Protoplasma (2013) 250:1273–1281 DOI 10.1007/s00709-013-0509-8

ORIGINAL ARTICLE

Anatomical and biochemical studies of bicolored development in latifolium

Yinyan Qi & Qian Lou & Huibo Li & Juan Yue & Yali Liu & Yuejin Wang

Received: 1 March 2013 /Accepted: 6 May 2013 /Published online: 16 May 2013 # Springer-Verlag Wien 2013

Abstract The of the broad-leafed grape hya- respectively, with the difference being nonsignificant. cinth, Muscari latifolium, shows an interesting, two-tone HPLC results indicate that the dihydroflavonol and flavonol appearance with the upper being pale blue and the contents are also very similar in the two sorts of flower. lower ones purple. To elucidate the mechanism of the dif- However, the upper flowers contained only delphinidin, ferential color development, anatomical research was car- whereas the lower flowers also contained cyanidin. The total ried out and a cytological study of the colored protoplasts in anthocyanin content in the lower flowers was 4.36 mg g−1, which the shapes of the cells accumulating anthocyanin which is approximately seven times higher than in the upper were observed by scanning electron microscopy. Next, vac- flowers, while the delphinidin content is four times higher. uolar pH was recorded using a pH meter with a micro Quantitative real-time PCR analysis established that the combination pH electrode, and the sap’s metal-ion content two-tone flower was a result of different expressions of the was measured by inductively coupled plasma mass spec- F3′5′H, F3′H and DFR genes, and these lead to different trometry. The anthocyanin and co-pigment composition was amounts of anthocyanin. determined by high-performance liquid chromatography (HPLC). Chemical analyses reveal that the difference in Keywords Anthocyanin . Color development . Metal ions . metal-ion content of the two parts was not great. The vacu- Grape hyacinth . Protoplast olar pHs of the upper and lower flowers were 5.91 and 5.84, Abbreviations RHSCC Royal Horticultural Society color chart Handling Editor: Hanns H. Kassemeyer CIE International Commission on Illumination Yinyan Qi and Qian Lou contributed equally to this work. Y. Qi : Q. Lou : Y. Wang College of Horticulture, Northwest A&F University, Yangling, Introduction Xianyang 712100 Shaanxi, People’s Republic of China : : H. Li J. Yue Y. Liu The Muscari, commonly known as grape hyacinths, is a College of Forestry, Northwest A&F University, Yangling, genus of ornamental bulbous named after their spikes ’ Xianyang 712100 Shaanxi, People s Republic of China of dense, urn-shaped flowers resembling bunches of grapes. Y. Qi : Q. Lou : H. Li : J. Yue : Y. Liu : Y. Wang They are excellent as garden, bedding, and ground-cover Key Laboratory of Biology and Genetic Improvement of plants because of their profuse flowering, adaptability, at- Horticultural Crops(Northwest Region), Ministry of Agriculture, tractive fragrance, and distinctive blue colors in spring. In Yangling, addition, they have important cultural significances as they Xianyang 712100, People’s Republic of China are referred to in many of the world’s literatures. There are : : : : : Y. Qi Q. Lou H. Li J. Yue Y. Liu (*) Y. Wang (*) about 40 different species of Muscari, and their flower color State Key Laboratory of Crop Stress Biology in Arid Areas, is most commonly a shade of blue (Doussi et al. 2002). Northwest A&F University, Yangling, Muscari latifolium is distinguished from the other grape 712100, Xianyang, Shaanxi, People’s Republic of China e-mail: [email protected] hyacinth species by the production from each of a e-mail: [email protected] single, bicolored, racemose inflorescence. The inflorescence 1274 Y. Qi et al. also shows an unusual two-toned appearance being purple to delphinidin, while the reddish or pale purple colors are over most of its slender length, but with a soft pale blue attributable to a mixture of cyanidin and delphinidin. “cap” (Fig. 1). To determine more precisely the causes of bicolor devel- In nature, a number of species can exhibit more than one opment in M. latifolium , we analyzed tepal color in a flower, but these are rare. Bicolored inflorescences structures, protoplast features, and the anthocyanin and fla- attract considerable interest among researchers and breeders, vonol compositions of the flower perianths in relation to and much research has been carried into this, but the mech- their different colors. We also investigated certain physico- anisms involved have not been fully elucidated because of chemical properties of the flowers which could have been their complexity and also the scarcity of multicolored associated with color development, such as pH and metal- flowers. In some species, different anthocyanin chromo- ion content. Through this analysis, we hoped to determine phores are the primary contributors to the distinctively color the mechanism of color development in their distinctive phenotypes. These occur in Anagallis monelli (Quintana et upper and lower flowers. al. 2007), Delphinium (Hashimoto et al. 2002), and Hydrangea macrophylla cv. Hovariatrade mark “Homigo” (Yoshida et al. 2008). In the case of some tulips, Tulipa Materials and methods gesneriana (Momonoi et al. 2009, 2012; Shoji et al. 2007, 2010) and H. macrophylla (Ito et al. 2009; Schreiber et al. material 2010, 2011), metal ions are required for specific color development. In Ipomoea tricolor, the color variation in Grape hyacinth (M. latifolium) were purchased from different regions of the same tepal is linked to vacuolar pH Zhejiang Hongyue Seeds Co., Ltd. (Zhejiang, People’s differences (Yoshida et al. 2005, 2009a, b). Thus, it is Republic of China) and planted in an experimental plot at reasonable to infer that the two-tone effect observed in the the Northwest A&F University, Xi’an, Shaanxi, People’s differently colored flowers of the M. latifolium is the Republic of China. The flowers were harvested from the result of certain biochemical or physiological differences spikes of each cultivar just after anthesis and were used but, at this stage, the precise mechanisms involved are immediately for the morphological and anatomy observa- uncertain. Comprehensive studies of color formation in the tions and the preparation of protoplasts. To measure antho- Muscari, or other flowers of the same family, have not yet cyanin, flavonol, and pH, the flowers were cut in half been reported to our knowledge. At this stage, there are a lengthways using a razor blade. The tepals were then care- few reports of anthocyanin composition in 13 genotypes of fully collected from straight-cut flowers and treated with a Muscari ssp. Mori et al. (2002) summarize these, stating that small puffer to blow off any dust or pollen adhering to the the varying shades found in the blue flowers are attributable epidermis. Next, the tepals were quickly frozen in liquid

Fig. 1 Phenotypes of M. latifolium. Upper flowers (a) with pale blue tepals and lower flowers (b) with purple tepals. Bar=2mm Studies of bicolored flower development in M. latifolium 1275 nitrogen and stored at −80°C pending flavonoid and pH 80 mM KCl, and 20 mM MES-Tris (pH 6.0). The solution measurement. Additional tepals were collected and dried was warmed to 55 °C for 10 min to enhance enzyme for metal-ion analysis. These were rinsed in sterile water, solubility, then cooled to room temperature, and 10 mM before being placed in a drying oven at 105 °C for 30 min, CaCl2 and 0.1 % BSA were added. The final enzyme solu- followed by one at 80 °C for 5 h. tion was filtered through a 0.45-μm syringe filter device into a Petri dish. Fresh tepals (0.5 g) were taken from a well- Measurement of tepal color opened flower at 9:00 am. Tepal strips (0.5–1 mm wide) were cut from the middle part of a tepal using a sharp razor Tepal color was determined in the mid portion of the urn- blade. They were transferred quickly and gently and im- shaped flowers at noon, indoors and in a north light (i.e., no mersed in enzyme solution. After vacuum infiltration for direct sunlight) using the Royal Horticultural Society Color 30 min without shaking, digestion was continued in the dark Chart (RHSCC). About 60–100 fully pigmented flowers for 3 h at room temperature. The reaction mixture were randomly sampled and placed in a single layer on a containing free protoplasts was filtered through a 75-μm six-well culture plate (Corning Costar, Sigma-Aldrich, nylonmeshandwashedwithbuffer[20mMMES-Tris USA). Their color parameters were measured with a chroma (pH 6.0), 0.8 M mannitol], then centrifuged at 100×g for meter (CR-400, Konica Minolta Investment Ltd., China) 3 min to form a pellet of protoplasts in a 30-ml round- based on the CIE (International Commission on bottomed tube. As much supernatant liquid as possible Illumination) L∗a∗b∗ scale (Hanbury and Serra 2002; was removed, and the protoplast pellet was re-suspended Zhang et al. 2008). Five replicates were recorded. in 2 ml of fresh buffer by gentle swirling. The centrifugal separation/re-suspension procedure was repeated three Microscopic observation of epidermal cells and transverse times, and the released protoplasts were checked under the sections microscope. All the procedures after the filtration were conducted below 4 °C. Fresh urn-shaped flowers were cross-sectioned, and an epi- dermal layer was peeled off by hand using a razor blade and Measurement of metal elements in colored tepals placed on a glass slide with a drop of water. The fresh sections were immediately observed using a light micro- Dried tepals were finely ground to a powder. Next, a 1-g scope (Eclipse 50i, Nikon, Japan), equipped with a DS sample of dried material was digested in 5 ml of concen- cooled camera head with FNIS-Elements image processing trated HNO3 and 1 ml of H2O2, followed by treatment in a software. high-performance microwave digestion unit (CEM, Mars, USA). Settings used were: time (minutes)/power Scanning electron microscopy (watts)/temperature (°C): 5/1,200/120, 10/1,200/160, and 20/1,200/180. After complete digestion and acid removal, Mid portions of urn-shaped flowers were fixed in glutaral- the samples were diluted with double-distilled water for dehyde buffer (4 % glutaraldehyde in 0.2 M phosphate measurement. Sample solutions were analyzed for elements buffer, pH 6.8) under vacuum (30 min) and then incubated by ICP-MS (Varian, 820-MS, USA). The parameters for for 6 h at 4 °C. Next, the fixed tissues were dehydrated in an analysis were plasma power 1,400 W, plasma flow ascending aqueous ethyl alcohol series (30, 50, 70, 80, 90, 18 l/min, auxiliary flow 1.8 l/min, and sampling depth was and 100 %) followed by 100 % acetone and two, 30-min 7.5 mm. treatments with isopentyl acetate. Lastly, the solvents in the samples were replaced with liquid carbon dioxide by a Measurement of cell sap pH critical-point drying method (Boyde et al. 1969). The dry tepals were mounted on a specimen stub and sputter-coated Fresh pale blue or purple perianths of small flowers taken with gold before examination in the scanning electron mi- just after anthesis were carefully collected within a 20-min croscope (JSM-6360LV SEM, JEOL Ltd., Japan). period. They were immediately frozen in liquid nitrogen and stored at −80 °C pending pH measurement. After grinding Preparation of free protoplasts from colored tepals in liquid nitrogen with a pestle and mortar, the tepals of about 25–40 flowers (0.5 g) were centrifuged at 18,407×g The preparation of free protoplasts was done using a method for 5 min. The supernatant was immediately transferred to a slightly modified from that of Yoshida et al. (2003a, b). An new, 1.5-ml reaction tube, and the pH of the solution was enzyme solution (10 ml) was prepared containing 1.3 % measured at 25 °C using a Mettler Toledo FE20 FiveEasy (w/v) cellulase R-10 (Yakult, Japan), 0.3 % (w/v) pH meter with a micro combination pH electrode (InLab® macerozyme R-10 (Yakult, Japan), 0.8 M mannitol, Ultra-Micro, Mettler Toledo Ltd., Switzerland). For each 1276 Y. Qi et al. flower color, the average cell sap pH was calculated from 12 Results independent, replicate, extractions carried out on different days. Two-tone flowers

Analysis of anthocyanidins According to the RHSCC, the colors of the upper flowers were pale blue and the lower ones were purple (Table 1). The tepals of each flower color were sampled for flavonoid Tepal color was measured with a chroma meter and recorded analysis using high-performance liquid chromatography as a three-dimensional CIE L∗a∗b∗ value (Hanbury and (HPLC). The methods of extraction were as previously Serra 2002; Zhang et al. 2008). In these, the parameter L∗ described (Albert et al. 2009) with some modifications. describes lightness of color, ranging from black (L∗=0) to Flavonoids were extracted from 50 mg dry weight (DW) white (L∗=100). A low L∗ value for the purple flowers of freeze-dried flowers from each flower color in 2 ml of means a darker color. The parameter a∗ represents the bal- 1 % MeOH/HCL for 72 h at 4 °C. The supernatant was ance between red and green, and the parameter b∗ represents removed and the pellet re-extracted overnight in 2 ml 1 % the balance between yellow and blue. As expected, in both MeOH/HCl (v/v) at 4 °C. The extraction was repeated once. pale blue and purple flowers, parameter a∗ takes a positive A 10-μl sample of each combined supernatant was quanti- value for reddish colors, and parameter b∗ takes a negative fied by HPLC (Hitachi L-2400, Japan) where flow rate was value for the bluish ones. Chroma parameter C∗ represents 0.5 ml/min, column temperature was 40 °C, and detection the vividness of the color; hence the lower value of param- was by diode array detector (Hitachi L-2455, Japan). Using eter C∗ for purple flowers means a grayer color. The hue a 4.6×250-mm column of C18 and a linear gradient (95– angle (h=arctan(b∗/a∗)) represents the basic color (Biolley 0 %) of Solvent A [0.04 % formic acid/water (v/v)] in and Jay 1993). A red color has h around 0°(360°), yellow is Solvent B (acetonitrile) for 40 min. Next with 100 % described by h around 90°, green has h around 180°(−180°), Solvent B for 20 min. Concentrations were expressed in while blue colors has h values around 270°(−90°). By using milligrams per gram dry weight. the hue angles, it is easy to define the color (Torskangerpoll et al. 2005). Total RNA isolation and quantitative real-time PCR analysis Spatial location of pigment Total RNAwas extracted from upper and lower flowers. RNA (1 μg) was used for first-strand cDNA synthesis with the To study the effects of tissue structure on bicolored flower PrimeScript RT Reagent Kit (TaKaRa). The cDNA was dilut- development, the spatial location of the pigment within the ed approximately 12-fold for quantitative real-time PCR tissues of both upper and lower flowers was examined. The (qRT-PCR). The qRT-PCR was run on an IQ 5 real-time lower flowers are slightly bigger than the upper flowers PCR Cycler (Bio-Rad Laboratories, USA) with SYBR (Fig. 2a, b). As shown in Fig. 2c, d, the cross section of a Green I dye. Reactions used the following profile: 3 min at tepal of M. latifolium shows a typical structure of an upper 95 °C, 45 cycles of 5 s at 95 °C, 30 s at 55 °C, and data epidermis, palisade mesophyll, spongy mesophyll, and a acquisition at 55 °C for 15 s with 81 cycles. Threshold values lower epidermis. In the pale blue flowers, the blue cells (CT) were generated from the IQ 5 software tool (Bio-Rad are located only in the palisade mesophyll—not in the Laboratories). The relative mRNA ratios were calculated spongy mesophyll or in the upper or lower epidermes. The according to the method of Livak and Schmittgen (2001). degree of pigmentation varied considerably between indi- Actin was used as the internal control to normalize gene vidual palisade cells. In contrast, the location of pigment in expression. Mean values and SDs were obtained from two the purple flowers was particularly interesting. Different technical and three biological replicates. PCR amplification shades of amaranth (a reddish rose color) and blue cells was performed with the primers listed in Table 4. are located mainly in the lower epidermis and the sub-

Table 1 Tepal colors and color parameters of M. latifolium

Type of flower Tepal color RHSCC CIE L∗a∗b∗ color coordinates

L∗ a∗ b∗ C∗ h

Upper flowers Violet-blue 93B 39.746±0.998 13.178±0.554 −24.400±1.014 28.318±1.057 297.724±0.183 Lower flowers Purple 79A 33.596±0.465 4.286±0.081 −4.200±0.131 6.03±0.116 315.818±1.657

RHSCC Royal Horticultural Society color chart; L∗ lightness; a∗ , b∗ chromatic components; C∗ brightness; h (hue angle)=arctan (b∗ /a∗ ) Studies of bicolored flower development in M. latifolium 1277

Fig. 2 Cellular features of upper and lower flowers of M. latifolium. Single flower from upper (a) and lower part (b)of the same raceme. Bar=5mm. Cross sections of upper (c) and lower tepals (d). Bar=100μm. Epidermal cells of upper (e) and lower tepals (f). Bar=100μm. Scanning electron micrograph (SEM) of papillate cells from the outer epidermis of tepals of upper (g) and lower (h). Bar= 20μm. Protoplasts from upper (i) and lower tepals (j). Bar= 30μm 1278 Y. Qi et al.

Table 2 Concentration of metal elements in tepals of M. latifolium

Type of flower Concentration of metal elements (mg·kg−1)

Mg Al Ca Mn Fe Cu Zn Cd

Upper flowers 1,742.30±51.71 72.30±5.76 3,977.44±78.87 32.03±2.15 440.01±28.19 10.38±0.51 42.36±0.35 0.073±0.007 Lower flowers 1,856.52±63.87 51.49±1.88 2,291.90±52.32 22.91±0.52 120.50±6.02 12.62±1.12 34.33±2.15 0.242±0.014

The concentration of metal elements in the tepals was determined by the ICP method epidermal layer of the outer (abaxial) face of the tepal. preparing protoplasts from pale blue and purple tepals. As In addition, sporadic colored cells also appear in the shown in Fig. 2i, a large blue central vacuole and some inner (adaxial) epidermis and spongy mesophyll which chloroplasts are located in the protoplast of the upper may contribute to the much deeper color of the lower flowers. However, the purple color of the lower flowers flowers. did not arise from a single kind of protoplast but from a As we have seen, the colored cells in the upper mixture of amaranth and blue protoplasts (Fig. 2j). This fits flowers are located mainly in the palisade layer, while with the observations from transverse sections. the epidermal cells are colorless or almost colorless (Fig. 2e). In the lower, purple flowers the colored cells Element analysis of two-tone flowers are of a variety of colors, ranging from amaranth to purple to blue (Fig. 2f). Therefore, this indicates that It is generally understood that one of the three major mech- the bluing and paling effect in the color of the upper anisms involved in the formation of bluish colors in flowers flowers of the raceme (compared with the lower is the formation of metal complexes (Yoshida et al. 2009a, flowers) is caused at least in part by differences in the b). Using inductively coupled plasma mass spectrometry, spatial location of the colored cells within the tissues. we recorded the eight metal elements (Al, Ca, Cd, Cu, Fe, As we know (Noda et al. 1994), the shape of the cells Mg, Mn, and Zn) considered to contribute to blue color accumulating anthocyanin pigments influences their optical development in bicolor flowers (Yoshida et al. 2009a, b). properties and thereby affects our sensation of color. To The contents of these elements are listed in Table 2. The analyze interactions between color and surface geometry, results show that Ca, Mg, and Fe are the three most abun- the epidermal cell shapes of pale blue and purple tepals were dant elements in both pale blue and purple tepals. Although examined by scanning electron microscopy. The cell shape the Fe content was three times higher in the upper than the of the two parts was similar. All epidermal cells are incon- lower flowers, the Cd content differences were the other spicuous papillate, which confers the properties of low light way around. The remaining elements, Mg, Zn, Cu, Al, Mn, absorption and a velvety sheen (Fig. 2g, h). This suggests and Ca, did not show significant differences between upper that the flower-color difference in the inflorescences of M. and lower flowers. latifolium is not associated with epidermal cell shape. Measurement of sap pH Colored protoplasts of two-tone flowers To examine the effect of pH on upper and lower flower To clarify the mechanism for the unique two-toned flowers, color, the sap pH of pale blue and purple tepals was mea- the cell features according to flower color were analyzed by sured. The mean pH value of purple tepals was 5.84±0.02

Table 3 Concentration of flavonoid and sap pH value of M. latifolium tepals

Type of flower pH Anthocyanidin (mg g−1) Dihydroflavonol (mg g−1) Flavonol (mg g−1)

Del Cya Pet Pel Mal DHM DHQ My Qu Km

Upper flowers 5.91±0.05 0.613±0.103 n.d. n.d. n.d. n.d. 0.031±0.010 n.d. 0.149±0.127 n.d. 0.471±0.253 Lower flowers 5.84±0.02 2.449±0.174 1.906±0.168 n.d. n.d. n.d. 0.043±0.025 n.d. 0.116±0.120 n.d. 0.440±0.193

Del delphinidin, Cya cyanidin, Pet petunidin, Pel pelargonidin, Mal malvidin, DHM dihydromyricetin, DHQ dihydroquercetin, My Myricetin, Qu Quercetin, Km Kaempferol, n.d. not detected; milligrams per gram of dry tepals Studies of bicolored flower development in M. latifolium 1279

Table 4 Primers used for q RT-PCR analysis

Target gene Forward primer sequence (5′ to 3′) Reverse primer sequence (5′ to 3′)

CHS AGCATGAGGCCCGACTCGCGTAGCT CCCTGCAAGGACCACATCCAGAACC CHI GGGGACTGGGTGAAGAGGATGGATGAG TGACCGGCCAGCAGTACAGCGAGAAGG F3′5′H TAAGCCGGATGGTGCTGGGCAAGAAG AGCCACGGGATGGAGTCGCCGATGTT F3′H GGCCGAGCACATCGCCTACAACTACCA CCCTTGCGAACGTGCATGAAGCGTAT DFR1 TTGTCTCGGAAGGCTGGATGTAGGT TTGACTAGCCAGGATGCGAAGAAGC DFR2 CGCCTCATCAAACATCTCCTCCATT GCAAGTATCCTCAGTATCACATCCCACA ANS GAGCTGCTGATGCAGATGAAGA CGTTGTGGAGGATGAAGGAGA Actin GCCTGCCATGTATGTTGCG CGGAGCTTCCATTCCGATC

(n=12) and that of pale blue tepals was 5.91±0.05 (n=12; flowers showed a red color and contained delphinidin Table 3). This does not represent a significant difference (56 %) and cyanidin (44 %) (Table 3). The total anthocyanin between the pH values of the two cells. contentinthelowerflowerswas4.36mgg−1,whichis approximately seven times higher than that in the upper Anthocyanidin analysis flowers (0.613 mg g−1), and that of delphinidin was four times higher in the lower flowers than in the upper ones. The two Flavonoids, including anthocyanins, are responsible for a flower types contained similar amounts of both wide range of flower and fruit colors including pale yellow, dihydroflavonol and flavonol (Table 3). scarlet, red, magenta, violet, and blue (Tanaka et al. 2009, 2008). To determine the composition of anthocyanins and co- Expression analysis of flavonoid biosynthetic genes pigments in the two-tone flowers, the flavonoids were in the upper and lower flowers extracted and analyzed by HPLC. Upper flowers had a mauve extract and contained only delphinidin, while the lower To investigate the expression level of the flavonoid biosyn- thetic genes, qRT-PCR was performed on upper and lower flowers with the primers listed in Table 4. The tested genes Upper flower of M. latifolium included chalcone synthase (CHS), chalcone isomerase (CHI), Lower flower of M. latifolium ′ ′ ′ ′ ′ 600 flavonoid 3 ,5 -hydroxylase (F3 5 H), flavonoid 3 -hydroxy- lase (F3′H), dihydroflavonol 4-reductase (DFR), and 500 anthocyanidin synthase (ANS). The key enzymes involved in 400 the synthesis of delphinidin and cyanidin are F3′5′HandF3′H, 300 respectively (Holton and Tanaka 1994). The expression levels 200 of all genes were higher in the lower flowers relative to the 100 upper flowers except for ANS. Specifically, F3′5′H, F3′H,and DFR2 were expressed strongly in the lower flowers (Fig. 3). 3.0 Their marked expressions caused a substantial accumulation 2.5 of delphinidin and cyanidin in the lower flowers, and the 2.0 combination of the two pigments produced an amaranth color. 1.5 Relative level of transcripts These results suggest that the formation of the bicolor in the 1.0 upper and lower parts is attributable to the different expres- 0.5 sions of F3′5′H, F3′H,andDFR2 leading to corresponding 0.0 CHS CHI F3'5'H F3'H DFR1 DFR2 ANS differences in the amounts of anthocyanin accumulating. Flavonoid biosynthetic genes

Fig. 3 Expression analyses of flavonoid biosynthetic genes in upper Discussion and lower flowers of M. latifolium by quantitative real-time PCR. Actin was used as an internal control. The transcript of each gene was quantified by densitometry using Bio imaging systems software The expression of tepal color involves many different mech- (Syngene). Mean values and SDs were obtained from two technical anisms (Yoshida et al. 2009a, b). Particular interest has been and three biological replicates. CHS chalcone synthase gene, CHI focused on those associated with the development of the chalcone isomerase gene, F3′5′H flavonoid 3′,5′-hydroxylase gene, F3′H flavonoid 3′-hydroxylase gene, DFR dihydroflavonol 4-reductase blue colors. In the blue dayflower Commelina communis, gene, ANS anthocyanidin synthase gene Kondo et al. (1992) revealed the structure of commelinin to 1280 Y. Qi et al. be a metalloanthocyanin. Meanwhile, cyanidin, which is flowers. At the same time, we note that DFR2 displayed a generally red, was shown to exhibit a blue coloration with high expression level in the lower flowers, whereas there Fe3+ and Mg2+ in Centaurea cyanus (Kondo et al. 1994, was no detectable expression of DFR1in either the upper or 1998; Shiono et al.2005; Takeda et al. 2005) and in the lower flowers. It was previously reported that the type of Himalayan blue poppy Meconopsis grandis (Yoshida et al. anthocyanidin accumulating depends on the substrate spec- 2006). Yoshida et al. (1995, 2005) also revealed that the ificity of DFR (Forkmann et al. 1999; Katsumoto et al. mechanism underlying blue color development in the morn- 2007). Therefore, we speculate that DFR1 takes ing glory, Ipomoea tricolor cv. Heavenly Blue, during flow- dihydroquercetin (DHQ) as substrate, while DFR2 catalyzes er opening was an increase in vacuolar pH from 6.6 to 7.7. dihydromyricetin (DHM). In some plants, anthocyanin chromophores were the key It seems reasonable to conclude that the different expres- factor deciding phenotype color, such as in A. monelli sions of key genes lead to a diversity in anthocyanin accu- (Quintana et al. 2007), Delphinium (Hashimoto et al. mulation, thus forming the two-tone flower. Future studies 2002), and H. macrophylla cv. Hovariatrade mark will focus on the selectivity of DFR for substrates. “Homigo” (Yoshida et al. 2008). Moreover, the intra- or inter-molecular stacking of co-pigments such as flavonols Acknowledgments We would like to thank Dr. Weirong Xu at and aromatic organic acids also leads to a blue shift of Ningxia University for helpful suggestions. This work was supported by the National Natural Science Foundation of China (grant no. flower coloration (Goto and Kondo 1991). 31170652). Based on these factors, to elucidate the cause of “two- toned” flowers, we first undertook anatomical research, and Conflict of interest We declare that we have no conflict of interest. then prepared free protoplasts, observed the shapes of the anthocyanin containing cells, investigated vacuolar pH and tissue metal ion content, and the compositions of anthocyanin and co-pigments. These results indicate that in the different References colored cells, cell shape and vacuolar pH are similar (Fig. 2, Table 3). Also, chemical analyses reveal that the metal ion Albert NW, Lewis DH, Zhang H, Irving LJ, Jameson PE,Davies KM compositions and the dihydroflavonol and flavonol contents (2009) Light-induced vegetative anthocyanin pigmentation in Pe- tunia. Journal of Experimental Botany 60:2191–2202 are also similar. Only the anthocyanins show large differences Biolley JP, Jay M (1993) Anthocyanins in modern roses: chemical and both of composition and also of concentration. The upper colorimetric features in relation to the colour range. J Exp Bot flowers contained only delphinidin, whereas the lower flowers 44:1725–1734 contain both delphinidin and cyanidin (Table 3). Thus, our Boyde A, Wood C (1969) Preparation of animal tissues for surface- scanning electron microscopy. J Microsc 90:221–249 results indicate that flower color is correlated with the visible Doussi MA, Thanos CA (2002) Ecophysiology of seed germination in anthocyanidin content. Mediterranean geophytes. 1. Muscari spp. Seed Sci Res 12:193–201 Data from quantitative real-time PCR revealed that, al- Forkmann G, Heller W (1999) Biosynthesis of flavonoids. In: Sankawa though all the genes tested were expressed both in the upper U (ed) Comprehensive natural products chemistry, volume 1. Polyketides and other secondary metabolites including fatty acids and lower flowers of M. latifolium, CHS, DFR1are and their derivatives. Elsevier, Oxford, pp 713–748 expressed at a low level, especially in the upper flower, it Goto T, Kondo T (1991) Structure and molecular stacking of is noted that the expressions of F3′H and F3′5′H were anthocyanins-flower color variation. Angew Chem Int Ed Engl relatively higher in the lower flowers than that in the upper 30:17–33 ′ ′ ′ Hanbury A, Serra J (2002) Mathematical morphology in the CIELAB flowers (Fig. 3). F3 H and F3 5 H determine the hydroxyl- space. Image Anal Stereol 21:201–206 ation pattern of the B-ring, so they likely play a crucial role Hashimoto F, Tanaka M, Maeda H, Fukuda S, Shimizu K, Sakata Y in the determination of flower color (Tanaka et al. 2013). (2002) Changes in flower coloration and sepal anthocyanins of The presence of F3′5′H leads to trihydroxylated delphinidin- Cyanic delphinium cultivars during flowering. Biosci Biotechnol Biochem 66:1652–1659 based anthocyanins that tend to have violet/blue colors, Holton TA,Tanaka Y (1994) Blue roses—a pigment of our imagina- whereas F3′H accumulates cyaniding-based anthocyanins tion? Trends Biotechnol. 12:40–42 having red/magenta colors (Tanaka et al. 2008). Our qRT- Ito D, Shinkai Y, Kato Y, Kondo T, Yoshida K (2009) Chemical studies PCR data indicated that F3′5′H, F3′H were expressed more on different color development in blue- and red-colored sepal cells of Hydrangea macrophylla. Biosci Biotechnol Biochem powerfully in the lower flowers than in the upper ones. 73:1054–1059 Therefore, substantial amounts of delphinidin and cyanidin Katsumoto Y, Fukuchi-Mizutani M, Fukui Y, Brugliera F, Holton TA, accumulate in the lower parts of the inflorescence showing Karan M et al (2007) Engineering of the rose flavonoid biosyn- an amaranth color. Although the expression of F3′H is thetic pathway successfully generated blue-hued flowers accumu- ′ ′ lating delphinidin. Plant Cell Physiol 48:1589–1600 higher than of F3 5 H in both regions, the production of Kondo T, Yoshida K, Nakagawa A, Kawai T, Tamura H, Goto T (1992) delphinidin is still greater than that of cyanidin, even though Structural basis of blue-colour development in flower petals from we did not examine the cyanidin content in the upper Commelina communis. Nature 358:515–518 Studies of bicolored flower development in M. latifolium 1281

Kondo T, Ueda M, Isobe M (1998) A new molecular mechanism of Takeda K, Osakabe A, Saito S, Furuyama D, Tomita A, Kojima Y et al blue color development with protocyanin, a supramolecular pig- (2005) Components of protocyanin, a blue pigment from the blue ment from cornflower, Centaurea cyanus. Tetrahedron Lett flowers of Centaurea cyanus. Phyotochemistry 66:1607–1613 49:8307–8310 Tanaka Y, Brugliera F, Chandler S (2009) Recent progress of flower Kondo T, Ueda M, Tamura H, Yoshida K, Isobe M, Goto T (1994) colour modification by biotechnology. Int J Mol Sci 10:5350–5369 Composition of protocyanin, a self-assembled supramolecular Tanaka Y, Ohmiya A (2008) Seeing is believing: engineering antho- pigment form the blue cornflower Centaurea cyanus. Angew cyanin and carotenoid biosynthetic pathways. Curr Opin BiotechI Chem Int Ed Engl 33:978–979 19:190–197 Livak KJ, Schmittgen TD (2001) Analysis of relative gene expression Tanaka Y, Brugliera F (2013) Flower colour and cytochromes P450. data using real-time quantitative PCR and the 2(T)(−Delta Delta C) Phil Trans R Soc B 368:20120432 method. Methods 25:402–408 Torskangerpoll K, Nørbæk R, Nodland E, Øvstedal DO, Andersen ØM Momonoi K, Tsuji T, Kazuma K, Yoshida K (2012) Specific expres- (2005) Anthocyanin content of Tulipa species and cultivars and its sion of the vacuolar iron transporter, TgVit, causes iron accumu- impact on tepal colours. Biochem Syst Ecol 33:499–510 lation in blue-colored inner bottom segments of various tulip Yoshida K, Ito D, Shinkai Y, Kondo T (2008) Change of color and petals. Biosci Biotechnol Biochem 76:319–325 components in sepals of chameleon hydrangea during maturation Momonoi K, Yoshida K, Mano S, Takahashi H, Nakamori C, Shoji K, and senescence. Phytochemistry 69:3159–3165 Nitta A, Nishimura M (2009) A vacuolar iron transporter in tulip, Yoshida K, Kawachi M, Mori M, Maeshima M, Kondo M, Nishimura TgVit1, is responsible for blue coloration in petal cells through M, Kondo T (2005) The involvement of tonoplast proton pumps iron accumulation. Plant J 59:437–447 and Na+(K+)/H+exchangers in the change of petal color during Mori S, Asano S, Kobayashi H, Nakano M (2002) Analyses of flower opening of morning glory, Ipomoea tricolor cv. Heavenly anthocyanidins and anthocyanins in flowers of Muscari spp. Blue. Plant Cell Physiol 46:407–415 Niigata Daigaku Nogakubu Kenkyu Hokoku 55:13–18 Yoshida K, Kitahara S, Ito D, Kondo T (2006) Ferric ions involved in Noda K, Glover BJ, Linstead P, Martin C (1994) Flower colour the flower color development of the Himalayan blue poppy, intensity depends on specialized cell shape controlled by a Myb- Meconopsis grandis. Phytochemistry 67:992–998 related transcription factor. Nature 369:661–664 Yoshida K, Mori M, Kondo T (2009a) Blue flower color development Quintana A, Albrechtova´ J, Griesbach RJ, Freyre R (2007) Anatom- by anthocyanins: from chemical structure to cell physiology. Nat ical and biochemical studies of anthocyanidins in flowers of Prod Rep 26:884–915 Anagallis monelli L. (Primulaceae) hybrids. Scientia Hortic- Yoshida K, Miki N, Momonoi K, Kawachi M, Katou K, Okazaki Y, Amsterdam 112:413–421 Uozumi N, Maeshima M, Kondo T (2009b) Synchrony between Schreiber HD, Jones AH, Lariviere CM, Mayhew KM, Cain JB (2011) flower opening and petal-color change from red to blue in morn- Role of aluminum in red-to-blue color changes in Hydrangea ing glory, Ipomoea tricolor cv. Heavenly Blue. P Jpn Acad A- macrophylla sepals. Biometals 24:1005–1015 math 85:187–197 Schreiber HD, Swink AM, Godsey TD (2010) The chemical mecha- Yoshida K, Osanai M, Kondo T (2003a) Mechanism of dusky reddish- nism for Al3+ complexing with delphinidin: a model for the bluing brown “Kaki” color development of Japanese morning glory, of hydrangea sepals. J Inorg Biochem 104:732–739 Ipomoea nil cv. Danjuro. Phytochemistry 63:721–726 Shiono M, Matsugaki N, Takeda K (2005) Phytochemistry: structure of Yoshida K, Toyama Y, Kameda K, Kondo T (2003b) Sepal color varia- the blue cornflower pigment. Nature 436:791 tion of Hydrangea macrophylla and vacuolar pH measured with a Shoji K, Miki N, Nakajima N, Momonoi K, Kato C, Yoshida K (2007) proton-selective microelectrode. Plant Cell Physiol 4:262–268 Perianth bottom-specific blue color development in tulip cv. Yoshida K, Kondo T, Okazaki Y, Katou K (1995) Cause of blue petal Murasakizuisho requires ferric ions. Plant Cell Physiol 48:243–251 colour. Nature 373:291 Shoji K, Momonoi K, Tsuji T (2010) Alternative expression of vacu- Zhang J, Wang LS, Gao JM, Shu QY, Li C, Yao J, Hao Q, Zhang JJ olar iron transporter and ferritin genes leads to blue/purple color- (2008) Determination of anthocyanins and exploration of rela- ation of flowers in tulip cv. ‘Murasakizuisho’. Plant Cell Physiol tionship between their composition and petal coloration in crape 51:215–224 myrtle (Lagerstroemia hybrid). J ntegr Plant Biol 50:581–588