In this study, we attempted to develop a new biotechnological method for the efficient modification of floral traits. Because transcription factors play an important role in determining floral traits, chimeric repressors, which are generated by attaching a short transcriptional repressor domain to transcription factors, have been widely used as effective tools for modifying floral traits in many plant species. However, the overexpression of these chimeric repressors by the Cauliflower mosaic virus 35S promoter sometimes causes undesirable morphological alterations to other organs. We attempted simultaneously to generate new floral traits and avoid such quality loss by examining five additional floral organ-specific promoters, one Arabidopsis thaliana promoter and four Torenia fournieri promoters, for the expression of the chimeric repressor of Arabidopsis TCP3 ( AtTCP3 ), whose overexpression drastically alters floral traits but also generates dwarf phenotypes and deformed leaves. We found that the four torenia promoters exhibited particularly strong activity in the petals but not in the leaves, and that the combination of these floral organ-specific promoters with the chimeric repressor of AtTCP3 caused changes in the color, color patterns and cell shapes of petals, whilst avoiding other unfavorable phenotypes. Interestingly, each promoter that we used in this study generated characteristic and distinguishable floral traits. Thus, the use of different floral organ-specific promoters with different properties enables us to generate diverse floral traits using a single chimeric repressor without changing the phenotypes of other organs.
Molecular breeding has enabled the alteration of floral traits that are impossible to isolate by traditional cross-breeding. In Japan, genetically modified (GM) blue-colored flowers, such as blue roses ( Katsumoto et al. 2007 ) and blue carnations ( Tanaka et al. 2009 ), are now commercially available, and blue chrysanthemums ( Noda et al. 2013 ), blue dahlia and blue phalaenopsis (Dr. Masahiro Mii, Chiba University, personal communication) have also been generated, although these are not yet available on the market. The modification of floral traits other than flower color, such as petal shapes and color patterns, has also been attempted through molecular breeding, with transcription factors (TFs) having been used in various plant species ( Mitsuda et al. 2011a , Ohtsubo 2011 ).
TFs play important roles in the development of floral organs ( Soltis et al. 2007 , Ó’Maoiléidigh et al. 2013 ) and floral morphology, such as flower symmetry ( Preston and Hileman 2009 , Hileman 2014 ). The genome of Arabidopsis ( Arabidopsis thaliana ) contains 1,717 TF loci (PlantTFDB 3.0; http://planttfdb.cbi.pku.edu.cn/index.php?sp=Ath ; Jin et al. 2014 ), which control the expression of downstream target genes ( Mitsuda et al. 2009 ). To investigate the functions of plant TFs, plant-specific chimeric repressor technology (CRES-T; Hiratsu et al. 2003 ) has been applied to various ornamental crops, such as chrysanthemum, cyclamen, gentian, morning glory, rose and torenia ( Torenia fournieri ) ( Gion et al. 2011 , Nakatsuka et al. 2011 , Narumi et al. 2011 , Sage-Ono et al. 2011 , Shikata et al. 2011 , Tanaka et al. 2011 ). In CRES-T, a short repressor domain consisting of 12 amino acids (SRDX; Hiratsu et al. 2003 ) is attached to the C-terminal region of the plant TFs, which alters the transcriptional activators into dominant chimeric repressors ( Mitsuda et al. 2011b ) that suppress the functions of other redundant plant TFs ( Hiratsu et al. 2003 ). In torenia, 92 Arabidopsis chimeric repressors have previously been introduced for the modification of floral traits, such as color patterns, curled petal margins and wavy petals ( Shikata et al. 2011 ). However, some of these chimeric repressors also caused the alteration of leaf morphology. Among these was a chimeric repressor of Arabidopsis TCP3 ( AtTCP3 ), a member of the TEOSINTE BRANCHED1, CYCLOIDEA and PCF (TCP) gene family, which drastically altered the petal morphology to produce wavy petals with serrated margins ( Narumi et al. 2011 ). However, the overexpression of AtTCP3-SRDX caused an undesirable loss in quality, such as dwarf phenotypes and deformed leaves.
AtTCP3 is homologous to the CINCINNATA ( CIN ) gene of Antirrhinum , which is also a TCP transcription factor ( Martin-Trillo and Cubas 2010 ). CIN controls cell differentiation and growth in the petals and leaves of Antirrhinum ( Nath et al. 2003 , Crawford et al. 2004 ). Twenty-four TCP genes have been identified in the Arabidopsis genome, which are classified into two classes (I and II) according to the amino acid sequence of the TCP domain ( Martin-Trillo and Cubas 2010 ). Each single mutant of the CIN -like TCP genes did not affect the phenotype, but quintuple mutants in which TCP3 , TCP4 , TCP5 , TCP10 and TCP13 genes were targeted showed severe defects in leaf formation ( Koyama et al. 2010 ), and serrated and wavy petals ( Koyama et al. 2011 ). Therefore, redundant CIN-like TCPs complement the functions of other CIN-like TCPs, even though a mutation was caused in only one TCP gene. Conveniently, AtTCP3-SRDX has been found to be effective in generating phenotypes that are similar to the multiple mutant of the CIN -like TCP genes in Arabidopsis ( Koyama et al. 2007 , Koyama et al. 2010 ). In addition, a recent study revealed that CIN -like TCP5 also participates in petal growth in Arabidopsis ( Huang and Irish 2015 ).
The aim of this study was to develop a new biotechnological method using a chimeric repressor for generation of useful floral traits without affecting the morphology of other organs. Since a change in the morphology of vegetative organs also affects the cultivation method required, it is desirable to develop a molecular breeding technique that only alters the floral organs, allowing many different cultivars to be cultivated using the same methods. Since the overexpression of some TFs that are under the control of the 35S ( Cauliflower mosaic virus 35S) promoter has been shown to result in undesirable morphological alterations of structures other than the floral organs ( Narumi et al. 2011 , Shikata et al. 2011 ), we used floral organ-specific promoters for the expression of a chimeric repressor. In addition, it was expected that various floral traits, such as petal shapes, colors and color patterns, could be generated through the combination of a single chimeric repressor and several floral organ-specific promoters with different expression timings, levels and positions. AtTCP3 was considered a good candidate for testing this idea because it has previously been shown that expression of AtTCP3-SRDX by the 35S promoter generates various floral traits in each transgenic line ( Narumi et al. 2011 ). Therefore, in this study, the AtTCP3-SRDX gene was expressed in torenia using six promoters: four floral organ-specific promoters of torenia, and the 35S and Arabidopsis APETALA1 ( AtAP1 ) promoters, which we have reported previously ( Sasaki et al. 2011 ). We then examined whether the combination of floral organ-specific promoters and AtTCP3-SRDX could be used to generate novel floral traits without producing undesirable phenotypes through the alteration of other organs.
Temporal expression patterns of floral organ-specific genes in torenia
To isolate floral organ-specific promoters in torenia, we focused on the class-B MADS-box genes and anthocyanin biosynthesis-related genes. The torenia class B genes DEFICIENS ( TfDEF ) and GLOBOSA ( TfGLO ) play important roles in the development of petals and are specifically expressed in the floral organs of torenia ( Sasaki et al. 2010 , Sasaki et al. 2014 ). Similarly, the torenia anthocyanin biosynthesis-related genes, dihydroflavonol 4-reductase ( TfDFR ) and flavanone hydroxylase ( TfF3H ), are also specifically expressed in the floral organs ( Sasaki et al. 2010 ). The expression of these four genes in various torenia organs was first examined using reverse transcription–PCR (RT–PCR) analyses ( Supplementary Fig. S3 ), which showed that their expression was floral organ specific, confirming our previous findings ( Sasaki et al. 2010 , Sasaki et al. 2014 ). However, we also found that they had different expression patterns from each other in several floral organs. The expression of TfDEF was only detected in the floral organs and was particularly strong in the petals. TfGLO had a similar expression pattern to TfDEF but was not expressed in the carpels. TfDFR was only expressed in immature petals, while TfF3H was expressed in immature and mature petals and stamens.
To examine further the temporal expression patterns of these genes in the petals, the growth of torenia petals was divided into eight stages (from −1 to 6; Fig. 1 A). These stages were mainly classified according to petal size, but stage 0 was defined as flowers in which the stamens were not separated from the petals, and so experimental materials from stages −1 and 0 contained immature stamens. The mRNA accumulation of these genes was examined by quantitative real-time-PCR (qRT-PCR) analysis, using petals from these eight growth stages ( Fig. 1 B). This demonstrated that these four genes differed in their expression patterns and in the amount of their transcripts in the petals. The expression of TfDEF and TfGLO was found in every stage, but there was greater accumulation of the TfDEF transcript than the TfGLO transcript. In contrast, the expression of TfDFR and TfF3H was not detected at stages −1 and 0, and the expression of TfDFR was only transiently detected from stages 1 to 5. Thus, these analyses revealed that the four torenia genes have different expression properties.
Expression characteristics of the floral organ-specific promoters
As outlined in the previous section, the four torenia genes exhibited different expression properties in torenia floral organs ( Fig. 1 ; Supplementary Fig. S3 ), and so we hypothesized that the utilization of these promoters may be effective for generating novel floral traits without changing the phenotype of other organs. Therefore, we examined their promoter activities and compared them with those of the 35S and AtAP1 promoters, which have previously been analyzed in torenia ( Sasaki et al. 2011 ). To do this, the promoter regions of the TfDEF (2,083, bp), TfGLO (1,394, bp), TfDFR (1,786, bp) and TfF3H (2,007, bp) genes were isolated and connected to the reporter β-glucuronidase gene ( GUS ) ( Fig. 2 A). The promoter activities were then compared by analyzing the GUS activity in the petals and leaves of transgenic torenia plants ( Fig. 2 B).
The analysis of GUS activities revealed that these four torenia promoters specifically functioned in the petals of torenia ( Fig. 2 B, left), and that TfDEF , TfDFR and TfF3H exhibited higher promoter activity than the 35S and AtAP1 promoters in the petals. In contrast, TfGLO had similar promoter activity levels to the 35S and AtAP1 promoters in the petals. No activity of these four torenia promoters was detected in the leaves ( Fig. 2 B, right). Thus, the activity of these four torenia promoters was more strictly restricted to the floral organs than that of the AtAP1 promoter, which we previously considered a floral organ-specific promoter ( Sasaki et al. 2011 ).
Spatial expression of the floral organ-specific promoters in petals
To visualize the spatial expression patterns of these six promoters in torenia, histochemical GUS analysis was performed using the flowers and leaves of transgenic plants ( Fig. 3 ). The GUS staining in the flowers demonstrated that all six promoters were active in the petal lips but were not active in the tube region of petals ( Fig. 3 , Flower and Flower side view). Strong GUS staining in the petal lips was particularly observed in the TfDEF , TfGLO , TfDFR and TfF3H torenia promoters, all of which were similar to each other. For the TfDEF and TfGLO promoters, strong GUS staining was also detected in the stamens ( Fig. 3 , red arrowheads), which is consistent with the expression patterns observed in the RT–PCR analysis ( Supplementary Fig. S3 ). In addition, the 35S promoter induced strong GUS staining and the AtAP1 promoter induced faint GUS staining in the leaves of transgenic torenia plants ( Fig. 3 , Leaf). However, no GUS staining was detected in the leaves when the four torenia promoters were used, which is also consistent with the results of the RT–PCR analysis ( Supplementary Fig. S3 ). Despite the lack of difference in the GUS staining patterns in the mature petal lips, it is possible that promoter-specific (spatiotemporal) characteristics may be observed by monitoring differences in activity during petal development.
We then examined differences in the GUS staining patterns at the cellular level in the petals, because these promoters were expected to be active in different layers of cells. Previous studies have shown that the transcripts of TfDEF and TfGLO occur in all of the petal cell layers of the floral meristem during the early development of petals in torenia ( Niki et al. 2012 ), while anthocyanin-related genes are generally expressed in the epidermal cells ( Jackson et al. 1992 , Martin and Gerats 1993 ). Thus, cross-sections of the GUS-stained petals were further observed with a digital microscope ( Fig. 3 , Petal cross-section). With the 35S promoter, GUS staining was weak but detected in all sections of the petal, particularly the vascular systems (red arrowheads, Supplementary Fig. S4 ). With the AtAP1 and TfGLO promoters, the petal tissues were also entirely stained, with the GUS staining detected in both the epidermal and parenchyma tissues. In contrast, with the TfDEF , TfDFR and TfF3H promoters, GUS staining was mainly observed in the epidermal tissues. The result for the TfDEF promoter was unexpected, as we predicted that this promoter would be active in all of the petal tissues; however, it is possible that different expression patterns would be observed if we used different petal growth stages in the analysis. These observations indicated that the expression patterns of these six promoters could be classified into two main types: 35S, AtAP1 and TfGLO promoters, which were active in all cell layers; and TfDEF , TfDFR and TfF3H promoters, which were mainly active in the epidermal layer.
Floral traits generated by each floral organ-specific promoter
We previously reported that overexpression of AtTCP3 - SRDX induced changes to not only the floral traits of torenia and chrysanthemum but also plant posture and growth ( Narumi et al. 2011 ). Therefore, in the present study, floral organ-specific promoters were used to express AtTCP3 - SRDX to avoid these undesirable phenotypes in torenia and these were compared with the 35S and AtAP1 promoters ( Fig. 4 A). These six chimeric repressor constructs were introduced into wild-type torenia ( Fig. 4 B), resulting in the generation of > 30 transgenic plants per construct, and the changes in floral traits were then observed ( Supplementary Fig. S5 ). Interestingly, these six constructs exhibited different floral traits from each other, despite all of the transgenic plants expressing the same AtTCP3 chimeric repressor ( Fig. 4 C; Supplementary Fig. S5 ). In contrast, consistent changes in the floral phenotype were observed within the same constructs, although there was some variation in petal color. The main characteristics of the floral organs of transgenic plants containing each of these promoters were as follows: (i) 35S promoter—finely serrated petal margins, scraped and variegated color patterns; (ii) AtAP1 promoter—roughly serrated petal margins and pale coloration; (iii) TfDEF promoter—wavy, fuzzy borders and faintly colored; (iv) TfGLO promoter—roughly serrated petal margins and partially decolorized border; (v) TfDFR promoter—narrowed border and scraped; and (vi) TfF3H promoter—wavy petal margins, faint coloration and color variation. Each transgenic line stably bears the flowers with identical phenotype all through the flowering periods under repetitive vegetative reproduction, while the transgenic lines expressing the same promoter construct exhibit various petal hues with the same tendency as shown in Fig. 4 and Supplementary Fig. S5 .
Despite the wide variation in flower colors and shapes, no morphological changes were observed in the leaves of transgenic torenia plants expressing AtTCP3 - SRDX with the floral organ-specific torenia promoters ( Fig. 4 D). It was previously shown that overexpression of AtTCP3 - SRDX caused serrated margins not only in the petals but also in the leaves of torenia, chrysanthemum, rose and Arabidopsis ( Koyama et al. 2007 , Gion 2011 , Narumi et al. 2011 ). Similarly, in the present study, 35Spro: AtTCP3 - SRDX resulted in both moderate and severe changes in the phenotypes of the leaves ( Fig. 4 D), as well as dwarf phenotypes ( Supplementary Figs. S6 and S7 ); and plants containing the AtAP1 promoter also occasionally exhibited a moderate serrated margin phenotype. In contrast, the floral organ-specific expression of AtTCP3 - SRDX had no effect on the growth of transgenic torenia plants ( Supplementary Fig. S6 ). These results demonstrate that the utilization of floral organ-specific promoters for the expression of AtTCP3 - SRDX generated diverse floral phenotypes without affecting the phenotypes of other organs in torenia.
Epidermal cell shapes in the AtTCP3-SRDX transgenic plants
As outlined above, the use of six different types of promoters for the expression of AtTCP3 - SRDX resulted in distinctive floral traits in torenia ( Fig. 4 C; Supplementary Fig. S5 ). Therefore, we examined the shape of the cells within the petals in each transgenic plant using scanning electron microscopy (SEM) to determine whether this correlated with the distinctive floral traits. We also examined the transgenic lines of TfDEF pro: AtTCP3 - SRDX and TfF3H pro: AtTCP3 - SRDX plants, because the flowers generated with these two promoter constructs exhibited very different colored petals (purple and pale purple, in contrast to dark violet in the wild type).
The epidermis of the petal lips of wild-type torenia generally contained cone-shaped cells ( Fig. 5 A; Sasaki et al. 2010 , Sasaki et al. 2014 ). The 35Spro: AtTCP3 - SRDX and AtAP1 pro: AtTCP3 - SRDX plants also contained some large papilla-shaped cells in the same region, and the number of the cells was larger in the 35S construct ( Fig. 5 B, C). Interestingly, two of the lines of TfDEF pro: AtTCP3 - SRDX plants (lines-5 and -15) exhibited different epidermal cell shapes in the petals ( Fig. 5 D): line-15, which had pale purple-colored petals, contained cone-shaped cells that were the same as those of wild-type torenia ( Fig. 5 D; right), while line-5, which had purple-colored petals, contained dome-shaped cells in the petals ( Fig. 5 D; left). The TfGLO pro: AtTCP3 - SRDX plants also exhibited large papilla-shaped cells but at a lower number than in 35Spro: AtTCP3 - SRDX plants ( Fig. 5 E), while the epidermal cells of the petals in the TfDFR pro: AtTCP3 - SRDX plants were cone shaped ( Fig. 5 F). The TfF3H pro: AtTCP3 - SRDX plants also exhibited two different shapes of epidermal cells in two transgenic lines (lines-13 and -16): line-13, which had pale purple-colored petals, contained cone-shaped cells ( Fig. 5 G; left), while line-16, which had purple-colored petals, exhibited dome-shaped cells ( Fig. 5 G; right). Several independent transgenic lines of TfDEF pro: AtTCP3 - SRDX and TfF3H pro: AtTCP3 - SRDX plants with purple-colored petals also exhibited dome-shaped cells in their petals ( Supplementary Fig. S8 ). Thus, SEM analysis revealed that the combination of floral organ-specific promoters and AtTCP3-SRDX resulted in alteration of the shape of petal lip cells to produce cone-shaped, papilla-shaped and dome-shaped cells, which were partially correlated with the petal traits ( Supplementary Fig. S9 ).
Expression of anthocyanin biosynthesis-related genes in the transgenic torenia plants
The six promoters used in this study affected not only the morphology but also the color of transgenic torenia plants ( Fig. 4 B), despite all of these plants expressing the AtTCP3-SRDX transgene. The TfDEF pro: AtTCP3 - SRDX and TfF3H pro: AtTCP3 - SRDX plants were particularly different, exhibiting purple- and pale purple-colored petals. We hypothesized that these different colored phenotypes were linked to differences in the expression levels of AtTCP3-SRDX and/or anthocyanin biosynthesis-related genes. Therefore, we examined the expression of AtTCP3-SRDX and the torenia anthocyanin biosynthesis-related genes chalcone synthase ( TfCHS ), TfF3H and chalcone isomerase ( TfCHI ) using qRT-PCR analysis ( Fig. 6 ). For this analysis, we used four independent transgenic lines of the TfDEF pro: AtTCP3 - SRDX and TfF3H pro: AtTCP3 - SRDX plants, including purple- and pale purple-colored petal lines, as well as three independent transgenic lines of 35Spro: AtTCP3-SRDX .
The expression of the AtTCP3-SRDX transgene was first examined using flower buds ( Fig. 6 A), which showed that all of the transgenic torenia plants contained similar amounts of the AtTCP3-SRDX transcript, with the exception of TfF3H pro: AtTCP3 - SRDX -13. The TfF3H pro: AtTCP3 - SRDX -13 plants contained very low levels of the transgene transcript, despite the introduction of the transgene being confirmed by genomic PCR, and the AtTCP3 - SRDX transgene appearing to have been expressed when the mRNA accumulation of the transgene was further confirmed by expanding the data of wild-type torenia and TfF3H pro: AtTCP3 - SRDX -13 shown in Fig. 6 A ( Supplementary Fig. S10 ). All three types of transgenic plants contained lower amounts of mRNA of the TfCHS , TfF3H and TfCHI genes than the wild-type torenia. However, the amount of AtTCP3-SRDX mRNA did not appear to be correlated with that of the anthocyanin biosynthesis-related genes. In addition, the amount of these transcripts was not correlated with the pale-colored phenotype in TfDEF pro: AtTCP3 - SRDX -15 and TfF3H pro: AtTCP3 - SRDX -13, and a severe phenotype observed in 35Spro: AtTCP3 - SRDX- 21 ( Supplementary Fig. S7 ). In contrast, the expression of the class B genes TfDEF and TfGLO , which are important for petal identity and development ( Soltis et al. 2007 , Ó’Maoiléidigh et al. 2013 ), was not affected by the AtTCP3 chimeric repressor ( Supplementary Fig. S11 ). The qRT-PCR analysis indicated that AtTCP3-SRDX expression reduced mRNA accumulation of the anthocyanin biosynthesis-related genes. However, the phenotypic severity and differences were not explained by the amount of the AtTCP3-SRDX transcript. Further analysis of the localized expression of the transgene and/or the accumulation of the transgene protein at various petal growth stages may elucidate this relationship.
The expression of TfF3H was not detected in TfF3H pro: AtTCP3 - SRDX -13 ( Fig. 6 C), and the level of transgene expression did not appear to be correlated with the alteration of the petal color phenotype. These findings were different from those observed for the other transgenic plants generated in this study, and so we further examined the TfF3H pro: AtTCP3 - SRDX -13 plants. We hypothesized that the reduction in expression of the transgene ( Fig. 6 A) and endogenous TfF3H ( Fig. 6 C) was caused by transcriptional gene silencing due to methylation of the TfF3H promoter in the TfF3H pro: AtTCP3 - SRDX -13 plants ( Eamens et al. 2008 ). Methylation of the TfF3H promoter region was confirmed by the observation that the TfF3H promoter was not digested by methylated DNA-sensitive restriction enzymes ( Supplementary Fig. S12 ). Bisulfite sequencing analysis targeting a part of the TfF3H promoter (−1,730 to − 1,490 bp) also revealed that DNA hypermethylation occurred in the TfF3H pro: AtTCP3 - SRDX -13 plants ( Supplementary Fig. S13 ). The cytosine of the CG motif, which is maintained by METHYLTRANSFERASE1 ( MET1 ) (Mathieu et al. 2007), was highly methylated in the TfF3H promoter of both wild-type torenia and TfF3H pro: AtTCP3 - SRDX line-13 ( Table 1 ). However, the other cytosines in the TfF3H promoter in TfF3H pro: AtTCP3 - SRDX -13 were also highly methylated compared with wild-type torenia. Thus, methylation of the TfF3H promoter region in TfF3H pro: AtTCP3 - SRDX -13 caused the gene silencing of TfF3H and the pale purple-colored phenotype in the petals.
|CG (%)||Other C (%)|
|TfT3H pro: AtTCP3-SRDX line-13||100.00||90.00|
|CG (%)||Other C (%)|
|TfT3H pro: AtTCP3-SRDX line-13||100.00||90.00|
Generation of diverse floral traits without affecting growth properties
In this study, we established a method for simultaneously generating many diverse floral traits through the expression of a single chimeric repressor in combination with various floral organ-specific promoters that exhibit different expression characteristics. As can be seen in Fig. 4 and Supplementary Fig. S5 , a surprisingly wide range of novel floral traits was obtained through the combination of these floral organ-specific promoters and the AtTCP3-SRDX chimeric repressor. So the question is, how does the expression of a single chimeric repressor result in such diverse floral traits? One of the major reasons was that the six promoters used (four of which were floral organ-specific genes) had different characteristics from each other in terms of the timing of expression ( Fig. 1 B), the expression levels ( Figs. 1 B, 2B), the expression patterns in the floral organs ( Supplementary Fig. S3 ) and the expression site in the tissue layers of the petals ( Fig. 3 ). Thus, these floral organ-specific promoters exhibited temporospatial variation in the transient suppression of the functions of torenia homologs for AtTCP3 at the time of development and formation of the floral organs, causing diversity in various floral traits, including petal colors, color patterns and cell shapes ( Figs. 4, 5 ; Supplementary Fig. S5 ). Previous reports showed that utilization of a petal-specific promoter for expression of a chimeric repressor avoided unfavorable phenotypes in other organs in Arabidopsis ( Azuma et al. 2016 ) and torenia ( Sasaki et al. 2011 ); however, such diversity in floral phenotypes was not observed in these transgenic plants. In addition, such diversity in floral phenotypes could not be obtained through mutation. This study may also provide a novel analytical method for studying other TFs with temporospatially specific functions, which were not analyzed in the present study due to the function of these genes being entirely knocked out in the mutants.
Different floral traits were also generated within the same constructs ( Fig. 4 C; Supplementary Fig. S5 ), although the type of alteration to their floral phenotypes was similar. The relationship between the degree of phenotypic change and the expression levels of the AtTCP3-SRDX transgene was not observed in qRT-PCR analysis; however, this analysis was performed using the immature petals of only one developmental stage. Therefore, further detailed analyses of the localized expression of the transgene at the tissue or cellular level and/or of protein accumulation of the transgene at various developmental stages may elucidate this relationship. The partial phenotypic differences within the same construct may result from differences in the insertion site of the transgene at a level that affected the transgene expression level and/or timing but did not fundamentally change the properties of the promoter.
The six promoters harbored different patterns in terms of the activities in different layers of petal cells ( Fig. 3 ). These patterns were correlated with the epidermal cell shapes ( Fig. 5 ) in the petals of AtTCP3-SRDX transgenic torenia ( Supplementary Fig. S9 ). The 35S , AtAP1 and TfGLO promoters were active in all of the cell layers in the petals, and the expression of AtTCP3-SRDX under the control of these promoters resulted in papilla-shaped cells. In contrast, the TfDEF and TfF3H promoters were mainly active in the epidermal cells, and the expression of AtTCP3-SRDX under the control of these promoters caused dome-shaped cells. Some of the transgenic lines of TfDEF pro: AtTCP3-SRDX and TfF3H pro: AtTCP3-SRDX , whose petals were a pale purple color, and TfDFR pro: AtTCP3-SRDX transgenic plants exhibited normal cell shapes in their petals. Although the diversity of floral traits observed in this study could not be entirely explained, the range of petal cell shapes caused by the promoters used in this study was partly related to the cell layer in which the transgene was expressed as a result of the promoters.
The aim of the present study was to find floral organ-specific promoters for the expression of AtTCP3 - SRDX that did not result in undesirable phenotypes being generated in structures other than the floral organs. We found that the combination of floral organ-specific promoters and AtTCP3-SRDX not only led to the generation of diverse new floral traits ( Fig. 4 C; Supplementary Fig. S5 ) but also avoided these undesirable phenotypes in the vegetative organs ( Fig. 4 D; Supplementary Fig. S6 ). Phenotypic alterations in structures other than the floral organs, such as is caused by cross-fertilization, would require the cultivation methods that were used for the parental variety to be optimized for the new cultivars. The establishment of a cultivation method for new cultivars requires great care and increases the cultivation costs, and so it is important to save on such efforts and costs when producing new flower cultivars. We have previously reported that the AtAP1 promoter is preferentially expressed in the floral organs when compared with the 35S promoter ( Sasaki et al. 2011 ). However, the promoter activities of the four torenia genes TfDEF , TfGLO , TfDFR and TfF3H in the floral organs were more specific than those of the AtAP1 promoter, because their activities were not detected in the leaves ( Fig. 2 B). The GUS activity provided by the AtAP1 promoter construct was low but detectable in the leaves, and this low activity corresponded to the serrated but mild leaf phenotype caused by AtTCP3-SRDX ( Fig. 4 D). Therefore, the floral organ-specific promoters TfDEF , TfGLO , TfDFR and TfF3H would be more effective for avoiding undesirable phenotypes in other organs, allowing the same cultivation method to be used as for the parental variety. The InMYB1 promoter from Japanese morning glory exhibited petal-specific activity in several floricultural crops ( Azuma et al. 2016 ), and the promoter may be also effective for the similar usage in torenia.
Why did the expression of AtTCP3-SRDX cause diverse floral phenotypes in torenia?
The expression of AtTCP3-SRDX under the control of the six promoters provided diverse floral traits in torenia ( Fig. 4 C; Supplementary Fig. S5 ), which was surprising given that only one chimeric repressor was used. It has already been mentioned that one of the major reasons for this diversity was the different characteristics of the promoters (see above). Another reason would be the properties of AtTCP3 itself.
Two possible factors may affect phenotypical diversity when using AtTCP3: redundancy of the TCP genes and multiple functions of the TCPs. The Arabidopsis genome contains 24 TCP genes, eight of which, including AtTCP3 , are classified as CIN -like TCP genes ( Martin-Trillo and Cubas 2010 ), which have several functions in the petals of Arabidopsis and Antirrhinum . For example, CIN controls cell differentiation and growth in the petals, and a cin mutant of Antirrhinum exhibited flattened epidermal cells instead of conical cells in the petals and smaller petal lobes ( Crawford et al. 2004 ), while multiple CIN -like TCP mutants of Arabidopsis exhibited serrated and wavy petals ( Koyama et al. 2011 ). The CIN -like TCP genes also suppress the expression of the CUP-SHAPED COTYLEDON ( CUC ) genes through the promotion of two different networks, miR164s expression and ASSYMETRIC LEAVES1 ( AS1 ) expression ( Specht and Howarth 2014 ). In addition, the CIN -like TCP mutants exhibited severe, dose-dependent morphological changes in the leaves ( Koyama et al. 2010 ). This demonstrates that several redundant TCPs have multiple functions, which play important roles in the development of the petals and/or leaves not only in Arabidopsis but also in torenia. Although these multiple functions could not be entirely confirmed and distinguished through mutant analysis, these TCPs may play redundant but partially different roles by changing the timing and region of expression in the petals.
In this study, the use of six promoters with different characteristics ( Figs. 2, 3 ) will have contributed to the temporospatial variation in the suppression of these TCP functions in torenia; and the diverse expression timings, levels and positions of AtTCP3-SRDX will have affected the endogenous torenia TCPs that play important roles in cell differentiation and growth in different ways, altering the cellular state of the transgenic torenia plants. The alteration of the expression of other TFs that play important roles in the development and formation of the floral organs would be also useful for the generation of many diverse floral traits.
Even when the same promoter construct was introduced, two different cell shapes were observed in the petals of TfDEF pro: AtTCP3 - SRDX plants: dome-shaped cells and normal cone-shaped cells ( Fig. 5 E). In the Antirrhinum CIN mutant, the epidermal cells of the petals were flattened ( Crawford et al. 2004 ), which is similar to the dome-shaped cells observed in TfDEF pro: AtTCP3 - SRDX -5. However, the TfDEF pro: AtTCP3 - SRDX -15 plants exhibited the same petal cell shapes as wild-type torenia ( Fig. 5 ), despite line-15 appearing to generate a more severe phenotype based on the petal color in this line being paler than that of line-5 ( Fig. 5 D). Thus, the AtTCP3 chimeric repressor multilaterally suppressed the functions of the torenia TCP homologs, resulting in two types of cell shapes and colors being generated by the same construct.
Lines TfDEF pro: AtTCP3 - SRDX -5 and -15 produced similar amounts of the AtTCP3 - SRDX -5 transcript but may have differed in the timing of expression of the transgene. The transgenic torenia used in this study will have had different mechanisms and/or developmental timing of action for generating petal colors and petal morphology, resulting in various petal colors being exhibited within the same construct despite the type of change to the floral phenotype being fundamentally the same ( Fig. 4 C; Supplementary Fig. S5 ). Because the TCPs control cell differentiation, growth and division ( Crawford et al. 2004 , Specht and Howarth 2014 ), the color and morphology of petals may be affected by different timings and/or levels of suppression of the TCP functions.
Translational efficiency of floral organ-specific promoters in torenia
The TfDEF and TfF3H promoters did not produce significantly different levels of transgene transcripts in the petals from the 35S promoter ( Fig. 6 A). The amount of transgene transcript that was accumulated with the TfDEF promoter was almost the same as that with the 35S promoter, while the amount accumulated with the TfF3H promoter was 1.7- to 2.7-fold higher than that with the 35S promoter. In contrast, the GUS activity levels provided by these floral organ-specific promoters were 10- to 25-fold higher than those of the 35S promoter in the petals ( Fig. 2 B). These findings suggest that the transcripts of the transgenes derived from these TfDEF and TfF3H promoters were more efficiently translated in the petals than those derived from the 35S promoter.
These floral organ-specific promoters may contain not only a floral organ-specific cis -element but also a translational enhancing sequence. In this study, effective translational enhancers, such as an omega element ( Gallie 1993 ) and/or the 5'-untranslated region (UTR) of alcohol dehydrogenase (ADH; Satoh et al. 2004 , Sugio et al. 2008 ), were not attached to any of the promoter constructs, including the 35S promoter construct. Therefore, the translational efficiency of each promoter could be evaluated by comparing the ratio between the transcription level of the transgene ( Fig. 6 A) and the GUS activity level ( Fig. 2 B). Since the TfDEF and TfF3H promoters did not contain a homologous sequence to a known translational enhancer, such as the 5'-UTR of ADH , these promoters may contain new translational enhancing sequences.
We also found that the promoter activity of TfDFR was >20-fold higher than that of the 35S promoter ( Fig. 2 B). The expression level of the TfDFR gene in the petals of wild-type torenia was not as high as those of TfF3H and TfDEF ( Fig. 1 B), but a large amount of protein was still accumulated as a result of the TfDFR promoter, suggesting that this promoter may also contain translational enhancing sequences. Future analysis may elucidate the translational enhancers in these floral organ-specific promoters or their 5'-UTR. These results also suggested that these floral organ-specific promoters would be useful for accumulating a protein of interest without affecting plant traits other than the floral organs.
Future perspectives for organ specific promoters in the study of plant TFs
The use of floral organ-specific promoters in combination with a chimeric repressor may also be useful for studying the function of TFs, because the phenotypes observed in the transgenic plants generated by this method would indicate the temporospatially confined functions of the TF of interest. The accumulation of these temporospatially specific functions of a particular TF at a particular time could then be combined to understand the total function of the TF. Mutant phenotypes of a particular TF would be derived from a loss of the total function of the TF.
In this study, floral organ-specific promoters were utilized, which provided information about the temporospatially specific functions of plant TFs at the time of development and formation of the floral organs. The introduction of the AtTCP3 chimeric repressor caused the alteration of floral traits such as petal color, petal color patterns and petal cell shapes in torenia, suggesting that torenia TCP3 homologs have temporospatially multilateral functions during the development and formation of the floral organs. In contrast, these diverse floral traits would not be produced in the mutants. If a different combination of organ-specific promoters and TFs was used, the function of endogenous TFs other than torenia TCPs would also be temporospatially suppressed by the chimeric repressors whose expression was controlled by these specific promoters.
In this study, we established a novel and useful molecular breeding method for simultaneously generating many diverse floral traits through the combination of floral organ-specific promoters and a single chimeric repressor which may represent a novel analysis tool for studying the functions of TFs in the near future.
Materials and Methods
Torenia fournieri Lind. ‘Crown Violet’ was used as the wild-type torenia plant. Transgenic torenia plants containing a GUS coding region (S69414) that is under the control of either the 35S promoter or the 1.7 kbp AtAP1 promoter have previously been generated ( Sasaki et al. 2011 ). The wild-type and transgenic torenia plants were maintained as described in Yamaguchi et al. (2011) .
Isolation of floral organ-specific promoters of torenia
In this study, the 35S promoter and five promoters derived from plant genes, AtAP1 (AT1G69120), TfDEF (AB359951), TfDFR (AB548587), TfGLO (AB548150) and TfF3H (AB548588), were used for expression of the transgenes. The AtAP1 and TfGLO (1,394, bp) promoters have been isolated previously ( Sasaki et al. 2008 , Sasaki et al. 2011 ). Therefore, in this study, we isolated the TfDEF , TfDFR and TfF3H promoters.
Since the sequences of the 5'-UTR of TfDFR and TfF3H had not been determined for the torenia cultivar that we used, we first isolated the highly conserved regions of the two genes using RT–PCR and then isolated their 5'-UTR using the 5'RACE (rapid amplification of cDNA ends) method. The sequences of the primers for the first RT–PCR ( Supplementary Table S1 ) were designed based on the sequences of TfDFR (AB012924) and TfF3H (AB211958), which have been isolated from related torenia cultivars. The sequences of the primers for the 5'RACE PCR ( Supplementary Table S2 ) were then designed based on the sequences that were isolated using the first RT–PCR analysis. For the 5'RACE method, total RNA was isolated from immature petals using TRIzol® (Invitrogen). cDNA was then synthesized from the total RNA using a cDNA synthesis kit (TOYOBO).
The promoter regions of TfDEF , TfDFR and TfF3H were isolated using a GenomeWalker™ kit (Clontech Laboratories, Inc.) according to the manufacturer’s instructions using gene-specific primers ( Supplementary Table S3 ), the sequences of which were designed based on the sequences of the 5'-UTR of each gene. Torenia genomic DNA was prepared from the leaves using an ISOPLANT II kit (Nippon Gene Co., Ltd.) to obtain a genomic PCR template for isolation of the promoter. This resulted in isolation of the promoter regions of TfDEF (2,083, bp), TfDFR (1,786, bp) and TfF3H (2,007, bp).
Plasmid construction for the generation of transgenic torenia plants
The promoter regions of the four torenia genes TfDEF , TfGLO , TfDFR and TfF3H were amplified using individual primer sets ( Supplementary Table S4 ). The amplified fragments were digested with Hin dIII and Bam HI, and cloned into the corresponding site of p35SSRDX ( Mitsuda et al. 2006 ) to produce p TfDEF proSRDX, p TfGLO proSRDX, p TfDFR proSRDX and p TfF3H proSRDX. The coding region of GUS (S69414) was then digested with Bam HI and Sac I of the binary vector pBI121 and cloned into the corresponding site of p TfDEF proSRDX, p TfGLO proSRDX, p TfDFR proSRDX and p TfF3H proSRDX to produce the p TfDEF pro: GUS , p TfGLO pro: GUS , p TfDFR pro: GUS and p TfF3H pro: GUS plasmids, respectively. A coding region of AtTCP3 (AT1G53230) was amplified using primer sets ( Supplementary Table S4 ). The amplified fragment was then digested with Bam HI and Sac I and cloned into the corresponding sites of p35SSRDX ( Mitsuda et al. 2006 ), p AP1 proSRDX ( Sasaki et al. 2011 ), p TfDEF proSRDX, p TfGLO proSRDX, p TfDFR proSRDX and p TfF3H proSRDX to produce p35S AtTCP3 - SRDX , p AP1 pro: AtTCP3 - SRDX , p TfDEF pro: AtTCP3 - SRDX , p TfGLO pro: AtTCP3 - SRDX , p TfDFR pro: AtTCP3 - SRDX and p TfF3H pro: AtTCP3 - SRDX plasmids, respectively. Following confirmation of the amplified sequences, the region corresponding to each transgene was transferred into the plant binary vector pBCKK using the Gateway system (Invitrogen) to produce pBCKK- TfDEF pro: GUS , pBCKK- TfGLO pro: GUS , pBCKK- TfDFR pro: GUS , pBCKK- TfF3H pro: GUS , pBCKK-35S: AtTCP3 - SRDX , pBCKK- AP1 pro: AtTCP3 - SRDX , pBCKK- TfDEF pro: AtTCP3 - SRDX , pBCKK- TfGLO pro: AtTCP3 - SRDX , pBCKK- TfDFR pro: AtTCP3 - SRDX and pBCKK- TfF3H pro: AtTCP3 - SRDX binary vectors, respectively.
Generation of transgenic torenia plants
The binary vectors were introduced into the Agrobacterium tumefaciens strain EHA105 by electroporation, which was then used to transform torenia leaf disks. The transgenic torenia plants were then screened and regenerated using previously described methods ( Aida and Shibata 1995 , Aida 2008 ). The expression of the transgenes was then checked ( Supplementary Figs. S1, S2 ), following which these transgenic torenia plants were used for this study.
Expression analysis using quantitative real-time RT-PCR
Total RNA was isolated from mature and immature petals using TRIzol (Invitrogen), and cDNA was synthesized from the total RNA using a cDNA synthesis kit (TOYOBO). qRT-PCR was performed using SYBR® Premix Ex Taq™ II (TIi RNaseH Plus; TAKARA) and signals were detected on the Thermal Cycler Dice® Real Time System TP800 (TAKARA) according to the manufacturer’s instructions. Gene-specific regions, including the partial 3'-UTR, were utilized to determine the primer sequences for the qRT-PCR ( Supplementary Table S5 ).
Histochemical and fluorometric GUS assays
GUS activity was histochemically and fluorometrically analyzed according to Kosugi et al. (1990) , with the following modifications. For histochemical GUS staining, plant tissues were incubated in the GUS reaction mixture, which contained 1 mM 5-bromo-4-chloro-3-indolyl-β-d-glucuronide, 50 mM potassium phosphate buffer (pH 7.0), 10−20% (v/v) methanol and 1 mM dithiothreitol (DTT) at 37 °C for approximately 16–20 h. The reaction was then stopped by replacing the GUS reaction buffer with 70% ethanol, and the pigments and Chls were removed using repeated 70% ethanol treatments.
For the quantitative analyses, each plant organ was homogenized in the GUS assay buffer (50 mM potassium phosphate, 10 mM EDTA, 0.1% Triton X-100, 0.1% Sarkosyl and 2 mM DTT), and an aliquot of the supernatant was incubated in the buffer with 4-methylumbelliferyl-β-d-glucuronide as a substrate at 37 °C for 30 min. The amount of 4-methylumbelliferone formed in each GUS reaction was then determined using a fluorescence spectrophotometer (VersaFluor™ fluorometer; Bio-Rad), and the protein concentration was determined using a Coomassie protein assay kit (Bio-Rad) with bovine serum albumin (BSA) as the standard.
GUS-stained petals were cut lengthwise using a Microslicer™ DTK-1000 (D.S.K.). The resulting cross-sections were then observed using a digital microscope (VH-8000C; Keyence Co.) without fixing the sections.
Scanning electron microscopy
The structure of the epidermal cells of the torenia petals was observed with a scanning electron microscope (VE-7000, Keyence Co.) without fixing the materials. The margins of the dorsal petals were observed in wild-type and transgenic torenia plants.
DNA methylation analysis
Genomic DNA was extracted from torenia leaves using an ISOPLANT II kit (Nippon Gene Co., Ltd.). Bisulfite-treated DNA was prepared from 5 μg of the genomic DNA using a MethylEasy™ Xceed (Human Genetic Signatures) according to the manufacturer’s instructions. This was then amplified using TAKARA EpiTaq™ HS (for bisulfite-treated DNA) (TAKARA). The amplified PCR products were cloned into pCR2.1 using a TOPO® TA cloning® Kit (Invitrogen) and 10× A-attachment mix (TOYOBO), following which a sequencing analysis was performed using primer sets ( Supplementary Table S6 ).
Supplementary data are available at PCP online.
This work was supported by the Scientific Technique Research Promotion Program for Agriculture, Forestry, Fisheries and Food Industry (Japan).
We thank Ms. Satoko Ohtawa for generating and maintaining the transgenic torenia plants, and Ms. Hiroko Yamada, Ms. Miho Seki, Ms. Yasuko Tanij and Ms. Yoshiko Kashiwagi for assistance with the molecular biological work as well as for maintaining the torenia plants used in the study.
The authors have no conflicts of interest to declare.
Arabidopsis APETALA 1
- BRANCHED 1
CYCLOIDEA and PCF Transcription Factor 3
chimeric repressor gene-silencing technology
quantitative real-time PCR
rapid amplification of cDNA ends
Cauliflower mosaic virus 35S
scanning electron microscopy