The CmYAB1‐CmCUC2 transcriptional module shapes the iconic ray florets in Chrysanthemum morifolium

Summary

• Sympetaly (fused petals) is a major innovation driving floral diversity, yet its genetic basis in the highly successful Asteraceae family is elusive. We therefore investigated the molecular mechanisms controlling this trait in Chrysanthemum morifolium.

• We conducted functional analyses of candidate genes, examined their spatiotemporal expression patterns, and performed biochemical assays to investigate the interaction between the transcription factors CmYAB1 and CmCUC2.

• Abaxially expressed CmYAB1 regulates floret fusion: its suppression enhanced fusion, while its overexpression caused petal splitting and curling. CmCUC2, a NAM/ATAF1/2/CUC2 (NAC) transcription factor and homolog of the Arabidopsis boundary‐specific genes CUC1/2, was downregulated in CmYAB1‐silenced lines. Interestingly, its expression shifts dynamically from early floral organ boundaries to later abaxial petal domains, overlapping CmYAB1. CmCUC2 knockdown also increased fusion, while its overexpression led to unfused, curved petals. CmYAB1 directly activates CmCUC2 by binding its promoter, and transient CmCUC2 knockdown in CmYAB1‐overexpressing plants rescued the unfused phenotype. Furthermore, silencing either gene upregulated and sustained CmCYC2c expression in dorsal petals, correlating with their elongation.

• This study delineates the CmYAB1‐CmCUC2 regulatory pathway, thereby revealing a novel module of polarity and boundary genes that controls sympetalous development in Asteraceae, advancing our understanding of floral diversity, and enabling the targeted engineering of ornamental traits.

The CmYAB1–CmCUC2 regulatory module controls petal fusion by modulating CmCYC2c expression in chrysanthemum ray florets.

Introduction

Sympetaly (fused petals/perianth) is one of the most important morphological innovations in the evolution of angiosperms. The degree and timing of petal fusion, as well as whether the petal is fused with other floral organs, contribute to the diversity of sympetalous flowers, which enabled remarkable specialization in flower structure and morphology (Zhong & Preston, 2015). Sympetaly is pervasive in the Asterids – a clade comprising nearly a quarter of all angiosperms and characterized by highly complex floral structures – and has undergone repeated independent evolution with modifications in size and shape (Tähtiharju et al., 2012; Yang et al., 2012, 2015). The study of sympetaly has focused on classical model plants including snapdragon (Antirrhinum majus L.), petunia (Petunia × hybrida), and morning glory (Ipomoea nil). In these species, petal primordia originate as separate structures but later fuse through basal meristem fusion, forming the corolla tube – a process classified as late sympetaly. Mutation of FEATHERED (FE, orthologs of Arabidopsis KANADI1 (KAN1)) in Japanese morning glory results in petal fusion defects (Iwasaki & Nitasaka, 2006). Repression of the function of TEOSINTE BRANCHED1, CYCLOIDEA, AND PCF TRANSCRIPTION FACTOR5 (TCP5) using CRES‐T system caused divided corollas in Ipomoea nil (Ono et al., 2012). In petunia, loss of MAEWEST (MAW, homolog of Arabidopsis thaliana L. WUSCHEL‐related homeobox 1 (WOX1)) activity disrupted the corolla tube (Iwasaki & Nitasaka, 2006; Vandenbussche et al., 2009). In snapdragon, CUP‐SHAPED COTYLEDON (CUC) orthologue CUPULIFORMIS (CUP) was implicated in petal fusion by clearing its activity at petal junctions (Rebocho et al., 2017). Conversely, the reduced expression of petunia NOAPICAL MERISTEM (NAM, orthologue of CUC1/2‐like) and NAM‐like 16 (NH16, orthologue of CUC3) using virus‐induced gene silencing (VIGS) caused petal unfusion (Zhong et al., 2016). In the monkeyflower (Mimulus lewisii), the tasiRNA‐Auxin Response Factor 3/4 (ARF3/4) pathway is involved in corolla tube formation by altering auxin homeostasis (B. Ding et al., 2020). While late sympetaly has been extensively studied in model plants, early sympetaly – characterized by petal fusion at primordia initiation – represents a distinct developmental pattern that remains largely unexplored.

It has been observed that most genes implicated in sympetalous flower development are those involved in regulating leaf polarity and organ boundary development in dialypetalous flower Arabidopsis, such as KAN1, WOX1, ARF3/4, CUC1/2, and CUC3. In Arabidopsis leaf, WOX1 and PRESSED FLOWER (PRS)/WOX3, expressed along the adaxial–abaxial juxtaposition and the leaf margins, is suppressed by abaxial determinants KAN1/2 and ARF3/4 and promoted by abaxial polarity protein YABBY1 (YAB1/FIL) (Vandenbussche et al., 2009; Nakata et al., 2012; Guan et al., 2017). Mutations of one or more members of the YAB family produced decreased floral organs and filamentous structures in place of flower organs (Siegfried et al., 1999; Sarojam et al., 2010). Mutation of YAB1 homologous GRAMINIFOLIA in A. majus causes smaller but still united petals (Golz et al., 2004). The expression levels of Solanum lycopersicum YAB1 and YAB3 are negatively correlated with flower size (Filyushin et al., 2018). YAB genes not only function in the maintenance of polarity but also are required for lateral growth. However, the role of YABs in the development of sympetalous flowers remains unclear.

Morphological boundaries, narrow bands of cells having limited cell proliferation, separate an organ primordium from surrounding meristem and organs (Breuil‐Broyer et al., 2004). Boundaries between petal primordia promote separation, whereas clearance of boundary gene activities enables petal fusion. Arabidopsis CUCs are extensively studied as boundary genes. Arabidopsis CUC1 and CUC2, targets of microRNA164 (miR164), function redundantly in organ fusion. Mutation of either the CUC1 or CUC2 single gene occasionally shows a weak phenotype in cotyledon and flower organs, while the cuc1 cuc2 double mutant displays severely fused cotyledons and fused sepals and stamens (Aida et al., 1999; Takada et al., 2001). Another Arabidopsis CUC gene, CUC3, is also involved in the establishment of the organ boundary but is not cleaved by miR164 (Vroemen et al., 2003; Hibara et al., 2006). However, none of the cuc mutants produce sympetalous flowers in Arabidopsis, suggesting that other genes or mechanisms exist in the regulation of sympetalous flower development. OsCUC1 (the homolog of Arabidopsis CUC1 and CUC2) and OsCUC3 function partially redundantly in boundary specification in rice (Wang et al., 2021). CUC2 homologs are also necessary for leaf margin dissection in simple leaves and leaflet development in compound leaf species, such as S. lycopersicum, Lotus japonicus, Medicago truncatula, and Vigna radiata (Nikovics et al., 2006; Blein & Laufs, 2008; Berger et al., 2009; Cheng et al., 2012; Wang et al., 2013; Jiao et al., 2019). Although silencing of NAM and NH16 by VIGS causes defects of petal fusion in petunia, flowers occasionally produced on escape shoots of nam mutants had normal corolla tubes (Souer et al., 1996; Zhong et al., 2016). Additionally, no NAM expression in the boundaries between petals is detectable, although it is expressed in the boundaries of other floral organs in Petunia axillari (Preston et al., 2019). Although the function of those boundary and polarity genes is studied in some sympetalous plants, no fused petals were observed in the polypetalous Arabidopsis mutants of these genes, suggesting that the regulation mechanisms of sympetalous flower development are not fully revealed.

Asteraceae, one of the largest families of angiosperms, encompasses numerous species of high ornamental and economic value (e.g. Helianthus annuus, Gerbera hybrida, Tagetes erecta, and Chrysanthemum morifolium). This family exhibits the highest incidence of sympetalous flowers and represents the primary family in which early sympetaly occurs (Endress, 2011; Schonenberger & von Balthazar, 2013). However, the regulatory mechanisms underlying the early‐sympetalous corolla tube fusion in Asteraceae remain poorly understood. Notably, C. morifolium is an economically important crop in the Asteraceae family, valued for its ornamental, medicinal, beverage, and edible uses, and widely cultivated world‐wide. Its distinctive capitulum comprises peripheral ray florets and central disc florets: The disk florets consistently display tubular corollas, while ray florets exhibit a morphological gradient – from unfused flat type to fused tubular type. This continuum of petal fusion patterns substantially enhances inflorescence complexity and ornamental value (Song et al., 2018). Therefore, chrysanthemum serves as an excellent model system for investigating the regulatory mechanisms of sympetaly.

Our previous study showed that the ray petal primordia initiate jointly, forming a ring primordium at the early stage of flower development in chrysanthemum. Subsequently, differential growth patterns were observed between petal morphotypes: In both flat and spoon‐shaped ray petals, the dorsal region of the annular primordium exhibited significantly slower growth compared to the ventral region, whereas tubular ray petals demonstrated synchronous growth throughout the entire annular primordium (Broholm et al., 2008). TCP transcription factors CYCLOIDEA2s (CYC2s) specify the floral identity and affect the petal fusion in Asteraceae plants, such as G. hybrida, Senecio vulgaris, Chrysanthemum lavandulifolium (Chapman et al., 2008, 2012; Kim et al., 2008; Huang et al., 2016; Chen et al., 2018; Shen et al., 2021; Rong et al., 2024; Zhang et al., 2024). Ectopic expression of the GhCYC2 in G. hybrida induced elongation of disk flower petals with enhanced ligulate morphology and reduced organ fusion (Broholm et al., 2008). Knocking down the transcripts of ClCYC2g in C. lavandulifolium caused the transition of a ray flower toward the disk‐like form, increasing the degree of petal fusion (Shen et al., 2021). Our previous research suggested that CmYAB1 and CmCUCs genes are responsible in petal morphogenesis in C. morifolium (Ding et al., 2019), the underlying mechanism is still unclear. In this study, we generated overexpression and knockdown transgenic plants of CmYAB1 and CmCUC2 and found that they are involved in ray petal fusion. In situ hybridization analysis revealed partial overlap in the expression domains of the two genes within the ray petals. Biochemical assays showed that CmYAB1 directly promotes the expression of CmCUC2. Transient transformation assays further indicated that CmCUC2 acts genetically downstream of CmYAB1 in regulating ray petal fusion. These findings provide new insights into the regulatory network of sympetalous flower development and offer strategic targets for directed breeding of floral morphology.

Materials and Methods

Plant materials and growth conditions

The chrysanthemum cv ‘Jinba’ used in this study was acquired from the Chrysanthemum Germplasm Resource Conservation Center at Nanjing Agricultural University (Nanjing, China). Stably transgenic lines used for morphological analysis were cultivated in a glasshouse with routine management protocols (Cheng et al., 2020). For transient transformation assays, plants were grown in a growth chamber under conditions as previously reported (Zhang et al., 2022).

Phylogenetic analysis and sequence alignment

Full‐length amino acid sequences of YAB and CUC family proteins were retrieved from NCBI databases and the Chrysanthemum Genome Database (http://210.22.121.250:8880/asteraceae/homePage), except for CmYAB1 and CmCUC2, which were cloned from chrysanthemum in this study. Detailed sequences are listed in Supporting Information Notes S1. Phylogenetic trees were reconstructed in MEGA X using the maximum‐likelihood algorithm, based on MUSCLE‐aligned protein sequences, the Jones‐Taylor‐Thornton (JTT) matrix‐based model, and 1000 bootstrap replicates for statistical support.

Subcellular localization of CmYAB1 and CmCUC2

The full‐length coding sequence (CDS) of CmYAB1 without the stop codon was cloned into the plasmid pMDC43 vector to generate a 35S::GFP (Green Fluorescent Protein)‐CmYAB1 fusion construct. The recombinant vector and the empty 35S::GFP control were separately co‐infiltrated with a 35S::D53‐RFP (Red Fluorescent Protein) nuclear marker into Nicotiana benthamiana leaf epidermal cells via Agrobacterium‐mediated transformation. After incubation in darkness at 22°C for 16 h, GFP and RFP fluorescence signals were observed and captured using a Zeiss LSM 780 confocal microscope (Zeiss, Jena, Germany). The primers are listed in Table S1.

Transcriptional activity analysis of CmYAB1

The transcriptional activity of CmYAB1 was evaluated using a dual‐luciferase assay in Arabidopsis protoplasts as previously described (Yoo et al., 2007). The 35S::GAL4DB‐CmYAB1 effector plasmid was constructed via LR recombination (Invitrogen) from the pENTR1A‐CmYAB1 entry vector. Three independent transfection groups were set: positive control (35S::GAL4DB‐AtARF5 + 5×GAL4‐LUC), negative control (35S::GAL4DB + 5×GAL4‐LUC), and test group (35S::GAL4DB‐CmYAB1 + 5×GAL4‐LUC). The luciferase activity was measured as previously reported (Song et al., 2013). The primers are listed in Table S1.

RT‐qPCR

Total RNA was isolated from outmost whorl ray petals of wild‐type (WT) and transgenic lines at the early blooming stage using an RNA isolation kit (Huayueyang, Beijing, China). First‐strand cDNA was synthesized using the PrimeScript RT reagent Kit (Takara, Dalian, China). Reverse transcription quantitative polymerase chain reaction (RT‐qPCR) was performed on a Lightcycler 96 Real‐Time PCR System (Roche, Basel, Switzerland) using SYBR Premix Ex Taq II kit (Takara) following the manufacturer's protocol. Relative transcript levels were calculated using the 2^−ΔΔCT method (Livak & Schmittgen, 2001), with CmEF1α (KF305681) as the internal reference gene.

LMD and RT‐qPCR

The 2‐mm flower buds were fixed in Farmers Fixative, dehydrated, embedded, and sectioned as described previously (Zhao et al., 2016). LMD was performed using a fully automatic laser microdissection (LMD) system (LMD7000; Leica Microsystems, Wetzlar, Hesse, Germany). Total RNA was extracted using the PicoPure RNA Isolation Kit (KIT0204), treated with DNase I (Qiagen), and amplified using the Ovation RNA‐Seq System V2 kit (NuGEN Technologies, Redwood City, CA, USA). Conventional RT‐qPCR was then conducted to quantify gene expression levels in specific tissues (Hollender et al., 2014; Zhao et al., 2016).

In situ hybridization

Chrysanthemum inflorescences of various developmental stages were collected and fixed in 3.7% formal‐acetic‐alcohol (3.7% formaldehyde, 5% glacial acetic acid, and 50% ethanol). Sample fixation, sectioning, and hybridization were performed as described previously (Ding et al., 2015). Digoxigenin (DIG)‐labeled sense and antisense RNA probes were synthesized using a DIG RNA Labeling Kit (SP6/T7; Roche) with gene‐specific primers. The primers are listed in Table S1.

Genetic transformation of chrysanthemum

The overexpression (35S::CmYAB1, 35S::mCmCUC2) and knockdown (amiR‐CmYAB1, amiR‐CmCUC2) constructs were introduced into Agrobacterium tumefaciens strain EHA105 via electroporation. Stable transgenic chrysanthemum plants were generated via the leaf disk transformation method (Ding et al., 2019). Positive transgenic lines were screened on selective rooting medium and verified by PCR at both the DNA level and the RNA level. The primers are listed in Table S1.

Morphological and histological analysis

The inflorescences of WT and CmYAB1 and CmCUC2 transgenic plants at the full blooming stage were photographed and used for phenotype observation. Corolla tube length (CTL) and ray floret length (RFL) of outermost ray florets were measured as defined previously (Song et al., 2018). For scanning electron microscopy (SEM), inflorescences (2–3, 4–6, and 8–10 mm in diameter) or only the outermost ray florets were sampled and fixed with 2.5% glutaraldehyde, dehydrated, dried to the critical point, sputter‐coated with gold, and observed using SEM (SU8100; Hitachi High‐Tech Corporation, Tokyo, Japan). Fresh samples were also observed under an S8APO optical microscope (Leica Camera AG, Germany).

Yeast one‐hybrid assay

For the yeast one‐hybrid (Y1H) assay, the full‐length CDS of CmYAB1 was cloned into the pGADT7 vector, and the promoter fragments of CmCUC2 were cloned into the pHIS2 bait vector. Combinations of recombinant plasmids were cotransformed into yeast strain Y187 and cultured on SD/−Trp/−His/−Leu medium supplemented with 50 mM 3‐AT for 3 d at 28°C (Zhou et al., 2022). The primers are listed in Table S1.

Electrophoretic mobility shift assay (EMSA)

The coding region of CmYAB1 CDS was cloned into the pET32a vector to express His‐tagged fusion protein in Escherichia coli BL21 (DE3). Recombinant His‐CmYAB1 protein was purified using Ni‐NTA magnetic beads (Thermo Fisher Scientific, Shanghai, China). Probes containing WT or mutated WATNAT motifs were synthesized and labeled with biotin at the 3′ end. Electrophoretic mobility shift assay (EMSA) was performed using the LightShift™ EMSA Optimization and Control Kit (Thermo Fisher Scientific) and Chemiluminescent Nucleic Acid Detection Module Kit (Thermo Fisher Scientific) following the manufacturer's instructions (Zhou et al., 2022). The primers are listed in Table S2.

Dual‐luciferase (LUC) assay

The 2100‐bp CmCUC2 promoter fragment was cloned into the pGreenII 0800‐LUC reporter vector as a reporter. The CDS of CmYAB1 was inserted into the pORE‐R4 vector under the control of the 35S promoter to generate the effector construct. The reporter and effector were cotransfected into chrysanthemum mesophyll protoplasts as previously described (Yoo et al., 2007). Firefly luciferase (LUC) and Renilla luciferase (REN) activities were measured using the Dual‐Luciferase Reporter Assay System (Promega), and the LUC/REN ratio was calculated (Zhou et al., 2022; Zhang et al., 2022). The primers are listed in Table S1.

ChIP‐qPCR assay

The 35S::CmYAB1‐GFP and 35S::GFP (as the control) transgenic plants were subjected to ChIP‐qPCR assay. Briefly, the shoot apexes (including two leaves) were crosslinked with 1% formaldehyde under vacuum for 30 min and then ground into powder in liquid nitrogen. Chromatin complexes were immunoprecipitated using Pierce™ ChIP‐grade Protein A/G Magnetic Beads (Thermo Fisher Scientific) and GFP recombinant rabbit monoclonal antibody (Thermo Fisher Scientific). Enriched DNA fragments were quantified by RT‐qPCR with primers targeting specific regions of the CmCUC2 promoter (Table S3).

Transiently transformation of chrysanthemum

A virus‐based artificial microRNA system was used to transiently silence CmCUC2 in OX‐CmYAB1 transgenetic plants (Guan et al., 2022; Huang et al., 2022). The amiR‐CmCUC2 fragment was inserted into CaLCuV vector to generate CaLCuV‐amiR‐CmCUC2. The 30‐d‐old rooted seedings of OX‐CmYAB1 plantlets were immersed in infiltration buffer containing pCVB and CaLCuV‐amiR‐CmCUC2 (or empty vector control) and vacuum‐infiltrated at 0.8 MPa for 10 min. Infected plants were kept at 10°C in the dark for 2 d, then transplanted into pots and grow under long‐day conditions for 4 wk before floral induction under short days. Phenotypes were observed at full blooming stage (Zhang et al., 2022).

Statistical analysis

All statistical analyses were conducted by the Spss v.19.0 software. Student's t‐test was used to compare differences between transgenic lines and WT in ChIP‐qPCR assays. One‐way ANOVA followed by Tukey's multiple range test was used for other comparisons. Significance levels are indicated as follows: *, 0.01 < P < 0.05; **, 0.001 < P < 0.01.

Results

CmYAB1 mediates petal fusion in chrysanthemum

CmYAB1 encodes 221 amino acids and contains a zinc finger protein domain at the N terminus and a typical YAB domain at the C terminus (Fig. S1A). Phylogenetic analysis showed that CmYAB1 belonged to the YAB3/FIL subfamily and formed a clade with other YABs of Asteraceae plants (Fig. S1B). CmYAB1 localizes to the nucleus and has transcriptional activation activity (Fig. S2).

Our previous study found that overexpression of CmYAB1 in chrysanthemum resulted in a transformation of inflorescence morphology from decorative to pompon, with the ray floret petals even rolling outward (Ding et al., 2019). In addition, we observed that the distal end of the ray floret petals was split. Therefore, we speculated that CmYAB1 might also be involved in the regulation of petal fusion. To test this, we conducted detailed morphological analyses on newly generated overexpression lines. Expression analysis of the CmYAB1 overexpression lines revealed a 1.7‐ to 3.9‐fold upregulation relative to its already high level in the WT (Figs 1a(a1–d3), S3). In WT plants, only 15.1% of the outermost ray floret petals exhibited splitting (Fig. 1b). In the CmYAB1 overexpression lines (OX‐15, OX‐17, and OX‐44), the split petal percentages increased to 51.4%, 61.8%, and 66.6%, respectively (indicated by arrows in Fig. 1a(b3–d3)). These changes were positively correlated with transgene expression levels, all showing statistically significant increases compared to the WT (Fig. 1b). In addition to the splitting phenotype, we observed protrusions along the sides of the ray florets (indicated by dots in Fig. 1a(b3–d3)). Furthermore, we analyzed the petal types of the outermost ray florets in these lines. Based on the ratio of CTL to the RFL, the petals were divided into flat petals (CTL/RFL < 0.2), spoon petals (0.2 ≤ CTL/RFL < 0.6), and tubular petals (0.6 ≤ CTL/RFL ≤ 1) (Song et al., 2018). In the WT, the proportion of flat petals was highest, accounting for up to 92% (Fig. 1c). In the overexpression lines OX‐15, OX‐17, and OX‐44, the proportion of the flat petal type increased markedly to 97%, 96%, and 98%, respectively (Fig. 1c). These findings demonstrated that CmYAB1 negatively regulates petal fusion in chrysanthemum.

Fig. 1.

Phenotype analysis and morphological observation of CmYAB1 transgenetic chrysanthemum. (a) Inflorescence and floret morphology of wild‐type (WT) and CmYAB1 ectopic expression and knockdown lines at the full blooming stage. (a1–g1) Lateral view of the inflorescence; (a2–g2) top view of the inflorescence; (a3–g3) the top view of ray florets in WT (a3), overexpression lines (b3–d3) and knockdown lines (e3–g3). The red arrows indicate petal splitting, and the yellow dots indicate protrusions along the sides of the ray florets. Bars, 2 cm. (b) The histogram of the frequency of splitting petals of ray florets. (c) The bar chart of the frequency of occurrence flat, spoon, and tubular florets. (d) Morphological observation of CmYAB1 transgenic plants when the inflorescences were 4–6 mm in diameter. (a1, a3, a4, b1, b3, b4, c1, c3, c4) The inflorescences of WT (a1, a3, a4), overexpression lines (b1, b3, b4) and knockdown line (c1, c3, c4,) with phyllaries removed. Panels (a2, a5, b2, b5, c2, c5) depict lateral views of the ray florets, while panels (a6, b6, c6) provide top views. Together, these perspectives illustrate both the degree of petal fusion and the splitting phenotype observed in the petals. The red range lines (a2, a5, c2, c5) indicate the fused part of the ray petals; the arrows indicate the split part. Bars, 500 μm. Data are presented as mean ± SD. *, 0.01 < P < 0.05; **, 0.001 < P < 0.01.

To validate the inhibitory role of CmYAB1 in petal fusion, we generated artificial microRNA (amiR) silencing lines. Three independent lines (amiR‐61, ‐62, and ‐64) with CmYAB1 expression reduced to 46–54% of the WT level were selected for further analysis (Fig. S3). The outermost ray flower petals of the transgenic lines exhibited different extents of fusion, even becoming tubular petals (Fig. 1a(e1–g3)). Then, the fusion degree and fusion rate of the outermost ray floret petal in knockdown lines were analyzed. In the amiR‐61, 62, and 64 lines, the split petal percentages in the outermost ray florets decreased to 6.1%, 5.7%, and 4.6%, respectively, which were significantly lower than those of the WT (Fig. 1b). Correspondingly, 85%, 78%, and 81% of the outermost whorl of ray flower in amiR‐61, 62, and 64 were spoon and tubular type, which were significantly higher than the 8% fusion rate (spoon and tubular type) of the WT (Fig. 1c). By contrast, the disk florets of YAB1 transgenic plants showed no significant difference compared to the WT. Furthermore, no polarity defects, as reported in other species such as Arabidopsis (Sawa et al., 1999; Eshed et al., 2004), were observed in these transgenic lines (Fig. S4). These results indicate that CmYAB1 suppresses petal fusion.

To clarify the period when CmYAB1 exerts its function, flower buds of 2–3, 4–6, and 8–10 mm in diameter were taken from the WT and transgenic lines, and their morphology was examined under stereoscopic and SEM. The results showed that when the flower buds developed to 2–3 mm, the five petal primordia of the ray florets in the transgenic lines developed simultaneously, with no significant difference in petal morphology from the WT (Fig. S5A). When the flower buds developed to 4–6 mm, the three ventral petal primordia of the WT ray elongated simultaneously, forming a tongue‐like shape, while the other two dorsal petal primordia stopped growing shortly after initiation, with lengths shorter than the ventral ones (Fig. 1d(a1–a6)). Whereas, the distal end of the three ventral petal primordia of the overexpression line showed splitting (Fig. 1d(b1–b6)). However, in the amiR plant, the two dorsal petal primordia continued to grow and formed ray flowers with increased fusion degree compared to the WT (Fig. 1d(c1–c6)). When the buds developed to 8–10 mm, the phenotypes of petal splitting and fusion were more evident in the transgenic lines (Fig. S5B). In summary, CmYAB1 participates in the regulation of chrysanthemum floret morphology by affecting the fusion degree of petals and exerts its function when the flower buds develop to 4–6 mm.

CmCUC2 is involved in regulating the sympetalous ray floret development

To screen for regulatory factors involved in petal fusion, we performed LMD to isolate ray floret primordia from the flat‐petaled cultivar ‘Nannong Taibai’ and the tubular‐petaled cultivar ‘Jinsongyue’. RNA was extracted and subjected to quantitative RT‐PCR analysis of flower development‐related genes (Ma et al., 2024). The results showed that the expression level of a CUC homolog – a boundary‐specification gene involved in regulating organ fusion – was significantly lower in ‘Jinsongyue’, measuring only 6% of that in ‘Nannong Taibai’. This result suggests that this gene may play a regulatory role in petal fusion (Fig. S6). Evolutionary phylogenetic analysis indicated that this CUC homolog clusters within the same subclade as AtCUC2 and was therefore designated CmCUC2 (Fig. S7). The CmCUC2 gene has a CDS of 1014 bp in length, encoding 337 amino acids and containing an NAC conserved domain (Fig. S8A). CmCUC2 localizes to the nucleus and has transcriptional activation activity (Fig. S8B,C). Based on the observation that CmYAB1‐silenced lines exhibited enhanced petal fusion, with some ray florets even developing tubular petals resembling those of the ‘Jinsongyue’ cultivar, we hypothesized that CmCUC2 expression might also be downregulated in these CmYAB1‐silenced plants. To test this hypothesis, we quantified CmCUC2 expression levels in the amiR‐CmYAB1 lines. The results demonstrated a significant reduction in CmCUC2 expression, reaching only 0.02‐, 0.05‐, and 0.11‐fold of WT levels (Fig. 2a). These findings suggest that CmCUC2 may function downstream of CmYAB1 in regulating petal fusion.

Fig. 2.

Phenotype characterization and morphological observation of CmCUC2 transgenetic chrysanthemum. (a) Reverse transcription quantitative polymerase chain reaction analysis of CmCUC2 in amiR‐CmYAB1 transgenic lines. (b) Inflorescence and floret morphology of wild‐type (WT) and CmCUC2 knockdown and ectopic expression lines at the full blooming stage. (a1–g1) Lateral view of the inflorescence; (a2–g2), top view of the inflorescence; The top view of ray florets in WT (a3), knockdown lines (b3–d3), and overexpression lines (e3–g3). The red arrows indicate petal splitting, and the yellow dots indicate protrusions along the sides of the ray florets. (c) The bar chart of the frequency of occurrence flat, spoon and tubular florets. (d) The histogram of the frequency of splitting petals of ray florets. (e) Morphological observation of CmCUC2 transgenic plants when the inflorescences were 4–6 mm in diameter. (a1, a3, a4, b1, b3, b4, c1, c3, c4) The inflorescences of WT (a1, a3, a4), overexpression lines (b1, b3, b4), and knockdown line (c1, c3, c4) with phyllaries removed. Panels (a2, a5, b2, b5, c2, c5) depict lateral views of the ray florets, while panels (a6, b6, c6) provide top views. Together, these perspectives illustrate both the degree of petal fusion and the splitting phenotype observed in the petals. The red range lines (a2, a5, c2, c5) indicate the fused part of the ray petals; the arrows indicate the split part. Bars, 500 μm. Data are presented as mean ± SD. Different letters denote significant differences by two‐way ANOVA with Tukey's multiple comparisons test (P < 0.05). **, 0.001 < P < 0.01.

To functionally characterize CmCUC2, we generated and analyzed three representative silencing lines (amiR‐CmCUC2‐76, ‐107, and ‐126), which displayed a reduction in CmCUC2 expression to 0.56–0.72 of WT levels (Fig. S9A). Similar to the phenotypic characteristics of the CmYAB1 knockdown lines, CmCUC2 amiR lines exhibited partially fused ray petals (Fig. 2b(b1–d3)). Statistical analysis revealed a pronounced increase in ray floret fusion rates across the amiR‐CmCUC2 lines (line 76: 97%; 107: 82%; 126: 80%) compared to WT plants (8%) (Fig. 2c). This was accompanied by a marked shift from WT's predominantly flat petals (92%) to fused spoon‐shaped and tubular petals in the transgenic lines, with line 76 exhibiting the most severe phenotype, including 44% tubular petals (Fig. 2c). Moreover, the degree of splitting in ray florets was significantly reduced in the overexpression lines (Fig. 2c). Subsequently, we engineered a miR164‐resistant version of CmCUC2 (mCmCUC2) by introducing synonymous mutations into its predicted binding site and overexpressed it in chrysanthemum. Three representative transgenic lines (OX‐6, ‐7, and ‐9), which showed significant CmCUC2 upregulation (Fig. S9B), were selected for phenotypic analysis (Fig. S9B). Overexpression of CmCUC2 not only promoted a higher proportion of flat‐petaled ray florets and altered inflorescence morphology through abaxial curling of the ray florets but also induced conspicuous splitting at the distal ends of the petals (Fig. 2b(e1–g3),c). Statistical analysis of the outermost whorl ray florets revealed a stark contrast between WT and CmCUC2‐overexpressing lines. While < 20% of WT petals exhibited any splitting, over 90% of transgenic petals were affected (91.3%–96.7%; Fig. 2d). Furthermore, CmCUC2‐OX transgenic plants displayed petal splitting in disk florets (white arrow, Fig. 2e(b3)), and the petals of these splitting disc florets were white, a trait typical of ray florets (Fig. S10). Although both CmCUC2 and CmYAB1 overexpression caused outward‐curling and split petals, CmCUC2 lines indicate a more complete failure of fusion. These results demonstrate that CmCUC2 suppresses petal fusion and is essential for sympetalous ray floret development.

We determined that the regulation of petal separation by CmCUC2 also initiates as early as the 4‐ to 6‐mm floral bud stage. At this stage, overexpression prevented the fusion of petal primordia, while silencing led to enhanced fusion and prolonged growth of dorsal primordia (Fig. 2e). These effects, which intensified during later development (Fig. S11), establish that CmCUC2 promotes petal separation to define chrysanthemum flower shape, functioning similar to CmYAB1.

CmCUC2 has a similar abaxial expression pattern with CmYAB1

In order to detect the spatial and temporal expression pattern of CmYAB1 and CmCUC2, in situ hybridization was applied to various sizes of inflorescence buds. Conservatively, transcripts of CmYAB1 are detected in the abaxial domain of the bract (Fig. 3a). At stage I of floret development, CmYAB1 is diffusively expressed in the floret meristem and gathers at the incipient petal primordia (Fig. 3a). At stage II, the transcripts of CmYAB1 are progressively restricted to the abaxial domain of petal primordia (Fig. 3b). As the primordia continued to develop, this abaxial expression pattern became increasingly pronounced. At stage III, CmYAB1 transcripts are detected both in the abaxial region of petal primordia and emerging stamen primordia (Fig. 3c). At stage IV, the signal of CmYAB1 was also detected in the abaxial domain of emerging pistil primordia (Fig. 3d). At stage V, with the development of stamen stagnating, the CmYAB1 expression level in the stamen primordia was weakened or even imperceptible, but maintained in the abaxial region of petal primordia and pistil primordia (Fig. 3e). At stage VI and beyond, CmYAB1 was expressed in the abaxial domain of the petal and style and highly transcribed in the lateral region of the styles (Fig. 3f). Moreover, in disk florets, CmYAB1 is also expressed in the abaxial domain of lateral floral organs, consistent with its expression pattern in ray florets (Fig. S12).

Fig. 3.

RNA in situ hybridization analysis of CmYAB1 in the chrysanthemum inflorescences and florets. (a) Longitudinal sections of chrysanthemum inflorescence. FM, floret meristem; IM, inflorescence meristem. (b–e) Longitudinal section of floret at the stage II (b), stage III (c), stage IV (d), and stage V (e). (f) Transverse section of floret at the stage VI. (g) Longitudinal section of floret with CmYAB1 sense probe as negative control. D‐rpe, dorsal ray petal primordium; rpi, ray pistil primordium; rst, ray stamen primordium; V‐rpe, ventral ray petal primordium. Bars, 100 μm.

For CmCUC2, the strip expression pattern was detected at the boundary between floret meristems in the inflorescence (Fig. 4a). CmCUC2 is also expressed at the boundary between the petal primordia and meristem (Fig. 4b), between the stamen primordia and meristem (Fig. 4c). Intriguingly, in contrast to the previously reported CUC2 boundary expression pattern, CmCUC2 transcripts are also detected in the petal primordia at stage II (Fig. 4b), and are mainly detected in the abaxial domain of ventral petal primordia at stage III (Fig. 4c). The abaxial expression pattern is maintained in the elongated ventral petal primordia until stage V, while the distal domain expression pattern was observed in the dorsal petal primordia at stages IV and V (Fig. 4d,e). In addition, the CmCUC2 signal is detected in the abaxial region of pistil primordia at stages IV and V, and in the stamen primordia at stage III and IV and undetected at stage V, similar to the YAB1 expression pattern in stamens (Fig. 4c–e). At stage VI and beyond, CmCUC2 was expressed in the abaxial domain of the petal and mainly transcribed in the lateral region of the styles (Fig. 4f). CmCUC2 expression is also observed at organ boundaries and in the abaxial domain within disk florets (Fig. S12). Collectively, our findings reveal that CmCUC2 exhibits a novel abaxial expression domain in lateral organs, in addition to its conserved boundary‐specific expression. The partially overlapping expression patterns of CmCUC2 and CmYAB1 suggest a potential direct interaction between these two regulators, indicating their functional interplay in coregulating floral morphogenesis in chrysanthemum.

Fig. 4.

RNA in situ hybridization analysis of CmCUC2 in the chrysanthemum inflorescences and florets. (a) Longitudinal sections of chrysanthemum inflorescence. IM, inflorescence meristem; FM, floret meristem. (b–e) Longitudinal section of floret at the stage II (b), stage III (c), stage IV (d), and stage V (e). (f) Transverse section of floret at the stage VI. (g) Longitudinal section of floret with CmCUC2 sense probe as negative control. D‐rpe, dorsal ray petal primordium; rpi, ray pistil primordium; rst, ray stamen primordium; V‐rpe, ventral ray petal primordium. Bars, 100 μm.

CmYAB1 promotes the expression of CmCUC2 by binding to its promoter

Given the overlapping expression and similar transgenic phenotypes of CmYAB1 and CmCUC2, we hypothesized that they function in a common genetic pathway regulating petal fusion. To test whether CmCUC2 is a direct target of CmYAB1, we first analyzed an c. 2000‐bp sequence upstream of its start codon. In silico promoter analysis predicted five conserved WATNAT motifs – the core cis‐elements for YAB binding – concentrated within the region from −879 to −136 bp. To focus on this core regulatory region, we cloned a 1361‐bp fragment of the CmCUC2 promoter for subsequent functional validation (Fig. 5b) (Franco‐Zorrilla et al., 2014). This bioinformatic evidence prompted us to hypothesize that CmYAB1 might regulate CmCUC2 through direct promoter binding. Experimental validation commenced with Y1H assays, which confirmed CmYAB1's physical interaction with the CmCUC2 promoter (Fig. 5a). To resolve binding specificity, EMSAs were performed using affinity‐purified GST‐CmYAB1 protein and biotin‐labeled probes spanning predicted WATNAT elements. The results showed that the biotin‐labeled normal CmCUC2 promoter probe at the fourth element (CATCATT) (−928 to −935 bp) can bind to the CmYAB1 protein. As the concentration of the unlabeled probe increased, the band intensity gradually weakened. By contrast, the addition of unlabeled mutant probe did not significantly alter the band intensity. On the other hand, the probe with a mutation in the binding element IV could not bind to the CmYAB1 protein (Fig. 5b,c). The other four predicted binding elements were unable to bind to the CmYAB1 protein (Fig. S13), indicating that CmYAB1 can specifically bind to the cis‐acting element IV on the CmCUC2 promoter under in vitro conditions. For further analysis, a dual‐luciferase (LUC) transaction assay in chrysanthemum protoplasts was performed. The ratio of LUC/REN was significantly increased upon cotransformation with 35S::CmYAB1 + ProCmCUC2:LUC as compared to the vector control (Fig. 5d). It indicated that CmYAB1 positively regulated the transcription of CmCUC2.

Fig. 5.

CmYAB1 binds to the promoter and promote the expression of CmCUC2. (a) Yeast one‐hybrid experiment of the binding of CmYAB1 to CmCUC2 promoter. (b) Schematic diagram of the CmCUC2 promoter used in electrophoretic mobility shift assay (EMSA) showed the binding elements. (c) EMSA of the binding of CmYAB1 protein to the fourth WATNAT elements of CmCUC2 promoter in vitro. MT, muted probe. (d) Dual‐luciferase (LUC) assay of 35S::CmYAB1 effector with CmCUC2 promoter firefly luciferase reporter constructs in chrysanthemum protoplasts. (e) ChIP‐qPCR assay showed relative binding of CmYAB1 to the CmCUC2 promoter. Data are presented as mean ± SE of three biological replicates. Student's t‐test: **, 0.001 < P < 0.01. Data are presented as mean ± SD. Different letters denote significant differences by two‐way ANOVA with Tukey's multiple comparisons test (P < 0.05). **, 0.001 < P < 0.01.

To validate in planta binding, chromatin immunoprecipitation (ChIP) was performed using 35S::CmYAB1‐GFP transgenic lines. According to the WATNAT elements on the CmCUC2 promoter, fragments P1–P5, corresponding to YAB‐I to V, and fragment P7 containing both YAB‐III and IV, along with the P6 fragment – which lacks a WATNAT element and serves as a negative control – were selected for qPCR primer design. ChIP‐qPCR assay demonstrated that the promoter of CmCUC2 was significantly enriched in the P3, P4, and P7 regions through recognizing CmYAB1‐GFP protein antibodies, but not enriched in the P1, P2, P5, and P6 regions (Fig. 5e). Therefore, CmYAB1 can directly bind to the fourth WATNAT element of the CmCUC2 promoter. Together, CmYAB1 directly binds to the promoter of CmCUC2 to promote its expression in vitro and in vivo.

Genetic analysis of the relationship between CmYAB1 and the downstream gene CmCUC2

To further obtain genetic evidence for the relationship between CmYAB1 and CmCUC2, a viral vector containing artificially interfering small RNAs CmCUC2: CaLCuV‐amiR‐CmCUC2 was constructed for transient transformation in the OX‐CmYAB1 transgenic background. The empty vector‐transformed OX‐CmYAB1 plants exhibited a phenotype of outward petals and splitting distal end petals (Fig. 6a(a1–a3)), whereas the OX‐CmYAB1 amiR‐CmCUC2 exhibited a phenotype of increased petal fusion and decreased petal curvature (Fig. 6a(b1–d3)). RT‐qPCR analysis showed that the expression of CmCUC2 in OX‐CmYAB1 amiR‐CmCUC2‐1#, 2#, and 5# plants was significantly lower than that in the empty vector‐transformed plants (Fig. 6b). Further statistical analysis showed that the proportion of split ligules in the control was 50.3%, whereas in lines OX‐CmYAB1 amiR‐CmCUC2‐1#, 2#, and 5#, it decreased to 18.0%, 14.3%, and 11.3%, respectively (Fig. 6c). Moreover, the fusion rate (spoon and tubular petal) of the outermost whorl florets of OX‐CmYAB1 amiR‐CmCUC2‐1#, 2# and 5# transiently transformed lines exceeded 78%, which was significantly higher than the 18% fusion rate in the control plant, and more than 64% were spoon‐like petals (Fig. 6d). The results described previously indicate that transient knockdown of CmCUC2 expression can rescue the phenotype of outward and splitting petals in OX‐CmYAB1 transgenic lines, demonstrating that CmYAB1 is upstream of CmCUC2 and affects chrysanthemum flower type by regulating CmCUC2.

Fig. 6.

Phenotype analysis of CaLCuV‐amiR‐CmCUC2 infected OX‐CmYAB1 chrysanthemum at the full blooming stage. (a) Inflorescence and floret morphology of pCVA‐Empty infected OX‐CmYAB1 and CaLCuV‐amiR‐CmCUC2 infected OX‐CmYAB1 lines. (a1, a2, b1, b2, c1, c2, d1, d2) The lateral and top view of inflorescence of pCVA‐Empty and CaLCuV‐amiR‐CmCUC2 infected OX‐CmYAB1 lines 1, 2, and 5. (a3, b3, c3, d3) The top view of ray florets of pCVA‐Empty and CaLCuV‐amiR‐CmCUC2 infected OX‐CmYAB1 lines. Arrow showed unfused and fused petal. The arrows indicate the split petals; the arrowheads indicate the fused site of the petals. Bars, 2 cm. (b) the expression analysis of CmCUC2 in the transiently transformation lines by reverse transcription quantitative polymerase chain reaction. (c) The histogram of the frequency of splitting petals of ray florets. (d) The column diagram of the frequency of occurrence flat, spoon, and tubular florets. Data are presented as mean ± SD. Different letters denote significant differences by two‐way ANOVA with Tukey's multiple comparisons test (P < 0.05).

Distinct spatiotemporal expression architecture of CmCUC2 in tubular‐petal cultivars

The experiments described previously demonstrated that CmYAB1 is involved in corolla tube formation in chrysanthemum by regulating CmCUC2. In these amiR‐CmYAB1 and amiR‐CmCUC2 lines, the degree of petal fusion increased, resulting in spoon or tubular petals. This prompted us to investigate the spatiotemporal expression patterns of both CmYAB1 and CmCUC2 in tubular‐petal cultivars. To this end, we analyzed their expression in two tubular‐petal cultivars, ‘Jinsongyue’ (Fig. S6B) and ‘Anastasia White’ (Fig. S14). The spatiotemporal expression pattern of CmYAB1 in tubular‐petal cultivars was similar to that in flat‐petal types, with specific expression detected in the abaxial side of floral organ primordia, including petal, stamen, and pistil primordia (Fig. 7a–j). By contrast, the expression pattern of CmCUC2 showed notable alterations. In ‘Jinsongyue’ and ‘Anastasia White’, CmCUC2 was primarily expressed at the boundaries between the meristem and petal primordia, between petal and stamen primordia, and between stamen and pistil primordia (Fig. 7k–t). Unlike its expression pattern in the flat‐petal cultivar ‘Jinba’, little to no expression was observed in the abaxial regions of floral organ primordia in tubular‐petal cultivars, and only weak expression was detected in the distal domain of dorsal petals (Fig. 7k–t). These findings indicate that the expression pattern of the CmCUC2 varies significantly across different petal types. Moreover, silencing CmCUC2 resulted in the transformation of flat petals into tubular ones, suggesting that the differential expression of CmCUC2 may regulate petal‐type differentiation and serve as a key genetic factor underlying the diversity of petal morphology in chrysanthemum.

Fig. 7.

RNA in situ hybridization analysis of CmYAB1 and CmCUC2 in the tubular‐petal cultivars ‘Jinsongyue’ and ‘Anastasia White’. (a–j) CmYAB1 was expressed in the abaxial domain of petal, stamen, and pistil primordia in ‘Jinsongyue’ (a–e) and ‘Anastasia White’ (f–j). (k–t) The transcripts of CmCUC2 were primarily detected at the boundary between the meristem and petal primordia, between petal and stamen primordia, and between stamen and pistil primordia in ‘Jinsongyue’ (k–o) and ‘Anastasia White’ (p–t). D‐rpe, dorsal ray petal primordium; rpi, ray pistil primordium; rst, ray stamen primordium; V‐rpe, ventral ray petal primordium. Bars, 100 μm.

Sustained expression of CmCYC2c in the dorsal petal primordia in CmYAB1 ‐ and CmCUC2 ‐knockdown lines

Some studies have demonstrated that CYC serves as a pivotal regulator of petal fusion in Asteraceae species (Chapman et al., 2008, 2012; Kim et al., 2008; Huang et al., 2016; Chen et al., 2018; Shen et al., 2021; Rong et al., 2024; Zhang et al., 2024). To investigate whether CmYAB1 and CmCUC2 influence petal fusion by modulating CmCYC2c expression, the CmCYC2c spatiotemporal expression pattern was analyzed via in situ hybridization in CmYAB1 and CmCUC2 knockdown lines. Consistent with our earlier reports, WT plants exhibited stage‐specific CmCYC2c expression: From stage III onward, expression was predominantly localized to the distal regions of ventral petal primordia, while gradually diminishing in dorsal petal primordia (Fig. 8a–c) (Broholm et al., 2008; L. Ding et al., 2020). Sustained CmCYC2c expression in ventral primordia promoted their elongation, whereas reduced expression in dorsal primordia correlated with growth termination (Broholm et al., 2008; L. Ding et al., 2020; Qiu et al., 2023). In CmYAB1‐knockdown lines, the expression of CmCYC2c in dorsal primordia was significantly upregulated at stage III, with detectable signals persisting through stages IV and V, indicating a prolonged duration of CmCYC2c expression in the dorsal primordia (Fig. 8d–f). This alteration in gene expression corresponded to morphological changes: The dorsal primordia exhibited sustained elongation, ultimately developing into spoon‐shaped or tubular petals with increased fusion. Similarly, CmCUC2 knockdown lines displayed persistent CmCYC2c signals in both dorsal and ventral primordia during stage III‐V, with elevated expression in dorsal primordia compared to WT (Fig. 8g–i). Consequently, elongation of normally suppressed dorsal primordia occurred, forming highly fused spoon‐shaped or tubular petals. These findings demonstrate that the knockdown of either CmYAB1 or CmCUC2 upregulates CmCYC2c expression in the distal domain of dorsal petal primordia, which corresponds to the observed elongation of the dorsal petals. Together, the results suggest that the CmYAB1‐CmCUC2 module acts upstream of CmCYC2c in regulating petal fusion in chrysanthemum. However, the precise regulatory mechanisms require further investigation.

Fig. 8.

Spatiotemporal expression analysis of CmCYC2c in the amiR‐CmYAB1 and amiR‐CmCUC2 transgenic plants by RNA in situ hybridization. The signals of CmCYC2c were detected in WT (a–c), amiR‐CmYAB1 (d–f), amiR‐CmCUC2 (g–i). And the longitudinal section of floret at the stage III (a, d, g), stage IV (b, e, h), and stage IV (c, f, i). D‐rpe, dorsal ray petal primordium; V‐rpe, ventral ray petal primordium. The arrow shows the signals of CmCYC2c in the ventral ray petal primordium. Bars, 100 μm.

Taken together, our findings demonstrate that petal fusion in chrysanthemum ray florets is fine‐tuned by a dorsal‐specific regulatory module. This module is initiated by abaxially expressed CmYAB1, which activates distal CmCUC2 expression. The resulting CmYAB1‐CmCUC2 module then suppresses dorsal petal growth – potentially by downregulating factors such as CmCYC2c to precisely control the extent of petal fusion and determine final petal shape (Fig. 9).

Fig. 9.

Working model of the genetic regulatory network controlling sympetalous flower development in Chrysanthemum morifolium. In the dorsal petals of the ray florets in chrysanthemum, CmYAB1 exhibits specific expression in the abaxial domain. In flat‐shaped petals, CmYAB1 directly promotes the distal expression of CmCUC2, subsequently suppressing the distal expression of CmCYC2c within the dorsal petal. When the expression of CmYAB1 is reduced, the expression of CmCUC2 correspondingly decreases. This attenuation of suppression allows for increased expression of CmCYC2c, which corresponds to the observed elongation of the dorsal petal. This enhanced growth ultimately produces petals with higher fusion, manifesting as spoon‐shaped or even tubular petals. Light red indicates the CmYAB1 expression domain; blue represents the CmCUC2 expression domain; yellow denotes the CmCYC2c expression domain; dark red marks the co‐expression domain of CmYAB1 and CmCUC2.

Discussion

CmYAB1 is implicated in the development of sympetalous flower in chrysanthemum

The YAB genes have various functions in plant growth and development, including polar establishment of lateral organs, lateral organ growth and expansion, flower development, and plant architecture (Sawa et al., 1999; Goldshmidt et al., 2008; Bonaccorso et al., 2012; Boter et al., 2015; Ding et al., 2021; Wang et al., 2022). In Arabidopsis, the vegetative YABs (YAB1, 2, 3, and 5) are specifically expressed in the abaxial domain of developing lateral organ primordia and specify the abaxial cell fate (Siegfried et al., 1999; Sarojam et al., 2010). The Arabidopsis fil mutant displays clusters of filamentous structures and flowers with altered numbers and shapes of floral organs (Sawa et al., 1999; Siegfried et al., 1999). Antirrhinum GRAM (YAB1 homologous) mutation disrupt the adaxial–abaxial asymmetry and growth of lateral organs, and occasionally produce needle‐like leaves (Golz et al., 2004). In tomato, the FAS gene (YAB2 homolog) negatively regulates carpel number during flower development, promoting fruit size. Different from the abaxial expression pattern in Arabidopsis and Antirrhinum, FAS transcripts were detected in stamen and carpel primordia and placental tissue without polarity (Cong et al., 2008). Our previous study found that chrysanthemum CmDRP is widely expressed in multiple tissues without polar localization and regulates plant height by directly suppressing the expression of the gibberellin biosynthesis gene CmGA3ox1 (Zhang et al., 2022). In rice, YAB genes are transcribed in the lateral organs in a nonpolar manner and function divergent with Arabidopsis (Jang et al., 2004; Yamaguchi et al., 2004; Dai et al., 2007). TONGARI‐BOUSHI genes, belonging to the FIL clade of the YAB family, are expressed in bract and lateral organ primordia and redundantly regulate the maintenance and fate of all reproductive meristems, ensuring proper inflorescence and spikelet development in rice (Tanaka et al., 2012, 2017). The rice OsYAB1 regulates plant architecture through feedback regulation of GA biosynthesis (Dai et al., 2007). Similarly, ectopic expression of OsYABBY4 in rice results in a semi‐dwarf phenotype and alterations in the number of floral organs by modulating the gibberellin pathway (Yang et al., 2016). In maize (Zea mays), members of the YAB gene family are transcribed adaxially in leaf primordia and promote lateral organ outgrowth rather than determining cell fate (Juarez et al., 2004). Therefore, the expression patterns and functions of YABs differ between dicots and monocots, and even between dicots. In this study, we found that CmYAB1 is transcribed in the abaxial region of floral organ primordia, similar to the expression pattern in Arabidopsis. And CmYAB1 ectopic expression in chrysanthemum exhibits downward curling of petals. A similar phenomenon was observed in A. thaliana leaves, where overexpression of FIL led to epinastic leaves. However, no obvious abaxialized petals were detected in our study (Fig. S4) (Bonaccorso et al., 2012; Boter et al., 2015). By contrast, fusion defects were observed in CmYAB1 transgenic chrysanthemum (Fig. 1) and have not been reported in other species. The flower morphogenesis of chrysanthemum is distinct from that of Arabidopsis, asymmetrical growth of dorsal‐ventral petal resulted in a flat petal, while symmetrical growth (synchronous growth) led to a tubular petal, intermediate will grow into a spoon petal. Overexpression of CmYAB1 displays unfused petals; knockdown plants show more spoon‐like and tubular florets, suggesting CmYAB1 inhibiting symmetry growth of petals and the development of sympetalous flower and regulates flower morphogenesis in chrysanthemum. Taken together, YABs have divergent functions in different species, probably because of variations in petal morphology.

Context‐dependent effect of CmCUC2 on flower morphogenesis

CUCs function redundantly to establish boundaries between adjacent organs and at the meristem‐organ junction by modulating auxin distribution to inhibit growth at boundaries (Takada et al., 2001). Mutations in CUC genes cause sepals and stamens fusion but induce no fusion petal defects (Ishida et al., 2000; Nahar et al., 2012; Irepan Reyes‐Olalde et al., 2013; Kamiuchi et al., 2014). The absence of petal fusion phenotypes in these mutants suggests novel regulatory networks underlying petal boundary formation. In Petunia, NAM (CUC homologous) transcripts are detectable in corollas by qPCR (Zhong et al., 2016), though spatial expression patterns in corolla primordia remain uncharacterized by RNA in situ hybridization. Moreover, virus‐induced silencing of NAM resulted in increased petal separation rather than an increase of sympetalous flower. In the nam mutant, interwhorl fusions between stamens and petals can be found, but no intrawhorl fusions are observed (Souer et al., 1996). By contrast, our study revealed that CmCUC2 overexpression induces petal splitting, while amiR‐CmCUC2 lines develop fused petals (increased spoon/tubular types). In the flat‐petal cultivar ‘Jinba’, CmCUC2 mRNA is localized not only at floral organ boundaries but also in the abaxial regions of petals (Fig. 4). However, in tubular‐petal cultivars, its expression is markedly restricted, primarily confined to organ boundaries, virtually absent in the abaxial regions, and weak in the distal domain of the dorsal petal (Fig. 7). Our results demonstrate that reduced CmCUC2 activity can promote sympetalous flower development, which has not been reported. Context‐dependent gene expression is responsible for modifying function. For example, CUP gene expression at the boundary and in serrations suppresses growth, whereas its expression in the intermediate region promotes growth (Nikovics et al., 2006; Blein & Laufs, 2008; Rebocho et al., 2017). The different expression patterns of CmCUC2 may contribute to the variations of function. Moreover, CmCUC2 is expressed in the distal domain of the inhibited growth dorsal petal primordia and transcribed in the abaxial region of elongated ventral petal primordia, indicating that it has divergent roles in the dorsal and ventral petal. We hypothesize that its distal expression on the dorsal petal may mediate growth suppression, whereas its abaxial expression on the ventral petal may facilitate growth promotion. The context‐dependent functions for boundary genes are beneficial to the development and evolution of sympetaly (Zhong & Preston, 2015). Besides the fusion defects in CmCUC2 transgenic plants, changes in inflorescence morphogenesis were observed; the petals of the CmCUC2 overexpression line are curved outward, similar to CmYAB1 overexpression lines, suggesting that CmCUC2 together with CmYAB1 promotes abaxial–adaxial asymmetry growth of elongated ventral petals. In summary, CmCUC2 plays a crucial role in the development of sympetalous flowers and inflorescence morphology.

CmYAB1‐CmCUC2 module elaborates flower development in chrysanthemum

YABs regulate shoot apical meristem activity in a non‐cell‐autonomous manner through boundary LATERAL SUPPRESSOR, which is expressed at the boundary between lateral organs and the SAM (Greb et al., 2003; Goldshmidt et al., 2008). YABs physically interact with LEUNIG (LUG), LEUNIG_HOMOLOG (LUH), the LUG‐associated coregulator SEUSS, and related SEUSS‐LIKE proteins, forming the LUG‐YAB complex (Stahle et al., 2009). Given that similar cuc1 cuc2 mutant phenotypes including fused cotyledons and loss of the embryonic SAM (Shoot Apical Meristem) were observed in seu slk2 mutants, Stahle et al. (2009) speculate that the LUG‐YAB complex may regulate CUC activity. So far, further studies on the relationship between YABs and boundary genes have not been reported. In this study, we found that the CmYAB1 transgenic plant showed fusion defects and the expression of CmCUC2 is significantly downregulated in CmYAB1 knockdown lines (Figs 1, 2). Genetic and molecular experiments have fully demonstrated that YAB1 binds to the CmCUC2 promoter and promotes its expression (Figs 2, 3, 4, 5, 6).

CUC2 is needed to promote the formation of auxin convergence points by regulating PIN‐FORMED1 (PIN1). CUC1 and CUC2 expression is repressed by miR164, and that of CUC2 is also repressed by auxin. CUC2 promotes the formation of auxin maxima by reorienting the auxin efflux carrier PIN1, which in turn downregulates CUC2 expression, creating a powerful feedback loop (Bilsborough et al., 2011; Hu et al., 2024). On the other hand, YABs regulate leaf development by limiting auxin flow. Loss of YAB activity greatly affects auxin distribution and PIN1 localization (Sarojam et al., 2010). In this study, we found that CmYAB1 directly promotes CmCUC2 expression to suppress dorsal petal growth during sympetalous flower development. In addition, the auxin gradient affects the development of distinct florets (flat ray florets and tubular disk florets) in Asteraceae capitulum (Zoulias et al., 2019). Thus, the CmYAB1‐CmCUC2 module may exert its function depending on PIN1‐mediated auxin distribution, but detailed research is needed to prove this.

CYC was the first identified key transcription factor controlling floral symmetry (Luo et al., 1999). In A. majus, CYC is exclusively expressed in dorsal petals but absent in ventral petals, indicating that its spatiotemporal expression pattern is closely associated with floral symmetry (Luo et al., 1999). Furthermore, comparative studies of Arabidopsis thaliana and its relative Iberis amara revealed that persistent expression of IaTCP (a CYC homolog) in dorsal regions drives prolonged dorsal petal elongation, forming zygomorphic flowers, whereas AtTCP1 is expressed only during early floral primordia stages in Arabidopsis, resulting in uniform dorsal‐ventral growth and radial symmetry (Cubas et al., 2001). This highlights the critical role of CYC expression timing in determining floral symmetry. Studies of CYC‐RAD‐DIV homologs across Gesneriaceae, Plantaginaceae, and Leguminosae species further demonstrate the relative conservation of floral symmetry regulatory networks (Feng et al., 2006; Gao et al., 2008; Yang et al., 2012; Reardon et al., 2014). The regulatory mechanisms controlling CYC expression have long been a key focus of research. In Asteraceae plants, CYC not only regulates petal symmetry but also influences stamen development, thereby affecting floret identity (Broholm et al., 2008; Chapman et al., 2012). In this study, we observed that in CmYAB1‐ and CmCUC2‐silenced lines, the expression of CmCYC2c in dorsal petal primordia was significantly upregulated compared to WT, with sustained expression throughout development and corresponding dorsal petal elongation (Fig. 8). The altered spatiotemporal expression of CmCYC2c mediated by the CmYAB1‐CmCUC2 module suggests that this module acts upstream to suppress CmCYC2c expression in dorsal petals. However, the precise molecular interplay between CmYAB1‐CmCUC2 and CmCYC2c remains to be elucidated.

Competing interests

None declared.

Author contributions

FC and LD designed the research. LD, XZ, DJ, YZ, WX and SL performed the experiments. LD and XZ wrote the manuscript. AS, JJ, SC and FC supervised the research and revised the manuscript. All authors have approved the final manuscript. LD and XZ contributed equally to this work.

Disclaimer

The New Phytologist Foundation remains neutral with regard to jurisdictional claims in maps and in any institutional affiliations.

Supporting information

Fig. S1 Alignment of amino acid sequences and phylogenetic analysis of CmYAB1 with other YAB proteins.

Fig. S2 Subcellular localization and transcriptional activity analysis of CmYAB1.

Fig. S3 The relative expression level of CmYAB1 in wild‐type and transgenic lines.

Fig. S4 Epidermal cell morphology on adaxial and abaxial surfaces of chrysanthemum ray floret petals observed by SEM.

Fig. S5 Morphological analysis of early inflorescence development in CmYAB1 transgenic lines.

Fig. S6 Screening of CmCUC2 by LMD and RT‐qPCR.

Fig. S7 The phylogenetic tree of CmCUC2 with other CUC proteins.

Fig. S8 Alignment of amino acid sequences, subcellular localization and transcriptional activity analysis of CmCUC2.

Fig. S9 The expression analysis of CmCUC2 in wild‐type and transgenic line by RT‐qPCR.

Fig. S10 Observations of disc floret phenotypes in OX‐mCmCUC2 plants.

Fig. S11 Morphological analysis of early inflorescence development in CmCUC2 transgenic lines.

Fig. S12 RNA in situ hybridization analysis of CmYAB1 and CmCUC2 in the chrysanthemum disc florets.

Fig. S13 Electrophoretic mobility shift assay (EMSA) of CmYAB1 protein binding.

Fig. S14 Images of inflorescence and ray florets in chrysanthemum cultivars ‘Anastasia White’. Images of inflorescence and ray florets in chrysanthemum cultivars ‘Anastasia White’.

Notes S1 Protein sequence.

Table S1 Primer sequences of vector construction.

Table S2 Probe sequences for EMSA.

Table S3 Primer sequences for ChIP‐qPCR.

Please note: Wiley is not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

Data availability

The data that support the findings of this study are available within the article and its Supporting Information (Figs S1–S11; Tables S1, S2; Notes S1). The accession numbers for the sequence data from this article can be found in the database of the National Center for Biotechnology Information: CmYAB1 (PX528644), CmCUC2 (PX528645), CmCYC2c (PX528646).