The Role of EjSPL3, EjSPL4, EjSPL5, and EjSPL9 in Regulating Flowering in Loquat (Eriobotrya japonica Lindl.)

The age pathway is important for regulating flower bud initiation in flowering plants. The major regulators in this pathway are miR156 and SPL transcription factors. To date, SPL genes have been identified in many species of plants. Loquat, as a woody fruit tree of Rosaceae, is unique in flowering time as it blooms in winter. However, the study of its SPL homologous genes on the regulation mechanism of flowering time is still limited. In this study, four SPL homologs—EjSPL3, EjSPL4, EjSPL5, and EjSPL9—are cloned from loquat, and phylogenetic analysis showed that they share a high sequence similarity with the homologues from other plants, including a highly conserved SQUAMOSA promoter binding protein (SBP)-box domain. EjSPL3, EjSPL4, EjSPL5 are localized in the cytoplasm and nucleus, and EjSPL9 is localized only in the nucleus. EjSPL4, EjSPL5, and EjSPL9 can significantly activate the promoters of EjSOC1-1, EjLFY-1, and EjAP1-1; overexpression of EjSPL3, EjSPL4, EjSPL5, and EjSPL9 in wild-type Arabidopsis thaliana can promote flowering obviously, and downstream flowering genes expression were upregulated. Our work indicated that the EjSPL3, EjSPL4, EjSPL5, and EjSPL9 transcription factors are speculated to likely participate in flower bud differentiation and other developmental processes in loquat. These findings are helpful to analyze the flowering regulation mechanism of loquat and provide reference for the study of the flowering mechanism of other woody fruit trees.


Introduction
Flowering is regulated by extensive factors, such as plant hormones, photoperiod, vernalization, age, temperature, light quality, sucrose, etc. [1][2][3][4][5][6]. Genes regulating flowering time have been characterized in many species, and they interact to form a complicated regulatory network [7][8][9]. FLOWERING LOCUS T (FT) is a crucial integrator in this network [10], and the flowering time regulating gene SUPPRESSOR OF OVEREXPRESSION OF CO1 (SOC1), the floral meristem identity genes APETALA1 (AP1), LEAFY (LFY), and FRUITFULL (FUL) are located at downstream of the regulatory network [11,12]. The aging pathway is reported to be a universal approach to control the transition discovered the roles of EjSPL3, EjSPL4, EjSPL5, and EjSPL9 on flowering regulation, and provides further understanding of the unknown transcription factors attending reproductive growth in cultivated loquat.

Results
2.1. Identification of SPL3, SPL4, SPL5, and SPL9 Homologs from Loquat (Eriobotrya japonica Lindl.) The genes of EjSPL3, EjSPL4, and EjSPL5 from the cultivated loquat "JFZ" were annotated within the genome-wide scope of "JFZ" genome (from our unpublished data), and their coding sequences were isolated. EjSPL3, EjSPL4, and EjSPL5 had 303, 570, and 576 nucleotides encoding 100, 189, and 191 amino acids, respectively (Sequence S1). Clustering analysis between the predicted amino acid sequences of the above EjSPLs and their homologous genes in other plants was conducted, the phylogenetic tree showed that EjSPL3, EjSPL4, and EjSPL5 were clustered in different branches, and the closest homologs to them were from apple or pear: EjSPL3 and MdSBPB, EjSPL4 and MdSPL1 are the closest, and EjSPL5 is closest to PbrSPL1a (Figure 1a). From the sequence alignment, there was a conserved SBP-box domain in each of them, including two zinc finger structures and a nuclear localization signal (Figure 1b). By blast alignment, EjSPL3 showed 40% and 53% identity compared with AtSPL3 and MdSBPB; EjSPL4 had 51% and 94% identity compared to AtSPL4 and MdSPL1; EjSPL5 had 48% and 96% identity with AtSPL5 and PbrSPL1a, respectively. In conclusion, it is shown that the EjSPLs had higher similarity with those from apple and pear compared to other reported SPLs. The genes of EjSPL3, EjSPL4, and EjSPL5 from the cultivated loquat "JFZ" were annotated within the genome-wide scope of "JFZ" genome (from our unpublished data), and their coding sequences were isolated. EjSPL3, EjSPL4, and EjSPL5 had 303, 570, and 576 nucleotides encoding 100, 189, and 191 amino acids, respectively (Sequence S1). Clustering analysis between the predicted amino acid sequences of the above EjSPLs and their homologous genes in other plants was conducted, the phylogenetic tree showed that EjSPL3, EjSPL4, and EjSPL5 were clustered in different branches, and the closest homologs to them were from apple or pear: EjSPL3 and MdSBPB, EjSPL4 and MdSPL1 are the closest, and EjSPL5 is closest to PbrSPL1a (Figure 1a). From the sequence alignment, there was a conserved SBP-box domain in each of them, including two zinc finger structures and a nuclear localization signal (Figure 1b). By blast alignment, EjSPL3 showed 40% and 53% identity compared with AtSPL3 and MdSBPB; EjSPL4 had 51% and 94% identity compared to AtSPL4 and MdSPL1; EjSPL5 had 48% and 96% identity with AtSPL5 and PbrSPL1a, respectively. In conclusion, it is shown that the EjSPLs had higher similarity with those from apple and pear compared to other reported SPLs.
Except EjSPL3, EjSPL4 and EjSPL5 genes, an EjSPL9 gene was also isolated from cultivated loquat "JFZ", and it contained 1140 nucleotides encoding 379 amino acids. The phylogenetic tree showed that EjSPL9 was closely related to SPL9 of pear and farthest from AtSPL9 and AtSPL15 ( Figure 2a). From the sequence alignment, the homology of SPL9 is much higher than that of SPL3, SPL4, and SPL5 in woody plants, such as apple, pear, birch, etc., and the similarity is relatively low in herbaceous plants. In addition, they had a highly conserved SBP-box domain also (Figure 2b). By blast alignment, EjSPL9 showed 44% and 96% similarity compared with AtSPL9 and PbrSPL9, respectively.   Except EjSPL3, EjSPL4 and EjSPL5 genes, an EjSPL9 gene was also isolated from cultivated loquat "JFZ", and it contained 1140 nucleotides encoding 379 amino acids. The phylogenetic tree showed that EjSPL9 was closely related to SPL9 of pear and farthest from AtSPL9 and AtSPL15 (Figure 2a). From the sequence alignment, the homology of SPL9 is much higher than that of SPL3, SPL4, and SPL5 in woody plants, such as apple, pear, birch, etc., and the similarity is relatively low in herbaceous plants. In addition, they had a highly conserved SBP-box domain also (Figure 2b). By blast alignment, EjSPL9 showed 44% and 96% similarity compared with AtSPL9 and PbrSPL9, respectively.
(b) Amino acid sequence alignment of SPL3, SPL4, and SPL5. Amino acids with black and gray represent 100% and 50% similarity, respectively; the underline indicates the SBP-box domain. Identical and similar amino acid residues are represented by black and grey shadows, respectively. The accession number of each gene is as follows:

Analysis of EjSPLs Expression Patterns in Different Tissues
In order to identify the role of the EjSPL3, EjSPL4, EjSPL5, and EjSPL9 in loquat flowering, we analyzed their transcription levels in the samples including leaves, buds, and flowers in different

Analysis of EjSPLs Expression Patterns in Different Tissues
In order to identify the role of the EjSPL3, EjSPL4, EjSPL5, and EjSPL9 in loquat flowering, we analyzed their transcription levels in the samples including leaves, buds, and flowers in different growing periods. Different tissues of the "JFZ" were sampled, including root, stem, mature leaf, leaf bud, flower bud, flower, fruit, and seed. The results showed that all EjSPLs were highly expressed in leaves, buds (leaf buds and flower buds), and flowers compared to other tissues. The expression level of EjSPL4, EjSPL5, and EjSPL9 in leaf buds was higher than that in flower buds, and EjSPL3 was expressed relatively higher in flower buds. However, EjSPL5 and EjSPL9 were also expressed in stem, and the transcription level of EjSPL9 was detectable in root and seed ( Figure 3). The results suggested that EjSPL3, EjSPL4, EjSPL5, and EjSPL9 may be involved in the development of leaves, buds, flowers, stems, roots, and seeds. growing periods. Different tissues of the "JFZ" were sampled, including root, stem, mature leaf, leaf bud, flower bud, flower, fruit, and seed. The results showed that all EjSPLs were highly expressed in leaves, buds (leaf buds and flower buds), and flowers compared to other tissues. The expression level of EjSPL4, EjSPL5, and EjSPL9 in leaf buds was higher than that in flower buds, and EjSPL3 was expressed relatively higher in flower buds. However, EjSPL5 and EjSPL9 were also expressed in stem, and the transcription level of EjSPL9 was detectable in root and seed ( Figure 3). The results suggested that EjSPL3, EjSPL4, EjSPL5, and EjSPL9 may be involved in the development of leaves, buds, flowers, stems, roots, and seeds.

Expression Pattern of EjSPLs during Growth and Development of Loquat
To further clarify the roles of EjSPLs in loquat, the leaves, buds, and flowers of different developmental stages were sampled for detecting the EjSPLs expression levels. In our previous observation of the flower bud paraffin of "JFZ" in Guangzhou [50] the loquat flower bud differentiation happens during the period from 23 June to 7 July, and inflorescence is forming during the end of August to the beginning of September. In the leaves of different periods, the expression of EjSPL3, EjSPL4, and EjSPL5 were all in a quite low level at the beginning of the year, and followed by a significant increase till the start of flower bud differentiation (the end of June). Furthermore, the expression level of EjSPL4 and EjSPL5 gradually decreased to a low level by the end of the year, similar as that in the beginning of the year. The expression level of EjSPL9 came to a peak one month earlier than that of EjSPL4 and EjSPL5, and was kept in a relatively high level during the whole period of flower development; it just declined to a very low level after loquat finished flowering. The level of the EjSPL3 transcripts showed the first peak at the beginning of March, which was even earlier than that of EjSPL9, and also it showed another distinct peak during the early stage of flower bud development, and after that, decreased obviously to a low level by the end of the year (Figure 4a).
The expression levels of the EjSPLs in the leaves of different developmental stages ( Figure 4b) showed that the expression levels of EjSPL3 and EjSPL4 were low in the young leaves and increased

Expression Pattern of EjSPLs during Growth and Development of Loquat
To further clarify the roles of EjSPLs in loquat, the leaves, buds, and flowers of different developmental stages were sampled for detecting the EjSPLs expression levels. In our previous observation of the flower bud paraffin of "JFZ" in Guangzhou [50] the loquat flower bud differentiation happens during the period from 23 June to 7 July, and inflorescence is forming during the end of August to the beginning of September. In the leaves of different periods, the expression of EjSPL3, EjSPL4, and EjSPL5 were all in a quite low level at the beginning of the year, and followed by a significant increase till the start of flower bud differentiation (the end of June). Furthermore, the expression level of EjSPL4 and EjSPL5 gradually decreased to a low level by the end of the year, similar as that in the beginning of the year. The expression level of EjSPL9 came to a peak one month earlier than that of EjSPL4 and EjSPL5, and was kept in a relatively high level during the whole period of flower development; it just declined to a very low level after loquat finished flowering. The level of the EjSPL3 transcripts showed the first peak at the beginning of March, which was even earlier than that of EjSPL9, and also it showed another distinct peak during the early stage of flower bud development, and after that, decreased obviously to a low level by the end of the year (Figure 4a).
The expression levels of the EjSPLs in the leaves of different developmental stages (Figure 4b) showed that the expression levels of EjSPL3 and EjSPL4 were low in the young leaves and increased to a higher level as the leaves developed; after that, they began to decline to a low level again when leaves matured. EjSPL5 mRNA level was high in the young leaves and started decreasing from L3 stage; EjSPL9 mRNA level was kept in a relatively high level in all the stages compared to other EjSPLs (Figure 4c).
to a higher level as the leaves developed; after that, they began to decline to a low level again when leaves matured. EjSPL5 mRNA level was high in the young leaves and started decreasing from L3 stage; EjSPL9 mRNA level was kept in a relatively high level in all the stages compared to other EjSPLs (Figure 4c). AP1 is a flower meristem identity gene, located in downstream of the flowering regulation network [11], and has a high expression only in shoot apical meristem [50]; so the expression level of AP1 is generally employed as a marker gene of flowering, especially for confirming the period of floral bud initiation and development. In this study, the expression level of EjAP1 showed a distinct rise from the end of June, reached the highest level in mid-August, and then declined, which is highly consistent with the period of flower bud differentiation and development in loquat (Figure 5a). In the shoot apical meristem of different periods, the expression levels of EjSPL3, EjSPL4, and EjSPL5 all came to a relatively small peak during spring, and compared with them, EjSPL9 was abundantly expressed at this period. During floral bud initiation, expression levels of all the EjSPLs started rising rapidly from 9 or 23 June, got to an obvious peak, which is ten days earlier than that of EjAP1, and then decreased immediately; therefore, it is speculated that they may be involved in the process of loquat flower bud differentiation and are upstream of EjAP1.
In addition, flower samples were collected from macroscopic flower buds to the start of fruit setting ( Figure 5b). As one ABC developmental model gene, the expression level of EjAP1 was continuously increased in almost all the stages except the end of October. Importantly, EjSPL3, EjSPL4 ,and EjSPL5 mRNA levels showed similar trend as that of EjAP1, and they started decreasing from the end of November or the beginning of December. However, EjSPL9 showed some different or even opposite expression trends (Figure 5c). Based on these results, we speculated that EjSPL3, EjSPL4, and EjSPL5 may be involved in the growing process from the initial floral bud differentiation to the end of flowering, and EjSPL9 is hypothesized to be important for floral bud differentiation and the early stages of flower organ development. AP1 is a flower meristem identity gene, located in downstream of the flowering regulation network [11], and has a high expression only in shoot apical meristem [50]; so the expression level of AP1 is generally employed as a marker gene of flowering, especially for confirming the period of floral bud initiation and development. In this study, the expression level of EjAP1 showed a distinct rise from the end of June, reached the highest level in mid-August, and then declined, which is highly consistent with the period of flower bud differentiation and development in loquat (Figure 5a). In the shoot apical meristem of different periods, the expression levels of EjSPL3, EjSPL4, and EjSPL5 all came to a relatively small peak during spring, and compared with them, EjSPL9 was abundantly expressed at this period. During floral bud initiation, expression levels of all the EjSPLs started rising rapidly from 9 or 23 June, got to an obvious peak, which is ten days earlier than that of EjAP1, and then decreased immediately; therefore, it is speculated that they may be involved in the process of loquat flower bud differentiation and are upstream of EjAP1.
In addition, flower samples were collected from macroscopic flower buds to the start of fruit setting ( Figure 5b). As one ABC developmental model gene, the expression level of EjAP1 was continuously increased in almost all the stages except the end of October. Importantly, EjSPL3, EjSPL4, and EjSPL5 mRNA levels showed similar trend as that of EjAP1, and they started decreasing from the end of November or the beginning of December. However, EjSPL9 showed some different or even opposite expression trends (Figure 5c). Based on these results, we speculated that EjSPL3, EjSPL4, and EjSPL5 may be involved in the growing process from the initial floral bud differentiation to the end of flowering, and EjSPL9 is hypothesized to be important for floral bud differentiation and the early stages of flower organ development. In summary, these findings suggested that EjSPL3, EjSPL4, EjSPL5, and EjSPL9 genes were possibly involved in the process of leaf development and flowering (from floral bud initiation to flower development) in loquat.

Subcellular Localization of EjSPL3, EjSPL4, EjSPL5, and EjSPL9
In order to verify the localization of EjSPLs-encoded proteins in cells, we constructed an EjSPLs fusion expression vector carrying 35S promoter and a green fluorescent protein (GFP)-tagged protein, and introduced them into the epidermal cells of tobacco by Agrobacterium tumefaciens infection, and then their fluorescence signals were observed. The results showed that EjSPL3, EjSPL4, and EjSPL5 In summary, these findings suggested that EjSPL3, EjSPL4, EjSPL5, and EjSPL9 genes were possibly involved in the process of leaf development and flowering (from floral bud initiation to flower development) in loquat.

Subcellular Localization of EjSPL3, EjSPL4, EjSPL5, and EjSPL9
In order to verify the localization of EjSPLs-encoded proteins in cells, we constructed an EjSPLs fusion expression vector carrying 35S promoter and a green fluorescent protein (GFP)-tagged protein, and introduced them into the epidermal cells of tobacco by Agrobacterium tumefaciens infection, and then their fluorescence signals were observed. The results showed that EjSPL3, EjSPL4, and EjSPL5 were localized in the cytoplasm and nucleus, and EjSPL9 was localized only in the nucleus (Figure 6), which suggested that EjSPL3, EjSPL4, EjSPL5, and EjSPL9 may have different functions in loquat. were localized in the cytoplasm and nucleus, and EjSPL9 was localized only in the nucleus (Figure 6), which suggested that EjSPL3, EjSPL4, EjSPL5, and EjSPL9 may have different functions in loquat.

The regulation of EjSPL3, EjSPL4, EjSPL5, and EjSPL9 on the expression of EjLFY, EjAP1, and EjSOC1
In the previous experiment, the expression level of EjSPLs in the apical buds showed the possibility that they may regulate the expression of the downstream flowering genes. It has been previously reported that some SPLs can directly bind to LFY, AP1, and SOC1 promoters to regulate their transcription in Arabidopsis, and the specific binding site sequence is GTAC [35,51]. Therefore, 2.5. The regulation of EjSPL3, EjSPL4, EjSPL5, and EjSPL9 on the expression of EjLFY, EjAP1, and EjSOC1 In the previous experiment, the expression level of EjSPLs in the apical buds showed the possibility that they may regulate the expression of the downstream flowering genes. It has been previously reported that some SPLs can directly bind to LFY, AP1, and SOC1 promoters to regulate their transcription in Arabidopsis, and the specific binding site sequence is GTAC [35,51]. Therefore, in order to verify whether EjLFY, EjAP1, and EjSOC1 can be regulated by EjSPLs, we cloned the promoters of EjLFYs, EjAP1s, and EjSOC1s from the genomic DNA of "JFZ" loquat, respectively. The lengths of these six promoter regions are all about 2000 bp (upstream of each EjSPL genes initiation codon ATG). We used the PlaceCARE website to perform cis-acting element prediction on the promoter regions of these genes; some binding site sequences (GTAC) were found in their promoter regions (Sequence S2).
We constructed the reporter vectors, in which, CAMV35S drives renilla luciferase (REN), and the EjLFY-1pro, EjLFY-2pro, EjSOC1-1pro, EjSOC1-2pro, EjAP1-1pro, and EjAP1-2pro drive firefly luciferase (LUC). EjSPL3, EjSPL4, EjSPL5, and EjSPL9 were constructed into the effector vector driven by CAMV35S (Figure 7a). The two vectors were co-transformed into tobacco by transient expression by different combinations to analyze whether EjSPL3, EjSPL4, EjSPL5, and EjSPL9 could transcriptionally activate the downstream genes promoters. From the luciferase signal, it was showed that EjSPL3, EjSPL4, EjSPL5, and EjSPL9 obviously activated the expression of LUC, which was driven by the promoters of EjLFY-1 and EjSOC1-1; and EjSPL4, EjSPL5, and EjSPL9 remarkably activated the expression of LUC driven by the EjAP1-1 promoter compared to the control (Figure 7b). in order to verify whether EjLFY, EjAP1, and EjSOC1 can be regulated by EjSPLs, we cloned the promoters of EjLFYs, EjAP1s, and EjSOC1s from the genomic DNA of "JFZ" loquat, respectively. The lengths of these six promoter regions are all about 2000 bp (upstream of each EjSPL genes initiation codon ATG). We used the PlaceCARE website to perform cis-acting element prediction on the promoter regions of these genes; some binding site sequences (GTAC) were found in their promoter regions (Sequence S2). We constructed the reporter vectors, in which, CAMV35S drives renilla luciferase (REN), and the EjLFY-1pro, EjLFY-2pro, EjSOC1-1pro, EjSOC1-2pro, EjAP1-1pro, and EjAP1-2pro drive firefly luciferase (LUC). EjSPL3, EjSPL4, EjSPL5, and EjSPL9 were constructed into the effector vector driven by CAMV35S (Figure 7a). The two vectors were co-transformed into tobacco by transient expression by different combinations to analyze whether EjSPL3, EjSPL4, EjSPL5, and EjSPL9 could transcriptionally activate the downstream genes promoters. From the luciferase signal, it was showed that EjSPL3, EjSPL4, EjSPL5, and EjSPL9 obviously activated the expression of LUC, which was driven by the promoters of EjLFY-1 and EjSOC1-1; and EjSPL4, EjSPL5, and EjSPL9 remarkably activated the expression of LUC driven by the EjAP1-1 promoter compared to the control (Figure 7b). EjSPL3, EjSPL4, EjSPL5, and EjSPL9 activated the EjLFY, EjAP1, and EjSOC1 promoters. Error bars indicating SE from at least six biological replicates. The asterisk indicates a significant difference compared to the empty effector (Student's t-test, * p < 0.01, ** p < 0.001, *** p < 0.0001, **** p < 0.00001).

EjSPL3, EjSPL4, EjSPL5, and EjSPL9 Promote Flowering in Arabidopsis
To investigate the genetic functions of EjSPL3, EjSPL4, EjSPL5, and EjSPL9, overexpression vectors containing full length coding sequences of the four EjSPL genes were constructed and transformed into wild-type Arabidopsis thaliana (Col-0). The homozygous transgenic lines were screened out, and the flowering time and the expression of downstream flowering genes were (b) EjSPL3, EjSPL4, EjSPL5, and EjSPL9 activated the EjLFY, EjAP1, and EjSOC1 promoters. Error bars indicating SE from at least six biological replicates. The asterisk indicates a significant difference compared to the empty effector (Student's t-test, * p < 0.01, ** p < 0.001, *** p < 0.0001, **** p < 0.00001).

EjSPL3, EjSPL4, EjSPL5, and EjSPL9 Promote Flowering in Arabidopsis
To investigate the genetic functions of EjSPL3, EjSPL4, EjSPL5, and EjSPL9, overexpression vectors containing full length coding sequences of the four EjSPL genes were constructed and transformed into wild-type Arabidopsis thaliana (Col-0). The homozygous transgenic lines were screened out, and the flowering time and the expression of downstream flowering genes were analyzed in these transgenic lines. All transgenic lines with different genes showed early flowering phenotype compared to wild-type Col-0 (Figure 8a). Under the same growing condition, Col-0 had about 12 rosette leaves when flowering, while the transgenic lines with 35S: EjSPL3 and 35S: EjSPL5 had only seven to eight rosette leaves; 35S: EjSPL4 had nine to ten rosette leaves; 35S: EjSPL9 had 10 to 11 rosette leaves (Figure 8b). The expression levels of the related flowering genes including AtAP1, AtLFY, and AtSOC1, showed different degrees of improvement ( Figure S1). There was no obvious different morphological characteristic of the above transgenic lines observed compared to that of Col-0, such as flower organs, leaf shapes, siliques, and cauline leaves. From the above results, EjSPL3, EjSPL4, EjSPL5, and EjSPL9 all have the function of promoting flowering in Arabidopsis. analyzed in these transgenic lines. All transgenic lines with different genes showed early flowering phenotype compared to wild-type Col-0 (Figure 8a). Under the same growing condition, Col-0 had about 12 rosette leaves when flowering, while the transgenic lines with 35S: EjSPL3 and 35S: EjSPL5 had only seven to eight rosette leaves; 35S: EjSPL4 had nine to ten rosette leaves; 35S: EjSPL9 had 10 to 11 rosette leaves (Figure 8b). The expression levels of the related flowering genes including AtAP1, AtLFY, and AtSOC1, showed different degrees of improvement ( Figure S1). There was no obvious different morphological characteristic of the above transgenic lines observed compared to that of Col-0, such as flower organs, leaf shapes, siliques, and cauline leaves. From the above results, EjSPL3, EjSPL4, EjSPL5, and EjSPL9 all have the function of promoting flowering in Arabidopsis.

Discussion
To date, remarkable achievements have been made in the research of flowering regulation mechanism in plants, including photoperiod pathway, gibberellin pathway, autonomous pathway, aging pathway, vernalization pathway, ambient temperature pathway, and so on, which integrate each other to form a huge and complex regulatory network [52]. Among the above pathways, SPL transcription factors play a very important role in the control of flowering time. Our results show that all members of the SPL gene family contain the SBP-box domain, which was conserved in a certain degree in different species (Figures 1 and 2). In this study, it was suggested that EjSPL3, EjSPL4, EjSPL5, and EjSPL9 are involved in the regulation of flowering time in loquat and are highly conserved with their homologous genes. Differential expression levels and subcellular localization implied that they may have different patterns in regulating loquat growth and development.
Functional verification of SPL3, SPL4, SPL5, and SPL9 homologous genes in other plants has also been reported. For example, overexpression of SPL3 in Arabidopsis wild-type results in early flowering, and knocking out SPL3, SPL4, SPL5, and SPL9 delays flowering time and produces more rosette leaves [13]. MdSPL3 of apple encodes 189 amino acids and is localized in the nucleus; overexpressing it in Arabidopsis wild-type turns into fewer trichomes and promotes morphological differentiation, the MdSPL3 to be highly expressed in the rosette leaves and stem of transgenic Arabidopsis, and finally significantly improves the expression of downstream genes (AtFT, AtSOC1, AtLFY, AtAP1, AtFUL) [53]. In citrus, overexpression of CiSPL5 removed the binding site of miR156 in the 3'UTR, to Arabidopsis wild-type, leading to early flowering and fewer small rosette leaves [54]. PtSPL9 is expressed in various organs, with the highest expression in stems, and showed a nuclear localization [55]. In grape, VpSBP11 (a homolog of AtSPL3/4/5) encoded 170 amino acids, contained a highly conserved SBP domain, and its protein was localized in the nucleus with the transcriptional activity; overexpressing it (removed 3' UTR, the binding site of miR156) into the Arabidopsis wildtype advanced flowering, and promoted the expression of FUL, AP1, LFY, and some other genes; and also, the phenotypic characteristics in transgenic plants were transformed [36]. Combined with our

Discussion
To date, remarkable achievements have been made in the research of flowering regulation mechanism in plants, including photoperiod pathway, gibberellin pathway, autonomous pathway, aging pathway, vernalization pathway, ambient temperature pathway, and so on, which integrate each other to form a huge and complex regulatory network [52]. Among the above pathways, SPL transcription factors play a very important role in the control of flowering time. Our results show that all members of the SPL gene family contain the SBP-box domain, which was conserved in a certain degree in different species (Figures 1 and 2). In this study, it was suggested that EjSPL3, EjSPL4, EjSPL5, and EjSPL9 are involved in the regulation of flowering time in loquat and are highly conserved with their homologous genes. Differential expression levels and subcellular localization implied that they may have different patterns in regulating loquat growth and development.
Functional verification of SPL3, SPL4, SPL5, and SPL9 homologous genes in other plants has also been reported. For example, overexpression of SPL3 in Arabidopsis wild-type results in early flowering, and knocking out SPL3, SPL4, SPL5, and SPL9 delays flowering time and produces more rosette leaves [13]. MdSPL3 of apple encodes 189 amino acids and is localized in the nucleus; overexpressing it in Arabidopsis wild-type turns into fewer trichomes and promotes morphological differentiation, the MdSPL3 to be highly expressed in the rosette leaves and stem of transgenic Arabidopsis, and finally significantly improves the expression of downstream genes (AtFT, AtSOC1, AtLFY, AtAP1, AtFUL) [53]. In citrus, overexpression of CiSPL5 removed the binding site of miR156 in the 3 UTR, to Arabidopsis wild-type, leading to early flowering and fewer small rosette leaves [54]. PtSPL9 is expressed in various organs, with the highest expression in stems, and showed a nuclear localization [55]. In grape, VpSBP11 (a homolog of AtSPL3/4/5) encoded 170 amino acids, contained a highly conserved SBP domain, and its protein was localized in the nucleus with the transcriptional activity; overexpressing it (removed 3 UTR, the binding site of miR156) into the Arabidopsis wild-type advanced flowering, and promoted the expression of FUL, AP1, LFY, and some other genes; and also, the phenotypic characteristics in transgenic plants were transformed [36]. Combined with our results, it is shown that the SPL genes are both functionally conserved and functionally differentiated during plant development.
The regulatory network of SPL is extremely complicated, and to date, the model plant Arabidopsis is the most thoroughly studied. Most of the SPL genes in Arabidopsis are targeted by miR156 and miR157 [14,56,57], and positively regulate downstream flowering genes directly, such as SOC1, LFY, AP1, FUL, and so on [35,51]. This regulation pattern has also been verified in other species [28,53,58,59]. In this study, we found some SPL binding sites (GTAC) in the promoter regions of EjLFY, EjSOC1, and EjAP1 (Sequence S2); and the dual-luciferase reporter assays verified that EjSPLs activate their expression, in which, EjSPL3, EjSPL4, EjSPL5, and EjSPL9 effectively activated EjLFY-1 and EjSOC1-1 expression; and EjSPL4, EjSPL5, and EjSPL9 significantly improve the expression of EjAP1-1 (Figure 7). In addition, we found that the expression levels of the floral meristem identity genes AtAP1 and AtLFY were significantly different among different transgenic lines ( Figure S1). These results suggested that the SPL genes in loquat may be coordinated with each other to act on different downstream genes to jointly regulate flowering.
There are also some novel regulatory models found in other plants. SlSBP3 and SlSBP15 of tomato are regulated by the activity of PROCERA/DELLA, which can activate the SINGLE FLOWER TRUSS (SFT) gene in the leaves and the AP1/MC in the shoot apex to promote flowering [60]. In the Arabidopsis thaliana, SODIUM POTASSUIM ROOT DEFECTIVE1 (NaKR1) was found to respond to the changes in potassium, and it could regulate the expression of FT through the miR156-SPL3 module, which also affected the transport of FT protein in the phloem [61]. The nuclear factor CmNF-YB8 is a direct upstream gene of cmo-miR156, which can directly bind to its promoter region. Silencing this gene led to down-regulation of cmo-miR156 and up-regulation of CmSPL3, CmSPL5, and CmSPL9 [58]. Interestingly, EjFT2 showed involvement in the process of flower bud differentiation in loquat [62]. In order to ascertain whether there is a similar pattern of flowering regulation in the EjSPL genes in loquat, we need more in-depth research.

Plant Materials and Growth Conditions
Cultivated loquat (Eriobotrya japonica Lindl.) was used in this study. Adult loquat trees were grown under natural conditions in the loquat germplasm resource preservation garden, South China Agricultural University, Guangzhou, China. The wild-type of Arabidopsis, Arabidopsis thaliana Columbia-0 (Col-0), was used for gene transformation. Nicotiana benthamiana was cultivated for the assays of transient expression. The seeds of the Arabidopsis Col-0 and tobacco used in this study were provided by Xingliang Hou. Both of Arabidopsis and tobacco were grown under long-day conditions (16 h light/8 h dark) at 22 • C in a controlled environment room. Samples (loquat tissues and Arabidopsis plants) for qRT-PCR were immediately frozen in liquid nitrogen and stored in an ultra-low temperature refrigerator at −80 • C until use.

Gene Cloning and Sequence Analysis
The coding region of EjSPL3, EjSPL4, EjSPL5, and EjSPL9 were amplified through PCR using PrimeSTAR ® Max DNA Polymerase (TaKaRa, shiga, Japan) with specific primers (Table S1). The PCR products were connected with pGEM-T vector (Promega, Wisconsin, USA). Then the cloned products were sequenced and blasted with the other homologous sequences in the NCBI (https: //www.ncbi.nlm.nih.gov/). DNAMAN 6.0, ClustalX 2.0.12, GeneDoc 2.7, and MEGA 6 software were employed for the amino acid sequences alignment and the construction of the phylogenetic tree of SPL proteins, respectively.

Gene Expression Analysis with qRT-PCR
The total RNA of loquat and Arabidopsis was extracted by EASYspin Plus plant RNA extraction kit (Aid lab, Beijing, China), and the cDNA was synthesized using PrimeScript TM RT reagent Kit with gDNA Eraser (TaKaRa, shiga, Japan). Quantitative real-time PCR (qPCR) was performed using iTaq TM universal SYBR Green Supermix (Bio-Rad, Hercules, CA, USA) in the LightCycler 480 (Roche, Basel Switzerland). Ejβ-actin [63] and TUB2 (AT5G62690) were amplified as the internal control genes for normalization in loquat and Arabidopsis thaliana, respectively. Three technical and biological replicates were applied and data were analyzed with previous methods [64]. Semi-quantitative RT-PCR was employed for detecting the expression of exogenous gene in the overexpression lines of Arabidopsis.

Subcellular Localization Analysis
The full length coding sequences of EjSPL3, EjSPL4, EjSPL5, and EjSPL9 without stop codon were separately cloned into the pGreen-35S-GFP vector [65]. These pGreen-35S-EjSPL-GFP or negative control vectors were transformed into the Agrobacterium tumefaciens strain GV3101::psoup by heat induction, and the transformed Agrobacterium cells were cultured and injected into epidermic cells of tobacco (Nicotiana benthamiana) leaves for transient expression following the previously described methods [66]. After about 48 h of injection, GFP signal was observed using the fluorescence microscope (Observer.D1, Zeiss, Jena, Germany).

Promoter Cloning and Analysis
The promoters of EjLFY, EjAP1 and EjSOC1 were cloned from the DNA of "JFZ" using PrimeSTAR ® Max DNA Polymerase (TaKaRa) with specific primers. Putative cis-acing elements in each promoter region were found in the PlantCARE database (http://bioinformatics.psb.ugent.be/webtools/plantcare/ html/).
The recombined effectors and reporters were co-transformed into tobacco leaves with different groups by Agrobacterium-mediated protocol, as described above. After about 48 h of transformation, we quantified the activity of LUC and REN using the dual luciferase assay kit (Promega) with appropriate modification [70] was quantified on the Luminoskan Ascent Microplate Luminometer (Thermo Fisher Scientific). The binding activity was calculated by the LUC to REN ratio. Each assay was made by at least six biological duplicates.

Data Analysis
All data analyses were performed by Excel and GraphPad Prism 6.01. Significance of differences between data was evaluated by Student's t test.

Conclusions
In summary, the Eriobotrya japonica transcription factors EjSPL3, EjSPL4, EjSPL5, and EjSPL9 are speculated to participate in flower bud differentiation and flower organ development by activating the expression of EjSOC1, EjLFY, or EjAP1.