Introduction

Chrysanthemum (Chrysanthemum morifolium Ramat.) is an important ornamental species around the world, and flowers with a range of petal colors (such as white, pink, purple, and pale yellow to deep orange) have been developed. Kishimoto et al. (2004, 2007) reported that, in chrysanthemum cultivars, the yellow color of the ray petals is derived from the presence of yellow carotenoids, mainly lutein derivatives, and the orange color is derived from a mixture of red anthocyanins and yellow carotenoids. The petal color has generally evolved to attract insects, birds, and other pollinators (Paige and Whitham 1985), so petal color is an important element in the plant reproduction. In addition, these characteristics are important determinants of the commercial value of the flowers in horticultural markets.

The existence of a dominant gene that suppresses carotenoid biosynthesis was suggested by Hattori (1991), but subsequent research demonstrated that the carotenoid biosynthesis pathway works almost equally well at the transcriptional level, both in plants with white ray petals and those with yellow ray petals (Kishimoto and Ohmiya 2006). Recently, a candidate gene was identified from a white-flowered chrysanthemum cultivar (Ohmiya et al. 2006). The gene shows high similarity to a gene for carotenoid cleavage dioxygenase (CCD), designated CmCCD4a. CmCCD4a is expressed specifically in the ray petals of chrysanthemum cultivars. Severe repression of CmCCD4a transcription by means of RNA interference (RNAi) has been shown to convert white ray petals into yellow ray petals in the transformants (Ohmiya et al. 2006). Thus, CmCCD4a expression degrades carotenoids into colorless compounds in the ray petals of cultivars that contain this gene, resulting in white ray petals.

Nine genes in the CCD family are found in the Arabidopsis genome, and the products of most of these genes have been demonstrated to perform enzymatic roles and to modify the physiological functions of their apocarotenoid products (Tan et al. 2003; Schwartz et al. 2004; Schmidt et al. 2006; see review: Auldridge et al. 2006). However, the biological function of AtCCD4, which shows about 61% identity to CmCCD4a, remains totally unknown. Recently, it has been shown that recombinant CCD4 proteins in different plant species cleave (apo)-carotenoids at the 9,10 (9′,10′) positions to yield β-ionone (Rubio et al. 2008; Huang et al. 2009). It is interesting to note that AtCCD4 and CmCCD4a have different substrate specificity: CmCCD4a cleaves β-carotene to produce β-ionone, whereas AtCCD4 does not exhibit enzymatic activity against β-carotene but exhibits enzymatic activity against 8′-apo-β-carotene-8′-al. This finding suggests that AtCCD4 and CmCCD4a have different biological functions.

Because wild chrysanthemums have some useful characteristics that chrysanthemum cultivars lack, and can interbreed with chrysanthemum cultivars, they are often used as breeding parents (Jong and Rademaker 1989, Tanikawa et al. 2006). Modern chrysanthemum cultivars are thought to have originated from hybrids between white- and yellow-flowered wild chrysanthemums (Dai et al. 2005). It is reasonable to assume that wild species and chrysanthemum cultivars with white ray petals possess the same mechanism for causing white ray petal coloration, and that this mechanism may be derived from the degradation of carotenoids.

Generally, chrysanthemums form compound flowers that are composed of many disc florets, which possess both pistils and stamens, at the center of the flower and an outer ring of ray petals. The capitulums of Chrysanthemum pacificum Nakai and Chrysanthemum shiwogiku Kitam. are composed of only the disc florets, and lack the ray petals. It is unknown whether they have CmCCD4a orthologs because they lack the ray petals in which CmCCD4a is specifically expressed.

In this paper, we report the results of a study that demonstrate that white-flowered wild chrysanthemums have CmCCD4a orthologs, and that the translated products of these orthologs contribute to the development of white ray petal coloration in wild species, as is the case for CmCCD4a in chrysanthemum cultivars. Furthermore, we found that the apetalous wild species also have functional CmCCD4a orthologs that are responsible for carotenoid degradation, independent of whether they form ray petals.

Materials and methods

Plant materials

Wild chrysanthemum species and the yellow-flowered chrysanthemum (C. morifolium) cultivar ‘Squash’ were obtained from the NIAS Genebank Project of the National Institute of Agrobiological Science (Tsukuba, Ibaraki, Japan). We used 10 wild species with white ray petals: C. wakasaense line 2-12, C. ornatum line 4-11, C. crassum line 5-21, C. zawadskii line 8912, C. weyrichii line 8913, C. zawadskii var. latilobum line 10, C. yoshinaganthum line 12, C. japonense line 13, C. makinoi line 18 and C. yezoense line 19, two apetalous wild species that lacked ray petals: C. shiwogiku line 4-8, C. shiwogiku line 8808, C. pacificum line 11 and C. pacificum line 8706, and two wild species with yellow ray petals: C. indicum line 2-3 and C. seticuspe line 1.

Southern blot analysis

Genomic DNA was isolated from chrysanthemum mature leaves using the cetyl trimethyl ammonium bromide method (Murray and Thompson 1980). BamHI-digested genomic DNA (25 μg) was electrophoresed in 0.7% agarose gel and blotted onto Hybond-N+ nylon membrane (GE Healthcare Biosciences) using 20× SSC, followed by baking at 80°C for 2 h. The probe was prepared using PCR digoxigenin probe synthesis kit (Roche Diagnostics). Based on the sequence of CmCCD4a from the chrysanthemum cultivar ‘Paragon’, the following primer sequences were designed: 5′-GCAGACCCTAGGAAGGTTGCAC-3′ (1117-F), 5′-AGATTCCGGATTGAAAGAGGGTACC-3′ (1370-R). The filter was hybridized with a digoxigenin-labeled CmCCD4a probe and washed twice with 2× SSC plus 0.1% SDS at room temperature for 15 min, and then twice with 0.1× SSC plus 0.1% SDS at 68°C for 15 min. Luminescence from the reaction between anti-digoxigenin-alkaline phosphatase Fab fragments and CDP-Star (Roche Diagnostics) was detected using the Light-Capture AE-6962C (ATTO).

Interspecies crossing between apetalous wild species and ‘Squash’

We chose two apetalous wild species, C. shiwogiku (line 4-8) and C. pacificum (line 8706), and the cultivar ‘Squash’, which was chosen as a representative of yellow-flowered cultivars because it lacks the CmCCD4a gene. Interspecies crosses were performed between both wild species and ‘Squash’. Four combinations between the ovary parent and the pollen parent were produced: C. shiwogiku line 4-8 × ‘Squash’, C. pacificum line 8706 × ‘Squash’, ‘Squash’ × C. shiwogiku line 4-8, and ‘Squash’ × C. pacificum line 8706. These crossings produced 51, 22, 23, and five F1 plants, respectively.

Colorimetry of F1 progenies and scatterplots of color values

The color values for the ray petals of each F1 plant were measured using a CM-2600d spectrophotometer (Konica Minolta), with values provided in the L*a*b* color space. The values of the a* and b* color components were plotted on the x- and y-axes, respectively, of scatterplots created using Excel software (Microsoft).

Genomic PCR using DNA from each progeny

CmCCD4a orthologs were detected by means of PCR using 10 ng of genomic DNA from the progenies as templates, and Advantage 2 DNA polymerase (Clontech) with 5% dimethyl sulfoxide in 1× solution of the reaction buffer provided by the manufacturer. The PCR procedure was as follows: 95°C for 2 min for initial denaturation, followed by 40 cycles of 95°C for 20 s, 65°C for 20 s, and final phase at 72°C for 1 min. The following primer sequences were used for this PCR reaction: 5′-CGGAGATACTATTGTGATGGTGGCG-3′ (CCD4a-1245F); 5′-AATAATCAAAGCGTTGTTAGGTATT-3′ (CCD4a-1805R). For the PCR control, the actin gene was amplified using the following primer set: 5′-CTTGCGTTTGGATCTTGCTGGTCGTGA-3′ (Actin-291F), and 5′-AGCAGCTTCCATCCCAATCATAGACGG-3′ (Actin-556R).

Results and discussion

Ohmiya et al. (2006) reported that CmCCD4a was found in the genome of all examined white-flowered cultivars, whereas most of the yellow-flowered cultivars lacked this gene. To clarify whether this was also true for wild chrysanthemum species, we performed Southern blot analysis. Genomic CmCCD4a has two BamHI sites: one at −464 bp and the other at 1,565 bp from the start codon (Ohmiya et al. 2006). These sites are well conserved in CmCCD4a orthologs among wild chrysanthemum species (unpublished data). Then, the genomic DNA was digested with BamHI and probed with a digoxigenin-labeled CmCCD4a fragment. We detected signals that corresponded to CmCCD4a orthologs (approximately 2.0 kbp) in the lane of each white-flowered species, but no such band was observed in the lanes of yellow-flowered species (Fig. 1). This result was consistent with that obtained for chrysanthemum cultivars, suggesting that CmCCD4a orthologs are present in a wide range of chrysanthemum genera with white-colored ray petals.

Fig. 1
figure 1

The results of the Southern blot analysis for wild chrysanthemums (Chrysanthemum spp.). BamHI-digested genomic DNA of each species (25 μg) was electrophoresed. Lanes 1–10, white-flowered wild species: 1, C. wakasaense line 2-12; 2, C. ornatum line 4-11; 3, C. crassum line 5-21; 4, C. zawadskii line 8912; 5, C. weyrichii line 8913; 6, C. zawadskii var. latilobum line 10; 7, C. yoshinaganthum line 12; 8, C. japonense line 13; 9, C. makinoi line 18; 10, C. yezoense line 19. Lanes 11–12′, species that lacked ray petals: 11 and 11′, C. shiwogiku lines 4-8 and 8808; 12 and 12′, C. pacificum lines 11 and 8706; lanes 13 and 14, yellow-flowered wild species: 13, C. indicum line 2-3; 14, C. seticuspe line 1. Lane 15, the yellow-flowered cultivar ‘Squash’

The wild species that lacked ray petals, C. shiwogiku (lines 4-8 and 8808) and C. pacificum (lines 11 and 8706), were also analyzed. Surprisingly, CmCCD4a orthologs were detected in C. shiwogiku lines 4-8 and 8808 and C. pacificum line 11 (Fig. 1, lanes 11, 11′, and 12), but not in C. pacificum line 8706 (Fig. 1, lane 12′). Line 8706 thus appears to be a genetic variant of the original C. pacificum (which has CmCCD4a orthologs), because line 11 of the same species has the orthologs. This result demonstrated that CmCCD4a orthologs also exist in the apetalous wild species.

To confirm whether the proteins encoded by the CmCCD4a orthologs have enzymatic activity and can degrade carotenoids, interspecies crossing was performed between C. shiwogiku line 4-8 and the yellow-flowered cultivar ‘Squash’. Line 8706 of C. pacificum was also crossed with ‘Squash’ to provide a negative control (Fig. 2a). When these two apetalous species were crossed with ‘Squash’, all progenies had ray petals. The same phenomenon was previously reported by Jong and Rademaker (1989) for interspecies crossings between chrysanthemum cultivars and C. pacificum ‘IVT 78173’. Thus, the formation mechanism for ray petals appears to reflect the effects of dominant genes.

Fig. 2
figure 2

a The combinations used in the interspecies crossing. The apetalous wild species C. shiwogiku line 4-8, which has CmCCD4a orthologs, and C. pacificum line 8706, which lacks these orthologs, were crossed with the yellow-flowered cultivar ‘Squash’ to obtain F1 progenies. The white bars in each image represent 1 cm. b The scatterplot of the a* (x-axis) and b* (y-axis) color values for ray petals of the F1 progenies. Circles and triangles indicate the results for the progenies from the crosses with C. shiwogiku line 4-8 and C. pacificum line 8706, respectively; the symbols correspond to that of the reciprocal crosses illustrated in a. The a* and b* values were defined as the color system by CIE in 1976

We then measured the color characteristics of the ray petals of the F1 progenies obtained from the four-way crosses using a spectrophotometer, and plotted the a* and b* color values (Fig. 2b). The ray petals of the F1 progenies divided completely into two separate clusters, with white or yellow petals, depending on whether CmCCD4a orthologs were present in the parents of the wild species. The ray petals of all F1 progenies with C. shiwogiku line 4-8 as the parent were white; this indicates that proteins encoded by CmCCD4a orthologs in C. shiwogiku line 4-8 exhibit enzymatic activity against carotenoids. In addition, the CmCCD4a orthologs of this line are homozygous. When C. pacificum line 8706, which lacks a CmCCD4a ortholog, was crossed with ‘Squash’, all progenies had yellow ray petals. Given the previous results (Jong and Rademaker 1989), in which the ray petals of F1 progenies from C. pacificum ‘IVT 78173’ showed ray petals with various colors, depending on the crossing partner, C. pacificum ‘IVT 78173’ also appears to lack a CmCCD4a ortholog. Furthermore, all F1 progenies obtained from the interspecies crossing between ‘Squash’ and all of the 10 white-flowered wild species in Fig. 1 had white ray petals (Hattori 1991: C. japonense; Sumitomo, unpublished data). The fact indicated that CmCCD4a orthologs played an important role in determining the ray petal color of these chrysanthemum genera. And enzymatic activities of translated proteins of CmCCD4a against carotenoids were confirmed not only by the results of previous RNAi experiments (Ohmiya et al. 2006) but also by our crossing experiments. We wanted to investigate whether the white coloration of ray petals in the progenies was associated with the presence of CmCCD4a orthologs. For this purpose, we used all the F1 progenies (Fig. 2) and performed genomic PCR for the detection of specific genes. Figure 3 shows one representative progeny for each crossing combination. There was an evident relationship between the presence of CmCCD4a orthologs and the white coloration of ray petals. The bands that corresponded to CmCCD4a orthologs were observed in all white-flowered progenies but not in any yellow-flowered progenies. Thus, our results strongly suggested that proteins encoded by CmCCD4a orthologs are involved in the white coloration of petals of all examined chrysanthemums.

Fig. 3
figure 3

The results of genomic PCR analysis of the parents used for the interspecies crossing and of their progenies. Lanes 1–3 represent the parent species: 1, C. shiwogiku line 4-8; 2, C. pacificum line 8706; 3, ‘Squash’ cultivar. Lanes 4–12 represent the progenies obtained by interspecific hybridization: 4–6, C. shiwogiku line 4-8 × ‘Squash’; 7 and 8, ‘Squash’ × C. shiwogiku line 4-8; 9 and 10, C. pacificum line 8706 × ‘Squash’; 11 and 12, ‘Squash’ × C. pacificum line 8706. Actin was used as the PCR control

CmCCD4a orthologs exist in a wide range of cultivars and wild species of chrysanthemums with white ray petals. Although cultivars and wild species are diverse, both chrysanthemum species groups were able to cross with little difficulty. Our results suggest that the chrysanthemum genera might have differentiated from a single ancestor that had one or more CmCCD4a orthologs. Wild species without ray petals might also have differentiated from species with white petals. Both white coloration and the formation of ray petals were shown to be dominant characteristics (Jong and Rademaker 1989; Hattori 1991). Thus, C. pacificum line 8706 may have lost the ability to produce ray petals as a result of evolutionary processes, and accidentally lost its CmCCD4a orthologs as a result of a mutation in a local population. Our results thus suggest that the translated proteins from CmCCD4a orthologs would maintain CCD function in chrysanthemums regardless of the existence of the ray petals.

To add useful characteristics from wild species to cultivars, C. pacificum and C. shiwogiku have been used for decades in interspecific crossbreeding (Shibata et al. 1988; Jong and Rademaker 1989; Tanikawa et al. 2006). The following beneficial characteristics have been reported: the development of several flower heads per leaf axil, significant stem elongation, prolific reproduction, frequent branching that avoids lodging, and resistance to disease and pest. Formerly, when these apetalous species were used as breeding parents, it was not possible to predict the ray petal color of the progenies before flowering. Our study revealed that the presence of CmCCD4a orthologs is useful because it allows the prediction of ray petal colors of the F1 progenies before breeding.