The flavonoid/anthocyanin biosynthetic pathway is well understood, and the relevant main enzymes and genes have been characterized. However, it is not clear how the simple precursors found in the cytosol of the plant cell are converted into the final complicated pigment compounds in the vacuole. The questions of whether there is protein–protein interaction to form metabolic channels and how plant cells transport complicated flavonoids into vacuoles need to be answered in order to achieve efficient engineering of the pathway. Furthermore, only limited knowledge is available regarding regulation of the vacuolar pH, and even less is known about the transport of metal ions from roots to petal vacuoles.
Betalain biosynthesis has been partly characterized in a limited number of species, and a key enzyme, tyrosine hydroxylase, has not yet been isolated. The vacuolar transport of betalains and the transcriptional regulation of betalain biosynthesis are unknown, and engineering of its pathway is yet to be achieved. The reason why betalains and anthocyanins do not co-exist is an interesting unresolved question.
There are many carotenoids whose biosynthesis has not been characterized. Understanding their biosynthesis and their transcriptional regulation should accelerate engineering of the pathway, which has only been partly achieved. Carotenoid engineering is expected to contribute to human health, as carotenoids are important nutrients as well as pigments. A good example is ‘Golden Rice’, which expresses the bacterial desaturase gene and the daffodil or maize PSY gene (Ye et al., 2000 and Paine et al., 2005, respectively).
The plant pigments described here are highly diverse with regard to structures and colors, reflecting the diversity of plants and their metabolic pathways. Obviously, further analysis of more plant species will add novel structures and possibly novel pathways. Characterization of relevant genes should provide a better knowledge of plant evolution. The study of plant pigments will also contribute to the floricultural, food and chemical industries. For example, extensive biochemical characterization, including the determination of crystal structures and structure–function relationships, will be useful to generate novel enzymes to synthesize versatile compounds that cannot be synthesized by non-enzymatic reactions. Generating ornamental plants with novel colors by engineering plant pigment pathways has been successful and will continue to benefit from plant pigment research.
We apologize to the authors whose work has not been cited because of the limitations on the length of this manuscript. We thank the editor and anonymous reviewers for their valuable suggestions to improve this article. The authors acknowledge Dr Tsukasa Iwashina (Tsukuba Botanical Garden, National Science Museum, Japan) for providing the photographs of Meconopsis horridula and Strongylodon macrobotrys, and Mr Masahiro Uchida (http://www9.plala.or.jp/mosimosi) for the photograph of Adonis aestivalis for Figure 1. We sincerely thank Dr Kichiji Kasahara (Yokohama, Japan) for providing the Mirabilis jalapa plant materials, and for permission to use his original photographs in Figure 1.