Flavonoid compounds such as anthocyanins and proanthocyanidins (PAs; so-called condensed tannins) have a multitude of functions in plants. They must be transported from the site of synthesis in the cytosol to their final destination, the vacuoles. Three models have been proposed for sequestering anthocyanins in vacuoles, but the transport machinery for PAs is poorly understood. Novel Arabidopsis mutants, transparent testa 19 (tt19), which were induced by ion beam irradiation, showed a great reduction of anthocyanin pigments in the vegetative parts as well as brown pigments in the seed coat. The TT19 gene was isolated by chromosome walking and a candidate gene approach, and was shown to be a member of the Arabidopsis glutathione S-transferase (GST) gene family. Heterologous expression of a putative ortholog, petunia anthocyanin 9 (AN9), in tt19 complemented the anthocyanin accumulation but not the brown pigmentation in the seed coat. This suggests that the TT19 gene is required for vacuolar uptake of anthocyanins into vacuoles, but that it has also a function different from that of AN9. The depositional pattern of PA precursors in the mutant was different from that in the wild type. These results indicate that TT19 participates in the PA pathway as well as the anthocyanin pathway of Arabidopsis. As involvement of GST in the PA pathway was previously considered unlikely, the function of TT19 in the PA pathway is also discussed in the context of the putative transporter for PA precursors.
Flavonoids are secondary metabolites that are unique to the plant kingdom, with a multitude of functions such as acting as pigments, UV-B protectants, agents for signaling between plants and microbes, and regulators of auxin transport (Winkel-Shirley, 2001). Some plant flavonoids are also known to have beneficial effects on human health by their strong antioxidant activities (Bagchi et al., 2000) and to prevent bloat in ruminants (Gruber et al., 1999). Two flavonoids, anthocyanins and proanthocyanidins (PAs; so-called condensed tannins) are exclusively accumulated in the vacuole, although they are synthesized in the cytosol (Mol et al., 1998; Figure 1). Until vacuolar compartmentalization is completed, these compounds are not stabilized, and are unable to function (e.g. as pigments or UV-B protectants) in plant cells (Winkel-Shirley, 2001).
Although much is known about flavonoid synthesis (Mol et al., 1998; Winkel-Shirley, 2001), the steps involved in their sequestration are poorly understood. Three models have been proposed for sequestering anthocyanins in vacuoles: (I) direct transport of non-modified anthocyanins; (II) transport after their modification; and (III) transport mediated by a carrier protein (Mueller and Walbot, 2001). For model I, Klein et al. (1996) showed that barley flavonoid-glucosides were taken up without their further modification into the isolated vacuoles. For model II, acylation was shown to be required for the effective uptake of anthocyanins by isolated vacuoles of carrot suspension cells (Hopp and Seitz, 1987). In addition, maize bronze 2 (BZ2) and petunia AN9 encode glutathione S-transferase (GST) proteins, and their corresponding mutants have anthocyanin-less phenotypes (Alfenito et al., 1998; Marrs et al., 1995). The GSTs were previously thought to form glutathione-conjugates of anthocyanidin 3-glucosides (model II; Marrs et al., 1995), but they are presently believed to bind to anthocyanidin 3-glucosides in order to sequester them into vacuoles. Thus, they are thought to operate as anthocyanin carrier proteins rather than glutathionation agents (model III; Mueller and Walbot, 2001; Mueller et al., 2000). These models are difficult to reconcile, and therefore it is hypothesized that different species utilize different sequestration mechanisms (Mueller and Walbot, 2001).
Even less is known about the vacuolar sequestration mechanisms for PAs or their precursors. The current hypothetical model for the PA accumulation mechanisms largely relies on data from Douglas fir (Stafford, 1989). PAs are polymers that are composed of two monomer precursors, flavan-3-ols and flavan-3,4-diols. Flavan-3-ols are believed to be employed as start units, to which flavan-3,4-diols are sequentially added as extension units in a condensation reaction. However, another pathway involving 2,3-cis-flavan-3-ols as extension units has been recently suggested (Xie et al., 2003). These three types of precursor are probably taken up into the vacuolar membrane as monomers (Debeaujon et al., 2001; Stafford, 1989). After completing the condensation, PAs are oxidized and result in a brown color (Figure 1). The condensation and oxidation steps are probably accomplished enzymatically, although non-enzymatic reactions are also possible (Stafford, 1989). Some mutants, presumably involved in the condensing and/or accumulation steps, were isolated in barley (Gruber et al., 1999) and Arabidopsis (Abrahams et al., 2002). However, identification of the constituent proteins of their transport, polymerization, and oxidation machinery is still lacking.
Recently, one putative constituent of the transport machinery for PA precursors was isolated as Arabidopsis TRANSPARENT TESTA 12 (TT12) protein (Debeaujon et al., 2001). TT12 encodes a putative transporter that belongs to a multidrug and toxic compound extrusion (MATE)-type transporter family, and is at least partly responsible for the vacuolar sequestration of PA precursors but not anthocyanins (Debeaujon et al., 2001). We previously obtained two novel tt mutants induced by ion beam irradiation of Arabidopsis (Shikazono et al., 2003). One is a tt18 mutant (originally referred to as tt19 in Winkel-Shirley, 2001), in which a gene encoding a putative leucoanthocyanidin dioxygenase (LDOX) is impaired. The other is defined as a tt19 mutant, but the nature of tt19 has not been elucidated. Here, we report molecular and physiological analyses of two tt19 alleles and isolation of the TT19 gene, which is involved in vacuolar compartmentalization of flavonoids.
Phenotypic characterization of tt19 mutants
Two M2 lines with very similar phenotypes were isolated by ion beam irradiation of wild-type Columbia (Col). An allelism test showed that these were allelic and were named tt19-1 and tt19-2 (Shikazono et al., 2003). Purple pigmentation because of anthocyanins was not visible in the basal region of the stem in tt19 (Figure 2a). Measurements with a spectrophotometer quantitatively showed a great reduction of anthocyanin pigments in tt19 (data not shown). The seed coat of tt19 was pale brown at the ripening stage, in contrast to brown in Col and yellow in tt4 (Figure 2b, upper panel). However, the brown color of the seed coat of tt19 became deeper with increasing length of the desiccation period, eventually darkening as much as that of Col after long-term desiccation (Figure 2b, lower panel). Other phenotypic traits seemed normal in the tt19 mutant.
Naringenin feeding experiment
Sugars are known to induce anthocyanin pigmentation in Arabidopsis seedlings (Tsukaya et al., 1991). Before the naringenin feeding experiment, the proper concentration of sucrose for the induction of anthocyanin accumulation was determined. In the Col seedlings, as the sucrose concentration increased to 5%, the anthocyanin pigmentation became markedly deeper in the upper hypocotyls and abaxial and marginal regions of the cotyledons. Ten per cent sucrose delayed development, and 20% inhibited germination in Col (data not shown). Two tt19 mutant lines also showed retardation of the germination and seedling development on plates with more than 10% sucrose. Therefore, 5% of sucrose was used in this experiment.
In the Col seedlings, addition of naringenin reinforced anthocyanin pigmentation to some extent (Figure 3a, upper panels). Although some effects were observed in seedling development on the 5% sucrose media, the tt4 mutant exhibited anthocyanin pigmentation as a result of naringenin feeding (Figure 3a, lower panels), as reported previously by Shirley et al. (1995). On the other hand, tt19 did not accumulate anthocyanins despite the naringenin feeding (Figure 3a, middle panels). These results indicate that the TT19 gene functions at a step downstream of the chalcone isomerase (CHI) reaction in the anthocyanin biosynthetic pathway (see Figure 1).
In general, the brown color of the Arabidopsis wild-type testa is thought to be mainly because of the oxidation of PAs (Chapple et al., 1994). To determine the presence and/or distribution of the PA precursors, the tt19 seeds were stained with vanillin (Figure 3b). Vanillin reacts with monomer units of PA precursors and terminal units of PAs in acidic conditions, resulting in the deposition of red pigments in the sites where the PA precursors are accumulated (Deshpande et al., 1986). No reddish color was observed in unstained immature seeds of Col and tt19 (data not shown). At 2–3 days after flowering (DAF), Col and tt19 seeds had clearly different patterns of red pigments: in the Col testa, red pigments appeared to accumulate in large vacuoles in cells of the endothelium layer, while in tt19, they were restricted to a few smaller vacuoles. At 5 DAF, the vacuole of Col was fully expanded within each constitutive cell, whereas such an expanded vacuole was not detected in tt19 (Figure 3b, upper panels). A difference in the distribution pattern of red pigments was continuously detected until 9–10 DAF, after which whole-mount observation was difficult probably because of hardening of the seed coat. To examine the distribution of PA precursors in such a hardened seed coat, we tried to dissect seed coat segments and their most inner (endothelium) layer was observed. Examples at 17 DAF are shown in Figure 3(b, lower panels). In Col, the red pigmentation was very weak and marginal in the constitutive cells of the endothelium layer. In contrast, the red pigmentation was observed throughout each endothelium cell in the tt19 mutants.
Molecular cloning and characterization of the TT19 gene
The TT19 gene was molecularly mapped using F2 individuals derived from crosses of Landsberg erecta with tt19-1 or tt19-2 mutants. Rough mapping suggested that TT19 was localized close to some flavonoid genes such as TT4, TT7, and flavonol synthase 1. However, allelism tests and sequence analyses showed that none of these genes was the TT19 gene (Shikazono et al., 2003). Further mapping was therefore performed using additional F2 individuals. For the tt19-1 allele, DNA markers at 20.6, 23.7, and 25.3 cM of chromosome 5 on the recombinant inbred (RI) map showed a gradual reduction of recombination frequencies, and the TT19 gene was localized at around 29.5 cM. On the other hand, recombination values of the 42.2 and 50.5 cM markers localized the TT19 gene at around 35.5 cM. In addition, recombination was not detected in any of 45 F2 individuals in the region between these two possible locations of the TT19 gene. This phenomenon is often found in mutants induced by ion beams, and suggests that inversion has taken place in this region with the breakpoints around 29.5 and 35.5 cM in the tt19-1 mutant. For the tt19-2 allele, a gradual decline of recombination values into a chromosomal site at around 35 cM was obtained among 103 F2 individuals. These results suggested that the TT19 gene is located in the vicinity of 35 cM, and that in the tt19-1 mutant, one of the breakpoints of the assumed large inversion coincided with the TT19 locus.
We first focused on about five bacterial artificial chromosome (BAC) or P1 clones, to which the map position of the tt19-2 allele has been narrowed down. As the primary characteristic of tt19 mutants is that they have little or no anthocyanins, the TT19 gene appears to be involved in the synthesis and/or accumulation of anthocyanins. In support of this hypothesis, several TT19 candidates could be found from DNA sequences of the five BAC or P1 clones.
Based on the mapping data from the tt19-2 allele, we started to analyze the most probable candidate (identified as a GST-like gene on P1 clone MKP11). The GST-like gene was amplified by PCR using four primer sets covering the whole gene (Figure 4a). Two out of four fragments (f1-r1 and f2-r2) were not amplified in tt19-1, suggesting that these non-amplified DNA regions of the tt19-1 mutant include a breakpoint of the putative large inversion. To isolate the DNA fragment including the rearranged point, thermal asymmetric interlaced (TAIL)-PCR was carried out. The sequence of the TAIL-PCR product demonstrated that the region downstream of the GST-like gene was joined to sequences that are completely homologous to those of BAC F18O22 on chromosome 5 in the reverse direction, with a filler-DNA-like 13-bp sequence at the border. The other rejoining point of the inversion was amplified by TAIL-PCR and sequenced, and was found to contain F18O22 and MKP11 sequences. A 7-bp filler-DNA-like segment is present at the border. Therefore, we concluded that in the tt19-1 mutant, inversion had occurred with the breakpoints in F18O22 and in the second intron of GST-like gene on MKP11 (Figure 4a). The size of this inversion was estimated at about 1000 kbp based on the Arabidopsis genome database. Genomic sequence data indicate that no gene is located at the breakpoint on BAC F18O22.
PCR analysis was also carried out for the GST-like locus in the tt19-2 mutant. The f1 and r1 primers did not amplify a product, suggesting that the tt19-2 mutant underwent DNA rearrangement with a breakpoint in the f1-r1 region. TAIL-PCR revealed that the nucleotide at position −53 (based on defining the translation initiation site as +1) was rejoined to a region 16.7 kbp further upstream (Figure 4a). No other mutation was found on the GST-like gene from the rejoined site to 60 bp downstream of exon 3. As in the case of the tt19-1 mutant, the genomic sequence data indicate that there is no gene on the upper breakpoint of 16.7-kbp fragment of the tt19-2 allele.
As both tt19 mutants have mutations in the GST-like gene, molecular complementation of the tt19 mutants was carried out using the wild-type GST-like gene with its authentic promoter (about 2.4 kbp in length). In each of the five independent T1 plants (named the tt19/TT19 line), anthocyanin in the seedlings and brown pigmentation in the testa at the ripening stage were both restored to wild-type level (Figure 4b). This is conclusive evidence that disruption of the GST-like gene is responsible for the flavonoid-deficient phenotype of the tt19 mutants. In the following, this gene is referred to as TT19. RT-PCR revealed that expression of the TT19 gene was abolished in the two tt19 mutants (data not shown).
Plant GSTs are divided on the basis of sequence identity into five types (Phi, Tau, Theta, Zeta, and Lambda; Dixon et al., 2002). As the TT19 gene is classified as a type Phi GST gene, the deduced amino acid sequence of wild-type TT19 gene was compared with that of other Arabidopsis Phi GSTs, such as an auxin-inducible GST (GST6; Chen et al., 1996), a dehydration-inducible GST (early-responsive to dehydration (ERD)13; Kiyosue et al., 1993), and an expressed sequence tags (EST) clone H36860 protein (Alfenito et al., 1998). Some flavonoid-related GSTs from other species were also included in the analysis. Petunia AN9 and maize BZ2 are involved in anthocyanin accumulation (Alfenito et al., 1998; Marrs et al., 1995). Parsley GST (PcGST1) is a putative regulator that stimulates chalcone synthase (CHS) expression in the presence of UV light (Loyall et al., 2000). The Arabidopsis H36860 protein is a functional GST with a high amino acid identity to AN9, and was previously found not to complement bz2 (Alfenito et al., 1998). A phylogenetic analysis indicated that TT19 primarily clustered with H36860, and secondarily clustered with AN9, with high bootstrap values (Figure 4c). BZ2 and PcGST1 formed a distinct cluster, which is consistent with their previous classification as type Tau GSTs (Loyall et al., 2000).
Functional complementation of tt19 mutation with petunia AN9
The results of the naringenin feeding experiment and the phylogenetic analysis suggest that TT19 is a putative ortholog of petunia AN9. To determine the function of TT19, a construct containing AN9 with the cauliflower mosaic virus (CaMV) 35S promoter was introduced into the tt19 mutants. All surviving T1 seedlings (tt19/35S:AN9 line) exhibited anthocyanin pigmentation on the selection media (Figure 5a). However, the seed color at the ripening stage remained the same as that of tt19 in all transgenic plants (Figure 5b; Table 1). RT-PCR revealed expression of AN9 mRNA even in developing siliques from these T1 plants (data not shown). A control construct of TT19 cDNA with the CaMV 35S promoter was separately introduced into the mutants (tt19/35S:TT19 line). This control transgenic experiment confirmed the ability of TT19 to complement the seed color of tt19 even under the CaMV 35S promoter (Figure 5b), although three plants with pale brown seeds were found in this control transgenic line (Table 1).
Table 1. Complementation of tt19 by the TT19 or AN9 genes
No. of seedlings with anthocyanins
No. of individuals with seed color
tt19 ban double mutant analysis
BANYULS (BAN) encodes anthocyanidin reductase (ANR), one of the enzymes involved specifically in the PA biosynthetic pathway in Arabidopsis (Xie et al., 2003; Figure 1). Loss of function of the BAN gene results in precocious accumulation of anthocyanins and no accumulation of flavan-3-ols in the seed coat (Albert et al., 1997; Devic et al., 1999), which is caused by the change of the flow of competitive substrate (anthocyanidins) from PA synthesis to anthocyanin synthesis (see Figure 1). To analyze the function of TT19 in the flavonoid pathway, a double mutant with tt19 and ban was constructed and its phenotype was characterized.
Pigmentation of anthocyanins was not visible in leaves and stems of the tt19 ban double mutant. The immature seed coat of the double mutant had a very weak coloration but no conspicuous accumulation of anthocyanins (Figure 6a). A vanillin assay did not detect PA precursors in the immature seed coat of the double mutant, as was the case in the ban mutant (Devic et al., 1999) and the ban-4 mutant used here (unpublished data). The seed color of the double mutant at the ripening stage (Figure 6b, third from left) was about the same shade of brown as that of tt19 (second from left), although it was a little grayer than tt19. The gray coloration of the double mutant seems to be because of the leakage of some of the anthocyanins during the immature stage (Figure 6a). One of the characteristics of tt19 is a darkening of seed color during long-term desiccation. In contrast, no change in seed color was observed in the double mutant after an additional desiccation period (Figure 6c).
TT19 is essential for vacuolar accumulation of anthocyanins
Molecular complementation of tt19 with wild-type GST-like gene provided conclusive proof that a defect of this GST-like gene was responsible for the tt19 phenotype such as a reduced level of anthocyanin in the whole plant body and low content of brown pigments in the testa. Two kinds of GSTs are known to be involved in the flavonoid and/or anthocyanin pathway. The first type is the parsley GST (PcGST1), which appears to act in the early steps of a UV light-dependent signal transduction pathway leading to CHS expression (Loyall et al., 2000). Although a Pcgst1 mutant has not been isolated, loss of function of the gene may result in a reduced level of flavonoid pigments. The other type of GST, which includes maize BZ2 (Marrs et al., 1995) and petunia AN9 (Alfenito et al., 1998), participates in the last step of anthocyanin accumulation. The naringenin feeding experiment in the present study indicated that a defect in tt19 mutants occurs in a step downstream of naringenin synthesis by CHI (see Figure 1). In addition, we observed normal expression of CHS mRNA in tt19 (data not shown), indicating that TT19 is not necessary to stimulate CHS expression. Furthermore, the amino acid identity between TT19 and AN9 (about 50%) is quite high when compared with the typically low identities of proteins in the Arabidopsis GST family (as low as 30%; Dixon et al., 2002). We therefore examined whether AN9 complemented the tt19 phenotype. All transformants accumulated anthocyanins at the seedling stage (Figure 5a), like the wild type on the normal sugar media (see Figure 3a). Therefore, we conclude that the function of TT19 in the Arabidopsis anthocyanin pathway, like that of AN9 in petunia (Alfenito et al., 1998; Mueller et al., 2000), is to take up anthocyanidin 3-glucosides into the vacuolar membrane (Figure 7).
The petunia an9 mutation was complemented by the maize BZ2, and the maize bz2 mutation was complemented by the petunia AN9 (Alfenito et al., 1998). However, the maize bz2 mutation was not complemented by Arabidopsis H36860, which has about 50% amino acid identity to AN9 (Alfenito et al., 1998). (We presume that Arabidopsis H36860 was used in that study because, of the Arabidopsis genes known at that time, it was the one that was most homologous to AN9.) The failure of H36860 to complement bz2 suggested that Arabidopsis might use different transport mechanisms without using GST (Mueller et al., 2000; Winkel-Shirley, 2002). Examples of such transport mechanisms are direct transport (Klein et al., 1996) and transport after modification of anthocyanins (Hopp and Seitz, 1987), both of which are suggested to be functioning by the in vitro analyses. Our mutant analysis, with other mutant reports (Alfenito et al., 1998; Marrs et al., 1995), leads us to expect that the in vivo role of GST for anthocyanin (model III) is relatively conserved in many plant species.
The finding that TT19 is a GST with anthocyanin affinity indicates an interesting evolution of this gene. The Arabidopsis genome has 47 GST genes (Dixon et al., 2002), including TT19. Of the remaining 46, the one that is most closely related to TT19 is H36860 (TT19 and H36860 correspond to GSTF12 and GSTF11, respectively; Dixon et al., 2002). Our phylogenetic analysis showed that TT19 and H36860 evolved after AN9 diverged from a common ancestral protein (Figure 4c). The anthocyanin affinity of AN9 is conserved in TT19, but is likely reduced or lost in H36860; the reduction or loss of anthocyanin affinity in H36860 was suggested by a great reduction of anthocyanin pigments in the tt19 mutant (see Figure 2a), which has an intact H36860 gene. The reduction of affinity was also suggested by the inability of H36860 to complement the bz2 mutation (Alfenito et al., 1998). Although the function of H36860 is unknown, comparative sequence analysis of TT19 and H36860 might help to elucidate the molecular mechanisms underlying the affinity of GSTs for anthocyanin as a substrate. Flavonols, another flavonoid subclass, are also known to be transported into vacuoles in the same cells (Mol et al., 1998). Although no experiments for the subclass were performed in the current study, TT19 might also be involved in flavonol accumulation in Arabidopsis.
Implication of TT19 function in PA pathway
In the present study, a mutation in a GST-like gene, which unequivocally participates in anthocyanin accumulation in the vegetative parts, also provoked a great reduction of pigments in the seed coat (Figure 2). It is unlikely that only a lack or reduction of anthocyanin pigments in the seed coat causes the tt phenotype, because the brown pigments in mature Arabidopsis seeds are thought to rely mainly on the oxidation of PAs (Chapple et al., 1994; Debeaujon et al., 2001). To examine the possibility that the PA pathway is also affected by the tt19 mutation, we used a vanillin assay (Debeaujon et al., 2000, 2001; Nesi et al., 2001). The depositional patterns of PA precursors in seeds (Figure 3b) indicate that the mechanism by which PA precursors are accumulated in the central vacuoles is affected by the tt19 mutation, while the precursors would be normally synthesized.
The 35S:TT19 cDNA construct could complement both anthocyanin and seed color in tt19, except three plants that produced pale brown T2 seeds were found (Table 1). As all plants in the tt19/35S:TT19 line accumulated anthocyanins in the vegetative part, co-suppression or unstable expression of the transgene(s) might have occurred in a tissue-specific manner in those plants. On the contrary, heterologous expression of AN9 in tt19 complemented the anthocyanin pigmentation but not the brown pigmentation in the testa in all 18 transgenic plants (Figure 5; Table 1), despite the presence of the AN9 mRNA in developing siliques. These results indicate that the inability of AN9 to complement the seed color of tt19 reflects a difference in gene function between TT19 and AN9. These results also suggest that the brown pigments of wild-type seeds indeed rely mainly on PA derivatives, and that TT19 functionally participates in the PA pathway of Arabidopsis. In petunia, two distinctive accumulation pathways for anthocyanin and PAs may utilize different GST proteins.
One of the characteristics of tt19 mutants is that long-term desiccation increases brown pigmentation in the testa to as much as wild-type level, despite the clear pale brown coloration at the ripening stage (Figure 2b). This is probably caused by oxidation of PAs (and their precursors) by aerobic conditions. This characteristic has also been observed in the tt10 and tt14 mutants (Debeaujon et al., 2000), but neither of these mutants has been characterized in detail. This characteristic disappeared in the tt19 ban double mutant (Figure 6c). BAN encodes ANR, which participates in the biosynthesis of 2,3-cis-flavan-3-ols (Xie et al., 2003; Figure 1). As other functions of the Arabidopsis BAN gene have not been suggested in planta, it is conceivable that 2,3-cis-flavan-3-ols such as (–)-epicatechins are essential for the browning of tt19 testa during long-term desiccation. Therefore, it is reasonable that the pale brown color at the ripening stage in tt19 testa is because of a lack of participation of 2,3-cis-flavan-3-ols in PA synthesis until the ripening stage. The results of the vanillin assay appear to support this idea (Figure 3b).
Relationship between TT19 function and specific transporters
Our findings clearly show that TT19 encodes a GST-like protein, and is functionally implicated in vacuolar transport and/or accumulation of both anthocyanins and PAs. The GST- or glutathione-GST-bound anthocyanins are thought to be taken up into vacuoles through an ATP-dependent glutathione-specific pump (Edwards et al., 2000; Martinoia et al., 1993), which is classified as belonging to the superfamily of ATP-binding cassette (ABC) transporters (Rea, 1999). In fact, two multidrug resistance-associated protein (MRP)-type ABC transporters of Arabidopsis were found to be able to mediate the vacuolar uptake of anthocyanin–glutathione conjugates in yeast (Lu et al., 1997, 1998). On the other hand, a GST does not appear to be involved in the PA pathway because the sequence of TT12 protein indicates it is not a member of the ABC transporter family, which is the only transporter family that is involved in the transport of GST- or glutathione–GST conjugates. TT12 is the only protein known to be involved in the PA sequestration machinery so far. It belongs to the MATE-type transporter family and is thus assumed to recognize some moieties such as glycosyl, acyl, or malonyl moieties, but not glutathione (Debeaujon et al., 2001). This raises the question why both accumulation pathways, which seem to recognize distinct forms, are affected by a single mutation in the TT19 locus. In the tt12 mutant, large amounts of PA precursors were spread diffusely over the cytoplasm rather than accumulated in the vacuoles (Debeaujon et al., 2001). On the contrary, in the immature seed coat of the tt19 mutant, PA precursors were obviously wrapped with a membrane-like structure (Figure 3b, upper panels). At 17 DAF, a positive reaction of vanillin with PA precursors was detected throughout each endothelium cell in tt19, in contrast to no reaction in the endothelium cells in Col (Figure 3b, lower panels). These results indicate that most PA precursors in tt19 are not only protected from the cellular machinery with a membrane-like structure, but also remain non-degraded and non-condensed during seed maturation. These observations suggest that vacuolar uptake of PA precursors itself is not abolished by the tt19 mutation. Some flavonoids were proposed to be transported from the cytosol to the vacuole using small membrane-like structures (so-called vesicles) containing flavonoids (Grotewold, 2001). It is thus tempting to speculate that TT19 acts after PA precursors are taken up into the membrane by TT12 (Figure 7), for example at the stage of intracellular vesicle trafficking of flavonoids. As described above, the function of TT19 in the anthocyanin pathway appears to be the vacuolar uptake of anthocyanidin 3-glucosides via a membrane-localized specific transporter. Although implication of TT19 in normal accumulation of PA precursors is evident, its function might be different from that in the anthocyanin pathway. Further biochemical and cell biological investigations using the TT19 protein would help to elucidate the current hypothetical model, including the possible involvement of TT19 in flavonol accumulation (Figure 7).
Mutagenesis of Arabidopsis thaliana ecotype Col and isolation of two tt19 mutants (named tt19-1 and tt19-2) have been previously described by Shikazono et al. (2003). ast (Tanaka et al., 1997) and tt4(C1) (Shikazono et al., 2001) have a Col background. A preliminary experiment showed that ast was not complemented by the ban mutant, indicating that ast was allelic to ban (Winkel-Shirley, 2001), and had a deletion of 49 bp (+114 to 162 nt) in BAN, resulting in a null mutant. Therefore, ast is referred to as ban-4.
Quantification of anthocyanin content
From rosette leaves (100 mg) harvested from about 45-day-old plants, anthocyanins were extracted with 1% HCl:methanol and measured with a spectrophotometer, as described previously by Tanaka et al. (1997). The average values for the absorption spectrum from 300 to 700 nm were obtained from five independent experiments.
Naringenin feeding and vanillin treatment
Dry seeds were sterilized and sown on MS/sucrose/agar (0.8%) plates with or without 0.1 mm naringenin as described by Shirley et al. (1995). The sucrose concentration was 0, 1, 2, 5, 10, or 20%. Following vernalization at 4°C for 5 days, these plates were incubated in a growth chamber set at 23°C with continuous light and observed every day by stereomicroscope (Stemi SV11, Zeiss, Germany).
Immature seeds were treated with vanillin essentially according to the method of Debeaujon et al. (2000). Samples up to 10 DAF were investigated by whole-mount observation. After 10 DAF, seed coat segments from vanillin-treated seeds were dissected, and the endothelium layers were observed using a light microscope (Axioskop, Zeiss, Germany).
Molecular mapping and cloning of the TT19 gene
Molecular mapping was performed using F2 genomic DNA in relation to the linkage with the cleaved amplified polymorphic sequence (CAPS) and single sequence length polymorphysms (SSLP) markers according to the standard methods by Bell and Ecker (1994) and Konieczny and Ausubel (1993).
Four combinations of primer (TT19-f0 and TT19-r0, TT19-f1 and TT19-r1, TT19-f2 and TT19-r2, and TT19-f3 and TT19-r3; Table 2) were designed around the putative TT19 locus. The amplified fragments were recovered, purified, and sequenced using standard procedures.
The points of DNA rearrangements in two tt19 mutants were determined by TAIL-PCR as described previously by Liu and Whittier (1995) and Liu et al. (1995a). Two sets of three nested specific primers were used, respectively, for isolation of two junction sequences of inverted DNA in tt19-1 (MKP11-R4, MKP11-R5, and MKP11-R6 for the downstream junction; MKP11-F7, MKP11-F8, and MKP11-F9 for the upstream junction; Table 2). Three oligonucleotides (bCC5-8-R1, bCC5-8-R2, and bCC5-8-R3; Table 2) were used as the nested specific primers in tt19-2. Two arbitrary degenerate primers (AD2 and AD3) were synthesized according to the sequences described by Liu et al. (1995a). The sequence of another AD primer (AD1) was 5′-GTN CGA (G/C)(A/T)C ANA (A/T)GT T-3′.
The sequence of an EST clone (169M6, derived from Newman et al., 1994) was determined for both strands, and was confirmed to consist of the complete length of the TT19 cDNA.
Expression of TT19 and other genes was determined by RT-PCR using 500 ng of total RNA from 6-week-old plants and a Takara RNA LA PCR Kit (ver. 1.1, Takara). The PCR program consisted of an initial denaturation step of 94°C for 2 min and 30 cycles of 94°C for 0.5 min, 57°C for 0.5 min, and 72°C for 1.5 min, and final extension step of 72°C for 7 min. For TT19 expression, TT19-RT/f2 and TT19-RT/r1 were used for the specific primers (Table 2). Expression for the elongation factor 1αA4 was determined as an internal control using a primer pair reported by Nesi et al. (2000). For expression of other flavonoid genes and petunia AN9, the primers shown in Table 2 were used.
A KpnI/SacI fragment (about 2.4 kbp) containing the wild-type TT19 gene with its authentic promoter was isolated from P1 clone MKP11 (Liu et al., 1995b). The TT19 cDNA was recovered from the EST clone 169M6. Using RNA from floral buds of petunia V26 line, which is kindly provided from Dr Akira Kanazawa (Hokkaido University), the AN9 cDNA was amplified by RT-PCR as described above. Sequencing of the TA-cloning products allowed us to identify the nucleotide change in exon 3 in all AN9 cDNA clones. This mutation resulted in the substitution Val80 → Asp80. The genomic TT19 gene and two of its cDNAs (TT19 and AN9) were isolated and introduced into the binary vectors pBI101 and pBI121 (Jefferson et al., 1987), respectively. Three kinds of binary vectors were separately transformed into Agrobacterium GV3101 by electroporation. Agrobacterium clones possessing the binary vector were infected to tt19 mutants by the floral dip method (Clough and Bent, 1998). The T1 seedlings were screened by growing them in medium containing kanamycin (50 mg l−1) and Claforan (166 mg l−1). The transformants were grown, and their phenotypes were observed mainly in relation to flavonoid pigmentation.
The authors wish to thank Yutaka Oono, Ayako Sakamoto, and Yoshihiro Hase for their helpful suggestions for the molecular analyses, and Chihiro Suzuki for technical assistance with the sequence analysis. We also thank the Arabidopsis Biological Resource Center for providing cDNA clone 169M6, the Kazusa DNA Research Institute for providing P1 clone MKP11, Akira Kanazawa for his kind gift of petunia V26 seeds and for helpful suggestions, and Loïc Lepiniec for providing the name of the ban allele. We are grateful to James Raymond for careful review of the manuscript.
Nucleotide accession numbers: AB111443 (the upstream junction sequence of the large inversion in the tt19-1 mutant, containing part of the tt19-1 gene); AB111444 (the downstream junction sequence of the large inversion in the tt19-1 mutant, containing part of the tt19-1 gene); AB111445 (the tt19-2 allele containing the rejoining point).