Functional Divergence of AP1 and FUL Genes Related to Flowering Regulation in Upland Cotton

The AP1/FUL transcription factors are important for oral development, but the underlying molecular mechanisms remain unclear. In this study, we cloned and identied two AP1/FUL-like genes, GhAP1.1 and GhFUL2, in upland cotton, which is a commonly cultivated economically valuable crop. Sequence alignments and phylogenetic analyses indicated GhAP1.1 and GhFUL2, which are encoded by genes in the AP1/FUL clade, have conserved N-terminals, but diverse C-terminal domains. A quantitative real-time PCR analysis revealed that GhAP1.1 and GhFUL2 were expressed in the ower and root, and had the opposite expression patterns during different shoot apical meristem stages. The upregulated expression of GhAP1.1 in Arabidopsis and the silencing of GhAP1.1 did not induce signicant changes to the owering time or oral organs, but the transcript levels of the origen FT gene and the AP1 homolog GhMADS42 increased. The overexpression of GhFUL2 in Arabidopsis delayed owering and promoted bolting by decreasing the FT and LFY expression levels. Silencing GhFUL2 in cotton dramatically increased the expression of GhFT and GhMADS42 and promoted owering. Additionally, yeast two-hybrid and bimolecular uorescence complementation assays indicated that GhAP1.1 can interact with the SVP homolog GhSVP1, whereas GhFUL2 can form heterodimers with SEP1, SEP4 homologs, and GhSVP1. Therefore, we proved that the functional divergence of GhAP1.1 and GhFUL2, which involved changes in sequences and expression patterns, inuenced the regulation of cotton ower development.


Introduction
Gene duplications, which are prevalent events during plant evolution, occur via whole-genome duplications or tandem duplications. Several studies con rmed the high frequency of gene duplications in land plants, including Arabidopsis, cotton, wheat, and tobacco (Flagel and Wendel 2009;Sun et al. 2014;Yang et al. 2018;Vision et al. 2000). Gene duplications during evolution increase the number of genes in the genome and may be associated with functional divergence (Huang et al. 2020;Yuan et al. 2015). The MADS-box transcription factors play important roles in gene regulatory networks controlling oral transitions as well as ower and fruit development Yang and Jack 2004;Preston and Kellogg 2006;Nishikawa et al. 2009).
The angiosperm APETALA1 (AP1)/FRUITFULL (FUL) gene lineage is an important MADS-box clade in which the number of genes increased because of duplication events. During the key gene duplication event, the AP1/FUL clade was divided into the core eudicots and non-core eudicot species. The core eudicot genes included Arabidopsis AP1, Arabidopsis FUL, and Antirrhinum sp. SQUAMOSA, whereas the non-core eudicot genes were similar to the Arabidopsis FUL gene (Litt and Irish 2003). Although the AP1/FUL sequences are similar, with both encoding the conserved MADS-box and K-box sequences, and are derived from the same paralog, the AP1 and FUL proteins vary regarding the motifs in the C-terminal domain (Zahn et al. 2005;Shan et al. 2007). The functional divergence between AP1 and FUL resulted from changes in important transcriptional regulators and coding sequences, especially unique conserved motifs (McCarthy et al. 2015). Thus, changes to the gene sequences encoding C-terminal domains and other protein regions may lead to functional divergence.
The functions of the AP1/FUL-like genes diverged during plant evolution. These genes have important functions in uencing plant growth and developmental processes, including oral transitions, fruit ripening, and the opening of the apical hook Chen et al. 2015;Li et al. 2016;Führer et al. 2020). In Arabidopsis, AP1 affects the positioning of organs as well as sepal and petal identity (Irish and Sussex 1990), whereas FUL inhibits cell division and promotes cell expansion, while also regulating the transcription of cellular differentiation-related genes during fruit development (Gu et al. 1998).
Additionally, AP1 combined with the ful mutation leads to a non-owering phenotype and altered in orescence architecture (Ferrándiz1 et al. 2000). In rice, the AP1/FUL transcription factor gene OsMADS18 and its two paralogs OsMADS14 and OsMADS15 have different functions related to plant development. Both OsMADS14 and OsMADS15 specify the palea and lodicule identities (Wu et al. 2017), whereas OsMADS18 is involved in delaying seed germination, altering plant architecture, and decreasing the number of tillers (Yin et al. 2019). In Brachypodium distachyon (Poaceae species), four AP1/FUL paralogs have been identi ed, with only BdVRN1 expressed normally during a prolonged cold treatment; the ectopic expression of the other three genes reportedly leads to early owering and severe morphological alterations to oral organs (Li et al. 2016). In tomato, the overexpression of the AP1/FUL homolog MBP20 results in simple leaves and modulated leaf development (Burko et al. 2013). Functional analyses of transgenic tomato plants revealed that FUL2 overexpression might result in fruit and leaf morphological changes, whereas the suppression of FUL1 and FUL2 expression can inhibit fruit ripening through ethylene biosynthesis and the regulation of ripening-related gene expression during tomato fruit ripening processes .
Cotton is an economically important crop cultivated worldwide. The MADS-box family genes have been identi ed in several cotton species. For example, 11 AP1/FUL clade genes were identi ed in tetraploid cotton and were subsequently analyzed in terms of their structures and expression pro les (Nardeli et al. 2018;Jiang et al. 2014). Among the genes in the AP1/FUL clade, GhAP1.7 promotes owering in Arabidopsis and is regulated by GhLFY in speci c pathways (Cheng et al. 2021b). A recent study con rmed GhAP1 interacts with GhCAL to form a heterodimer that may regulate GhAP1 expression, with GhAP1-silenced plants exhibiting signi cantly late owering (Cheng et al. 2021a). Although a few AP1 genes have been functionally characterized, other AP1/FUL homologs and the associated regulatory mechanisms remain unknown. In this study, two AP1/FUL homologs were cloned from cotton cultivars exhibiting differential maturation. Moreover, the expression of these homologs in shoot meristems at several stages was analyzed. The results of this study may be useful for clarifying the functional divergence of genes in cotton and related species.

Plant materials and growth conditions
Cotton cultivars CCRI50 (late maturing with a normal fruit branch) and ZAO1 (early maturing with a clustered fruit branch) were grown at 28 °C in a greenhouse at the Henan Institute of Science and Technology. Different tissues were sampled from CCRI50 plants. Shoot meristems were harvested from both CCRI50 and ZAO1 plants from the four-leaf expansion stage to the seven-leaf expansion stage. For phytohormone treatments, cotton seeds were sown in plastic pots, which were then incubated at 28 °C with a 14-h light/10-h dark cycle until seedlings reached the two-leaf expansion stage (about 4 weeks). The cotton seedlings were treated with 100 µM abscisic acid, gibberellin, or salicylic acid. The leaves were harvested at 1, 3, 6, 12, and 24 h post-treatment, immediately frozen in liquid nitrogen, and stored at −80°C until analyzed. The wild-type (WT) and GhAP1.1and GhFUL2-overexpressing transgenic Arabidopsis lines were grown in a culture room at 22 °C with a 16-h light/8-h dark cycle. Tobacco plants were grown in a greenhouse at 25 °C with a 14-h light/10-h dark cycle.
Phylogenetic and gene sequence analysis A phylogenetic tree was constructed according to the neighbor-joining method using MEGA5.05. The GhAP1.1 and GhFUL2 coding sequences were ampli ed from the cDNA of CCRI50.

Gene expression pro le analysis
Total RNA was extracted from samples using the RNA Pure Plant Plus kit (Tiangen). The puri ed RNA served as the template for synthesizing rst-strand cDNA using the PrimeScript RT reagent Kit with gDNA Eraser (Takara). A quantitative real-time (qRT)-PCR assay was performed in 384-well plates using the ABI Q6 system (ABI) and SYBR Green Premix Ex Taq (Takara). The GhACTIN and AtUBQ7 genes were selected as the internal controls for cotton and Arabidopsis, respectively. The 2 −ΔΔCt method was used to calculate relative gene expression levels.

Virus-induced gene silencing (VIGS) assay
For the VIGS assay, a 315-bp fragment of GhAP1.1 and a 291-bp fragment of GhFUL2 were ampli ed from the cDNA of CCRI50 and inserted into the pCLCrVA vector to construct recombinant plasmids. The pCLCrVA-GhAP1.1 and pCLCrVA-GhFUL2 plasmids were inserted into separate Agrobacterium tumefaciens GV3101 cells. The GV3101 cells containing pCLCrVA-GhAP1.1, pCLCrVA-GhFUL2, pCLCrVA (negative control), or pCLCrVA-PDS (positive control) were mixed with cells carrying pCLCrVB (1:1 ratio). The cells were cultured until the OD 600 reached 2.0, after which they were injected into 10-day-old cotton cotyledons. After 8 weeks, the cotton plants were analyzed regarding their gene expression pro les and phenotypes. The inoculated cotton plants were grown in a greenhouse at 22 °C with a 16-h light/8-h dark cycle. The analysis was repeated at least three times.
Yeast two-hybrid (Y2H) assay The GhAP1.1 and GhFUL2 coding sequences were cloned into the pGBKT7 vector according to the In-Fusion cloning method to construct the bait vectors. The coding sequences of eight GhSEP genes and four GhSVP genes were ampli ed from cotton cDNA and ligated to pGADT7 to construct the prey vectors. The bait and prey vectors were used to transform Y2HGold yeast cells as described in the yeast transformation system user manual (Clontech). Protein interactions were screened on SD/−Trp/−Leu double dropout medium (DDO) and SD/−Trp/−Leu/−His/−Ade quadruple dropout medium (QDO) supplemented with aureobasidin A (AbA) and X-α-galactosidase (X-α-Gal) (Clontech). The primers used are listed in Supplementary Table S1.

Bimolecular uorescence complementation (BiFC) assay
To verify protein interactions in vivo, the GhAP1.1 and GhFUL2 coding sequences were inserted into the pUC-SPYNE vector, whereas partial sequences of the SEP and SVP homologs were inserted into the pUC-SPYCE vector. The resulting recombinant plasmids were inserted into A. tumefaciens GV3101 cells, which were then introduced into Nicotiana benthamiana leaves using needleless syringes for the subsequent coexpression analysis. More speci cally, the transiently transformed tobacco plants were incubated for 3 days (post-injection) at 22 °C with a 14-h light/10-h dark cycle. The uorescence in the lower epidermal cells of tobacco leaves was observed 72 h later using a confocal microscope (Zeiss LSM780).

Results
Sequence alignment and phylogenetic analysis of AP1/FUL genes in cotton An analysis of upland cotton genomic data revealed ve AP1 and four FUL orthologs of the Arabidopsis AP1 and FUL genes (Fig. 1a). The encoded amino acid sequences included a conserved MADS domain, a K domain, and a FUL motif. With the exception of GhAP1.2, the cotton AP1 orthologs encoded the AP1 motif. In contrast, the four FUL genes in cotton encoded the paleoAP1 motif. Additionally, a comparison with the AtAP1 protein sequence indicated the GhAP1.1 and GhAP1.2 were more similar than GhAP1.3 (Fig. 1b). These results suggest the diversity in the C-terminal domain encoded by MIKC-type genes may be associated with functional divergence.
Expression pro les of AP1/FUL genes at different shoot meristem stages and in response to phytohormone treatments Cotton cultivars ZAO1 and CCRI50 were used to analyze shoot meristem expression levels from the fourleaf expansion stage to the seven-leaf expansion stage. The data revealed that GhAP1.1 was more highly expressed in ZAO1 than in CCRI50, whereas the opposite pattern was observed for GhFUL2 ( Fig. 2a and  2c). The GhAP1.1 and GhFUL2 expression levels in different cotton tissues and after phytohormone treatments were examined in a qRT-PCR assay. The results indicated GhAP1.1 was highly expressed in the ower and apical bud, whereas GhFUL2 was highly expressed in the root and apical bud (Fig. 2b). The three phytohormone treatments upregulated the GhAP1.1 and GhFUL2 expression levels. The GhAP1.1 transcript levels peaked at 3 h, after which they decreased to control levels (Fig. 3a). The GhFUL2 expression levels following the abscisic acid and salicylic acid treatments also peaked at 3 h and then rapidly decreased. In contrast, following the gibberellin treatment, GhFUL2 expression gradually increased and peaked at 24 h (Fig. 3b).

Ectopic expression of GhAP1.1 and GhFUL2 in Arabidopsis
To functionally characterize GhAP1.1 and GhFUL2, binary vectors were constructed for the subsequent examination of the constitutive expression of these genes in transgenic Arabidopsis (ecotype Columbia-0). Eleven and nine independent kanamycin-resistant T 1 transgenic plant lines were generated for 35S::GhAP1.1 and 35S::GhFUL2, respectively. Three T 3 GhAP1.1 lines owered 2 days earlier and produced 2-3 fewer rosette leaves than the WT plants under long-day conditions, whereas the phenotypes of the other lines did not differ from the WT phenotype (Table 1). However, six GhFUL2overexpressing transgenic plants owered 5-6 days later and produced 2-3 more rosette leaves than the WT plants under long-day conditions (Fig. 4a-d and Table 1). The expression of GhAP1.1, GhFUL2, and owering-related genes in transgenic plants was analyzed by qRT-PCR. The results revealed that FT, AP1, and FUL expression levels were upregulated and the LFY expression level was downregulated in the three GhAP1.1-overexpressing transgenic Arabidopsis lines that owered earlier (Fig. 4e). In the GhFUL2overexpressing transgenic Arabidopsis lines, the expression of both FT and LFY was signi cantly downregulated compared with the corresponding expression in the WT plants, whereas FUL expression was signi cantly upregulated (Fig. 4f).
Silencing GhAP1.1 and GhFUL2 expression promoted cotton owering Virus-induced gene silencing assays were completed to verify the GhAP1.1 and GhFUL2 functions. The empty pCLCrVA vector was used as a negative control, whereas pCLCrVA-PDS served as the positive control. The appearance of white leaves and stems indicated the VIGS was successful (Fig. 5a). A qRT-PCR analysis con rmed the GhAP1.1 and GhFUL2 expression levels decreased signi cantly in the VIGS plants ( Fig. 5c and 5f). Compared with the control plants, the owering time was accelerated only for the plants in which GhFUL2 was silenced (Fig. 5e). According to the qRT-PCR assay data, the GhAP1.1 and GhFUL2 expression levels were signi cantly lower in the VIGS plants than in the negative control plants.

Interactions between GhAP1.1/GhFUL2 and GhSVP/GhSEP-like proteins in Y2H and BiFC assays
The AP1/FUL subfamily proteins may interact with SOC1/SVP/SEP to form heterodimers with regulatory roles. The cotton genome contains more of these MADS-box genes than the Arabidopsis genome. Previous research proved that GhAP1/FUL can interact with GhSOC1 in vivo and in vitro (Zhang et al. 2016). In the current study, eight SEP-like genes and four SVP-like genes were cloned to verify interactions in Y2H and BiFC assays. The Y2H assay results indicated GhAP1.1 can interact with GhSEP5 and GhSVP1, whereas GhFUL2 can interact with GhSEP2, GhSEP3, GhSEP4, GhSEP5, GhSEP6, GhSEP7, and GhSVP1 (Fig. 6). These interactions were veri ed by BiFC assays involving 1-month-old tobacco leaves.
The YFP signals detected in the nucleus and/or membranes re ected the interactions of GhAP1.1 with GhSVP1 and of GhFUL2 with GhSEP2, GhSEP3, GhSEP4, GhSEP5, GhSEP6, and GhSVP1 (Fig. 7). These ndings demonstrated that both GhAP1.1 and GhFUL2 can form heterodimers with GhSVP1, but only GhFUL2 can form a heterodimer with SEP1 and SEP4.

Discussion
Upland cotton is an economically valuable tetraploid crop. Polyploidization, which has been a major force driving plant evolution, has resulted in gene duplications and the expansion of gene families. Gene duplication events have been accompanied by the development of new gene functions in uencing plant development as well as gene functional divergence and redundancy. The MADS-box genes encode important transcription factors modulating reproductive and vegetative developmental processes.
In upland cotton, 106 MIKC-type MADS-box genes have been identi ed, which is three-fold more than the number of corresponding genes in Arabidopsis. Moreover, these genes include eight AP1/FUL homologs (Nardeli et al. 2018). In the current study, we investigated the expression patterns and functions of two AP1/FUL genes in cotton. In an earlier study on Arabidopsis, FUL, which was negatively regulated by AP1, was undetectable in the young oral primordia, but it accumulated in the walls of the developing carpel (Mandel and Yanofsky 1995). Other studies con rmed that both AP1 and FUL genes encode ABCclass proteins that affect oral transitions, ower organ identity, and fruit development (Pelaz et al. 2001;Ellul et al. 2004;Lin et al. 2009;Shimada et al. 2009;Shulga et al. 2011;Pabon-Mora et al. 2012;Pabon-Mora et al. 2013). Among monocots, FUL-like genes in wheat and barley are responsive to vernalization (Trevaskis et al. 2007;Distelfeld et al. 2009), whereas rice FUL-like genes, including OsMADS14, OsMADS15, OsMADS18, and OsMADS20, have important functions related to inflorescence and floral meristem identity (Kobayashi et al. 2012b;Wu et al. 2017). In cotton, AP1 and FUL genes are reportedly differentially expressed in various tissues (Jiang et al. 2014). Consistent with this earlier observation, GhAP1.1 was highly expressed in the ower and apical bud, whereas GhFUL2 was highly expressed in the root and apical bud. The analyses of the shoot meristem transcriptomes of the two cotton cultivars examined in this study revealed differences in the GhAP1.1 and GhFUL2 expression pro les. More speci cally, GhAP1.1 expression was upregulated in ZAO1, whereas GhFUL2 expression was upregulated in CCRI50. These observations re ect the divergence in AP1/FUL expression during shoot apical meristem development in tetraploid cotton species.
The upland cotton genome contains homologs of Arabidopsis AP1, including GhMADS42, which can promote owering and alter oral organs when expressed in transgenic Arabidopsis plants (Zhang et al. 2016). The expression of another homolog, GhAP1.7, can induce precocious flowering. Additionally, LEAFY, which affects cotton oral meristem identity, can bind to the GhAP1.7 promoter and negatively regulate expression (Cheng et al. 2021b;Li et al. 2013). In a recent study, GhCAL, which is another AP1 homolog in cotton, was identi ed, and anti-GhCAL transgenic cotton plants exhibited delayed oral bud differentiation and owering (Cheng et al. 2021a). In contrast to these earlier results, we observed that the overexpression of GhAP1.1 in Arabidopsis did not induce early owering or alter the oral organ phenotype. Accordingly, GhAP1.1 may be a redundant gene in tetraploid cotton. The expression of a FUL ortholog, GhMADS22, in transgenic Arabidopsis can promote owering and result in the production of abnormal owers (Zhang et al. 2013), but the overexpression of GhFUL2 in Arabidopsis can delay owering and promote bolting. In Arabidopsis, the function of the FUL gene mimics the function of the SMALL AUXIN UPREGULATED RNA 10 (SAUR10) gene in stems and inflorescence branches (Bemer et al. 2017). In rice, the downregulated expression of OsMADS18 reportedly delays seed germination and young seedling growth, whereas the upregulated expression of OsMADS18 decreases the number of tillers (Yin et al. 2019). Considering the tillering in wheat is similar to the bolting in Arabidopsis, this result indicates that GhFUL2 may have functions affecting plant architecture. In our study, the upregulation of GhFUL2 expression decreased the expression of FT and LFY genes in Arabidopsis. Furthermore, FT expression was signi cantly upregulated in GhFUL2-silenced cotton plants. A previous study demonstrated that rice AP1/FUL-like genes (OsMADS14, OsMADS15, and OsMADS18) function upstream of the FT homologs Hd3a and RFT1 (Kobayashi et al. 2012a). In wheat, the protein encoded by the AP1/FUL-like gene VRN1 binds to the CArG-box of the FT-like gene WFT to upregulate expression (Tanaka et al. 2018). These results imply that FUL MADS-box transcription factors have diverse functions in uencing the owering time, oral organ formation, and the regulation of plant architecture in cotton.
The interactions among MADS-box proteins during oral development have been relatively well elucidated. Previous research proved that AP1 and SEP3 interact to control sepal development (Pelaz et al. 2001). Another study indicated FUL does not interact with SVP, but both AP1 and FUL interact with SOC1 and SEP4 (de Folter et al. 2005). Interestingly, in cotton, GhAP1.1 can interact with the SVP homolog GhSVP1, but not with SEP proteins, whereas GhFUL2 can form heterodimers with SEP1, SEP4 homologs, and GhSVP1. Three FUL proteins in Platanus acerifolia can interact strongly with the E-class (SEP-like) proteins (Zhang et al. 2019). The interaction between FUL and SVP has been observed in other plants (van Dijk et al. 2010;Balanz et al. 2014). Considering the interaction of SEP1, SEP4, or GhSVP1 with GhFUL2, we speculate that GhFUL2 may function cooperatively with SEP1, SEP4 homologs, or GhSVP1 to control the owering time and oral organ formation or the plant architecture.
In this study, we proved that GhFUL2 can function as a transcription factor with important roles related to oral development and plant architecture. In contrast, GhAP1.1 may be a redundant gene. Future investigations should focus on the SEP and SVP regulatory mechanisms in uencing oral organs and plant architecture. Doing so will expand our understanding of GhFUL2 functions mediating cotton development.

Declarations
Author contribution statement: XZ QM and SF conceived and designed research. XZ and GH conducted experiments and wrote the manuscript. HW and ZR revised the manuscript. All authors read and approved the manuscript.
Funding: This work was nancially supported by the State Key Laboratory of Cotton Biology (CB2018A08) and National Key project of Science and Technology (2020ZX08009-12B).  Expression patterns of GhAP1.1 and GhFUL2 in cotton in response to various hormone treatments.
Hydroponic cotton seedlings at the two-leaf expansion stage were treated with 100 µM ABA, GA, or SA, after which leaf and stem samples were harvested at different time-points. Total RNA was extracted and a qRT-PCR assay was performed to determine GhAP1.1 and GhFUL2 expression levels, with an Actin gene used as the internal control. (a) Expression pro les of GhAP1.1 following ABA, GA, and SA treatments. (b) Expression pro les of GhFUL2 following ABA, GA, and SA treatments. The signi cance of the data was determined using Student's t-test (*P < 0.05, **P < 0.01)   Bimolecular uorescence complementation (BiFC) assays in tobacco epidermal cells. The GhAP1.1 and GhFUL2 fragments were inserted into the pSPYNE vector, whereas GhSEP2, GhSEP3, GhSEP4, GhSEP5, GhSEP6, GhSEP7, and GhSVP1 fragments were inserted into the pSPYCE vector. The resulting recombinant pSPYNE and pSPYCE plasmids were transiently co-expressed in tobacco lower epidermal cells for an examination by uorescence microscopy. The detection of YFP uorescence indicated a positive interaction. Scale bar, 20 μm

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.