A root chicory MADS box sequence and the Arabidopsis flowering repressor FLC share common features that suggest conserved function in vernalization and de-vernalization responses


For correspondence (e-mail cperilleux@ulg.ac.be).


Root chicory (Cichorium intybus var. sativum) is a biennial crop, but is harvested to obtain root inulin at the end of the first growing season before flowering. However, cold temperatures may vernalize seeds or plantlets, leading to incidental early flowering, and hence understanding the molecular basis of vernalization is important. A MADS box sequence was isolated by RT-PCR and named FLC-LIKE1 (CiFL1) because of its phylogenetic positioning within the same clade as the floral repressor Arabidopsis FLOWERING LOCUS C (AtFLC). Moreover, over-expression of CiFL1 in Arabidopsis caused late flowering and prevented up-regulation of the AtFLC target FLOWERING LOCUS T by photoperiod, suggesting functional conservation between root chicory and Arabidopsis. Like AtFLC in Arabidopsis, CiFL1 was repressed during vernalization of seeds or plantlets of chicory, but repression of CiFL1 was unstable when the post-vernalization temperature was favorable to flowering and when it de-vernalized the plants. This instability of CiFL1 repression may be linked to the bienniality of root chicory compared with the annual lifecycle of Arabidopsis. However, re-activation of AtFLC was also observed in Arabidopsis when a high temperature treatment was used straight after seed vernalization, eliminating the promotive effect of cold on flowering. Cold-induced down-regulation of a MADS box floral repressor and its re-activation by high temperature thus appear to be conserved features of the vernalization and de-vernalization responses in distant species.


Plants have evolved a complex regulatory network that ensures that flowering occurs at the correct time of year. They therefore rely on environmental cues such as photoperiod, that is the relative duration of day and night. Many plants in temperate climates also use prolonged exposure to cold, which increases the ability to flower in winter annuals, biennials and perennials (Chouard, 1960; Bernier et al., 1981). Distinguishing features of this process, known as vernalization, are its quantitative nature, and, in most cases, the temporal uncoupling between stimulus and effect. Vernalization is often combined with a long-day (LD) requirement, so that flowering occurs after winter, that is in spring or early summer. Adverse conditions encountered after vernalization, for example short days (SD) or high temperature, may reduce the promotive effect of cold, and hence the vernalized state is a reversible process.

The vernalization requirement is an important trait in agriculture, and determines the adaptation range of domesticated species. For example, the vernalization requirement of winter cereals allows them to start growing in the fall/autumn, take maximum advantage of the spring before flowering, and reach early maturity to avoid yield-limiting summer heat. Another type of crop with a strong vernalization requirement are biennials that are harvested at the end of the first growing season for their vegetative parts, for example fleshy roots that store food reserves before winter. Root chicory (Cichorium intybus L. var. sativum) is one such plant. It is cultivated mainly in Western Europe for its tap root, which accumulates inulin-type fructans. It was used in the past for production of an ersatz coffee, and is presently used in functional foods as a source of pre-biotic fibers. Other varieties of chicory are grown for their foliage: radicchio (C. intybus L. var. silvestre) is cultivated for its colored leaves, and witloof (C. intybus L. var. foliosum) is cultivated for its etiolated leaves, produced by forcing the roots. All these chicory crops are biennials, requiring cold and LD conditions for flower induction (Paulet, 1985; Pimpini and Gianquinto, 1988; Gianquinto, 1997; Demeulemeester and De Proft, 1999; Dielen et al., 2005). They normally remain vegetative during the first growing season, but there is a risk that the plants will experience cold and become vernalized if sown too early. This is especially critical for root chicory, which is prone to spring vernalization (Dielen et al., 2005). A low sensitivity to vernalization – commonly referred to as ‘resistance to bolting’ – is therefore a major trait used in breeding programs. Such breeding programs would be greatly facilitated by a better understanding of the molecular mechanisms involved. Although the availability of chicory sequences has considerably increased recently, no genetic determinant of bolting has been identified so far in root chicory, and commercial varieties are still produced as synthetic populations characterized by a high genetic diversity (Van Cutsem et al., 2003).

The molecular basis of vernalization has been investigated in depth in the annual plant Arabidopsis thaliana, for which the vernalization requirement of winter accessions was found to correlate with the level of expression of a gene encoding a potent inhibitor of flowering, FLOWERING LOCUS C (FLC) (Michaels and Amasino, 1999; Sheldon et al., 1999). The FLC protein, a MADS domain transcription factor, blocks at multiple points the activation cascade that normally triggers flowering via up-regulation of the FLOWERING LOCUS T (FT) gene under favorable LD conditions. The FT protein is synthesized in the leaves, moves to the shoot apical meristem and triggers changes in gene expression that initiate the switch from vegetative to reproductive growth (Turck et al., 2008). FLC not only inhibits production of FT in the leaves, but also impairs the response to the FT signal in the shoot meristem (Michaels and Amasino, 2001; Helliwell et al., 2006; Searle et al., 2006). Two pathways promote transition to flowering through repression of FLC expression: the so-called ‘autonomous pathway’, which involves constitutive repressors of FLC, and the vernalization pathway, which down-regulates FLC via a complex series of mechanisms including synthesis of long non-coding transcripts and epigenetic chromatin modifications (Ietswaart et al., 2012).

Allelic variation at the FLC locus is associated with flowering time variation in Arabidopsis, but only in epistatic interaction with the upstream regulatory gene FRIGIDA (FRI), which activates FLC. Late-flowering ecotypes contain functional alleles of FRI and FLC, whereas rapid-cycling accessions have evolved through loss of FRI function and/or attenuation of FLC activity (Johanson et al., 2000; Gazzani et al., 2003; Michaels et al., 2003). This is the case in Columbia, which lacks a functional FRI (and hence is hereafter referred to as Col fri), and is therefore fast-flowering independently of vernalization because of low FLC expression. However, introgression of an active FRI allele causes an increase in FLC expression and so confers a vernalization requirement to the resulting Col FRI line (Lee and Amasino, 1995). Naturally occurring weak alleles of FLC are not usually caused by a difference in their coding region but by a difference in the regulatory elements contained in their promoter (or 5′ untranslated) region or the first intron (Sheldon et al., 2000; Michaels et al., 2003; Liu et al., 2004). An flc null mutant eliminates the FRI late-flowering phenotype, but retains some sensitivity to vernalization, indicating that vernalization is able to promote flowering via FLC-independent mechanisms (Michaels and Amasino, 2001). These alternative routes involve other MADS box genes such as the MADS AFFECTING FLOWERING genes (MAF1–5) that belong to the FLC clade, and AGAMOUS-LIKE19 and 24 (AGL19 and AGL24) (Alexandre and Hennig, 2008).

Orthologs of Arabidopsis FLC (hereafter AtFLC) have been found in other Brassicaceae, such as Arabis alpina, Arabidopsis halleri, Brassica napus, B. oleracea, B. rapa, Capsella rubella and Sinapis alba (Schranz et al., 2002; D'Aloia et al., 2008; Wang et al., 2009; Aikawa et al., 2010; Guo et al., 2012). Identification of the AtFLC ortholog PERPETUAL FLOWERING 1 (PEP1) in Arabis alpina highlighted a criticial mechanism by which the regulation of flowering differs between related annual and perennial plants. In contrast to AtFLC, which is stably repressed by cold in annual A. thaliana, re-activation of PEP1 after a return to warm temperature blocks flowering of all meristems that did not undergo flowering during the vernalization treatment, and confers perenniality to Arabis alpina (Wang et al., 2009).

FLC orthologs initially appeared to be restricted to the Brassicaceae family (Becker and Theissen, 2003). For example, in winter cereals, vernalization down-regulates an inhibitor of FT as in Arabidopsis, but this inhibitor, called VERNALIZATION2 (VRN2), is not orthologous to AtFLC and is not directly repressed by cold (Dennis and Peacock, 2009). In sugar beet (Beta vulgaris), the obligate vernalization requirement of biennial cultivated sub-species is due to a homozygous recessive mutation at the long-sought ‘bolting gene B‘. This was recently shown to be due to partial loss of function of an activator of FT that causes reduced sensitivity to photoperiod that is restored by vernalization (Pin et al., 2012). Sugar beet is the first case in which an AtFLC homolog has been characterized outside the plant family Brassicaceae (Reeves et al., 2007), but, although BvFL1 repressed flowering in transgenic Arabidopsis and was down-regulated in response to cold, how this integrates with the action of the ‘bolting gene B’ remains to be investigated. Interestingly, the phylogenetic approach used by Reeves et al. (2007) indicated evolutionary conservation of FLC-like genes in the three major eudicot lineages: rosids (including Brassicales such as Arabidopsis), caryophyllids (including Caryophyllales such as sugar beet) and asterids. The latter clade includes Asterales species such as root chicory, and we provide here functional evidence supporting this prediction. We isolated the root chicory CiFL1 sequence, which falls in the FLC/MAF clade of MADS box genes and functions like AtFLC as a repressor of flowering in transgenic Arabidopsis. CiFL1 is down-regulated in response to cold in root chicory and re-activated at de-vernalizing temperature. We show that this de-vernalization response is also correlated with resumption of AtFLC activity in Arabidopsis, and hence may be a common feature of the flowering response to temperature in distant species.


Root chicory is an obligate vernalization-requiring plant

In a first set of experiments, experimental conditions were established to grow root chicory up to flowering in controlled cabinets. Two cultivars were tested: Crescendo, which is categorized as resistant to vernalization, and Fredonia, which is more sensitive. Vernalization was applied to imbibed seeds at 4°C, and, as germination occurred at that temperature, seeds were sown on soil and placed in a lightened cabinet to allow seedling establishment during the cold treatment (Figure 1a). To mimic winter conditions, a short photoperiod and low light were provided [8 h day−1 (SD conditions), 25 μmol m−2 sec−1]. At the end of the vernalization treatment, plantlets were transplanted and exposed to ‘summer-like’ conditions: 20°C, 16 h day−1 (LD conditions), 300 μmol m−2 sec−1. Plants were observed every 2 days to detect bolting and record anthesis. All bolted plants initiated inflorescences, and hence we consider bolting and flowering without distinction in this paper.

Figure 1.

Development of Cichorium intybus var. sativum in phytotronic cabinets. (a) Seedling development during vernalization at 4°C under 8 h SD conditions (25 μmol m−2 sec−1). (b) Adult plant vernalized for 4 weeks, then grown for approximately 100 days at 20°C under 16 h LD conditions (300 μmol m−2 s−1). The numbers of nodes are shown for a representative experiment (cv. Fredonia).

Both cultivars showed a strict requirement for vernalization: no flowering occurred in non-vernalized plants, whereas a high percentage of bolting plants was recorded after 4 weeks of vernalization. In seven independent experiments, the mean proportion of flowering plants was 69% in Crescendo and 98% in Fredonia. The development rate of the plants that bolted was amazingly synchronized. In general, 1 month elapsed between the end of vernalization and visible bolting (Table S1). The adult plant consisted of (in acropetal order): (i) rosette leaves and a complex elongated stalk bearing (ii) cauline leaves, (iii) lateral flowering branches, (iv) sessile capitula in clusters of decreasing size (from four capitula to one capitulum), and (v) a terminal capitulum (Figure 1b). Anthesis of the first capitulum occurred 1 month after the start of bolting, and was observed at mid-height of the flowering stalk, in upper type 3 or lower type 4 nodes (Table S1). Successive antheses were then observed in both acropetal and basipetal directions, except for the terminal capitulum, which opened its flowers at about the same time as median nodes of type 4.

These experiments showed that vernalization starting at the imbibed seed stage allows root chicory to be grown as an annual, and that flowering up to anthesis may be achieved within growth chambers. However, under natural conditions the cultivated root chicory behaves as a biennial, and vernalization occurs at the adult rosette stage. This suggests that the conditions for vernalization of the imbibed seeds may be set up artificially, but are not usually met in nature, or that the vernalized state is not maintained during the first growing season. We therefore analysed the effects of vernalization duration, seedling age and post-vernalization conditions. We focused on the Fredonia cultivar, which showed the highest sensitivity to vernalization and lowest variability.

First we observed that 4 weeks of vernalization were necessary for full induction of flowering. The flowering response dropped to 44.0% when vernalization was limited to 2 weeks, and 1 week of cold was not enough to obtain a single flowering plant (Figure 2). Second, we found that when the 4-week vernalization treatment was not given at the imbibed seed stage, but to 7-day-old plantlets, the percentage of bolting plants was reduced to 33.3%, that is one-third of the flowering induction that was obtained with vernalized seeds. This is consistent with a previous report indicating that a short developmental window gates the response to vernalization in root chicory in the stages immediately following seed germination (Dielen et al., 2005). An extension of the vernalization period compensated for the post-germination decrease in sensitivity: plantlets that were vernalized for 6 weeks resulted in as many bolting plants as imbibed seeds that had been vernalized for 4 weeks.

Figure 2.

Flowering response of Cichorium intybus var. sativum (cv. Fredonia) to various vernalization treatments and post-vernalization temperatures. Imbibed seeds (left) or 7-day-old plantlets (right) were vernalized for 1–6 weeks at 4°C (black). After vernalization, plants were transferred to 17–20°C (white) or 25°C (gray) under 16 h LD conditions. The proportion of flowering plants (% bolted plants) was evaluated a maximum of 100 days later. The numbers in parentheses indicate the number of independent experiments from which results were calculated. Each experiment was performed with 12–18 plants per batch.

We next studied the effect of post-vernalization growing conditions, and observed that an increase in temperature had an unfavourable effect on flowering. Compared with the standard temperature of 17–20°C, the percentage of bolting plants was less at 25°C whether vernalization was applied to imbibed seeds or 7-day-old plantlets (Figure 2). The treatment at 25°C was inhibitory to flowering even if transient, as may been seen from the adverse effect of 2 weeks at 25°C before transfer to 17–20°C. Most interestingly, 1 week at 17–20°C limited this negative effect when applied between the vernalization and 25°C treatments. These results indicate that the temperature just after vernalization is critical for the flowering response: 25°C exerted some de-vernalizing effect, but 17–20°C stabilized the vernalized state.

Identification of CiFL1

To identify FLC homologs in root chicory, degenerated primers were designed based on the sequences of the FLC/MAF-like proteins found in Asteraceae databases. This allowed cloning of a 624 bp cDNA, referred to as CiFL1 (Cichorium intybus FLC-LIKE 1). The predicted amino acid sequence of CiFL1 had the same MIKC (for MADS-, Intervening-, Keratin-like-, and C-terminal domains) structure as AtFLC (Figure 3a). The sequence of the MADS domain of CiFL1 showed very high conservation when aligned against the consensus sequence of FLC-like proteins from Brassicaceae (Figure 3b), whereas conservation of other domains was much weaker (Figure S1). Most interestingly, CiFL1 was found to contain an acidic residue (E) in position 30 of the MADS domain that is specific to FLC/MAF transcription factors in Arabidopsis (Ratcliffe et al., 2001). Phylogenetic analyses were performed and confirmed that CiFL1 fell within the FLC/MAF sub-tree of MADS box proteins (Figure 3c), and that the nucleotide sequence of its MADS domain was the most similar to Asteraceae ESTs, as expected (Figure 3d). Surprisingly, this analysis showed that previously reported chicory sequences annotated as ‘MADS FLOWERING LOCUS C-LIKE (MFL)’ (Locascio et al., 2009) were actually much closer to AtFLC than to Asteraceae ESTs. These sequences were not included in our phylogenetic tree because, upon further analysis, they appeared to be chimeric (Figure S2).

Figure 3.

Phylogenetic analyses of CiFL1. (a) Structure of AtFLC protein showing the conserved domains of MIKC-type MADS box transcription factors (modified from Kaufmann et al., 2005). (b) Alignment of the deduced amino acid sequence of the MADS box of CiFL1 (CCO61905.1) against the MADS box consensus sequence (≥ 70% conservation) and the sequence logo of FLC- and MAF-like proteins from Brassicaceae. (c) Arbitrarily rooted maximum-likelihood tree (LG+Γ4 model) of the MADS box domain of MIKC proteins from selected organisms (Arabidopsis thaliana, Beta vulgaris, Brassica napus, Helianthus annuus, Lactuca sativa and Oryza sativa). The FLC & FLC-like sub-tree is detailed whereas sub-trees of other groups of MADS box proteins were collapsed, and are annotated with representative proteins from Arabidopsis thaliana or Oryza sativa. (d) Arbitrarily rooted maximum-likelihood tree (GTR+Γ4 model) of the MADS box domain of FLC-like nucleotide sequences found in National Center for Biotechnology Information nucleotide and EST databases. In both trees, the numbers on branches correspond to statistical support values (≥ 50%) obtained from analysis of 500 bootstrap replicates. Arrows indicate CiFL1.

Over-expression of CiFL1 delays flowering in Arabidopsis

In order to obtain further insight into the biological function of CiFL1, the cDNA sequence was over-expressed in Arabidopsis Col fri and the flc-3 mutant. Three 35S:CiFL1 transgenic lines were obtained for Col fri and one for flc-3. The flowering time of T3 lines was recorded under 10 h SD conditions after various durations of seed vernalization. Col FRI was used as a control for late flowering and high sensitivity to vernalization associated with a high level of AtFLC expression.

Over-expression of CiFL1 in the Col fri background had little effect on flowering time without vernalization, but a significant delay was recorded for vernalized plants because the transgene weakened the flowering response to cold (Figure 4a). By contrast, over-expression of CiFL1 in the flc-3 mutant delayed flowering in vernalized and non-vernalized plants. These results suggest that a delay in flowering is caused by CiFL1 when AtFLC was not active (vernalized Col fri or loss-of-function flc-3 mutant).

Figure 4.

Effect of the 35S:CiFL1 transgene on the flowering time of Arabidopsis. (a) Flowering time under 10 h SD conditions without (NV) or after 2, 4 or 6 weeks of vernalization (V2, V4 and V6). Flowering time was recorded as final leaf number. (b) CiFL1 and AtFLC transcript levels were estimated by semi-quantitative RT-PCR in aerial parts of seedlings harvested 15 days after the end of vernalization, under 10 h SD conditions. ELONGATION FACTOR (EF1) was used as a constitutive control gene. (c) Flowering time under 16 h LD conditions without (NV) or after 6 weeks of vernalization (V6). Flowering time was recorded as days from end of vernalization to macroscopic appearance of flower buds. In (a) and (c), three transgenic lines in the Col fri background and one transgenic line in the flc-3 mutant are shown. Asterisks indicate significant differences between transgenic lines and their untransformed background (two-tailed t tests, < 0.05, = 12). In (b), data for one representative 35S:CiFL1 Col fri line are shown.

These phenotypes were correlated with the analyses of AtFLC and CiFL1 transcript levels 15 days after the end of vernalization (Figure 4b). Vernalization caused AtFLC silencing in Col fri and Col FRI control plants as expected, but also in transgenic plants over-expressing CiFL1, indicating that the transgene did not prevent down-regulation of endogenous AtFLC by cold. By contrast, the abundance of CiFL1 transcripts remained high throughout the vernalization period in the transgenic lines. This is consistent with the fact that the transgene contained the coding sequence only. In Arabidopsis, the cis-regulatory elements of AtFLC are located in the promoter (or 5′ untranlated) region and the first intron (Sheldon et al., 2002).

The 35S:CiFL1 Arabidopsis lines were also grown under 16 h LD conditions, and their delay in flowering was much stronger than under 10 h SD conditions, with appearance of the first flower bud taking almost twice as long as in non-transgenic control plants (Figure 4c). Over-expression of CiFL1 also completely over-rode the effect of a 6-week vernalization treatment on flowering. One of the 35S:CiFL1 Col fri lines even flowered as late as non-vernalized Col FRI plants, indicating that the chicory transgene repressed flowering under LD conditions to the same extent as high expression of AtFLC does in a Col FRI background.

The fact that the retarding effect of CiFL1 under SD conditions was mostly visible when AtFLC was not active (Figure 4a) suggests that competition may exist between the encoded proteins for the same partners or targets. We therefore analysed whether CiFL1 affected the repressive action of AtFLC on its target FT (hereafter referred to as AtFT). As AtFT is up-regulated under LD conditions, we subjected the plants to a 48 h light period before performing the expression analyses. All samples were collected at the same time of the subjective day (Zeitgeber time 8, ZT8). Expression of AtFT after the photoperiodic treatment was clear in vernalized Col fri, Col FRI and flc-3 plants (Figure 5a). Most interestingly, over-expression of CiFL1 completely abolished activation of AtFT in the Col fri and flc-3 backgrounds, indicating that CiFL1 is able to repress photoperiodic induction of AtFT in Arabidopsis. Again consistent with these molecular data, flowering after the 48 h light period was much delayed in the CiFL1 over-expressors compared with the untransformed plants (Figure 5b).

Figure 5.

Effect of the 35S:CiFL1 transgene on the Arabidopsis response to LD conditions. Three transgenic lines in the Col fri background and one transgenic line in the flc-3 mutant were used, together with Col fri, flc-3 and Col FRI as controls. After 6 weeks of vernalization, plants were grown under 10 h SD conditions for 4 weeks, followed by a 48 h light period. (a) AtFT transcript levels were estimated by quantitative RT-PCR in aerial parts of seedlings harvested 32 h after the start of the 48 h light period (ZT8). Data were normalized using ACTIN2 and TUBULIN2 as internal controls. (b) Flowering time expressed as days from the end of vernalization to macroscopic appearance of flower buds. Asterisks indicate significant differences between the transgenic lines and their untransformed background (two-tailed t tests, < 0.001, = 10–12).

CiFL1 is repressed in chicory during vernalization but repression is not maintained

Expression analyses of CiFL1 in root chicory were performed during vernalization. We observed expression of CiFL1 in the seeds and a continuous decrease in transcript abundance during their exposure to cold, indicating that CiFL1 was down-regulated during vernalization (Figure 6a). In the next step, we analysed the expression level of CiFL1 after vernalization treatment. Surprisingly, we observed a strong up-regulation of CiFL1 at both the stabilizing (17–20°C) and de-vernalizing (25°C) temperatures, indicating that the repressed state induced by vernalization was not maintained (Figure 6b).

Figure 6.

Steady-state levels of CiFL1 transcripts in aerial parts of Cichorium intybus var. sativum (cv. Fredonia) during vernalization of imbibed seeds at 4°C (a) and after 4 weeks of vernalization (black) and transfer to 17–20°C (white) or 25°C (gray) under 16 h LD conditions (b). UBIQUITIN CONJUGATING ENZYME E2 (UBC) was used as a constitutive control gene.

We performed the same analyses using 7-day-old plantlets, and observed that expression of CiFL1 was down-regulated during cold and resumed to high levels after return to either 17–20°C or 25°C, as observed with imbibed seeds (Figure S3).

Re-activation of AtFLC also occurs in Arabidopsis at de-vernalizing temperature

The instability of CiFL1 repression after cold treatment contrasts with what is known about the AtFLC gene in Arabidopsis, which is stably repressed by vernalization providing that the cold treatment is long enough (Gendall et al., 2001). However, maintenance of AtFLC repression is usually evaluated in Arabidopsis after ‘return to warm conditions’, that is 20–23°C, as used in this study (Figure 4). We therefore performed a complementary experiment in which Arabidopsis was exposed to various temperatures after the vernalization period. Seeds were vernalized for 5 weeks, then immediately transferred to 20°C or exposed to 1 week at 30°C before the transfer to 20°C (Figure 7a). Seedlings were harvested 15 days after transfer to 20°C for expression analyses of AtFLC. The winter accession Stockholm (St-0) containing active alleles of FRI and FLC was used in addition to Col fri and Col FRI. As expected, AtFLC was down-regulated by vernalization and repression was stable after transfer of seedlings to 20°C (Figure 7b). By contrast, when the seeds spent 1 week at 30°C after vernalization, the expression of AtFLC was re-activated to levels similar to those detected before vernalization. This was correlated with physiological suppression of the vernalization effect: when the seeds were exposed to 30°C just after the cold treatment, plants of the St-0 and Col FRI genotypes flowered as late as those that had not been vernalized (Figure 7c). The de-vernalizing effect of the treatment was greatest in Col FRI because of the large variation in AtFLC expression. In the next experiment, the effect of 1 week at 30°C was compared in Col FRI when applied to non-vernalized seeds, straight after vernalization, or 2 weeks later (Figure 8a). Re-activation of AtFLC was seen only when the 30°C treatment was applied straight after vernalization (Figure 8b), indicating that heat interfered with the mechanisms of AtFLC regulation occurring just after cold treatment. Consistent with these expression analyses, the vernalized plants that were exposed to 30°C straight after cold were as late flowering as non-vernalized plants and did not all flower within the duration of the experiment (100 days). By contrast, when the 30°C treatment followed 2 weeks at 20°C, plants flowered 55.5 ± 7.3 days after the end of vernalization, at the same time as non-treated vernalized plants (55.6 ± 3 days), indicating that the period at 20°C stabilized the vernalized state.

Figure 7.

Effects of post-vernalization treatment at 30°C on AtFLC transcript levels and flowering time in Arabidopsis. (a) Experimental design. After stratification or 5 weeks of vernalization (black), non-vernalized (NV) and vernalized (V5) seedlings of Col fri, Col FRI and St-0 genotypes were immediately transferred to 20°C under 16 h LD conditions (white) or exposed to 1 week at 30°C in darkness (gray) before transfer. (b) Steady-state levels of AtFLC transcripts were estimated by quantitative RT-PCR in the aerial part of the plants 15 days after transfer to 20°C. Data were normalized using ACTIN2 and TUBULIN2 as internal controls. (c) Flowering time expressed as days from transfer to 20°C to macroscopic appearance of floral buds. Asterisks indicate significant differences between vernalized (V5) and non-vernalized (NV) plants (two-tailed t tests, < 0.001, = 15–17).

Figure 8.

Effects of a treatment at 30°C on AtFLC transcript levels in Arabidopsis Col FRI. (a) Experimental design. After stratification or 5 weeks of vernalization (black), non-vernalized (NV) and vernalized (V5) seedlings were immediately transferred to 20°C under 10 h SD conditions (white) or exposed to 1 week at 30°C (gray) before transfer. One batch received the 1-week treatment at 30°C after 2 weeks at 20°C. Arrows indicate time of harvest for expression analyses. (b) Steady-state levels of AtFLC transcripts were estimated by quantitative RT-PCR in the aerial part of the plants. Data were normalized using ACTIN2 and TUBULIN2 as internal controls.


Biennial root chicory is cultivated as an annual but bolting is incidentally observed, especially in cases of early sowing. This undesired flowering may be attributed to cold encountered during spring. Root chicory is an obligate vernalization-requiring crop: in no case did we observe flowering of the plants without vernalization, as reported before for other cultivars (Dielen et al., 2005). Vernalization may occur during the first growing season at the very early stages of seed germination (Dielen et al., 2005), and may lead to full flowering under favorable conditions as shown here. By contrast, we observed that plantlets are much less sensitive to cold and require longer vernalization, as reported in the literature for other cold-requiring plants (Bernier et al., 1981). Together, these results suggest that bolting of chicory during the first growing season in the field remains incidental because cold, if experienced, may not coincide with the stage of highest sensitivity and/or may not be of long enough duration for vernalization.

We isolated the MADS box sequence CiFL1 to investigate the molecular mechanisms of vernalization in root chicory. Phylogenetic analysis of CiFL1 unambiguously showed that it falls within the FLC clade, and functional analyses revealed a number of common features with AtFLC. First, we observed down-regulation of CiFL1 during vernalization, and the repression increased with the duration of cold, as reported in Arabidopsis (Sheldon et al., 2000). Second, over-expression of CiFL1 in Col fri and the flc-3 null mutant of Arabidopsis delayed flowering, especially under LD conditions, and made the plants resistant to vernalization. Over-expression of AtFLC in early accessions of Arabidopsis also caused late flowering (Michaels and Amasino, 1999). Third, we showed that CiFL1 reduced the activation of AtFT by LD conditions in transgenic Arabidopsis, indicating that CiFL1 is able to perform at least some of the biological functions of AtFLC in Arabidopsis. As AtFLC directly binds to the promoter of AtFT (Helliwell et al., 2006), we speculate that CiFL1 has the same ability and represses FT in chicory. Such a hypothesis is consistent with the fact that root chicory requires LD conditions after vernalization to enter floral transition.

Together, these results suggest that CiFL1 acts like AtFLC as a repressor of flowering, but further experiments are required to confirm its biological function when expressed under the control of its own regulatory elements rather than a constitutive promoter. It also remains to be demonstrated whether the high expression level of CiFL1 in non-vernalized chicory plants is actually the cause of their absolute vernalization requirement for flowering. In the absence of chicory mutants, the genetic diversity of various cultivars should be explored to see whether variations in CiFL1 activity are associated with the resistance to bolting.

A divergence between CiFL1 and AtFLC is that repression of CiFL1 by cold in root chicory was not maintained after return to a normal growth temperature of 17–20°C, although the vernalized state was stabilized and flowering ensued. These results may indicate that the flowering response to vernalization in chicory does not exclusively proceed through CiFL1, in the same way as AtFLC-independent branches play a role in the vernalization pathway of Arabidopsis (Alexandre and Hennig, 2008). Nevertheless, more precise quantifications of CiFL1 activity are required. Indeed, antisense and sense non-coding transcripts have been implicated in the regulation of AtFLC activity during vernalization (Ietswaart et al., 2012), and hence qualitative assessment of CiFL1 transcripts should be performed. Furthermore, it was shown in Arabidopsis that maintenance of the AtFLC repressed state occurs only in cells in which a certain threshold of epigenetic marks have accumulated in the cold (Angel et al., 2011) and requires mitotic activity (Finnegan and Dennis, 2007). We cannot exclude the possibility that CiFL1 is stably silenced in chicory seedlings at the stabilizing temperature in a sub-population of cells, possibly in the shoot meristem, that represent a small proportion of the sample. Re-activation of CiFL1 in other tissues may be linked to the lifecycle of root chicory, as expression of other FLC homologs was found to be restored after cold in various biennial or perennial plants. In sugar beet, the instability of BvFL1 repression was correlated with the propensity for inflorescence reversion observed in this species (Reeves et al., 2007), but we did not observe similar trend in root chicory. In the perennial Arabis alpina, the instability of PEP1 repression was correlated with the polycarpic growth habit of the plant, which requires some meristems stay vegetative after vernalization to ensure plant growth continuation (Wang et al., 2009). Such an explanation does not hold for root chicory, which is a monocarpic species, and, under our growth conditions at least, does not undergo floral transition during vernalization. In cabbage, repression of BoFLC occurs only when cold is applied to leaf rosettes and not to seeds, but, when the gene is expressed in Arabidopsis, repression occurs when cold is applied to seeds but is not maintained (Lin et al., 2005). Together, these results clearly indicate that repression of AtFLC homologs and maintenance of the repressed state are under strong developmental control. Root chicory appears to be a very flexible system to study as it behaves as a winter annual plant when seeds are vernalized and as a true biennial when vernalized at the adult rosette stage. The switch from one lifecycle to another may involve spatial regulation of gene expression. Chicory is a species in which vernalization of the adult plant is perceived by the root collar and not by the bud, as reported for the witloof variety (Joseph et al., 1985). The bienniality of chicory may therefore involve a combination of developmental and spatial regulation of genetic repressors of flowering such as CiFL1.

De-vernalization by high temperature has been reported for a number of cold-requiring plants (Bernier et al., 1981) including radicchio, another chicory variety (Pimpini and Gianquinto, 1988; Gianquinto and Pimpini, 1995) and Arabidopsis (Napp-Zin, 1957), but had not been studied at the molecular level. We report here that, in both root chicory and Arabidopsis, the de-vernalization effect of heat is correlated with re-activation of CiFL1 and AtFLC, respectively. We also show in both species that the vernalized state may be stabilized by application of ‘normal’ temperature between the cold period and the transfer to heat. At the molecular level, these results are consistent with the fact that vernalization in Arabidopsis proceeds in two steps: repression of AtFLC is established during the cold treatment, then maintained during subsequent growth at 20–23°C (Angel et al., 2011). These steps correspond to the accumulation of repressive chromatin modifications in a ‘nucleation’ site at the start of AtFLC during the cold treatment, followed by spreading of the modifications across the AtFLC locus upon return to warm temperature. Our results clearly suggest that heat prevents spreading of the repressive chromatin modifications, which conditions maintenance of the repressed state of AtFLC, and may completely eliminate the promotive effect of vernalization on flowering. Given the considerable impacts that climate change may have on plant reproduction, de-vernalization certainly deserves more attention.

Experimental Procedures

Plant materials and growth conditions

Root chicory

The experiments were performed with two root chicory cultivars: Fredonia (SAREA, http://www.sarea.at) and Crescendo (Chicoline, http://www.cosucra.com). Achenes were sown at 1 cm depth in soil cores prepared in plastic tubes (height 80 mm; diameter 20 mm) filled with a mixture of peat compost and sand (3:1 v/v). Seedlings were placed under 8 h SD conditions for 1 day (‘imbibed seed’ stage) or 7 days (‘plantlet’ stage) at 20°C (day and night), 150 μmol m−2 sec−1, before transfer to 4°C (day and night), 25 μmol m−2 sec−1, for vernalization. After 1–6 weeks of vernalization, plantlets were returned to 20°C (day and night), 150 μmol m−2 sec−1, for 3 days before being transplanted into two-liter pots (height 180 mm; diameter 140 mm) containing the same substrate as above. The growing conditions were then changed to 16 h days (LD conditions), 300 μmol m−2 sec−1, at a temperature of 17-20°C or 25°C (day and night). Plants were watered daily with tap water, and fed once a week with 0.5 g l−1 SAP Vegetora (N-P-K 16-18-21; Comptoir Franco Belge des Engrais et Produits Chimiques, Spa, Belgium). All experiments were performed in controlled cabinets.


Experiments were performed with Arabidopsis accessions Columbia (Col fri) and Stockholm (St-0), obtained from the Nottingham Arabidopsis Stock Centre (http://arabidopsis.info/). The Col FRI line and the flc-3 null mutant were kindly provided by R. Amasino (Department of Biochemistry, University of Wisconsin, Madison, WI). Seeds were placed on wet filter paper in Petri dishes, or directly sown in trays of 1.5 × 1.5 cm compartments filled with sifted peat compost. The seeds were then placed at 2–4°C, in complete darkness, for 3 days (stratification) or for 5–6 weeks (vernalization). All seeds were thereafter sown in trays, and the trays were moved to a growth cabinet under 8 h SD, 10 h SD or 16 h LD conditions, 100 μmol m−2 sec−1, at 20°C (day and night). In the de-vernalization experiments, a 1-week treatment at 30°C was given to seeds straight after stratification or vernalization, or to seedlings after 2 weeks of growth under 10 h SD conditions at 20°C. Plants kept for flowering record were transplanted into larger containers (either individual 7 cm pots or one-liter tanks for six plants), containing a mixture of peat compost and sand (3:1 v/v) and were watered daily with tap water.

Flowering record

Plants were observed every 2 days to count rosette leaves, and to record visible bolting (chicory) or the macroscopic appearance of flower buds (Arabidopsis), as well as anthesis (chicory). These features were visible with the naked eye. The upper limit of the experiment duration was 100 days post-vernalization, when the percentage of flowering plants was calculated for each experimental batch of 12–18 individuals.

RNA isolation and RT-PCR analyses

RNA isolation from chicory and Arabidopsis seedlings

The whole aerial part was harvested from 10 to 12 plants, pooled and stored at –80°C until used. Samplings were performed at approximately 4 pm. Total RNA was extracted from biomass ground in liquid nitrogen, using TRIzol (Invitrogen, http://www.invitrogen.com) according to the manufacturer's instructions.

RNA isolation from chicory seeds

Approximately 100 mg of imbibed seeds were ground in liquid nitrogen, and total RNA was isolated as described by Vergara and Ismail (2008). This method uses 1% N-lauroyl sarkosinate (sarkosyl) as detergent and a double phenol/chloroform extraction to remove polysaccharides.

cDNA synthesis and RT-PCR

After DNase treatment of total RNA (0.2 U DNase μg−1), first-strand cDNA was synthesized from 1.5 μg RNA, using MMLV reverse transcriptase and oligo(dT)15 according to the manufacturer's instructions (Promega, http://www.promega.com/). Aliquots were used as templates for PCR with gene-specific primers (Table S2). In semi-quantitative analyses, PCR products were electrophoresed on a 2% agarose gel and stained with ethidium bromide. In quantitative PCR analyses, reactions were performed in triplicate using SYBR Green I (Eurogentec, http://www.eurogentec.com) in 96-well plates with an iCycler IQ5 (Bio-Rad, http://www.bio-rad.com). Quantification cycle (Cq) values were extracted using the instrument software, and individual quantitative PCR runs were imported into qbasePLUS 2.0 (http://www.qbaseplus.com). For normalization, a geNormPLUS analysis was performed with ten commonly used housekeeping genes. ACTIN2 and TUBULIN2 were selected as constitutive genes (geNorm M value <0.3), and the geometric mean of their Cq values was used to calculate the normalization factor as described by Vandesompele et al. (2002).

Cloning of CiFL1

Degenerated primers were designed based on the sequence of the FLC/MAF-like proteins found in Asteraceae databases. The RNA sample used for amplification was total RNA extracted from 7-day-old non-vernalized seedlings of the Crescendo cultivar. One pair of degenerated primers (5′-GAGRAARSTYGAAATDAAGCGGATC-3′ and 5′-AAGYTCTTSTTCWAGTTGTGTCAT-3′) allowed amplification of a 397 bp fragment that showed high homology with an EST from Cichorium endivia (accession number EL359100) and started 9-bp downstream of the start codon. This C. endivia EST was aligned with very high conservation with other ESTs from Asteraceae, and hence was used to design a second pair of primers (5′-TTGAAATTAAGCGGATCGAG-3′ and 5′-AGTCGAAGAAATTTAGCCA TTAAAGA-3′), which allowed cloning of the 3′ end and extension of the 397 bp amplicon to a 624 bp cDNA, named CiFL1.

Construction of 35S:CiFL1 transgenics

As a consequence of the cloning strategy described above, the cDNA sequence of CiFL1 lacks 9 bp at the 5′ end. As this region is highly conserved among all MADS box proteins (see Figure 3), an oligonucleotide (5′-ATGGGGCGA-3′) encoding the missing amino acids (MGR) was added to obtain a full-length cDNA. The resulting sequence was cloned in a binary vector derived from pPZP200 (Hajdukiewicz et al., 1994) between the constitutive CaMV 35S promoter and the octopine synthase terminator. The T-DNA was engineered to contain an herbicide (glufosinate ammonium, Basta S (R), AgrEvo, Diegem, Belgium) resistance cassette. The resulting binary vector was introduced into Agrobacterium tumefaciens strain GV3101, and Arabidopsis plants were transformed using the floral-dip method (Clough and Bent, 1998). Primary transformants and homozygous progeny were selected by treatment of two-leaf seedlings with 0.15% Basta S (R).


All statistical analyses were performed in R version 2.15.1. Data were tested for homogeneity of variances using the F statistic [var.test() function of R]. The standard t test or Welch's approximation were used in case of equal or unequal variances, respectively, using the t.test() function of R. Results for which the P value is less than 0.05 are considered significant.

Phylogenetic analyses

EMBOSS stretcher (http://www.ebi.ac.uk/Tools/psa/) was used for pairwise alignments (Figure 3b and Figure S1) and percentage identity calculations (Figure S2). Multiple alignments were performed in Seaview 4 (Gouy et al., 2010) using Clustal Omega (Sievers et al., 2011). Domain sequence logos and consensus sequences were computed from the alignment of annotated FLC- and MAF-like sequences from Brassicaceae species, namely Arabidopsis thaliana (AFU51419.1, AAO65320.1, AAO65315.1, Q9AT76.1, AAO65310.1 and Q9FPN7.2), Arabidopsis arenosa (AAZ92552.1 and AAZ92550.1), Arabidopsis halleri subsp. gemmifera (BAJ08315.1), Arabidopsis suecica (AAZ92553.1), Arabidopsis lyrata subsp. lyrata (XP_002873441.1), Sinapis alba (ABP96967.1), Eutrema wasabi (ADK92387.1), Eutrema halophilum (AAY34137.1), Raphanus sativus (BAK55645.1), Brassica oleracea var. capitata (AAP31677.1 and AAQ76273.1) and several sub-species of Brassica rapa (AAP31681.1, ABL96240.1, ADA70732.1, ABO40820.1, ABI30000.1, ACR25199.1, ABI29999.1 and ADA83718.1). Protein sequence logos were created using WebLogo 3.3 (Crooks et al., 2004).

For phylogenetic analysis of the MADS box at the amino acid level (Figure 3c), a list of MADS box sequences was compiled from the results of multiple BLASTP searches (Altschul et al., 1997) against the National Center for Biotechnology Information non-redundant (nr) protein database (E-value cut-off of 1) using the MADS box sequences of AtFLC, a diverging AtFLC BLAST hit (E-value of approximately 1.10−11) and the CiFL1 sequence as queries. The resulting list was restricted to MIKC proteins, and a maximum-likelihood tree was created using PhyML 3.0 (Guindon and Gascuel, 2003) with the LG+Γ4 model (Yang, 1993; Le and Gascuel, 2008). In contrast, analyses of the MADS box at the nucleotide level (Figure 3d and Figure S1) were performed using a two-step process. In the first step, a list of approximately 6500 full-length MADS box sequences was assembled from three successive TBLASTN searches (E-value cut-off of 1) against the National Center for Biotechnology Information nucleotide collection and EST databases, as described above. These sequences were used to create a large preliminary neighbor-joining tree from a matrix of observed distances. In a second step, sequences within the FLC-like sub-tree were used for alignment and tree building in Seaview 4 (Gouy et al., 2010) with PhyML 3.0 using the GTR+Γ4 model (Lanave et al., 1984). For all trees, statistical support was evaluated through 500 bootstrap replicates (Felsenstein, 1985), and final trees were drawn using FigTree (http://tree.bio.ed.ac.uk/software/figtree/).


The research on root chicory was funded by the Service Public de Wallonie DG03 (subventions D31-1062, D31-1123 and D31-1175) and by Chicoline, a division of Cosucra Groupe Warcoing SA. The authors are indebted to J. -M. Kinet (Université Catholique de Louvain, Louvain-la-Neuve, Belgium) for having initiated the project. The Arabidopsis work was funded by the Interuniversity Attraction Poles Program initiated by the Belgian Science Policy Office P6/33. F.B. is research fellow of the Fonds de la Recherche Scientifique-FNRS (F.R.S.-FNRS, Belgium). The authors are very grateful to D. Baurain (Department of Life Sciences, University of Liège, Belgium) for his guidance on the phylogenetic analyses of CiFL1, G. Bernier (Department of Life Sciences, University of Liège, Belgium) for critical reading of the manuscript, and N. Ormrod for English editing.


The Genbank accession number for the Cichorium intybus FLC-LIKE 1 sequence is HF549158.