GhD14 regulates plant architecture and fiber development in cotton


 Strigolactone (SL) signaling is essential in regulating plant development. DWARF14 (D14), the SL receptor, interacts with the F-box in MORE AXILLARY GROWTH (MAX2) to modulate SL signaling. However, the biological function of D14 protein is still unknown in cotton. Here, we identified GhD14s in Gossypium hirsutum and resolved its function in cotton plant architecture and fiber development. The GhD14D protein was localized to both the cytoplasm and nucleus. GUS staining assay showed that GhD14D was mainly expressed in leaf primordium, inflorescence, axillary bud and stem and expression analysis revealed that GhD14A/D was highly expressed in stem, flower and fiber cells at 20 days post-anthesis (DPA). Silencing GhD14A/D gene expression in upland cotton signiﬁcantly increased branch angle and reduced ﬁber length as well as the transcripts of secondary cell wall biosynthesis related genes. In addition, overexpression of GhD14D in Atd14 mutant successfully rescued the phenotype of the d14-1 mutant with much shoot-branching and short plant height. Our findings suggest that the GhD14 gene contributes to shoot branch development and fiber cell development in cotton. This study deepens our understanding of the biological role of SL signaling in cotton and providing guidance for modifying cotton plant architecture and improving fiber development using genetic engineering to help us breed better cotton varieties in the future.


Introduction
Plant architecture refers to the organization of the plant body, includes shoot branching, branch length, branch angle, position of organs and so on in plants (Wang et al. 2008). Higher plants display various plant architectures for their different requirements. For crops, the crop yield and quality could be affected by plant architecture. After the Green Revolution, crop yields dramatically increased due to higher planting density, which was made possible by changes in plant architecture leading to shorter and more compact cultivars (Peng et al. 1999). The plant architecture is in uenced by many environmental factors such as moisture and light. The drought stress reduces plant height and grain yield (Todaka et al. 2015). The dark and cold condition cause the plant to be dwarfed. Phytohormones play important roles in plant development process (Blázquez et (Kameoka et al. 2018). The biosynthesis pathway of SLs is well documented (Alder et al. 2012). The SLs biosynthesis is derived from the carotenoid pathway: rst, Dwarf 27 (D27) carotenoid isomerase catalyzes the conversion of all-trans-β-carotene into 9-cis-β-carotene (Alder et al. 2012). Subsequently, Carotenoid Cleavage Dioxygenase 8 (CCD8), encoded by More Axillary Growth 4 converts 9-cis-β-apo-10'carotenenal into carlactone (Alder et al. 2012). Carlactone can be further converted into 5-deoxylstrigol and other bioactive SLs via a cytochrome P450 monooxygenase, which is encoded by More Axillary Growth 1 (Seto et al. 2014). Meanwhile, the biosynthesis of SLs can be regulated by zaxinone, an endogenous carotenoid-drived molecule (Wang et al. 2019;Ablazov et al. 2020). The perception for the strigolactone DWARF14 (D14), is an α/β-hydrolase (Seto et al. 2019).
The strigolactones control many aspects of plant development and abiotic stress response. The SL biosynthesis gene (HIGH TILLERING AND DWARF 1/DWARF17) increases tiller number of rice and improves rice grain yield ). In Arabidopsis, three key repressors of the strigolactone (SL) signaling pathway (SMXL6/SMXL7/SMXL8) directly interact with two transcriptional factors and suppress their transcriptional activation of a key repressor of branching (BRC1), thus promoting branching (Xie et al. 2020). The SL could trigger polyubiquitination and degradation of SMXL2, then regulating the gene expression and hypocotyl elongation ). The SL also control the degradation of cytokinin via activating the transcription of CYTOKININ OXIDASE/DEHYDROGENASE 9 in rice (Duan et al. 2019).
The SL receptor, Dwarf14 (D14), is a key receptor that interacts with an F-box protein to form a Skp1-Cullin-F-BOX (SCF) complex and regulates SL signaling pathway (Nakamura et al. 2013). In soybean, the GmMAX2 interacted with the GmD14 and then forming active D14/KAI-SCF MAX2 complexes for function (Ahmad et al. 2020). The SCF complex then recruits SL repressor proteins to be ubiquitinated and then the SL be degraded, hence the SL repressor proteins could regulate the expression level of SL-dependent genes (Marzec. 2016). A Yoshimulactone Green (YLG) based in vitro assay of a high-throughput chemical screening identi ed a novel small molecule DL1 as a potent inhibitor of D14 that competes with endogenous SLs and increases shoot branching in Arabidopsis and rice (Yoshimura et al. 2018). In Dendranthema grandi orum, DgMAX2, a key regulatory gene in SL signal transduction, can restore atmax2-1 mutant branching to wild-type (WT) in Arabidopsis (Dong et al. 2013). In Populus, there are two PtD14 genes, but only PtD14a was able to recover the Atd14 mutant (Zheng et al. 2016). However the D14 in G. hirsutum has not been investigated.
In this study, a putative SL receptor, GhD14, was identi ed in G. hirsutum and it was localized to both the cytoplasm and nucleus. The expression analysis showed that GhD14 was primarily expressed in buds, stems, owers and 20 DPA bers. GhD14-silenced plants had a phenotype of increased branch angle, reduced ber length and the transcripts level of secondary cell wall biosynthesis related genes.
Meanwhile, the GhD14 rescued the phenotype of the d14-1 mutant with much shoot-branching and short plant height. These result shows that GhD14 is involved in regulating cotton architecture and ber development, which lays a foundation for manually controlling cotton plant architecture in the future.

Plant materials and growth conditions
G. hirsutum cultivar Xuzhou 142 was planted in a climate-controlled greenhouse with the same condition as previously reported (Lin et al. 2019). The bers at 0, 3, 5, 10, 15, and 20 days after anthesis (DPA), stems, leaves, owers from four months plants and roots form one month cotton plants were collected and frozen in liquid nitrogen immediately, then stored at -80°C until use. Seeds of wild-type A. thaliana (Columbia, Col-0) and d14 mutant (Waters et al. 2012) were placed at 4°C vernalization for 48-72 hours, then surface-sterilized and germinated on 1/2 strength Murashigeand Skoog (1/2 MS) agar plates. Then, A. thaliana seedlings were transferred from the medium to soil pots grown in chambers at 24°C 14 h /10 h, light/dark cycle. The seeds of Nicotiana Benthamiana are uniformly scattered in the soil and culture conditions are the same as the G. hirsutum.

Bioinformatics analysis of D14 proteins
The open reading frame (ORF) of GhD14 was identi ed by ORF Finder (https://www.ncbi.nlm.nih.gov/or nder/). The physicochemical properties were calculated by ExPASy online software (Gasteiger et al. 2003). The GhD14 protein secondary structure was predicted by SOPMA software with secondary structure prediction method (Geourjon et al. 1995). Three dimensional (3D) structure of GhD14D protein was constructed by SWISS-MODEL (Waterhouse et al. 2018). The DNAMAN software was used to further annotate the protein structure of GhD14s. The protein sequences of AtD14, OsD14 and GhD14s were submitted to DNAMAN software to perform multiple protein sequences alignment.
The D14 genes from two monocots (Zea mays and Oryza sativa) and 13 dicots (Theobroma cacao, Corchorus olitorius, Cicer arietinum, Solanum pennellii, Capsicum annuum, Juglans regia, Arachis duranensis, Glycine max, Abrus precatorius, Phtheirospermum japonicum, Striga hermonthica, Orobanche cernua and Arabidopsis thaliana) were selected for phylogenetic analysis. The full-length protein sequences of these homologous were obtained from the National Center for Biotechnology Information (NCBI). The primers were designed by OLIGO 7.0 (Rychlik. 2007). For expression analysis, the ubiquitin gene GhUBQ7 was used as a housekeeping gene (Xiao et al. 2016). All primers in this work are listed in Supplementary Table S1. The qRT-PCR experiments were performed using SYBR® Premix Ex Taq™ II (Takara, Japan) kit on a Bio-Rad Real Time PCR detection system (Bio-Rad CFX96Touch, USA) with the following reaction parameters: 94°C for 2 min, followed by 40 cycles of 94°C for 30 s and 60°C for 30 s. The 2 −∆∆CT method was used to calculate the relative expression levels of the target genes. All data were statistically analyzed using by One-way ANOVA method by SigmaStat software with default parameters (Ji et al. 2003).

Cloning, plasmid construction and transformation
The protein sequence of AtD14 was downloaded from TAIR database, and used as a query sequence to against the G. hirsutum protein genome database (ZJU_v2.1). The most similarity genes to AtD14 in G.hirsutum were recognized as GhD14s. The gene was ampli ed from G. hirsutum cDNA with PrimeSTAR® Max DNA Polymerase (TaKaRa, Japan) and sequenced by TSINGKE company (Beijing, China).
The 1715 bp upstream genomic DNA sequence of the initiation codon (ATG) from GhD14D was cloned as the promoter sequence from Xuzhou 142 DNA and full-length CDS of GhD14 were ampli ed from Xuzhou 142 cDNA. For overexpression of GhD14D experiment, the full-length of GhD14D CDS sequence was assembled into pCAMBIA1305 by In-Fusion® HD Cloning Kit (Vazyme Biotech, China) according to the instructions. For β-glucuronidase (GUS) staining analysis, the GhD14D promoter sequences was introduced into pCAMBIA2300 to drive the GUS gene expression. All the nal constructs were transformed into A. thaliana and N. benthamiana by Agrobacterium tumefaciens mediated transformation method as previously described (Cheng et al. 2018). In addition, the full-length CDS of GhD14D was cloned into pTF486-GFP and pCHF3-GFP to generate the 35S::GhD14-GFP construct for subcellular localization assays. Two primers in pCAMBIA1305 were used for identi cation of whether A. thaliana contain pCAMBIA1305: GhD14D. All the primers used in vector construction are listed in Supplementary Table S1. Meanwhile, one month old A. thaliana seedlings was used for observation of the rosette leaves of seedlings and leaves phenotype and Two and a half month A. thaliana was used for the whole plants phenotype observation.

GUS staining analysis
For GUS staining analysis, the GhD14Dpro::GUS construct was introduced into A. tumefaciens strain GV3101 and subsequently transformed into A. thaliana using the oral dip method (Lloyd et al. 1986). Transgenic plants were selected on solid half-strength MS media plates containing 50 µg/mL chloramphenicol. The selected transgenic seedlings were further validated by genomic PCR. Various tissue at different developmental stages were collected, and stained with GUS staining solution according to the instructions. The microscop (Nikon, Japan) was used to observe the GUS staining result.

Transient expression of GhD14D-GFP
For subcellular localization of GhD14D, the fused GhD14D-GFP vectors were transiently expressed in A. thalinan protoplast and N. benthamiana according to previous studies (Abel et al. 1994;Cheng et al. 2018). The two fusion constructs pTF486-GFP and pCHF3-GFP were used for transient expression GhD14D-GFP in A. thaliana protoplast and in N. benthamiana epidermis cells, respectively. In addition, the nuclei staining, the chemical reagent 4', 6'-diamidino-2-phenylindole (DAPI) staining method was used to con rm the nuclei localization of GhD14D (Liang et al. 2018). The positive transgenic plants were used for in vivo subcellular localization analysis using a confocal microscope (Nikon, Japan).

Virus
The binary CLCr vectors pCLCrVA and pCLCrVB were used for virus-induced gene silence (VIGS) analysis (Gu et al. 2014). Three fragments of GhD14A/D gene were ampli ed using cDNA with the primers listed in Supplementary Table S1. The PCR fragments were inserted into pCLCrVA and these vectors were transformed into A. tumefaciens GV3101 by electroporation. The A. tumefaciens culture containing pCLCrVA or the constructed vectors were mixed with an equal volume of an A. tumefaciens culture containing pCLCrVB to generate the mixed A. tumefaciens solutions for in ltration. The mixed solutions were in ltrated into fully expanded cotyledons of 10-day-old cotton seedlings with 15 plants through vacuum in ltration (Gao et al. 2011). The inoculated seedlings were then transferred to a dark climatecontrolled greenhouse at 25°C. The one month old seedling leaves and 25 DPA bers from negative control and GhD14A/D-VIGS cotton plants were chosen to assess the silence effects of GhD14A/D gene by qRT-PCR experiment. The sixth branches from bottom to top were used for detecting the angle of the monopodial branch 3 Results

Identi cation of the GhD14 in G. hirsutum
The AtD14 protein sequence was used as the query sequence to against the G. hirsutum genome (Hu et al. 2019) to obtain the D14 protein in G. hirsutum. The sequences GH_A02G1790.1 and GH_D03G0270.1 in the cotton genome were the most similar to AtD14 and were chosen as the candidate orthologs of AtD14 and named GhD14A and GhD14D, respectively. The GhD14A and GhD14D have one intron, contain 816 base pairs (bp) ( Supplementary Fig. 1A), 271 amino acids (Fig. 1A) and have a molecular weight of 19.9 kDa and 30.0 kDa, respectively.

Conserved domain and promoter analysis of GhD14s
The AtD14 protein structure in A. thaliana has been reported as an α/β hydrolase (Li et al. 2020). We compared the protein sequences of GhD14A, GhD14D and AtD14, and identi ed eight α-helixes, ve ηhelixes and seven β-strands in D14 protein (Fig. 1B, Supplementary Fig. 1B, C). There are three catalytic residues, S95, D217 and H246 in GhD14A and GhD14D protein sequences (Fig. 1B). The sequence similarities of AtD14 with GhD14A and GhD14D were 52.7% and 53.5%, respectively. As GhD14A and GhD14D have 97.4% similarity in coding sequences, we cloned GhD14D from G. hirsutum cultivar Xuzhou 142 and used it as the representative GhD14 gene for cis-elements distribution, subcellular location, GUS and overexpression analysis.
The cis-element distribution of 1,715 bp GhD14D promoter sequence and promoter driven GUS expression were investigated to study the potential function of GhD14D. The GhD14D promoter has high A and T content, and typical cis-elements TATA-box and CAAT-box (Supplementary Figure S2), which are consistent with the characteristics of other plant promoters (Zhang et al. 2016). The auxin responsiveness cis-element AUXRR-core and jasmonic acid methyl ester (MeJA) responsiveness regulatory element CGTCA-motif were also present in the GhD14D promoter region (Supplementary Figure  S2), indicating that auxin and MeJA might function in regulating GhD14D gene expression via combine the corresponding cis-elements present in the promoter of GhD14D.

Phylogenetic analysis of GhD14s
To explore the evolutionary relationships among GhD14D and other fteen D14s from typical higher plant species, we performed phylogenetic analysis of D14 proteins from G. hirsutum and other fteen higher plant species using MEGA 7.0 software with Neighboring-Joining (NJ) method and 1000 bootstrap replications. The phylogenetic tree showed that the D14s in sixteen higher plants have the same evolutionary origin. The D14s was resolved into two evolutionary branches in phylogenetic tree and both of which contain D14 proteins from monocotyledons and dicotyledons. Meanwhile, GhD14s was more closely related to eudicots, and was most closely related to T. cacao, with moderate bootstrap support of 79% (Supplementary Figure S3).

GhD14D is localized to the cytoplasm and nucleus
The pTF486-GhD14D:GFP vector was constructed to investigate the subcellular location of GhD14D protein and DAPI staining was used to test whether the GhD14D protein was localized in the nucleus. The pTF486-GhD14D:GFP was transiently expressed in A. thaliana protoplasts and the cells were stained with DAPI, and then examined using confocal microscopy. The results showed that GhD14D-GFP fusion protein was accumulated throughout the nucleus and cytoplasm and was also co-localized with DAPI ( Fig. 2A). The pCHF3-GhD14D:GFP was bombarded into tobacco epidermal cells by agroin ltration to further con rm the subcellular localization of GhD14D protein. We found the green uorescence of GhD14D-GFP distributed in the cytoplasm and nucleus and the uorescence of GhD14D-GFP coincided with the DAPI uorescence, con rming that GhD14D protein is co-localized in the cytoplasm and nucleus (Fig. 2B). Both GhD14D-GFP fusion protein and DAPI methods in A. thaliana protoplast and tobacco epidermal cells clearly showed strong nucleus and cytoplasm localization, indicating that the hydrolase GhD14D functions in the cytoplasm and nucleus.

Expression patterns of GhD14A/D
Expression patterns of GhD14A/D in bers at 0, 3, 5, 10, 15 and 20 days post-anthesis (DPA) and root, stem, leaf, and ower were investigated to explore the function of GhD14 using quantitative real-time PCR (qRT-PCR). Since GhD14A and GhD14D have 97.43% coding sequence similarity and they can't separate with speci c primer, the expression level of GhD14A/D was investigated. As shown in Fig. 2, GhD14A/D gene transcripts were increased during the ber development with peak value at 20 DPA (Fig. 3A), and abundantly expressed in ower and stem (Fig. 3B), indicating that GhD14A/D may function in ber, ower and stem development.
In order to test the tissue speci c expression of GhD14D, the GhD14D promoter driving GUS gene expression in A. thaliana was visualized by histochemical staining of transgenic A. thaliana. The color of transgenic A. thaliana seedlings represents the promoter driven GUS gene expression. As shown in Fig. 4, the GhD14D promoter-driven GUS gene was highly expressed in buds, stems and owers (Fig. 4). These observations demonstrated that GhD14D may play important role in buds, stems and owers.

GhD14A/D gene silencing in cotton increased branch angles and reduced ber length
Virus-induced gene silencing (VIGS) is an e cient and rapid method to reduce gene transcripts and investigate gene functions in plants (Gu et al. 2014). To further uncover potential functions of GhD14 in cotton, the VIGS strategy was used to reduce GhD14A/D transcript levels in G. hirsutum. Positive control of GhPDS gene silence shows in Supplementary Figure S5. Expression levels of the GhD14A/D gene were detected using qRT-PCR strategy and the result showed that GhD14A/D gene expression was signi cantly reduced in GhD14A/D-silenced plants compared with that in the control plants (CLCrVA) (Fig. 5A). The phenotype of branch angles of control and VIGS plants (GhD14A/D-V1, GhD14A/D-V2 and GhD14A/D-V3) showed that the monopodial branch angles were signi cantly increased in GhD14A/D-silenced plants compared with the negative control plants (CK). The branch angles of GhD14A/D-silenced plants were increased about two times compared with the CK (Fig. 5B, C, Supplementary Figure S4). In order to explore the potential functions of GhD14A/D in cotton ber development. The ber length was observed after GhD14A/D silencing, and the result showed that the cotton ber length was signi cantly reduced in GhD14A/D-silenced plants (Fig. 5D, E), indicating the GhD14A/D gene may play a critical role in cotton ber elongation. In general, reducing GhD14A/D gene expression led to wider branch angles and decreased ber length, suggesting that GhD14A/D functions in cotton architecture and ber development.

GhD14A/D gene silencing in cotton reduce the transcripts of secondary cell wall biosynthesis genes
The cotton ber is produced by the speci c elongation of cell in ovule epidermal and cell wall biosynthesis is required during the ber elongation process. To further investigate the mechanism of GhD14A/D in regulating ber development, the relative expression level of secondary cell wall biosynthesis genes (Sun et al. 2017) was investigated in GhD14A/D-silenced plants. As shown in Fig. 6, The six genes (GhLBD30, GhCesA7, GhMYB46, GhXCP1, GhIRX8 and GhXCP2) related secondary cell wall biosynthesis were down regulated in GhD14A/D-silenced plant, while other three genes (GhIRX10, GhCesA8 and GhCesA4) related secondary cell wall biosynthesis have no signi cant difference in expression levels compare with the negative control (CLCrVA) (Fig. 6). This result demonstrates that downregulating of GhD14 gene expression reduced the transcripts of genes involved in secondary cell wall biosynthesis.

Overexpression of Ghd14d rescued the d14 mutant phenotype in Arabidopsis
Overexpression of GhD14D in the d14 Arabidopsis mutant, a heterologous complementation approach, was performed to further explore the function of GhD14D in stem development and shoot branching, since AtD14 is the receptor of SLs and has essential functions in the SL signal transduction process (Chevalier et al. 2014). The pCAMBIA1305-35S-GhD14D was constructed and transformed into A. thaliana mutant d14-1. The seven transgenic lines were obtained after PCR detection (Supplementary Figure S6). The third generation of six transgenic lines with stable and highly expressed GhD14A/D was used for further analysis. As shown in Fig. 6, mutant d14-1 plants had smaller length-width ratio of leaf (Fig. 7A, B) and more shoot branches as well as shorter plant height (Fig. 6C, D, E) compared with wildtype A. thaliana plants. Overexpression of GhD14D in A. thaliana d14-1 mutant restored the leaf phenotypes and the number and length of branches to the phenotype of wild-type plants (Fig. 7). The phenotype of the d14-1 mutant with more shoot branches and short stature could be rescued by GhD14D and there was no signi cant phenotype difference between wild-type A. thaliana lines and d14-1/35S:GhD14D transgenic lines (Fig. 7). These results suggested that GhD14D and AtD14 may have similar functions in regulating branching number, plant height, petiole length, and length-width ratio of leaf.

Discussion
In China, modern agricultural cultivation of cotton requires moderately short and compact varieties to adapt to mechanical operations (Su et al. 2018). The short and compact characteristics are mainly controlled by plant architecture, which is regulated by plant hormones, such as auxins and strigolactones Branching is a highly plastic determinant of plant shape to allow plants to respond to environmental stresses (Evers et al. 2011). Shoot branching is a ubiquitous phenomenon in higher plants and a basic characteristic of plant growth, that is essential in determining plant architecture.
Cotton is an important source of protein and oil and cotton ber is commonly used as natural ber in the textile industry, hence, regulating growth and controlling branching are essential for cotton cultivation.
Several key genes involved in the SL biosynthesis signaling pathway and participating in plant architecture development, especially shoot-branching have been identi ed (Beveridge et al. 2010). In this study, a homolog of AtD14 was cloned from upland cotton cultivar Xuzhou 142 and named GhD14s. GhD14s showed high identity with AtD14 (53.5%). The D14 functional domains were conserved in cotton and the model plants rice (monocotyledon) and Arabidopsis (dicotyledon) (Fig. 1). Phylogenetic analysis showed that the D14s in higher plants have a close relationship with D14s from other higher plants (Supplementary Figure S3). The conservation of D14 among different species indicates it play critical role in higher plants.
Subcellular localization results showed that GhD14D is localized to the nucleus and cytoplasm, which is consistent with the localization results of the D14 genes in Arabidopsis and rice (Chevalier et al. 2014; Yao et al. 2018). The subcellular localization results provided evidence that GhD14D functions in both the nucleus and cytoplasm. GUS staining showed that D14 in rice was mainly expressed in parenchyma cells surrounding the xylem in leaves, stems and axillary buds (Arite et al. 2009). In this study, the GhD14D pro::GUS gene expression was mainly observed in the buds, stems and owers (Fig. 4). The transcripts of GhD14D in cotton were also signi cantly accumulated in stem, leaf, ower and 20 DPA bers (Fig. 3). These results demonstrate that GhD14D functions in stem, leaf, ower and ber development. Previous studies in other plants found that D14 was involved in the development of stem, leaf and ower: The d14 mutant had a larger stomatal aperture in leaf, slower abscisic acid (ABA)-mediated stomatal closure, lower anthocyanin and reduced plant senescence under drought stress (Li et al. 2020). OsMADS57 interacts with TEOSINTE BRANCHED1 (OsTB1) and targets OsD14 to control the outgrowth of axillary buds in rice (Guo et al. 2013).
The OsD14 is the receptor of SL protein, which acts as a new component of the SL-dependent branching inhibition pathway and inhibits rice tillering (Arite et al. 2009). AtD14 hydrolyzes SLs into a D-ring-derived intermediate CLIM and irreversibly binds CLIM to trigger SL signal transduction, thus regulating shoot branching in Arabidopsis (Yao et al. 2016). Furthermore, overexpression of GhD14D in the d14 mutant, was performed to further explore the function of GhD14D in stem development and shoot branching. As a result, overexpression of GhD14D in AtD14 mutant (d14-1) reduced the mean number of branches and restored the plant height (Fig. 7). Meanwhile, the branching and leaf phenotypes were restored to wildtype plants phenotype and there was no signi cant difference between wild-type plants and d14-1/35S::GhD14D transgenic lines. The phenotype of the d14-1 mutant with fewer shoot branches and higher stature could be rescued via GhD14D overexpression in d14 mutant, suggesting that GhD14D and AtD14 may have similar functions in regulating branching number, plant height, petiole length, and lengthwidth ratio of leaf. Consistently, the OsD14 gene also restored the phenotype of d14 to the wild type both in plant height and tiller development in rice (Yao et al. 2018).
The GhD14A/D had high expression levels in ower, stem, leaf, and 20 DPA bers, but was weakly expressed in cotton root (Fig. 3). In rice, transcripts of D14 were highly accumulated in leaves and the rst leaf buds, but not in root tip (Arite et al. 2009). In petunia, high expression levels of DAD2 were observed in axillary bud and leaf, but not in root (Hamiaux et al. 2012). In chrysanthemum, DgD14 had the highest expression level in stem, followed by node, and was only weakly expressed in root (Wen et al. 2015). These results re ect that the D14 gene may not be involved in root structural development. The GhD14A/D showed high expression level at 20 DPA bers (Fig. 3), and this stage is the secondary cell wall thickening stage (Zhang et al. 2015). Meanwhile, silencing the GhD14A/D gene expression reduced the ber length (Fig. 5) and the expression levels of the secondary cell wall biosynthesis related genes (Fig. 6). These results indicating that GhD14 might play a role in ber development through impact secondary cell wall biosynthesis of cotton ber. In the future, molecular mechanisms and regulatory relationships between GhD14 and secondary cell wall biosynthesis needs to be further investigated to deep our understanding of the strigolactone and secondary cell wall biosynthesis.
The GhD14A/D gene silencing enhanced fruit branch angles and reduced ber length in cotton (Fig. 5), indicating that the GhD14A/D transcripts accumulation level affects cotton plant architecture and ber length. The plant architecture is also regulated by auxin, cytokinin, and gibberellic acid (GA). D14 is the receptor of SL, which involved in plant stature and inhibition of plant shoot branching (Arite et al. 2009).
The strigolactone signal can regulate the auxin polar transport and the cytokinin content in the stem (Ferguson et al. 2009). In the future, molecular mechanisms and regulatory relationships between GhD14 and auxin or cytokinin should be examined to deepen our understanding of strigolactone signaling in cotton architecture. Although we did not investigate this, we speculate that GhD14 might regulate branch number and angles through auxin and GA in cotton. In the future, molecular mechanisms and regulatory relationships between GhD14 and auxin or GA should be examined to deepen our understanding of SL signaling in cotton architecture.

Conclusion
Strigolactone (SL) signaling functions directly in regulating plant development. DWARF14 (D14), a noncanonical SL receptor, interacts with the F-box protein MAX2 to modulate SL signaling in rice and Arabidopsis. In this study, the GhD14D was cloned from G. hirsutum, which containing a hydrolase fold motif, and had similar protein domains as typical D14 proteins. Subcellular localization analysis revealed that GhD14D was located in the cytoplasm and nucleus. qRT-PCR results showed that GhD14A/D was highly expressed in stem, ower and 20 DPA ber cells; GUS staining assay indicated that GhD14D was mainly expressed in leaf primordium, in orescence, axillary bud and stem. Reducing the transcripts of GhD14A/D in cotton signi cantly increased branch angle and reduced ber length and the transcripts of secondary cell wall biosynthesis related genes. In addition, overexpression of GhD14D successfully rescued the phenotype of Arabidopsis d14 mutant with reduced shoot-branching, length-width ratio of rosette leaves and plant height. These results indicate that GhD14D contributes to ber development and plant architecture development in cotton, which will lay a foundation for manipulating cotton plant architecture by genetic engineering and provide a candidate gene for producing cultivars with ideal architectures in the future.

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