Introduction

Gene duplication and loss events are significant drivers of gene family evolution1. Duplications result from two primary mechanisms: whole-genome duplication (WGD) and small-scale duplications (SSD). These small-scale duplications include tandem, segmental, and transposon-mediated duplications2. These mechanisms seem to represent separate modes of expansion, as gene families evolving through WGDs seldom undergo SSD events, contributing to the maintenance of gene family expansion3.

The defense responses mediated by resistance (R) genes represent a significant source of resistance against plant pathogens, including viruses4. Among the major families of R genes, nucleotide-binding leucine-rich repeat (NLR) genes are prominent5. The Nucleotide Binding Site Leucine-Rich Repeat (NLR) gene family originated and diverged into at least three subclasses. Two of these subclasses, each characterized by either a Toll/Interleukin-1 Receptor (TIR) or Coiled-Coil (CC) domain in the N-terminal, play crucial roles as major immune receptors for effector-triggered immunity (ETI) in plants6. Simultaneously, the third subclass, distinguished by an N-terminal Resistance to the Powdery Mildew8 (RPW8) domain, functions as a component for signal transfer within this system7.

Plant NLRs, akin to their animal counterparts, are modular proteins typically comprised of three fundamental components: an N-terminal domain and a central NB-ARC domain, which is often referred to as the Nucleotide-Binding Adaptor shared with APAF-1, plant resistance proteins, and CED-4. It also contains a C-terminal domain rich in leucine repeats (LRR)8,9. The central domain of animal NLRs is also known as the NACHT domain (named after NAIP, CIITA, HET-E, and TP1)10. The NACHT domain, while structurally similar to the plant NB-ARC domain, exhibits distinct features specific to animal NLRs11,12.

Plant-NLRs exhibit a unique feature with the utilization of either a TOLL/interleukin 1 receptor (TIR) domain or a coiled-coil (CC) domain at the N-terminus. This feature distinguishes two major types of plant NLRs: the TIR-type NLRs (TNLs) and the CC-type NLRs (CNLs), respectively13,14. Nonetheless, deciphering the structures of N-terminal domains in numerous plant NLRs poses a significant challenge. This challenge arises primarily due to their structural variability and the lack of substantial similarity to well-established protein structures15. Therefore, NLRs with an N-terminus other than the TIR domain are occasionally classified as non-TIR-type NLRs (nTNLs). This categorization distinguishes them from TNLs.

The NLR family has greatly expanded in many plants, resulting in one of the largest and most variable plant protein families16. This is in contrast to vertebrate NLR repertoires, which typically consist of around 20 members17. Many surveyed plant genomes have exhibited large NLR repertoires, and recently, a database, ANNA: an Angiosperm NLR Atlas, was constructed. It contains over 90,000 NLR genes from 304 angiosperm genomes, including 18,707 TNL genes, 70,737 CNL genes, and 1847 RNL genes18, similar to the 2012 NBS encoding genes found in wheat19.Notably, bryophytes like Physcomitrella patens and lycophytes like Selaginella moellendorffii represent ancestral land plant lineages20. They possess relatively small NLR repertoires, with around 25 NLRs in the case of Physcomitrella patens and 2 NLRs in the case of Selaginella moellendorffii21. This indicates that substantial gene expansion has primarily occurred in flowering plants22. Recent research has uncovered that many microRNAs target the nucleotide sequences encoding conserved motifs within NLRs, including the P-loop, in a variety of flowering plants23. This theory suggests that the comprehensive control of NLR transcripts may enable a plant species to maintain extensive NLR repertoires without exhausting functional NLR loci24,25. As microRNA-mediated transcriptional suppression of NLR transcripts could potentially offset the fitness costs associated with NLR maintenance, this mechanism might contribute to the sustained existence of large NLR repertoires23,26.

Currently, in Pakistan, almost all cotton varieties of Gossypium hirsutum are vulnerable to cotton leaf curl disease (CLCuD)27. CLCuD is induced by Begomoviruses from the Geminiviridae family. The disease is transmitted by the whitefly insect vector, scientifically known as Bemisia tabaci28. G. arboreum, also known as “desi cotton,” represents a high level of resistance against insect pests and diseases including CLCuD, while Mac7 (G. hirsutum) is highly tolerant, and Coker-312 (G. hirsutum) is highly susceptible to CLCuD29,30. NLR is the main class of resistance genes that showed a response to viral disease31,32. Thus, this study aimed to evaluate the NBS domain associated with host plant resistance genes, that might be a potential source of disease resistance genetic elements.

Methodology

Genome assemblies and data collection

In the current study, we have selected 39 land plants ranging from green algae to higher plant families including Amborellaceae, Brassicaceae, Poaceae, Citrus, Cucurbitaceae, Malvaceae, Marchaceae, Fabaceae, Nelumbonaceace, Salicaceae, Rosaceae, and Araceae families. In addition, the selection of plants was also made based on ploidy level (haploid, diploid, and tetraploid) for further detailed evolutionary study. The latest genome assemblies (Table S1) were downloaded from publicly available respective genome databases, NCBI, Phytozome, and Plaza genome databases33,34.

Identification, classification, and comparison among land plants

To screen the NBS (NB-NRC) domain-containing genes, the PfamScan.pl HMM, search script was used with default e-value (1.1e-50) using background Pfam-A_hmm model35. All genes having NB-ARC domain were considered NBS genes and filtered for further analysis. In addition, the additional associated decoy domains were also observed through the domain architecture of NBS genes by following the Hussain et al.36 classification method. In this classification system, similar domain-architecture-bearing genes were placed under the same classes. Furthermore, a comprehensive comparison of classes was also made among land plants.

Evolutionary study; orthogrouping, and duplication analysis

To provide a deep understanding of the evolution and diversification of NBS genes in land plants, we used OrthoFinder v2.5.1 package tools37. In this package, the DIAMOND tool was used for fast sequence similarity searches among NBS sequences38. The clustering of genes was done using the MCL clustering algorithm. The orthologs and orthogrouping were carried out with DendroBLAST39. For multiple sequence alignment, MAFTT 7.0 was used40. A gene-based phylogenetic tree was also constructed by the maximum likelihood algorithm in FastTreeMP with a 1000 bootstrap value41.

Transcriptomic analyses of NBS genes

To ascertain the differential expression and responsiveness of NBS genes in various tissues and stresses, we have retrieved RNA-seq data from the IPF database (http://ipf.sustech.edu.cn/pub/)42, (Arabidopsis, maize, soybean, upland cotton, and wild cotton) from different databases including Arabidopsis RNA-seq database (http://ipf.sustech.edu.cn/pub/athrna/)42, maize RNA-seq database (http://ipf.sustech.edu.cn/pub/zmrna/)42, cotton RNA-seq database (http://ipf.sustech.edu.cn/pub/cottonrna/)42, soybean RNA-seq database (http://ipf.sustech.edu.cn/pub/soybean/)42, Cotton Functional Genomics Database (CottonFGD) (https://cottonfgd.net/)43 and Cottongen database (https://www.cottongen.org/)44. The Fragments Per Kilobase of transcript per Million mapped reads (FPKM) values were retrieved from respective databases using gene accession as query IDs. The extracted FPKM values the categorized into biotic and abiotic stresses. Besides the FPKM data retrieved from the above-mentioned databases, we also collected additional RNA-seq data from NCBI BioProjects (Bio-projects PRJNA490626 and PRJNA594268, (PRJNA390823) and PRJNA398803)45,46,47. The RNA-seq data is then categorized into three data types; (1) tissue-specific (leaf, stem, flower, pollen, endosperm, pollen, and seed, etc.), (2) abiotic stress-specific (dehydration, cold, drought, heat, dark, osmotic, salt, wounding, etc.) and (3) biotic-stress specific (Blumeria graminis, Botrytis cereal, Collettrichum tofieldiae, Heterodera schachti nematodes, bacterial strain, Fusarium graminearum, Rhizotonia solani, etc.) expression profiling. The RNA-seq data was processed through transcriptomic pipelines, as mentioned by Zahra. et al.47. The final heat map was drawn using the TBTool package under heatmap construction with a Log2Base value of FPKM48.

Genetic marker prediction in susceptible and tolerant cotton accessions

To find important genetic markers in NBS genes, we have selected two contrast accessions of G. hirsutum cotton. The Mac7, a highly tolerant G. hirsutum to CLCuD, and Coker 312, a highly susceptible accession to CLCuD. The whole genome resequencing data of these plants were collected from NCBI BioProject: PRJNA75643549 and PRJNA542238 respectively. The NGS Raw reads were mapped to NBS genic regions of the G. hirsutum TM-1 reference genome and identified variants (SNPs/InDels). Furthermore, the identified variants were annotated and characterized based on variant type. The variants associated with genes were further compared between the two-accession using the Venny-Bioinfo Tool.

Gene ontology, KEGG pathways, and cis-regulatory elements analysis

To find the functional analysis of NBS genes, we performed gene ontology and KEGG pathway enrichment analysis using the Gene Ontology resource and KEGG Pathway Database50. The protein features including the number of amino acids, molecular weight, theoretical pI (isoelectric point), and Grand Average Hydropathy (GRAVY), of G. hirsutum were studied using the ProtParam tool–Expasy tool. For the identification of the cis-regulatory elements of NBS genes in G. hirsutum, we have retrieved 2000b upstream of NBS genes and subjected them to Plant CARE databases51.

Protein modeling, molecular docking, and target gene mining

To determine the NBS protein’s structure and its binding activity to CLCuD viral proteins52 and with ATP and ADP molecules, we have used the protein modeling and molecular docking approaches. Based on the biotic stress RNA-seq, we have selected NBS genes (upregulated in tolerant; Mac7, while downregulated in susceptible, Coker 312) from wild cotton (G. arboreum; naturally resistant to CLCuD) and upland cotton (G. hirsutum, susceptible to CLCuD) data. The 3D structure of viral and host proteins was predicted using an I-TASSER server with default values53. The molecular docking was done using Auto-dock Vina54 and MDock web-based server55. The 2D and 3D structures were visualized with LigPlus and Discovery Studio, respectively56. The binding Gibs free energy was calculated with the ClusPro tool57. For further functional analysis, we selected Gar06G24920 (OG2 member), previously reported as a differentially expressed gene under CLCuD in G. arboreum.

Plant seed material

A naturally immune Gossypium sp., G. arboreum variety FDH-228 was used for the functional analysis of the NBS gene. The seeds of G. arboreum variety FDH-228 utilized in this study were obtained through a legal and ethical process. The seeds were collected from Gene Isolation Lab, NIBE, Faisalabad, Pakistan by all applicable regulations and guidelines governing the collection and use of plant genetic resources.

Virus-induced gene silencing (VIGS) protocol

The selected gene (Gar06G24920_OG2) structural features were retrieved from the Cottongen database, and the gene-specific primer was designed with a 500 bp amplicon size of the product. The tobacco rattle virus (TRV) system (pTRV-RNA1 and pTRV-RNA2) was used for VIGS analysis as previously reported by30. A 500 bp fragment was PCR amplified from G. arboreum cDNA and cloned in a pTRV-RNA2 vector at EcoR1 and Kpn1 restriction sites. The pTRV-RNA2 clones were confirmed through sequencing. The TRV-GrCLA1 clone was used as a positive VIGS control. The VIGS clones were named TRV: G2, and transformed in Agrobacterium strain GV3101 for inoculation. For the VIGS assay, 10 plants of FDH-228 were inoculated for each of the VIGS construct and control at the cotyledonary leaf stage as described by Gao et al.58. The inoculated plants were placed at 26 ± 1 °C temperature with a 16/8 h light–dark period in a glasshouse containment facility. After the development of a completely bleached phenotype on TRV-GrCLA1 inoculated cotton plants, gene silencing was analyzed using RT-qPCR and results were compared with the TRV:00 inoculated control plants. The VIGS plants of FDH-228 were inoculated with CLCuD through grafting or by exposure to viruliferous whiteflies. For each condition, five VIGS and control plans were challenged and considered for qPCR-based estimation of virus titer.

CLCuD inoculation through grafting and viruliferous whitefly exposure method

To define the function of OG2 genes (Gar06G24920_OG2), we have selected a naturally immune Gossypium sp, G. arboreum variety FDH-228, and used the VIGS approach to silence the selected gene. The silenced plant is then treated with CLCuD through grafting. For graft inoculation, 4–5-inch long CLCuD-infested scions of G. hirsutum were grafted with FDH-228 plants. The bottle-neck grafting technique was implemented as described by Akhtar et al.59 for successful virus transmission. The diseased scions of G. hirsutum were kindly provided and maintained by KP Akhter (PS) at the grafting facility of NIAB. Faisalabad, Pakistan.

The whiteflies were initially collected from the cotton fields of the NIAB. The whitefly culture was maintained on potted cotton plants at 7–10 leaf stages (G. hirsutum) in a separate room specified for insect rearing within insect-free cages under the above-mentioned controlled conditions. The maintained culture of whiteflies was assumed to be B. tabaci Asia II 1, the most prevalent whitefly species in a cotton-growing zone of Punjab, Pakistan28,60. In each cage, both VIGS and control plants were placed side by side and 20–30 adult whiteflies were released per cage. The CLCuD harboring cotton cuttings were placed inside the cages and plants were frequently shuffled for uniform whitefly-mediated virus transmission. The whitefly growth room and glasshouse with grafting experiments were set at 26–28 °C temperature, 68–70% relative humidity, and a 16/8 h light–dark period.

RNA extraction and qPCR assay

To assess the VIGS and the viral titer, a total RNA was extracted from VIGS-silenced and control plants. For RNA isolation, leaf tissue samples were taken from the upper three leaves of each plant and RNA was isolated using Trizol (Invitrogen). Purified RNA was used to synthesize cDNA using Revert Aid's first-strand cDNA synthesis kit (Thermo Scientific, USA). RT-qPCR was performed on at least three independent samples of VIGS and TRV:00 inoculated control plants to measure gene silencing. The results were analyzed by the ΔCt method and the 18S gene was used to normalize the corresponding Ct values30. Gene-specific primers were used for the expression analysis (Table S33). Twenty-five days post CLCuD inoculation, the leaf discs (~ 3–4 discs/leaf) from the top three leaves of each whitefly exposed and grafted cotton plants were excised and pooled for CTAB-mediated DNA extraction. The purified DNA was nanodroped and diluted to a final concentration of 10 ng/µl. For qPCR analysis and standard preparation previously optimized protocol by Shafiq et al.61,62 was followed. The dilutions of plasmids were made in the range from 2, 0.2, 0.02 to 0.002 ng for standard curve preparations.

Statement on ethical collection of plant materials

The seed materials utilized in this study are commercially approved varieties and are readily available in the market. Specifically, Gossypium arboreum variety FDH-228 was sourced from the Gene Isolation Lab, NIBGE, Faisalabad, Pakistan. All requisite permissions and licenses for specimen collection have been acquired, in strict adherence to regulatory guidelines. This study is committed to maintaining ethical standards, respecting intellectual property rights, and ensuring transparent documentation of the sources throughout the research process.

Results

Genome-wide identification

The whole-genome screening analysis identified a total of 12,822 NBS-encoding genes across 34 species. The species-based genome-wide identification demonstrated that the number of genes present in each genome was independent of its genome size. The genome size to the number of NBS gene regression (R2) values was observed as below the cutoff value (R2 = 0.015). Interestingly, we observed a correlation of genome size to the number of NBS genes in Gossypium sp. (R2 = 0.669). For instance, G. hirsutum with a genome size of 2.4 Gb has the highest number of NBS genes (708 NBS genes), followed by G. barbadense (622 NBS genes) with a genome size of 2.3 Gb, and G. arboreum, a diploid genome with a 1.7 Gb genome, has 365 NBS genes. Similarly, G. raimondii with a 0.7 Gb genome has 323 NBS genes (Fig. 1, Tables S1 and S2).

Figure 1
figure 1

Genome-wide identification and classification of NBS genes in Land plants. The length of the bar represents the total number of genes and different colors demonstrate the top 20 subclasses, based on the domain architecture.

Classification of NBS genes in land plants

The Pfam domain analysis identified several decoy domains in addition to the NBS domain. The deep analysis of domain architecture reported several functional protein domains associated with NBS domains (Table S2). So, based on the conserved domains, motifs, and their sites in the primary sequences of NBS proteins, all 12,822 genes were divided into 168 classes concerning the presence /absence and copy number of associated domains (Table S3). Of these classes, the 33rd class (only the NBS domain) has 8591 NBS genes, showing the largest class of NBS superfamily and most of the classes were highly specific with few numbers of NBS genes. So, based on the observation, it can be estimated that the main superfamily of NBS proteins contains only the NBS domain. The species-wise distribution and diversity of identified NBS classes were also interesting. As the lower plants possessed only either one or two specific classes e.g., Selaginella moellendorffii belong to a lycophyte, and possessed only 33th classes (only NBS single domain) with 16 NBS genes. Similarly, Spirodela polyrhiza is a species of duckweed known by the common name common duckmeat and has only two classes 33rd and 60th (NBS and NLR). In contrast, the evolutionary development and adaptation process causes drastic changes in the genome of higher land plants. As moved from simple to complex plants, the NBS classes became more complex. For instance, Hordeum vulgare, barley, a member of the grass family, is a major cereal grain grown in temperate climates globally and has the highest number of NBS classes and several unique NBS domain architectures. Similarly, Vitis vinifera, the common grape vine, a species of flowering plant, native to the Mediterranean region, also has twenty NBS classes. Among all land plants, the most common domain architectures were detected as NBS, NLR, TIR-NBS, TIR-NBS—LRR, and RPW8—NBS domain architectures. However, we also observed highly species-specific classes like Methyltransf_11—NBS, NBS-Glyco_transf_8, TIR-NBS-Lectin_legB-Pkinase, TIR-NBS-RHD3, and NB-LRR-NB-LRR-NBS-Retrotran_gag_2. Overall, out of 168 classes, 94 classes were specific-specific, some classes were family-specific, and were genus-specific like the Gossypium species showed a close relationship regarding the classes. The tetraploid species (G. barbadense and G. hirsutum) of Gossypium sp. have more diversity rather than other diploid species (Fig. 1, Tables S3 and S4).

Orthologs analysis: orthogrouping, duplication, and overlapping,

The orthologs groping analysis of 12,822 NBS in 33 land plants demonstrated that 91.8% of total NBS proteins were assigned orthogroups. However, only 8.2% of NBS proteins were highly unique and could not fit into any orthogroups. A total of 664 orthogroups were predicted and out of these, 347 orthogroups (holding 1135 NBS proteins) were species-specific orthogroups and the mean and median orthogroup size was 18 and 3 genes, respectively (Table S5). The species-based comparative genomics overexposed the genetic makeup of different land plants regarding NBS genes. We identified many common/core genes as well as species-specific unique genes based on sequence divergence. For instance, in the case of Arabidopsis layrata, a total of 209 NBS genes were predicted and categorized into 37 orthogroups (with 2 species-specific orthogroups) covering 206 NBS genes. The remaining three genes did not fit any orthogroups due to their clear sequence divergence as compared to other NBS genes. Similar results were also observed for other land plants. For instance, in the Gossypium sp. the Gar, Gba, Ghi, Ghe, Gth, and Gra did not show any species-specific unique orthogroup or ortholog. But observed some unique NBS genes were categorized as “unassigned orthogroup genes” (Table S6). The number of genes in orthogroups of each species was also assessed and identified that the “OG000” orthogroup possessed the highest number with 2535 NBS proteins and shared all species except some lower plants including Gbi, Mpo, and Smo. In addition, some orthogroups were highly specific to species like OG400 was only present in Al with ALNBS134 and ALNBS135 genes. Similarly, the OG024 was specific to Mtr with more than 15 NBS genes and the OG454 (GhirNBS444 and GhirNBS445), and OG455 (GhirNBS626 and GhirNBS627) were specific to G. hirsutum. Similar unique genes were also identified in other land plants (Tables S7, S8, S9 and S10). The orthogroups duplication event analysis conformed several duplication events during the evolutionary process e.g., the OG000 duplicated 1932 times with 100% confidence and 1600 times with a 50% confidence level. Similarly, other orthogroups were also passed by some duplication events (Tables S11 and S12).

To find the relationships among different species the orthogroups and orthologs overlapping were also assessed and compiled into clusters using dendrogram. The result demonstrated that Osa, Hvu, Zmy, Bra, and Sbi formed one clade, and Sly, Stu, and Sme formed another clade. Similarly, the Gossypium sp. Including Gth, Gtu, Gra, Gba, Ghi, Gar, and Ghe formed one clade. In addition, the Gar (A-genome) and Ghe (A2-genome) shared sister branches, and the Ghi (AD-genome), and Gba (AD2-genome) formed another sister branches showing the highest genome similarity index (Figs. S1, S2, and Table S13).

Phylogenetic tree analysis: duplication events at nodes and gene-based phylogenetic tree

The duplication events at internal and terminal nodes indicated the evolutionary relationship of NBS genes in land plants. The duplication events at terminal nodes demonstrated that Ginko biloba was highly conserved and did not demonstrate any duplication at the terminals as compared to other land plants. Entering the Gossypium genus, the N24 node further duplicated more than 210 times and separated other Gossypium sp. from G. raimondii (Table S14). In conclusion, gene duplication events at internal and terminal nodes demonstrated the expansion of the NBS gene family from lower plants to higher plants. The duplication events also cause the gain and loss of NBS-associated terminal domains, as we observed in the domain architecture analysis. This diversity in NBS proteins might be an adaptation of land plants (Fig. 2).

Figure 2
figure 2

Evolutionary study of NBS genes in land plants. (A) Basic statistics of orthogroups with abbreviated species names. (B) Gene duplication events at terminal nodes of the phylogenetic tree, and (C) gene duplication events at internal nodes of a phylogenetic tree.

The gene-based phylogenetic tree was divided into 282 sub-clusters and each cluster was correspondence to the orthogroups. The highest number of genes was observed in cluster OG000, which gradually decreased to and decreased to OG282. In addition, some clades were highly species-specific, and some were genus while very few were family specific. Due to the large number of genes in top orthogroups only OG2 was present in the phylogenetic tree (Fig. S3).

RNA-based expression profiling

For a deep understanding of NBS genes in different tissues under different biotic and abiotic stresses, we have selected three plant species including A. thaliana, Z. mays, G. arboreum, and G. hirsutum. Based on the RNA-seq expression data most important putative genes were further filtered for detail study. The tissue-specific expression profiling of NBS genes in A. thaliana generally demonstrated that most of the genes are differentially expressed in leaf, shoot, seedling, flower, and silique. However, very little or negligible expression is observed in pollen, endosperm, embryo, and seed. At the orthogroups (OGs) level, some OGs were highly specific to some tissues like OG6 (At4G33300OG6 and At5G04720 OG6), OG15 (AT5G45490 OG15) had the highest expression in root tissue (Table S15). Similarly, the OG2 (AT1G61300OG2, AT5G63020 OG2, and AT1G61190 OG2) showed similar patterns in leaf, shoot, and seedling. The cladogram among tissues and genes also demonstrated a co-expression network of genes. In the case of Zmy tissues, three major co-expression pattern networks were observed among NBS genes (Fig. 3, Tables S16, S19, S22 and S25).

Figure 3
figure 3

RNA-seq-based expression profiling of NBS gene in different tissues. The OGs associated with gene IDs represent the orthogroups. (A) A. thaliana, (B) G. hirsutum, (C) Z. mays and (D) G. arboreum.

The RNA-seq data of abiotic stresses included oxidative stress, UV, nutrient deficiency, dark, salt, cold, drought, heat, ozone, and wounding stresses. Generally, it was observed that the OG members mostly co-expressed under various abiotic stresses. In A. thaliana, the NBS genes are most commonly expressed in all stresses except oxidative and UV stresses. Among OGs, the OG6, OG11, and OG1 showed similar expression patterns with the highest expression under most of the stresses. Similarly, the OG6, OG0, and OG351 formed one co-expression clade in the Z. mays plant under various abiotic stresses (Fig. S4, Tables S17, S20 and S23).

The biotic stress expression profiling included different pathogens from aphids to viruses (aphids, nematodes, fungi, bacteria, and viruses). The analyses identified several NBS gene responses under different pathogenic stresses in Arabidopsis, Z. mays, G. arboreum, and G. hirsutum. For instance, the OG1 (AT1G66090OG1, AT4G19520OG1, AT5G41750OG1, and AT1G72900OG1) significantly upregulated under Slerotinia sclerotiorum (Fig. 4A). Similarly, under viral disease, the OG1, OG5, and OG6 showed putative responses during viral infection in Arabidopsis and Z. mays (Fig. 4B). The G. arboreum is naturally resistant to several viral diseases, so we have taken the CLCuD grafted RNA-seq for the identification of the NBS gene’s role in the presence of viruses. We identified several differentially expressed NBS genes in G. arboreum under grafted-CLCuD. The OG2 (Gar06G25610OG2, GarUnG04050OG2), OG6 (Gar11G21680OG6), OG4 (Gar11G229450OG4) and OG115 (Gar10G07690OG115) showed significantly upregulation under CLCuD (Fig. 4C). The Gar06G25610OG2 was further validated through gene silencing approaches in this study (will discuss later) (Fig. S5). As we are so interested in the identification of most putative NBS genes and their role in G. hirsutum under cotton leaf curl disease, which is one of the major challenges in Pakistan, we presented a two-contrast agronomic trait associated accession (Mac7 (a highly CLCuD tolerant G. hirsutum accession developed by USDA, but low production) and Coker312 (highly susceptible, most regenerative accession, commonly used for tissues culturing, also posseted important agronomic traits) RNA-seq data. The results were surprising, we found differential regulation of NBS genes in tolerant accession (Fig. 5, Tables S18, S21, S24 and S26). In summary, we found that the OG0, OG2, OG5, OG6, and OG15 have a highly responsive role in different tissues and under various biotic and abiotic stresses in Arabidopsis, Z. mays, G. arboreum and G. hirsutum.

Figure 4
figure 4

RNA-seq based expression profiling of NBS gene under different biotic stresses. The OGs associated with gene IDs represent the orthogroups. (A) A. thaliana, (B) Z. mays and (C) G. arboreum.

Figure 5
figure 5

Expression profiling of G. hirsutum NBS genes in different tissues under various biotic stresses. (A) Cotyledon tissues, (B) leaf tissues, (C) expression of NBS gene in two contrasting accessions, 1; Mac7 (tolerant to CLCuD), 2; Coker312 (highly susceptible to CLCuD) under CLCuD infection in leaf. (D) under various root-associated pathogens in root tissue.

Variants detection in tolerant and susceptible accessions

The genome-wide genetic diversity of Mac7 (G. hirsutum tolerant) and Coker312 (G. hirsutum susceptible) with reference to TM-1 G. hirsutum reference genome, identified several SNPs and InDels in NBS genomics regions of Mac7 (InDels: 2989, SNPs: 3594) and Coker 312 (InDels: 2646, SNPs: 2527). The identified variants were characterized into four impact levels i.e., impact as high (affecting splice-sites, stop and start codons), moderate (non-synonymous), low (synonymous coding/start/stop, start gained), and modifier (upstream, downstream, intergenic, UTR). A comparative study of high-impact SNPs and InDels associated genes between Mac7 and Coker identified several unique and common variants e.g., Coker 312 (7 SNPs, 10 InDels) and Mac 7 (22 SNPs, 26 InDels) unique variants. While 4 genes were common between the two accessions. Similarly, the modifier and moderate and low-impact variants were also observed in the two accessions (Fig. 6, Tables S27, S28 and S29).

Figure 6
figure 6

Genetic variations of NBS genes between Mac7 and Coker 312. (A) variants distribution region Based on impact level of variants on gene functions. (A) HI; High Impact, (B) MFI; Modifier Impact, (C) MI; Moderate Impact, (D) LI; Low Impact, comparison at SNPs (single nucleotide polymorphisms) and InDels (Insertion and Deletion).

Physiochemistry, gene ontology, and metabolic pathways

The amino acid (AA) sequence analysis of Gossypium hirsutum showed a length ranging from 500 to 1500 amino acid (AA), molecular weight (MW) ranged from 50 to 150 kDa, isoelectric point (pI) value ranged from 2 to 6, charge ranged from + 55.5 to − 44.5 and the Grand Average of Hydropathy ranged from 0 to − 0.2. The gene ontology comprises three components; biological process, molecular function, and cellular components. In the case of molecular function, the results demonstrated that the NBS is mainly involved in the Adenosine diphosphate (ADP) and Adenosine triphosphate (ATP) binding activity with signal transduction. The KEGG pathway analysis indicated the role of the NBS gene in plant-pathogen interaction (PPI) and signaling pathways (SP). The promoter analysis identified several stress-responsive elements in the promoter region of NSB genes. A total of 72 cis-regulatory elements were identified some of the responsive elements were present in all NBS genes like TATA-box, CAAT-box, Box, ARE, G-box, GT1-motif, ABRE, CGTCA-motif, TGACG-motif, TCT-motif, TCA- element, MBS, GATA-motif, CAT-box, and O2-site. Most of the cis-regulatory elements belonged to stress responses (Fig. S5, Table S30).

Molecular docking, and protein–protein interaction

For the proteins modeling, we selected genes based on upregulated in G. arboreum (naturally resistant to CLCuD) under grafted-CLCuD and whitefly induced in Mac7 (G. hirsutum, tolerant to CLCuD) covering OG0, OG2, OG15, and OG43. The selected genes’ translated proteins were used for 3D structure prediction using the I-TASSER server. The PDB database proteins 6J5T-C,6S2P-N, 7CRB-A,4TZH-A, and 4U09-A were used as template sequences. The 3D molecules of ATP and ADP were downloaded from the cheminformatics database. The docking results demonstrated a stable interaction of NBS genes with ATP and ADP with a range of − 7 kcal/mol to 8.2 kcal/mol, except Gar09G25760_OG43 and Ghir_A12G021600_OG15, which showed below − 6.8 kcal/mol and − 6.6 kcal/mol, respectively. The binding affinity of Gar12G23120_OG0 showed that it has a more stable interaction with ATP as compared to the ADP molecule. Similarly, the OG2 genes (Gar01G01860 OG2, Gar06G24920 OG2) showed more affinity to the ATP molecule as compared to the ADP molecule. In contrast to this, the OG0 (Gar11G29700OG0, Ghir_D13G021900OG0) showed more stable interaction with the ADP molecule. The interacting residues of NBS proteins varied from ATP to ADP e.g. the ATP binds with Lys179, His240, Trp139, Val261, Gln 264, Glu142, and Val 284, whereas the ADP binds with Asp143, Glu142, Asn289 and Lys286 in Gar12G23120_OG0 protein. Similarly, other binding results also demonstrated similar patterns (Fig. S6, Table S31). For further detailed molecular interaction of OG2 genes (Gar06G24920_OG2) with CLCuD viral proteins (AC1, AC2, AC3, AC4, AV1, and AV2), we assessed the interaction level by Gibs free energy. The Gar06G24920_OG2 protein demonstrated the highest stability value (− 1352 kcal/mol) with AC1 viral proteins followed by AV2 (− 1103.8), AV1 (− 1026.4), and so on. Similarly, another member of OG2 group protein (Gar06G25610/Gohir.A06G19220/Ghir_A06G020580), also demonstrated high interaction value with AC1 (− 1506.6 kcal/mol) and AV1(− 1255.8 kcal/mol), AC3 (− 1230.8 kcal/mol). The active residues were also changed with the viral protein complexes. (Fig. S7, Table S32).

Silencing of OG2 (G2) genes enhanced virus titer in G. arboreum

TRV-based VIGS assay in cotton is a well-established approach to conducting functional studies of OG2 (Gar06G24920_OG2)genes. FDH-228 plants were inoculated with VIGS constructs to silence the subjected genes, 15 days post infiltration, a completely bleached phenotype was visible on TRV-GrCLA1 inoculated cotton plants, at this time, RT-qPCR based gene silencing was performed (Figs. S8, S9). The results demonstrated a considerable reduction in the expression level of the G2 gene in FDH-228 silenced plants, compared to the TRV:00 inoculated control plants (Fig. 7A). The graft-mediated CLCuD inoculation has been well demonstrated by Ullah et al.58 in G. arboreum plants. Since this breakthrough, it has become possible to identify the resistance imparting genes of G. arboreum against CLCuD. After assessing the gene silencing, an equal number of silenced and control plants were inoculated with CLCuD by grafting CLCuD harboring scions, and exposing them to viruliferous whitefly. 25 days post-CLCuD exposure, enhanced virus concentration was witnessed in G2 (Gar06G24920_OG2) silenced plants in comparison to TRV:00 inoculated controls both under graft and viruliferous whitefly exposure (Fig. 7B). Minor disease symptoms were evident only on graft-inoculated G2 (Gar06G24920_OG2) silenced plants (Fig. 7C). These results suggested the likely involvement of the G2 (Gar06G24920_OG2) gene in CLCuD resistance in FDH-228 plants.

Figure 7
figure 7

qPCR-based estimation of gene expression and CLCuD Symptoms development on VIGS plants. Panel (A), Orange represents a decreased expression of the G2 gene in G. arboreum plants. Green shows showing expression of G2 in empty vector inoculated control plants. Typical CLCuD symptoms of lower grade appeared on G2 VIGS plants. Panel (B) shows qPCR-based estimation of virus titer in both graft and whitefly exposed VIGS plants of G. arboreum. ** is for a significant difference. Panels (C) is showing minor vein thickening on G2 VIGS plants and TRV:00 inoculated FDH-228 plant leaf with no symptoms. The plants were picturized using a microlens-assisted camera of the Apple iPhone 11 pro max model a2161.

Discussion

In the last two decades, several types of resistance (R) genes have been discovered. The majority of R genes found so far are members of the nucleotide-binding site (NBS)-leucine-rich repeat (LRR) receptor (NBS-LRR, also known as NLR) gene family63. NLR genes have been found in plants and their origin may be traced back to the common ancestor of all green plants. Before the split of green plants, phylogenetic studies revealed that NLR genes had diverged into separate subclasses20,64. According to genome-wide studies, different species of plants possess a varying number of NLR genes. It was also observed that the number of gene was correspondence to the ploidy level of plant Species like the tetraploid cotton, Gossypium hirsutum (AD1), and Gossypium barbadense (AD2), had a higher number of NBS genes compared to their progenitors G. arboreum (A2), G. herbaceum (A1), and G. raimondii (D5) (diploid species). Similar results were also observed in other plant species; for instance, bread wheat (Triticum aestivum), a hexaploid, possessed 2012 NLRs19., while its progenitor wild species, Aegilops, had half of the detected wheat, with 742, 800, and 1030 candidate NLRs in Ae. longissima, Ae. sharonensis, and Ae. speltoides, respectively65.Additionally, the simpler genome had fewer NBS candidate genes, like Jia-Yu et al. identified 43 NLR in Marchantia polymorpha66 while our study listed 54 NLR (this study) in the same plants and similar contrast was also found in a few other plants like 443 (literature)67, and 636 (this study) in Solanum tubersum. A list of detailed comparisons of NBS from the literature and this study give significant differences in the number of NBS genes (Table S34). The variation in this study and previous ones might be due to the use of the latest genome assemblies as well as the improved algorithms, which was the main focus of this study. Furthermore, we have also observed that some of the green plant ancestors like Coccomyxa subellipsodiea, Klebsormidium flaccidium, Micromonas pusilla, Physcomitrella Patens, and Volvox carteri do not possess NBS genes, indicating the lack of such resistance gene in ancestor plants. NLR gene expansion and contraction can support the “birth-and-death hypothesis” that predicts fast NLR evolution in land plants, which might be due to duplication events at the genome, and segmental levels68,69.

NLR genes are frequently clustered in complex clusters, a structure that may favor NLR dynamic development and diversification to cope with rapidly evolving pathogens70. The detection of certain pathogen effectors activates NLRs, resulting in a robust immune response that is frequently linked with localized programmed cell death, known as the hypersensitive response (HR)71. Most NLRs feature an N-terminal extension consisting of a Toll/interleukin-1 receptor (TIR) domain, a coiled-coil domain (CC), or a divergent coiled-coil domain (CCR) identical to the Resistance to powdery mildew 8 (RPW8) domain72. Generally, there are three kinds of NLRs based on the N-terminal domain and evolutionary history of the NB-ARC which are (1) TIR-NLRs (TNLs), (2) CC-NLRs (CNLs), and (3) RPW8-NLRs (RNLs)15,73. Our classification system was unique and identified several N-terminal and C-terminal helper domains, in addition to NBS domains, e.g., Zf-BED, WRKY, BRX, LIM, DA1-like, Plant_tran, zf-RVT, Myb_DNA-binding, AP2, DDE_Tnp_4, Thioredoxin, Jacalin, zf-BED, etc. These additional domains are known as immunological receptors with integrated domains (IDs) that resemble pathogen targets and are activated in response to effector change74. Previous literature data divided NLRs into two functional groups: (1) direct/indirect sensor NLRs that detect invasion and (2) helper NLRs that are genetically necessary for immunological activation by other NLRs75,76. Based on the presence and absence of the 2nd group, we have classified all 12,822 NBS genes into 168 classes. The number of different classes in different plant species also gives genetic features of land plants and their evolutions. Some of these identified classes are reported in the literature and well-characterized. As the RNL clade is usually characterized by a low copy number77,78, except Gymnosperms (there are 31 ADR homologs in spruce)79. RNLs exhibit extraordinary intron conservation in Amborella and dicots, which share four introns, but monocots have three introns (the second is missing)80,81. NRG and ADR were separated before angiosperms diverged and they are still retained in syntenic blocks across flowering plants and several lineages have lost NRG genes but not ADR genes82. The TNLs are divided into two subfamilies: (1) TIR1 and (2) TIR283, while in some plants such as monocots, only TIR2 NLRs are maintained83,84. Many dicot species have TIR1 NLRs, however, other dicot lineages are devoid of them85. So, the presence and absence of these newly identified classes in our study further required functional characterization.

The evolutionary study of NBS genes in 33 land plants provided several common patterns of NLR evolution that were not visible in studies of a single species or plant family. A total of 664 orthogroups (OGs) were predicted among the land plants with only 8.2% of the total genes with unique sequences. Of these, we found many core and conserved OGs in plant species-specific lineage. The OG00 was shared with all plants except some lower plants, showing orthologs reduction due to their small and simple genome. The OG2 was another orthogroup that was found as the core orthogroup among land plants. The genome and tandem duplication events also cause the extension of OGs in the genome and cause genomic diversity86,87. The diversity analysis of NBS at the genic and genomics level also demonstrated duplication events at internal and terminal nodes of the phylogenetic tree during the evolutionary process..

The RNA-seq-based expression profiling of the NBS gene in A. thaliana, Z. mays, G. hirsutum, and G. arboreum demonstrates the significant role of the NBS gene in plant growth and development88,89. As in the control of NLR activity, gene transcription is an early regulatory step. To acquire resistance, proper NLR gene transcription is necessary as excessive transcription can cause programmed cell death, which is detrimental to plant growth and development. Overall expression profiling suggested that the OG2, OG6, and OG15 members showed the highest putative role in different tissues, under various biotic and abiotic stresses. The highest expression of these genes in the given plant species under biotic and abiotic stresses might be due to the presence of stress-responsive cis-regulatory elements in the promoter region, which was observed in the promoter sequences of G. hirsutum e.g. Box, ARE, G-box, GT1-motif, ABRE, CGTCA-motif, TGACG-motif, TCT-motif, TCA- element, MBS, GATA-motif, CAT-box and O2-site. In addition to the presence/absence of stress-responsive elements, epigenetic markers such as histone post-translational modifications and DNA methylation can also influence NLR gene expression by modulating their chromatin structure. NLR gene expression may be influenced by DNA methylation at the promoter, as the methylation in the promoter of the Arabidopsis TNL gene (MG1), upregulated flg22 treatment, implying dynamic DNA methylation in NLR promoters during biotic stress. Additionally, the genetic variation in the UTRs, splicing sites, and exonic regions of NBS genes in the susceptible and tolerant G. hirsutum also depicted their role in the two lines under cotton leaf curl disease.

For long-term strategies to control disease in cotton and other important crops, the deployment of genomics tools including, NGS sequencing, exome sequencing, and genotype by sequencing, are helping to identify genetic variations (InDels, SNPs, SSR) associated with the functional diversity, which is further in marker-assisted selection (MAS) approach during breading of wild (resistant) and cultivars (susceptible)46,69. In the current study, we have made a comparison of genetic variation between a highly virus-resistant G. hirsutum accession and a highly susceptible cultivar and found several SNPs and InDels in tolerant accession. As breeding of virus-resistant cotton varieties with sufficient genetic diversity has been suggested as a durable strategy for controlling the disease effectively90,91 and the genetic basis of resistance and its inheritance are the key components for designing breeding strategies. In cotton plants, very little is known about the genetic variants associated with cotton leaf curl disease resistance. Therefore, our identified genetic variants might help develop CLCuD-resistant varieties, which is our next project.

Plant NLR proteins have been discovered to have a central NBS domain that resembles the AAA-ATPase family92,93. The ADP and ATP are the two molecules that have a putative role in the activation and deactivation of the NBS domain94. The binding of ADP with the NBS domain, causes deactivation, while its exchange with ATP opens the active site of the NBS domain and activates the molecular mechanism95. After activation, two intracellular proteins CC-NLR and TIR-NLR recognize viral effector molecules to activate the defense responses96. So, to undertrained the deep mechanism of NBS protein activation in cotton plants, we have presented a comparative molecular interaction of ADP and ATP with NBS proteins and found that the OG2 genes (Gar01G01860OG2, Gar06G24920OG2) showed more affinity to ATP molecule as compared to ADP molecule, suggesting their active role in defense response. Furthermore, the functional validation of OG2 genes (Gar01G01860OG2, Gar06G24920OG2), in G. arboreum (a naturally immune landrace) through virus-inducing gene silencing also increased the viral disease system in silenced plants as compared to control plants.

Conclusion

The identification and analysis of resistance genes in plants have revealed interesting patterns of their evolution and diversification. The majority of these genes belong to the NLR receptor gene family, which can be traced back to the common ancestor of all green plants. The number and type of NLR genes vary greatly among different species, indicating their adaptation to cope with rapidly evolving pathogens. NLRs are frequently clustered in complex clusters, which may facilitate their diversification and development. Different classes of NLRs have been identified based on their N-terminal and C-terminal helper domains, some of which resemble pathogen targets and are activated in response to effector change. While the study of NBS genes in different plant species has revealed common patterns of evolution, the presence and absence of newly identified classes require further functional characterization. Furthermore, the NBS gene plays a significant role in plant growth and development, as demonstrated by RNA-seq expression profiling in various plant species. Proper transcription of the NLR gene is necessary for acquiring resistance, as excessive transcription can cause programmed cell death that is harmful to plant growth. The OG2, OG6, and OG15 members showed the highest putative role in different tissues, under various biotic and abiotic stresses, possibly due to stress-responsive cis-regulatory elements in the promoter region. Finally, the functional validation of OG2 (Gar01G01860OG2, Gar06G24920OG2) in resistant cotton plants demonstrated its putative role in plant resistance during cotton leaf curl disease infection. Overall, these findings provide insights into the genetic features and evolution of land plants.