The Role of EjSOC1s in Flower Initiation in Eriobotrya japonica

The MADS-box transcription factor SUPPRESSOR OF OVEREXPRESSION OF CONSTANS1 (SOC1) integrates environmental and endogenous signals to promote flowering in Arabidopsis. However, the role of SOC1 homologs in regulating flowering time in fruit trees remains unclear. To better understand the molecular mechanism of flowering regulation in loquat (Eriobotrya japonica Lindl.), two SOC1 homologs (EjSOC1-1 and EjSOC1-2) were identified and characterized in this work. Sequence analysis showed that EjSOC1-1 and EjSOC1-2 have conserved MADS-box and K-box domains. EjSOC1-1 and EjSOC1-2 were clearly expressed in vegetative organs, and high expression was detected in flower buds. As observed in paraffin-embedded sections, expression of the downstream flowering genes EjAP1s and EjLFYs started to increase at the end of June, a time when flower bud differentiation occurs. Additionally, high expression of EjSOC1-1 and EjSOC1-2 began 10 days earlier than that of EjAP1s and EjLFYs in shoot apical meristem (SAM). EjSOC1-1 and EjSOC1-2 were inhibited by short-day (SD) conditions and exogenous GA3, and flower bud differentiation did not occur after these treatments. EjSOC1-1 and EjSOC1-2 were found to be localized to the nucleus. Moreover, ectopic overexpression of EjSOC1-1 and EjSOC1-2 in wild-type Arabidopsis promoted early flowering, and overexpression of both was able to rescue the late flowering phenotype of the soc1-2 mutant. In conclusion, the results suggest that cultivated loquat flower bud differentiation in southern China begins in late June to early July and that EjSOC1-1 and EjSOC1-2 participate in the induction of flower initiation. These findings provide new insight into the artificial regulation of flowering time in fruit trees.


INTRODUCTION
Plant evolution has resulted in a variety of endogenous and exogenous factors that form a complex and sophisticated regulatory network to accurately respond to internal and external signals and integrate them to promote blooming at the most favorable time. The molecular genetic mechanisms at play in annual flowering plants, such as Arabidopsis thaliana, involve multiple regulatory pathways, including photoperiod, age, autonomic, vernalization, and gibberellin pathways (Moon et al., 2005;Amasino, 2010;Srikanth and Schmid, 2011;Teotia and Tang, 2015). These pathways precisely regulate flowering in Arabidopsis through major integrated genes such as FLOWERING LOCUS T (FT), LEAFY (LFY), and SUPPRESSOR OF OVEREXPRESSION OF CONSTANS1 (SOC1).
MADS-box genes are a key components of flower development networks. In addition to the MADS-box domain, MIKC C -type MADS-box genes contain three other domains, the I-domain, K-box and C-terminal domain; although the MADS-box is highly conserved, the degree of conservation of the I-domain and C-domain is relatively low (Theissen et al., 1996;Parenicova et al., 2003;Vandenbussche et al., 2003;Smaczniak et al., 2012;Chen et al., 2017). SOC1 is a member of the MIKC C -type gene family and encodes a type II MADS-box protein that contains the highly conserved MADS-box, K-box, and a C-terminal SOC1 motif (Vandenbussche et al., 2003). SOC1 plays a vital role in regulating plant development and flower organogenesis by integrating photoperiod, age, and gibberellin signals (Parcy, 2005;Lee and Lee, 2010;Teotia and Tang, 2015).
Loquat (Eriobotrya japonica Lindl.) is an evergreen fruit tree belonging to the family Rosaceae that is cultivated mainly in tropical and subtropical regions. In Rosaceae, flower initiation and flowering typically occur in different years in species including apple, pear, plum, strawberry, and raspberry . However, flower bud initiation and flowering occur within the same year in loquat, with the former generally occurring from July to September in China (Lin, 2007) and the latter mainly from October to January; there is also slight variability depending on the cultivar and environment. To date, 26 Eriobotrya species have been identified, and each wild species has a different flowering time that includes the months of November to June of the next year for some (Lin, 2017). For example, cultivated loquat (E. japonica Lindl.) blooms in fall or early winter, whereas E. deflexa Nakai blooms from May to June (Gu and Spongberg, 2003).
In this study, the flower initiation time of cultivated loquat ("Jiefangzhong") in Southern China was confirmed. Two SOC1-like genes from cultivated loquat were identified and named EjSOC1-1 and EjSOC1-2. To elucidate their roles in regulating flowering time in loquat, their expression patterns and subcellular localizations were analyzed. In addition, we examined their function using transgenic Arabidopsis and explored the effects of short-day (SD) and GA 3 treatments on bud differentiation.

Plant Materials and Growth Conditions
Material was collected from 12-year-old "Jiefangzhong" loquat (E. japonica Lindl.) trees grown under natural conditions in the loquat germplasm resource preservation garden, South China Agricultural University, Guangzhou, China (N23 • 09 N,113 • 20 E). The trees used in the experiments were grafted, and they had grown to the flowering stage. Leaf and shoot apical meristem (SAM) tissues were randomly sampled from three sites on the trees, and tissues was collected at 16:00. Wild-type A. thaliana ecotype Col-0 and the soc1-2 mutant were used for genetic transformation. Nicotiana benthamiana was grown for transient expression. Arabidopsis and Nicotiana were grown under long-day conditions (16 h light/8 h dark) at 22 • C.

RNA Isolation, cDNA Preparation, Gene Isolation, and Sequence Analysis
Frozen mature loquat leaves or other tissues were ground to a powder in a mortar with liquid nitrogen. Total RNA was extracted using EasySpin Plus (Aidlab, China) and digested with recombinant RNase-free DNase I (Aidlab, China). Firststrand cDNA was generated from loquat leaf RNA using the PrimeScript TM RT (TAKARA, Japan) reagent kit and gDNA Eraser (TAKARA, Japan), the experiment was proceeded according to the manufacturer's instructions.
The full-length coding sequences of EjSOC1-1 and EjSOC1-2 were obtained from the completed loquat de novo genome sequencing project, which has not yet been published. The two sequences were isolated from mature loquat leaf cDNA using Phusion DNA Polymerase (TAKARA, Japan). The gene-specific primers used for cloning were listed in Supplementary Table S1. Alignment of the deduced protein sequences was performed using ClustalX 2.0.12 and GeneDoc 2.7. Phylogenetic trees were constructed with MEGA 6.06 using the Neighbor-Joining (N-J) method with 1,000 bootstrap replicates.

Gene Expression Analysis
Primers for qPCR were designed using Primer 5 software, and their specificity was confirmed by melting curve analysis and sequencing. qPCR was carried out in triplicate using a LightCycler R 480 system (Roche, United States) with iTaq TM universal SYBR Green Supermix (Bio-Rad, United States). The relative expression levels of target genes were evaluated using the Ct (cycle threshold) method. β-Actin was used as an internal reference gene for loquat (Shan et al., 2008). AtPP2AA3 (AT1G13320) was used as an internal control for Arabidopsis (Hong et al., 2010). Semi-quantitative reverse transcription PCR (RT-PCR) was used for detecting exogenous gene expression in transgenic Arabidopsis lines. The primers used for RT-PCR were identical to the cloning primers (removal of the stop codon). The primers used are listed in Supplementary Table S2.

Arabidopsis Transformation
35S:EjSOC1-1-HA and EjSOC1-2-HA were introduced into Agrobacterium tumefaciens GV3101::psoup and then transformed into Arabidopsis Col-0 using the floral dip method (Zhang et al., 2006). Transgenic lines were screened on soil by Basta. For each construct, more than 10 independent transgenic lines were screened out, and two homozygous T3 generation lines of each genotype were used for checking ectopic gene expression.

Short-Day and GA 3 Treatments
An awning (Supplementary Figure S1) was set up to cover the tree to provide 8 h (10:00-18:00) of natural light and 16 h of darkness (18:00-10:00 [the next day]) each day. Control trees were grown under normal conditions. The experimental period lasted from May 18th to August 10th.
For GA 3 treatment, trees were sprayed with 300 mg L −1 GA 3 (Guangzhou DingGuo Biology Company, China) aqueous solution containing 0.1% (v/v) phosphoric acid and 0.025% (v/v) Triton X-100 as a surfactant. Control trees were sprayed with a solution containing only 0.1% (v/v) phosphoric acid and 0.025% (v/v) Triton X-100. All leaves and top buds were sprayed every 2 weeks from May 18th to August 10th.

Subcellular Localization Analysis
Agrobacterium-mediated transient transformation of N. benthamiana leaves (Sparkes et al., 2006) was used to observe the subcellular localization of EjSOC1-1 and EjSOC1-2. Green fluorescent protein (GFP) fluorescence signals were detected using a fluorescence microscope Observer D1 (Zeiss, Germany). A GFP-free construct was used as a negative control.

Data Analysis
Significant differences between data were evaluated by Student's t-test. Calculations were carried out using GraphPad Prism 6 software.

Observation of Flowering and Determination of Flowering Initiation in Loquat
Continuous year-round observation of loquat SAM development was conducted. The panicle of "Jiefangzhong" loquat in Guangzhou was clearly visible from the end of August to early September (Figures 1A,B). In addition, analysis of paraffin-embedded sections of the SAM from June to September revealed no obvious inflorescence primordium formation in apical tissue before June 23rd, with multiple inflorescence primordia in the bottom of panicle appearing on July 7th ( Figure 1A).
Furthermore, the expression levels of the floral meristem identity genes EjAP1-1, EjAP1-2, EjLFY-1, and EjLFY-2 at different developmental stages of apical tissues were analyzed by qPCR. The results showed a high level of expression for both EjAP1s and EjLFYs began on June 23rd that was maintained from July to September (Figure 1C), except for EjLFY-2, which maintained a peak until November. AP1 and LFY determine flower meristem characteristics and are key genes for flower induction and morphology (Lohmann et al., 2001). These results indicate that "Jiefangzhong" loquat flower bud differentiation in Guangzhou begins in late June to early July.

Cloning and Identification of SOC1-Homologous Genes
We cloned two genes homologous to SOC1, EjSOC1-1 and EjSOC1-2, using unpublished loquat genome sequence data. EjSOC1-1 and EjSOC1-2 CDSs are 642 and 648 bp and encode 213 and 215 amino acids, respectively (Supplementary Figure S2). Their sequences are highly similar, with nucleotide sequence identity of 93.36%. The predicted protein amino acid sequences of EjSOC1-1 and EjSOC1-2 are similar to those of other SOC1 orthologs from apple, soybean, rapeseed and Arabidopsis (Figure 2A). Sequence analysis showed that EjSOC1-1 and EjSOC1-2 harbor highly conserved MADS-box, K-box and SOC1-motif domains (Figure 2A).
Based on phylogenetic analysis of EjSOC1s and other plant SOC1 sequences, EjSOC1s and the other SOC1s from Rosaceae grouped into a large clade, with apple sequences forming a small clade with a high genetic relationship to the large clade ( Figure 2B). EjSOC1s show the highest sequence similarity to MdSOC1 homologs (97.18% identity for EjSOC1-1 and MdSOC1A and 97.21% identity for EjSOC1-2 and MdSOC1B) (Figure 2A). These results confirm that EjSOC1-1 and EjSOC1-2 are MADS-box genes and SOC1 orthologs in loquat.

Expression Analysis of EjSOC1s in Different Tissues
To understand the potential function of EjSOC1-1 and EjSOC1-2 in loquat, we employed qPCR to examine the expression patterns of EjSOC1-1 and EjSOC1-2 in various tissues of "Jiefangzhong" loquat, including roots (from rootstock), leaves, shoots, leaf buds, flower buds, flowers, and fruits ( Figure 3A). EjSOC1-1 and EjSOC1-2 were mainly expressed in vegetative organs, and for both, the highest expression was observed in flower buds. In particular, the expression level of EjSOC1s in early flower buds was significantly higher than that in blooming flowers, with scant expression in fruits ( Figure 3B). These results suggest that EjSOC1s participate in the development of vegetative organs and flower initiation.

Expression of EjSOC1s During the Growth and Development of Loquat
To further investigate the functions of EjSOC1-1 and EjSOC1-2 during vegetative and reproductive developmental processes, we  examined their expression at different developmental stages of leaves, buds and flowers, as well as leaves of different maturities in the same period (Supplementary Figure S3), using qPCR.
In leaves, the tendency of EjSOC1-1 expression was similar to that of EjSOC1-2: their expression began to increase on June 23rd and reached the highest level by July 14th (Figure 4A). In addition, there was no significant difference in expression between the genes in mesophyll tissue or in veins ( Figure 4C).
With regard to the SAM in different periods, the expression levels of EjSOC1-1 and EjSOC1-2 began to increase sharply on June 23rd and reached the highest level around July 7th ( Figure 4D). Moreover, expression of these genes gradually decreased as flower bud development progressed (Figure 4E).
For different flower parts (Supplementary Figure S4), EjSOC1-1 and EjSOC1-2 showed relatively high expression levels in receptacles, but only EjSOC1-2 was highly expressed in pistils. Little expression of either was found in petals and stamens ( Figure 4F).
In summary, EjSOC1-1 and EjSOC1-2 may function to induce flowering and are involved in the growth and development of early flower organs in loquat.
Interestingly, we found that EjSOC1-1 and EjSOC1-2 exhibited different expression trends in leaves at different levels of maturity in the same period. EjSOC1-1 showed high expression in L3, L4 and L5 but relatively low expression in L1, L6, and L7 ( Figure 4B), though EjSOC1-2 did not display this trend. The results indicate that EjSOC1-1 might also be involved in leaf development.

EjSOC1s Are Inhibited by Short-Day and Exogenous GA 3 Treatments
After we analyzed the possible roles of EjSOC1-1 and EjSOC1-2 in loquat flowering, further exploration of the function of EjSOC1s under SD and exogenous GA 3 was proceeded. SOC1 can integrate the photoperiod and gibberellin pathways to regulate flower bud differentiation in Arabidopsis, and the loquat flower bud differentiation time coincides with the longest day of the year (summer solstice). Thus, we designed two experiments to alter growth conditions to explore whether EjSOC1s are affected by photoperiod and gibberellin. Interestingly, EjSOC1-1 and EjSOC1-2 were affected by SD and GA 3 treatments, with lower levels of expression during the critical period of flower bud differentiation (late June and early July). EjSOC1-1 and EjSOC1-2 were abundantly expressed in the normal growth group and the control group at the end of June and early July ( Figure 5A). More importantly, the SD-treated and GA 3treated loquat trees did not produce visible inflorescences in September compared to the trees in the normal growth group and the control group ( Figure 5B). Furthermore, according to qPCR, the floral meristem identity genes EjAP1-1 and EjAP1-2 were hardly expressed (Figure 5A), as well as EjLFY-1 and EjLFY-2 were consistently expressed at a low level (Figure 5A). Based on the above results, we conclude that EjSOC1-1 and EjSOC1-2 are regulated by photoperiod and GA 3 and that flower bud differentiation does not occur under SD conditions or after GA 3 exposure. The results suggest that EjSOC1s can initiate flower bud differentiation by integrating photoperiod and gibberellin signaling.

Subcellular Localization of EjSOC1s
To examine the subcellular localization of EjSOC1-1 and EjSOC1-2, 35S-EjSOC1-1-GFP, and 35S-EjSOC1-2-GFP fusion proteins were generated and transiently expressed in leaf epidermal cells of N. benthamiana. Fluorescence from the 35S-GFP control was detected in both the cytoplasm and nucleus, whereas fluorescence from the 35S-EjSOC1-1-GFP and 35S-EjSOC1-2-GFP fusions was detected only in the nucleus ( Figure 6). These results indicate that EjSOC1-1 and EjSOC1-2 were nuclear-localized proteins. These subcellular localization patterns were similar to AtSOC1 in Arabidopsis (Lee et al., 2008).

Functional Analysis of EjSOC1s in Arabidopsis
To examine whether EjSOC1-1 and EjSOC1-2 encode functional homologs of AtSOC1, we generated 35S:EjSOC1-1-HA and 35S:EjSOC1-2-HA constructs and introduced them into Arabidopsis wild-type ecotype Col-0 and the late flowering mutant soc1-2. Although the wild-type plants flowered when 13 or 14 rosette leaves appeared, the 35S:EjSOC1-1-HA and 35S:EjSOC1-2-HA transgenic lines flowered with only 6-10 rosette leaves (Figures 7A-D). In addition, soc1-2 mutant, which showed an obviously delayed phenotype compared to wild-type, flowered with 17 or 18 rosette leaves. However, in 35S:EjSOC1-1-HA/+soc1-2-and 35S:EjSOC1-2-HA/+soc1-2overexpressing lines, flowering occurred at a comparable or even lower number of rosette leaves compared to wild-type (Figures 7E-H). We also detected expression of EjSCO1s in the transgenic plants and found high levels in the respective lines (10 and 20 days) (Supplementary Figures S5A-C). Compared to Col-0, both AtAP1 and AtLFY were relatively highly expressed in the 35S:EjSOC1s-HA transgenic lines (Supplementary Figures S5D,E), and the expression levels of AtAP1 and AtLFY in the 35S:EjSOC1s-HA/+soc1-2 transgenic line and Col-0 were similar or even higher than the expression level in Col-0 (Supplementary Figures S5F,G). These results suggest that EjSOC1-1 and EjSOC1-2 both have a conserved role in accelerating flowering in Arabidopsis and that they may have a significant function in inducing flowering in loquat.
Interestingly, different from the Col-0 phenotypes (Figures 8A,E), a number of flower and silique phenotypes were observed in the 35S:EjSOC1-2-HA transgenic lines. For example, some of the petals were green or lilac in color; hypogenetic stamens were also observed, and the calyx showed anomalous growth (Figures 8B-D) and was not shed after maturity ( Figure 8F). Additionally, the surface of some siliques was lilac (Figures 8D,F).

DISCUSSION
The phenomenon of flowering in autumn and harvesting in spring is very unique in rosaceous plants. In spring, other fresh fruits are rarely sold, or the price of fruit is relatively high, and the range of choices for fruits is greatly reduced. Loquat is undoubtedly a relatively healthy and delicious choice in the fruit market at this time of year. In addition, loquat fruit is affected by storage and transportation. However, loquat fruits are easily injured and only remain fresh for 10 days at normal temperatures (Lin, 2007). Therefore, the price of loquat fruits is usually high, and their transport time to market is short. Overall, determining the flowering time and flowering mechanism of loquat will provide a means to successfully advance or delay it. If that were the case, loquat may be cultivated for a longer period of time and would be more stable in the fruit market for a longer duration. Notably, Eriobotrya species vary in flowering time, but each species of this genus can be hybridized, and fertile offspring can be obtained (Lin, 2017). This fact suggests that the flowering period of loquat is flexible. Research on the flowering mechanism of loquat can greatly benefit loquat growers and, more importantly, provide us with new insight into perennial fruit breeding.
SOC1, which can integrate the gibberellin pathway, photoperiod pathway and age pathway, is a key integration factor in flowering. In the age pathway, miRNA156 is  highly expressed in juveniles and inhibits transcription of the SBP family transcription factor SQUAMOSA PROMOTER BINDING PROTEIN LIKE (SPL). As plants mature, expression of miRNA156 decreases, the transcriptional level of SPLs increases, and SPL15 binds to SOC1 in a GA-dependent manner, recruiting MED18 and RNAPII to induce expression of the downstream MADS-box flowering gene FRUITFULL (FUL), which promotes flowering (Hyun et al., 2016). In the photoperiod pathway, FT activates expression of the flowering integration gene SOC1 and the floral meristem gene APETALA (AP1) to initiate flower bud differentiation and flower development (Abe et al., 2005;Wigge et al., 2005). In the gibberellic acid (GA) pathway, the MADS-box transcription factor AGAMOUS-LIKE24 (AGL24) interacts with SOC1, resulting in direct transcriptional upregulation of both (Liu et al., 2008). Under SD conditions, FT expression is low, and AGL24 interacts with SOC1 to promote Arabidopsis flowering  (Liu et al., 2008;Tao et al., 2012). In addition, nuclear factor Y (NF-Y) interacts with the photoperiod transcription factor CO and the GA pathway transcription factor DELLA, directly binding to a unique cis-element within the SOC1 gene, regulates H3K27 methylation levels of SOC1, and affects flowering time (Hou et al., 2014). Unlike in Arabidopsis, studies on woody fruit trees have shown that GA 3 inhibits floral bud induction (García-Pallas et al., 2001;Lenahan et al., 2006;Nakagawa et al., 2012;Goldberg-Moeller et al., 2013;.
In this study, the possibility that EjSOC1s are involved in loquat flower formation was identified, and we found that EjSOC1 expression is regulated by photoperiod and GA 3 . More importantly, our results show that SD and GA 3 treatment can inhibit flower differentiation in loquat. A detailed analysis of the gibberellin pathway and photoperiod pathway will help us to better understand the biological mechanism of flower bud differentiation in loquat and provide new insight into artificially delaying flowering in woody fruit trees.
There is increasing evidence that the initiation of flower buds is mainly regulated by AP1 and LFY, of which AP1 is mainly regulated by FT and LFY mainly by SOC1 (Abe et al., 2005;Lee et al., 2008). Furthermore, chromatin immunoprecipitation analysis indicated that the SOC1 protein can directly bind to the CArG domain in the LFY promoter (Lee et al., 2008;Liu et al., 2008). In this study, expression of EjAP1-1, EjAP1-2, EjLFY-1, and EjLFY-2 began to increase in late June and early July, and observation of paraffin-embedded sections showed that the leaf buds began to differentiate into flower buds from late June to early July. These results are consistent and show that EjAP1-1, EjAP1-2, EjLFY-1, or EjLFY-2 may be used as markers for identifying flower bud differentiation in loquat. In addition, the expression trends of EjSOC1s (Figures 4D,E) and EjLFYs (Figure 1C) in the SAM were similar, with EjSOC1-1 and EjSOC1-2 beginning to be highly expressed only 10 days earlier than EjLFYs. In addition, heterologous overexpression of EjSOC1s in Arabidopsis significantly upregulated expression of AtLFY.
Furthermore, it is worth noting that, EjSOC1-1 and EjSOC1-2 were differentially expressed at different developmental stages in leaves, EjSOC1-1 transcription level increased obviously as the young leaves getting mature, and decreased in the late stage of leaf development, this implied EjSOC1-1 might attend the regulation of leaf development, however, the expression of EjSOC1-2 did not show distinct variation. It was reported that aging transcription factor AtSPLs can up-regulate AtSOC1, and therefore promote flowering in Arabidopsis (Wang et al., 2009). In loquat flower, EjSOC1-1 and EjSOC1-2 were mainly expressed in receptacle and pistil, respectively. We speculated that EjSOC1s have a positive effect on the development of floral organs. In the overexpressed transgenic Arabidopsis, phenotypes including color changed petal and silique suggested that EjSOC1-2 might interrupt normal flower development through the abnormal regulation, in addition, the changed color of petals showed possible function of EjSOC1-2 on secondary metabolism, which is worth to investigate in the future. These results provide a theoretical basis for further exploration of the function and mechanism of EjSOC1s in loquat growth and development.
In recent research, EjFT1 has been shown to possibly have to do with bud sprouting and leaf development, whereas EjFT2 has been shown to possibly be involved in flower bud induction (Reig et al., 2017). In this study, EjSOC1-1 and EjSOC1-2 showed different expression trends in leaves with different levels of maturity in the same period ( Figure 4B). Therefore, it is speculated that EjSOC1-1 is involved in the processes of flower development and leaf growth. In addition, the expression levels of EjSOC1-1 and EjSOC1-2 differed in various flower tissues ( Figure 4F). Clearly, EjSOC1-1 and EjSOC1-2 have some different functions. In the model plant Arabidopsis, SOC1 integrates the photoperiod pathway through FT, which is transported to the SAM and interacts with FD to upregulate SOC1 (Lee and Lee, 2010). Similarly, in loquat (Eriobotrya deflexa Nakai f. koshunensis), EdFT can interact with both EdFD1 and EdFD2 . These interesting and meaningful findings provide a basis for further studies on the growth and development of loquat and a reference for such studies in other species.