Physical interaction of floral organs controls petal morphogenesis in Arabidopsis thaliana

Flowering plants bear beautiful flowers to attract pollinators. Petals are the most variable organs in flowering plants with their color, fragrance, and shape. In Arabidopsis thaliana , petal primordia arise at a similar time to stamen primordia, and elongate at later stages through the narrow space between anthers and sepals. While many of the genes involved in regulating petal identity and primordia growth are known, the molecular mechanism for the later elongation process remains unknown. We found a mutant, folded petals 1 ( fop1 ), in which normal petal development is inhibited during their growth through the narrow space between sepals and anthers, resulting in formation of folded petals at maturation. During elongation, the fop1 petals fuse to the sepal surface at several sites. The conical-shaped petal epidermal cells are flattened in the fop1 mutant, as if they had been pressed from the top. Surgical or genetic removal of sepals in young buds restores the regular growth of petals, suggesting that narrow space within a bud is the cause of petal folding in the fop1 mutant. acyltransferase gene family, WSD11, which is expressed in elongating petals and localized to the plasma membrane. These results suggest that the FOP1/WSD11 products synthesized in the petal epidermis may act as a lubricant, enabling uninhibited growth of the petals as they extend between the sepals and the anthers. that is in the elongation of as though in


Introduction
Floral organs usually develop sequentially from the outermost whorl towards the inner in the order of sepals, petals, stamens and carpels. Sepals arise first and form a tight external covering, which functions as a barrier to protect the developing internal floral organs from physical or biological attacks from the outside. Petals initiate when sepals cover the flower bud and grow rapidly at later stage. Petal development can be divided into several stages, each of which has been well described at a molecular level in Arabidopsis thaliana (Smyth et al., 1990;Irish, 2008). Petal and stamen primordia arise simultaneously at developmental stage 5. Stamens grow faster than petals until stage 8, and the anthers fill the upper internal space created by the protective dome-like closed sepals. After stage 9, petal growth is accelerated, and petals elongate through a narrow space generated by the sepals and anthers.
Floral organ identity is established in a concentric pattern by MADS and AP2/ERF transcription factors, which is described by the floral ABCE or quartet model (Theissen and Saedler, 2001). Petal identity is fixed in the second whorl by the combined function of class A, B and E genes (Bowman et al., 1989;Weigel and Meyerowitz, 1994;Krizek and Fletcher, 2005). Suppression of AGAMOUS activity in the perianth whorls is important for petal growth, and this process is controlled by APETALA2 (AP2), AINTEGUMENTA (ANT), LEUNIG (LUG), SEUSS (SEU), RABBIT EARS (RBE), ROXY1, and STERILE APETALA (SAP) (Liu and Meyerowitz, 1995;Byzova et al., 1999;Conner and Liu, 2000;Krizek et al., 2000;Franks et al., 2002;Sridhar et al., 2004;Xing et al., 2005;Krizek et al., 2006;Grigorova et al., 2011). Petal primordia arise at four loci in the second whorl, and this positioning is established independently of the process that determines organ identity (Griffith et al., 1999;Brewer et al., 2004;Takeda et al., 2004;Xing et al., 2005;Lampugnani et al., 2013). After initiation, the growth of petals depends on the activity of cell division and 7 ( Figure 1B), with an outward fold in the medial region and an inward fold in the distal portion.
When fully expanded, the size, shape, color, and vascular pattern of a mature petal of the fop1-1 mutant were almost the same as those of the wild type, except for its folded pattern (Figures S1A,B,O and P), suggesting that the fop1 mutation is responsible for the physical process of growth, rather than cell proliferation or expansion. This is supported by the expression pattern of a cell-proliferation marker gene, HISTONE 4, in petals, which was not significantly different from wild type (Figures S1C-J;Krizek, 1999;Gaudin et al., 2000;Dinneny et al., 2004). These data suggest that the overall petal shape and growth are not altered in fop1-1. We examined the phenotype of two T-DNA insertion mutants for the FOP1 gene after gene identification ( Figure S3, see below), and found that they showed a similar phenotype to fop1-1 ( Figures 1C and D). On the basis of this phenotypic and genetic similarity, we used fop1-1 for further analysis.
Morphological changes in developing petals were examined by sectioning of the bud at several developmental stages. Petal primordia of the fop1-1 mutant formed at stage 5 grew normally up to stage 9, when the petal tip reached the base of an anther (Figures 1E,F,K and L). At stage 10, when wild-type petals elongate through the space between a sepal and an anther ( Figure 1G), the mutant petals did not show smooth elongation, becoming stuck between an anther and a sepal ( Figure 1M). Dissection of unopened buds revealed that when the tip of wild-type petals reached the top of the long stamens ( Figure 1H), the mutant petals failed to undergo straight elongation ( Figure 1N). The degree of folding increased at points where the mutant petal was stuck between the anther and sepal ( Figures 1O and P), whereas wild-type petals remained straight ( Figure 1I), even though their middle part could contact the anther and sepal ( Figure 1J). These observations suggest that the FOP1 is involved in the smooth elongation of petals as they grow though the narrow space in a floral bud. fop1 petals make contact with, and fuse to, the sepal surface Next we examined the contact region between petals and sepals using scanning electron microscopy to identify where the folding starts in the mutant. In the wild type, at the stage where the petals are equal in height to the anthers, no fusion to the sepals was observed were flattened as if they had been pressed and rubbed from above ( Figure 2R). The flattened cells were frequent on the abaxial surface in the distal part of a folded petal facing a sepal ( Figure S1L). Similar flat-tip cells were also found in wild-type petals, but the number of such cells was comparatively few ( Figure S1K). Cuticle formation per se is not altered in fop1 because petal cells had epicuticular nanoridges, although they were affected after its formation. The surface of epidermal cells of anthers or sepals were not altered in the mutant (data not shown). Taken together, the flattened surface of the epidermal cells in fop1-1 mutant petals could be the result of strong pressure and friction between the petals and sepals.
Sepal removal restores the straight growth of fop1 petals 9 These data indicate that the petals on the fop1 mutant do not easily extend through the space between the sepals and the anthers. To confirm this further, we examined whether regular petal elongation is restored in 'open buds' where sepals do not form a tight covering, so that the physical contact between petals and sepals or anthers would not be strong. First, we removed one or two sepals from a fop1-1 mutant bud before petal folding started, and let the buds grow. After three days, the petals grew flat and did not show the folding phenotype

ACYLTRANSFERASE family
We mapped the FOP1 gene on chromosome 5 and identified a mutation in At5g53390 ( Figure   S3A). We found that the guanine at nucleotide 820 of the coding sequence was substituted with adenine in fop1-1, replacing glycine with arginine in the mutant protein ( Figure S3B).
We examined the border sequences of three T-DNA insertion lines found in the SIGnAL database (http://signal.salk.edu/cgi-bin/tdnaexpress) within or near the gene ( Figure S3B; Alonso et al., 2003), and confirmed that the three lines had a small deletion at the T-DNA insertion sites ( Figure S3B). Two of them, SALK_093133 (named fop1-2) and SALK_137481 (fop1-3), showed a similar phenotype to fop1-1 ( Figures 1B-D). The SALK_149804 line, in which T-DNA was inserted at 31 bp upstream from the ATG initiation codon, was indistinguishable from wild type and did not show the petal folding phenotype.
To confirm whether At5g53390 is FOP1, a genomic fragment comprising the 2.2 kb promoter, 2.2 kb open reading frame (ORF) and 1.6 kb 3′ region, was transformed into the fop1-1 mutant. We obtained 80 independent T 1 plants carrying the FOP1 genomic fragment and all lines complemented the petal defects, indicating that At5g53390 is FOP1. We cloned a full-length FOP1 cDNA using a rapid amplification of cDNA ends (RACE) strategy and identified the region covering the 48 bp 5′-UTR, 217 bp 3′ UTR, and seven exons ( Figure   S3B). grew through the space between anthers and sepals ( Figure 4E), and the expression expanded widely in the distal part of petals ( Figure 4F). FOP1 had lower expression in sepals and anthers than in petals, indicating that function of FOP1 in petals is required for straight elongation. The FOP1-GFP fusion protein was localized at the periphery of petal epidermal cells ( Figure 4G). We further examined the cellular localization of FOP1 protein by expressing the GFP:FOP1 fusion gene transiently in suspension cultured cells of Arabidopsis thaliana ( Figure 4H). The GFP signal was detected in the periphery of the cell, which overlapped with the FM4-64 signal ( Figure 4I and J), indicating that FOP1 is localized to the plasma membrane. Together, these data suggest that FOP1 is involved in wax ester synthesis at the plasma membrane in the petal epidermis.

Discussion
Study of the fop1 mutant directed attention to an aspect of petal growth in a floral bud, namely to how petals elongate through the narrow space between sepals and anthers. Figure   5A describes the schematic model highlighting the differences in the petal elongation process between wild type and fop1. Growth of petal primordia is normal up to stage 9, but after stage 9 the petal tip becomes stuck between sepals and anthers in the mutant, whilst in the wild type, petals elongate straight. Our data suggest that the products of FOP1 act as a lubricant on the petal surface, enabling petals to elongate smoothly through the narrow space in a floral bud. It is also possible that the FOP1 products are involved in making cuticle rigid, so that in the fop1 mutant the petal surface intensity is lower and easy to get rubbed off, enhancing the petal folding.  Figures S1M and N). We also examined the difference of the hexane-soluble surface contents of floral buds between wild type and fop1-1, but found no significant change ( Figure   S4). Arabidopsis WSD1, a homolog of FOP1, has high wax synthase activity and lower but cause organ fusion (Li-Beisson et al., 2009;Panikashvili et al., 2009;Bessire et al., 2011).
Compared to these mutants, fop1 petals form nanoridges on the petal epidermis, and the degree of organ fusion is only subtle (Figure 2). Because the petal folding starts before the nanoridges deposit on the surface of petal epidermis, we propose that FOP1 is involved in the synthesis of wax-related products in the petal epidermis before nanoridge deposition ( Figure   5B). Similar petal-folding defects are found in mutants of the DESPERADO (DSO)/AtWBC11/ABCG11 and ABCG13 genes, both of which encode members of ATP-binding cassette transporters (Panikashvili et al., 2007;Panikashvili et al., 2011), suggesting that they are involved in the similar process to FOP1 ( Figure 5B).
Is the petal elongation mechanism conserved among flowering plants? Similar petal-folding phenotype is also known in a breed of Japanese morning glory (Ipomoea nil).
Corollas of the crepe (cp) mutant fold twice like those of fop1, forming an additional tube-like structure surrounding the stamens and carpels ( Figure S5) (Miyake and Imai, 1927). Similar to our results, the removal of calyx restored the formation of funnel-shaped corollas (Nishino and Goto, 2002). These characters suggest that the cp phenotype might be caused by a similar mechanism to that of fop1 petals. Molecular identification of the CP gene will provide an answer.

Conclusion
Here we propose a mechanism for the smooth elongation of petals. After primordia initiation, petals elongate through the narrow space generated by anthers and sepals. FOP1/WSD11 is expressed in elongating petals, and may catalyze the wax ester biosynthesis at the epidermal plasma membrane. The FOP1 products may act as a lubricant, which enables petals to grow straight in a narrow space in a bud. Identifying the substrates and products of FOP1/WSD11 is the next challenge in the effort to understanding the relationship between the cell surface components and petal morphogenesis.

Plant Growth Conditions
The Wassilewskija (WS) and Columbia (Col)

Histology and Microscopes
For scanning electron microscopy, samples were prepared as previously described  Table S1.
cDNA cloning was performed by both 5′-RACE and 3′-RACE using the SMART RACE cDNA Amplification Kit (Clontech, USA). Sequencing was performed using the ABI BigDye Terminator Cycle Sequencing Ready Reaction Kit and an ABI Prism 3100 genetic analyzer (Applied Biosystems, USA).

Complementation Test
The genomic fragment including the 2.2 kb promoter, 2.2 kb ORF and 1.6 kb 3' regions of pFgf211NP, respectively. These constructs were introduced into the fop1-1 mutant by a vacuum infiltration procedure with the Agrobacterium strain C58C1. Transgenic plants were screened on an agar medium containing 30 μ g/ml kanamycin and 100 μ g/ml carbenicillin.
Sequences of oligonucleotide primers used for cloning are listed in Table S1.

RNA Isolation and RT-PCR
Total RNA was isolated with the Isogen reagent (Nippon Gene) or RNeasy Plant mini kit (QIAGEN). One microgram of total RNA was reverse-transcribed with the SUPERSCRIPTII reverse transcription kit (Invitrogen). The oligonucleotide primers used for RT-PCR are as follows: FOP1, 53390F1a and 53390R1; ACT8; ACT8F and ACT8R. The sequences of the oligonucleotide primers are listed in Table S1.
mRNA in situ hybridization mRNA in situ hybridization was performed as previously described (Matsumoto and Okada, 2001). The HISTONE 4 probe was prepared as previously described (Dinneny et al., 2004).

Histochemical Analysis of FOP1
The HindIII fragment from pFpSK was subcloned into pBI101 (Clontech, USA) to generate pFpGUSBI. pFpGUSBI was transformed into Col plants and screened as described above.

Generation of FOP1p::FOP1-GFP
The G3GFP and FOP1 cDNAs were cloned into pPZP211 and pPZP211NP to generate pFcG21135 and pFcG211NP, respectively. The HindIII fragment from pFpSK was subcloned into pFcG21135 and pFcG211NP to generate pFpFcG21135 and pFpFcG211NP, respectively.
Transformation and screening were performed as described above.

Gas Chromatography
Surface fatty acids were extracted from flowers of 6-week old plants with hexane. C19COOH