The MADS transcription factor GhAP1 coordinates the flowering regulatory pathway in upland cotton (Gossypium hirsutum L.)

Background: MADS-box gene family plays an important role in the molecular regulatory network of flower development. APETAL1 (AP1), a MADS-box gene, plays an important role in the development of flower organs. Although many studies about MADS-box family genes have been reported, the function of AP1 is still not clear in cotton. Results: In this study, GhAP1 (Gh_D03G0922), a candidate gene for cotton flower time and plant height obtained from our previous studies, was cloned from CCRI50 cotton variety and functionally characterized. Subcellular localization demonstrated that GhAP1 was located in nucleus. Infection test of Arabidopsis revealed that GhAP1 could cause precocious flowering and virus-induced gene silence (VIGS) assay demonstrated that GhAP1 could lead to delayed flowering of cotton plants. Yeast one-hybrid assays and transient dual-luciferase assays suggested that floral meristem identity control gene LEAFY (LFY) can bind the promoter of GhAP1 and negatively regulate it. Yeast two-hybrid assays suggested that GhAP1 can interact with pyridoxal phosphate (PLP)-dependent transferases superfamily protein PSAT2. Conclusions: Our research indicated that GhAP1 might work as a positive regulator in plant flowering. Moreover, GhAP1 may interact with GhPSAT2 and be negatively regulated by GhLFY in the regulatory pathways. This work laid the foundation for subsequent functional studies of GhAP1.

. Usually, the flowers of typical dicotyledonous plants consist of sepal, petal, stamen, pistil and carpel.
These five floral organs are controlled by organ characteristic genes. The flower growth ABC model revealed the molecular mechanism underlying flower development which was accepted generally [2]. With the deepening of research, D functional genes were found in Petunia hybrida and E functional genes were found in Gerbera hybrida and Oryza sativa [3][4][5] . I domain is less conservative, which is related with combination of DNA and protein [11]. According to different structures of I domain, these genes can be subdivided into MIKC C and MIKC * [12,13]. K domain is a semi-conservative region, it is homologous to the helix of keratin and has about 70 amino acids [14] and is the signature sequence of MIKC type MADS-box genes. Expect for individual genes of maize, all the MADS-box genes contain K domain in plants [15]. The C domain is the least conservative region comes after the K domain, which is a transcriptional activation region composed of hydrophobic amino acids [16]. Different C domains can lead to different functions [17]. The MADS-box gene family encodes transcription factors (TFs) that shares a highly conserved domain which can bind CArG box in promoter region of the MADS-box protein target genes [18]. Early maturity is one of the important traits of cotton. Early-maturing cotton is helpful for realizing both grain and cotton maturity, reducing the land conflict between grain and cotton. Flowering is one of the most important traits of early maturity. Many studies have shown that MADS TFs play important roles in all the stages of plant growth, especially flowering. Therefore, it is important to study the functional and molecular mechanism of MADS genes in cotton. In our previous studies of our lab, a MIKC type MADS gene, GhAP1 was identified as a candidate gene for cotton flower time [34] and plant height [35]. In this study, GhAP1 was cloned and characterized. The overexpression lines of GhAP1 in Arabidopsis and the VIGS plants of GhAP1 in cotton were used to explore its biological function. Yeast one-hybrid assays, transient dual-luciferase assays and yeast two-hybrid assays were performed to examine the molecular mechanism of GhAP1. Our findings help to enrich flowering regulatory pathways and provide genetic resources for genetic improvement of cotton varieties.

Gene structure and protein sequence analysis of GhAP1
Based on the results of genome-wide association study (GWAS) in our previous studies, GhAP1 (Gh_D03G0922) was selected according to gene annotation, and its Arabidopsis homologs contains MADS-box domains [35]. GhAP1 was located on D03 chromosome and its Arabidopsis homolog gene is Agamous-like MADS-box protein AGL8 (AT5G60910). The protein sequence similarity between GhAP1 and AGL8 was 56.2%. The genome DNA length was 5452 bp and the CDS length was 690 bp, encoding 229 amino acids (Fig. 1a). In addition, GhAP1 contained eight exons and seven introns (Fig. 1a). Protein sequence analysis showed its homologous genes of different species shared the high conserved MADS domain and the signature Kdomain of MIKC type MADS-box genes. A putative nuclear localization signal (NLS) sequence (ALKRIRSRKNQLM) was predicted in GhAP1 protein sequence (Fig. 1b).

Expression analysis of GhAP1 in different tissues and different varieties
Quantitative real-time PCR ( qRT-PCR) was performed to detect the expression levels of GhAP1 in roots, stems, leaves, buds, fiber, petals, and sepals. GhAP1 was found to be differentially expressed in different tissues. GhAP1 was strongly expressed in sepals and leaves but was weakly expressed in petals, roots, and fiber (Fig. 2a). To evaluate the expression pattern of GhAP1 during different growth stages, the expression level of GhAP1 was examined in the leaves of two early maturing varieties (CCRI50 and CCRI74) and two late maturing varieties (BM and G11) at different growth stages. The results revealed that the expression level of GhAP1 reached its maximum level in CCRI50 plants at the three-leaf stage. In addition, the expression levels of GhAP1 increased gradually and then decrease during growth stages and were higher in early maturing varieties than in late maturing varieties ( Fig. 2b).

GhAP1 was localized to the nucleus
GhAP1 was predicted by subcellular location software Plant-mPloc (http://www.csbio.sjtu.edu.cn/bioinf/plant-multi/) and the results showed that the GhAP1 protein was localized to the nucleus. To confirm our prediction, the 35S-GhAP1::GFP vector was constructed and injected into tobacco leaves. Using a fluorescence microscope, we found that the tobacco leaves harboring the 35S-GhAP1::GFP construct emitted green fluorescence predominantly in nuclei under blue light (Fig. 3a, 3b).

Overexpression of GhAP1 promoted flowering in Arabidopsis
To further study the function of GhAP1, the overexpression vector 35S::GhAP1 was constructed and transformed into Arabidopsis. qRT-PCR results showed that GhAP1 gene was hardly expressed in WT, but was highly expressed in transgenic plants, which was significantly higher in transgenic plants than that of non-transgenic Arabidopsis plants (Fig. 4a). Early flowering was observed in GhAP1-OE transgenic Arabidopsis plants (Fig. 4b). The flowering time was measured using the days after sowing (DAS) to first flowering. Through investigation, we found that WT plants started flowering at 29 DAS and had 11 rosette leaves, whereas transgenic lines started flowering at 24 DAS and had 8 rosettes. The transgenic Arabidopsis bloomed about 5 days earlier than the WT and had 3 rosette leaves less than WT (Table 1). SGN-VIGS Tool (http://vigs.solgenomics.net/) was used to analyze the GhAP1 sequence and 250 bp of GhAP1 was selected to construct pCLCrVA-GhAP1 vector.
pCLCrVA-PDS vector was served as a positive control and pCLCrVA vector was served as an empty control. The cotton plants harboring pCLCrVA-PDS showed an albino phenotype, suggesting that the VIGS assay was successful. qRT-PCR was performed to evaluate the effect of gene silencing, the results showed that the expression level of GhAP1 in positive plants was significantly lower than that in VA empty control (Fig. 5a). When flowering was observed in VA plants, the positive plants were not (Fig. 5b). Compared to VA plants, we found that positive plants flowered later.

GhAP1 interacted with GhPSAT2
A yeast library was constructed using the bud samples from two-leaf, three-leaf and five-leaf stages of early maturing variety CCRI50. pGBKT7-GhAP1 vector was build and transformed into Y2H Gold yeast cells. After three days both of pGBKT7 control and pGBKT7-GhAP1 can grow on SD-Trp medium plate, but can't grow and become blue on SD-Trp/-His/-Ade and SD-Trp/-His/-Ade/X-α-gal medium plate (Fig. 6a), indicating that GhAP1 had no self-activation activity. According to gene annotation and yeast two-hybrid validation, a pyridoxal phosphate (PLP)-dependent transferases superfamily protein GhPSAT2 (Gh_D07G1721) was identified to be interacted with GhAP1 (Fig. 6b). The research on GhPSAT2 was poor, but its homologous gene in soybean was proved to encode mitochondrial isozyme of aspartate aminotransferase (AAT4) when expressed in E. coli [36].
GhLFY bound to the promoter of GhAP1 in the yeast one-hybrid assay.
Three GhLFY binding sites (CCAATGG) were founded in promoter of GhAP1, which located at -80, -1500 and -1700 bp. In order to validate which site can interact with GhLFY, approximately 51 bp around these sites were selected respectively and inserted into the pHis2 carrier vector with three series repeats (called pHis2-L plasmid). When 120 mM 3-amino-1,2,4-triazole (3-AT) were added, their selfactivation could be inhibited. pGADT7-GhLFY vector was constructed and cotransformed into Y187 yeast cells with pHis2-L vector. The yeast cells contained pHis2-1500bp and pGADT7-GhLFY vector could growth on the SD-Trp/-His/-Leu/120mM 3-AT medium. These results indicate that GhLFY can directly bind the promoter of GhAP1 at the -1500 bp site (Fig. 7).  [42]. Overexpression of PaAP1 in Arabidopsis can trigger early flowering in the 35S::PaAP1 transgenic Arabidopsis lines [43]. GmAP1 could cause early flowering and alteration of floral organs when it was ectopically expressed in tobacco [44].

The expression of
These studies are similar to the results of this study. In the VIGS assay, compared with the control group, the experimental group showed a late flower phenotype, and the expression level of GhAP1 in the experimental group was significantly lower than that of the VA plants. Silencing the SBP-box gene in Snapdragon by VIGS technology will down-regulate the expression of multiple MADS-box genes, such as SQUA, AmFUL, etc., resulting in late flowering or even no flowering of plants [45].
Silencing of TaAG-A and TaAG-B in a fertile wheat line resulted in green and yellow striped leaves, emaciated spikes and reduced seed setting rate [46]. These studies increased the credibility of our study.

Molecular mechanism of GhAP1
Proteins interact with each other to diversify their functions. Yeast hybrid system is an efficient and rapid method for analyzing protein interactions. For example, OsMADS18 can interact with the seed-specific protein NF-YB by yeast two-hybrid system [47]. In addition, MADS-box proteins can regulate the temporal and spatial expression of target genes, and its own expression is also strictly regulated.
Previous reports showed that AP1 can be directly and positively regulated by LFY [48]. Promoter sequence analysis revealed that three GhLFY-binding sites (CCAATGG) were present in the promoter region of GhAP1. Yeast one-hybrid assay showed that GhLFY can interacte with the -1500 bp promoter site of GhAP1, indicating that GhLFY can regulate GhAP1 by binding to the -1500 bp site. By dualluciferase assay, it was found that the value of LUC/REN in the experimental group was significantly lower than that of the control group, indicating that GhLFY might negatively regulate GhAP1 expression. This is in contrast to the reporting pathway in Arabidopsis [ 49]. This may be due to the functional redundancy of GhLFY and GhAP1, which is considered to be implemented by a set of shared genes. Moreover, studies have shown that GhLFY and GhAP1 have antagonistic effects on target gene regulation [48]. Therefore, we suspected that this is an antagonist effect that leads to this result. As a complex biological phenomenon of plants, flowering was regulated by numerous genes. Therefore, it's understandable that there are different results, but this conjecture remains to be further proved.
As a TF, AP1 can regulate the expression of target genes, but most studies on signaling pathways comes to an abrupt end with AP1. However, little research has been done on its mechanism of action in cotton. In our study, the library screening experiment was carried out by using yeast two-hybrid. We found that a protein GhPSAT2 can interact with GhAP1. It is a mitochondrial isozyme of aspartate aminotransferase AAT4 in soybean [36]. Currently, there are few reports on GhPSAT2, indicating that it may be a new gene, and its mechanism of action on plant growth still needs to be further studied.

Conclusions
GhAP1 is a TF associated with flowering. It has a highly conserved MADS domain and belongs to the MADS-box gene family. GhAP1 is located in the nucleus.
Overexpressed has an obvious character of early maturity [50]. The growth period of CCRI74 is 100 days and it has obvious early maturity [51]. The growth period of G11 is about 123 days, which was identified by Guoxin rural technical service association [52]. BM plants have the characteristics of late emergence, tall and loose, it belongs to late maturing varieties [53]. As Gu described, these varieties are often used for testing and preservation in our laboratory [54]. All samples were quickly frozen in liquid nitrogen and stored at −80°C for subsequent experiments. In addition, two earlymaturing varieties, CCRI50 and CCRI74, and two non-early-maturing varieties, G11 and BM, were used for expression pattern analysis. The cotton varieties were planted in the field of the Cotton Research Institute of the Chinese Academy of Agricultural Sciences (Anyang, Henan, China). Different tissues, including roots, stems, leaves, buds, fiber, petals and sepals were harvested from CCRI50 [54].

Gene cloning and sequence analysis
To amplify the CDS and promoter of GhAP1 (Gh_D03G0922), we designed primers using Oligo7. The primers used in this study are listed in Additional file 1: Table S1.
The full-length CDS and promoter fragment of GhAP1 was cloned from cDNA and genes. The 2 -ΔΔCT method was applied to calculate relative expression levels [55].
All reactions were performed with three technical replicates.

Subcellular localization
The CDS sequence of GhAP1 without termination code was amplified from CCRI50 cDNA using PCR. GhAP1 was cloned into PBI121-GFP to construct 35-GhAP1::GFP vector and transformed into Agrobacterium tumefaciens strain LBA4404.
Approximately 200 μl LBA4404 strains were added into LB liquid medium containing kanamycin, rifampicin and streptomycin, shaking to OD600=1.8-2.0 at 28℃. The culture was centrifuged for 10 min at 4000 rpm. Adjust the OD600=1.5 using the transformation medium. The bacteria were left in the dark at room temperature for 3 h before injected into the tobacco leaves. The fluorescent signal was observed by fluorescence microscope after 2 days (dark culture for 24 h and normal culture for 24 h).

Genetic transformation of Arabidopsis
The CDS sequence of GhAP1 was inserted into PBI121 vector to construct 35S::GhAP1 and transformed into the Agrobacterium tumefaciens strain LBA4404 chemically competent cells. Arabidopsis was infected by dipping flower method [56]. as previously described [57,58].

Yeast two-hybrid
To identify the interaction proteins of GhAP1, the buds and leaves of CCRI50 from the two-leaf, three-leaf and five-leaf stages were used to construct yeast two-hybrid library. GhAP1 was cloned into pGBKT7 to construct pGBKT7-GhAP1 plasmids and transformed into Y2H yeast receptor cells. Three colonies were randomly selected for self-activation detection. The pGBKT7 empty vector was used as a negative control. The three clones were diluted with water and placed on the defect medium of yeast growth, namely, SD/-Trp, SD/-Trp/-His/-Ade and SD/-Trp/-His/-Ade/X-a-gal.
The mating method was used to screen yeast library on SD/-Trp/-His/-Ade/-Leu plates to identify the proteins that can interact with GhAP1. The Y2H yeast transformants containing the correct pGBKT7-GhAP1 was used to prepare the yeast competent cells, and the library plasmid pGBKT7-cDNA was transferred into it.
Screening was conducted on the defective culture plate of SD/-Trp/-His/-Ade/-Leu plates, and the normally growing spots were coated on the culture plate of SD/-Trp/-His/-Ade/-Leu with X-a-Gal for further screening. The spots that could grow normally and turn blue were considered to be positive spots. The positive clones were amplified via PCR and sequenced. Cotton genome blast analysis was performed to obtain the coding genes of potential proteins [59]. Field studies were conducted in accordance with local legislation. The plant materials used in this study were previously preserved in our laboratory.

Consent for publication
Not applicable.

Availability of data and materials
The data sets supporting the results of this article are included within the article and its additional file.

Funding
This research was support by the Shangdong TAISHAN Industry Leading Talent Program to Shuxun Yu (LJNY201608). The funders had no role in the design of the study, the collection, analysis, and interpretation of data, and in writing the manuscript.   Overexpressed GhAP1 in Arabidopsis and qRT-PCR results. a qRT-PCR analysis of the transcri  GhLFY bound to the promoter of GhAP1 in the yeast one-hybrid assay. pHis2-ProGhAP1 and p Figure 8 GhLFY inhibited the expression of GhAP1 in the dual-luciferase reporter assay. a Sketch of th