Background

Sulfur is one of the most basic elements in the plant life. Its assimilation in higher plants and the decrease of metabolically important sulfur compounds are key factors in plant growth, vigor and stress resistance [1]. Sulfur plays an important role in the structure, regulation and catalysis of proteins. According to the previous study, sulfation is essential for nodulation factors of rhizobia to signal to plants in bacteria [2]. In mammals, sulfation contributes to the homeostasis and regulation of many endogenous compounds with biological activity [3]. In plants, the sulphate conjugation reaction appears to play an important part in plant growth, development and stress adaptation [4]. Sulfate must be activated by two subsequent activation steps to form adenosine-5′-phosphosulfate (APS) and 3′-phosphoadenosine-5′-phosphosulfate (PAPS) before being used for biochemical conversion [5].

Sulfotransferases (SOTs) (EC 2.8.2.-) catalyze the transfer of a sulfate group from PAPS to a hydroxyl group of different substrates [6]. The first plant SOT gene was cloned from Flaveria species (Asteraceae), which was related to the sulfation reaction of flavonol [7]. Subsequently, the cDNA encoding sulfotransferase was isolated from Arabidopsis thaliana and its deduced 302 amino acid polypeptide was highly correlated with plant flavonol sulfotransferase [8]. SOTs are widespread among higher plants, animals and eubacteria [1, 9]. Based on previous studies, SOT proteins were involved in the regulation of diverse physiological and biological processes, such as growth, development, adaptation to land, stomatal closure, drought tolerance and pathogen infection [1, 3, 8,9,10,11,12,13,14,15,16,17,18]. SOTs of Flaveria species were well characterized by means of molecular biology and biochemistry and used as a general model of plant SOTs [7]. These SOTs accept different flavonols as sulfate receptors, which may be involved in adaptation to stress or polar auxin transport. When Arabidopsis seedlings were treated with hormones or stress-related compounds, SOT protein expression was significantly induced by salicylic acid and methyl jasmonate. In addition, the accumulation of SOTs was also observed in the leaves or cell suspensions of mature plants after infection with bacterial pathogens [8]. Several other reports revealed that SOTs can directly catalyze thioglucosate, brassinosteroid, jasmonate, flavonoids and salicylic acid, and directly or indirectly participated in defense signaling, development and stress responding [1, 10, 12, 16, 19].

Cotton (Gossypium) is a major industrial crop that provides important natural fibers and edible oil in the world. The genus contains 45 diploid and 5 tetraploid species. Among them, Gossypium hirsutum L. has been cultivated worldwide and currently accounts for the vast majority of the world’s fiber output (> 90%) of the world’s fiber production [20,21,22]. The cotton fiber is a unique elongated cell, which is helpful to study cell differentiation. Cotton fibers are single-cell trichomes differentiated that has undergone four major developmental stages, including initiation, elongation, secondary cell wall synthesis, and maturity [23]. The development of cotton fibers in elongation and secondary cell wall synthesis determines the length and strength characteristics of the fiber [24]. In addition, fiber development is a complex process involved in many pathways, including various secondary metabolism, hormone, signal transduction and transcriptional regulatory components [25, 26]. For example, one of the flavonoids, naringenin has been verified to be negatively correlated with fiber development [26, 27]. Auxin and brassinosteroid promoted the fiber initiation as well as elongation; gibberellin acid and ethylene played a positive role during the fiber elongation phase [25,26,27,28,29]. On the other hand, cytokinin, abscisic acid played an opposite role [11]. Jasmonic acid participates in various developmental processes. Different concentrations of jasmonic acid play different roles and high concentration of jasmonic acid inhibits fiber initiation [30, 31]. Similarly, jasmonate inhibited cotton development to some extent by inhibiting gibberellin signal [32]. Overaccumulation of jasmonic acid inhibited both lint and fuzz fiber initiation, reduced the fiber length, and lead to a fiberless phenotype in cotton seeds [33].

Considering that SOTs directly catalyze brassinosteroid, jasmonate, flavonoids and salicylic acid, which are related to growth, cotton fiber development and stress adaptation, it is necessary to understand the information of SOT gene family in Gossypium in order to better understand the relationship between sulfation reaction and physiological processes. However, as far as we know, there is no systematic study of the SOT family in Gossypium. In this study, we identified 46, 46, 76 and 77 SOT genes from G. arboreum, G. raimondii, G. barbadense, and G. hirsutum, respectively, and then looked into the features such as chromosomal locations, phylogenetic evolutionary relationships, gene structures, conserved motifs, tissue and subcellular localization, as well as expression patterns. Our study provided a comprehensive analysis of the Gossypium SOT gene family and the results might be useful in understanding the role of SOT in plant development.

Results

Identification, characterization and chromosomal distribution of SOT genes in four cotton species

According to the results of BLASTP and HMMER 3.1, a total of 245 SOT genes were identified from four cotton species, including 46 genes of G. arboreum, 46 genes of G. raimondii, 76 genes of G. barbadense and 77 genes of G. hirsutum. Protein sequence analysis indicated that all SOT gene proteins encoded a wide range of amino acids ranging from 60 to 672, with an average molecular weights (Mw) at 32.48 kDa and isoelectric points (pI) at 6.58. Subcellular localization analysis showed that 70.6% of 245 SOT genes were localized in the cytoplasm, which may be consistent with their functions as transferases. The SOT gene names, locus IDs and other characteristics were listed in Additional file 1: Table S1.

245 SOT genes distributed unevenly on the chromosomes in four cotton species (Fig. 1). Chr09 of G. raimondii contained the largest number of SOT genes (11). By contrast, Chr03/ Chr08 of G. arboreum, Chr04/ Chr05/ Chr10 of G. raimondii, A03/ A08/ D02/ D08 of G. barbadense and A03/ A08/ D02/ D08 of G. hirsutum contained none of SOT genes. In addition, the distribution of SOT genes in G. barbadense and G. hirsutum showed some similarities. So, we further analyzed the collinearity of the SOT gene across these four genomes.

Fig. 1
figure 1

Chromosome distribution of SOT genes of four cotton species. The cotton species name was on the left of graphic and chromosome name was at the top of each bar. The vertical scale on the left showed the size of chromosomes and black lines indicated the corresponding position of genes. The scale of the chromosomes was millions of base pairs (Mb). The gene names was correspond to those in Additional file 1: Table S1

Full size image

Collinearity and duplication analysis of SOT genes

We found out all the homologous genes among these four cotton genomes to analyze the collinearity relationships of SOT genes (Fig. 2 and Additional file 1: Table S2). Among all the 77 SOT genes of G. hirsutum, 39 GhSOTs had intergenomic homologous genes in G. arboretum, 37 homologous genes in G. raimondii and 49 homologous genes in G. barbadense, respectively. In total, we identified 32 pairs of common homologous SOT genes in the four cotton species.

Fig. 2
figure 2

Collinearity analyses of SOT genes among G. hirsutum, G. barbadense, G. arboretum and G. raimondii. From top to bottom, three graphics displayed the collinear relationship between G. hirsutum and G. arboretum, G. hirsutum and G. raimondii, G. hirsutum and G. barbadense, respectively. Grey lines in the background showed the collinear relationship across the whole genome, while the red lines predominantly displayed the collinear SOT gene pairs

Full size image

Previous studies in Gossypium showed that gene families always expanded through tandem, whole-genome and segmental duplications [34, 35]. In G. hirsutum, 20 pairs of tandem duplication gene pairs (32 genes) distributing on 12 chromosomes were found (Fig. 3a and Additional file 1: Table S3). In addition, 16 gene pairs of replications were categorized as WGD/segmental duplicates. The remaining gene replication mechanisms were detected as proximal or dispersed. As a result, tandem replication might primarily contribute to the expansion of the SOT gene family during the evolution of G. hirsutum. In order to understand the collinearity of the SOT gene family between G. hirsutum and two diploid cottons ancestors, we also identified these linked gene pairs (Fig. 3b). 56 collinear gene pairs were identified between G. hirsutum and G. arboretum, and 29 of them belonged to At subgroup in G. hirsutum. 48 collinear gene pairs were also found between G. hirsutum and G. raimondii, and 22 genes were Dt subgroup in G. hirsutum.

Fig. 3
figure 3

Duplication and synteny of SOT genes among G. hirsutum, G. arboretum and G. raimondii. a Localization and duplication of SOT genes on G. hirsutum chromosomes. Tandem duplication gene pairs were marked with red curve lines. b The synteny of SOT genes between G. hirsutum and two diploid cottons, G. arboretum and G. raimondii. Red lines connected the homologous genes between G. hirsutum and G. arboretum, blue lines connected the homologous genes between G. hirsutum and G. raimondii, respectively

Full size image

Phylogenetic analysis of SOT genes

From the phylogenetic tree constructed by all members of the SOT genes (Fig. 4), 170 of the 245 SOT genes were distributed in 4 subfamilies, and the remaining 75 were separated into two clades. The subfamilies VII and VI were the largest two subfamilies, containing 78 and 75 members, respectively. Subfamily V was the smallest one, including only five genes. The SOT genes from four cotton species were more closely related than the genes from Arabidopsis. In addition, at the end of the branch, there were many clades where three genes were clustered together. Generally speaking, of the three genes, two genes are from the At subgroup of tetraploids, one from G. arboretum; or two genes from the tetraploid Dt subgroup, one gene from G. raimondii. This was consistent with the fact that tetraploids came from two diploids [36]. However, after the formation of tetraploids, the relationship between the two tetraploids was closer than that between their ancestors.

Fig. 4
figure 4

Phylogenetic tree of 266 SOT genes from Gossypium and Arabidopsis. Neighbor-Joining tree was constructed by Mega X program with the full-length amino acid sequence of the SOT genes. The number of subfamily was marked according to the classification results of genes in Arabidopsis. Genes marked in red came from Arabidopsis. Genes marked in purple indicated that the clustered genes came from two tetraploids and one diploid

Full size image

Structural characterizations and conserved motif analyses of GhSOT genes

The gene structure of SOT genes was analyzed according to the gene annotation files and displayed in Fig. 5. Results showed that the exon numbers ranged from 1 to 6, with an average of 1.5. The great majority of genes contained less than 3 exons, and most contained only one exon. Classically, genes in the same evolutionary branch had similar structures, which shared a conserved gene structure pattern in terms of intron/exon number and intron/ exon length.

Fig. 5
figure 5

Conserved motif and gene structure of GhSOT genes. The phylogenetic tree was generated using protein sequences of 77 GhSOT genes. Intron/exon structure of SOT genes was analyzed by GSDS. Green boxes standed for exons; grey lines for introns. 20 conserved motifs were identified by MEME. Different color boxes with number represented different motifs

Full size image

20 conserved motifs of GhSOT genes were identified through the MEME program (Fig. 5 and Additional file 1: Table S4), with a width ranged from 11 to 50 amino acids. The number of conserved motifs in different genes varied from 2 to 14, however, in the same branch of the phylogenetic tree, the number and type of conserved motifs were similar. Motif 4 appeared in 66 genes and was common to almost all GhSOT genes, followed by Motif 5, 3, 10, 7, 1 (appearing in more than 60 genes). The gene structures and conserved motifs of the four genes on the same evolutionary clade, GH_D03G0217, GH_A02G1840, GH_D10G1036 and GH_A10G0926, were different from other genes, which may lead to changes in evolutionary speed and function.

RNA-Seq expression profile of GhSOT genes

Firstly, 21 GhSOT genes with expression levels less than 1 at 10 different stages were eliminated. The raw data of the remaining 56 GhSOT genes were normalized to log2FPKM and the heatmap of the expression was shown in Fig. 6. Most genes exhibited characteristics that were specifically expressed during the different stages. 16 genes were constitutively expressed in 10 tissues, especially the expression values of GH_A04G0111 and GH_D04G1212 were more than 1 at all stages. Most of the GhSOT genes exhibited a high level of expression in the stem, leaf, and the initial stage of fiber development (− 3, 0, 3 dpa ovule). This indicated that SOT genes might be associated with fiber development in cotton. As reported in previous study [37], a lot of loci related with fiber quality were clustered on chromosomes D11. In this study, there were two SOT genes located on chromosomes D11, and one of them was specifically expressed in several tissues. So, we further performed experiments to understand the characteristics and functions of GhSOT67 (GH_D11G2586).

Fig. 6
figure 6

Transcriptome expression of GhSOT genes in different tissues and developmental stages (FPKM> 1). Expression levels were illustrated by graded color scale, red indicated high FPKM value, blue indicated low abundance, while grey indicated none expression. Genes marked in red represented constitutively and highly expression in 10 tissues

Full size image

Tissue and subcellular localization analysis of GhSOT67

To investigate the tissue localization of GhSOT67, a recombinant vector of pGhSOT67::GUS was constructed and transformed into Arabidopsis mediated by Agrobacterium tumefaciens cells (GV3101). Multiple positive transformants were screened, soaked in the GUS staining solution and the most typical one was shown in Fig. 7a and b. The results showed that the staining in blue color was found in the stem and leaf of the transformant plant, which was consistent with the expression of the transcriptome expression of GhSOT67 (Fig. 7c). This expression pattern had also been reported in Arabidopsis [8].

Fig. 7
figure 7

Tissue and subcellular localization of GhSOT67. a Transformants were planted on half-strength MS medium containing 50 μg/mL kanamycin, the red circle showed the positive plants. b GUS staining analysis of the positive transgenic lines. c The FPKM value of GhSOT67 according to the transcriptome data. d Subcellular localization of RFP fusion proteins of GhSOT67 in infected tobacco leaves

Full size image

According to the online tool CELLO, GhSOT67 was predicted to be localized in the cytoplasm (Additional file 1: Table S1). To verify this, full-length CDS of GhSOT67 without initial condon was ligated with pBinRFP vector. The control empty vector pBinRFP was present all over the cell, including the nucleus, membrane and cytoplasm (Fig. 7d). By contrast, the GhSOT67::RFP fusion protein was mainly localized in cytoplasm, confirming the previously predicted result.

Virus-induced gene silencing (VIGS) of GhSOT67 in cotton

In order to investigate the relationship between GhSOT67 gene and fiber development, we performed VIGS on a cotton variety, J02. The empty vector pYL156 was used as a negative control. The recombinant vector pYL156:CLA1 could induce a leaf bleaching phenotype, therefore, it was served as a positive control to indicate the success of gene silencing.

17 days after the induction, the albino phenotype occurred on the positive control plants (Fig. 8a), proving that VIGS was successful. The expression of GhSOT67 after gene silencing was firstly verified by PCR compared with Histon3. Subsequently, the results of qRT-PCR revealed that the level of gene expression of most GhSOT67 silenced plants decreased by more than 80% (Fig. 8b). As shown in Fig. 8c, after 1 month of the treatment, the number of stem hairs in GhSOT67 silenced plant decreased evidently, comparing with the blank and negative control plants. In the meantime, the length of stem and leaf hairs of GhSOT67 silenced plants was obviously shorter than that of control plants (Fig. 8c and d). The stem and leaf hairs, as well as cotton seed fiber, were originated from the single cell layer, which might have similar fiber differentiation and development mechanisms [38,39,40]. Accordingly, the results suggested that GhSOT67 might be involved in the fiber development process.

Fig. 8
figure 8

Agrobacterium-mediated VIGS of GhSOT67 in cotton. a Phenotypes of gene silencing plants. The plants from left to right were CLA1-silenced, blank control, negative control and GhSOT67-silenced. b The expression levels of GhSOT67 in the negative control and silenced cotton plants conducted through qRT-PCR. N referred to the negtive control (Agrobacterial culture suspension of pYL156 only). c The stem hair of cotton after the induction. Left: blank control; Middle: negative control; Right: three lines of GhSOT67-silenced plant. d The leaf hairs of cotton observed under an optical microscopy after the induction. Upper: blank control under 10-fold visual field, the right one was the twice magnified leaf hair; Middle: negative control under 15-fold visual field, the right one was the double enlarged leaf hair; Lower: GhSOT67-silenced plant under 15-fold visual field, the right one was the fourfold enlarged leaf hair

Full size image

Discussion

In recent years, the nuclear genome sequences of G. arboreum, G. raimondii, G. hirsutum, G. barbadense and G. hirsutum have been published successively [41,42,43,44], further deepening the understanding of cotton genomics and genetics, which provides a possibility for exploring SOT gene family members and their phylogenetic relationships. Here, we identified a total of 245 SOT genes from four cotton species, according to the sequence identity of proteins. The number of GhSOT and GbSOT genes were more than that of SOT genes in two diploid cotton, possibly due to the polyploidization event occurred in two tetraploid cotton about 1.5 million years ago (Mya) [36].

Gene duplication is considered to be the main driver of evolution, leading to functional differentiation and diversification [45]. Gene duplication mainly includes three forms such as tandem, whole-genome and segmental duplications. In this study, we found that tandem replication might primarily contribute to the expansion of the GhSOT gene family, as well as several other replication methods exist. On the bases of the previous reports in Arabidopsis, SOTs had been divided into 8 groups [1, 9]. Phylogenetic analysis demonstrated that 245 SOT genes from Gossypium were cluster with SOTs from Arabidopsis into 4 clades, except for 75 SOTs from Gossypium. The convergence of three genes at the end of the evolutionary branch was consistent with previous studies that two diploids were the ancestors of tetraploids [36]. The difference in the number of exons and conserved motifs between genes indicated that the gain and loss of exons may lead to the functional diversity of SOT genes closely related to the evolution of SOT gene family.

To date, only a few Arabidopsis SOTs were functionally characterized. At5g07000 from group VI was proved to catalyze the sulfation of 12-hydroxyjasmonates, thus causing inactivation of jasmonic acid in plants [16]. For another Arabidopsis SOT, At3g45070 from group II, had been found to specifically bind to flavonols [1]. For the GhSOT gene members, we paid particular attention to those that might play crucial roles in plant growth or fiber development. Combining the transcriptome expression of GhSOT genes with the fiber-quality-related loci reported previously [37], GhSOT67 was selected to further understand its characteristics and functions. For the localization analysis, GhSOT67 was estimated to express in cytoplasm and locate in stem and leaf tissue. These features would be related to its function as a catalyst [8]. Transcriptome expression showed that GhSOT67 was specifically expressed in several tissues and the initial stage of fiber development (− 3, 0, 3 dpa ovule). In addition, GhSOT67-silenced plants treated by VIGS showed a shorter length of stem and leaf hairs than that of control plants. According to the results of phylogenetic cluster, GhSOT67 belonged to group VI, it might have similar function to At5g07000 that can catalyze the inactivation of jasmonic acid. So we speculated that when GhSOT67 was silenced, jasmonic acid could not be sulfated and accumulated in the plant, then the length of stem and leaf hairs was shortened. Taken together, these results suggest that GhSOT67 may involve in cotton fiber development. However, the detailed correlation between SOTs, jasmonic acid and fiber development remains to be further verified.

Conclusion

In this study, a comprehensive analysis including chromosomal location, collinearity and duplication, gene structure and expression patterns of the SOT gene family in Gossypium was first performed. To summarise, we isolated a total of 245 SOT genes in the genome of G. arboreum, G. raimondii, G. barbadense and G. hirsutum, and further classified the SOT genes into four groups based on the orthologous relationships comparing with Arabidopsis. Tandem replication primarily contributed to the expansion of SOT gene family in G. hirsutum. Expression profiles of the GhSOTs in various tissue and developmental stages implied that GhSOTs might be involved in the fiber development. In addition, gene silencing by VIGS significantly induced the expression of GhSOT67 and shortened the length of stem and leaf hairs. Taken together, these findings indicated that SOT genes might be associated with fiber development in cotton.

Methods

Database search and sequence retrieval

The genome files and protein sequences of two diploid cottons [41, 46] (G. arboreum L., G. raimondii Ulbr.) and two tetraploid cottons [44] (G. hirsutum L., G. barbadense L.) were downloaded from the Cotton Functional Genomics Database (CottonFGD) (https://cottonfgd.org/) [47]. The protein sequences of Arabidopsis thaliana (L.) were obtained from the Arabidopsis Information Resource (TAIR) (https://www.arabidopsis.org/). Based on the sequence similarity of the translated products, the Arabidopsis whole genome contains 21 genes encoding the SOT protein (AtSOT) [1] and all 21 Arabidopsis SOT proteins were extracted using TBtools (https://github.com/CJ-Chen/TBtools/releases) [48].

Two methods were used to search SOT genes in four cotton species. Firstly, 21 Arabidopsis SOT proteins were used as query sequences against the four cotton protein sequences files with default parameters (e-value <1e-5) through BLAST algorithm for Proteins (BLASTP) search. The candidate SOT genes of each cotton species were named separately, such as GhSOT from G. hirsutum and GbSOT from G. barbadens. Secondly, the hidden Markov model seed file (Stockholm format) of sulfotransferase domain (PF00685) were acquired from Pfam (http://pfam.xfam.org/) and used as a query sequence searching for candidate SOT protein sequences against the four cotton protein sequences files by Hmmer 3.0 (http://hmmer.org/), with default parameters. The SOT protein sequences with e-value less than 15 were preserved. Then, we merged all hits obtained above and discarded the repetitive sequences. All non-redundant protein sequences were further checked the conserved domains of the protein using the NCBI Conserved Domain Database (https://www.ncbi.nlm.nih.gov/cdd) in automatic mode (threshold = 0.01, maximum hits =500).

Finally, the candidate SOT genes were further manually confirmed to eliminate the pseudo sequences and the position in the cell was predicted according to the online tool CELLO v2.5 (http://cello.life.nctu.edu.tw/) [49]. The molecular weight (Mw) and isoelectric points (pI) of the candidate SOT genes were predicted using the online ExPASy server (http://web.expasy.org/compute_pi/) [50].

Chromosomal mapping and phylogenetic analysis

Chromosomal position and gene structure information of SOT genes were obtained from four cotton gene annotation files, and these SOT genes were mapped separately on the corresponding chromosomes using the MapChart software (https://www.wur.nl/en/show/Mapchart/).

The full-length amino acid sequence of the SOT genes from both Arabidopsis and Gossypium were saved as a fasta format file and used to perform multiple sequence alignments using the ClustalW program with the default settings. Subsequently, we constructed the neighbor-joining (NJ) tree in MEGA X, the parameters were set as follows: 1000 bootstrap replicates, Jones-Taylor-Thornton (JTT) substitution model, and partial gap deletion mode with a cut-off value of 80%.

Intron/exon distribution and conserved motif analysis

The gene structure of SOT genes was analyzed using Gene Structure Display Server 2.0 (GSDS, http://gsds.cbi.pku.edu.cn/) [51]. The conserved domain motifs of the SOTs were determined by Multiple Em for Motif Elicitation (MEME) (http://meme-suite.org/tools/meme) [52] according to the following parameters: site distribution was set at 0 or 1 occurrence per sequence, the width of motifs ranged from 6 to 50, the maximum number of motifs was 20. All the characteristic results of SOT genes were visualized and integrated into graphics by Tbtools.

Gene expression analysis

The fragments per kilobase of exon per million fragments mapped (FPKM) values were acquired from the transcriptome data of G. hirsutum cv. TM-1 [53]. The expression values of three different tissues and seven different stages of fiber development, − 3 dpa (day post anthesis) ovule, 0 dpa ovule, 3 dpa ovule as well as 5, 10, 20, and 25 dpa fibers, were considered and the genes with FPKM values more than 1 at least one stage were further analyzed. The expression of the SOT gene was estimated to be normalized in the form of log2FPKM and displayed in the heat map.

Plant materials

A cotton variety, J02, was provided by Germplasm Repository of Institute of Cotton Research, Chinese Academy of Agricultural Sciences (CRI of CAAS, Anyang, Henan province, China) only for scientific research purpose. J02 was sown in mixed soil (vermiculite:humus = 1:1) and cultured in an incubator with a 16 h /8 h (light/ dark) photoperiod at 28 °C and 25 °C respectively till the cotyledons were fully unfolded.

Arabidopsis thaliana ecotype Colombia (Col-0) and tobacco (Nicotiana benthamiana) were also provided by CRI of CAAS and grown as recipient materials in the following ways. The seeds were grown on agar-solidified Murashige and Skoog (MS) medium by dropper, and after 48 h of hypothermia, the culture dishes were placed in an incubator with a 16 h / 8 h (light / dark) photoperiod at 24 °C and 22 °C respectively. When the cotyledons were unfolded, the seedlings were transplanted into sterile mixed soil (vermiculite:humus = 1:1).

Construction of target gene vectors and their inoculation treatment

In order to perform the tissue location of GhSOT67, 1500 bp promoter sequence upstream of the gene was amplified and inserted into the two restriction sites (HindIII and BamHI) of pBI121 vector. The Agrobacterium tumefaciens cells (GV3101) containing constructed vector was transformed into Arabidopsis plants according to the floral dip method [54]. The wildtype and transgenic plants were grown under conditions mentioned above. Positive transformants were screened by planting on half-strength MS medium containing 50 μg/mL kanamycin and confirmed by PCR and β-glucuronidase (GUS) staining.

The CDS of GhSOT67 without initial codon was inserted into the SalIrestriction site of the pBinRFP vector [55] to construct the translational RFP fusion constructs. The recombinant plasmid was transformed into Agrobacterium tumefaciens strain LBA4404 and inoculated into the second or third leaves on top of the tobacco according to the protocols [56]. The vector of pBinRFP (RFP alone) was also transformed into the tobacco leaves which was planted at the same time and in the same condition as the control. Finally, the infected tobacco leaves were wrapped in tinfoil, placed in a dark environment for 24–48 h and observed under an optical microscopy with CCD camera (Leica Microsystems, Germany) [57].

For the virus-induced gene silencing (VIGS) experiment, an specific 300-bp sequence selected from the GhSOT67 was amplified with two restriction sites at both ends (SpeI and AscI). Firstly, the PCR amplification product was cloned into pMD19 T vector. Both the resultant construct and pYL156 were digested with SpeI and AscI, and connected through ligation buffer solutionI to form pYL156:GhSOT67. The plasmid was transformed into Agrobacterium tumefaciens LBA4404 for infecting cotton. Agrobacterial culture suspension of pYL192 was respectively mixed with others equally as an auxiliary carrier. Agrobacterial culture suspension of pYL156 (negtive control), pYL156:CLA1 (positive control) and pYL156:GhSOT were separately injected into fully expanded cotyledons of cotton variety, J02, before the true leaves hadn’t yet emerged. Ten strains of J02 were reserved for wild type (blank control), 10 strains were injected with pYL156 and pYL156:CLA1 respectively, and 45 strains were injected with pYL156:GhSOT67. Experimental procedures and methods of operation were used as described by ref. [58].

Collections, RNA isolation and qRT-PCR analysis

About 2 weeks post infiltration, when true leaves appeared albino phenotype, the leaves of the J02 were put into the liquid nitrogen immediately and stored at − 80 °C for RNA isolation and analysis. Total RNA was extracted via the RNA extraction kit (TIANGEN, Beijing, China). First-strand cDNA was synthesized using PrimeScript™ RT reagent Kit with gDNA Eraser (TaKaRa, Japan). The quantitative real-time (qRT)-PCR analysis was completed on 7500 Fast Real-Time PCR system (Applied Biosystems, Inc., California USA) with SYBR Premix Ex Taq (TaKaRa, Japan). The Histon3 gene were used as an endogenous control to normalize gene expression. The relative expression levels of GhSOT67 gene after infiltration was calculated using the 2-ΔΔCT method [59].

All the gene-specific primers used for amplifications or vector constructions were listed in Additional file 1: Table S5.