Arabidopsis thaliana ecotype Col-0 (WT) served as both the wild-type control and transgenic recipient for GhPI4P5K-D04–2 overexpression studies. Arabidopsis seeds were surface-sterilized with 70% ethanol for 2 min followed by 5% sodium hypochlorite (NaClO) treatment for 15 min, then rinsed thoroughly with sterile water. Three independent overexpression lines were generated through Agrobacterium tumefaciens-mediated floral dipping of flower buds at the rosette stage. Seeds from WT and transgenic lines were stratified on MS solid medium containing 0 mM, 125 mM NaCl under 4 °C vernalization for 2 days to synchronize germination. Germinated seedlings were transferred to a controlled growth chamber maintained at 22 °C with 16 h light/8 h dark photoperiod and 60% relative humidity for phenotypic analysis.

Taking the cotton (Gossypium hirsutum L.) line CCRI24 obtained from the Cotton Research Institute of the Chinese Academy of Agricultural Sciences as the research object. To analyze the expression of GhPI4P5K-D04–2 after treatment with NaCl (150 mM), cotton seeds were wrapped in wet filter paper and placed vertically in a water basin containing 1L of water. Place the water basin in a 25 °C incubator under 16 h light/8 h dark photoperiods. After the cotton seeds germinate, they are transferred to Hoagland’s solution for growth until the three-leaf stage. Then, leaf samples were collected after processing 0, 1, 3, 6, 12, and 24 hours under high salt (150 mM NaCl) conditions. The samples are frozen in liquid nitrogen and stored at -80 °C for RNA isolation and cDNA preparation.

The genomic data for G. arboretum (G. arboreum_V1.0, CRI, https://www.cottongen.org/species/Gossypium_arboreum/CRI-A2_genome_v1.0), G. raimondii (G.raimondii_221_V2.0, JGI, https://www.cottongen.org/species/Gossypium_raimondii/jgi_genome_221), G. barbadense (Hai7124_V1.1, ZJU, https://www.cottongen.org/data/download/genome_tetraploid/AD2) and G. hirsutum (TM-1_V2.1, ZJU, https://www.cottongen.org/species/Gossypium_hirsutum/ZJU-AD1_v2.1) were downloaded from the CottonGen (https://www.cottongen.org/), including coding sequences (CDS), gene annotations, and protein sequences. The Pfam number PF01504 of the PI4P5K gene family was searched through the Pfam38.0 database (http://pfam-legacy.xfam.org/) (Mistry et al., 2021), and the hidden Markov model (HMM) of the PI4P5K gene family was downloaded. The sequences containing PI4P5K protein domains in the protein files of G. hirsutum were searched via HMMER3.3.2 software (http://www.HMMER.org/) (Potter et al., 2018). The E value was set to 1e-10 to screen candidate protein sequences. The two major non-functional homologues, including pseudogenes and erroneous sequences, are often considered noise that has no significant effect on the results. Therefore, non-functional homologues sequences in gene family identification were removed using seqrutinator with default parameter (Amalfitano et al., 2024). The retained sequences were validated further the conserved domain of PI4P5K protein in NCBI-CDD (https://www.ncbi.nlm.nih.gov/cdd/). Using the same method, we obtained other three predicted cotton PI4P5K genes from G. arboreum, G. raimondii, and G. barbadense.

In order to analyze the genetic diversity of the cotton PI4P5K family, multiple sequence alignment of G. arboreum, G. raimondii, G. barbadense and G. hirsutum were performed using Clustal W. Phylogenetic trees using the neighbor-joining (NJ) method with the bootstrap value setting 1000 replicates were constructed for 4 cotton species and G. hirsutum in MEGA7.0 software, respectively.

Exon-intron structure representations based on basic coding information were displayed using the online program GSDS2.0 (http://gsds.cbi.pku.edu.cn/). Protein domains of the cotton PI4P5K family were annotated using online SMART software (http://smart.embl-heidelberg.de/), and visualized using Gene Structure View in TBtools software (Liu et al., 2024). Physicochemical properties of PI4P5K proteins were calculated and subcellular localization predictions obtained using ProtParam (https://web.expasy.org/protparam/) and CELLO RESULTS (http://cello.life.nctu.edu.tw/) online software. In addition, the hydrophilicity (GRAVY) of PI4P5K protein was analyzed using ProtParam online tool (http://web.expasy.org/protparam/).

Chromosome location information of PI4P5K gene family members was extracted from upland cotton genome annotation files. The distribution map of the PI4P5K genes on the upland cotton chromosomes was drawn in the TBtools software according to their specific physical locations (bp). Collinearity of homologous genes of PI4P5K proteins were analyzed using default parameters by MCScanX (https://github.com/wyp1125/MCScanX), and gene duplication events between the two were visualized using TBtools (Liu et al., 2024).

Relative expression of 48 upland cotton PI4P5K genes for 11 tissues (Ovule, Fiber, Root, Stem, Leaf, Torus, Petal, Sepal, Epicalyx, Anther, and Pistil) and salt stress with 200 mM NaCl treatment of upland cotton “TM-1” PI4P5K genes could be download by website page from the cottonomics database (http://cotton.zju.edu.cn/10.rnasearch.html). The downloaded data have been normalized and averaged by cottonomics (Dai et al., 2022). The expression of upland cotton PI4P5K genes under salt stress at 0 h, 1 h, 3 h, 6 h, 12 h and 24 h was analyzed and a heatmap will be drawn using TBtools software (Liu et al., 2024).

RNA was extracted by RNA isolation kit (Cat # DP441, Tiangen, Beijing, China) after samples were collected, and cDNA were synthesized using cDNA synthesis kit (Cat # RR037A, TaKaRa, Dalian, China). The RT-PCR and qRT PCR methods are based on our previous research (Chen et al., 2021). RT-PCR and qRT-PCR were performed using AtUBQ10 and GhHIS3 as internal references, respectively.

The GhPI4P5K-D04–2 gene coding sequence was amplified by specific primers and inserted into p6MYC to generate an overexpression vector. The restriction sites used in the vector were Kpn I/Sac I. Arabidopsis transformation is carried out through the flower soaking method (Clough and Bent, 1998). Transgenic plants were select on MS solid plates containing 50 µg/L kanamycin. The primers used for cloning GhPI4P5K-D04–2 in this study are as follows: forward (F) (5′- GGTACCTATGTCTGGCCCCGTGGTC-3′) and reverse (R) (5′- GAGCTCTCAACCTTTAATGGAGTTTTGA -3′).

The tobacco rattle virus (TRV2:00) plasmid was digested with restriction enzyme Sac I and Xba I, which were combined with the target fragment to generate TRV2:GhPI4P5K-D04–2 by specifically designed primers. Subsequently, the correctly sequenced TRV2:GhPI4P5K-D04–2 plasmid was transformed into Agrobacterium tumefaciens GV3101, was mixed with auxiliary bacteria, and was injected into two fully unfolded cotton seedlings. Salt treatment (200 mM) was performed on cotton at the three leaf stage, and the silencing efficiency was confirmed by qRT-PCR experiments. The specific primers for constructing the GhPI4P5K-D04–2 VIGS vector are as follows: forward (F) (5′- TCTAGAATGTCTGGCCCCGTGGTC-3′) and reverse (R) (5′- GAGCTCAGAGTCAATATATGTCCCAGTA -3′). Three biological replicates were analyzed.

We identified putative PI4P5K genes in four cotton species. A total of 146 PI4P5K genes were identified, including 25 in G. arboreum, 24 in G. raimondii, 49 in G. barbadense and 48 in G. hirsutum (Supplementary Table S1). To understand the evolutionary relationship of the PI4P5K protein family, a phylogenetic tree was constructed using the PI4P5K protein amino acid sequences from G. arboreum, G. raimondii, G. barbadense and G. hirsutum by the NJ method of MEGA7.0 software. The results showed that all the PI4P5K proteins can be divided into three subgroups: subgroup I, subgroup II, and subgroup III (Figure 1). The number of PI4P5K genes in G. arboreum and G. raimondii in each subgroup was basically half the number in G. hirsutum and G. barbadense in each subgroup. The results indicated that the evolution of PI4P5K genes in cotton were relatively conserved. Moreover, the subgroup I has the largest members of PI4P5K genes, while the subgroup II contains fewest PI4P5K gene members in cotton.

In order to gain a deeper understanding of the basic information of the PI4P5K family in cotton, we compared the full-length protein sequences of PI4P5K genes from four cotton species (G. arboreum, G. raimondii, G. barbadense and G. hirsutum) and constructed a phylogenetic tree containing conserved domains, and exons and introns of the genes (Supplementary Figure S1). Specially, we focused on analyzing the domains and gene structure of PI4P5Ks in G. hirsutum species (Figure 2). The result showed that three types of conserved domains were identified in PI4P5K protein family of G. hirsutum, and all proteins contain PIPKc superfamily domain, making it the core domain of the GhPI4P5K family (Figure 2B). What’s more, only one domain could be found in GhPI4P5K proteins of subgroup I and subgroup II. While, in subgroup III, one GhPI4P5K protein has one domain, six GhPI4P5K proteins each contain two domains, and thirteen GhPI4P5K proteins each possess three domains (Figure 2B). These results suggest that GhPI4P5K proteins of subgroup III might have a more important and specific functions in cotton growth and development. The structural diversity was explored to gain insights into the structural evolution of the GhPI4P5K genes. The results revealed that the GhPI4P5K genes contain exon numbers ranging from 7 to 12, and the intron numbers ranging from 6 to 11 (Figure 2C). Moreover, within the same subgroup, most members have significant similarities in gene structure (Figure 2C).

Further, the physicochemical properties of all the upland cotton PI4P5K proteins were analyzed, and the results showed that all the upland cotton PI4P5K proteins were very different in terms of protein length, protein molecular weight (MW) and isoelectric point (pI) (Table 1). In upland cotton, PI4P5K proteins had an average length of 1125 amino acids and varied in length from 384 (GhPI4P5K-D12-1) to 1853 (GhPI4P5K-D06-2) amino acids, with molecular weights varying from 44.35 to 208.63 KDa. The isoelectric point (pI) values of the upland cotton PI4P5K proteins ranged from 5.32 to 9.16. The overall mean of the hydrophilicity (GRAVY) scores for all PI4P5K proteins was negative, indicating that PI4P5K proteins were hydrophilic. Subcellular localization prediction results showed that 24 PI4P5K proteins were displayed in the nuclear and 18 PI4P5K proteins were displayed in the cytoplasm. The remaining six PI4P5K proteins were predicated to localize in the nuclear and cytoplasm, which was consistent with hydrophilic of the overall mean of the GRAVY scores.

There was a total of 48 GhPI4P5K genes distributed on 24 chromosomes of G. hirsutum, of which 24 genes were distributed on 12 chromosomes of the A subgenome and D subgenome, respectively (Figure 3). While there were no GhPI4P5K genes distributed on A11 and D11 chromosomes. Except the chromosome 02, 03, 04 and 05 of subgenome A and D had different gene numbers, all the other corresponding chromosomes of the two subgenomes contained the same gene numbers, and the highest number of GhPI4P5K genes were on chromosome A05 and D05. In addition, observations had shown that the majority of GhPI4P5K genes were distributed at both ends of chromosomes. This telomere-proximal distribution might be associated with higher recombination rates and gene density in these regions, which could facilitate gene family expansion and functional diversification. Furthermore, such a distribution pattern may influence the regulation and expression of these genes. Whatever, the GhPI4P5K gene distribution on the A and D subgenomes was relatively similar, indicating a strong correlation between the two subgenomes. The chromosomal location results of GhPI4P5K genes were in line with the evolutionary relationship of polyploid cotton.

We further analyzed the collinearity relationships for all PI4P5K genes on the chromosomes At and Dt of G. hirsutum with other Gossypium species. Results showed that 24 PI4P5K genes on A subgenome had intergenomic homologous genes on the chromosomes D subgenome of G. hirsutum. In-depth studies found that PI4P5K genes on chromosomes A01/D01, A04/D04, A05/D05, A06/D06, A07/D07, A08/D08, A09/D09, A10/D10, A12/D12 and A13/D13 exhibited strong one-to-one collinearity (Figure 4). Interestingly, genes of GhPI4P5K-A02-1, GhPI4P5K-A03-1, GhPI4P5K-A03–2 and GhPI4P5K-A05–6 exhibited collinearity with GhPI4P5K-D03-1, GhPI4P5K-D02-1, GhPI4P5K-D02-2, and GhPI4P5K-D04-1, respectively. To explore the evolutionary relationship of PI4P5K genes in cotton, we performed synteny analysis of genome of G. arboreum and G. hirsutum A subgenomes, genome of G. raimondii and G. hirsutum D subgenome, G. hirsutum and G. barbadense. Among 24 PI4P5K genes on A and D subgenome had intergenomic homologous genes in G. arboreum and G. raimondii, respectively (Supplementary Figures S2, S3). The 48 PI4P5K genes in G. hirsutum and the 49 PI4P5K genes in G. barbadense had intergenomic homologous relationship, including GhPI4P5K-A05–6 exhibited collinearity with GbPI4P5K-A05–6 and GbPI4P5K-A05-7 (Supplementary Figure S4). The results emphasized PI4P5K genes were highly conserved for gene quantity during polyploidization, while afterward same directional evolution for different allotetraploid cotton species.

Gene expression patterns are usually related to gene function (Zhao et al., 2021). We investigated the expression patterns of GhPI4P5K genes in eleven tissues of cotton (Ovule, Fiber, Root, Stem, Leaf, Torus, Petal, Sepal, Epicalyx, Anther and Pistil). The results showed that most of the 48 selected GhPI4P5K genes exhibited tissue expression specificity (Supplementary Figure S5). For example, GhPI4P5K-D05-2, GhPI4P5K-A05-2, GhPI4P5K-A01–1 and GhPI4P5K-D01–1 had the highest expression level in ovule of 10 dpa (day post-anthesis), GhPI4P5K-A03–1 and GhPI4P5K-A03–2 in fiber of 20 dpa, GhPI4P5K-A01–3 and GhPI4P5K-D01–3 in root, and GhPI4P5K-D09–2 in pistil. PI4P5K gene family plays an important role in plant growth, development, and stress response. So, we focused on the analysis of the expression patterns of GhPI4P5K genes under salt stress, and the results showed that the expression levels of many genes have changed with most exhibiting upregulation, among which five genes (GhPI4P5K-A05-2, GhPI4P5K-A01-3, GhPI4P5K-D01-3, GhPI4P5K-A01-1, GhPI4P5K-D01-1) were significantly upregulated, and their expression levels reached their highest after 12 h of salt treatment (Figure 5). In addition, GhPI4P5K-A05–4 and GhPI4P5K-A05–1 had the highest expression level at the time point of 3h after salt treatment, and GhPI4P5K-A03–2 occurred at 6h later after salt stress (Figure 5). Moreover, GhPI4P5K-D04–2 was a typical representative of the PI4P5K gene family in subgroup I in G. hirsutum, and its expression level had changed at 3 h, 6 h and 12 h obviously in upland cotton (Figure 5). Therefore, investigating the function of GhPI4P5K-D04–2 gene under salt stress is representative.

To analyze the expression pattern of GhPI4P5K-D04–2 in cotton, we subjected the three-leaf stage cotton to salt stress treatment and took its leaves for qRT-PCR analysis. The results showed that GhPI4P5K-D04–2 showed a slight upregulation followed by a decrease and then upregulation trend, and the expression peak appeared 24 hours after salt stress treatment (Figure 6A). The above results indicated that GhPI4P5K-D04–2 was involved in the response of plants to salt stress. To verify whether GhPI4P5K-D04–2 could enhance plant salt tolerance, three transgenic Arabidopsis lines (38-15, 42–27 and 45-10) of T3 generation were obtained and checked by semi-quantitative PCR (Figure 6B). Arabidopsis seeds from three transgenic lines and the control Col-0 in the same culture dish were planted for growth. It was found that the transgenic strains exhibited stronger salt tolerance under treatment with 125mM NaCl (Figure 6C). In addition, the results of germination rate and greening rate showed that the transgenic strains were higher than those of Col-0 under salt stress (Figures 6D–G). All the results indicated that GhPI4P5K-D04–2 positively regulated the salt resistance in transgenic Arabidopsis.

PI4P5K genes plays a key enzyme role in the inositol signal transduction system and has essential functions in plants in terms of growth, development, and stress responses. Therefore, we performed a virus-induced gene silencing (VIGS) experiment to verify its function under salt stress. After approximately 14 days of bacterial infection, TRV2:PDS cotton leaves showed an albino phenotype (Figure 7A). The silencing efficiency of GhPI4P5K-D04–2 was evaluated using qPCR, and the results showed that the expression level of GhPI4P5K-D04–2 gene in TRV2:GhPI4P5K-D04–2 plants was significantly lower than that in TRV2:00 plants (Figure 7B). The above results indicated that the GhPI4P5K-D04–2 gene was successfully silenced in cotton through the VIGS system. Plants of TRV2:GhPI4P5K-D04–2 and TRV2:00 were subjected to 200 mM NaCl treatment. The results showed that there was no significant difference in growth status between TRV2:GhPI4P5K-D04–2 and TRV2:00 plants under non-stressed conditions. Nevertheless, after approximately 48 h of exposure to 200 mM NaCl stress, TRV2:GhPI4P5K-D04–2 plants exhibited more severe leaf wilting compared with TRV2:00 plants (Figure 7C). These results preliminarily indicated that silencing of GhPI4P5K-D04–2 impaired the tolerance of plants to salt stress.