• In Arabidopsis thaliana, the ability to flower is mainly related to a floral repressor, FLOWERING LOCUS C (FLC), which is regulated through the vernalization pathway. The genes controlling the vernalization pathway seem to be only partially conserved in dicots other than the Brassicaceae. Cichorium intybus (chicory) is a biennial species belonging to the Asteraceae family, and it shows an obligate vernalization requirement for flowering.
• Cichorium intybus MADS (MCM1, Agamous, Deficiens, SRF) FLC-like (CiMFL) sequences were isolated in C. intybus by RT-PCR and their expression patterns characterized during plant development and in response to vernalization. The biological function of CiMFL was analysed by complementation of A. thaliana FRIGIDA (AtFRI);flc3. Resetting of MFL expression after vernalization was analysed during microsporogenesis.
• Before vernalization, CiMFL is mainly expressed in the axils of young leaves. Vernalization induced CiMFL down-regulation under a long-day photoperiod but not under a short-day photoperiod. Furthermore, together with a decrease in CiMFL transcripts, cold conditions induced changes in the morphology of the shoot apical meristem and in the transition to flowering. The biological function of CiMFL was found not to be conserved.
• Our results show that the regulation of CiMFL expression in time and space and in relation to environmental conditions is only partially conserved with respect to FLC isolated from A. thaliana. A model for flowering repression by CiMFL is proposed.
Plants undergo several developmental transitions during their life cycle, the most important of which is the transition from the juvenile phase to the adult vegetative phase, in which they are able to respond to floral inductive signals.
The winter annual behaviour of some ecotypes of A. thaliana is conferred by the expression of the MADS (MCM1, Agamous, Deficiens, SRF)-box transcription factor gene FLOWERING LOCUS C (FLC) during the first growing season (Michaels & Amasino, 1999).
FLC acts as the main flowering repressor. It down-regulates the expression of systemic flowering signals in the leaf (e.g. Flowering Locus T (FT)) and in response to these signals represses other genes at the meristem (e.g. Suppressor of Constans1 (SOC1) and Flowering Locus D (FD)). Vernalization represses FLC, making the meristem responsive to the flowering signals and allowing the leaves to produce these signals (Lee et al., 2000; Sheldon et al., 2000; Searle et al., 2006). Following a cold treatment, the repression of FLC is maintained by an epigenetic mechanism that involves histone modification in FLC chromatin (Sung & Amasino, 2004; Dennis & Peacock, 2007; Schmitz et al., 2008). In contrast, its expression is enhanced by the gene FRIGIDA (FRI) (Clarke & Dean, 1994).
FLC-like sequences have not been extensively characterized in taxa other than the Brassicaceae family. There are a few examples of the identification of FLC-like genes in species such as Beta vulgaris, Vitis vinifera, Solanum tuberosum and Lycopersicon esculentum, but their functions have not yet been fully characterized (Reeves et al., 2007).
Cichorium intybus (chicory) is an important leaf vegetable belonging to the Asteraceae family. In the wild, it is a perennial species, while the cultivated types behave as strict biennials. In this species, the requirement of a long-day photoperiod and a period of cold seems to be absolute in order to enable the plant to flower. Sensitivity to low temperature increases with plant age, so that chicory plants gain the competence to respond to vernalization treatment only after the third true leaf has unfolded (Pimpini & Gianquinto, 1988).
The genetic control of flowering induction and differentiation in this specie is unknown. Several studies reported the effects of environmental conditions on bolting and flowering, but this information mainly came from in vitro cultivation of root tissues (Badila et al., 1985; Demeulemeester et al., 1995).
In this study we used a molecular approach to isolate MADS FLC-LIKE (CiMFL) sequences in C. intybus to obtain a better understanding of the genetic mechanism governing flowering induction. Given that AtFLC is the main repressor of flowering in A. thaliana and is regulated by vernalization, we were interested in characterizing its homologue in chicory. Our analyses were therefore mainly focused on the effects of vernalization and photoperiod on the regulation of CiMFL expression.
In comparing AtFLC and CiMFL regulation following vernalization treatments, this work provides new insights into the molecular programme controlling flowering in chicory and indicates that CiMFL is a homologue of FLC with a diverged function.
Materials and Methods
Plant growth conditions, vernalization treatment and flowering induction determination
Plants of wild chicory (Cichorium intybus L. (FB)) and cv. Treviso (TVT) were grown in 16-h long days (16h-LDs), under cool white fluorescence light, at 22°C. Vernalization, at 4°C and under a photoperiod of 16h-LDs or 8-h short days (8h-SDs), started when plants had developed the third true leaf. Vernalization treatments were conducted for a minimum of 7 d and a maximum of 40 d, respectively, in LD and SD conditions. After vernalization, in order to stabilize the cold treatment, the plants were transferred to a growth chamber maintained at 16°C with a 16h-LD photoperiod for 15 d. Subsequently they were moved to a growth chamber maintained at 22°C (also with a 16h-LD photoperiod) until flowering occurred.
To determine flowering induction, embedded chicory apexes were cut as described in the section ‘In situ hybridization and cytological observations’ below, and SAM sections were observed under the microscope.
RNA extraction, cDNA synthesis and sequence analysis
Total RNA was extracted from leaves, apical meristems and flowers of chicory FB and TVT, using the RNeasy Plant Mini Kit (Qiagen) according to the manufacturer's instructions. DNA-free RNA was obtained by DNase treatment (RNAse free DNase; Qiagen). First-strand cDNA was synthesized using SuperScript® II Reverse Transcriptase (Invitrogen) and oligo(dT)12–18.
CiMFL cDNA was amplified by reverse transcription–PCR (RT-PCR) using degenerated primers designed based on the conserved regions obtained by aligning FLC sequences from Arabidopsis thaliana (National Center for Biotechnology Information (NCBI) GenBank accession no. NM_001085094) and Brassica napus (NCBI GenBank accession no. AY036888); the forward primer was 5′-RNG GTCTCGTTGAGAAAGCTCG-3′ and the reverse primer was 5′-TATTATCAGCTT CGGCTCCCGYA-3′. PCR amplifications of the cDNA were performed with Bio-X-Act Short DNA polymerase (Bioline, London, UK) with the following cycles: 94°C for 2 min, and 30 repetitions of 94°C for 10 s, 68°C for 1 min and 70°C for 2 min. Amplicons were cloned into the pGEM-Teasy vector (Promega) and sequenced (GenBank accession numbers FJ347969, FJ347970, FJ347971, FJ347972, FJ347973 and FJ347974). The full-length sequences were obtained by Rapid Amplification of cDNA end (RACE) reaction (Roche) using the CiMFL-specific primer 5′-CTCCGTGACTAGAGCCAAGAAGACCGAA-3′ to amplify the 3′-end. Sequences were edited and aligned using the software CLC Sequence Viewer 4 (CLC, Katrinebjerg, Denmark) and Lasergene DNAStar (Madison, WI, USA). Amino acid sequences were deduced using the software available online at http://www.expasy.org/tools/dna.html. A search for homology was performed using the BLASTn algorithm, with which the GenBank/European Molecular Biology Laboratory databases were queried. Phylogenetic relationships were deduced using the CLC sequence viewer workbench (http://www.clcbio.com), with application of the neighbour-joining algorithm. Values were estimated from 500 bootstrap replicates.
Southern blot hybridization
Genomic DNA was extracted from young leaves of chicory according to the Hexadecyltrimethyl-ammonium bromide (CTAB) protocols (Doyle & Doyle, 1987). During the DNA extraction, RNase (10 ng µl−1; Sigma-Aldrich) was added. Up to 10 µg of DNA extracted from leaves of different chicory cultivars (Treviso, Castelfranco, Chioggia, Lusia Adige and Verona) was digested with EcoRI overnight at 37°C, separated on a 0.8% (w/v) agarose gel and blotted onto a positively charged nylon membrane (Hy-bond N+; Amersham). In order to specifically identify the MADS-box transcription factor gene CiMFL, a CiMFL probe was synthesized without the MADS-domain conserved region and labelled with dUTP-DIG (Roche). The MADS domain was removed from the CiMFL Coding Sequence (CDS) through amplification with the primer combination 5′-ACAGCTTCTCCTCCGGCGATAA-3′ and 5′-ATCTGGCTAGCCAAAACCTGGTTC-3′. The amplicon (423 bp) was used as a probe for hybridization. Blotted DNAs were hybridized overnight at 60°C. Filter washing and detection of anti-DIG antibody linked to alkaline phosphatase (Roche) with CDP Star (Roche) were performed according to the manufacturer's instructions.
A second Southern blotting experiments was planned to detect changes in CiMFL DNA methylation after vernalization. Each DNA sample of TVT chicory was digested overnight with one of two enzymes manifesting different degrees of sensitivity to cytosine methylation, in any of the cases in which methylation falls in N-CG-N (were N can be any nucleotide) and CNG sequences, Sau3AI and NdeII (Promega). DNAs were hybridized as previously described above using a DNA genomic probe of 438 bp, spanning the region from nucleotide 117 to 261 in the MADS domain, taking CiMFL2 sequence as a reference for the nucleotide positions, plus 293 bp of intronic sequence.
In situ hybridization and cytological observations
Plant material (shoot apexes and pollinated and unpollinated flowers) was collected and fixed in 4% paraformaldehyde (Sigma) in 0.1 M phosphate buffer, pH 7.2, and incubated for 16 h at 4°C. Tissues were dehydrated by washing and incubation in a mixture of increasing ethanol-xylene concentration, according to Varotto et al. (2003). Dehydrated tissues were embedded in paraffin (Paraplast Plus; Sigma) and cut into 6–10-µm-thick sections using a rotary microtome (RM 2135 Leica). Sections were collected on SuperFrost Plus Slides (Menzel-Glazer, Braunschweig, Germany) and were de-waxed and treated with 10 µg ml−1 proteinase K (Sigma). Sense and antisense riboprobes were obtained by in vitro transcription using T7 and SP6 RNA polymerases (Roche) and labelled with Digoxygenin RNA labelling mix (Roche). The CiMFL probes were synthesized from a cDNA sequence lacking the MADS domain, as described for the Southern blot hybridization. Hybridization was conducted overnight in 50% formamide at 48°C. DIG detection and signal visualization were carried out with NBT-BCiP (Roche) following the manufacturer's instructions. After staining, slides were dried and mounted with DPX Mountant for histology (Sigma-Aldrich).
For cytological observations, shoot apexes were collected, fixed in 4% paraformaldehyde (Sigma-Aldrich) and embedded in Paraplast (Sigma-Aldrich) as described for the in situ hybridization procedure. All vernalized samples were collected 4 wk after moving the plants to 24°C under LDs. Sections of 6 µm for each sample were produced using a rotary microtome and collected on SuperFrost slides. After de-waxing and drying, slides were mounted with VectaShield® Mounting Medium containing 4-6-diamidino-2-phenylindole (DAPI) (Vector Laboratories, Burlingame, CA, USA).
Slides were observed under epifluorescent light or bright field with a Leica DM4000B microscope. Images were captured with a Leica DC300F camera and processed with Adobe Photoshop 6.0.
Chicory MFL::GFP expression construct
CiMFL2 and CiMFL3 coding sequences (CDS) were amplified by RT-PCR using the forward primer 5′-AGATCTATGATTGTAATACGACTCACT-3′ and the reverse primer 5′-GGTACCTCTAGCTATGCATCCAACGCGT-3′. Amplicons were then separated in 1.5% (w/v) agarose and gel-purified. BglII and KpnI restriction sites (in bold characters in the primer sequences) were incorporated at their 5′-ends and the stop codon was removed from the CDS through point mutations inserted in the reverse primer. PCR products were digested with BglII and KpnI, and cloned into the corresponding sites of the vector pTZ-19U (Stratagene, LaJolla, CA, USA), in place of the β pre-sequence originally present in the vector. In the construct, the green fluorescent protein (GFP) cDNA was framed by a strong promoter (EN50PMA4) and a nos terminator.
Chicory protoplast transformation
Protoplasts were isolated as in Varotto et al. (2001). Plantlets of chicory cv. Treviso TVT were maintained on B5 medium under controlled environmental conditions, in a growth chamber at 22°C with a 12-h day:night photoperiod. Young leaves (4–5 cm in length) were cut into very small strips and placed in a filtered sterilized enzyme solution containing 0.1% (w/v) Cellulase Onozuka R10, 0.05% (w/v) Driselase and 0.02% (w/v) Macerozyme (all from Duchefa, Haarlem, The Netherlands) in WS9M medium (27 mg l−1 KH2PO4, 1.48 g l−1 CaCl2⋅2H2O, 100 mg l−1 KNO3, 250 mg l−1 MgSO4 and 90 g l−1 mannitol, pH 5.6). Strips were incubated overnight in this enzyme solution, in the dark at 28°C. Gentle shaking was maintained.
The protoplast suspension was then filtered with 200-µm nylon mesh (Sigma-Aldrich) and centrifuged at 130 g for 10 min. The pellet was re-suspended in FS13S buffer (27 mg l−1 KH2PO4, 1.48 g l−1 CaCl2⋅2H2O, 100 mg l−1 KNO3, 250 mg l−1 MgSO4 and 130 g l−1 saccharose) and centrifuged again (100 g for 10 min). The pellet was then washed in WS9M medium as described above and re-suspended in calcium-mannitol buffer (MaCa3) (0.5 M mannitol, 20 mM CaCl2 and 0.1% MES, pH 5.8, with KOH). After centrifugation at 100 g for 5–10 min the protoplasts were re-suspended in MaCa3 buffer, and 10–20 µg of the pTZ-19U vector carrying the CiMFL2::GFP or CiMFL3::GFP cassette was added. After 5 min of incubation at room temperature, a fresh solution of Polyethylene Glicol (PEG) 40% was added. Transformed protoplasts were centrifuged at 100 g for 5 min and the pellet was re-suspended in WS9M medium. At this point, protoplasts were cultured in the dark, overnight (18–20 h) at room temperature for the expression of the fusion protein.
Semiquantitative RT-PCR for CiMFL mRNA quantification
For quantification of CiMFL transcripts, cDNA was prepared from young leaves. Leaves were collected from 10 plants for each treatment (LD or SD) in the growth chamber at either 4 or 24°C. Serial dilutions of the concentrated first-strand cDNA reaction were prepared.
A gene-specific primer combination (5′-CAGTCACGACGTTGTAAAACGACGGC-3′ with 5′-CGCCAAGCTATTTAGGTGACACT-3′) was used to amplify four CiMFL sequences at the same time, while the primer combination 5′-CAGTCACGACGTTGTAAAACGACGGC-3′ with 5′-GAGCGACGGATGCGTCACAGAGAACAGAA-3′ was used for CiMFL2-specific amplification.
PCR conditions were controlled by 18S amplification with the 18S primers forward 5′-GGAGCCATCCCTCCGTAGTTAGCTTCTT-3′ and reverse 5′-CCTGTCGGCCAAGGCTATATACTCGTTG-3′.
All the PCRs were performed with Bio-X-Act Short DNA polymerase (Bioline), with the following cycles: 94°C for 2 min, followed by either 26 cycles (for 18S) or 30 cycles (for FLC) of 94°C for 10 s, 58°C for 12 s and 72°C for 45 s, in a PerkinElmer 9600 thermocycler (PerkinElmer, Waltham, MA, USA). Amplified fragments were separated on a 1.5% (w/v) agarose gel.
For the analysis of CiMFL2 expression during the day, leaf samples were collected throughout the day every 4 h, starting at 07:00 h (time 0). The cDNA was diluted and each dilution was amplified with selective primers for CiMFL2 using the above cycling conditions. Images were captured using Kodak Molecular Imaging Software.
Production, selection and analysis of transgenic A. thaliana plants
The CiMFL2 sequence, including its 3′ untranslated region (UTR) (703 bp), was cloned into the pENTRTM/D-TOPO vector (Invitrogen) to enable use of the GatewayTM recombination system (Invitrogen) for insertion of CiMFL2 into the expression vector pMDC32, which carried the 2X35S promoter and nos terminator (ABRC; The Ohio State University). The construct was introduced into Escherichia coli strain TOP10 (Invitrogen) for sequencing. Agrobacterium tumefaciens strain C58i_pMD990 (kindly provided by R. Amasino's laboratory) was subsequently transformed with the over-expression construct through electroporation. An A. thaliana FLC null mutant in the FRI background, flc-3 (Redei, 1962; Lee et al., 1993; Michaels & Amasino, 1999), was transformed with the selected A. tumefaciens using the floral dip method (Clough & Bent, 1998). Transformants were selected by supplying 25 mg l−1 hygromycin on Murashige and Skoog (MS; Duchefa) medium. Plants were grown in a growth chamber under long-day conditions at 24°C. Hygromycin-resistant plants were transferred to soil and grown for analysis of flowering time and phenotype. The presence of the transgene in the A. thaliana genome was verified by PCR, using primers annealing to the 35S promoter and the nos terminator sequences flanking the CiMFL2 CDS in the cassette (forward primer 5′-CTATCCTTCGCAAGACCCTTCCTCT-3′ and reverse primer 5′-AATCAT CGCAAGACCGGCAACAGGATTC-3′).
Time to flowering was measured in T1 plants as the total leaf number in the rosette at flowering, when the inflorescence stem reached approximately 1–3 cm in length.
Identification of CiMFL cDNAs
Chicory FLC-like sequences were identified by PCR-based cloning using primers designed on the basis of the FLC sequences of A. thaliana and B. napus available in GenBank, and named CiMFLs. Specifically, degenerated primers were designed based on the MADS-box conserved region and on the CDS 3′-end. CiMFL transcripts were amplified by RT-PCR in a homozygous wild accession of chicory (FB line) which showed perennial behaviour. Subsequently, we isolated CiMFL sequences in the self-incompatible heterozygous cultivar Treviso (TVT), which showed biennial behaviour and required vernalization to flower.
Three partial cDNA sequences homologous to AtFLC were identified in the FB wild chicory line and were designated CiMFLa, CiMFLb and CiMFLc; the lengths of the cDNA fragments were as follows: CiMFLa, 503 bp; CiMFLb, 236 bp; and CiMFLc, 182 bp (GenBank accession numbers FJ347972, FJ347973 and FJ347974). Sequence analysis led to the identification of a putative open reading frame (ORF) for CiMFLa, while premature stop codons were retrieved inside the CDS of CiMFLb at positions 212 and 217, and in CiMFLc at positions 109 and 124 (data not shown).
Four CiMFL sequences were isolated in TVT chicory. Because the 5′-CDS was already covered, gene-specific primers were designed to obtain the missing 3′-end of the four amplicons by 3′ RACE reactions. Full-length cDNA sequences were gel-purified and named CiMFL1, CiMFL2, CiMFL3 and CiMFL4 (GenBank accession numbers FJ347968, FJ347969, FJ347970 and FJ347971). Details of these sequences are given in Fig. 1.
A nucleotide comparison between the partial CiMFL sequences cloned from wild chicory and the full-length sequences isolated from the cultivated accession TVT showed that two of the transcripts were shared between the FB line and TVT. Specifically, the deduced amino acid sequence of MFLa matched that of MFL2 (91% identity), while the sequence of MFLb matched that of MFL4 (100% identity). MFL1 and MFL3 seem to be unique to TVT, while CiMFLc appears to be characteristic of the wild accession. Interestingly, CiMFL2 possesses a distinctive insertion of 14 amino acids into the MADS domain when compared with FLC proteins from other species available in the GenBank database (Fig. 2a).
Phylogenetic analysis of FLC and CiMFL2 sequences was carried out using the aligned deduced amino acid sequences, both considering the whole coding sequence and considering only the MADS domain (Fig. 2b,c). MADS Affecting Flowering (MAF) sequences were inserted into this alignment as an out-group. MAF genes constitute a group of transcriptional factors that are very similar to FLC genes, but are classified as a separate group supported by a high bootstrap value (Tadege et al., 2001). Phylogeny construction using the whole FLC coding sequence showed that CiMFL2 is grouped together with A. thaliana FLC, supported by a high bootstrap value of 491 out of 500, and is separated from the MAF genes (Fig. 2b). However, the neighbour-joining tree created by aligning only the MADS-box domain revealed that CiMFL2 is neither in the cluster of the MAF family nor in the group including AtFLC (Fig. 2c). These results support the view that the MADS-domain sequence in CiMFL2 with the additional amino acids in its first trait of sequence (Fig. 2a) represents a peculiar characteristic of the gene that could explain the evolution of a diverged function in spite of its high sequence homology with AtFLC.
The presence of MFL in the genome of chicory was confirmed by Southern blot analysis. Genomic DNA from different cultivars of chicory and from A. thaliana was digested and hybridized with a CiMFL2 cDNA probe lacking the MADS domain. Three major hybridizing fragments were detected (Supporting Information Fig. S1).
A second Southern blot hybridization was performed in which the TVT chicory genome was hybridized with a 439-bp DNA probe corresponding to a fragment of a genomic sequence comprising an intron and isolated by genome walking. At least four hybridization signals were observed.
Taken together, our results (data not shown) support the conclusion that more copies of CiMFL are present in the genome of C. intybus.
Effect of vernalization and photoperiod on CiMFL down-regulation
To determine whether chicory MFLs were down-regulated by vernalization, the pattern of CiMFL expression in response to vernalization was analysed using semiquantitative RT-PCR. Seedlings of TVT chicory were vernalized at 4°C under both SD and LD photoperiods for 7, 15 and 30 d. As shown in Fig. 3, under LD conditions the variants CiMFL2, CiMFL3 and CiMFL4 were down-regulated after 7 d of treatment; at the same time, CiMFL1 maintained a constitutive level of expression. In addition, we observed that, while 7 d of treatment in LDs was enough to trigger down-regulation of CiMFLs, vernalization had no effects on CiMFL expression under the SD photoperiod and particularly did not have an effect on CiMFL2 expression (Fig. 3, left side).
Given that vernalization under the SD photoperiod did not promote CiMFL2 down-regulation, we tested the possibility that its regulation might be dependent on the photoperiod rather than on vernalization. To explore this hypothesis, we monitored CiMFL2 expression by semiquantitative RT-PCR and found that, without a vernalization treatment, the LD photoperiod is not sufficient to induce CiMFL2 down-regulation (Fig. 3, bottom).
The response of CiMFL2 towards vernalization, strictly depends upon daylength, which stimulated our interest in the effect of this environmental factor on CiMFL2 regulation. To investigate the possibility of a regulatory effect of photoperiod on CiMFL2 expression, transcript levels were monitored by RT-PCR throughout the day. We found that the level of expression for this gene remained high and stable throughout the day and night (data not shown).
The data indicate that vernalization promotes down-regulation of CiMFL, but cold conditions exert their effect only under LD conditions. In fact, in plants vernalized under SDs, the expression level of CiMFL was maintained, and when the same plants were shifted to LDs they did not flower. In addition, one of the CiMFL transcripts analysed (CiMFL1) responded differently to vernalization, because it was not affected by cold induction in LD conditions either. We assume that CiMFL1 is transcribed from a different MFL locus than MFL2 in the C. intybus genome and is not regulated by vernalization.
Effect of CiMFL down-regulation on floral transition
To shed light on the involvement of CiMFL down-regulation upon cold treatment in floral transition, we investigated changes in the SAM profile during inflorescence meristem (IM) initiation. In C. intybus the vegetative SAM had a concave shape and its three cell layers, L1, L2 and L3, were well defined (Fig. 3f). Conversely, in a plant vernalized for 7 d, a month after the end of the cold treatment (15 d after fixation of the vernalization treatment) the SAM had a completely different profile. It was ‘flattened’, and its three-layer (L1, L2 and L3) structure had lost its original organization, as a thick cell layer was formed by anticlinal divisions of the meristematic cells. An increase in the meristem diameter was also observed (Fig. 3d; 7 days of vernalization (dV) and 30 dV). Similarly to A. thaliana and other dicots, this cytological rearrangement characterizes the early stages of IM organization. Later, differentiation of the IM was evident when the cells in the thick layer began to assume a new organization: the profile of the apex appeared superficially wrinkled, mirroring the effect of a new pattern of cell division in the inner cells of the meristem (Fig. 3d; 30 dV T 40).
SAM profiles were also analysed in plants vernalized for 40 d under the SD photoperiod, and then shifted to LDs to await flowering. DAPI staining of SAM sections demonstrated that vernalization in SDs was not sufficient to induce the formation of the IM (Fig. 3b).
To determine whether apical meristem induction was promoted by cold, we analysed the SAM profile of chicory grown for a month in LDs at 22°C. The SAM maintained its vegetative structure, indicating that an LD photoperiod cannot induce IM differentiation (data not shown).
In addition, we verified that the repression of CiMFL2 induced by vernalization in LDs was maintained after the treatment ended and led to changes in the organization of the IM (Fig. 3c).
These results indicate that CiMFL2 down-regulation is associated with a reorganization of SAM morphology and with IM differentiation.
Expression analysis of CiMFL at the transcript level
To better characterize the MFL sequences isolated from chicory, we analysed expression of the gene in different plant tissues and organs by RT-PCR. CiMFL2 expression was detected in leaves, in shoot apexes and in both unpollinated and pollinated flowers (data not shown).
To determine CiMFL expression at the tissue level, we performed in situ hybridization experiments. With our probe we were not able to selectively hybridize a specific CiMFL variant (i.e. CiMFL1, CiMFL2 or CiMFL3). In fact, the main differences among the sequences of CiMFL1, CiMFL2 and CiMFL3 are located in the MADS domain, which we had to remove from the probe in order to make the hybridization MFL-specific. Thus, from here onwards, ‘CiMFL expression’ is used to refer to the general expression of a pool of multiple sequences.
CiMFL expression was detected in the embryonic apical meristem (Fig. 4a) and in the epidermal tissue of the cotyledons (Fig. 4b). In 4-wk-old nonvernalized plants the hybridization signal was detected in the leaf axils and in the adaxial surface of the leaf (Fig. 4c). Shoot apexes and leaves of vernalized plants did not show any MFL expression (Fig. 4d). In flower sections the hybridization signal was localized in the pollen grains (Fig. 4e,f).
CiMFL2 and CiMFL3 protein localization
The localization of CiMFL proteins was assessed using GFP reporter fusions with CiMFL2 and CiMFL3 CDS (CiMFL2::GFP and CiMFL3::GFP). We selected these two CiMFL cDNAs to investigate whether the deletion of 14 amino acids inside the MADS domain of CiMFL3 could be responsible for a different cellular localization of the protein or could affect translation. In transformed chicory TVT protoplasts, the CiMFL2 protein was localized in the nucleus (Fig. S2). The same localization was observed for CiMFL3 (data not shown), thus excluding a specific role for the 14 amino acid sequence insertion in protein localization.
Resetting of CiMFL expression after vernalization
Semiquantitative RT-PCR and in situ hybridization experiments showed that CiMFL was down-regulated by cold treatment and that after flowering the expression of CiMFL was up-regulated in pollen (Figs 3,4). Moreover, in Southern blot analysis we found that methylation of CiMFL chromatin is not involved in its silencing (data not shown).
To determine at which stages during gametogenesis CiMFL expression was resumed, in situ hybridizations were carried out on chicory flowers at different developmental stages during microsporogenesis. This allowed us to follow the expression of CiMFL transcripts in the anther during meiosis. In chicory pollen, ontogeny has been divided into five developmental stages: pollen mother cells (PMCs), meiocytes, tetrads, microspores and pollen (Varotto et al., 1996).
In flower buds from vernalized chicory plants, the expression of CiMFL was detected in the anthers starting at the pollen mother cells stage (Fig. 5a,b). In particular, during pre-meiosis expression was also detected in the meiocytes as well as in the surrounding tapetum (Fig. 5c–e). In post-meiosis, CiMFL expression was observed in dyads, in tetrads, in the surrounding tapetum, in young microspores and in pollen (data not shown), as in nonvernalized plants (Fig. 4). This result suggests a possible resetting of CiMFL expression in the male reproductive structures of vernalized plants, in the sporogenous tissues before meiosis, and in the highly specialized nourishing cells of the tapetum before their degeneration.
Functional complementation of CiMFL2 in A. thaliana
The A. thaliana flc3 null mutant (Michaels & Amasino, 1999) shows an early flowering phenotype as a result of the knockdown of FLC expression. To assess the biological function of CiMFL2, the transcript was ectopically expressed in the A. thaliana FLC null mutant flc3. The sequence, driven by the 35S promoter, was inserted into a binary vector containing a gene for hygromycin resistance for transgenic selection. The construct was transformed into A. tumefaciens strain C58i_pMD990. Arabidopsis thaliana plants were stably transformed using the floral dip method. Transgenic plants were selected by antibiotic resistance and analysed for the presence of the transgene by PCR, using construct-specific primers (data not shown). The 35S::CiMFL2 lines of A. thaliana did not show a typical late-flowering phenotype (measured by counting the total number of leaves in T1 plants; Fig. 6) although they expressed the CiMFL2 transgene. Nonetheless, we were struck by the phenotype of the plants and particularly by the arrangement and morphology of the leaves. In fact, as shown in Fig. 6, seedlings of the transformants had unusually curled leaves, with irregular margins. Even after transplanting, transformants maintained the leaf phenotype, which was, however, lost after flower stalk emergence. This phenotype was observed in 80% of the independent transformants analysed (60 plants). The A. thaliana mutant transformed with the empty vector showed a regular phenotype (Fig. 6).
Using gene-specific primers for transgene amplification, mRNA expression in transgenic A. thaliana was confirmed by RT-PCR (data not shown). The positive transgenic lines showed the expected CiMFL2 transcript, but while the level of expression was high in young plants, in 10-d-old nonvernalized plants transcription was drastically reduced. CiMFL2 was detectable, albeit at a low level, in all the transformants and its expression did not explain the earliness of the phenotype. Taken together, these results indicate that CiMFL2 is not a functional FLC homologue in C. intybus.
In this work we investigated the transition to flowering mediated by MFL down-regulation through the vernalization pathway in C. intybus.
Chicory cannot be considered a model plant for molecular biology studies because of its strong auto-incompatibility mechanism which makes it difficult to obtain inbred lines (Varotto et al., 1995). Moreover, its long life-cycle, its unsequenced genome and the absence of established technical procedures hamper the use of this species in successful breeding programmes. For all these reasons, we encountered several problems in our efforts to isolate and characterize the MFL gene.
Identification of CiMFL sequences
Four CiMFL transcripts were identified and sequenced in the heterozygous cultivar Treviso (TVT) and named CiMFL1, CiMFL2, CiMFL3 and CiMFL4. Of these, CiMFL2 and CiMFL3 were investigated in more detail, as they were initially considered putative orthologues of the A. thaliana FLC. In this study, the variants CiMFL1 and CiMFL4 were not characterized in relation to flowering induction. In fact, CiMFL1 does not respond to vernalization treatment and shows a divergent sequence at the 3′-end of the CDS with respect to the other MFL sequences isolated; in contrast, CiMFL4 has a stop codon at the beginning of its sequence and lacks many of the conserved regions of MADS-box genes (i.e. the MADS-box, K-box and I-region).
The overlapping patterns of localization and the effectiveness of the vernalization treatment in down-regulating CiMFL2 and CiMFL3 supported our hypothesis that CiMFL2 and CiMFL3 are two expressed alleles of the same putative CiMFL locus (95.8% amino acid identity). A comparison between the amino acid sequence of CiMFL2 and those of FLC-like genes of other species showed that there is a 14 amino acid region within the MADS domain that is unique to C. intybus. Three other CiMFL-like sequences were isolated in a wild accession of chicory (FB) and were named CiMFLa, CiMFLb and CiMFLc. Alignment of the variants isolated in TVT with those identified in FB revealed a high degree of amino acid identity between CiMFL2 and CiMFLa (91%), suggesting that CiMFLa could be the functional allele of the MFL locus in the FB homozygous line.
The results from the Southern hybridization suggested that more sequences of CiMFL are present in the genome of chicory (Fig. S1). Two inverted repeats at the 3′ end of the transcript characterize CiMFL1 and the computational analyses showed that this variant forms a hairpin structure, which may be involved in miRNA biogenesis. Moreover, CiMFL1 is neither down-regulated by the vernalization treatments and nor involved in the induction of flowering mediated by cold treatments. These observations suggest that CiMFL1 might be codified by a different MFL genomic sequence and differently regulated at the transcriptional level from the CiMFL2 and CiMFL3 alleles.
CiMFL down-regulation and IM differentiation
In this study, for the first time we analysed the vernalization response of C. intybus, both monitoring CiMFL expression and observing SAM morphology during and after cold treatments. We showed that the vernalization treatment was effective in down-regulating CiMFL2 and that the repression was stably maintained. A period of 7 d of cold treatment at 4°C, under LDs, was enough to induce full down-regulation of all CiMFLs, apart from CiMFL1, for which constitutive expression was maintained, demonstrating that CiMFL is clearly an essential component of the vernalization response (Fig. 7).
CiMFL down-regulation was associated with a change in the morphology of the SAM, which was transformed from a ‘concave’ profile, corresponding to a three-layer (L1, L2 and L3) organization characteristic of a vegetative meristem, to a ‘flat’ apex in which the three-layer organization was lost and the meristem appeared to be committed to flowering.
We observed that the LD photoperiod itself was not able to induce flowering in chicory and that CiMFL2 expression was unaffected by the rhythms of the circadian clock.
However, vernalization treatments were completely ineffective under SDs and neither CiMFL2 down-regulation nor changes in SAM morphology were observed. These results are apparently in contrast to those previously reported for A. thaliana (Lee & Amasino, 1995), Raphanus sativus (J.-Young Yun et al., unpublished data), Brassica napus (Fowler et al., 2001) and Sinapis alba (D’Aloia et al., 2008). In fact, in A. thaliana and related species, LD conditions allowed more effective vernalization, with longer periods of vernalization being needed to down-regulate FLC expression under SDs.
Along with a photoperiodic effect on IM differentiation in vernalized plants, we noted a determinant effect of temperature on the differentiation of floral structures. A decrease of 2–4°C had marked consequences for the flowering behaviour of vernalized plants. Even if the apex presented all the characteristics of an IM and the photoperiod was favourable, the emergence of the floral stalk was arrested and resumed only when the temperature increased again. The critical value of the temperature for flowering was over 23°C (A. Locascio et al., unpublished results). Taken together, our results for the vernalization treatments confirm that CiMFL is a key gene for the transition to flowering in C. intybus, and that both photoperiodic and vernalization pathways contribute to CiMFL silencing, but they also indicate that CiMFL down-regulation itself induces differentiation of the IM. However, this last process is not sufficient to trigger flowering. We hypothesize that there is an additional mechanism controlling floral structure differentiation after IM formation in chicory.
Given the similarity in behaviour and sequence between AtFLC and CiMFL2, also confirmed by the phylogenetic analysis, we investigated whether CiMFL2 could rescue the null mutant AtFRI;flc3. CiMFL2 was not able to rescue this null mutant. Instead, we observed a peculiar ‘leaf phenotype’ in all the independent transformants, and this altered phenotype was maintained until the plants flowered. Although in these plants the level of transcript was higher compared with that of AtFLC, they still showed an early flowering phenotype. One hypothesis is that the additional sequence present in the CiMFL2 MADS domain impairs the formation of the MADS complex. A second hypothesis is that the transgene is recognized as a foreign sequence and a mechanism of silencing would block CiMFL2 expression. A third possibility is that CiMFL2 is not homologous in function to AtFLC and has a new function: the regulation of leaf orientation. An irregular phenotype was also observed during BnFLC characterization. In particular, in transgenic lines of B. napus inflorescence development was disturbed and fertility slightly reduced as a result of irregular anther development (Tadege et al., 2001).
Our idea that CiMFL2 might control leaf orientation is supported by several lines of evidence. First, it did not appear to be the only determinant of flowering except under certain conditions (i.e. certain photoperiod and temperature conditions). Secondly, its pattern of expression in nonvernalized chicory adult plants was limited to the leaf axils. Thirdly, chicory plants committed to flowering drastically changed their leaf organization to allow floral stalk emergence.
To test this last hypothesis, we are currently analysing chicory mutants in which CiMFL2 is either over-expressed or silenced. Preliminary results show that over-expression of CiMFL2 inhibits differentiation of transformed calli, while the silenced line shows a leaf phenotype. Nevertheless, better characterization of these mutants is required.
As FLC in A. thaliana, CiMFL2 down-regulation is maintained after the cold treatment ended. A study on the resetting of FLC repression produced by vernalization in A. thaliana has recently been published (Sheldon et al., 2008). The authors reported that both paternally and maternally derived FLC genes were reset in the progeny. In particular, in vernalized plants the paternal FLC allele was transcribed in the anthers, and also in somatic tissues and in the sporogenous PMCs, but there was no transcription in mature pollen. In chicory, during microsporogenesis, the CiMFL transcript was localized in the meiocytes and in the tapetum, and then in the dyads and tetrads in free microspores and in pollen grains. These results are consistent with the pattern of resetting found in A. thaliana.
The authors thank M. A. Blazquez, V. Rossi and N. Carraro for critical reading of the manuscript, members of the laboratory for helpful comments, and particularly A. Vannozzi for help with the vernalization experiment, S. Canova for help with the Southern blots and M. Salmaso for expert assistance with the phylogenetic analyses. We also thank C. Nicoletto for providing the TVT seeds, T. Pengo for taking care of the plant material and A. Garside for English language revisions.