GhBES1 mediates brassinosteroid regulation of leaf size by activating expression of GhEXO2 in cotton (Gossypium hirsutum)

We proposed a working model of BR to promote leaf size through cell expansion. In the BR signaling pathway, GhBES1 affects cotton leaf size by binding to and activating the expression of the E-box element in the GhEXO2 promoter region. Brassinosteroid (BR) is an essential phytohormone that controls plant growth. However, the mechanisms of BR regulation of leaf size remain to be determined. Here, we found that the BR deficient cotton mutant pagoda1 (pag1) had a smaller leaf size than wild-type CRI24. The expression of EXORDIUM (GhEXO2) gene, was significantly downregulated in pag1. Silencing of BRI1-EMS-SUPPRESSOR 1 (GhBES1), inhibited leaf cell expansion and reduced leaf size. Overexpression of GhBES1.4 promoted leaf cell expansion and enlarged leaf size. Expression analysis showed GhEXO2 expression positively correlated with GhBES1 expression. In plants, altered expression of GhEXO2 promoted leaf cell expansion affecting leaf size. Furthermore, GhBES1.4 specifically binds to the E-box elements in the GhEXO2 promoter, inducing its expression. RNA-seq data revealed many down-regulated genes related to cell expansion in GhEXO2 silenced plants. In summary, we discovered a novel mechanism of BR regulation of leaf size through GhBES1 directly activating the expression of GhEXO2.


Introduction
Leaf size strongly influences the efficiency of photosynthesis in plants (Gonzalez et al. 2012). Brassinosteroids (BRs) are essential phytohormones that promote plant growth and a characteristic phenotype of BR-related mutants is reduced leaf size (Mao and Li 2020;Li and Chory 1997;Li et al. 1996;Praveena et al. 2020;Yang et al. 2011;Clouse and Sasse 1998;Nolan et al. 2017;Vert et al. 2005). In BR biosynthesis mutant constitutive photomorphogenesis and dwarfism (cpd), the leaves are small, round, and dark-green (Noguchi et al. 1999;Szekeres et al. 1996). Another BR-deficient mutant, deetiolated-2 (det2), has leaves that are small and round and twice as numerous as the wild type (Azpiroz et al. 1998;Chory et al. 1991). The transmembrane receptor kinase brassinosteroid insensitive1 (BRI1), has been associated with BR responses (Guo et al. 2013). Mutations in BRI1 result in BR-insensitivity and a morphological phenotype almost identical to that of the BR biosynthesis mutant cpd (Clouse et al. 1996). In rice, M107 is a gain-of-function mutant of the BR biosynthetic gene P450 CYP724B1 with a typical BR excess phenotype, including long and narrow leaves with greatly increased leaf angles (Wan et al. 2009). In addition, the positive regulatory protein complexes of the BR signaling pathway reduced leaf angle 1 (RLA1) and BRASSINAZOLE RESISTANT1 (OsBZR1) have mutants with a distinct, erect leaf phenotype (Qiao et al. 2017). In Gossypium hirsutum, GhPAG1 is highly homologous to AtCYP734A1, and its expression is activated in pagoda1 (pag1) mutants exhibiting dwarfism and smaller leaf size due to inhibition of cell expansion (Yang et al. 2014).
BR regulates expression of downstream genes mainly through the transcription factor (TF) BES1/BZR1 (BRI1-EMS-SUPPRESSOR 1/BRASSINAZOLE-RESISTANT 1) interacting with key elements, including E-box motif (CANNTG) and BR response element (BRRE; CGTGC/ TG) . BES1/BZR1 can directly bind to the promoter regions of CPD, dwarf4 (DWF4), and BR-6-oxidase (BR6OX) to regulate cell expansion in leaves (He et al. 2005). The bes1-D mutant, whose BES1 protein is widely expressed by an amino acid mutation, exhibits a constitutive BR response phenotype including long, bending petioles and curled leaves. Notably, multiple BR-induced genes are up-regulated or hyperresponsive to BR in bes1-D, including EXORDIUM (EXO), which is described as an AtPhi-1 phosphate-induced protein (Yin et al. 2002). BZR1/BES1 genes have been identified from many species such as Chinese cabbage , maize (Yu et al. 2018b), rice (Bai et al. 2007), and cotton ) and they are reported to have a conserved role in regulating BR signal transduction pathway.
The EXO gene was first reported as a plant growth regulator expressed in tissues with high auxin concentrations (Farrar et al, 2003). In addition, as a BR response gene, EXO expression is induced after exogenous 2,4-epibrassinolide treatment in Arabidopsis (Schröder et al. 2009;Coll-Garcia et al. 2004). Proteomics approaches identified the EXO, EXL1, EXL2, EXL3, and EXL4 proteins as part of the cell wall proteome and determined that they promote plant growth, cell expansion, and carbon starvation response through mediating BR (Schröder et al. 2012(Schröder et al. , 2009Coll-Garcia et al. 2004). In Arabidopsis, rosette leaves became larger in EXO overexpression transgenic plants compared with wild type, and increased transcript levels of BR response genes were observed (Coll-Garcia et al. 2004). However, elongation of abaxial epidermal and palisade cells in the subepidermal layer were inhibited in the exo knockout mutant based on scanning electron microscopy (SEM) showing a diminished leaf and root growth phenotype and a diminished response to BL (Schröder et al. 2009). OsEXO knockout plants exhibited partial dwarfism and had smaller cells in the culms, promoting cell expansion by regulating microtubule organization (Aya et al. 2014). EgEXO overexpression in Eucalyptus globulus promoted plant height and increased leaf biomass. In addition, the presence of two G-boxes in the promoter region suggests that it is regulated by the BES1 family of TFs (Sousa et al. 2020).
Cotton (mainly upland cotton, G. hirsutum) is a primary fiber crop and an excellent model to study polyploidization, genome evolution, and cell expansion (Malik et al. 2018;Ali et al. 2021). Cotton leaf size is an important agronomic trait that affects plant architecture, yield, and stress tolerance (Andres et al. 2016). Cotton leaf size has significant phenotypic diversity in crops and has a unique role in cotton development (Dolan and Poethig 1991). Only a few genes affecting cotton leaf development have been functionally characterized, such as LMI1 (LATE MERISTEM IDENTITY1, an HD-Zip transcription factor) and GhARF16 (Auxin response factor) (He et al. 2021;Andres et al. 2017). However, the mechanisms of cotton leaf development remain largely unknown.
Previous studies have shown that BR primarily regulates leaf size by affecting cell expansion, but the genetic mechanisms involved are unclear. In this study, the major TF GhBES1 in the BR signaling pathway was shown to act as a key regulator that binds to the GhEXO2 promoter and increases its expression, enhancing cell expansion and ultimately affecting leaf size. In brief, our data suggest that GhEXO2, the downstream target gene of GhBES1, functions in expanding cotton leaf cells in BR signaling to regulate leaf size.

Plant materials and growth conditions
Upland cotton (Gossypium hirsutum) mutant pag1 (BRdefective) and the corresponding wild-type (WT) CRI24 were obtained from the Institute of Cotton Research of the Chinese Academy of Agricultural Sciences. In this study, cotton plant materials were grown in a greenhouse with a photoperiod of 16 h light/8 h darkness at 28 °C. 1/2 Murashige and Skoog (MS) medium was used to germinate Arabidopsis thaliana seeds with long-day conditions at 23 °C (16 h light/8 h darkness). The Columbia-0 ecotype of Arabidopsis was used as WT. Arabidopsis seeds were germinated on 1/2 MS medium followed by growth in a greenhouse with long-day conditions as described above.

Phylogenetic tree construction, conserved motif and promoter cis-acting elements analysis
A significantly down-regulated gene was identified in the unpublished cotyledon transcriptome data of pag1. We downloaded the protein sequences from the CottonFGD database (https:// cotto nfgd. org/). The retrieved protein sequences were used to identify eight EXO genes in Arabidopsis from The Arabidopsis Information Resource (TAIR) database (https:// www. arabi dopsis. org/). A phylogenetic tree was generated in MEGA 7 using the NJ (neighborjoining) method (Saitou et al. 1987;Kumar et al. 1994). We performed conserved amino acid residue analysis via Jalview software after multiple alignments of the conserved Phi_1 domain and the signal peptide region using ClustalW (Waterhouse et al. 2009;Thompson et al. 2003). The protein domains were predicted with the online tool MEME (http:// meme-suite. org/ tools/ meme) as described previously (Yu et al. 2018a;Wang et al. 2018). The promoter sequence 2,000 bp upstream of the start codon was downloaded from the cottonFGD database (https:// cotto nfgd. org/). The cisacting elements in the promoter region were predicted on the web tools PlantCARE (http:// bioin forma tics. psb. ugent. be/ webto ols/ plant care/ html/) and PlantTFDB (http:// plant tfdb. gao-lab. org/ predi ction. php).

BL induction treatment assay, RNA extraction and qRT-PCR analysis
For BL treatment, 10 μM/L BL to treat cotton seedlings growing the three-four leaf stage for 1, 3 and 5 h. An equal volume of absolute ethanol (solvent) was added to deionized water as the Mock. Samples of plant tissues were immediately stored in liquid nitrogen, and the extracted RNA was stored at -80℃. Each experiment was conducted with three biological replicates. We extracted total RNA using the RNA prep Pure Plant Kit (TSINGKE, Beijing, China) based on the manufacturer's instructions. EasyScript® (One-Step gDNA Removal) (TransGen Biotech, Beijing, China) was used to synthesize cDNA from 1 μg of RNA. The internal controls were Actin2 (AT3G18780.1) and GhHistone3 (AF024716). AceQ qPCR SYBR Green Master Mix (Low ROX Premixed) (Vazyme, Nanjing, China) was used for Quantitative Real-time PCR (qRT-PCR) analysis on a Light-Cycler 480 (Roche Diagnostics, Germany). Values of relative expression patterns were calculated using the 2 −△△Ct method (Livak and Schmittgen 2001). 2 × Taq Plus Master Mix (Dye Plus) (Vazyme, Nanjing, China) was used for semi-quantitative RT-PCR analysis (Phillips et al. 1993).

Subcellular localization assay
We incorporated the full-length coding sequence of GhEXO2 into a vector containing a GFP region to analyze the GhEXO2 protein expression localization. Nicotiana benthamiana leaves were co-injected with Agrobacterium GV3101 containing the GhEXO2-GFP vector coupled with p19. The tobacco plant was placed in the dark for 16 h then returned to the light. A confocal microscope (OLYMPUS FV1200) was used to laser scan the epidermal cells of infiltrated tobacco leaves after 48 h of inoculation before which the leaves were stained with membrane dye FM 4-64 for 8 min for co-localization.

Generation of overexpression transgenic lines and virus-induced gene silencing
To verify the function of GhEXO2, we used Arabidopsis Col-0 plants to produce transgenic GhEXO2 lines. To construct the overexpression vector, the full-length coding sequence of GhEXO2 was amplified with gene-specific primers from cDNA and cloned into pCAMBIA-2300 with the 35S promoter. Arabidopsis plants were infected with Agrobacterium GV3101 containing 35S-GhEXO2-pCAM-BIA-2300 vector. The inflorescences of Arabidopsis plants were dipped in an Agrobacterium suspension to produce transgenic Arabidopsis plants (Clough and Bent 1998). For VIGS (virus-induced gene silencing), 300 bp highly specific coding sequences were cloned into the PTRV2 vector, which was then ligated into GV3101 before injection into cotton plants (CRI24) (Dinesh- Kumar et al. 2003).
For overexpression and RNA interference vector, the GhBES1 coding sequence was amplified with the corresponding primers, and separately constructed into pCAM-BIA-2300 and pBI121 vector driven by the 35S promoter. Notably, overexpression of the GhBES1.4 coding sequence introduces a point mutational modifications designed by PrimerX (http:// www. bioin forma tics. org/ prime rx/ cgibin/ DNA_1. cgi), as we previously reported . Both vectors were transformed into Agrobacterium LBA4404. The hypocotyls of 6-day-old sterile cotton seedlings were infected with Agrobacterium LBA4404 as explants. Then, the transgenic plants were created in the greenhouse through callus induction, proliferation, embryogenic callus induction, embryo differentiation, and plant regeneration stages. PCR and qRT-PCR were used to detect transgenic plants at the genome and transcription level, respectively. All primers used were presented in Table S2.

Observation of cotton tissue morphology
The selected tissue was immersed in FAA fixation solution before dehydrating it with an alcohol gradient. The tissues were made transparent with xylene and then embedded in paraffin. The paraffin-embedded tissues were then sectioned into 10 micron thick slices (Leica Instruments GmbH, Wetzlar, Germany). The sections were stained with toluidine blue and observed under a Leica M165FC epifluorescence stereomicroscope (Leica Instruments GmbH). Each sample was repeated at least three times.

Electrophoretic mobility shift assay
EMSA was performed using the LightShift Chemiluminescent EMSA Kit (Thermo Fisher Scientific, Shanghai, China) according to the previous description ). Biotin-labeled probe containing E-box (CANNTG) and mutant probe (AAA AAA ) from the GhEXO2 promoters was incubated with purified recombinant His-GhBES1 in binding buffer. Non-labeled probes were used as cold competitors. The reaction components were electrophoresed on a polyacrylamide gel using TBE buffer at 4 °C. The DNA in the gel was electroblotted onto a nitrocellulose membrane and detected by ChemiScope 6300 (CLINX, Shanghai, China).

Transcriptional activation assay
The 2000 bp promoters of GhEXO2, both the natural promoter and that mutated in the E-box region, were separately inserted into the pGWB435 vector with the luciferase reporter gene. Agrobacterium GV3101 containing GhEXO-2pro-LUC and GhEXO2pro-mutant-LUC were mixed with the above-mentioned GhBES1-pCAMBIA2300 vector coinjected into N. benthamiana leaves, and the empty vector pCAMBIA2300 was used as a control. The tobacco plants were grown in the dark for 24 h and then switched to a 16 h light/8 h dark cycle for 48 h at 23 °C. The fluorescein signal was detected in the Tanon 5200 Multi automatic chemiluminescence/fluorescence image analysis system (Tanon, Shanghai, China).

Biolayer interferometry assay
The Octet RED96 System (ForteBio) was used to assess the kinetics of DNA-protein binding using streptavidin-coated biosensors. GhBES1 protein and biotin-labeled probe on GhEXO2 promoter were used as described in the EMSA. After the sensor was immersed in the buffer, the biotinylated protein was bound according to Baseline, Loading and Wash, and then the binding of different sample wells and the sensor was detected according to Baseline, Association and Dissociation. All operations were as described in the manufacturer's instructions. Data were analyzed using OriginPro 2021 software.

RNA-sequencing and GO and KEGG analyses
For RNA-seq data analysis, the leaves of GhEXO2 silenced cotton plants were collected 30 days after germination. Screening of clean reads was performed after the removal of low-quality and contaminated reads as well as adaptors. Next, we mapped clean reads in the GRAND database (http:// grand. crica as. com. cn/) by Hisat2. StringTie was used to measure the expression levels and the values were normalized by FPKM (fragments per kilobase of transcript per million fragments mapped) (Pertea et al. 2015). Raw reads count for coding genes were obtained using String-Tie subscript prepDE.py (Kukurba and Montgomery 2015). R package 'EdgeR' was used to identify the differentially expressed genes (DEGs) between two groups (Robinson et al. 2010). To fine-tune the P-value for the false detection rate, Benjamini and Hochberg's approaches were used. DEGs were defined by fine-tuning P-value < 0.05 and absolute fold change value > 2. The GO and KEGG enrichments were conducted in the R package 'clusterProfiler' (Yu et al. 2012). RNA-seq data have been uploaded to the NCBI database, BioProject: PRJNA834593.

Expression of the GhEXO2 gene is down-regulated in pag1 mutants
pag1 exhibited BR-defective related phenotypes, particularly in smaller leaf size (Fig. 1A) (Yang et al. 2014). Cytological observation and microscopic analysis showed that cell expansion was inhibited in the leaves of pag1 mutant plants compared to wild-type (WT) CRI24 plants (Figs. 1B and S1A, D and G). A significantly down-regulated gene was identified in the hormone-related differentially expressed genes in cotyledon transcriptome data of pag1 (Fig. S2A). Sequence analysis showed that this gene has the highest similarity to AtEXL2 and has a conserved phi-1 structural plants. D Expression patterns of GhEXO2 in response to BL treatment as determined by qRT-PCR analysis. Each experiment was performed with three biological repeats and the error bars indicate the standard deviation among these replicates. Student's t-test: *P < 0.05, ***P < 0.001 domain (Fig. S3A), hence we named it GhEXO2. Expression of GhEXO2 was monitored in cotton plants, including CRI24 and BR deficient mutant pag1 in different tissues (roots, stem, leaves, and 20 days post-anthesis (DPA) fiber tissues). This gene had significantly down-regulated expression in pag1 plants compared to CRI24 cotton plants at all experimental stages (Fig. 1C). Exogenous application of BL induced its expression, which decreased after the peak at 1 h of treatment (Fig. 1D). These results suggested that GhEXO2 responds to BR treatment.

GhBES1.4 promotes cotton leaf size and regulates
GhEXO2 expression BES1 is the core TF regulating expression of downstream response genes in BR signaling to influence plant development and respond to environmental stress ). Our previous study identified 22 GhBES1 genes in upland cotton and found that GhBES1.4 was a functional gene mediating BR response . In order to explore if GhBES1 participates in BR regulation of leaf size, we created GhBES1 RNA interference and GhBES1.4 overexpression transgenic cotton in CRI24. We selected several GhBES1s with high homology to GhBES1.4 and verified the effectiveness of RNA interference in the GhBES1-RNAi plants ( Fig. S4A and B). In addition, the transcription level of GhBES1.4 was significantly increased in overexpression plants ( Fig. S4C and D). Three independent transgenic lines with the most significant down-regulation in RNAi lines (lines 3, 7, and 8) and different levels of up-regulation in overexpression lines (lines 2, 4, and 8) were selected for further analysis. Suppressing the expression level of GhBES1s significantly reduced leaf size. In contrast, overexpression of GhBES1.4 increased leaf size and plant height compared to the WT ( Fig. 2A and S5). Observation of cotton leaf morphology showed smaller cells in GhBES1-RNAi plants and bigger cells in GhBES1.4 overexpression plants than WT (Figs. 2B and S1B). Then we counted the number of cortical cells per 20,000 μm 2 and the number of all cortical cells in the above two materials (Fig. S1E and H). We found that GhBES1.4-overexpressing plants had fewer cells and GhBES1-RNAi plants had more cells than WT under the same area. However, by counting all cortical cell numbers, no significant difference was found between the three samples, suggesting that GhBES1.4 has an important role in cell expansion. Moreover, qRT-PCR analysis showed that GhEXO2 increased expression in GhBES1.4 overexpression plants and decreased expression in GhBES1-RNAi plants ( Fig. 2C and D). At the same time, we also found that the expression of GhEXO2 was up-regulated in the unpublished transcriptome sequencing of leaves of GhBES1.4 overexpressing plants, which was consistent with the q-RT-PCR results (Fig. S2B).These results suggested that GhBES1.4 positively regulated leaf size and induced the transcription of GhEXO2. Through further analysis, we also found that other family genes (for example, CESA) were up-regulated in the transcriptome data, suggesting that GhBES1.4 regulates cell size not only by affecting GhEXO2 genes, but also possibly other cell expansion-related genes.

GhEXO2 encodes a membrane protein containing BES1 binding sites in its promoter
Phylogenetic analysis (Fig. 1A) showed that GhEXO2 closely clustered with the Arabidopsis EXO family gene AtEXL2, which suggested that GhEXO2 might have a similar function with AtEXL2.To elucidate the subcellular localization of GhEXO2, a construct containing GhEXO2-GFP was infiltrated into tobacco (Nicotiana benthamiana) leaves, which were subjected to confocal imaging microscopy. Confocal imaging microscopy confirmed the presence of GhEXO2-GFP in the cell membrane compared to the red cell membrane marker dye FM 4-64 (Fig. 3B). Analysis of the 2,000 bp promoter region upstream of GhEXO2 start codon revealed some elements related to hormone response, including auxin (AuxRE), methyl jasmonate (MeJA, CGTCA-motif), and BR (E-box and G-box) (Fig. 3C). Furthermore, this E-box has been included in the BES1 binding motif multiple times as predicted by PlantTFDB (Tian et al. 2020). These findings confirmed that GhEXO2 might have a function similar to EXO in Arabidopsis and that it is a potential target gene of GhBES1. These results indicated that GhEXO2 is a cell membrane-associated protein, which mediates BR signaling and may function in changing cotton leaf size through cell expansion.

Ectopic overexpression of GhEXO2 in Arabidopsis promotes leaf size
Previous studies have shown that constitutive overexpression of AtEXO under 35SCaMV promoter control in Arabidopsis enhanced vegetative growth (Coll-Garcia et al. 2004), and that both exo knock-out mutant growth and cell size were reduced (Schröder et al. 2009). To identify the biological functions of GhEXO2, we ectopically overexpressed GhEXO2 in Arabidopsis (Col-0). The resulting Col-0/GhEXO2 transgenic lines had enhanced vegetative growth compared to WT (Col-0) plants, like AtEXO (Fig. 4A). Most obviously, the leaf area of Col-0/GhEXO2 transgenic plants was more than twice as large as in WT plants (Fig. 4B). Moreover, the high relative expression of GhEXO2 in Col-0/GhEXO2 transgenic plants was verified by semi-quantitative RT-PCR and qRT-PCR analysis ( Fig. 4C and D) indicating the validity of the phenotype at the transcription level.
Expression analysis of cell elongation-related genes by qRT-PCR, including AtAGP4, AtEXP5, AtKCS1, and Atδ-TIP, was conducted in Col-0/GhEXO2 transgenic plants and WT plants (Coll-Garcia et al. 2004). The results indicated that expression of all cell elongation-related genes was significantly up-regulated in three independent Col-0/GhEXO2 transgenic lines compared to WT plants ( Fig. 4E-H). Among them, AtAGP4, AtKCS1, and Atδ-TIP were up-regulated by 2-to eightfold in the three transgenic lines, and the expression level of AtEXP5 was up-regulated by 15-to 20-fold. These gene expression results indicated that enhanced vegetative growth in Col-0/GhEXO2 transgenic plants altering leaf size might result from cell expansion.

Silencing of GhEXO2 suppresses leaf cell expansion in cotton
To further investigate the role of GhEXO2 in the regulation of leaf size, we knocked down its expression through virusinduced gene silencing (VIGS) using the pTRV system. As predicted, significant suppression of GhEXO2 expression Each experiment was performed with three biological repeats and the error bars indicate the standard deviation among these replicates. Student's t-test: *P < 0.05, **P < 0.01,***P < 0.001 and less vegetative growth was detected in silenced plants (Fig. 5A) as well as altered leaf size (Fig. 5B) compared to the vector control. Statistical analysis found that the leaf area of all three independent lines of GhEXO2 silenced plants was significantly less than the vector control (Fig. 5E). The qRT-PCR analysis demonstrated that GhEXO2 expression was significantly depressed in observed silenced lines relative to vector control, indicating that GhEXO2 was effectively silenced in VIGS lines (Fig. 5C).
To fully understand the mechanism of altered leaf size in GhEXO2 silenced plants, histological microscopic analysis was performed. Microscopic images showed that the leaves of GhEXO2-silenced plants had a smaller leaf area and a higher number of cells in the same area than leaves of vector control ( Fig. S1C and F), indicating that GhEXO2 is required for cell expansion (Fig. 5D). Similarly, cell expansion was inhibited in the leaves of pag1 mutant plants compared to CRI24 plants. GhEXO2 silenced plants showed smaller leaf sizes due to the inhibition of cell expansion compared with vector control. These findings followed our previous results. The overexpression of GhEXO2 in Arabidopsis resulted in increased leaf size with higher leaf area compared to WT plants (Fig. 4A).

GhBES1.4 directly binds GhEXO2 promoter to activate its expression
Cis-acting element analysis of GhEXO2 promoter and detection of GhEXO2 expression level in GhBES1.4 transgenic plants indicate that GhEXO2 may be a downstream target gene of GhBES1.4. To verify this hypothesis, an electrophoretic mobility shift assay (EMSA) was performed (Fig. 6A). The result indicated that GhBES1.4 fusion protein directly binds to the biotin-labeled probe but fails to bind to the mutant probe. Furthermore, non-labeled probes (cold competition probes) markedly reduced the binding efficiency of the biotin-labeled probe and GhBES1.4 fusion protein. The binding affinity in real-time of GhBES1.4 and the E-box element in the GhEXO2 promoter was measured by biolayer interferometry (BLI) to clarify their interaction (Fig. 6B). The fragment of the GhEXO2 promoter had a stronger binding affinity to GhBES1.4 fusion protein compared with mutant fragments. It weakened with the decrease of the fusion protein concentration. These results indicated that GhBES1.4 specifically binds to the E-box element in the GhEXO2 promoter region in vitro.
However, it is unclear whether GhBES1.4 activates GhEXO2 transcription in vivo. Promoters (2000 bp) of GhEXO2, both natural and mutated in the E-box region, were separately inserted into the pGWB435 vector with the luciferase (LUC) reporter gene. Tobacco leaves coinfected with GhEXO2pro-LUC, and 35S-GhBES1.4 showed brighter fluorescence signals, while leaves co-infected with GhEXO2pro-mutant-LUC and 35S-GhBES1.4 had weaker fluorescence compared with the vector control (Fig. 6C). LUC activity analysis also indicated that the LUC reporter gene was activated by GhBES1.4 driven by the E-box in the GhEXO2 promoter (Fig. 6D). These results revealed that GhBES1.4 directly binds to the E-box element in the promoter region of GhEXO2 to activate its expression, which promotes cotton leaf size.

GhEXO2 affects the expression of various cell expansion genes
To explore the mechanism by which GhEXO2 in cotton altered leaf size, we performed RNA-seq analysis of GhEXO2 silenced plants. A total of 919 genes were  Table S1). Here, GhEXO2 integrated multiple pathways to induce plant growth and development with altered leaf size. Gene ontology (GO) enrichment analysis of RNA-seq data indicated that more genes (97 and 48 genes) enriched in cell wall organization and biogenesis and polysaccharide biosynthetic processes were down-regulated than upregulated (only four and six genes), respectively ( Fig. 7B and Table S1). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis indicated that more DEGs (55 genes) involved in plant hormone signal transduction were down-regulated than up-regulated (14 genes) in GhEXO2 silenced plants (Fig. 7C and Table S1). Further, we generated a heatmap of expression of selected genes related to cell expansion, including fatty acid elongation, cell wall organization and biogenesis, cell periphery, cell wall biosynthetic process, BR biosynthesis, and MAPK pathway. Most of these selected genes were down-regulated in GhEXO2 silenced plants (Fig. 7D).
Some key DEGs were selected to validate our results, and qRT-PCR analysis was conducted in two lines with significantly down-regulated GhEXO2 expression levels. Except for GhCPD and GhDWF4, which are related to BR biosynthesis, the other seven genes are directly or indirectly related to cell expansion (Wanjie et al. 2005;Todd et al. 1999;Goh et al. 2012;Jensen et al. 2011;Lee et al. 2006;Liu et al. 2021a;Chung and Choe 2013). The results of the qRT-PCR analysis indicated that GhKCS1, GhEXPA4, GhEXPA8, GhIRX15-L, and GhLNG1 were down-regulated while the two BR biosynthesis genes GhCPD and GhDWF4 were up-regulated in GhEXO2 silenced plants (Fig. S7). These qRT-PCR results followed the results of RNA-seq data analysis, indicating the validity of our findings.

Discussion
The leaf is an essential organ of plant photosynthesis and transpiration. Variability in leaf size is regulated by the external environment and localized delivery of the phytohormone (Nikolov et al. 2019). BRs are essential steroid hormones that function in different developmental processes in plants, chiefly in plant growth. Here, we demonstrated that BR regulates cotton leaf size by cell expansion. Further, GhBES1 mediates BR regulation of leaf size through directly activating the expression of GhEXO2, which was confirmed as a cell expansion regulator. In this study, we revealed the novel mechanism by which BR regulates leaf size.

BR regulates leaf size in plants
In previous studies, the BR signaling pathway was shown to be a relatively clear signal transduction pathway in plants, based on combining multiple research approaches (Guo et al. 2013;Kim et al. 2009;Kour et al. 2021;Li 2010). Analysis of BR deficient mutants and BR-related transgenic plants showed that BRs positively regulate leaf size in a variety of plants. In Arabidopsis, dwf4 and cpd plants are reduced in stature, and have dark green colored leaves due exclusively to inhibition of cell elongation increasing chloroplast density (Szekeres et al. 1996;Azpiroz et al. 1998). In BR-insensitive mutant, bri1, applying exogenous BRs does not rescue the phenotype in an extreme dwarf with short rosette leaves (Clouse et al. 1996). In rice, brd3-D enhances CYP734A4 expression and shows dwarf and erect leaf phenotypes. In addition, knockout of the CYP734A4 gene by CRISPR/Cas9 rescues the brd3-D phenotype (Qian et al. 2017). In barley, the BR receptor gene HvBRI1 mutant uzu1 has an erect structure, an acute angle of the leaf blade, and a wavy leaf margin (Dockter et al. 2014). Overexpression of GmMYB14 in soybean reduced the endogenous BR content, resulting in decreased plant height and smaller leaf area, and exogenous application of BL partially restored its phenotype (Chen et al. 2021). GhPAG1, a BR catabolic gene, produces BR deficient mutant (pag1) in cotton, which showed less fiber elongation, a dwarf phenotype, and markedly decreased leaf size due to inhibited cell elongation ( Fig. 1 A and B) (Yang et al. 2014). The transcription factor BES1 regulates the expression of thousands of target genes in response to BRs (Jiang et al. 2015). In our study, Silencing of GhBES1 gene reduces leaf size, whereas GhBES1.4 over-expression increased the leaf size ( Fig. 2A). Moreover, GhBES1 can bind to the GhEXO2 promoter region and activate its expression (Fig. 6). In addition, silencing the GhEXO2 gene in cotton reduced leaf size (Fig. 5B). Ectopic overexpression of GhEXO2 in Arabidopsis resulted in larger leaf size compared to control (Fig. 4A).Thus, BR regulation of leaf size is conserved in plants.

BES1 is a critical factor for BR to promote leaf size
BES1 is a basic helix-loop-helix (bHLH) protein, a superfamily of TFs essential in plant growth and development. Activation of BES1 causes curled leaves and long and curved petioles in Arabidopsis bes1-D mutant (Yin et al, 2002). Conversely, a hextuple mutant (bzr-h) that disrupts all BZR1s in Arabidopsis showed a phenotype that was more severely curled and crinkled, and had small and rounded leaves (Chen et al. 2019). In rice, BES1 promotes the bHLH TF INCREASED LEAF INCLINATION 1 (ILI1) transcription by binding to its promoter region and significantly increasing leaf tilt ). In cotton, the bHLH TF 1 3 FIBRE-RELATED PROTEIN 1 (GhFP1) promotes fiber elongation by directly binding to promoter sequences of BR synthesis genes to activate its expression (Liu et al. 2020).
Our previous work found that GhBES1.4 is a functional TF inducing BR response . In our study, inhibition of GhBES1s expression suppressed cell expansion and decreased leaf size. Overexpression of GhBES1.4 showed larger leaf size via accelerating cell expansion in cotton ( Fig. 2A and B). Therefore, BES1 functions directly in BR regulation of leaf size.
An important mechanism in BR signaling depends on the direct regulation of target genes by BZR1/BES1 via binding E-box (CANNTG) in the promoter Xie et al. 2011;. Absence of AtMYB30 function exhibited decreased BR responses, because BES1 binds to E-box to directly regulate the transcription of AtMYB30 and activates downstream target genes to amplify BR signals via recruiting AtMYB30 (Li et al. 2009). BES1 directly binds to MYELOBLASTOSIS FAMILY TRANSCRIPTION FACTOR-LIKE 2 (MYBL2), repressing its transcription to regulate cell elongation (Ye et al. 2012). In our study, a significantly down-regulated gene, GhEXO2, was found in pag1 with the E-box in the promoter region, and exogenous application of BL induced GhEXO2 expression (Figs. 1C, D, and 3B). GhCaM7-like was also downregulated in the pag1 mutant as in a previous study (Yang et al. 2014). Silencing of GhCaM7-like expression suppressed fiber length and leaf size, and multiple E-boxes were predicted in the promoter region. These results indicated that GhCaM7-like is a potential target gene for BES1 (Cheng et al., 2016). In addition, silencing of GhEXO2 resulted in inhibition of stem cell expansion similar to pag1 and GhCaM7-like-silenced plants ( Fig. S7A and B). Further, protein and DNA binding experiments in vivo and in vitro showed that GhBES1.4 binds to the E-box motif in the promoter region of GhEXO2 and activates its expression. Thus, GhBES1 regulated leaf size and affected GhEXO2 expression by activating its transcription ( Fig. 2C and D). Overall, BES1 mediated BR signaling affects leaf size by regulating downstream genes.
GhEXO2 promotes cell expansion and affects cotton leaf size, thus mediating BR signal EXO, a protein localized to the cell membrane, is closely associated with the cell wall and EXL1 and EXL2 in Arabidopsis and is expressed in dividing cells (Farrar et al. 2003;Schröder et al. 2009;Coll-Garcia et al. 2004). Arabidopsis EXO/EXL mediates plant growth and development through cell expansion as a downstream BR signaling gene, and overexpression in BR-deficient dwf1-6 and det2 mutants does not normalize the dwarf phenotype (Schröder et al. 2009;Coll-Garcia et al. 2004). In addition, exogenous application of auxin and BL strongly induced their expression, while cytokinin inhibited it (Farrar et al. 2003;Coll-Garcia et al. 2004). Further, bes1-D, a gain of function mutant of BES1, showed constitutive expression of EXO genes (Yin et al. 2002). 35S::EXO had larger rosette leaves and increased transcription levels of BR-responsive genes (AtAGP4, AtEXP5, AtKCS1, and Atδ-TIP) associated with cell walls (Coll-Garcia et al. 2004). exo knockout mutant showed a dwarfed phenotype with inhibited cell expansion, and reduced transcript levels of AtKCS1 and AtEXP5 recovered with increasing age (Schröder et al. 2009).
Our results demonstrate that GhEXO2 promotes cotton leaf size by positively promoting cell expansion in the BR pathway. First, GhEXO2, like members of the AtEXO family, is a cell membrane-associated protein (Fig. 3B). The conserved Phi_1 domain showed maximum similarity with Arabidopsis AtEXL2 (Figs. 3A, S3A and B). These results imply that GhEXO2 functions in the cell wall. Next, overexpression of GhEXO2 in Arabidopsis showed enhanced leaf size (Fig. 4A). Higher expression patterns of cell elongation-related genes AtAGP4, AtEXP5, AtKCS1, and Atδ-TIP in transgenic lines indicated that GhEXO2 induces enhanced leaf size through cell expansion (Fig. 4E-H) like AtEXO. Furthermore, silencing GhEXO2 by VIGS showed reduced leaf size and limited cell expansion ( Fig. 5B and D). Therefore, GhEXO2 acts as a cell expansion regulator to promote leaf size.
In addition, RNA-seq analysis in cotton indicated that silencing GhEXO2 affects multiple pathway genes to suppress leaf size ( Fig. 7B-D). First, in silenced GhEXO2 plants, different genes were markedly affected in fatty acid elongation and cell wall organization or biogenesis pathways compared to Arabidopsis (Coll-Garcia et al. 2004;Shang et al. 2016). In the fatty acid elongation pathway, the KCS1 gene is affected in both Arabidopsis and cotton, but other fatty acid elongation genes such as KCS11 are also affected in cotton. In addition, EXP5, which is affected in Arabidopsis, is not a DEG, while EXPA4 and EXPA8 genes from the same family are significantly down-regulated in cell wall organization or biogenesis pathway. We also found that CPD and DWF4, which are reported to differ little in Arabidopsis, were enriched in the BR biosynthetic pathway in the up-regulated cotton DEGs. Moreover, we identified multiple genes that have not been reported in Arabidopsis that were enriched in the novel pathways. In the cell wall macromolecule biosynthetic process, the expression of the xylan biosynthesis gene GhIRX15-L and the longitudinal leaf cell elongation gene GhLNG1 was down-regulated, which has not been found in Arabidopsis. This result may Fig. 4 Ectopic overexpression of GhEXO2 in Arabidopsis. A Phenotype of ectopically overexpressed GhEXO2. B Leaf area in WT and GhEXO2 transgenic lines. C RT-PCR analysis for transgene validation in GhEXO2 transgenic lines and wild type (WT). D qRT-PCR analysis of relative expression pattern analysis for transgene validation in GhEXO2 transgenic lines and WT. E-H qRT-PCR analysis of AtAGP4, AtEXP5, AtKCS1, and Atα-TIP genes in three independent Col-0/GhEXO2 transgenic lines and WT plants. Each experiment was performed in three biological repeats and the error bars indicate the standard deviation among these replicates. Student's t-test: *P < 0.05, **P < 0.01,***P < 0.001 GhEXO2 silenced and control plants. E Graphical representation of leaf area in GhEXO2 silenced and control plants. Each experiment was performed with three biological repeats and the error bars indicate the standard deviation among these replicates. Student's t-test: *P < 0.05, **P < 0.01 be due to interspecific differences between tetraploid cotton and diploid Arabidopsis. KCS1, a BZR1 target gene involved in cell wall biosynthesis, has been reported to be an early BR-responsive gene (Sun et al. 2010). Expansion protein is the first discovered and most widely reported cell wall loosening protein. The alpha-expansin group members GhEXPA4 and GhEXPA8 enriched in cell wall organization contributed to early lint fiber initiation. Moreover, GhEXPA8 improves mature cotton fiber length and micronaire value (Bajwa et al. 2015;Liu et al. 2021b). CPD and DWF4 are key genes of BRs biosynthetic enzymes, which can be activated by feedback Fig. 6 Analysis of GhBES1 and GhEXO2 interactions in vivo and in vitro. A EMSA showed that GhBES1 protein binds to the E-box element of the GhEXO2 promoter region. B Real-time binding analysis of GhBES1 protein to the E-box element of the GhEXO2 promoter region. C Luc activity analysis of the interaction of the E-box element of the GhEXO2 promoter and GhBES1. D Quantification of relevant Luc activities. The error bars represent the SD of three biological repeats. The asterisks indicated significant differences as determined by t-test (**P < 0.01) from reduced BR levels (Szekeres et al. 1996;Choe et al. 1998). IRX15-L encodes xylan biosynthesis, a major component of plant cell wall hemicellulose. irx15 irx15-L double mutants have irregular secondary cell wall edges and low xylan aggregation in fibroblasts. LNG1 promotes polar elongation of cells and its dominant mutant longifolia1-1D (lng1-1D) presents a long petiole, narrow and extremely long leaf blade, and a serrated margin phenotype (Jensen et al. 2011;Lee et al. 2006). Taken together, these results imply that GhEXO2 affects multiple pathways that influence leaf size by mediating cell expansion.
Collectively, we proposed a working model of BR to promote leaf size through cell expansion. In the BR signaling pathway, GhBES1 affects cotton leaf size by binding to and activating the expression of the E-box element in the GhEXO2 promoter region. Overexpression of GhEXO2 promotes leaf size, and suppression of its expression inhibits leaf size by altering the expression of cell elongation-related genes (Fig. 8).