Characterization and Functional Analysis of PEBP Family Genes in Upland Cotton (Gossypium hirsutum L.)

Upland cotton (Gossypium hirsutum L.) is a naturally occurring photoperiod-sensitive perennial plant species. However, sensitivity to the day length was lost during domestication. The phosphatidylethanolamine-binding protein (PEBP) gene family, of which three subclades have been identified in angiosperms, functions to promote and suppress flowering in photoperiod pathway. Recent evidence indicates that PEBP family genes play an important role in generating mobile flowering signals. We isolated homologues of the PEBP gene family in upland cotton and examined their regulation and function. Nine PEBP-like genes were cloned and phylogenetic analysis indicated the genes belonged to four subclades (FT, MFT, TFL1 and PEBP). Cotton PEBP-like genes showed distinct expression patterns in relation to different cotton genotypes, photoperiod responsive and cultivar maturity. The GhFT gene expression of a semi-wild race of upland cotton were strongly induced under short day condition, whereas the GhPEBP2 gene expression was induced under long days. We also elucidated that GhFT but not GhPEBP2 interacted with FD-like bZIP transcription factor GhFD and promote flowering under both long- and short-day conditions. The present result indicated that GhPEBP-like genes may perform different functions. This work corroborates the involvement of PEBP-like genes in photoperiod response and regulation of flowering time in different cotton genotypes, and contributes to an improved understanding of the function of PEBP-like genes in cotton.

Plants have developed mechanisms to control the vegetative and reproductive growth by integration of environmental and developmental cues. The plant life cycle is divided into distinct developmental phases based on the morphological and functional characteristics of the organs that differentiate from the shoot apical meristem. In Arabidopsis thaliana, flowering is initiated by four regulatory pathways (photoperiodic, temperature, age and gibberellins) that converge on three integrator genes, namely FT, SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1) and LEAFY (LFY) [17,18]. Photoperiodic pathway integrate the circadian clock and light, and are involved in the expression and regulation of genes such as CON-STANS (CO), FLAVIN BINDING, KELCH REPEAT, F-BOX 1 (FKF1) and so on [19][20][21]. Activation of FT is achieved largely through regulation of the transcription factors CO, TAR-GET OF EAT1 (TOE1) and other genes involved in photoperiodic pathways [22,23]. The FT protein is a major component of the florigen signal, which is transported in the phloem from leaves to the shoot apex where it promotes flowering and interacts with the bZIP transcription factor FLOWERING LOCUS D (FD) to activate floral meristem identity genes APETALA1 (AP1) and LFY [24][25][26][27]. In Monocotyledon, rice FT homologue protein Hd3a also interacts with 14-3-3 and OsFD1 proteins to form a florigen activation complex, which induces expression of the Arabidopsis AP1 homologue gene OsMADS15 [28,29].
Most plant PEBP-related genes have been identified from mutants that show altered inflorescence architecture. In Arabidopsis, two FT-like genes, FT and TWIN SISTER OF FT (TSF), are floral activators whose mutants are late-flowering [18,30]. Under long-day conditions, TSF and FT expression is up-regulated in phloem companion cells in the leaves [31]. Transcripts of TFL1, the paralogous of an additional Arabidopsis PEBP gene ARABIDOPSIS THALIANA CENTRORADIALIS (ATC), are weakly accumulated in the inner cells of matures shoot meristemsin the vegetative phase, whereas the transcript level increases following the switch to flowering [32,33]. The TFL1 protein is a mobile signal that is translocated from the inner shoot meristems cells to the outer cells and coordinates shoot meristem identity [34]. With regard to flowering, TFL1 and FT are functionally antagonist, because FT is a floral activator, whereas TFL1 is a floral repressor, but both proteins are able to interact with FD to regulate the FDdependent transcription targets [35]. BROTHER OF FT AND TFL1 (BFT) modulates the function of the FT-FD module and may provide an adaptation strategy that fine-tunes photoperiodic flowering under high salinity [15]. MFT is specifically induced in the radical-hypocotyl transition zone of the embryo in response to ABA, and mft loss-of-function mutants show hypersensitivity to ABA in seed germination. In germinating seeds, MFT expression is directly regulated by transcription factors ABA-INSENSITIVE3 (ABI3) and ABI5 in the ABA signaling pathway. In addition, MFT promotes embryo growth via a negative feedback loop by directly repressing ABI5 expression in the ABA signaling pathway during seed germination [36].
The Gossypium genus comprises more than 50 species and includes many important cotton species [37]. Cotton is one of the most important natural textile fiber crops. The seeds are also a source of oil and protein meal. Novel insights into Gossypium biology have been provided by whole-genome sequencing of Gossypium raimondii U., Gossypium arboretum L. and G. hirsutum [38][39][40]. Most of studies of Gossypium species have focused on the allopolyploids Gossypium hirsutum and Gossypium barbadense (2n = 52) [39,[41][42][43][44], and the diploids Gossypium arboreum and Gossypium raimondii (2n = 26) [40,45]. Gossypium hirsutum is a tetraploid species considered to have originated through hybridization of the D-genome species G. raimondii and the A-genome species G. arboreum. Meanwhile, Gossypium hirsutum consists of semi-wild races and domesticated cultivars adapted to diverse geographical and ecological environments and has a long evolutionary history [46]. Cultivated upland cotton consists of several semi-wild races, such as 'marie-galante', 'paimerii', 'morrilli', 'punctatum', 'yucatanense', 'richmondii' and 'latifolium'. Semi-wild upland cotton shows superior resistance to insects, drought and disease than cultivated germplasm. In addition, the semi-wild race and cultivated upland cotton differ in photoperiod sensitivity. The semi-wild races flower only under short-day conditions, whereas flowering of cultivated upland cotton is insensitive to day length. In order to study the flowering-related genes in cotton, we applied different photoperiod treatments to selected semi-wild and cultivated genotypes of upland cotton. The aim of this study was to analyze genotype-and maturity-specific expression of PEBP-like genes in upland cotton germplasm and undertake ectopic transgenic analysis of GhFT and GhPEBP2 genes and their promoters. We demonstrated the interaction of two PEBP-like proteins and the FD-like bZIP transcription factor GhFD, which is conserved in Arabidopsis and other species. This work highlights the involvement of PEBP-like genes in photoperiod response and regulation of flowering time in different cotton genotypes, and contributes to an improved understanding of the function of PEBP-like genes in cotton.

Plant materials and growth conditions
Seeds of the upland cotton Semi-wild race 'latifolium' were abtained from the National Wild Cotton Nursery, Sanya, China. Samples were collected from plants of the three early-flowering cultivars 'CCRI36', 'CCRI74' and 'CCRI50' and three late-flowering cultivars 'CCRI41', 'CCRI60' and 'Lu28' of upland cotton cultivated on the experimental farm of the Cotton Research Institute of Chinese Academy of Agricultural Sciences (CAAS), Anyang, Henan. All plants were grown in the field on the experimental farm of Anyang. We sampled roots, stems, leaves and shoot apicals with two fully expanded leaves stage, buds at 5mm in length, flowers at anthesis and fibers 10 days after flowering from plants of the cultivar 'CCRI36'. We collected leaves and shoot apicals samples from two fully expanded cotyledons to four fully expanded leaves at 08:30 in the early-and late-maturing cultivars experiment.
Plants of the semi-wild race 'latifolium' and the cultivar 'CCRI36' subjected to different photoperiod treatments were grown initially in the phytotron under long days (14/10 h light/dark photoperiod, 08:00-22:00 light). Half of the plants were treated with short day (10/14 h light/ dark, 08:00-18:00 light) once the plants had developed two fully expanded leaves. In the phytotron, the seeds germinated and the cotyledons were fully expanded after about seven days, and the seedlings developed new leaves every six days. We collected leaves and shoot apicals samples from two fully expanded cotyledons to five fully expanded leaves at 08:30.Wild type Arabidopsis thaliana accession Colombia 0 (Col-0) plants was grown in the greenhouse at 22°C under long days (16/8 h light/dark photoperiod).

Gene cloning, vector construction and transformation
For nine GhPEBP-like genes, we designed specific primers to amplify the open reading frame using cDNA templates prepared from different tissues of CCRI36. For the GhFT and GhPEBP2 genes, we designed a pair of infusion primers to amplify the open reading frame and cloned to pBI121vector using XbaI and SacI enzyme sites. We also designed a pair of infusion primers to amplify the promoters of GhFT and GhPEBP2 sequences from the genomic DNA of CCRI36 and cloned to pBI121vector using HindIII and XbaI enzyme sites. PrimeSTAR 1 GXL DNA polymerase (TaKaRa Tokyo, Japan) was used to amplify the two genes with the following cycling profile: 98°C for 1 min, and 30 cycles of 98°C for 10 s, 55°C for 15 s, and 68°C for 1 min. The amplified products were cloned into vector pBI121 (Clontech, Palo Alto, CA, USA) and sequenced from both ends. Arabidopsis plants were transformed using the Agrobacteriummediated gene transfer method described previously [47]. Transgenic Arabidopsis plants were selected with kanamycin. Flowering date, the number of rosette leaves, and the number of cauline leaves on the main inflorescence of individual plants were recorded for transgenic lines in the T 3 generation. Samples for quantitative real-time PCR (qRT-PCR) were harvested 14 days after sowing (DAS). Statistical test was used by one-way analysis of variance Duncan's method.

Sequence alignment and phylogenetic analysis
The amino acid sequences of the proteins analyzed in this study were downloaded from Gen-Bank. Accession numbers of all species are listed in S1 Table. Multiple sequence alignment was performed with ClustalW (http://www.ebi.ac.uk). A phylogenetic tree was constructed using the neighbor joining method with Molecular Evolutionary Genetics Analysis (MEGA) software MEGA5.05 [48]. Branch support was estimated using bootstrapping with 1000 replicates.

Quantitative Real-Time PCR
Total RNA was isolated from samples using a plant RNA purification kit (Tiangen, Beijing, China). Reverse transcription-PCR was carried out using a SuperScript™ III First-Stand Synthesis System for RT-PCR (Invitrogen, Carlsbad, USA). Transcript levels were then determined by qRT-PCR using the 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) and SYBR Premix Ex Taq (2×) (TaKaRa). Gene-specific primer pairs used for the PCR amplifications are listed in S2 Table. To normalize variance among samples, actin was used as an endogenous control. Determination of reaction specificities and data processing were performed as described previously [49]. Three biological replicates were analyzed. The data were analyzed with Graph Pad Prism 5.

Yeast two-hybrid and bimolecular fluorescence complementation analysis
The full-length sequence of GhFD was amplified and cloned into pGBKT7 vector, a bait protein is expressed as a fusion to the Gal4 DNA-binding domain (DNA-BD). Tested for autoactivation activity and toxicity were conducted. In addition, GhFT and GhPEBP2 were amplified by PCR and cloned into the pGADT7 vector, prey proteins are expressed as fusions to the Gal4 activation domain (AD) (Clontech, USA). The resulting recombinant plasmid pGBKT7-GhFD was introduced into yeast strain Y2H, and the recombinant plasmids pGADT7-GhFT and pGADT7-GhPEBP2 were introduced into yeast strain Y187 respectively. When bait and prey fusion proteins interact, the DNA-BD and AD are brought into proximity to activate transcription of four reporter genes. Additionally, GhFD was also cloned into pGADT7 vector, and GhFT / GhPEBP2 were cloned into the pGBKT7 vector respectively. Two-hybrid interactions were assayed on selective SD/-Trp/-Leu (DDO) and SD/-Ade/-His/-Leu/-Trp (QDO) media supplemented with X-α-Gal and Aureobasidin A.
To measure in vivo interactions, open reading frames of full-length GhFD, GhFT and GhPEBP2 genes coding sequences were cloned into vectors pSPYNE and pSPYCE, which contain DNA encoding the N-terminal and C-terminal regions of yellow fluorescent protein, respectively. Molecular techniques of BiFC were performed using protocols according to [51,52]. Fully-expanded rosette leaves of Arabidopsis Col-0 plants grown for 4 weeks under short-day conditions were collected for the protoplast isolation. Protoplasts was transformed with 10μg of plasmid DNA and incubated at 22°C. Protoplasts isolated from Arabidopsis leaves were transformed with the following combinations of plasmids: pSPYNE-GhFD, pSPYNE-GhFT, pSPYNE-GhPEBP2, pSPYCE-GhFD, pSPYCE-GhFT and pSPYCE-GhPEBP2. The samples were observed with a FV1000 confocal laser scanning microscope (Olympus, Tokyo, Japan).

Identification of PEBP-like genes in Gossypium hirsutum
To identify PEBP-like family genes in upland cotton, we searched databases containing genomic sequences for G. raimondii, G. arboreum and G. hirsutum and identified nine putative GhPEBP-like genes (Table 1). We then amplified cDNAs based on the annotated coding sequences and compared their structures. Analysis of genomic sequences revealed that the aforementioned seven genes conserved the characteristic genomic organization, containing four exons and three introns in identical positions, and the latter two genes showed a different structure with two exons and one intron (S1 Fig). Comparisons of amino acid sequences with FT/TFL1-like sequences from Arabidopsis, in conjunction with phylogenetic analysis of the multiple sequence alignment (see below), enabled interpretation of the cloned sequences as either a FT orthologue, MFT orthologue, TFL1 orthologue and PEBP orthologue. Seven FT/ MFT/TFL1 genes were shown similarity with the flowering plant species PEBP family, and two PEBP sequences showed conserved with PEBP genes of bacteria and archaea. Based on the Abbreviations are as follows: Chr, chromosome; A, D and AD, Gossypium arboreum, Gossypium raimondii and Gossypium hirsutum genomes. Chr and identity indicate the chromosomal location and identity of a given gene. AD indicate the G. hirsutum genome references from BGI-CGP.

Phylogenetic relationships of the PEBP-like genes
To study the phylogenetic relationships of GhPEBP-like genes and other PEBP homologues, we compiled a data set of 104 PEBP amino acid sequences from 14 species, comprising a bryophyte (Physcomitrella patens H.), a gymnosperm (Picea abies L.) and angiosperms (Antirrhinum majus L., Populus nigra L., Arabidopsis thaliana L., Vitis vinifera L., Solanum lycopersicum L., Malus domestica B., Medicago truncatula G., Glycine max L., Nicotiana tabacum L., Gossypium hirsutum L., Zea mays L. and Hordeum vulgare L. ). An unrooted phylogenetic tree derived from the amino acid sequences was constructed using the neighbor-joining method and consisted of four subclades (Fig 1). The FT-subclade comprised GhFT, Arabidopsis FT and TSF, as well as other FT orthologous proteins identified in other species. GhFT displayed the characteristic features of the FT protein subclade, which include conservation of amino acid residues Tyr85 and Gln139 (S1 Fig). The second distinctive monophyletic clade consisted of the two GhMFT proteins, Arabidopsis MFT and putative orthologous proteins identified in other dicotyledons. The third subclade contained the four GhTFL1 proteins, Arabidopsis BFT, TFL1 and ATC. The TFL1 proteins shared conserved amino acid residues His88 andAsp144 in similar positions with Arabidopsis TFL1. The fourth subclade consisted of proteins that contained the conserved PEBP domain with a structure specific to bacteria and archaea, but not a eukaryote-specific sequence identified in previous plant studies. Most GhPEBP-like proteins exhibited a closer relationship to PEBP proteins from other dicotyledons than to those from monocotyledons. The gymnosperm Picea abies had MFT subclade and TFL1 subclade, and the bryophyte Physcomitrella patens only had MFT subclade. These findings indicated that PEBP proteins may be highly conserved in different species. Protein structure analysis is also predicted that GhFT and GhTFL1 proteins were structurally similar to its homologous proteins of Arabidopsis and rice [14,28], whereas GhPEBP-like proteins was more similar to predicted PEBP protein of Escherichia coli M.

Tissue-specific expression of GhPEBP-like genes in Gossypium hirsutum
GhPEBP-like transcription levels were examined by quantitative real-time PCR (qRT-PCR) in samples of roots, stems, leaves, shoot apicals, buds, flowers and fibers of the short-season Chinese upland cotton cultivar 'CCRI36'. All the samples were collected from plants growing on the farm of Anyang. To differentiate individual GhPEBP-like genes, we designed gene-specific primers by avoiding highly conserved sequence regions. Transcript levels for GhFT and GhPEBP1 were higher in buds and flowers compared with vegetative organs (Fig 2). Accumulation of GhTFL1a and GhTFL1c transcripts was highest in roots, whereas GhTFL1b and GhTFL1d transcript levels were highest in shoot apicals and flowers. GhMFT1 and GhMFT2 were transcribed mainly in flowers with low transcript levels observed in other tissues. The highest levels of GhPEBP2 transcripts were detected in leaves, shoot apicals, buds and flowers.

Expression of GhPEBP-like genes under different photoperiod treatments
Due to the different photoperiod sensitivities of semi-wild races and cultivars of upland cotton, we selected the semi-wild race 'latifolium' and cultivar 'CCRI36' as representative genotypes to analyze GhPEBP-like expression patterns under different photoperiod treatment. Interestingly, under short days compared with long days, and GhFT especially showed hundreds of fold enrichment (Fig 3). The other identified genes were not unambiguously detected under both long-and short-day conditions. For the cultivar 'CCRI36', GhFT, GhMFT1, GhPEBP2 and GhTFL1a expression were increased under long days relative to that under short days. GhTFL1b showed the strongest expression induction and was increased under short days ( Fig  4). It would be interesting in the future to examine in more detail the expression of PEBP-like genes, e.g. in leaves and shoot apical meristems, in diverse upland cotton genotypes.

Expression of flowering-related genes in different maturing upland cotton cultivars
Although previous studies have described flowering-related genes of G. hirsutum, these studies have reported tissue-specific expression in a single cultivar [56][57][58][59] or in different Gossypium species such as G. hirsutum and G. arboreum [60]. The relative expression of flowering-related genes in early-and late-maturing cultivars of upland cotton has not been investigated previously. Thus, we tested the expression patterns of GhFT, GhPEBP2, GhFD and GhSOC1 genes in whole aboveground tissues of early-and late-maturing cultivars from the cotyledon to the four-leaf stages. As shown in Fig 5, GhFT transcript levels were higher in early-maturing cultivars than in late-maturing cultivars throughout the experimental period. GhSOC1 showed the strongest induction, with transcript levels increasing gradually from the two fully expanded leaves stage, and expression was higher in early-maturing cultivars. Neither GhPEBP2 nor GhFD transcript levels showed discernible variation in the aboveground tissues of plants of either maturity class.

Ectopic expression and promoter analysis of GhFT and GhPEBP2
To confirm the functional roles of GhFT and GhPEBP2 in the regulation of flowering time, we constructed an overexpression vector using the cauliflower mosaic virus 35S promoter to drive In addition, we analyzed the GhFT and GhPEBP2 promoter sequences 1500bp upstream of the ATG start codon, based on the available Gossypium genome sequences, using the Plant-CARE and PLACE databases. Both GhFT and GhPEBP2 promoters contained common elements such as a CAAT box, TATA box and many light-responsive elements, and elements such as different binding motifs specific for transcriptional factors (S5 Fig). The promoter of GhFT was predicted to show shoot-specific expression responsiveness, abscisic acid (ABA) and auxin responsiveness. The ABA-responsive motif, ethylene-induced motif and MYB binding site involved in drought-inducibility elements were identified in the GhPEBP2 promoter. We also analyzed three transgenic Arabidopsis lines overexpressing pGhFT::GUS and pGhPEBP2:: GUS. The GhFT promoter was most highly expressed in shoot apicals, followed by leaves and roots, but not in cotyledons and stems (Fig 6D-6F). In contrast, GhPEBP2 promoter was mainly expressed in cotyledons, stems and the axial roots (Fig 6G-6I). These results indicated that GhFT may play an important role in plant development, whereas GhPEBP2 may play roles in abiotic stress response.

GhFD interaction with cotton FT and PEBP2 proteins
FT can interact with the bZIP transcription factor FD in the shoot apical meristem, in which the resultant FD-FT complex is essential for AP1 and LFY induction to promote the flowering transition. In our cotton cDNA database [61], we identified a sequence in shoot apicals encoding a predicted protein that showed high similarity to Arabidopsis FD (Fig 7A and 7B) and contains the conserved bZIP domain and C-terminal SAP (Ser-Ala-Pro) motif (Fig 7B). RT-PCR analysis indicated that GhFD showed a higher expression level in shoot apicals than in other organs (Fig 7C). In order to investigate the interaction of GhFD and cotton FT and PEBP2, we designed a yeast two-hybrid assay. We first tested the autoactivation activity and toxicity of GhFD and determined that this protein exhibited autoactivation activity. Then, we cloned GhFT and GhPEBP2 genes into pGBKT7 vector respectively and investigated the interaction of these two proteins and GhFD in yeast cells, which were grown on selective media with Aureobasidin A and 5-bromo-4-chloro-3-indolyl-α-D-galactopyranoside (X-α-gal). The assay indicated that GhFD could interact with GhFT protein but not with GhPEBP2 (Fig 7D).
We next conducted bimolecular fluorescence complementation (BiFC) assays to evaluate whether GhFD and GhFT form protein complexes in plant cells. After transformation for 12-16 h, GhFD was revealed to have interaction with GhFT (Fig 8), providing evidence for GhFD-GhFT complex interaction in plant cells.

Discussion
With the availability of genome sequences for G. raimondii, G. arboreum and G. hirsutum, genetics and molecular functional studies on Gossypium have entered a new era. Previous studies have investigated fiber development, leaf and floral development, biotic and abiotic stress and so on [57,[62][63][64][65][66]. Meanwhile, there were few studies making use of molecular marker techniques to detect QTL for cotton earliness traits, such as growth period, growth stages, node of first fruiting branch, and height of node of first fruiting branch [67][68][69][70]. In this study, GhPEBP-like family genes were cloned from upland cotton and shown to cluster in four subclades, namely the FT subclade, MFT subclade, TFL1 subclade and PEBP subclade. According to the chromosome locus in genomic, the GhPEBP-like family genes could not be anchored with cotton earliness traits QTLs in previous studies [67][68][69][70]. Compared with the Arabidopsis FT/TFL1 family, both FT and TSF showed similar functions in flowering time [30,71], and some of the FT family genes had no TSF and two FT in other species [72]. We identified five FT/TFL1-like genes, comprising one GhFT and four GhTFL1s, in upland cotton. For MFT, only one gene has been identified in Arabidopsis and other species, whereas two MFT genes  were identified in G. hirsutum. These results indicated that differences in gene numbers have arisen during species evolution. Meanwhile, increases or decreases in the number of homologous genes may have effects on gene expression, gene function and stoichiometric balance. The spatiotemporal expression patterns of PEBP genes have been determined in many species such as Arabidopsis, grape, and rice [7,11,73]. However, no study of PEBP genes expression in semi-wild and cultivated genotypes of upland cotton has been undertaken previously. We observed that transcript levels of GhFT had higher transcript levels in buds and flowers than in vegetative organs. The result was very similar to the previous study and the expression pattern of FT in Arabidopsis [6,59]. Although GhTFL1s shared similar structures, the expression induction was also completely different. GhTFL1a and GhTFL1c were most highly expressed in roots, whereas transcript levels of GhTFL1b and GhTFL1d were highest in shoot apicals and flowers. This multiple and differently expression pattern of GhPEBP-like genes may also contributed to distinct gene function specifically. We then examined the contribution of GhPEBPlike genes to the photoperiod responsiveness in different upland cotton germplasm and in early-and late-maturing upland cotton cultivars. For semi-wild type cotton, expression levels of GhFT, in particular, and GhSOC1 strongly increased under short days. This induction was photoperiod specific and together with no apparent expression trend in cultivated cotton. GhTFL1d and GhPEBP2 were induced and expression levels increased in both semi-wild race and cultivated upland cotton under long days, of which GhPEBP2 showed especially high expression enrichment. Only GhMFT1 expression was increased in semi-wild race, whereas expression was decreased in cultivated cotton, under short days. These result indicated GhFT and GhSOC1 are important flowering related genes in semi-wild upland cotton. Additionally, the function of GhPEBP2 requires further analysis to determine whether it can promote flowering under long-day condition. Semi-wild race of upland cotton could flower only under shortday conditions, whereas cultivated upland cotton has been selected to be photoperiod-insensitive during domestication and crop evolution. Thus, the key flowering genes showed dispensable effect in the photoperiod pathway. In addition, a shift in the function of these genes over the course of evolution and selection is evident.
In addition, we investigated the expression pattern of flowering related genes in early-and late-maturing cultivars. GhFT and GhSOC1 showed higher transcript levels in earlier-maturing cultivars than in late-maturing cultivars, whereas GhFD and GhPEBP2 showed no significant differences. The existence of PEBP family members that we observed in cotton here had also been described previously. For example, Pea FT family genes showed different expression patterns among developmental timing, tissue specificity and photoperiod response [9]. In upland cotton, GhSOC1 also plays an important role in promoting flowering and foral organ variations [74]. Both GhSOC1 and GhFT had higher expression level in early maturing cultivars from the two fully expanded leaf stage, in which the floral bud primordia have already emerged and continued to differentiate during the third fully expanded leaf stage [58]. These result indicated that GhFT may be an important flowering-related gene similar to GhSOC1. Under long day conditions it is suggested that, although GhFT might play fewer roles in regulating gene expression concomitant with the long domestication of cultivated cotton from semi-wild races, it still can distinguish cultivars belonging to different maturity classes.
Studies of Arabidopsis and other plant species had reported the effects of FT family genes on flowering. The Arabidopsis FT gene promotes flowering in heterologous plant species, and FT protein can interact with FD to regulate other flowering related genes [75,76]. The tomato FT orthologue also triggers systemic signals to regulate growth and flowering [77]. In the present study, we identified two PEBP genes, GhFT and GhPEBP2, and transformed them into wildtype Arabidopsis. Ectopic expression of GhFT in Arabidopsis accelerated flowering under both long-and short-day conditions. The expression patterns observed for flowering-related genes in Arabidopsis demonstrated that GhFT could promote flowering by up-regulating the expression of FT. However, overexpression of GhPEBP2 had no obvious phenotypic impact. Molecular evolution studies suggest that plasticity at exon four contributes to the divergence of FT-like function in floral promotion [78]. This also may explain why GhPEBP-like genes, which contain two exons, may not contribute to flowering. In plants, the florigen FT interacts with the bZIP transcription factor FD and promotes flowering [24,28]. Thus, we analyzed the interaction of GhFD and cotton FT and PEBP2 proteins. We demonstrated that GhFD interacted with GhFT, but not GhPEBP2. The present analyses of protein structure, expression pattern and function support the conclusions that GhFT was flowering related promoter and GhPEBP2 has no consistent effect on flowering.
In conclusion, this study highlights the important role of cotton PEBP-like genes in specifying photoperiod responsiveness and regulating the time of flowering. The results provide a method with which to distinguish cotton cultivars of different maturity classes through monitoring expression of important flowering-related genes. In addition, the findings provide insights into the mechanism by which the genes GhFT may regulate flowering time. Broadly, our results provide a better understanding for in-depth analyzing upland cotton breeding and guiding future work.