Molecular cloning and characterization of GhERF105, a gene participated in the regulation of gland formation from cotton (Gossypium hirsutum L.)

Abstract


Abstract
Background Gossypium hirsutum L. (cotton) is one of the most economically important crops globally. Cottonseed is the signi cant source of ber, feed, foodstuff, oil and biofuel products. However, the utilization of cottonseed was limited by the presence of small and darkly pigmented glands that contain large amounts of gossypol, which is toxic to human beings and other non-ruminant animals. To date, there has been some progress in the pigment gland formation, but the underlying molecular mechanism of pigment gland formation was still complicated and unclear.

Results
In this study, we identi ed an AP2/ERF transcription factor named GhERF105 (Gh_A12G1784), which was involved in the regulation of gland pigmentation, from comparative transcriptome analysis of the leaf of two pairs of glanded and glandless accessions, which are CCRI12 and CCRI12XW, L7 and L7XW. It encoded an ERF protein localized in the nucleus with transcriptional activation activity containing a conserved AP2 domain, and showed the high expression in glanded cotton accessions that contained much gossypol. Virus-induced gene silencing against GhERF105 caused the dramatic reduction in the number of glands and signi cantly lowered levels of gossypol in cotton leaves. GhERF105 showed the patterns of spatiotemporal and inducible expression in the glanded plants.

Conclusions
These results suggest that GhERF105 participates in the pigment gland formation and gossypol biosynthesis in partial tissue of glanded plant. It also provides a potential molecular basis to generate 'glandless-seed' and 'glanded-plant' cotton cultivar.

Background
Cotton (Gossypium spp.) is a globally appreciated crop for its economic value of the textile ber, feed, foodstuff, oil and biofuel products in the world [1][2]. There are approximately 50 species in the Gossypium genus, four of which are cultivated in agriculture, including two diploid cottons (G.herbaceum and G.arboreum) and two allotetraploid cottons (G.hirsutum and G.barbadense) [3][4][5]. G.hirsutum, the upland cotton, is most widely grown today and dominates world cotton commerce with more than 90% of the annual cotton ber production [5][6]. However, the potential of nutrition sources of cottonseed cannot be used su ciently due to the presence of gossypol stored mainly in the small darkly pigmented lysigenous glands in plants and seeds, which is toxic to human beings and other non-ruminant animals [7][8]. Gossypol, as phytoalexin, is a yellowish phenolic compound that serves as a protective function against various pests, diseases and abiotic stresses in certain species of cotton plants of the family Malvaceae [9][10][11]. Therefore, developing cotton with low-gossypol seeds and high-gossypol plants has become an interesting area of cotton breeding for researchers.
The pigment glands, which are also called 'gossypol glands', 'internal glands' and 'black glands' located in certain of the subepidermal layer of hollow organs in many parts of the plant, originate from a cluster of cells in the ground meristem, which differ from other cells in that they have a high-density gossypol and related terpenoids [7]. Research on the molecular genetic mechanisms of pigment gland in the cotton plant began in lines of 'Hopi Moencopi' in the 1950s [12][13][14]. So far, the research of cotton inheritance has indicated that the gland formation is controlled by a combination of at least six independent loci such as gl 1 , gl 2 , gl 3 , gl 4 , gl 5 and gl 6 , the different combinations of dominant(Gl) and recessive (gl) alleles modulate gland formation in different organs [14][15][16][17][18][19]. The completely glandless phenotype was controlled by two pairs of duplicate homozygous recessive genes (gl 2 gl 2 gl 3 gl 3 ) in the allotetraploid G.hirsutum [13][14], while the dominant alleles (Gl 2 , Gl 2 , Gl 3 , Gl 3 ) in any combination produced the glanded phenotype with variable distribution in different organs [14,20]. In 1965, the gl 2 and gl 3 gene were located on chromosome (chr.) A t 12 and D t 12 of G.hirsutum, respectively [16,[21][22]. Alleles gl 4 and gl 5 decrease the number of glands while gl 6 have the weaker effects on gland formation compared with gl 1 [23][24]. Subsequently, gl 2 arb , gl 2 b , gl 3 dav , gl 3 thur , gl 3 rai , gl 3 b [7], Gl 2 s [25], Gl 2 e [26][27] and Gl 2 b [28] related to pigment gland formation were identi ed. Among them, Gl 2 e is the most critical gene that controls glandless character of the whole plant. A single completely dominant glandless G. barbadense mutant(Gl 2 e ) named 'Bahtim 110' ( G. barbadense L), which is a dominant allele of Gl 2 that shows epistatic effect on Gl 3 , was originally discovered in Egypt by the irradiation mutagenesis of the sea-island cotton 'Giza 45' seeds with 32 P, and could e ciently inhibit the formation of pigment gland [29][30][31][32]. Since then, several genes for gland formation have been discovered gradually by researcher. In 2016, GoPGF gene(Gossypium Pigment Gland Formation gene),which encodes a basic helix-loop-helix transcription factor as the critical gene, was identi ed through map-based cloning approach and located on chromosome A t 12 [33][34]. CGF3 (Cotton Gland Formation), which is identical to GoPGF gene, not only controls the gland morphogenesis directly, but also regulates gossypol biosynthesis indirectly [35]. CGP1 (Cotton Gland Pigmentation 1), which interacted with GoPGF, was identi ed through comparative transcriptome analysis of glanded and glandless cotton accessions and was involved in the regulation of gossypol biosynthesis but not gland formation [36]. In addition, the novel RanBP2 zinc nger protein (ZFP) and GauGRAS1, which played the roles in the development of the cotton gland, were identi ed using suppression subtractive hybridization (SSH) from upland cotton 'Xiangmian 18' [9,[37][38][39]. During the past three decades, there has been some progress in the molecular mechanism of gland formation and the relationship between gossypol and pigment gland. However, the speci c mechanism of pigment gland formation remains complicated and still unclear up to now.
Here, we identi ed an Ethylene Response Factor named GhERF105, which was involved in the regulation of gland pigmentation, from a comparative transcriptome analysis of the leaf of two pairs of glanded and glandless allotetraploid cotton accessions, which are CCRI12 and CCRI12XW, L7 and L7XW (Fig.S1).
It encoded an ERF protein localized in the nucleus with transcriptional activation activity containing a conserved AP2 domain and showed the high expression in glanded cotton accessions that contained much gossypol. Silencing of GhERF105 by VIGS not only resulted in the drastic reduction of gland, but also decreased the accumulation of gossypol in the leaves of the treated plants. Moreover, GhERF105 showed a temporal and spatial pattern of expression in various hollow organs of glanded and glandless cotton plants, including Cotyledon, Hypocotyl, Petiole, True leaf and Stem, and demonstrated the inducible expression under ethylene treatment. In addition, GhERF105, CGF, CGP1 and GoPGF genes were highly expressed in the leaves and stems in glanded CCRI12 and L7 but were indeed substantially lower expression in CCRI12XW, CCRI12YW and L7XW.
These results provide a reference for the comprehensive analysis of the molecular mechanism of formation of gland and gossypol biosynthesis in cotton. However, the diversity of gland trait inheritance indicates the regulation complexity of gland formation. Further studies are needed to better understand the molecular mechanisms underlying gland development.

Results
Sequence analysis of the full-length GhERF105 gene In this study, we identi ed an Ethylene Response Factor named GhERF105 (Gh_A12G1784) through a comparative transcriptome analysis of the leaves of two pairs of glanded and glandless accessions, which are CCRI12 and CCRI12XW, L7 and L7XW. The GhERF105 gene, which was cloned from the leaves of CCRI12 is 711bp in length with an open reading frame (ORF) of containing initial code (ATG) and terminal code (TAA) with no intron (Fig.1). The predicted protein comprised of 236 amino acids with relative molecular weight of 26.3kDa and isoelectric point of 7.72 containing an ERF conserved DNA binding domain. The cotton GhERF105 belongs to the largest AP2/ERF family of transcription factors that plays an important role in plant development and environmental stress responses, as well as hormone signaling and pathogen defense [40][41][42].
The expression analysis of GhERF105 gene in many cotton accessions Because GhERF105 gene was identi ed from the comparative transcriptome analysis of the leaf of two pairs of Near Iso-genic Lines (NILs) with glanded and glandless phenotype. Therefore, the expression levels of GhERF105 were analyzed in two pairs of Near Iso-genic Lines (NILs) and other cotton accessions, the results show that GhERF105 was highly expressed in the leaves and stem of glanded G.hirsutum. (CCRI12, L7 and TM-1) but was indeed substantially lower expression in CCRI12XW, L7XW and CCRI12YW (Fig.2). Based on the different expression pattern of GhERF105 in partial tissues of six cotton accessions, GhERF105 may be related to the formation of glands. However, its function and regulatory mechanism in pigment gland development need further be investigated using VIGS technology and other technology.
Silencing of GhERF105 reduced gland formation and gossypol biosynthesis Here, In order to further ascertain the function of the GhERF105 during pigment gland formation. Agrobacterium-mediated VIGS systems was constructed using a TRV-based VIGS vector for silencing GhPDS and GhERF105 genes in the cotton seedlings. The results were as follows: (1)Silencing the expression of the endogenous phytoene desaturases gene(PDS), which is commonly used marker gene for VIGS, causes loss of chlorophyll and carotenoids [43]. A photobleaching phenotype in cotton plants in ltrated with GhPDS-expressing agrobacteria was observed 14-21 days after in ltration in true leaves, compared to the leaves in plants in ltrated with negative gene-expressing agrobacteria (Fig.3a). (2) To assess its function, we cloned the 289bp fragment of GhERF105 from CCRI12 plant and inserted it into pTRV2 for virus-induced gene silencing (VIGS) to suppress the expression of endogenous in cultivated glanded allotetraploid cottons. Compared with that in the untreated CCRI12 as the negative control ( Fig.3b1-b2), The GhERF105-silenced CCRI12 plants exhibited the dramatic reduction in gland numbers in the new true leaf of 14-21d after in ltration ( Fig.3 b3-b6). The transcript levels of GhERF105 in pTRV-GhERF105 leaves were prominently lower than those in the untreated CCRI12 but still higher than those in the untreated CCRI12XW (Fig.3c). However, the veins of the new emerging true leaves had fewer dotted glands and the stems had thickly dotted glands ( Fig.3 b5-b6, Fig.S2). These data suggested that GhERF105 regulated the formation of glands in leaf but not stem, in contrast to GoPGF, which results in glandless phenotype in all tissues, including the leaves and stems [29]. (3) we conducted HPLC analysis to measure the level of gossypol in the leaves gossypol content was reduced by over 78% in the GhERF105-silenced leaves compared with the untreated CCRI12 leaves but still higher than those in the untreated CCRI12XW (Fig.3d). In all, the results suggest that GhERF105 may be involved in the pigment gland formation and gossypol biosynthesis.

Spatiotemporal expression analysis of GhERF105 gene
The pigment glands are located on the surfaces of the stem, leaves, sepals, petals, and stigma [17]. GhERF105 gene was associated with the development of cotton pigment gland. Therefore, the transcription level of GhERF105 gene was detected by qRT-PCR in different organs of gland development in glanded and glandless cotton accessions. The result showed the mRNA levels in Cotyledon, Hypocotyl, Petiole, True leaf and Stem of the gland plant were increased to 3.5, 10.5, 15, 8.7 and 4.0 folds of that in glandless plant, respectively. That is to say, the mRNA levels of GhERF105 in the above organs of the glanded plants was higher than that in the glandless plants. At the same time, the expression level of GhERF105 was highest in the leaf of glanded plants but wasn't signi cant differences between the leaf and other organs of glandless plants (Fig.4). Therefore, the GhERF105 gene is of a highly different expression pattern between the glanded and glandless cotton plants in pigment gland formation.
Nuclear Localization and Revealed Transcription Activity of GhERF105 protein The green uorescent protein (GFP) reporter, which is a vital marker for protein subcellular localization, showed a very strong uorescence signal under the control of the constitutive CaMV35S promoter, and the signal was uniformly and diffusely distributed throughout the cell. Based on functional annotation information, GhERF105 is believed to act as a transcriptional factor. Therefore, the nuclear localization should be essential for the function of GhERF105. To test this possibility, the coding sequence (CDS) of GhERF105 was fused to the green uorescent protein (GFP) reporter gene. After introducing the construct ( Fig. 5a,S3)into the tobacco cells by agro in ltration, GhERF105-GFP, the transcription factor fused to GFP, was expressed transiently and located exclusively in the nucleus of tobacco epidermal cells (Fig.5b).This con rms that GhERF105-GFP is a nuclear localized protein that is mainly involved in nuclear transport.
The yeast strains transformed with the pGBKT7-GhERF105 were able to grow blue colonies on the selected medium SD/-Trp/-X-a-gal while those stains with empty vector pGBKT7 could grow white colonies (Fig.6). This result indicated that pGBKT7-GhERF105 is of the transcriptional activity, implicating a role of GhERF105 as a transcription activator.
Expression pattern of GhERF105 gene in cotton under ethylene treatment The ERFs, which are important plant-speci c transcription factors in the ethylene signal transduction pathway, have been shown to play an important regulatory role in modulating the expression of speci c stress-related genes [44][45][46]. Ethylene interacting with other plant hormone, regulated the programmed expression of pathogenesis-related (PR) genes in the ethylene-mediated signaling pathways [47].
Programmed cell death (PCD) plays an important role during the development of pigment glands in Gossypium hirsutum leaf tissue [48]. Ethylene, which is the upstream signal moleculars during PCD process, mediates the PCD signal by ROS [49].Therefore, It is meaningful to investigate the expression pattern of GhERF105 gene in response to stress hormone ethylene stimuli. In this study, qRT-PCR analysis was employed to detect GhERF105 expression level changes in leaves at different time after ethylene treatment. Compared to that in the water-treated plants, The GhERF105 mRNA was rapidly accumulated and peaked within 8h, and then declined to the original level in the ethylene-treated plants, The result suggested that the level of the mRNAs of GhERF105 gene was inducible at the early stage of ethylene treatment and maintained the high level from 6h to 10h by the stress hormone ethylene in leaf (Fig.7).
However, the expression pattern of GhERF105 did not positively correlate with the length of time after ethylene treatment. The result indicates that the expression of GhERF105 is was characteristically responsive to ethylene treatment at the transcriptional level and GhERF105 may be related to ethylene signal transduction pathways or defense/stress signaling pathways. At the same time, It is tempting to speculate that gland formation and gossypol synthesis in cotton may be induced and regulated directly or indirectly by ethylene.

Expression Patterns of genes involved in gland formation
To see whether there is the relationship between GhERF105 and other genes (such as CGF1, CGF2, CGP1 and GoPGF) that are involved in the gland formation [16,21,29]. The expression levels of GhERF105, CGF1, CGF2, CGP1 and GoPGF were analyzed in the leaf and stem of ve cotton accessions by qRT-PCR, including glanded G. hirsutum(CCRI12, TM-1 and T582), dominant glandless CCRI12XW and recessive glandless CCRI12YW. Results obtained from qRT-PCR analysis con rmed that GhERF105, CGF1, CGF2, CGP1 and GoPGF were highly expressed in the leaves and stems in glanded CCRI12 and TM-1 but were indeed substantially lower expression in CCRI12XW and CCRI12YW (Fig.8). In addition, we also observed that the expression level of the genes was signi cantly higher in the leaves than in the stems for G. hirsutum (T582) (Fig.9). The results showed that GhERF105 has the similar expression pattern as GoPGF, CGF1, CGF2 and CGP1 in some cotton accessions. This results further concluded that GhERF105 was associated with cotton pigment gland development and indicated that the relationship was unknown between GhERF105 gene and GoPGF, CGF1, CGF2 and CGP1 gene in gland formation.

Discussion
To date, developing cotton varieties, which produce low-gossypol seeds and high-gossypol plants, has become an important topic of cotton breeding for researchers. Therefore, it is very signi cant to understand the molecular mechanism of the pigment gland formation and the relationship between gossypol and gland in cotton.
In the recent years, the considerable efforts have been made by researchers to accumulate knowledge and to identify a series of genes related to pigment gland formation and gossypol synthesis. GoPGF/CGF3, which plays the most critical and direct role in gland development, independently regulates gland morphogenesis and indirectly affect gossypol biosynthesis possibly through regulating the expression of gossypol-related genes through binding to the G-box motif by yeast one-hybrid assays [34]. CGF1 shows similar functions to CGF3, and CGF2 regulates the density of pigment glands [35]. Silencing GoPGF results in the apparent absence of glands in all tissues of gland cotton and leads to an almost complete lack of gossypol [34][35]. CGP1 modulates gossypol accumulation but not gland morphogenesis, Knockout of CGP1 by CRISPR/Cas9 and VIGS produces a strong reduction in gossypol levels [36]. Silencing GauGRAS1 by VIGS leads to glandless stems and petiole and didn't change the form of glands in the leaves in G. australe. Moreover, the gossypol content in the stem of the GauGRAS1silenced plants was signi cantly reduced [38]. However, the molecular mechanism for pigment gland formation remains complicated and unclear, which has limited progress in low-gossypol breeding of cotton. Therefore, the further exploring of molecular mechanism for gland formation may facilitate the genetic improvement of cotton.
Here, this study provides several evidences that GhERF105 gene was associated with gland formation of partial tissues in the glanded plant. First, GhERF105 gene was identi ed through the comparative transcriptome analysis of the leaf of two pairs of glanded and glandless cotton accessions. Second, GhERF105 was highly expressed in the glanded accession. Third, GhERF105 knockout via VIGS markedly resulted in the drastic reduction of visible pigmented glands and decreases the content of the gossypol in the leaves but didn't change the density of gland on the stem of the cotton. In addition, the expression pattern of GhERF105 was similar with that of another known genes related to gland development (such as GoPGF and CGF) in some glanded and glandless accessions [Fig. 7]. These ndings further indicated that GhERF105 may be involved the gland formation in cotton. Nevertheless, the regulatory mechanism on the pigment gland was somewhat different between GhERF105 and GauGRAS1. Ma et al. (2016) have con rmed that GoPGF can bind to G-box motif by Yeast one-hybrid assays [34].The promoter region of GhERF105 includes G-box cis-acting elements. It is therefore tempting to speculate that GoPGF likely regulates the expression of GhERF105 by binding to G-box cis-acting elements in the nucleus and modulate the expression of gossypol-related genes by binding to the related cis-acting elements of their promoter directly and indirectly (Fig. 10), the speculation will be needed further be veri ed by the results of related experiments.
In conclusion, the cloning and characterization of GhERF105 both provide new information to study the molecular mechanism of gland formation and its functions in upland cotton.  [50]. 'TM-1', which is widely used as a genetic standard, is the glanded accession of the seeds and the whole plant, 'T582' is an accession with glandless-stem and glanded-leaf of plant traits. All materials were stored by self-crossing for several years in our lab.

Conclusions
The seeds were immersed in water and followed by germination in a high humidity environment at 28°C in the dark for 2d. Well-germinated seeds were subsequently planted in 0.3 litres pots of 7cm diameter with one seed per pot in a commercially available sand/soil/fertilizer mix and grown for two to three weeks at 28°C (16h light and 8h dark) with LED lamps (Opple lighting Zhongshan China ) in a greenhouse.

Extraction of total RNAs
Samples from different tissues of the cotton plants, including Cotyledon, Hypocotyl, Petiole, True leaf and Stem of one or many different gland accessions served as the source of total RNA were immediately frozen in liquid nitrogen and stored at -80°C. For each sample, total RNAs were isolated from 100 mg of leaf ground with liquid nitrogen using the RNAprep Plant RNA kit (polysaccharides&polyphenolics-rich) (TIANGEN BIOTECH (BEIJING)CO., LTD) according to the manufacturer's instructions. The quantity and purity of RNAs were assessed according an absorbance ratio of OD 260/280 (1.9-2.1) using a NanoDrop One C Microvolume UV-Vis Spectrophotometer with Wi-Fi (Thermo Fisher Scienti c Inc., Waltham, MA, USA) ultraviolet spectrophotometer, and was con rmed using 1.0% (w/v) denatured formaldehyde agarose gel electrophoresis to investigate its integrality.
Synthesis of the rst-strand cDNA RNA was reversely transcribed into 1st strand cDNA in a 20μL reaction volume using the PrimeSeript TM 1stStrand cDNA Synthesis Kit (TaKaRa Bio, Dalian, China) following the manufacturer's protocol of Reverse Transcription System. Firstly, two micrograms(2μg) of total RNA was mixed with 1.0μL Oligo dT Primer(50μm), 1.0μL dNTP mixture(10mM each), then RNase free ddH 2 O was added to make the whole reaction volume up to 10μL, then, A total 10μL reaction volume was incubated at 65℃for 5min and placed on ice for 2 min to denature probable RNA secondary structure. Secondly, the rst-strand cDNA synthesis mixture was prepared by adding following components to the above 10μL reaction volume in the indicated order, 4μL 5xPrimeScript II Buffer, 0.5μL RNase Inhibitor(40U/μL), 4.5μL RNase free ddH 2 O, and1μL Primescript 1I RTase(200U/μL). The rst-strand cDNA synthesis mixture was incubated at 30℃for 10 min, 42℃for 60min and terminated at 95℃for 5 min.

Gene Expression Analysis by Quantitative Real-Time PCR
The RNA sequencing samples that were isolated were used to perform real-time quantitative (qRT-PCR) analysis using the ABI Quantstudio 5 Detection System (Applied Biosystems, Carlsbad, CA). Actin (GenBank accession numbers: AY305733) was used as reference gene. The gene-speci c primers with about 215bp product size were designed using the Primer 5.0 software or online in NCBI website (https://www.ncbi.nlm.nih.gov/tools/primer-blast/index.cgi?LINK_LOC=BlastHome) and listed in Table  11. The speci city of each primer set was validated by melt-curve analysis, and the e ciency was calculated by analyzing the standard curves with a tenfold cDNA dilution series (Bustin et al., 2009). The 20μL qPCR experiment was carried out on the ABI Quantstudio 5 Detection System with TB Green Premix Ex Taq TM (Tli RNaseH Plus) (TaKaRa Bio, Dalian, China). The reaction volume contains 0.5μL of each primer (10μM), 0.4μL ROX Reference DyeII (50x), 1μL above synthesized cDNA template, and 7.6μL of sterilized ddH 2 O. The qPCR thermal cycling conditions were 95°C for 5 min to pre-denature cDNA template; 40 ampli cation cycles of 95℃ for 5S, 55℃ for 30S, and 72℃ for 30S; and followed by 15s at 95°C, 1 min at 60°C, and 15 s at 95°C. Each sample was run in triplicate, each biological replicate was assessed three times. The relative expression level of the genes were calculated according to the 2 −ΔΔCT method [51]. For the reference gene used in this experiment, their geometrical mean was operated at rst, and then the relative transcript level of target gene was calculated following the method of one reference gene. Results were generally expressed mean ± standard error (ER) from values three independent tests.
PCR ampli cation of the GhERF105 gene was performed in a reaction volume of 15μL containing 1μL of template cDNA, 1.2μL of dNTP Mixture 2.5mM , 3μL of 5×PrimeSTAR GXL Buffer, 0.3μL of Prime STAR GXL DNA Polymerase, 0.4μL of forward primer and 0.4μL of reverse primer(10μΜ for each), and 8.7μLddH 2 O. The ampli cation product was achieved using the following pro le: 5 min at 98°C; 35cycles of 10 s at 98 °C, 15s at 55℃,1 min at 68°C and a nal cycle of 5 min at 68°C; hold at 10°C. The PCR products were then puri ed following the instructions in the QIAquick PCR Puri cation Kit (250) (Qiagen, Düsseldorf, Germany) and eluted in a nal sample volume of 35µL of Qiagen EB buffer. three microliters of each PCR product were assessed by size on a 1% agarose gel to select fragments in the range of 700bp ± 50 bp.

VIGS procedure
The VIGS (Virus-induced gene silencing) vector tobacco rattle virus (TRV) invades a wide range of hosts and is able to spread vigorously throughout the entire plant but produces only mild symptoms [52]. Therefore, VIGS system has been proven to be a powerful tool in elucidating gene function and functional genomics in cotton [10, 34, 35-38, 53, 54]. To knockdown the expression of GhERF105, The pTRV-VIGS vectors were constructed using a previously published method [53,55]. Brie y, cDNA fragments of cotton PDS (GhPDS1, 327bp, GenBank accession numbers: HQ441184) and Pigment gland formation GhERF105 (337bp) were ampli ed using Prime STAR GXL DNA Polymerase (TaKaRa) from CCRI12 by PCR with gene speci c primers (listed in the table 2). The resulting products were cloned into pTRV2 with BamHI and KpnI to produce recombinant vectors named pTRV2::PDS and pTRV2::GhERF105, respectively. These recombinant vectors and the empty vector (pTRV2::00) were then introduced into the Agrobacterium strain GV3101(Weidi Bio, Shanghai, China) by Heat shock method, For the VIGS assay, the transformed Agrobacterium colonies containing pTRV1 and pTRV2-GhPDS, pTRV2-GhERF105 were grown overnight at 28℃ in an antibiotic selection medium containing rifampicin, Gentamicin and kanamycin 50mg/ml, and suspended in the solutions (10mM 2-(N-morpholino) ethane sulfonic acid, 10mM MgCl 2 and 400µM acetosyringone (AS)) to the nal optical densities as OD values of 1.5 at 600nm and then left at 25℃for 4h without shaking in the dark. Before in ltration, Agrobacterium cultures containing pTRV1 and pTRV2 or its derivatives were mixed in 1:1 ratios. Seedlings with the fully expanded cotyledons but without a visible true leaf of CCRI12 were in ltrated by inserting the Agrobacterium suspension containing pTRV1 and pTRV2, pTRV2-GhPDS, pTRV2-GhERF105 into the cotyledons via a syringe. Plants were grown in the pots at 25℃ in a growth chamber under a 16h light/8h dark photoperiod with 70% humidity. To analyze silencing e ciency, RNA was extracted and qRT-PCR was performed. The Actin (GenBank accession numbers: AY305733) and GhERF105 was ampli ed as reference gene and target gene, respectively [56]. Leaves were numbered sequentially such that number 1 refers to the rst true leaf initiated after the cotyledons. In this study, leaves 2-3 were investigated and collectively referred to as total foliage [57].

Gossypol detection and analysis
The total gossypol concentration in the leaves from CCRI12, GhERF105-silenced CCRI12 and CCRI12XW plants was determined by high-performance liquid chromatography (HPLC) (Agilent 1100, Agilent, Santa Clara USA). Each 100 mg plant sample, which was freeze-dried and ground into powder with liquid nitrogen. was dissolved with 2ml leaf extraction (acetonitrile/water/phosphoric acid=80:20:0.1). The leaf extraction was centrifuged at 10000rpm for 10 min and then the supernatant was carefully transferred into a new EP tube at room temperature. The eluent was ltered through a 0.45μm nylon lter into a vial for HPLC analysis with Agilent Zorbax Eclipse Plus C18 analytical column (250mm×4.6mm, 5micron). The sample was analyzed at a wavelength of 235nm. The concentration was calculated using Agilent 1100 system by comparing to the gossypol standard curve. A gossypol reference standard was purchased from Sigma Chemical Co. Ltd.

Subcellular Localization of GhERF105 Protein
To study the subcellular localization of GhERF105 protein, the coding regions of GhERF105 was ampli ed with stop codon removed by Primers listed in Table 3, which contained a XbaI and SmaI site (underlined) through polymerase chain reaction (PCR), The resulting fragments were cloned between the XbaI and SmaI site of the transient expression pBI121-GFP vector, which harbors an ORF encoding the green uorescent protein (GFP) under the control of the CaMV35S promoter, and construct the recombinant plasmid p35S-GhERF105-GFP. p35S-GFP was used as positive control. The plasmids of GFP-GhERF105 and GFP were then introduced into tobacco leaves (Nicotiana benthamiana) respectively via Agrobacterium-mediated transformation and incubated at 25℃ under light for 48-72h. The green uorescence signals were observed and the localization of the fusion protein was determined using a confocal laser scanning microscope (Leica TCS SP8, Germany).

Transactivation Activity Assay of GhERF105 Protein
To study the transactivation activity of GhERF105 protein, GhERF105 cDNA was ampli ed with by Primers listed in Table 4 and cloned into the EcoRI and NotI sites of pGBKT7 vector to generate pGBKT7-GhERF105 construct. This plasmid with empty vector control was then transformed into yeast strain AH109 to analysis the transactivation activity. Yeast transformants with OD600 of 0.1, 0.01and 0.001were plated on the selective media, SD/-Trp and SD/-Trp/-X-a-gal, and incubated at 30°C for 4 days.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.

Competing Interests
The authors declare there are no competing interests. Figure 1 Ampli cation of the full-length cDNA of GhERF105. 1: DNA marker; 2-3: the full-length cDNA of GhERF105   Functional characterization of GhERF105 by VIGS. a The photo-bleaching phenotypes of cotton seedlings in CCRI12 inoculated with pTRV::GhPDS and empty vectors. TRV::GhPDS and TRV::00 are the positive control and negative control, respectively. b Relative transcript levels of GhERF105 of leaf inoculated with pTRV::GhERF105 or empty vector control. c The gossypol content in empty vector (TRV::00) and in the GhERF105-silenced leaves of CCRI12.and CCRI12XW, Actin was used as an internal control. d d1-d2

Figures
Phenotypes of Gossypium hirsutum CCRI12 inoculated with pTRV::00 vector. d3-d6 Phenotypes of Gossypium hirsutum CCRI12 inoculated with pTRV:: GhERF105 vector. d7-d8 Phenotypes of Gossypium hirsutum CCRI12XW. d1-d8 are enlarged versions of the positions indicated by the yellow box in Fig.3. d1 correspond to d2, d3 correspond to d4, d5 correspond to d6 and d7 correspond to d8. the red arrow indicates the location of the glands on the leaf. Scale bars, 1mm. Each bar value represents mean ± SD of three independent experiments. Error bars are the SD of three biological repeats.   Expression analysis of GhERF105, GoPGF, CGF1, CGF2 and CGP1 genes in the leave and stem of T582 cotton cultivar.