Genome-wide characterization and expression analysis of Erf gene family in cotton

Background AP2/ERF transcription factors are important in a variety of biological activities, including plant growth, development, and responses to biotic and abiotic stressors. However, little study has been done on cotton’s AP2/ERF genes, although cotton is an essential fibre crop. We were able to examine the tissue and expression patterns of AP2/ERF genes in cotton on a genome-wide basis because of the recently published whole genome sequence of cotton. Genome-wide analysis of ERF gene family within two diploid species (G. arboreum & G. raimondii) and two tetraploid species (G. barbadense, G. hirsutum) was performed. Results A total of 118, 120, 213, 220 genes containing the sequence of single AP2 domain were identified in G. arboreum, G. raimondii, G. barbadense and G. hirsutum respectively. The identified genes were unevenly distributed across 13/26 chromosomes of A and D genomes of cotton. Synteny and collinearity analysis revealed that segmental duplications may have played crucial roles in the expansion of the cotton ERF gene family, as well as tandem duplications played a minor role. Cis-acting elements of the promoter sites of Ghi-ERFs genes predict the involvement in multiple hormone responses and abiotic stresses. Transcriptome and qRT-PCR analysis revealed that Ghi-ERF-2D.6, Ghi-ERF-12D.13, Ghi-ERF-6D.1, Ghi-ERF-7A.6 and Ghi-ERF-11D.5 are candidate genes against salinity tolerance in upland cotton. Conclusion Overwhelmingly, the present study paves the way to better understand the evolution of cotton ERF genes and lays a foundation for future investigation of ERF genes in improving salinity stress tolerance in cotton. Supplementary Information The online version contains supplementary material available at 10.1186/s12870-022-03521-z.

Zafar et al. BMC Plant Biology (2022) 22:134 regulators in various biological processes are the most important target for crop engineering [3]. The AP2/ERF family is a large group of plant-specific transcription factors. This gene family having AP2/ERF-type DNAbinding domain of about 60-70 amino acids was first discovered in Arabidopsis homeotic gene, APETALA2 (AP2) [4]. AP2, ERF, RAV, and DREB are four subfamilies of AP2/ERF gene family. AP2 subfamily is different from the other 3 (ERF, DREB, and RAV) subfamilies as it has double AP2/ERF domains while all three have a single AP2/ERF domain whereas RAV is different from ERF and DREB subfamily by having an additional B3 DNAbinding domain. Over the last 20 years, ERF family genes caught attention as overexpression of ERF genes in different plants directed to abiotic stress tolerance and pathogen resistance in transgenic plants [5,6].
The ERF subfamily members regulate the expression of PR (pathogenesis-related) genes through binding to GCC-boxes (AGC CGC C). Further, ERFs are intricated in signaling pathways including salicylic acid, jasmonic acid, and ethylene pathways which are important for stress response and plant development [7][8][9]. Due to their coordinating ability with multiple signaling and hormonal pathways, ERFs are considered excellent entities for engineering biotic/abiotic stress tolerance in plants [9]. Several studies have shown the involvement of ERF gene expression under different stress tolerance conditions and tissues in plants [10][11][12].
Cotton is the main agro-industrial crop and is the source of the most important natural fiber used in textile production [13]. This plays a vital role in a global economy and is grown in more than seventy countries. The genus Gossypium contains 46 diploid (2n = 2x = 26) and 5 well-established and 1 purported tetraploid (2n = 4x = 52) species. It has been proposed that all diploid cotton species may have evolved from a common ancestor that subsequently diversified to produce eight groups, including groups A-G and K3 [14]. However, biotic and abiotic stresses badly affect the production and growth of cotton. So, the struggles to explore the molecular mechanism of stress to support stress tolerance in plants is real and of fundamental importance for cotton production [15]. Considering the importance of ERF family genes in crop improvement, genome-wide investigation of ERF gene-family in cotton can help us to understand the molecular mechanisms of resistance to stress, and thus aid in the development of cotton varieties, using transgenic technology, with greater tolerance to many adverse environments. The release of different cotton whole-genome sequence data, including Gossypium arboreum L. [16], Gossypium raimondii [17], Gossypium hirsutum L. [18], and Gossypium barbadense L. [19] has made it possible to systematically identify and analyze the cotton ERF genes on a genome-scale level. Here, we performed a comprehensive analysis of cotton ERF genes, including their gene structure, motif compositions, chromosome distribution, duplication patterns, and expression profiles. This study will provide valuable clues for the functional characterization of the ERF gene family in cotton.

Database and sequence retrieval
Gene sequences of Arabidopsis AtERF were retrieved from the TAIR (Arabidopsis thaliana database) [20]. Cotton genome database Cotton FGD was used for retrieval of G. barbadense (NAU), G. hirsutum (CRI), G. raimondii (JGI) and G. arboreum (CRI) genome sequences [21]. ERF domain sequence was used as a template to retrieve the probable domain homologs from the whole-genome sequence of G. barbadense, G. hirsutum, G. raimondii, and G. arboreum [22] through BLASTP at CottonGen (https:// www. cotto ngen. org). All non-redundant hits with less than 1E-5 E-value were taken. Non-targeted and overlapping sequences were removed. Pfam30.0 (http:// pfam. xfam. org/) database was used to retrieve hidden Markov Model (HMM) profile of the U-box domain (PF00847) [22], and retrieved results were used as a template to find out the candidate ERFs from the cotton genome protein database using HMMER3.0 [23]. Protein sequences, CDSs (coding domain sequences), and corresponding full-length sequences in the genome were obtained by using BLAST2.2.31+ (ftp:// ftp. ncbi. nlm. nih. gov/ blast/ execu tables/ blast+/ LATEST/). Further, SMART (http:// smart. embl-heide lberg. de/) and Pfam 30.0 [22] databases were used for additional analysis to ensure that each candidate protein contained a U-box domain and BUSCA was used for subcellular localization prediction [24,25].
Gene structure and conserved motif analyses MEME (Multiple Em for Motif Elicitation) version 5.3.0 (http:// meme-suite. org/ tools/ meme/), an online tool was used for the identification of conserved motifs of ERF proteins. TBtools software was used to predict gene structure and integrate phylogenetic trees and conserved motifs.

Analysis of Cis-acting regulatory elements in promoter regions
A 1.5 kb of promoter sequence upstream from the transcription start site in each Ghi-ERF gene was extracted from the G. hirsutum genome database and analyzed using PlantCARE online software (http:// bioin forma tics. psb. ugent. be/ webto ols/ plant care/ html/) to predict the putative cis-acting regulatory elements [26]. Zafar et al. BMC Plant Biology (2022) 22:134 Physical location of ERFs on the chromosome and phylogenetic tree Mapcahrt was used to generate the distribution of cotton ERF on the chromosome. Genome annotation files were used to retrieve the GFF (general feature format) information of the cotton ERFs. The 671 protein sequences of all four species of cotton and 60 protein sequences of Arabidopsis were used for the phylogenetic tree. The Clustal Omega (https:// www. ebi. ac. uk/ Tools/ msa/ clust alo/ [27]; was used to align ERF protein sequence and MEGA v7.0 [28] program was used to construct a neighbor-joining phylogenetic tree with 1000 bootstrap replicates.

Gene duplication and micro-synteny analysis in G. arboreum, G. raimondii, G. barbedense and G. hirsutum L
To investigate the collinearity and to analyze the syntenic relationship among the four cotton species, the complete genome sequences of these cotton species along with respective genome annotation files were subjected to MCScanX tool [29]. TBtools software was used to visualize the obtained results. Multiple collinear scanning toolkits (MCScanX) and Dual Synteny Plotter software (https:// github. com/ CJ-Chen/ TBtoo ls/) [30] were used for the analysis of gene duplication events [29] and the syntenic relationship between the ERF genes among cotton species, respectively. Some previous studies showed the identification of homologous gene pairs according to MSA (multiple sequence alignment). Advanced circos was used to visualize the collinearity of homologous genes based on the homology between each species and their positions on the genome [30].

Transcriptome data analysis expression analysis
To identify the expression pattern of identified ERF genes in various tissues of all the four cotton species, the RNA-seq expression data of G. arboreum, G. raimondii, G. barbadense, and G. hirsutum were downloaded from NCBI. The raw expression reads data from various anatomical tissues and various time-points of ovule and fiber of G. arboreum cultivar Shixiya1 and G. hirsutum accession TM-1 were retrieved from NCBI-Bioproject PRJNA594268 [31]. Furthermore, RNA-seq expression data of G. hirsutum accession TM-1 and cultivar Hai7124 were retrieved from NCBI-Bioproject PRJNA490626 [32] for G. barbadense. Similarly, raw expression data of strain Ulbr of G. raimondii were obtained from NCBI-Bioproject PRJNA171262 [33]. The program hisat2 [34], was used for mapping reads, cufflinks (version: 2.2.1) [35], were used to analyze gene expression levels, and fragments per kilobase million values were used to normalize gene expression levels, then the results were logtransformed and a heatmap was generated by MeV [36].

qRT-PCR
Cotton genotype (Eagle-2) was obtained from Four Brothers hybrid seed company Lahore, Pakistan and permission was granted. The cotton (Eagle-2) seeds were germinated on a wet filter paper for 3 days at 25 °C, and then transferred to a Hoagland nutrient solution (Supplementary file 1) [37] in a greenhouse with 60-70% humidity, 14 h photoperiod, and 28 °C day/night temperature. Over a period of 6-8 weeks, the plants were placed in a greenhouse under controlled conditions. The selected cultivar Eagle-2 was exposed to salt stress at 0 dS/m, 10 dS/m, and 15 dS/m at two leaves stage. Newly emerged leaves of the cultivar Eagle-2 were taken after a regular interval of 1 h, 3 h, 6 h, 12 h, and 24 h for the extraction of RNA (RNAprep Pure Plant Kit by Tiangen, Beijing, China) and freeze in − 80 °C. RNA concentration and integrity were observed on NanoDrop 2000 spectrophotometer (Thermo Scientific, USA) and 1% agarose gel electrophoresis and cDNA was prepared (PrimeScript ® RT Reagent Kit, Perfect Real Time, Takara Biotechnology Co., Ltd., Dalian, China). Technically, three replicates of each sample were used. The mean expression of mRNA was measured using qRT-PCR (Maxima SYBR Green)/ ROX qPCR Master mix (2X), cat#K0221, Thermo scientific, USA. Gene-specific primers were used for the amplification of all the genes and GAPDH was used as an internal control for gene normalization.

Results
Identification, sequence analysis and phylogenetic tree of ERF genes in G. arboreum, G. raimondii, G. barbadense and G. hirsutum A total of 118, 120, 213, 220 genes were identified in G. arboreum, G. raimondii, G. barbadense, and G. hirsutum respectively. To identify ERF genes, the sequence of ERF domain was a blast against the whole genome sequence of G. arboreum, G. raimondii, G. barbadense, and G. hirsutum. All non-redundant ERF genes were obtained from each species. The amino acid sequence of these 671 ERF genes was evaluated by Pfam software to confirm their reliability and the presence of ERF domains. Genes that lack ERF domain in the encoding protein sequences or truncated genes and the genes that were not annotated in their respective genome were deleted. The detailed information of selected genes for G. arboreum, G. raimondii, G. hirsutum, and G. barbadense are listed in (Supplementary file 1). According to locations on chromosomes the ERF genes in cotton were renamed, Gar-ERF-1A.1-13A in G. arboreum, Gar-ERF-1D.1-13D in G. raimondii, Gba-ERF-1A.1-13D in G. barbadense, Ghi-ERF-1A.1-13D in G. hirsutum. In G. barbadense, 98 ERF genes were discovered on At sub-genome, and 115 genes were identified on Dt sub-genome. In G. hirsutum, 109 genes were identified in At sub-genome and 111 genes were identified in Dt subgenome. A phylogenetic (neighbour-joining) tree was constructed to determine the evolutionary relationship of ERF genes by using amino acid sequences of identified ERF proteins in G. arboreum, G. raimondii, G. barbadense, and G. hirsutum with corresponding 60 genes of Arabidopsis (Fig. 1). The evolutionary tree classified the ERF genes into 8 clades with well-supported bootstrap values. The clade one is the largest clade followed by clade VI and clade VIII has the lowest number of genes followed by clade IV. The results showed that ERF genes of four species of cotton and Arabidopsis were unevenly distributed in all clades. Only 2 genes from G. arboreum, and 2 from G. raimondii, were found in clade VIII.

Chromosome distribution and gene duplication analysis
To discover the ERF genes distribution on chromosomes, each ERF gene was mapped on their corresponding chromosome according to gene information of the respective genome database. To further examine the evolution of ERF genes in four species of cotton, genome duplication events were investigated for WGD or segmental and tandem duplications.
In G. raimondii, 120 ERF genes were distributed unequally on all 13 chromosomes (Fig. 2b). The number of ERF genes from chromosome D01 to D13 was 14,5,8,15,8,8,15,17,11,7,7,2, and 3 respectively. The chromosomes D08 and D04 contain the highest number of genes 17 and 15 respectively. While chromosome D12 and D13 exhibited the lowest number of genes 2 and 3 respectively. Among120 ERF genes 74 genes have WGD or segmental duplications and were located unevenly on all chromosomes (Fig. 3B). The chromosomes D08 and D01 contain the highest WGD or segmentally duplicated genes 14 and 10 respectively. On chromosomes D12 and 13 only a single gene have segmental duplications (Fig. 3B). The genes Gra-ERF-1D.6, Gra-ERF-1D.13, Gra- In G. barbadense, 213 ERF genes were scattered on all 26 chromosomes of At and Dt sub-genomes. The 98 and 115 ERF genes were identified in At and Dt sub-genomes respectively (Fig. 2C). In At genome of G. barbadense, chromosome A12 had maximum (15) ERF genes whilst chromosome A01 and A13 have a minimum number of ERF genes 2 and 3 respectively. In Dt sub-genome, the chromosomes D08 and D11 contain the greatest number of (15) ERF genes and chromosome D04 had the lowest (3) ERF genes (Fig. 2C). The 182 ERF genes of G. barbadense, have segmental duplications and these were scattered unevenly on all 26 chromosomes. Out of these 182 genes, 42.75 and 57.14% belonged to At and Dt subgenome respectively (Fig. 3C). Chromosome A13 has only single genes Gba-ERF-13A. 3  In G. hirsutum, 220 ERF genes were unevenly distributed on all 26 chromosomes. Among these 220, 109, and 111 ERF genes were located in At and Dt sub-genome respectively (Fig. 2d). In At genome of G. hirsutum, the chromosomes A08 and A12 have maximum (15) ERF genes whilst the chromosomes A10 and A13 have the minimum number of ERF genes 3 and 4 respectively. In Dt sub-genome, chromosome D12 contains the greatest number of (15) ERF genes and chromosomes D01 and D04 have the lowest (3) ERF genes (Fig. 2d). The segmental duplications were found in 94.55% ERF genes of G. hirsutum, 49.52, and 50.48% genes were located in At and Dt sub-genomes respectively (Fig. 3D and Supplementary file 2). Tandem duplications were found in 3.18% ERF genes and these genes were located on chromosomes A07, A08, D07, D08, D09, and D11.
The G. hirsutum and G. barbadense are evolved due to hybridization between an A-genome species (G. herbaceum or G. arboreum) and a D-genome genome (G. raimondii) [38]. To understand the evolutionary relationships of ERF genes, a relative syntenic map of ERF genes from the four cotton species was fabricated (Fig. 4). Synteny analysis showed several gene loci that are highly conserved between the At and Dt sub-genomes of both tetraploid cotton species. According to our MCScan analysis, 237, 105 duplication gene pairs were found between diploid G. arboreum and tetraploid G. barbadense, G. hirsutum respectively. 393, 307 duplication gene pairs were found between diploid G. raimondii and tetraploid G. hirsutum, G. barbadense respectively. The location of ERF genes on D11, D01, D04, and D06 in G. raimondii have a good collinear relationship with the ERF genes present on the homologous chromosomes of G. hirsutum and G. barbadense (Fig. 4). The ERF genes on chromosome A11 in G. arboreum also have a good collinear relationship with the ERF genes present on the homologous chromosomes of G. hirsutum and G. barbadense. The synteny map revealed that small deletion, duplication and reshuffling of the chromosome may have occurred during evolution. To elucidate the divergence throughout cotton evolution, we examined the orthologous of cotton ERF genes between G. arboreum, G.  For more understanding into the evolution and structural diversity of the ERF family ineach cotton species, we analyzed the gene structure and conserved motif of the ERF genes. Motif 1 and 2 were conserved in all ERF genes across all four species. In G. arboreum, based on the evolutionary tree the ERF family genes are categorized into 3 groups. As shown in (Fig. 5A) 1). To gain more insight into the divergence and functional relationship of Gar-ERF proteins, a total of 12 conserved motifs in the cotton ERF were recognized by MEME software, and the height of each letter in the logo was proportional to the conservation level of amino acid in all sequences analyzed.
As shown in (Fig. 5A) all the Gar-ERF proteins contain motifs 1 and 2. The motifs in different groups indicated that varying degrees of divergence among them. The genes Gar-ERF-11A.12 and Gar-ERF-12A.7 were distinct in group 1 due to the presence of motif 12. The presence of motif 3 in some genes of group 2 and 3 make them different from other genes. In general, the G. arboreum, ERF proteins in the same group usually contained similar motifs, which indicates that they may play similar roles in the development and growth of G. arboreum.   15, Gra-ERF-4D.10, Gra-ERF-8D.6, Gra-ERF-8D.14 were distinct due to the presence of motif 10. In group 2, only one gene was unique due to the presence of motif 3 (Fig. 5B). The

Analysis of putative Cis-acting elements in Ghi-ERFs promoters
To investigate the further biological activity of Ghi-ERFs, the 1.5 kb upstream promoter regions of all Ghi-ERFs were obtained and analyzed the cis-acting regulatory elements by using the PlantCARE database. Various types of elements were found in the promoter sites of ERF genes in G. hirsutum (Fig. 6 & Supplementary file 3). The cis-acting elements which were the binding regions of transcription factors play important role in regulating gene expression. The distribution of cis-acting elements among different genes was different, as shown in (Fig. 6 & Supplementary file 3) a large number of cis-acting elements were predicted to be related to transcription, various hormones, and stresses response, cell cycle, and development. The promoter sites of all Ghi-ERF genes contained both CAAT-box and Unnamed-4 cis-acting regulatory elements. The CAAT box plays a key role in the regulation of nopaline synthase promoters [39]. The core element TATA-box was found in most of the motifs is important for transcription. There exist multiple phytohormone responsive elements including ABA response element (ABRE), ethylene-responsive element (ERE), MeJA-responsive element (CGTCA-motif, MYC and TGACG-motif ), GA-responsive element (P-box), and salicylic acid-responsive element (TCA-element), suggesting that Ghi-ERFs expression were regulated by different phytohormones. Various light-related elements such as (Sp1, I-box, 3-AF1 binding site, TCT-motif, and Gap-box) were also observed. The stress-responsive elements were detected in the promoter of some Ghi-ERFs, like ARE (anaerobic induction element), STRE (stressresponsive element), LTR (low temperature-responsive elements), dehydration-responsive element (DRE1 and DRE-core), TC-rich repeats (defence and stress-responsive element), and wound-responsive element (WRE3 and WUN-motif ). Some growth and development elements were also predicted, including endosperm-expression element (AACA-motif and GCN4-motif ), and meristem-expression element (CAT-box and CCG TCC -motif ). The results indicated that Ghi-ERFs are involved in plant growth and development as well as environmental stress responses.

Expression analysis from Transcriptome data
Various expression patterns of different cotton ERF genes were detected across different tissues like seed, stem, leaves, flower, and fiber. RNA-seq data for fiber (0 DPA (Days Post Anthesis) ovules, 1 DPA ovules, 3 DPA ovules, 5 DPA and 10 DPA fibers) and seed ( . 6 Predicted cis-elements in the promoter regions of G. hirsutum ERF genes DPA, and 40 DPA) of G. arboreum, was used for expression analysis. The Gar-ERF genes were clustered in 5 pattern groups based on the heat map of fiber and seed. Generally, a similar expression pattern was observed within groups. The genes of groups 1, 2, 3, 4 and 5 were highly expressed in ovule 5DPA, 3DPA, 1DPA, 0DPA and fiber 10DPA respectively. From group first only Gar-ERF-11A.7 were upregulated in seed, ovule and fiber development. Whereas, Gar-ERF-4A.1 and Gar-ERF-5A.1 were upregulated in fiber and ovule whilst downregulated in seed development (Fig. 7a). Twenty-one genes in G. arboreum did not express in any tissues.
In the case of G. raimondii, the genes of groups 1, 2, 4, and 5 were highly expressed in seed 20 DPA, 10 DPA, 40 DPA and 30 DPA respectively. The genes of group 3 were highly expressed in seed at 40 DPA and 30 DPA. In the case of seed, the heat map of Gra-ERF genes was clustered in six pattern groups (Fig. 7b). As for as seed and ovule were concerned, most of the genes of groups 1 and 5 were highly expressed at 20DPA whereas, 2 and 4 groups exhibited higher expression at 30 DPA, additionally, groups 3 and 6 demonstrated high expression at 40 and 10 DPA respectively. The genes of groups 1, 2, 3, 4, and 5 were highly expressed in fiber 10DPA, fiber 20DPA, ovule 1DPA, 0DPA, and 3DPA respectively. Similarly, Gar-ERF-7D.12 exhibited relatively up-regulation in seed, ovule, and fiber development (Fig. 7b). Nineteen genes have null expression across all the tissues.
The expression profile of Gba-ERF genes in ovule (0DPA, 1DPA, 3DPA, 5DPA, 10DPA, 20DPA) organs (root, stem, leaf epicalyx, receptacles), flower (sepal, petal, filament, anther), and fiber (10 DPA ovules, 20 DPA ovules, 25 DPA ovules) were investigated. The expression pattern of Gba-ERF genes in the ovule clustered the genes in 5 groups (Fig. 7c). The genes of groups 1, and 2 were expressed in ovule at 0DPA, and 20DPA respectively. The genes of group 3 were mostly expressed in 5DPA followed by 3DPA and 1DPA respectively. Most of the genes of group 4 were expressed at 10DPA. The genes of group 5 were highly expressed in 0DPA and 1DPA. The heat map of Gba-ERF genes in organs clustered the genes into 4 groups based on their expression pattern (Fig. 7c). The genes of 1st group were highly expressed in the receptacle and a few genes of 1st group in leaf epicalyx. Most of the genes of group 2 were highly expressed in the root. The genes of groups 3 and 4 were expressed in stem and leaf epicalyx respectively. In the case of flower and fiber the heat map also clustered the genes into 4 groups. In flower, the genes of groups 1, 2, 3 and 4 were highly expressed in petal, anther, sepal and filament respectively (Fig. 7c). Most of the genes of groups 1 and 3 were highly expressed in fiber at 25DPA but the genes Gba-ERF-5D.8, Gba-ERF-3A.4, Gba-ERF-7A.2, Gba-ERF-11D.13 from 1st group were expressed in fiber at 20 DPA. The genes of group 2 were expressed in fiber at 20 DPA. The genes of group 4 were highly expressed in fiber at 10 DPA (Fig. 7C). Specifically, Gba-ERF-2A.4 exhibited up-regulation in ovule development, fiber development, root, stem, petal, filament, and anther.
In G. hirsutum, the Ghi-ERF genes expression was examined in organs (root, stem, leaf, filaments, anther), ovule (10DPA, 15DPA, 20DPA, 25DPA), fiber (10DPA, 15DPA, 20DPA, 25DPA), salinity, heat, cold, and PEG stress for (1 h, 3 h, 6 h, 12 h and 24 h) respectively. Based on Ghi-ERF genes expression in organs (root, stem, leaf, filaments and anther), the genes were clustered in 5 groups (Fig. 8A). Most of the genes of 1, 2, 3, 4 and 5 were highly expressed in anther, filaments, stem, leaf and root respectively. An ovule and fiber, based on heat map Ghi-ERF genes were clustered in 4 expression pattern groups (Fig. 8A). For ovule, most of the genes of groups 1, 2, 3 and 4 were highly expressed at 25DPA, 20DPA, 10DPA, and 15DPA respectively. Most of the genes of groups 1, 2, 3 and 4 were highly expressed in fiber at 10DPA, 20DPA, 15DPA, and 25DPA respectively. Under salt stress, the heat map clustered the genes in 3 groups (Fig. 8A). Most of the Ghi-ERF genes of group 1 were highly expressed at NaCl 1 h, while some genes of group 1 were also expressed at NaCl 24 h. The Ghi-ERF genes of 2nd group were differentially expressed at NaCl 3 h, NaCl 6 h, and NaCl 12 h. Most of the genes of group 3 were highly expressed at NaCl 12 h. The heat map of Ghi-ERF genes in heat stress divides the genes into three groups. The genes of group 1 were highly expressed at 37 C 1 h. Most of the genes of 2nd and 3rd groups were highly expressed at 37 C 6 h and 37 C 3 h respectively (Fig. 8A). Some genes of group 3 were also expressed at 37 C in 24 h and 12 h. The expression of ERF genes under cold and PEG stress the heat map clustered these genes in 5 groups. In the case of cold stress, most of the genes of group 1 were expressed at 4C in 1 h and 12 h. The genes of groups 2, 3, 4 and 5 were highly expressed at 4C in 24 h, 3 h, 6 h and 24 h respectively. In the case of PEG stress, most of the genes of groups 1, 2, 3, 4 and 5 were highly expressed in 24 h, 6 h, 3 h, 12 h and 1 h respectively (Fig. 8A). Interestingly, Ghi-ERF-2D.6, Ghi-ERF-6D.1, Ghi-ERF-11D.5, Ghi-ERF-12D.13, and Ghi-ERF-7A.6 demonstrated comparatively up-regulation under various tissues.

Examination of the expression of ERF genes by qPCR
The quantitative measurement of mRNA expression in the leaf of cultivar Eagle-2 at different salt stress levels after regular intervals was performed. GAPDH was used as an internal control of gene normalization. In Fig. 8B, the genes Ghi-ERF-2D.6, Ghi-ERF-12D.13, Ghi-ERF-6D.1, Ghi-ERF-7A.6 and Ghi-ERF-11D.5 showed

Discussion
In plants, the (AP2/EREBP) superfamily is one of the important and largest transcription factor (TF) families. AP2/ERF domain specifies different members of the (AP2/EREBP) family. The domain consists of up to 60-70 amino acids [4]. These members have necessary roles in the growth and development of plants, and in responding to a variety of environmental stresses involving heat, drought, salinity, cold and various pathogen infections. The roles are exerted either by direct response to the stresses or by regulation of expression of target genes downstream [40]. Numerous studies have recognized different numbers of ERF family members in plants; like in soybean 120 ERF family members [41], cucumber 103 (Hu & Liu, 2011) tomato 85 [42], rice 139 [43] and 147 Arabidopsis [44] respectively. Arabidopsis, used as the model plant comprises 147 AP2/EREBP genes divided into 4 subfamilies which are; APETALA2 (AP2), Ethylene-Responsive Factor (ERF), Dehydration-Responsive Element Binding protein (DREB), and Related to ABI3/VP1 (RAV) subfamilies [45]. Subfamily members of ERF and DERB containing single conserved domain AP2 and making the largest groups in superfamily AP2/EREBP, perform important roles in various plant processes and variety of stress responses. The sequences containing one AP2 domain have the biggest number of members in the superfamily AP2/EREBP, which have been further divided into 2 major subfamilies; ERF subfamily and DREB subfamily [46].
Cotton is an important economic crop in the world providing the leading share of the world's natural textile fiber and a substantial amount of edible oil. Studies revealed the formation of the allotetraploid Gossypium species occurred about 1-1.5 million years back through a polyploidization event that involved two species; one being A-genome species called G. arboreum and the other was paternal D-genome species named G. raimondii [18]. WGD or polyploidy is being acknowledged as a source of evolution in plants, a significant force resulting in enormous silencing and removal of duplicated genes [47]. In this study, we identified 118, 120, 213, 220 ERF genes in G. arboreum, G. raimondii, G. barbadense and G. hirsutum respectively. The evolutionary tree showed that more ERF proteins appeared in pairs and clustered together with 2 Gba-ERF, 2 Ghi-ERF, 1 Gar-ERF and 1 Gra-ERF which supported the cotton species polyploidization event that occurred 1.5 million years ago [48].
The tetraploid cotton should contain ERF genes in a number equal to ERF genes contained in G. arboreum plus ERF genes in G. raimondii. Though the number of ERF genes actually in allotetraploids (G. barbadense 213 and G. hirsutum 220) appeared lesser than those in two diploids (G. arboreum 118 and G. raimondii 120), implying that the chromosomes doubling and then genome rapid sequence arrangement would produce gene loss of different degrees in polyploidization process [18]. Gene duplication is a significant source of evolution in the genome and turn genetic system by producing new gene subfamilies [49]. Polyploidy, segmental, and Tandem duplications chiefly help to generate new gene families [49]. ERF genes in G. arboreum, (13), G. barbadense (15), G. hirsutum (7) and G. raimondii, (13) have tandem duplications. Segmental duplications frequently take place in plants as most plants occur as diploidized polyploids [50]. In G. arboreum (92), G. hirsutum (208), G. barbadense (182), and G. raimondii (74) ERF genes have undergone segmental or WGD duplications.
There's a possibility of deletion of some of the preexisting genes or newly generated genes during the process of cotton evolution. Collinearity and chromosomal locations showed the significant role of segmental duplications in ERF genes expansion in cotton which was following the results reported in rice, B. napus, A. thaliana and other plants [10,42,51,52]. The synteny analysis results could be used to reveal the evolutional and functional connections among the cotton species. In these studies, 237, 105 duplication gene pairs were established between diploid G. arboreum and tetraploid G. barbadense, G. hirsutum correspondingly. And there were 393, 307 duplication gene-pairs established between diploid G. raimondii and tetraploid G. hirsutum, G. barbadense correspondingly. Our study analysis showed a higher homology between ERF genes of tetraploid and diploid cotton species. Some ERF genes did not find any orthologous gene pairs, which could be attributed to chromosomes rearrangement or fusion during the evolution [53,54].
Gene structure analysis plays a crucial role in revealing the function of genes [51]. Here, our results suggested that 13.56, 19.16, 37.55 and 13.64% genes possess introns in G. arboreum, G. raimondii, G. barbadense and G. hirsutum respectively. The ERF members demonstrated similar gene structures within the same group. Loss of introns occurred more rapidly in genes than intron acquisition after segmental duplication [55]. Also, some studies revealed that intron number and distribution were related to evolution in plants [56] in a way that introns possibly have been lost during evolution from ERF family genes in higher plants. Our [57][58][59]. Transcriptional output can be delayed due to long multiple introns, possibly causing suppression of genes expression in adverse conditions. Contrarily, the genes containing small or fewer introns might have efficient expression while responding to stress environments [60]. Therefore many intron-less ERF genes might react quickly to the external environment variations [51]. The transcription factors domains and motifs perform essential roles during transcriptional activity, proteins interaction, and DNA binding [61].
Here, a total of 12 conserved motifs in the G. arboreum, G. raimondii, G. barbadense and G. hirsutum for ERF genes were identified. Gene function diversity could be affected by different numbers and types of motifs present in ERF proteins, in all four species. Motif 1 and motif 2 in all ERF genes of cotton were highly conserved. The ERF family members, in general, shared similar structures of gene and motif compositions within the same group, which proposes the possibility of their similar roles in plant growth and development. In promoter sites, Cisregulatory elements of genes play key roles in plant stress responses [62]. Various cis-acting elements responsible for stresses, phytohormones, growth and development were discovered in the promoter sites of Ghi-ERFs, stipulating the potential role of Ghi-ERFs in regulating several responses to phytohormones, environmental stresses, and development. For example, (ABRE, CGTCA-motif, MYC, TGACG-motif, ERE, P-box, and TCA-element), (ARE, STRE, LTR, DRE1, DRE-core, TC-rich repeats, WRE3 and WUN-motif ) and (CAT-box, CCG TCC -motif, AACA-motif and GCN4-motif ), are associated to phytohormones, stresses, growth and development respectively [63][64][65][66][67]. Based on these results, we speculate that the Ghi-ERFs genes are probably involved in abiotic stress responses.
The study of expression patterns of genes is used to predict the function of genes [53]. In this study the ERF genes showed high expression in ovule (0DPA, 1DPA, 3DPA, 5DPA, 10DPA, 20DPA) organs (root, stem, leaf epicalyx, receptacles), flower (sepal, petal, filament, anther) and fiber (10 DPA, 20 DPA, 25 DPA) across the species, suggesting that these genes have key roles in growth and development of cotton plant. Moreover, the expression of cotton ERF in different tissues or diverse stages, showed that these genes could be more stable than those that only expressed in specific tissues or one stage of an organ. Most of the ERF genes in G. hirsutum were also expressed in salt and heat stress.
Ghi-ERF-2D.6, Ghi-ERF-12D.13, Ghi-ERF-6D.1, Ghi-ERF-7A. 6 and Ghi-ERF-11D.5 identified as candidate genes against salinity tolerance in upland cotton by RT-PCR. According to previous studies, some ERF genes were involved in various stresses responses in plants, such as high-salt, low-temperature, and drought stress [46,57]. Overall, the above findings provide a foundation to further investigate the potential function of cotton ERF genes. These analyses are not only helpful in selecting valuable candidate ERF genes for further functional studies but also has important implication for genetic improvement for agricultural production and stress tolerance in cotton crop.

Conclusions
In this study, a genome-wide analysis of cotton ERF genes was performed. The 671 ERF genes in four species were identified and classified in detail. Protein lengths, molecular weights, and theoretical isoelectric points of cotton ERF vary greatly. The evolutionary characteristics, expression patterns of ERF genes in various cotton organs and growth stages, and their response to abiotic stress were studied. The expression profile analysis suggested that the ERF gene family of upland cotton may be important in stress responses. We proved that Ghi-ERF-2D.6, Ghi-ERF-12D.13, Ghi-ERF-6D.1, Ghi-ERF-7A. 6 and Ghi-ERF-11D.5 are responsive to salt stress by qRT-PCR. These results will be helpful to understand the biological role of the ERF genes in cotton growth and development.