RoKSN, a floral repressor, forms protein complexes with RoFD and RoFT to regulate vegetative and reproductive development in rose

Authors

  • Marie Randoux,

    1. INRA, Institut de Recherche en Horticulture et Semences (INRA, AGROCAMPUS-OUEST, Université d'Angers), SFR 4207 QUASAV, Beaucouzé Cedex, France
    2. Agrocampus Ouest, Institut de Recherche en Horticulture et Semences (INRA, Agrocampus-OUEST, Université d'Angers), SFR 4207 QUASAV, Angers, France
    3. Institut de Recherche en Horticulture et Semences (INRA, AGROCAMPUS-OUEST, Université d'Angers), Université d'Angers, SFR 4207 QUASAV, PRES LUNAM, Beaucouzé Cedex, France
    Search for more papers by this author
  • Jean-Michel Davière,

    1. Unité Propre de Recherche 2357, CNRS, Institut de Biologie Moléculaire des Plantes, Conventionné avec l'Université de Strasbourg, Strasbourg Cedex, France
    Search for more papers by this author
  • Julien Jeauffre,

    1. INRA, Institut de Recherche en Horticulture et Semences (INRA, AGROCAMPUS-OUEST, Université d'Angers), SFR 4207 QUASAV, Beaucouzé Cedex, France
    Search for more papers by this author
  • Tatiana Thouroude,

    1. INRA, Institut de Recherche en Horticulture et Semences (INRA, AGROCAMPUS-OUEST, Université d'Angers), SFR 4207 QUASAV, Beaucouzé Cedex, France
    Search for more papers by this author
  • Sandrine Pierre,

    1. INRA, Institut de Recherche en Horticulture et Semences (INRA, AGROCAMPUS-OUEST, Université d'Angers), SFR 4207 QUASAV, Beaucouzé Cedex, France
    Search for more papers by this author
  • Youness Toualbia,

    1. INRA, Institut de Recherche en Horticulture et Semences (INRA, AGROCAMPUS-OUEST, Université d'Angers), SFR 4207 QUASAV, Beaucouzé Cedex, France
    Search for more papers by this author
  • Justine Perrotte,

    1. INRA, Institut de Recherche en Horticulture et Semences (INRA, AGROCAMPUS-OUEST, Université d'Angers), SFR 4207 QUASAV, Beaucouzé Cedex, France
    Search for more papers by this author
  • Jean-Paul Reynoird,

    1. INRA, Institut de Recherche en Horticulture et Semences (INRA, AGROCAMPUS-OUEST, Université d'Angers), SFR 4207 QUASAV, Beaucouzé Cedex, France
    2. DNM Plant Breeding, Institut Polytechnique - LaSalle Beauvais, Beauvais, France
    Search for more papers by this author
  • Marie-José Jammes,

    1. INRA, Institut de Recherche en Horticulture et Semences (INRA, AGROCAMPUS-OUEST, Université d'Angers), SFR 4207 QUASAV, Beaucouzé Cedex, France
    Search for more papers by this author
  • Laurence Hibrand-Saint Oyant,

    1. INRA, Institut de Recherche en Horticulture et Semences (INRA, AGROCAMPUS-OUEST, Université d'Angers), SFR 4207 QUASAV, Beaucouzé Cedex, France
    Search for more papers by this author
  • Fabrice Foucher

    Corresponding author
    1. INRA, Institut de Recherche en Horticulture et Semences (INRA, AGROCAMPUS-OUEST, Université d'Angers), SFR 4207 QUASAV, Beaucouzé Cedex, France
    Search for more papers by this author

Summary

  • FT/TFL1 family members have been known to be involved in the development and flowering in plants. In rose, RoKSN, a TFL1 homologue, is a key regulator of flowering, whose absence causes continuous flowering. Our objectives are to functionally validate RoKSN and to explore its mode of action in rose.
  • We complemented Arabidopsis tfl1 mutants and ectopically expressed RoKSN in a continuous-flowering (CF) rose. Using different protein interaction techniques, we studied RoKSN interactions with RoFD and RoFT and possible competition.
  • In Arabidopsis, RoKSN complemented the tfl1 mutant by rescuing late flowering and indeterminate growth. In CF roses, the ectopic expression of RoKSN led to the absence of flowering. Different branching patterns were observed and some transgenic plants had an increased number of leaflets per leaf. In these transgenic roses, floral activator transcripts decreased. Furthermore, RoKSN was able to interact both with RoFD and the floral activator, RoFT. Protein interaction experiments revealed that RoKSN and RoFT could compete with RoFD for repression and activation of blooming, respectively.
  • We conclude that RoKSN is a floral repressor and is also involved in the vegetative development of rose. RoKSN forms a complex with RoFD and could compete with RoFT for repression of flowering.

Introduction

Flowering is a key event in the life of a plant, representing the first step in sexual reproduction. The correct timing of this event is an essential parameter in agriculture, horticulture and plant breeding. To successfully reproduce, plants sense multiple environmental and endogenous signals to decide the exact time at which to flower (Srikanth & Schmid, 2011).

In the monocarpic plant model, Arabidopsis thaliana, the transition from the vegetative to the reproductive phase is controlled by environmental (photoperiod and temperature) and internal (age, autonomous and GA) pathways. These pathways converge to regulate the floral integrator genes (FLOWERING LOCUS T (FT) and SUPRESSOR OF OVEREXPRESSION OF CONSTANS1 (SOC1)) and the floral identity genes (LEAFY (LFY), APETALA1 (AP1), CAULIFLOWER (CAL) and FRUITFULL (FUL); for a review, see Andres & Coupland (2012)). Among them, FT and its paralogue, TWIN SISTER OF FT (TSF), belong to the phosphatidylethanolamine binding protein (PEBP) family. FT and TSF are floral activators whose mutants are late-flowering (Kardailsky et al., 1999; Kobayashi et al., 1999; Yamaguchi et al., 2005). Both genes have redundant functions such as the double ft tsf mutant, which flowers later than the single ones (ft or tsf; Yamaguchi et al., 2005). The activation of FT is an essential step in the blooming process. Under long-day (LD) conditions, TSF and FT expression is up-regulated in phloem companion cells in the leaves (Jang et al., 2009). FT and TSF were identified as a component of florigen. Both proteins are produced in the leaves, moving through the phloem into the shoot apical meristem to induce flowering (Corbesier et al., 2007; Jaeger & Wigge, 2007; Notaguchi et al., 2008). In the apex, they interact with the bZIP transcription factor (FD). In turn, the FT/FD complex up-regulates the expression of floral genes such as LFY, AP1, CAL and SOC1 (Abe et al., 2005; Wigge et al., 2005; Kaufmann et al., 2010; Hanano & Goto, 2011). In rice, FT interacts with 14-3-3 proteins. The FT/14-3-3/FD complex is necessary to translocate the transcription factor into the nucleus where the complex induces expression of floral meristem identity genes (Taoka et al., 2011). Floral buds are only produced on the inflorescence meristem flanks where LFY and AP1 are expressed (Conti & Bradley, 2007).

The indeterminate growth of the inflorescence meristem is maintained by TERMINAL FLOWER 1 (TFL1), which belongs to the PEBP family. TFL1 controls flowering time and inflorescence architecture. The tfl1 mutants flower earlier and their inflorescences are converted into determinate ones with the rapid production of a terminal flower (Shannon & Meeks-Wagner, 1991; Alvarez et al., 1992; Bradley et al., 1997; Ratcliffe et al., 1999). At the vegetative stage, TFL1 mRNA is weakly accumulated in the apical rib meristem. After flower induction, TFL1 transcripts are highly accumulated (Shannon & Meeks-Wagner, 1991; Bradley et al., 1997; Ratcliffe et al., 1999). However, the TFL1 protein is present throughout the meristem, except on its flanks where floral primordia can develop (Conti & Bradley, 2007). TFL1 inhibits the expression of AP1 and LFY at the centre of the shoot apex and maintains the indeterminate growth of the inflorescence meristem (Ratcliffe et al., 1999; Parcy et al., 2002; Conti & Bradley, 2007; Hanano & Goto, 2011). Concerning flowering, TFL1 and FT are functionally antagonist, as FT is a floral activator, whereas TFL1 is a floral repressor. A single amino acid substitution, H88Y in TFL1 and Y85H in FT, converts TFL1 into a floral activator and FT into a floral repressor (Hanzawa et al., 2005). Both proteins are able to interact with FD. The FT/FD complex acts as a floral activator, whereas the TFL1/FD complex acts as a floral inhibitor by negatively regulating the FD-dependent transcription targets (Ahn et al., 2006; Hanano & Goto, 2011), suggesting that the TFL1/FT ratio might determine the transition between vegetative and floral development.

In A. thaliana, it was shown that TFL1 and FT were mainly involved in the floral process. Functions of flowering gene homologues are partially conserved between monocarpic and polycarpic species. Polycarpic plants undergo several cycles of reproduction during their lifetime. The main characteristic of polycarpics is to commit some meristems to reproductive development and to maintain vegetative growth in the remaining meristems (Battey & Tooke, 2002; Albani & Coupland, 2010; Bergonzi & Albani, 2011). In polycarpic plants, new functions of FT/TFL1 genes have been described. In Arabis alpina, a polycarpic that is a close relative of Arabidopsis, the TFL1 homologue prevents precocious flowering of young buds after vernalization, and therefore maintains vegetative buds for further development (Wang et al., 2011). In poplar, FT homologues were proposed to control juvenility, seasonal flowering and initiation of bud dormancy (Böhlenius et al., 2006; Hsu et al., 2006). In tomato, SINGLE FLOWER TRUST (SFT), the FT homologue, is involved in the determinate growth of the primary shoot and in inflorescence development (Molinero-Rosales et al., 2004). SELF PRUNING, the TFL1 homologue, is involved in the development of the sympodial unit. In the sp mutants, the number of nodes of the sympodial unit progressively diminishes until two inflorescences are formed (Pnueli et al., 1998). Both SFT and SP are able to interact with the tomato FD homologue SPGB (SELF PRUNING G-BOX PROTEIN), and a 14-3-3 protein (Pnueli et al., 2001; Lifschitz & Eshed, 2006). In tomato, the local SFT/SP ratio may control the balance of determinate/indeterminate growth in different processes such as the primary/sympodial shoot or composed leaf development (Shalit et al., 2009).

Rose is an economically important ornamental genus. Flowers represent the major value of roses, and the control of blooming is an important issue. Rose is a polycarpic plant with different modes of blooming. Once-flowering (OF) roses have a single annual flowering period, whereas continuous-flowering (CF) roses flower continuously during the year. Recently, a rose TFL1 homologue, RoKSN, was shown to be a regulator of flowering. The absence of RoKSN transcript accumulation (as a result of a mutation) causes continuous flowering. In the woodland strawberry, Fragaria vesca, the mutation of the orthologue, FvKSN, leads to continuous flowering and alters the response to photoperiod (Koskela et al., 2012). In spring, in OF roses, the floral repressor, RoKSN, is weakly accumulated during floral induction, allowing blooming with RoFT transcript accumulation (Iwata et al., 2012; Randoux et al., 2012). This weak expression of RoKSN could be the result of low GA content (Roberts et al., 1999; Remay et al., 2009). The exogenous application of GA3 leads to RoKSN transcript accumulation and, consequently, floral inhibition (Randoux et al., 2012). After the first flowering, RoKSN transcripts are highly accumulated and shoots remain vegetative until the following spring. During this period, no RoFT transcript is accumulated. In CF roses, RoKSN transcripts are not accumulated as a result of the insertion of a copia retrotransposon in RoKSN, and consequently roses flower continuously (Iwata et al., 2012). In rose, mechanisms underlying the molecular function of RoKSN remain elusive. Our aim is to provide functional evidence that RoKSN is a floral repressor and to understand how RoKSN inhibits flowering. Our results further indicate that RoKSN is a floral repressor, able to interact with RoFD. Furthermore, RoKSN may compete with RoFT to interact with RoFD, providing the basis of the molecular mechanism responsible for flowering regulation in rose.

Materials and Methods

Materials

The Arabidopsis thaliana (L.) mutant tfl1-11 (Shannon & Meeks-Wagner, 1991) was obtained from the Nottingham Arabidopsis Stock Centre (N6235). Plants were grown under LD conditions (16 : 8 h, light : dark; 20°C). Nicotiana benthamiana Domin plants were maintained in a glasshouse (20°C, 16 h light), and 5- to 7-wk-old plants were used for agroinfiltration. A CF rose, Rosa hybrida E.H.L.Krause RI, was used for rose transformation. After a period of acclimation, roses were grown in a glasshouse (20°C) under natural light conditions.

Plasmid constructions

All genes (RoKSN, RoFT and RoFD) were amplified from cDNA of Rosa wichurana using the Finnzymes Phusion High Fidelity DNA polymerase (Thermo Fisher Scientific, Waltham, MA, USA). PCR products were cloned using the Gateway® system (Invitrogen) with TOPO isomerase mix, into the pENTR™/DTOPO® entry vector. For plant transformation, RoKSN was introduced by LR Clonase recombination with the following destination vectors: pK7WG2D for RoKSN with a STOP codon, and pK7FWG2 for RoKSN without a STOP codon to generate 35S::RoKSN and 35S::RoKSN::GFP, respectively (Supporting Information, Table S1; Karimi et al., 2002, 2007). For yeast two-hybrid experiments, pLex10- and pGADT7-derived plasmids were a gift from Y. Duroc (Institut Jean Pierre Bourgin, INRA, France) and were described in Drevensek et al. (2012): pGADT7 contains the activation domain whereas pLex10 contains the binding domain. Plasmids were used in LR reactions (Gateway) with the entry vector pENTR™/DTOPO® containing RoFD, RoKSN or RoFT. The activation or binding domains were fused to the N-terminal part of the interest protein. For tobacco transient assay, each coding gene sequence was inserted into pDONR207 (Invitrogen) using Gateway cloning methods and then recombined with the plant binary vectors pB7FWG2, pB7RWG2 and pGWB14 (Karimi et al., 2007). The primers and vectors used are listed in Tables S1 and S2, respectively.

Rose regeneration

Shoot cultures of Rosa hybrida RI were established from axillary shoots collected on glasshouse-grown plants as described by Dohm et al. (2001). Once a month, shoots were periodically subcultured to a fresh shoot proliferation medium (SPM; Dohm et al., 2001). The leaves from these proliferating shoot cultures were used as explants for regeneration by somatic embryogenesis.

To obtain the somatic embryos, the protocol used was adapted from Li et al. (2002). Callus from wounded immature leaves was induced on callus induction medium (CIM; Table S3). The calli were then isolated, squeezed and transferred to embryo induction medium (EIM; Table S3). The growth and the differentiation of embryos were performed on a differentiation medium (DM; Table S3). The differentiated plantlets were transferred to liquid rooting medium (RM, Table S3). The rooting plants were then acclimated as described by Vergne et al. (2010).

Plant transformation

A binary vector was introduced by electroporation into Agrobacterium tumefasciens EHA105 (Hood et al., 1993) containing the plasmid pbbR. A. thaliana, tfl1-11 mutants were transformed using the floral dip method as described by Clough & Bent (1998). Transformed plants were selected on 1 × Murashige and Skoog medium (Murashige & Skoog, 1962) supplemented with 1% (w/v) sucrose, 0.8% (w/v) agar and 50 mg l−1 kanamycin.

For rose, the transformation was adapted from Dohm et al. (2001) and Vergne et al. (2010) with the A. tumefasciens strain EHA105 (Hood et al., 1993) containing the plasmid pbbR. A suspension of EHA105 containing the plasmid of interest (OD600, c. 0.6) in LB media supplemented with spectinomycin (50 μg l−1) and rifampicin (50 μg l−1) was centrifuged and then resuspended at OD600 c. 0.8 in MinA medium(Svab et al., 1975), pH 5.6, containing acetosyringone (100 μM). Bacteria were cultivated for 2 h at 28°C with agitation of 150 rpm to induce Vir genes.

Approximately 100 whole calli showing some somatic embryos were dried on sterile paper and wounded in the bacterial suspension with a wire brush for 10 min. They were then dried on sterile filter paper and plated on cocultivation medium, COM (Table S3), without selection for 3 d in the dark. The calli were washed several times with MS salts and a vitamin medium containing cefotaxime (500 mg l−1) to eliminate bacteria. The calli were recovered on a filter, dried on sterile filter paper and plated on a selective EIM, pH 5.5, containing cefotaxime (350 mg l−1) and kanamycin (50 mg l−1). The calli were repeatedly transferred to a fresh selective DM until the differentiation of embryos. When the plantlets were well formed, root induction and glasshouse acclimatisation were performed.

Plant scoring

The mean size of internodes was measured by scoring the number of internodes and the size of the stem. The mean number of leaflets along the stem and the percentage of branching in the group of plants and for each bud position were recorded. For each plant, between three and four stems were scored. A statistical analysis between each group of plants was carried out using a Kruskal–Wallis test in R software (http://www.R-project.org).

Gene expression analysis

For RNA isolation, samples were harvested and frozen in liquid nitrogen. Total RNA was isolated using the NucleoSpin® RNA plant kit (Macherey-Nagel, Düren, Germany) following the manufacturer's recommendations. The reverse transcription, real-time PCR experiment and analysis were carried out as described by Randoux et al. (2012). The amount of RNA was normalized using TCTP and UBC genes (Klie & Debener, 2011), and the relative expression level was calculated according to Pfaffl (2001), on the basis of three technical repetitions. Data presented are obtained from two biological replicates.

Yeast two-hybrid assays

For protein interaction assays, the L40 yeast strain was used (MATa trp1 leu2 his3 ade2 LYS2::lexA-HIS3 URA3::lexA-lacZ). Yeast transformations were carried out using LiAC methods as described by Gietz & Schiestl (2007). Yeast samples transformed with each bait construct were placed (along with the empty prey vector) on minimal medium lacking Trp, Leu and His, with increasing concentrations of 3-amino-1,2,4-triazole (up to 200 mM) to evaluate the self-activation background levels of the HIS3 gene. As no self-activation of HIS3 gene was detected, the yeast samples containing both bait and prey vectors were plated on minimal medium lacking Trp, Leu and His to study pairwise interaction. At least three independent transformations were realized and three clones per transformations were used to evaluate the protein interaction.

Transient expression

Each expression vector was introduced into A. tumefasciens VG3101 by electroporation. Agrobacterium bacterial suspensions were incubated overnight at 28°C with agitation. Each culture was centrifuged, washed and resuspended in infiltration buffer (0.01 M MgCl2, 1.95 g l−1 2-(N-morpholino)ethanesulfonic acid, pH 5.7) to an OD600 of 0.4. Before agroinfiltration, acetosyringone (200 μM) was added, and bacterial suspensions were incubated at room temperature for 30 min. The suspension was infiltrated into N. benthamiana leaves. To enhance transient expression, a culture of A. tumefasciens strain (OD600 of 0.5) containing the viral suppressor of gene silencing p19 (Voinnet et al., 2003) was coexpressed in N. benthamiana leaves. For coinfiltration experiments, equal volumes of the different tested cultures were mixed before agroinfiltration.

Coimmunoprecipitation assays

Coimmunoprecipitation assays of RoKSN, RoFT and RoFD proteins were performed on N. benthamiana agroinfiltrated leaves with the different constructions of interest. Total proteins were extracted in lysis buffer containing 50 mM Tris-HCl, pH 8, 150 mM NaCl, 1% (v/v) Triton X-100, supplemented with EDTA-free protease inhibitors (Roche) and then incubated for 2 h at 4°C with 50 μl of anti-green fluorescent protein (GFP) antibody conjugated to microbeads (μMACS GFP-tagged beads; Miltenyi Biotec, Bergish Gladbach, Germany). Beads (μMACS GFP-tagged) were retained and washed onto a magnetic column system (M columns; Miltenyi Biotec) to recover the immunoprotein complexes, according to the manufacturer's recommendations. RoKSN, RoFT and RoFD proteins bound to the beads were resolved by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) and detected by anti-GFP-horseradish peroxidase (anti-GFP-HRP) (Miltenyi Biotec) or anti-red fluorescent protein (RFP) antibody (Clontech, Madison, WI, USA) followed by a goat anti-rabbit-HRP antibody (Invitrogen) or anti-haemagglutinin (anti-HA) antibody (Sigma) followed by a goat anti-mouse-HRP (Sigma).

FLIM analyses

Fluorescence lifetime imaging microscopy (FLIM) was performed using a Nikon TE2000 microscope (Nikon, Tokyo, Japan) connected to a LiFA FLIM system (Roden, the Netherlands). Fluorescence lifetime was measured using LiFLIM software, version 1.2.8, on N. benthamiana agroinfiltrated leaves expressing RoFD-GFP (control) and coexpressing RoFD-GFP and RoKSN-RFP, RoFT-RFP, RoFD-GFP, or RoFT-RFP and RoKSN-HA proteins. At least 40 nuclei per condition were analysed and at least two biological replicates were carried out.

Southern blot

Genomic DNA was extracted from 200 mg of leaves, according to the procedure of Kobayashi et al. (1998), with modification of buffer 2: CTAB buffer (2% CTAB, w/v); 100 mM Tris-HCl, pH 8, 20 mM EDTA, pH 8, 1.4 M NaCl, 2% polyvinyl polypyrrolidone (w/v), dithiothreitol (5 mM). gDNA was treated with 1 μl RNAse (12 mg ml−1; Qiagen), for 1 h at 37°C. Genomic DNA was digested with EcoRV (Fermentas, Vilnius, Lithuania). The restricted DNA was separated on a 0.8% (w/v) agarose gel overnight and transferred onto a nitrocellulose membrane (Hybond XL; GE Healthcare, Little Chalfont, UK). A GFP probe was amplified by PCR with the primers eGFP-F1 and eGFPP-R1 (Table S2), and was labelled using the Ready to go™ DNA labelling beads (-dCTP; GE Healthcare). Probe and a hybridized membrane were incubated (65°C) overnight. The membrane was washed twice in 2 × saline sodium citrate (SSC), 0.1% SDS for 15 min, once in 1 × SSC, 0.1% SDS for 15 min and then in 0.5 × SSC, 0.1% SDS for 5 min. The membrane was exposed for 3 d and the signal was detected with Personal FX Phosphorimarger and Quantity One 1-D software (Bio-Rad, Hercules, CA, USA).

Western blot

One peptide consisting of amino acids 125–139 (KQKRRQSVNPPSSRD) of RoKSN was synthesized, conjugated to Keyhole limpet hemocyanin (KLH), and used to generate rabbit polyclonal antibody (Eurogentec, Liège, Belgium). Total proteins were extracted from a sample of 200 mg of leaves, according to the protocol used by Wang et al. (2006). After denaturation (94°C, 5 min), 20 μg of protein were separated on an SDS polyacrylamide gel (12%, w/v). Proteins were then transferred onto a polyvinylidene difluoride (PVDF) membrane (Biotrace PVDF; VWR, Radnor, PA, USA). After incubation with the RoKSN antibody (1/500), proteins were detected using a goat peroxidase conjugated anti-rabbit antibody (1/5000; Sigma) and visualised using ECL chemiluminescence (BioRad, Hercules, CA, USA).

Results

Ectopic expression of RoKSN in Arabidopsis tfl1 mutant delayed flowering and restored indeterminate growth of the inflorescence

To estimate the effects of RoKSN on flowering time and inflorescence morphology, we overexpressed RoKSN into A. thaliana tfl1-11 mutants that show early flowering and determinate growth. We obtained 12 independent transgenic lines (Table 1). Transgenic Arabidopsis were analysed under LD conditions (16 : 8 h, light : dark).

Table 1. Phenotype of Arabidopsis thaliana tfl1-11 mutants that ectopically expressed RoKSN
GenotypeTransformationNumber of rosette leavesLength of the stem (cm)Production of seedsTerminal flowerNumber of plantsClass
  1. The phenotype was observed at the T1 generation and confirmed at the T2 generation for plants that produced seeds (Table S5). The number of rosette and cauline leaves were counted. +, plants that produced terminal flowers; −, plants with indeterminate growth of the inflorescence. Plants were classified into three phenotypic classes as described in Fig. 1. WT, wild-type; NT, nontransformed plants. The number of rosette leaves and the length of the stem are shown as an average ± SD.

WT (Columbia)NT8.1 ± 1.226.9 ± 5.6+15 
tfl1_11_T1NT7.1 ± 1.68.4 ± 3.5++15 
Empty vector8.3 ± 2.19.8 ± 2.0++6 
35S::RoKSN_3 725+11
35S::RoKSN_4 938+11
35S::RoKSN_5 1232+11
35S::RoKSN_7 627+11
35S::RoKSN_8 534+11
35S::RoKSN_14 931+11
35S::RoKSN_1 1934+12
35S::RoKSN_2 2040+12
35S::RoKSN_9 2522+12
35S::RoKSN_10 > 505413
35S::RoKSN_11 > 504613
35S::RoKSN_13 363613

Plants that overexpressed RoKSN were classified into three classes according to their phenotype (see Fig. 1, Table 1). The wildtype A. thaliana Columbia produced 8.1 ± 1.2 rosette leaves and the stem measured 26.9 ± 5.6 cm, whereas the mutant tfl1-11 produced 7.1 ± 1.6 rosette leaves, with a stem of length 8.4 ± 3.4 cm. Class 1 plants behaved almost like the Col-0 Arabidopsis. They flowered later than tfl1-11 mutants and presented an indeterminate growth of the inflorescence. They had five to 12 rosette leaves and the stem measured 31 ± 4.7 cm (Table 1, Fig. 1a). The phenotypes of the class 2 plants were stronger than those of class 1. Plants flowered later, they had 19–25 rosette leaves, and the stem measured c. 33 ± 8 cm (Table 1, Fig. 1a). As in the case of class 1, the inflorescence presents an indeterminate growth. Class 3 transgenic plants have a long vegetative phase between 36 and > 50 rosette leaves, the stem measured c. 45.3 ± 9 cm, and they never produced seed (Table 1, Fig. 1b–f). In class 3 transgenic plants, leaf-like organs were produced instead of floral organs (Fig. 1d). This phenotype was similar to the lfy mutant described by Liljegren et al. (1999). Moreover, as presented in Fig. 1(e,f), some class 3 plants produced inflorescences with leaf-like organ shoots that never developed flowers, as described in the triple ap1, ful, cal mutant (Ferrandiz et al., 2000). The observed phenotypes for class 1 and 2 plants were maintained in T2 progeny for kanamycin-resistant plants.

Figure 1.

Complementation of Arabidopsis thaliana tfl1-11 mutants by RoKSN. (a) From left to right: A. thaliana Columbia (Col); tfl1-11 mutant; the three phenotypic classes (referred to as 1, 2 and 3) of tfl1-11 35S::RoKSN plants. The three classes are described in Table 1. (b–f) Detailed phenotype of class 3. (b) Left: ectopic expression of RoKSN in tfl1-11 plant class 3; right: tfl1-11 mutant. (c) Rosette leaves of the class 3 plant. (d–f) Inflorescence details of class 3 plants.

In conclusion, ectopic expression of RoKSN in A. thaliana tfl1-11 mutants restored late-flowering phenotypes and the indeterminate growth of the inflorescence. The terminal flower observed in the tfl1-11 mutant was never observed in tfl-11 35S::RoKSN plants.

Ectopic expression of RoKSN blocks flowering in CF roses

To investigate the role of RoKSN in flowering and other developmental processes in rose, we constitutively expressed RoKSN in a CF rose, RI. The ‘RI’ genotype is a tetraploid CF rose that presents only the ‘copia’ alleles at the RoKSN locus (Fig. S2). As previously demonstrated for CF roses, this plant does not accumulate RoKSN transcripts or RoKSN protein (Fig. 3; Iwata et al., 2012). The ‘RI’ genotype was transformed with 35S::RoKSN proID::GFP, 35S::RoKSN::GFP, or with the empty vector (35S::GFP). Rose transformation and regeneration are difficult, long and inefficient (Debener & Hibrand-Saint Oyant, 2009). We succeeded in regenerating and acclimating two plants transformed with 35S::RoKSN proID::GFP (D8.1 and D8.2 plants), five plants transformed with 35S::RoKSN::GFP (D6.1 to D6.5 plants) and six plants transformed with the empty vector (D7.1, D7.2, D7.3, D7.4, D7.5 and D5.1 plants).

First, we tested whether or not regenerated plants are transgenic. The presence and the number of transgenes were evaluated by Southern blot using a GFP probe. No copy of the GFP gene included in the T-DNA was detected in the RI nontransformed plant (Fig. 2). For the plants transformed with the empty vector, we have tested two plants (D7.1 and D5.1), which integrated one copy of the transgene, as only one fragment is visible on the Southern blot (Fig. 2). The two plants transformed with 35S::RoKSN proID::GFP (D8.1 and D8.2) presented the same patterns with at least seven clear bands that correspond to seven insertions of the T-DNA. According to these results, these two plants were considered as not being independent. Only plant D8.2 was used for further study. Plants transformed with the 35S::RoKSN::GFP (except D6.3) presented different patterns and had integrated one copy of the T-DNA (Fig. 2). We failed to obtain a pattern for D6.3; however, this plant was considered to be transgenic, as it accumulated transcripts and proteins (see later; Fig. 3). In conclusion, we obtained a least five independent plants that present at least one insertion of the T-DNA.

Figure 2.

Southern blot of transgenic roses that ectopically expressed RoKSN. DNA was digested with EcoRV restriction enzyme. After DNA transfer, the membrane was hybridized with a GFP probe. One band is predicted for each insertion of the transgene. Lane 1, nontransformed (NT) ‘RI’ rose; lanes 2–3, RI 35S::RoKSN 35S::GFP roses (D8) ; lanes 4–5, RI roses transformed with the empty vector (35S::GFP) ; lanes 6–10, RI 35S:RoKSN::GFP.

Figure 3.

RoKSN transcript and RoKSN protein accumulation in transgenic roses that ectopically expressed RoKSN under the 35S promoter. (a) Transcript accumulation was monitored by quantitative reverse transcription polymerase chain reaction in young expanding leaves. Asterisks indicate that no transcript was detected. The scale is not the same between D6 and D8 transformants. The error bars are ± SE. (b) RoKSN protein accumulation in young leaves, detected by Western blot using an RoKSN antibody. For 35S::RoKSN and 35S::RoKSN::GFP plants, 19.5 and 47 kDa proteins are predicted, respectively. ‘RI’ refers to the nontransformed ‘RI’ genotype. ‘L’ refers to the molecular weight marker. The size of the protein marker (in kDa) is indicated on the left side of the gel.

In order to confirm that these plants ectopically expressed RoKSN, we studied RoKSN transcript and RoKSN protein accumulation in young leaves by quantitative PCR (qPCR) and Western blot analyses, respectively. No RoKSN transcripts or RoKSN proteins were detected in the nontransformed plant (RI) or in plants transformed with the empty vector (D5.1 and D7.1-5; Fig. 3). Roses transformed with 35S::RoKSN::GFP (D6.1–D6.5) accumulated RoKSN transcripts and RoKSN proteins. The transcript accumulation is comparable among the five plants (ranging from 1 to 2.5). The D8.2 transgenic rose (35S::RoKSN proID::GFP) presents a 1000-fold RoKSN transcript accumulation when compared with the D6 plants (Fig. 3a). This high accumulation could be correlated with the high number of T-DNA insertions (Fig. 2). The roses transformed with 35S::RoKSN::GFP plasmid (D6.1 to D6.5) accumulate a protein of 47 kDa at a comparable level, corresponding to the predicted size of the fused RoKSN::GFP protein (Fig. 3b). The D8.2 transformed rose accumulated a protein of 19.5 kDa, corresponding to RoKSN protein size.

After 12–18 months of in vitro growth, all the plants were acclimated in a glasshouse. The most visible phenotype is the absence of flowering after 12 months in the glasshouse, whereas nontransformed plant or plants transformed with the empty vector produced a terminal flower. The plants that ectopically expressed RoKSN have long vegetative shoots. For detailed phenotyping, plants were pruned and new emerging shoots (around four per plant) were scored. Although nontransformed plants (RI) and plants transformed with the empty vector (D5.1 and D7.1) flowered after 70 d, roses that ectopically expressed RoKSN (D6.1–D6.5 and D8.2) did not flower after 12 months in the glasshouse (Fig. 4a).

Figure 4.

Ectopic expression of RoKSN leads to nonflowering in continuous-flowering roses. (a) Photographs of the transgenic roses maintained for 12 months under glasshouse conditions after acclimation: D6.2 and D6.4 (35S::KSN::GFP), D8.2 (35S::KSN proID::GFP) and controls (‘RI’ for nontransformed and D7.4 for transgenic roses transformed with the empty vector, 35S::GFP). (b) Branching pattern of the transgenic roses along the shoot, expressed as a percentage of bud outgrowth at the different node positions of the first-order shoot. (c) Mean internode length (cm) of the primary shoot for the different transgenic roses. (d) Mean number of leaflets per leaf. The number of shoots (n) per plant is as follows: = 2 for RI-1, D7.4 and D6.4; = 3 for RI.2, D7.2, D7.5, D5.1, D6.1, D6.5 and D8.2; and = 4 for D7.1, D7.3, D6.2 and D6.3. A Kruskal–Wallis test (< 0.05) was performed to compare the group of transgenic plants. Data with different letters (a, b or c) indicate a significant difference. The error bars are ± SE.

We further tested whether or not the nonflowering phenotype could be transmitted by grafting. D6.2, D8.2 and RI nontransformed plants were used as rootstock, and the CF ‘Old Blush’ plants were used as scions. In all of the conditions tested, the rootstock did not influence the flowering of the scions (Fig. S1a). The number of primary shoot nodes and the branching of these shoots were not modified (Fig. S1a,b). Furthermore, using Western blot, we were unable to detect the presence of RoKSN protein in ‘Old Blush’ scions, whereas the protein is detected in 35S::RoKSN rootstock (Fig. S1c).

In order to determine whether or not RoKSN ectopic expression can affect other developmental processes, we analysed the internode size, the branching pattern and the leaf development (number of leaflets per leaf). We analysed the branching pattern along the main stem. In control plants (RI, D7 and D5), flowering stems present an acrotonic branching pattern. After blooming, axillary buds of the nodes below the terminal flower rapidly developed to form flowering stems. Approximately 60–100% of the axillary buds burst (Fig. 4b). A different branching pattern was observed in 35S::RoKSN plants. The D8.2 and D6 plants presented a low percentage of branching (c. 0% for D8.2 and 20% for D6; Fig. 4b). For D8.2 plants, only basal nodes branched, whereas for D6 plants, buds in the median part of the shoot undergo outgrowth. As the shoot remains vegetative, we can speculate that an apical dominance prevents bud burst. Concerning internode size, the average length was variable within the control plants, from 2.0 to 4.2 cm (RI, D7 and D5 transgenic plants; Fig. 4c). The same variability is observed in plants that ectopically expressed RoKSN. D8.2 and D6.1 plants have a mean internode length of c. 2.2 cm, whereas D6.4 has an internode length of 3.8 cm (Fig. 4c). RoKSN overexpression does not modify internode length.

Next, we studied the complexity of the compound leaves by analysing the average number of leaflets per leaf. Control plants have compound leaves with an average of around four or five leaflets (Fig. 4d). Interestingly, all five of the D6 plants have between six and seven leaflets; this value is statistically different from control plants (Fig. 4d). The D8.2 plant has only 4.5 leaflets, just like the control plants.

Ectopic expression of RoKSN modified the transcript accumulation of floral genes

To examine whether the inhibition of flowering observed in roses that expressed RoKSN ectopically could be associated with the modification of floral gene transcript accumulation, qPCR analyses were performed on two independent transgenic plants (D8.2 and D6.2) and nontransformed plants (RI). These plants were multiplied vegetatively by cuttings. Each cutting developed one shoot. When five leaves were fully opened, the terminal part of the shoot was dissected (removal of leaves). In this tissue, the accumulation of RoFT, RoLFY, RoAP1 and RoSOC1 transcripts was significantly reduced in plants that ectopically expressed RoKSN (Fig. 5). RoFT was not detected in D8.2 and D6.2 roses. For RoAP1 and RoLFY, the repression was more pronounced in D8.2 plants than in D6.2 plants. This difference might be linked to the difference in RoKSN accumulation (Fig. 3). In contrast to the other floral genes tested, the accumulation of RoFD was up-regulated in D8.2 and D6.2 plants (Fig. 5).

Figure 5.

Transcript accumulation of floral genes was followed by quantitative PCR in shoots before the floral initiation for RoKSN, RoFT, RoFD, RoSOC1, RoAP1 and RoLFY. The transcript accumulation levels are expressed in relation to the continuous-flowering rose (Rosa hybrida E.H.L. Krause RI, black column) for each gene according to the Pfaffl ratio (Pfaffl, 2001) (base value = 1). Light grey column, D6.2 plants; dark grey column, D8.2 plants. RI-NT, D6.2 and D8.2 were multiplied vegetatively. The scale for RoKSN is different from that of the other genes. Data with different letters (a, b or c) indicate a significant difference (Kruskal–Wallis test; < 0.05). The error bars are ± SE.

RoKSN and RoFT interacted together and with RoFD

In previous studies, FT and TFL1 were reported to interact with FD in Arabidopsis (Abe et al., 2005; Hanano & Goto, 2011; Wang et al., 2011). We investigated the interactions of RoKSN and RoFT with the bZIP transcription factor RoFD. First, we demonstrated by ft mutant complementation that RoFT is a floral activator. Ectopic expression of RoFT in ft1 mutants restored an early-flowering phenotype (Table S4). Furthermore, RoFD was able to partially and weakly complement the fd1 mutant (Table S4). Ectopic expression of RoFD accelerated the flowering in fd mutants, as previously shown in Arabidopsis by Abe et al. (2005).

Using yeast two-hybrid assays, we showed an interaction of RoFD with both RoKSN and RoFT, whereas RoKSN and RoFT failed to interact together (Fig. 6a). To confirm these results, we performed FLIM analyses in transient transformed N. benthamiana cells (Fig. 6b). We showed the interaction of RoFD with RoKSN and RoFT, as the GFP lifetime significantly decreases in the presence of both proteins (1.9 and 2.08 ns, respectively; Fig. 6b). This interaction can be quantified using the fluorescence resonance energy transfer (FRET) percentage, the transfer of energy from GFP protein to RFP protein after excitation. The FRET percentages between RoFD/RoKSN and RoFD/RoFT were 22.93 and 15.92%, respectively (Fig. 6b). The strong interaction between RoFT and RoFD was in agreement with previous reports in Arabidopsis (Abe et al., 2005; Hanano & Goto, 2011; Wang et al., 2011). Moreover, coimmunoprecipitation studies further confirmed an in vivo interaction between RoKSN and RoFD and between RoFT and RoFD (Fig. 6c).

Figure 6.

RoKSN and RoFT interact with RoFD: (a) RoKSN and RoFT interact with RoFD in the yeast two-hybrid system. Transformed yeast cells were grown on selective media. The X-gal test was performed to characterize the activity of the LACZ gene (blue). AD, activating domain; BD, binding domain; SD, synthetic dextrose minimal medium. Vectors are described in Table S1. (b) Fluorescence lifetime analyses (in ns) of RoFD-GFP with different vector combinations of RoKSN-GFP in combination with RoFT-RFP and mean fluorescence resonance energy transfer (FRET) value (%) in Nicotiana benthamiana Domin agroinfiltrated leaves. (c) Coimmunoprecipitation studies. Total protein extract from N. benthamiana Domin agroinfiltrated leaves with the different vector combinations were immunoprecipitated with anti-green fluorescent protein (GFP) antibody. The coimmunoprecipitated proteins were detected by anti-red fluorescent protein (RFP) antibody or anti-haemagglutinin (HA) antibody. The experiment was repeated twice with several technical replicates. CoIP, coimmunoprecipitated proteins; IP, immunoprecipitated proteins: nt, not tested; input, total protein extract.

Interestingly, in contrast to results obtained by the yeast two-hybrid assays (Fig. 6a), FLIM analysis allowed us to detect a direct interaction between RoFT and RoKSN, as the GFP lifetime significantly decreased. The FRET percentage between RoKSN and RoFT was 17.77% (Fig. 6b). This interaction was further validated by immunoprecipitation (Fig. 6c).

We tested the hypothesis of competition between RoFT and RoKSN for the interaction with RoFD. To demonstrate this competition, we performed FLIM analysis on nuclei of transiently transformed N. benthamiana cells with RoFD-GFP, RoFT-RFP and RoKSN-HA protein fusion. We tested different concentrations of A. tumefasciens (DO600 nm = 0.4 or 0.6) to increase the accumulation of RoKSN-HA (Fig. 6c). Using FLIM analysis, we demonstrated that RoFT was able to interact with RoFD in the absence of RoKSN, with FRET percentages of 15.92% between RoFT and RoFD, and 11.91 and 9.92% in the presence of RoKSN (DO = 0.4 and 0.6, respectively; Fig. 6b). We verified by Western blot that proteins are accumulated under these conditions (Fig. 6c). This decrease of FRET suggests a competition between RoFT and RoKSN for the interaction with RoFD (see the 'Plasmid constructions' section).

Discussion

RoKSN is a floral repressor

A previous study demonstrated that continuous flowering in rose is associated with RoKSN, a TFL1 homologue (Iwata et al., 2012). We conducted a functional analysis of RoKSN by overexpression approaches in A. thaliana tfl1 mutant and in CF rose.

In Arabidopsis, ectopic expression of RoKSN in tfl1-11 mutant was sufficient to restore the indeterminate growth of the inflorescence and the late flowering (Fig. 1, Table 1). These results suggest that RoKSN acts as a floral repressor and controls indeterminate growth of the inflorescence meristem in transgenic Arabidopsis. TFL1 function is conserved in the heterologous Arabidopsis system. TFL1 homologues were shown to be able to complement the tfl1 mutant, as is the case for woodland strawberry (Koskela et al., 2012), apple (Mimida et al., 2001; Kotoda & Wada, 2005), grapevine (Carmona et al., 2007) and rice (Nakagawa et al., 2002).

We tested the ectopic expression of RoKSN in CF roses. CF roses did not accumulate RoKSN transcripts because of a retrotransposon insertion at the RoKSN locus (Iwata et al., 2012). We obtained at least five independent lines that accumulated RoKSN transcripts and RoKSN protein (Figs 2, 3). These lines never flowered after 18 months; the terminal meristem remained vegetative. The control plants flowered rapidly (Fig. 4a). This strong repression of flowering by RoKSN was consistent with the role of TFL1-like genes as a floral repressor. However, in other plants, overexpression of the TFL1 homologue only delayed flowering, as is the case for Arabidopsis (Ratcliffe et al., 1998), rice (Nakagawa et al., 2002), Brachypodium distachyon (Olsen et al., 2006), tobacco (Amaya et al., 1999), tomato (Pnueli et al., 1998) and kiwifruit (Varkonyi-Gasic et al., 2013). In poplar, most of the trees did not flower, and only rare blooming could be observed (Mohamed et al., 2010). In rose and woodland strawberry, ectopic expression completely blocked flowering (this study and Koskela et al. (2012)). These results suggest that TFL1 acts as a strong floral inhibitor in these species whose ectopic expression completely blocked flowering. However, we cannot completely exclude the possibility that ectopic expression of RoKSN in rose extends the juvenile phase and that transgenic roses will flower later. In apple and pear, the extinction of the TFL1 homologue reduces the juvenility period and leads to precocious flowering (Flachowsky et al., 2012; Freiman et al., 2012). In rose, the complete functional validation of RoKSN can be obtained by down-regulation of RoKSN in once-flowering roses by RNAi. However, the genetic transformation of OF roses is not possible at this time and protocols have only been developed on CF roses (Debener & Hibrand-Saint Oyant, 2009).

Is RoKSN a floral mobile signal?

Grafting experiments represent a simple but powerful method for establishing the transmission of a floral signal through the phloem. As demonstrated in tomato, kiwifruit and Cucurbita moschata, FT acts as a mobile signal to induce flowering in the scions placed in noninductive conditions (Lifschitz & Eshed, 2006; Lin et al., 2007; Varkonyi-Gasic et al., 2013). We tested whether or not the floral repression driven by RoKSN can be mobile and transmitted by grafting. Using different graft combinations, we were unable to demonstrate a transmission of the signal (Fig. S1). Similarly, in apple, TFL1 RNAi and 35S::FT plants flower earlier; however, the floral stimulus is not graft-transmissible (Trankner et al., 2010; Flachowsky et al., 2012). In Arabidopsis, TFL1 was assumed to move only locally within the apical meristem (Conti & Bradley, 2007). In contrast to FT, TFL1 seems not to be a systemic antiflorigen mobile signal.

RoKSN affects other developmental processes

As constitutive overexpression of TFL1/FT homologues from different species has been shown to cause pleiotropic effects, we investigated the vegetative growth in transgenic roses. Overexpression of RoKSN in rose modified leaf development with more complex compound leaves (Fig. 4). In tomato, mutation of SFT, the homologue of FT, led to an increase in the number of leaflets per leaf, whereas overexpression of SFT led to a decrease (Lifschitz & Eshed, 2006; Lifschitz et al., 2006; Shalit et al., 2009). It was hypothesized that the local ratio of SFT/SP controls the balance of determinate/indeterminate growth, floral transition, primary and sympodial unit formation and leaf complexity (Shalit et al., 2009). Overexpression of RoKSN in rose could lead to a small ratio of FT/TFL1 that induces indeterminate growth and increases the number of leaflets per leaf (D6 plant; Fig. 4).

Furthermore, branching patterns along the stem were also modified in RoKSN transgenic roses: less axillary buds had undergone bud outgrowth in transgenic plants, and the position of the growing buds was different between the transgenic plants (Fig. 4). This new branching pattern can be explained by the nonflowering behaviour of the shoots and the apical dominance exerted. In nontransgenic plants, the formation of a terminal flower released apical dominance, and nodes, below the inflorescence, develop. However, we cannot exclude a direct action of RoKSN in the bud. In poplar, for example, overexpression of TFL1 also reduced secondary branching (Mohamed et al., 2010). Furthermore, in Arabidopsis, it was recently proposed that the PEBP gene is involved in branching control as a result of the interaction between FT and BRANCHED 1 (Hiraoka et al., 2012; Niwa et al., 2013).

How does RoKSN act as a floral repressor?

As demonstrated in different species, TFL1 and FT play an opposite role in flowering. Interaction of FT and TFL1 with a common partner, FD, appears to be a conserved mechanism (Pnueli et al., 2001; Abe et al., 2005; Wigge et al., 2005; Danilevskaya et al., 2010; Varkonyi-Gasic et al., 2013). In A. thaliana, the TFL1/FD complex acts as a transcriptional repressor complex, whereas the FT/FD complex acts as a transcriptional activator (Hanano & Goto, 2011). In this article, we provide evidence that RoKSN acts at the transcriptional level by repressing floral genes. We demonstrated that overexpression of RoKSN in transgenic roses led to the repression of floral activators such as RoFT, RoSOC1, RoAP1 and RoLFY (Fig. 5). The same repression was observed when RoKSN was accumulated after a GA3 treatment in once-flowering rose, where GA3 treatment induced the accumulation of RoKSN and inhibited flowering via floral gene repression (Randoux et al., 2012). Owing to the accumulation of RoKSN, FT and downstream targets of FT (such as SOC1, AP1 and LFY) were not induced and flowering therefore did not occur (Randoux et al., 2012). In our two systems, GA3 exogenous application or RoKSN ectopic expression, the accumulation of RoKSN led to a repression of FT, SOC1, AP1 and LFY. In orchid, overexpression of OnTFL1 in Arabidopsis led to the repression of AP1 (Hou et al., 2009). In Arabidopsis, overexpression of TFL1 led to a transcriptional repression of LFY and AP1 (Ratcliffe et al., 1998, 1999). AP1 and TFL1 expression were antagonistic. AP1 repressed TFL1 by directly binding with TFL1 regulatory elements (Kaufmann et al., 2010). Recently, a model of regulation of TFL1 by FT was proposed: FT up-regulates TFL1 to maintain the vegetative centre of the apical meristem (Jaeger et al., 2013). Our results suggest a regulation of RoFT accumulation by RoKSN, as ectopic expression of RoKSN blocked RoFT transcript accumulation. Furthermore, our results suggest a different regulation between Fragaria vesca and rose. In F. vesca, FvSOC1 is a positive regulator of FvTFL1, the orthologue of RoKSN (Mouhu et al., 2013), whereas in rose RoKSN may negatively regulate RoSOC1, as RoKSN ectopic expression leads to RoSOC1 repression (Fig. 5).

Furthermore, RoKSN acts at the transcriptional level by interacting with RoFD. A binding competition between RoKSN and RoFT might exist. Indeed, both proteins, RoFT and RoKSN, are able to interact strongly with RoFD (FRET value > 10%, Fig. 6b). In the presence of RoKSN, the interaction between RoFT and RoFD is weaker (a significant decrease in the FRET value; Fig. 6b). This evidence suggests a binding competition for the interaction with RoFD. However, we still need to validate the biological significance of this competition at the molecular level (effect on floral gene identity induction) or at the plant level (induction or inhibition of flowering).

Using coimmunoprecipitation and FLIM analysis, we were able to detect an interaction between RoKSN and RoFT, not initially detectable in the yeast two-hybrid system (Fig. 6). Interestingly, this interaction was not previously reported in other plants such as Arabidopsis or rice. The interaction was obtained in tobacco leaves, a heterologous system. A hypothesis for this interaction is a possible indirect interaction via 14-3-3 and FD proteins. 14-3-3 and FD homologues might be expressed in tobacco leaves. In tomato, a Solenaceae-like tobacco, FD and 14-3-3 homologues are expressed in the leaves (Pnueli et al., 2001). In rice, it was demonstrated that 14-3-3, FD and FT homologue proteins form a hexameric structure, referred to as the ‘florigen activation complex’ (Taoka et al., 2011). We can speculate that this hexameric complex may contain FT and TFL1 homologues. The interaction shown in this study could be explained by such a scenario. If this is the case, it would suggest multiple combinations that could lead to flowering activation or repression complexes with possible competition between FT and TFL1. This hypothesis needs to be validated by further experiments, such as yeast-three hybrid experiment with coexpression of RoFD.

In conclusion, we demonstrated that RoKSN is a floral inhibitor. Its ectopic expression completely blocks flowering in CF rose. RoKSN can interact with FD and FT, suggesting complex interactions leading to flowering activation or repression complexes.

Acknowledgements

We thank G. Michel for plant maintenance in the confined glasshouse; Société Nouvelle des Pépinières et Roseraies Georges Delbard for providing the plant material; P. Satour and R. Cournol for their help with protein extraction and Western blot analysis; J. Mutterer from the IBMP of Strasbourg for his help with the FLIM/FRET experiment; M. Banfield from JIC of Norwich for his help with peptide design for antibody production; and G. Wagman for correcting the English. This work was supported by the French Ministry of Agriculture and Fisheries (project CTPS C 07-08 – INRA – section Ornementales). The research of M.R. was supported by a grant from Angers Loire Métropole (France). This work was supported by the program Investments for the Future (grant ANR-11-BTBR-0006-GENIUS) managed by the French National Research Agency.

Ancillary