Regular Paper Violet/Blue Chrysanthemums Metabolic Engineering of the Anthocyanin Biosynthetic Pathway Results in Novel Petal Colors Filippa Brugliera 1,2, Guo-Qing Tao 1,3, Ursula Tems 1, Gianna Kalc 1,2, Ekaterina Mouradova 1,2, Kym Price 1, Kim Stevenson 1, Noriko Nakamura 4, Iolanda Stacey 1,2, Yukihisa Katsumoto 4, Yoshikazu Tanaka 4, * and John G. Mason 1,3 1 Florigene Pty Ltd., VABC, 1 Park Drive, Bundoora, Victoria, Australia 2 La Trobe Institute of Molecular Sciences, La Trobe University, Department of Biochemistry, Kingsbury Drive, Bundoora, 3086, Victoria, Australia 3 DEPI Biosciences Research Division, AgriBio, Centre for AgriBiosciences, 5 Ring Road, La Trobe University, Bundoora, 3083, Victoria, Australia 4 Research Institute, Suntory Global Innovation Center Ltd., Shimamoto, Osaka, 618-8503 Japan *Corresponding author: E-mail, Yoshikazu_Tanaka@suntory.co.jp; Fax, +81-75-962-3791. (Received May 16, 2013; Accepted July 29, 2013) Chrysanthemums (Chrysanthemummorifolium Ramat.) are an important cut-flower and potted plant crop in the horticultural industry world wide. Chrysanthemums express the flavonoid 3 0 -hydroxylase (F3 0 H) gene and thus accumulate anthocyanins derived from cyanidin in their inflorescences which appear pink/red. Delphinidin-based anthocyanins are lacking due to the deficiency of a flavonoid 3 0,5 0 -hydroxylase (F3 0 5 0 H), and so violet/blue chrysanthemum flower colors are not found. In this study, together with optimization of transgene expression and selection of the host cultivars and gene source, F3 0 5 0 H genes have been successfully utilized to produce transgenic bluish chrysanthemums that accumulate delphinidin-based anthocyanins. HPLC analysis and feeding experiments with a delphinidin precursor identified 16 cultivars of chrysanthemums out of 75 that were predicted to turn bluish upon delphinidin accumulation. A selection of eight cultivars were successfully transformed with F3 0 5 0 H genes under the control of different promoters. A pansy F3 0 5 0 H gene under the control of a chalcone synthase promoter fragment from rose resulted in the effective diversion of the anthocyanin pathway to produce delphinidin in transgenic chrysanthemum flower petals. The resultant petal color was bluish, with 40% of total anthocyanidins attributed to delphinidin. Increased delphinidin levels (up to 80%) were further achieved by hairpin RNA interference-mediated silencing of the endogenous F3 0 H gene. The resulting petal colors were novel bluish hues, not possible by hybridization breeding. This is the first report of the production of anthocyanins derived from delphinidin in chrysanthemum petals leading to novel flower color. Keywords: Anthocyanin Chrysanthemum Delphinidin Flavonoid Flower color Metabolic engineering. Abbreviations: BA, 6-benzylaminopurine; CaMV, Cauliflower mosaic virus; CHS, chalcone synthase; DFR, dihydroflavonol 4-reductase; DHK, dihydrokaempferol; DHM, dihydromyricetin; DHQ, dihydroquercetin; F3 0 H, flavonoid 3 0 -hydroxylase; F3 0 5 0 H, flavonoid 3 0, 5 0 -hydroxylase; GUS, b-glucuronidase; hprnai, hairpin RNA interference; LB, left border; MS medium, Murashige and Skoog medium; NAA, napthaleneacetic acid; RNAi, RNA interference; RT PCR, reverse transcription PCR; T-DNA, transferred DNA; RHSCC, Royal Horticultural Society Color Chart. Introduction Flower color is an important characteristic for attracting pollinators and therefore influences plant reproductive success. Flower color and a pollinator s color vision have co-evolved (Rausher 2006), which has resulted in a limited flower color range in wild plant species. In the flower and ornamental plant industry, the development of novel colored varieties is of particular interest. However, limited color and genetic variation of wild species inhibits such developments. In general, three classes of pigments contribute to flower color: flavonoids, carotenoids and betalains (Tanaka et al. 2008). The flavonoids and their colored class of compounds, anthocyanins, are the most common and confer a wide range of flower colors. The flavonoid biosynthetic pathway has been well characterized in terms of genetics, biochemistry and molecular biology, and consists of enzymes from several Plant Cell Physiol. 54(10): 1696 1710 (2013) doi:10.1093/pcp/pct110, available online at www.pcp.oxfordjournals.org! The Author 2013. Published by Oxford University Press on behalf of Japanese Society of Plant Physiologists. All rights reserved. For permissions, please email: journals.permissions@oup.com 1696 Plant Cell Physiol. 54(10): 1696 1710 (2013) doi:10.1093/pcp/pct110! The Author 2013.
Violet blue chrysanthemums by metabolic engineering protein families that have been characterized in many plant species (Davies and Schwinn 2006, Grotewold 2006, Tanaka and Brugliera 2006). Anthocyanins are water-soluble pigments that are stored in vacuoles of petal epidermal cells or in some cases, vacuoles of subepidermal cells of leaves. Anthocyanin color is dependant upon chemical structure, vacuolar ph, copigmentation (usually by flavones and/or flavonols) and metal ion complexation. Final petal color is achieved by genetic regulation of these factors. Blue or violet flower colors can be achieved through the accumulation of delphinidin-based anthocyanins often modified with aromatic acyl groups, higher (neutral) vacuolar ph and the presence of co-pigments and/or metal ions (Yoshida et al. 2009). The pattern of B-ring hydroxylation of anthocyanins plays a key role in determining their color. Dihydrokaempferol (DHK) can be hydroxylated at the 3 0 position to produce dihydroquercetin (DHQ) and at the 3 0 and 5 0 position to produce dihydromyricetin (DHM) (Fig. 1). DHQ leads to the generation of cyanidin-based anthocyanins contributing to red and pink flower color. The production of DHM leads to the generation of delphinidin-based anthocyanins. Cyt P450 enzymes catalyze these hydroxylations, flavonoid 3 0 -hydroxylase (F3 0 H) (classified to CYP75B in Cyt P450 nomenclature) and flavonoid 3 0,5 0 - hydroxylase (F3 0 5 0 H) (mainly CYP75A), and therefore the presence or absence of these enzymes in different plant species contributes to many of the petal color variations observed in nature (Tanaka and Brugliera 2013). Roses, carnations and chrysanthemums, which are the most important horticultural cut-flower crops, do not generally accumulate anthocyanins derived from delphinidin due to the absence of F3 0 5 0 H activity. Genes encoding F3 0 5 0 H were first isolated from petunia (Holton et al. 1993) and have now been cloned from a number of species and successfully used to engineer the flavonoid biosynthetic pathway to produce 3 0, 5 0 -hydroxylated flavonoids and therefore modulate petal color. In carnation and rose, this has led to the commercial production of violet/blue-colored flowers (Katsumoto et al. 2007, Tanaka et al. 2009, Tanaka and Brugliera 2013). Chrysanthemums (Chrysanthemum morifolium Ramat.) are herbaceous, perennial plants with large flower heads, which are white, yellow or pink in the wild species. All cultivated chrysanthemums are allohexaploid (2n = 6x = 54) and their ancestry includes 10 or more primarily hexaploid species (Anderson 2007). Transformation of chrysanthemum has been reported by many groups (Deroles et al. 2002, da Silva 2003), with low levels of gene expression generally obtained when using the Cauliflower mosaic virus (CaMV) 35S promoter to control transgene expression, with Takatsu et al (2000) reporting that b-glucuronidase (GUS) gene expression decreased gradually and elapsed 1 year after transformation. Higher GUS expression was achieved using the promoter of a gene for chrysanthemum Chl a/b-binding protein; however, expression was poor in petals (Aida et al. 2004). It was therefore considered necessary to optimize expression of heterologous genes in chrysanthemum to achieve consistent and significant delphinidin production and petal color change. The anthocyanins generally found in chrysanthemum petals are cyanidin 3-mono or di malonylglucosides (Saito et al. 1988, Nakayama et al. 1997), and the predominant flavonoids are flavones [apigenin, acacetin (4 0 -methoxyapigenin), luteolin and diosmetin (4 0 -methoxyluteolin)] (Fig. 1) (Schwinn et al. 1994) which can act as co-pigments with anthocyanins to enhance flower color intensity. Pelargonidin-based anthocyanins are rarely found in chrysanthemum, but when their florets were incubated in tetcyclasis (a Cyt P450 inhibitor and thus an inhibitor of F3 0 H), anthocyanins derived from pelargonidin were produced, suggesting that the dihydroflavonol 4-reductase (DFR) specificity was not a factor in the lack of pelargonidinbased anthocyanins in chrysanthemums (Schwinn et al. 1994) as is the case in petunia (Forkmann and Ruhnau 1987), but was due to the competing F3 0 H activity. Yellow colors found in chrysanthemum petals are derived from carotenoids (Jordan and Riemann-Philipp 1983, Kishimoto et al. 2004), with bronze shades being the combination of the accumulation of red anthocyanins and yellow carotenoids in petals. One reason why violet/blue-colored chrysanthemum flowers have not been generated by classical breeding practices is due to the lack of a F3 0 5 0 H activity. This is the first report of the genetic engineering of delphinidin production in chrysanthemums. Production of delphinidin was achieved by transformation with a chimeric pansy F3 0 5 0 H gene under the control of floral-specific promoters, resulting in violet/blue-colored petals. Furthermore, overexpression of F3 0 5 0 H in conjunction with RNA interference (RNAi)-mediated silencing of endogenous F3 0 H expression led to an increase in delphinidin-based anthocyanins in petal tissue and a further shift to violet/blue hues. The 3 0, 5 0 -hydroxylated flavone, tricetin, and anthocyanins derived from pelargonidin, both of which are not normally found in chrysanthemum, were also produced in petals. Results Screening of chrysanthemum cultivars by HPLC and precursor feeding assays The flavonoid compositions of 75 chrysanthemum cultivars were analyzed by HPLC. The results of 24 of the analyzed cultivars are shown in Table 1. The only anthocyanidin detected in petals was cyanidin, and the flavones detected were luteolin, apigenin, diosmetin and acacetin, with predominance of luteolin and apigenin (see Fig. 1, Table 1). The pink varieties analyzed tended to have relatively low cyanidin to high flavone ratios. Cultivars with orange or bronze shades that are derived from carotenoids were avoided as targets due to the potential impact of these pigments on any engineered colors. Precursor feeding experiments of chrysanthemum petals with DHM, a precursor of delphinidin, resulted in the coloration of petal segments (Fig. 2), and HPLC analysis of the segments confirmed the production of delphinidin. The colors on the cut Plant Cell Physiol. 54(10): 1696 1710 (2013) doi:10.1093/pcp/pct110! The Author 2013. 1697
F. Brugliera et al. Fig. 1 Schematic representation of the general flavonoid biosynthesis pathway leading to colored anthocyanins with emphasis on the flavonoid pathway in chrysanthemum petals. Enzymes involved in the pathway have been indicated as follows: PAL, phenylalanine ammonia lyase; C4H, cinnamate 4-hydroxylase; 4CL, 4-coumarate:CoA ligase; CHS, chalcone synthase; CHI, chalcone isomerase; F3H, flavanone 3-hydroxylase; F3 0 H, flavonoid 3 0 -hydroxylase; F3 0 5 0 H, flavonoid 3 0,5 0 -hydroxylase; DFR, dihydroflavonol 4-reductase; ANS, anthocyanidin synthase; 3GT, UDP-glucose:anthocyanidin 3-glucosyltransferase; 3MalT, malonyl-coa:anthocyanin 3-malonyltransferase. Classically bred chrysanthemums only accumulate cyanidin-based anthocyanins such as cyanidin 3-malonylglucoside. Lack of pelargonidin-based anthocyanins is primarily due to the activity of the F3 0 H, and the absence of anthocyanins derived from delphinidin and associated 3 0,5 0 -hydroxylated flavones (tricetin) is due to the lack of F3 0 5 0 H activity. Although not shown in this figure, F3 0 5 0 H can utilize eriodictyol and dihydroquercetin to produce pentahydroxyflavone and dihydromyricetin, respectively. edges of the petals obtained after incubation in DHM ranged from pink to brown to violet. Sixteen of the 53 cultivars assessed by precursor feeding experiments resulted in a shift to mauve or purple after incubation with DHM, and a subset of these was targeted for transformation (Fig. 3). White cultivars that were deficient only in DFR or F3 0 H as had been discovered in carnation (Tanaka and Brugliera 2013) were not found (data not shown). 1698 Plant Cell Physiol. 54(10): 1696 1710 (2013) doi:10.1093/pcp/pct110! The Author 2013.
Violet blue chrysanthemums by metabolic engineering Table 1 HPLC analysis of anthocyanidins and flavones detected in the petals of chrysanthemum cultivars Cultivar Color Cya Lut (3 0 4 0 ) Dios (3 0 4 0 ) Api (4 0 ) Aca (4 0 ) Total flavone Flavone/anthocyanidin ratio Improved Reagan Pink 0.05 1.33 0.01 1.67 0.07 3.16 63 Dark Splendid Reagan Dark pink 0.29 1.30 0.01 1.64 0.1 3.12 11 Sei Aida Red 1.17 1.97 0.18 0.26 0.15 2.57 2 Sei 050-0382 Pink 0.01 0.28 0.01 0.66 0.09 1.05 105 Sei Faust Pink 0.02 1.32 0.13 2.02 0.51 3.98 199 Sei Figaro Pink 0.01 1.56 0.17 0.64 0.21 2.58 258 Sei Titan Pink 0.06 2.1 0.18 1.06 0.33 3.70 62 Sei Florea Pink 0.06 2.64 0.04 1.32 0.08 3.70 62 Sei Spire Pink 0.02 1.23 0.02 1.35 0.05 2.65 132 Sei Amelie Pink 0.001 0.72 0.13 0.15 0.15 1.15 1150 Sei Titan 406 Pink 0.03 2.61 0.28 1.85 0.66 5.39 180 Sei Titan 410 Salmon 0.13 2.56 0.18 1.17 0.42 4.44 34 Sei Amelie Rose Pale pink 0.07 0.86 0.10 0.14 0.13 1.22 17 Sei Opera Pink Pink 0.00 0.34 0.13 0.77 0.43 1.67 1670 Mari Fusha Pink 0.02 1.46 0.28 1.85 0.66 2.80 155 Sei Opera Pink 0.001 0.29 0.12 0.82 0.46 1.69 1690 Mai Fusha Pink/white 0.01 0.80 0 2.33 0.66 3.80 475 Sei Lakme Pink 0.22 1.74 0 1.87 0 3.60 16 Sei Iolanta Pink 0.11 1.33 0.16 0.26 0.24 1.99 17 Sei Otello Dark Pink 0.11 2.30 0.27 0.75 0.42 3.74 35 Sei Carmen (Seiko Kyouka) Dark Pink 0.04 2.21 1.18 3.40 82 Seko no Uta Pale Pink 0.09 2.29 1.81 4.10 25 Sei Rusalka Pink 0.25 4.17 0.88 5.06 16 920-0686 Red/apricot 0.29 6.15 0.67 2.96 1.24 11.02 38 Amounts are in mg g 1 of fresh petal weight. Cya, cyanidin; Lut, luteolin, Dios, diosmetin; Api, apigenin; Aca, acacetin (5, 7 di-oh, 4 0 OMe); Total flavone is the sum of luteolin, diosmetin, apigenidin and acacetin. Fig. 2 An example of precursor feeding experiments. (A) Improved Reagan inflorescence. (B) Change in petal color observed after incubation of cut petal segments in water and DHM (dihydromyricetin). (C) Color change coded using the RHSCCs. The color was a good indicator of petal color of transgenic plants producing delphinidin-based anthocyanins. Plant Cell Physiol. 54(10): 1696 1710 (2013) doi:10.1093/pcp/pct110! The Author 2013. 1699
F. Brugliera et al. Fig. 3 A selection of chrysanthemum cultivars highlighting those deemed suitable for transformation to achieve blue coloration, based upon flavonoid analysis and precursor feeding experiments. Cultivars marked with a cross were deemed unsuitable. Cultivars with names highlighted were successfully transformed with the F3 0 5 0 H gene. Table 2 Results of analysis of transgenic Improved Reagan chrysanthemum petals containing F3 0 5 0 H genes Construct Transgenes #tg # CC HPLC RHSCC RHSCC group Improved Reagan - 75A Purple F3 0 5 0 H genes pcgp2205 pdcans:pansyf3 0 5 0 H:tDcANS 12 1 3 18% (n = 8) 84C Violet pcgp2788 pcamv35s:pansyf3 0 5 0 H:t35S 14 0 0.5 3% (n = 2) 75A Purple pcgp3141 Cineraria gf3 0 5 0 H 50 2 3 20% (n = 11) 77D Purple pcgp2217 prhchs:pansyf3 0 5 0 H:tnos 37 5 0.1 37% (n = 26) 84C Violet prhchs: pansyf3 0 5 0 H:tnos and hprnai F3 0 H cassettes pcgp3424 prhchs: pansy F3 0 5 0 H:tnos; p35s:dsf3 0 H:tocs 50 11 3 41% (n = 10) 84B Violet pcgp3429 prhchs: pansy F3 0 5 0 H:tnos; prhchs: ds F3 0 H*:tnos 50 14 26 53% (n = 10) N82D Purple violet pcgp3618 prhchs: pansy F3 0 5 0 H:tnos; prhchs:dsf3 0 H**:tnos 49 39 32 80% (n = 28) N82B/C Purple violet Construct, transformation binary vector; Transgenes, transgenes contained in the transformation vector (besides the 35S: SuRB selectable marker gene); #tg, total number of transgenic events produced and flowered; #CC, number of individual transgenic events that produced inflorescences with a significant shift in color (as determined by eye) compared with that of a non-transformed plant; HPLC, ranges in percentage levels of delphinidin detected in total petal anthocyanidins of a selected number of transgenic events where n = number analysed by HPLC; RHSCC, Royal Horticultural Society Color Chart code of the ray petals from the inflorescences of chrysanthemum with the most significant color changes (as determined by eye) as compared with the control petals; RHSCC group, color group associated with the petal color according to the Royal Horticultural Society. An increase in RHSCC of the transgenic compared with the control suggests that the transgenic is bluer than the control. The chrysanthemum cultivars chosen for transformation based on the results of the precursor feeding assays and on the presence of flavones (a co-pigment that can enhance bluing) included Improved Reagan, Dark Splendid Reagan, Sei Titan, Sei Titan406, Sei Figaro, Sei Florea, Sei 050-0382 and Sei Spire (Fig. 3). Due to ease of transformation, Improved Reagan was used as a model cultivar to test the efficacy of various F3 0 5 0 H transgenes. Optimization of transgene expression Different promoter fragments and coding sequences were assessed for their functionality in chrysanthemum petals as not all transgenes that led to high delphinidin accumulation in other species such as carnation, petunia and rose (Brugliera et al. 2004, Tanaka and Brugliera 2013) were similarly effective in chrysanthemum (data not shown). A subset of promoter and F3 0 5 0 H sequence combinations that resulted in delphinidin production in Improved Reagan petals is shown in Table 2. Of the F3 0 5 0 H chimeric transgenes assessed in Improved Reagan, transferred DNA (T-DNA) of pcgp2217 (Fig. 4) containing a pansy F3 0 5 0 H coding sequence under the control of a promoter fragment from a rose chalcone synthase (CHS) gene (prhchs) with a terminator fragment from the nopaline synthase gene (tnos) of Agrobacterium tumefaciens resulted in the bluest shift in petal color and highest percentage of delphinidin of total anthocyanidins in the petals (Table 2). In general, petals 1700 Plant Cell Physiol. 54(10): 1696 1710 (2013) doi:10.1093/pcp/pct110! The Author 2013.
Violet blue chrysanthemums by metabolic engineering Fig. 4 Schematic representation of the T-DNA components of selected binary plasmid vectors used for plant transformation. LB, left border; RB, right border; p35s, Cauliflower mosaic virus 35S promoter; SuRB, chlorsulfuron resistance gene encoding a mutated acetolactate synthase (ALS) with its own terminator from Nicotiana tabacum; F3 0 5 0 H, 1.8 kb pansy F3 0 5 0 H cdna clone; prhchs, promoter region from a CHS gene of Rosa hybrida; tnos, terminator region from the nopaline synthase (nos) gene of A. tumefaciens; dsf3 0 H, double-stranded chrysanthemum partial F3 0 H coding sequences in sense and antisense orientations spaced by intron 1 of the petunia DFR-A gene; dsf3 0 H*, double-stranded chrysanthemum partial F3 0 H coding sequences in sense and antisense orientations; dsf3 0 H**, double-stranded chrysanthemum partial F3 0 H coding sequences in sense and antisense orientations spaced by intron 1 of a cineraria F3 0 5 0 H gene; EcoRI restriction enzyme sites are indicated with vertical arrows. Probes used for genomic DNA blot analysis are also highlighted; the green bar represents the 0.8 kb HindIII SuRB fragment used as a probe and the blue bar highlights the 0.84 kb XhoI/SalI F3 0 5 0 H fragment used as a probe. The SurB probe would hybridize to different sized EcoRI genomic bands depending on the integration of the T-DNAs into the genome. The F3 0 5 0 H probe would hybridize to a 2 kb EcoRI genomic band in transgenic lines containing the T-DNAs from pcgp3429 or pcgp3618 and different sized EcoRI genomic bands depending on the integration of the T-DNAs from pcgp2217 or pcgp3424 into the genome. containing at least 15% of total anthocyanins derived from delphinidin showed a shift in color towards violet. Significant shifts in color were observed in petals with >30% of total anthocyanins derived from delphinidin (Fig. 5). The highest percentage of delphinidin detected in the petals of Improved Reagan/pCGP2217 transgenic inflorescences was 37% (Table 2, Fig. 6). Based on this result, the F3 0 5 0 H transgene from pcgp2217 was used as the base for further construct design and transformations of other cultivars. In order to shift the petal color further towards blue, higher delphinidin levels were required. It was thought that the endogenous F3 0 H would compete with the introduced F3 0 5 0 H for substrates. Therefore, three constructs (pcgp3424, pcgp3429 and pcgp3618) that incorporated a doublestranded (ds) chrysanthemum F3 0 H transgene aiming at down-regulation of the endogenous F3 0 H via hairpin RNAi- (hprnai) mediated silencing (Waterhouse et al. 1998, Smith et al. 2000) were designed (Fig. 4). Using these combinations, higher percentage levels of delphinidin in petals were achieved. The highest delphinidin level detected was 80% in an Improved Reagan/pCGP3618 line (Fig. 5) containing the prhchs:pansyf3 0 5 0 H:tnos transgene of pcgp2217 along with the prhchs:dsf3 0 H**:tnos transgene containing partial sense and antisense fragments of the chrysanthemum F3 0 H cdna with an intron from a cineraria F3 0 5 0 H gene (Table 2, Fig. 4). Transformation of other chrysanthemum cultivars Based on the results obtained with the model cultivar Improved Reagan, a further seven cultivars, Dark Splendid Reagan, Sei Spire, Sei Figaro, Sei Titan, Sei Titan406, Sei Florea and Sei 050-0382, were transformed with the T-DNAs contained in the selected transformation vectors pcgp2217, pcgp3424, pcgp3429 and pcgp3618. The results obtained are summarized in Table 3. The F3 0 5 0 H transgene contained in the construct pcgp2217 (Fig. 4) resulted in significant petal color changes (as determined by eye) in all cultivars with the highest delphinidin percentages across the cultivars, ranging from 12% in DSR to 50% in Sei Titan and Sei 050-0382 (Table 3). In addition, constructs with dsf3 0 H transgenes (pcgp3424, 3429 or 3618) aiming at down-regulation of the endogenous F3 0 H, resulted in an increase in delphinidin percentages in most Plant Cell Physiol. 54(10): 1696 1710 (2013) doi:10.1093/pcp/pct110! The Author 2013. 1701
F. Brugliera et al. Fig. 5 Inflorescence color changes with the production of delphinidin-based anthocyanins (A H). The host is on the left and the transgenic on the right. The percentage of delphinidin (of total anthocyanidins) detected in hydrolyzed petal extracts is also given under the transgenic inflorescence. IR, Improved Reagan; DSR, Dark Splendid Reagan. The transgenic line number is given next to the cultivar/construct. 1702 Plant Cell Physiol. 54(10): 1696 1710 (2013) doi:10.1093/pcp/pct110! The Author 2013.
Violet blue chrysanthemums by metabolic engineering Fig. 6 Percentage amounts of total anthocyanidins (A) and flavones (B) detected in the petals of the selected transgenic Improved Reagan (IR) lines for each of the constructs pcgp2217, pcgp3424, pcgp3429 and pcgp3618. An IR control is shown as a comparison. A photograph of each inflorescence is shown below the flavone chart. Pel, pelargonidin; Cya, cyanidin; Del, delphinidin; 4 0 Api, 4 0 -hydroxylated flavones (apigenin and the methylated derivative acacetin); 3 0 4 0 Lut, 3 0, 4 0 -hydroxylated flavones (luteolin and the methylated derivative diosmetin); 3 0 5 0 Tri, 3 0,4 0,5 0 -hydroxylated flavone (tricetin). Pelargonidin, delphinidin and tricetin are not found in non-transgenic chrysanthemum petals. cultivars (Table 3). Inflorescences with the greatest shift to blue for each cultivar (excluding Sei Florea) are shown in Fig. 5. Such novel bluish colors have not been achieved by traditional breeding. The results of HPLC analysis of the petals of the bluest Improved Regan transgenic lines with each construct of Fig. 4 are represented in a chart in Fig. 6, highlighting that the transgenes contained in the construct pcgp3618 (Fig. 4) resulted in the highest delphinidin levels in line 31721. The 3 0,5 0 -hydroxylated flavone tricetin was also produced in some transgenic petals (Fig. 6), indicating that the introduced F3 0 5 0 H was temporally and spatially available and able to utilize the flavones, luteolin and apigenin, and/or that the endogenous flavone synthase (FNS) was able to utilize the newly formed pentahydroxyflavone (Fig. 1). Although Fig. 6 highlights that the selected lines for pcgp3424 and pcgp3429 did not produce tricetin, there was no strict correlation between the construct used and production of tricetin. Other lines transformed with the T-DNAs of these constructs did produce tricetin (data not shown). Moreover, pelargonidin (not normally found in chrysanthemum petals) was detected in transgenic lines transformed with the dsf3 0 H expression cassettes (in pcgp3424, pcgp3429 and pcgp3618) aiming at down-regulation of the endogenous F3 0 H via hprnai-mediated silencing (Fig. 6). Molecular analysis of transgenic chrysanthemum plants. Integration of T-DNA into the chrysanthemum genome was analyzed in selected transgenic lines (Fig. 7). Shoots that produced roots on media containing 5 mgl 1 chlorsulfuron were confirmed as containing the SuRB selectable marker gene. A range of copy numbers, from single to multiple copies of at least 16, was obtained (Table 4). Moreover, rearrangements/deletions or transfer of truncated T-DNAs were also detected, as is indicated by detection of F3 0 5 0 H hybridizing bands other than the expected 2 kb band (see Fig. 4) in the EcoRI-digested genomic DNA of lines 31719 (Improved Regan/3618) and 31646 (Dark Splendid Regan/3429). The F3 0 5 0 H probe hybridized to the expected 2 kb EcoRI fragment along with a larger 9 kb fragment in 31719 and a 3 kb fragment in 31646 (Fig. 7B). There appeared to be no strict correlation between delphinidin levels accumulating and the copy number of transgenes (Fig. 7). For example, lines 31645, 31646 and 31649 all accumulated around 50% delphinidin but varied in copy number from 1 to >5 (Fig. 7). Under the transformation methods used, around 14% of the transgenic Improved Regan lines analyzed (with a change in petal color) were estimated as having a single copy of the SuRB transgene integrated into the genome (Table 4). Further analysis using probe fragments directed to the transformation vector backbone revealed that integration of vector plasmid sequences outside of the left border (LB) of the T-DNA occurred in some lines. Of the 189 lines analyzed, 57 (30%) were found to have vector DNA just outside the LB of the T-DNA (data not shown). This occurred in a selection of lines with multiple copies of T-DNA integrated. RNA blot analysis and/or reverse transcription PCR (RT PCR) of total RNA extracted from a selection of transgenic chrysanthemum petals revealed the accumulation of F3 0 5 0 H and SuRB transcripts (Fig. 8). The production of delphinidinbased anthocyanins in petals correlated well with the detection of an intact F3 0 5 0 H transcript. In lines that also contained a dsf3 0 H transgene (pcgp3618 and pcgp3429), a reduction of the endogenous F3 0 H transcripts was detected in some (Fig. 9), indicating that the endogenous F3 0 H transcripts were successfully targeted as intended. The CHS probe was used an internal control to show that the RNA was intact. In the lines analyzed, increased delphinidin levels were observed where the Plant Cell Physiol. 54(10): 1696 1710 (2013) doi:10.1093/pcp/pct110! The Author 2013. 1703
F. Brugliera et al. Table 3 Results of analysis of transgenic chrysanthemum petals containing F3 0 5 0 H genes Cultivar Construct Transgenes #tg # CC HPLC RHSCC RHS color group DSR control 70B Red purple DSR pcgp2217 prhchs:pansyf3 0 5 0 H:tnos 25 2 3 12% (n = 10) N78B/C Purple DSR pcgp3424 prhchs:pansy F3 0 5 0 H:tnos; p35s dsf3 0 H:tocs 49 9 2 33% (n = 10) N80C Purple violet DSR pcgp3429 prhchs:pansyf3 0 5 0 H:tnos; prhchs:dsf3 0 H*:tnos 34 16 27 58% (n = 8) N81B Purple violet DSR pcgp3618 prhchs:pansyf3 0 5 0 H:tnos; prhchs:dsf3 0 H**:tnos 10 5 53% (n = 1) N81C Purple violet Sei Figaro Control 65D Red purple Sei Figaro pcgp2217 prhchs:pansyf3 0 5 0 H:tnos 25 3 4 23% (n = 6) 76C Purple Sei Figaro pcgp3424 prhchs:pansyf3 0 5 0 H:tnos; p35s:dsf3 0 H:tocs 50 3 11 37% (n = 10) 76C Purple Sei Florea Control 75A/B Purple Sei Florea pcgp2217 prhchs:pansy3 0 5 0 H:tnos 15 5 12 20% (n = 4) N80C Purple violet Sei Spire Control 75C/D Purple Sei Spire pcgp2217 prhchs:pansyf3 0 5 0 H:tnos 42 7 8 39% (n = 11) N80D Purple violet Sit Titan Control 75B Purple Sei Titan pcgp2217 prhchs:pansyf3 0 5 0 H:tnos 25 7 10 50% (n = 9) N80D Purple violet Sei Titan pcgp3424 prhchs:pansyf3 0 5 0 H:tnos; p35s:dsf3 0 H:tocs 62 7 9 46% (n = 10) N81B Purple violet Sei Titan pcgp3429 prhchs:pansyf3 0 5 0 H:tnos; prhchs:dsf3 0 H*:tnos 48 7 7 31% (n = 7) N80B Purple violet Sei Titan pcgp3618 prhchs:pansyf3 0 5 0 H:tnos; prhchs:dsf3 0 H**:tnos 19 8 1 64% (n = 8) N82C Purple violet Sei Titan406 Control 75C/D Purple Sei Titan406 pcgp2217 prhchs:pansyf3 0 5 0 H:tnos 25 4 13 33% (n = 5) 84C Violet Sei Titan406 pcgp3424 prhchs:pansyf3 0 5 0 H:tnos; p35s:dsf3 0 H:tocs 50 3 5 24% (n = 9) 76B Purple Sei 050-0382 Control 75C/D Purple Sei 050-0382 pcgp2217 prhchs:pansyf3 0 5 0 H:tnos 7 4 50% (n = 1) 76B Purple Cultivar; chrysanthemum cultivar that was transformed. DSR, Dark Splendid Reagan. See Table 2 for other abbreviations. endogenous F3 0 H levels had decreased as compared with the control (Fig. 9). Discussion In this study, chrysanthemum cultivars with suitable biochemical backgrounds for delphinidin production and a shift in color towards the blue spectrum were sourced from commercial growers. Although the constitutive CaMV 35S promoter was suitable for controlling the expression of the SuRB selectable marker gene, thereby resulting in transgenic plant production, it was the use of a promoter fragment from a floral-specific flavonoid pathway gene (rose CHS) that appeared to be important to achieve significant delphinidin production in the petals of the selected cultivars. This along with the use of specific F3 0 5 0 H sequences from pansy led to the production of close to 40% delphinidin-based anthocyanins in Improved Reagan chrysanthemum petals. A further increase in delphinidin production was achieved by also down-regulating the endogenous F3 0 H gene. Moreover, when the dsf3 0 H transgene was under the control of the floral-specific rose CHS promoter (such as in pcgp3618 and pcgp3429), a significant effect with respect to color shift and delphinidin levels was achieved, with the production of 80% of the total anthocyanidin being delphinidin. Selection of the cultivar Our HPLC analysis of the anthocyanidins and flavones that accumulate in chrysanthemum petals (Table 1) is in line with other studies showing that cyanidin is the predominant anthocyanidin and the co-pigments are flavones, with the predominant flavones being luteolin and apigenin (Saito et al. 1988, Nakayama et al. 1997, Chen et al. 2012). Ideal criteria for the development of blue hues in rose petals were to select cultivars that produced flavonols, had a relatively high vacuolar ph and that accumulate pelargonidin rather than cyanidin (i.e. less F3 0 H activity) (Katsumoto et al. 2007). In carnation, white lines that were deficient in DFR or lines that accumulated pelargonidin and so were deficient in F3 0 H activity were targeted (Tanaka and Brugliera 2013). However in chrysanthemum, the anthocyanins are derived from cyanidin whilst pelargonidin-based anthocyanins are rarely found and white lines that were only deficient in DFR were also not discovered (data not shown). Chrysanthemum cultivars were therefore selected on the color obtained when excised petals were incubated with DHM (a precursor to delphinidin). This allowed selection of cultivars that (i) contained endogenous enzymes that were able to utilize the delphinidin precursors and transport delphinidin-based anthocyanins to the vacuole and (ii) contained appropriate vacuolar environments (ph, co-pigments, etc.) that would result in a bluer color with delphinidin production. Moreover, 1704 Plant Cell Physiol. 54(10): 1696 1710 (2013) doi:10.1093/pcp/pct110! The Author 2013.
Violet blue chrysanthemums by metabolic engineering chrysanthemum cultivars were selected on the production of flavones (that can act as co-pigments to enhance bluing). From this, it was obvious that dark red, bronze, red or yellow petals were not appropriate for the development of bluer petal colors for chrysanthemum. The most appropriate cultivars tended to be those with pink petals (Fig. 3). Selection of transgene elements to increase delphinidin contents The selection of the promoter region and selection of the F3 0 5 0 H sequence were critical in chrysanthemum, as has been observed in rose (Brugliera et al. 2004, Tanaka and Brugliera 2013). Although the CaMV 35S promoter was effective at controlling the selectable marker gene (SuRB) in developing shoots and roots through the transformation process, it did not result in the same level of efficacy in petals, as seen by the results obtained with the F3 0 5 0 H transgene (under the control of the CaMV 35S promoter) of pcgp2788 (Table 2). Previous results in our laboratory have shown that the same promoter Fig. 7 Genomic DNA blot analysis of selected transgenic lines and controls of Improved Reagan (IR) and Dark Splendid Reagan (DSR). The genomic DNA has been digested with EcoRI. The cultivar and construct number (IR/2217, IR/3618 and DSR/3429) are given above the individual transgenic line numbers. The percentage of delphinidin (Del%) detected in the petals is also presented above the transgenic line numbers. (A) Hybridization with the SuRB probe to detect the SuRB transgene (see Fig. 4). The number of hybridizing bands gives an estimation of the copy number. Bands with stronger signals contain more copies of the SuRB gene. (B) Hybridization with a F3 0 5 0 H probe to detect the F3 0 5 0 H transgene (see Fig. 4). In constructs pcgp3618 and pcgp3429, the F3 0 5 0 H probe hybridizes to an EcoRI band of around 2 kb (Fig. 4). In lines 31719 (IR/3618) and 31646 (DSR/3429), further bands are detected, suggesting rearrangements/deletions/partial transfer of the T-DNA. The number of hybridizing bands in lines 26974 and 26981 (IR/pCGP2217) gives an estimation of the copy number. Bands with stronger signals contain more copies of the F3 0 5 0 H gene. Molecular weight marker sizes (in kb) are shown to the left of the autoradiographs. Fig. 8 RT PCR analysis of selected transgenic lines of Improved Reagan (IR) and Dark Splendid Reagan (DSR), to detect transcripts of the SuRB and F3 0 5 0 H transgenes. The cultivar and construct identifiers (IR/2217, IR/3618 and DSR/3429) are given above the individual transgenic line numbers. The control lines (IR and DSR) show no SuRB or F3 0 5 0 H products. Amplification of 18S was used as an internal positive control. Photographs of inflorescences are shown below. Table 4 Estimated number of SuRB transgenes integrated into selected transgenic chrysanthemum lines with novel color changes Cultivar Number Estimated SuRB transgene copy number Single Two Three or more Range Improved Regan 66 9 5 52 1 16 Dark Splendid Regan 26 1 11 14 1 12 Sei Titan 34 2 3 29 1 12 Sei Figaro 6 1 0 5 1 10 Sei Spire 3 0 1 2 2 9 Sei Titan406 5 1 0 4 1 11 Sei Florea 3 0 0 3 5 9 Sei 050-0382 4 0 0 4 4 12 Number, number of transgenic lines analysed by genomic DNA blot analysis. Plant Cell Physiol. 54(10): 1696 1710 (2013) doi:10.1093/pcp/pct110! The Author 2013. 1705
F. Brugliera et al. Fig. 9 RNA blot analysis of selected Improved Reagan (IR) and Dark Splendid Reagan (DSR) transgenic lines and controls. Hybridization with 32 P-labeled chrysanthemum F3 0 H or CHS fragments to total RNA isolated from petals of IR/3618 lines 31719 and 31722 and DSR/3429 line 31645 along with the respective controls (IR and DSR). Photographs of inflorescences are shown below, along with the percentage of delphinidin detected in hydrolyzed petal extracts of each line (Del%). The CHS probe was used as an internal positive control on RNA quality and expression of the flavonoid pathway. controlling GUS expression resulted in GUS activity and stable GUS activity in leaves and petals over many cycles of propagations (data not shown). This is contrary to a previous report showing that GUS expression driven by the CaMV 35S promoter was inactivated in transgenic chrysanthemum (Takatsu et al. 2000). However, in the cultivars that were transformed, the promoter fragment from the rose CHS gene controlling the expression of the F3 0 5 0 H transgene resulted in the highest levels of delphinidin accumulation especially when coupled with the hprnai F3 0 H cassette directed at down-regulating the endogenous F3 0 H under the control of the same rose CHS promoter fragment. The primary transgenic plants with novel violet/blue petal colors were maintained under greenhouse conditions for 2 3 years with stable flower color, and secondary propagated plants produced inflorescences of the same color as the primary transgenic event (data not shown). These results suggest that the rose CHS promoter appears to allow for stable expression in chrysanthemum, at least in controlled greenhouse conditions. This study highlights and reconfirms, as was the case with the production of delphinidin in rose (Brugliera et al. 2004, Katsumoto et al. 2007), that transgenes can perform differently in different species and in some cases different cultivars or perhaps depending on the transformation protocol. Efficient delphinidin production in chrysanthemum using a campanula F3 0 5 0 H gene under the control of the chrysanthemum flavanone 3-hydroxylase promoter attached to a tobacco alcohol dehydrogenase translation enhancer in a few chrysanthemum cultivars (Noda et al. 2013) has not been successfully repeated in Improved Reagan in our lab (data not shown). Noda et al. used a chrysanthemum variety of the decorative background, whereas our analysis has predominantly been in a daisy-type chrysanthemum cultivar. We targeted promoter fragments of flavonoid pathway genes as these promoters would be more likely to have the temporal and spatial specificities required for anthocyanin manipulation. In carnation, the ideal F3 0 5 0 H gene expression cassettes were F3 0 5 0 H cdna sequences (from pansy, petunia or salvia) under the control of a promoter fragment from either the snapdragon CHS gene or the carnation ANS gene (Tanaka and Brugliera 2013). However, the same F3 0 5 0 H transgenes did not result in delphinidin production in rose (Brugliera et al. 2004). In rose, the most effective F3 0 5 0 H transgene was the pansy F3 0 5 0 H cdna sequence under the control of the CaMV 35S promoter fragment. The petunia F3 0 5 0 H cdna sequence that had been effective in petunia and carnation resulted in a degraded transcript in rose petals (Brugliera et al. 2004). Similarly, the snapdragon CHS and carnation ANS promoters were ineffective at driving high levels of gene expression in rose (Brugliera et al. 2004). In the case of chrysanthemum, it appears that of the transgenes assessed to date, the use of a rose CHS promoter linked with the pansy F3 0 5 0 H cdna sequence resulted in F3 0 5 0 H activity that altered the anthocyanin biosynthetic pathway most efficiently and, conversely, the same CaMV35S:F3 0 5 0 H transgene so effective in rose petals was not as effective in chrysanthemum petals (Table 2). The flavonoid pathway has been described as acting as a multichannel grid where many of the enzymes can act on different groups of flavonoids (Fig. 1) (Davies and Schwinn 2006). This leads to the competition between enzymes for substrates. For diversion of the anthocyanin pathway to the production of delphinidin, a F3 0 5 0 H activity is required. However, the introduced F3 0 5 0 H enzyme needs to out-compete effectively the endogenous enzymes such as F3 0 H and DFR for substrate. In the case of carnation, a white carnation deficient in both F3 0 H and DFR was a suitable cultivar for the production of 100% delphinidin upon the introduction of a pansy or petunia F3 0 5 0 H and a petunia DFR transgene (Tanaka and Brugliera 2013). The fact that petunia DFR was not able to utilize DHK was used to advantage in the production of novel colored carnations. In the case of rose, appropriate DFR mutants were not discovered and so roses that were deficient in F3 0 H were utilized. The endogenous rose DFR was silenced using the hprnai strategy, and an iris DFR (with similar properties to the petunia DFR above) was introduced (Katsumoto et al. 2007). In chrysanthemum, appropriate F3 0 H and/or DFR mutants were not discovered. Instead the endogenous F3 0 H was down-regulated using hprnai-mediated silencing. A series of three constructs (pcgp3424, pcgp3429 and pcgp3618) were tested in our model cultivar (Improved Regan), with the most effective being pcgp3618 containing the dsf3 0 H partial coding sequences of the chrysanthemum F3 0 H in sense and antisense directions with an intervening sequence from intron 1 of the cineraria F3 0 5 0 H gene under the control of the rose CHS promoter. Moreover, in lines containing the dsf3 0 H transgenes, pelargonidin was detected, confirming that the chrysanthemum DFR was able to utilize DHK to produce pelargonidin (Figs. 1, 6), as was suggested by in vitro P450 inhibition studies 1706 Plant Cell Physiol. 54(10): 1696 1710 (2013) doi:10.1093/pcp/pct110! The Author 2013.
Violet blue chrysanthemums by metabolic engineering (Schwinn et al. 1994). However, it appears that the introduced pansy F3 0 5 0 H was able to compete effectively with the endogenous DFR for substrate. With only a dsf3 0 H transgene, higher levels of pelargonidin have been produced in our cultivars (unpublished data). The results obtained here confirm that speciesspecific and possibly cultivar-specific tactics are required to achieve high delphinidin content and novel color changes. Southern analysis of the transgenic lines with significant color changes has revealed that most if not all contain multiple copies of the T-DNA. An expectation may be that this would lead to possible methylation and silencing of the transgenes. This has not been observed in the transgenic plants that have been studied to date; however, a detailed analysis was not performed as done in Arabidopsis transgenic lines (Schubert et al. 2004). We only analyzed transgenic lines that had significant shifts to blue in petal color and that would be of potential commercial value. Since extensive molecular analysis is often required to obtain commercialization of genetically modified plants, fewer copies of the T-DNA insert are desirable. This is also true when incorporating a transgenic plant into hybridization breeding. Moreover, regulatory requirements in some jurisdictions restrict the release of transgenic lines if vector DNA containing the antibiotic resistance genes used for selection in bacteria is incorporated into the genome (Chandler and Tanaka 2007). Our methods led to a relatively high incidence of multiple copies of T-DNA integrations. A subset of these transgenic plants (30%) contained vector backbone integrations covering the tetracycline resistance gene (located outside the LB of the T-DNA). In Agrobacterium-mediated transformation, transfer of the T-DNA region occurs from the right to the left T-DNA borders (Wang et al. 1984). Therefore, the present result confirms that incomplete termination occurred at the LB T-DNA junction. Development of a transformation protocol resulting in fewer copies of transgenes and correct transfer of the T-DNA between the left and right borders would therefore be useful, possibly by reducing co-cultivation times, changing T-DNA border sequences, changing binary vector backbones and changing A. tumefaciens strains. However, these changes may also impact the transformation efficiency. Currently, selecting chrysanthemum lines with single-copy integrations from a large number of transgenic plants produced is practical as the transformation efficiency is around 5% (for Improved Reagan). However, it would not be feasible if the transformation efficiency was 0.2% as in the case for Sei 050-0382. Future challenges The development of chrysanthemums with novel violet hues has been an exciting step in development of blue-colored chrysanthemums. However, in order to achieve bluer color in transgenic chrysanthemum, strategies based on the compositions of native blue flowers (Yoshida et al. 2009) will need to be further explored. These include modification of delphinidin-based anthocyanins with aromatic acyl groups, elevation to neutral vacuolar ph, accumulation of flavone C-glucoside and incorporation of metal ions. Another challenge will be the commercialization of transgenic chrysanthemums in some jurisdictions. Having large amounts of viable pollen and being easy to hybridize to wild chrysanthemums, obtaining commercialization approval for the color-modified chrysanthemum will be more difficult than for roses and carnations, at least in Japan, where the chrysanthemum holds special cultural significance. Materials and Methods Plant materials and plant tissue culture Chrysanthemum varieties were obtained as tissue-cultured plantlets from Clean Grow (Australia) or Seikoen (Japan). Flavonoid analysis and precursor feeding experiments Anthocyanidins and flavonoids of petals of >75 chrysanthemum cultivars and their transgenic plants were analyzed as described previously (Katsumoto et al. 2007). Eight cultivars were selected for transformation. The cultivars chosen included Improved Reagan and Dark Splendid Reagan (Chrysanthemum Breeder s Association, The Netherlands) and Seikoen cultivars including Sei Titan, Sei Titan 406, Sei 050-0382, Sei Spire, Sei Florea and Sei Figaro (Seikoen, Japan). Petals from early stage flower buds were dissected and placed in water or 2 mg ml 1 DHM (The New Zealand Institute for Plant & Food Research) and incubated for 40 h under lights. Petals were then examined under a stereomicroscope and color coded using the Royal Horticultural Society Color Charts (RHSCCs). The pigmented petals were subjected to HPLC analysis. Binary plasmid construction A summary of the construction of four binary vectors in Fig. 4 is given below. The transformation vector pcgp2217 contains the prhchs:pansyf3 0 5 0 H:tnos expression cassette in a tandem orientation with respect to the 35S:SuRB selectable marker cassette of the Ti plasmid pcgp1988 [based upon pwtt2132 (DNAP)] with the multicloning site from pneb193 (New England Biolabs) (Brugliera et al. 2003). prhchs is an 2.8 kb fragment containing the promoter region from a CHS gene of Rosa hybrida cv. Kardinal (DNA data accession No. FW556946), pansy F3 0 5 0 H is a 1.8 kb F3 0 5 0 H cdna clone #18 from Viola wittroxkiana cv. Black Pansy (FW55690), tnos is the terminator region from the nopaline synthase (nos) gene of A. tumefaciens (Depicker et al. 1982), SuRB is the chlorsulfuron resistance gene (encodes acetolactate synthase) with its own terminator from Nicotiana tabacum (Lee et al. 1988). The transformation vector pcgp3424 contains the prhchs: pansyf3 0 5 0 H:tnos expression cassette in a tandem orientation with respect to the 35S:SuRB selectable marker cassette of the transformation vector pcgp2217 (described above) along with a CaMV 35S:dsF3 0 H tocs cassette directed at downregulating expression of the endogenous chrysanthemum Plant Cell Physiol. 54(10): 1696 1710 (2013) doi:10.1093/pcp/pct110! The Author 2013. 1707
F. Brugliera et al. F3 0 H, where dsf3 0 H is the ds chrysanthemum F3 0 H fragment harboring an 1.0 kb sense partial chrysanthemum F3 0 H cdna fragment (X79723), a 180 bp petunia DFR-A intron 1 fragment (Beld et al. 1989) and an 1.1 kb antisense partial chrysanthemum F3 0 H fragment (HV782742) with the aim of formation of a double-stranded (hairpin loop) RNA molecule to induce silencing of the endogenous chrysanthemum F3 0 H through hprnai, and tocs is the terminator region from the octopine synthase gene of A. tumefaciens. The transformation vector pcgp3429 contains a prhchs:pansyf3 0 5 0 H:tnos expression cassette in tandem orientation with respect to the 35S:SuRB selectable marker cassette of pcgp2217 along with a prhchs:dsf3 0 H*:tnos cassette, where dsf3 0 H* is ds chrysanthemum F3 0 H harbouring a 1.1 kb sense partial chrysanthemum F3 0 H cdna fragment and a 0.8 kb antisense partial chrysanthemum F3 0 H fragment. The transformation vector pcgp3618 contains a prhchs:pansyf3 0 5 0 H:tnos expression cassette in a tandem orientation with respect to the 35S:SuRB selectable marker cassette of pcgp2217 along with a prhchs:dsf3 0 H**:tnos cassette, where the dsf3 0 H** fragment is similar to dsf3 0 H described above except that the petunia DFR intron is replaced by a 570 bp fragment harboring intron 1 of the genomic F3 0 5 0 H clone from cineraria (FW570902). The transformation vector pcgp2205 contains a chimeric pdcans pansyf3 0 5 0 H:tDcANS gene cassette in a tandem orientation with respect to the 35S:SuRB selectable marker gene cassette of the Ti plasmid pcgp1988. pdcans incorporates around 2.3 kb of the promoter region of the anthocyanidin synthase gene (GZ637195) from Dianthus caryophyllus cv. Laguna. and tdcans is a 0.8 kb terminator fragment from the same ANS gene (GZ637196). The transformation vector pcgp2788 contains the CaMV35S:pansyF3 0 5 0 H:t35S expression cassette in a tandem orientation with respect to the 35S:SuRB selectable marker cassette of the Ti plasmid pcgp1988. Construction of pcgp2788 has been described (Brugliera et al. 2003). The transformation vector pcgp3141 contains a 4.6 kb fragment containing the Cineraria F3 0 5 0 H gene (FW570902) in a tandem orientation with respect to the 35S:SuRB selectable marker cassette of the Ti plasmid pcgp1988. Transformation of chrysanthemum and assessment of flower color Chrysanthemum cultivars Improved Reagan and Dark Splendid Reagan were sourced as in vitro plantlets from CleanGrow, Australia. Sei cultivars were kindly provided by Seikoen, Japan. In vitro plantlets were established and grown to a height of 5 7 cm and maintained in subdued light to produce longer internodes. Stem segments of 1 2 mm or petiole segments of 2 4 mm, or, in some cases, leaf segments were used as explants. Explants were immersed in the Agrobacterium inoculum immediately after dissection for 30 40 min and then blotted thoroughly on sterile filter paper. The inoculated explants were then placed onto a sterile filter disc placed on top of co-cultivation medium [Murashige and Skoog mineral salts and vitamins (MS) (Murashige and Skoog 1962)] supplemented with 0.5 mg l 1 BA (6-benzylaminopurine), 0.2 mg l 1 NAA (napthaleneacetic acid), 200 mm acetosyringone and 0.3% Gelrite TM (Merck), and incubated in a growth room in the dark for 2 4 d at a temperature of about 20 C. Explants were then transferred to Petri dishes containing post co-cultivation medium (MS supplemented with 0.5 mg l 1 BA, 0.2 mg l 1 NAA, 300 mg l 1 timentin and 0.3% Gelrite TM ), and incubated in a growth room for 2 4 d under subdued light. Explants were then transferred to regeneration and selection medium [MS supplemented with 0.5 mg l 1 BA, 0.2 mg l 1 NAA, 300 mg l 1 timentin, 1 mgl 1 Glean Õ (chlorsulfuron) (DuPont) and 0.3% Gelrite] and incubated for 10 14 d, after which they were transferred to a selection medium with a higher concentration of selecting agent (MS containing 0.5 mg l 1 BA, 0.2 mg l 1 NAA, 300 mg l 1 timentin, 3 mgl 1 Glean Õ and 0.3% Gelrite) and further incubated for 10 14 d in a growth room under lights. Welldeveloped shoots were excised and transferred to ACA5G plates [MS supplemented with 5 mgl 1 Glean Õ and 0.5% Agargel TM (Sigma-Aldrich)] and incubated for a further 14 30 d, subculturing as necessary to remove necrotic tissue and small side shoots which obstruct contact with the medium. Shoots that developed roots in the 5 mgl 1 Glean Õ -containing media were selected and transferred to glass jars containing fresh ACA5G medium. Shoots grown to around 4 cm in height with a vigorous root growth were transferred to soil and grown under greenhouse conditions until flowering. Using this transformation method, the time from co-cultivation to soil was between 8 and 12 weeks, and the transformation efficiency varies between cultivars and explant source from 0.1% to 20%. Typically a transformation efficiency of 5% was obtained with Improved Reagan. The disarmed A. tumefaciens strain AGL0 was used for transformations (Lazo et al. 1991). The cultivar Improved Reagan was used as a model cultivar to test the effectiveness of the transgenes. Around 25 50 individual transgenic events were generated for each construct and grown to flowering in enclosed greenhouses under controlled environments. Inflorescences were examined, and colors of the ray or ligulate petals were described using RHSCCs. Plasmid DNA was introduced into A. tumefaciens strain AGL0 (Lazo et al. 1991) essentially as described (Brugliera et al. 1994). Cells of A. tumefaciens carrying the plasmid were selected on Luria Bertani medium agar plates containing 50 mgml 1 tetracycline. The presence of the binary vector plasmid in A. tumefaciens was confirmed by restriction endonuclease mapping of DNA isolated from the antibiotic-resistant transformants. Stages of flower development Chrysanthemum inflorescences were harvested at developmental stages defined as follows: stage 1, closed bud, petals not visible; stage 2, inflorescence buds opening, tips of petals visible; stage 3, tips of nearly all petals exposed; stage 4, outer 1708 Plant Cell Physiol. 54(10): 1696 1710 (2013) doi:10.1093/pcp/pct110! The Author 2013.