Generation of Yellow Flowers of the Japanese Morning Glory by Engineering Its Flavonoid Biosynthetic Pathway toward Aurones

Abstract

Wild-type plants of the Japanese morning glory (Ipomoea nil) produce blue flowers that accumulate anthocyanin pigments, whereas its mutant cultivars show wide range flower color such as red, magenta and white. However, I. nil lacks yellow color varieties even though yellow flowers were curiously described in words and woodblocks printed in the 19th century. Such yellow flowers have been regarded as ‘phantom morning glories’, and their production has not been achieved despite efforts by breeders of I. nil. The chalcone isomerase (CHI) mutants (including line 54Y) bloom very pale yellow or cream-colored flowers conferred by the accumulation of 2′, 4′, 6′, 4-tetrahydoroxychalcone (THC) 2′-O-glucoside. To produce yellow phantom morning glories, we introduced two snapdragon (Antirrhinum majus) genes to the 54Y line by encoding aureusidin synthase (AmAS1) and chalcone 4′-O-glucosyltransferase (Am4′CGT), which are necessary for the accumulation of aureusidin 6-O-glucoside and yellow coloration in A. majus. The transgenic plants expressing both genes exhibit yellow flowers, a character sought for many years. The flower petals of the transgenic plants contained aureusidin 6-O-glucoside, as well as a reduced amount of THC 2′-O-glucoside. In addition, we identified a novel aurone compound, aureusidin 6-O-(6″-O-malonyl)-glucoside, in the yellow petals. A combination of the coexpression of AmAS1 and Am4′CGT and suppression of CHI is an effective strategy for generating yellow varieties in horticultural plants.

Introduction

Flower color plays an important role in attracting pollinators and is a crucial characteristic of floricultural crops. Plant breeders have successfully expanded flower color varieties through extensive hybridization programs. For example, yellow-cultivated roses were produced from hybridizations of the Persian yellow rose (Rosa foetida). Yellow flower pigments have been classified into the three groups: flavonoids, carotenoids and betaxanthins (a class of betalains) (Tanaka et al. 2008). Flavonoids are subclassified into a dozen groups of which chalcones, aurones and flavonols provide yellow pigmentation in flowers. The yellow pigmentation is due to a λmax around 365–430 nm (Brouillard and Dangles 1994). Aurones confer the clearest yellow color of flavonoid compounds, as is the case for Scrophulariaceae [e.g. snapdragon (Antirrhinum majus)], Compositae [e.g. dahlia (Dahlia variabilis) and cosmos (Cosmos bipinnatus)] (Ono and Nakayama 2007). These flavonoids are usually glucosylated and accumulate in the vacuoles of petal epidermal cells. The yellow color of roses and chrysanthemums are derived from carotenoids (Tanaka et al. 2008), whereas the yellow of portulacas and cockscombs are derived from betaxanthins (Gandia-Herrero and Garcia-Carmona 2013).

A schematic diagram of the biosynthesis pathway for flavonoids is shown in Fig. 1A. 2′, 4′, 6′, 4-tetrahydoroxychalcone (THC) is the first compound in the biosynthesis pathway, which is then converted to aurones and flavanones and then to other classes of flavonoids, including anthocyanins. A typical aurone in the yellow petals of snapdragon is aureusidin 6-O-glucoside (Fig. 1A). Its biosynthesis pathway and its two key enzymes [aureusidin synthase (AmAS1) and chalcone 4′-O-glucosyltransferase (Am4′CGT)] have been well characterized in A. majus (Nakayama et al. 2000, Sato et al. 2001, Ono et al. 2006). Cytosolic Am4′CGT catalyzes glucosylation of the 4′-hydoroxyl group of THC to yield THC 4′-O-glucoside, which is transported into vacuoles and then converted to aureusidin 6-O-glucoside by the catalysis of AmAS1 (Ono et al. 2006).

Fig. 1

(A) Simplified flavonoid biosynthesis pathway. The enzyme names are presented in bold: CHS (chalcone synthase), CHI (chalcone isomerase), F3H (flavanone 3-hydroxylase), AS (aureusidin synthase) and 4´CGT (chalcone 4´-O-glucosyltransferase). (B) Binary vectors. The transgenes are regulated by the constitutive enhanced Cauliflower Mosaic Virus 35S promoter (pEL2-35S) (Mitsuhara et al. 1996). tNOS indicates the Agrobacterium nopaline synthase terminator. (C) Transgene expression analyzed with RT-PCR. The I. nil γ-subunit of F₁F₀-ATP synthase was used as an internal control.

Torenia (Torenia hybrida), belonging to the family Linderniaceae, mainly accumulates anthocyanins in its petals. The coexpression of the Am4′CGT and AmAS1 genes results in the production of aureusidin 6-O-glucoside in torenia petals (Ono et al. 2006). Yellow torenia flowers are generated by the coexpression of both genes and the downregulation of anthocyanin production through RNA interference of the flavanone 3-hydroxylase (F3H) or dihydroflavonol 4-reductase (DFR) gene. Transgenic torenia overexpressing the AmAS1 gene alone fails to produce aurones (Ono et al. 2006). The successful generation of yellow by the coexpression of the Am4′CGT and AmAS1 genes has been not reported in other species. Flower color modification by the production of betaxains or carotenoids has also not been reported, although introducing the betaxanin biosynthetic pathway to tobacco BY-2 cells results in yellow cells (Nakatsuka et al. 2013, Polturak et al. 2017).

The Japanese morning glory (Ipomoea nil), in the family Convolvulaceae, is a traditional and popular horticultural plant in Japan and a common botanical material used in Japanese elementary schools. Its petals are blue due to the accumulation of highly modified peonidin-based anthocyanin modified with three aromatic acyl groups and six glucose moieties under a high vacuolar pH (Lu et al. 1992, Fukada-Tanaka et al. 2000, Yamaguchi et al. 2001).

Many spontaneous mutations exhibiting the alteration of flower pigmentation have been isolated (Hagiwara 1956, Lu et al. 1992). Thanks to such mutations, I. nil possesses cultivars with wide range petal colors, such as red, magenta and white (Morita and Hoshino 2018). For example, the cultivars carrying a mutation of the flavonoid 3′-hydroxylase gene (Magenta), which accumulates pelargonidin-based anthocyanins, show magenta petal color, whereas the cultivars having a mutation of the cation/H⁺ gene (Purple) exhibit a purple petal color (because they fail to elevate pH in the vacuole) (Fukada-Tanaka et al. 2000, Hoshino et al. 2003). Although I. nil lacks bright yellow cultivars today, bright yellow flowers were repeatedly recorded in block painted picture books published in the 19th century (Shijian 1817, Shusoen 1853). A few I. nil cultivars and experimental lines currently exhibit pale yellow flowers at most. Among them, the I. nil line 54Y (Fig. 2A) carries the speckled mutation that is a transposon insertion into the CHI gene (Hoshino et al. 2001). Due to this mutation, its petals accumulate chalcone glucoside (THC 2′-O-glucoside) as a major pigment and a trace amount of aurone (aureusidin 4-O-glucoside) (Abe et al. 1997, Hoshino et al. 1997, Saito et al. 2011). The flowers of 54Y often fail to fully open and produce yellowish brown features in their flower petals that may consist of necrotic cells (Fig. 2A).

Fig. 2

Flower phenotypes. (A) Comparison of petal colors of the AmAS1/Am4´CGT expressing line (left) and the host line 54Y (right). White arrowheads indicate necrotic cells on the pale cream yellow petal of the host line. (B) Petal of the AmAS1/Am4´CGT line superimposed against the yellow group 5 of the Royal Horticultural Society Color Chart. These photographs were taken using the same camera and light settings. (C) Comparison of petal colors the AmAS1/Am4´CGT line and snapdragon cv. Floral Carpet Yellow (lower left, Sakata Seed). (D) Stigmas of the host and AmAS1/Am4´CGT line. The picture was taken using a VHX-100 digital microscope (Keyence, Osaka, Japan). Scale bar represents 1 mm.

We hypothesize that enhancement of aurone accumulation in the CHI mutant will result in a clear yellow coloration in its petals. The overexpression of the Am4′CGT and AmAS1 genes in the 54Y line should successfully result in aurone accumulation in fully open yellow flowers.

Results

Generation and phenotype of the I. nil CHI mutant expressing the AmAS1 and Am4′CGT genes

To increase aurone accumulation, we introduced the Antirrhinum genes AmAS1 and Am4′CGT into the 54Y line using an Agrobacterium-mediated transformation. Three binary vectors: pSFL304 to express both AmAS1 and Am4′CGT genes constitutively, pSPB211 to express the AmAS1 gene and pSFL209 to express the Am4′CGT gene were constructed (Fig. 1B). We obtained 17, 20 and 16 transgenic lines transformed with the vectors pSFL304, pSPB211 and pSFL209, respectively. From these lines, we randomly chose 6, 3 and 3 lines that carried the AmAS1/Am4′CGT, AmAS1 and AmAS1 genes, respectively, to grow in greenhouses. We confirmed the expression of the transgenes with RT-PCR (Fig. 1C), and further confirmed a second time using quantitative reverse transcription–PCR (RT–qPCR) (Table 1). Line #70 of AmAS1 and Line #36 of Am4′CGT showed very low expression of the transgenes.

Table 1

Expression and flavonoid analysis of the transgenic petals

		mRNAa		Flavonoidb
		AmAS1	Am4′CGT	Aureusidin 6-O-glucoside		Aureusidin 6-O-(6′-O-malonyl)-glucoside		THC2′G
Transgene	Line	%	%	Area	Area %	Area	Area %	Area	Area %
AmAS1/Am4′CGT	1-1-2	81.3	38.6	604,604	26.5	770,956	29.5	131,669	5.0
	2-4	102.4	22.4	699,846	24.8	836,854	29.6	147,296	5.2
	3-1	100.0	100.0	744,761	24.7	857,690	28.4	219,084	7.3
	4-1	31.7	21.4	307,196	20.9	302,446	20.6	304,815	20.7
	5-1	134.4	24.0	622,210	25.9	801,411	33.4	60,964	2.5
	DD	32.2	44.5	344,832	20.3	524,715	30.9	97,437	5.7
AmAS1	39	75.7	NT	144,729	9.2	217,586	13.8	540,441	34.3
	46	57.4	NT	211,480	9.9	246,457	11.5	474,466	22.1
	70	1.6	NT	0	0	0	0	547,998	67.1
Am4′CGT	36	NT	0.1	0	0	0	0	680,638	63.8
	60	NT	21.9	0	0	0	0	538,862	47.7
	74	NT	12.7	0	0	0	0	714,651	68.2
No transgenes	54Y	NT	NT	0	0	0	0	1,047,062	64.3
	54Yc	NT	NT	0	0	0	0	1,309,885	59.4

NT indicates not tested.

a

Relative expression levels. mRNA levels of the AmAS1/Am4′CGT overexpressing line #3–1 were set as 100.

b

The flavonoid contents were analyzed by HPLC and are shown in areas and area %.

c

The plant is regenerated from cultured cells prepared from immature embryo.

Among the greenhouse-grown transgenic plants, the AmAS1/Am4′CGT overexpressing lines showed clear color change from pale yellow to bright yellow in petals. The flower rays showed stronger yellow pigmentation than other parts of petal (Fig. 2A, Supplementary Fig. S1). The intensity of yellow color varied, depending on the line, but Line #3–1 showed the most yellow petals. Its petals corresponded to 5D in the yellow group of the Royal Horticultural Society Color Charts (Fig. 2B) and dorsal petals of a yellow snapdragon cultivar (Fig. 2C). Although the host line often exhibits shriveled flowers with necrotic cells, most of the AmAS1/Am4′CGT lines produced fully opened flowers with few necrotic cells (Fig. 2A, Supplementary Fig. S1). Stigmas of the AmAS1/Am4′CGT lines were bright yellow, while stigmas of the host lines were whitish (Fig. 2D). The petal colors of the 11 chamber-grown lines were pale to bright yellow depending on lines and flowers, and most of the petals of the 11 lines had a lighter color than the petals of the six greenhouse-grown lines (Supplementary Fig. S2). The chamber-grown lines were not subjected to further studies because their flower colors were inconsistent.

In contrast, none of the transgenic lines harboring either AmAS1 or Am4′CGT gene exhibited an obviously yellow coloration (Supplementary Fig. S1). These plants produced the same pale yellow petals as the host line and the Am4′CGT lines sometimes bloomed whiter flowers. Both lines produced shriveled flowers with necrotic cells.

Aurone accumulation in the AmAS1/Am4′CGT expressing lines

To characterize the pigments accumulated in the transgenic plants, we conducted an high performance liquid chromatography (HPLC) analysis (Fig. 3A–G), which showed that petals of the host plant contained a large amount of THC 2′-O-glucoside having retention time of 13.09 min with a characteristic spectrum (λmax at 368 nm) of chalcone compounds (Table 1, Fig. 3B). This HPLC result was consistent with the previous HPLC analysis of the host plant by Saito et al. (2011). The yellow petals of all of the AmAS1/Am4′CGT transgenic lines we produced contained a reduced amount of THC 2′-O-glucoside and accumulated compounds having retention times of 10.49 and 11.61 min, neither of which we detected in the host plant (Table 1, Fig. 3C). The compounds exhibited absorption spectra characteristic of aurones (λmax at 404 and 403 nm, respectively) (see Fig. 4D, Supplementary Fig. S3D). The first peak was identified as aureusidin 6-O-glucoside, based on its retention time and mass spectrum (described below). The second peak (Compound A) was subjected to structural studies as discussed below. Aureusidin 6-O-glucoside and small amount of Compound A were also detected in the yellow stigmas of the transgenic morning glories (Fig. 3G). We confirmed the aglycones of the petal and stigma pigments by an HPLC analysis of pigments after hydrolysis (Fig. 3H–N). The main flavonoid aglycons of the AmAS1/Am4′CGT transgenic lines were aureusidin and naringenin, spontaneously isomerized forms of THC (Fig 3J, N).

Fig. 3

HPLC chromatograms of petal and stigma pigments detected at 390 nm. (A–G) Pigments extracted from lyophilized samples. (H–N) Aglycones of the pigments. (A, H) Standards, (B, I) petals of the host line 54Y, (C, J) petals of the AmAS1/Am4´CGT line (#4–1), (D, K) petals of the AmAS1 line (#39), (E, L) petals of the Am4´CGT line (#46), (F, M) stigmas of 54Y and (G, N) stigmas from the six AmAS1/Am4´CGT greenhouse-grown plants. Bracteatin is 4, 6, 3´, 4´, 5´-pentahydroxylated aurone.

Fig. 4

Yellow petals of the AmAS1/Am4´CGT transgenic lines produced the novel aurone, aureusidin 6-O-(6″-O-malonyl)-glucoside. (A) Positive ion mode of MS and MS/MS spectra showed molecular ion at m/z 535.10 [M + H]⁺ and product ion at m/z 287.05 [M–86 (Mal)–162 (Glc)+H]⁺. (B) Negative ion mode of MS and MS/MS spectra showing fragment ion at m/z 489.08 [M–44 (CO₂)–H]⁻ and product ion at m/z 285.03 [M–44 (CO₂)–42 (Ac)–162 (Glc)–H]⁻. (C) Chemical structure and proposed fragmentation patterns (dotted arrows) of aureusidin 6-O-(6´´-O-malonyl)-glucoside. (D) Absorption spectrum obtained with a PDA detector. Moieties: Mal (malonyl), Glc (glucosyl) and Ac (acetyl).

The flavonoids in the petals were further characterized using liquid chromatography–mass spectrometry (LC-MS) (Fig. 4, Supplementary Fig. S3). The mass spectrum of the 20.63 min compound suggests that it is aureusidin attached to one hexose moiety (Supplementary Fig. S3A–C); its mass spectrum and retention times matched those of the authentic aureusidin 6-O-glucoside. The molecular ion peaks for the compound with a retention time of 23.59 min (Compound A) was larger than aureusidin 6-O-glucoside by m/z 86, which corresponds to the mass of a malonyl group (Fig. 4A, Supplementary Fig. S3A). The compound at 30.45 min was a chalcone, based on its absorption spectrum (Supplementary Fig. S3H). The LC-MS analysis indicated that this compound is chalcone mono hexoside (Supplementary Fig. S3E–G).

The AmAS1 transgenic petals also accumulated small amounts of aureusidin 6-O-glucoside and Compound A in addition to THC 2′-O-glucoside. Nevertheless, their petals exhibited no obvious changes in color (Table 1, Supplementary Fig. S1, Fig. 3D, K). The amounts of aureusidin 6-O-glucoside and Compound A were much lower than the concentrations in AmAS1/Am4′CGT transgenic petals. None of the transgenic lines, including the Am4′CGT transgenic lines, showed significant accumulation of THC 4′-O-glucoside (Fig. 3C–E).

Production of a novel malonylated aureusidin in the yellow morning glory

Our LC-MS analysis indicated that the molecular formula of Compound A (extracted from AmAS1/Am4′CGT transgenic petals) was C₂₄H₂₂O₁₄ (Fig. 4A, B). Our MS/MS analysis suggested that the compound consisted of aureusidin, hexose and a malonyl group (Fig. 4A–C). We performed nuclear magnetic resonance (NMR) spectroscopy experiments to determine the structure of Compound A. The ¹H and ¹³C chemical shift values of compound A were in good agreement with those of aureusidin 6-O-glucoside, except for the values at the 5″- and 6″-positions of the glucose moiety (Supplementary Table S1, Fig. 4C). In addition, the two signals of glucose H-6″ correlated with the signal for the malonyl group C-1‴ in the HMBC spectrum, which indicates that Compound A is malonylated at the 6″-position of the glucose. The spectra data indicate that Compound A is aureusidin 6-O-(6″-O-malonyl)-glucoside (Fig. 4C). Absorption spectra for aureusidin 6-O-(6″-O-malonyl)-glucoside were very similar to the spectra for aureusidin 6-O-glucoside and differed from those for THC 2′-O-glucoside (Fig. 4D, Supplementary Fig S3D, H).

Ipomoea nil homolog of Am4′CGT

The presence of aureusidin 6-O-glucoside and aureusidin 6-O-(6″-O-malonyl)-glucoside in petals expressing AmAS1 gene prompted us to speculate that the Japanese morning glory may have an enzyme catalyzing 4′CGT activities. A BLSTP search using Am4′CGT as a query against the Japanese morning glory genome (Hoshino et al. 2016) revealed that the glucosyltransferase encoded by INIL05g22870 (XP_019200220) locus has the highest amino acid sequence identity (48%) to Am4′CGT (Supplementary Fig. S4). Based on the RNA-seq data published in the NCBI genome database, INIL05g22870 is typically expressed in flowers. These results suggest that the glucosyltransferase encoded by INIL05g22870 may be involved in aureusidin biosynthesis as 4′CGT in the AmAS1 transgenic lines.

Discussion

In the present study, we successfully generated yellow flowers of the Japanese morning glory, which have been long regarded as phantom morning glories, by engineering the flavonoid biosynthetic pathway using snapdragon aurone biosynthetic genes. We succeeded in intensifying the pale yellow color of the flower petals of the CHI mutant line (54Y) to visible yellow via the coexpression of AmAS1 and Am4′CGT genes. The yellow petals of the transgenic lines not only accumulated aureusidin 6-O-glucoside, but they also accumulated aureusidin 6-O-(6″-O-malonyl)-glucoside, which had not been reported from any plant species, including the Japanese morning glory. Aureusidin accumulation in I. nil suggests that an endogenous malonyltransferase unphysiologically catalyzes malonylation of aureusidin 6-O-glucoside to yield aureusidin 6-O-(6″-O-malonyl)-glucoside. In plants, phenolic compounds, including flavonoids, are acylated by BAHD and serine carboxy peptidase-like (SCPL) acyltransferase families (Bontpart et al. 2015). Ipomoea nil genome encodes a number of acyltransferases belonging to these families. Further studies are required to identify the endogenous aureusidin glucoside malonyltransferase. Yellow pigmentation and aureusidin 6-O-glucoside accumulation were also observed in stigmas of the AmAS1/Am4′CGT transgenic lines, whereas I. nil, including flower color mutants, have white stigmas. This indicates that stigmas of I. nil produce THC and do not express malonyltransferase, which catalyzes the malonylation of aureusidin 6-O-glucoside in petals.

Because the transgenic lines expressing only AmAS1 produced a small amount of aurones (Table 1), I. nil may express an endogenous 4′CGT activity in petals. Saito et al. (2011) reported that the CHI mutant of I. nil accumulates the least amount of aureusidin 4-O-glucoside; therefore, I. nil may also express an endogenous AS activity in petals. It can be assumed that the yellow flowers, which appeared in the 19th century, were derived from the endogenous 4′CGT and AS activities. Alternatively, the yellow I. nil flower might accumulate a large amount of THC 2′-O-glucoside, which confers a pale yellow color to some plants including CHI mutants of carnation (Itoh et al. 2002, Saito et al. 2011). Fortuitous and specific physiological and environmental conditions would be required for the reproduction of the yellow morning glory flowers.

Our study underscores the advantages of the metabolic engineering of heterologous genes to generate novel flower colors in floricultural crops. New blue hues have been produced in transgenic carnations, roses and chrysanthemums, hues that native plants of these species lack, by engineering plants to express flavonoid 3′,5′-hydroxylase genes (Katsumoto et al. 2007, Brugliera et al. 2013, Noda et al. 2013, Tanaka and Brugliera 2013). Furthermore, the additional expression of the butterfly pea anthocyanin 3′,5′-glucosyltransferase gene further has shifted the flower color in chrysanthemum to an even deeper blue (Noda et al. 2017).

Among the 17 AmAS1/Am4′CGT transgenic lines, 11 were cultured in growth chambers with normal fluorescent lights. The yellow coloration of the petals of the chamber-grown lines was inconsistent and generally paler than those of the six greenhouse-grown lines. This difference can be attributed to the variation of transgene expression among the transgenic lines and to environmental conditions. Because the expression of the CHS gene in many plant species is stimulated by sunlight, especially by UV-B wavelengths (Jenkins 1997, Müller-Xing et al. 2014), sunlight is likely to be necessary for stable and higher aureusidin accumulation to generate clearly visible yellow coloration in petals of the AmAS1/Am4′CGT transgenic lines.

The Am4′CGT transgenic lines accumulated no detectable THC 4′-O-glucoside and sometimes produced whiter petals than petals of the host lines. This suggested that THC 4′-O-glucoside produced in these lines is converted to other colorless compounds in petal cells. The lack of THC 4′-O-glucoside accumulation is consistent with the fact that THC 4′-O-glucoside has not been found in the Japanese morning glory, including acyanic flower mutants, despite the expression of INIL05g22870 in flowers. In contrast, torenia expressing the Am4′CGT gene alone does accumulate THC 4′-O-glucoside (Ono et al. 2006). The intrinsic function of INIL05g22870 in I. nil flowers is yet to be elucidated.

The yellow flowers of the AmAS1/Am4′CGT lines fully open, whereas the host line often produces small and partially opened flowers that have yellowish brown parts composed of necrotic petal cells. Because fewer and smaller yellowish brown parts appeared in the yellow petals of the AmAS1/Am4′CGT lines, this suggests that THC 2′-O-glucoside (or its derivatives) are toxic for petal cells and that removal of these compounds due to aurone production results in the petals of the transgenic plants fully opening. Because the toxicity of THC 2′-O-glucoside has not been reported despite its presence in yellow petals (Forkmann and Dangelmayr 1980), we speculate that the ectopic accumulation of THC 2′-O-glucoside (and/or its derivatives) is harmful to plant cells. In the torenia AmAS1/Am4′CGT transgenic lines, no changes in flower shape due to the transgene expression have been reported (Ono et al. 2006). Neither necrotic petal cells nor defects in flower shape were observed in their host lines in which F3H and DFR genes were repressed by RNA interference.

In torenia, knock-down of the F3H and DFR genes by RNA interference were necessary to produce yellow flowers via the coexpression of AmAS1 and Am4′CGT genes (Ono et al. 2006). However, the coexpression of AmAS1 and Am4′CGT genes in the DFR-B mutant having white flower petals failed to generate yellow flowers (Supplementary Fig. S5A). All 13 AmAS1/Am4′CGT transgenic plants derived from a mutant of DFR-B grown in greenhouses showed no obvious changes in petal color and flavonoid accumulation compositions (Supplementary Fig. S6) despite of the fact that expression of the transgenes was confirmed (Supplementary Fig. S5B). Endogenous F3H possibly catabolized THC more efficiently than exogenous Am4′CGT; therefore, THC was not significantly converted to THC 4′-O-glucoside. To generate yellow flower varieties in a wide range of floricultural plants by the coexpression of AmAS1 and Am4′CGT genes, choosing CHI mutants as the hosts would be the first choice. Many important floricultural crops, such as cyclamen and begonia, only have very pale or no yellow varieties. This study has paved the way to produce yellow petal varieties in such crops.

Materials and Methods

Transformation of the Japanese morning glory

We chose the CHI mutant line of 54Y as a host plant, described by Abe et al. (1997), Hoshino et al. (1997) and Saito et al. (2011). The binary vectors, pSPB211, pSFL209 and pSFL304 carrying the cDNA of the AmAS1, Am4′CGT and both Am4′CGT and AmAS1 genes, respectively, were used for the overexpression of these genes (Fig. 1B). All of the A. majus gene’s cDNA were driven by the enhanced cauliflower mosaic virus 35S promoter (Mitsuhara et al. 1996). These binary vectors were introduced into the Agrobacterium tumefaciens strain EHA105. Transformation procedure of I. nil has been previously descried by Takatori et al. (2015). All plants were grown in chambers at their early growth stage under regular fluorescent lamps. We moved randomly selected plants to greenhouses to continue growth and those plants were subjected to analyses.

HPLC analysis of flavonoids

We collected the petals of fully opened flowers from each plant and lyophilized them. We then extracted flavonoids with 4 ml of 50% acetonitrile containing 0.1% TFA derived from 0.5 g of lyophilized petals. We subjected 10 μl of each extract to a column of Shim-pack FC-ODS (φ4.6 × 150 mm) using a HPLC system (LC-2030C, Shimadzu, Kyoto, Japan) attached to a photodiode array detector (SPD-M20A, Shimadzu). The flow rate was 0.6 ml/min, and the column temperature was 40 °C. Water containing 0.1% TFA (solvent A) and 90% acetonitrile containing 0.1% TFA (solvent B) were used as the mobile phase under gradient elution conditions. The gradient solution program was as follows: a linear gradient from 20% to 100% solvent B for 16 min, 100% solvent B for 6 min and 20% solvent B for 2 min. We prepared THC-2′-O-glucoside from yellow carnations and aureusidin 6-O-glucoside from snapdragons and stored them as authentic flavonoids in an in-house bank (Suntory Global Innovation Center). Flavonoid aglycones were prepared by hydrolysis of petal extracts as described by Katsumoto et al. (2007).

LC-MS analysis

Flavonoids were extracted from the lyophilized petals with 10 time volume of 50% acetonitrile containing 0.1% TFA. We filtered extract of the petals through Millex-LH membranes (0.45 μm, Merck, Darmstadt, Germany), which was then subjected to filtrate to LCMS-IT-TOF (Shimadzu) mass spectrometer to which we attached a photodiode array detector (SPD-M20A, Shimadzu). Six microliters of each extract was applied to Inertsil ODS-4 column (φ4.6 × 250 mm, GL Sciences, Tokyo, Japan). Water containing 0.1% formic acid (solvent A) and 90% acetonitrile containing 0.1% formic acid (solvent B) were used as mobile phase under gradient elution conditions. The gradient solution program was as follows: from 10% to 45% solvent B for 35 min, from 45% to 100% solvent B for 10 min and 100% solvent B for 2 min at a flow rate of 0.6 ml/min. The column temperature was 40°C. MS and MS/MS data were obtained in positive and negative ion electrospray ionization modes. We performed full MS and MS/MS scans in the range of 250–1,250 m/z for MS(+), 270–1,250 m/z for MS(−) and 100–1,250 m/z for MS/MS(+, −).

Purification of the novel aurone

We ground 31.2 g of the freeze-dried petals, derived from 524 flowers, into powder under liquid nitrogen. We then extracted their flavonoids with 700 ml of 50% acetonitrile containing 0.1% formic acid and sonicated the extraction for 30 min. We centrifuged the crude extract for 10 min at 1, 700 g and recovered the supernatant. After centrifuging for another 5 min at 1,700×g, we filtered the supernatant through 7 and 4 μm filter papers (Kiriyama Glass, Tokyo, Japan). We then concentrated the filtered liquid to the half volume under a vacuum and subjected the concentrate to the column chromatography with 200 ml of Diaion HP-20SS resin (Mitsubishi Chemical, Tokyo, Japan) equilibrated with 0.5% TFA. The resin was washed with 400 ml of 0.5% TFA and eluted with 400 ml of 10% acetonitrile containing 0.1% TFA and then with 25% acetonitrile containing 0.1% TFA. We recovered two yellow color fractions (total 743.6 mg) containing the aurone compounds.

We subjected the two yellow fractions to Shimadzu preparative HPLC system using Inertsil ODS-4 (φ20 × 250 mm). The gradient solution program was as follows: 10% solvent B for 5 min, from 10% to 45% solvent B for 15 min, from 45% to 60% solvent B for 40 min and 60% solvent B for 20 min. We further purified the fractions containing the compound under a different gradient regime: from 40% to 50% solvent B for 45 min, from 50% to 60% solvent B for 15 min and from 60% to 100% solvent B for 5 min. The purified aurone (Compound A) of 4.7 mg was subjected to NMR.

NMR experiments

We dissolved the purified aurone (Compound A) in methanol-d₄ or DMSO-d₆. We acquired the NMR spectra using methanol-d₄ on an AVANCE DMX 750 spectrometer, whereas the spectra using DMSO-d₆ was acquired on an AVANCE III HD 800 spectrometer (Bruker BioSpin, Billerica, MA, USA). All spectra were measured at 298 K. All NMR data were derived from DQF-COSY, TOCSY, NOESY, HSQC, HMBC and H2BC experiments. Chemical shifts were reported in parts per million (ppm) relative to peaks of methanol-d₄ (δ_H 3.31 and δ_C 49.00 ppm) and DMSO-d₆ (δ_H 2.50 and δ_C 39.52 ppm).

Expression analysis

We extracted total RNA from flower petals 12 h before flower opening using an RNeasy mini kit (Qiagen, Hilden, Germany). We synthesized first strand cDNA from 200 ng of total RNA using ReverTra Ace qPCR Master Mix with a gDNA Remover kit (Toyobo, Osaka, Japan) for RT-PCR and a SuperScript III reverse transcriptase (Thermo Fisher Scientific, Waltham, MA, USA) for RT-qPCR. The I. nil γ-subunit of F₁F₀-ATP synthase (Ohnishi et al. 2005) was used as an internal control.

Thermal cycle condition used for the PCR amplification was 30 cycles of 95°C for 10 s, 62°C for 30 s and 72°C for 2 min. The primer sets we used were SYP8-7 Fw3 (5′-TACAACTTCCTCAGCTAGTTGT-3′) and SYP8-7 Rv4 (5′-GATCAACTGTAGGTTGAGAAGA-3′) for AmAS1, Am4CGT-Fw1 (5′-TGGGAGAAGAATACAAGAAAACA-3′) and Am4CGT-Rv1 (5′-ACGAGTGACCGAGTTGATGA-3′) for Am4′CGT, and ipgap-forward (5′- GCTTTAAGCCTCCGCCATGGG-3′) and ipgap-reverse (5′- ACGTTGGAAGCAATAAGCCCTTAAGCAG-3′) for the internal control.

We measured the expression levels of transgenes with a 7500 real-time PCR system (Thermo Fisher Scientific) using THUNDERBIRD SYBR qPCR mix (Toyobo) according to the manufacturer’s instructions. Standard curves were drawn using dilution series of pSFL304 and a cDNA clone of γ-subunit of F₁F₀-ATP synthase. The PCR condition was as follows: 95°C for 1 min, followed by 40 cycles of 95°C for 15 s and 60 °C for 1 min. The primer sets we used were SYP8-7 Fw5 (5′- AGGGGAAGAGGACAAAGACG-3′) and SYP8-7 Rv6 (5′- TTCTCGGAACCAAAGTGAC-3′) for AmAS1, Am4CGT-Fw2 (5′- GGAGTCGGTAAAGGGGAAAG-3′) and Am4CGT-Rv2 (5′-TTCTCCAAAGAAGCCAAGGA-3′) for Am4′CGT, and gATPase-Fw1 (5′-GATCGCCTCACCCTCACTTA-3′) and gATPase-R1 (5′-TTTAATCAGCCCTCCAATGC-3′) for the internal control.

Acknowledgments

We thank Seiko Watanabe, Ryoko Nakamura, Tomoyo Takeuchi, Kazuyo Ito, the late Chisato Matsuda and the late Chieko Nanba for technical assistance, and the NIBB Model Plant Research Facility and the NIBB Functional Genomics Facility for their technical support. We dedicate this work to the late Dr. Yukihide Abe who established the line 54Y.

Disclosures

The authors have no conflicts of interest to declare.