Expression Patterns of Key Genes in the Photoperiod and Vernalization Flowering Pathways in Lilium longiflorum with Different Bulb Sizes

Lilium longiflorum is a wild Lilium, and its flowering transition requires a long period of cold exposure to meet the demand of vernalization. The responses of different sized bulbs to cold exposure and photoperiod are different, and the floral transition pathways of small and large bulbs are different. In this study, small and large bulbs were placed in cold storage for different weeks and then cultured at a constant ambient temperature of 25 °C under long day (LD) and short day (SD) conditions. Then, the flowering characteristics and expression patterns of key genes related to the vernalization and photoperiod pathways in different groups were calculated and analyzed. The results showed that the floral transition of Lilium longiflorum was influenced by both vernalization and photoperiod, that vernalization and LD conditions can significantly improve the flowering rate of Lilium longiflorum, and that the time from planting to visible flowering buds’ appearance was decreased. The flowering time and rate of large bulbs were greatly influenced by cold exposure, and the vernalization pathway acted more actively at the floral transition stage. The floral transition of small bulbs was affected more by the photoperiod pathway. Moreover, it was speculated that cold exposure may promote greater sensitivity of the small bulbs to LD conditions. In addition, the expression of LlVRN1, LlFKF1, LlGI, LlCO5, LlCO7, LlCO16, LlFT1, LlFT3 and LlSOC1 was high during the process of floral transition, and LlCO13, LlCO14 and LlCO15 were highly expressed in the vegetative stage. The expression of LlCO13 and LlCO14 was different under different lighting conditions, and the flowering induction function of LlCO9 and LlFT3 was related to vernalization. Moreover, LlFKF1, LlGI, LlCO5, LlCO16, LlSOC1 and LlFT2 were involved in the entire growth process of plants, while LlCO6, LlCO16 and LlFT1 are involved in the differentiation and formation of small bulblets of plants after the inflorescence stage, and this process is also closely related to LD conditions. This study has great significance for understanding the molecular mechanisms of the vernalization and photoperiod flowering pathways of Lilium longiflorum.


Introduction
Flowering is essential to most plants, and the flowering process is regulated by environmental and internal signals [1]. At present, five classic flowering pathways have been identified in Arabidopsis, among which vernalization and photoperiod pathways have been studied extensively. In Arabidopsis, the FLOWERING LOCUS T (FT) gene is the key regulator of flowering time [2,3], and the FT gene regulates flowering time by integrating both environmental and endogenous signals from multiple pathways. The FT gene belongs to the FT/TFL1(TERMINAL FLOWER1) gene family. The FT protein is synthesized in the leaves and then transported to the shoot apical meristem (SAM). It interacts with the b-ZIP transcription factor FLOWERING LOCUS D(FD) to form FT/FD protein complexes in the SAM [4], after which it induces flowering by promoting the expression of SUPPERSSOR OF OVEREXPRESSION OF CONSTANS1(SOC1), APETALA1(AP1), FRUITFULL(FUL) and LEAFY (LFY) [5]. To date, the FT gene family has been found in many plants, and the numbers and functions of FT genes vary in different plants [6][7][8][9][10][11][12][13]. Yan et al. (2021) identified three Lilium FLOWERING LOCUS T (LFT) family members, and LFT1 and LFT3 were highly expressed at the floral transition stage and may promote floral induction, while LFT2 may be involved in the differentiation of bulblets [14].
FLOWERING LOCUS C (FLC) is the most studied and the most important gene in the vernalization pathway; it encodes a potent MAD-box gene that inhibits blooming in Arabidopsis. FLC can directly inhibit or induce the FRIGIDA (FRI) gene to inhibit the expression of floral meristem formation genes. The expression of FT was inhibited by FLC at the beginning of winter, and the expression of FLC was downregulated after a period of vernalization [15]. The genes VERNALIZATION 1(VRN1) and VERNALIZATION 2 (VRN2) that encode zinc-finger proteins are also key genes for vernalization in Arabidopsis, and they can promote flowering by continuously inhibiting FLC expression [16,17]. In wheat, the flowering repressor is the VRN2 protein, and VRN1 is one of the genes inhibited by VRN2. VRN1 encodes a protein that promotes plant flowering [18,19]. In cereals, the VRN1 gene plays a crucial and dual role in the process of flowering; it can not only induce the expression of FT homologues by vernalization, but can also be used as a floral meristem identity gene [19,20]. Liu et al. (2014) identified LoVRN1, which can promote the early flowering and change the flower type in Arabidopsis, in the Oriental lily 'Sorbonne' [21].
In Arabidopsis, the CONSTANS/FLOWERING LOCUS T (CO/FT) module plays a central role in the photoperiodic flowering pathway [22]. In the photoperiodic regulation of flowering, GIGANTEA (GI) and FLAVIN-BINDING KELCH REPEAT F-BOX 1 (FKF1) respond to changes in light length and regulate the expression of CO [23,24]. The CO gene regulates flowering time by promoting transcription and affecting the expression of the downstream genes SOC1 and FT [5]. The CONSTANS/CONSTANS-LIKE (CO/COL) family members play a very important role in the CO-FT pathway [25,26]. COL proteins generally contain two conserved domains:the B-box(BBX) and CCT (CO, CO-LIKE and TOC1) structural domains [27]. CO/COL regulates flowering both in Arabidopsis and rice [28], and its homologous gene sequence and function are highly conserved and present in many varieties of plants [29]. Most of the COL genes are similar in structure during plant evolution, and all of them are regulated by circadian rhythms and photoperiods, but some of them also have different functions [28]. For instance, in addition to playing a negative regulatory role in the photoperiod, AtCOL3 also promotes the development of lateral roots [30], and AtCOL7 promotes stem branching in environments with a high ratio of red light/far-red light [31]. Recent studies have identified many CO homologues in other plants, such as PhalCOL in Phalaenopsis hybrida, MaCOL1 in banana, StCO in potato, VvCOL1 and VvCOL5 in grape, PbCOL8 in pear, and NnCOL5 in Nelumbo nucifera [28,[32][33][34][35][36][37][38].
Lilium longiflorum belongs to Liliaceae, and is a horticultural plant [39]. Flowering of plants is the key to their successful reproduction, and many plants need to go through a cold phase to break dormancy to bloom [40]. Studies have shown that the flowering transition of Lilium longiflorum requires a long period of cold exposure to meet the demand of vernalization. Vernalization is a necessary condition for the flowering of Lilium longiflorum, and it is also the main factor that affects the flowering period and quality [41]. Bulbs of Lilium longiflorum (cultivar 'White Heaven') grown under a constant temperature of 25 • C without vernalization formed only leaves and did not bloom for more than 15 months, which proved the necessity of vernalization of this cultivar [42]. During the entire growth process of the plant, from the cold exposure stage to flower initiation, all stages have the ability to induce the vernalization signal [43]. The study of morphological and biochemical markers showed that cold storage of bulbs can promote flowering [44]. In Lilium longiflorum 'White Heaven', cold exposure greatly promoted stem elongation and shortened the time from planting to flowering [42,45]. However, when the bulb sizes are different and under certain conditions or during special periods of development, the LD photoperiod can promote the flowering transition of lily and may replace cold storage as the factor inducing flower formation [41,43,46]. The latest research shows that small bulbs of nonvernalized Lilium longiflorum can bloom under long-day (LD) conditions, indicating that small bulbs of Lilium longiflorum flowered by the photoperiodic pathway, rather than vernalization [46]. However, the key genes involved in vernalization and photoperiod flowering pathways and the developmental stages at which they act warrant further studies.
In geophytes, such as lilies, tulips, lycoris, irises and onions, the size of the bulbs during planting affects their subsequent growth and development as well as their induction by vernalization and florogenesis [47][48][49][50]. When the bulb reaches the critical size required for flowering transition, the quality of flowering is usually positively correlated with bulb size [51]. The heavier the bulb mass is, the greater the ability of Lilium oriental and Leucocoryne to produce superior flowers [52]. At present, studies on lily bulbs of different sizes mainly focus on morphology, starch metabolism, soluble sugar and endogenous hormones [46,[53][54][55][56], and there are no reports on the effects of bulb size on flowering nor on the underlying molecular mechanisms. According to the research group's study, eight LfCOL genes were obtained from lily, the LfCOL5, LfCOL6 and LfCOL9 genes participate in flowering induction under long days, and the overexpression of those genes rescued the late flowering phenotype of the co-1 mutant in Arabidopsis [13]. Therefore, to elucidate the molecular mechanism of flowering under different conditions for different bulb sizes of Lilium longiflorum and to lay a foundation for exploring the relationship among lily bulb size, cold exposure time and photoperiod, the flowering characteristics and the expression patterns of key genes in the vernalization and photoperiod pathways were analysed in Lilium longiflorum.

Flowering Characteristics of Lilium Longiflorum under Different Conditions
The bulbs of all treatment groups were planted on the same day and the plants in all treatment groups emerged one week after planting. The flowering time and rate of different groups were statistically analyzed (plants that did not flower remained in the vegetative growth stage) ( Figure 1, Table 1). For the small bulbs without vernalization grown under SD conditions, the time from planting to visible flowering bud (VB) was 12 weeks. The time from planting to VB of the 3-, 5-, 7-and 9-week cold exposure treatment groups was 10 weeks ( Figure 1A). Under SD conditions, with increasing cold storage time, the flowering rate of small bulbs increased from 10% to 42.1% (Table 1). For the small bulbs planted under LD conditions, when the time of cold-storage increased, the time from planting to VB appearance decreased from 9 weeks to 5 weeks ( Figure 1A), and the flowering rate increased from 52.1% to 93.25%. Therefore, compared with SD conditions, LD conditions increased the flowering rate of nonvernalized small bulbs by 42.1% (52.1−10%) and the 9 weeks cold-exposed small bulbs by 51.15% (93.25−42.1%); moreover, compared with the nonvernalized small bulbs, 9 weeks of cold exposure increased the flowering rates of small bulbs by 32 nonvernalized large bulb group by 15.6% (48.3−32.7%) and the 9-week cold exposed large bulb group by 17% (100−83%). At the same time, compared with the nonvernalized smal bulbs, 9 weeks of cold exposure for large bulbs increased flowering rates by 50.3% (83−32.7%) under SD and 51.7% (100−48.3%) under LD conditions (Table 1). It should be noted that the flower number per plant of large bulbs could increase with increasing cold storage time and LD conditions ( Figure 1D).    For the small bulbs, the combination of LD and vernalization can significantly improve the flowering rate, and LD conditions played an important role in the flowering of both nonvernalized and cold-exposed small bulbs (Table 1). For the small bulbs planted under LD conditions, the time required for flowering was negatively correlated with the length of cold-storage time ( Figure 1A). Notably, the time from planting to visible flowering bud (VB) of the small bulbs without vernalization grown under LD conditions was 9 weeks, and the time from planting to VB was 10 weeks in the 3-, 5-, 7-and 9-week cold exposure groups under SD; therefore, LD conditions promoted the flowering transition of small bulbs more than cold exposure ( Figure 1A). In addition, there was no obvious difference in the plant height and the number of internodes among different weeks of cold storage in small bulbs grown under LD ( Figure 1B,C); however, the plant height and internode number under SD conditions were higher than those under LD conditions ( Figure 1B,C). For the flower number, all small bulb plants had one flower ( Figure 1D).
For the large bulbs planted under SD conditions, the time from planting to VB appearance was inversely proportional to the cold-storage time. When the time of cold-storage increased, the time from planting to VB appearance decreased from 11 weeks to 5 weeks ( Figure 1A); at the same time, the flowering rate increased from 32.7% to 83% (Table 1). For the large bulbs planted under LD conditions, with the increase in cold storage time, the weeks from planting to VB appearance decreased from 7 weeks to 4 weeks ( Figure 1A), and the flowering rate increased from 48.3% to 100% (Table 1). For the large bulbs, compared to plants grown under SD, LD conditions increased the flowering rate of the nonvernalized large bulb group by 15.6% (48.3−32.7%) and the 9-week cold exposed large bulb group by 17% (100−83%). At the same time, compared with the nonvernalized small bulbs, 9 weeks of cold exposure for large bulbs increased flowering rates by 50.3% (83−32.7%) under SD and 51.7% (100−48.3%) under LD conditions (Table 1). It should be noted that the flower number per plant of large bulbs could increase with increasing cold storage time and LD conditions ( Figure 1D).

Expression of Vernalization Related Genes in Scales during Cold Exposure in Lilium Longiflorum
The expression patterns of LlFLC, LlFRI3, LlFRI5, LlVRN1, LlVRN2, LlSOC1, LlFT1, LlFT2 and LlFT3 in scales during cold exposure are shown in Figure 2. The expression levels of LlFLC, LlFRI3, and LlFRI5 in both small and large bulbs were significantly decreased during the process of vernalization (4 • C/dark), and the decrease in expression levels in large bulbs was more obvious (Figure 2). For bulbs exposed to 25 • C/dark condition, the expression levels of LlFLC, LlFRI3, and LlFRI5 showed a slight downwards trend but no regular changes ( Figure 2). However, in all treatment groups, the expression of LlVRN1 and LlVRN2 showed opposite trends, and the expression of LlVRN2 gradually decreased, while the expression of LlVRN1 showed a significant upwards trend ( Figure 2). Among them, the expression of LlVRN1 in the cold exposed large bulb group increased the most ( Figure 2). The expression of LlSOC1 in the 25 • C /dark treatment groups showed no obvious trends ( Figure 2). At the same time, the expression of LlSOC1, LlFT1 and LlFT3 in both small and large bulbs constantly increased during the process of cold exposure (4 • C/dark). Moreover, it should be noted that the expression of LlFT3 increased significantly from 3 weeks during the process of vernalization (4 • C/dark), especially in the large bulbs ( Figure 2). In the 25 • C /dark treatment groups, there was an irregular change in the small bulbs. The expression of LlFT2 in all groups showed a downwards trend, and the decrease was more obvious in the large bulbs and the cold exposure (4 • C/dark) group ( Figure 2).

Expression of Key Flowering Genes in Leaves of Lilium Longiflorum after Planting
To gain insights into the expression patterns of key flowering genes in Lilium longiflorum, the expression patterns of LlCOLs and key flowering genes in leaves grown under SD/LD conditions were measured for cold exposed (4 °C/Dark/9 weeks) and nonvernalized (25 °C/Dark/9 weeks) large and small bulbs (Figures 3 and 4). According to our previous study [13,14], the critical point of floral transition for Lilium was accompanied by an increase in LFT1 or LFT3 expression, and LFT1 may be associated with photoperiod, while LFT3 may be associated with vernalization [14]. For the cold exposed (4 °C/Dark/9 weeks) large bulb groups, the time from planting to VB under LD and SD conditions was 4 weeks and 5 weeks, respectively. The LD condition had no obvious effect on flowering time, and the expression of LlFT3 increased significantly from 3 weeks during the process of vernalization. The results indicate that the cold exposed (4 °C/Dark/9 weeks) large bulbs had completed the floral transition before planting. However, for the small bulbs, the time from planting to VB of the cold-exposed (4 °C/Dark/9 weeks) group grown under LD and SD conditions was 5 weeks and 10 weeks, respectively, and the LD condition greatly shortened the flowering time; therefore, the cold-exposed (4 °C/Dark/9 weeks) small bulbs entered the floral transition stage after planting. Moreover, the nonvernalized (25 °C/Dark/9 weeks) small and large bulbs, both completed the floral transition after planting. On these grounds, the increased expression of LlFT1 or LlFT3 before visible bud (VB) appearance was considered the floral transition stage of the small bulb and nonvernalized (25 °C/Dark/9 weeks) large bulb group.
According to previous research results, LD conditions played an important role in the flowering of both nonvernalized and cold-exposed small bulbs. For the nonvernalized (25 °C/Dark/9 weeks) small bulbs, under LD conditions, the expression of LlFT1, LlSOC1, LlFKF1, LlGI, LlCO5, LlCO6, LlCO7, LlCO9, LlCO13, LlCO14 and LlCO16 was high during the process of flowering transition (Figures 3 and 4). Among them, the expression levels of LlCO7, LlCO13 and LlCO16 were particularly high before and during the VB stage. However, the expression levels of LlCO9, LlCO14 and LlCO15 were high in the vegetative stage ( Figure 4). For cold-exposed (4 °C/Dark/9 weeks) small bulbs, the expression of LlFT1, LlFT3, LlSOC1, LlFKF1, LlGI, LlCO6, LlCO7, LlCO13 and LlCO16 was upregulated under LD conditions at the flowering transition stage (Figures 3 and 4). For the LlCOL

Expression of Key Flowering Genes in Leaves of Lilium Longiflorum after Planting
To gain insights into the expression patterns of key flowering genes in Lilium longiflorum, the expression patterns of LlCOLs and key flowering genes in leaves grown under SD/LD conditions were measured for cold exposed (4 • C/Dark/9 weeks) and nonvernalized (25 • C/Dark/9 weeks) large and small bulbs (Figures 3 and 4). According to our previous study [13,14], the critical point of floral transition for Lilium was accompanied by an increase in LFT1 or LFT3 expression, and LFT1 may be associated with photoperiod, while LFT3 may be associated with vernalization [14]. For the cold exposed (4 • C/Dark/9 weeks) large bulb groups, the time from planting to VB under LD and SD conditions was 4 weeks and 5 weeks, respectively. The LD condition had no obvious effect on flowering time, and the expression of LlFT3 increased significantly from 3 weeks during the process of vernalization. The results indicate that the cold exposed (4 • C/Dark/9 weeks) large bulbs had completed the floral transition before planting. However, for the small bulbs, the time from planting to VB of the cold-exposed (4 • C/Dark/9 weeks) group grown under LD and SD conditions was 5 weeks and 10 weeks, respectively, and the LD condition greatly shortened the flowering time; therefore, the cold-exposed (4 • C/Dark/9 weeks) small bulbs entered the floral transition stage after planting. Moreover, the nonvernalized (25 • C/Dark/9 weeks) small and large bulbs, both completed the floral transition after planting. On these grounds, the increased expression of LlFT1 or LlFT3 before visible bud (VB) appearance was considered the floral transition stage of the small bulb and nonvernalized (25 • C/Dark/9 weeks) large bulb group.
According to previous research results, LD conditions played an important role in the flowering of both nonvernalized and cold-exposed small bulbs. For the nonvernalized (25 • C/Dark/9 weeks) small bulbs, under LD conditions, the expression of LlFT1, LlSOC1, LlFKF1, LlGI, LlCO5, LlCO6, LlCO7, LlCO9, LlCO13, LlCO14 and LlCO16 was high during the process of flowering transition (Figures 3 and 4). Among them, the expression levels of LlCO7, LlCO13 and LlCO16 were particularly high before and during the VB stage. However, the expression levels of LlCO9, LlCO14 and LlCO15 were high in the vegetative stage ( Figure 4). For cold-exposed (4 • C/Dark/9 weeks) small bulbs, the expression of LlFT1, LlFT3, LlSOC1, LlFKF1, LlGI, LlCO6, LlCO7, LlCO13 and LlCO16 was upregulated under LD conditions at the flowering transition stage (Figures 3 and 4). family members, except for LlCO6, the others showed consistent expression with that in the nonvernalized small bulbs (Figure 4).

Vernalization of Lilium Is a Complex Process
Flowering is a very complex process in which plants transition from the vegetative growth stage to the reproductive stage, and flowering is a critical stage for the development of plants. Many factors can affect the floral transition [22,57]. To ensure that the plant blooms under favorable conditions, many plants need to be cold exposed for a period of time, that is vernalization [16,17]. Vernalization is regulated by a series of genes, and the vernalization and flowering processes are greatly affected by the external environmental conditions [1]. In Arabidopsis, the key factor in the response to vernalization is FLC. FLC is a flowering suppressor that mainly encodes a class of MADS proteins. Other genes regulate flower formation by promoting or suppressing FLC, and FLC can inhibit the genic activity of flowering inducible factors such as FT and SOC1 [15,16]. When plants are under cold exposure at low temperatures, the high expression of SOC1 and FT inhibits the expression of the FLC gene and promotes vernalization and flowering in plants. Two other key vernalization genes, VRN1 and VRN2, also play an important role in the vernalization process. Studies in wheat have shown that three VRN homologous genes jointly control the response to vernalization. VRN1 has a certain homology with AP1/Cal/FUL in Arabidopsis [58]. VRN2 mainly encodes a class of TOC1(CCT), CO-like and CO structure Zinc finger proteins and is highly homologous with ZCCT (ZCCT1 and ZCCT2) gene sequences [59]. Moreover, VRN2 is thought to have the same function as FLC as a flowering inhibitor, but it is not evolutionarily relevant. Yan et al. (2006) found that the VRN3 gene in wheat was highly homologous to FT in Arabidopsis [19]. According to the results of this study, the flowering of Lilium longiflorum with large bulbs was mainly affected by vernalization. It is speculated that LlFLC can significantly inhibit the formation of flowers in large bulbs. After vernalization, the expression of LlFLC can be rapidly reduced to promote flowering (Figure 2), and the expression of LlVRN1 increased from the initial stage to the completion of the vernalization process ( Figure 2); therefore, LlVRN1 may inhibit the expression of LlFLC to promote flowering. However, the expression of LlVRN2 showed a similar decreasing trend during cold exposure as that of LlFLC, and we speculated that LlVRN2 could inhibit flowering by promoting the expression of LlFLC ( Figure  2). Previous studies have shown that the LoVRN1 gene plays an active role in the response to low-temperature-induced vernalization in Lilium Oriental hybrids 'Sorbonne', LoVRN1 can promote early flowering in Arabidopsis, and the LoVRN1 gene also plays a role in the change in lily flower type [21]. In addition, when Lilium responds to low-temperature conditions, FLM (FLOWERING LOCUS M) may replace FLC to inhibit the expression of the FT and SOC1 genes, thereby regulating the vernalization pathway; however, it is not clear how the MAF gene plays a role in the vernalization pathway [21]. Vernalization of Lilium is a complex process, and there may be other unique molecular regulatory mechanisms to complete vernalization in Lilium in addition to the similar regulatory mode of vernalization as that in Arabidopsis. The expression peaks of LlFT1, LlCO5, LlCO7, LlCO16 and LlSOC1 appeared in each group during the floral transition stage and inflorescence stage (Figures 3 and 4). For the small bulbs grown under LDs, the expression of LlGI was higher in the cold-exposed treatment group than in the nonvernalized group, and it was speculated that cold exposure may promote greater sensitivity of the small bulbs to LD conditions ( Figure 4). LlCO9 was highly expressed during the flowering induction period in nonvernalized group, and there was no peak expression level at the floral transition stage of cold-exposed plants, so we speculated that its flowering induction function was related to vernalization ( Figure 4). The LlCO13 expression level peaked at the flowering induction stage in the LD group, and its participation in flower formation may be closely related to light (Figure 4). The expression of LlFT3 was high or peaked in the floral transition stage in the vernalized or SD groups for all sizes of bulbs. The results suggest that LlFT3 was involved in flowering induction in other flowering pathways, such as the vernalization pathway and autonomic pathway, in addition to the photoperiod pathway ( Figure 3).
In addition to the floral induction stage, LlCO5 was highly expressed in the vegetative stage and the postflowering stage, suggesting that it was involved in the whole growth process of plants. The expression levels of LlCO13, LlCO14 and LlCO15 were high in the vegetative stage, so it is speculated that these genes are related to the vegetative growth of the plants. Among them, LlCO15 was highly expressed in each group of lilium at the seedling stage ( Figure 4). In the early vegetative stage of large bulbs of Lilium longiflorum plants, LlCO14 was highly expressed only in the plants that were grown under SD conditions, while it did not have a high expression level under LD conditions, so we speculated that its expression was closely related to the light (Figure 4). The expression of LlGI, LlFKF1 and LlSOC1 was also high in the vegetative stage and the postflowering stage, suggesting that they were involved in the growth process of plants (Figures 3 and 4). LlCO6, LlCO16 and LlFT1 showed peak expression levels after anthesis in the LD groups, and they may be involved in the differentiation and formation of small bulblets of plants after the inflorescence stage. The differentiation process of small bulblets may also be closely related to the LD conditions (Figures 3 and 4). The expression peaks of LlFT2 showed no obvious pattern in different groups at different stages, but it had peaks at the vegetative stage, floral transition stage and postflowering stage, showing that it was widely involved in the whole growth process of Lilium longiflorum (Figures 3 and 4).

Vernalization of Lilium Is a Complex Process
Flowering is a very complex process in which plants transition from the vegetative growth stage to the reproductive stage, and flowering is a critical stage for the development of plants. Many factors can affect the floral transition [22,57]. To ensure that the plant blooms under favorable conditions, many plants need to be cold exposed for a period of time, that is vernalization [16,17]. Vernalization is regulated by a series of genes, and the vernalization and flowering processes are greatly affected by the external environmental conditions [1]. In Arabidopsis, the key factor in the response to vernalization is FLC. FLC is a flowering suppressor that mainly encodes a class of MADS proteins. Other genes regulate flower formation by promoting or suppressing FLC, and FLC can inhibit the genic activity of flowering inducible factors such as FT and SOC1 [15,16]. When plants are under cold exposure at low temperatures, the high expression of SOC1 and FT inhibits the expression of the FLC gene and promotes vernalization and flowering in plants. Two other key vernalization genes, VRN1 and VRN2, also play an important role in the vernalization process. Studies in wheat have shown that three VRN homologous genes jointly control the response to vernalization. VRN1 has a certain homology with AP1/Cal/FUL in Arabidopsis [58]. VRN2 mainly encodes a class of TOC1(CCT), CO-like and CO structure Zinc finger proteins and is highly homologous with ZCCT (ZCCT1 and ZCCT2) gene sequences [59]. Moreover, VRN2 is thought to have the same function as FLC as a flowering inhibitor, but it is not evolutionarily relevant. Yan et al. (2006) found that the VRN3 gene in wheat was highly homologous to FT in Arabidopsis [19]. According to the results of this study, the flowering of Lilium longiflorum with large bulbs was mainly affected by vernalization. It is speculated that LlFLC can significantly inhibit the formation of flowers in large bulbs. After vernalization, the expression of LlFLC can be rapidly reduced to promote flowering (Figure 2), and the expression of LlVRN1 increased from the initial stage to the completion of the vernalization process ( Figure 2); therefore, LlVRN1 may inhibit the expression of LlFLC to promote flowering. However, the expression of LlVRN2 showed a similar decreasing trend during cold exposure as that of LlFLC, and we speculated that LlVRN2 could inhibit flowering by promoting the expression of LlFLC (Figure 2). Previous studies have shown that the LoVRN1 gene plays an active role in the response to low-temperature-induced vernalization in Lilium oriental hybrids 'Sorbonne', LoVRN1 can promote early flowering in Arabidopsis, and the LoVRN1 gene also plays a role in the change in lily flower type [21]. In addition, when Lilium responds to low-temperature conditions, FLM (FLOWERING LOCUS M) may replace FLC to inhibit the expression of the FT and SOC1 genes, thereby regulating the vernalization pathway; however, it is not clear how the MAF gene plays a role in the vernalization pathway [21]. Vernalization of Lilium is a complex process, and there may be other unique molecular regulatory mechanisms to complete vernalization in Lilium in addition to the similar regulatory mode of vernalization as that in Arabidopsis.

The Functions of COL Homologous Gene Family Members in Lilium Are Different
It has been reported that photoperiod induced flowering and bolting in plants often involve members of the COL gene family in multiple species, including dicotyledonous and monocotyledonous plants [60]. There are often many members of COL gene family in plants, and the number of the COL family members varies among different plants. There were 17 CO family members in Arabidopsis, 16 in O. sativa, 12 in soybean, 1 in pea, 3 in tomato and 17 in peanut [27,[60][61][62]. In the photoperiod pathway, the COL gene can induce the expression of the downstream genes FT and LFY and then promote plant flowering [5]. The function of CO homologous genes in different plants was also different, and some COL genes in Arabidopsis promoted flowering under LD conditions. The CO homologous gene Hd1 in rice inhibited flowering under LD conditions, but promoted flowering under SD conditions [63,64]. In legumes, ectopic expression of COL5 could promote the flowering of Arabidopsis co-mutants. The overexpression of GmCOL1a in soybean led to late flowering, while the GmCOL1b mutation promoted flowering [65,66]. Li et al. (2018) obtained eight LfCOLs homologous genes from Lilium × formolongi, and LfCOL5, LfCOL6 and LfCOL9 were upregulated in the induction phase of flower formation under LD conditions. At the same time, a diurnal expression pattern was found under both LD and SD conditions, and each of these three genes complemented the late-flowering phenotype of the Arabidopsis co-mutants and promoted flowering [67]. In this study, LlCO5 was highly expressed in the flowering induction stage, and may promote the flowering of Lilium longiflorum in all groups ( Figure 4). Moreover, LlCO9 s promotion of flowering was related to vernalization, the expression of LlCO13 was involved in LD conditions ( Figure 5), and LlCO15 was related to growth in the vegetative stage of Lilium (Figures 3 and 4). These results indicated that the functions of COL homologous genes in Lilium were diverse, and conditions such as cold exposure and LDs could affect the expression patterns and functions of COL family members in Lilium longiflorum.
eight LfCOLs homologous genes from Lilium × formolongi, and LfCOL5, LfCOL6 and LfCOL9 were upregulated in the induction phase of flower formation under LD conditions. At the same time, a diurnal expression pattern was found under both LD and SD conditions, and each of these three genes complemented the late-flowering phenotype of the Arabidopsis co-mutants and promoted flowering [67]. In this study, LlCO5 was highly expressed in the flowering induction stage, and may promote the flowering of Lilium longiflorum in all groups ( Figure 4). Moreover, LlCO9′s promotion of flowering was related to vernalization, the expression of LlCO13 was involved in LD conditions ( Figure 5), and LlCO15 was related to growth in the vegetative stage of Lilium (Figures 3 and 4). These results indicated that the functions of COL homologous genes in Lilium were diverse, and conditions such as cold exposure and LDs could affect the expression patterns and functions of COL family members in Lilium longiflorum.

The Functions of LFT Gene Family Members in Lilium Are Different
FT genes from different species can promote the flowering of Arabidopsis [68], and the function of FT is generally highly conserved in plants. However, in the evolution of plants, the FT homologous genes in different plants have also evolved species-specific functions [69]. In Arabidopsis, promoting flowering is the main function of the FT gene. Moreover, FT also maintains the stability of the flower meristem and promotes inflorescence development, FT promotes flowering and thus inhibits vegetative growth of plants [5], and the interaction of AtFT and AtTSF can regulate lateral branch growth after flowering [70]. AcFT1 and AcFT4 in onion promote and inhibit bulb formation, respectively, [11]. SD conditions could promote the expression of SP6A, which is an FT homologous gene in potato, and thus promote the formation of tubers [71]. SINGLE-FLOWER TRUSS(SFT) in tomato not only regulates flowering, but also promotes the formation of fruit [72,73]. In previous studies, we found 3 Lilium FT(LFT) homologous genes in Lilium, and the functions of different LFT family members were also different [14]. In this study, the expression of LlFT1 peaked in all groups before the emergence of visible flower buds (Figure 4), which suggested its involvement in the induction of Lilium flower formation, consistent with Li's research results in Lilium × formolongi [67]. However, LlFT2 is widely involved in the whole growth process of Lilium longiflorum, and LlFT3 may be involved in floral induction pathways other than the photoperiodic pathway (Figures 3-5). These results indicate that the functions of LFT gene family members in Lilium are different, and the molecular regulatory mechanism between LFT and upstream LCO gene family members in Lilium's flowering process needs to be further studied.

Vernalization and Photoperiod Significantly Promoted Flower Formation in Lilium Longiflorum
There is a strong interaction between vernalization and the photoperiod, and LD conditions can be equivalent to a 2-week-cold-storage for Lilium; that is, there is a '2-week effect'. Roh and Wilkins' study in 1977 showed that LDs could replace cold exposure before planting [43]. Lazare and Zaccais' research on Lilium longiflorum 'White Heaven' showed that Lilium longiflorum had a high requirement for vernalization, and the nonvernalized large bulbs of Lilium longiflorum remained in a vegetative growth state; however, the small bulbs of L.longiflorum could blossom under LD conditions with cold exposure. At the same time, vernalization and LD conditions could accelerate the flowering of Lilium longiflorum [46]. The results of this study showed that LDs could promote the formation of flowers in Lilium with different sized bulbs; among them, small bulbs of Lilium longiflorum were more affected by LDs, and large bulbs were mainly affected by vernalization, but all the treatment groups of Lilium longiflorum could bloom, including the different sizes of bulbs under SDs and the nonvernalized Lilium (Figure 1). These results are different from those of Lazare and Zaccai (2016), who showed that different cultivars of Lilium longiflorum need different conditions for flower formation. The 'White Heaven' Longiflorum hybrid has a greater need for vernalization, but the wild lily Lilium longiflorum used in this study may have a weak demand for vernalization. However, these results proved that vernalization and photoperiod significantly promoted flower formation in Lilium Longiflorum, and vernalization can significantly promote the flowering of large bulb plants, while LD conditions can significantly promote the flowering of small bulb plants. For Lilium longiflorum, both defoliation and lack of light can lead to bud abortions [74,75], indicating that insufficient accumulation of nutrients may block Lilium flower formation. In this study, LD conditions greatly promoted the flowering rate of small bulbs (Table 1). Prolonging the cold exposure time of Lilium can promote the plant's growth rate and floral transition [45], and the accumulation of carbohydrates can reduce the proportion of bud abortions [76]; however, the results of this study showed that under SD conditions, increasing the cold exposure time could not promote the flower formation of small bulbs. At the same time, the proportion of large bulbs showing flower formation was significantly higher than that of small bulbs, which also proved the importance of nutrient accumulation for Lilium flowering. Some of small bulb plants always stayed at the vegetative growth stage (Figure 1 and Table 1). The size and number of flowers are important for the commercial production of Lilium and will affect the quality of flowering in Lilium oriental hybrids if the bulb diameter is less than 6 cm [52]. The results of this study showed that all the small bulbs of Lilium longiflorum had only one flower, while the number of flowers of large bulb plants was higher than that of small bulb plants; therefore, for the production of Lilium longiflorum, if grown in a cold climate where vernalization can occur naturally, using large bulbs for production is better than using small bulbs. However, in warm climates, which usually have LD conditions, the flower formation rate of nonvernalized small bulbs is higher than that of large bulbs, and it is cheaper to use small bulbs for production.

Plants Growth Conditions and Samples Collection
Lily bulbs ('Qing chun') were harvested after the aboveground parts withered during autumn (October 2016; temperature average of 25 ± 2 • C), and then bulbs were divided into two groups according to their diameter: 1.5-2.5 cm small bulbs and 4-5 cm large bulbs. All bulbs were placed in moist peat in the dark, the cold exposure groups were placed at 4 • C for 3, 5, 7 and 9 weeks, and the nonvernalized groups were placed at 25 • C for 9 weeks. Samples of central scales (containing the shoot meristem) of bulbs were taken on the first day of every week at 14:00-16:00 before planting. After 9 weeks, all bulbs were planted on the same day and grown in incubators under SD (light/dark:8/16 h) or LD (light/dark: 16/8 h) conditions and a constant ambient temperature of 25 • C. Twenty plants of each group were selected for observation and recording.
During the planting period, the time from planting to visible flower buds, the number of internodes and plant height at visible flower bud, and the average number of flowers were recorded. After all the bulbs were planted, the middle leaves of cold-exposed (4 • C/Dark/9 weeks) and nonvernalized (25 • C/Dark/9 weeks) large and small bulb plants were sampled every week from 16:00 to 18:00, immediately frozen in liquid nitrogen and for further use. All the samples were collected from three biological replicates.

CDNA Preparation and QRT-PCR Analysis
Total RNAs was extracted from the leaves using kits from Aidlab Biotechnology Co., Ltd. (Beijing, China), total RNAs of the scales was extracted with an EasySpin Plus Complex Plant RNA Kit (RN53, Aidlab Biotechnologies Co., Ltd., Beijing, China) according to the manufacturer's instructions, and then reverse transcribed cDNA with ReverTra Ace qPCR RT Master Mix in conjunction with gDNA Remover (FSQ-301, Toyobo, Japan), according to the manufacturers' instructions. RNA quality was evaluated by 1% agarose gel electrophoresis and subsequently verified using an ND-1000 spectrophotometer (Nan-oDrop, Wilmington, DE, USA). The standards applied were 1.8 ≤ OD260/280 ≤ 2.2 and OD260/230 ≥ 1.8.
Specific qRT-PCR primers were designed via Premier 6 on the basis of the coding sequences of LlVRN1, LlFRI3, LlFRI5, LlFKF1, LlGI, LlCO5, LlCO6, LlCO7, LlCO9, LlCO13, LlCO14, LlCO15, LlCO16, LlFT1, LlFT2, LlFT3 and LlSOC1 obtained by our research group from the Lilium transcriptome, and the specific qRT-PCR primers of LlFLC and LlVRN2 were referred to Liu' study in 2014 [13,21]. qRT-PCR was performed on a CFX96 Real-time PCR Detection System (Bio-Rad, Hercules, CA, USA) in conjunction with a SYBR Premix EX Taq II Kit (TaKaRa, Beijing, China) according to the manufacturers' protocols, and the reaction system components were shown in Table S1. The PCR program was as follows: 3 min at 95 • C, followed by 40 cycles of 10 s at 95 • C, 30 s at the annealing temperature, and 15 s at 72 • C, all the samples were run in triplicate in 96-well optical reaction plates. The resulting data were analyzed by CFX Manager software (Bio-Rad). Lilium × formolongi EF-1a was selected as a reference gene to standardize the results because it was stably expressed at different developmental stages in Lilium according to our previous experiment [13]. The sequence information of the primers was listed in Table S2. The relative expression levels in all the figures were measured via qRT-PCR (y-axis), and the data represent the average of three biological replicates, with three technical replicates.
All statistical analyses in this paper were performed using SPSS (Statistical Product and Service Solutions) software.

Conclusions
The flowering time and rate of large bulbs were greatly influenced by cold exposure, and the vernalization pathway acted more actively at the floral transition stage. The floral transition of small bulbs was affected more by the photoperiod pathway. Moreover, it was speculated that cold exposure may promote greater sensitivity of the small bulbs to LD conditions.

Conflicts of Interest:
The authors declare no conflict of interest.