BIGPETALp, a bHLH transcription factor is involved in the control of Arabidopsis petal size

Judit Szécsi, Caroline Joly, Karim Bordji, Emilie Varaud, J Mark Cock, Christian Dumas, Mohammed Bendahmane

Author Affiliations

  1. Judit Szécsi,
  2. Caroline Joly,
  3. Karim Bordji,
  4. Emilie Varaud,
  5. J Mark Cock§,
  6. Christian Dumas and
  7. Mohammed Bendahmane*,1
  1. 1 IFR128 BioSciences Lyon‐Gerland, Reproduction et Développement des Plantes UMR INRA‐CNRS‐UCBL‐ENSL, Ecole Normale Supérieure, Lyon cedex, France
  1. *Corresponding author. RDP, UMR5667, Ecole Normale Supérieure, 46 allée d'Italie, 69364 Lyon cedex 07, France. Tel.: +33 4 7272 8611; Fax: +33 4 7272 8600; E-mail: mbendahm{at}
  • Present address: EVIR, INSERM 758, ENS, 46 allée d'Italie 69364 Lyon, France

  • Present address: GIP Cyceron, UMR CNRS 6185, BP 5229F, 14074 Caen cedex, France

  • § Present address: UMR 7139 CNRS ‐UPMC, Végétaux Marins et Biomolécules, Station Biologique, BP74, 29682 Roscoff, France


In Arabidopsis, APETALA1, PISTILLATA, APETALA3 and SEPALLATA interact to form multimeric protein complexes required to specify petal identity. However, the downstream events that lead to petal specific shape and size remain largely unknown. Organ final size can be influenced by cell number or cell expansion or both. To date, no gene that specifically limits petal size by controlling postmitotic cell expansion has been identified. Here we have identified a novel petal‐expressed, basic helix‐loop‐helix encoding gene (BIGPETAL, BPE) that is involved in the control of petal size. BPE is expressed via two mRNAs derived from an alternative splicing event. The BPEub transcript is expressed ubiquitously, whereas the BPEp transcript is preferentially expressed in petals. We demonstrate that BPEp is positively regulated downstream of APETALA3, PISTILLATA, APETALA1 and PISTILLATA3 and is negatively regulated downstream of AGAMOUS. Plants that lack the petal‐expressed variant BPEp have larger petals as a result of increased cell size, showing that BPEp interferes with postmitotic cell expansion. BPEp is therefore a part of the network that links the patterning genes to final morphogenesis.


During flower development, many proteins interact as components of the molecular networks that control biological processes such as the floral meristem emergence and the acquisition of flower organ identity (Krizek and Fletcher, 2005). Once a plant has been induced to flower, the floral organs (sepals, petals, stamens and carpels) develop sequentially on the flanks of the floral meristem under the combinatory action of four classes of organ identity genes (A, B, C and E) following the ABCE model (Krizek and Fletcher, 2005). In this model, A+E specify sepals, A+B+E petals, B+C+E stamens and C+E carpels. Over the past few years, these homeotic genes have been shown to be master regulatory genes that trigger the developmental programs required for flower organogenesis (Jack, 2004; Krizek and Fletcher, 2005). In Arabidopsis thaliana, these genes are APETALA1 (AP1) and APETALA2 (AP2) for the A class, PISTILLATA (PI) and APETALA3 (AP3) for the B class; AGAMOUS (AG), the only known class C gene and SEPALLATA 1, 2, 3, 4 (SEP1, SEP2, SEP3, SEP4) for the E class (Krizek and Fletcher, 2005). All of these genes (except for AP2) encode MADS‐box transcription factors (TFs) that have been proposed to interact and form four different tetrameric complexes according to the ‘quartet model’ (Honma and Goto, 2001; Theissen and Saedler, 2001). This directly links the action of these four different tetrameric complexes to floral organ identity. AP3, PI, SEP3 and either AP1 or AG have been shown to interact to form multimeric protein complexes required to specify petal (in whorl 2) or stamen (in whorl 3) identity, respectively (Honma and Goto, 2001; Theissen and Saedler, 2001). In whorl 2, PI, AP3, AP1 and SEP presumably regulate a set of downstream structural genes (so called ‘realizators’) that encode proteins required for the cell division and differentiation events that lead to petal organogenesis. These downstream proteins must function in a timely, coordinated way, leading to a constant final organ shape and size. Thus, understanding petal development and morphogenesis requires the identification of those genes whose expression is regulated downstream organ identity genes. Efforts have been made to perform reverse genetic screens to identify petal‐expressed genes using microarray approaches (Zik and Irish, 2003; Wellmer et al, 2004). These studies identified only a very small number of genes that were petal specific. To date, only a few genes have been clearly shown to be involved in petal development. NAP, whose expression is controlled by AP3/PI, has been suggested to function in the transition between growth by cell division and cell expansion in petals and stamens (Sablowski and Meyerowitz, 1998). RABBIT EARS, whose expression is under the control of AP1, is involved in second whorl organ development (Takeda et al, 2004). PETAL LOSS has been shown to be involved in petal organ initiation and orientation (Griffith et al, 1999). Very recently, an E3 ubiquitin ligase‐encoding gene, BIG BROTHER (BB), has been shown to limit plant organ size by controlling cell proliferation (Disch et al, 2006).

In this study, we used cDNA‐AFLP‐differential display (DD) to identify genes that act downstream of petal organ identity genes in A. thaliana. We identified a basic helix‐loop‐helix TF encoding gene (BIGPETAL or BPE) that is expressed via two mRNA transcripts derived from an alternative splicing event (BPEp and BPEub). We demonstrate that (i) BPEp is preferentially expressed in petal and is derived from a ubiquitously expressed gene via an alternative splicing event, (ii) BPEp acts downstream of petal organ identity genes and (iii) that BPEp regulates the size of A. thaliana petals by restricting cell expansion.


Identification and expression analysis of petal‐expressed genes

We carried out a cDNA‐AFLP‐DD to search for genes that are expressed specifically in petals. This screen was performed by comparing gene expression profiles in inflorescences of two A. thaliana floral homeotic mutants, pistillata (flowers composed of sepals and carpels) and agamous (flowers composed of sepals and petals) (Bowman et al, 1989). Genes upregulated in agamous flowers, in comparison to pistillata flowers, are therefore expected to be petal‐expressed. The nucleotide sequences of the cDNA‐AFLP bands that were differentially amplified in agamous corresponded to 13 A. thaliana genes.

RT–PCR analysis of the expression pattern of each of these 13 genes showed that eight were expressed specifically or preferentially in flowers compared to leaves (Figure 1). Based on their putative protein sequences, these genes are predicted to encode a basic helix‐loop‐helix TF (that we named BPE for BIGPETAL, see below), an F‐box protein (F‐box), a Histone2A (His2A), a Phosphatase 2C (Phos 2C), a Chalcone Synthase (ChS), a Receptor‐Like‐protein Kinase (RLK), a endo‐β‐1, 4‐glucanase (β‐gluc) and a PolyA‐Binding protein (PABP), respectively. Moreover, BPE, His2A and F‐box‐encoding genes were expressed in agamous and wt flowers, but showed no or background level expression in pistillata flowers (lack petals), indicating that they are likely to be petal‐expressed. The remaining five genes, which encode a putative translationally controlled tumor protein (TCTP), an early‐responsive to dehydration stress protein‐3 (ERD‐3), a photosystem I subunit XI precursor, a vacuolar ATPase or α tonoplast intrinsic protein, showed ubiquitous expression (Figure 1 and data not shown).

Figure 1.

RT–PCR expression analyses of selected identified genes in flowers of wt, pistillata (pi) and agamous (ag) and in leaves (L) of wt A. thaliana. F‐box, putative F‐box protein; His2A, putative Histone2A; Phos 2C, putative phosphatase 2C protein; ChS, Chalcone Synthase; RPK, putative receptor‐like protein kinase; β‐gluc, similar to endo‐β‐1,4‐glucanase; PABP, putative polyA‐binding protein; TCTP, translationally controlled tumor protein; ERD‐3, early‐responsive to dehydration stress protein‐3. Gene identifiers are listed.

BIGPETAL encodes two basic helix‐loop‐helix TFs

Sequence similarity searches showed that the BPE corresponds to AtbHLH031 (At1g59640) which belongs to subgroup XII of the large family of basic helix‐loop‐helix TFs in Arabidopsis (Heim et al, 2003). Cloning of BPE cDNAs revealed that this gene encodes two transcripts (Figure 2A and B): a short transcript of 1215 nucleotides that we named BPEub (ub for ubiquitous, see below) and a longer, transcript of 1623 nucleotides, BPEp (p for petal‐expressed). Both transcripts have identical 5′ nontranslated regions of 136 nucleotides. BPEp and BPEub differ in their 3′ coding and nontranslated regions because of an alternative splicing event that leads to the retention of the 408 nucleotide fifth intron in the BPEp transcript and a change in the reading frame. Thus, BPEub and BPEp are predicted to encode two different proteins of 29 and 38 kDa, respectively. The first 221 amino acids of the two proteins are identical and contain the highly conserved bHLH domain, known to be involved in DNA binding and in protein–protein interactions. The 43 amino acids corresponding to the C‐terminal domain of BPEub are replaced by a stretch of 122 amino acids in BPEp that is translated from intron 5. These two C‐terminal regions do not show any similarity to each other, nor to other known protein domains. BPEp and BPEub proteins fused to the GAL4‐binding domain were able to activate transcription of a reporter gene in yeast (EV & MB, unpublished data) consistent with their predicted functions as transcription factors.

Figure 2.

(A) Genome organization of BPE transcripts. Exons are presented as boxes numbered in roman characters. Introns are indicated in arabic numbers. Identical amino‐acid sequences in both proteins are marked as white boxes. Amino‐acid sequence present in BPEp only is shown as a black box and that present in BPEub by a gray box. Dashed lines show the position of the two regions used to generate RNAi constructs. Positions of primers used in RT–PCR experiments are indicated by arrows. (B) Nucleotide sequence of the pre‐messenger RNA of BPE. The 5′ and 3′ nontranslated regions as well as introns are presented in lower case characters, exons in capital characters. The differentially spliced region that is expressed in BPEp and not in BPEub is marked in italicized capital characters. Stop codons are marked by stars. BPEp and BPEub proteins are shown in red and blue, respectively. The basic region is underlined and the helices of the helix‐loop‐helix region are marked in bold characters. The site of the T‐DNA insertion in the BPE knockout line is indicated by ‘†’. BPEub and BPEp sequences have been submitted to the DDBJ/EMBL/GeneBank databases under accession numbers AM269753 and AM269754, respectively. (C) RT–PCR analysis of BPEp (top) and BPEub (middle) mRNA accumulation in wt A. thaliana inflorescence (i), flower buds (b), open flowers at the onset of petal senescence (o), rosette leaves (Lr), cauline leaves (Lc) and inflorescence stem (s). The constitutively expressed gene TCTP (bottom) was used as a control. (D) Analysis of BPEub mRNA accumulation using RT–PCR in flowers of mutant pistillata (pi) and agamous (ag) and of wt A. thaliana. (E) Expression analysis of the BPEp transcript in flowers of wt and mutant A. thaliana (lines pi, ag, apetala1 [ ap1], ap2, and ap3) using RT‐QPCR. A.U., arbitrary units. (F) BPE promoter activity as monitored in the PromoterBPE:GUS‐expressing line. GUS staining in (a) the inflorescence, (b) the rosette leaves (right: promoterBPE:GUS plant, left: wild‐type control plant) and (c) in a flower at development stage 13. (d) GUS staining in wild‐type flower. (G) Analysis of BPEp expression during early (1–7), mid (8–12) and at open flower development stages (13 and 14).

The two BPE transcripts have different expression profiles

We designed primers to discriminate between the BPEp and BPEub transcripts, in RT–PCR analyses. The primers 12–5 and 12–3 (Figure 2A) were used to specifically detect BPEp transcript and allow amplification of DNA fragment of 905 bp (note that these are the primers we used to detect BPE in Figure 1). As there is no nucleotide sequence that discriminates the shorter BPEub transcript from the BPEp transcript, the assay for the former was based on the size difference of the PCR products (by amplifying a region spanning intron 5). The primers 12‐5 and 3C (Figure 2A) allow the amplification of DNA fragments of 804 or 1212 bp from cDNA template corresponding to BPEub or BPEp, respectively. As shown in Figure 2C, the BPEub transcript was expressed in flowers, stems and in cauline and rosette leaves of wt A. thaliana and this transcript was expressed at similar levels in flowers of both pistillata and agamous (Figure 2D). Thus BPEub is ubiquitously expressed. In contrast, the BPEp transcript accumulated in wt A. thaliana inflorescences, and flower buds, but no or very weak expression was observed in inflorescence stems and in rosette and cauline leaves. Figure 1 shows that the BPEp transcript accumulated in agamous flowers but not in pistillata flowers. These results indicate that BPEp has a petal‐ or petal and stamen‐specific expression profile. To quantify the abundance of the BPEp transcript, we performed real time‐quantitative PCR (RT‐QPCR; Figure 2E). This sensitive assay detected low levels of BPEp transcripts in pistillata and apetala3 flowers and a very high level of BPEp transcript in agamous flowers compared to wt flowers, thus agreeing with the RT–PCR results (Figure 1). Similar experiments showed that BPEp was very weakly expressed in apetala1 or apetala2 flowers (which possess stamens but lack petals; Figure 2E). Taken together, these results indicate that BPEp preferentially accumulates in petals.

Global gene expression during development in diverse organs of A. thaliana has been investigated using the Affymetrix ATH1 arrays (Zimmermann et al, 2004; Schmid et al, 2005). These microarray experiments indicate a ubiquitous pattern of expression for BPE, they do not agree with our data in as far as they indicate that BPE expression in petals and stamens is stronger than in other plant organs, with particularly high levels in petals (Zimmermann et al, 2004). These microarray data do not allow discrimination between the two splice variants BPEp and BPEub because the probe sets corresponding to BPE target both BPEub and BPEp, but it is very unlikely that the high level of expression detected in petals is owing to BPEp accumulation because this transcript is significantly less abundant than BPEub, even in petals (Figure 2C). Therefore, in contrast to our observations, the Affymetrix data suggest that BPEub is expressed to a higher level in petals and stamens compared to other organs.

To investigate these conflicting results further, we fused 1317 bp of the BPE promoter sequence to the GUS reporter gene and examined X‐gluc staining in transgenic plants carrying this fusion. Nine independent PromoterBPE:GUS lines were generated and showed similar GUS staining patterns. BPE expression was detected ubiquitously in vegetative organs and in flowers at different developmental stages, with no evidence of a higher level of expression of BPEub in petals compared to other floral organs or leaves (Figure 2F). Hence, the GUS‐fusion data corroborated the RT–PCR experiments (Figure 2C and D) and we therefore propose that the BPEub transcript is expressed ubiquitously.

Attempts to investigate the BPEp expression pattern by in situ hybridization failed, most likely owing to the low level of expression of this transcript. Instead, we analyzed its expression at different flower development stages (Smyth et al, 1990); Figure 2G). A very low level of BPEp expression was observed at early development stages (1–7, corresponding to floral meristem determination, organ identity and initial development). The BPEp transcript accumulated from stage 8 and a very high relative abundance was observed at stage 10 and above (at which stage cell differentiation occurs in petals). BPEp expression decreased and was at a background level in flowers at the onset of petal senescence (Figure 2C).

In summary, BPE encodes two transcripts whose relative abundance is regulated at the post‐transcriptional level, with BPEub showing ubiquitous expression and BPEp exhibiting an expression in petals mainly at flower development stages above 10.

BPEp expression is upregulated downstream of PI/AP3, SEP2, SEP3 and AP1

As BPEp was negatively regulated in pistillata and in apetala3 flowers, we investigated its expression in A. thaliana lines that overexpressed PI (35S:PI) or AP3 (35S:AP3). In flowers of both lines, we observed a five‐fold increase in the abundance of BPEp compared to wt flowers (Figure 3A). These results indicate that BPEp expression is positively regulated downstream of the B class genes PI and AP3.

Figure 3.

RT‐QPCR analysis of BPEp mRNA expression in flowers and leaves of A. thaliana lines overexpressing (A) the B class proteins PI (line 35S:PI) or AP3 (line 35S:AP3); (B) the E class proteins SEP2 (line 35S:SEP2) or SEP3 (line 35S:SEP3). (C) BPEp expression in flowers of line 35S:AP1‐GR that overexpress an inducible activity of the A class protein AP1. Flower samples were collected at 6 h post‐treatment with DEX or with CHX or simultaneously with DEX and CHX. A.U., arbitrary units.

It has been shown that, to mediate their function as organ identity proteins, PI, AP3 and SEP3 interact with AP1 in whorl 2 to specify petal identity and with AG in whorl 3 to specify stamen identity (Honma and Goto, 2001). We therefore investigated the expression of BPEp in lines that overexpress AP1, SEP3 or SEP2. The latter two genes are known to function redundantly (Pelaz et al, 2000, 2001b). Interestingly, BPEp abundance increased by about 10‐fold or 6‐fold in flowers of the 35S:SEP3 or the 35S:SEP2 lines, respectively (Figure 3B). To investigate the role of AP1 in BPEp regulation, we followed the latter's expression in an apetala1 line that carried an inducible AP1 sequence fused to the steroid‐binding domain of the rat glucocorticoid receptor (GR; line 35S:AP1‐GR; Yu et al, 2004). Following treatment with the GR substrate dexamethasone (DEX) to activate AP1, an approximately six‐fold increase in BPEp expression was observed in flowers within 6 h after treatment (Figure 3C). These data indicate that BPEp is positively regulated by AP1. To investigate whether this activation of BPEp by AP1 is direct or indirect, inflorescences of the 35S:AP1‐GR line were simultaneously treated with DEX and with cycloheximide (CHX, a protein synthesis inhibitor). At 6 h post‐treatment, no accumulation of the BPEp transcript was observed, indicating a requirement for intermediate protein synthesis (Figure 3C). Hence, AP1 is a positive regulator of BPEp accumulation and this regulation is indirect, requiring de novo protein synthesis. A small increase in BPEp transcript accumulation was also observed in leaves of lines ectopically expressing PI, AP3, SEP2 or SEP3 (Figure 3A and B). Taken together, these results indicate that BPEp is activated downstream of the petal organ identity PI, AP3, AP1 and SEP MADS‐box TFs.

BPEp is negatively regulated downstream of AGAMOUS

BPEp accumulation was 8–10‐fold higher in agamous flowers compared to wt flowers (Figure 2E), suggesting that BPEp is negatively regulated by AG. To further investigate this regulation process, we followed the expression of BPEp in an agamous line that overexpresses an inducible AG fused to GR (35S:AG‐GR) (Ito et al, 2004). As expected, in the absence of DEX treatment, high levels of BPEp were detected, similar to those found in the agamous mutant (Figure 4A). When inflorescences of line 35S:AG‐GR were treated with DEX, a significant decrease in BPEp mRNA expression was observed between 6 and 24 h post‐treatment. BPEp expression continued to decrease at 36 h post‐treatment. As expected, DEX treatments did not modify BPEp accumulation in leaves (Figure 4A). These findings supported our previous observation that BPEp mRNA accumulation is negatively regulated downstream of AG. Interestingly, DEX treatment did not modify BPEub expression in leaves indicating that BPEp regulation of expression is likely to be post‐transcriptional (Figure 4B). One way to investigate whether repression of BPEp by AG is direct or indirect, would be to simultaneously treat the inflorescences of the 35S:AG‐GR line with DEX and CHX. However, 24 h of CHX treatment (corresponding to the time at which we saw a significant repression of BPEp by AG) induces severe physiological phenotypes (toxic effects) and, therefore, this test was not possible (data not shown).

Figure 4.

Expression analysis of BPEp (A) and BPEub (B) transcripts in flowers and leaves of A. thaliana line 35S:AG‐GR without or with DEX treatments using RT‐QPCR. Flower samples were collected at 0, 6, 24 or at 36 h post‐treatment and leaves at 24 h post‐treatment. ‘c’, nontreated control flower samples. ‘−’, control samples treated with the buffer alone and harvested at 24 h post‐treatment. A.U., arbitrary units.

BPEp limits petal organ size

We identified a line (bigpetal‐1 or bpe‐1 allele) carrying a T‐DNA insertion in the fourth intron of the BIGPETAL gene. No BPEp nor BPEub transcripts were detected, using RT–PCR, in plants homozygous for the bpe‐1 allele (data not shown). In parallel, we generated two RNA interference (RNAi) lines by expressing inverted repeats of nucleotide sequences corresponding to two different regions in the BPE transcripts under the control of the CaMV 35S promoter (Figure 2A). The target nucleotide sequence for the first RNAi (RNAi‐bpe) spans the first and second exons that are present in both BPE transcripts, whereas the second RNAi (RNAi‐bpep) is within intron 5. As intron 5 is spliced out of BPEub but retained in BPEp, RNAi‐bpep should specifically target the latter. No or very little, BPEub and BPEp transcripts were detected in RNAi‐bpe‐expressing lines (data not shown). However, in the RNAi‐bpep‐expressing lines, some plants exhibited repression of both transcripts (e.g. plant 2 in Figure 5A) whereas other plants showed no accumulation of BPEp transcript while the BPEub transcript was not affected (e.g. plant 1 in Figure 5A). The latter plants are referred to as RNAi‐bpep/BPEub.

Figure 5.

(A) Expression of BPEp and BPEub in two individual plants of an RNAi‐bpep‐expressing line and of wt control. TCTP was used as control. (B) Flower phenotype in the KO line bpe‐1 and in lines expressing RNA interference constructs directed against BPE transcript. (a) Flowers of wt, bpe‐1 and RNAi‐expressing lines. (b) Flower organs of the bpe‐1 line compared to those of wt plants. Note the larger size of petals in the bpe‐1 line. (c) Petals of the RNAi lines compared to those of wt plants. (C) Petal length and width in RNAi‐BPEp plants was measured and compared to wt petals. The data are means±s.d. A total of 60 flowers from 12 plants were examined for each genotype. *, Significantly different from wt at P<0.05. (D) SEM views of the petal adaxial surface cells from wt (a and d), RNAi‐bpe (b and e) and RNAi‐bpep/BPEub (c and f) plants at flower developmental stage 15. (E) (a) Means±s.d. of cell size in the distal adaxial surface of petals in the RNAi‐bpep and wt plants. (b) Cell number per 500 μm2 surface area in the distal adaxial surface of petals in the RNAi‐bpep and wt plants.

Phenotypic analyses of RNAi and bpe‐1 loss‐of‐function mutants showed that petals in these plants were significantly increased in size compared to wild‐type petals (Figure 5B, panels a–c). The other flower organs did not exhibit any phenotypic modifications (Figure 5B, panel b). Petal length and width (distal region of petal blade) measurements showed that in the mutant lines, petals were about 24% larger in size compared to the wt (Figure 5C). A Student's t‐statistical test confirmed that the increase in petal size was statistically significant (P<0.05; Figure 5C). The RNAi‐bpep/BPEub plants also showed an increased petal size similar to that observed in the bpe‐1 line, demonstrating the role of the petal‐expressed splice variant BPEp (and not BPEub) in influencing petal size.

BPEp limits final cell size

To check whether the increase in petal size associated with reduced BPEp expression is a result of failure to maintain correct cell proliferation or cell expansion or both, petals epidermis cell size and number per surface area were analyzed and compared to the wild type. Scanning electron microscopy (SEM) analyses showed that, in the bpe‐1 line and in RNAi‐bpe‐expressing line, the elongated rectangular cells in the proximal region of petals retained their longitudinal form and orientation (Figure 5D, panels d–f). However, these cells appeared to be more irregular with a more pronounced surface curvature compared to the wild type. SEM visualization of the adaxial petal epidermis revealed that in the bpe‐1 line and in the RNAi‐bpe‐expressing line, the conical differentiated cells were larger than the wild‐type cells, but retained their conical form and tight packing (Figure 5D, panels a and b). The RNAi‐bpep/BPEub plants also showed a similar increase in the size of petal conical cells (Figure 5B, panel c), in agreement with the role of BPEp in the control of petal size.

In the RNAi‐expressing plants, the petal adaxial epidermis cells were about 30% larger compared to the wt (Figure 5E, panel a). This increase in cell size corresponded to a reduction in cell number over the same surface area in the RNAi mutant compared to the wt (Figure 5E, panel b). These data are in agreement with a role for BPEp in limiting petal size by restricting postmitotic cell expansion. No significant difference in the sizes of leaves was observed between the wt and the knockout or the RNAi mutant plants (data not shown).


We used a cDNA‐AFLP‐DD strategy to identify genes involved in petal organogenesis downstream of organ identity genes. Three of the genes we identified encode putative petal‐, or petal and stamen‐specific proteins. Our results are in agreement with previous studies using microarray strategies, in the sense that a relatively small number of genes were found to be petal upregulated or petal‐specific (Zik and Irish, 2003; Wellmer et al, 2004). A possible explanation would be because the anatomy of the petal is less complex, composed of a smaller number of cell types compared to stamens or carpels, petal‐specific genes may be less abundant. Alternatively, many of the structural genes involved in petal organogenesis may also have roles in developmental processes in other floral or vegetative organs and would thus not be identified by a differential screen based on gene expression. However, it should be noted that none of the genes we identified in the present study were found in the microarray screens and among the petal upregulated genes identified in the microarray screens, only one (GDLS‐motif lipase/hydrolase like encoding gene) was identified in both studies (Zik and Irish, 2003; Wellmer et al, 2004). Therefore, it is likely that these studies have identified only a subset of the petal‐expressed genes and hence underestimate the number of genes required for petal organogenesis.

In this work, we identified BPE that encodes two bHLH TFs (BPEp and BPEub) originating from an alternative splicing event (intron retention in BPEp). This post‐transcriptional event is correlated with differential expression profiles of these two transcripts: BPEp shows a petal‐expression pattern, while BPEub is ubiquitously expressed. The presence of a ubiquitous BPEub transcript explains why the BPE gene was not identified in previous microarray studies (Zik and Irish, 2003; Wellmer et al, 2004).

BPEp mRNA accumulates to equally high relative abundance in flowers of both 35S:PI and 35S:AP3 lines. It is known that PI and AP3 form a heterodimer and work together in an autoregulatory loop to maintain each other's expression (Goto and Meyerowitz, 1994; Jack et al, 1994; Krizek and Meyerowitz, 1996; Riechmann et al, 1996; Tilly et al, 1998). Ectopic expression of AP3 results in persistent fourth whorl expression of the endogenous PI gene. Similarly, in plants overexpressing PI (35S:PI), AP3 autoactivates its expression in whorl 1 and in other regions where PI is active (Goto and Meyerowitz, 1994; Jack et al, 1994; Krizek and Meyerowitz, 1996). Therefore, BPEp presumably acts downstream of the PI/AP3 heterodimer during petal development. It should be noted that the endogenous AP3 and PI were highly expressed in flowers, but expressed at very lower levels in leaves of the 35S:PI and 35S:AP3 lines, respectively (data not shown), in agreement with previously reported work (Goto and Meyerowitz, 1994; Jack et al, 1994; Krizek and Meyerowitz, 1996). This explains why a small but significant increase in BPEp accumulation was observed in leaves compared to wt. These data provide additional evidence that PI and AP3 are able to induce BPEp accumulation.

AP3 is expressed in sepals but PI is not. As a result, the first whorl organs of a 35S:PI line have more functional AP3/PI than 35S:AP3 plants and are mosaic petal–sepal organs (Krizek and Meyerowitz, 1996). As there is more functional AP3/PI in the first whorl organs of the 35S:PI line compared to the 35S:AP3 line (Krizek and Meyerowitz, 1996), we expected to have more of the splice variant BPEp in 35S:PI than in 35S:AP3. However, our data show that BPEp expression is as highly induced in both overexpressing lines. Although, we do not have a clear explanation for this observation, the data suggest that other regulatory factors may be involved. The fact that in the 35S:PI line we observe a five‐fold greater accumulation of the BPEp transcript in flowers compared to leaves represents another argument that other factors whose expression is restricted to flowers are needed for efficient activation of BPEp expression.

Analyses of A. thaliana plants overexpressing SEP3 or SEP2 demonstrated an increase of BPEp expression in flowers, indicating the involvement of SEP in BPEp activation. To our knowledge, there are no known targets of SEP proteins downstream of flower organ identity genes. The only known targets of SEP3 are the organ identity genes AP3 and AG (Castillejo et al, 2005). Therefore, it is possible that (i) BPEp is activated downstream of SEP3 via AP3 or (ii) that BPEp accumulation is activated downstream of the petal organ identity complex formed by AP3, PI, SEP and AP1 (Honma and Goto, 2001; Theissen and Saedler, 2001). However, although ectopic expression of SEP3 is sufficient to highly activate AP3 in leaves (Castillejo et al, 2005), we did not observe a strong accumulation of BPEp transcript in leaves of the 35S:SEP2 or 35S:SEP3 lines (in contrast to the situation in flowers; Figure 3B). Thus, the second possibility involving regulation via the MADS protein complex is likely to be correct. Moreover, we found that BPEp expression is activated by AP1 in an indirect manner. During early floral stages, AP1 is required to specify floral meristem identity. At later floral stages, AP1 is required to specify the identity of sepals and petals by activating B class genes via the UNUSUAL FLORAL ORGANS gene (UFO) (Mandel et al, 1992; Ng and Yanofsky, 2001; Durfee et al, 2003). Therefore, BPEp may be activated downstream AP1 via AP3 and PI. As the tissues used in this experiment were mostly from flower buds older than stages 5 and 6, BPEp accumulation is most likely to be controlled by the protein complex formed by AP3, PI, SEP and AP1 during petal organogenesis. These data taken together indicate that the regulation of BPEp accumulation is likely to occur downstream of floral organ identity MADS protein complexes.

As BPEp is positively regulated downstream of AP3, PI and SEP3, we would expect it to be highly expressed in the second and third whorls (petals and stamens, respectively). However, the BPEp transcript accumulated to a low level in flowers of apetala1 and apetala2 mutants which possess stamens and lack petals. This low level of expression in stamens could be explained by the analyses of BPEp expression in the flowers of agamous and 35S:AG‐GR plants which showed that BPEp is negatively regulated by AG. It must be noted that at 24 h post‐DEX treatment, the flowers do not show any phenotypic difference from the non‐DEX‐treated plants: flowers with repeated sepal–petal–petal. This is in agreement with the timing of flower organ development (Smyth et al, 1990). It takes more than 7 days from the identity specification of petals to stage10 at which BPEp starts to be highly expressed. Taking this into account, it is likely that the activation of AG in petal cells leads to the almost immediate negative regulation of BPEp observed. Our results demonstrate the involvement of AG in the negative regulation of a petal preferentially expressed transcript. These results are in agreement with a previous report of an AG‐dependent inhibitory pathway where an unknown factor produced in whorl 3 has been proposed to block petal development (Durfee et al, 2003). It is therefore possible that the negative regulation of BPEp by AG is mediated by this unknown factor.

BPEp is produced by an alternative splicing event, thus involving control at the post‐transcriptional level. Such regulation is likely to occur via intermediate proteins acting downstream of PI, AP3, SEP and/or AP1 in agreement with the fact that regulation downstream AP1 requires de novo protein synthesis. Alternative splicing is a major mechanism for expanding functional diversity of animal and plants genes (Kriventseva et al, 2003; Ner‐Gaon et al, 2004). Thus, BPEp represents another example of functional diversification by means of alternative splicing, and the elucidation of its origin at the evolutional level would be interesting.

All loss‐of‐function mutants (bpe‐1 and RNAi‐bpe lines where both BPEp and BPEub expression were affected) showed a significant increase in petal size compared to wt plants. The RNAi‐bpep/BPEub‐expressing plants (which exhibit downregulation of BPEp, whereas BPEub is not effected) showed increased petal size compared to the wt. These results demonstrated clearly the role of the petal‐expressed splice variant BPEp in influencing petal size. In the BPE loss‐of‐function mutants, the flower organs (other than petals) exhibit no size phenotypic modifications. These data constitute another argument toward the specificity of BPEp involvement during petal development. Organ growth is a well‐coordinated process. Therefore, cellular proliferation and expansion must be tightly controlled to give a remarkably constant final size of petals. Failure in maintaining correct cell proliferation and/or expansion will therefore result in organ size modification. To date, loss‐of‐function analyses have identified few genes involved in plant floral and/or vegetative organ morphogenesis. In one recent example, the E3 ubiquitin ligase‐encoding gene BIG BROTHER was shown to limit plant organs size by controlling cell proliferation (Disch et al, 2006). AINTEGUMENTA (ANT) gene has been shown to control organ growth by promoting cell proliferation, thus in an antagonist manner to BIG BROTHER (Krizek, 1999; Mizukami and Fischer, 2000). Similarly, JAGGED and FRILL genes have been described to control organ growth mainly by affecting cell cycle activity, by controlling either the mitotic state or the endo‐reduplication processes, respectively (Hase et al, 2000; Dinneny et al, 2004).

Petals of the BPEp loss‐of‐function mutants possess less cells for the same surface area compared to the wt. These results demonstrate that BPEp controls A. thaliana petal size by restricting cell growth. Therefore, BPEp is the first protein that specifically limits petal organ size by controlling the postmitotic rate of cell growth and expansion. These data are in agreement with the expression profile of BPEp starting at stages when cell differentiation occurs (maximum expression at stages 10 and above; Figure 2G).

The two predicted BPE‐encoded proteins possess C‐terminal stretches that do not show similarities to each other. The significance of these sequences to the biological function of these two putative TFs is yet to be determined. Three TFs belonging to the No Apical Meristem (NAM) family, namely NAM, CUC and NAP (act downstream AP3/PI), and one bHLH (SPATULA) have also been shown to be involved in flower organogenesis (Souer et al, 1996; Aida et al, 1997; Sablowski and Meyerowitz, 1998; Heisler et al, 2001). Hence, bHLH and NAM TFs may be important components in the genetic control of flower organogenesis downstream of the organ identity genes. In turn, BPEp and NAP could be intermediate TFs responsible for regulating the downstream structural genes that directly mediate petal and/or stamen morphogenesis. It is worth noting that published studies suggest that certain bHLHs interact with MYB TFs and WD40 repeat proteins to form complexes involved in multiple cellular differentiation pathways leading to diverse cell fates (Ramsay and Glover, 2005). In a similar way, the bHLHs involved in flower organ development may act in complexes with other proteins (MYB, WD40, and/or other proteins) to activate the downstream structural genes. Identifying, on the one hand the interacting proteins and on the other hand the target genes of BPEp and NAP will, therefore, help to unravel the network of structural genes (realizators) involved in petal morphogenesis.

Materials and methods

Constructions and plant lines

Arabidopsis mutants agamous‐1, pistillata‐1, apetala3‐3, apetala1 and apetala2 and the lines overexpressing PI (35S:PI) or AP3 (35S:AP3) or SEP2 (35S:SEP2), SEP3 (35S:SEP3), AG (35S:AG‐GR) or AP1 (35S:AP1‐GR) have been described (Bowman et al, 1989; Jack et al, 1992; Mandel et al, 1992; Jack et al, 1994; Jofuku et al, 1994; Krizek and Meyerowitz, 1996; Pelaz et al, 2001a, 2001b; Ito et al, 2004; Yu et al, 2004).

bigpetal‐1 (bpe‐1) T‐DNA insertion knockout line for BPE was identified in a screen of Wisconsin A. thaliana mutant collections. The two RNA interfering lines (RNAi‐bpe and RNAi‐bpep) were generated by expressing inverted repeats of nucleotide sequence corresponding to two different regions (Figure 2A) in the target gene under the control of the CaMV 35S promoter in the vector pK7GWIWG2 (Karimi et al, 2002).

The 1317 bp DNA sequences up to and including the BPE start codons (corresponding to nucleotides 62 136–61 453 in the A. thaliana genomic sequence, GenBank accession AC009317) was PCR amplified and cloned upstream of the GUS sequence in the pMDC162 vector (Curtis and Grossniklaus, 2003). Resulting constructs harboring RNAi or PromoterBPE:GUS expression cassette were transformed into A. thaliana Col‐0 plants. GUS staining was performed as previously described (Hill et al, 1998). All plants were kept in growth chambers with a condition of 16/8 h day/night at 22°C and 70 μE/m2/s light.

Treatment of plants with DEX and CHX

Plants were treated with 10 μM (35S:AG‐GR) or with 1 μM (35S:AP1‐GR) DEX solution containing 0.01% DMSO, 0.1% (v/v) ethanol and 0.01% (v/v) Silwet‐L77. The same solution without DEX was applied to control plants under the same conditions. CHX treatment was performed by adding 10 μg/ml of CHX to the DEX solution.

RNA preparation and cDNA synthesis

Total RNA was prepared using the Trizol reagent (Invitrogen, Groningen, NL) according to the manufacturer's instructions. Contaminating DNA was removed using the DNA‐free™ kit (Ambion, Cambridgeshire, UK). Total RNA was used for cDNA synthesis using a Revert Aid M‐MuLV Reverse transcriptase (Fermentas, Ontario, Canada) according to the manufacturer's recommendations.

cDNA‐amplified fragment length polymorphism‐differential display (AFLP‐DD)

The cDNA‐AFLP‐DD was performed using the DisplayPROFILETM kit (Q‐BIOgene, Irvine, CA) following the manufacturer's guidelines. Amplified reaction products were separated side‐by‐side on a 6% polyacrylamide gel under denaturing conditions. DNA bands were excised from the gel, cloned, their nucleotide sequence was determined and their corresponding gene was identified in the A. thaliana genomic sequence.

Mapping of the 5′ and 3′ ends of the BIGPETAL mRNA transcripts

The Marathon™ cDNA amplification method was used to map the exact 5′ and 3′ ends of BPEp and BPEub mRNA from wt A. thaliana Columbia flowers, according to manufacturer's recommendations (Clontech, Mountain View, CA).

Gene expression analysis

Primers (available upon request) specific to each gene were used for expression analysis by RT–PCR and RT‐QPCR. RT‐QPCR was performed with the qPCR Core Kit for SYBR Green I Quick Gold Star (Eurogentec, Liege, Belgium) using the Gene Amp 5700 Sequence Detector System (Applied Biosystems, Foster City, CA). Reactions were run in duplicate and quantified against a relative standard curve made from a serially diluted stock cDNA containing the target sequence. Data collection and analysis were performed using GeneAmp 5700 SDS Software package (Applied Biosystems). Results were expressed using the ΔΔCt calculation method in arbitrary units by comparison to the control (flower of a wild‐type plant). The housekeeping gene TCTP we identified as constitutively expressed (this study), was used as a control in all gene expression analyses.

SEM and organ and cell size measurements

Petals from flowers at stage 14–15 (maximum expansion (Smyth et al, 1990) were carefully dissected and mounted flat on their abaxial surface. Samples were imaged using a Hitachi S 3000N scanning electron microscope.

Two adjacent petals were removed from flowers at developmental stage 14 (Smyth et al, 1990) and their length and width (distal region of petals blade) were measured using digital images. For cell size measurements, petals were cleared overnight in a solution containing 86% ethanol–14% acetic acid followed by two times 4 h incubation in ethanol 70%. Digital images of the distal region of the cleared petals were used to measure the adaxial epidermis cell size. Cell number was calculated per 500 μm2 surface area using SEM images and digital images of the cleared petals.


We thank P Chambrier for technical help and P Vergne, A Dabois and J Traas for critical reading of the manuscript, the Salk Institute Genomic Analysis Laboratory for providing the sequence‐indexed Arabidopsis T‐DNA insertion mutants and the ABRC and NASC for providing the lines. This work was funded from the ACI‐Young Investigator Program from the French Ministry of Research and by the French National Institute for Agronomic Research (INRA) to MB. JS and CJ were funded by fellowships from the Biologie Végétale department of INRA and ACI‐YI program, respectively.