Introduction

Carotenoids are compounds that typically composed of several isoprene units and had multiple conjugated double bonds. They are ubiquitous in plants, fungi, bacteria, and animals and performed important biological functions. Carotenoid cleavage oxygenase (CCO) are vital for the metabolism of carotenoids, which specifically catalyzes the cleavage of carotenoid to produce various types of cleavage products. The products known as apocarotenoids and their derivatives possess essential biological functions in plant development and physiological response (Akemi et al. 2009; Walter and Strack 2011; Kloer et al. 2005). Based on whether the substrates are forming epoxy structure, CCO can be further classified into nine-cis-epoxycarotenoid dioxygenases (NCEDs) and carotenoid cleavage dioxygenases (CCDs) subfamily (Giuliano et al. 2003).

The vp14 gene isolated from maize was the first identified CCO gene (Schwartz et al. 1997). Subsequently, Tan et al. (2003) identified a total of nine homologs of vp14 in Arabidopsis thaliana. Four of them encoded CCDs (CCD1, CCD4, CCD7, and CCD8), while five encoded NCEDs (NCED2, NCED3, NCED5, NCED6, and NCED9) (Auldridge et al. 2006a). The orthologs of CCO identified in other plants were then named typically according to orthologs in Arabidopsis (Kim et al. 2016; Wang et al. 2013). Recently, CCD-like, a new subgroup of CCDs, was identified in tomato, strawberry, apple, tobacco, and sugarcane (Chen et al. 2018; Su et al. 2021; Wang et al. 2017; Wei et al. 2016; Zhou et al. 2019). To date, CCO gene family had been comprehensively analyzed in several species (Chen et al. 2018; Kim et al. 2016; Su et al. 2021; Tan et al. 2003; Vallabhaneni et al. 2010; Wang et al. 2013; Wei et al. 2016; Zhao et al. 2021; Zhou et al. 2019).

CCDs cleave a variety of trans-carotenoid substrates (Bouvier et al. 2003; Sun et al. 2008). CCD1 and CCD4 mainly contributed to the cleavage of apocarotenoids to generate compounds related to scent of flowers and fruits (Adami et al. 2013; Rubio et al. 2008; Ilg et al. 2009; Simkin et al. 2004; Song et al. 2016). Wang et al. (2013) reported that CCD1 and CCD4 genes in soybean showed strong responses to abiotic stress. CCD7 and CCD8 take part in metabolism of strigolactone which is involved in axillary shoot growth as well as drought and salt response in plants (Alder et al. 2012; Ha et al. 2014; Kim et al. 2016; Kloer and Schulz 2006; Seo and Koshiba 2002; Umehara et al. 2008). So far, little attention had been paid on specific biological functions of CCD-like genes.

NCEDs are key enzymes that catalyze the rate-limiting step in the biosynthesis of abscisic acid (ABA) (Gavassi et al. 2021; Schwartz et al. 2004; Vishwakarma et al. 2017). Wang et al. (2021) revealed that both PpNCED1 and PpNCED5 genes cooperatively regulate ABA biosynthesis in peach fruits. AcNCED1 might serve a major function in the early fruits softening of kiwifruit (Gan et al. 2020). AtNCED3, an ortholog of the best-characterized maize VP14, played a prominent role in drought tolerance (Tan et al. 2003; Iuchi et al. 2000). Overexpression of TaNCED increased drought resistance and delayed seed germination in Arabidopsis (Tong et al. 2017). NCED3 and NCED5 of rice have taken part in abiotic stress response by regulating ABA content (Huang et al. 2018, 2019). AtNCED6 and AtNCED9 were involved in seed development by regulating ABA biosynthesis (Lefebvre et al. 2006). Overall, NCED genes play an essential role in several biological processes in plants, including seed dormancy, plant growth and reproduction development, and abiotic stress response (Chernys and Zeevaart 2000; Leng et al. 2014).

Cotton, as a kind of important industrial crops for many countries around the world, serves natural fiber for the textile industry. Abiotic stress imposes a significant negative impact on yield and quality of cotton fiber. Genetic engineering provided an effective way for cotton genetic improvement on stress resistance. However, information on cotton CCO genes is still unknown and identification of stress responsive CCO genes has received little attention. In this study, characterization of CCO genes from three cotton species was conducted. Phylogenetic relationship, gene structure, putative cis-elements, collinearity relationship, and gene expression patterns under multiple stress conditions were further comprehensively performed. GaNCED3a, the ortholog gene of multiple stress-responsive gene GhNCED3a_A/D, was selected to perform functional analysis under drought and salt stress by VIGS in G. arboreum. Results of GaNCED3a silencing showed evidence that NCED3 in cotton not only played a role in drought stress but also taken part in salt stress. The results shed light on deep understanding of characteristics of cotton CCO genes and taking advantage of them in cotton stress-responsive genetic improvement.

Materials and methods

Identification of cotton CCO genes

The genomic data of the diploid cotton Gossypium raimondii (D5), Gossypium arboreum (A2), and tetraploid cotton Gossypium hirsutum L. acc. TM-1 (AD1) were downloaded from https://www.cottongen.org/. A hidden Markov model (HMM) profile of CCO domain (Pfam accession number: PF03055) was downloaded from Pfam (El-Gebali et al. 2019) and used to scan the protein databases with HMMER v3.0 (http://hmmer.org/) with E value < 0.01. The screened proteins were further confirmed against SMART (Schultz et al. 1998) and InterPro (Mitchell et al. 2019) databases. The identified cotton CCO genes were named according to their homology with Arabidopsis genes. Physiochemical characteristics were identified using ExPASy on http://www.expasy.org/tools/.

Multiple alignment and phylogenetic analysis

Multiple alignment was performed by DNAMAN software. Phylogenetic trees were conducted by MEGA 7.0 using the neighbor-joining (NJ) method. Sequences of CCO proteins of Arabidopsis thaliana, Zea mays, and Solanum lycopersicum were published by Tan et al. (2003), Vallabhaneni et al. (2010), and Wei et al. (2016).

Structure analysis and cis-element prediction

Motif analysis was carried out with MEME Suite (Bailey et al. 2009) and displayed using TBtools software (Chen et al. 2020). Gene structure was retrieved from GFF3 files of the genomes. The sequences 1500 bp upstream of the translation initiation site were extracted, and cis-elements were predicted by PlantCARE on http://bioinformatics.psb.ugent.be/webtools/plantcare/html/. Gene Structure Display Server (GSDS) 2.0 (Hu et al. 2015) was employed to show gene structure and cis-element distribution.

Chromosomal location and collinearity analysis

Chromosomal distribution of cotton CCO genes was mapped by MapInspect software. Gene duplication was identified using the MCScanX program. The nonsynonymous substitution rates (Ka) to synonymous substitutions rates (Ks) (Yadav et al. 2015) were calculated with TBtools software.

Gene expression pattern analysis of cotton CCOs

Data of the accession codes PRJNA248163 of G. hirsutum L. acc. TM-1 were fetched from SRA databases (Zhang et al. 2015). Various tissues (leaf, stem, root, stamen, petal, ovule of − 3, 0, and 3 DPA (days post-anthesis), fiber of 5, 10, 20, and 25 DPA) and abiotic stress (drought, salt, and cold stress for 1, 3, 6 and 12 h) transcriptome datasets were employed. The fragments per kilobase of exon model per million mapped reads (FPKM) were estimated with cufflinks and normalized to evaluate the expression levels. Heat maps were generated using the TBtools software.

Plant materials and treatments

Seedlings of upland cotton cultivar “Jimian 2016” were cultivated in Hogland solution in the artificial incubator under controlled conditions (25℃ for 16-h light/22℃ for 8-h dark). After 3 weeks, seedlings at trefoil stage were subjected to drought (17% PEG6000), salt (200 mM NaCl), and cold (4℃), respectively. Samples were collected at desired time points and stored at − 80℃. Samples from untreated seedlings were selected as control.

Quantitative real-time PCR analysis

Total RNA extraction, cDNA synthesis, and quantitative real-time PCR (qRT-PCR) were performed according to the manufacturer’s instruction of the RNAprep pure plant kit (TIANGEN, China), the PrimeScript™ RT reagent kit (TaKaRa, China), and the TB Green® Premix Ex Taq™II (TaKaRa, China), respectively. qRT-PCR reactions were conducted with three replicates on CFX96 Real-Time PCR System (Bio-Rad, USA). The histone3 (AF024716) was amplified as internal control. Relative expression levels were evaluated using 2−ΔΔCt method (Livak and Schmittgen 2001). Specific primers designed by Primer-BLAST (http://www.ncbi.nlm.nih.gov/tools/primer-blast/) are listed in Table S1.

Virus-induced gene-silencing assay

GaNCED3a, the ortholog gene of stress-related gene GhNCED3a_A/D, was selected for functional analysis under drought and salt stress by VIGS in G. arboreum. A 300-bp PCR product of GaNCED3a was amplified using cDNA from G. arboreum cv. “Shixiya-1” and inserted to the TRV vector. Primer used for pTRV2:GaNCED3a vector conduction is listed in Table S1. Plasmid with three vectors, pTRV2:00, pTRV2:CLA1, and pTRV2:GaNCED3a, were transformed to Agrobacterium tumefaciens strain GV3101. VIGS injection was performed according to Yang et al. (2019). Samples were collected from plants 14 days after inoculation and qRT-PCR analysis was conducted to evaluate the efficiency of VIGS silencing. Plants with expression level of GaNCED3a less than 40% of control were used to exam phenotypic and physiological variations. TRV2:00-tageted plants were taken as control. At 21 days post-inoculation, plants were subjected to drought and salt stress, and leaves were sampled for malondialdehyde (MDA), proline (PRO), and water loss assay according to the instruction of corresponding kit (Suzhou Comin Biotechnology Co., Ltd.). Water loss was evaluated by weighing the leaves per hour and water loss rate represented the leaf water loss weight to the leaf fresh weight. For drought and salt stress treatment, plants were stopped watering and watered with 300-mM NaCl instead, respectively. Plants irrigated with water were taken as mock.

Results

Identification of cotton CCO members

A total of 15, 15, and 30 CCO members were identified from G. arboreum, G. raimondii, and G. hirsutum, respectively. The number of CCO of tetraploid cotton G. hirsutum was twice of diploid cotton G. arboreum and G. raimondii, which was consistent with their corresponding genomes. Length of encoded amino acid residues of cotton CCOs varied from 353 to 647. Molecular weight of cotton CCOs proteins ranged from 39.00 to 72.13 kDa, and isoelectric point (PI) was distributed from 5.39 to 8.94. Information of cotton CCO genes is detailed in Table 1.

Table 1 Information of cotton CCO genes
Full size table

Phylogenetic analysis of CCO proteins

Multiple sequence alignments indicated that CCO proteins in Gossypium showed high sequence identity (Fig. S1). To further analyze the evolutionary relationship of CCO family proteins, a phylogenetic tree was constructed using sequences of CCO from monocot Z. mays, dicot A. thaliana, S. Lycopersicum, and Gossypium (Fig. 1). CCO proteins from six plant species were divided into NCED and CCD groups, and the CCD group was further classified into five subgroups, CCD1, CCD4, CCD-like, CCD7, and CCD8, as previously reported (Chen et al. 2018; Kim et al. 2016).

Fig. 1
figure 1

Phylogenetic tree of CCO proteins from Arabidopsis thaliana (At), Zea mays (Zm), Solanum lycopersicum (Sl), G. ramondii (Gr), G. arboreum (Ga), and G. hirsutum (Gh)

Full size image

In NCED group, NCED9 did not exist in cotton and S. lycopersicum. NCED3 in cotton were clustered with AtNCED3 and SlNCED, while NCED5 in cotton were clustered with SlNCED2 but not AtNCED5. Furthermore, NCED6 in cotton were orthologous to AtNCED6 and SlNCED3. Orthologous of NCED5 were not found in maize. Five members of maize NCEDs, ZmNCED1, ZmNCED2, ZmNCED3a, ZmNCED3b, and ZmNCED9, clustered in a single branch, indicating a conservation of NCED sequences in monocot maize. Compared with the fact that other plants contained only one copy of NCED3 and NCED5 (Zhou et al, 2019; Wang et al. 2013; Tan et al. 2003), there are three copies of NCED3 and NCED5 in cotton, NCED3a/3b/3c and NCED5a/5b/5c. This suggested that a duplication event had occurred during the evolution process.

In CCD group, orthologous of CCD-like were found in Gossypium, S. lycopersicum (SlCCD-like), and Z. mays (ZmCCD8b), and were not found in A. thaliana. Obviously, ZmCCD8b were not clustered into CCD8 subgroup but CCD-like subgroup in the present study. There were two CCD4 subgroup members, CCD4a and CCD4b, in Z. mays, S. lycopersicum, and Gossypium. Only one member was found in A. thaliana. These results implied that CCO members in different plant species had great divergency in the evolution process. CCD1, CCD4, and NCEDs had more close relationship in the NJ phylogenetic tree, suggesting their divergence from a common ancestor.

Analysis of gene structure and motif

Exon/intron structures were analyzed to further understand the evolution of CCO gene family in cotton (Fig. 2). NCEDs in cotton shared a simple gene structure. All NCEDs contained only one exon with conserved length except for GrNCED5a and GrNCED5b. While exon number of cotton CCDs varied greatly, with the range of 1 to 15. CCD4 in cotton contained 1–2 exons. CCD 7 and CCD8 contained 6 exons (except for GaCCD8a and GrCCD8a), but with different distribution of exon and intron length. CCD1 contained variable numbers of exons from 12 to 15. CCD-likea and CCD-likeb in cotton shared similar gene structures with different exon numbers and intron lengths. The results revealed that genes grouped in the same clades processed a similar gene structure, indicating a close correlation between the phylogeny and exon/intro structure.

Fig. 2
figure 2

The CCO gene family from three cotton species. a Phylogenetic relationship of cotton CCO proteins. b Extron-intron structure analysis of cotton CCO genes. c Motifs in cotton CCO proteins

Full size image

Ten conserved motifs were analyzed by MEME software (Fig. S2, Table S2). The conserved motif in all 60 cotton CCO protein sequence was plotted by TBtools software (Fig. 2). It is clearly that CCOs grouped in the same cluster had similar conserved motifs, which was consistent with the results of the phylogenetic analysis. Cotton NCED3 and NCED5a proteins contained all ten conserved motifs and their relative positions were also conserved. Motif 9 and motif 10 were not found in GhNCED5b_A/D, GhNCED5c_A, and GrNCED5c. Only four motifs were detected in GhNCED5c_D, which might result from the gene fragment loss in the process of evolution or genome assembly errors. Cotton NCED6 protein did not contain motif 7, implying the potential divergence functions of NCEDs in cotton. In cotton CCD subgroup, CCD1, CCD4, and CCD-like had nine conserved motifs except motif 7. Motif 7 were only found in N-terminal of CCD8 members. CCD8 had seven motifs (motif 2–8) and CCD7 had only four motifs (motif 2, motif 5, motif 8, and motif 9). Both CCD7 and CCD8 did not contain motif 1 and motif 10. Motif numbers and distributions of CCD7 and CCD8 were highly different from other CCOs, suggesting that there was a high sequence variation. On the whole, cotton NCEDs contained more conserved motifs than CCDs.

Promoter analysis of cotton CCO genes

According to the potential functions of cotton CCO, cis-elements involved in plant hormone responsiveness and biotic and abiotic stress were analyzed. The distribution of putative cis-elements were drawn by GSDS (Fig. 3, Fig. S3). Eighteen types of cis-elements related to hormone and stress were found in 60 cotton CCO promoter regions. The information is briefly summarized in Table S3. Seven types of stress-related cis-elements including MBS involved in drought stress, LTR, and DRE core that involved in cold response, TC-rich repeats involved in defense and stress response, W box involved in abiotic stress, WUN-motif involved in wound-response, and ARE involved in anaerobic induction were found in the promoters. These results of cis-elements identification implied that cotton CCO genes take a great part in plant multiple stress response. Phytohormones response cis-elements, such as ABA-responsive element (ABRE), ET responsive element (ERE), MeJA-responsive elements (TGACG-motif and CGTCAmotif), SA-responsive element (TCA-element), and gibberellin-responsive element (P-box) were also found abundantly in the promoter sequence. There were 12 out of 15 promoters contained ABA-responsive element (ABRE), indicating cotton CCO’s potential role in ABA-mediated signals. Overall, the diverse cis-regulatory elements in the promoters related to stress and plant hormone may imply their diverse functions in cotton growth.

Fig. 3
figure 3

Predicted cis-elements related to stresses and hormones in G. hirsutum

Full size image

Chromosomal distribution and synteny analysis

The chromosomal distribution of CCO from three cotton species were determined based on their corresponding genomic information (Fig. 4). The chromosomal distribution of cotton CCOs was uneven. For G. hirsutum, 30 CCOs were distributed on 18 chromosomes and two CCOs (GhNCED5b_A and GhNCED3b_D) were located on two scaffolds. The distribution of CCO on chromosomes of A sub-genome exhibited a good correspondence with those on chromosomes of D sub-genome. There was no CCO gene located on Ga_chr2, Ga_chr4, Ga_chr12, Gr_chr3, Gr_chr5, Gr_chr7, and Gr_chr12. The localization of CCO genes between G. raimondii and G. arboreum did not exhibit a good correspondence, which might be owing to the evolution of different species.

Fig. 4
figure 4

Physical mapping of CCO family genes in cotton. The chromosomes of A-subgenome of G. hirsutum, D-subgenome of G. hirsutum, G. arboreum, and G. raimondii are drawn in green, blue, yellow, and red, respectively

Full size image

To understand the expansion patterns of cotton CCO genes, gene duplication events were investigated by genome synteny analysis (Fig. 5). Only one pair of tandem duplication cluster (GhCCD-likea_D and GhCCD-likec_D) was identified. One segmental duplication pair was found within the genome of G. raimondii, 14 pairs within G. hirsutum, and 29, 26, and 14 pairs between G. raimondii and G. hirsutum, G. arboreum and G. hirsutum, and G. arboreum and G. raimondii, respectively (Table S4). These results implied polyploidization and segmental duplication were the major processes during gene expansion. The ortholog genes of GaNCED5b and GaCCD-likea were not found in G. hirsutum, while 3 genes in tetraploid cotton (GhNCED5b_A, GhCCD8b_A, and GhCCD-likec_D) had no orthologous gene in the two diploid cottons. These results implied that some CCO genes might undergo gene loss, gain, or rearrangement during the polyploidization process. To explore the selection constraints, the Ka/Ks ratio was calculated (Table S4). Only one pair (GaNCED6 and GhNCED6_A) had a positive pressure (Ka/Ks > 1), while the other pairs experienced a purifying selection process.

Fig. 5
figure 5

Syntenic analysis of CCO genes between G. raimondii, G. arboreum, and G. hirsutum. Green curves denote syntenic regions between G. arboreum and G. raimondii, red curves between G. arboreum and G. hirsutum, blue curves between G. raimondii and G. hirsutum, brown curve within G. raimondii, and orange curve within G. hirsutum

Full size image

Genomic comparison of different organisms is a way to understand the origins, evolutionary history, and new gene functions (Lyons et al. 2008). Arabidopsis as a model plant had been well characterized, and some CCO genes had been given deep research. To better explore the origin and evolutionary history of CCO genes in G. hirsutum, a syntenic map between G. hirsutum and Arabidopsis was generated (Fig. S4). Sixteen pairs of ortholog between G. hirsutum and Arabidopsis were syntenic genes. Eight pairs of orthologous genes (AtNCED6/GhNCED6_A, AtNCED6/GhNCED6_D, AtNCED5/GhNCED5a_A, AtNCED5/GhNCED5a_D, AtCCD7/GhCCD7_A, AtCCD7/GhCCD7_D, AtCCD4/GhCCD4b_A, AtCCD4/GhCCD4b_D) were single one-to-one correspondence between Arabidopsis and A-subgenome or D-subgenome of G. hirsutum, indicating that they may be derived from a common ancestor. Such results laid foundation on understanding the roles of cotton CCO genes.

Expression profiles of CCO genes in G. hirsutum

Transcription levels of CCO genes in various tissues varied greatly (Fig. 6a). In CCD subgroup, GhCCD1_A/D has shown high expression level in all tested tissues (FPKM > 15). GhCCD4a_A and GhCCD4b_A/D exhibited high expression level in stamen; other CCDs showed high expression in stem. GhNCED3a_A/D exhibited high expression level in root; GhNCED3b_A/D, GhNCED3c_A/D, and GhNCED6_A/D have shown high expression level in leaf, and GhNCED5a_A and GhNCED5c_A/D in petal.

Fig. 6
figure 6

Gene expression pattern of CCO members in specific tissues of G. hirsutum. a Expression pattern in different tissues. b Expression patterns in fiber development stages. c Expression pattern under abiotic stress. Data represented log2-transformed FPKM that obtained from SRA dataset PRJNA248163

Full size image

Dynamic changes in expression level of CCO genes in different stages of fiber development were observed (Fig. 6b). GhNCED3c_A/D was high-expressed in fiber initiation stage of 3 DPA ovules and 5 DPA fibers, while GhNCED3b_A/D high-expressed in late stage of fiber development of 25 DPA fibers. The expression profiles of GhNCED3a_A/D were dynamic, which with relatively higher expression in 3 DPA ovules and 25 DPA fibers. Only one CCD member, GhCCD4b_A, had high expression level in 10, 20, and 25 DPA fibers. Other CCD genes showed a sustained low expression level in different fiber developmental stages (FPKM < 1). The specific expression patterns of cotton CCO in fiber implied their potential functions in different fiber developmental stages.

To explore their functions under abiotic stress, a comprehensive analysis of expression profiles under drought, salt, and cold conditions was conducted (Fig. 6c). Notably, GhNCED3a_A/D and GhNCED3c_A/D were upregulated by all imposed stress, and peaked at 12 h after treatments. GhCCD4b_A was induced strongly during the first 6 h and then decreased under drought and salt, while was repressed under cold condition. GhCCD4b_D showed a similar expression pattern with GhCCD4b_A under drought and cold. GhCCD1_A/D were slightly downregulated under all treatments. Almost no expression was detected in the other genes (FPKM < 1).

Seven CCO genes that significantly responded to three abiotic stress were selected for qRT-PCR to validate the differential expression patterns (Fig. 7). The results of qRT-PCR for candidate CCO genes basically coincide with the results from RNA-seq data. Of note, the results indicated the selected NCED genes were upregulated at each time points under various treatments. GhNCED3a_A/D under drought and salt stress were upregulated, with a maximum 11-fold increase. Moreover, GhNCED3c_A/D were strongly induced by all imposed adversity stress and the expression level peaked at 6 h under salt, whereas peaked at 12 h under drought and cold. It is deserved to be mentioned that the expression level of GhNCED3c_D under cold condition at 12 h was up to 389-fold than in control. The high fold-change expression levels indicated their important roles in stress response.

Fig. 7
figure 7

Expression analysis of 7 CCO genes in leaves under drought, salt, and cold stress in “Jimian2016” by qRT-PCR. * indicated significant difference between treatment and CK (p < 0.05), ** indicated highly significant difference between treatment and CK (p < 0.01)

Full size image

VIGS assay of GaNCED3a in G. arboreum under drought and salt stress

To further identify the function of abiotic stress responsive gene, we selected GaNCED3a, the ortholog gene of GhNCED3a_A/D, for reverse genetics in diploid cotton G. arboreum via VIGS method. About 2 weeks after VIGS injection, plant leaves that inoculated with pTRV2:CLA1 vector (VIGS-CLA1) were observed to turn white (Fig. 8a), which was coincident with previous findings (Li et al. 2018; Yang et al. 2019). Meanwhile, the plant leaves inoculated with pTRV2:00 and pTRV2:GaNCED3a were sampled for qRT-PCR to evaluate the silencing efficiency of GaNCED3a. Figure 8b shows that silencing of GaNCED3a was highly effective. Plants of VIGS-NCED3a-1/2/3/4/5/7/10/11/12 with expression level of GaNCED3a less than 40% of control plants were employed for further study on phenotypic and physiological variations.

Fig. 8
figure 8

Function validation of GaNCED3a by VIGS. a Phenotypes of seedlings after 14d inoculation. b Silencing efficiency of GaNCED3 gene via VIGS. c–d Phenotypes of WT, VIGS-TRV2, and VIGS-GaNCED3a seedlings under drought and salt treatment, respectively. e Water loss rate, MDA, and proline assay of VIGS-TRV2 and VIGS-GaNCED3a seedlings. f Expression analysis of biotic stress responsive genes RD29A, DREB1A, and SOS1 in WT, VIGS-TRV2, and VIGS-GaNCED3a seedlings under drought and salt stress. * indicated significant difference between treatment and its corresponding CK (p < 0.05); different lowercase letter a and b represented significant difference between the expression levels of genes in seedlings under different treatments (p < 0.05)

Full size image

Then, plants of TRV2:GaNCED3a (VIGS-GaNCED3a), TRV2:00 (VIGS-TRV2), and wild type (WT) were exposed to drought and salt stress. GaNCED3a-silenced plants showed more obvious symptoms of wilting and drooping under both drought and salt stress than control plants (Fig. 8c–d). Water loss rate and MDA content in silenced plants were significantly higher than control plants (P < 0.05), while PRO content in silenced plants was significantly lower than control plants (P < 0.05) (Fig. 8e). The results of VIGS assay of GaNCED3a indicated decreased drought and salt stress tolerance in silenced plants.

To learn how GaNCED3a responded to drought and salt stress in the silenced plants, expression patterns of abiotic stress response genes DREB1A, RD29A, and SOS1 were evaluated. Under drought and salt stress, DREB1A and RD29A exhibited significantly high expression in both silenced and control plants (P < 0.5) (Fig. 8f). However, all three stress response genes showed significantly downregulated expression in VIGS-silenced plants. Silencing GaNCED3a reduced drought and salt tolerance, which may be related to the inhibition of expression activity of abiotic stress response genes DREB1A, RD29A, and SOS1.

Discussion

Characterization of cotton CCO genes

Prior to this study, CCO gene family in rice, soybean, pepper, Brassica, and apple had been systematically identified and named according to Arabidopsis orthologs (Chen et al. 2018; Kim et al. 2016; Tan et al. 2003; Vallabhaneni et al. 2010; Wang et al. 2013). A total of 15, 15, and 30 CCO were identified from G. arboreum, G. raimondii, and G. hirsutum, respectively. Cotton CCO genes could be divided into two groups, NCEDs and CCDs, which was in consistent with other plants (Giuliano et al. 2003). In NCED subgroup, there was one orthologous gene to AtNCED6 in cotton. NCED3 and NCED5 each had three copies in cotton (NCED3a, 3b, 3c, and NCED5a, 5b, 5c), while no orthologs of AtNCED2 and AtNCED9 genes were found in cotton. According to these results, it is speculated that cotton NCED3 and NCED5 had experienced a gene duplication event in the evolution process and their duplication may be the primary factor for the expansion of CCO gene family. Meanwhile, the specific enzymatic role of cotton NCED members remained elusive. Two members of cotton CCD-like genes clustered together with SlCCD-like (Fig. 1), a new subgroup found in some plant species (Chen et al. 2018; Wang et al. 2017; Wei et al. 2016; Zhou et al. 2019), while CCD-like homologs were not found in Arabidopsis. CCD1 and CCD7 had only one member each in cotton, while CCD4 and CCD8 had two members each (CCD4a, 4b, and CCD8a, 8b), which was consistent with previous studies (Vallabhaneni et al. 2010; Wang et al. 2013, 2017).

Numbers of exons and introns of cotton CCO members varied significantly. NCEDs and CCD4 in cotton were found intron deficient (intron ≤ 1), in accordance with other species, such as Rosa damascene, Osmanthus fragrans, and Fragaria vesca (Huang et al. 2009; Wang et al. 2017). This intron-deficient structure used to be considered essential for rapid response to stress through ABA synthesis and ABA-mediated signal transductions in plants (Wang et al. 2017). NCEDs in cotton had only one exon and encoded more conserved motifs than other CCO members, and was similar to the results described in other reports (Zhou et al. 2019).

The number of CCO genes in cotton was more than that in most of other plants, indicating that cotton CCO gene experienced extensive expansion in the evolution process. Polyploidization contributed a great part in duplication, while segment duplication also made valuable contributions to the expansion of gene families (Li et al. 2015; Paterson et al. 2012). Eighty-four pairs of segmental duplicated genes were observed in this study, implying that segmental duplication had profound effects on evolution of CCO gene family in cotton. Sixteen pairs of ortholog between G. hirsutum and Arabidopsis were detected, suggesting that they may derive from a common ancestor and have similar functions (Xu et al. 2012).

Expression patterns of CCO genes

Tissue-specific expression patterns implied a high functional diversification within CCO family. GhCCD1_A/D has shown high expression in all samples (Table S5, Fig. 6), which was in line with the results in other plants (Auldridge et al. 2006b; Chen et al. 2018; Simkin et al. 2004). This result suggests the multiple effects of CCD1 and its orthologs in plant growth and development (Ohmiya et al. 2006; Walter et al. 2007). In contrast with the fact that maize CCD genes preferentially high expressed in leaves (Vallabhaneni et al. 2010), prevalent CCD genes of upland cotton predominantly high-expressed in stem. However, GhCCD4b_A showed highest expression level in petal, stamen, and samples of fiber elongate stage (10 and 20 DPA), suggesting a putative function in flower and fiber development. The predominant expression of CCD4 in floral organs were also found in other plants (Brandi et al. 2011; Adami et al. 2013). The function of GhCCD4a_A, which expressed slightly in stamen, followed by root and stem, and no expression in any other tissues, remained subtle. GhCCD7_A/D barely expressed in petal, stamen, and fibers, but exhibit high expression in stem and leaf, indicating CCD7 genes might mainly take part in vegetative growth instead of reproduction growth in upland cotton, while this was inconsistent with the results in early studies that root and fruit preferentially accumulated CCD7 (Vogel et al. 2010; Wei et al. 2016; Chen et al. 2018). GhNCED3a_A/D has shown high expression level in roots, while GhNCED3b_A/D and GhNCED3c_A/D in leaves, suggesting that NCED3 genes in upland cotton exhibited diverged functions involved in plant growth. The GhNCED5c_A/D had the highest expression in the petal, and followed by the stamen, but barely expressed in ovule and fiber samples. Early studies reported that AtNCED5 induce seed dormancy (Frey et al. 2012), while our results implied specific functions of GhNCED5c_A/D involving in floral organ development.

The expression pattern analysis under abiotic stress provided useful information to gain a deep understanding of the roles of cotton CCO. Based on the results of RNA-seq data analysis and qRT-PCR, NCED3a and NCED3c in cotton were probably responsible for stress response. It is well known that abiotic stress, such as drought, salt, cold, and heat, can induce accumulation of ABA level in plants (Roychoudhury et al. 2013; Zhang et al. 2006) and NCEDs are mainly involved in ABA-mediated stress response and plant growth by regulating ABA synthesis (Huang et al. 2019; Hwang et al. 2018; Pedrosa et al. 2017; Yang et al. 2018). NCED3, which is the key player in ABA biosynthesis pathway, have been widely proved to take a prominent part in abiotic stress response, especially drought stress (Frey et al. 2012; Gavassi et al. 2021; Iuchi et al. 2000; Tan et al. 2003). GhNCED3a_A/D and GhNCED3c_A/D exhibited strong stress responses, indicating their potential roles in stress tolerance. In this study, NCED3 and NCED5 genes in upland cotton exhibited different expression pattens, implying that they had experienced divergency in function. This phenomenon was also found in other plant species (Iuchi et al. 2000; Wang et al. 2013). Frey et al. (2012) demonstrated that there was an interaction between NCED3 and NCED5 in drought stress response. Whether cotton NCED5 interacted together with NCED3 involving in ABA biosynthesis in stress response deserved to be further determined.

Silencing of GaNCED3a by VIGS reduced drought and salt stress resistance

NCED3 was demonstrated to be greatly involved in drought stress response in previous studies (Tan et al. 2003; Fujita et al. 2011; Huang et al. 2018). In this study, results of silencing GaNCED3a in G. arboreum indicated that NCED3 not only are involved in drought stress tolerance but also take an important role in salt stress response.

DREB1 (dehydration-responsive element-binding protein 1) transcription factors are involved in responsiveness to drought, salt, and low temperature stress by regulating a series of downstream genes related to abiotic stress (Gilmour et al. 2000; Yamaguchi-Shinozaki and Shinozaki 1994). RD29A is proved to have a positive effect on enhancing abiotic stress tolerance in plants (Msanne et al. 2011). SOS1, which encodes a plasma membrane Na+/H+ antiporter, is assumed to be involved in salt tolerance improvement (Yuesen et al. 2012; OlÍas et al. 2009). In this study, stress response genes DREB1A, RD29A, and SOS1 were demonstrated to be significantly downregulated in GaNCED3a-silenced cotton plants, which implied that GaNCED3a had a potential function in enhancing stress tolerance. Previous studies showed that some stress-responsive genes were upregulated by ABA under stress condition (Agarwal and Jha 2010; Gavassi et al. 2021; Ingram and Bartels 1996). What is more, mutants of nced3 and nced5 in rice were sensitive to salt and drought stress, and OsNCED3 and OsNCED5 were proved to be responsible for endogenous ABA accumulation (Huang et al. 2018, 2019). Therefore, our findings indicated that inhibited transcription level of GaNCED3 in the silenced plants may lead to decreased ABA biosynthesis, and further caused the inhibition of stress-responsive genes expression, which finally resulted in decreased drought and salt stress tolerance in GaNCED3-silenced plants. It is speculated that GaNCED3, as the orthologous of NCED3 genes in other plants, played an essential role in regulation of ABA content in abiotic stress response. However, detailed molecular mechanism of GaNCED3a in the regulation of multiple abiotic stress still deserved to be evaluated in further study.

Conclusions

We performed a genome-wide characterization of CCO gene family in three cotton species. CCO genes from A. thaliana, Z. mays, and S. lycopersicum and three species of cotton were divided into 6 subgroups. The proteins were found to be conserved. Gene expansion analysis indicated that polyploidization and segmental duplication were the major processes during cotton CCO genes expansion. Expression profile analysis revealed that a certain number of CCO genes, such as GhCCD1_A/D, GhNCED3a_A/D, and GhNCED3c_A/D may be closely involved in stress response. Silencing of GaNCED3a, the ortholog gene of stress-related gene GhNCED3a_A/D, by VIGS reduced drought and salt stress resistance. Our findings paved the way for the researches focused on clarifying the function of cotton CCO genes.