Components and Functional Diversification of Florigen Activation Complexes in Cotton

Abstract

In shoot apex cells of rice, a hexameric florigen activation complex (FAC), comprising flowering locus T (FT), 14-3-3 and the basic leucine zipper transcription factor FD, activates downstream target genes and regulates several developmental transitions, including flowering. The allotetraploid cotton (Gossypium hirsutum L.) contains only one FT locus in both of the A- and D-subgenomes. However, there is limited information regarding cotton FACs. Here, we identified a 14-3-3 protein that interacts strongly with GhFT in the cytoplasm and the nuclei, and five FD homoeologous gene pairs were characterized. In vivo, all five GhFD proteins interacted with Gh14-3-3 and GhFT in the nucleus. GhFT, 14-3-3 and all the GhFDs interacted in the nucleus as well, suggesting that they formed a ternary complex. Virus-induced silencing of GhFD1, -2 and -4 in cotton delayed flowering and inhibited the expression of floral meristem identity genes. Silencing GhFD3 strongly decreased lateral root formation, suggesting a function in lateral root development. GhFD overexpression in Arabidopsis and transcriptional activation assays suggested that FACs containing GhFD1 and GhFD2 function mainly in promoting flowering with partial functional redundancy. Moreover, GhFD3 was specifically expressed in lateral root meristems and dominantly activated the transcription of auxin response factor genes, such as ARF19. Thus, the diverse functions of FACs may depend on the recruited GhFD. Creating targeted genetic mutations in the florigen system using Clustered regularly interspaced short palindromic repeats (CRISPR) and their associated proteins (Cas) genome editing may fine-tune flowering and improve plant architecture.

Introduction

As sessile organisms, plants are more vulnerable to environmental fluctuations. They have evolved a variety of ways to orchestrate endogenous signals and exogenous stimuli to ensure flowering at the correct time. In the model plant species Arabidopsis, floral induction is controlled by several regulatory pathways: photoperiod, vernalization, ambient temperature, age, autonomous and gibberellin (Fornara et al. 2010). The flowering pathways finally converge in the florigen–antiflorigen hormone system. Florigen is encoded by Arabidopsisflowering locusT (FT), tomato single flower truss and rice Heading date 3a (Hd3a), while antiflorigen is encoded by terminal flower1 (TFL1) and self-pruning (SP) family members (Tamaki et al. 2007, Lifschitz et al. 2014, Eshed and Lippman 2019).

Centroradialis (CEN)/TFL1/SP of the CETS gene family, also known as the phosphatidylethanolamine-binding protein (PEBP) gene family, contains three main subfamilies: Mother of FT and TFL1, FT and TFL1 (Hedman et al. 2009). Among them, FT and TFL1/SP constitute the core members of the florigen–antiflorigen system, which is essential for maintaining the determinate and indeterminate growth of the meristem (Eshed and Lippman 2019). Thus, these genes are key regulators of plant architecture, controlling the flowering transition and meristem destiny (Bradley et al. 1997, Pnueli et al. 1998, Kardailsky et al. 1999, Kobayashi et al. 1999).

In plant species that respond to day length, leaves perceive photoperiod changes and FT is expressed in phloem companion cells. The synthesized FT protein is loaded into phloem sieve elements and transported to the shoot apical meristem (SAM) through vascular bundles over a long-distance transport (Corbesier et al. 2007, Tamaki et al. 2007). In SAM cells, FT interacts with 14-3-3 proteins in the cytoplasm, forming a subcomplex that translocates into the nucleus and binds to the basic leucine zipper (bZIP) transcription factor FD, which is mainly expressed in the SAM, to yield a ternary florigen activation complex (FAC) (Abe et al. 2005, Wigge et al. 2005, Taoka et al. 2011). FAC activates the downstream floral homeotic genes, such as apetala1 (AP1) and fruitfull (FUL), to regulate flowering transition (Abe et al. 2005, Wigge et al. 2005).

Interestingly, in many plants, FT homologs are also involved in the control of diverse developmental transitions other than flowering induction, such as heterosis and crop yield (Krieger et al. 2010), tuberization transition in potato (Navarro et al. 2011), bulb formation in onion (Lee et al. 2013), and reiterative growth and termination of shoots in tomato (Shalit et al. 2009). Thus, florigen is a plant hormone that functions as a universal growth hormone. However, knowledge regarding the molecular nature and functional diversity of FACs in plants remains limited.

In FAC, FD, as a bZIP transcription factor, is an important FAC component, because it determines the selection of downstream target genes. FD proteins share a short and conserved motif at the C-terminus, the Thr/Ser-Ala-Pro (T/SAP) motif, which resembles the FT-interacting motif (Abe et al. 2005, Wigge et al. 2005, Taoka et al. 2011). S or T must be phosphorylated in the SAP motif by calcium-dependent protein kinases, and alanine substitutions disrupt the interactions with FT (Taoka et al. 2011, Tsuji et al. 2013, Kawamoto et al. 2015). The conserved C-terminal domain and interactions with FT homologs suggest that the functions of FD homologs are conserved in plants. However, several studies have shown the functional diversity of FD homologs. For example, Arabidopsis FD and FDP have different functions in flowering control and abscisic acid responses (Romera-Branchat et al. 2020). Two FD-like paralogs, FDL1 and FDL2 in aspen, have dual functions in photoperiodic growth control and adaptive response pathways (Tylewicz et al. 2015) and the rice FD paralog OsFD2 controls leaf development, suggesting that FD displays neofunctionalization (Tsuji et al. 2013). In potato (Solanum tuberosum), three FD paralogs, StFD and StFDL1a/b, form a turberigen activation complex with the FT homolog StSelf pruning 6A. However, StFDL1a/b promote tuber formation, while StFD does not (Navarro et al. 2011, Teo et al. 2017). Thus, the transcription factor in the FAC may be exchanged, resulting in a change in FT function and in the modulation of diverse developmental transitions, depending on the developmental context (Tsuji 2017). Moreover, a study identified many previously undiscovered FD target genes involved in multiple aspects of plant development, including in hormone signaling pathways (Collani et al. 2019). On the basis of these results, we hypothesized that FD acts as a central regulator during floral transition and integrates multiple signaling pathways in the SAM.

The main cultivated cottons, Upland cotton (AD)₁ (G. hirsutum L.) and Sea island cotton (AD)₂ (Gossypium barbadense L.) are allotetraploids composed of two homoeologous genomes, defined as the A- and D-subgenomes (Senchina et al. 2003). Recent efforts to sequence and update cotton genomes have made comparative and functional genomics studies possible (Hu et al. 2019, Wang et al. 2019a, Huang et al. 2020). A genome-wide analysis indicated that the A₂ and D₅ diploid genomes each contain only one FT locus, and the (AD)₁ and (AD)₂ genomes each contain one FT homoeologous gene pair (McGarry et al. 2016, Zhang et al. 2016, Wang et al. 2019b). Furthermore, genetic and molecular analyses revealed that GhFT and GhSP homologs encode cotton florigen–antiflorigen systems to control flowering transition and plant architecture, respectively (Guo et al. 2015, Li et al. 2015, McGarry et al. 2016, Liu et al. 2018, Prewitt et al. 2018, Si et al. 2018, Chen et al. 2019). GhFT directly interacts with one cotton FD homolog (Zhang et al. 2016, Chen et al. 2019) or Arabidopsis FD (Prewitt et al. 2018) as assessed by biochemical assays.

Although current efforts in cotton have not deciphered the molecular mechanism of GhFT in as much detail as its homologs in Arabidopsis and rice, they suggest that the cotton FT–FD complex has both conserved and diversified functions. Functional divergence among FD paralogs in rice, aspen and potato validated the hypothesis, which was proposed by Tsuji et al. (2013), that FACs functions depend on the bound FD. However, detailed characterizations of FD paralogs are still limited in all species (Tsuji et al. 2013), and in particular, limited attention has been paid to the allopolyploid FD paralogs. Moreover, as an allotetraploid plant with a genome size of ∼2.3 Gb (Zhang et al. 2015, Huang et al. 2020), G. hirsutum has many 14-3-3 proteins (Sun et al. 2011), which led us to investigate the composition and possible functions of FACs in cotton. We initially aimed to explore the nature of the molecular components in the FACs, and we hope in the future to apply the florigen–antiflorigen system to cotton improvement. Here, we describe a molecular analysis of cotton FACs aimed at answering the following questions: (i) In addition to GhFT, what are the 14-3-3 and FD components that comprise cotton FACs? (ii) Do multiple FD genes of a polyploid species form FACs in vivo? (iii) What are the similarities and differences among FD paralog functions? and (iv) Does the function of the FACs depend on the incorporated FD? We identified a 14-3-3 protein that interacts with GhFT in the cytoplasm and the nucleus, and we characterized five FD homoeologous gene pairs in the G. hirsutum genome. GhFT, Gh14-3-3 and GhFDs form FACs that are involved in floral induction and lateral root development. Thus, the pleiotropic functions of FACs and functional diversification of cotton FDs are important in processes other than flowering and support the hypothesis that FACs activities are modulated by FDs, depending on the developmental context.

Results

Cotton FT protein interacted with 14-3-3 in vitro and in vivo

To understand more precisely the molecular functions of FT in cotton, we used the yeast two-hybrid (Y2H) system to screen for proteins interacting with GhFT. We first constructed a complementary DNA (cDNA) library from the mixed samples of shoot apices and ovules at a titer of 1.2 × 10⁶ colony forming units ml^–1. Polymerase chain reaction (PCR) amplification of randomly picked clones revealed that the insert size ranged from ∼500 to ∼3,000 bp, with an average of ∼1,500 bp (Supplementary Fig. S1). Full-length GhFT was used as the bait to screen this library, and many colonies were identified on the selective medium (data not shown). Among these, one clone appeared at least three times in the screening. Sequencing and a nucleotide alignment identified it as GH_D04g1722, which contained a full-length coding sequence (CDS) encoding 14-3-3 protein 6-like with the highest similarity to Arabidopsis thaliana general regulatory 2 (14-3-3 protein) encoded by At1G78300 (Supplementary Fig. S2). Consequently, this interacting protein was named Gh14-3-3. A swapping experiment using the full-length GhFT and Gh14-3-3 either as bait or as prey further confirmed that the GhFT and Gh14-3-3 proteins interacted in the yeast system (Fig. 1A). Furthermore, GhFT interacted directly with Gh14-3-3 in glutathione S-transferase pull-down assays (Fig. 1B). Our previous studies detected fluorescence from GhFT tagged with green fluorescence protein (GFP) in both the cytoplasm and the nuclei of Arabidopsis root tip cells (Guo et al. 2015) or Nicotiana benthamiana leaf epidermal cells (Li et al. 2015). Here, fluorescence from GhFT tagged with GFP and from Gh14-3-3 tagged with GFP was observed in both the cytoplasm and the nuclei of Arabidopsis protoplast (Fig. 1C). Also, the GhFT–Gh14-3-3 interaction was monitored using bimolecular fluorescence complementation (BiFC) assays. It was detected in both the cytoplasm and the nuclei in N. benthamiana leaf epidermal cells (Fig. 1D). Gh14-3-3 was highly expressed in root, stem, leaf, flower and shoot apex tissues (Supplementary Fig. S3). Moreover, the ectopic overexpression of Gh14-3-3 in Arabidopsis delayed flowering (Supplementary Fig. S4). These data indicated that the GhFT–Gh14-3-3 complex is involved in floral transition in cotton. To date, however, there are few reports on the FD protein family in cotton. We are interested in further clarifying the molecular roles of FT, 14-3-3 and FD during the floral transition and throughout cotton development. Because we did not identify cotton FD homologs using the Y2H screening, we next attempted a genome-wide identification of the FD family members of cotton.

Identification and analysis of the Gossypium FD gene family members

The genome-wide analysis identified four loci in Gossypium arboreum (A₂) and five loci in Gossypium raimondii (D₅), as well as 10 loci in G. hirsutum (AD)₁ and G. bardadense (AD)₂ genomes with two homoeologous genes in each A_t and D_t subgenomes (Supplementary Table S1, Supplementary text). Compared with G. raimondii, there are only four FD genes in the G. arboreum genome; therefore, a Basic Local Alignment Search Tool (BLAST) analysis was performed using the whole-genome shotgun contigs database with GrFD as the query. One whole-genome shotgun sequence (GenBank accession number:AYOE01032781.1)from the G. arboreum cultivar Shixiya1 was obtained, which contains one gene highly similar to GrFD4. Therefore, the extant diploid, G. arboreum, also has five FD paralogs. A total of 30 FD genes were identified from cotton genomes. The basic information on the candidate GhFD genes is shown in Supplementary Table S2.

To analyze the phylogenetic relationship among the FD gene family members, we first identified 13 FD genes from six plant species, including A. thaliana (2), Oryza sativa (2), S. tuberosum (3), Populus trichocarpa (2), Corchorus olitorius (2) and Theobroma cacao (2) (Supplementary Table S3, Supplementary text), by searching public genomic sequence databases. Conserved amino motifs in these proteins were identified using sequence alignments by aligning and visual inspection. A phylogenetic tree using the amino acid sequences of 43 FD proteins was constructed, and the result indicated that evolutionarily, the cotton FD genes were closest to those of T. cacao and C. olitorius (Supplementary Fig. S5). Both T. cacao and C. olitorius genomes contain two FD homologs. Further phylogenetic analyses of FD members in more closely related plants, including Gossypium, T. cacao and C. olitorius clustered the cotton FD genes into five subclades. TcFD1 and CoFD1 were evolutionarily closest to cotton FD1 and FD2, while TcFD2 and CoFD2 were closest to cotton FD3, -4, and -5 (Supplementary Fig. S5).

A phylogenetic analysis of two Arabidopsis FDs and 10 upland cotton FDs indicated that GhFD1 and GhFD2 were evolutionarily closest to AtFD (Supplementary Fig. S6A). Additionally, cotton FDs share a similar intron–exon distribution with AtFD and AtFDP, with three exons and two introns in each FD gene (Supplementary Fig. S6B. The similarity in gene structure indicates the functional conservation of FD genes in cotton. Among these 12 FDs, the lengths of the first exon differed significantly, while the lengths of the second and third exons were similar. Amino acid sequence alignment revealed that cotton FD proteins share a high sequence similarity to Arabidopsis FDs (Supplementary Fig. S7). An analysis of domain structures showed that all the cotton FD proteins contained five conserved motifs found in Arabidopsis FD, motif A [(M/V)EEVWKDINLSXLHD], LSL [T(A/V)LSLN], basic region [N-X₇-R/K], leucine zipper [L-X₆-L-X₆-L] and T/SAP [RXX(S/T)(A/T)F]. The LSL motifs in the GhFD1, GhFD2 and GhFD3 homoeologous proteins share a strong similarity, whereas the LSLs in GhFD4 and GhFD5 homoeologs varied. The bZIP domain in plants consists of two structural features, a basic region and a heptad L repeat (Dröge-Laser et al. 2018). GhFD homoeologous pairs contain a highly similar basic region of 16 amino acids that include the invariant N-X₇-R/K motif and a leucine zipper (L-X₆-L-X₆-L) (Supplementary Fig. S7), which suggests that the bZIP protein’s functions are conserved, including localization to the nucleus and the ability to form dimers. The C-terminal T/SAP motif is a key region for the interactions between 14-3-3 and FD homologs. S/T residues within the SAP motif may be phosphorylated by calcium-dependent protein kinases, and this phosphorylation is essential for the recognition of OsFD1 by 14-3-3 proteins (Abe et al. 2005, Taoka et al. 2011, Kawamoto et al. 2015). All the GhFD proteins share a T residue at the 207th position with Arabidopsis FD and FDP, and only in GhFD5 is the alanine (A) at the 208th position replaced by T (Supplementary Fig. S7).

Expression profiling of GhFD homoeologous genes

To explore the possible roles of the FD gene family during cotton development, we first investigated their expression patterns in various organs and during developmental stages of ovules and fibers using the publicly available RNA sequencing (RNA-seq) data (Hu et al. 2019). All the genes, except for GhFD-3, were expressed [as fragments per kilobase of exon model per million mapped reads (FPKM) > 1] in 15 vegetative and reproductive tissues (Fig. 2A). GhFD1 and GhFD5 showed obviously biased expression patterns in the tested tissues. GhFD1-A was weakly expressed at 0 day post-anthesis (DPA) in ovules and had a very low expression level in other tissues, whereas GhFD1-D was expressed highest at 0–3 DPA in ovules and at 25 DPA in fibers. Interestingly, GhFD2 homoeologs showed clearly biased expression levels between A- and D-subgenomes. GhFD2-A was predominantly expressed in petals and filaments. However, GhFD2-D was highly expressed in leaves, sepals, pistils and ovules at 0–5 DPA, and fibers at 25 DPA. GhFD4 homoeologs showed similar high expression levels in vegetative and floral tissues, and the expression levels of GhFD4-A in ovules and fibers, except the latter at 20 DPA, were significantly greater than those of GhFD4-D. GhFD5-A exhibited very low expression levels in various tissues, while GhFD5-D was highly expressed in all the tested tissues, including a significantly higher expression in ovules at 0–5 DPA. The differential expression profiling of FD homoeologs indicated that the functions of cotton FD transcription factors diverged during cotton domestication, and the high expression levels of GhFD1-D, GhFD2-D, GhFD4-A and GhFD5-D in vegetative and reproductive tissues suggested potential roles in regulating developmental transitions in cotton.

Cotton is highly recalcitrant in terms of regeneration, and it is difficult to generate transgenic plants. Therefore, we next used Arabidopsis (Col-0) as a transgenic receptor to explore the spatial expression patterns of GhFDs. We obtained five GhFDproβ-glucuronidase (GUS) reporter constructs in which the ∼2.0 kb sequence upstream from the start codon of each GhFD-D was independently fused to the GUS gene. More than 10 independent T₁ transgenic Arabidopsis plants for each construct were created. Histochemical GUS staining was analyzed in T₂ plants at two developmental time points: 7 days after germination (DAG) of seedlings and 14 DAG of seedlings. At 7 DAG, the GUS activities were mainly observed in shoot apices, cotyledons, hypocotyls and the primary roots in GhFD1pro:GUS and GhFD2pro:GUS transgenic lines (Fig. 2B, C). In GhFD3pro:GUS transgenic lines, GUS was mainly expressed in cotyledons, but it was also weakly expressed in the shoot apices and at the lateral root meristems (Fig. 2D). In contrast, GUS genes were weakly expressed in the vasculature of cotyledons and the primary roots in GhFD4pro:GUS and GhFD5pro:GUS transgenic seedlings (Fig. 2E, F). However, strong GUS staining was observed in the whole seedlings at 14 DAG, including roots, shoot apices, cotyledons and the expanding leaves of GhFD1pro:GUS plants (Fig. 2G). Strong GUS staining was also observed in cotyledons, shoot apices, hypocotyls and primary roots, but GUS was weakly expressed in the vasculature and the distal tip of expanding leaves in GhFD2pro:GUS plants (Fig. 2H). In GhFD3pro:GUS plants, GUS was strongly expressed in cotyledons and lateral root primordia, moderately expressed at the shoot apices, but weakly expressed at the apices of true leaves (Fig. 2I). In GhFD4pro:GUS and GhFD5pro:GUS transgenic seedlings, the GUS genes were mainly expressed in the leaf vasculatures, the shoot apices and the roots (Fig. 2J, K). The strong GUS activities in the roots of the older seedlings indicated that the FDs play important roles in regulating root development, and the varied GUS activities in young shoot apices suggested differential regulatory behaviors among GhFD orthologous genes.

In Arabidopsis, FD is preferentially expressed in the shoot apices, where the FD protein interacts with FT to form a complex required for FT function (Abe et al. 2005, Wigge et al. 2005). No RNA-seq data are available for shoot apices; therefore, we further validated the expression profiling of GhFDs in roots, stems, leaves, flowers and shoot apices using quantitative real-time PCR (qRT-PCR). The amplification curve analysis revealed that the absolute value of the slope for each gene is close to zero (Supplementary Fig. S8), verifying that the amplification efficiencies of each GhFD and the reference are the same. The expression patterns of GhFDs detected by qRT-PCR were similar to those detected in the RNA-seq data, suggesting that the gene expression values of GhFDs obtained by RNA-seq were reliable (Fig. 3, Supplementary text). Furthermore, all five GhFDs were differentially expressed in shoot apices, with GhFD1 expression being the highest, followed by GhFD2, which suggested that they play important roles in controlling meristem activities.

Silencing of GhFDs affected flowering time and lateral root development

We first used virus-induced gene silencing (VIGS) to explore the putative functions for five GhFDs in cotton development. To explore the possible roles of GhFD genes in regulating flowering, we observed the phenotypic changes in the mature VIGS plants. A 67 days post-agroinfiltration, buds emerged in succession in the TRV:00 (control), TRV:GhFD3 and TRV:GhFD5 plants (Supplementary Fig. S9A–C). In contrast, the TRV:GhFD1-, TRV:GhFD2- and TRV:GhFD4-infected plants exhibited more late buds and showed more vegetative growth than the control (Supplementary Fig. S9D–F). A statistical analysis of the numbers of the nodes of the first fruit branches revealed that the architectures in TRV:GhFD3 and TRV:GhFD5 plants resembled that of the control, whereas plants in which GhFD1, -2, and -4 were silenced independently exhibited more vegetative growth (Supplementary Fig. S9G, H). A transcript analysis using qRT-PCR verified that each of the GhFD and GhAP1.1 (GH_D13G0875) genes were reduced in these plants (Supplementary Fig. S9H, I). The results indicated that the functions of GhFD1, -2, and -4 are involved in cotton flowering.

At 15 days post-agroinfiltration, the newly emerging true leaves showed photobleaching phenotype in plants with GhCLA1-expressing agrobacteria silencing the cotton cloroplastosalterados 1 gene (Gao et al. 2011), indicating that VIGS had been successful (Supplementary Fig. S10A). A gene expression analysis using qRT-PCR revealed that the expression levels of GhFDs in young roots decreased by more than 60% compared with the control (Supplementary Fig. S10B–F). To investigate the effects of GhFD silencing on root formation, the roots of these VIGS plants were subjected to a root scan system (Fig. 4A). No significant phenotypic differences were observed between the control lines and the plants treated with TRV2:GhFD1 and TRV2:GhFD2. However, silencing GhFD3, -4, and -5 independently significantly affected root growth and development. In an examination of the lateral roots traits of these three VIGS plants, TRV2:GhFD3 plants showed 58% and 55% decreases in total root length and total root surface area, respectively, compared with the control lines. TRV2:GhFD4 and TRV2:GhFD5 plants showed 47% and 37% reductions in total root length, respectively, compared with the control lines, and 46% and 33% reductions in total root surface area, respectively (Fig. 4B, C). In Arabidopsis, the formation of lateral roots is regulated by two auxin response factors, ARF7 and ARF19, which directly regulate the transcription of auxin-induced genes (Okushima et al. 2007). The ARF homologous genes in G. hirsutum, GhARF7 and GhARF19, were downregulated in these three VIGS plant lines (Fig. 4D, E), suggesting that GhFD3, -4, and -5 activate the transcription of ARF genes in cotton. Moreover, the expression levels of GhARF7 and GhARF19 in GhFD3 VIGS plants were only 0.5 and 0.3 times greater, respectively, than in control lines. Thus, the reduction of ARF gene expression in cotton resulted in the decrease in lateral root development. At 28 days post-agroinfiltration, we scanned the roots of these plants again and obtained results similar to those obtained at 15 days post-agrofiltration (Supplementary Fig. S11). To test whether silencing of GhFT and GhARF19 affected root development, we next used VIGS of GhFT and GhARF19 in cotton. The silenced plants were selected and their roots were scanned (Supplementary Fig. S12A–C). Silencing of GhFT and GhARF19 significantly decreased the lengths of lateral roots (Supplementary Fig. S12D, E), suggesting that GhFT, GhFD and GhARF are involved in the lateral root developmental pathway.

Effects of GhFD genes’ overexpression on Arabidopsis growth and development

To explore whether the heterologous expression of GhFD genes affects the growth of Arabidopsis, we generated transgenic Arabidopsis plants independently overexpressing each of five cotton FD genes having C-terminal FLAG tags under the control of the constitutive cauliflower mosaic virus 35S promoter. We selected a representative line harboring each transgene for further analyses. A western blot analysis using anti-FLAG revealed that FD1–FLAG, FD2–FLAG, FD3–FLAG, FD4–FLAG and FD5–FLAG merged proteins were expressed in the transgenic plants (Supplementary Fig. S13). We observed that the ectopic overexpression of each FD-FLAG did not affect the development of A. thaliana (Col-0), but affected the flowering time. Under long day (LD) conditions, 35S:GhFD1-FLAG and 35S:GhFD2-FLAG transgenic plants flowered significantly earlier than Col-0, and 35S:GhFD4-FLAG transgenic plants weakly promoted flowering (Fig. 5A, B). The numbers of rosette leaves in these three transgenic lines were also less than in Col-0 (Fig. 5C). In contrast, the overexpression of GhFD3–FLAG and GhFD5–FLAG did not promote flowering in Arabidopsis (Fig. 5A-C). Furthermore, a gene expression analysis using qRT-PCR indicated that the flower meristem identity genes, AtAP1 and AtFUL, were significantly upregulated in GhFD1, -2, and -4-overexpressing lines (Fig. 5D).

In addition to affecting flowering time, there were no obvious phenotypic differences in the aerial parts among these transgenic plants. We next observed the phenotypes of the underground parts of these transgenic plants (Supplementary Fig. S14A). Interestingly, overexpressing GhFDs in 14-day-old seedlings of Arabidopsis affected lateral root development. Among these transgenic plants, 35S:GhFD3–FLAG and 35S:GhFD4–FLAG plants had more lateral roots than the control. The number of lateral roots increased more significantly in 35S:GhFD3–FLAG seedlings [13.7 ± 1.2 (mean ± standard deviation)] in Col-0 and 19.4 ± 1.2 in 35S:GhFD3–FLAG (n > 20, P < 0.001) and in 35S:GhFD4–FLAG seedlings [16.2 ± 4.0 (n > 20), P < 0.001] (Supplementary Fig. S14B). Furthermore, the lengths of lateral roots in 35S:GhFD3–FLAG [65.6 ± 4.7 (n > 20), P < 0.001] and 35S:GhFD4–FLAG [50.3 ± 1.9 (n > 20), P < 0.001] were significantly longer than those in Col-0 plants (29.6 ± 2.1) (Supplementary Fig. S14C). By comparison, GhFD3 not only significantly increased the lateral root number, but also increased the lateral root length, suggesting its function in controlling lateral root development. Gene expression patterns, as assessed by qRT-PCR, showed that AtARF7 and AtARF19 were upregulated the most among all the transgenic plants, except AtARF19 in 35S:GhFD5–FLAG plants (Supplementary Fig. S14D), suggesting that GhFDs directly or indirectly activate the transcription of ARF genes. Two genes in 35S:GhFD3–FLAG had the highest expression levels, being ∼2.5-fold and 3.5-fold the wild-type levels. Thus, GhFD1 and GhFD2 appear to share conserved functions involved in flowering, whereas GhFD3 is mainly involved in lateral root development.

Subcellular localizations of GhFDs and the interactions of GhFT, Gh14-3-3 and GhFDs

The family of the FD bZIP transcription factor predominantly localizes in the nucleus (Abe et al. 2005, Taoka et al. 2011). In this study, we monitored the subcellular location of each GhFD–GFP fusion protein in the leaf epidermal cells of N. benthamiana using a confocal laser scanning microscope. We observed the fluorescence signals of five GhFDs tagged with GFP exclusively in the nuclei, unlike the cells expressing GFP alone, which had signals in both the cytoplasm and the nuclei (Supplementary Fig. S15). The characteristics of nuclear location indicated that cotton FDs, as transcription factors, regulate specific transcriptional events in the cell, including the induction of flowering.

We obtained a 14-3-3 protein that interacts with GhFT using Y2H screening with GhFT protein as the bait (Fig. 1A), and this Gh14-3-3 interacted with GhFT in both the cytoplasm and the nuclei in N. benthamiana epidermal cells (Fig. 1D). We next determined whether GhFDs can interact with GhFT and Gh14-3-3 using BiFC assays. When constructs encoding nYFP–Gh14-3-3 and each cYFP–GhFD were co-expressed in N. benthamiana epidermal cells, each Gh14-3-3–GhFD BiFC signal was exclusively detected in the nuclei (Supplementary Fig. S16A), where GhFDs were mainly localized (Supplementary Fig. S15). Similarly, we found that all five GhFT–GhFDs’ BiFC signals also exclusively appeared in the nucleus (Supplementary Fig. S16B). Furthermore, the GhFT–Gh14-3-3 BiFC signals appeared both in the cytoplasm and in the nuclei (Fig. 1B), suggesting that the interactions of GhFT with GhFDs were mediated by an endogenous 14-3-3 protein to form the FACs (FAC–GhFDs). Moreover, Y2H experiments similarly demonstrated that the five GhFDs physically interacted with Gh14-3-3 and GhFT in vivo (Supplementary Fig. S16C, D). Thus, the dimerized GhFT–Gh14-3-3 may potentially interact with each GhFD to form FACs in cotton apical cells.

Transcription of GhAP1.1 and GhARF19 is activated by the GhFT–GhFDs complexes

The transcription of Arabidopsis floral homeotic genes, such as AP1, is dependent on FT and FD (Abe et al. 2005, Wigge et al. 2005), while the transcription of the rice AP1 ortholog, MADS15, is dependent on Hd3a and OsFD1 (Taoka et al. 2011). We first used a dual-luciferase (LUC) reporter in an Arabidopsis protoplast system, in which the transcriptional activation of the cotton AP1 homolog, GhAP1.1, was used as a transcriptional read-out for the ability of FT and FD to act together (Fig. 6A). Compared with the GAL4-BD negative control, when transfected with an effector plasmid that encoded GhFD1 protein alone, the LUC activity significantly increased (2.8 times of the control, P < 0.05) (Fig. 6B), suggesting that GhFD1 activated GhAP1.1 transcription. However, transfection with each of the other four GhFDs independently barely activated GhAP1.1 transcription. Thus, GhFD1 alone may have an activation ability, and therefore, acts as an activator of transcription through binding the promoter of GhAP1.1. Next, GhFT and each GhFD were expressed together. When GhFT was co-expressed with GhFD1, GhFD2 or GhFD4, the LUC activity was significantly higher than without GhFT. The relative LUC activity values were 4.8-, 3.6- and 2.7-fold, respectively, compared to that of the control (Fig. 6B), indicating that the three GhFT–GhFD complexes increased GhAP1.1 transcription. In contrast, when GhFT was co-expressed with GhFD3 or GhFD5, the LUC activity was very close to that produced by GhFT alone, indicating that the GhFT–GhFD3 or GhFT–GhFD5 complex did not promote GhAP1.1 transcription. Thus, although five GhFDs interact with GhFT, as demonstrated by Y2H and BiFC assays (Supplementary Fig. S16), the functions of the five GhFT–GhFD complexes differed from each other as assessed by the dual-LUC assays.

Next, we analyzed the transcriptional activation of the cotton ARF homolog, GhARF19 (GH_D04G0088) by GhFDs, using the same experimental method. When co-expressed with GhFT, all five GhFDs activated GhARF19 transcription. Among the five GhFDs, the GhFD3 protein had the strongest activation effect (Fig. 5C). GhFD3 alone significantly increased LUC activities (to ∼4.5 times that of the control), suggesting that GhFD3 activated GhARF19 transcription. When GhFT was co-expressed with GhFD3, the LUC activity value was 6.4-fold that of the control (Fig. 6C), indicating that the GhFT–GhFD3 complex increased GhARF19 transcription. Importantly, the differences among the GhFT–GhFD complexes coincided with the observed phenotypic differences among the five GhFD-overexpressing lines (Fig. 5, Supplementary Fig. S4A).

Discussion

The published research revealed that there is only one FT gene locus in each genome (A and D) of cotton. However, the TFL1/SP genes in cotton have three loci, which is different from many other plants that contain multiple FT homologs. This implies that cotton genomes contain more complex florigen–antiflorigen system. Therefore, we aim to investigate the functional basis of GhFT, explore FAC components and analyze the molecular nature of FACs in cotton in this study.

GhFT interacts with one cotton 14-3-3 protein in the cytoplasm and the nucleus

The 14-3-3 proteins are important for binding and regulation, and they participate in various physiological processes. After the phosphorylation of the target protein by serine or threonine, 14-3-3 proteins bind the phosphorylated motif. Thus, 14-3-3s act as a scaffold capable of bridging two proteins (de Boer et al. 2013). We have repeatedly screened the same 14-3-3 protein for strong interactions with GhFT in cDNA libraries and found that both the GhFT and Gh14-3-3 proteins were localized in the cytoplasm and the nucleus (Fig. 1C). In vivo and in vitro assays confirmed that GhFT interacts with Gh14-3-3 (Fig. 1A, B), and a BiFC experiment showed that the two proteins interact strongly in the cytoplasm and the nucleus (Fig. 1D). This interaction does not use a similar mode to that of Hd3a–GF14b, which occurs in the cytoplasm, where GF14b is mainly localized (Taoka et al. 2011). Gh14-3-3 is constitutively expressed in all tissues of cotton, similar to the expression patterns of rice GF14b and potato 14-3-3 (Taoka et al. 2011, Teo et al. 2017). The ectopic expression of Gh14-3-3 in Arabidopsis delayed flowering (Supplementary Fig. S4), and a similar result was found by overexpressing GF14c in rice (Purwestri et al. 2009), indicating that Gh14-3-3 interacts with GhFT as a negative regulator of flowering. The 14-3-3 proteins have large numbers of isoforms in plants, and different 14-3-3 isoforms are redundant or play specialized roles in certain developmental processes (de Boer et al. 2013). Cotton is also rich in 14-3-3 isoforms (Sun et al. 2011), and they may interact with GhFT, because the four 14-3-3 proteins (GF14b, c, e and f) in rice all interact with Hd3a and are functionally redundant in regulating flowering (Taoka et al. 2011). Therefore, it is necessary to clarify whether cotton 14-3-3 isoforms interact with GhFT during flowering transition or other processes. However, here, we observed that this 14-3-3 interacted strongly with GhFT, suggesting that the GhFT–Gh14-3-3 complex recruit different FD components to form FACs, prompting us to identify the FD members in cotton.

FD gene expansion, conserved structures and FD-like proteins′ functional diversification

The bZIP transcription factors have evolved conservatively in eukaryotic organisms. The bZIP family in Arabidopsis genome has 78 known members, which are categorized into 13 groups (Dröge-Laser et al. 2018). FD and FD paralogs belong to the A-bZIPs subgroup, which was first identified as flowering regulators (Abe et al. 2005, Wigge et al. 2005). Our genome-wide analysis identified five FD homoeologous gene pairs from cotton AD genomes that were group A members, which is consistent with the results of Wang et al. (2020). Compared with two close relatives, T. cacao and C. olitorius, each having two FD genes, the cotton FD genes have undergone genome expansion during evolution (Supplementary Fig. S5). Cotton’s five FD homoeologous pairs not only have similar gene structures (Supplementary Fig. S6), but they are also highly similar at the amino acid level. The five motifs are also highly conserved, especially the bZIP and S/TAP motifs (Supplementary Fig. S7).

Because transcription factor FD binds the promoter regions of target genes and activates their expression, the expansion of cotton FD genes suggests a plethora of functions related to developmental transitions other than flowering. First, the five homoeologous pairs show diversified expression profiles in a wide variety of tissues, and most of the GhFD homoeologs are biased toward the D_t subgenome (Fig. 2). GhFD1 and GhFD2 were highly expressed in SAM, whereas other GhFDs showed only weak expression, suggesting that not all five GhFDs function in flowering regulation. Moreover, a phylogenetic analysis showed that five GhFDs are divided into two subclades, with GhFD1-A/D and GhFD2-A/D being in the same branch as AtFD, while the remaining GhFD genes were in another branch (Supplementary Fig. S5). Acting as the key component of the FT–FD complex, Arabidopsis FD is required for FT function in flowering promotion, and the overexpression of FD in Arabidopsis significantly induces flowering (Abe et al. 2005, Wigge et al. 2005). In this study, the silencing of GhFD1, -2, and -4 independently by VIGS delayed flowering, whereas silencing GhFD3 and GhFD4 reduced lateral root formation in cotton seedlings (Fig. 4, Supplementary Fig. S10). Interestingly, VIGS of GhFT and GhARF19 also decreased lateral growth (Supplementary Fig. S11), suggesting that a pathway containing FT, FD and ARF homologous proteins in cotton regulates lateral development. Although FD proteins found in the flowering plants examined contain functionally important conserved C-terminal S/TAP motifs, functional diversification has been reported in several plants. For example, FD1 in rice promotes flowering, but FD2 promotes leaf development (Tsuji et al. 2013). The FD-like genes in hybrid aspen display neofunctionalization, with only FDL1 being involved in the photoperiodic control of growth (Tylewicz et al. 2015). In potato, StFDL1a and StFDL1b RNA interference (RNAi) significantly delayed tuberization, but the phenotype of StFD RNAi plants was not affected (Teo et al. 2017).

Ectopic expression of GhFD genes in Arabidopsis affects flowering time and lateral root development

In this study, the ectopic expression of GhFD1 and GhFD2 independently significantly promoted flowering in Arabidopsis, while that of GhFD4 weakly promoted flowering. However, the ectopic expression of GhFD3 and GhFD5 did not affect flowering time (Fig. 5). The heterologous expression suggested that the FD functions have diverged among these paralogs. GhFD1 and GhFD2 mainly promote flowering, and the remaining GhFDs may function in other developmental transitions. Only the overexpression of GhFD1 and GhFD2 significantly promoted flowering, indicating that other cotton FDs function in developmental transitions other than flowering. Not unexpectedly, GhFD3 overexpression in A. thaliana more obviously promoted lateral root numbers and lengths (Supplementary Fig. S14). The ectopic expression assays revealed that GhFD1 and GhFD2 affected flowering times in Arabidopsis, whereas GhFD3 overexpression affected lateral root development.

GhFT forms a ternary complex with each FD protein through Gh14-3-3

The rice FAC assembly mode explains the molecular basis of florigen in that Hd3a interacts with the receptor 14-3-3 and then binds to FD1 to form a FAC that activates flowering genes and induces flowering (Taoka et al. 2011). Because the ectopic expression of GhFD1 and GhFD2 significantly promoted flowering in Arabidopsis, we speculated that they interacted with GhFT and Gh14-3-3 to form FACs that activate floral homeotic genes.

Subcellular localization analyses showed that GhFT and rice Hd3a share a subcellular location as do Gh14-3-3 and rice GF14b (Fig. 1C). BiFC assays showed the GhFT–Gh14-3-3 BiFC signal occurred both in the cytoplasm and in the nuclei (Fig. 1D), indicating that Gh14-3-3 is an intracellular receptor in cotton. Moreover, five GhFDs were localized exclusively in the nuclei (Supplementary Fig. S15), which is consistent with the localization of rice FD1 (but OsFD2 is localized in the cytoplasm) (Taoka et al. 2011, Tsuji et al. 2013). The interactions between Gh14-3-3 and each GhFD occurred mainly in the nuclei (Supplementary Fig. S16A). In contrast, Gh14-3-3 is localized in both the cytoplasm and the nuclei, indicating that GhFDs promote the translocation of Gh14-3-3 from the cytoplasm to the nucleus, which is similar to FD1 in rice. Similarly, the interactions between each GhFT and GhFD only occurred in the nuclei (Supplementary Fig S16B). Because FT cannot bind to DNA, this implies that in response to GhFDs, the GhFT–Gh14-3-3 complexes in the cytoplasm translocate from the cytoplasm to the nucleus, where they interact with GhFDs and the existing GhFT–Gh14-3-3 complexes to form a larger FAC. The data shows that the GhFT–Gh14-3-3 complexes in the nuclei recruit different GhFDs to from different FACs, which perform related functions. This model is very similar to, but not the same as, that of rice FAC–OsFD1 (Taoka et al. 2011), in which the transcription factor FD is illustrated to be a core FAC member. Moreover, GhFD3 silencing significantly decreased the lateral root formation in cotton (Fig. 4). These results were in agreement with the expression profile of GhFD3 which was specifically expressed in the lateral root meristems (Fig. 2D, I), and they indicated that FAC–GhFD3 functions in lateral root formation, although this needs further exploration. Thus, the five FACs are formed in the apical cells and are involved in a plethora of functions in cotton.

Furthermore, the five GhFDs contain the conserved bipartite nuclear localization signal and leucine zipper motifs, suggesting similar nuclear localizations and bound DNA sequences. However, two amino acids in the C-terminal basic regions of GhFD3 and GhFD5 differed from the others (Supplementary Fig. S7), and this slight difference may alter the target genes bound by these two FDs, resulting in differential functions during flowering. The S/TAP motif at the C-terminus of the FD protein is a key region in which 14-3-3 interacts with FD through the phosphorylation of S or T residues (Abe et al. 2005, Kawamoto et al. 2015). The cotton FDs, except for FD5, have the same conserved TAP motif as the Arabidopsis FD. TAP is replaced by TTP in GhFD5, but this mutation did not affect the interaction between Gh14-3-3 and GhFD5, which indicates that FAC–GhFD5 binds to different target genes than the other FACs.

Transcriptional activation experiments showed that only GhFD1 strongly activated GhAP1.1 transcription (Fig. 6). In rice, however, FD1 or Hd3a alone does not activate MADS15 transcription, and only co-expression of Hd3a and FD1 activates the expression of MADS15 (Taoka et al. 2011). In this study, when GhFT was co-expressed with GhFD1, GhFD2 or GhFD4, but not with GhFD3 or GhFD5, GhAP1.1 transcription was significantly activated (Fig. 6B), which is consistent with the flowering promotion phenotype that occurred in Arabidopsis when each of these three genes was overexpressed (Fig. 5). Interestingly, GhFD3 alone and the GhFT–GhFD3 complex strongly activate GhARF19 transcription (Fig. 6C), which is consistent with the lateral root reduction in TRV:GhFD3 plants (Fig. 4). These data indicate that the recruitment of a different FD to the FAC results in the targeting of different genes. OsFD2 also forms another FAC in rice, but the co-expression of OsFD2 and Hd3a did not induce OsMADS15 transcription (Tsuji et al. 2013). These results strongly suggest that the five cotton FD transcription factors have significantly different targets genes. GhAP1.1 may be the main target gene of GhFD1 and GhFD2, whereas GhARF19 is the main target gene of GhFD3. The target genes of the other GhFDs may be other MADS-box genes or genes in other pathways. Therefore, we speculate that the five cotton FDs form FACs with different target genes and modes of action, which explains the complexity of the functions and mechanisms of florigen–antiflorigen systems in polyploids.

Thus, the five GhFDs interacted with GhFT and Gh14-3-3 to from five FACs. In SAM, in the absence of GhFT, only GhFD1 activated GhAP1.1, indicating that GhFD1 promotes flowering independent of GhFT. Additionally, the transcriptional activation of GhFD2 and GhFD4 is dependent on GhFT. When forming ternary complexes in the nuclei, FAC–GhFD1, FAC–GhFD2 and FAC–GhFD4 each activate GhAP1.1, leading to floral induction. Therefore, GhFDs also have FT-dependent functions, and these three FAC–GhFDs control flowering with partially functional redundancy. On the basis of phenotypic differences, FAC–GhFD1 and FAC–GhFD2 play important roles in controlling flowering. In addition, in the nuclei of the root cells, FACs (formed with GhFD3, -4, and -5) activate the transcription of key regulators of lateral root development such as ARF19 (Fig. 6B). In comparison, FAC–GhFD3 mainly regulates lateral root development, but how the FT protein migrates to roots needs further investigation. Further research using techniques such as gene editing to explore the exact roles of each GhFD will offer a deep insight into the molecular nature of FACs. Manipulating the balance between florigen and antiflorigen signals in cotton further, through gene-editing techniques, holds promise for fine-tuning maturation processes and improving plant architecture, farming practices and field performances.

Materials and Methods

cDNA library construction and the yeast two-hybrid assay

The seeds of G. hirsutum cultivar ‘Xinluzao 33’ were field-grown under natural conditions in Shihezi (Xinjiang, China). To construct the cDNA library, shoot apexes were harvested at the fourth-true-leaf expanding stage in approximately 4-week-old seedlings after planting. Cotton bolls were also sampled at the following time points: 3 DPA and both 3 and 5 DPA from ovules containing fibers. All samples collected were immediately frozen in liquid nitrogen and stored at −80°C. Total RNA for each tissue was extracted using the Polysaccharides & Polyphenolics-rich RNAprep Pure Plant kit (Tiangen, Beijing, China) in accordance with the manufacturer’s protocol. PolyA mRNA was isolated using the Oligotex mRNA kit (Qiagen, Hilden, Germany) following the manufacturer’s protocol using mixed RNA containing an equal amount of total RNA from each tissue. A cDNA library was constructed in the pGADT7 vector using the above mix mRNA, and this library represented 10⁷ independent cDNA clones. As bait, the full-length GhFT was inserted into the PstI/EcoRI sites of the GAL4-binding domain in the pGBKT7-GAL4 vector (Clontech, California, America), and the resulting vector was introduced into the AH109 yeast line. The screening of GhFT interactors was performed as described previously (Taoka et al. 2011). The screening was performed on synthetic dropout (SD) medium lacking, histidine, leucine, tryptophan and adenine (SD–His–Trp–Leu–Ade) with 3 mM 3-aminotriazole. Plasmids extracted from positive yeast colonies were transformed into Escherichia coli (DH5α) for amplification, and plasmids isolated from positive E. coli colonies were used for the retransformation of yeast. The obtained positive plasmids were transformed into DH5α for sequencing.

Genome-wide identification of FD genes in cotton

Putative cotton FD gene members were identified by BLASTP searches against the Cotton Functional Genomics Database (http://www.cottonfgd.org) using Arabidopsis FD protein sequences (GenBank accession numberNP_195315.3), which were downloaded from the Arabidopsis Information Resource 10 database (http://www.arabidopsis.org), as query. The cotton FD genes with e-values less than 1e^–10 were used for further analyses. All the retrieved putative proteins were further submitted to PFAM (http://pfam.xfam.org/), SMART (http://smart.embl-heidelberg.de/smart/batch.pl) and InterProScan (http://www.ebi.ac.uk/interpro/) for annotations of domain structures. MEGA5.0 (Tamura et al. 2011) was used to perform a phylogeny reconstruction analysis using the neighbor-joining method and Poisson correction distance model. A bootstrap analysis was performed to estimate nodal support based on 1,000 re-samplings. Exon–intron structures were constructed by aligning genomic DNA and full-length cDNA sequences using the Gene Structure Display Server (GSDS2.0) (http://gsds.cbi.pku.edu.cn/).

Expression profiling of GhFD genes

To analyze the expression patterns of the GhFD homoeologous gene pairs in different tissues and fiber developmental stages, the RNA-seq Atlas of seven tissues, roots, stems, leaves, sepals, petals, pistils and filaments; ovules from 0, 1, 3 and 5 DPA; and fibers from 10, 15, 20 and 25 DPA, were download from PRJNA490626 (Hu et al. 2019). Subsequently, the values of FPKMs were used to measure the gene expression level for each FD with Cufflinks (Version 2.2.2) (http://cufflinks.cbcb.umd.edu/) (Trapnell et al. 2012). A heatmap of the expression levels was constructed using R software with log-transformed FPKM [log2(FPKM)] the values. The method for the qRT-PCR analysis of gene expression levels in tissues is provided in Supporting information.

Ectopic expression of cotton FD genes in Arabidopsis

A full-length CDS for each FD gene was amplified by RT-PCR using gene-specific primers (Supplementary Table S4), subcloned into the Gateway entry vector, pDONR-Zeo using the BP recombination reaction, and sequenced. The subsequent products were recombined with the destination pB35GWF (Gou et al. 2010) vector by an LR recombination reaction to generate 35S:GhFD-FLAGs. The resulting 35S:GhFD-FLAG plasmids were introduced independently into Agrobacterium tumefaciens GV3101. The seeds of A. thaliana were sown in pots containing soil:vermiculite:perlite (2:1:1) and cultivated in a phytotron at 22°C under LD conditions (16 h light/8 h dark, 200 µmol m^–2 s^–1). Transgenic plants were generated using the floral dip method (Clough and Bent 1998) and screened on half-strength MS culture medium containing 50 µg ml^–1 Basta. The homozygous transgenic lines of T₃ were identified and selected for phenotypic and molecular analyses. Detailed methods for comparing flowering times, lateral root numbers and lateral root lengths between control and transgenic plants are provided in Supporting information.

VIGS assays of GhFD genes

Seeds of G. hirsutum were planted in pots containing soil:vermiculite (2:1) and cultivated in a phytotron at 22°C under LD conditions. For the VIGS assays, five fragments of GhFDs were cloned by PCR and inserted separately into the XbaI and KpnI sites of pTRV2 vectors to generate pTRV-GhFDs. The pTRV-GhCLA1 was used as a control (Gao et al. 2011). The transformation and cultivation of Agrobacterium strain GV3101, as well as the VIGS assays, were performed as previously described (Gao et al. 2011, Si et al. 2018). When the albino phenotype of pTRV-GhCLA1 cotton leaves became visible, the total root lengths and surface areas were scanned and analyzed using a plant analysis system (WinRHIZO STD4800 LA2400, Canada).

BiFC assay

The CDSs of GhFDs, GhFT and Gh14-3-3 were separately amplified and cloned into the pDONRZeo vector (Invitrogen) using the BP recombination reaction. GhFT and Gh14-3-3 were fused with PVYNE by LR reaction, and the GhFD coding region was amplified and fused into the PSCYCE vector (Waadt et al. 2008). The resulting plasmids were transformed independently into A. tumefaciens GV3101, which was then infiltrated into 3-week-old N. benthamiana leaves for transient expression. The BiFC assays and the observations of cell fluorescence were performed as previously described (Si et al. 2018). Primers used to construct the BiFC plasmids are listed in Supplementary Table S4.

Transactivation activity analysis of Arabidopsis protoplasts

For effector constructs, the full-length CDSs of GhFD and GhFT were amplified by PCR, and the resulting fragments were fused independently to a downstream sequence encoding the GAL4-binding domain (GAL4BD) in the BamHI and KpnI sites of 35S-GAL4 BD as effectors. The empty GAL4BD was used as the negative control. For reporter constructs, approximately 1.8 kb of the promoter sequences of GhAP1.1 (GH_D13G0875) and GhARF19 (GH_D04G0088) were amplified independently by PCR. The resulting fragments were fused into the EcoRV and XbaI sites of GAL4-LUC, containing the firefly luciferase reporter gene driven by the 35S minimal TATA box and 5× GAL4BD, generating reporter plasmids containing GhAP1.1 or GhARF19 promoters fused to luciferase. Arabidopsis protoplasts were prepared as described previously (Yoo et al. 2007). In cotransfection assays, equal amounts of effector and reporter plasmids (6 μg) were used. For the normalization of values, 0.5 μg of the Renilla reniformis luciferase construct driven by the 35S promoter was used as an internal control in each reaction. After culturing in the dark for 16 h, luciferase assays were performed with the dual-LUC reporter assay system (Promega, E1960) and the GloMax™ 20–20 Luminometer, in accordance with the manufacturers’ protocols. The primer sequences used are shown in Supplementary Table S4.

Supplementary Data

Supplementary dataare available at PCP online.

Data Availability

The data underlying this article are available in the article and in its online supplementary material.

Acknowledgements

We thank Professor Xiangdong Fu (Institute of Genetics and Developmental Biology, Chinese Academy of Sciences) for the critical comments on the manuscript. This work was supported by grants from the National Natural Science Foundation of China (Nos. 31860393 and 31360366), the Talent Introduction Start-up Fund Project of Anhui Science and Technology University (No. NXYJ202001), and the Technological Innovation Leading Talents of Xinjiang Production and Construction Corps (No. 2006BC001).

Disclosures

The authors have no conflicts of interest to declare.

References