Introduction

Floral morphology of angiosperms can be classified into two major symmetry types: bilateral symmetry, which possesses a single plane of symmetry, and radial symmetry, which possesses two or more planes of symmetry. Bilateral symmetry is thought to have evolved from radial symmetry to accommodate specific pollinators1. From a commercial perspective, particularly in the international floriculture market, morphological traits such as floral symmetry are important factors that influence consumer preferences.

The morphological diversity in floral symmetry has been previously characterized. Radially symmetrical mutants have been identified in species with bilaterally symmetrical flowers such as Antirrhinum majus1 and Sinningia speciosa2. Scaevola aemula, a member of the family Goodeniaceae in the order Asterales, is an herbaceous perennial or annual plant native to Australia. It exhibits vigorous growth, high environmental adaptability, an extended flowering duration, and prolific flowering3. These traits have made S. aemula a commercially valuable ornamental plant for urban landscaping, horticulture, and floriculture, particularly in North America and Europe4.

The family Goodeniaceae, which includes S. aemula, comprises 12 genera and over 420 species distributed across Australia and is divided into three major clades: Scaevola s.l., Brunonia Sm. ex R. Br., and Goodenia Sm. s.l5,6. These clades exhibit morphological variations in floral symmetry. Members of the Scaevola group display dorsoventral bilateral symmetry, with five morphologically similar petals arranged downward. In contrast, the monotypic genus Brunonia exclusively exhibits radial symmetry and lacks the characteristic dorsal slit found in other Goodeniaceae species. The Goodenia group exhibits a range of floral symmetries, typically with three ventrally oriented petals and two morphologically distinct dorsal petals7. S. aemula has one ventral petal, two lateral petals, and two dorsal petals, forming a fan-shaped bilaterally symmetrical flower with the family-specific dorsal cleft entirely turned inward (Fig. 1)7. Lineage-specific morphological variation suggests recent genetic events that arranged petal patterns in Scaevola spp.; however, the molecular basis remains unclear.

Fig. 1
Fig. 1
Full size image

Flower and whole plant of wild type, bilateral symmetrcal flower (A, B, C) and radially symmetric flower (peloric) (D, E, F). Dorsal slit is made during St3 to St4 (G). d, dorsal petal; l, lateral petal; v, ventral petal. (Scale bars: 1 cm.).

CYCLOIDEA (CYC)-like genes are crucial regulators of floral symmetry in several plant taxa. The CYC gene was the first isolated from A. majus and has since been extensively studied1,8. CYC belongs to a plant-specific gene family encoding TCP transcription factors named after the founding members TEOSINTE BRANCHED1 (TB1), CYC, and PROLIFERATING CELL FACTORS (PCFs). These transcription factors are key regulators of meristematic activity and cell proliferation9. Phylogenetic analysis has identified two major classes of TCP proteins: class I (PCF-type) and class II (CYC/TB1 and CINCINNATA, CIN)10. The CYC/TB1 lineage, also referred to as the ECE clade, is further divided into three subclades — CYC1, CYC2, and CYC3 — within the core eudicots11.

TB1 is a major determinant of maize domestication that suppresses lateral branching and promotes apical dominance12. Homologs of TB1 have also been shown to inhibit axillary bud outgrowth in both rice and Arabidopsis10,13. In Antirrhinum, two TCP genes, CYC and DICHOTOMA (DICH), control floral symmetry by regulating genes associated with cell division1,8,9,14. Wild-type Antirrhinum flowers exhibit a clear dorsiventral axis of symmetry with distinct dorsal, lateral, and ventral organ types. The expression of CYC and DICH is restricted to the dorsal domain of the floral meristem, where they promote bilateral symmetry and suppress growth of the dorsal stamen1,8. CIN, another TCP gene, regulates marginal leaf growth in Antirrhinum and affects petal lobe development by controlling cell proliferation15.

In Asteriaceae, Senecio RAY1 and RAY2 are Gerbera CYC genes that have been reported to influence floral symmetry16,17. Although floral symmetry diversity is recognized in Goodeniaceae, including the bilaterally symmetric flower S. aemula, the role of symmetry-related genes such as CYC remains unclear. In the present study, we evaluated a radially symmetrical mutant line of S. aemula to assess the involvement of CYC-like genes in floral symmetry and to uncover the genetic basis underlying this morphological transition.

Material & methods

Plant material

The breeding program for S. aemula conducted by the author began in 1997. A total of 67 wild and horticultural varieties of S. aemula were subjected to crossbreeding, open pollination, gamma irradiation, and other mutation techniques. The primary breeding objectives were the diversification of flower coloration in bilaterally symmetric flowers and the improvement of plant architecture. Among the resulting lines was #7482, a wild-type strain with bilateral symmetry. In 2014, the world’s first S. aemula line, designated #7952, which exhibited radially symmetrical flowers, was obtained through mutation breeding. Line #7952 initially displayed an unstable intermediate phenotype in which only the early stage flowers were radially symmetrical, whereas the later-developing flowers reverted to the wild-type bilaterally symmetrical form (Fig. S1). Continuous crossbreeding and selection using line #7952 eventually led to the development of line #11,361 in 2016, which exhibited a stable, radially symmetrical flower phenotype. Three representative lines with distinct genetic backgrounds were used as plant materials: #7482 (wild type; bilaterally symmetric type), #7952 (intermediate type), and #11,361 (radially symmetric type). For comparison of floral organ morphology between wild-type and radially symmetric lines at 15 weeks after grafting, three fully developed flowers were collected from each of the following lines: wild type #7482, radially symmetrical type #11,361, and intermediate type #7952. Corolla length and width and petal length and width were measured. Seedlings were grown in a greenhouse under controlled conditions (15–25 °C) by using a soil mix consisting of peat moss, perlite, and akadama.

Inheritance test of the radial symmetry trait

Plants were grown in a greenhouse at 15–25 °C. Crosses were performed among ten bilaterally symmetric lines, three radially symmetric lines (including #11361), and six lines with intermediate phenotypes (Table S1). Seeds were collected from the crosses. After sowing and raising the seedlings, floral symmetry was evaluated in 299 individuals, consisting of 110 individuals derived from ten cross combinations between radially symmetric lines and wild-type lines, 33 individuals from three cross combinations among radially symmetric lines, including self-pollination of #11,361, and 156 individuals obtained from bulk crossing among six intermediate lines.

RNA-seq, de Novo transcriptome assembly, and identification of S. aemula CYC homologs

Total RNA was extracted from flower buds using the RNeasy Plant Kit (QIAGEN, Hulsterweg, NLD) according to the manufacturer’s instructions. The extracted RNA was treated with RNase-free DNase (Promega, Madison, WI, USA) to remove residual genomic DNA. Libraries for RNA-seq analysis were prepared using TruSeq Stranded mRNA Library Prep Kit (Illumina, San Diego, CA, USA) and sequenced using NextSeq 500 (Illumina) in pair-end, 150 bp mode. RNA-sequencing (RNA-seq) analysis was outsourced to Nippon Genetics Co. Ltd. and performed using their eukaryotic mRNA-seq service (https://www.n-genetics.com/products/search/detail.html?product_id=6281). RNA-seq reads from #7055 with bilaterally symmetrical flowers were processed using Trimmomatic v0.39 to removal adapter sequences and low-quality reads18. The processed RNA-seq reads were used for de novo transcriptome assembly using Trinity v2.15.119. A TBLASTN search was conducted against the de novo assembled transcriptome using A. majus CYC (AmCYC) protein sequence (CAA76176.1) as a query. The sequences obtained from the BLAST search were compared with known TCP gene sequences to identify sequences belonging to the CYC clade.

Molecular cloning and sequence analysis

PCR primers for the amplification of the full-length S. aemula CYC1, CYC2, and CYC3 homologs were designed using the de novo assembled CYC1, CYC2, and CYC3 sequences described above. First-strand cDNA was synthesized from 300 ng of total RNA using ReverTra Ace (TOYOBO) following the manufacturer’s protocol. Genomic DNA was extracted from young leaves using NucleoSpin Plant II (Macherey-Nagel). PCR reaction mixtures included 1.25 U of Tks Gflex DNA polymerase (Takara bio), 0.2 µM of each primer, and 1 µl first-strand cDNA for RT-PCR or 10 ng genomic DNA for genomic PCR as templates. PCR conditions were as follows: 94 °C for 2 min, followed by 30 cycles at 98 °C for 15 s, 60 °C for 20 s, and 68 °C for 30 s. The PCR products were cloned using the Zero Blunt TOPO PCR Cloning Kit (Invitrogen) and sequenced. The primer sequences are shown in Table S2.

Phylogenetic analysis of CYC-like genes

A phylogenetic tree was constructed using the BioNJ method based on ClustalW multiple alignments of the amino acid sequences of CYC-related proteins using Seaview ver.4.5.420. Multiple sequences are provided in Tables S3 and S4.

Gene expression analysis using RT-qPCR

Quantitative RT-PCR was performed to determine the expression levels of SaCYC1, SaCYC2, SaCYC3A and SaCYC3B in the dorsal, lateral, and ventral petals of S. aemula. The total RNA of each petal from the flower buds at stages St1 to St2 from #7482, #7952, and #11,361 was extracted separately using the RNeasy Plant Mini Kit (Qiagen). cDNA was synthesized from 300 ng of total RNA using ReverTra Ace reverse transcriptase (Toyobo). Quantitative PCR was performed using PowerUp SYBR Green Master Mix (Applied Biosystems) and a StepOnePlus real-time PCR system (Applied Biosystems), following the manufacturer’s instructions. The primer concentration was 500 nM and the reaction conditions were as follows; 50 °C for 2 min, 95 °C for 2 min, then 40 cycles of 95 °C for 3 s and 60 °C for 30 s. The primer sequences are listed in Table S2. Relative expression levels were determined by normalizing the PCR threshold cycle number of each gene with that of actin using the 2−ΔCt method.

Confirmation of floral symmetry and genotype using cleaved amplified polymorphic sequence (CAPS) markers

Sequence analysis of the four isolated CYC genes from S. aemula revealed a nonsense mutation upstream of SaCYC2 that introduced a premature stop codon. Therefore, we developed a DNA marker to detect the mutation. Genomic DNA was extracted using a NucleoSpin Plant II kit (Macherey-Nagel), according to the manufacturer’s instructions. To detect nonsense mutations associated with floral symmetry, a 151-bp PCR fragment was amplified and subsequently digested with the restriction enzyme MseI (New England Biolabs). The digested fragments were analyzed using electrophoresis to distinguish between individuals that did not carry this mutation, those that were heterozygous, and those that were homozygous for the mutation. The PCR primers used for the CAPS marker assay were: CYC2-2F: 5’-CCATGTCTGCCCTCCTTCT-3’ and CYC2-152R: 5’-AACATTCTCCATAACCTGAGGA-3’ PCR amplification was performed using Tks Gflex DNA Polymerase (Takara Bio) under the following cycling conditions: 98 °C for 15 s, 60 °C for 20 s and 68 °C for 30 s for a total of 30 cycles. PCR products were digested with MseI for 16 h at 37 °C.

Statistical analysis

The goodness of fit was assessed using the chi-square test in Microsoft Excel. For multiple comparisons, the Tukey Honestly Significant Difference (HSD) test was performed using Microsoft Excel.

Results

Floral organ morphology of radial symmetry in S. aemula

The flowers of wild-type S. aemula line #7482 consisted of one ventral petal, two lateral petals, and two dorsal petals. It developed fan-shaped, bilaterally symmetrical flowers with a cleft between the two dorsal petals, which is a characteristic feature of the Goodeniaceae family (Fig. 1A–C). In contrast, Line #11,361 developed radially symmetrical flowers that lacked a cleft between the dorsal petals (Fig. 1D–F). Line #7952, the first strain to exhibit radially symmetrical flowers, developed radially symmetrical flowers during early growth. However, as the plant grew, most flowers became bilaterally symmetrical, with only a few radially symmetrical flowers, indicating intermediate traits. There were no significant differences in the vertical and horizontal diameters of the petals among the wild-type, intermediate, and mutant strains (Table S5). Furthermore, no morphological differences were observed in the dorsal, lateral, or ventral petal regions of either the wild or radially symmetric type (data not shown). To investigate the development of the dorsal cleft, which is a characteristic of the Goodeniaceae family, we evaluated flower buds at various developmental stages. In the wild-type line, no dorsal cleft was observed from stages St1–St3; however, a dorsal cleft formed between stages St3 and St4, resulting in a bilaterally symmetrical flower type (Fig. 1G).

Heritability of the radially symmetric trait

To clarify the heritability of the floral symmetry trait, the wild-type, radially symmetric type, and F1 progeny (between the wild and mutant types) were investigated (Table S1). All 110 seedlings obtained from the crossing of radially symmetric line #11,361 with ten wild-type lines exhibited an intermediate phenotype similar to that of line #7952. All 33 seedlings obtained from three radially symmetric × radially symmetric combinations, including a self-cross of line #11,361, flowered with radially symmetrical flowers. In the progeny from the six combinations using bulk crosses of the same intermediate-type line as #7952, the segregation of bilaterally and radially symmetrical flowers was observed in all combinations, with the wild-type occurrence rate ranging from 61 to 82%. A chi-square test was performed for each cross combination, assuming that radial symmetry is a recessive trait determined by a single gene, with a wild-type: radially symmetrical flower segregation ratio of 3:1. Although the theoretical and measured values deviated more significantly in the 16–184 × bulk pollen combination (six lines), the actual occurrence rates for the other combinations agreed with the theoretical segregation ratio of 3:1. These results suggest that the radial symmetry is a recessive monogenic trait.

Characterization of S. aemula CYC-like genes

To isolate floral symmetry-related CYC genes from S. aemula, we conducted RNA-seq analysis of floral buds and mature flowers. De novo transcriptome assembly was performed using Trinity, and candidate genes with high homology to AmCYC were screened using BLAST. We identified three CYC-like genes, SaCYC1, SaCYC2, and SaCYC3A in the transcriptome dataset (Fig. S2). Berger et al.21 used genome-skimming analysis to predict the presence of four CYC genes in the Goodeniaceae family members S. collaris, S. phlebopetala, S. porocarya, and S. tomentosa. Based on these findings, the authors proposed that Scaevola possesses four CYC genes21. To support this, we examined the published genome sequence of the closely related species S. taccada and identified four CYC homologs: StCYC1, StCYC2, StCYC3A, and StCYC3B. These results led us to predict that S. aemula also contains four CYC genes: SaCYC1, SaCYC2, SaCYC3A, and SaCYC3B (Fig. S3). We then attempted to isolate the paralog from S. aemula that gave rise to StCYC3B. We designed primers based on the StCYC3B sequence and successfully amplified genomic SaCYC3B from S. aemula DNA. The estimated genome lengths of the coding sequences of SaCYC1, SaCYC2, SaCYC3A, and SaCYC3B were 1170 bp, 1005 bp, 1212 bp, and 1152 bp, respectively (Fig. S3, Table S4). Sequence identity analysis revealed that the conserved TCP domains among these four homologs shared 71–92% identity, whereas the R domains showed 65–88% identity (Table S6).

To predict the functions of full-length SaCYC1, SaCYC2, SaCYC3A, and SaCYC3B sequences, we constructed a phylogenetic tree using the TCP Class II CYCLOIDEA clade of closely related species, including S. taccada, Gerbera hybrida, Senecio vulgaris from the Asteraceae family, A. majus from the Plantaginaceae family, A. thaliana (Brassicaceae), Torenia fournieri, (Linderniaceae), and Chirita pumila (Gesneriaceae) (Fig. 2). The analysis showed that CYC-like proteins were broadly classified into four clades, with A. thaliana AtTCP1 separately placed as an outgroup. Clade 1 included SaCYC1 and its closely related homologs StCYC1 and GhCYC1 from the Asteraceae family. Clade 2 included SaCYC2, StCYC2, GhCYC2, SvRAY2, GhCYC4, GhCYC3, and SvRAY1 of Asteraceae, which are topologically similar to Goodeniaceae. Clade 3 contained SaCYC3A, StCYC3A, and StCYC3B. Clade 4 consisted of TfCYC2, CpCYC2, AmDICH, and AmCYC from Lamiales. These results suggest that the CYC-like genes in clades 1, 2, and 3 are specifically shared among Asteraceae lineages. Notably, SvRAY and GhCYC genes, the CYC-like genes regulating floral symmetry of Asteraceae, were clustered with SaCYC2/StCYC2 in Clade 2, suggesting that these CYC-like genes are involved in floral symmetry of Scaevola spp.

Fig. 2
Fig. 2
Full size image

A phylogenetic tree of CYCLOIDEA-related genes. A neighbor-joining tree is shown with bootstrap values (%) on the branches (1000 replicates). The scale bar indicates 0.05 amino acid substitutions per site.

Sequence analysis of SaCYC genes

We compared the sequences of SaCYC1, SaCYC2, SaCYC3A, and SaCYC3B in wild-type #7482, intermediate #7952, and radially symmetric #11,361 lines of S. aemula. No mutations resulting in an apparent loss of function, such as frameshift or nonsense mutations in SaCYC1, SaCYC3A, or SaCYC3B were found. However, in SaCYC2 of radially symmetrical line #11,361, a nonsense mutation was detected at the 99 bp position, where a cytosine (C) was replaced by an adenine (A), resulting in a premature stop codon, we named this mutant gene as SaCYC2m (Fig. 3A). Additionally, RNA-seq data showed that, in contrast to sequences corresponding to SaCYC2 in the wild-type lines, all sequences in radially symmetric #11,361 lines containedSaCYC2m. Intermediate #7952 possessed both SaCYC2 and SaCYC2m alleles at approximately equal allele frequencies (Fig. 3A). These results suggest that #7952 is heterozygous for SaCYC2 and SaCYC2m, whereas #11,361 is homozygous for SaCYC2m. SaCYC2 contains both TCP and R domains, which are hallmarks of TCP Class II transcription factors, located at positions 333–507 bp and 696–747 bp, respectively. In the SaCYC2m, a nonsense mutation occurred upstream of these domains, likely resulting in a truncated protein lacking both the TCP and R domains. Collectively, these results suggest that the mutation causes a loss of function in SaCYC2, thereby causing a recessive genetic defect in bilateral floral development in #11,361.

Fig. 3
Fig. 3
Full size image

Gene structure of CYCLOIDEA2 of S. aemula. (A) Radially symmetric line #11,361 has nonsense mutation in SaCYC2 at the 99 bp position, C to A. Intermediate line #7952 is heterozygous for SaCYC2 and SaCYC2m, containing both the wild and mutant alleles. The frequencies of the wild and mutant alleles were visualized using IGV (Integrative Genomics Viewer). (B) CAPS marker for radial symmetry in S. aemula. The restriction enzyme Msel detects only mutated allele.

Gene expression analysis using RT-qPCR

We further investigated the relationship between floral symmetry and the expression patterns of the four SaCYC genes in the dorsal, lateral, and ventral regions of S. aemula (Fig. 4). In all petal regions, significantly higher SaCYC1 expression was exhibited in the radially symmetric lines, followed by the intermediate type, and then the wild-type. SaCYC2 was specifically expressed in the dorsal region regardless of floral symmetry but was barely expressed in the lateral and ventral regions. SaCYC3A showed significantly lower expression in the wild-type than in the intermediate and mutant types in all petal regions. Similarly, SaCYC3B exhibited significantly lower expression in the wild type than in the radially symmetric type across all petal regions. SaCYC3B was more highly expressed in the dorsal and lateral regions than in the ventral regions in the intermediate type. Because wild type #7482, intermediate type #7952, and radially symmetric type #11,361 have different genetic backgrounds, the expression levels of SaCYC1, SaCYC2, SaCYC3A, and SaCYC3B varied depending on the petal region. However, all lines consistently exhibited dorsal-specific expression of SaCYC2.

Fig. 4
Fig. 4
Full size image

The results of RT-qPCR of SaCYC1, SaCYC2, SaCYC3A and SaCYC3B on dorsal, lateral and ventral petal at St1 to St2 in S. aemula. Only SaCYC2 was expressed specifically in the dorsal region. Error bars of gene expression are ± 1 SD from three biological replicates. Different letters indicate statistically significant differences at p <0.05 according to Tukey’s HSD test.

Confirmation of floral symmetry and genotype using CAPS markers

Based on polymorphisms within SaCYC2 alleles, we constructed CAPS markers and applied them to lines #7482, #7952, and #11,361 (Fig. 3B). Line #7482 displayed a single band at approximately 151 bp, whereas line #11,361 exhibited two bands, estimated at 100 bp and 50 bp, when digested with the restriction enzyme MseI (Fig. 5A). Line #7952 showed three bands corresponding to those appearing in both #7482 and #11,361. The genotypes were classified as N/N for the wild type, N/n for the intermediate type, and n/n for the mutant type. To assess the versatility of the CAPS marker, it was also applied to 24 bilaterally symmetrical wild-type flowers (including horticultural varieties), 12 mutant types, and three F1 lines resulting from crosses between the wild-type and radially symmetric types (Fig. 5B; Table 1, and Table S7). The genotypes of these samples were consistent with the phenotypes, with 24 wild types showing N/N, 3 intermediate types showing N/n, and 12 radially symmetric types showing n/n, thereby confirming the association between the genotype and phenotype.

Fig. 5
Fig. 5
Full size image

CAPS maker identifying radially symmetrical flower in S. aemula. (A) N/N, wild, BL shows band in approximately 151 bp, n/n, radially symmetrical line, RD shows bands in around 55 bp and 96 bp using this CAPS marker. N/n, hetero type shows bands both N/N and n/n. (B) Genotype using CAPS marker identified phenotype of BL, RD and IM. BL, bilateral symmetrical flower; RD, radially symmetrical flower; IM, intermediate flower.

Table 1 Relationship of phenotype of flower symmetry and genotype by CAPS maker in S. aemula. BL, bilateral symmetrical flower; IM, intermediate flower between BL and RL; RL, radially symmetrical flower; N, wild type, BL allele; n, mutant type, RL allele.
Full size table

Discussion

In this study, we investigated the floral phenotype and inheritance pattern of a newly discovered radial symmetry trait in S. aemula and sought to identify its genetic basis, focusing on the TCP transcription factor CYC-like gene, which is known to be associated with floral symmetry. We found SaCYC2 was expressed specifically in the dorsal region of S. aemula and that the sequence in the radially symmetrical line had a nonsense mutation. Thus, we hypothesized that SaCYC2 mainly controls flower symmetry in S. aemula. Among four CYC-like genes in S. aemula, we identified SaCYC2 as a regulator of floral symmetry, similar to the CYC-like genes SvRAY and GhCYC in Asteraceae. We found that only SaCYC2 is preferentially expressed in dorsal region of the petal among the four paralogs, which further supports this hypothesis. In Antirrhinum, AmCYC plays a critical role in bilateral symmetry and dorsal stamen suppression1 and is similarly expressed early in the dorsal region. Similarly, in Torenia, dorsal-specific expression has also been reported for TfCYC1, TfCYC2, and TfCYC322. The dorsal-specific expression of SaCYC2 mirrors these species, supporting its functional importance in establishing bilateral symmetry in S. aemula.

The striking structural similarity between SaCYC2 and StCYC2 in Goodeniaceae suggests that StCYC2 may determines floral symmetry of S. taccada. These results support the functional conservation of CYC-like genes of clade 2 for floral symmetry in Asterales. However, few peloric flowers of Scaevola spp. have been naturally observed, indicating a strict natural selection of floral morphology through pollinator attraction7. Furthermore, we did not observe significant differences in petal size (dorsal, lateral, and ventral) between wild-type and radially symmetric type. In contrast, in Gerbera and Torenia, GhCYC2 and TfCYC2 knockout influences floral symmetry and often results in smaller petals compared to those in the wild type17,22. This phenotypic difference suggests a lineage-specific differentiation of CYC-like gene functions between Asteraceae and Goodeniaceae.

The results of this study indicate that radial symmetry in S. aemula is controlled by a single recessive gene. In Gloxinia (Gesneriaceae), when the bilaterally symmetrical flower ‘Pink Flower’ was crossed with the radially symmetrical mutant ‘White Bell,’ all F1 progeny exhibited the wild-type phenotype, while the F2 generation segregated in a 3:1 ratio of wild type to mutant, indicating that the radial symmetry phenotype is caused by a recessive mutation23. In S. aemula, although the F1 progeny of wild type and radially symmetric type crosses display intermediate traits rather than a purely wild-type phenotype, the segregation pattern in the next generation supports the conclusion that the radial symmetrical flower is controlled by a single recessive gene.

We isolated four CYC-like genes from S. aemula — SaCYC1, SaCYC2, SaCYC2m, SaCYC3A, and SaCYC3B — using floral and bud transcriptome analyses. Although SaCYC3B was absent from the transcriptome, we isolated its genomic sequence based on the sequence of a closely related species, S. taccada. It is important to note that the gene expression of SaCYC3A and SaCYC3B was trivial during petal development. Neither SaCYC1, SaCYC3A, nor SaCYC3B showed any regional specificity of expression in the dorsal, lateral, and ventral dimensions, regardless of floral symmetry (Fig. 4). In addition, there were no genetic polymorphisms such as loss of function, frameshift, or nonsense mutations in the three SaCYC-like genes between the wild-type and peloric mutants. These results highlight the preferential expression of SaCYC2 in the dorsal region of the petal and further support its specific role in floral symmetric development.

Phylogenetic analysis grouped SaCYC2 with SvRAY1, SvRAY2, GhCYC2, GhCYC3, GhCYC4, and StCYC2 (from S. taccada), all of which belong to the order Asterales and have been implicated in promoting bilateral symmetry. In Gerbera, GhCYC2 is critical for ray flower symmetry but is not expressed in radially symmetrical disc flowers. Suppressing GhCYC2 expression produces radially symmetrical ray flowers9,17. Similarly, S. aemula plant lacking SaCYC2 function produce radially symmetrical flowers without dorsal slits, akin to the tubular flowers of Gerbera, whereas the wild-type forms fan-shaped flowers with a dorsal slit, similar to ray flowers. However, in Gerbera, symmetrical shifts were accompanied by petal size changes, suggesting more complex CYC functions than those in S. aemula. Notably, CYC genes of S. aemula (Asterales) formed a clade distinct from those of Antirrhinum, Torenia, and Chirita (Lamiales), suggesting independent evolutionary paths. In Antirrhinum, AmCYC and AmDICH are expressed in the dorsal slit and silenced in radially symmetrical mutants1,8. In Chirita, overexpression of CpCYC1/2 induces dorsal petal formation, whereas double knockout promotes ventralization24. Similarly, in Torenia, TfCYC2 overexpression alters pigmentation in dorsal petals, whereas its knockout causes ventralization22. TfCYC1 plays a supporting role and influences pigmentation. GhCYC1, in the same clade as SaCYC1, and its homolog HaCYC1a/b in Helianthus, are expressed mainly in vegetative tissues and may regulate branching or structural maintenance, although their specific roles remain unclear25,26. In the CYC3 clade, KmCYC3B from Knautia shows dorsal-specific expression, suggesting its role in floral symmetry, although its function remains poorly understood21. In summary, SaCYC2 plays a central role in the regulation of floral symmetry in S. aemula, as supported by its expression patterns, sequence data, and phylogenetic relationships. Other CYC homologs, such as SaCYC1, SaCYC3A, and SaCYC3B, may play auxiliary roles; however, detailed functional analyses are required. These findings suggest that while CYC genes are evolutionarily conserved regulators of floral symmetry, their diversification and functional specialization differ between plant lineages, such as Asterales and Lamiales.

Floral symmetry is a key trait in ornamental flowers. Previous studies on radially symmetrical flower traits have been reported in several species, including Antirrhinum, Senecio, Gerbera, Mimulus, and others1,16,22,27,28. Introducing diversity into the floral symmetry types could potentially enhance their commercial value7. Within the Goodeniaceae family, to which S. aemula belongs, floral symmetry is diverse: monotypic Brunonia exhibits radial symmetry, Goodenia includes both radially and bilaterally symmetrical species, and the Scaevola group consists solely of bilaterally symmetrical species. Thus, most commercial varieties of S. aemula are limited to bilateral floral symmetry. Here, we developed CAPS markers that perfectly corresponded to the floral symmetry phenotype and genotype of the mutant progeny, thereby confirming the critical role of CYC2-like genes in regulating symmetry in S. aemula. Furthermore, these markers offer valuable tools for the efficient development of novel cultivars in Scaevola breeding programs.

In this study, we demonstrated that SaCYC2 plays a central role in establishing bilateral floral symmetry in S. aemula, and that the loss of its function leads to a developmental defect in dorsal slit formation, resulting in a transition to radially symmetrical flowers. However, the mechanisms through which heterozygous individuals exhibit intermediate phenotypes remain unclear. Further functional studies of other paralogs, such as SaCYC1, SaCYC3A, and SaCYC3B will are also necessary to elucidate their contributions to the floral development of Scaevola. Introducing Scaevola CYC-like genes into plants with radial flowers such as Arabidopsis will address this issue.