Dynamic Expression, Differential Regulation and Functional Diversity of the CNGC Family Genes in Cotton

Cyclic nucleotide-gated channels (CNGCs) constitute a family of non-selective cation channels that are primarily permeable to Ca2+ and activated by the direct binding of cyclic nucleotides (i.e., cAMP and cGMP) to mediate cellular signaling, both in animals and plants. Until now, our understanding of CNGCs in cotton (Gossypium spp.) remains poorly addressed. In the present study, we have identified 40, 41, 20, 20, and 20 CNGC genes in G. hirsutum, G. barbadense, G. herbaceum, G. arboreum, and G. raimondii, respectively, and demonstrated characteristics of the phylogenetic relationships, gene structures, chromosomal localization, gene duplication, and synteny. Further investigation of CNGC genes in G. hirsutum, named GhCNGC1-40, indicated that they are not only extensively expressed in various tissues and at different developmental stages, but also display diverse expression patterns in response to hormones (abscisic acid, salicylic acid, methyl jasmonate, ethylene), abiotic (salt stress) and biotic (Verticillium dahlia infection) stimuli, which conform with a variety of cis-acting regulatory elements residing in the promoter regions; moreover, a set of GhCNGCs are responsive to cAMP signaling during cotton fiber development. Protein–protein interactions supported the functional aspects of GhCNGCs in plant growth, development, and stress responses. Accordingly, the silencing of the homoeologous gene pair GhCNGC1&18 and GhCNGC12&31 impaired plant growth and development; however, GhCNGC1&18-silenced plants enhanced Verticillium wilt resistance and salt tolerance, whereas GhCNGC12&31-silenced plants had opposite effects. Together, these results unveiled the dynamic expression, differential regulation, and functional diversity of the CNGC family genes in cotton. The present work has laid the foundation for further studies and the utilization of CNGCs in cotton genetic improvement.


Introduction
Cyclic nucleotide-gated channels (CNGCs) are nonselective cation channels first identified in animals; they form heterotetrameric complexes consisting of two or three different types of subunits, and are opened by the direct binding of cyclic nucleotides (cNMPs; cAMP and cGMP) and modulated by Ca 2+ /calmodulin and phosphorylation; their strong permeability for Ca 2+ provides an intracellular Ca 2+ signal that is crucially important for both excitation and adaptation, and thus, for the channel's function to mediate light adaptation and chemosensation, as well as playing roles in neuronal pathfinding or synaptic plasticity [1,2]. Cyclic nucleotides and Ca 2+ are among the most well established intracellular second messenger molecules present in almost all living organisms [3]. The unique position of CNGCs as ligand-gated Ca 2+ -permeable channels suggests that they function at key sites where cNMPs and Ca 2+ signaling pathways interact [4]. In plants, calcium nuclear envelope [23]. In Arabidopsis thaliana, AtCNGC19 was localized in the vacuole membrane, while AtCNGC20 seemed to reside in the Golgi body and may target the vacuole membrane via co-expression with AtCNGC19 [34]. Interestingly, it was noted that a considerable portion (8/20) of the CNGC family in G. herbaceum was predicted to target chloroplasts, and such a case for the CNGC family has not been reported in other plant species. All five Gossypium species in this study have evolved independently in diverse geographic regions, and it was evident that the allotetraploid formation has preceded the speciation of G. arboreum and G. herbaceum [31][32][33]35]. The CNGC family of G. herbaceum, a species native to the semi-arid regions of sub-Saharan Africa and Arabia, may have acquired distinctive functions relevant to chloroplast activities during evolution.

Phylogenetic Relationships of the CNGC Family Members in Cotton
To obtain insights into the evolutionary relationships of CNGC family members in cotton species, a neighbor-joining phylogenetic tree was constructed, together with the CNGC family members in a representative dicot species Arabidopsis thaliana and a monocot species Oryza sativa. As shown in Figure 1, the cotton CNGCs were clearly divided into four groups (I, II, III, IV) and two distinguished subgroups (IV-A and IV-B) within group IV, conforming to the classification of the CNGC family in both Arabidopsis and rice, as reported previously [4,15]. These results confirmed the highly evolutionary conservation of the CNGC family in plants. The cotton CNGCs tended to cluster together with that from dicot species A. thaliana, rather than the monocot species; moreover, it was obvious that CNGCs from the five cotton species were more closely related. Thus, the CNGC family genes may have undergone apparent sequence divergence between dicot and monocot species, and they may acquire sequence differentiation between different species during evolution.
All three diploid cotton species (G. arboreum, G. herbaceum and G. raimondii), each containing 20 CNGC family members, have the same distribution pattern among the five phylogenetic groups, and Group IV-B contains 4 CNGC members for each species (Supplementary Figure S1); in contrast, Arabidopsis thaliana contains 20 CNGC genes, but Group IV-B only contains two members, indicating the diversification of the CNGC family between cotton and Arabidopsis. Both tetraploid species G. hirsutum and G. barbadense evolved from natural interspecific hybridization between two diploid species, resembling G. arboreum and G. raimondii [36,37]. Accordingly, the CNGC family members from the tetraploid tended to form 20 gene pairs from the A t -and D t -subgenome in a phylogenetic tree, exhibiting very high sequence similarity between the paired paralogous genes (Supplementary Figures S2 and S3). Similarly, the CNGC family members from G. hirsutum and G. barbadense largely formed 40 orthologous gene pairs (except GhCNGC6 versus GbCNGC40 and 41) of high sequence similarity in a phylogenetical tree (Supplementary Figure S4). These results were consistent with the notions that all the species of Gossypium genus originate from a common ancestor [30]; and that G. hirsutum and G. barbadense originated from a common allotetraploid ancestor and diverged recently (~0.4-0.6 million years ago) [31]. The CNGC family in the Gossypium genus is extremely conserved during evolution, with little expansion or contraction during the process of speciation and polyploidization, which may implicate an essential function of this protein family in cotton.

Chromosomal Localization, Gene Duplication, and Synteny Analysis of the CNGC Family Genes in Cotton
Chromosomal locations of the CNGC family genes in the cotton species are shown in Figure 2. The CNGC genes in the diploid species G. arboreum, G. herbaceum, and G. raimondii were located on a total of 10, 11 and 11 chromosomes, respectively, except that GaCNGC20 resides on an unassembled scaffold; in contrast, the CNGC genes in the tetraploid species G. hirsutum and G. barbadense were located on a total of 19 and 20 chromosomes, respectively, except seven genes (GhCNGC38, GhCNGC39, GhCNGC40, GbCNGC38, GbCNCG39, GbCNGC40 and GbCNGC41) on unassembled scaffolds. It was noted that CNGC genes Figure 1. Phylogenetic relationships of the CNGC family members in cotton. The amino acid se quences of cyclic nucleotide-gated channels (CNGCs) were aligned with ClustalW, and a consensu tree was generated by the neighbor-joining (NJ) method, with 1000 bootstraps using MEGA X soft ware [38]. The cotton CNGCs were clustered into four groups (I, II, III, IV) and two subgroups (IV A and IV-B), conforming to the classification of CNGCs in both Arabidopsis thaliana and Oryza sativ [4,15]. G. herbaceum, GheCNGC1-20; G. arboreum, GaCNGC1-20; G. raimondii, GrCNGC1-20; G. hir sutum, GhCNGC1-40; G. barbadense, GbCNGC1-41; A. thaliana, AtCNGC1-20; O. sativa, OsCNGC1 16.
1 Figure 2. Chromosome distribution of the CNGC family genes in different cotton species. The chromosomes of five Gossypium species are represented by vertical bars of different colors, and the chromosome numbers are indicated at the top of each bar. The tetraploid species G. hirsutum and G. barbadense have a very similar distribution pattern of the CNGC family genes. Among the three diploid species, the distribution patterns are more similar between G. arboreum and G. herbaceum, compared to G. raimondii.
Gene families commonly arise as a result of gene duplication events, mainly including tandem, segmental, and whole-genome duplications [39,40]. The analysis of duplication events identified 6,8,4,36, and 19 duplicated gene pairs in G. arboreum, G. herbaceum, G. raimondii, G. hirsutum and G. barbadense, respectively (Supplementary Table S7); most of them were determined as segmental duplications, except that five gene pairs (GhCNGC12 and GhCNGC13; GhCNGC31 and GhCNGC32; GbCNGC13 and GbCNGC14; GheCNGC14 and GheCNGC15; GrCNGC6 and GrCNGC7) were determined as tandem duplications using MCScanX [41]. The nonsynonymous (Ka) to synonymous (Ks) substitution rate ratio (Ka/Ks) was used to serve an estimator for selective pressure on DNA sequence evolution, wherein the Ka/Ks ratio = 1, >1 and <1 implies the genes under neutral selection, positive selection, and purifying or stabilizing selection, respectively [42,43]. Consequently, the Ka/Ks ratios of CNGC duplications in the cotton species averaged 0.1896 ± 0.1302 (standard deviation) and ranged from 0.0319 to 0.6001, indicating that the CNGC family expansion in cotton was subjected to purifying selection, which would lead to losses of redundant genes [44]. When the Ks values were used to calculate the approximate date of duplication events using the formula t = Ks/2r, assuming the neutral substitution rate r = 2.6 × 10 −9 substitutions/synonymous site/year for cotton [31], the divergence times of duplicated genes were estimated at least 72 million years ago (MYA) in the three diploid species (G. arboreum, G. herbaceum, G. raimondii) (Table S7). It is believed that the Gossypium genus originated from the paleo-hexaploidy of a eudicot progenitor and subsequent diversification, mainly derived from the events of diploid species divergence around 5~10 MYA and interspecific hybridization around 1~2 MYA [31][32][33]35]. Thus, the CNGC family expansion seemed to occur much earlier, likely involving the paleo-hexaploidization, than the diploid speciation divergence within Gossypium genus, which was well retained after allopolyploid formation. Based on a total of 338 orthologous CNGC pairs between the cotton species (Table S7), 97% of them had the Ka/Ks values < 1, confirming the importance of purifying selection for maintaining the functions of CNGC family genes during the evolution of these cotton species; nine gene pairs (GaCNGC18 and GheCNGC19; GhCNGC12 and GaCNGC14; GhCNGC12 and GheCNCG14; GhCNGC16 and GheCNCG19; GhCNGC19 and GrCNCG3; GbCNGC24 and GrCNGC19; GhCNGC2 and GbCNGC2; GhCNGC12 and GbCNGC13; GhCNGC33 and GbCNGC33) had the Ka/Ks ratios > 1, suggesting that they were subjected to positive selection for adaptive evolution.
The density plots of Ks distribution peaked at 0.1425 for the CNGC orthologs between G. arboreum and G. herbaceum, 0.0463 between G. arboreum and G. raimondii, and 0.0474 between G. herbaceum and G. raimondii, indicating a much larger evolutionary distance of the CNGC orthologs between the two cultivated diploid species, which might be attributed to an impact of domestication, except for the more variable A genome compared to D genome [32]; in contrast, similar peak positions of Ks distributions were observed for both G. hirsutum and G. barbadense relative to the three diploid species ( Figure 3A). However, the Ks distribution of the CNGC orthologs between the A t -subgenome of G. hirsutum and the genomes of G. arboreum and G. herbaceum peaked at 0.0134 and 0.0204, respectively; in contrast, the peaks appeared at 0.0206 and 0.0136 for the CNGC orthologs between the A t -subgenome of G. barbadense and the genomes of G. arboreum and G. herbaceum, respectively, whereas the peak positions were similar (0.0128 vs. 0.0108) for the CNGC orthologs between both the D t -subgenomes of G. hirsutum and G. barbadense and G. raimondii genome ( Figure 3B). It has been shown that the A t -and D t -subgenome may have undergone positive selection for fiber improvement and stress tolerance traits, respectively [32,33]. Thus, the CNGC family members may potentially have involved in the differentiation of fiber traits between G. hirsutum and G. barbadense. The Ks distribution of the CNGC paralogs between the A t -and D t -subgenome of G. hirsutum shared a similar peak position (0.0432 vs. 0.0522) with that between the A t -and D t -subgenome of G. barbadense, which agreed with the notion that G. hirsutum and G. barbadense originated from a common allotetraploid ancestor that diverged around 0.4~0.6 MYA and independently evolved in diverse geographic regions [31]; however, the Ks distribution of the CNGC paralogs between the A t -subgenomes of G. hirsutum and G. barbadense peaked at 0.0133, in comparison with the peak at 0.0062 between the two D t -subgenomes, supporting the finding that sequence divergence was more common in the A t subgenome than in the D t subgenome [32]. The phylogenetic mechanism of CNGC family among the five cotton species was further investigated by syntenic gene analysis. Among the three diploid species, 12, 21, and 22 CNGC gene pairs (Table S7) showed syntenic relationships between G. arboreum and G. herbaceum, G. arboreum and G. raimondii, G. herbaceum and G. raimondii, respectively. A total of 70 syntenic CNGC gene pairs (Table S7) were identified between the two tetraploid species G. hirsutum and G. barbadense. There were 45, 32, and 42 syntenic CNGC gene pairs between G. hirsutum and each of the diploid species G. arboreum, G. herbaceum and G. raimondii, respectively ( Figure 3D; Table S7); in contrast, there were 31, 31, and 32 syntenic gene pairs between G. barbadense and each of the diploid species G. arboreum, G. herbaceum and G. raimondii, respectively ( Figure 3E; Table S7). In general, the CNGC family genes showed very similar patterns of syntenic relationships between the two allotetraploid species and the three diploid species ( Figure 3D vs. Figure 3E), suggesting a highly conserved evolution of the CNGC family within the Gossypium genus.

Gene Structures of the CNGC Family Members in Cotton
Structural variations in genic regions and alterations in gene expression play critical roles for speciation and the evolutionary history of cotton species [31,35]. Two paralogs or orthologs were regarded as structurally divergent if they had different numbers of exons, or if they had the same number of exons, but the lengths of at least one pair of homologous exons were different [45]. The CNGC family genes in the cotton species were analyzed for structural diversity, per the exon-intron arrangement ( Figure 4). For the two tetraploid species, the CNGC family genes in both G. hirsutum and G. barbadense have an average of about 7 exons, ranging from 2 to 13 exons in each gene; however, the gene structures were more diversified in G. hirsutum than G. barbadense, as evidenced by the more variable exon/intron lengths ( Figure 4A vs. Figure 4B), as well as the larger coefficient of variation (CV; 27.6% vs. 22.2%) for exon numbers. For the three diploid species, the CNGC family genes in G. arboreum, G. herbaceum and G. raimondii each have an average of about 7, 6 and 7 exons, ranging from 6 to 12, 2 to 13, and 6 to 12 exons in each gene, respectively; the gene structures were more variable in G. herbaceum compared to G. arboreum and G. raimondii, per the exon/intron lengths ( Figure 4D vs. Figure 4C,E) and the CVs (38.7% vs. 17.3%, 18.2%) for exon numbers. Generally, the most closely related CNGC members in the phylogenetic tree had a more similar gene structure (Figure 4), underlying their functional similarity. Intriguingly, it was noted that the phylogenetic group IV-A of the CNGC family represented the orthologous genes featuring a length much longer than other CNGC genes in each species except G. hirsutum, which featured the much longer gene GhCNGC30, belonging to the phylogenetic group I ( Figure 4A). These CNGC genes of exceptionally long length may be potentially subjected to distinct regulation. It has been shown that paralogous genes, which are derived from gene duplication, and initially have identical sequences and functions, tend to diverge in regulatory and coding regions, which may result in shifts in the expression pattern and the acquisition of new functions [45]. Altogether, it was postulated that the greater gene structure variations of the CNGC family in G. hirsutum and G. herbaceum may be of significance for environmental adaptation during evolution and domestication.

Composition of Cis-Acting Elements in the Promoters of CNGC Genes in Upland Cotton
The Upland cotton, G. hirsutum, is the predominant species for cultivation and cotton production in the world, accounting for over 90% of commercial cotton production worldwide [31]. Thus, the CNGC family genes in this species were further analyzed in the following sections. To understand the regulatory mechanisms governing the expression of GhCNGCs, the conserved cis-acting regulatory DNA elements in the promoter region of 1000 bp upstream from the translation start site for each gene (except GhCNGC40 with an available region of 187 bp) were determined using the PLACE database [46]. Consequently, a variety of cis-acting elements were identified in the promoters of each CNGC gene ( , and abiotic stress (CATGTG; CANNTG; TAACTG). Thus, the GhCNGC genes may be under the regulatory control of phytohormones, developmental cues, and environmental conditions. It was noted that most of these genes contain more copies of cis-acting elements related to ethylene, biotic, abiotic stresses, and WRKY transcription factor, suggesting their predominant functions likely involving environmental responses. However, the CNGC family genes each showed distinct patterns, per the composition and copy number of cis-acting elements, and even the most closely related paralogous gene pairs in the phylogenetic tree did not display an identical pattern ( Figure 5), suggesting that the CNGC family members may be regulated with difference in varying degrees. In line with this notion, it was documented that the closest paralogs AtCNGC2 and AtCNGC4 of the CNGC family in Arabidopsis thaliana exhibited very similar functions using mutant plants, but also showed subtle differences in the gene-for-gene resistance response [47,48].

Composition of Cis-Acting Elements in the Promoters of CNGC Genes in Upland Cotton
The Upland cotton, G. hirsutum, is the predominant species for cultivation and cotton production in the world, accounting for over 90% of commercial cotton production worldwide [31]. Thus, the CNGC family genes in this species were further analyzed in the following sections. To understand the regulatory mechanisms governing the expression of However, the CNGC family genes each showed distinct patterns, per the composition and copy number of cis-acting elements, and even the most closely related paralogous gene pairs in the phylogenetic tree did not display an identical pattern ( Figure 5), suggesting that the CNGC family members may be regulated with difference in varying degrees. In line with this notion, it was documented that the closest paralogs AtCNGC2 and AtCNGC4 of the CNGC family in Arabidopsis thaliana exhibited very similar functions using mutant plants, but also showed subtle differences in the gene-for-gene resistance response [47,48]. .

Functional Protein Association Network of the CNGC Family in Upland Cotton
The regulatory mechanisms of CNGC family genes in Upland cotton were further explored by analyzing protein interactions using the STRING database at the highest confidence score [49], resulting in the functional protein association network ( Figure 6). A total of 27 GhCNGCs were shown to have a strong interaction with FLS2 (FLAGELLIN SENSING 2), which is an important regulatory receptor kinase at the plasma membrane to activate immune signaling [50]; 7 GhCNGCs (GhCNGC1, 14,15,18,27,34,35) interacted with RSTK (receptor serine/threonine kinases), which play a central role in signaling during pathogen recognition [51]; additionally, 4 GhCNGCs (GhCNGC7, 23, 33, 39) interacted with MOL (mildew resistance locus O), which modulates plant disease defense and cell death [52]. Thus, these results may suggest a prominent role of GhCNGCs in plant immune responses. However, both RSTK and MOL have been implicated in morphological and developmental control [51,52], whereas 4 GhCNGCs (GhCNGC4, 17, 25, 37) interacted with CLV2 (CLAVATA2), which is involved in the regulation of SAM (shoot apical meristem) and RAM (root apical meristem) maintenance, affects organ development, and functions in plant-microbe interactions [53]; additionally, GhCNGC14 and GhCNGC34 interacted with TAD3 (tRNA-specific adenosine deaminase) which is essential for embryo develop-ment, and impacts plant growth [54]. Clearly, GhCNGCs may be critical players during growth and development. Among other interaction partners of GhCNGCs were a set of transporter or exchanger proteins: CAT9 (CATIONIC AMINO ACID TRANSPORTER 9) mediates cellular nitrogen-dependent amino acid homeostasis [55]; SUC3 (sucrose transporter 3) conducts energy-dependent sucrose/maltose transport, and plays a role in the sucrose import into sink tissues, as well as in the generation of osmotic gradients [56,57]. Both PLT3 and PLT6 (polyol/monosaccharide transporters) are implicated in response to environmental stimuli [58], and PLT6 is induced upon endogenous cAMP elevation [7]; NHX4 (sodium hydrogen exchanger 4) is critical for the maintenance of cellular cation homeostasis, and contributes to growth and development, as well as mediating plant stress acclimation [59,60]; MHX1 (magnesium/proton exchanger 1) is a vacuolar transporter that is important for mediating the adequate homeostasis of several divalent metal cations (i.e., Mg 2+ , Zn 2+ , Fe 2+ ), which are required for many enzymatic reactions, and has been found to affect the proton homeostasis of cells and plant growth [61,62]; additionally, TIP3-2 (tonoplast intrinsic protein) is implicated in the transport of water and small neutral substrates such as urea, ammonia, and hydrogen peroxide (H 2 O 2 ), and acts to modulate the response to abscisic acid (ABA) and maintain seed longevity under the control of ABI3 (ABSCISIC ACID INSENSITIVE 3), as well as affecting seed dormancy and germination in response to stress [63,64]. No doubt, the linkages of GhCNGCs with the above genes of various transport activities may be of significance for modulating plant responses to genetic and environmental stimuli. Finally, GhCNGC11, 15, 29, and 35 interacted with GhCNGC14 and 34. Recent studies have clearly shown that plant CNGCs may form both homomeric and heteromeric channels via dynamic interactions [65][66][67]. Thus, dynamic interactions between GhCNGC family members may greatly contribute to their functional diversity and regulatory complexity.

Expression Profiles of the CNGC Family Genes during Growth and Development of Upland Cotton
The expression profiles of the CNGC family genes in Upland cotton were investigated across different tissues and developmental stages using the estimates of FPKM (fragments per kilobase of exon model per million mapped fragments) from transcriptome sequencing data [31]. As shown in Figure 7A, these genes were expressed at varying levels across different tissues (root, stem, leaf, anther, petal, pistil, filament, bract, sepal, torus, ovule, and fiber) and developmental stages (−3 to 25 days post anthesis (DPA)); of them, GhCNGC1, 9,13,17,18,21,28,30,32,37 and 40 showed higher expression levels in most tissues and at various developmental stages. Thus, GhCNGCs may function during the life cycle. These genes exhibited varying expression patterns between them, which may suggest functional differentiation. Generally, most of the highly expressed GhCNGCs showed higher levels in reproductive tissues than vegetative tissues, implicating that GhCNGCs may be critical to cotton reproduction. For example, GhCNGC1, 13, 18, and 32 were expressed most highly in ovules (0, 1, 20 DPA), anther, filament, petal and sepal; their closest homologs are AtCNGC2 and AtCNGC4 in Arabidopsis (Figure 1), which play important functions in the regulation of floral transition [66,68], pollen growth [69], thermal sensing and acquired thermotolerance [66], senescence and programmed cell death [70][71][72], and innate immune response [65,73]. Interestingly, GhCNGC17, 33, 37, and 39 increased expression levels during the developmental stage from 10 to 25 DPA ( Figure 7A), likely contributing to fiber development. Cotton fibers are highly elongated and thickened single seed epidermal cells, resembling the tip-growing cells, such as pollen tubes and root hairs [74]. In Arabidopsis, AtCNGC5, the closest homolog of GhCNGC17 and 37, is essential for constitutive root hair growth [75]; in contrast, AtCNGC16, the closest homolog of GhCNGC33 and 39, is critical for pollen tube growth and fertility, and has been suggested to specifically impact cell walls or membrane dynamics [76].

Expression Profiles of the CNGC Family Genes during Growth and Development of Upland Cotton
The expression profiles of the CNGC family genes in Upland cotton were investigated across different tissues and developmental stages using the estimates of FPKM (fragments per kilobase of exon model per million mapped fragments) from transcriptome sequencing data [31]. As shown in Figure 7A, these genes were expressed at varying levels across different tissues (root, stem, leaf, anther, petal, pistil, filament, bract, sepal, torus, ovule, and fiber) and developmental stages (−3 to 25 days post anthesis (DPA)); of them, GhCNGC1, 9,13,17,18,21,28,30,32,37 and 40 showed higher expression levels in most tissues and at various developmental stages. Thus, GhCNGCs may function during the life cycle. These genes exhibited varying expression patterns between them, which may We performed quantitative RT-PCR (qRT-PCR) to confirm the expression of GhCNGCs using leaf and root tissues ( Figure 7B). The results indicated the relative higher expression of GhCNGC1, 10,18,20,22,30,34,37 and 40 in leaves, whereas GhCNGC9, 10, 14, 15, 20, 21, 28, 29, 34, 37 and 40 exhibited higher expression levels in roots. The relative expression levels of GhCNGCs in both leaf and root tissues detected by qRT-PCR were largely in agreement with the FPKM estimates by transcriptome sequencing (Figure 7A), despite some differences of the tested tissues in genotypes, growth stages, and conditions. creased expression levels during the developmental stage from 10 to 25 DPA ( Figure 7A), likely contributing to fiber development. Cotton fibers are highly elongated and thickened single seed epidermal cells, resembling the tip-growing cells, such as pollen tubes and root hairs [74]. In Arabidopsis, AtCNGC5, the closest homolog of GhCNGC17 and 37, is essential for constitutive root hair growth [75]; in contrast, AtCNGC16, the closest homolog of GhCNGC33 and 39, is critical for pollen tube growth and fertility, and has been suggested to specifically impact cell walls or membrane dynamics [76].

Hormonal Control of GhCNGCs Expression during Seedling Growth of Upland Cotton
Given the above finding that the promoters of GhCNGCs contain a variety of cis-acting regulatory elements responding to various phytohormones (  Figure 8D). These results confirmed that many GhCNGCs were highly responsive to phytohormones; moreover, some of them (GhCNGC1, 8, 9, 14, 21, 36, 37, 38 and 40) responded to different phytohormones, which may suggest the important roles of these genes in the coordination of the hormone signaling network. Similarly, it has been previously reported that the CNGC family genes showed significant responses to exogenously applied hormones in rice [15] and wheat [20]. More recently, it was demonstrated that AtCNGC5, 6 and 9 are involved in the auxin signaling of root hairs in Arabidopsis [75]. hirsutum "0-153" were sprayed with water solution containing the phytohormone or mock control (CK), and the above-ground tissue samples were collected after 24 h treatment for quantitative RT-PCR detection. Relative expression levels were normalized to house-keeping gene GhUBQ7 and calculated using the 2 −∆∆Cq method. Data are mean ± SD (n = 3), two-tailed Student's t-test * p < 0.05.

Regulation of GhCNGCs Expression under Abiotic and Biotic Stresses in Upland Cotton
CNGCs have been found to play extensive functions in responses to salt, drought, cold, heat, and heavy metal stresses, as well as pathogen infection in Arabidopsis [25]. We investigated expression profiles of GhCNGCs under the treatment of salt stress (200 mM NaCl), indicating that almost half of them showed the significant alteration of expression levels, including the upregulation of GhCNGC25 and the downregulation of 17 genes . Three-week-old seedlings of G. hirsutum "0-153" were sprayed with water solution containing the phytohormone or mock control (CK), and the above-ground tissue samples were collected after 24 h treatment for quantitative RT-PCR detection. Relative expression levels were normalized to house-keeping gene GhUBQ7 and calculated using the 2 −∆∆Cq method. Data are mean ± SD (n = 3), two-tailed Student's t-test * p < 0.05.

Regulation of GhCNGCs Expression under Abiotic and Biotic Stresses in Upland Cotton
CNGCs have been found to play extensive functions in responses to salt, drought, cold, heat, and heavy metal stresses, as well as pathogen infection in Arabidopsis [25]. We investigated expression profiles of GhCNGCs under the treatment of salt stress (200 mM NaCl), indicating that almost half of them showed the significant alteration of expression levels, including the upregulation of GhCNGC25 and the downregulation of 17 genes (GhCNGC2, 4, 5, 6, 8, 11, 12, 13, 15, 16, 18, 20, 24, 27, 29, 30 and 35) (Figure 9A). GhCNGC4 and 25 are the paralogous gene pair from the A-and D-subgenome of G. hirsutum, respectively; they showed the closest homology with AtCNGC5 (Figure 1), which is required for constitutive root hair growth in Arabidopsis [75]; intriguingly, they were regulated in an opposite direction under salt stress. In Arabidopsis, both AtCNGC19 and 20 were induced in response to salt stress [77], whereas AtCNGC10 was dramatically inhibited after exposure to 200 mM NaCl, and it negatively regulated salt tolerance by mediating Na + transport [78]. GhC-NGC11, 15, 29 and 35 seemed to exhibit similar salt stress responses to their closest ortholog AtCNGC10; however, this phenomenon was not observed with GhCNGC14 and 34, which are the closest orthologs of AtCNGC19 and 20. Thus, the orthologous genes of the CNGC family members in different plant species may have evolved to play species-specific roles.
Expression profiles of GhCNGCs were also studied in young seedlings by inoculation with Verticillium dahliae strain Vd991, an isolate from G. hirsutum [79]. Consequently, the fungal infection caused the obvious suppression of GhCNGC6, 10, 15, 17, 20, 21, 24, 25 and 26. At least four members of the CNGCs family in Arabidopsis have been demonstrated to play critical roles in pathogen defense responses, including AtCNGC2, 4, 11 and 12. Loss-offunction mutants of these Arabidopsis genes displayed impaired hypersensitive response (HR), the constitutive expression of SA, changes in the expression pattern of pathogenesisrelated (PR) genes, and the alteration of plant responses to avirulent pathogens [27]. Like AtCNGC11 and 12, GhCNGC15 and 21 belong to group I in the phylogenetic tree (Figure 1), which may support similar functions between them in disease resistance; however, the corresponding G. hirsutum orthologous genes of AtCNGC2 and 4 did not show significant expression induction by Vd991 infection. constitutive root hair growth in Arabidopsis [75]; intriguingly, they were regulated in an opposite direction under salt stress. In Arabidopsis, both AtCNGC19 and 20 were induced in response to salt stress [77], whereas AtCNGC10 was dramatically inhibited after exposure to 200 mM NaCl, and it negatively regulated salt tolerance by mediating Na + transport [78]. GhCNGC11, 15, 29 and 35 seemed to exhibit similar salt stress responses to their closest ortholog AtCNGC10; however, this phenomenon was not observed with GhCNGC14 and 34, which are the closest orthologs of AtCNGC19 and 20. Thus, the orthologous genes of the CNGC family members in different plant species may have evolved to play species-specific roles.

Modulation of GhCNGCs Expression Associated with cAMP Signaling during Cotton Fiber Development
Cotton fiber is a unique elongated cell, and its development has been well known to involve sugar metabolism, hormones, secondary metabolites, and the cytoskeleton during ovule culture [80]. Very coincidentally, we recently found that cAMP signaling is predominantly linked with these biological processes in Arabidopsis [7]. Given the notion that CNGCs are directly regulated by cAMP [4,25], we wonder if cAMP signaling might modulate GhCNGCs expression during ovule culture. For this, we tested the effects of a set of commonly used drugs for cAMP signaling pathway studies, including a specific inhibitor (2 ,3 -dideoxyadenosine; DDA) or activator (forskolin) of adenylate cyclase, as well as the membrane permeable cAMP analog (8-Br-cAMP). Adenylate cyclase is the sole enzyme responsible for the cellular production of cAMP, which stimulates cAMPdependent protein kinase A (PKA) to phosphorylate cAMP response-element bindingprotein (CREB), and subsequently activates the transcription of a variety of target genes, resulting in multiple physiological functions [81]. Both DDA and forskolin are commonly used in biological process or pathway studies involving adenylyl cyclase activity and cAMP pool modulation, whereas 8-Br-cAMP is often used to imitate cAMP activation agents, both in plants and animals [73,[82][83][84]. Based on our RNA sequencing results, 13 GhCNGCs showed differential expression at the threshold of an absolute value of log2(fold change) > 1 and adjusted p < 0.05 under treatments of DDA or forskolin, or 8-Br-cAMP during ovule culture (Table 1), which included GhCNGC39 being the closest homolog of AtCNGC16 that has been found to be critical for pollen tube growth by specifically impacting cell walls or membrane dynamics [76]. Basically, the DDA-mediated inhibition of adenylate cyclase activities caused the suppression of most GhCNGCs, except the upregulation of GhCNGC5 and 16 under condition of higher concentration (100 µM); of them, the paralogous gene pair GhCNGC1 and 18 were all significantly down-regulated in a dose-dependent manner (Table 1), highlighting a prominent role during fiber development. In contrast, the forskolin-mediated activation of adenylate cyclase activities seemed to up-regulate most GhCNGCs in a general trend, but only GhCNGC5 and GhCNGC26 showed significant up-regulation at the above artificially set criteria. These results may be evident for the modulation of GhCNGCs mediated by adenylate cyclase activities; however, the exogenous application of 8-Br-cAMP seemed to have no significant impact on GhCNGCs during ovule culture. Interestingly, recent advances supported the notion that adenylate cyclase activities in higher plants are embedded in multidomain proteins which usually have distinctive functions in development and environmental responses [85]. For examples, both AtKUP5 and AtKUP7 contain adenylate cyclase activities for the production of cAMP, but they are essential genes for K + transport in Arabidopsis [86,87]; AtLRRAC1 possesses adenylate cyclase activity and is a leucine-rich repeat (LRR) protein implicated in immune response [88]. Thus, plant adenylate cyclase activities and cAMP production seemed to be tightly coupled with other distinctive functions for development and environmental adaptation, which may pose a great challenge to elucidating cAMP signaling in plants. and indicate significant up-and down-regulation in comparison with the mock control (CK), respectively; ↑ and ↓ indicate significant up-and down-regulation compared to the treatments of lower concentrations, respectively.

Functional Characterization of GhCNGC1&18 and GhCNGC12&31 in Upland Cotton
Gossypium hirsutum represents a true allotetraploid species that evolved from natural interspecific hybridization between the A-and D-genome diploid species in the New World [31,33,35]. GhCNGCs from the A-and D-subgenome of G. hirsutum formed about 20 gene pairs of very high sequence similarity in a phylogenetic tree (Supplementary Figure S2), underlying redundant functions between the paired genes. Thus, the tobacco rattle virus (TRV)-mediated virus-induced gene silencing (VIGS) assay, a robust and effective reverse genetic tool commonly used in cotton [89], was performed to simultaneously knock down the paired genes for functional analysis. For this, the silencing target fragments were designed to specifically disrupt the expression of the paired genes in the G. hirsutum genome. GhCNGC1&18 are the paired genes sharing closest homology with AtCNGC4 in Arabidopsis (Figure 1), and we obtained the expression levels reduced by 83% for GhCNGC1 and 79% for GhCNGC18 in the silenced plants (TRV::GhCNGC1&18) compared to mock control plants (TRV::00) ( Figure 10A). The GhCNGC1&18-silenced plants grew smaller leaves with darker colors and a downward curled margin, as well as displaying stunted growth phenotype ( Figure 10B). When subjected to salt stress treatment by growing in a deionized water solution containing 200 mM NaCl, GhCNGC1&18-silenced plants were able to maintain a largely normal growth phenotype after 6 days' treatment, whereas mock control plants were severely withered ( Figure 10C), suggesting that GhCNGC1&18 are negatively implicated in salt stress resistance. When infected with Verticillium dahliae strain Vd991, GhCNGC1&18-silenced plants remained an almost healthy status at 16 days post-inoculation (dpi), whereas mock control plants developed obvious yellowing leaves, representing a typical Verticillium wilt symptom ( Figure 10D); by cutting the stems of these plants to examine the vascular wilt symptoms, it was found that the extent of vascular browning was much weaker in GhCNGC1&18-silenced plants than control plants ( Figure 10E); moreover, the fungal biomass analysis of the stem tissues indicated that GhCNGC1&18-silenced plants developed significantly lower fungal biomass than mock control plants ( Figure 10F). These results clearly showed that silencing GhCNGC1&18 resulted in an enhanced resistance to fungal infection in Upland cotton. To further confirm GhCNGC1&18-mediated Verticillium wilt resistance, the activation of pathogenesis-related (PR) genes was investigated, which indicated that both the SA pathway marker gene PR1 and the JA (jasmonic acid) pathway marker gene PR3 were significantly upregulated in the GhCNGC1&18-silenced plants upon Vd991 inoculation ( Figure 10G). It was noted that, under the mock control condition without infection, expression levels of both PR1 and PR3 genes in the GhCNGC1&18-silenced plants were significantly higher than those of the non-silenced control plants ( Figure 10G), suggesting that the silencing of GhCNGC1&18 caused the constitutive activation of PR genes, which are in agreement with those findings from mutants of orthologous CNGCs in Arabidopsis [48,68]. Altogether, the above results indicated that GhCNGC1&18 are essential for growth and development, and they play negative roles during abiotic and biotic stress responses in Upland cotton.
Similarly, we characterized functions of GhCNGC12&31 as the paired genes sharing the closest homology with AtCNGC2 in Arabidopsis (Figure 1). Expression of GhCNGC12 was not detectable in leaf samples, but GhCNGC31 transcripts decreased by 89% more in GhC-NGC12&31-silenced plants than in mock control plants ( Figure 11A). Given that GhCNGC12 was clearly expressed at varying levels during ovule/fiber development ( Figure 7A), we cannot completely rule out the possibility that GhCNGC12 may be expressed in other tissues (e.g., apical meristems) or induced upon stimulation, or even expressed at a barely detectable level by quantitative RT-PCR. GhCNGC12&31-silenced plants displayed few changes in phenotypes, except that the leaves appeared to have darker colors at an earlier growth stage ( Figure 11B). Under the condition of salt stress, GhCNGC12&31-silenced plants started to exhibit obvious wilting symptoms after 43 h treatment, whereas mock control plants remained normal ( Figure 11C), suggesting that GhCNGC12&31 play positive roles during salt stress. Under the condition of Vd991 infection, GhCNGC12&31-silenced plants developed more severe Verticillium wilt symptoms (i.e., yellowing leaves, defoliation, and wilting) than mock control plants at 20 dpi ( Figure 11D); the extent of vascular browning was much stronger in the stems of GhCNGC12&31-silenced plants than mock control plants at 24 dpi ( Figure 11E); accordingly, fungal biomass in the stems of GhCNGC12&31-silenced plants was significantly higher than that in mock control plants ( Figure 11F). These results supported that GhCNGC12&31 positively contribute to Verticillium wilt disease resistance. A further detection of PR genes confirmed the significant elevation of both PR1 and PR3 genes in the GhCNGC12&31-silenced plants upon Vd991 inoculation; under mock control condition without infection, the GhCNGC12&31-silenced plants showed a significantly higher level of PR1 expression compared to the non-silenced control plants, whereas PR3 was increased without significant difference ( Figure 11G). A comparison between GhC-NGC12&31and GhCNGC1&18-silenced plants indicated that Vd991 infection induced a similar level of both PR1 and PR3 expression in the GhCNGC12&31-silenced plants compared to that in the infected non-silenced control plants ( Figure 11G); in contrast, Vd991 infection induced the highly significant elevation of both PR1 and PR3 expression in the GhCNGC1&18-silenced plants compared to that in the infected non-silenced control plants  In Arabidopsis, AtCNGC2 and AtCNGC4 are the only two members comprising the IV-B group of the CNGC family, and null mutants of both genes exhibited almost identical phenotypes, including severe dwarfing in stature, delayed flowering, loss of hypersensitive response (HR) cell death, and constitutive systemic resistance [48,68,90,91]; in addition, it was evident that both AtCNGC2 and AtCNGC4 may work in the same signaling pathway to regulate pathogen defense and floral transition [66]. While our above results demonstrated that GhCNGC12&31 and GhCNGC1&18 are required for growth and development, and they do cause similar leaf phenotypes, as well as the constitutive activation of PR genes by gene silencing; however, GhCNGC12&31 and GhCNGC1&18 played opposite roles during abiotic and biotic stress responses in Upland cotton. It was documented that the plant-pathogen interaction pathway played important roles in cotton defense response to V. dahliae infection, and that genes encoding the RLKs (receptor-like protein kinases) family members, including GhFLS2 (LRR receptor-like serine/threonine-protein kinase) and GhGsSRK (G-type lectin S-receptor-like serine/threonine-protein kinase), are highly up-regulated upon V. dahliae infection; however, VIGS experiments showed that GhGsSRKsilenced plants exhibited more severe symptoms compared with the vector control plants, whereas GhFLS2-silenced plants did not compromise cotton resistance to V. dahlia [92]. FLS2 has the most functional connection with GhCNGCs ( Figure 6), which may be of significant importance to the coordinated regulation or complexity of pathogen responses in plants.
Our results suggested that the closest orthologs of CNGCs in different plant species may play different roles in the specific genome background, which indicate the necessity of addressing the functions of CNGCs in a specific plant species for their potential application in breeding and improvement.

Plant Materials and Growth Conditions
Upland cotton cultivars "0-153" and "Han 8266" were used in the present study. Moreover, "0-153" is derived from introgressive hybridization between G. hirsutum ("Damian 2") and G. arboreum ("Jinxian zhongmian"), and it has excellent fiber quality traits (an average fiber strength of 33.70 cN/tex, fiber length of 30.28 mm, and micronaire of 4.52). "Han 8266" is a commercial transgenic cultivar developed through the cross between a conventional variety "Han 4849" and a Bt (Bacillus thuringiensis) transgenic insect-resistant variety "Han 5158", and it has high-yielding potential and wide adaptability, with resistance to insect and tolerance to Fusarium wilt (FW index 6.3~11.8) and Verticillium wilt (VW index 18.0~21.0). Cotton seeds were grown in potting soil (Pindstrup Mosebrug A/S, Ryomgård, Denmark) mixed with 20% (v/v) vermiculite, or grown hydroponically in Hoagland's nutrient solution, at 25 • C, with 16 h light/8 h dark photoperiod in a growth room, except that the samples for cotton ovule culture were collected from "0-153" grown in an experimental field at Zhengzhou Research Base, State Key Laboratory of Cotton Biology, Zhengzhou, China.
All cotton CNGC genes were named according to their positions on the chromosomes in the genome. Protein isoelectric point (pI) and number of amino acids (aa) were calculated using ExPasy (https://web.expasy.org/ (accessed on 27 May 2020)). The protein subcellular location was predicted using WoLF PSORT (https://wolfpsort.hgc.jp/ (accessed on 28 May 2020)) [97].

Chromosomal Localization, Gene Duplication, and Phylogenetic Analysis
The chromosomal locations of cotton CNGC genes were determined according to the genome annotation data; the positions and relative distances on the chromosomes were visualized using TBtools software [98]. Synteny, collinearity, and gene duplication were analyzed using MCScanX (http://chibba.pgml.uga.edu/mcscan2/ (accessed on 6 June 2021)) [41]. Gene duplication was determined according to two criteria: (a) the length of the shorter aligned sequence covered > 70% of the longer sequence, and (b) the two aligned sequences shared > 70% amino acid sequence similarity [99]. Two genes separated by less than five intermediate genes in the 100 kb chromosomal fragment are considered to have undergone tandem duplication [100]. To detect the mode of selection forces acting on the protein, the ratio of the number of nonsynonymous substitutions per nonsynonymous site (Ka) to the number of synonymous substitutions per synonymous site (Ks) was calculated using TBtools with the Nei-Gojobori model [101]; generally, a Ka/Ks ratio > 1 indicates positive selection, the ratio < 1 implying negative or purifying selection, while the ratio = 1 indicates neutral evolution [43]. The density plots of Ks were analyzed and visualized using the lattice package in RStudio [102]. The estimated divergence time (T) of each duplicated gene pair was calculated as T = Ks/2r, in accordance with the neutral substitution rate of cotton (r = 2.6 × 10 −9 ) [31]. The synteny relationship of orthologous CNGC genes in different species was constructed using the Dual Synteny Plotter (https://github.com/CJ-Chen/TBtools (accessed on 8 June 2021)) [103], and the results were visualized and optimized via RCircos [104]. The phylogenetic analysis was performed with the full-length amino acid sequences of CNGCs using MEGA X software (http://www.megasoftware.net/ (accessed on 4 July 2020)), wherein multiple sequence alignment was conducted using the ClustalW and the phylogenetic tree was generated using the neighbor-joining (NJ) method with 1000 bootstraps.

Gene Structure, Cis-Acting Regulatory DNA Elements and Protein Interaction Network
For the analysis of gene structure, intron/exon structure information was collected from the annotations of cotton genomes and analyzed using GSDS2.0 (http://gsds.cbi.pku. edu.cn/ (accessed on 2 September 2021)). To identify cis-acting regulatory DNA elements, the 1000 bp promoter sequence upstream from translation start site of the CNGC gene was analyzed using the PLACE database (http://dna.affrc.go.jp/PLACE/?action=newplace/ (accessed on 7 September 2021)). To construct a functional protein association network, protein-protein interactions were determined using STRING (v11.0; https://string-db.org/ cgi/input.pl (accessed on 19 September 2021)) with the highest confidence score >0.9, and the results were depicted by a network using Cytoscape (v3.8.2; https://cytoscape.org/ (accessed on 20 September 2021)).

Plant Treatments with Hormones and Stress Conditions
For hormonal treatment, 3-week-old seedlings (G. hirsutum "0-153") were sprayed with a sterile distilled water solution containing salicylic acid (SA), methyl jasmonate (MeJA), 1-aminocyclopropane-1-carboxylic acid (ACC), or abscisic acid (ABA) at the indicated concentration, and then immediately covered overnight using a transparent dome; the above-ground tissue samples were collected 24 h after spraying, along with the mock control. For fungal infection, the stock of Verticillium dahliae stain Vd991 was first activated by growth on PDA (potato dextrose agar) for 5-7 days at 25 • C under the dark condition; then, the mycelia were collected and cultured in liquid Czapek's medium with shaking (150 rpm); finally, conidia were harvested by centrifugation, washed with sterile water, and adjusted to a final concentration of 1 × 10 7 conidia/mL sterile water using a hemocytometer. Three-week-old seedlings (G. hirsutum "Han 8266") grown in a 9-ounce paper cup were inoculated by root dipping in 30 mL conidial suspension per seedling plant, and whole plant samples were collected 3 days post-inoculation (dpi). For salt stress treatment, cotton seeds (G. hirsutum "0-153") were germinated for 3 days, and then grown hydroponically in Hoagland's nutrient solution containing 200 mM NaCl for 2 weeks before the collection of whole plant samples.

VIGS Analysis and Detection of Stress Resistance
An agrobacterium-mediated virus-induced gene silencing (VIGS) assay was performed using the tobacco rattle virus (TRV) system [89,105]. Specifically, VIGS target fragments of GhCNGCs were determined to ensure the specificity by BLAST search against G. hirsutum genome sequence, and approximately 400-500 bp fragments were amplified from G. hirsutum genomic DNA using gene-specific primers (Supplementary Table S1). Each of the fragments was cloned into vector pTRV2, generating pTRV2 derivatives (pTRV2::GhCNGCs), which were subsequently introduced into Agrobacterium tumefaciens GV3101. For achieving a high silencing efficiency, two target fragments from different regions of each gene were separately cloned to generate the derived Agrobacterium cultures, which were used by a mixture of equal proportions. Agrobacterium was cultured in LB medium containing 50 mg/L kanamycin and 20 mg/L rifampicin overnight at 28 • C in a shaker (200 rpm), and then sub-cultured in the same fresh medium overnight (OD 600 = 0.8~1.2). Cells were pelleted by centrifugation at 4000× g for 10 min, resuspended with OD 600 of 1.0 in infiltration buffer containing 10 mM MgCl 2 , 10 mM 2-(N-morpholino)-ethanesulfonic acid (MES), and 200 µM acetosyringone. Cell suspensions of Agrobacterium carrying pTRV2 or pTRV2 derivatives were mixed with pTRV1 in a 1:1 ratio, and incubated at room temperature for at least 3 h. Cotton seedlings with fully expanded cotyledons were used to infiltrate the cotyledons with the mixed culture using a 1 mL needleless syringe. Immediately after infiltration, plants were watered, covered with a plastic dome, and shaded from light using a piece of black plastic cloth overnight. The effectiveness of the VIGS assay was evaluated by silencing the phytoene desaturase gene (PDS) as a positive control, resulting in visible leaf photo-bleaching [106]. When PDS-silenced plants displayed white leaves at approximately 2 weeks post-infiltration, experiments for the evaluation of stress resistance were conducted. The second true leaf samples were collected for RNA extraction and interference efficiency detection by quantitative RT-PCR through the comparison of gene expression levels in the silenced plants and mock control plants. All the experiments were conducted at least three times independently.
For the evaluation of salt stress resistance, VIGS-treated plants at the three leaves stage were transplanted with roots soaking in a water solution containing 200 mM NaCl, as reported previously [107]. For examining fungal disease resistance, VIGS-treated plants, for about two to three weeks (i.e., after two true leaves fully-expanded), were inoculated with Vd991 as described above, and Verticillium wilt symptoms were investigated; seedling shoots were cut to investigate vascular wilt symptoms under a microscope [108]; relative fungal biomass in the stem tissues was quantified by quantitative PCR using primers specific to the internal transcribed spacer region (ITS) of 5.8S ribosomal RNA gene in V. dahliae and GhUBQ7 (GenBank: DQ116441) gene in G. hirsutum for sample equilibration (Supplementary Table S1), as we have described previously [7].

Quantitative RT-PCR
Total RNA was isolated using the EASYspin Plus Plant RNA Kit (Aidlab, Beijing, China) following manufacturer's instructions. Reverse transcription (RT) was performed using the HIScript ® III RT SuperMix for qPCR (+gDNA wiper) Reagent Kit (Vazyme Biotech, Nanjing, China). A quantitative PCR analysis was carried out using the ChamQ TM Universal SYBR ® qPCR Master Mix (Vazyme Biotech) and gene-specific primers (Supplementary Table S1) in a LightCycler ® 480II PCR system (Roche, Basel, Switzerland). The cycling conditions were 30 s at 95 • C, 40 cycles of 10 s at 95 • C, and 30 s at 60 • C. The specificity of amplified products was monitored by melting curve analysis, and verified by agarose gel electrophoresis. Relative expression levels of genes were normalized to GhUBQ7 as an internal control and calculated using the 2 −∆∆Cq method. When necessary, the house-keeping gene GhACT7 was used for plate-to-plate equilibration.

Ovule Culture, Drug Treatments and Transcriptomic Profiling
The chemicals forskolin (FSK), 8-Br-cAMP, and 2 ,3 -dideoxyadenosine (DDA) were purchased from Sigma-Aldrich, Shanghai, China. Then, 8-Br-cAMP was solubilized as a 50 mM stock in water and filter sterilized; FSK and DDA were prepared as 50 mM and 200 mM stock in DMSO, respectively. The stock solutions were aliquoted and stored at −80 • C.
Cotton ovules culture was conducted as previously described with minor modification [109]. Briefly, the bolls at the stages of 0 or 1 day post anthesis (DPA) were collected with anthocaulus from cotton plants (G. hirsutum "0-153"); ovaries were surface-sterilized and dissected under sterile conditions; intact ovules were immediately placed into liquid Beasley and Ting (BT) medium containing 18 g/L glucose, 3.6 g/L fructose, 5 µM indole-3-acetic acid (IAA) and 0.5 µM GA 3 , and then incubated at 32 • C in the dark without agitation. The fiber should be easily visible after 4~5 days of culture. For drug treatments, DDA (50 or 100 µM), FSK (10 or 50 µM), or 8-Br-cAMP (10 or 50 µM) were added with the indicated final concentrations in the medium. After 6 days of culture, the ovules were harvested for RNA extraction. Three biological replicates were performed.
Total RNAs were isolated using the EASYspin Plus Plant RNA Kit, as described above. Preparation of sequencing libraries, instrumental platform, raw reads cleaning, transcripts assembly and alignment, annotation and quantification were all performed as we previously described [7], except that the reference genome of G. hirsutum 'TM-1' (NAU-NBI_v1.1_a1.1; https://www.cottongen.org/ (accessed on 20 September 2020)) was used. FPKM (fragments per kilobase of exon model per million mapped fragments) for each gene model were calculated for an estimate of expression level. Differential gene expression was determined with an absolute value of log2 (fold change) >1 and a false discovery rate (FDR) adjusted p-value < 0.05.
RNA sequencing data of G. hirsutum 'TM-1' during growth and development were downloaded from the NCBI Sequence Read Archive under accession number PRJNA490626 [31], and used for analyzing gene expression patterns.

Statistical Analysis
Data for quantification analyses were presented as mean ± standard deviation (SD) with three biological replicates. Student's t test or analysis of variance (ANOVA) followed by Tukey's multiple comparisons test were performed to determine significant differences between the means using GraphPad Prism 8.0.2 (https://www.graphpad.com/ (accessed on 14 October 2021)).

Conclusions
In the present study, genome-wide analyses have identified a total of 40, 41, 20, 20, and 20 CNGC genes in G. hirsutum, G. barbadense, G. herbaceum, G. arboreum, and G. raimondii, respectively. These genes are highly conserved during evolution across all the five cotton species above, and they are classified into five phylogenetic groups (I, II, III, IV-A and IV-B), conforming to the CNGC family in other plant species. Most members of the CNGC family in cotton are localized to plasma membrane, except some of them residing in the nucleus, mitochondria, and chloroplasts; however, a considerable portion (8/20) of the CNGC family in G. herbaceum may target chloroplasts. The CNGC genes are distributed on most chromosomes in the cotton genome, with a highly similar pattern across different species, and the family expansion is mainly derived from segmental duplication under purifying selection, and show very similar patterns of syntenic relationships between the species. Generally, the most closely related CNGC family members in the phylogenetic tree tend to have a more similar gene structure, while the CNGC family in G. hirsutum and G. herbaceum seems to display greater gene structure variations among the five cotton species. Further analyses of the CNGC family genes in G. hirsutum confirmed that they are extensively expressed in various tissues, and at different developmental stages. Each GhCNGC gene contains a variety of cis-acting elements residing in the promoter regions, which are implicated in response to phytohormones, biotic and abiotic stimuli; however, the CNGC family genes each showed distinct patterns, per the composition and copy number of cis-acting elements. Accordingly, quantitative RT-PCR detection unveiled diverse and altered expression patterns of GhCNGCs upon treatments of hormones (ABA, SA, MeJA and ethylene), salt stress, and V. dahlia infection in cotton plants; additionally, a set of GhCNGCs were identified in response to cAMP signaling during cotton fiber development. A functional protein association network of the CNGC Family in G. hirsutum was established, demonstrating the linkages of GhCNGCs with a few crucial proteins in plant immune responses (FLS2, RSTK and MOL), growth and development (CLV2 and TAD3), as well as with various transporter and exchanger proteins implicated in response to genetic, hormonal, and environmental cues, in addition to the dynamic interactions between different GhCNGCs. The silencing of both the homoeologous gene pair GhCNGC1&18 and GhCNGC12&31 impaired plant growth and development; however, GhCNGC1&18-silenced plants enhanced Verticillium wilt resistance and salt tolerance, whereas GhCNGC12&31-silenced plants showed the opposite effects. Collectively, these findings enrich our knowledge on the CNGCs family and its association with cAMP signaling, about which almost nothing is known currently in cotton, and thus pave the foundation for elaborating the biological functions and utilization of CNGCs in cotton breeding and genetic improvement.