Introduction

The evolution in Asteraceae of false but functional flowers (pseudanthium), characterized by combination of different shaped ray and disk florets, in the same inflorescence (radiate capitulum), has a pivotal ecological role (Knowles 1978; Harris 1995). The showy appearance of the sunflower (Helianthus annuus L.) inflorescence depends on the outermost whorl of large ray flowers with monosymmetric (zygomorphic) corolla because hermaphrodite disk flowers are smaller and polysymmetric (actinomorphic) (Fig. 1a). Monosymmetric ray flowers have evolved an attraction role as they are sterile because they retain only filamentous remnants of aborted stamens and empty ovaries (Knowles 1978). A variation of ray flower morphology has been described in the tubular ray flower (turf) mutant where a single nuclear recessive mutation causes the development of atypical capitula characterized by ray flowers with a nearly polysymmetric tubular-like corolla (Fig. 1b, c; Berti et al. 2005; Fick 1976). These flowers, arranged in the outmost whorl, maintain their positional identity because they are bigger than the polysymmetric disk florets, but also achieve the ability to differentiate both stamens and pistils (Fig. 1d; Berti et al. 2005). The turf plants produce tubular-like ray flowers in which stamen filaments, anthers and styles display some developmental defects (Berti et al. 2005), but morphological and genetic analysis demonstrate that both male and female organs are functional (Berti et al. 2005). The turf mutation is unstable, and reversions to heads with monosymmetric ray flowers (with about 1 % frequency) have been described (Fambrini et al. 2007).

Fig. 1
figure 1

Morphological characteristics of the tubular ray flower (turf) mutant of sunflower (H. annuus L.). a Heterogamous inflorescence of normal (TURF) sunflower. b A normal ray flower (left) and turf tubular-like ray flower (right). c Inflorescence of the turf mutant. d Longitudinal section of the turf tubular-like ray flower 1 week before the anthesis; a anther, co corolla, f filament, o ovule, ov ovary, s style, st stigma. Scale bar = 1.7 mm

Full size image

In several eudicots, the CYCLOIDEA-(CYC)-like genes, which belong to the TCP class of transcription factors, are key players for floral symmetry (Costa et al. 2005; Donoghue et al. 1998; Martín-Trillo and Cubas 2010; Preston and Hileman 2009). The acronym TCP derives from the first four genes isolated, TEOSINTE-BRANCHED1 (Zea mays), CYC (Antirrhinum majus) and PROLIFERATING CELL FACTOR 1 and 2 (Oryza sativa) (reviewed in Martín-Trillo and Cubas 2010; Preston and Hileman 2009). All TCP genes contain a 59-amino acid basic helix-loop-helix (bHLH) motif common to all the members that allows DNA binding and protein–protein interactions (Cubas et al. 1999a). This function is consistent with the role for TCP genes as transcriptional regulators of cell division and growth (Martín-Trillo and Cubas 2010). Members of this family are divided into two classes (I and II) based on their structure and TCP domain. In addition to the bHLH motif, CYC-like genes, which belong to the TCP class II, have a specific motif, an 18–20 residue arginine-rich motif (the R domain) and a glutamic acid-cysteine-glutamic acid stretch (also known as an ECE motif). Through phylogenetic and sequence analysis, members of class II can be divided into CYC/TB1-like and CINCINNATA-like clades. The first clade is further divided into CYC1, 2 and 3 subclades which have evolved due to a series of duplication events (Howarth and Donoghue 2006; Martín-Trillo and Cubas 2010; Preston and Hileman 2009). Analysis of different angiosperm species shows that most genes of the CYC2 clade were recruited to function in floral symmetry control (Citerne et al. 2000, 2003; Cubas et al. 1999b; Damerval et al. 2007; Luo et al. 1996, 1999). More complicated cases of CYC2 recruitment appear to have occurred in members of the Asteraceae and Dipsacaceae families characterized by radiate inflorescences (Abbott et al. 2003; Broholm et al. 2008; Busch and Zachgo 2009; Carlson et al. 2011; Kim et al. 2008). In particular, the radiate species of Dipsacaceae have undergone more extensive diversification of CYC-like genes than discoid species suggesting the hypothesis that these genes could be implicated in the evolution of radiate heads (Carlson et al. 2011). To date, in Asteraceae, the molecular dissection of radiate inflorescences in Senecio and Gerbera has demonstrated that flower identity within the inflorescence depends on the expression of CYC-like genes (Broholm et al. 2008; Kim et al. 2008). In sunflower, ten CYC-like genes have been isolated and the diversification of this ample CYC gene family depends on duplication events probably followed by functional divergence (Chapman et al. 2008). Recently, a comparative study has been conducted in gerbera and sunflower to explore CYC/TB1 gene family evolution and diversification in Asteraceae (Tähtiharju et al. 2012). In these two species, marginal florets (ray and trans flowers) showed a strong expression level of the six members of CYC2 clade in gerbera and the five genes in sunflower. Tähtiharju et al. (2012) hypothesized for these genes early functions in flower type differentiation and late functions in reproductive organ development. Moreover, the null or very low level of expression in disk florets of one gene in gerbera (GhCYC3) and two in sunflower (HaCYC2d and HaCYC2c) could be indicative to their role in controlling ray flower identity (Chapman et al. 2008; Tähtiharju et al. 2012). Functional specificity for CYC2 clade members in sunflower has been supposed by flower-type-specific expression and protein–protein interactions (Chapman et al. 2008; Tähtiharju et al. 2012) but the control of ray flower identity is unclear.

Mutants with altered inflorescence morphology can provide information on the roles of CYC-like genes in radiate inflorescence evolution (Coen et al. 1995). In this study, we report the molecular characterization of turf. Our results identify HaCYC2c as a key regulator of ray flower monosymmetry. Also, HaCYC2c loss-of-function promotes the developmental switch from sterile to hermaphrodite flowers, revealing a novel and unexpected role for a CYC-like gene in the repression of female organs.

Materials and methods

Plant material and growth condition

Sunflower seedlings (H. annuus L.) from wild type (TURF), mutant tubular ray flowers (turf), F1 (turf × TURF) progenies, F2 progenies, turf plants reverted to TURF phenotype and progenies derived from self-fertilization of revertants were grown in field conditions (Experimental fields, University of Pisa, Pisa, Italy). Plant growth conditions were as previously described (Fambrini et al. 2007). Briefly, progenies were grown using 50 cm inter-row spacing with 25–30 cm between plants (about 8–9 plants m−2).

Histological analysis

Ray flowers were fixed as previously described (Fambrini et al. 2006). Paraffin-embedded tissues were sectioned using a rotary microtome (Reichert, Vienna, Austria). Longitudinal sections (10 μm thick) were stained with a mixture of three dyes as described by Graham and Trentham (1998) and examined under a light microscope (DMRB Leitz, Wetzlar, Germany).

Isolation of the HaCYC2c gene in wild type (TURF) and turf mutant

Expression analysis previously reported in sunflower (Chapman et al. 2008), suggested that HaCYC2c, a CYC-like gene, was the probable candidate for the TURF locus. The GenomeWalker DNA walking approach was used to extend the sequence to 5′ untranslated region of the HaCYC2c gene (Fambrini et al. 2011; Siebert et al. 1995). DNA was extracted from leaves of 30-days-old plants of both TURF and mutant plants with the NucleoSpin® Plant II DNA extraction and purification protocol for plant tissue, according to the manufacturer’s instructions (Macherey–Nagel GmbH & Co., Düren, Germany). Genomic DNA digested with the DraI restriction enzyme was ligated to the Genome Walker Adaptor according to the manufacturer’s instructions (Clontech Laboratories, Inc., Mountain View, CA, USA). The first PCR was performed with the AP1 primer provided in the kit and the HaCYC2c-specific primer RAG5R (reverse, 5′-ACCCAGACCCTTCATGATCAAGCAGCTGGT-3′) (see Supplemental Fig. 1 for primer position in both TURF-HaCYC2c and turf-HaCYC2c sequences). The nested PCR was performed with the nested adaptor primer (AP2) and the nested HaCYC2c-specific primer RAG6R (reverse, 5′-TGGAGGTAGGGTTGATGGTTGTGGAGCA-3′). A PCR product of about 1,750 bp was obtained. All PCRs were performed in conditions recommended by manufacturer (Clontech).

Genomic DNA and first strand cDNA from TURF H. annuus were used as templates in PCR reactions to identify the introns/exons regions of HaCYC2c. Total RNA was extracted from immature ray flowers (3–4 mm long) with the TRIzol® Reagent, according to the manufacturer’s instructions (Invitrogen™ life technologies, Carlsbad, CA, USA). Total RNA (4 μg), was used with the Superscript pre-amplification kit (Invitrogen™), to produce, the first strand cDNA in conditions recommended by the manufacturer. Two gene-specific primers were used: CYC10 (forward, 5′-CAGTGTAAAGCTTAAGAGGGGTGG-3′) placed 67 bp before the start codon and CYC11 (reverse, 5′-CCAATCAACTAGTGCCGAGAAGTGG-3′) placed 179 bp after the stop codon (Supplemental Fig. 1). PCRs were performed in 20 μl of 1× buffer (Promega Italia, Milano, Italy) containing 0.2 mM dNTPs, 0.5 μM of each primer, 2 mM MgCl2, 0.5 U GoTaq DNA polymerase (Promega) and 50 ng of genomic DNA or 1 μl of first strand cDNA. From DNA a PCR product of 1,392 bp was obtained (Supplemental Fig. 2). From cDNA a PCR product of 1,310 bp was obtained. The PCR conditions were: 95 °C for 4 min, 35 cycles (30 s at 94 °C, 1 min 20 s at 60 °C, 30 s at 72 °C), with a final extension of 7 min at 72 °C.

When using mutant DNA as a template, no amplified products were obtained with the primer combination CYC10/CYC11. However, several other primer combinations were able to produce PCR products (Supplemental Fig. 1). In the 5′ region of HaCYC2c, a product of 404 bp was obtained with the primer combination CYC10/RAG5R while in the 3′ region, a product of 778 was obtained with the primer combination RAG14F (forward, 5′-ATTGGCTCTTTACCAAGTCCAAGAAGGCGA-3′)/CYC11. The sequences were identical to those of the TURF gene (Supplemental Fig. 3a, b). These results suggested that, within the 210 bp-long region between the primers RAG5R and RAG14F of the turf-HaCYC2c gene, a chromosomal rearrangement occurred.

The GenomeWalker DNA walking approach was used to find genomic DNA sequences between the RAG5R and RAG14F region of the turf-HaCYC2c gene (Supplemental Fig. 1). In the 3′ direction genomic DNA digested with the DraI restriction enzyme was ligated to Genome Walker Adaptor according to the manufacturer’s instructions (Clontech). The first PCR was performed with the AP1 primer provided in the kit and the HaCYC2c-specific primer CYC6F (forward, 5′-TGCTCCACAACCATCAACCCTACCTCCAGT-3′). The nested PCR was performed with the nested adaptor primer (AP2) and the nested HaCYC2c-specific primer CYC5F (forward, 5′-ACCAGCTGCTTGATCATGAAGGGTCTGGGTT-3′). A PCR product of about 1,500 bp was obtained. In the 5′ direction genomic DNA digested with the EcoRV restriction enzyme was ligated to Genome Walker Adaptor according to the manufacturer’s instructions (Clontech). The first PCR was performed with the AP1 primer provided in the kit and the HaCYC2c-specific primer RAG15R (reverse, 5′-CCTGATCCTCCCTTGATGGTCTCTAGGAA-3′). The nested PCR was performed with the nested adaptor primer (AP2) and the nested HaCYC2c-specific primer RAG14R (reverse, 5′-TCGCCTTCTTGGACTTGGTAAAGAGCCAAT-3′). A PCR product of about 5,000 bp was obtained. All PCRs were performed in conditions recommended by manufacturer (Clontech).

All PCR products were separated by electrophoresis on a 1.5 % TAE-agarose gel and visualized with ethidium bromide under UV light. All the amplified products were purified using the Wizard® SV Gel and PCR Clean-UP System (Promega Italia, Milano, Italy), ligated into the pGEM-T Vector (Promega), and transformed in Escherichia coli JM109 competent cells (Promega). Plasmid DNA was prepared using Wizard® Plus Minipreps DNA Purification Kit (Promega). Several clones were sequenced on both strands. Sequence data from this article have been deposited in GenBank under the following accession numbers: HE604335 and HE604336.

Database search and sequences analysis

Database searches were carried out using the BLAST program at the National Center for Biotechnology Information (NCBI) (Altschul et al. 1997). PROSITE and PFAM databases were searched to identify conserved domains (Bateman et al. 2002; Falquet et al. 2002). The databases for Triticeae repetitive elements (TREP, http://wheat.pw.usda.gov/ITMI/Repeats) were also screened with the BLASTN and BLASTX algorithms. The chosen sequences were aligned using ClustalW version 1.7 (Thompson et al. 1994), and the EMBOSS pairwais alignment algorithms from EMBL-EBI (European Bioinformatics Institute; http://www.ei.ac.uk/emboss/align/).

Co-segregation analysis

Seedlings from wild type (TURF), mutant (turf), F1 (turf × TURF) progenies, one F2 progeny, mutant plants reverted to wild type and one progeny derived from self-fertilization of a revertant were grown in field conditions as described above. DNA was extracted from leaves of 20-days-old plants as described above. The primer combination CYC10/CYC11 was used for genotype screening identifying TURF/TURF and heterozygous (TURF/turf) material. In the turf genome (turf/turf), the PCR conditions were unable to amplify the ample sequence (7,179 bp) spanning from CYC10 to CYC11 (Supplemental Fig. 1 and Supplemental Fig. 4). The primer combination RAG19 (forward, 5′-GGTTGCATTTGAGTGTAACAAAGG-3′)/CYC11 amplified a product of 1,259 bp. The sequence of the primer RAG19 is placed in the transposable element and therefore, the PCR product was expected only in homozygous (turf/turf), and heterozygous (TURF/turf) genotypes (Supplemental Fig. 1 and Supplemental Fig. 4). The PCR conditions for both primer combinations were: 95 °C for 4 min, 30 cycles (30 s at 94 °C, 1 min 20 s at 60 °C, 30 s at 72 °C), with a final extension of 7 min at 72 °C. The PCR products were separated by electrophoresis on a 1.5 % TAE-agarose gel and visualized with ethidium bromide under UV light.

Results and discussion

A transposable element is inserted in the HaCYC2c gene of the turf mutant

The HaCYC2c gene from wild type (TURF/TURF) seedlings was isolated using PCR and Genome walking approaches. Sequence analysis showed a 1,146 bp coding sequence (CDS) (Fig. 2a and Supplemental Fig. 1 and Supplemental Fig. 2). The putative TURF-HaCYC2c peptide was 381 amino acids long with a calculated molecular mass of 43,570 Da. No nucleotide polymorphisms were detected between the CDS sequences of TURF-HaCYC2c and the HaCYC2c previously isolated in sunflower (GenBank accession number EU088370; Chapman et al. 2008). However, in the HaCYC2c gene isolated from the USDA inbred line Ames 3963 (Chapman et al. 2008), the intron is positioned before the stop codon (see Fig. 2, Chapman et al. 2008). By contrast, the TURF-HaCYC2c intron (82 bp) is localized outside the CDS, within the 3′-UTR region, 11 bp after the stop codon (Fig. 2a, Supplemental Fig. 1 and Supplemental Fig. 2). The intron position was deduced from a comparison between HaCYC2c cDNA and the genomic DNA sequences. The 3′-UTR position of this intron was confirmed in an assortment of inbred lines of our germplasm collection (data not shown). Indeed, intron position is well conserved within the same species; however, we cannot exclude differences related to gene organization between distinct inbred lines. The 404 bp 5′ region as well as the 778 bp 3′ regions amplified from the turf genomic DNA were identical with the nucleotide sequence of TURF-HaCYC2c (Supplemental Fig. 3a, b). However, PCR experiments suggested a probable chromosomal rearrangement in proximity of the TCP domain of the turf-HaCYC2c gene (see also Materials and Methods). Genome walking was used to explore this genomic region. A large insertion (5,787 bp) was found in the basic motif of the TCP domain, 427 nucleotides after the start codon. Sequence analysis performed in public databases and the database for Triticeae repetitive elements (TREP, http://wheat.pw.usda.gov/ITMI/Repeats) suggested that this insertion was a part of a transposable element (TE) belonging to the CACTA superfamily of transposons (En/Spm, Tgm, and Tam) (Fig. 2b, Supplemental Fig. 4 and Supplemental Fig. 5). CACTAs are class II TEs that utilize a cut-and-paste mechanism for transposition, which requires a transposase enzyme (Feschotte and Pritham 2007; Muszynski et al. 1993). The name ‘CACTA’ refers to the flanking terminal inverted repeats (TIRs), which are normally 10–28 bp long and terminate in a conserved 5′-CACTA-3′ motif (Wicker et al. 2003). The TIRs serve as recognition sequences for the transposase enzyme (Lewin 1997). CACTA elements also have characteristic sub-terminal repetitive (sub-TRs) regions of 10–20 bp units that are repeated in direct and inverted orientations (Feschotte and Pritham 2007). In the HaCYC2c gene, the insertion of this sequence created a perfect three bp (ATA) target site duplication (TSD), characteristic of the CACTA superfamily (Fig. 2b and Supplemental Fig. 4). Its 5′ and 3′ ends carried imperfect TIRs flanking the conserved CACTA sequence: a 16 bp TIRs contained one mismatch (Fig. 3a), while the outer 15 bp of the 3′ termini of the sunflower CACTA-like element were identical to those of the maize En/Spm TE (Fig. 3b). Sub-terminal repeats regions were also identified (Fig. 3c); we named this element Transposable element of turf1 (Tetu1).

Fig. 2
figure 2

Schematic representations of HaCYC2c gene structure in the wild type (TURF, a), and mutants (turf1, b and turf2, c) of sunflower (H. annuus L.). GenBank accession numbers are HE60435 and HE60436. The exons are boxed in light-blue. The introns are shown as yellow triangles. The class II TCP conserved domains, TCP and R, are boxed in green and brown, respectively. The unstable turf1 allele b presents the CACTA-like transposable element, Tetu1 inserted in the basic motif of the TCP domain (427 bp after the start codon). The stable turf2 allele c shows a footprint (four bp) generated by an imperfect excision of Tetu1. The partial 3′- and 5′-UTR are shown in grey. The primers used for molecular analysis are indicated. The arrows indicate the 5′–3′-primer direction

Full size image
Fig. 3
figure 3

Characteristics of the CACTA-like transposable element (TE) Tetu1 inserted in the HaCYC2c gene of the tubular ray flower (turf) mutant of sunflower (H. annuus L.). a Alignment of terminal inverted repeats (TIRs) of Tetu1. b Alignment of TIRs comparing Tetu1 and the CACTA TE En/Spm. Asterisks indicate identical nucleotides. The 3′ transposon terminal sequence of Tetu1 is in reverse-complement counterpart. c Schematic representation of TIRs and sub-terminal repeat (sub-TRs) regions of Tetu1. The red arrows indicate the TIRs; the coloured triangles indicate the sub-TRs regions in direct and inverted orientations. The TIRs and sub-TRs were deduced by sequence analysis of Supplemental Fig. 4

Full size image

The insertion changes the reading frame of turf-HaCYC2c for the encoded protein and led to differences in the amino acid sequence as well as to a premature stop codon (Supplemental Fig. 5). Sequence analyses indicated that Tetu1 does not contain a transposase encoding gene(s). We hypothesized that Tetu1 is a non-autonomous truncated and/or rearranged version of a CACTA TE, most likely generated by an imperfect excision of a complete element. Alignment of the whole turf-HaCYC2c sequence to the TURF-HaCYC2c sequence using the EMBOSS pairwise alignment showed that, except for the Tetu1 insertion, the two allelic sequences were identical (Supplemental Fig. 6).

The sunflower genome is extremely repetitive with large number of retrotransposons (Kane et al. 2011). Some transposon superfamilies have been identified and their activity is an important component of reorganizations in the genome of interspecific hybrids (Anisimova et al. 2009). However, turf is the first spontaneous mutant in sunflower characterized by a CACTA element insertion in a CYC-like gene with a distinct role in development of radiate inflorescence. Nevertheless, transposon-mediate mutations of CYC genes affecting flower symmetry were described in the model plant A. majus (Carpenter and Coen 1990; Luo et al. 1999). Two independent recessive mutants (cycloidea-608 and cycloidea-609) characterized by a more radially symmetrical flower than the wild type were induced by a transposon-mediated strategy of mutagenesis using CACTA elements (Carpenter and Coen 1990). Moreover, in the spontaneous mutant backpetals a transposon inserted in the regulatory region of the CYC gene caused ectopic expression in the lateral and ventral regions of the flower (Luo et al. 1999). This mutation is semidominant and genetically unstable; in particular, the transposon excision in revertants demonstrated that the insertion was indispensable for the development of dorsalized petals. Furthermore, in Antirrhinum, Roccaro et al. (2005) reported that the binding factor ROSINA (RSI) can bind the promoter of the MADS-box DEFICIENS gene involved in petal and stamen identity. The genomic organization of several RSI copies proved that this DNA binding factor is a part of a CACTA TE (Roccaro et al. 2007).

The mutation in the HaCYC2c gene co-segregates with the turf mutant phenotype

Co-segregation analysis of parental, F1, F2 and populations of self-fertilized plants derived from reversion of turf to wild type phenotype was performed using a PCR approach with the primer combinations CYC10/CYC11 and RAG19/CYC11 (Fig. 2a, b and Supplemental Fig. 1). We observed full correlation with the segregating phenotypes and inheritance of turf alleles (Fig. 4a–c and Supplemental Fig. 7a, b). In the wild type (TURF/TURF, TT), the only primer combination that gave a PCR product was CYC10/CYC11 (Fig. 4a). By contrast, in the turf mutant (tt) only the RAG19/CYC11-PCR product was obtained. As expected, from heterozygous (Tt) F1 plants both PCR products were amplified (Fig. 4a). Within one F2 progeny of 48 plants, 38 individuals showed normal monosymmetric ray flowers, 13 of which displayed a single CYC10/CYC11-PCR product (genotype TT) and 25 of which displayed both PCR products (genotype Tt). From the 10 individuals with polysymmetric tubular-like flowers, only a RAG19/CYC11-PCR product was obtained (genotype tt), in agreement with the expected 1:2:1 genotypic segregation (χ2 = 0.458; P = 0.70–0.80; df = 2). A previous genetic analysis demonstrated that reversion of turf to wild type affected a single allele (Fambrini et al. 2007). Here, the molecular analysis of one progeny (44 plants) derived from a self-fertilized reverted plant (Fig. 4c), unmistakably validated that the TURF locus was the gene HaCYC2c. Indeed, in homozygous (TT) and heterozygous (Tt) individuals, the CYC10/CYC11-PCR products showed identical nucleotide sequences to wild type HaCYC2c-TURF gene (Fig. 4c and Supplemental Fig. 7a). As expected, identical nucleotide alignment was also detected in RAG19/CYC11-PCR products obtained from homozygous (tt) and heterozygous (Tt) individuals and the RAG19/CYC11-turf sequence (Fig. 4c and Supplemental Fig. 7b). Of the 44 individuals, 12 were homozygous (TT) and 24 heterozygous (Tt) both with a normal inflorescence, and 8 were homozygous (tt) with a turf phenotype, in agreement with the expected 1:2:1 ratio (χ2 = 1.091; P = 0.50–0.70; df = 2). Therefore, the turf mutant co-segregated with the mutation in the HaCYC2c gene and the Tetu1 excision was responsible for the reversion of the mutant to wild type phenotype.

Fig. 4
figure 4

Co-segregation analysis. a PCR screening of genotypes with the primer combination CYC10/CYC11 (lanes 1, 3 and 5) and RAG19/CYC11 (lanes 2, 4 and 6) in homozygous normal plants (TURF), homozygous mutant plants (turf1) and F1 hybrid (TURF × turf1) (for primer position see also Fig. 2, Supplemental Fig. 1, Supplemental Fig. 2 and Supplemental Fig. 4). b, c Co-segregation analysis performed as described in a in a F2 progeny (b) and in a progeny derived from a self-fertilized reverted plant (c). The primer combinations (PC) are indicated on the left of gels. M and N indicate the mutant and normal phenotype, respectively. TT, Tt and tt indicate the genotypes for the alleles TURF (T) and turf (t). N* and Tt* indicate the phenotype and the genotype of a reverted plant, respectively. The PCR products were resolved on ethidium bromide-stained 1.5 % (w/v) TAE-agarose gels and visualized under UV light. The size of amplified PCR products is indicated. Ma = 1 kb DNA ladder (BioLabs Inc., New England, Ipswich, MA, USA)

Full size image

The CACTA excision generates stable turf mutant a new wild type phenotype

Excision of the Tetu1 element generated footprints in the HaCYC2c gene resulting in stable mutant phenotypes with frame shift mutations (see for example the sequence of turf2 generated by a four bp footprint and depicted in Fig. 2c). Notably, an homozygous reverted wild type phenotype displayed an excision of the Tetu1 that generated a deletion of three bp (GTA) in proximity of the TSD and therefore, located in the basic region of the TCP domain (Supplemental Fig. 8a, b). It is likely that the serine in the bHLH motif was not essential for the activity of the HaCYC2c gene. In HaCYC2c alleles of other three reverted wild type plants, Tetu1 excision only generated a canonical three bp (ATA) TSD (data not shown). Indeed, the presence, in some reverted wild type plants, of ray flowers with additional petaloid structures attached at the base of the corolla and filaments with small sterile anthers (Fambrini et al. 2007) could be related to footprints generated by Tetu1 excision that weakly influenced the HaCYC2c amino acid sequence. Also, the absence of chimerism in inflorescences reverted to a wild type phenotype suggested a preponderance and/or exclusive germinal excision of Tetu1, in agreement with the exact mendelian (3:1) segregation detected in progenies of self-fertilized reverted plants (Fambrini et al. 2007).

Although in Tetu1 a transposase gene is lacking, these results clearly suggest that it is still an active CACTA TE. It is possible that the non-autonomous Tetu1 element could be mobilized in trans by related transposases as long as it contains the sequence motifs recognized by those transposases that are essential for mobilization (e.g. TIRs and sub-TRs).

Conclusion

Our data clearly show that HaCYC2c is a key inducer of monosymmetry in ray flowers of sunflower. Also, according to Chapman et al. (2008), the HaCYC2c transcript accumulation in sunflower inflorescence is restricted to petals and ovaries of ray flowers. When HaCYC2c function is lost, the morphology of H. annuus inflorescence is redesigned in a similar manner to transgenic Senecio plants overexpressing the CYC-like gene RAY2 in the radiate background (Kim et al. 2008). In Gerbera hybrida loss and/or gain of function of GhCYC2 gene had marked effects on petal length and disc flowers morphology without change of corolla symmetry in ray flowers (Broholm et al. 2008). In particular, constitutive expression of the GhCYC2 gene resulted in shorter zygomorphic petals of ray flowers while disk flower petals showed a distinct ligular structure that resembled the bilaterally symmetric shape of ray and trans flowers. Moreover in one of the overexpression lines, all petals were fused to give peculiar tubular disk flowers (Broholm et al. 2008). In transgenic lines with reduced GhCYC2 expression, the length of trans flower petals was shorter than in wild type and occasional splitting of the trans flower ligules was observed (Broholm et al. 2008). Taken together these results suggest that in Asteraceae, species-specific mechanisms are involved in the molecular dissection of inflorescence development. Loss-of-function of HaCYC2c affects symmetry and fertility of ray flowers but the floret identity is maintained in turf mutant confirming a probable involvement of other genes, e.g. HaCYC2d of the CYC2 clade (Tähtiharju et al. 2012). In reproductive organs of mature disk flowers, significant levels of transcription for all genes of the CYC2 clade has recently been reported by Tähtiharju et al. (2012) suggesting late functions; however, the specific role of the HaCYC2c gene remains doubtful because morphology and fertility of disk flowers of turf are unaffected. Unexpectedly, the transposon-inactivation of the HaCYC2c gene was sufficient to promote stamen and pistil differentiation and to recover selfed seeds from outermost whorl of flowers (Fambrini et al. 2007). The relationship between the expression of CYC-like genes and stamen abortion has been documented in several species (Cubas et al. 1999a; Luo et al. 1996; reviewed in Preston and Hileman 2009) but their involvement in pistil development is a novel role for a gene of the CYC2 sub-clade.

This work is the first molecular characterization of a sunflower mutant characterized by an altered floral symmetry. Also, these results open a question about a probable interaction between TCP and MADS box genes in flower development of sunflower. The regulation of stamen and carpel growth by HaCYC2c activity should be mediated by MADS box gene with B, C and/or SEPALLATA function (Krizek and Fletcher 2005; Pelaz et al. 2000). On the basis of these data, we retain that further characterization of the turf mutant could be very useful to shed light on new aspects about the functional divergence of HaCYC2c in H. annuus.