Obtaining plants that flower over a long period is the goal of many gardeners, so as to be able to achieve year-round flower production. The flowering period can be lengthened by exploiting the blossoming duration or number of flowering cycles. Most plants have a single annual flowering period. Some perennial plants have the ability to flower again during the year: they can flower continuously during the favourable season or they may only have a second bloom later in the season, which may be occasional. Some Rosaceae plants, such as Rosa hybrid, have the ability of flowering more than once or recurrent flowering in a year, which is an important horticultural and economical characteristics, but their molecular background is not clear     .
There are three gene classes such as inflorescence meristem genes, floral meristem genes and flower organ identity genes involved or interacted in flower development. The inflorescence meristem genes control the transition of shoot apical meristem from vegetative to reproductive (inflorescence), forcing or delaying the flowering time. The TERMINAL FLOWER 1 (TFL1) is an inflorescence meristem identity gene, and encodes phosphatidylethanolamine binding protein (PEBP) belongs to the TFL/CENTRORADIALIS (CEN) family. TFL1 was first identified from Arabidopsis thaliana mutants, followed by cloning of its homologous genes in Antirrhinum majus, Nicotianin tobacco, Rose chinensis, Malus domestica, Fragaria × ananassa, Pyrus pyrifolia, Pyrus communis, Prunus armeniaca and other plants      ruminate rather than indeterminate, and early flowering occurs due to the shortened vegetative growth. In recent years, several researchers believed that the TFL1 may be a regulator of continuous flowering in roses. Lina Wang et al. (2012) examined the relationship between the recurrent flowering character and the expression patterns of TFL1 homologs in Rosa multiflora, R. rugosa, R. chinensis, six other rose species and nine rose cultivars  . TFL1 was expressed at high levels at all four flowering stages in non-recurrent flowering species, such as R. multiflora and R. rugosa, whereas it was barely detected at any stage in recurrent flowering species, such as R. chinensis. These results suggest that the recurrent flowering habit in roses results from lower expression of TFL1, which may be related to recurrent flowering character in roses. Hikaru Iwata et al. (2012) found a similar phenomenon in roses and strawberries  . Therefore, TFL1 probably demonstrates a new role of modifying flowering seasonality in perennial plants. But, it still needs more experimental evidence.
The Rosa rugosa is a famous traditional Chinese flower. It is fragrant as well as resistant to cold, drought, pest, disease, salt, and alkali    . It is not only a kind of fine plant material in landscaping, but also one of the oldest natural spice plants, which is the raw material of perfume, cosmetics and other chemical products, food and precious Chinese herbal medicines. Therefore, the R. rugosa is a collection of ornamental, medicinal, edible, beauty in a beautiful flowers and trees. But the vast majority of R. rugosa varieties only once a year to spend flowers, flowering are in April-May, and the flowering period is only about 2 weeks, which severely limits the application of R. rugosa. After years of observation, it was found that the wild Rosa rugosa which was introduced from Muping by our laboratory had a unique recurrent flowering trait, but their molecular background is not clear.
In order to determine if the TFL1 is related with the continuous flowering phenotype of wild Rosa rugosa from Muping, we isolated the TFL1 homolog from wild Rosa rugosa from Muping, and comparatively analyzed its bioinformatics characteristics.
2. Material and Methods
2.1. Plant Material
The plant material, Chinese wild Rosa rugosa from Muping, was from the R. rugosa germplasm resources garden at Shandong Agricultural College. The wild Rosa rugosa from Muping is the most representative traditional Rosa rugosa in China.
2.2.1. Flower Buds Collection
Between March 2016 and May 2016, the flower buds in different developmental stages (Vegetative growth period; Growth cone transformation period; Receptacle development period; Pistil and stamen morphogenesis period) of wild Rosa rugosa from Muping were collected, flash frozen with liquid nitrogen and then stored in a −80˚C freezer.
2.2.2. Total RNA Extraction and cDNA Synthesis
An EASYspin plant RNA Rapid Extraction Kit from Adlai Biotechnology Co., Ltd. was used to extract and mix the wild R. rugosa from Muping at different stages of flower bud RNA. Agarose gel electrophoresis and spectrophotometer were used to determine the quality and concentration of the mixed total RNA. An EasyScript First-Strand cDNA Synthesis SuperMix Kit from Beijing TransGen Biotech Co., Ltd. was used to synthesize the first-strand cDNA.
2.2.3. Cloning of the Middle Fragment
According to the reported TFL1 sequences of Rosaceae, the degenerate primers F1 (5’-CCTCTTGTKGTTGGRAGAGT-3’) and R1 (5’-AGRTADGGRTCACT- RGGDCCBGG-3’) were designed with Primer Premier 5.0 PCR amplification was conducted using the synthesized cDNA in Section 2.2.2 as a template and F1 and R1 as the primers. The reaction system included 1 µL cDNA, 1 µL F1 primer (10 µmol/L), 1 µL R1 primer (10 µmol/L), and 12.5 µL PCR MIX, with ddH2O added to a total volume of 25 µL. The reaction conditions were: 94˚C for 3 min; 94˚C for 30 s, 55˚C for 30 s, and 72˚C for 30 s for a total of 36 cycles; and then extension at 72˚C for 10 min. Next, 1% agarose gel electrophoresis was used to detect the PCR products. The target PCR fragment was recovered with the MiniB-EST Agarose Gel DNA Extraction Kit Ver. 3.0 (TaKaRa). The recovered fragment was ligated to the pMD18-T vector and then transformed into E. coli DH5a. The positive clones were selected and sent to BGI for sequencing.
2.2.4. 3’ RACE and 5’ RACE
The 3’RACE specific primers MG1 (5’-TGTCACTTACAACACCAAACTCG- TCTTC-3’) and MG2 (5’-CACCGCCAAACCTAGAGTTGAGATT-3’) and the 5’RACE specific primers GSP1 (5’-TAGGTTTGGCGGTGAC-3’), GSP2 (5’-AT- GTCCATTGAAGACGAGTTTG-3’) and GSP3 (5’-CAATCATTTTAGTAGTT- GGGGT-3’) were all designed with Primer Premier 5.0. Nested PCR was conducted using 3’1022-1, 3’1022-2, and the SMARTer™ RACE cDNA Amplification Kit (Clontech) in order to obtain the 3’-terminal sequence of the target gene. Nested PCR was also conducted using GSP1, GSP2, GSP3, and the 5’ RACE System for Rapid Amplification of cDNA Ends (Version 2.0, Invitrogen) in order to obtain the 5’-ter-minal sequence of the target gene.
2.2.5. Full-Length Gene Sequence Splicing and Verification
DNAstar software was used to splice the middle fragment, the 5’-terminal sequence, and the 3’-terminal sequence in order to obtain the full-length cDNA sequence of the gene. The 5’- and 3’-primers for the spliced sequence were designed with Primer Premier5 as follows: F2 (5’-ATGGACATGTCGGAACC-3’) and R2 (5’-CTAGCGTCTTCTTGCTGCC-3’). The spliced sequence was amplified using the re-verse transcription product of cDNA as a template, and then, it was further validated and verified.
2.2.6. Bioinformatics Analysis of Gene
BLASTX (NCBI) was used to study the homology of the nucleotide sequence and the deduced amino acid sequence. The ORF finder (NCBI) was used to search for an open reading frame, and the Conserved Domains database (NCBI) was used to analyze the conserved domains. The ProtParam Tool was used to analyze protein physical and chemical properties. Post Prediction, WOLF PSORT, and SubLocv were used to predict protein sub-cellular localization. Furthermore, ProtScale was used to predict hydrophilic or hydrophobic protein proper-ties. The SignalP 4.0 Server was used to predict the protein signal peptide. The TMHMM Server v2.0 was used to predict the protein transmembrane domain. The NetPhos 2.0 Server was used to predict potential protein phosphorylation sites, and the NetNGlyc 1.0 Server and NetOGlyc 3.1 Server were used to predict potential protein glycosylation sites. ExPaSy-SOPMA was used to predict protein secondary structure. SWISS-MODEL was used to predict protein tertiary structure. DNAMAN5.2.2 was used to conduct multiple sequence alignment. The Neighbor-Joining method from Mega5 was used to create the phylogenetic tree.
3. Results and Analysis
3.1. Cloning of the Rosa rugosa TFL1 Gene
The cloned middle fragment is 200 bp (Figure 1(a)), the cloned 3’-terminal fragment is 650 bp (Figure 1(b)), and the cloned 5’-terminal fragment is 278 bp (Figure 1(c)). These three fragments were spliced together with DNAstar in order to obtain a 973 bp cDNA sequence. The spliced sequence was then validated
Figure 1. PCR amplification of RrTFL1 cDNA. (a) Intermediate fragment; (b) 3’-RACE; (c) 5’-RACE; (d) Full-length fragment.
by PCR amplification (Figure 1(d)). In addition, the Blast analysis confirmed that all its homologous genes are the TFL1 gene and named RrTFL1 (GenBank accession number: KY928070).
3.2. Bioinformatics Analysis of the RrTFL1 Gene
The RrTFL1 gene has a full length of 973 bp, an open reading frame of 519 bp, a 5’ UTR of 153 bp, and a 3’ UTR of 301 bp, encoding 172 amino acids. The derived protein has a molecular weight of 19.48 kD, an isoelectric point of 9.13, a (c100227) conserved domain at position1-172. Thus RrTFL1 protein belongs to the PEBP family. Furthermore, the subcellular localization prediction result indicated that the protein is probably located at the cytoplasm. The hydrophilicity analysis further showed that the overall average hydrophobic index is −0.302, thus indicating a hydrophilic protein. The signal peptide prediction result demonstrated that no signal peptide cleavage site, thus indicating a non-secretory protein. The transmembrane domain analysis showed that no transmembrane domain exists. The phosphorylation site prediction results demonstrated that there are five Ser phosphorylation sites, seven Thr phosphorylation sites, and three Tyr phosphorylation sites, thereby providing a reference for the future study of the regulation of gene expression and protein modification. The glycosylation site prediction results showed that there is one O-glycosylation site and no N-glycosylation sites. The secondary structure prediction results demonstrated that there is 24.42% α-helix, 36.63% random coil, 27.91% extended peptide chain, and 11.05% β-corner. The tertiary structure prediction indicated that RrTFL1 and other homologous TFL1 gene proteins are composed of a large number of random coils, no spiral coil, the overall structure is similar. The BLAST results showed that the protein shares 96% homology with the TFL1 amino acid sequences of Rosaceae Plants including Rosa chinensis (ADO64261.1), Fragaria × ananassa (AFA42328.1), Prunus mume (XP 008241250.1), Prunus armeniaca (ADL62862.1), Spiraea cantoniensis (AEO72025.1), Photinia serratifolia (AEO72024.1). The multiple sequence alignment result demonstrated that the RrTFL1 protein and the above plant TFL1 amino acid sequences all have a c100227 conserved domain, two highly conserved modules D-P-D-x-P, G-x-H- R, and two functional sites His, Asp (Figure 2). Furthermore, the constructed phylogenetic tree revealed that RrTFL1 is closely related to TFL1 from the same family members Rosa chinensis, Fragaria × ananassa and Prunus mume, whereas it is relatively distant from Morusnotabilis and Citrus trifoliata, which are from different families, consistent with the traditional classification results (Figure 3).
TFL1 inhibits flower development by inhibiting the expression and activity of LFY and AP1 in the central region of the stem to maintain the characteristics of inflorescence meristems. By now, Some homologs of TFL1 have been isolated from many plants, include GMtfl1 in soybean, VvTFL1 in grapevine, Self- Pruning (SP) in tomato, CET in tobacco, CsTFL1 in citrus, MdTFL1 in apple, and PcTFL1 in pear  -  . In this research, the cDNA sequence of TFL1 gene was obtained from the flower buds of wild Rosa rugosa from Muping at different developmental stages for the first time, and its biological information was predicted. At present, the report on TFL1 gene in R. rugosa is very rare. The results
Figure 2. Multiple alignment of the RrTFL1 with other TFL1. Notes: The black underlined line is the amino acid sequence of RrTFL1; The red box area is labeled with two functional modules (D-p-D-x-p, G-x-H-R) and two functional sites (His, Asp) of the TFL1 gene.
Figure 3. The phylogenetic tree derived from the alignment of amino acid sequences of RrTFL1 and other TFL1.
of this study laid a foundation for further study of the expression, regulation and flowering patterns of RrTFL1, but also cultivating new varieties of R. rugosa which can flower continuously by gene engineering.
Previous studies have shown that the different expression pattern of TFL1 gene results in the different functions of TFL1 gene in different plants. CEN mutant became short, inflorescence became limited inflorescence in Antirrhinum majus, indicating that TFL1 acts as an inflorescence meristem-specific gene and is capable of maintaining the characteristic of the infinite growth of inflorescence; Arabidopsis thaliana TFL1 deletion mutants showed early flowering and multiflorous character, indicating that the TFL1 gene was an maintenance gene of inflorescence meristem   ; Excessive expression of the TFL1 gene leads to a flowering delay by prolonging the time of vegetative growth and inflorescence growth in Antirrhinum majus, as well as a delayed flowering transition in Nicotianin tobacco  . These results show that TFL1 gene is widely involved in vegetative growth and reproductive tissue development. In recent years, several researchers believed that the TFL1 may be a regulator of continuous flowering inroses  . But, it still needs more experimental evidence. Previous studies in our laboratory have found that the wild Rosa rugosa introduced from Muping had a unique recurrent flowering trait, but their molecular background is not clear. Therefore, we cloned the gene from the wild Rosa rugosa from Muping, and the gene was analyzed by bioinformatics. Next, we will further to verify whether the gene has continuous flowering regulation function.
The TFL1 gene cloned in this study has a conserved domain (c100227), two highly conserved modules, and two functional sites of His and Asp, which have a PEBP protein family compared with other species such as rose. But the BEBP protein family is divided into two subfamily TFL1 gene and FT gene, which play the opposite roles in the flowering process. The continuous flowering in R. rugosa may also be controlled by multiple genes. This study laid the foundation for further study on the mechanism of continuous flowering of R. rugosa. In addition, we further constructed the phylogenetic tree based on the TFL1 genes and found that the RrTFL1 gene had the smallest phylogenetic distance from the TFL1 genes derived from the species of the same family and the longest phylogenetic distance from the TFL1 genes derived from the species of different families. This agrees with the conventional plant classification. It is inferred that the evolution of the TFL1 gene corresponds with the phylogenetic relationship among the plant species from which the gene is derived.
In this study, the full-length cDNA sequence of TFL1 gene was cloned successfully from R. rugosa and named as RrTFL1. The bioinformatics characteristics of the RrTFL1 gene were analyzed, which provided the basis for further research on the function of RrTFL1 gene.
This work was funded by the National Science Foundation of China (NSFC) (31200524).
 Debener, T., Malek, B.V., Mattiesch, L., et al. (2001) Genetic and Molecular Analysis of Important Characters in Roses. Acta Horticulturae, 547, 45-49.
 Liang, H., Zhebentyayeva, T., Olukolu, B., et al. (2010) Comparison of Gene Order in the Chromosome Region Containing a TERMINAL FLOWER 1 Homolog in Apricot and Peach Reveals Microsystem across Angiosperms. Plant Science, 179, 390-398.
 Mouhu, K., Kurokura, T., Koskela, E.A., et al. (2013) The Fragaria vesca Homolog of SUPPRESSOR OF OVEREXPRESSION OF CONSTANS1 Represses Flowering and Promotes Vegetative Growth. Plant Cell, 25, 3296-3310. https://doi.org/10.1105/tpc.113.115055
 Nakano, Y., Higuchi, Y., Yoshida, Y., et al. (2015) Environmental Responses of the FT/TFL1 Gene Family And Their Involvement In Flower Induction in Fragaria × ananassa. Journal of Plant Physiology, 177, 60-66. https://doi.org/10.1016/j.jplph.2015.01.007
 Mimida, N., Kotoda, N., Ueda, T., et al. (2009) Four TFL1/CEN-Like Genes on Distinct Linkage Groups Show Different Expression Patterns to Regulate Vegetative and Reproductive Development in Apple (Malus × domestica Borkh.). Plant & Cell Physiology, 50, 394-412.
 Esumi, T., Kitamura, Y., Hagihara, C., et al. (2010) Identification of a TFL1, or Theology in Japanese Apricot (Prunusmume). Scientia Horticulture, 125, 608-616.
 Wang, L.-N., Liu, Y.-F., Zhang, Y.-M., et al. (2012) The Expression Level of Rosa Terminal Flower 1 (RTFL1) Is Related with Recurrent Flowering in Roses. Molecular Biology Reports, 39, 3737-3746. https://doi.org/10.1007/s11033-011-1149-8
 Iwata, H., Gaston, A., Remay, A., et al. (2012) The TFL1 Homologue KSN Is a Regulator of Continuous Flowering in Rose and Strawberry. The Plant Journal, 69, 116-125.
 Koskela, E.A., Sotheby, A., Flachowsky, H., et al. (2016) Terminal Flower Is a Breeding Target for a Novel Overbearing Trait and Tailored Flowering Responses in Cultivated Strawberry (Fragaria × ananassa Duch.). Plant Biotechnology Journal, 14, 1852-1861.
 Kim, W., Park, T.I., Yoo, S.J., et al. (2013) Generation and Analysis of a Complete Mutant Set for the Arabidopsis FT/TFL1 Family Shows Specific Effects on Thermo-Sensitive Flowering Regulation. Journal of Experimental Botany, 64, 1715-1729. https://doi.org/10.1093/jxb/ert036
 Ito, A., Saito, T., Sakamoto, D., et al. (2015) Physiological Differences between Bud Breaking and Flowering after Dormancy Completion Revealed by DAM and FT/TFL1 Expression in Japanese Pear (Pyrus pyrifolia). Tree Physiology, 36, 109.
 Esumi, T., Tao, R. and Yonemori, K. (2005) Isolation of Leafy and Terminal Flower 1 Homologues from Six Fruit Tree Species in the Subfamily Maloideae of the Rosaceae. Sexual Plant Reproduction, 17, 277-287. https://doi.org/10.1007/s00497-004-0239-3
 Coelho, C.P., Minow, M.A., Chalfunjúnior, A., et al. (2014) Putative Sugarcane FT/TFL1 Genes Delay Flowering Time and Alter Reproductive Architecture in Arabidopsis. Frontiers in Plant Science, 5, 221. https://doi.org/10.3389/fpls.2014.00221
 Yoo, S.J., Chung, K.S., Jung, S.H., et al. (2010) Brother of FT and TFL1 (BFT) Has TFL1-Like Activity and Functions Redundantly with TFL1 in Inflorescence Meristem Development in Arabidopsis. Plant Journal, 63, 241-253. https://doi.org/10.1111/j.1365-313X.2010.04234.x
 Zeng, H.Y., Lu, Y.T., Yang, X.M., et al. (2015) Ectopic Expression of the BoTFL1-Like Gene of Bambusa oldhamii Delays Blossoming in Arabidopsis Thaliana and Rescues the tfl1 Mutant Phenotype. Genetics & Molecular Research, 14, 9306-9317.
 Baumann, K., Venail, J., Berbel, A., et al. (2015) Changing the Spatial Pattern of TFL1 Expression Reveals Its Key Role in the Shoot Meristem in Controlling Arabidopsis Flowering Architecture. Journal of Experimental Botany, 66, 4769. https://doi.org/10.1093/jxb/erv247
 Randoux, M., Jeauffre, J., Thouroude, T., et al. (2012) Gibberellins Regulate the Transcription of Thecontinuous Flowering Regulator, RoKSN, a Rose TFL1 Homologue. Journal of Experimental Botany, 63, 6543-6554. https://doi.org/10.1093/jxb/ers310