Biochemistry and molecular biology of the late-stage of biosynthesis of anthocyanin: lessons from Perilla frutescens as a model plant


  • Kazuki Saito,

    Corresponding author
    1. Department of Molecular Biology and Biotechnology, Graduate School of Pharmaceutical Sciences, Chiba University, Yayoi-cho 1–33, Inage-ku, Chiba 263–8522, Japan
    Search for more papers by this author
  • Mami Yamazaki

    1. Department of Molecular Biology and Biotechnology, Graduate School of Pharmaceutical Sciences, Chiba University, Yayoi-cho 1–33, Inage-ku, Chiba 263–8522, Japan
    Search for more papers by this author

Author for correspondence: Kazuki Saito Tel: +81 43 290 2904 Fax: +81 43 290 2905 Email:


Although substantial progress has been made on the molecular genetics of anthocyanin biosynthesis, the biochemistry of some components, such as anthocyanidin synthase, are not fully understood. To explore anthocyanin formation in more detail, and in particular, the late-stage of the biosynthetic pathway, Perilla frutescens (Labiatae) was chosen as a model plant. Two chemo-varietal forms exist in P. frutescens, the pigmented red form and, in striking contrast, the non-pigmented green form, which contains only a trace amount of anthocyanin in the leaves and stems. Using this plant, we investigated the biochemical characteristics of anthocyanidin synthase and two anthocyanin glycosyltransferases, and in addtion we used this plant to investigate the expression and regulation of flavonoid biosynthesis genes. P. frutescens represents a good model plant for investigating anthocyanin biosynthesis. Further exploitation of this model system will require the establishment of a suitable transformation system for P. frutescens. Future work will be directed towards further characterization of the chemo-varietal forms and investigating their evolution from the ancestral form.


  • I.Introduction9
  • II.Biosynthetic enzymes and their genes11
  • III.Regulation of gene expression and regulatory genes19
  • IV.Conclusions and future prospects21
  • References21

I. Introduction

Of all plant secondary compounds, anthocyanins have been investigated most extensively in the areas of chemistry, biochemistry and genetics. Anthocyanins are the main pigments in flowers and fruits and they serve as visual signals that attract insects and animals for pollination and seed dispersal. They also play a role in photoprotection in autumn foliage and rapidly developing shoots of tropical trees. The antioxidant activity of anthocyanins gives cause for a variety of medicinal usage: prevention of cancer, anti-inflammatory activity and antiarteoscelosis activity, etc. (Fauconneau et al., 1997; Nijveldt et al., 2001).

The enzymes and genes involved in anthocyanin biosynthesis are most commonly investigated in petunia, snapdragon and maize as model plant species, resulting in the accumulation of knowledge regarding elucidation of anthocyanin biosynthetic pathway (Heller & Forkmann, 1988; Forkmann, 1993; Holton & Cornish, 1995). Most genes for the biosynthetic enzymes have been isolated, and the biochemical reactions catalyzed by these enzymes from those model plants have been characterized. In addition, the regulatory proteins and their genes were also isolated through analysis of genetic mutants, which exhibit altered flower colour. In most cases, these regulatory genes encode transcriptional factors to control the expression of the genes for biosynthetic enzymes (Mol et al., 1996; Mol et al., 1998; Winkel-Shirley, 2001).

‘Aka-jiso’ (red perilla) is a form of Perilla frutescens Britton var. crispa Thuab (perilla, wild coleus, beef-steak plant) belonging to the Labiatae family. This plant exhibits a dark red or purple color in the leaf and stem during all developmental stages (Fig. 1), and it is widely used as a red food coloring and as a traditional medicine in Japan, China and other Asian countries (Duke, 1985; Iwatsuki et al., 1993). The main anthocyanin pigment of red perilla is a cyanindin-type compound, malonylshisonin (cyanidin 3-O-(6′′-O-(E)-p-coumaryl)-β-D-glucopyranoside-5-O-(6′′′-O-malonyl)-β-D-glucopyranoside) (Fig. 2). In P. frutescens var. crispa, there is another form, ‘Ao-jiso’ (green perilla), which contains only a trace amount of anthocyanin in its leaves and stems, in striking contrast to red perilla (Fig. 1). In spite of the economical and medicinal importance of the anthocyanin pigment and its intriguing form-specific accumulation in P. frutescens, no investigation on the molecular biology of anthocyanin biosynthesis was carried out for this plant until the late 1990s. Only one report on classical genetics regarding a locus that could be responsible for anthocyanin accumulation was published (Koezuka et al., 1985). P. frutescens may provide a new model system for the biochemistry and molecular regulation of anthocyanin biosynthesis as well as the well-known model plants. In fact, by using P. frutescens, the mechanism of anthocyanidin synthase (ANS) has been clarified, and the gene encoding anthocyanin 5-O-glucosyltransferase has been isolated, respectively, for the first time as described below. In this review we examine, from a biochemistry and molecular biology viewpoint, late-stage steps of anthocyanin biosynthesis and its regulation that has been explored by recent studies using P. frutescens as a model plant.

Figure 1.

The phenotypes of (a) anthocyanin-pigmented red and (b) nonanthocyanin accumulated green chemo-varietal forms of Perilla frutescens. The accumulation of anthocyanin is genetically controlled.

Figure 2.

Chemical structure of the main anthocyanin in the red form of Perilla frutescens. Malonylshisonin (cyanidin 3-O-(6′′-O-(E)-p-coumaryl)-β-D-glucopyranoside-5-O-(6′′′-O-malonyl)- β-D-glucopyranoside).

II. Biosynthetic enzymes and their genes

1. General biosynthetic pathway of flavonoids

Figure 3 illustrates the biosynthetic pathway of flavonoids (anthocyanin and flavones) in P. frutescens. As is usual for flavonoid formation, two precursors, p-coumaroyl-CoA and malonyl-CoA, are derived from phenylalanine and acety-CoA, respectively. The 10 enzymes are presumed to catalyze successively the biosynthetic reactions for anthocyanins. These involve chalcone synthase (CHS), chalcone isomerase (CHI), flavanone 3-hydroxylase (F3H), flavonoid 3′-hydroxylase (F3′H), dihydroflavonol reductase (DFR), anthocyanidin synthase (or leucoanthocyanidin dioxygenase) (ANS), flavonoid 3-O-glucosyltransferase (3-GT), anthocyanin acyltransferase (ACT), anthocyanin 5-O-glucosyltransferase (5-GT) and anthocyanin malonyltransferase (MAT). Out of these 10 enzymes, 9 enzymes and their cDNAs (except CHI) have been isolated and characterized from P. frutescens in the last several years (see sections below). Anthocyanins are the main flavonoid products, however, flavones are also formed through a branched pathway involving flavone synthase II (FSII) and F3′H (see below).

Figure 3.

The biosynthetic pathway of flavonoids in Perilla frutescens. CHS, chalcone synthase; CHI, chalcone isomerase; F3H, flavanone 3-hydroxylase; F3′H, flavonoid 3′-hydroxylase; DFR, dihydroflavonol reductase; ANS, anthocyanidin synthase (leucoanthocyanidin dioxgenase); 3-GT, flavonoid 3-glucosyltransferase; ACT, anthocyanin acyltransferase; 5-GT, anthocyanin 5-glucosyltransferase; MAT, anthocyanin malonyltransferase; flavone synthase II (FSII).

2. Anthocyanidin synthase (ANS)

Reaction mechanism of ANS In general, the first colored compound in the biosynthetic pathway of anthocyanin is anthocyanidin (pelargonidin, cyanidin and delphinidin), which is derived from colorless leucoanthocyanidin. In P. frutescens, leucocyanidin is converted into cyanidin (Fig. 3). From leucoanthocyanidin (flavan-3,4-cis-diol) to anthocyanidin (flavylium ion), dehydrogenation from C-2 and dehydration from C-3,4 takes place formally (Fig. 4). However, actual reactions involved in this critical step for coloring of anthocyanin have not been clarified in vitro, because attempts to demonstrate the cell-free activity of conversion from leucoanthocyanidin to anthocyanidin or its 3-glucoside have been unsuccessful (Heller & Forkmann, 1993).

Figure 4.

Reaction mechanism from leucoanthocyanidin to anthocyanidin 3-glucoside catalyzed by (a) anthocyanidin synthase (ANS) and (b) 3-GT, and transport into vacuoles (adopted from Nakajima et al., 2001).

It has been suggested that ANS catalyzes the reaction(s) from leucoanthocyanidin to anthocyanidin. The cDNA encoding putative ANS was isolated from maize mutant (a2) by transposon tagging; and the involvement of the A2 gene in the coloring of anthocyanin was confirmed by the complementation of the a2 mutant by transient expression of the intact A2 gene (Menssen et al., 1990). Although the deduced amino acid sequence of A2 cDNA exhibited significant similarities with those of a family of 2-oxoglutarate-dependent oxygenases, no biochemical investigation was carried out until the first study using the recombinant enzyme from P. frutescens (Saito et al., 1999). This has been ‘a missing link’ where molecular biology failed to meet biochemistry.

A cDNA encoding ANS was isolated from red and green forms of P. frutescens by differential display of mRNA (Yamazaki et al., 1997). As well as recombinant perilla ANS (Saito et al., 1999), ANS proteins of petunia, maize, snapdragon and torenia were produced as fusion proteins with maltose-binding-protein (Nakajima et al., 2001). By using these recombinant proteins, the formation of anthocyanidins from leucoanthocyanidins was detected after the incubation in the presence of Fe2+, 2-oxoglutarate, molecular oxygen and ascorbate, being followed by acidification (pH 1–5). As an important feature of the reaction catalyzed by ANS, no additional enzyme(s) such as dehydratase was required in spite of the involvement of a formal dehydration step.

These results suggest that the reaction mechanism from leucoanthocyanidin to anthocyanidin was catalyzed by ANS as illustrated in Fig. 4(a). The reaction catalyzed by 2-oxoglutarate-dependent oxygenases includes two steps. In the first step, ANS binds with ferrous ion, which acts as a catalytic center of the reaction, and composes a complex with molecular oxygen and 2-oxoglutarate followed by the formation of an oxoferryl enzyme complex, succinate and CO2 (reaction (i) in Fig. 4). This first step is common to all 2-oxoglutarate-dependent oxygenases. In the second step, the oxoferryl species is used for the hydrogen radical abstraction from C-2 and C-3, or for hydroxylation at C-2 or C-3 followed by spontaneous dehydration that yields 2-flaven-3,4-diol and H2O (reaction (ii)). Subsequently, isomerization of the hydroxyl group and double bond of 2-flaven-3,4-diol occurs spontaneously to yield the thermodynamically more stable 3-flaven-2,3-diol (pseudobase) at cytosolic pH (reaction (iii)). The enzyme 3-GT catalyzes 3-O-glucosylation of the pseudobase form under cytosolic neutral conditions (reaction (iv)) (see below).

In the flavonoid biosynthetic pathway, there are four reactions which are catalyzed by 2-oxoglutarate-dependent oxygenases, ANS, F3H (Britsch, 1990b; Britsch et al., 1992), flavonol synthase (Holton et al., 1993) and flavone synthase I (Britsch, 1990a). All these reactions are concerned with oxidation at the C-2 and/or C-3 position of the flavonoid skeleton. Since no stable hydroxylated intermediates were detected in the reactions of ANS, flavonol synthase and flavone synthase I, the mechanism involving direct 2,3-desaturation seems to be the most likely route for the formation of 2-flaven-3,4-diol (Fig. 4a). Regarding the reaction mechanism of flavone synthase I, a radical mechanism was proposed for the direct 2,3-dehydrogenation of flavanone (Britsch, 1990a).

Expression of ANS gene The ANS mRNA was expressed in leaves and stems of the red form of P. frutescens but not in the green form, although the ANS gene is present also in green perilla. The ANS protein accumulation and anthocyanin content were in the same manner to that of mRNA level (Fig. 5). The upper and lower epidermal cells were the sites of accumulation of ANS protein indicated by an immunohistochemical study (Fig. 6a). These tissue-specific accumulations of mRNA and protein of ANS in upper and lower epidermis correspond well to the tissue-specific accumulation pattern of anthocyanin (Fig. 6d), suggesting the important involvement of ANS in anthocyanin formation.

Figure 5.

Expression of the anthocyanidin synthase (ANS) mRNA, accumulation of ANS protein and anthocyanin production in red and green Perilla frutescens (adapted from Saito et al., 1999). (a) Northern blot analysis of ANS mRNA. Total RNA from plant tissues was electrophoresed on an agarose gel, transferred to a nylon filter, and then hybridized with ANS cDNA as the probe. (b) Total soluble protein was separated in SDS-PAGE and transblotted onto a nylon filter. Immunoreactive protein against anti-ANS rabbit serum was visualized. (c) Anthocyanin content in plant tissues was measured in the acidic methanol extracts by absorbance at 517 nm. Data are mean values of triplicate tests ± SD.

Figure 6.

Cell-specific accumulation of anthocyanin and the anthocyanidin synthase (ANS) protein in the red form of Perilla frutescens (Yamazaki et al. unpublished). (a) Stem sections of red perilla were immuno-reacted with anti-ANS antibodies and visualized by secondary antibodies labeled with fluoescein isothiocyanate (FITC). Immunoreactive ANS protein is localized in the epidermis cells. (b) Stem sections of red perilla were immuno-reacted with anti-ANS antibodies preabsorbed to recombinant ANS protein as a negative control. (c) Stem sections of green perilla were immuno-reacted with anti-ANS antibodies. No ANS protein is localized in the sections of green perilla. (d) Leaf sections of red perilla under bright field. Anthocyanins are accumulated in upper and lower epidermis cells.

3. Possible alternative pathway from dihyroflavonols to anthocyanidins

The conversion steps from dihydroflavonol to 2-flaven-3,4-diol in the natural pathway involve reduction of 4-keto group by nicotinamide adenine dinucleotide (NADH) and dehydrogenation of C-2 and C-3 (or equivalently oxidation of C-3–3-keto group) as shown in Fig. 7 (pathway A); however, these steps might be performed simply by consecutive keto-enol isomerization as illustrated in Fig. 7 (pathway B). Pathway B is simpler than pathway A, involving reduction and re-oxidation, and does not consume any NADH and 2-oxoglutarate. However, no evidence has been provided to support the presence of this second pathway. Moreover, DFR and ANS are vital for the formation of anthocyanin, since the mutants lacking either of these two enzymes do not produce anthocyanin. Why doesn’t nature use this alternative simple pathway? For pathway B, involving consecutive keto-enol isomerization of dihydroflavonol, 3-oxo-flavan-4-ol must be formed in equilibrium with dihydroflavonol and 3-flaven-3,4-diol. However, 3-oxo-flavan-4-ol is thermodynamically very unfavorable compared with dihydroflavonol, because the conjugation of keto group and aromatic A-ring present in dihydroflavonol that is contributed to the thermodynamic stability is lost in 3-oxo-flavan-4-ol. The results of energy calculation using PM3 (Spartan) of the compounds also support this idea: the conversion route from dihydroflavonol to 3-oxo-flavan-4-ol is very unlikely to be a result of the calculated thermodynamic energies of the compounds (Fig. 7). This means that two consecutive keto-enol isomerization reactions (pathway B) would be thermodynamically impossible. Thus, to cope with the thermodynamic difficulty of the reactions from dihydroflavonol to 2-flaven-3,4-diol we see consumption of NADH, oxygen and 2-oxoglutarate through reduction by DFR and re-oxidation of ANS (pathway A). This is a smart way to manage the thermodynamically unfavorable reactions by consumption of NADH, oxygen and 2-oxoglutarate.

Figure 7.

Two possible routes from dihydroflavonol to anthocyanidin. (a) Route involving the reduction by dihydroflavonol reductase (DFR) and re-oxidation by anthocyanidin synthase (leucoanthocyanidin dioxgenase) (ANS) found in nature. (b) Presumptive route involving two consecutive keto-enol isomerization. Thermodynamic energies of dihydrokaempferol and the derived compounds shown in parentheses are calculated by the program PM3 (Spartan).

4. Flavonoid 3-O-glucosyltransferase (3-GT)

3-GT catalyzes the second half of the reaction in Fig. 4, namely the formation of anthocyanidin 3-glucoside from anthocyanidin. However, the detail of this reaction has also been not completely clarified in vitro. From inter-tissue complementation assays in maize (Reddy & Coe, 1962) it has been generally assumed that leucoanthocyanidin first is converted to anthocyanidin by ANS, and then glucosylated by 3-GT (Heller & Forkmann, 1988). However, no in vitro biochemical experimental evidence has been provided to confirm the reaction sequence from leucoanthocyanidin to anthocyanidin 3-glucoside, although 3-GT activity has been detected in crude enzyme preparations of various plant species and in preparations containing recombinant 3-GT including P. frutescens (Tanaka et al., 1996; Gong et al., 1997; Ford et al., 1998).

The in vitro reconstitution experiments were conducted using recombinant ANS and 3-GT (Nakajima et al., 2001). The results indicate that the reaction sequence from leucoanthocyanidin to anthocyanidin 3-glucoside involves anthocyanidin formation from leucoanthocyanidin followed by 3-glucosylation of anthocyanidin, as illustrated in Fig. 4(b), and 3-GT catalyzes 3-O-glucosylation of the pseudobase form under cytosolic neutral conditions (reaction (iv)). The pseudobase form of anthocyanidin 3-glucoside is then transported into vacuoles; and under the moderate acidic condition of vacuoles (c. pH 5), the colored flavylium ion of anthocyanidin 3-glucoside is formed by removal of the hydroxyl anion from the C-2 position of the pseudobase form (reaction (v)). For the net formation of colored anthocyanidin 3-glucoside (flavylium ion form) from leucoanthocyanidin, only two enzymes, ANS and 3-GT, are therefore required, and the change of pH following transport of anthocyanidin 3-glucoside into vacuoles is sufficient and essential for the coloring of anthocyanidin 3-glucoside (or more modified anthocyanins).

It has been suggested that ANS and 3-GT might work as a multienzyme complex (Heller & Forkmann, 1993). If an ANS-3-GT complex was formed to catalyze the sequential reactions, the intermediate of the reactions, anthocyanidin, can then be rapidly channeled between these two enzymes. However, no evidence has been provided for the channeling of the intermediate and apparent interaction between ANS and 3-GT from P. frutescens. Since a protein–protein interaction between several anthocyanin biosynthetic enzymes of Arabidopsis thaliana has recently been reported (Burbulis & Shirley, 1999), a more detailed investigation will be required.

5. Anthocyanin 5-O-glucosyltransferase (5-GT)

Anthocyanins from a variety of plants, for example, P. frutescens (Yoshida et al., 1997), Petunia hybrida (Jonsson et al., 1984), Verbena hybrida (Toki et al., 1995) and Arabidopsis thaliana (Bloor & Abrahams, 2002), are 5-glucosylated. Glycosylation at the 5-position is believed to allow for more stable complexes in copigmentation of anthocyanins, resulting in flowers with bright reddish-purple color from dull violet or pure red when 5-O-glycosylation does not take place (Jonsson et al., 1984; Teusch et al., 1986; Martin & Gerats, 1992). Therefore, 5-O-glycosylation is one of the most important steps modifying flower color, in particular, for creating purple hue. The enzyme catalyzing this step, UDP-glucose:anthocyanin 5-O-glucosyltransferase (5-GT), has been partially characterized in several plant species (Kamsteeg et al., 1978; Kamsteeg et al., 1980; Seyffert, 1982; Jonsson et al., 1984; Teusch et al., 1986), however, no cDNA had been isolated until the 5-GT cDNA was cloned from P. frutescens by cDNA differential display (Yamazaki et al., 1999).

The expression of structural genes encoding anthocyanin biosynthetic enzymes was co-ordinately regulated in an anthocyanin forma-specific manner in perilla: the expression was observed only in the red form and not in the green (see sections below) (Gong et al., 1997). The cDNA differential display technique was applied to isolation of new genes, which were expressed in an anthocyanin form-specific manner. Through the differential display strategy, 5-GT cDNA was cloned from perilla (Yamazaki et al., 1999).

The recombinant proteins of perilla 5-GT expressed in the yeast catalyzed the conversion of anthocyanidin 3-O-glucosides into corresponding anthocyanidin 3,5-O-diglucosides using UDP-glucose as a cofactor. Several biochemical properties shown in Table 1 were similar to those reported previously for partially purified native 5-GTs from petunia (Jonsson et al., 1984) and Matthiola incana (Teusch et al., 1986).

Table 1.  Biochemical properties of the recombinant 5-GT from Perilla frutescens
Optimum pH8.0–8.5
Km for cyanidin 3-O-glucoside31.4 µM
Km for UDP-glucose940 µM
InhibitorsSensitive to p-chloromercuribenzoate, Mn2+ and Co2+

The 5-GT cDNAs were also cloned from petals of Verbena hybrida (Yamazaki et al., 1999) and Petunia hybrida (Yamazaki et al., 2002) using the perilla cDNA as the probe. As summarized in Table 2, recombinant enzymes of P. frutescens and V. hybrida exhibited 5-GT activities transferring glucose moiety from UDP-glucose to the 5 position of various anthocyanins, in which the 3 position is glycosylated or (acyl)glycosylated. Although cyanidin 3-O-glucoside was the best substrate for both 5-GTs, they exhibited a broad substrate specificity. However, no activities were detected towards cyanidin and cyanidin 3,5-O-diglucoside. Acyltransferase from perilla catalyzing acylation of glucose moiety of anthocyanidin 3-O-glucoside utilizes both anthocyanidin 3-O-glucoside and anthocyanidin 3,5-O-diglucoside as substrates (Yonekura-Sakakibara et al., 2000). These results suggest that the biosynthetic pathway from anthocyanidin 3-O-glucoside to anthocyanidin 3-O-(acyl)glucoside-5-O-glucoside forms a ‘metabolic grid’ comprising two bypass routes in perilla as illustrated in Fig. 8.

Table 2.  The substrate specificities of recombinant 5-GT from Perilla frutescens, Verbena hybrida and Petunina hybrida
Substrate5-GT activity (pkat/mg protein) (relative activity percentage)
P. frutescens (%)V. hybrid (%)P. hybrida (%)
  1. The assay was carried out with the recombinant 5-GT produced in yeast. The data are from Yamazaki et al. (1999) for P. frutescens and V. hybrida and from Yamazaki et al. (2002) for P. hybrida. The strict substrate specificity of the recombinant P. hybrida to delphinidin 3-O- (p-coumaroyl)rutinoside is in good agreement with that reported for crude protein extract from P. hybrida by Jonsson et al. (1984). n.d.; not determined.

Pelargonidin-3-O-glucoside28.3(20.9)11.7(24.1)< 0.1(0)
Cyanidin-3-O-glucoside135.9(100)48.3(100)< 0.1(0)
Delphinidin-3-O-glucoside92.5(68.1)17.5(36.2)< 0.1(0)
Cyanidin-3-O-(p-coumaroyl)glucoside26.0(19.1)6.0(12.4)< 0.1(0)
Cyanidin-3-O-(cafferoyl)glucoside13.3(9.8)8.7(17.9)< 0.1(0)
Cyanidin-3,5-diglucosiden.d. n.d. < 0.1(0)
Delphinidin-3-O-rutinosiden.d. n.d. < 0.1(0)
Delphinidin-3-O-(p-coumaroyl)rutinosiden.d. n.d. 25.8(100)
Cyanidin< 0.1(0)< 0.1(0)< 0.1(0)
Figure 8.

The grid-type pathway for 5-GT and anthocyanin acyltransferase (ACT) in anthocyanin biosynthesis in Perilla frutescens.

By contrast, the recombinant 5-GT from petunia exhibited a strict substrate specificity towards the proper substrate, comparing with other 5-GTs from perilla and verbena exhibiting a broad specificity. The recombinant petunia 5-GT produced in yeast exhibited a strict substrate specificity towards delphinidin 3-O-(p-coumaroyl)-rutinoside, being consistent with the reported results for crude enzyme from the petunia plant (Jonsson et al., 1984) (Table 2). Other compounds examined did not serve as substrates for petunia 5-GT. By contrast, the recombinant 5-GT proteins from P. frutescens and V. hybrida could catalyze transfer the glucose moiety to the 5-O-position of various anthocyanins, in which the 3-O-position is glycosylated or acyl-glycosylated (Yamazaki et al., 1999). Such strict substrate specificity of petunia 5-GT towards only the acylated rutinoside of anthocyanin compared with those of perilla and verbena 5-GTs seems to be suitable to synthesize the particular petunia anthocyanins efficiently. A number of amino acid residues, which are different between petunia 5-GT and perilla 5-GT/verbena 5-GT, could be the candidates to be changed for genetic engineering of altering substrate specificity of 5-GT.

6. Acyltransferase

Acylation of anthocyanin is supposed to serve a number of functions: to stabilize the anthocyanin structure resulting in enhancing blue color; to increase water solubility; to protect against degradation by glucosidase; and to uptake anthocyanins into vacuoles. This is also the case for perilla with the importance of acyltransferase for anthocyanin pigmentation.

Since the major anthocyanin fund in perilla is malonylshisonin, cyanidin 3-O-(6′′-O-(E)-p-coumaryl)-β-D-glucopyranoside-5-O-(6′′′-O-malonyl)-β-D-glucopyranoside, two acyltransfer reactions, p-coumaroyl transfer and malonyl transfer, should be involved. Anthocyanin 3-aromatic transferase was purified from P. frutescens (Fujiwara et al., 1998), and subsequently its cDNA clone was isolated (Yonekura-Sakakibara et al., 2000). This enzyme is a soluble monomeric 50 kDa protein belonging to a family of CoA-dependent acyltransferases containing a common sequence motif (St-Pierre et al., 1998). It catalyzes the transacylation of p-coumaroyl or cafferoyl moieties from the corresponding CoA-derivatives to 6′′-O-position of 3-O-glucoside of anthocyanidin 3-O-glucosides or anthocyanidin 3,5-O-diglucosides. Because Km for cyanidin 3-O-glucoside is 17 times lower than that for cyanidin 3,5-O-diglucoside (Fujiwara et al., 1998), cyanidin 3-O-glucoside is the preferable substrate. However, the relative 5-GT activities towards cyanidin 3-O-glucoside and cyanidin 3-O-(p-coumaroyl)-glucoside were 100 : 19, preferable to cyanidin 3-O-glucoside again. These suggest that 5-GT and ACT presumably compete for cyanidin 3-O-glucoside as a better substrate for both reactions in vivo.

Anthocyanin malonyltransferase was purified and cloned for the first time from Salvia splendens (salvia) (Suzuki et al., 2001), and then from perilla by using a salvia clone as the probe (Nakayama et al. in preparation). This enzyme catalyzes the transfer of malonyl moiety from malonyl-CoA to 6′′′-O-position of 5-O-glucose of shisonin, cyanidin 3-O-(6′′-O-p-coumaroyl)-β-D-glucoside-5-O-β-D-glucoside, and also belongs to a CoA-dependent acyltransferase family.

7. P450 monooxygenase

Anthocyanins and flavones, two subgroups of flavonoids, are biosynthesized from a common intermediate, flavanone (naringenin) (Fig. 9). The color of anthocyanin is influenced by the presence of copigments such as flavones and flavonols. In the biosynthetic pathway of cyanidin-type anthocyanins and flavones in P. frutescens, two cytochrome P450s, flavonoid 3′-hydroxylase (F3′H) and flavone synthase II (FSII) are involved. The hydroxylation of the B-ring of flavonoids is an important reaction for determining the flower color. F3′H and flavonoid 3′,5′-hydroxylase (F3′5′H), which belong to the CYP75 family, are the essential enzymes involved in hydroxylation of the flavonoid B-ring (Heller & Forkmann, 1993). FSII, which belongs to CYP93B subfamily, catalyses the reaction that introduces the double bond between C-2 and C-3 of the flavanones to produce flavones (Heller & Forkmann, 1993). Since P. frutescens accumulates cyaniding-type anthocyanins and flavones, it would be interesting to clarify the mode of regulation of two P450s, F3′H and FSII, involved in the formation of respective products.

Figure 9.

Two cytochrome P450s involved in biosynthesis of anthocyanin and flavone in Perilla frutescens. FSII, flavone synthase II CYP93B6; F3′H, flavonoid 3′-hydroxylase CYP75B4.

The F3′H-encoding perilla cDNA was isolated from a cDNA library of red P. frutescens using a F3′H-encoding cDNA from petunia as a heterologous probe. The FSII-encoding cDNA was isolated also from red P. frutescens cDNA library using cDNAs from Antirrhinum majus and Glycyrrhiza echinata as a mixed heterologous probe (Kitada et al., 2001). The F3′H cDNA encoded 57.5 kDa protein designated CYP75B4, and FSII cDNA did 57.1 kDa protein designated CYP93B6. Recombinant F3′H CYP75B4 catalyzed 3′-hydroxylation of flavanones to the corresponding compounds with Km values of 18–20 µM, and recombinant FSII CYP93B6 converted flavanones to flavones with Km values of 8.8–11.9 µM. The F3′H CYP75B4 transcript was predominantly expressed in the red form of P. frutescens, and its expression was induced by light in conjunction with other transcripts of biosynthetic enzymes of anthocyanin. However, the FSII CYP93B6 transcript accumulated to an equal level in leaves of both red and green forms of P. frutescens, in agreement with the accumulation pattern of flavones in the leaves. These results indicate that genes of a set of anthocyanin biosynthetic enzymes including F3′H are expressed co-ordinately only in the red form of P. frutescens and not in the green form, whilst FSII gene expression is controlled in a similar manner in red and green forms of P. frutescens. Since no delphinidin-type anthocyanins are detected in perilla, the presence of F3′5′H is unlikely.

8. Transport and storage

Uptake transport and storage of anthocyanin into vacuoles are important issues for anthocyanin metabolism. Several possible mechanisms for vacuolar uptake of anthocyanins have been proposed (Martinoia et al., 2000): H ± potential dependent uptake system that includes acylated anthocyanin-specific transport in carrot (Hopp & Seitz, 1987) and presumably multidrug and toxic compound extrusion (MATE) protein-mediated system (Debeaujon et al., 2001); MgATP-energized transport by ATP-binding cassette (ABC) protein that contains multidrug resistance-associated(MRP)- protein (Lu et al., 1998) acting for glutathione S-conjugates (Marrs et al., 1995; Mueller et al., 2000) and glucuronide conjugates (Klein et al., 2000). Nevertheless, little is known about the mechanism and regulation of uptake into the storage organella in perilla. By differential display technique of red and green chemo-varieties of perilla, we have isolated a cDNA designated 8R6 clone that encodes anthocyanin-related membrane (ANM) protein and might be involved in uptake of anthocyanins into vacuoles (Yamazaki et al., unpublished). The mRNA expression pattern of 8R6 was identical to that of other structural genes for anthocyanin biosynthesis (see the section below), suggesting the possible involvement of 8R6 in anthocyanin formation and/or storage. The investigation of transgenic plants expressing the fusion proteins of 8R6 with green fluorescent protein (GFP) indicated the localization of the fusion proteins in tonoplasts of tobacco and Arabidopsis, also implying the possible function of 8R6 in vacuole membranes. The transgenic Arabidopsis over-expressing 8R6 gene accumulated a higher amount of anthocyanin under high-sucrose stressed condition than nontransformed control. Moreover, the Arabidopsis protoplasts producing 8R6 protein could uptake cyanidin 3-O-glucoside in vitro more efficiently than the control protoplasts. Recently we have identified the possible orthologues of 8R6 in Arabidopsis. The expression patterns of these Arabidopsis genes, Anm, also suggest their function being correlated to anthocyanin accumulation. These results suggest that ANM protein is presumably involved in the uptake of anthocyanin into vacuoles in red perilla. Detailed analysis is currently being undertaken.

III. Regulation of gene expression and regulatory genes

1. Gene expression by a form-specific manner

As well as the genes described previously, cDNA clones encoding the enzymes, CHS, F3H and DFR were isolated from cDNA libraries in the leaves of a red form of P. frutescens by screening with partial fragments amplified by means of polymerase chain reaction (PCR) and heterologous cDNAs as probes (Gong et al., 1997). The deduced amino acid sequences of these four genes exhibited 40–90% identity with those reported for the corresponding gene from other unrelated species.

Southern blot analyses for these structural genes indicated that each gene comprises a small multigene family. The expression of eight genes (F3H, F3′H, DFR, ANS, 3-GT, ACT, 5-GT and 8R6), with exception of the CHS gene, was detected only in red leaves of the red form of P. frutescens; not in green leaves of the green form. The CHS gene was expressed in both red and green leaves, but 10-fold more in red leaves than in green leaves. These results show that the expression of all structural genes is co-ordinately regulated in a form-specific manner. Under weak-light conditions, the accumulation of both anthocyanin and mRNAs of biosynthetic enzymes were lower than those under the strong-light conditions in leaves of the red form. High-intensity white light co-ordinately induced the accumulation of transcripts of all genes examined in the mature leaves of red P. frutescens. Taking these results together, one can presume the presence of regulatory gene(s) that control the expression of all genes directly committed in anthocyanin formation, as extensively described in maize, snapdragon and petunia (Holton & Cornish, 1995; Mol et al., 1996). Several regulatory genes, which are presumably involved in anthocyanin biosynthesis in perilla, have been isolated as summarized in Table 3.

Table 3.  Regulatory genes obtained from Perilla frutescens
Gene/proteinOrthologue proteinExpressionEnhanced anthocyanin in transgenic plantsOne/two- hybrid assayReference
  1. ++, very strong expression; +, strong expression; ±, weak expression; −, no expression.

Myc-rp/gpDelila+ ++ +Yes (tobacco, tomato)Activation of DFR promoterGong et al. (1999a)
MYC-RP/GP(Snapdragon) Jaf 13 (Petunia)   Interaction with PFWD and MYB-P1 
Myc-F3G1AN1 (petunia)+ +Under investigationUnder investigationUnpublished
MYC-F3G1TT8 (Arabidopsis)     
Myb-P1Myb305+ +±No (tobacco)Binding to DFR promoterGong et al. (1999b)
MYB-P1(Snapdragon)   Interaction with MYC-RP 
PfwdAN11 (Petunia)++Yes (Arabidopsis)Interaction with MYC-RPSompornpailin et al. (2002)
PFWDTTG1 (Arabidopsis)     

2. MYC transcriptional factors

A gene (Myc-rp) encoding a transcriptional factor MYC-like protein containing basic helix-loop-helix (bHLH) motif was isolated from perilla red form, and its deduced amino acid sequence shows 64% identity with that of Delila from snapdragon (Gong et al., 1999a). However, the Myc-rp gene was expressed in leaves and roots of both red and green P. frutescens equally. Comparison of deduced amino acid sequence of MYC-RP with that of MYC-GP, the second allele isolated from a green form of P. frutescens, indicates that the 132nd amino acid, alanine, existing in MYC-RP was changed to serine in MYC-GP. The heterologous expression of these two alleles of Myc-like gene in tobacco and tomato resulted in an increase of the anthocyanin contents in flowers of tobacco and vegetative tissues and flowers of tomato. These data indicate that this Myc-like gene presumably functions in enhancement of anthocyanin biosynthesis similarly in different tissues of dicot plants.

In yeast, the MYC-RP/GP and Delila protein exhibited transactivation activity on a yeast promoter and the promoter of gene encoding DFR from P. frutescens. A transactivation domain of MYC-RP/GP and Delila could be located in the region between 193rd-420th amino acids of MYC-RP/GP proteins. Moreover the study of extensive scanning mutations of Delila suggested the critical role of alanine-161 for transactivation activity of Delila (Gong et al., 2000).

By cDNA differential display technique, the second Myc-like gene (Myc-F3G1) was recently isolated from perilla (Saito et al. unpublished). The deduced amino acid sequence of this cDNA is closely similar to those of AN1 of petunia (Spelt et al., 2000) and TT8 of Arabidopsis (Nesi et al., 2000), indicating the direct involvement in the regulation of anthocyanin biosynthesis. Because of chemo-variety specific expression and close sequence similarity to AN1 and TT8, this gene would be a good candidate for directly determining the chemo-varietal forms of perilla.

3. MYB transcriptional factors

Screening of a set of differentially expressing Myb-related genes in red and green perilla resulted in isolation of the Myb-p1 gene (Gong et al., 1999b). The expression of Myb-p1 was 10-fold abundant in red perilla than in green perilla, and was remarkably induced by light. Interaction between MYB-P1 and MYC-RP was indicated by yeast two-hybrid system. MYB-P1 was able to bind the perilla DFR gene promoter in yeast. These data suggest that Myb-p1 may also be involved in regulation of anthocyanin biosynthesis and subsequently may be one of the responsible factors for determination of anthocyanin-form in red and green P. frutescens.

4. WD-repeat protein

Pfwd, a perilla cDNA, whose deduced amino acid sequence is homologous to WD40-repeat regulatory proteins, AN11 from petunia (de Vetten et al., 1997) and TTG1 from A. thaliana (Walker et al., 1999), was cloned from the red leaves P. frutescens (Sompornpailin et al., 2002). Nucleotide sequence analysis revealed that Pfwd cDNA contains 1411 base pairs, which encode a predicted protein of 333 amino acids designated PFWD. The deduced amino acid sequence shows 81% identity with that of petunia AN11. The N-terminal part contains one of basic residues similar to nuclear localization signal. The C-terminal part encodes a region containing WD (tryptophan and aspartic acid) repeat that is highly conserved in several proteins even from the species that do not produce anthocyanin such as yeast, nematodes and mammals. RNA gel blot analysis showed that Pfwd is expressed in leaves and roots of both red and green perilla. No difference is detected between deduced amino acid sequences in red and green perilla. Fusion protein between PFWD and a bacterial β-glucuronidase was localized in the cytosol under darkness and light, in epidermal cells of onion bulbs; however, coexpression of PFWD and MYC-RP resulted in nuclear localization of the PFWD protein. The ectopic overexpression of Pfwd in Arabidopsis caused the enhanced anthocyanin formation in hypocotyls of young seedlings, although the plants expressing Pfwd could not survive until setting seeds presumably because of the lethal effects of its ectopic expression. The yeast two-hybrid experiments indicated the interaction of PFWD protein with MYC-RP protein. These results suggested the involvement of Pfwd in the signal transduction of anthocyanin biosynthesis, although it is not a direct determining factor of anthocyanin chemotypes of perilla.

IV. Conclusions and future prospects

Substantial progress has been made on biochemical mechanism and molecular regulation of anthocyanin biosynthesis in perilla as a model plant in the last several years, in particular, by applying a differential strategy to chemo-varietal forms regarding anthocyanin. The biochemical mechanism of ANS was clarified by the recombinant perilla enzyme, and the successful cDNA cloning of 5-GT was also carried out with perilla. However, establishment of the transformation protocol in P. frutescens will be needed for more detailed molecular investigation by reverse genetics approach. In particular, the definitive factor responsible for chemo-varietal forms and how it evolves from the ancestor form is still in question. Addressing these questions would be intriguing for the understanding of the function of anthocyanin in plants.


The authors thank all colleagues who were involved in our study and whose names are in the papers cited here. They also thank to Dr T. Nagamitsu, Kitasato University, for energy calculation of the intermediates. The study by the authors’ group was supported, in part, by Grant-in-Aids from Ministry of Education, Culture, Sports, Technology and Sciences, Japan; CREST of Japan Science and Technology, Japan; a grant from HOANSHA, Osaka, Japan; and San-Ei-General Foundation, Osaka, Japan.