Background

Cotton is an important cash crop in many develo** countries and frequently grown in dry lands or on supplementary irrigation [1], because agricultural water consumption can no longer be expanded thanks to water competition among domestic, industrial and agricultural users [2]. The quantity and quality of fiber produced by cotton plants are directly related to water available to them during their developmental stages. When suffered from water deficits, especially during the period of flowering and fructification, cotton would show significant yield loss, sometimes up to 50% reduction compared to those that have been irrigated [3, 4].

In the long-term evolutionary history, plants have formed a complex gene-metabolic network to accommodate a variety of environmental changes. As an important metabolite, lignin plays vital roles in defense against biotic and abiotic stresses [5,6,7]. Lignin is synthesized through the phenylpropane pathway. 4-coumarate-CoA ligases (4CL, EC 6.2.1.12) is the main branch point enzyme of the phenylpropanoid pathway, which catalyzes cinnamic acid to generate corresponding CoA thioesters [8]. Products of 4CL subsequently serve as substrates of various oxygenases, reductases and transferases for biosynthesis of lignin, flavonoids, anthocyanins, aurones, stilbenes, coumarins, suberin, cutin, sporopollenin, and others [9]. The 4CL gene family has been characterized in many plants, such as Arabidopsis [10], rice [11] and aspen [12]. Genes of the 4CL family in dicots can be classified into two distinct groups, type I and type II [8]. Type I genes are mainly involved in lignin biosynthesis whereas type II genes are involved in biosynthesis of phenylpropanoids other than lignin. Some additional genes containing the same conserved motifs of 4CLs and showing high similarity with the 4CL proteins are classified as 4CL-like genes [13].

Studies have shown that 4CL genes play momentous roles in plants, such as regulation of growth and development, protection against biotic and abiotic stresses [11, 14, 15]. In Arabidopsis, At4CL1, At4CL2, and At4CL4 were found to be involved in lignin formation, the 4 cl1 4 cl2 double and 4 cl1 4 cl2 4 cl3 triple mutant plants exhibited a dwarf and bushy phenotype [10]. In rice, Os4CL2 was specifically expressed in anthers and induced by UV irradiation [16]. Plagiochasma appendiculatum thallus plants showed downregulation of Pa4CL1 when treated with abscisic acid (ABA), and showed upregulation of Pa4CL1 when treated with salicylic acid and MeJA [17]. In both poplars and Arabidopsis, the expression levels of 4CL genes were induced by salt stress and wounding [7]. The 4CL-like genes may also play a role in response to abiotic stresses [18, 19]. Overexpression of Fm4CL-like1 in tobacco increased drought tolerance due to increasing lignin accumulation and the activities of antioxidant enzymes, and upregulating the expression levels of stress-related genes [19]. Nevertheless, our knowledge of the 4CL gene family in cotton is very limited.

To gain insights into the cotton 4CL gene family and its role in abiotic stress tolerance, in this study, we did genome-wide identification of 4CL genes in G. hirsutum and analysed their expression changes in response to various abiotic stresses based on publically available RNA-seq datasets. We identified 34 putative Gh4CL genes in G. hirsutum and 26 of them were found to be induced by at least one stress, including Gh4CL7 that was up-regulated under polyethylene glycol (PEG) osmotic stress. We further investigated the function of Gh4CL7 in drought tolerance by silencing its expression in cotton using virus-induced gene silencing (VIGS) and generating transgenic Arabidopsis plants overexpressing Gh4CL7. Our results indicated that Gh4CL7 functions positively in response to drought stress and is a potential candidate gene for improving drought resistance of cotton by genetic engineering.

Results

Genome-wide identification and bioinformatics analysis of Gh4CL genes

Using the approach described in Materials and Methods, we identified 34 Gh4CL genes in G. hirsutum. They are randomly distributed on 22 chromosomes and an unanchored scaffold that was not assigned to a particular chromosome (Fig. 1a, Table 1). We named them Gh4CL1 to Gh4CL34 based on their chromosomal location. Two pairs of Gh4CL genes, Gh4CL10/11 and Gh4CL21/22, are in tandem configuration on chromosomes A09 and D03, respectively. Segmental duplication could be involved in generation of 12 Gh4CLs based on MCScanX analysis. Phylogenetic analysis using the 34 Gh4CL genes and 4CL genes from A. thaliana, G. max, P. tremuloides, P. trichocarpa, R. idaeus, and I. tinctoria showed that they were clustered into three groups (Fig. 1b). The Ka/Ks ratio of each homologous/paralogous Gh4CL pair is < 1 (Additional file 2: Table S1), suggesting that these Gh4CL genes have experienced purifying selective pressure during their evolution history to eliminate deleterious mutations.

Fig. 1
figure 1

Phylogenetic, chromosomal distribution and interchromosomal relationships of the Gh4CL genes. a Chromosomal distribution of Gh4CL genes. The gray lines indicate all synteny blocks in the genome of G. hirsutum and the red lines indicate interchromosomal relationships of Gh4CLs. The first circle represents individual chromosomes length. The second circle represents the gene density of each chromosome by color-coded short bars. The chromosome number were shown inside the circle. b The neighbor-joining phylogenetic tree was generated using 4CL protein sequences from G. hirsutum and six other plants

Full size image
Table 1 Characteristics of the 34 Gh4CL genes in G. hirsutum
Full size table

The protein length of Gh4CLs is between 129 and 576 amino acids (aa) with ORF from 390 to 1731 bp, molecular weight from 14.11 to 63.08 KD, and pI from 5.3 to 9.77. Most Gh4CLs seem to be associated with various biomembranes based on subcellular localization prediction (Table 1). Analyses of gene structures and motifs showed that each Gh4CL has multiple exons, introns and motifs (Additional file 1: Figure S1; Additional file 2: Table S2). All the Gh4CL proteins have two structural domains, a putative AMP-binding domain “SSGTTGLPKG” (Box I) and a conserved domain “GEICIRG” (Box II) (Additional file 1: Figure S2).

Cis-elements in combination with transcription factors regulate the transcription level of a gene. To identify potential cis-elements involved in regulation of transcription of Gh4CL genes, we scanned cis-elements in the promoter region (2 kb upstream of ATG) of each Gh4CL gene using the online tool PlantCARE [20]. Many Gh4CL genes harbored plant hormone-responsive and/or stress-responsive elements, including ABA responsive elements (ABREs), auxin responsive elements (AuxRR-core, TGA-elements and TGA-box), MeJA-responsive elements (CGTCA-motif, TGACG-motif), gibberellin-responsive elements (TATC-box, GARE-motif and P-box), salicylic acid responsive elements (TCA-elements), low-temperature responsive elements (LTR), defense and stress responsiveness elements (TC-rich repeats) and drought-responsive elements (MBS) (Additional file 1: Figure S3).

Tissue specific expression patterns of Gh4CL genes

The expression patterns in various tissues provide clue for the possible biological functions of genes of interest. We thus analysed the transcript abundance of the Gh4CL genes in different tissues (root, stem, leaves, flower, ovule and fibers at 5, 10, 15 and 20 days-post-anthesis (DPA)) under normal growth conditions using the publically available RNA-seq data (BioProject Accession: PRJNA248163) [21]. We found that 10 Gh4CL genes were expressed in all the tested tissues [base on fragments per kilobase of transcript per million mapped reads (FPKM) ≥ 1], and 4 genes (Gh4CL3, Gh4CL5, Gh4CL18 and Gh4CL27) showed weak or no expression in all tissues analysed (Fig. 2). In addition, 8 Gh4CL genes (Gh4CL2, Gh4CL4, Gh4CL8, Gh4CL12, Gh4CL17, Gh4CL24, Gh4CL29 and Gh4CL30) were highly expressed (FPKM ≥20) in stem, with the highest expression level observed for Gh4CL17 (FPKM ≥84) and 6 genes (Gh4CL7–8, Gh4CL12, Gh4CL20 and Gh4CL30–31) were strongly expressed in leaves, with the highest expression level observed for Gh4CL20 (FPKM ≥202).

Fig. 2
figure 2

Expression pattern of the Gh4CL genes in different tissues of G. hirsutum based on the RNA-seq data. The heatmap was generated by using the pheatmap package

Full size image

Expression analysis of Gh4CL genes under different abiotic stress conditions

Since 4CL genes are capable of responding to biotic and abiotic stresses in various plant species, we further investigated the transcript abundance of the Gh4CL genes under cold, heat, PEG and salt stresses using the transcriptomic data of G. hirsutum (BioProject Accession: PRJNA248163) [21]. We found that 26 Gh4CL genes were induced significantly by one or more stresses, and the remaining 8 Gh4CL genes (Gh4CL3, Gh4CL5, Gh4CL10, Gh4CL18–19, Gh4CL23, Gh4CL28 and Gh4CL34) were not induced by either of the four stresses (Fig. 3a). Comparing the four stress conditions, more Gh4CL genes showed altered expression in response to salinity, cold and heat stresses than to PEG stress. Notably, ten Gh4CL genes (Gh4CL2, Gh4CL7–9, Gh4CL11–13, Gh4CL17, Gh4CL22, Gh4CL25 and Gh4CL31) showed increased expression (treatment FPKM/control FPKM ≥1.5) in response to PEG stress over the 3 h to 12 h time period. To verify these results, we investigated the expression patterns of the selected Gh4CL genes under the simulated drought treatment using quantitative real-time polymerase chain reaction (qRT-PCR). As shown in Fig. 3b, the expression levels of Gh4CL7–8, Gh4CL12–13, Gh4CL17, Gh4CL22 and Gh4CL24 were up-regulated in cotton leaves over the time period of 3 h to 24 h after PEG stress, consistent with the RNA-seq based results.

Fig. 3
figure 3

Expression profile of the Gh4CL genes in response to different abiotic stresses. a Heatmap showing the relative expression levels of each Gh4CL gene in each treatment showing at the bottom of the image. b The relative expression level of the selected Gh4CL genes under 20% PEG stress based on qRT-PCR. The GhUBQ7 gene was used as internal control. Error bars represents the standard deviation calculated from three biological replicates

Full size image

Gh4CL7 plays an important role in lignin biosynthesis

Based on the above analysis results of promoter cis-elements and expression patterns under drought stress, three Gh4CL genes, including Gh4CL7, Gh4CL8 and Gh4CL13, were considered as candidate genes with a potential role in the regulation of drought stress response in cotton. In this study, we selected Gh4CL7 for further functional analysis by silencing its expression in cotton and overexpression in Arabidopsis thaliana.

We used VIGS to silence the expression of Gh4CL7 using the TRV vector (TRV:Gh4CL7; Additional file 1: Figure S4). TRV:GhCHLI was used as a positive control of the VIGS experiment (Additional file 1: Figure S5). Arabidopsis thaliana plants overexpressing Gh4CL7 (Gh4CL7-OE) were obtained by the floral dip method. Gh4CL7 belongs to class I whose genes have been shown to regulate lignin biosynthesis [10, 22]. We thus first investigated whether or not Gh4CL7 is also involved in lignin biosynthesis by comparing the lignin contents of the Gh4CL7-OE Arabidopsis lines and TRV:Gh4CL7 cotton plants with that of their corresponding control plants. The lignin content increased by approximately 10% in the Gh4CL7-OE lines compared with wild-type (WT) plants (Fig. 4a), while decreased by approximately 20% in the TRV:Gh4CL7 plants compared with the TRV:00 plants (Fig. 4b). Additionally, the stem of the TRV:Gh4CL7 plants were sectioned and stained with phloroglucinol-HCl to detect the presence of lignin (Fig. 4c). We found that the stem section of the TRV:Gh4CL7 plants with reduced lignin content exhibited light red color, but the TRV:00 plants displayed typically purple-red color after phloroglucinol-HCl staining. These results suggested that Gh4CL7 is related to lignin synthesis. We also analysed the expression level of the phenylpropane pathway genes that are related to lignin biosynthesis, including GhPAL, GhCCoAOMT1, GhCOMT1, GhCOMT2, GhCOMT3, GhCCR1, GhCCR2, and GhCAD. The relative expression level of these genes were lower in the TRV:Gh4CL7 plants than in TRV:00 (Fig. 4d), indicating that Gh4CL7 could affect the accumulation of lignin by regulating the transcription level of these downstream genes of the lignin biosynthesis pathway.

Fig. 4
figure 4

Analyses of lignin contents and the expression level of genes related to lignin biosynthesis. a-b Comparison of lignin contents in Gh4CL7-OE Arabidopsis plants (a) and in Gh4CL7-silencing cotton plants (b). c Stem sections from the TRV:Gh4CL7 and TRV:00 cotton plants stained with phloroglucinol-HCl. d The relative expression levels of genes related to lignin biosynthesis. Data were represented as the mean ± SE of three biological replicates; asterisks indicate levels of significance based on t-test (* P < 0.05, ** P < 0.01)

Full size image

Silencing of Gh4CL7 compromises tolerance of cotton to drought stress

Phenotypic difference between the TRV:Gh4CL7 and TRV:00 plants was observed after 20 days of water deficiency treatment. Compared to the TRV:00 plants, the TRV:Gh4CL7 plants displayed severe wilting and yellowing leaves (Fig. 5a), consistent with a lower leaf relative water content (RWC) (Fig. 5b) and a decrease chlorophyll contents (Fig. 5c). Besides, it was also found that the size and the ratio of width to length of stomata significantly increased in the TRV:Gh4CL7 plants (Fig. 5d-f), which might accelerate the transpiration rate under drought conditions, consistent with the observed higher water loss relative (WLR) (Fig. 5g). The hydrogen peroxide (H2O2) content and malondialdehyde (MDA) level were measured to reflect the cell damage or injury in TRV:Gh4CL7 and TRV:00 plants. During drought stress, the TRV:Gh4CL7 plants accumulated more MDA (Fig. 5h) and H2O2 (Fig. 5i) compared to the TRV:00 plants. The activities of superoxide dismutase (SOD), peroxidase (POD) and catalase (CAT) in the TRV:Gh4CL7 and TRV:00 plants were also measured to explore the function of Gh4CL7 in the modulation of antioxidant enzymes (Fig. 5j). As expected, under drought stress conditions, the TRV:Gh4CL7 plants displayed a significantly reduced activity of SOD, POD and CAT as compared to the TRV:00 plants. Additionally, six stress-related genes (GhABI4, GhABF4, GhLEA14, GhRD22, GhRD29 and GhNCED1) were down-regulated in the TRV:Gh4CL7 plants after drought treatment (Additional file 1: Figure S6). These results suggested that silencing of Gh4CL7 decreases tolerance of cotton to drought stress.

Fig. 5
figure 5

Drought tolerance analysis of the Gh4CL7-silencing cotton plants. a Representative phenotypes of the TRV:00 (CK) and TRV:Gh4CL7 (VIGS) plants after 3 weeks of drought treatment. b-c Leaf RWC and chlorophyll contents of CK and VIGS plants under drought stress. d-e Comparison of stomata in the CK and VIGS plants. f Water loss rate of detached leaves from the CK and VIGS plants. g-h Comparison of H2O2 and MDA contents in the CK and VIGS plants. i Activities of antioxidative enzymes in the CK and VIGS plants. Mock, normal growth conditions; drought, 3 weeks of water deficit conditions. Statistical significance was determined by the student test. * and ** represent significant at P < 0.05 and P < 0.01, respectively

Full size image

Overexpression of Gh4CL7 in Arabidopsis enhances drought tolerance

We further investigated the function of Gh4CL7 in response to drought stress using Arabidopsis plants overexpressing Gh4CL7. Three independent Gh4CL7-OE lines that showed an elevated level of Gh4CL7 (Fig. 6a) were selected for phenoty** under the drought stress conditions. Compared to the WT plants, the three Gh4CL7-OE lines had a decreased germination rate (Fig. 6b), but showed a significantly increased root length under the mannitol stress conditions (Fig. 6c, d). Three-weeks-old seedlings of Gh4CL7-OE and WT were used for water deficiency treatment. No obvious phenotypic difference was observed between Gh4CL7-OE and WT by the mock treatment. However, the Gh4CL7-OE plants showed much less damage than WT after 10 days of water deficiency (Fig. 6e). Under drought stress conditions, the H2O2 content and MDA level in the Gh4CL7-OE plants were relatively lower than that in WT (Fig. 6f-g), but the SOD, POD and CAT activities were significantly higher (Fig. 6h). Additionally, the size and the ratio of width to length of stomata significantly decreased in the Gh4CL7-OE Arabidopsis plants (Fig. 7a-b), consistent with a lower WLR observed in those plants (Fig. 7c). To further elucidate the possible mechanism of Gh4CL7 in response to drought stress, the transcript levels of four known ABA-responsive genes (AtRD29B, AtRD22, AtABI4, AtCOR15A) and two ABA-biosynthesis genes (AtNCED3 and AtNCED5) were analyzed in the Gh4CL7-OE lines and WT plants after drought stress treatment. The qRT-PCR data showed that the expression levels of these genes were induced in Gh4CL7-OE, but not or just slightly induced in WT by drought stress (Additional file 1: Figure S7). These results indicated that overexpression of Gh4CL7 could enhance the tolerance of transgenic Arabidopsis plants to drought stress.

Fig. 6
figure 6

Drought tolerance analysis of the Gh4CL7-overexpressing Arabidopsis plants. a The relative expression level of Gh4CL7 in three independent Gh4CL7-OE Arabidopsis plants. b Germination rate of the Gh4CL7-OE Arabidopsis seeds on 1/2 MS supplemented with 0 and 300 mM mannitol. c-d Root elongation of the 6-days-old Gh4CL7-OE Arabidopsis seedlings on 1/2 MS supplemented with 0, 200, and 300 mM mannitol. Bar = 1 cm. e Phenotypic assay of the Gh4CL7-OE and WT Arabidopsis plants during drought treatment. f-h The contents of H2O2 and MDA and antioxidative enzyme activity in Gh4CL7-OE and WT plants under normal and water deficit conditions. Ten-days-old seedlings were transplanted to soil and regularly watered for 2 weeks. For drought treatment, irrigation was terminated for 2 weeks. For the rehydration treatment, the plants were re-watered 3 days after the drought treatment

Full size image
Fig. 7
figure 7

Stomatal opening in the WT and Gh4CL7-OE Arabidopsis plants under drought stress. a-b Comparison of stomata in the WT and Gh4CL7-OE plants after 14 days of water-withholding stress. c WLR of detached leaves from WT and Gh4CL7-OE Arabidopsis plants. Data were represented as the mean ± SE of three biological replicates. Statistical significance was determined by the student test. * and ** represent significant at P < 0.05 and P < 0.01, respectively

Full size image

Discussion

The 4CL gene family has been characterized in several plants, including Arabidopsis thaliana, Populus trichocarpa, Oryza sativa and Glycine max [11, 23,24,25]. Genes of this family have been reported to function not only in plant growth and development [16, 22, 26], but also in response to biotic and abiotic stresses [7, 27]. However, no comprehensive analysis of the 4CL genes has been documented in G. hirsutum. In this study, we did genome-wide identification of 4CL genes in G. hirsutum and investigated their expression profiles in various tissues and under different stress conditions with an aim to identify 4CL gene(s) with a potential role in stress tolerance. In total, 34 Gh4CL members were identified in the G. hirsutum genome (Table 1). In other plant species, such as A. thaliana, 4CL genes were divided into three classes, i.e. class I, class II and 4CL-like [13]. The 34 Gh4CL genes could also be clustered into three classes. We named the 4CL-like class as class III, which contains the largest number of Gh4CL genes (in total 25) together with At4CL6–9, At4CL11, and Ii4CL1 (Fig. 1b). The class III Gh4CLs cannot catalyze any of the hydroxycinnamic acid substrates into the corresponding CoA esters, their function is different from that of class I (related to the lignin biosynthesis) and class II (related to the biosynthesis of flavonoids) 4CL genes [13, 44], as well as many other factors related to drought responses, including stomata closure and stress-gene regulation [45, 46]. Under drought stress conditions, the closed stomata can decrease transpiration rate that helps plants to resist adverse environmental conditions. Our results showed that the size of stomatas was bigger in the TRV:Gh4CL7 plants, suggesting a positive role of Gh4CL7 in reducing transpiration rate that allows cotton to maintain a more favourable water balance, and effectively improves drought tolerance. This was supported by the observation of a higher WLR in leaves of the TRV:Gh4CL7 plants than those of WT plants.

Lignin is the second largest polymer in plants after cellulose [47]. It provides mechanical support to plants by increasing cell wall hardness and enhancing compressive strength of cells [48, 49]. We found that repressing the expression level of Gh4CL7 in G. hirsutum reduced the lignin content and led to a reduction in drought resistance, consistent with the result of rice plants with a decreased lignin content being more prone to drought stress [50]. Studies in Fraxinus mandshurica also showed that decreased lignin content resulted in drought resistance reduction [19]. The hydrophobicity of lignin is thought to have an inhibitory effect on the transpiration of plant tissue under drought conditions [51], that could be the reason for Arabidopsis plants overexpressing Gh4CL7 with an increased level of lignin content being more resistant to drought.

Conclusions

The findings of this study demonstrate that the Gh4CL7-silencing cotton plants had an increased sensitivity of drought stress while overexpressing Gh4CL7 enhanced tolerance of drought stress in Arabidopsis. Gh4CL7 conferred tolerance to drought stress by increasing lignin content, improving the antioxidant system, closing stomata, and up-regulating the transcription levels of ABA-responsive genes. Although the exact mechanism of Gh4CL7-mediated drought tolerance is still yet to be uncovered, our results provide evidence for the role of Gh4CL7 in combating drought stress.

Methods

Identification of the 4CL family genes in Gossypium hirsutum

The annotated protein sequences of G. hirsutum [21] were downloaded from CottonGen (https://www.cottongen.org/). The hidden Markov model file corresponding to the AMP-binding domain (PF00501) was downloaded from the Pfam protein family database (http://pfam.xfam.org/) and used as query (P < 0.001) [52] to search for the 4CL genes in G. hirsutum with HMMER 3.0 [53]. The existence of the AMP-binding domain sequences was examined using the Pfam, SMART (http://smart.embl-heidelberg.de/), and National Center for Biotechnology Information (NCBI) Conserved Domains (http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi) databases [54, 55].

Gene structure, conserved motif and promoter analyses

The length, molecular weight (MW), and isoelectric point (pI) of the identified Gh4CL proteins were calculated using the ExPasy website tools (http://web.expasy.org/protparam/) [56]. PSORT software (https://psort.hgc.jp/) was used for predicting subcellular localization [57]. Gene Structure Display Server 2.0 (GSDS, http://gsds.cbi.pku.edu.cn/) was used for intron and exon analysis [58]. The conserved motifs in the Gh4CL protein sequences were identified using the Multiple Expectation Maximization for Motif Elicitation (MEME) program (version 5.0.5, http://meme-suite.org/tools/meme) [59]. The potential cis-elements in the promoter sequences (up to 2000-bp upstream ATG) of Gh4CL genes were identified using the PlantCARE program (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/).

Phylogenetic tree, chromosomal distribution and syntenic relationship analyses

The multiple sequence alignment of Gh4CLs was done by Clustal X [60] and DNAMAN (version 5.2.2). The 4CL homologous protein sequences of Arabidopsis thaliana (At4CL1: OAP14948; At4CL2: OAP07084; At4CL3: AEE34324; At4CL4: AY376731; At4CL5: AY250839; At4CL7: AY376733; At4CL9: AF360250 At4CL11: AY376735), Glycine max (Gm4CL1: AF279267; Gm4CL2: AF002259; Gm4CL3: AF002258; Gm4CL4: X69955), Rubus idaeus (Ri4CL1: AF239687; Ri4CL2: AF239686; Ri4CL3: AF239685), Populus tremuloides (Pt4CL1: U12012; Pt4CL2: U12013), and Isatis tinctoria (Ii4CL1: ADG46006; Ii4CL2: KC430622; Ii4CL3: KC430623) were downloaded from the NCBI (http://www.ncbi.nlm.nih.gov/, accessed on 7 May 2018) and used for the phylogenetic tree analysis by using the neighbor joining method (NJ) in MEGA 6.0 [61] with 1000 repetitions for the bootstrap test.

All the Gh4CL genes were mapped to G. hirsutum chromosomes, based on their physical location information, using TBtools [73].

RNA extraction and quantitative real-time PCR analysis

To investigate gene expression patterns, total RNA was extracted from leaves of G. hirsutum and Arabidopsis with the EASYspin Plus plant RNA kit (Aidlab, Bei**g, China). RNA was reverse transcribed into cDNA using the M-mlv reverse transcript system (TAKARA, Da Lian, China). The qRT-PCR was performed using the Power SYBR Green PCR Master Mixture (Roche, Rotkreuz, Switzerland) on a Light Cycler® 480 II system (Roche, Rotkreuz, Switzerland) under the following conditions: initial pre-incubation at 95 °C for 5 min, followed by 40 cycles at 94 °C for 10 s, 59 °C for 10 s, and 72 °C for 10 s. The relative expression level of genes was analyzed by the 2-△△Ct method. The results were presented as the mean of three biological replications. The G. hirsutum histone3 gene and Arabidopsis EF-lα gene were used as the reference genes. All the primers used in this study were designed using the NCBI primer designing tool (https://www.ncbi.nlm.nih.gov/tools/primer-blast/, accessed on 27 August 2018) and listed in Additional file 2: Table S3.

Statistical analyses

Statistical analyses and data plotting were performed using SPSS and Graphpad Prism 5, respectively. ** and * represent significant differences at P < 0.01 and P < 0.05, respectively.