Functional diversification and molecular mechanisms of FLOWERING LOCUS T/TERMINAL FLOWER 1 family genes in horticultural plants

Flowering is an important process in higher plants and is regulated by a variety of factors, including light, temperature, and phytohormones. Flowering restriction has a considerable impact on the commodity value and production cost of many horticultural crops. In Arabidopsis, the FT/TFL1 gene family has been shown to integrate signals from various flowering pathways and to play a key role in the transition from flower production to seed development. Studies in several plant species of the FT/TFL1 gene family have revealed it harbors functional diversity in the regulation of flowering. Here, we review the functional evolution of the FT/TFL1 gene family in horticulture plants and its unique regulatory mechanisms; in addition, the FT/TFL1 family of genes as an important potential breeding target is explored.


Introduction
Flowering is an important stage in the life history of higher plants that includes the processes of flower bud differentiation, development, and the opening of flowers (Parmar et al. 2017;Xu et al. 2019). An optimal flowering is of great significance for plants to complete their life cycle under suitable environmental conditions (Su et al. 2019). Horticultural plants are critical components of agricultural production; they include fruits, flowers, vegetables, spices, medicinal, and aromatic plants (Karkute et al. 2017). Understanding how environmental factors influence the flowering transition of horticultural plants, as well as the underlying mechanisms involved, can help to improve the commercial value, lower production costs, and augment the annual production and seasonal supply of horticultural products (Higuchi 2018;Matsoukas et al. 2012).
Flowering time in the model plant Arabidopsis is regulated by integrating vernalization, temperature, photoperiod, hormones, age, autonomic pathways, and other floral transition signal transduction pathways (Srikanth and Schmid 2011;Cho et al. 2017). In 1936, M.K. Chailakhyan observed a type of flowering stimulator in Chrysanthemum that is produced in its leaves and transported to the shoot apical meristem (SAM) after photoperiod induction; it was designated 'florigen' (Chailakhyan and Krikorian 1975). This flowering element was later identified in Arabidopsis as a product of the FLOWERING LOCUS T (FT) gene (Kardailsky et al. 1999;Tsuji and Taoka 2014;Tsuji 2017). In Arabidopsis, TERMINAL FLOWER 1 (TFL1) of the FT/TFL1 family of proteins has been identified as a local floral inhibitor expressed in the SAM (Shannon and Meeks-Wagner 1991;Bradley et al. 1997). FT/TFL1 encodes a pair of flowering regulators that are homologous to phosphatidylethanolamine-binding proteins (PEBPs) (Ahn et al. 2006;Karlgren et al. 2011). The PEBP gene family in Arabidopsis includes six members: FT (Kim et al. 2013;Xu et al. 2012), TWIN SISTER OF FT (TSF) (Yamaguchi et al. 2005;Michaels et al. 2005;D' Aloia et al. 2011;Song et al. 2015), and MOTHER OF FT AND TFL1 (MFT) (Xi et al. 2010;Yoo et al. 2004) all promote flowering, whereas TFL1 (Kim et al. 2013), Arabidopsis Thaliana CENTRORADIALIS HOMOLOG (ATC) (Yoo et al. 2010;Huang et al. 2012) and BROTHER OF FT AND TFL1 (BFT) (Yoo et al. 2010) have function that differ from flowering.
FT protein was induced in Arabidopsis leaf vascular tissue phloem companion cells and transferred to the SAM by interacting with FT-INTERACTING PRO-TEIN 1 (FTIP1), QUIRKY (QKY), and SYNTAXIN OF PLANTS121 (SYP121) (Mathieu et al. 2007;Liu et al. 2012;Putterill and Varkonyi-Gasic 2016;Liu et al. 2019). Long-distance transmission of the FT protein is blocked by its interaction with negatively-charged phosphatidylglycerol (PG) at low temperatures (Liu et al. 2020;Susila et al. 2021). After being transported to the SAM, the FT protein forms a complex with the bZIP transcription factor FD and induces the expression of the floral meristem-identity genes APETALA1 (AP1) and FRUIT-FULL (FUL) (Abe et al. 2005;Wellmer and Riechmann 2010;Taoka et al. 2013). The interaction of environmental, endogenous, and hormonal signals precisely regulates the spatiotemporal expression of the FT gene in leaf phloem companion cells and the flowering in Arabidopsis (Fig. 1A). CONSTANS (CO) reflects the correspondence between external light signals and endogenous biological circadian clock, to activate the expression of FT at the right time to induce flowering (Imaizumi and Kay 2006;Song et al. 2015;Goralogia et al. 2017). Moreover, CYCLING DOF FACTORs (CDFs) directly bind to the proximal Block A region of the FT promoter to inhibit the transcription of FT (Imaizumi et al. 2005;Goralogia et al. 2017). Genes related to the circadian clock, temperature, and blue-light signals, such as GIGANTEA (GI) (Sawa and Kay 2011), BR ENHANCED Production 1 (BEE1) , PHYTOCHROME INTER-ACTING FACTOR 4 (PIF4) (Kumar et al. 2012), and CIB (cryptochrome-interacting basic-helix-loop-helix) (Liu et al. 2008), bind upstream from the transcription start site (TSS) of the FT gene, triggering its expression. TEM-PRANILLO (TEM) (Castillejo and Pelaz 2008), TAR-GET OF EAT 1 (TOE1), TOE2, SCHAFLMüTZE (SMZ), SCHNARCHZAPFEN (SNZ) (Mathieu et al. 2009), and SHORT VEGETATIVE PHASE (SVP)  respond to ambient temperature or photoperiod to directly repress FT expression. Further, several MADS transcription factors, namely FLOWERING LOCUS C (FLC), SVP, FLOWERING LOCUS M (FLM), and MADS AFFECTING FLOWERING (MAF), can inhibit transcription by binding to the first intron of FT at low temperatures or before vernalization (Luo et al. 2021). Other hormone signals also play a role in controlling the initiation of flowering. For example, ERF1, a key member of the ethylene signal transduction pathway, binds directly to the FT's promoter and inhibits its transcription (Chen et al. 2021). Furthermore, polycomb group (Pc-G) proteins reportedly mediate epigenetic gene regulation, which maintains the identity of the inflorescence and floral meristems after floral induction (Müller-Xing et al. 2014). The simultaneous occurrence of H3K27me3 at FT has also been demonstrated, using a sequential ChIP analysis (Jiang et al. 2008). The genes of PcG subunits, including EMBRYONIC FLOWER 2 (EMF2), EMF1, CURLY LEAF (CLF), MULTICOPY SUPPRESSOR OF IRA 1 (MSI1) and LIKE HETEROCHROMATIN PRO-TEIN 1 (LHP1), deposit H3K27me3 in the chromatin of FT to repress its expression (Jiang et al. 2008;Schatlowski et al. 2008;Mozgova and Hennig 2015;Merini and Calonje 2015). Unlike other PcG target genes in Arabidopsis, modification by H3K27me3 occurs in the promoter, coding region, and downstream region of the FT gene (Turck et al. 2007).
In Arabidopsis, the antagonism of TFL1 and floral meristem-specific genes controls flower initiation and the ensuing inflorescence structure. TFL1, which is expressed in the central region of the apical meristem but can move to the meristem layer L1, is required for movement into the SAM to regulate floral transition (Conti and Bradley 2007;Goretti et al. 2020). Once in the SAM, TFL1 interacts with the bZIP transcription factor FD via the 14-3-3 protein, and FT and TFL1 compete for binding FD to regulate the downstream floral meristem identification genes LEAFY (LFY), APETALA1 (AP1), and CAULIFLOWER (CAL), which maintains meristem indeterminacy (Zhu et al. 2020;Goretti et al. 2020) (Fig. 1B). Under long day (LD) conditions, SOC1 and AGL24 bind to TFL1's chromatin regions and directly activate its transcription in the SAM (Azpeitia et al. 2021). LFY binds to the TFL1 promoter and directly activates TFL1 transcription, a regulatory loop which ensures that flower formation occurs only when AP1/CAL levels are sufficiently high to repress TFL1 expression and trigger the genetic program required for flower development (Serrano-Mislata et al. 2017). The proteins encoded by the FT/TFL1 family of genes have small differences in conformation, giving them opposite functions in plants. Hanzawa et al. (2005) showed that changing only a single amino acid in the Arabidopsis TFL1 protein can render TFL1 to function as a floral activator, and vice versa.
Recent reports in gymnosperms indicate that the ancestor of FT functioned in a TFL1-like manner (Karlgren et al. 2011). For example, PaFTL1 and PaFTL2 in Norway spruce (Picea abies) both repress this conifer tree's growth, and their heterologous expression in Arabidopsis also delays the onset of flowering (Karlgren et al. 2011;Klintenäs et al. 2012). The data to date suggests that the function of FT and TFL1 diverged after the evolutionary separation in different plant species. Moreover, FT/TFL1 family genes are reportedly involved in other developmental processes and feature functional diversification in regulating flowering time in several species (Karlgren et al. 2011;Klintenäs et al. 2012). This review summarizes the functional diversification and molecular mechanisms of the FT/TFL1 family members in horticultural plants, to provide a timely reference for further research on these pivotal genes in horticultural plants.

FT/TFL1 family genes regulate flowering time in horticultural plants
In horticultural plants, the number of FT/TFL1 family genes varies yet they have conserved functions in the process of floral transition (Table 1). For example, in ornamental plants of the Asteraceae-the largest flowering plant family contains the greatest number of speciesthree FT-like genes have been identified in Chrysanthemum seticuspe: CsFTL1, CsFTL2, and CsFTL3, which are flowering inducers (Oda et al. 2012;Mao et al. 2016;Sun et al. 2017;Wang et al. 2020a). Mao et al. (2016) found that the archetypal and alternative splicing (AS) forms of CmFTL1 (C. morifolium cultivar 'Jimba') has the function of complementing the late-flowering phenotype of the Arabidopsis ft-10 mutant, and CmFTL1 can induce In vegetable crops, the FT-like protein StSP3D is essential for flowering in potato (Solanum tuberosum) (Navarro et al. 2011). Tomato (Lycopersicon esculentum) is the second most globally important vegetable crop (after potato), whose flowering time is jointly controlled by the flowering inducer SINGLE FLOWER  Jiang et al. 2013). SFT is an ortholog of FT-like that is expressed in mature leaves and systematically promotes flowering, while SP is the ortholog of TFL1-like that is instead expressed in young leaves and shoot tips, and inhibits flowering (Shalit et al. 2009). Recently, FTL1, which regulates flowering time in tomato, was located and sequenced through map-based cloning. In a Chinese continuous-flowering rose plant cultivar, the TFL1-like gene RoKSN is a flowering suppressor whereas RoFT is a floral inducer (Iwata et al. 2012;Otagaki et al. 2015). The insertion of a retrotransposon in RoKSN inhibits RoKSN expression in roses, thereby facilitating their continuous flowering (Randoux et al., 2014, b). Collectively, these reports suggest the function of FT/TFL1 family genes is generally conserved in horticultural plants.

Functional diversification of FT/TFL1-like in horticultural plants
The functions of proteins encoded by homologous FT/ TFL1-like genes are not entirely conserved in horticultural plants, in that they show functional diversification in regulating flowering time (Table 1).
Homologous genes of FT could contribute to inhibiting flowering. In vegetable crops, the two FT-like homologous genes BvFT1 and BvFT2 in sugar beet (Beta vulgaris) function antagonistically in flowering. Under non-vernalized or SD conditions, the flowering inhibitor BvFT1 inhibits flowering by limiting the expression of the flowering-inducing factor BvFT2 (Pin et al., 2010). In cucumber (Cucumis sativus), the structural types in the upstream region (UR) of CsFT have differential effects on flowering induction; the 'short-1' UR CsFT and 'short-2' UR CsFT accelerate the onset of flowering, whereas the 'long' UR CsFT delays flowering (Wang et al., 2020b). Four FT-like homologous genes have been identified in tomato: SP3D/SFT has a florigen function, whereas SP5G, SP5G2, and SP5G3 are characterized by flowering inhibitory activity (Cao et al., 2016). AcFT4 in onion (Allium cepa) and both StSP5G and StSP5G-like in potato are also able to inhibit flowering (Navarro et al., 2011;Lee et al., 2013). In ornamental plants, three PEBP genes were isolated in tulip (Tulipa gesneriana): TgFT1, TgFT2, and TgFT3. Overexpression of TgFT2 in Arabidopsis resulted in an early-flowering phenotype, while TgFT1 and TgFT3 overexpression resulted in a late-flowering phenotype (Leeggangers et al., 2018). The PhFT6 in Phalaenopsis and HaFT1 in sunflower (Helianthus annuus) can also repress their flowering (Li et al., 2014;Blackman et al., 2010). The two FT-like genes PtFT1/PtFT2 in poplar, a woody perennial species, also have opposing flowering regulatory functions (Hsu et al., 2011;Mohamed et al., 2010). PtFT1 has a florigen function, whereas PtFT2, it induced by LDs and high temperature, reduces the level of GA via the GA 13-hydroxylation pathway and maintains the vegetative growth of poplar to preclude flowering (Gómez-Soto et al., 2022).
In addition to regulating flowering time, members of FT/TFL1-like genes are involved in a variety of other processes in horticultural plants. CsTFL1 inhibits determinate growth and terminal flower formation in cucumbers (Zhao et al., 2018;Wen et al., 2019;Njogu et al., 2020). Navarro et al. (2011) reported that overexpressing the FT homologous gene Hd3a in potato enabled it to grow more tubers than the wild type, and that the endogenous gene StSP6A also had a similar function, thus indicating that FT promotes tuber formation. In onion, LDs induced the downregulation of AcFT4 expression but the upregulation of AcFT1 expression, which promoted the formation of bulbs and increased the yield (Lee et al., 2013). In tomato sft mutants, the inflorescence differentiated into only one flower, the sepals were enlarged, and leaves have excess intercalary leaflets; however, the leaves became smaller blades and lack folioles after the overexpression of SFT (Shalit et al., 2009;Lifschitz et al., 2014). In Dendrobium Chao Praya Smile, DoFT-RNAi transgenic lines also displayed abnormal inflorescence development and delayed pseudobulb formation, suggesting that DOFT may have evolved with unknown functions related to the regulation of storage organs and flower development . PtFT2 promotes vegetative growth and shoot dormancy in poplar trees (Mohamed et al., 2010); similarly, FT and CEN are involved in the regulation of kiwifruit plant growth by integrating developmental and environmental signals (Varkonyi-Gasic et al., 2013). Taken together, these reports show that the functions of members of the FT/TFL1 gene family have evolved dynamically over the course of horticultural plants' evolution.

Regulation of FT/TFL1 family genes in horticultural plants
The photoperiodic pathway is the most important and most conserved of the floral induction pathways, and some of the key loci and mechanisms are shared even among distantly related plant species, whereas others are not conserved and give rise to crucial species differences (Matsoukas et al., 2012). The autumn flowering chrysanthemum cultivars are short day plants that require a repeated SD photoperiod for successful flowering, because CsFTL3 expression increases with such repeated SDs before successful flowering occurs, but their vegetative growth can be strictly maintained under LD or night-break (NB) conditions. When SDs switch to LDs before the involucre-forming stage, those plants do not initiate florets on the apical receptacle, or their capitulum development is strongly suppressed (Higuchi, 2018;Nakano et al., 2019). Recent studies have revealed the transcriptional regulation mechanism of the FT/ TFL1 family genes in chrysanthemum (Fig. 2). ClCRY2 facilitates floral transition in C. lavandulifolium by finetuning the expression of circadian clock-related genes, such as the downregulation of LHY and overexpression of GI (Yang et al., 2018). By downregulating CsFTL3 and CsAFT, CsLHY-SRDX induced a photoperiod-insensitive floral transition (Oda et al., 2017). The transcription of another circadian-clock-related gene, CsGI, has been shown to increase the necessary night length for blooming, chiefly by maintaining lower levels of CsAFT (Oda et al., 2020). In chrysanthemums, CsPHYB-mediated light signaling upregulates CsFTL3 but downregulates CsAFT to determine their obligate photoperiodic blooming response (Higuchi et al., 2013). Furthermore, gibberellins function critically in floral induction in response to LDs (Porri et al., 2012). CmBBX24 inhibits the expression of CmFTL3, which regulates flowering primarily through effects on the GA pathway under LDs (Yang et al., 2014). Recently, the role of NF-Y proteins in the aging pathway in chrysanthemum was identified, in that CmNF-YB8 influences flowering time by directly upregulating the expression of cmo-MIR156 in the aging pathway (Wei et al., 2017). More recently, the CO homologous protein CmBBX8 was discovered to target CmFTL1 for flowering regulation in chrysanthemum (Wang et al., 2020a).
Nevertheless, the functions of CO1 and CO2 in poplar do not overlap with those in Arabidopsis. The growth of CO2 RNAi transgenic poplar stopped when LDs transitioned to SDs, and its shoots formed earlier. Overexpression of CO1 and CO2 in poplar did not induce the upregulated expression of FT2 under SD conditions, and its timing of flowering and bud formation did not change (Hsu et al., 2012). Another SD-dependent FT2 inhibition pathway mediated by LHY2 was recently discovered in poplar. Under SD conditions, LHY2 is induced to express and directly bind to the homeopathic element at the 3′ end of FT2 to inhibit its expression, resulting in the arrested growth of poplar (Fig. 3A). But under LD Some FT homologs regulate flowering and aspects of development in response to temperature except for photoperiod in horticultural plants. In sugar beet, the expression of the flowering inhibitor BvFT1 was inhibited by both vernalization and LD conditions and BvFT2 was induced under LDs (Pin et al., 2010) (Fig. 3B). In strawberry, FvFT1 was specifically upregulated in mature leaves and this promoted the upregulation of FvSOC1 in shoot tips, which activated FvTFL1 expression to inhibit flowering under LDs (Mouhu et al., 2013;Rantanen et al., 2014). Furthermore, FvTFL1 was regulated by a temperature-dependent pathway independent of photoperiod-dependent regulation (Rantanen et al., 2015). Floral transition of lily (Lilium longiflorum) is also induced by low-temperature conditions and is not regulated by photoperiod. In lily, LlFT, as a flowering activator, is significantly induced by a period of lowtemperature treatment and this promoted flowering; however, without the vernalization treatment, overexpression of LlFT also led to a bloom, indicating that LlFT is the main regulatory factor controlling flowering in the vernalization pathway (Lazare and Zaccai et al., 2016). After switching from vegetative to reproductive growth, LlFT expression was further reduced in floral meristems and small flower buds. Therefore, LlFT is thought to be involved in switching the meristem to a flowering state during vernalization, but it does not act as a flowering inducer (Leeggangers et al., 2018). Further, the flowering regulation of Narcissus tazetta and tulip is induced by high temperature and does not depend on the photoperiod and vernalization pathways. In narcissus, high temperature induces the transcription of Narcissus FLOWERING LOCUST1 (NFT1), which promotes the expression of the downstream LFY homolog gene NLF and induces flowering (Li et al., 2013;Noy-Porat et al., 2013). The floral transition of tulip occurs in the bulb, for which high temperature induces the expression of TgFT2 and inhibits the expression of TgTFL1, which then induces the floral transition (Leeggangers et al., 2018) (Fig. 3C). On the contrary, the chrysanthemum flowering is severely delayed by high temperature during the summer, when the reduction of CsFTL3 expression at high temperatures are involved in flowering's retardation in C. seticuspe (Higuchi, 2018). Interestingly, exogenous ethylene induces the upregulated expression of FT-like and AP1-like genes, which promotes transition to flower formation in pineapple (Liu and Fan, 2016;Liu et al., 2018) (Fig. 3D), this starkly differing from their inhibitory effect on Arabidopsis flowering (Chen et al., 2021). These findings indicate that not only the functions of members of the FT/TFL1 family of genes, but also their upstream regulators, have evolved drastically and in cases also divergently in horticultural plants.
The transport of florigens in horticultural plants has also been studied. In Cucurbita moschata, LD-induced transport of FT proteins from its leaves to shoot tips promoted the transition into flowering (Lin et al., 2007). Moreover, floral promotion via the graft transmission of FT has been demonstrated in woody plants. For a recent example, when the scion of JcFT-RNAi transgenic Jatropha curcas was grafted onto SUC2:JcFT rootstock, FT protein was transported into the scion which promoted the transition into flowering, whose efficacy depended on the length of the scion (Tang et al., 2022). In trifoliate orange, its early flowering was induced in the transgenic tomato as well as trifoliate orange plants transformed with ToFT. However, the rootstocks of transgenic trifoliate orange could not induce flowering of grafted wild-type (WT) juvenile scions because of their low accumulation of total FT protein (Wu et al., 2022). That finings suggests the expression of FT must reach a certain threshold to induce flowering. A TFL1 homolog (RoKSN) in rose was found to be immobile, precluding its transmission via grafting experiments (Randoux et al., 2014, b). Yet when a WT chrysanthemum plant was grafted onto the CsTFL1-ox stock, the flowering of the WT scion was delayed vis-à-vis the WT/WT grafting . Spatial expression patterns of CsTFL1 showed that it was mainly expressed in shoot tips, with low expression levels in leaves Haider et al., 2020). These results suggest that CsTFL1 probably can move long distances through a grafting union as a floral repressor, to systemically regulate an indeterminate apical meristem. Because FT participates in vesicle trafficking (Liu et al., 2020), whether the transport of TFL1 occurs via a similar way awaits investigation.

Outlook
In summary, many studies of diverse horticultural plants have revealed the conserved functioning of members of the FT/TFL1 gene family, which have evolved dynamically over the course of horticultural plant evolution. Moreover, amino acid substitutions in FT/TFL1 family genes in Arabidopsis and horticultural plants such as sugar beet could cause a conversion in functionality, from having repressor activity to becoming a floral activator and vice versa (Ho and Weigel, 2014;Pin et al., 2010). A single base deletion or the products of a premature stop codon in TFL1 gene in strawberry facilitates their continuous flowering (Koskela et al., 2012). In cucumber and domesticated tomato, the short upstream region of CsFT and mutations in the cis-regulatory region of antiflorigen SP5G hasten their onset of flowering, respectively (Soyk et al., 2017;Wang et al., 2020b). These results suggest FT/TFL1 family genes are elite editing targets for manipulating gene structure, to change key flowering characteristics of horticultural plants, using genome editing technology, which is a powerful and precision-breeding approach, although there are legal/ethical concerns (Gao, 2021).
Because FT/TFL1 family genes integrate multiple regulatory pathways, such as photoperiod, vernalization, and ambient temperature pathway, to govern flowering, not only the functions of their members but also their upstream regulators have drastically evolved in horticultural plants. With ongoing global warming, the rise in ambient temperature is often accompanied by a greater concentration of carbon dioxide (CO 2 ), which is conducive to the accumulation of photosynthetic products (sugar and starch) in plants. This increase in CO 2 is apt to cause changes in tissues' sugar status or directly drives FT/TFL1 to regulate flowering (Jagadish et al., 2016). Although studies have found that FT can mediate nitrogen's control of flowering, its regulatory mechanism is still not well understood (Gras et al., 2018;Zhang et al., 2021). Therefore, elucidating in detail how the regulation mechanisms of FT/TFL1 family genes may respond to various environmental and endogenous stimuli would promote the development of an efficient and energysaving approach to regulate flowering. Due to the rapid adoption and spread of genomic sequencing technology applied to horticultural plants, genomic resources are becoming increasingly available. This combined with other techniques, namely high-throughput phenotyping, genomic selection, and gene function analysis, will enable us to obtain detailed knowledge of the FT/TFL1 gene family, so as to modify their action to meet the increasing demand for horticultural products in the future.