Non-cell-autonomous regulation of petal initiation in Arabidopsis thaliana

Flower morphology is important for attracting pollinators, such as flying insects or birds. The arrangement of floral organs in floral buds is roughly classified into two patterns: the spiral pattern, observed in several clades, including basal angiosperms (Endress, 2001; Endress and Doyle, 2007); and the whorled (or concentric) pattern seen in many other flowering plants. Flowers generally consist of four types of floral organ: sepals or outer tepals; petals or inner tepals; stamens; and carpels. In whorled flowers, different organ identities in each whorl of the flower are specified by distinct complexes of floral homeotic proteins. This regulation is explained by the floral ABC model or the more advanced ABCE model, which both describe the genetic mechanism for the establishment of floral identity in most flowering plants (Theissen et al., 2016).

Floral organs generally arise in a more or less equally spaced pattern within the whorl. In Arabidopsis thaliana, two medial sepals arise on the adaxial and abaxial sides of the inflorescence meristem, followed by two sepals at the lateral positions at floral stage 3 (Smyth et al., 1990). The second whorl contains petals, which usually occur in alternate positions to the sepals (or outer tepals). This suggests that positional information for the intersepal region is transmitted to the inner petal founder cells to fix the organ positions within the second whorl. However, the molecular mechanism for this positional signaling remains unknown.

Genetic approaches have identified regulators involved in the early development of second whorl organs in Arabidopsis. The petal loss (ptl) mutant has defects in sepal separation and petal initiation (Griffith et al., 1999). PTL encodes a GT2-clade trihelix transcription activator and is expressed in intersepal cells. These cells do not overlap with cells expressing the auxin-responsive DR5 reporter gene, suggesting that PTL regulates petal development indirectly (Brewer et al., 2004; Lampugnani et al., 2012). RABBIT EARS (RBE) encodes a C2H2-type zinc finger protein and its mutation results in the deformation or elimination of petals (Takeda et al., 2004). RBE is expressed in petal primordia cells, in which it negatively regulates the transcription of AGAMOUS, MIR164, TEOSINTE BRANCHED1/CYCLOIDEA/PROLIFERATING CELL FACTOR 4 (TCP4), TCP5 and PUB25 to facilitate the early development of petal primordia (Huang and Irish, 2015; Huang et al., 2012; Krizek et al., 2006; Li et al., 2016, 2021). When the ptl or rbe mutant was combined with the homeotic mutant apetala3 (ap3), the second and third whorl organs of which are replaced by sepals and carpels, respectively, the double mutants showed a reduced number of sepalloid second whorl organs (Griffith et al., 1999; Takeda et al., 2004), suggesting that PTL and RBE are involved in the development of second whorl organs independently of organ identity.

Several pieces of evidence suggest that PTL and RBE regulate petal initiation in the same pathway: first, both mutants show similar petal-loss phenotypes; second, the ptl rbe double mutant resembles the ptl single mutant (Lampugnani et al., 2013); and third, RBE expression is absent in the petal primordia of the ptl mutant (Takeda et al., 2004). These findings suggest that positional signal transduction occurs between the sepal boundary, where PTL is expressed, and the petal primordium, where RBE is expressed.

Here, we uncovered a molecular link between the sepal boundary and petal initiation site and the role of this link in petal initiation. Using fluorescent fusion proteins, we examined the spatiotemporal expression patterns of PTL and RBE in floral buds. PTL was detected in sepal boundaries and RBE was detected in petal initiation sites; the fluorescent signals were adjacent but not overlapping, confirming the notion that PTL regulates petal primordium initiation indirectly. We used the glucocorticoid induction system to enable temporal activation of PTL and determined that several genes were activated by PTL, including UNUSUAL FLORAL ORGANS (UFO). The ptl ufo double mutant resembled the ptl single mutant, suggesting that UFO acts downstream of PTL. Our findings shed light on the signal transduction of positional information between the sepal boundary and petal initiation sites, which is required for early petal development.

Previous studies showed that PTL is expressed in sepal boundary cells, whereas the ptl mutant either lacks, or has deformed, petals, suggesting that PTL regulates petal initiation in a non-cell-autonomous manner (Brewer et al., 2004; Griffith et al., 1999). To investigate whether PTL moves from intersepal cells to petal initiation sites, we generated a translational fusion of PTL and CFP expressed by the PTL own promoter and terminator (PTLg:CFP; Fig. S1). The construct restored petal development in ptl-1 and the fluorescent signal remained at intersepal cells (Fig. 1A,E,I), suggesting that the fusion protein was functional and that PTL functions in a cell-autonomous manner at the sepal boundary.

To elucidate whether the PTL-expressing region contains cells that give rise to petals, we generated GFP:RBEg PTLg:CFP plants. RBE is transcribed and translated in petal primordium cells and is essential for petal primordium initiation (Fig. 1B; Takeda et al., 2014). The expression domains of PTL and RBE did not overlap but were adjacent (Fig. 1C,D; Movies 1 and 2), indicating that the domains of the intersepal and petal initiation sites were neighboring but did not overlap.

To confirm the PTL expression at the sepal boundary, we crossed PTLg:CFP plants with a line expressing fluorescently labeled proteins in boundary domains. CUP-SHAPED COTYLEDON1 (CUC1) and CUC2 are expressed in the organ boundaries in the aerial organs of plants and regulate boundary identity (Aida et al., 1997; Ishida et al., 2000; Takada et al., 2001). The translational fusion of CUC1 and GFP in CUC1g:GFP plants was expressed in the boundary of the sepal and floral meristems (Fig. 1F), whereas the CUC2-GFP fusion protein in CUC2g:GFP plants was expressed in a broader region compared with CUC1 (Fig. 1J). Plants carrying both PTLg:CFP and CUC1g:GFP contained a few cells in which both signals were detected at the sepal boundary (Fig. 1G,H; Movie 3). By contrast, PTLg:CFP CUC2g:GFP plants contained more cells expressing both PTL and CUC2 at the sepal boundary (Fig. 1K,L; Movie 4). These data support the notion that PTL is transcribed and translated in the sepal boundary and, thus, regulates petal primordium development in a non-cell-autonomous manner.

To examine whether sepal boundary cells are required for petal initiation, we genetically ablated these cells using diphtheria toxin A chain (DT-A) (Bellen et al., 1992). DT ribosylates the EF2 translation initiation factor and inhibits protein synthesis, causing cell death. DT-A is a component of DT that kills cells in a cell-autonomous manner and has been used for tissue-specific genetic ablation in plants (Day et al., 1995; Nilsson et al., 1998; Takeda et al., 2004; Tsugeki and Fedoroff, 1999). We generated PTLg:DT-A plants in the wild-type (Colombia-0; Col-0) background. Among the 24 lines isolated by screening in the T1 generation, nine lines showed a small stature, with narrow, dark-green leaves, and subsequently developed a few flowers (Fig. 2A). Narrow leaves suggest cell death in the edge region, where PTL is expressed (Brewer et al., 2004). Eleven other lines were small, with small, green leaves and set of flowers with reduced numbers of petals and stamens (Fig. 2B; Table S2). The other two lines showed normal vegetative growth and set seeds, and their offspring generated flowers with fused sepals (Fig. 2C) and fewer petals compared with Col-0 (Fig. 2D), resembling the phenotypes of the ptl mutant. Some petals exhibited serration at their proximal regions (Fig. 2E). These results indicate that genetic ablation of PTL-expressing cells affects sepal separation, petal initiation and petal morphology, confirming the notion that the sepal boundary is required for petal initiation.

To identify genes involved in positional signal transduction, we examined genes regulated by PTL. We developed a chemical induction system of PTL using the rat glucocorticoid receptor (GR; Aoyama and Chua, 1997). A translational fusion of PTL and GR was expressed under the control of the PTL promoter and terminator (PTLg:GR) in the ptl-1 mutant background (Fig. S1). First, we applied mock or DEX solution onto each inflorescence once daily. The mock treatment did not restore petal formation in the mutant (Fig. 3A), whereas DEX-treated inflorescences developed petals after 7-10 days (Fig. 3B), indicating that the PTL-GR fusion protein was functional. Next, to determine whether the temporary action of PTL is sufficient for petal initiation, we applied mock or DEX treatment to inflorescences only once, finding that petals developed after 10 days in DEX-treated plants but not in mock-treated plants (Fig. 3C,D). According to the floral stages defined by Smyth et al. (1990), an open flower at stage 13 was at stage 5 (bud) 10 days earlier, when petal and stamen primordia arose. Therefore, when flowers at the petal initiation stage (i.e. around stage 5) were induced to express PTL by DEX treatment, they successfully initiated petal primordia. This further suggests that, once PTL is activated, subsequent stages of petal development progress to the final steps in morphogenesis. Our data highlight the importance of PTL as a master regulator of petal initiation and development.

Using this induction system combined with RNA-sequencing analysis, we identified genes that function downstream of PTL. We identified 42 upregulated and seven downregulated genes in the inflorescences 3 h after DEX treatment compared with mock control plants (Fig. 3E; Table 1). Among the upregulated genes, we focused on the F-box gene UNUSUAL FLORAL ORGANS (UFO)/At1g30950; the upregulation of this gene was confirmed by RT-qPCR (Fig. 3F). These results suggest that UFO plays a major role in petal development under PTL regulation. We also examined UFO expression in PTLg:GR plants by mRNA in situ hybridization. UFO is expressed in the center of the floral primordium at stage 2, in the cup-shaped domain during stage 3 and in four clusters at the base or abaxial side of the petal primordium at stage 4 (Durfee et al., 2003; Ingram et al., 1995; Laufs et al., 2003; Lee et al., 1997; Samach et al., 1999). In mock-treated plants, UFO was expressed in the cup-shaped domain at stage 3 (Fig. 4A) but was absent from four clusters at later stages (Fig. 4B). By contrast, in DEX-treated inflorescences, UFO was expressed in four clusters at stages 4 and 5 (Fig. 4C,D), supporting the notion that PTL regulates UFO expression specifically in these four clusters.

To confirm the genetic relationship between PTL and UFO, we generated ptl ufo double mutants. We used a weak ufo allele, ufo-6, to examine the effects of ptl on petal development (Lee et al., 1997; Levin and Meyerowitz, 1995). Flowers of the ptl-1 ufo-6 double mutant resembled those of the ptl-1 single mutant (Fig. 4E-H), which is consistent with our hypothesis that UFO acts downstream of PTL.

At1g11620, encoding an F-box protein in a family distinct from UFO (Kuroda et al., 2002), was also highly upregulated in DEX-treated PTLg:GR plants (Table 1; Fig. 3G). We examined the T-DNA insertion mutant for this gene (SALK_068307) but did not detect any differences from the wild type in whole plants, including flowers. Therefore, whether this gene is involved in petal initiation remains unknown.

Interestingly, TDF1/At3g28470 was also upregulated in DEX-treated PTLg:GR plants (Table 1; Fig. 3H). TDF1 encodes an R2R3 MYB transcription factor that affects late tapetal function and subsequent pollen development (Zhu et al., 2008). TDF1 expression is first detected in anthers at anther stage 5, corresponding to flower stage 9 (Sanders et al., 1999; Smyth et al., 1990; Zhu et al., 2008). PTL is expressed in lateral regions of the stamen primordium at floral stages 7-9 (Brewer et al., 2004). Although their expression patterns need to be examined in detail, it is possible that PTL activates TDF1 expression in the stamen primordium.

PTL is suggested to regulate petal initiation in a non-cell-autonomous manner and we confirmed this here by protein localization at the sepal boundary. Using a DEX induction system and transcriptome analysis, we identified 42 upregulated genes in DEX-treated PTLg:GR plants versus controls. Among these, we focused on the role of UFO in floral organ development. UFO was expressed in four clusters upon PTL induction (Fig. 4). The ptl-1 ufo-6 double mutant resembled the ptl-1 single mutant (Fig. 4), and ufo is epistatic to rbe-3 in the development of the second whorl organs (Krizek et al., 2006), suggesting that UFO mediates positional signaling between PTL and RBE.

UFO, an F-box protein belonging to the SCF ubiquitin complex, regulates multiple processes during floral organ development (Ingram et al., 1995; Laufs et al., 2003; Levin and Meyerowitz, 1995; Wilkinson and Haughn, 1995). One of the major roles of UFO is to activate class B homeotic genes, especially APETALA3 (AP3), with other components of the SCF complex (Lee et al., 1997; Ni et al., 2004; Samach et al., 1999; Wang et al., 2003; Wuest et al., 2012; Zhao et al., 2001). The activation of AP3 is mediated by the direct interaction of UFO with LFY, which directly binds to the AP3 promoter for transcriptional activation (Chae et al., 2008). AP3 binds to the terminator region of RBE and regulates its transcription (Wuest et al., 2012), suggesting that UFO is involved in activating RBE via AP3 activity. However, because AP3 is expressed in a whorled pattern, whereas RBE expression is restricted to the petal primordia (Takeda et al., 2004), another mechanism must exist for the positioning of gene expression to the organ primordia in the second whorl.

The other function of UFO is to regulate petal initiation: some ufo alleles result in flowers lacking petals (Durfee et al., 2003). UFO is expressed in floral meristems at stage 1, in three inner whorls at stage 2, in a cone-shaped region at stage 3, in four clusters lying on the abaxial side of the petal primordia at stage 4 and in petal primordia from stages 5 to 6 (Durfee et al., 2003; Ingram et al., 1995; Laufs et al., 2003). Therefore, UFO is likely involved in the position-dependent initiation of petal primordia at stage 4.

We propose a genetic mechanism for petal initiation in the second whorl involving two pathways: a pathway for the determination of organ identity in a whorled pattern; and a pathway for organ positioning (Fig. 5; Fig. S2; Table S3). LFY and UFO activate class B genes, which establish the petal identity together with class A genes in the second whorl. Simultaneously, PTL is activated in the sepal boundary region, likely by CUC1, given that transcriptome analysis showed that PTL is upregulated in 35S:CUC1 plants (Takeda et al., 2011). PTL expressed in the sepal boundary then activates UFO, and UFO activates RBE in the petal primordia. We examined UFO expression in PTLg:GR plants 3 h after treatment with cycloheximide and DEX treatment but did not detect its expression. Moreover, it appears, although not definitely so, that four clusters in which UFO is expressed at stage 4 include intersepal cells; thus, whether UFO is a direct transcriptional target of PTL remains unknown.

There may be crosstalk between these two pathways, because AP3 directly activates UFO and RBE (Wuest et al., 2012), and RBE suppresses class B genes (Takeda et al., 2014). The latter negative feedback regulation might enable temporal expression of RBE during petal initiation. Together, these signals might be integrated and lead to the initiation of petals at alternate positions from sepals in the second whorl at floral stage 5. Given that UFO is a component of the SCF complex, we propose that a protein degradation process occurs during petal initiation. Indeed, several F-box genes were upregulated in DEX-treated PTLg:GR plants (Table 1), suggesting that proteolysis is an important process for position-dependent petal initiation.

PTL is thought to promote auxin accumulation at petal initiation sites, and auxin-related factors, such as AUX1, PID and PIN, act downstream of PTL (Lampugnani et al., 2013). Expression of AUX1 at intersepal cells driven by the control of the PTL promoter restored petal development in the ptl mutant, suggesting that auxin biosynthesis at intersepal cells is required for petal initiation. However, in this study, we did not identify any upregulated auxin-related genes under PTL induction (Table 1). IAA29, an auxin-responsive gene, was downregulated in DEX-treated PTLg:GR plants based on transcriptome data, but RT-qPCR analysis showed that this difference was not significant (Table 1; Fig. S3). Therefore, although the PTL-dependent auxin pathway is crucial for petal initiation, we suggest that this type of regulation occurs later than the early response of transcriptional regulation by PTL.

The differentially expressed genes in PTL-induced plants included several genes encoding transcription factors, secondary metabolic enzymes and transporters (Table 1), but we did not identify genes that were reported to be involved in petal development. Given that PTL is expressed not only in the sepal boundary during early stages of flower development, but also in the lateral domains of sepals and petals (Brewer et al., 2004), the differentially expressed genes may include genes involved in the development of the lateral regions of perianth organs, such as TDF1 (involved in stamen development), as shown above.

In conclusion, we suggest that the F-box protein UFO mediates the signaling of positional information between the sepal boundary (PTL) and petal initiation sites (RBE). This idea might explain how PTL regulates petal initiation in a non-cell-autonomous manner and why petals are located at alternate positions from sepals. UFO orthologs are conserved in other flowering plants, including Antirrhinum and pea (Ingram et al., 1997; Taylor et al., 2001; Wilkinson et al., 2000). In Antirrhinum, the orthologous gene fimbriata (fim) is involved in the specification of organ positioning and the maintenance of organ boundaries (Ingram et al., 1997). This observation suggests that non-cell-autonomous regulation of petal initiation is conserved in flowering plants, a process mediated by a boundary gene, an F-box gene and a petal primordium gene.

A. thaliana plants were grown in vermiculite in small pots under long-day conditions (16 h light/8 h dark) at 23-25°C. Arabidopsis ecotypes Col-0 and Landsberg erecta (Ler) were used as the wild types. The ptl-1, rbe-1 and ufo-6 mutants were from laboratory stocks, which were originally obtained from a stock center (https://abrc.osu.edu/; ufo-6) or were a gift from Prof. David Smyth (Monash University, Melbourne, Australia; ptl-1).

The PTL constructs generated in this study are shown in Fig. S1. To generate genomic fusions of PTL with Cyan Fluorescent Protein (SECFP), a 6.3-kb genomic fragment that included the PTL promoter and the coding sequence without the stop codon was amplified from Col-0 using 5′-CGGGATCCGATATCATTACGTGTTTGTCGGCA-3′ and 5′-GGGGTACCCTGATTCTCTTCTTTACTGAGCCT-3′ primers, digested by BamHI and KpnI, and ligated to pAN19 to generate pPTLpg19. The 1.2-kb PTL terminator was amplified using 5′-GGAGTCGAGCTCGTAATTTCTCTTAATGAAGAAGAA-3′ and 5′-CGGAATTCTCTAGACCAAATCAAGATCAAACA-3′ primers, digested by SacI and EcoRI, and cloned into pSECFP19, in which SECFP had been subcloned into the pAN19 vector, to generate pPTLtSECFP19. The BamHI and KpnI fragment from pPTLpg19 was subcloned into pPTLtSECFP19 to generate PTLp:PTLg::SECFP::PTLt (designed as PTLg:CFP). The PTLg:GR plasmid was constructed in the same manner, except that GR was used, which was amplified using 5′-GGGGTACCCAGCAAGCCACTGCAGGAGTC-3′ and 5′-GGAGTCGAGCTCTCATTTTTGATGAAACAGAAG-3′ primers, digested by KpnI and SacI, and cloned into pAN19. For PTLg:DTA, the SECFP region of PTLgCFP was replaced by the gene encoding DT-A, which was amplified using 5′-GGGGTACCATGGATCCTGATGATGTTGTT-3′ and 5′-CGCGAGCTCTTAGAGCTTTAAATCTCTGTA-3′ and digested with KpI and SacI to generate PTLg:DTA. NotI fragments of PTLg:CFP, PTLg:GR, and PTLg:DTA were subcloned into the pBIN30 binary vector. For CUC2g:GFP, the CUC2 promoter and coding sequence without the stop codon were amplified using 5′-GGGGACAACTTTGTATAGAAAAGTTGACTAGAGGAAGAGTTAAGAGATG-3′ and 5′-GGGGACTGCTTTTTTGTACAAACTTGCGTAGTTCCAAATACAGTCAAG-3′ primers and cloned into pDONR P4-P1R using the Gateway system (Invitrogen). The CUC2 terminator was amplified using 5′-GGGGACAGCTTTCTTGTACAAAGTGGCATCACAAAAGAGGTGACTTATA-3′ and 5′-GGGGACAACTTTGTATAATAAAGTTGAAATCATCTAACCGAAGATTCG-3′ and cloned into pDONR P2R-P3 (Invitrogen). These two plasmids, together with pDONR207 carrying GFP, were transferred to the pGWB multisite vector to generate CUC2p:CUC2-GFP::CUC2ter (CUC2g:GFP). The plasmids were transformed into Agrobacterium tumefaciens (Rhizobium radiobacter) strain GV3101 (pMP90) and into Col-0 by the floral dip method (Clough and Bent, 1998). GFP:RBEg and CUC1g:GFP plants were generated as described previously (Gonçalves et al., 2015; Takeda et al., 2014). The plants carrying two constructs were generated by crossing, and their genotypes were checked by PCR.

For fluorescence microscopy, inflorescences were trimmed, mounted on glass slides with water and examined under an LSM780 confocal microscope (Carl Zeiss). Flower images were captured with an SPAP0 binocular microscope equipped with an EC3 digital camera system (Leica).

The PTLg:GR construct was transformed into the ptl-1 mutant. Twelve T1 lines showed resistance to 10 μg/ml glufosinate ammonium on selection medium (1% sucrose, 1% agar and 1×MS salt). Segregation and petal restoration were checked in the T2 generation. Seven lines showed restored petal development after treatment with DEX solution (10 μM dexamethasone in ethanol and 0.015% Silwet L-77). One homozygous line in the T3 generation was selected for further analysis. Mock treatment was performed with 0.1% ethanol and 0.015% Silwet L-77. After 3 h of DEX or mock treatment, total RNA was extracted from the inflorescences using an RNeasy Plant Mini Kit (QIAGEN). RNA integrity was confirmed using an Agilent RNA 6000 Nano Chip (Agilent Technologies). A library was constructed from a 0.5 µg RNA sample using an Illumina TruSeq Standard mRNA LT sample kit (Illumina). RNA sequencing was performed on the Illumina NextSeq 500 sequencing platform (Illumina), and data analysis was performed as described previously (Shimoki et al., 2021). We selected 42 and seven genes the expression of which was up- or downregulated, respectively, by summing those with a false discovery rate (FDR) <0.05, a sum (total number of mapped reads) >1, and a log₂FC >1 (upregulated) or <−1 (downregulated).

For RT-qPCR, cDNA synthesis was performed using ReverTra Ace qPCR RT Master Mix with gDNA remover (Toyobo). The reaction was performed with Thunderbird SYBR qPCR mix (Toyobo) and a Dice Real Time System Lite thermal cycler (Takara Bio). The relative expression level was calculated based on amplification of the internal control gene TUB4 (At5g44340). The primer sequences are listed in Table S1.

Inflorescences from PTLg:GR plants were treated with DEX or mock for 3 h and fixed with formalin/acetic acid/ethanol (FAA). mRNA in situ hybridization was performed as previously described (Takeda et al., 2004). The UFO antisense probe was generated from the pDW221.1 template (Lee et al., 1997) and labeled using a DIG RNA labeling kit (Merck).

We thank Drs Michitaka Notaguchi, Takamasa Suzuki, Noriko Inada and Yoichiro Fukao, Ms Kaori Kaminoyama, Prof. Kiyotaka Okada and Prof. Detlef Weigel for providing materials, technical support and data analyses.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.200684.reviewer-comments.pdf.