Decoding Plant Immunity: A Comprehensive Guide to NBS Disease Resistance Gene Architecture and Classification

Abstract

This article provides a systematic overview of the domain architecture and classification of Nucleotide-Binding Site Leucine-Rich Repeat (NBS-LRR) genes, the largest family of plant disease resistance genes. Tailored for researchers and scientists in plant pathology and genomics, it explores the foundational principles of NBS-LRR structure, from core domains to major subfamilies. It details cutting-edge methodological approaches for gene identification, from domain-based searches to deep learning tools, and addresses key challenges in genome annotation and data interpretation. The content further covers validation techniques and comparative evolutionary analyses across diverse plant species, synthesizing knowledge to empower the discovery and functional characterization of resistance genes for crop improvement.

The Building Blocks of Immunity: Core Domains and Major Classes of NBS-LRR Genes

Nucleotide-binding site leucine-rich repeat (NBS-LRR) proteins constitute the largest and most prominent class of disease resistance (R) proteins in plants, serving as critical intracellular immune receptors that mediate effector-triggered immunity (ETI). These proteins function as specialized surveillance systems that detect pathogen effector molecules, initiating robust defense signaling cascades that often culminate in hypersensitive response and programmed cell death to restrict pathogen spread. This technical guide comprehensively examines the domain architecture, classification, molecular mechanisms, and experimental methodologies central to NBS-LRR research, providing researchers with essential frameworks for understanding plant immunity at the molecular level. Through detailed analysis of structural features, signaling pathways, and genomic distribution across diverse plant species, we establish the fundamental principles governing NBS-LRR function in pathogen perception and defense activation.

Domain Architecture and Classification of NBS-LRR Proteins

Core Structural Domains and Functional Modules

NBS-LRR proteins represent some of the largest protein families in plants, characterized by a conserved tripartite domain architecture that enables their dual functions in pathogen recognition and defense signaling. These proteins typically range from approximately 860 to 1,900 amino acids in length and contain at least four distinct domains joined by linker regions [1]:

• Variable amino-terminal domain: Serves as a protein-protein interaction interface and determines subfamily classification
• Nucleotide-binding site (NBS) domain: Also known as the NB-ARC (nucleotide binding adaptor shared by NOD-LRR proteins, APAF-1, R proteins, and CED4) domain, which functions as a molecular switch through ATP/GTP binding and hydrolysis
• Leucine-rich repeat (LRR) region: Composed of tandem repeats that form a solenoid-shaped structure with a parallel β-sheet lining the inner concave surface
• Variable carboxy-terminal domains: Often involved in regulatory functions

The NBS domain contains several defined motifs characteristic of the 'signal transduction ATPases with numerous domains' (STAND) family of ATPases, including conserved sequences known as P-loop, RNBS-A, kinase-2, RNBS-B, RNBS-C, and GLPL motifs that are essential for nucleotide binding and hydrolysis [2]. The LRR domain typically consists of multiple repeats (averaging 14 LRRs per protein) that provide remarkable structural diversity for specific molecular recognition [1].

Classification Systems and Subfamily Organization

NBS-LRR proteins are classified into distinct subfamilies based on their N-terminal domain composition and architectural features. Two primary classification systems have emerged in the literature, reflecting different perspectives on the organizational principles of this diverse protein family [3] [4].

Table 1: NBS-LRR Classification Systems Based on Domain Architecture

Classification System	Subfamily	Domain Composition	Functional Role
Eight-subfamily system [3]	CNL	CC-NBS-LRR	Intracellular receptor in ETI
	TNL	TIR-NBS-LRR	Intracellular receptor in ETI
	RNL	RPW8-NBS-LRR	Defense signaling transduction
	CN	CC-NBS	Potential adaptors/regulators
	TN	TIR-NBS	Potential adaptors/regulators
	NL	NBS-LRR	Recognition and signaling
	N	NBS	Functionally diverse
	RN	RPW8-NBS	Signaling components
Six-subfamily system [4]	TNL	TIR-NBS-LRR	Pathogen recognition
	CNL	CC-NBS-LRR	Pathogen recognition
	NL	NBS-LRR	Recognition and signaling
	TN	TIR-NBS	Regulatory functions
	CN	CC-NBS	Regulatory functions
	N	NBS	Diverse regulatory roles

The classification based on N-terminal domains reveals two major evolutionary lineages: TIR-NBS-LRR (TNL) proteins containing Toll/interleukin-1 receptor domains and CC-NBS-LRR (CNL) proteins featuring coiled-coil motifs [5] [1]. An important evolutionary distinction exists between these subfamilies, as TNL proteins are completely absent from cereal genomes, suggesting lineage-specific loss during monocot evolution [1]. Additionally, a third minor subclass, RPW8-NBS-LRR (RNL), has been identified that functions primarily in downstream defense signaling rather than direct pathogen recognition [6].

Genomic Distribution and Evolutionary Dynamics

NBS-LRR genes represent one of the largest and most diverse gene families in plant genomes, with significant variation in copy number across species:

Table 2: NBS-LRR Gene Family Size Across Plant Species

Plant Species	Genome Type	NBS-LRR Count	Reference
Arabidopsis thaliana	Dicot model	150-207	[1] [7]
Oryza sativa (rice)	Monocot crop	400-505	[1] [7]
Nicotiana tabacum (tobacco)	Allotetraploid dicot	603	[3]
Nicotiana benthamiana	Dicot model	156	[4]
Capsicum annuum (pepper)	Dicot crop	252	[2]
Salvia miltiorrhiza	Medicinal dicot	196	[7]
Triticum aestivum (wheat)	Hexaploid cereal	2,151	[3]
Akebia trifoliata	Fruit crop	73	[3]

NBS-LRR genes are frequently organized in clusters throughout the genome, resulting from both segmental and tandem duplication events [1] [2]. In pepper (Capsicum annuum), 54% of NBS-LRR genes form 47 physical clusters, with chromosome 3 containing the highest number of clusters [2]. This clustered arrangement facilitates rapid evolution through unequal crossing-over and gene conversion, generating substantial diversity in recognition specificities [1]. Evolutionary analyses reveal heterogeneous rates of evolution across different protein domains, with the LRR region exhibiting the highest variability due to diversifying selection that maintains variation in solvent-exposed residues [1].

Molecular Mechanisms of Pathogen Recognition and Signaling Activation

Direct vs. Indirect Pathogen Detection Strategies

NBS-LRR proteins employ two principal strategies for pathogen detection, each with distinct molecular mechanisms and evolutionary implications. The direct recognition model involves physical binding between the NBS-LRR protein and pathogen effector molecules, while the guard model proposes indirect recognition through monitoring host proteins targeted by pathogen effectors [5].

Direct recognition is characterized by specific physical interaction between NBS-LRR proteins and pathogen-derived effectors. Key experimental evidence supporting this mechanism includes:

• The rice Pi-ta protein binding to the Magnaporthe grisea effector AVR-Pita via its LRR domain [5]
• Direct interaction between flax rust resistance proteins (L5, L6, L7) and corresponding fungal AvrL567 effectors in yeast two-hybrid systems [5]
• Association of Arabidopsis RRS1 protein with bacterial PopP2 effector in split-ubiquitin assays [5]

Indirect recognition involves surveillance of host cellular components that are modified by pathogen virulence factors. Well-characterized examples include:

• Arabidopsis RPM1 monitoring the phosphorylation status of RIN4 protein after modification by Pseudomonas syringae effectors AvrRpm1 and AvrB [5]
• Arabidopsis RPS2 detecting proteolytic cleavage of RIN4 by AvrRpt2 protease [5]
• Arabidopsis RPS5 recognizing PBS1 kinase cleavage by AvrPphB cysteine protease [5]
• Tomato Prf protein indirectly detecting AvrPto and AvrPtoB effectors through interaction with Pto kinase [5]

The indirect detection strategy provides an evolutionary advantage by allowing plants to monitor a limited number of key host targets rather than maintaining countless specific receptors for rapidly evolving pathogen effectors [5].

Signaling Activation and Conformational Dynamics

NBS-LRR proteins function as molecular switches that transition between inactive and active states through nucleotide-dependent conformational changes. In the absence of pathogens, these proteins maintain an auto-inhibited ADP-bound state. Upon pathogen recognition, conformational alterations in the amino-terminal and LRR domains promote exchange of ADP for ATP by the NBS domain, activating downstream signaling through mechanisms that remain incompletely understood [5].

The LRR domain plays a critical role in both effector recognition and maintaining autoinhibition. Structural models based on mammalian LRR domains suggest they form barrel-like structures with parallel β-sheets lining the inner concave surface, creating a versatile binding interface [5]. The remarkable diversity of LRR sequences, with 5-10 sequence variants for each repeat across the approximately 14 repeats typical in NBS-LRR proteins, enables recognition of extremely diverse pathogen molecules [1].

Recent evidence indicates that NBS-LRR activation may involve oligomerization, similar to mammalian NOD proteins. The tobacco N protein (a TNL) forms oligomers in response to pathogen elicitors, suggesting this may be a conserved mechanism for signal amplification [1]. Downstream signaling pathways differ between TNL and CNL subfamilies, indicating divergence in defense activation mechanisms despite similar overall architecture [1].

Diagram 1: NBS-LRR mediated immunity signaling pathway. The diagram illustrates the two-layer plant immune system, showing both direct and indirect pathogen recognition mechanisms that activate NBS-LRR proteins and lead to defense responses.

Experimental Approaches for NBS-LRR Gene Identification and Characterization

Genome-Wide Identification and Bioinformatics Pipelines

Comprehensive identification of NBS-LRR genes relies on integrated bioinformatics approaches that leverage conserved domain signatures and advanced computational tools. The standard workflow combines hidden Markov model (HMM) searches with domain validation and phylogenetic analysis [3] [4].

Core Identification Protocol:

• HMM Search Implementation

  • Obtain the NB-ARC domain (PF00931) HMM profile from the Pfam database
  • Perform HMMER searches against target proteomes with expectation value (E-value) thresholds < 1*10⁻²⁰ [4]
  • Extract candidate sequences containing the NBS domain for further validation

• Domain Architecture Analysis

  • Validate NBS domains using Pfam, SMART, and NCBI Conserved Domain Database (CDD) with E-value < 0.01 [4]
  • Identify additional domains (TIR, CC, LRR, RPW8) using respective HMM profiles:
    • TIR domains: PF01582, PF00560, PF07723, PF07725
    • LRR domains: PF12779, PF13306, PF13516, PF13855, PF14580
    • RPW8 domain: PF05659
  • Confirm coiled-coil domains using NCBI CDD or nCoil prediction tools [4]

• Classification and Phylogenetics

  • Classify sequences into subfamilies based on domain composition
  • Perform multiple sequence alignment using MUSCLE or Clustal W with default parameters [3] [4]
  • Construct phylogenetic trees using Maximum Likelihood method in MEGA7/MEGA11 with 1000 bootstrap replicates [4]

Advanced Computational Tools: Recent developments in machine learning have produced specialized tools for R gene prediction. PRGminer represents a deep learning-based approach that employs dipeptide composition analysis to identify resistance genes with 98.75% accuracy in training and 95.72% on independent testing, significantly outperforming traditional alignment-based methods [6].

Expression Analysis and Functional Characterization

Transcriptional profiling and functional validation constitute critical steps in establishing the roles of NBS-LRR genes in disease resistance pathways.

Expression Analysis Methodology: [3]

• RNA-Seq Data Processing

  • Obtain RNA-seq datasets from relevant databases (NCBI SRA)
  • Convert SRA files to FASTQ format using fastq-dump v2.6.3
  • Perform quality control with Trimmomatic v0.36, retaining reads with minimum length of 90bp
  • Map cleaned reads to reference genome using Hisat2

• Transcript Quantification

  • Calculate expression values (FPKM) using Cufflinks v2.2.1
  • Identify differentially expressed genes (DEGs) through Cuffdiff with appropriate statistical thresholds
  • Correlate expression patterns with pathogen challenge or specific treatments

Functional Validation Approaches:

• Virus-Induced Gene Silencing (VIGS): Particularly effective in Nicotiana benthamiana model system [4]
• Heterologous expression in model plants to test resistance specificity [3]
• Overexpression studies to evaluate potential for broad-spectrum resistance [3]
• Gene editing using CRISPR/Cas9 to create knockout mutants for functional analysis [3]

Diagram 2: Experimental workflow for NBS-LRR gene identification and characterization. The pipeline illustrates the integrated bioinformatics and experimental approaches used to identify, classify, and functionally validate NBS-LRR genes.

Research Reagent Solutions for NBS-LRR Studies

Table 3: Essential Research Reagents and Computational Tools for NBS-LRR Investigation

Category	Tool/Reagent	Specific Application	Function
Bioinformatics Tools	HMMER v3.1b2 [3]	Domain identification	Identifies NBS domains using PF00931 model
	Pfam Database [4]	Domain annotation	Validates protein domains and architecture
	MEME Suite [4]	Motif discovery	Identifies conserved protein motifs
	MCScanX [3]	Genome analysis	Detects gene duplication events and synteny
	PRGminer [6]	R gene prediction	Deep learning-based resistance gene identification
Experimental Resources	Virus-Induced Gene Silencing (VIGS) [4]	Functional validation	Rapid gene silencing in Nicotiana models
	CRISPR/Cas9 [3]	Gene editing	Targeted mutagenesis for functional studies
	Yeast two-hybrid systems [5]	Protein interaction	Detects direct effector-NBS-LRR interactions
Biological Materials	Nicotiana benthamiana [4]	Model system	Susceptible host for functional assays
	Arabidopsis T-DNA lines	Mutant resources	Readily available knockout mutants

NBS-LRR proteins represent a sophisticated plant immune surveillance system that has evolved through complex evolutionary processes to provide effective defense against diverse pathogens. Their modular domain architecture enables both specific pathogen recognition and activation of defense signaling, while their genomic organization in clusters facilitates rapid evolution and adaptation to changing pathogen pressures. The distinction between direct and indirect recognition mechanisms reveals strategic evolutionary solutions to the challenge of recognizing highly variable pathogen effectors.

Future research directions will likely focus on elucidating the structural basis of NBS-LRR activation through crystallographic studies, understanding the precise signaling mechanisms that differentiate TNL and CNL pathways, and harnessing natural diversity through pan-genome analyses to identify novel resistance specificities. The development of advanced computational tools like PRGminer demonstrates the growing integration of machine learning approaches to accelerate resistance gene discovery. As our understanding of NBS-LRR function deepens, these insights will directly inform crop improvement strategies aimed at developing durable disease resistance through pyramiding multiple R genes or engineering novel recognition specificities.

The nucleotide-binding site leucine-rich repeat (NBS-LRR) family represents the largest class of plant disease resistance (R) genes, encoding intracellular immune receptors that confer resistance to diverse pathogens including viruses, bacteria, fungi, nematodes, and oomycetes. These proteins typically exhibit a tripartite domain architecture consisting of a variable N-terminal domain, a central nucleotide-binding site (NBS) domain, and a C-terminal leucine-rich repeat (LRR) domain. This architectural configuration enables NBS-LRR proteins to function as sophisticated molecular switches that detect pathogen effectors and initiate robust defense signaling cascades. Understanding the structure-function relationships of these domains provides crucial insights into plant immunity mechanisms and offers opportunities for engineering disease-resistant crops through both traditional breeding and emerging biotechnological approaches.

Plant NBS-LRR proteins are some of the largest proteins known in plants, ranging from approximately 860 to 1,900 amino acids in length [1]. They function as critical sentinels in the plant innate immune system, directly or indirectly recognizing pathogen-derived effector proteins and activating defense responses that often include a form of localized programmed cell death termed the hypersensitive response (HR) [8] [9]. The modular architecture of NBS-LRR proteins has been evolutionarily conserved across land plants, with orthologs identified in non-vascular plants, gymnosperms, and angiosperms [1].

The N-terminal domain, which can be either a Toll/interleukin-1 receptor (TIR) domain or a coiled-coil (CC) domain, defines the two major subfamilies of NBS-LRR proteins: TNLs and CNLs [10] [1]. The central NBS domain (also referred to as NB-ARC) functions as a molecular switch through nucleotide-dependent conformational changes [1]. The C-terminal LRR domain typically contains multiple leucine-rich repeats that form a curved solenoid structure ideal for protein-protein interactions [9]. The number of LRR repeats varies considerably among NBS-LRR proteins, with Arabidopsis NBS-LRRs having a mean of 14 LRRs and a typical repeat length of 24 residues [9].

Table 1: Major Subfamilies of Plant NBS-LRR Proteins

Subfamily	N-Terminal Domain	Representative Members	Key Features	Distribution
TNL	TIR (Toll/Interleukin-1 Receptor)	N gene (tobacco), L6 (flax)	Signals via EDS1/PAD4; absent in monocots	Dicots only
CNL	CC (Coiled-Coil)	Rx (potato), RPS5 (Arabidopsis)	Signals via NRC proteins; widespread	All angiosperms
RNL	RPW8 (Resistance to Powdery Mildew 8)	NRG1, ADR1	Helper NLRs for signal transduction	Limited lineages

The N-Terminal Domain: Signaling and Specificity

The N-terminal domain of NBS-LRR proteins serves critical functions in determining signaling pathway specificity and engaging downstream components of the immune response. TIR-domain-containing NBS-LRR proteins (TNLs) and CC-domain-containing NBS-LRR proteins (CNLs) represent evolutionarily distinct lineages that utilize different signaling mechanisms [1].

TIR Domains

The TIR domain is named for its homology to the intracellular signaling domains of Drosophila Toll and human interleukin-1 receptors [9]. In plants, TIR domains are approximately 175 amino acids in length and contain four conserved motifs [1]. Polymorphisms in the TIR domain can affect pathogen recognition specificity, as demonstrated with the flax TNL protein L6 [1]. Additionally, many TNLs contain an alanine-polyserine motif immediately adjacent to the N-terminal methionine that may be involved in protein stability [1]. TIR domains are thought to function in protein-protein interactions, potentially with the proteins being "guarded" or with downstream signaling components [1].

CC Domains

The CC domain is a structural motif characterized by heptad repeats that facilitate protein oligomerization. In many CNLs, the CC motif spans approximately 175 amino acids N-terminal to the NBS domain [1]. However, some CNLs exhibit substantial variation in their N-terminal regions; for instance, the tomato Prf protein has 1,117 amino acids N-terminal of the NBS domain, much of which is unique to this protein [1]. Functional studies of the potato Rx protein demonstrated that its CC domain is both necessary and sufficient for complementing a version of Rx lacking this domain (NBS-LRR) when co-expressed in trans [8].

The NBS Domain: A Molecular Switch Mechanism

The NBS domain, also known as the NB-ARC (nucleotide binding adaptor shared by NOD-LRR proteins, APAF-1, R proteins and CED4) domain, serves as a molecular switch that regulates NBS-LRR protein activation through nucleotide-dependent conformational changes [1]. This domain contains several defined motifs characteristic of the 'signal transduction ATPases with numerous domains' (STAND) family of ATPases, which function as molecular switches in disease signaling pathways [1].

Conserved Motifs and Structure

The NBS domain contains multiple conserved motifs, including the kinase 1a (P-loop), kinase 2, and kinase 3a motifs common to a large variety of nucleotide-binding proteins [8]. In Arabidopsis, eight conserved NBS motifs have been identified, with the NBS domains of TNLs and CNLs distinguished by the sequences of three resistance NBS (RNBS) motifs within them (RNBS-A, RNBS-C, and RNBS-D motifs) [1]. Threading plant NBS domains onto the crystal structure of human APAF-1 has provided insights into the spatial arrangement and function of these conserved motifs [1].

Nucleotide Binding and Hydrolysis

Specific binding and hydrolysis of ATP has been demonstrated for the NBS domains of several plant NBS-LRR proteins, including the tomato CNLs I2 and Mi [1]. The current model suggests that in the resting state, the NBS domain is bound to ADP, and upon pathogen recognition, ADP is exchanged for ATP, resulting in conformational changes that activate downstream signaling [4]. This nucleotide-dependent switching mechanism is crucial for the proper regulation of NBS-LRR proteins, preventing inappropriate activation in the absence of pathogens while enabling rapid response upon pathogen detection.

Table 2: Conserved Motifs in the NBS Domain

Motif Name	Consensus Sequence	Functional Role	Subfamily Specificity
P-loop (Kinase 1a)	GxGGLGKT	Phosphate binding of nucleotide	Common to TNLs and CNLs
Kinase 2	LVLDDVW	Mg²⁺ binding and catalysis	Common to TNLs and CNLs
Kinase 3a	GSRII	nucleotide binding	Common to TNLs and CNLs
RNBS-A	FDxxDER	Domain structure/function	Divergent between TNLs and CNLs
RNBS-C	FLhMCfY	Domain structure/function	Divergent between TNLs and CNLs
RNBS-D	CFLYC	Redox regulation?	Divergent between TNLs and CNLs

The LRR Domain: Versatility in Recognition and Regulation

The C-terminal LRR domain represents one of the most versatile components of the NBS-LRR architecture, participating in multiple aspects of protein function beyond pathogen recognition. LRR domains are characterized by a conserved pattern of hydrophobic leucine residues and adopt a slender, arc-shaped structure with a high surface-to-volume ratio that maximizes interaction potential [9].

Structural Features

Each LRR typically consists of a β-strand followed by more variable sequences that form loops, with multiple repeats stacking to create a super-helical structure [9]. The β-strands align to form a continuous β-sheet along the concave surface of the arc, while regularly spaced leucine residues face inward to form a stable hydrophobic core [9]. The structure is further stabilized by the conserved asparagine residue in each repeat [9]. While no plant NBS-LRR proteins have had their structures fully resolved, modeling of the RPS5 LRR domain based on bovine decorin suggests compatibility with the characteristic LRR architecture, despite limited sequence identity (~14%) [9].

Functional Versatility

The LRR domain contributes to multiple aspects of NBS-LRR function:

• Pathogen Recognition: The LRR domain can directly bind pathogen effectors, as demonstrated by the interaction between the rice Pi-ta LRR domain and the fungal effector Avr-Pita [9]. Positive selection is often strongest in solvent-exposed residues of the β-sheet, consistent with direct pathogen recognition [9].

• Intramolecular Interactions: Studies with the potato Rx protein revealed that the LRR domain interacts physically with the CC-NBS region in planta, and this interaction is disrupted in the presence of the coat protein elicitor [8].

• Autoregulation: The LRR domain maintains NBS-LRR proteins in an autoinhibited state in the absence of pathogen effectors. Certain mutations in the LRR, such as the VLDL to VLEL mutation in Rx, can lead to constitutive activation of defense responses [9].

Domain Interactions and Activation Mechanisms

The coordinated interactions between the three major domains govern the activation and regulation of NBS-LRR proteins. Research on the potato Rx protein, which confers resistance to Potato Virus X (PVX), has provided particularly insightful models of these intramolecular relationships.

Complementation in Trans

Remarkably, co-expression of the CC-NBS and LRR regions of Rx as separate molecules results in a coat protein-dependent hypersensitive response, demonstrating that functional resistance can be reconstituted through physical interactions between domains [8]. Similarly, the CC domain alone can complement a version of Rx lacking this domain (NBS-LRR) when co-expressed in trans [8]. These interactions have been confirmed through co-immunoprecipitation experiments, which showed physical interactions between CC-NBS and LRR domains, as well as between CC and NBS-LRR domains [8].

Sequential Disruption Model

The current model for Rx activation proposes that pathogen recognition initiates a sequence of conformational changes involving the disruption of at least two intramolecular interactions [8]. In this model:

• The LRR domain interacts with the CC-NBS region to maintain an autoinhibited state
• Recognition of the PVX coat protein disrupts this interaction
• Subsequent conformational changes enable signaling through the CC domain
• The interaction between CC and NBS-LRR is dependent on a wild-type P-loop motif, whereas the interaction between CC-NBS and LRR is not [8]

This model highlights the sophisticated regulatory mechanisms that prevent inappropriate activation of these potent immune receptors while enabling rapid response upon pathogen detection.

Figure 1: NBS-LRR Activation Mechanism. The model depicts the transition from inactive ADP-bound state to active ATP-bound state upon pathogen recognition, leading to defense response activation.

Experimental Approaches for Studying NBS-LRR Architecture

Domain Complementation Assays

The functional relationships between NBS-LRR domains can be investigated through domain complementation assays, as exemplified by studies with the potato Rx protein:

Protocol: Transient Expression and HR Assay

• Clone individual domains (CC, NBS, LRR) or domain combinations (CC-NBS, NBS-LRR) into appropriate expression vectors with epitope tags (e.g., HA tag)
• Co-infiltrate Nicotiana benthamiana leaves with Agrobacterium strains expressing different domain combinations along with the pathogen elicitor (e.g., PVX coat protein)
• Monitor for hypersensitive response development over 24-72 hours
• Verify protein expression and interactions through co-immunoprecipitation and western blotting [8]

Key Findings from Rx Studies:

• Co-expression of CC-NBS and LRR as separate molecules resulted in CP-dependent HR
• CC domain alone complemented NBS-LRR for CP-dependent HR
• The LRR is required for activation of signaling domains, not just elicitor recognition
• Complementation specificity varies; N. benthamiana CC-NBS-LRR proteins do not complement Rx NBS-LRR, while potato homologs do [8]

Genome-Wide Identification and Characterization

Bioinformatic approaches enable comprehensive identification and classification of NBS-LRR genes across plant genomes:

Protocol: HMM-Based Identification

• Retrieve the hidden Markov model (HMM) profile for the NB-ARC domain (PF00931) from the Pfam database
• Perform HMMsearch against the target genome with an E-value cutoff (typically < 1×10⁻²⁰)
• Extract candidate protein sequences and verify the presence of NBS domains using the Pfam database (E-value < 0.01)
• Classify proteins into subfamilies using conserved domain databases (CDD) and coiled-coil prediction tools (e.g., Coiledcoil with threshold 0.5)
• Analyze gene structures, conserved motifs, phylogenetic relationships, and chromosomal distributions [10] [4]

Application Examples:

• Identification of 73 NBS genes in Akebia trifoliata (50 CNL, 19 TNL, 4 RNL) [10]
• Characterization of 156 NBS-LRR homologs in Nicotiana benthamiana (5 TNL, 25 CNL, 23 NL, 2 TN, 41 CN, 60 N) [4]
• Comparative analysis of 239 NBS-LRR genes across Vernicia fordii (90) and Vernicia montana (149) genomes [11]

Table 3: Research Reagent Solutions for NBS-LRR Studies

Reagent/Tool	Application	Function	Example/Reference
HMMER Software	Genome-wide identification	Identifies NBS domains using hidden Markov models	[10] [4] [11]
Agrobacterium tumefaciens	Transient expression	Delivers genetic constructs into plant cells for functional assays	[8]
Virus-Induced Gene Silencing (VIGS)	Functional characterization	Knocks down gene expression to assess function	[11]
Co-immunoprecipitation	Protein interaction studies	Validates physical interactions between domains	[8]
MEME Suite	Motif analysis	Identifies conserved protein motifs	[10] [4]
Epitope Tags (HA, FLAG)	Protein detection and purification	Enables tracking and isolation of expressed proteins	[8]

Figure 2: Genomic Identification Workflow. The pipeline depicts bioinformatic approaches for genome-wide identification and characterization of NBS-LRR genes.

Emerging Applications and Future Directions

CRISPR Activation for NBS-LRR Gene Regulation

CRISPR activation (CRISPRa) technology represents a promising approach for modulating NBS-LRR gene expression without introducing permanent genomic changes. This system employs a deactivated Cas9 (dCas9) fused to transcriptional activators to achieve targeted gene upregulation [12]. Unlike conventional CRISPR editing that introduces double-stranded breaks, CRISPRa allows quantitative and reversible gene activation while preserving the native genomic context [12].

Applications in Disease Resistance:

• CRISPRa has been successfully used to upregulate pathogenesis-related genes, enhancing defense against bacterial pathogens in tomato [12]
• Epigenetic reprogramming of defense genes through CRISPRa systems has improved somatic embryo induction and maturation in Micro-Tom tomato [12]
• The technology shows particular promise for gain-of-function studies that can reveal the roles of NBS-LRR genes in disease resistance, especially when functional redundancy obscures phenotypes in loss-of-function approaches [12]

Integration of NBS-LRR Genes in Crop Breeding

The modular architecture of NBS-LRR proteins presents opportunities for engineering novel disease resistance specificities in crop plants. Recent research has revealed that some NBS-LRR genes influence both disease resistance and agronomic traits, highlighting the importance of understanding their pleiotropic effects. For example, the rice GL6.1 gene encodes a CC-NBS-LRR protein that functions as a negative regulator of grain length while also interacting with OsWRKY53 to mediate disease resistance signaling [13]. This dual functionality suggests that breeding efforts must carefully balance resistance and yield traits.

The tripartite architecture of NBS-LRR proteins represents a sophisticated molecular framework for pathogen perception and defense activation in plants. The modular nature of these proteins, with distinct N-terminal, NBS, and LRR domains, enables both precise regulation in the absence of pathogens and rapid response upon pathogen detection. The functional independence yet cooperative interactions between these domains, as demonstrated by complementation assays, reveals the remarkable evolutionary optimization of these immune receptors. Emerging technologies such as CRISPR activation offer promising avenues for harnessing NBS-LRR genes in crop improvement, while advanced genomic approaches continue to reveal the diversity and evolution of this critical gene family. Future research elucidating the structural basis of domain interactions and activation mechanisms will undoubtedly provide new insights for engineering durable disease resistance in agricultural crops.

Plants rely on a sophisticated innate immune system to defend against pathogens. A critical component of this system is the nucleotide-binding leucine-rich repeat receptors (NLRs), which are intracellular immune receptors that recognize pathogen effector proteins and initiate effector-triggered immunity (ETI) [14]. NLRs represent the largest family of plant resistance (R) genes and are found across all land plants, with their origins tracing back to green algae [14] [15]. These proteins typically exhibit a modular domain architecture consisting of a central nucleotide-binding domain (NBD), a C-terminal leucine-rich repeat (LRR) domain, and a variable N-terminal domain that defines their classification into major subfamilies [14]. The NBD belongs to the STAND (signal transduction ATPases with numerous domains) family and acts as a nucleotide-dependent molecular switch, cycling between inactive ADP-bound and active ATP-bound states [14]. The LRR domain is involved in protein-protein interactions and often responsible for specific pathogen recognition [16]. This technical guide provides an in-depth analysis of the classification, structure, function, and evolution of the three major NLR subfamilies: CNL, TNL, and RNL, framed within the broader context of domain architecture and classification of NBS disease resistance genes research.

Classification and Domain Architecture

The classification of plant NLR genes is primarily based on the identity of their N-terminal domain, which determines their signaling mechanisms and downstream partners [14] [15]. The three major subfamilies are:

• CNL (CC-NBS-LRR): Characterized by an N-terminal coiled-coil (CC) domain [14] [16]
• TNL (TIR-NBS-LRR): Features an N-terminal Toll/Interleukin-1 receptor (TIR) domain [14] [16]
• RNL (RPW8-NBS-LRR): Contains an N-terminal resistance to powdery mildew 8 (RPW8) domain [14] [16]

Additionally, many NLR genes deviate from this canonical structure and may lack one or more domains, forming irregular types such as CN (CC-NBS), TN (TIR-NBS), and NL (NBS-LRR) proteins [4] [17]. These truncated forms often function as adaptors or regulators for the typical NLR types [4].

Table 1: Major NLR Subfamilies and Their Characteristics

Subfamily	N-terminal Domain	Primary Function	Signaling Pathway	Representative Species Distribution
CNL	Coiled-coil (CC)	Pathogen sensor	NDR1-dependent	All land plants
TNL	TIR (Toll/Interleukin-1 Receptor)	Pathogen sensor	EDS1-dependent	Most angiosperms (lost in monocots)
RNL	RPW8 (Resistance to Powdery Mildew 8)	Helper NLR	EDS1-dependent	All land plants

Structural and Functional Divergence

The structural differences between NLR subfamilies underlie their functional specialization. CNL and TNL proteins generally function as sensor NLRs that directly or indirectly recognize pathogen effectors, either through direct interaction with effectors or by monitoring host proteins targeted by effectors [14] [15]. In contrast, RNL proteins act as helper NLRs that assist in downstream immune signal transduction for both TNL and CNL sensors [14] [15]. Recent structural studies have revealed that upon activation, NLRs undergo conformational changes that enable them to form oligomeric complexes called resistosomes, which act as signaling hubs to initiate immune responses [14].

The central NBS domain contains several conserved motifs that are crucial for nucleotide binding and hydrolysis. These include the P-loop, RNBS-A, kinase-2, RNBS-B, RNBS-C, GLPL, RNBS-D, and MHDV motifs [18]. The kinase-2 motif may regulate ATP hydrolysis, while the P-loop, GLPL, and MHDV motifs are involved in nucleotide binding [18]. Mutations in these motifs, such as in the MHDV region of the tomato I-2 gene or the P-loop region of Arabidopsis RPM1, can lead to constitutive activation or inactivation of the protein [18].

Genomic Distribution and Evolution

NLR genes exhibit remarkable variation in abundance and composition across plant species, independent of genome size [19]. This diversity results from frequent gene duplication and loss events, which have shaped the NLR repertoire throughout plant evolution [15].

Table 2: NLR Gene Distribution Across Plant Species

Species	Total NLRs	CNL	TNL	RNL	Other/Partial	Reference
Arabidopsis thaliana	165	52	106	7	-	[15]
Nicotiana benthamiana	156	25 (CNL), 41 (CN)	5 (TNL), 2 (TN)	4 (across types)	23 (NL), 60 (N)	[4]
Solanum lycopersicum (tomato)	321	211 (full-length total across CNL, TNL, RNL)			110 (partial domains)	[17]
Akebia trifoliata	73	50	19	4	-	[10]
Prunus persica (peach)	286	153 (Subfamily I)	104 (Subfamily II)	11 (Subfamily III)	18 (Subfamily IV)	[18]
Oryza sativa (rice)	498	497	0	1	-	[15]

Evolutionary Patterns and Lineage-Specific Adaptations

The evolution of NLR genes in angiosperms has proceeded in two distinct stages: a period of relatively low gene numbers from the origin of angiosperms until the Cretaceous-Paleogene (K-Pg) boundary, followed by a dramatic expansion after the K-Pg boundary that led to the extensive NLR repertoires observed today [15]. Different plant families exhibit distinct evolutionary patterns: Brassicaceae shows "first expansion and then contraction," Poaceae displays a "contraction" pattern, while Fabaceae and Rosaceae maintain consistent expansion [15].

A significant evolutionary phenomenon is the differential loss of TNL genes in specific lineages. For instance, TNLs are absent in most monocots, including economically important crops like rice, as well as in several magnoliids and certain eudicot lineages like Ranunculales and Lamiales [20] [15]. Genomic evidence suggests that the loss of TNLs in monocots occurred through a process where non-TNL genes replaced the ancestral TNL subclass in syntenic genomic regions [20]. This loss is often associated with deficiencies in the corresponding immune signaling pathway components [15].

Experimental Identification and Classification Protocols

Genome-Wide Identification of NLR Genes

The standard workflow for identifying NLR genes from genome sequences involves a multi-step domain-based approach:

• Initial Domain Search: Perform HMMER searches using the NB-ARC domain model (PF00931) from the PFAM database with an expectation value (E-value) cutoff of <1*10⁻²⁰ [16] [4] [10].

• Domain Verification: Confirm the presence of additional domains using:

  • PFAM domains (TIR: PF01582; LRR: PF00560, PF07723, PF07725, PF12779, PF13306, PF13516, PF13855, PF14580; RPW8: PF05659) [16] [10]
  • NCBI Conserved Domain Database (CDD) for CC domains [16]
  • Coiled-coil prediction tools (e.g., Coiledcoil with threshold 0.5) [10]

• Sequence Validation: Remove duplicate genes and verify domain completeness through manual inspection [4].

• Classification: Categorize genes based on domain composition into CNL, TNL, RNL, and irregular types (CN, TN, NL, N) [16] [4].

Phylogenetic and Structural Analysis

Following identification, comprehensive characterization of NLR genes involves:

• Multiple Sequence Alignment: Using tools like MUSCLE or CLUSTALW with default parameters [16] [17].

• Phylogenetic Tree Construction: Employing maximum likelihood method in MEGA (v.11 or X) with 1000 bootstrap replications to assess evolutionary relationships [16] [4] [17].

• Motif Analysis: Predicting conserved motifs with MEME suite, typically set to identify 10 motifs with width lengths of 6-50 amino acids [4] [10].

• Gene Structure Analysis: Visualizing exon-intron structures using TBtools or GSDS2.0 based on GFF3 annotation files [4] [10].

• Cis-element Analysis: Identifying regulatory elements in promoter regions (1500 bp upstream) using PlantCARE database [4] [17].

Signaling Pathways and Immune Mechanisms

The activation mechanisms and signaling pathways differ significantly between NLR subfamilies. Sensor CNLs and TNLs generally employ a two-step mechanism for pathogen detection and immune activation [14] [4].

NLR Activation and Signaling Cascade

In the resting state, NLRs exist in an autoinhibited conformation with the LRR domain folding back onto the NBS domain, maintaining it in an ADP-bound state [14] [15]. Upon pathogen recognition, conformational changes enable ADP-ATP exchange, promoting oligomerization into resistosome complexes [14]. For CNL proteins like ZAR1, this oligomerization forms a calcium-permeable channel that triggers downstream immune responses [15]. TNL proteins, upon activation, often utilize the EDS1 (enhanced disease susceptibility 1) family proteins as central signaling components, which in turn activate helper RNLs (NRG1 and ADR1 lineages) to amplify the immune response [14].

Research Reagent Solutions and Experimental Tools

Table 3: Essential Research Reagents and Tools for NLR Studies

Reagent/Tool	Function/Application	Example Sources/References
HMMER v3.1b2	Hidden Markov Model searches for domain identification	[16] [4]
PFAM Database	Repository of protein domain families and HMM profiles	[16] [4]
NB-ARC Domain (PF00931)	Core domain model for initial NLR identification	[16] [4] [10]
MEME Suite	conserved motif discovery and analysis	[4] [10]
NCBI CDD	Conserved domain identification and analysis	[16] [17]
MEGA Software	Phylogenetic tree construction and evolutionary analysis	[16] [4] [17]
TBtools	Bioinformatics toolkit for visualization and analysis	[4] [10]
PlantCARE Database	cis-regulatory element prediction in promoter sequences	[4] [17]

Expression Analysis and Functional Validation

Understanding the expression patterns and functional roles of NLR genes is crucial for characterizing their biological significance. Most NLR genes are expressed at low levels under normal conditions, with some showing tissue-specific expression or induction upon pathogen infection [15] [10].

Transcriptomic Approaches

• RNA-seq Analysis: Process raw sequencing data (SRA format) using fastq-dump for format conversion, followed by quality control with Trimmomatic (minimum read length of 90 bp) [16]. Map cleaned data to reference genomes using Hisat2 and perform transcript quantification with Cufflinks with FPKM normalization [16].

• Differential Expression Analysis: Identify differentially expressed NLR genes using Cuffdiff, comparing infected vs. control samples [16].

• Validation: Confirm expression patterns through qPCR analysis of selected NLR genes under infection conditions [17].

In disease resistance studies, multiple NLR genes typically show upregulation upon pathogen infection. For instance, in peach, 22 NLR genes were significantly upregulated after Green Peach Aphid infestation [18]. Similarly, in tomato, specific NLR genes (Solyc04g007060 [NRC4] and Solyc10g008240 [RIB12]) showed consistent upregulation patterns in response to late blight infection [17].

The classification of NLR proteins into CNL, TNL, and RNL subfamilies reflects fundamental functional specializations within the plant immune system. While CNL and TNL proteins primarily act as pathogen sensors, RNL proteins serve as helper NLRs that amplify defense signals. The distinctive domain architecture of each subfamily dictates their specific signaling pathways and activation mechanisms. Genomic studies have revealed tremendous diversity in NLR composition across plant species, shaped by lineage-specific expansions and losses, particularly affecting the TNL subfamily. The experimental framework for NLR identification and characterization continues to evolve with advancements in bioinformatics and genomics, enabling researchers to better understand the complex roles of these critical immune receptors in plant defense. This classification system provides an essential foundation for future research aimed at elucidating the molecular mechanisms of plant immunity and developing novel strategies for crop improvement.

The canonical model for Nucleotide-Binding Site Leucine-Rich Repeat (NBS-LRR) genes defines three major classes based on N-terminal domains: TIR-NBS-LRR (TNL), CC-NBS-LRR (CNL), and RPW8-NBS-LRR (RNL). However, genome-wide studies across diverse plant species have revealed a substantial prevalence of atypical and truncated forms that deviate from this standard architecture. These non-canonical variants—classified as CN (Coiled-Coil NBS), TN (TIR-NBS), NL (NBS-LRR), and N (NBS-only) types—lack complete domain complements yet play significant functional roles in plant immune signaling networks. Their abundance suggests evolutionary plasticity within the NBS gene family and highlights the limitations of rigid classification systems that only recognize full-length proteins.

The identification of these truncated forms has accelerated with the increasing availability of high-quality plant genomes. For instance, studies in Akebia trifoliata identified 73 NBS genes with 50 CNL, 19 TNL, and 4 RNL genes, but also documented multiple truncated forms [21]. Similarly, analysis of Vernicia fordii and Vernicia montana revealed 90 and 149 NBS-LRR genes respectively, with notable occurrences of domain-deficient types including CC-NBS (37 in V. fordii) and NBS-only (29 in both species) [22]. These truncated genes are not genomic artifacts but functional components of plant immune systems, often involved in signal modulation, network regulation, and compensatory functions within complex resistance pathways.

Classification and Genomic Characteristics of Atypical NBS Genes

Defining Atypical NBS Architectures

Atypical NBS genes are characterized by the absence of one or more domains typically associated with full-length NBS-LRR proteins. The CN-type possesses a Coiled-Coil domain followed by an NBS domain but lacks the C-terminal LRR region. The TN-type contains TIR and NBS domains without LRRs. The NL-type has NBS and LRR domains but lacks a defined N-terminal signaling domain (TIR, CC, or RPW8). The N-type contains only the central NBS domain without either flanking domain. These structural variations likely correspond to functional specializations within plant immune networks.

CN-type (CC-NBS) genes typically retain the N-terminal coiled-coil domain known for mediating protein-protein interactions, coupled with the nucleotide-binding domain that functions as a molecular switch. In sunflower, genome-wide analysis identified 100 genes belonging to the CNL group including 64 genes with RX_CC-like domain, plus additional CN types [23]. The conservation of the CC domain suggests these truncated forms may retain signaling capabilities or function as regulators of full-length CNL proteins.

TN-type (TIR-NBS) genes maintain the TIR domain associated with signaling in the TIR-NBS-LRR class, along with the NBS domain, but lack the LRR region typically responsible for pathogen recognition. A striking example is TIR-NBS2 (TN2), an atypical NLR protein that lacks the LRR domain but remains functional in immunity [24]. Research demonstrates that TN2 interacts with EXO70B1, an exocyst complex subunit, and is required for activated disease resistance responses in Arabidopsis, proving that the LRR domain is not always essential for immune function [24].

NL-type (NBS-LRR) and N-type (NBS-only) genes represent progressively more minimal architectures. The NL-type retains the LRR domain potentially enabling pathogen recognition, while the N-type consists essentially of the core nucleotide-binding domain. In Vernicia species, N-types represent a significant portion of the NBS repertoire, with 29 identified in both V. fordii and V. montana [22].

Genomic Distribution and Evolutionary Patterns

Atypical NBS genes display distinctive genomic distribution patterns that provide insights into their evolutionary origins. Comparative analysis of four Gossypium species revealed that NBS genes are distributed nonrandomly across chromosomes, often forming clusters where typical and atypical genes frequently co-localize [25]. This clustering facilitates the generation of structural diversity through unequal crossing over and gene conversion events.

Table 1: Distribution of Atypical NBS Genes in Various Plant Species

Plant Species	CN-type	TN-type	NL-type	N-type	Total NBS Genes	Reference
Akebia trifoliata	Information not available in search results				73	[21]
Vernicia fordii	37	0	12	29	90	[22]
Vernicia montana	87	7	12	29	149	[22]
Gossypium arboreum	Present (quantity not specified)	Present (quantity not specified)	Present (quantity not specified)	Present (quantity not specified)	246	[25]
Gossypium hirsutum	Present (quantity not specified)	Present (quantity not specified)	Present (quantity not specified)	Present (quantity not specified)	588	[25]
Sunflower	100 CNL (includes 64 with RX_CC); CN types not quantified separately	Information not available in search results	Information not available in search results	162 NL	352	[23]

Evolutionary analyses indicate that atypical NBS genes arise primarily through duplication and divergence processes. A study of NBS genes in Vernicia species identified 43 orthologs between resistant V. montana and susceptible V. fordii, with distinct expression patterns suggesting functional differentiation [22]. The researchers noted that in the susceptible V. fordii, "no TIR domains were found in VfNBS-LRRs, indicating that none of the resistance genes in V. fordii belonged to the TIR class," highlighting how species-specific evolutionary pressures shape NBS gene repertoires [22].

Tandem and dispersed duplications represent the two main mechanisms generating NBS gene diversity. In Akebia trifoliata, these processes produced 33 and 29 genes respectively, continuously expanding and diversifying the NBS repertoire [21]. The high sequence similarity between atypical genes and their full-length counterparts suggests most truncations arise from relatively recent duplication events followed by domain loss.

Research Methodologies for Identifying and Characterizing Atypical NBS Genes

Genome-Wide Identification Pipeline

The standard workflow for identifying atypical NBS genes combines multiple bioinformatic approaches to ensure comprehensive detection. The typical process begins with Hidden Markov Model (HMM) profiling using the NB-ARC domain (Pfam accession: PF00931) as query against the entire predicted proteome of the target organism [23] [26] [22]. For example, in the sunflower genome study, this approach identified 352 NBS-encoding genes from 52,243 putative protein sequences [23].

The subsequent domain architecture analysis employs multiple tools: the NCBI Conserved Domain Database for detecting TIR (PF01582), RPW8 (PF05659), and LRR (PF08191) domains; and specialized tools like Coiled-coil prediction algorithms with a threshold P-value of 0.5 for identifying CC domains [21]. This multi-step verification ensures accurate classification of truncated forms that might be missed by single-method approaches.

Key experimental considerations for this workflow include:

• Using an E-value cutoff of 1.0 for initial HMM and BLAST searches to maximize sensitivity [21]
• Applying a second verification step with Pfam analysis at E-value 10^-4 to eliminate false positives [26] [21]
• Manually checking ambiguous cases by comparing results from multiple domain databases
• Validating gene models through transcriptome data where available to confirm expression

Recent advancements in this pipeline include the RGAugury automated tool that systematically identifies not only NBS-encoding genes but also receptor-like kinases (RLKs) and receptor-like proteins (RLPs), collectively termed Resistance Gene Analogs (RGAs) [23]. This automated approach facilitates comparative analyses across multiple genomes, enabling researchers to identify conserved and lineage-specific atypical NBS genes.

Expression Analysis and Functional Validation

Transcriptomic approaches provide critical insights into the functional relevance of atypical NBS genes. Research in Akebia trifoliata demonstrated that NBS genes generally express at low levels, with a few showing relatively high expression during later development in rind tissues [21]. This pattern suggests these genes may have specialized roles in specific tissues or developmental stages rather than constituting redundant components.

For functional validation, Virus-Induced Gene Silencing (VIGS) has emerged as a powerful technique. In a study of Vernicia montana resistance to Fusarium wilt, researchers used VIGS to silence a candidate NBS-LRR gene (Vm019719), demonstrating its essential role in disease resistance [22]. The experimental protocol involves:

• Cloning a 300-500 bp fragment of the target gene into TRV-based vectors
• Introducing constructs into Agrobacterium tumefaciens strain GV3101
• Infiltrating leaves of 2-3 week-old plants using needleless syringes
• Monitoring silencing efficiency 2-3 weeks post-infiltration via qRT-PCR
• Challenging silenced plants with pathogens and assessing disease symptoms

This approach confirmed that Vm019719, activated by the transcription factor VmWRKY64, confers resistance to Fusarium wilt in V. montana [22]. In the susceptible V. fordii, the allelic counterpart (Vf11G0978) showed an ineffective defense response due to a deletion in the promoter's W-box element, highlighting how regulatory mutations in atypical NBS genes can impact disease resistance.

Functional Mechanisms and Signaling Pathways

Integrated Immune Signaling Networks

Atypical NBS genes function within complex immune networks rather than as isolated components. The NRC (NLR-REQUIRED FOR CELL DEATH) immune receptor network provides a compelling example of this integration. In asterid plants, this network has evolved from a pair of linked genes into a genetically dispersed and phylogenetically structured network of sensor and helper NLR proteins [27]. Within this network, atypical members like NRCX modulate the activities of key helper NLR nodes during plant growth [27].

Diagram: Simplified NRC Immune Network Showing Atypical NBS Function

Research on NRCX demonstrates that systemic gene silencing of this atypical NBS gene in Nicotiana benthamiana markedly impairs plant growth, resulting in a dwarf phenotype [27]. This growth impairment is partially dependent on NRCX paralogs NRC2 and NRC3, indicating that NRCX maintains NRC network homeostasis by balancing immune responsiveness and growth [27]. This regulatory function exemplifies how atypical NBS genes can evolve modulatory roles within complex immune networks.

Molecular Functions and Interaction Partners

At the molecular level, atypical NBS genes engage in diverse interactions with host proteins. TN2 (TIR-NBS2), a TN-type gene, physically associates with EXO70B1, a subunit of the exocyst complex involved in secretory pathways [24]. This interaction provides a link between the exocyst complex and immune signaling, suggesting that TN2 may monitor EXO70B1 integrity as part of an immune surveillance mechanism [24].

Table 2: Documented Molecular Functions of Atypical NBS Genes

Gene/Type	Species	Molecular Function	Interaction Partners	Biological Role
TN2 (TN-type)	Arabidopsis thaliana	Exocyst complex monitoring; immune activation	EXO70B1 (exocyst subunit)	Activated disease resistance to powdery mildew [24]
NRCX (CNL-related)	Nicotiana benthamiana	Network homeostasis; modulation of helper NLRs	NRC2, NRC3 (helper NLRs)	Balancing growth and immunity; preventing autoimmunity [27]
Vm019719 (NL-type)	Vernicia montana	Pathogen recognition; defense activation	VmWRKY64 (transcription factor)	Fusarium wilt resistance [22]
CN-types	Various species	Signaling modulation; decoy function	Full-length CNL proteins	Regulation of immune signaling networks

The functional significance of domain composition in atypical NBS genes is exemplified by the discovery that the "MADA motif" in the α1 helix of ZAR1 and about one-fifth of angiosperm CC-NLRs functions as a death switch [27]. This motif is interchangeable between distantly related NLRs, indicating that the 'death switch' mechanism applies to MADA-CC-NLRs from diverse plant taxa [27]. In atypical forms, the presence or absence of this motif likely determines functional capabilities.

The Scientist's Toolkit: Essential Research Reagents and Methods

Table 3: Essential Research Reagents and Solutions for Studying Atypical NBS Genes

Reagent/Solution	Application	Function	Example Use
TRV-based VIGS vectors	Functional validation	Gene silencing in plants	Silencing NBS genes to assess function in disease resistance [22]
HMM profile (NB-ARC domain PF00931)	Bioinformatics identification	Identifying NBS domains in protein sequences	Genome-wide scans for NBS-encoding genes [23] [26]
qRT-PCR reagents	Expression analysis	Quantifying transcript levels	Measuring NBS gene expression under different conditions [22]
Agrobacterium tumefaciens GV3101	Plant transformation	Delivering genetic constructs into plant tissues	VIGS experiments; stable transformation [22]
Domain prediction tools (CDD, Pfam, SMART)	Protein classification	Identifying functional domains	Classifying NBS genes into CN, TN, NL, and N types [22] [28]
Phylogenetic analysis software	Evolutionary studies	Reconstructing gene families	Understanding evolutionary relationships among atypical NBS genes [21]

Atypical CN, TN, NL, and N-type NBS genes represent functionally significant components of plant immune systems rather than mere genomic artifacts. Their diverse domain architectures reflect evolutionary specialization for modulatory, regulatory, and compensatory functions within complex defense networks. The study of these genes challenges rigid classification paradigms and reveals the remarkable plasticity of plant immune systems.

Future research directions should prioritize structural characterization of atypical NBS proteins to elucidate how domain loss affects function, comprehensive interactome mapping to define their positions within immune networks, and translational applications in crop improvement. As demonstrated by the critical roles of TN2 in Arabidopsis immunity and NRCX in Solanaceae immune homeostasis, these atypical forms offer promising targets for engineering durable disease resistance without compromising plant growth and productivity. Their extensive diversity across plant lineages suggests we have only begun to appreciate the full functional repertoire of these non-canonical resistance genes.

Nucleotide-binding site leucine-rich repeat (NBS-LRR) genes constitute the largest family of plant disease resistance (R) genes and play a critical role in effector-triggered immunity (ETI) by encoding intracellular receptors that detect pathogen effectors [29] [4]. The domain architecture of these genes typically features a conserved NBS domain (NB-ARC, PF00931) alongside variable N-terminal (TIR, CC, or RPW8) and C-terminal (LRR) domains, enabling their classification into TNL, CNL, and RNL subclasses [30] [4]. Recent genome-wide comparative analyses across diverse plant taxa have revealed that NBS-LRR genes exhibit remarkable species-specific evolutionary patterns, with dramatic differences in gene family size, composition, and organization [30] [31] [32]. This dynamic evolution, characterized by frequent gene duplication and loss events, represents a genomic arms race between plants and their rapidly evolving pathogens [31] [33]. Understanding these species-specific evolutionary trajectories provides crucial insights into plant-pathogen coevolution and informs strategies for breeding durable disease resistance in crop species.

Evolutionary Patterns Across Plant Families

Quantifying NBS-LRR Diversity

Table 1: NBS-LRR Gene Distribution Across Plant Families

Plant Family	Species	Total NBS-LRR Genes	CNL	TNL	RNL	Evolutionary Pattern
Rosaceae	Fragaria vesca (strawberry)	144	84.03%	15.97%	-	"Expansion, contraction, further expansion" [30] [31]
Rosaceae	Malus × domestica (apple)	748	70.72%	29.28%	-	"Continuous expansion" [30] [31]
Rosaceae	Pyrus bretschneideri (pear)	469	52.88%	47.12%	-	"Early sharp expansion to abrupt shrinking" [30] [31]
Rosaceae	Prunus persica (peach)	354	63.84%	36.16%	-	"Early sharp expansion to abrupt shrinking" [30] [31]
Solanaceae	Nicotiana benthamiana	156	25 CNL-type	5 TNL-type	4 with RPW8	Not specified [4]
Euphorbiaceae	Vernicia montana (tung tree)	149	65.8%	8.1%	-	Resistance-specific expansion [22]
Euphorbiaceae	Vernicia fordii (tung tree)	90	54.4%	0%	-	Susceptibility-associated contraction [22]
Dioscoreaceae	Dioscorea rotundata (yam)	167	99.4%	0%	0.6%	Monocot-specific TNL absence [33]
Passifloraceae	Passiflora edulis (purple passion fruit)	25 CNL	100%	0%	0%	Family-specific CNL specialization [34]

The expansion and contraction of NBS-LRR genes display remarkable variation between and within plant families. In the Rosaceae, Malus × domestica (apple) possesses 748 NBS-LRR genes, while Fragaria vesca (strawberry) contains only 144 genes, representing a five-fold difference despite their phylogenetic relatedness [31]. This disparity is primarily driven by species-specific duplication events, with 61.81% of strawberry NBS-LRRs and 66.04% of apple NBS-LRRs derived from recent species-specific duplications [31]. Similarly, in the Euphorbiaceae, the resistant Vernicia montana contains 149 NBS-LRRs compared to only 90 in the susceptible Vernicia fordii, highlighting how differential evolutionary histories can directly impact disease resistance [22].

The evolution of NBS-LRR subclasses also demonstrates distinct trajectories. TNL genes generally evolve more rapidly than non-TNLs, as evidenced by significantly higher Ks and Ka/Ks values [31]. Furthermore, certain plant lineages have experienced complete loss of specific subclasses; monocots including Dioscorea rotundata and Oryza sativa lack TNL genes entirely, while some eudicots like Vernicia fordii and Sesamum indicum have also independently lost this subclass [22] [33].

Genomic Drivers of NBS-LRR Evolution

Table 2: Genomic Mechanisms Driving NBS-LRR Evolution

Mechanism	Impact on NBS-LRR Genes	Examples
Tandem duplication	Rapid expansion of clustered genes; creates sequence diversity	63% of cassava NBS-LRRs occur in 39 clusters [29]; Major mechanism in Dioscorea [33]
Segmental/WGD duplication	Large-scale expansion; preserves gene families	Whole genome triplication in Solanaceae [35]; 17 segmental duplication pairs in passion fruit [34]
Purifying selection	Maintains functional protein domains; Ka/Ks < 1	Most NBS-LRRs in five Rosaceae species [31]; Passion fruit CNLs [34]
Birth-and-death evolution	Continuous turnover of genes via duplication/diversification/loss	Solanaceae family evolution [35]
Positive selection	Drives adaptation to specific pathogens; Ka/Ks > 1	Specific solvent-exposed residues in LRR domains [30]

Whole genome duplication (WGD) events have played a particularly significant role in expanding NBS-LRR repertoires. The recent whole genome triplication in Solanaceae species contributed substantially to their NBS-LRR complement, with 819 genes identified across nine species [35]. Similarly, the high NBS-LRR numbers in apple (748) and pear (469) reflect their paleopolyploid origins [31]. Following duplication, NBS-LRR genes predominantly evolve under purifying selection (Ka/Ks < 1), which maintains functional protein domains while allowing for diversification in pathogen recognition specificities [31].

The genomic organization of NBS-LRR genes into clusters facilitates their rapid evolution through mechanisms such as unequal crossing-over and gene conversion. Approximately 63% of cassava NBS-LRR genes reside in 39 clusters across the genome [29]. These clusters are typically homogeneous, containing genes derived from recent common ancestors, which promotes the generation of novel recognition specificities through recombination between paralogs [29].

Methodological Framework for NBS-LRR Analysis

Standardized Gene Identification Pipeline

(NBS-LRR Identification Workflow)

The accurate identification of NBS-LRR genes requires a comprehensive bioinformatics approach combining multiple complementary methods. The standard workflow begins with HMMER searches using the hidden Markov model for the NB-ARC domain (PF00931) as query against target proteomes, typically with an E-value cutoff of 1.0 or more stringent thresholds (E-value < 1×10⁻²⁰) to ensure specificity [30] [4]. Parallel BLAST searches using known NBS-LRR sequences as queries provide additional candidates and help recover divergent family members [29].

Candidate genes subsequently undergo domain architecture validation using Pfam, CDD, and SMART databases to confirm the presence of characteristic N-terminal (TIR, CC, RPW8) and C-terminal (LRR) domains [22] [4]. For CC domains, which are poorly detected by conventional Pfam searches, tools like Paircoil2 with a P-score cutoff of 0.03 are essential for accurate prediction [29]. The final classification into TNL, CNL, and RNL subclasses requires manual curation to account for non-canonical domain arrangements and partial genes [30].

For complex polyploid genomes, specialized pipelines like DaapNLRSeek (Diploidy-assisted Annotation of Polyploid NLRs) have been developed to overcome challenges posed by genome duplication and high sequence similarity between homeologs [36]. This approach has proven effective for accurate NLR annotation in sugarcane and other polyploid crops.

Evolutionary and Phylogenetic Analysis

(Evolutionary Analysis Methodology)

Evolutionary analyses begin with multiple sequence alignment of the conserved NBS domain using tools like ClustalW or MAFFT, followed by manual curation with Jalview to trim poorly aligned regions [29] [4]. Phylogenetic reconstruction via Maximum Likelihood (e.g., Whelan and Goldman model) or Neighbor-Joining methods with bootstrap testing (1000 replicates) reveals evolutionary relationships and classifies sequences into major clades [29] [4].

For multi-species comparisons, OrthoFinder implements a robust orthogroup inference pipeline, using DIAMOND for fast sequence similarity searches and the MCL algorithm for clustering [32]. This approach identifies core orthogroups conserved across species and lineage-specific expansions. Duplication type analysis distinguishes tandem from segmental duplications by examining genomic coordinates and syntenic relationships [34].

The Ka/Ks ratio (non-synonymous to synonymous substitution rate) serves as a key metric for detecting selection pressures. Ka/Ks < 1 indicates purifying selection, Ka/Ks = 1 suggests neutral evolution, and Ka/Ks > 1 signifies positive selection [31]. Most NBS-LRR genes evolve under purifying selection, though specific solvent-exposed residues in LRR domains may experience positive selection associated with pathogen recognition specificity [30].

Table 3: Key Research Reagents and Computational Tools for NBS-LRR Studies

Category	Tool/Resource	Specific Function	Application Example
Domain Databases	Pfam (PF00931)	NBS (NB-ARC) domain identification	Core domain detection [30] [29]
	CDD/InterPro	Multi-domain architecture analysis	Supplementary domain verification [22] [34]
Search Algorithms	HMMER	Hidden Markov model-based searches	Initial genome-wide identification [30] [29]
	BLAST	Sequence similarity searches	Recovery of divergent homologs [29] [34]
Motif Analysis	MEME Suite	Conserved motif discovery	Identifying NBS subdomain structure [30] [4]
	WebLogo	Sequence logo generation	Visualizing conserved residues [30]
Phylogenetic Tools	OrthoFinder	Orthogroup inference	Multi-species comparative analysis [32]
	MEGA	Phylogenetic tree construction	Evolutionary relationship inference [29] [4]
Expression Validation	VIGS	Functional gene silencing	In planta validation of resistance function [22]
	RNA-seq	Transcriptome profiling	Expression analysis under stress [34] [33]

The evolutionary dynamics of NBS-LRR genes across plant genomes demonstrate a complex interplay of species-specific duplication events, selective pressures, and genomic mechanisms that collectively shape the plant immune repertoire. The striking variation in gene family size and composition between even closely related species highlights the adaptive nature of this gene family in response to pathogen pressures. Future research leveraging the methodologies and resources outlined in this review will continue to unravel the molecular basis of plant-pathogen coevolution and facilitate the development of crop varieties with enhanced disease resistance through molecular breeding approaches. The integration of comparative genomics, functional validation, and computational prediction represents a powerful framework for elucidating the principles governing NBS-LRR evolution and their application to agricultural improvement.

From Sequence to Function: Methodologies for Identifying and Classifying NBS Genes

Nucleotide-binding site (NBS) domain genes constitute the largest family of plant disease resistance (R) genes, playing crucial roles in innate immunity against diverse pathogens. This technical guide provides a comprehensive framework for identifying and characterizing NBS domains using HMMER and Pfam, contextualized within domain architecture and classification research. We present detailed experimental protocols, data analysis workflows, and visualization tools to enable researchers to systematically discover and annotate NBS genes across plant genomes. The integration of these bioinformatics approaches has revolutionized plant resistance gene studies, facilitating the development of disease-resistant cultivars through genome-wide identification of NBS-encoding genes.

NBS Domains in Plant Immunity

NBS domains form the core component of plant resistance proteins that function in effector-triggered immunity (ETI), providing protection against viruses, bacteria, fungi, nematodes, and insects [37]. These domains are characterized by conserved nucleotide-binding motifs that bind and hydrolyze ATP/GTP, serving as molecular switches in disease resistance signaling pathways [10]. The NBS domain is typically embedded within larger protein architectures, most commonly as part of NBS-LRR (leucine-rich repeat) proteins, which represent over 60% of cloned plant R genes [21]. The significance of NBS domains extends beyond individual pathogen recognition events, as their genomic distribution and evolution directly impact plant resilience to rapidly evolving pathogens.

Classification and Diversity of NBS Genes

NBS-encoding genes are classified into distinct subfamilies based on their N-terminal domains: TNL (TIR-NBS-LRR), CNL (CC-NBS-LRR), and RNL (RPW8-NBS-LRR) [10] [21]. The distribution of these subfamilies varies significantly across plant species, reflecting evolutionary adaptations to specific pathogen pressures. For instance, Akebia trifoliata possesses 50 CNL, 19 TNL, and 4 RNL genes [10], while cassava contains 228 NBS-LRR genes with 34 TNL and 128 CNL types [38]. This diversity underscores the importance of comprehensive domain-centric approaches for cataloging resistance genes across species with different evolutionary histories.

Materials and Methods: The Bioinformatics Toolkit

Table 1: Key Research Reagents and Databases for NBS Domain Discovery

Resource	Type	Function	Source/Access
Pfam Database	Protein Family Database	Provides HMM profiles for domain identification	https://pfam.xfam.org/ [39]
HMMER Suite	Software Toolkit	Sequence database searching using HMMs	http://hmmer.org/
NB-ARC Domain (PF00931)	HMM Profile	Primary query for NBS domain identification	Pfam Accession: PF00931 [38]
TIR Domain (PF01582)	HMM Profile	Identifies TIR-NBS-LRR subfamily	Pfam Accession: PF01582 [21]
RPW8 Domain (PF05659)	HMM Profile	Identifies RNL subfamily	Pfam Accession: PF05659 [21]
LRR Domain (PF08191)	HMM Profile	Identifies leucine-rich repeats	Pfam Accession: PF08191 [21]
NCBI CDD	Domain Database	Verifies conserved domain presence	https://www.ncbi.nlm.nih.gov/cdd/ [10]
Coiled-coil Prediction	Algorithm	Identifies coiled-coil domains not detected by Pfam	e.g., Paircoil2 [38]

Experimental Design and Workflow

The fundamental workflow for NBS domain discovery integrates multiple bioinformatics tools in a sequential pipeline to ensure comprehensive identification and accurate classification. The process begins with genome-wide scanning using HMMER with the NB-ARC domain profile, followed by domain architecture analysis, phylogenetic classification, and structural validation. This systematic approach enables researchers to overcome challenges associated with gene family diversity and evolutionary divergence.

Figure 1: Comprehensive workflow for NBS domain discovery and characterization

Technical Protocols for NBS Domain Identification

Primary Identification Using HMMER and Pfam

The core identification process employs HMMER tools with the NB-ARC domain profile from Pfam. Implementation requires careful parameter optimization to balance sensitivity and specificity:

Researchers should note that Pfam is now hosted by InterPro, and while the database remains accessible, all updates and current data are available through InterPro [39]. The E-value threshold of 1.0 provides an initial broad search, which should be refined in subsequent verification steps.

Classification and Domain Architecture Analysis

Following initial identification, comprehensive classification delineates NBS genes into subfamilies based on associated domains:

Classification should follow established standards: TNL (TIR-NBS-LRR), CNL (CC-NBS-LRR), and RNL (RPW8-NBS-LRR) [21]. Additional validation using NCBI Conserved Domain Database improves accuracy, particularly for divergent sequences.

Motif Identification and Structural Validation

Conserved motif analysis within NBS domains reveals evolutionary relationships and functional constraints:

Studies consistently identify eight conserved motifs within plant NBS domains, with variations distinguishing TNL and CNL subfamilies [10] [37]. The conserved order and amino acid sequences of these motifs facilitate functional predictions and evolutionary analyses.

Data Analysis and Interpretation

Quantitative Profiling of NBS Genes

Table 2: Comparative Analysis of NBS Genes Across Plant Species

Plant Species	Total NBS Genes	CNL	TNL	RNL	Clustered	Singleton	Reference
Akebia trifoliata	73	50	19	4	41	23	[10]
Manihot esculenta (Cassava)	228	128	34	-	63%	37%	[38]
Arabidopsis thaliana	~150	-	-	-	-	-	[37]
Oryza sativa (Rice)	>400	-	-	-	-	-	[37]
Gossypium hirsutum (Cotton)	12,820 (across 34 species)	-	-	-	-	-	[32]

Genome-wide analyses reveal substantial variation in NBS gene numbers, ranging from dozens in some species to over 2,000 in others [21]. This variation reflects species-specific evolutionary trajectories rather than direct correlations with genome size. Most NBS genes display non-random chromosomal distributions, preferentially clustering at chromosome ends where recombination rates are higher, facilitating rapid evolution of recognition specificities [10] [38].

Evolutionary Analysis and Gene Duplication

Evolutionary analyses indicate that tandem and dispersed duplications represent primary mechanisms for NBS gene expansion. In Akebia trifoliata, these mechanisms generated 33 and 29 genes respectively [10]. Phylogenetic relationships typically separate TNL and CNL proteins into distinct clades with different evolutionary patterns, informing functional predictions and comparative genomic studies.

Figure 2: Evolutionary analysis workflow for NBS genes

Advanced Applications and Validation

Expression Analysis and Functional Validation

Transcriptomic analyses reveal that NBS genes typically exhibit low baseline expression with specific induction during pathogen challenge. In Akebia trifoliata, most NBS genes showed low expression across fruit development stages, with a subset displaying relatively high expression during later development in rind tissues [10]. Similar patterns emerge in comparative studies of cotton NBS genes, where orthogroups OG2, OG6, and OG15 showed upregulated expression under biotic stress in tolerant genotypes [32].

Functional validation through virus-induced gene silencing (VIGS) demonstrates the critical role of specific NBS genes in disease resistance. Silencing of GaNBS (OG2) in resistant cotton increased susceptibility to cotton leaf curl disease, confirming its functional importance in antiviral defense [32]. These validation approaches bridge bioinformatics predictions with biological relevance, prioritizing candidates for breeding applications.

Structural Prediction and Molecular Modeling

Advanced structural modeling using tools like AlphaFold 3 incorporates HMMER-based template searches to generate accurate protein structures [40]. The NBS domain functions as a molecular switch, with conformational changes between ADP-bound (inactive) and ATP-bound (active) states regulating downstream signaling. Integration of structural predictions with evolutionary analyses illuminates structure-function relationships across NBS subfamilies.

Domain-centric bioinformatics using HMMER and Pfam provides a powerful framework for systematic discovery and characterization of NBS genes in plant genomes. The standardized protocols outlined in this guide enable comprehensive identification, classification, and evolutionary analysis of this crucial gene family. Future developments will likely include improved integration with structural prediction tools, expanded databases covering more plant lineages, and machine learning approaches for predicting recognition specificities. As genomic resources continue expanding, these methodologies will play increasingly vital roles in mining the genetic basis of disease resistance and accelerating the development of durable resistant crop varieties.

The identification of plant disease resistance (R) genes, particularly those with Nucleotide-Binding Site Leucine-Rich Repeat (NBS-LRR) domains, has long been a cornerstone of plant pathology and breeding research. Traditional methods for R-gene identification, which rely on sequence alignment and domain-based search tools like HMMER and InterProScan, face significant challenges when dealing with genes exhibiting low sequence homology or novel architectural patterns [6] [41]. The limitations of these conventional approaches have become increasingly apparent as researchers explore diverse plant species with complex genomes, where R-genes often reside in repetitive regions and exhibit low expression levels, making them difficult to annotate accurately [6].

The integration of machine learning (ML) and deep learning (DL) represents a paradigm shift in bioinformatics, enabling the prediction of R-genes based on complex sequence patterns rather than mere homology [41]. Among these advanced tools, PRGminer emerges as a specialized deep learning framework specifically designed for high-throughput prediction and classification of plant resistance genes. By harnessing sophisticated neural networks, PRGminer addresses critical gaps in traditional R-gene identification methods, offering researchers an powerful tool for elucidating the genetic basis of plant immunity [6]. This technical guide explores the architecture, implementation, and application of PRGminer within the broader context of NBS disease resistance gene research, providing researchers with comprehensive protocols for leveraging this cutting-edge computational tool.

PRGminer: Architectural Framework and Performance Metrics

Core Algorithmic Design and Implementation

PRGminer employs a structured two-phase deep learning framework specifically optimized for plant R-gene prediction. The tool's architecture reflects a sophisticated approach to handling the complexity and diversity of resistance gene sequences:

• Phase I - Binary Classification: The initial phase performs a critical filtering function, classifying input protein sequences as either R-genes or non-R-genes. This stage utilizes dipeptide composition as the primary feature representation, which has demonstrated superior performance compared to other sequence encoding methods. The model achieves this classification through a deep neural network architecture capable of capturing complex hierarchical patterns in protein sequences that transcend simple domain presence or absence [6].

• Phase II - Multi-Class Classification: Sequences identified as R-genes in Phase I proceed to this secondary classification stage, where they are categorized into one of eight distinct R-gene classes based on their domain architecture and functional characteristics. This phase employs a more specialized neural network trained to recognize subtle patterns indicative of specific R-gene subtypes [6] [42].

The implementation of PRGminer leverages multiple layers of neural networks to extract progressively higher-level features from raw encoded protein sequences. This approach enables the model to learn complex representations directly from the data rather than relying on manually engineered features, allowing it to identify novel R-gene candidates that might be missed by traditional alignment-based methods [6].

Performance Validation and Benchmarking

Rigorous validation experiments demonstrate PRGminer's robust performance across both phases of prediction:

Table 1: Performance Metrics of PRGminer's Prediction Phases

Prediction Phase	Validation Method	Accuracy	Matthews Correlation Coefficient (MCC)
Phase I (R-gene vs Non-R-gene)	k-fold cross-validation	98.75%	0.98
Phase I (R-gene vs Non-R-gene)	Independent testing	95.72%	0.91
Phase II (R-gene Classification)	k-fold cross-validation	97.55%	0.93
Phase II (R-gene Classification)	Independent testing	97.21%	0.92

The high Matthews Correlation Coefficient values, particularly the 0.91 MCC on independent testing in Phase I, indicate strong predictive performance that significantly reduces false positives and negatives - a common challenge in R-gene prediction [6]. This performance surpasses traditional machine learning approaches such as Support Vector Machines (SVM) and alignment-based methods, especially for sequences with low homology to known R-genes [6] [41].

Computational Methodology: Implementing PRGminer for R-gene Discovery

Data Acquisition and Preprocessing Protocols

The foundation of PRGminer's predictive capability lies in its comprehensive training dataset compiled from multiple public databases:

• Data Sources: Protein sequences were obtained from Phytozome, Ensemble Plants, and NCBI to ensure broad taxonomic coverage and sequence diversity [6]. This multi-source approach helps mitigate database-specific biases and enhances model generalizability.

• Sequence Representation: The dipeptide composition representation, which yielded optimal performance, calculates the frequency of all possible pairs of amino acids along the protein sequence. This representation captures local sequence order information while being insensitive to sequence length variations [6].

• Dataset Partitioning: The implementation follows standard machine learning protocols with separate training, validation, and independent test sets to prevent overfitting and provide unbiased performance estimation [6].

Operational Workflow: From Sequence to Prediction

The end-to-end workflow for utilizing PRGminer encompasses both the computational prediction phases and subsequent biological validation:

Diagram 1: PRGminer two-phase prediction and validation workflow. The process begins with sequence input, proceeds through binary classification, then multi-class categorization of R-genes, culminating in experimental validation.

For researchers implementing PRGminer, multiple input modalities are supported:

• Accession ID Input: Users can submit valid protein accession identifiers from NCBI or UniProt databases, enabling rapid analysis of known sequences [43].
• FASTA File Upload: Batch processing of multiple sequences is supported through FASTA file upload, facilitating genome-scale analyses [43].
• Direct Sequence Pasting: For individual sequences or small batches, users can directly paste FASTA-formatted sequences into the web interface [43].

The system processes sequences typically within approximately two minutes, though processing time may scale with the number and length of submitted sequences [42]. For large-scale analyses exceeding 10,000 sequences, local installation is recommended to optimize processing efficiency and enable pipeline integration [43].

Classification Schema: R-gene Categories and Domain Architectures

PRGminer classifies R-genes into eight distinct categories based on domain composition and functional characteristics:

Table 2: R-gene Classes Predicted by PRGminer Phase II Classification

R-gene Class	Domain Architecture	Functional Role in Plant Immunity
CNL	Coiled-Coil, NBS, LRR	Intracellular receptor; effector-triggered immunity [42]
TNL	TIR, NBS, LRR	Intracellular receptor; effector-triggered immunity [42]
RNL	RPW8, NBS, LRR	Signal transduction component in immunity [32]
RLP	LRR, Transmembrane domain	Membrane-bound pathogen recognition; lacks kinase domain [42]
RLK	LRR, Kinase domain	Membrane-bound receptor with kinase signaling activity [42]
LYK	LysM, Kinase, TM domain	Recognition of chitin and peptidoglycan fragments [42]
LECRK	Lectin, Kinase, TM domain	Carbohydrate recognition and signaling [42]
TIR	TIR domain only	Signaling component in immunity pathways [42]

This comprehensive classification system enables researchers to move beyond simple R-gene identification to functional inference based on class-specific characteristics. The domain architectures corresponding to these classes represent the structural foundation of plant immune perception systems, with CNL and TNL proteins comprising the majority of intracellular immune receptors, while RLK and RLP proteins function as membrane-bound pattern recognition receptors [6] [41].

Integration with NBS-LRR Research: Evolutionary and Structural Context

Genomic Distribution and Evolutionary Dynamics

The NBS-LRR gene family represents the largest and most diverse class of plant resistance genes, with significant variation in copy number across plant species:

• Taxonomic Distribution: Comprehensive surveys have identified 12,820 NBS-domain-containing genes across 34 plant species, ranging from mosses to monocots and dicots, revealing both conserved and lineage-specific evolutionary patterns [32].
• Architectural Diversity: These genes display remarkable structural variation, with 168 distinct domain architecture patterns identified, including classical configurations (NBS, NBS-LRR, TIR-NBS-LRR) and species-specific structural patterns (TIR-NBS-TIR-Cupin_1, TIR-NBS-Prenyltransf) [32].
• Genomic Organization: NBS-LRR genes are frequently organized in clusters of closely duplicated genes, though solitary genes also occur scattered throughout plant genomes. This arrangement contributes to the evolutionary plasticity of plant immune systems through mechanisms such as tandem duplication and whole-genome duplication events [6] [16].

Recent research has revealed intriguing patterns of NBS-LRR subfamily distribution across plant taxa. For instance, gymnosperms like Pinus taeda exhibit significant expansion of the TNL subfamily (comprising 89.3% of typical NBS-LRRs), while monocotyledonous species such as Oryza sativa have completely lost TNL and RNL subfamilies [7]. In medicinal plants like Salvia miltiorrhiza, researchers identified 196 NBS domain-containing genes, with a notable reduction in TNL and RNL subfamily members compared to other angiosperms [7].

Structural-Functional Relationships in NBS-LRR Proteins

The modular architecture of NBS-LRR proteins enables their dual functions in pathogen recognition and defense activation:

• N-terminal Domains: The CC, TIR, or RPW8 domains at the N-terminus mediate protein-protein interactions and initiate signaling cascades [32] [7].
• NBS Domain: The central nucleotide-binding site domain binds and hydrolyzes ATP, functioning as a molecular switch for immune activation [7].
• LRR Domain: The C-terminal leucine-rich repeat domain provides structural framework for specific pathogen effector recognition, with hypervariable regions determining recognition specificity [7] [16].

The functional significance of different NBS-LRR architectural types was demonstrated in a study of Nicotiana species, which found that among 1,226 identified NBS genes, approximately 45.5% contained only the NBS domain, while 23.3% belonged to the CNL class, and TNL members represented only 2.5% of the family [16]. This distribution reflects both evolutionary constraints and functional specialization within plant immune systems.

Experimental Validation and Functional Characterization

Transcriptional Profiling and Genetic Variation Analyses

Computational predictions from tools like PRGminer require experimental validation to confirm biological relevance:

• Expression Profiling: Studies of NBS gene expression patterns under various biotic and abiotic stresses have identified specific orthogroups (OG2, OG6, OG15) that show putative upregulation in tolerant versus susceptible cotton accessions facing cotton leaf curl disease [32].
• Genetic Variation Mapping: Comparative analysis between susceptible (Coker 312) and tolerant (Mac7) Gossypium hirsutum accessions revealed significant variation in NBS genes, with Mac7 exhibiting 6,583 unique variants compared to 5,173 in Coker 312, highlighting the potential genetic basis of resistance differences [32].
• Protein Interaction Studies: Protein-ligand and protein-protein interaction assays have demonstrated strong binding between putative NBS proteins and ADP/ATP, as well as core proteins of the cotton leaf curl disease virus, providing mechanistic insights into resistance protein function [32].

Functional Validation Through Gene Silencing

Virus-Induced Gene Silencing (VIGS) represents a powerful approach for validating computational predictions:

• Protocol Implementation: Silencing of GaNBS (OG2) in resistant cotton through VIGS demonstrated its putative role in virus tittering, confirming functional importance predicted through computational means [32].
• Phenotypic Correlation: The correlation between gene silencing and altered disease response provides critical evidence for establishing causal relationships between predicted R-genes and pathogen resistance [32].

These validation methodologies create an essential feedback loop for refining computational prediction tools, enabling iterative improvement of model accuracy and biological relevance.

Table 3: Key Research Reagents and Computational Resources for R-gene Studies

Resource/Reagent	Type	Function/Application	Example Sources
PRGminer	Deep Learning Tool	R-gene prediction and classification	https://kaabil.net/prgminer/ [43]
HMMER3	Bioinformatics Tool	Domain-based gene identification	http://hmmer.org/ [6]
InterProScan	Protein Domain Analyzer	Domain architecture characterization	https://www.ebi.ac.uk/interpro/ [41]
Phytozome	Genomic Database	Source of plant protein sequences	https://phytozome-next.jgi.doe.gov/ [6]
VIGS Constructs	Molecular Biology Reagent	Functional validation through gene silencing	Custom-designed [32]
PATRIC Database	Pathogen Database	Antimicrobial resistance gene references	http://www.patricbrc.org [44]
RNA-seq Datasets	Transcriptomic Data	Expression profiling under stress conditions	NCBI SRA, IPF Database [32] [16]

Comparative Analysis: PRGminer in the Computational Ecosystem

PRGminer operates within a broader ecosystem of computational tools for R-gene identification, each with distinct strengths and applications:

• Alignment-Based Tools: Methods such as those implemented in DRAGO2/3 and RGAugury rely on sequence similarity and domain searches using tools like BLAST, HMMER, and InterProScan. While these approaches remain valuable for detecting genes with clear homology to known R-genes, they often miss novel or highly divergent sequences [6] [41].

• Traditional Machine Learning Methods: Tools utilizing Support Vector Machines (SVM) and random forests extract various numerical features from protein sequences for classification. These methods represent an intermediate approach between alignment-based and deep learning methods [6] [45].

• Deep Learning Frameworks: PRGminer exemplifies this category, employing multiple neural network layers to automatically learn relevant features directly from sequence data, enabling identification of complex patterns that may not be captured by manual feature engineering [6].

The performance advantage of deep learning approaches is particularly evident when handling sequences with low homology, fragmented domains, or novel architectural arrangements that defy conventional domain-based classification methods [6].

Future Directions and Implementation Considerations

As computational approaches to R-gene discovery evolve, several emerging trends and considerations shape their development:

• Explainability and Interpretability: While deep learning models often function as "black boxes," ongoing research focuses on enhancing model interpretability through techniques like SHapley Additive exPlanations (SHAP), which help elucidate the contribution of specific sequence features to classification outcomes [44].

• Integration with Multi-Omics Data: Future iterations of R-gene prediction tools will likely incorporate transcriptomic, epigenomic, and pan-genomic data to provide more comprehensive functional predictions [41].

• Scalability and Computational Efficiency: As plant genome sequencing proliferates, tools must efficiently handle increasingly large datasets. PRGminer's standalone installation option addresses this need for large-scale analyses [43].

The continued refinement of deep learning tools like PRGminer promises to accelerate the discovery of novel resistance genes, enhance our understanding of plant immunity mechanisms, and ultimately contribute to the development of disease-resistant crop varieties through molecular breeding and genetic engineering strategies.

Diagram 2: Evolution of computational methods for R-gene prediction, from traditional alignment-based approaches to modern deep learning and future integrative frameworks.

This technical guide provides a comprehensive workflow for conducting genome-wide analysis of nucleotide-binding site leucine-rich repeat (NBS-LRR) genes, from initial identification using profile hidden Markov models (HMMs) through phylogenetic reconstruction. Within the broader context of domain architecture and classification of plant disease resistance genes, we present detailed methodologies, computational tools, and best practices for researchers investigating the evolution and function of this critical gene family. The pipeline integrates multiple bioinformatics approaches including sequence alignment, profile HMM searching, domain characterization, and evolutionary analysis to enable systematic classification of NBS-LRR genes across plant genomes.

NBS-LRR genes represent one of the largest and most important disease resistance gene families in plants, encoding proteins characterized by nucleotide-binding site (NBS) and leucine-rich repeat (LRR) domains [1]. These genes play a critical role in plant immune responses by recognizing diverse pathogens including bacteria, viruses, fungi, nematodes, and oomycetes [1]. The typical NBS-LRR protein architecture consists of three major domains: an variable N-terminal domain (either TIR or CC), a conserved NBS domain, and C-terminal LRR repeats [22] [1].

The classification of NBS-LRR genes is primarily based on their domain architecture, with two major subfamilies recognized: TIR-NBS-LRR (TNL) proteins containing Toll/interleukin-1 receptor domains and CC-NBS-LRR (CNL) proteins containing coiled-coil motifs [4] [1]. Additional variants include truncated forms that lack one or more domains, classified as TN, CN, NL, or N-types depending on which domains are present [4] [46]. This diversity in protein structure directly influences their functions in pathogen recognition and defense signaling [4].

The complete analysis workflow encompasses four major phases: (1) identification of NBS-LRR candidates using profile HMM searches, (2) multiple sequence alignment and domain validation, (3) classification based on domain architecture, and (4) phylogenetic reconstruction to elucidate evolutionary relationships. This pipeline enables researchers to systematically characterize the resistance gene landscape in plant genomes, providing insights into evolutionary patterns and potential applications in marker-assisted breeding for disease resistance.

Materials and Research Reagent Solutions

Table 1: Essential research reagents and computational tools for NBS-LRR genome-wide analysis

Category	Item	Function/Description	Example Sources
Software Tools	HMMER	Profile HMM analysis for identifying homologous sequences	[47] [48]
	MAFFT	Multiple sequence alignment	[47]
	MEME	Motif discovery and analysis	[4]
	Clustal W	Multiple sequence alignment for phylogenetic analysis	[4]
	MEGA	Molecular Evolutionary Genetics Analysis	[4]
Databases	Pfam	Protein family database containing HMM profiles	[4]
	SMART	Protein domain identification	[4]
	Conserved Domain Database	Domain annotation and classification	[4]
	PlantCARE	cis-acting regulatory element prediction	[4]
HMM Profiles	NB-ARC (PF00931)	Conserved NBS domain for initial identification	[4]
	TIR (PF01582)	TIR domain identification	[22]
	CC domains	Coiled-coil domain identification	[46]
	LRR domains	Leucine-rich repeat identification	[22]

Methodological Workflow

Phase I: Identification of NBS-LRR Genes Using HMMER

The initial identification of NBS-LRR genes begins with profile HMM searches against the target genome or proteome. This approach provides greater sensitivity and accuracy compared to simple sequence similarity tools like BLAST, as profile HMMs describe the probability distribution of residues at each position in a multiple sequence alignment [47].

Step 1: Obtain Conserved Domain HMM Profiles

• Download the NB-ARC (PF00931) HMM profile from the Pfam database (http://pfam.sanger.ac.uk/) [4]
• Additional HMM profiles for TIR, CC, and LRR domains may be included for comprehensive domain annotation

Step 2: Perform HMM Search

• Execute hmmsearch with an E-value threshold (typically < 1e-20) to identify candidate NBS-containing sequences [4]
• Command structure: hmmsearch --tblout output_file -E 1e-20 NB-ARC.hmm target_proteome.fasta
• Extract sequences from significant hits using tools like TBtools [4]

Step 3: Manual Verification and Refinement

• Submit candidate sequences to Pfam, SMART, and CDD for domain verification [4]
• Retain only sequences with complete NBS domains and E-values below 0.01
• Remove duplicate genes and sequences with incomplete domains

Table 2: HMMER search parameters used in recent NBS-LRR genome-wide studies

Parameter	Typical Setting	Rationale	Example Reference
E-value threshold	1e-20 to 1e-10	Balance between sensitivity and specificity	[4]
Domain verification	E-value < 0.01	Ensure domain completeness	[4]
Target database	Annotated proteome	Focus on coding sequences	[22]
Output format	Table (--tblout)	Facilitate downstream processing	[47]

Phase II: Sequence Alignment and Domain Characterization

Following identification, candidate sequences undergo multiple sequence alignment and detailed domain characterization to classify them into NBS-LRR subfamilies.

Step 1: Multiple Sequence Alignment

• Align protein sequences using MAFFT with auto parameter optimization: mafft --auto input_file > aligned_output_file [47]
• Alternatively, use Clustal W with default parameters for phylogenetic analysis [4]

Step 2: Identify Conserved Motifs and Domains

• Analyze conserved motifs using MEME suite with motif count set to 10 and width lengths from 6-50 amino acids [4]
• Validate domain composition using SMART tool and Conserved Domain Database [4]
• Classify sequences into subfamilies (TNL, CNL, NL, TN, CN, N) based on domain presence/absence [4]

Step 3: Assess Additional Features

• Predict subcellular localization using CELLO v.2.5 and Plant-mPLoc [4]
• Calculate physicochemical characteristics (molecular weight, pI) using EXPASY ProtParam [4]
• Analyze gene structure and exon-intron organization using GFF3 annotation files [4]

Phase III: Phylogenetic Analysis

Phylogenetic reconstruction elucidates evolutionary relationships among NBS-LRR genes and reveals patterns of gene family expansion and diversification.

Step 1: Sequence Preparation and Model Selection

• Use full-length protein sequences or conserved domains (NBS) for alignment [4]
• Trim aligned sequences to remove unreliable regions while preserving phylogenetic signals [49]
• Select appropriate evolutionary models based on sequence characteristics (e.g., Whelan and Goldman model for NBS domains) [4]

Step 2: Tree Construction Methods Several phylogenetic inference methods are available, each with distinct advantages and limitations:

Table 3: Comparison of phylogenetic tree construction methods for NBS-LRR analysis

Method	Principle	Advantages	Limitations	Applications in NBS-LRR Studies
Neighbor-Joining (NJ)	Minimal evolution: minimizes total branch length	Fast computation; suitable for large datasets	May reduce sequence information through distance conversion	Initial exploratory analysis of large NBS-LRR families [49]
Maximum Parsimony (MP)	Minimizes number of evolutionary steps	No explicit model assumptions; straightforward approach	Computationally intensive for large datasets; multiple equally parsimonious trees	Analysis of closely related NBS-LRR sequences with high similarity [49]
Maximum Likelihood (ML)	Maximizes likelihood value given evolutionary model	Statistically rigorous; accounts for evolutionary processes	Computationally intensive; requires careful model selection	Preferred method for distantly related NBS-LRR sequences [49] [4]
Bayesian Inference (BI)	Bayes theorem with MCMC sampling	Provides posterior probabilities; incorporates prior knowledge	Computationally intensive; complex implementation	Detailed analysis of specific NBS-LRR clades [49]

Step 3: Tree Evaluation and Visualization

• Assess branch support using bootstrap analysis with 1000 replicates [4]
• Construct consensus trees to represent relationships supported by multiple methods
• Visualize and annotate trees using MEGA or similar software [4]
• Interpret clades in context of domain architecture and chromosomal distribution

Phase IV: Advanced Analysis and Integration

Gene Duplication and Evolutionary Analysis

• Identify gene duplication events through synteny analysis [46]
• Calculate non-synonymous to synonymous substitution rates (dN/dS) to detect selection pressures [1]
• Analyze lineage-specific expansions and contractions in NBS-LRR subfamilies [22]

cis-Element Analysis and Expression Correlation

• Extract promoter regions (1500 bp upstream of start codon) [4]
• Identify regulatory elements using PlantCARE database [4]
• Correlate gene expression patterns with phylogenetic relationships [22]

Case Studies and Applications

Case Study: NBS-LRR Identification in Nicotiana benthamiana

A recent genome-wide analysis of NBS-LRR genes in Nicotiana benthamiana identified 156 NBS-LRR homologs using HMMER search with the NB-ARC domain (PF00931) [4]. The researchers applied an E-value cutoff of 1e-20 and manually verified domain composition, resulting in classification of 5 TNL-type, 25 CNL-type, 23 NL-type, 2 TN-type, 41 CN-type, and 60 N-type proteins [4]. Phylogenetic analysis using maximum likelihood method based on the Whelan and Goldman model with 1000 bootstrap replicates revealed three major clades, each containing mixtures of different NBS-LRR types, indicating complex evolutionary relationships [4].

Case Study: Comparative Analysis in Vernicia Species

A comparative analysis of NBS-LRR genes between Fusarium wilt-susceptible Vernicia fordii and resistant Vernicia montana identified 90 and 149 NBS-LRR genes, respectively [22]. The study revealed the complete absence of TIR domains in V. fordii, while V. montana contained 12 TIR-containing NBS-LRRs, suggesting domain loss events during evolution [22]. Chromosomal distribution analysis showed significant differences in NBS-LRR organization between the two species, with specific orthologous gene pairs potentially responsible for differential disease resistance [22].

Case Study: Brassica carinata RGA Analysis

A comprehensive analysis of resistance gene analogs (RGAs) in Brassica carinata identified 2570 RGAs, including 550 NBS-LRR genes and 2020 transmembrane leucine-rich repeat genes [46]. The study utilized the RGAugury pipeline for prediction and classification, revealing that 65.2% of RGAs were affected by gene duplication events [46]. Comparative analysis with diploid progenitors B. nigra and B. oleracea showed conservation of genomic features alongside extensive expansion of specific RGA classes, providing insights into polyploid genome evolution [46].

Troubleshooting and Quality Control

Common Issues in HMMER Searches

• Low specificity: Adjust E-value threshold (typically 1e-10 to 1e-50) based on reference sequences and orthogroups [47]
• Incomplete domains: Always verify domain composition using multiple databases (Pfam, SMART, CDD) [4]
• Sequence redundancy: Remove duplicate genes while preserving potentially recent duplicates

Phylogenetic Analysis Considerations

• Alignment quality: Ensure proper trimming of aligned sequences; insufficient trimming introduces noise while excessive trimming removes phylogenetic signal [49]
• Model selection: Choose evolutionary models appropriate for your sequences; model misspecification can lead to incorrect tree topologies [49]
• Branch support: Use bootstrap analysis (≥1000 replicates) to assess confidence in tree nodes [4]

Interpretation of Results

• Evaluate E-values in context: < 10^-3 indicates significant results, but stricter cutoffs (10^-50 to 10^-100) are often used for NBS-LRR identification [47]
• Consider both sequence bit scores and E-values: bit scores are independent of database size while E-values account for multiple testing [47]
• Interpret phylogenetic trees in conjunction with domain architecture and chromosomal location data [22]

The integrated workflow from HMM search to phylogenetic analysis provides a robust framework for genome-wide characterization of NBS-LRR genes. This pipeline enables systematic classification based on domain architecture and evolutionary relationships, facilitating insights into the expansion and diversification of plant disease resistance genes. The methodologies outlined in this guide, drawn from recent applications across multiple plant species, offer researchers comprehensive tools for investigating the genomic landscape of NBS-LRR genes and their role in plant immunity. As genomic resources continue to expand, this pipeline will remain essential for uncovering the molecular basis of disease resistance and informing marker-assisted breeding strategies for crop improvement.

Integrating Genomic and Transcriptomic Data for Expression and Functional Insights

The domain architecture and classification of Nucleotide-Binding Site Leucine-Rich Repeat (NBS-LRR) genes constitute a fundamental area of research in plant immunity. These genes encode the largest family of plant disease resistance (R) proteins and play a critical role in the effector-triggered immunity (ETI) system, which recognizes specific pathogen effectors and activates defense responses [50]. Recent advances in high-throughput sequencing technologies have enabled researchers to employ integrated multi-omics approaches—combining genomic, transcriptomic, epigenomic, and metabolomic data—to gain unprecedented insights into the expression patterns, evolutionary dynamics, and functional mechanisms of NBS resistance genes [51]. This technical guide provides a comprehensive framework for integrating genomic and transcriptomic data to advance the study of NBS disease resistance genes, with specific methodologies and examples relevant to this research domain.

The NBS Gene Family: Classification and Domain Architecture

NBS resistance genes are characterized by a conserved nucleotide-binding site (NBS) domain, also known as NB-ARC (Nucleotide Binding Apaf-1, R proteins, and CED-4), and a C-terminal leucine-rich repeat (LRR) region [10]. Based on their N-terminal domains, NBS-LRR genes are classified into three major subfamilies:

• TNL genes: Contain Toll/Interleukin-1 Receptor (TIR) domains
• CNL genes: Feature Coiled-Coil (CC) domains
• RNL genes: Possess Resistance to Powdery Mildew 8 (RPW8) domains [10] [4]

Monocot plants, including important cereal crops, typically lack TNL-type genes, which is considered a result of evolutionary degeneration [50]. The table below summarizes the distribution of NBS-LRR genes across various plant species:

Table 1: NBS-LRR Gene Distribution Across Plant Species

Plant Species	Total NBS Genes	CNL	TNL	RNL	Other	Reference
Arabidopsis thaliana	210	40	48	18	104	[50]
Dendrobium officinale	74	10	0	-	64	[50]
Helianthus annuus (Sunflower)	352	100	77	13	162	[23]
Akebia trifoliata	73	50	19	4	-	[10]
Nicotiana benthamiana	156	25	5	-	126	[4]

The NBS domain contains several conserved motifs, including P-loop, Kinase-2, RNBS-A, GLPL, and MHDL, which facilitate nucleotide binding and hydrolyze ATP to generate energy for pathogen defense mechanisms [52]. The LRR domain is involved in protein-protein interactions and pathogen recognition [52].

Analytical Frameworks for Multi-Omics Integration

Core Conceptual Workflow

The integration of genomic and transcriptomic data follows a structured workflow that transforms raw sequencing data into biological insights. The diagram below illustrates this conceptual framework:

Experimental Design Considerations

Effective integration of genomic and transcriptomic data requires careful experimental design:

• Sample Selection: Include multiple tissues, developmental stages, and treatment conditions (e.g., pathogen inoculation, salicylic acid treatment) to capture expression dynamics [53] [50]
• Biological Replicates: Essential for statistical power in differential expression analysis
• Time-Series Experiments: Capture dynamic changes in gene expression during pathogen infection [52]
• Multi-Tissue Analysis: Reveal tissue-specific expression patterns of NBS genes [53]

Computational Methodologies

Genome-Wide Identification of NBS Genes

The identification of NBS genes from genomic data involves a multi-step process:

• HMMER Search: Use Hidden Markov Model (HMM) profiling with the NB-ARC domain (PF00931) as query to identify candidate NBS genes [10] [4]
• Domain Validation: Verify the presence of characteristic domains (TIR, CC, RPW8, LRR) using Pfam, SMART, and CDD databases [10]
• Manual Curation: Remove redundant genes and verify the complete presence of NBS domains [4]

Table 2: Key Bioinformatics Tools for NBS Gene Analysis

Analysis Type	Tool	Function	Key Parameters
Domain Identification	HMMER	HMM-based domain search	E-value < 1×10⁻²⁰ [4]
Domain Verification	Pfam/ SMART	Domain confirmation	E-value < 0.01 [4]
Motif Discovery	MEME Suite	Conserved motif identification	Motif width: 6-50 aa [10]
Phylogenetic Analysis	MEGA	Evolutionary relationships	Bootstrap > 1000 replicates [4]
Gene Structure	TBtools	Exon-intron visualization	GFF3 annotation file [10]

Transcriptomic Analysis

Transcriptome sequencing and analysis provide expression insights:

• Library Preparation: Utilize either short-read (Illumina) or long-read (PacBio Iso-Seq) platforms depending on research objectives [54]
• Differential Expression: Identify significantly regulated genes using tools like DESeq2 or edgeR
• Co-expression Analysis: Apply Weighted Gene Co-expression Network Analysis (WGCNA) to identify gene modules associated with specific traits or conditions [50]

Multi-Omics Integration Approaches

• Correlation Analysis: Identify relationships between gene expression patterns and genomic features
• Pathway Enrichment: Map expressed NBS genes to known immune signaling pathways [50]
• Epigenomic Integration: Combine DNA methylation data with expression profiles to identify regulatory mechanisms [51]

Case Studies in NBS Gene Research

Panax japonicus: Integrated Transcriptomic and Metabolomic Analysis

A recent study on Panax japonicus var. major performed integrated transcriptomic and metabolomic analyses across four tissues (roots, stems, fruits, and leaves) [53]:

• Transcriptome Scale: Identified 216,108 unigenes, surpassing previous studies
• NBS Gene Expansion: Discovered 507 plant defense-related NBS genes, the highest number among Panax species
• Metabolite Profiling: Detected 1,161 metabolites and 792 differentially accumulated metabolites
• Biosynthetic Pathways: Identified 59 candidate genes involved in upstream pathways of saponins biosynthesis, plus 311 cytochrome P450 genes and 147 UDP-dependent glycosyltransferase genes

This integrated approach revealed that triterpenoid saponin biosynthesis upstream pathways occur in leaves, while downstream pathways occur in roots, demonstrating the value of multi-tissue analysis [53].

Rice Blast Resistance: Combining GWAS and RNA-Seq

A study on rice blast resistance integrated genome-wide association study (GWAS) and RNA sequencing to identify resistance genes [52]:

• Population Scale: 295 diverse rice genotypes evaluated for blast resistance
• QTL Identification: Mapped 11 Quantitative Trait Loci (QTLs) encompassing 233 genes
• Candidate Gene Validation: Identified OsLB2.2 as a negative regulator of blast resistance through knockout mutants
• Molecular Markers: Developed Kompetitive Allele-Specific PCR (KASP) markers for molecular breeding

This approach demonstrated how integrated genomic and transcriptomic data can identify key regulatory genes and facilitate molecular breeding programs [52].

Dendrobium officinale: Evolutionary Patterns and Signal Transduction

Research on Dendrobium officinale explored the evolutionary patterns of NBS genes and their role in signal transduction pathways [50]:

• Comparative Genomics: Identified 655 NBS genes across seven orchid species and Arabidopsis
• Domain Degeneration: Documented frequent degeneration of NB-ARC domains in Dendrobium species
• Salicylic Acid Response: Identified six NBS-LRR genes significantly upregulated under salicylic acid treatment
• Pathway Integration: Revealed that NBS-LRR genes participate in ETI systems, plant hormone signal transduction, and Ras signaling pathways

Table 3: Essential Research Reagents and Resources for NBS Gene Studies

Category	Specific Resource	Function/Application	Example from Literature
Sequencing Platforms	PacBio Iso-Seq Nanopore Illumina	Long-read transcriptome assembly Epigenetic modifications Short-read RNA sequencing	Full-length transcriptome for impatiens [54] RRBS for DNA methylation [51] Differential expression analysis [52]
Bioinformatics Tools	HMMER MEME Suite TBtools	Domain identification Motif discovery Genomic data visualization	NBS gene identification [4] Conserved motif analysis [10] Gene structure visualization [10]
Experimental Materials	Salicylic Acid Pathogen Strains Mutant Lines	Defense hormone induction Disease resistance phenotyping Functional validation	SA treatment in Dendrobium [50] M. oryzae ZD5 in rice [52] oslb2.2 knockout mutants [52]
Databases	NCBI Databases Pfam Database Phytozome	Sequence retrieval Domain annotation Genome access	Sunflower genome [23] NB-ARC domain (PF00931) [4] Plant genomic resources [23]

Technical Protocols for Key Experiments

Protocol: Genome-Wide Identification of NBS Genes

This protocol follows methodologies successfully applied in multiple studies [23] [10] [4]:

• Data Retrieval

  • Download genome sequence and annotation files from relevant databases (e.g., Phytozome, NCBI)
  • Example: Sunflower genome (H. annuus r1.2) was accessed from https://www.sunflowergenome.org [23]

• HMMER Search

  • Use HMMER software with NB-ARC domain (PF00931) as query
  • Set expectation value (E-value) threshold to < 1×10⁻²⁰ for high-confidence hits [4]
  • Command: hmmsearch --domtblout output.txt NB-ARC.hmm protein_fasta.fa

• Domain Validation

  • Validate candidate sequences using Pfam (http://pfam.sanger.ac.uk/) and SMART databases
  • Confirm presence of complete NBS domain with E-value < 0.01
  • Identify additional domains (TIR, CC, RPW8, LRR) using NCBI CDD

• Classification and Annotation

  • Classify genes into subfamilies (TNL, CNL, RNL) based on domain composition
  • Annote gene structures using GFF3 files
  • Identify conserved motifs using MEME Suite with default parameters [10]

Protocol: Integrated Transcriptomic Analysis of Disease Resistance

This protocol integrates approaches from impatiens downy mildew and rice blast studies [52] [54]:

• Experimental Design

  • Select resistant and susceptible cultivars at appropriate developmental stages
  • Include time-series sampling after pathogen inoculation (e.g., 0, 6, 12, 24, 48 hours post-inoculation)
  • Collect multiple biological replicates (minimum n=3)

• RNA Extraction and Sequencing

  • Extract high-quality RNA using commercial kits with DNase treatment
  • Prepare libraries using standardized protocols (e.g., Illumina TruSeq)
  • Sequence with sufficient depth (>20 million reads per sample)

• Differential Expression Analysis

  • Align reads to reference genome/transcriptome using STAR or HISAT2
  • Quantify gene expression with featureCounts or HTSeq
  • Identify differentially expressed genes using DESeq2 with adjusted p-value < 0.05

• Multi-Omics Integration

  • Correlate NBS gene expression patterns with genomic localization
  • Identify co-expression networks using WGCNA
  • Integrate with epigenomic data (e.g., DNA methylation) when available

Signaling Pathways and Molecular Mechanisms

NBS-LRR proteins function within complex signaling networks in plant immunity. The diagram below illustrates key pathways and interactions:

The molecular mechanisms of NBS-LRR proteins involve:

• Pathogen Recognition: Direct or indirect recognition of pathogen effectors through LRR domains [10]
• Nucleotide-Dependent Activation: Conformational changes from ADP-bound to ATP-bound states [4]
• Downstream Signaling: Activation of defense responses including hypersensitive response (HR), hormone signaling (salicylic acid, jasmonic acid, ethylene), and MAPK cascades [50]
• Transcriptional Reprogramming: Activation of defense-related genes through coordinated transcriptional networks

The integration of genomic and transcriptomic data provides powerful insights into the expression patterns and functional mechanisms of NBS disease resistance genes. This multi-omics approach has revealed:

• Evolutionary Dynamics: NBS gene families expand primarily through tandem and dispersed duplications, with frequent domain degeneration events [50] [10]
• Expression Specificity: NBS genes often show tissue-specific, developmentally regulated, and pathogen-induced expression patterns [53] [10]
• Network Interactions: NBS proteins function within complex immune signaling networks, interacting with hormone pathways and other defense components [50]

Future research directions should leverage emerging technologies such as single-cell transcriptomics to understand cell-type-specific immune responses, pangenomics to capture full NBS gene diversity across populations, and deep learning approaches to predict gene function from sequence and expression data. The continued integration of multi-omics data will accelerate the identification and functional characterization of NBS resistance genes, facilitating the development of disease-resistant crop varieties through molecular breeding and biotechnology.

Plant disease resistance is a complex biological process fundamentally mediated by a sophisticated innate immune system. Within this system, nucleotide-binding site leucine-rich repeat (NBS-LRR) proteins constitute the largest and most prominent class of intracellular immune receptors, playing a pivotal role in the plant's ability to recognize and respond to diverse pathogens [1]. These proteins function as specialized guards that monitor cellular integrity and directly or indirectly perceive pathogen effector molecules, triggering robust defense responses that often include a hypersensitive response and systemic acquired resistance [1]. The NBS-LRR gene family is characterized by remarkable diversity in sequence, structure, and function, with members classified based on their domain architecture into distinct subgroups such as TNL (TIR-NBS-LRR), CNL (CC-NBS-LRR), and RNL (RPW8-NBS-LRR), among others [16] [33].

The identification and characterization of NBS-LRR genes across crop species have accelerated with the advent of advanced genomic technologies, providing crucial insights into plant immunity mechanisms and offering valuable resources for molecular breeding programs. This technical guide presents comprehensive case studies of NBS-LRR gene identification in key crops, with detailed methodologies, data analysis frameworks, and practical tools for researchers investigating plant disease resistance genes. By examining the genomic organization, evolutionary patterns, and functional characterization of these genes, scientists can develop innovative strategies for enhancing crop resistance to devastating pathogens, ultimately contributing to global food security.

Methodological Framework for NBS-LRR Identification

Core Bioinformatics Pipeline

The identification of NBS-LRR genes across plant genomes follows a relatively standardized bioinformatics workflow that leverages conserved domain features and sequence homology. The foundational step involves Hidden Markov Model (HMM) searches using the PF00931 (NB-ARC) model from the Pfam database against the entire proteome of the target species [55] [29] [4]. This initial screening is typically conducted with HMMER software suite with stringent E-value cutoffs (e.g., < 1×10⁻²⁰) to ensure high-confidence matches [29] [4]. The resulting candidate sequences then undergo domain architecture analysis using multiple databases including Pfam, SMART, and NCBI's Conserved Domain Database (CDD) to identify associated domains such as TIR (PF01582), CC (detected by tools like Paircoil2), RPW8 (PF05659), and various LRR domains (PF00560, PF07723, PF07725, PF12799) [29] [16].

Following domain annotation, phylogenetic analysis is performed to classify the identified NBS-LRR genes into distinct clades and subgroups. This process typically involves multiple sequence alignment of the NB-ARC domains using tools such as ClustalW or MUSCLE, followed by tree construction with maximum likelihood methods implemented in MEGA software [55] [4] [16]. Bootstrap analysis with 1000 replicates is commonly employed to assess node support [4]. Additional analyses include motif detection using MEME suite to identify conserved sequence motifs beyond the core domains, gene structure analysis with TBtools to examine exon-intron organization, and cis-element prediction using PlantCARE database to identify potential regulatory elements in promoter regions [55] [4].

Advanced Genomic Analyses

For more comprehensive investigations, researchers often implement additional genomic analyses to understand the evolutionary dynamics and functional implications of NBS-LRR genes. Synteny and duplication analysis using MCScanX helps identify segmental and tandem duplication events that have contributed to the expansion of NBS-LRR gene families [16]. The calculation of non-synonymous (Ka) and synonymous (Ks) substitution rates using KaKs_Calculator 2.0 provides insights into selective pressures acting on different NBS-LRR genes and gene pairs [16]. Chromosomal distribution mapping reveals the clustering patterns of NBS-LRR genes, which is a hallmark of this gene family driven by rapid evolution in response to pathogen pressure [29] [22].

Expression profiling through RNA-seq data analysis offers functional insights by identifying NBS-LRR genes responsive to pathogen challenge. This typically involves quality control of sequencing reads with Trimmomatic, alignment to reference genomes using Hisat2, transcript quantification with Cufflinks, and differential expression analysis with Cuffdiff [16]. For functional validation, virus-induced gene silencing (VIGS) has emerged as a powerful tool to assess the role of candidate NBS-LRR genes in disease resistance, as demonstrated in studies of tung tree and tobacco [16] [22].

The experimental workflow for genome-wide identification and characterization of NBS-LRR genes can be visualized as follows:

Comprehensive Case Studies

Tobacco (Nicotiana benthamiana and Nicotiana tabacum)

Tobacco species serve as model systems for plant-pathogen interaction studies due to their experimental tractability and significance in plant virology research. A genome-wide analysis of Nicotiana benthamiana identified 156 NBS-LRR homologs representing approximately 0.25% of the total annotated genes in the genome [55] [4]. These genes were classified into six distinct architectural types: 5 TNL-type, 25 CNL-type, 23 NL-type, 2 TN-type, 41 CN-type, and 60 N-type proteins, revealing a remarkable diversity in domain composition [4]. Phylogenetic analysis clustered these 133 full-length NBS-domain genes into three major clades, each containing at least four different structural types, indicating substantial sequence and functional divergence [4].

Subcellular localization predictions using CELLO v.2.5 and Plant-mPLoc indicated that the majority of NBS-LRR proteins (121) were localized in the cytoplasm, with 33 in the plasma membrane, and 12 in the nucleus, suggesting distinct surveillance compartments within the cell [55] [4]. Gene structure analysis revealed that most NBS-LRR genes contained few introns, a characteristic feature of this gene family that may facilitate rapid evolution and functional diversification [55]. Regulatory element analysis identified 29 shared cis-element types and 4 elements unique to irregular-type NBS-LRR genes, providing insights into their transcriptional regulation [4].

A broader comparative analysis across three Nicotiana species (N. tabacum, N. sylvestris, and N. tomentosiformis) identified 1,226 NBS genes total, with N. tabacum containing 603 members, approximately the combined total of its parental species [16]. This comprehensive study revealed that 76.62% of N. tabacum NBS genes could be traced back to their parental genomes, with whole-genome duplication significantly contributing to NBS gene family expansion [16]. Expression analysis during disease resistance responses identified numerous NBS genes differentially expressed in resistance to black shank and bacterial wilt, including one potential multi-disease resistance gene [16].

Table 1: NBS-LRR Gene Distribution in Tobacco Species

Species	Total NBS Genes	TNL	CNL	NL	TN	CN	N	Key Findings
Nicotiana benthamiana	156	5	25	23	2	41	60	Diverse subcellular localization; clustered phylogenetically
Nicotiana tabacum	603	Not specified	Not specified	Not specified	Not specified	Not specified	Not specified	76.62% derived from parental genomes; WGD expansion
Nicotiana sylvestris	344	Not specified	Not specified	Not specified	Not specified	Not specified	Not specified	Parental species of N. tabacum
Nicotiana tomentosiformis	279	Not specified	Not specified	Not specified	Not specified	Not specified	Not specified	Parental species of N. tabacum

Cassava (Manihot esculenta)

Cassava represents a critically important food security crop in tropical regions, where its productivity is threatened by viral diseases such as Cassava Mosaic Disease (CMD) and Cassava Brown Streak Disease (CBSD). Genomic analysis of cassava identified 228 NBS-LRR type genes and 99 partial NBS genes, collectively representing nearly 1% of the total predicted genes in the cassava genome [29]. Among these, 34 contained an N-terminal TIR-like domain, while 128 contained an N-terminal coiled-coil domain, indicating a predominance of CNL-type genes in the cassava NBS-LRR repertoire [29].

A particularly notable finding was the clustered genomic organization of cassava NBS-LRR genes, with approximately 63% of the 327 R genes occurring in 39 clusters distributed across the chromosomes [29]. These clusters were predominantly homogeneous, containing NBS-LRRs derived from recent common ancestors, which facilitates the generation of diversity through unequal crossing-over and gene conversion events [29]. This clustered arrangement supports the birth-and-death evolution model for R genes, where duplication events create genetic raw material for functional innovation, followed by selection pressure that maintains beneficial variants while eliminating deleterious ones [1].

Table 2: NBS-LRR Genes in Root and Tuber Crops

Crop Species	Total NBS Genes	TNL	CNL	Other Types	Clustered Genes	Key Findings
Cassava (Manihot esculenta)	327 (228 full + 99 partial)	34	128	165 other/partial	63%	Homogeneous clusters; TIR and CC domains present
White Guinea Yam (Dioscorea rotundata)	167	0	166 (CNL)	1 RNL	74% (124 in clusters)	TNL absence typical of monocots; tandem duplication major driver
Tung Tree (Vernicia fordii)	90	0	49 CC-containing	41 other	Not specified	No TIR domains; LRR domain loss events
Tung Tree (Vernicia montana)	149	12 TIR-containing	98 CC-containing	39 other	Not specified	Resistant species; unique LRR domains

Yam (Dioscorea rotundata) and Tung Tree (Vernicia species)

White Guinea yam (Dioscorea rotundata) represents an important staple crop in tropical regions, where its productivity is constrained by various pathogens. Genomic analysis identified 167 NBS-LRR genes in the D. rotundata genome, accounting for approximately 0.6% of the total annotated genes [33]. Classification based on domain architecture revealed a striking pattern: 166 genes belonged to the CNL subclass, while only one belonged to the RNL subclass, and no TNL genes were detected—a pattern consistent with other monocot species that universally lack TNL genes [33]. The 167 genes were further classified into six groups based on domain combinations: 64 intact CNL genes, 28 NL genes (lacking CC domain), 30 CN genes (lacking LRR domain), 40 N genes (lacking both CC and LRR domains), one RNL gene, and four genes with complicated domain arrangements classified as "others" [33].

The genomic distribution analysis revealed that 124 (74%) of the NBS-LRR genes were arranged in 25 multigene clusters, while 43 genes were singletons [33]. Tandem duplication was identified as the major evolutionary mechanism driving this cluster formation, with segmental duplication detected for 18 NBS-LRR genes despite no documented whole-genome duplication in the species [33]. Expression profiling across four different tissues revealed generally low expression of most NBS-LRR genes, with relatively higher expression in tuber and leaf tissues compared to stem and flower tissues, reflecting their role in defending vulnerable organs against pathogens [33].

A comparative analysis between two tung tree species (Vernicia fordii and Vernicia montana) with contrasting resistance to Fusarium wilt identified 239 NBS-LRR genes across both genomes: 90 in the susceptible V. fordii and 149 in the resistant V. montana [22]. The domain architecture differed significantly between species, with V. montana possessing 12 TIR-containing NBS-LRRs while V. fordii had none [22]. Functional analysis identified an orthologous gene pair (Vf11G0978-Vm019719) with distinct expression patterns and functional differences—the V. montana allele was activated by VmWRKY64 and conferred resistance to Fusarium wilt, while the V. fordii allele contained a promoter deletion that rendered it non-functional [22].

Signaling Pathways and Functional Mechanisms

NBS-LRR proteins function as sophisticated intracellular immune receptors that activate defense signaling pathways upon pathogen perception. The signaling mechanisms differ between the major subfamilies, with TNL proteins typically signaling through ENHANCED DISEASE SUSCEPTIBILITY 1 (EDS1) and CNL proteins signaling through NON-RACE SPECIFIC DISEASE RESISTANCE 1 (NDR1) [1]. Recent research has revealed that some NBS-LRR proteins function in paired configurations, as demonstrated in wheat where atypical NLR pairs confer resistance to powdery mildew and stripe rust [56].

The molecular activation mechanism involves nucleotide-dependent conformational changes. In the resting state, NBS-LRR proteins exist in an autoinhibited ADP-bound conformation. Upon pathogen recognition, often through direct or indirect detection of pathogen effectors, the proteins undergo conformational changes to an ATP-bound state that activates downstream signaling [1]. This signaling typically triggers a hypersensitive response characterized by programmed cell death at the infection site, which restricts pathogen spread and establishes systemic immunity throughout the plant [29].

The NBS-LRR signaling pathway and functional partnerships can be visualized as follows:

Table 3: Essential Research Reagents and Bioinformatics Tools for NBS-LRR Gene Analysis

Category	Tool/Resource	Specific Application	Key Features
Genome Databases	Phytozome, NCBI Genome	Source of genomic sequences and annotations	Curated plant genomes; standardized annotation formats
Domain Identification	HMMER v3, Pfam database	Identification of NBS (PF00931) and associated domains	Hidden Markov Model searches; domain-specific HMM profiles
Motif Analysis	MEME Suite	Discovery of conserved protein motifs	Identifies ungapped sequence motifs; multiple motif models
Phylogenetic Analysis	MEGA11, ClustalW, MUSCLE	Multiple sequence alignment and tree building	Maximum likelihood methods; bootstrap support assessment
Gene Structure Visualization	TBtools	Gene structure schematics and data visualization	Integrative toolkit; user-friendly interface
Cis-Element Analysis	PlantCARE	Identification of regulatory elements in promoters	Database of cis-acting regulatory elements
Expression Analysis	Hisat2, Cufflinks, Trimmomatic	RNA-seq data processing and differential expression	Transcript quantification; quality control of sequencing data
Functional Validation	VIGS (Virus-Induced Gene Silencing)	Functional characterization of candidate genes	Rapid gene silencing; no stable transformation required

The comprehensive identification and characterization of NBS-LRR genes across crop species represents a fundamental step toward understanding the molecular basis of disease resistance in plants. The case studies presented in this technical guide demonstrate both conserved features and species-specific innovations in the NBS-LRR gene family. Common evolutionary patterns include clustered genomic organization, expansion through tandem duplication, and diversifying selection acting particularly on the LRR domains involved in pathogen recognition [29] [1] [33]. However, striking differences are also evident, such as the complete absence of TNL genes in monocot species like yam [33] and the unusual loss of TIR domains in certain eudicot species like Vernicia fordii [22].

Future research directions will likely focus on functional characterization of identified NBS-LRR genes using gene editing technologies, elucidating the specific pathogen recognition spectra of individual receptors, and exploiting natural variation in NBS-LRR genes for crop improvement. The discovery of atypical NLR pairs in wheat that confer broad-spectrum resistance [56] highlights the potential for engineering sophisticated immune receptors that provide durable disease control. Additionally, integrating multi-omics data will enable researchers to understand the regulatory networks that control NBS-LRR gene expression and the signaling pathways that these receptors activate upon pathogen perception.

As genomic technologies continue to advance, particularly with more affordable long-read sequencing and pan-genome analyses, our understanding of NBS-LRR gene diversity and evolution will deepen considerably. This knowledge will accelerate the development of crop varieties with enhanced and durable disease resistance, reducing reliance on chemical pesticides and contributing to more sustainable agricultural systems. The methodologies and case studies presented in this technical guide provide a robust framework for researchers undertaking NBS-LRR gene identification and characterization in crop species of interest.

Navigating Complex Genomes: Overcoming Challenges in NBS-LRR Annotation and Analysis

The study of nucleotide-binding site-leucine-rich repeat (NBS-LRR) genes, which constitute the largest family of plant disease resistance (R) genes, is fundamentally dependent on accurate genome annotation. These genes play a crucial role in effector-triggered immunity (ETI), enabling plants to recognize pathogens and initiate defense responses [32] [57]. However, the comprehensive analysis of NBS-LRR gene families across plant species faces significant technical challenges stemming from fragmented genome assemblies and complex repetitive sequences that complicate gene prediction and annotation.

Research has demonstrated that upwards of 40% of all gene families are inferred to have the wrong number of genes in draft assemblies due to annotation errors [58]. These inaccuracies manifest as both false positives (added genes) and false negatives (missing genes), directly impacting the identification and classification of NBS-LRR genes. The domain architecture analysis central to classifying NBS-LRR genes into subfamilies such as TNL (TIR-NBS-LRR) and CNL (CC-NBS-LRR) is particularly vulnerable to these annotation hurdles [32] [59]. As the field moves toward large-scale comparative genomics—such as the analysis of 12,820 NBS-domain-containing genes across 34 plant species—addressing these technical challenges becomes increasingly critical for valid biological interpretations [32].

Technical Challenges in NBS-LRR Gene Annotation

The Fragmented Gene Problem

Incomplete genome assemblies directly lead to gene fragmentation, where single genes are split across multiple contigs or scaffolds. This fragmentation occurs because assembly algorithms struggle to correctly resolve repetitive regions and complex genomic structures, resulting in "cleaved" genes that are erroneously annotated as separate entities [58].

Major consequences for NBS-LRR research include:

• Overestimation of gene family size: Fragmented genes are annotated as multiple separate genes, inflating the apparent number of NBS-LRR genes in a genome. Comparative analyses of chicken genome assemblies revealed that more than half of all genes had the wrong number of copies in draft genomes, with the majority of additional genes resulting from fragmentation [58].
• Disrupted domain architecture: Fragmentation within NBS-LRR genes can prevent the identification of complete domain structures, making it impossible to correctly classify genes into TNL, CNL, or other subfamilies based on their N-terminal domains [32] [59].
• Compromised phylogenetic analysis: Incomplete gene sequences due to fragmentation introduce errors in evolutionary studies and orthogroup analyses, potentially leading to incorrect conclusions about gene family expansion and contraction [32].

Table 1: Impact of Genome Assembly Quality on Gene Prediction Accuracy

Assembly Type	Coverage	Contig Count	Full-length Genes	Conserved Ortholog Completeness
Fosmid (2X)	2X	281,711	21,250	14.1%
454 (12X)	12X	45,554	36,210	Data not available
Reference (v4.0)	Multiple technologies	Data not available	Data not available	Data not available

Repetitive Sequence Complications

Repetitive DNA sequences constitute a substantial portion of plant genomes, with over two-thirds of the human genome consisting of repetitive elements [60]. These sequences present particular challenges for NBS-LRR gene annotation due to their diversity and abundance.

Key repetitive elements affecting NBS-LRR annotation:

• Tandem repeats: Short sequences repeated in head-to-tail arrangement including microsatellites (2-6 bp units) and minisatellites (10-100 bp units) [61]. These are enriched in centromeric and telomeric regions but also occur in gene-rich regions.
• Transposable elements (TEs): Class I (retrotransposons including LTR, LINE, SINE) and Class II (DNA transposons) elements that comprise approximately 45% of the human genome [60]. These can insert into or near NBS-LRR genes, disrupting their annotation.
• Gene duplication products: NBS-LRR genes themselves often occur in clusters resulting from tandem duplication, creating regions of high sequence similarity that challenge assembly algorithms [32].

The repetitive nature of the LRR domains within NBS-LRR genes themselves presents an additional complication, as their characteristic repeating units can be misidentified as genomic repeats rather than protein-coding sequences [57] [59].

Methodological Framework for Addressing Annotation Challenges

Experimental and Computational Strategies

Integrated Gene Prediction Pipeline

Accurate identification of NBS-encoding genes requires a multi-step approach that combines evidence from various sources:

Step 1: Initial Identification

• Use BLASTN searches against genome assemblies with known NBS-encoding sequences from related species as queries [59]
• Perform HMMER searches (hmmsearch) using the NB-ARC domain (Pfam: PF00931) profile Hidden Markov Model [32] [57]
• Apply stringent e-value thresholds (e.g., 1.1e-50) to reduce false positives [32]

Step 2: Domain Architecture Validation

• Validate candidate genes using NCBI's Conserved Domain Database (CDB) tool [57]
• Identify associated domains (TIR, CC, LRR, etc.) using PfamScan or similar tools [32]
• Classify genes into architectural classes (TNL, CNL, TN, CN, NL, etc.) based on domain composition [59]

Step 3: Structural Annotation Refinement

• Employ AUGUSTUS (version 3.3) or similar ab initio predictors to identify exon-intron structures [57] [62]
• Use TransDecoder (Release v5.5.0) to identify coding regions within transcript sequences [57]
• Incorporate RNA-Seq evidence where available to verify gene models and identify alternative splicing [58]

RNA-Seq Enhanced Annotation

The limitations of draft assemblies can be mitigated through the integration of transcriptomic data:

Experimental Protocol for RNA-Seq Enhanced Annotation:

• Sample Preparation: Collect tissue samples under appropriate conditions (e.g., pathogen-infected and control plants) to capture expressed NBS-LRR genes [63]
• RNA Extraction: Use commercial kits (e.g., RNeasy Plant Kit) with DNase treatment to obtain high-quality RNA [63]
• Library Preparation and Sequencing: Prepare stranded RNA-seq libraries and sequence using Illumina platforms (NovaSeq 6000) to obtain at least 6 GB of data per sample with Q30 > 80% [63]
• Transcript Assembly: Use alignment-free tools like Salmon (version 1.9.0) for transcript quantification [63]
• Gene Model Enhancement: Integrate transcript evidence with genome annotations using tools like AUGUSTUS with RNA-Seq hints to correct fragmented gene models [62]

This approach has been successfully implemented in banana blood disease resistance research, where RNA-seq revealed key defense genes, including NBS-LRR genes, through differential expression analysis [63].

Specialized Approaches for Repetitive Regions

Strategies for resolving repetitive sequence complications:

• Repeat masking and characterization: Use tools like RepeatMasker to identify and characterize repetitive elements before gene prediction [61]
• Long-read sequencing technologies: Employ PacBio or Oxford Nanopore sequencing to generate reads long enough to span repetitive regions and resolve complex loci [64]
• Comparative genomics: Leverage synteny with related species to identify conserved NBS-LRR gene clusters and distinguish genuine genes from recent duplicates or pseudogenes [32] [59]

Table 2: Research Reagent Solutions for NBS-LRR Gene Annotation

Reagent/Resource	Function in Annotation	Application Example
RNeasy Plant Kit (QIAGEN)	High-quality RNA extraction for transcriptome sequencing	RNA extraction from banana roots for blood disease resistance study [63]
NovaSeq 6000 (Illumina)	High-throughput RNA sequencing	Transcriptome analysis of banana blood disease resistance [63]
NB-ARC Domain (PF00931) HMM	Identification of NBS-encoding regions	HMMER search for NBS domain identification in grass pea [57]
AUGUSTUS (v3.3)	Gene structure prediction	Predicting alternative transcripts in grass pea NBS-LRR genes [57]
NCBI Conserved Domain Database	Domain architecture validation	Verification of NBS domains in candidate resistance genes [57]
OrthoFinder (v2.5.1)	Orthogroup analysis across species	Evolutionary study of NBS genes across 34 plant species [32]

Case Studies in NBS-LRR Gene Annotation

Cucumber NBS-LRR Gene Family Analysis

A comprehensive analysis of NBS-encoding genes in cucumber (Cucumis sativus) identified 57 NBS-encoding genes through a systematic annotation approach [59]. The researchers addressed annotation challenges by:

• Utilizing multiple genome assemblies ('Chinese Long' inbred line 9930 and gynoecious inbred line 'Gy14') to cross-validate gene models
• Implementing a three-step identification process involving BLAST searches, domain-based queries, and manual curation
• Classifying genes into seven distinct categories (TNL, CNL, TN, CN, N, NL, and RPW8-NL) based on domain architecture
• Identifying conserved motifs specific to TIR (TNBS-1) and CC (CNBS-1, CNBS-2) families to support phylogenetic classification

This careful annotation revealed that cucumber maintains both TIR and CC NBS-LRR families despite its relatively small NBS-encoding gene repertoire compared to other plants [59].

Grass Pea NBS-LRR Gene Identification

In grass pea (Lathyrus sativus), researchers identified 274 NBS-LRR genes from a genome assembly of 8.12 Gbp with an N50 of 59,728 bp [57]. The annotation strategy included:

• Using Local TBLASTN with a sequence similarity threshold of 90% and sequence length of 600 nucleotides
• Applying TransDecoder to predict coding regions from transcript sequences
• Validating domains through NCBI-CDD and custom HMM searches
• Analyzing gene structures, conserved motifs, and phylogenetic relationships

The study successfully classified 124 genes with TNL domains and 150 with CNL domains, providing a foundation for future resistance gene isolation and characterization [57].

Cross-Species NBS Gene Orthogroup Analysis

A large-scale study analyzing 12,820 NBS-domain-containing genes across 34 plant species demonstrated the power of comparative approaches for addressing annotation challenges [32]. Key methodological advances included:

• Orthogroup clustering: Using OrthoFinder v2.5.1 with the DIAMOND tool for sequence similarity searches and MCL for clustering, identifying 603 orthogroups with both core and species-specific groups
• Tandem duplication analysis: Identifying lineage-specific expansions through analysis of gene duplication patterns
• Expression validation: Utilizing RNA-seq data from multiple sources to verify putative gene models and assess functional relevance
• Genetic variation assessment: Comparing susceptible and tolerant cotton accessions to identify unique variants associated with disease resistance

This approach revealed significant diversification in NBS gene domain architectures, with several species-specific structural patterns discovered beyond the classical NBS domain combinations [32].

Addressing annotation hurdles posed by fragmented genes and repetitive sequences is essential for advancing research on NBS disease resistance genes. The methodologies outlined in this technical guide provide a framework for improving annotation accuracy through integrated computational and experimental approaches. As sequencing technologies continue to evolve, with long-read sequencing becoming more accessible and affordable, the resolution of complex NBS-LRR loci will improve significantly. Furthermore, the integration of multi-omics data and machine learning approaches holds promise for further enhancing the annotation of these critical disease resistance genes. The continued refinement of annotation methodologies will directly support the identification and characterization of NBS-LRR genes for crop improvement and sustainable agriculture.

Within the study of plant disease resistance, Nucleotide-Binding Site Leucine-Rich Repeat (NBS-LRR) genes represent the largest and most critical family of resistance (R) genes, directly responsible for effector-triggered immunity (ETI) following pathogen recognition [10] [32]. The domain architecture of these proteins fundamentally dictates their function and recognition specificity. Two major subclasses exist based on their N-terminal domains: those possessing a Coiled-Coil (CC) domain and those containing a Toll/Interleukin-1 Receptor (TIR) domain, classifying them as CNLs or TNLs, respectively [10] [65]. A third subclass, RNL, characterized by an RPW8 domain, often functions in downstream signaling [10] [32].

Accurate identification of CC and TIR domains is therefore paramount for correctly classifying NLR genes, predicting their function, and understanding plant immune mechanisms. This guide synthesizes current methodologies and best practices for optimizing domain prediction, framed within the broader context of domain architecture and classification research for NBS disease resistance genes.

Classification and Diversity of NBS Domain Architectures

The central NBS domain is highly conserved and contains characteristic motifs that facilitate its identification, while the variable N-terminal domains (CC or TIR) present the primary classification challenge [10] [65]. Genome-wide analyses across diverse plant species reveal significant variation in the composition and number of these NLR subfamilies.

Table 1: Distribution of NBS Gene Subfamilies in Select Plant Species

Plant Species	Total NBS Genes	CNL Genes	TNL Genes	RNL Genes	Reference
Akebia trifoliata	73	50	19	4	[10]
Dioscorea rotundata	167	166	0	1	[10]
Brassica napus	641	180	461	0	[10]
Wheat (Triticum aestivum)	~2,012	Not Specified	Not Specified	Not Specified	[32]
Arabidopsis thaliana	~150	Not Specified	Not Specified	Not Specified	[32]

Large-scale comparative studies have identified 12,820 NBS-domain-containing genes across 34 plant species, which can be classified into 168 distinct domain architecture classes [32]. Beyond the classical patterns (e.g., NBS, NBS-LRR, TIR-NBS, TIR-NBS-LRR), numerous species-specific patterns have been discovered, such as TIR-NBS-TIR-Cupin1 and Sugartr-NBS [32]. This remarkable diversification, driven primarily by tandem and dispersed gene duplications, underscores the need for robust and adaptable domain prediction strategies [10].

Integrated Workflow for Domain Identification and Classification

A comprehensive approach to identifying and classifying CC and TIR domains combines multiple bioinformatic tools and sequence analysis techniques. The following workflow integrates established methods with recent advancements.

Diagram 1: Domain identification and classification workflow.

Detailed Experimental Protocols

Protocol 1: Genome-Wide Identification of NBS Genes

This protocol is adapted from methodologies used in recent genome-wide analyses [10] [32].

• Sequence Dataset Preparation: Obtain the complete set of protein sequences for the target plant species from resources such as NCBI, Phytozome, or Plaza.
• NBS Domain Screening:
  • Use the Hidden Markov Model (HMM) profile for the NB-ARC domain (PF00931) to scan the proteome. Tools like PfamScan.pl or HMMER are suitable.
  • Set a conservative E-value threshold (e.g., 1.0) to cast a wide net for potential NBS-containing genes [10].
  • Perform a reciprocal BLASTP search using known NBS protein sequences as queries.
• Merge and Filter Candidates: Combine results from both methods and remove redundant entries. Verify the presence of the NBS domain in the non-redundant set by re-running against the Pfam database with a standard E-value threshold (e.g., 10⁻⁴).
• Classification via N-terminal Domain Identification:
  • TIR Domain: Use the NCBI Conserved Domain Database (CDD) or Pfam (PF01582) to scan for TIR domains [10].
  • CC Domain: Due to low sequence conservation, CC domains are often not reliably identified by Pfam searches. Use a dedicated coiled-coil prediction tool like COILS or DeepCoil with a defined probability threshold (e.g., 0.5) [10].
  • RPW8 Domain: Use the corresponding HMM profile (PF05659) to identify the RNL subclass.
• Validate Domain Architecture: Manually inspect the arrangement of identified domains (TIR/CC, NBS, LRR) to confirm a coherent architecture and classify genes into CNL, TNL, or RNL subfamilies.

Protocol 2: In Silico Prediction of NLR-Effector Interactions

Accurate domain prediction enables higher-level functional studies, such as predicting which NLRs recognize specific pathogen effectors. A structure-based approach using machine learning represents the current state-of-the-art [66].

• Identify Candidate NLRs: Use Protocol 1 to identify NLR genes, focusing on those with direct effector-recognition capability (often singletons with high amino acid diversity in the LRR domain) [66].
• Predict Protein Complex Structures: For NLR-effector pairs of interest, use AlphaFold2-Multimer to predict the 3D structure of the complex. The NLRLRR domain is typically used for this analysis [66].
• Calculate Binding Energetics: Use the predicted structures as input for machine learning models (e.g., Area-Affinity with 97 models) to calculate the binding affinity (BA) and binding energy (BE) for the complex [66].
• Classify Interactions: An Ensemble machine learning model can distinguish "true" interacting pairs from non-functional "forced" pairs based on BA and BE values. "True" interactions typically show binding affinities between -8.5 and -10.6 log(K) and binding energies between -11.8 and -14.4 kcal/mol [66].

Table 2: Key Research Reagent Solutions for NLR Domain Analysis

Tool / Resource	Type	Primary Function in Domain Prediction	Access
Pfam / HMMER	Database & Search Tool	Identifies conserved protein domains (NBS, TIR, LRR) using Hidden Markov Models.	EBI Website
NCBI CDD	Database	Conserved Domain Database for scanning sequences against domain models.	NCBI Website
COILS / DeepCoil	Prediction Server	Predicts coiled-coil (CC) domains with a configurable probability score.	ExPASy / Standalone
AlphaFold2-Multimer	AI Prediction Tool	Predicts 3D structures of protein complexes (e.g., NLR-Effector).	ColabFold / Local
MEME Suite	Motif Analysis	Discovers conserved motifs within identified domains.	Online Suite
OrthoFinder	Phylogenetic Tool	Infers orthogroups and evolutionary relationships among NBS genes.	Standalone Package
D-I-TASSER	Hybrid Prediction Tool	Integrates deep learning with physics-based simulations for high-accuracy structure prediction, particularly for multi-domain proteins [67].	Online Server

Advanced Strategies and Optimization Techniques

Leveraging Protein Structure Prediction

Over-reliance on primary sequence can lead to misclassification. Integrating protein structure prediction significantly enhances accuracy.

• Windowed MSA for Chimeric Proteins: When analyzing fused or engineered proteins (e.g., for functional validation), standard multiple sequence alignment (MSA) methods fail, degrading prediction accuracy. The Windowed MSA approach independently computes MSAs for the target peptide and scaffold protein, then merges them. This strategy restored prediction accuracy in 65% of tested fusion constructs [68].
• Hybrid Folding Simulations: For large, multi-domain NLR proteins, tools like D-I-TASSER, which integrate deep learning potentials with iterative threading assembly simulations, can outperform end-to-end learning tools like AlphaFold2 and AlphaFold3. Its domain-splitting protocol is particularly valuable for modeling complex NLR architectures [67].

Phylogenetic and Expression Analyses

Contextualizing predictions within evolutionary and functional data provides strong validation.

• Orthogroup Analysis: Clustering NBS genes from multiple species into orthogroups (OGs) can reveal core conserved lineages and species-specific expansions. Studies have identified 603 such OGs, with certain groups (e.g., OG0, OG1, OG2) being central [32].
• Expression Profiling: NBS genes are often expressed at low levels, but some show specific upregulation in certain tissues or under biotic/abiotic stresses. Analyzing RNA-seq data (e.g., FPKM values) helps prioritize functionally relevant NLR candidates for further study [10] [32].

Diagram 2: Simplified NLR signaling pathway.

Accurate prediction of CC and TIR domains is a foundational step in elucidating the function and evolution of the complex NBS-LRR gene family in plants. A successful strategy requires an integrated, multi-layered approach that moves beyond simple sequence scanning. Researchers are encouraged to combine standard domain profiling tools (HMMER, COILS) with advanced structure prediction methods (AlphaFold-Multimer, D-I-TASSER) and evolutionary context (OrthoFinder, expression analysis). Furthermore, specialized techniques like Windowed MSA are crucial for analyzing non-natural protein fusions used in experimental validation. As these computational methodologies continue to advance, they will profoundly deepen our understanding of plant immunity and accelerate the development of disease-resistant crops.

Gene duplication is a fundamental evolutionary mechanism that provides the raw genetic material for innovation, adaptation, and the acquisition of new biological functions [69]. In eukaryotic genomes, duplicated genes are not randomly distributed but often form distinct clustered architectures primarily generated through two principal mechanisms: tandem duplications and segmental duplications (SDs). These architectures present significant challenges for genomic analysis, from assembly and annotation to functional characterization, yet they represent hotbeds of genetic diversity and rapid evolution.

This technical guide examines the analytical frameworks required to resolve these complex genomic regions, with particular emphasis on the NBS-LRR gene family, a critical component of plant immune systems. The intricate domain architecture and dynamic duplication patterns of NBS-LRR genes make them an ideal model system for exploring the broader principles governing duplicated gene families. Understanding these architectures is essential for elucidating how plants evolve new disease resistance specificities and how these mechanisms can be harnessed for crop improvement [3] [4].

Mechanisms and Fates of Duplicated Genes

Duplication Mechanisms and Their Genomic Signatures

The formation of duplicated genes occurs through distinct molecular mechanisms, each imparting characteristic signatures on genome architecture [69]:

• Whole Genome Duplication (WGD): This mechanism involves the duplication of complete chromosome sets, creating ohnologs (paralogs formed by WGD). WGD events are particularly prevalent in plant evolution, with correlations observed between WGD and increased speciation rates [69]. Following WGD, genomes undergo fractionation (heavy loss of duplicated genes) and diploidization (chromosomal rearrangements and segment loss as the genome returns to a diploid state) [69].

• Tandem Duplications: These localized events create novel gene copies adjacent to their progenitors, producing tandemly arrayed genes (TAGs). The primary molecular mechanism involves unequal crossing over, which can occur through homologous recombination between sequences or non-homologous recombination via replication-dependent chromosome breakages [69] [70]. When multiple unequal crossovers occur, they can drive expanding or contracting copy numbers in gene families.

• Segmental Duplications (SDs): These are defined as blocks of homologous DNA greater than 1 kb in length with >90% sequence identity [71]. In humans, approximately 60% of SDs are interspersed—separated by more than 1 Mb within a chromosome or mapping to non-homologous chromosomes [71]. SDs contribute significantly to structural variation through non-allelic homologous recombination (NAHR) and represent some of the most challenging genomic regions to resolve.

Table 1: Characteristics of Major Gene Duplication Mechanisms

Mechanism	Definition	Primary Molecular Process	Genomic Signature	Evolutionary Impact
Whole Genome Duplication (WGD)	Duplication of complete chromosome sets	Polyspermy, non-reduced gametes, incomplete mitosis	Genome-wide paralogy; syntenic blocks across chromosomes	Major source of genetic novelty; associated with speciation
Tandem Duplication	Localized duplication creating adjacent gene copies	Unequal crossing over	Clustered genes in direct genomic arrays	Rapid expansion of gene families; adaptation to environmental pressures
Segmental Duplication (SD)	Duplicated blocks >1kb with >90% identity	Non-allelic homologous recombination (NAHR)	Interspersed intra- and inter-chromosomal repeats	Significant contributor to structural variation and disease

Evolutionary Fates of Duplicated Genes

Once fixed in a population, duplicated genes face several evolutionary trajectories [72]:

• Pseudogenization: One copy accumulates silencing mutations, becoming a non-functional pseudogene. This is considered the most common fate due to functional redundancy [72].
• Subfunctionalization: Both copies partition aspects of the original function, with neither gene retaining full functionality alone [72].
• Neofunctionalization: One copy maintains the original function while the other evolves a novel function through positive selection [72].

The Duplication-Degeneration-Complementation (DDC) model provides a framework for understanding how mutations in regulatory regions can lead to subfunctionalization, while classical population genetics models emphasize the role of beneficial mutations in driving neofunctionalization [72].

Analytical Frameworks for Resolving Duplication Architectures

Identification and Classification of Duplicated Genes

Resolving clustered gene architectures requires specialized bioinformatic approaches tailored to different duplication mechanisms [69]. For the NBS-LRR gene family, the following analytical pipeline has proven effective:

Step 1: Domain-Based Identification The initial identification phase employs hidden Markov model (HMM) searches using domain models (e.g., PF00931 for the NB-ARC domain from the Pfam database) to identify candidate genes [3] [4]. Subsequent validation through the NCBI Conserved Domain Database (CDD) confirms domain completeness and identifies auxiliary domains including TIR, CC, and LRR domains [3].

Step 2: Phylogenetic Classification Multiple sequence alignment of identified protein sequences (using tools such as MUSCLE) followed by phylogenetic reconstruction (e.g., with MEGA11) enables classification of NBS-LRR genes into distinct clades based on domain architecture [3] [4]. Standard classifications include:

• TNL: TIR-NBS-LRR
• CNL: CC-NBS-LRR
• NL: NBS-LRR
• TN: TIR-NBS
• CN: CC-NBS
• N: NBS-only [4]

Step 3: Duplication Type Assessment The application of synteny analysis tools such as MCScanX enables the discrimination of duplication types by identifying tandem duplicates (genes located within 10 kb of one another), intrachromosomal duplicates (same chromosome, >10 kb apart), and interchromosomal duplicates (different chromosomes) [3] [73].

Table 2: Quantitative Distribution of NBS-LRR Genes in Nicotiana Species

Species	Genome Type	Total NBS Genes	TNL	CNL	NL	TN	CN	N
N. tabacum	Allotetraploid	603	64	74	306	9	150	-
N. sylvestris	Diploid	344	37	48	172	5	82	-
N. tomentosiformis	Diploid	279	33	47	127	7	65	-

Data adapted from [3] showing the distribution of NBS-LRR gene types across three Nicotiana species, demonstrating the expansion of NBS genes in the allopolyploid N. tabacum.

Evolutionary Dynamics and Selection Pressures

Analysis of evolutionary dynamics involves calculating non-synonymous (Ka) and synonymous (Ks) substitution rates using tools like KaKs_Calculator 2.0 [3]. The Ka/Ks ratio serves as an indicator of selective pressure:

• Ka/Ks < 1: Purifying selection
• Ka/Ks = 1: Neutral evolution
• Ka/Ks > 1: Positive selection

Population genetic analyses, including Tajima's D, Fu & Li's tests, and the McDonald-Kreitman test, can further identify signatures of selection acting on duplicated genes [70]. For example, analysis of tandemly duplicated genes in Drosophila has revealed strong evidence of positive selection driving functional diversification [70].

Resolving Structural Polymorphisms in Segmental Duplications

Long-read sequencing technologies (PacBio HiFi, Oxford Nanopore) have revolutionized the resolution of SDs by enabling full haplotype phasing and assembly [71]. The analytical workflow includes:

• Assembly of haplotype-resolved genomes using HiFi data and specialized assemblers
• Identification of SDs through self-alignment and application of SD operational definitions (>1 kb, >90% identity)
• Classification into fixed versus polymorphic SDs based on population frequency
• Gene annotation within SDs to identify copy-number polymorphic genes

Recent studies of 170 human genomes have revealed that approximately 47.4 Mb of SD sequence was absent from the telomere-to-telomere reference genome, with intrachromosomal SDs displaying greater polymorphism than interchromosomal events [71]. African genomes harbor significantly more intrachromosomal SDs and recently duplicated gene families with higher copy numbers, highlighting the importance of population diversity in SD analysis [71].

The NBS-LRR Gene Family: A Case Study in Duplication Architecture

Domain Architecture and Functional Significance

The NBS-LRR gene family represents one of the most extensively duplicated gene families in plants, playing a critical role in disease resistance as intracellular immune receptors [3] [4] [6]. Their protein architecture typically consists of:

• N-terminal domain: Either a TIR (Toll/Interleukin-1 receptor) or CC (coiled-coil) domain involved in signaling and partner interaction
• NBS (NB-ARC) domain: A nucleotide-binding domain that functions as a molecular switch, alternating between ADP-bound (inactive) and ATP-bound (active) states
• LRR domain: A leucine-rich repeat region responsible for specific pathogen effector recognition [4] [6]

NBS-LRR genes are preferentially organized in clusters of closely duplicated genes, though they can also exist as singletons distributed throughout the genome [6]. This clustered arrangement facilitates the generation of diversity through unequal crossing over and gene conversion.

Experimental Workflow for NBS-LRR Gene Analysis

The following diagram illustrates a comprehensive workflow for identifying and characterizing NBS-LRR genes, integrating multiple bioinformatic approaches:

Diagram Title: Comprehensive NBS-LRR Gene Analysis Workflow

Advanced Computational Approaches

Recent advances in deep learning have enabled the development of tools like PRGminer, which employs dipeptide composition and convolutional neural networks to identify resistance genes with high accuracy (98.75% in k-fold testing) [6]. This approach circumvents limitations of homology-based methods, particularly for identifying novel R-genes with low sequence similarity to known candidates.

Research Reagent Solutions for Duplication Analysis

Table 3: Essential Research Reagents and Computational Tools for Duplication Analysis

Category	Tool/Reagent	Specific Function	Application Context
Domain Databases	Pfam (PF00931)	HMM models for NBS domain identification	Initial identification of NBS-LRR candidates [3] [4]
Alignment & Phylogenetics	MUSCLE v3.8.31	Multiple sequence alignment	Phylogenetic reconstruction of gene families [3]
Synteny Analysis	MCScanX	Identification of duplication types	Discriminating tandem, intrachromosomal, and interchromosomal duplications [3] [73]
Selection Analysis	KaKs_Calculator 2.0	Calculation of Ka/Ks ratios	Quantifying selective pressures on duplicated genes [3]
Gene Annotation	NCBI CDD	Domain verification and completeness	Confirming domain architecture in candidate genes [3]
Deep Learning	PRGminer	R-gene prediction and classification	Identifying resistance genes beyond homology-based methods [6]
Sequencing Technology	PacBio HiFi	Long-read sequencing	Resolving complex duplicated regions [71]

The resolution of clustered gene architectures arising from tandem and segmental duplications requires specialized analytical frameworks that integrate evolutionary theory, bioinformatic tools, and advanced sequencing technologies. The NBS-LRR gene family exemplifies how duplication mechanisms drive the evolution of critical biological functions, particularly in plant immunity. As long-read sequencing technologies continue to mature and computational approaches like deep learning become more sophisticated, our ability to resolve these complex genomic regions will expand, offering new insights into genome evolution and creating opportunities for engineering disease resistance in crop species. The analytical frameworks presented here provide a roadmap for navigating the complexities of duplicated gene families across diverse biological systems.

Troubleshooting Low-Expression Genes and Incomplete Genome Assemblies

In the field of nucleotide-binding domain and leucine-rich repeat (NLR) research, a comprehensive understanding of gene function is often hampered by two significant technical challenges: the inherent low expression of certain resistance genes and the prevalence of incomplete or fragmented genome assemblies. These obstacles are particularly problematic for the domain architecture and classification studies of NBS disease resistance genes, as they can lead to the omission of crucial gene family members and misinterpretation of functional capabilities.

Recent evidence challenges the long-held assumption that NLR genes are universally maintained at low expression levels to avoid autoimmunity. Studies have revealed that functional NLRs actually exhibit substantial expression in uninfected plants, with known resistance genes frequently appearing among the most highly expressed NLR transcripts [74]. This paradigm shift underscores the necessity of distinguishing between truly low-expression genes and those that appear under-expressed due to technical artifacts from incomplete genomic data.

This technical guide provides a systematic framework for addressing these challenges, offering detailed methodologies for accurate gene expression analysis, genome assembly completion, and functional validation specifically within the context of NLR gene research.

Technical Challenges in NLR Gene Analysis

The Problem of Low-Expression NLR Genes

The transcriptional regulation of NLR genes has traditionally been considered tightly constrained, with the pervasive hypothesis that low expression levels prevent deleterious autoimmune responses. However, recent transcriptomic analyses across multiple plant species have revealed that functional NLRs are often present among highly expressed transcripts in uninfected tissues [74].

Table 1: Documented Cases of Functional High-Expression NLR Genes

NLR Gene	Species	Expression Level	Pathogen Targeted	Functional Evidence
Mla7	Barley (Hordeum vulgare)	Requires multiple copies for function	Blumeria hordei (powdery mildew)	Multicopy transgene complementation [74]
ZAR1	Arabidopsis thaliana	Most highly expressed NLR in Col-0 ecotype	Multiple bacterial pathogens	Natural variant analysis [74]
Rpi-amr1	Solanum americanum	Highly expressed isoform is functional	Phytophthora infestans	Isoform-specific functional validation [74]
Mi-1	Tomato (Solanum lycopersicum)	High expression in leaves and roots	Aphids, whiteflies, nematodes	Tissue-specific activity confirmation [74]
Sr46, SrTA1662, Sr45	Aegilops tauschii	Highly expressed across accessions	Wheat stem rust (P. graminis)	Expression quantitative trait loci mapping [74]

The case of barley Mla7 illustrates a critical principle in NLR biology: some functional resistance genes require threshold expression levels for activity. In complementation experiments, single insertions of Mla7 driven by its native promoter were insufficient to confer resistance, whereas lines carrying two or more copies showed clear resistance to powdery mildew pathogens [74]. This gene natively exists as three identical copies in the haploid genome of barley cv. CI 16147, supporting the hypothesis that a specific expression threshold is required for function.

Limitations of Incomplete Genome Assemblies

Incomplete genome assemblies present substantial obstacles to accurate NLR gene characterization. These limitations manifest primarily in:

• Fragmented gene models: NLR genes are frequently fragmented across multiple contigs, preventing comprehensive domain architecture analysis.
• Missing gene family members: Tandemly duplicated NLRs are often underrepresented in draft genomes due to their sequence similarity and complex genomic organization.
• Misassembly of homologous regions: The high sequence similarity among paralogous NLR genes leads to systematic assembly errors, including gene collapse and chimeric constructs.

Short-read sequencing technologies exacerbate these challenges, particularly in regions with high sequence homology. Mapping accuracy decreases significantly when reads cannot be uniquely placed in a genomic context, such as in areas with paralogous genes or pseudogenes [75]. This problem affects both gene expression quantification (through ambiguous read mapping) and structural annotation.

Quantitative Assessment of Technical Challenges

Impact of Sequencing and Assembly Methods

Table 2: Effect of Read Length on Mapping Accuracy in Homologous Regions

Read Length	Correctly Mapped Reads	Incorrectly Mapped Reads	Unmapped Reads	Average Depth Coverage	Remedied Low-Depth Genes
75 bp	>99%	<1%	<1%	Low (higher variance)	Baseline
100 bp	>99%	Fewer than 75 bp	Fewer than 75 bp	Moderate	15/35 genes
150 bp	>99%	Fewest	Fewest	High (lower variance)	25/35 genes
250 bp	>99%	Minimal	Minimal	Highest (lowest variance)	35/35 genes with resolvable homology

Longer read lengths significantly improve mapping accuracy and depth coverage across homologous regions [75]. However, even 250 bp reads cannot resolve regions with extreme homology, such as the SMN1/SMN2 paralogs, which exhibit near-identical sequences [75]. In such cases, alternative approaches are necessary.

Experimental Protocols for Resolution

Protocol 1: Copy Number Validation for Low-Expression NLRs

Purpose: To determine whether apparently low-expression NLRs require threshold copy numbers for function, as demonstrated with barley Mla7 [74].

Materials:

• Plant expression vector with native NLR promoter
• Recipient plant line lacking the target NLR function
• Agrobacterium tumefaciens strain for plant transformation
• Selection agents appropriate for the transformation system
• qPCR reagents for copy number determination
• Pathogen isolates for phenotyping

Methodology:

• Clone the full-length NLR genomic sequence (including native promoter and terminator) into a plant expression vector.
• Transform recipient plants and select multiple independent transformants.
• Determine transgene copy number in T0 plants using quantitative PCR with transgene-specific primers.
• Propagate lines to generate families segregating for different copy numbers (0-4 copies).
• Measure expression levels of the transgene in each line using RNA-seq or RT-qPCR.
• Challenge plants with cognate pathogens and assess resistance phenotypes.
• Correlate copy number, expression level, and resistance phenotype.

Expected Outcomes: This protocol determines whether functional complementation requires multiple gene copies, indicating a threshold expression effect. As observed with Mla7, higher-order copies may be required for full resistance, with four copies needed to recapitulate native resistance levels [74].

Protocol 2: Genome Assembly Completion Using Hybrid Approaches

Purpose: To resolve incomplete NLR gene models in draft genome assemblies through integration of long-read sequencing and optical mapping.

Materials:

• High-molecular-weight genomic DNA
• Long-read sequencing platform (PacBio or Oxford Nanopore)
• Optical mapping system (Bionano Genomics)
• Computing infrastructure for hybrid assembly
• Gene-specific PCR primers for validation

Methodology:

• Prepare DNA libraries for long-read sequencing (>10 kb insert size).
• Generate 100-200x genome coverage with long reads.
• Perform de novo assembly using dedicated long-read assemblers (Canu, Flye).
• Extract high-molecular-weight DNA for optical mapping.
• Label specific sequence motifs and image single molecules.
• Assemble optical maps and integrate with sequence-based assembly.
• Resolve misassemblies and close gaps in NLR-containing regions.
• Validate completed NLR loci through PCR and Sanger sequencing.

Expected Outcomes: Hybrid assembly produces more contiguous genomes with complete NLR gene models, reducing fragmentation and missing gene family members. This approach is particularly valuable for resolving complex NLR clusters with tandem duplicates.

Protocol 3: Expression-Level Guided NLR Discovery

Purpose: To identify functional NLR candidates based on their steady-state expression levels in uninfected tissues [74].

Materials:

• RNA from appropriate plant tissues (considering pathogen infection sites)
• RNA-seq library preparation kit
• Sequencing platform
• Bioinformatics tools for transcriptome assembly and expression quantification
• NLR annotation pipeline

Methodology:

• Extract RNA from relevant plant tissues (leaf, root, etc.) in biological replicates.
• Prepare strand-specific RNA-seq libraries.
• Sequence to sufficient depth (>30 million reads per sample).
• Assemble transcripts de novo or align to reference genome.
• Quantify expression levels of all annotated NLR genes.
• Rank NLRs by expression level (TPM or FPKM).
• Prioritize highly expressed NLRs for functional validation.
• Validate expression patterns using RT-qPCR.

Expected Outcomes: This expression-guided approach enriches for functional NLR candidates. In Arabidopsis thaliana, known NLRs are significantly enriched in the top 15% of expressed NLR transcripts compared with the lower 85% (χ² test, P = 0.038) [74].

Visualization of Experimental Workflows

Workflow for troubleshooting low-expression NLR genes

Research Reagent Solutions

Table 3: Essential Research Reagents for NLR Gene Characterization

Reagent/Category	Specific Examples	Function/Application	Technical Notes
Expression Vectors	pLysS, pLysE, lysY host strains	Control basal expression in toxic protein expression	T7 lysozyme inhibits T7 RNA Polymerase; critical for threshold-dependent NLRs [76]
Tunable Expression Systems	Lemo21(DE3) with PrhaBAD promoter	Fine-tune expression levels for toxic proteins	L-rhamnose concentration inversely proportional to protein production [76]
Solubility Enhancers	pMAL vectors with MBP tag, GroEL/DnaK/ClpB chaperonins	Improve yield of properly folded NLR proteins	MBP fusion aids expression/solubility; chaperones assist complex folding [76]
Disulfide Bond Systems	SHuffle strains with cytoplasmic DsbC	Enable correct disulfide bond formation in cytoplasm	Mutations alter cellular redox; essential for NLRs requiring specific cysteine pairs [76]
Hybrid Assembly Tools	PacBio/Oxford Nanopore, Bionano, Hi-C	Resolve complex NLR clusters in genomes	Long reads span repeats; optical mapping validates large-scale structure [75]
Variant Calling Pipelines	GATK HaplotypeCaller, custom filters for homologous regions	Accurate SNP/indel detection in paralogous NLRs	Standard pipelines require modification for high-homology regions [75]

Discussion and Future Perspectives

The integration of expression-level analysis with advanced genomic solutions provides a powerful framework for overcoming long-standing challenges in NLR gene research. The recognition that functional NLRs are frequently highly expressed overturns conventional assumptions and offers a practical discovery tool for identifying new resistance genes.

Future developments in several technological areas will further enhance NLR characterization:

• Single-cell and spatial transcriptomics will enable precise determination of NLR expression patterns at cellular resolution, revealing expression heterogeneity within tissues.
• Telomere-to-telomere (T2T) genome assemblies will completely resolve complex NLR clusters, eliminating the gaps and misassemblies that plague current references.
• Protein structure prediction algorithms (e.g., AlphaFold2) combined with expression data will facilitate structure-function analyses of NLR domains.
• Genomic language models like Evo demonstrate potential for semantic design of novel NLRs based on genomic context [77], though this approach remains exploratory for plant resistance genes.

As these technologies mature, the research community will move closer to comprehensive classification of NLR domain architectures and their functional correlates, ultimately enabling more precise engineering of disease resistance in crop species.

The strategic application of the protocols and reagents described in this guide will accelerate the resolution of low-expression genes and incomplete genome assemblies, removing critical bottlenecks in NBS disease resistance gene research.

Benchmarking serves as a critical tool for evaluating the performance of computational methods in scientific research, yet significant challenges persist in ensuring its accuracy and real-world applicability. This technical guide examines the core principles of effective benchmarking, with a specific focus on methods for predicting the function and evolution of Nucleotide-Binding Site (NBS) disease resistance genes in plants. We analyze key metrics for assessing tool performance, detail experimental protocols for NBS gene identification and characterization, and identify prevalent limitations in current benchmarking approaches. Within the context of NBS gene domain architecture and classification, this review provides researchers with a framework for developing robust evaluation methodologies that generate biologically meaningful results and advance crop improvement efforts.

Benchmarking represents a structured process that compares key performance indicators against established business objectives or industry standards, enabling the objective assessment of how well a computational platform meets specific operational needs [78]. In the domain of plant genomics, effective benchmarking is particularly crucial for evaluating tools that predict the structure, function, and evolution of disease resistance genes, especially the NBS gene family which encodes proteins containing nucleotide-binding sites and C-terminal leucine-rich repeats (LRRs) [10] [32]. These genes constitute the largest family of plant resistance (R) genes, accounting for over 60% of detected and cloned R genes across all plant species and playing vital roles in effector-triggered immunity [10] [2].

The expansion of genomic resources for numerous plant species, including recently sequenced crops like pepper (Capsicum annuum L.) and various Dendrobium orchids, has created unprecedented opportunities for genome-wide analysis of NBS genes [2] [28]. However, this data abundance also presents significant benchmarking challenges. Current prediction methods must be evaluated on their accuracy in identifying classical NBS domain architectures (NBS, NBS-LRR, TIR-NBS, TIR-NBS-LRR) as well as species-specific structural patterns (TIR-NBS-TIR-Cupin1-Cupin1, TIR-NBS-Prenyltransf, Sugar_tr-NBS) that have been discovered through comparative genomics [32]. Moreover, with the integration of machine learning approaches in genomics, understanding the limitations of benchmarking these computational methods becomes increasingly important for research reliability and progress [79] [80].

This technical guide addresses the critical aspects of benchmarking tool performance within the context of NBS disease resistance gene research. It provides a comprehensive framework for evaluating prediction method accuracy, details standardized experimental protocols, identifies common benchmarking limitations, and offers visualization approaches for complex data relationships, ultimately empowering researchers to make informed decisions in tool selection and method development.

Key Metrics for Benchmarking Prediction Tools

Accuracy Dimensions in NBS Gene Prediction

Accuracy serves as the foundational metric for evaluating prediction tools in NBS gene research, encompassing multiple dimensions of correctness and relevance. For NBS gene identification, accuracy measurements should specifically assess:

• Tool calling accuracy: The system's ability to correctly identify NBS domains and associated architectural elements. Industry benchmarks for 2025 set expectations of 90% or higher for top-performing tools [78].
• Domain recognition precision: Accuracy in classifying NBS genes into appropriate subfamilies (TNL, CNL, RNL) based on N-terminal domains and architectural patterns [32] [2].
• Context retention: Capability to maintain contextual information across multi-step analyses, such as tracking gene family evolution through duplication events [78] [10].

Tools must be evaluated using real genomic datasets that reflect diverse use cases, comparing predictions against gold-standard sets of known-correct annotations [78]. For example, in pepper genomics, comprehensive benchmarking would involve verifying the identification of 252 NBS-LRR resistance genes against manual curation, with particular attention to the correct classification of 248 nTNLs (non-TIR NBS-LRR) versus 4 TNLs (TIR NBS-LRR) [2].

Speed and Throughput Considerations

While accuracy remains paramount, speed metrics determine the practical utility of prediction tools, especially as genomic datasets continue to expand. Key speed considerations include:

• Response time: The average duration from query submission to result display, with industry benchmarks targeting under 1.5 to 2.5 seconds for interactive analysis [78].
• Update frequency: How quickly new or modified genomic information becomes searchable and analyzable, with leading platforms supporting real-time or near-real-time indexing [78].
• Processing throughput: The volume of genomic data a tool can process within a specified timeframe, particularly important for pan-genome analyses involving multiple assemblies [81].

Different research contexts prioritize different speed dimensions. For evolutionary studies analyzing NBS gene clusters across multiple species, processing throughput may be most critical, whereas interactive genome annotation requires optimized response times [78] [32].

Comprehensive Evaluation Metrics

Effective benchmarking requires a multidimensional approach that incorporates both quantitative and qualitative metrics tailored to NBS gene research:

Table 1: Core Metrics for Benchmarking NBS Gene Prediction Tools

Metric Category	Specific Measurements	Application to NBS Gene Research
Accuracy	Tool calling accuracy, Domain recognition precision, Context retention	Correct identification of NBS subfamilies (CNL, TNL, RNL) and architectural variations
Speed	Response time, Update frequency, Processing throughput	Efficient analysis of large genomic datasets and multi-species comparisons
Resource Utilization	Memory consumption, CPU usage, Storage requirements	Practical constraints for analyzing complex plant genomes (e.g., soybean ~1Gb)
Usability	Interface intuitiveness, Documentation quality, Workflow integration	Accessibility for researchers with varying computational expertise
Biological Relevance	Functional prediction accuracy, Evolutionary pattern recognition	Correct inference of NBS gene expansion mechanisms (tandem/segmental duplications)

Experimental Protocols for NBS Gene Analysis

Genome-Wide Identification of NBS Genes

The comprehensive identification of NBS genes within plant genomes requires a multi-method approach to ensure complete coverage. The following protocol, adapted from methodologies used in recent studies of Akebia trifoliata, pepper, and Dendrobium orchids [10] [2] [28], provides a robust framework:

• Data Acquisition: Obtain the latest genome assembly and annotation files from relevant databases (NCBI, Phytozome, Plaza) [32]. For example, in the pepper genome study, researchers utilized these resources to access the chromosomal sequences [2].

• Initial Candidate Identification:

  • Perform BLASTP analysis against reference NBS proteins using the NB-ARC domain (PF00931) as query with an E-value threshold of 1.0 [10].
  • Conduct hidden Markov model (HMM) scanning using the NB-ARC domain profile (accession: PF00931) with the same E-value threshold [10] [32].
  • Merge candidate genes from both approaches and remove redundancies [10].

• Domain Validation and Classification:

  • Analyze non-redundant genes against the Pfam database to verify NBS domain presence (E-value: 10^-4) [10].
  • Classify validated NBS genes using the NCBI Conserved Domain Database to identify TIR (PF01582), RPW8 (PF05659), and LRR (PF08191) domains [10].
  • Identify coiled-coil (CC) domains using tools like Coiledcoil with a threshold value of 0.5, as these domains are not always detected by Pfam searches [10].

• Subfamily Categorization:

  • Categorize genes into main subfamilies: CC-NBS-LRR (CNL), TIR-NBS-LRR (TNL), and RPW8-NBS-LRR (RNL) [10] [2].
  • Document unusual domain architectures and species-specific structural patterns [32].

This protocol enabled the identification of 73 NBS genes in Akebia trifoliata (50 CNL, 19 TNL, 4 RNL) and 252 NBS-LRR genes in pepper, demonstrating its applicability across diverse plant species [10] [2].

Evolutionary and Expression Analysis

Understanding the evolutionary dynamics and functional roles of NBS genes requires additional analytical approaches:

• Phylogenetic Analysis:

  • Select conserved NBS domain sequences for multiple sequence alignment using MAFFT 7.0 [32].
  • Construct phylogenetic trees using maximum likelihood algorithms in FastTreeMP with 1000 bootstrap replicates [32].
  • Visualize evolutionary relationships between NBS subfamilies and identify orthogroups using OrthoFinder v2.5.1 with the MCL clustering algorithm [32].

• Genomic Distribution Mapping:

  • Map NBS candidates to chromosomes and identify distribution patterns using genetic linkage maps [2].
  • Identify gene clusters (defined as ≥2 NBS genes within 200 kb) and singleton genes [2].
  • Analyze duplication mechanisms (tandem, segmental, whole-genome) contributing to NBS gene expansion [10] [32].

• Expression Profiling:

  • Retrieve RNA-seq data from relevant databases (NCBI BioProjects, species-specific expression databases) [10] [32].
  • Process data through transcriptomic pipelines to obtain normalized expression values (e.g., FPKM) [32].
  • Categorize expression patterns into tissue-specific, abiotic stress-responsive, and biotic stress-responsive profiles [32] [28].
  • For specific treatments (e.g., salicylic acid), identify differentially expressed genes (DEGs) and conduct weighted gene co-expression network analysis (WGCNA) to elucidate regulatory relationships [28].

The following diagram illustrates the comprehensive workflow for NBS gene identification and characterization:

Diagram 1: Workflow for comprehensive NBS gene identification and characterization

Successful NBS gene research requires specialized computational tools, databases, and experimental resources. The following table catalogs essential components of the research toolkit, compiled from methodologies used in recent studies [10] [32] [2]:

Table 2: Essential Research Reagents and Resources for NBS Gene Analysis

Tool/Resource	Type	Primary Function	Application Example
HMMER	Software	Hidden Markov Model scanning for domain identification	Identifying NB-ARC domains (PF00931) in protein sequences [10]
Pfam Database	Database	Protein family classification and domain verification	Validating NBS domain presence (E-value: 10^-4) [10] [32]
OrthoFinder	Software	Orthogroup inference and comparative genomics	Identifying core and unique orthogroups across species [32]
MEME Suite	Software	Motif discovery and sequence analysis	Identifying conserved motifs in NBS domains (P-loop, RNBS-A, kinase-2, etc.) [10] [2]
RNA-seq Data	Data	Transcriptome profiling and expression analysis	Determining NBS gene expression under stress conditions [32] [28]
VIGS System	Experimental	Functional validation through gene silencing	Testing role of specific NBS genes in disease resistance [32]
Asm2sv Pipeline	Software	Assembly-based structural variation detection	Identifying gene-level SVs in soybean pangenome analysis [81]

Limitations of Current Benchmarking Methods

Technical and Methodological Constraints

Current benchmarking approaches for prediction tools in genomics face several significant technical limitations that can compromise their validity and utility:

• Benchmark saturation: Occurs when leading models achieve near-perfect scores on standardized tests, eliminating meaningful differentiation. This phenomenon is observed when state-of-the-art systems score above 90% on common benchmarks, prompting some platforms to exclude saturated benchmarks entirely from their evaluations [82].

• Data contamination: Undermines validity when training data inadvertently includes test questions, leading to memorization rather than genuine reasoning capability. Research on mathematical benchmarks revealed evidence of memorization, with some model families experiencing up to 13% accuracy drops on contamination-free tests compared to original benchmarks [82].

• Construct validity issues: Many benchmarks fail to measure what they claim to measure, particularly for complex concepts like "fairness" and "bias" in genomic analyses. Without clear definitions and stable ground truth, benchmarks may provide false certainty about tool performance [80].

• Rapid capability obsolescence: The swift advancement of AI and genomic tools means benchmarks struggle to maintain relevance. In some cases, models achieve such high accuracy that benchmarks become ineffective, while slow implementation frameworks fail to flag risks in a timely manner [80].

Domain-Specific Challenges in NBS Gene Research

Benchmarking prediction tools for NBS gene analysis presents unique challenges rooted in the biological complexity of disease resistance gene families:

• Evolutionary dynamism: The NBS gene family exhibits remarkable diversity across plant species, with numbers ranging from dozens to over 2,000 members in different plants [10] [32]. This variation complicates the development of standardized benchmarking datasets.

• Architectural diversity: Beyond classical NBS domain architectures, numerous species-specific structural patterns exist, including recently discovered configurations like TIR-NBS-TIR-Cupin1-Cupin1 and TIR-NBS-Prenyltransf [32]. Benchmarking tools must account for this architectural heterogeneity.

• Subfamily composition disparities: The composition of NBS subclasses (TNL, CNL, RNL) varies dramatically between species. Some plants like Dioscorea rotundata possess 166 CNLs but only one RNL and no TNLs, while Brassica napus contains 461 TNLs and 180 CNLs but no RNLs [10]. These taxonomic differences challenge tool generalization.

• Clustering and distribution patterns: NBS genes frequently distribute unevenly across chromosomes, often clustering at chromosome ends [10] [2]. Prediction tools must accurately identify both clustered arrangements and singleton genes to perform effectively.

The following diagram illustrates the primary limitations and their relationships in current benchmarking practices:

Diagram 2: Key limitations in benchmarking genomic prediction tools

Effective benchmarking of prediction tools for NBS disease resistance gene research requires a sophisticated approach that acknowledges both technical limitations and biological complexity. As genomic datasets expand and computational methods evolve, benchmarking frameworks must adapt to maintain relevance and utility. The structured evaluation metrics, standardized protocols, and comprehensive resource cataloging presented in this guide provide a foundation for robust tool assessment.

Future benchmarking efforts should prioritize several key areas: developing contamination-resistant evaluation datasets that refresh regularly with novel biological sequences; implementing multimodal assessment strategies that combine automated metrics with expert biological validation; and creating specialized benchmarks for emerging research domains like pangenome structural variation analysis [82] [80] [81]. Additionally, greater attention to evolutionary context in NBS gene benchmarking—accounting for species-specific differences in subfamily distribution, duplication mechanisms, and architectural diversity—will enhance the biological relevance of tool evaluations.

By adopting the comprehensive benchmarking approaches outlined in this technical guide, researchers can more effectively navigate the complex landscape of prediction tools, ultimately accelerating progress in understanding plant immunity mechanisms and developing disease-resistant crop varieties through informed method selection and continuous improvement of evaluation frameworks.

Beyond Prediction: Validating Function and Evolutionary Patterns of NBS Genes

The domain architecture and classification of nucleotide-binding site leucine-rich repeat (NBS-LRR) genes provide crucial insights into plant immune system evolution and function. As the largest family of plant disease resistance (R) genes, NBS-LRR proteins function as intracellular immune receptors that detect pathogen effector proteins and activate effector-triggered immunity (ETI). Approximately 80% of cloned plant R genes belong to the NBS-LRR family [83] [7], which can be subdivided into distinct classes based on their N-terminal domains: TIR-NBS-LRR (TNL) containing Toll/Interleukin-1 receptor domains, CC-NBS-LRR (CNL) containing coiled-coil domains, and RPW8-NBS-LRR (RNL) containing resistance to powdery mildew 8 domains [4] [1]. Additional atypical forms exist that lack complete domain combinations, including TN, CN, NL, and N-types that may function as adaptors or regulators in immune signaling networks [4].

Understanding the precise functions of these complex gene families requires sophisticated experimental validation techniques that can connect genomic information with biological function. This technical guide examines two powerful approaches for characterizing NBS-LRR genes: Virus-Induced Gene Silencing (VIGS) for functional analysis and protein interaction assays for mechanistic studies. These methodologies enable researchers to move beyond bioinformatic predictions to experimental validation within the context of plant immune responses.

Virus-Induced Gene Silencing (VIGS) for NBS-LRR Gene Validation

Fundamental Principles and Applications

Virus-Induced Gene Silencing (VIGS) is a powerful reverse genetics technique that leverages the plant's innate RNA interference (RNAi) machinery to achieve targeted gene knockdown. The method utilizes recombinant viral vectors carrying fragments (typically 200-500 bp) of the target plant gene to trigger sequence-specific mRNA degradation [84]. When introduced into plants, these modified viruses both replicate and activate the plant's post-transcriptional gene silencing mechanism, leading to degradation of mRNAs homologous to the inserted sequence [85].

For NBS-LRR research, VIGS provides several distinct advantages over stable genetic transformation. It enables rapid functional screening of candidate genes without the need for time-consuming stable transformation, which is particularly valuable for species with long life cycles or recalcitrant transformation systems [85] [84]. VIGS can be applied to study gene function in specific tissues and at specific developmental stages, allowing researchers to investigate genes whose complete knockout might be lethal. The technique is especially valuable for validating NBS-LRR genes identified through genome-wide analyses, where numerous candidates require functional characterization [11].

Recent applications demonstrate the power of VIGS in NBS-LRR research. In tung trees, VIGS was successfully employed to validate the role of Vm019719, a TNL-type NBS-LRR gene that confers resistance to Fusarium wilt [11]. Similarly, in soybean, a TRV-based VIGS system was used to silence the rust resistance gene GmRpp6907 and defense-related gene GmRPT4, confirming their functions in disease resistance [85].

Established VIGS Protocols for Recalcitrant Species

TRV-Based VIGS in Soybean

The Tobacco Rattle Virus (TRV) system has been optimized for soybean functional genomics through Agrobacterium tumefaciens-mediated delivery. The protocol centers on infection via cotyledon nodes, which enables systemic viral spread and effective silencing of endogenous genes throughout the plant [85].

Vector Construction: Target gene fragments (200-300 bp) are amplified from cDNA using gene-specific primers containing appropriate restriction sites (e.g., EcoRI and XhoI). The fragments are cloned into the pTRV2-GFP vector, and recombinant plasmids are verified by sequencing before transformation into Agrobacterium tumefaciens GV3101 [85].

Agroinfiltration Method: Conventional infiltration methods often show low efficiency in soybean due to thick cuticles and dense trichomes. The optimized protocol involves:

• Surface-sterilizing soybean seeds and soaking in sterile water until swollen
• Bisecting seeds longitudinally to obtain half-seed explants
• Immersing fresh explants for 20-30 minutes in Agrobacterium suspensions containing pTRV1 or pTRV2 derivatives
• Co-cultivating infected explants on sterile medium for 2-3 days before transplanting to soil [85]

Efficiency Assessment: Using this method, infection efficiency exceeding 80% has been achieved, reaching up to 95% for certain soybean cultivars. Silencing efficiency typically ranges from 65% to 95%, as confirmed by phenotypic observations and quantitative PCR [85].

VIGS in Woody Plant Tissues

For recalcitrant woody tissues like Camellia drupifera capsules, VIGS optimization requires special consideration of tissue accessibility and developmental stage. Researchers have developed a robust protocol through orthogonal testing of three key factors: silencing target, inoculation approach, and developmental stage [84].

Delivery Methods Comparison:

• Peduncle injection: Limited viral movement into capsules
• Direct pericarp injection: Moderate efficiency but tissue damage concerns
• Pericarp cutting immersion: Highest efficiency (~93.94%) by maximizing Agrobacterium contact
• Fruit-bearing shoot infusion: Variable results depending on vascular connections [84]

Developmental Stage Optimization: Silencing efficiency varies significantly with capsule maturity:

• Early stages: Optimal for genes involved pigmentation (69.80% efficiency for CdCRY1)
• Mid stages: Best for structural genes (90.91% efficiency for CdLAC15)
• Late stages: Reduced efficiency due to lignification and reduced cell division [84]

Troubleshooting and Optimization Guidelines

Successful VIGS implementation requires careful optimization of several parameters. The table below summarizes key optimization factors for different plant systems:

Table 1: VIGS Optimization Parameters Across Plant Systems

Parameter	Soybean [85]	Camellia [84]	Vernicia [11]
Delivery Method	Cotyledon node immersion	Pericarp cutting immersion	Leaf infiltration
Optimal Duration	20-30 minutes immersion	15-20 minutes immersion	2-3 minutes vacuum infiltration
Agrobacterium OD₆₀₀	0.9-1.0	0.8-1.0	0.5-0.8
Temperature Post-Inoculation	22-25°C	20-22°C	22-25°C
Time to Phenotype	14-21 days	21-30 days	14-21 days
Silencing Efficiency	65-95%	70-94%	60-80%

Additional optimization considerations include:

• Fragment Selection: 200-300 bp fragments with <40% similarity to non-target genes to ensure specificity [84]
• Plant Growth Conditions: Maintain moderate light levels (100-150 μmol/m²/s) and humidity (60-70%) to reduce stress while allowing viral spread
• Controls: Always include empty vector controls and visual markers (e.g., PDS for photobleaching) to monitor silencing progression [85]

Protein Interaction Assays for NBS-LRR Signaling Mechanisms

Investigating NBS-LRR Signaling Networks

Protein-protein interactions (PPIs) form the foundation of NBS-LRR-mediated immune signaling. These interactions govern how NBS-LRR proteins recognize pathogen effectors, transition between activation states, and communicate with downstream signaling components [86] [1]. Traditional PPI assays often rely on overexpression systems that may not accurately reflect native complex formation in plant cells. Recent advances in endogenous tagging and live-cell imaging now enable researchers to investigate these interactions under more physiologically relevant conditions [86].

NBS-LRR proteins function as molecular switches within immune signaling networks. Their NBS domains bind and hydrolyze ATP, facilitating conformational changes that regulate activity [1]. The LRR domains are involved in both effector recognition and intramolecular interactions that maintain autoinhibition in the absence of pathogens [1]. The N-terminal domains (TIR, CC, or RPW8) determine interaction specificity with downstream signaling partners. For example, TIR domains typically initiate signaling pathways requiring EDS1 and SAG101, while CC domains more commonly interact with NRG1 [1].

Advanced Bioluminescence-Based Interaction Assays

Bioluminescence technologies have revolutionized PPI detection by providing sensitive, quantitative measurements in live cells. NanoLuc Binary Technology (NanoBiT) and Bioluminescence Resonance Energy Transfer (NanoBRET) represent state-of-the-art approaches for studying NBS-LRR complex formation and dynamics [86].

NanoBiT Methodology: This system splits the NanoLuc luciferase into two complementary fragments (LgBiT and SmBiT) that reconstitute a functional enzyme only when brought together by interacting proteins. For studying NBS-LRR interactions:

• Fusion tags are integrated at endogenous loci via CRISPR-mediated genome engineering
• This preserves native expression levels and regulation, critical for properly studying NBS-LRR function
• Interactions are quantified by adding cell-permeable furimazine substrate and measuring luminescence output [86]

NanoBRET Applications: This technique detects proximity between a NanoLuc-tagged protein and a fluorescently-labeled interaction partner through energy transfer:

• NanoLuc-tagged NBS-LRR proteins serve as energy donors
• HaloTag-fused candidate interactors labeled with fluorescent ligands act as acceptors
• BRET signals indicate interactions within <10 nm distance, suitable for detecting direct binding [86]

Experimental Workflow for Endogenous PPI Detection:

• Design CRISPR cassettes for C-terminal tagging at native genomic loci
• Generate stable transgenic plant lines expressing tagged proteins
• Validate expression and function of tagged proteins
• Treat with pathogen extracts or purified effectors to stimulate immune activation
• Measure luminescence (NanoBiT) or BRET ratios (NanoBRET) over time
• Quantify interaction dynamics and calculate binding affinities [86]

Technical Considerations for NBS-LRR Interaction Studies

Studying NBS-LRR interactions presents unique challenges due to their large size, low abundance, and rapid activation kinetics. The following table outlines key reagents and solutions for successful interaction assays:

Table 2: Research Reagent Solutions for Protein Interaction Studies

Reagent Category	Specific Examples	Function/Application	Technical Considerations
Luciferase Systems	NanoBiT, NanoBRET	Quantitative PPI detection in live cells	NanoBiT offers superior sensitivity; NanoBRET provides distance information
Tagging Systems	HaloTag, SNAP-tag, HALO	Protein labeling with synthetic ligands	Enable specific labeling with various fluorophores for multiplexing
CRISPR Components	Cas9 nucleases, sgRNAs, repair templates	Endogenous tagging	Preserve native regulation; requires careful validation
Luminescence Substrates	Furimazine, Coelenterazine-h	Bioluminescence generation	Furimazine offers improved stability and signal duration
Effector Proteins	Pathogen lysates, purified Avr proteins	NBS-LRR activation	Quality and concentration critically impact activation kinetics

Additional methodological considerations include:

• Expression Level Validation: Always verify that endogenously-tagged proteins are expressed at wild-type levels and remain functional
• Compartmentalization Analysis: NBS-LRR proteins localize to different cellular compartments (cytoplasm, nucleus, membranes); consider this in experimental design [4]
• Activation State Controls: Include both inactive (ADP-bound) and active (ATP-bound) forms when studying conformation-dependent interactions [1]

Integrated Workflows for Comprehensive NBS-LRR Characterization

Connecting VIGS with Protein Interaction Studies

Combining VIGS with protein interaction assays creates a powerful pipeline for comprehensive NBS-LRR characterization. This integrated approach enables researchers to connect gene function with mechanistic signaling pathways. A typical workflow begins with genome-wide identification of NBS-LRR candidates, proceeds to functional validation through VIGS, and culminates in mechanistic studies through interaction mapping [11].

The diagram below illustrates this integrated experimental workflow:

This systematic approach confirmed the role of specific NBS-LRR genes in disease resistance and identified their regulatory mechanisms, including WRKY transcription factor binding and promoter variations that explain differential resistance between species [11].

NBS-LRR Family Classification and Distribution Patterns

Understanding NBS-LRR domain architecture provides essential context for designing functional experiments. The table below summarizes the classification and distribution of NBS-LRR genes across various plant species, highlighting the diversity researchers encounter when designing validation studies:

Table 3: NBS-LRR Gene Family Distribution Across Plant Species

Plant Species	Total NBS-LRR Genes	CNL Subfamily	TNL Subfamily	RNL Subfamily	Atypical Forms	Reference
Nicotiana benthamiana	156	25	5	Not specified	126	[4]
Salvia miltiorrhiza	196	61	0	1	134	[7]
Vernicia montana	149	98*	12*	Not specified	39	[11]
Vernicia fordii	90	49	0	Not specified	41	[11]
Perilla citriodora	535	104	Not specified	1	430	[83]
Arabidopsis thaliana	~150	Majority	Minority	Present	~58	[1]
Oryza sativa	~500	All	0	0	Present	[1]

Note: Vernicia montana numbers include genes with CC domains (98) and TIR domains (12), with some containing both domains.

This distribution data reveals important patterns that inform experimental design. For example, researchers working with monocot species like rice should focus exclusively on CNL-type NBS-LRR genes, while those studying gymnosperms may encounter predominantly TNL-type genes [7]. Species like Salvia miltiorrhiza show remarkable degeneration of TNL subfamilies, suggesting lineage-specific evolutionary paths in immune gene content [7].

The integration of Virus-Induced Gene Silencing and advanced protein interaction assays provides a powerful toolkit for elucidating NBS-LRR gene function within the framework of domain architecture and classification. VIGS enables rapid functional screening of candidate genes across diverse plant species, including recalcitrant woody plants, while modern bioluminescence-based interaction assays offer unprecedented insight into the mechanistic underpinnings of immune signaling. Together, these approaches facilitate a comprehensive research pipeline from gene identification to functional validation and mechanistic characterization. As these technologies continue to evolve, particularly with improvements in CRISPR-mediated endogenous tagging and sensitive detection methodologies, researchers will be increasingly equipped to unravel the complex signaling networks that underpin plant immunity, ultimately enabling the development of crops with enhanced disease resistance.

In the field of plant genomics, the discovery and characterization of disease resistance (R) genes represent a critical research area with significant implications for agricultural sustainability and food security. The NBS-LRR gene family, encoding proteins with nucleotide-binding site and leucine-rich repeat domains, constitutes the largest and most important class of plant resistance genes, accounting for approximately 60% of all characterized R genes in plant species [3] [41]. These genes enable plants to recognize pathogen-derived effectors and initiate robust immune responses, ultimately leading to pathogen restriction through mechanisms such as the hypersensitive response [4] [41]. The comprehensive identification and classification of NBS-LRR genes across plant species have been dramatically accelerated by computational biology approaches, particularly through cross-species comparative genomics frameworks that leverage synteny and orthogroup analyses [41].

The integration of comparative genomics with specialized computational tools has enabled researchers to move beyond single-reference genome analyses to a pan-genome perspective that captures the full diversity of R genes across multiple genotypes and species [87]. This review provides an in-depth technical examination of the conceptual frameworks, methodologies, and tools for conducting synteny and orthogroup analysis specifically within the context of NBS-LRR gene research, offering both foundational principles and practical implementation guidance for scientists investigating plant immunity mechanisms.

Conceptual Foundations: Synteny, Orthology, and Paralogy

Defining Core Concepts in Comparative Genomics

The interpretation of cross-species genomic comparisons requires precise understanding of key terminology that describes evolutionary relationships between genes and genomic regions:

• Orthologs: Genes in different species that evolved from a common ancestral gene by speciation, typically retaining similar functions [88]. Orthology forms the basis for most functional inferences in comparative genomics.
• Paralogs: Genes related by duplication within a genome, which often evolve new functions [88]. Paralogy can complicate comparative analyses when recent duplications exist.
• Synteny: Literally "same thread," referring to two or more genes located on the same chromosome within a species [88].
• Conserved Synteny: When orthologs of syntenic genes in one species are also located on a single chromosome in a second species, regardless of gene order [88].
• Conserved Segments: Genomic intervals where the order of multiple orthologous genes is preserved between species, also called "conserved linkages" [88].
• Orthogroup: A set of genes across multiple genomes derived from a single ancestral gene, encompassing both orthologs and paralogs [87].

Table 1: Evolutionary Relationships in Comparative Genomics

Term	Definition	Functional Significance
Orthologs	Genes in different species derived from common ancestor through speciation	Often retain similar biological functions; basis for functional inference
Paralogs	Genes related by duplication within a genome	May evolve new functions (neofunctionalization) or partition ancestral functions (subfunctionalization)
Homeologs	Paralogs derived from whole-genome duplication events	Common in polyploid plants; may exhibit subgenome dominance
Orthogroup	Set of genes across genomes from single ancestral gene	Provides evolutionary context for gene family expansion and contraction

Evolutionary Conservation and Functional Constraint

The fundamental premise underlying comparative genomics is that functional sequences tend to evolve at slower rates than nonfunctional sequences due to selective constraints [88]. This principle enables researchers to distinguish functionally important genomic elements, including protein-coding genes and regulatory sequences, through their conservation across evolutionary time. The phylogenetic distance between compared species determines the type of functional elements that can be identified:

• Distant comparisons (e.g., humans-pufferfish, ~450 million years divergence) primarily reveal coding sequences under strong purifying selection [88].
• Intermediate comparisons (e.g., humans-mice, ~40-80 million years divergence) identify both coding sequences and functional noncoding elements [88].
• Close comparisons (e.g., humans-chimpanzees) highlight recently evolved sequences that may underlie species-specific traits [88].

In plant NBS-LRR gene research, comparisons across varying evolutionary distances have revealed dramatic differences in gene family size and composition. For example, genome-wide analyses have identified 73 NBS genes in Akebia trifoliata [10], 252 in pepper (Capsicum annuum) [2], 603 in Nicotiana tabacum [3], and up to 2,151 in wheat (Triticum aestivum) [3], reflecting both biological differences and methodological approaches.

Computational Methodologies for Synteny and Orthogroup Analysis

Foundational Workflows for Comparative Genomics

The integration of synteny and orthogroup analysis follows a systematic workflow that progresses from data acquisition through multiple processing stages to biological interpretation. The following diagram illustrates a generalized pipeline for cross-species comparative genomics with emphasis on NBS-LRR gene discovery:

Genome-Wide Identification of NBS-LRR Genes

The accurate identification of NBS-LRR genes across multiple genomes represents the foundational step for subsequent comparative analyses. The standard methodology employs hidden Markov models (HMMs) based on the NB-ARC domain (PF00931) from the Pfam database, followed by rigorous domain architecture characterization [3] [2] [4].

Experimental Protocol: NBS-LRR Identification Pipeline

• HMM Search Implementation

  • Retrieve the NB-ARC (PF00931) HMM profile from the Pfam database
  • Perform HMMER searches (v3.1b2 or later) against target proteomes with E-value cutoff < 1×10⁻²⁰ [3]
  • Extract sequences and remove redundant entries

• Domain Architecture Characterization

  • Scan candidate sequences against Pfam databases for TIR (PF01582), RPW8, and LRR domains
  • Identify coiled-coil (CC) domains using NCBI Conserved Domain Database or Coiledcoil with threshold 0.5 [10]
  • Classify genes into structural subfamilies: CNL, TNL, RNL, CN, TN, N, NL based on domain composition [4]

• Manual Curation and Validation

  • Verify complete presence of NBS domains with E-values < 0.01
  • Remove pseudogenes and incomplete sequences
  • Cross-validate with known R-genes from specialized databases

Table 2: NBS-LRR Gene Subfamily Classification Based on Domain Architecture

Subfamily	N-Terminal Domain	Central Domain	C-Terminal Domain	Representative Species Count
CNL	Coiled-coil (CC)	NBS (NB-ARC)	LRR	150 in N. tabacum [3]
TNL	TIR	NBS (NB-ARC)	LRR	64 in N. tabacum [3]
RNL	RPW8	NBS (NB-ARC)	LRR	4 in N. benthamiana [4]
NL	None	NBS (NB-ARC)	LRR	23 in N. benthamiana [4]
CN	Coiled-coil (CC)	NBS (NB-ARC)	None	41 in N. benthamiana [4]
N	None	NBS (NB-ARC)	None	60 in N. benthamiana [4]

Orthogroup Inference with OrthoFinder

OrthoFinder applies a graph-based algorithm to infer orthogroups across multiple genomes, providing a fundamental organization of genes into hierarchical groups descended from a single ancestral gene in the last common ancestor [87]. The algorithm employs the following methodology:

• Sequence Similarity Search

  • Perform all-versus-all BLASTP of protein sequences across genomes
  • Apply conservative E-value cutoff (typically 1×10⁻⁵)
  • Generate pairwise similarity scores

• Orthogroup Inference

  • Construct graph where vertices represent genes and edges represent sequence similarities
  • Apply Markov Cluster Algorithm (MCL) to partition graph into orthogroups
  • Resolve complex many-to-many relationships across species

• Gene Tree Reconciliation

  • Infer gene trees for each orthogroup
  • Reconcile with species tree to distinguish orthologs from paralogs
  • Root trees to determine evolutionary relationships

For NBS-LRR genes, which often exhibit complex evolutionary patterns including tandem duplications and species-specific expansions, OrthoFinder provides crucial evolutionary context for interpreting functional conservation and divergence.

Synteny Detection with MCScanX and GENESPACE

Synteny detection algorithms identify regions of conserved gene order across genomes, providing critical evidence for orthology assignment and revealing evolutionary rearrangements. MCScanX remains a widely used algorithm, while GENESPACE represents a more recent integration of synteny with orthogroup information [87] [89].

Experimental Protocol: Synteny Block Identification

• Input Data Preparation

  • Generate BLASTP all-against-all results for target genomes
  • Prepare GFF/GTF annotation files with gene positions
  • Format data for specific synteny tool requirements

• MCScanX Implementation

  • Run MCScanX with parameters: MATCHSCORE, MATCHSIZE, GAPPENALTY, OVERLAPWINDOW
  • Set minimum anchor pairs (typically 5) to define syntenic blocks
  • Apply gap penalty to control maximum distance between anchors

• GENESPACE Workflow

  • Run OrthoFinder to establish orthogroups
  • Generate synteny blocks using only "potential anchor" genes within same orthogroup
  • Condense tandem arrays to single representatives to minimize copy number variation effects
  • Create pan-genome annotation integrating syntenic orthology networks

• Visualization and Interpretation

  • Generate dot plots and chromosomal maps using built-in visualization tools
  • Identify systemic regions and rearrangements
  • Correlate synteny breaks with structural variants

The integration offered by GENESPACE is particularly valuable for complex plant genomes with whole-genome duplications, as it resolves the circular problem where "a priori knowledge of gene copy number is needed to effectively infer orthology and synteny, yet measures of synteny and orthology are needed to infer copy number between a pair of sequences" [87].

Table 3: Essential Computational Tools for Synteny and Orthogroup Analysis

Tool/Resource	Primary Function	Application in NBS-LRR Research	Access Information
OrthoFinder	Orthogroup inference across multiple genomes	Evolutionary classification of NBS-LRR gene families	https://github.com/davidemms/OrthoFinder
MCScanX	Synteny detection and visualization	Identification of conserved NBS-LRR gene clusters	http://chibba.pgml.uga.edu/mcscan2/
GENESPACE	Integrative synteny and orthology	Pan-genome annotation of R-genes across cultivars/species	https://github.com/jtlovell/GENESPACE
JCVI Library	Comparative genomics utilities	Synteny visualization, graphics generation	https://github.com/tanghaibao/jcvi
HMMER	Domain identification using hidden Markov models	NB-ARC (PF00931) domain detection	http://hmmer.org/
Pfam Database	Protein family domain databases	Curated HMM profiles for NBS, TIR, LRR domains	http://pfam.xfam.org/
PRGminer	Deep learning-based R-gene prediction	Identification and classification of novel R-genes	https://github.com/usubioinfo/PRGminer

Case Study: NBS-LRR Gene Family Evolution in Nicotiana Species

A recent genome-wide analysis of NBS-LRR genes in three Nicotiana species (N. tabacum, N. sylvestris, and N. tomentosiformis) provides an illustrative example of integrated synteny and orthogroup analysis [3]. This study identified 1,226 NBS genes across the three genomes, with 603 in allotetraploid N. tabacum, approximately matching the combined total (623) of its parental species [3].

Methodology and Workflow Implementation

The research employed a comprehensive analytical pipeline:

• Genome-Wide Identification: HMMER searches with PF00931 followed by domain verification via NCBI CDD
• Phylogenetic Analysis: Multiple sequence alignment with MUSCLE and maximum-likelihood trees with MEGA11
• Duplication Analysis: Self-BLASTP and MCScanX for segmental and tandem duplication identification
• Synteny Analysis: Reciprocal BLASTP and MCScanX for collinearity detection
• Selection Pressure Analysis: Ka/Ks calculations using KaKs_Calculator 2.0

Key Findings and Biological Insights

The analysis revealed that 76.62% of NBS genes in N. tabacum could be traced to their parental genomes, demonstrating the power of synteny-based orthology assignment [3]. Furthermore, whole-genome duplication contributed significantly to NBS gene family expansion, with selection pressure analyses indicating purifying selection as the dominant evolutionary force [3]. This case study exemplifies how integrated comparative genomics approaches can decipher the evolutionary history of complex gene families and identify candidate genes for functional characterization.

Advanced Applications and Future Directions

Machine Learning and Deep Learning Approaches

Traditional similarity-based methods for R-gene identification face limitations when homology is low, particularly for newly sequenced species [6] [41]. Recent advances incorporate machine learning (ML) and deep learning (DL) to overcome these challenges:

• PRGminer: A deep learning-based tool that achieves 98.75% accuracy in R-gene prediction using dipeptide composition features [6]
• Convolutional Neural Networks (CNNs): Capture hierarchical patterns in protein sequences for improved classification [41]
• Multi-Layer Perceptrons (MLPs): Model complex sequence features beyond domain architecture [41]

These approaches represent a paradigm shift from alignment-dependent to alignment-free R-gene identification, particularly valuable for detecting rapidly evolving NBS-LRR genes with atypical domain architectures.

Pan-Genome Framework for R-Gene Discovery

The integration of synteny and orthogroup analysis enables a pan-genome perspective that transcends single-reference limitations. The GENESPACE approach creates "pan-genome annotations" that positionally anchor orthologs and paralogs across multiple genomes, facilitating the identification of presence-absence variation (PAV) and copy-number variation (CNV) in NBS-LRR genes [87]. This framework is particularly powerful for crop improvement programs, as it enables researchers to "examine all putatively functional variants within a genomic region of interest, even those in genes that are absent in the focal reference genome" [87].

Synteny and orthogroup analysis provide complementary and mutually reinforcing frameworks for comparative genomics research on NBS-LRR disease resistance genes. The integration of these approaches through tools like GENESPACE represents a significant methodological advance, particularly for complex plant genomes with abundant duplication and rearrangement events. As sequencing technologies continue to produce increasingly contiguous genome assemblies, and computational methods incorporate more sophisticated machine learning approaches, the research community is positioned to make accelerated progress in understanding the evolution and function of plant immune genes. These advances will directly support crop improvement programs through the identification of durable disease resistance genes that can be deployed in sustainable agricultural systems.

Nucleotide-binding site (NBS) genes represent the largest class of plant disease resistance (R) genes and are vital components of the plant immune system, enabling responses to both biotic and abiotic stresses. These genes encode proteins characterized by a conserved NBS domain and frequently C-terminal leucine-rich repeats (LRRs), forming the NBS-LRR family. The specific domain architecture of these proteins facilitates pathogen recognition and signal transduction, triggering robust defense mechanisms [50] [21]. The domain composition serves as the primary basis for classifying NBS genes into distinct subfamilies, including TIR-NBS-LRR (TNL), CC-NBS-LRR (CNL), and RPW8-NBS-LRR (RNL), with TNL and CNL proteins primarily responsible for recognizing specific pathogens [21].

Plant immunity involves a complex network of signaling pathways. Plants possess a two-tiered innate immune system. The first layer, pathogen-associated molecular pattern-triggered immunity (PTI), is activated upon recognition of conserved microbial patterns. The second layer, effector-triggered immunity (ETI), is initiated when specific R proteins, predominantly from the NBS-LRR family, directly or indirectly recognize pathogen effector proteins [50]. This recognition often induces a hypersensitive response (HR), limiting pathogen spread. Emerging research underscores that NBS-LRR genes are integral not only to biotic stress responses but also to abiotic stress adaptation, indicating a sophisticated crosstalk between different stress signaling pathways [90] [91]. This technical guide synthesizes current knowledge on NBS gene classification, their expression patterns under stress, and the experimental frameworks used to profile them, providing a resource for researchers and drug development professionals working on plant immunity.

Domain Architecture and Classification of NBS Genes

Conserved Domain Composition and Phylogenetic Classification

The classification of NBS-LRR genes is fundamentally grounded in their domain architecture. The central NB-ARC domain (Nucleotide-Binding Adaptor Shared by APAF-1, R proteins, and CED-4) is a conserved molecular switch that binds ATP/GTP and is essential for signal transduction [50] [21]. The C-terminal leucine-rich repeat (LRR) domain is involved in protein-protein interactions and is responsible for specific pathogen effector recognition [3]. Variations in the N-terminal domain define the major subfamilies:

• TNL (TIR-NBS-LRR): Characterized by an N-terminal Toll/Interleukin-1 Receptor (TIR) domain. Prevalent in dicots like Arabidopsis thaliana but absent in most monocots [50] [21].
• CNL (CC-NBS-LRR): Features an N-terminal coiled-coil (CC) domain. This is the most widespread subfamily found in both dicots and monocots [50] [3].
• RNL (RPW8-NBS-LRR): Contains an N-terminal Resistance to Powdery Mildew 8 (RPW8) domain. This subfamily is involved in downstream defense signaling and is subdivided into NRG1 and ADR1 lineages [21].

Additionally, many partial or incomplete NBS genes exist, lacking the LRR domain (e.g., TIR-NBS (TN) or CC-NBS (CN)) or other domains, yet still often retaining functionality in defense responses [90].

Table 1: NBS-LRR Gene Family Classification Based on Domain Architecture

Subfamily	N-Terminal Domain	Central Domain	C-Terminal Domain	Prevalence	Proposed Primary Function
TNL	TIR (Toll/Interleukin-1 Receptor)	NBS (NB-ARC)	LRR (Leucine-Rich Repeat)	Common in dicots, rare in monocots	Specific pathogen recognition
CNL	CC (Coiled-Coil)	NBS (NB-ARC)	LRR (Leucine-Rich Repeat)	Ubiquitous in angiosperms	Specific pathogen recognition
RNL	RPW8	NBS (NB-ARC)	LRR (Leucine-Rich Repeat)	All angiosperms	Downstream signal transduction
TN	TIR	NBS	—	Varies by species	Defense signaling, often incomplete
CN	CC	NBS	—	Varies by species	Defense signaling, often incomplete

Genomic Distribution and Evolution

NBS genes are often distributed unevenly across chromosomes, frequently clustered at telomeric regions, and have expanded primarily through local gene duplication events, including tandem and segmental duplications [3] [21]. For instance, 76.62% of NBS genes in the allotetraploid Nicotiana tabacum could be traced back to its parental genomes (N. sylvestris and N. tomentosiformis), with whole-genome duplication significantly contributing to the family's expansion [3]. This dynamic evolution leads to considerable variation in NBS gene number across plant species, from 73 in Akebia trifoliata [21] to 603 in Nicotiana tabacum [3], enabling a vast repertoire for pathogen recognition.

Expression Profiling of NBS Genes Under Biotic Stress

Biotic stresses, such as infections by fungi, bacteria, and viruses, trigger specific and rapid changes in the expression of NBS genes. Profiling these expression patterns is key to identifying functional R genes.

Response to Fungal Pathogens

Studies across multiple species have identified specific NBS genes activated in response to fungal challenges.

• In Brassica oleracea (cabbage), transcriptome analysis identified 17 TNL genes from heat-shock treated plants. Subsequent qRT-PCR validation revealed that eight of these genes showed significant responses to Fusarium oxysporum infection. Three genes (Bol007132, Bol016084, and Bol030522) exhibited dramatically higher expression in a F. oxysporum-resistant line compared to intermediate and susceptible lines, and were physically linked to known F. oxysporum resistance gene clusters [90].
• In Dendrobium officinale, a transcriptome analysis of salicylic acid (SA) treatment, a key hormone in defense signaling, identified 1,677 differentially expressed genes (DEGs). Among them, six NBS-LRR genes (Dof013264, Dof020566, Dof019188, Dof019191, Dof020138, and Dof020707) were significantly up-regulated. Furthermore, weighted gene co-expression network analysis (WGCNA) indicated that Dof020138 was closely associated with pathogen identification pathways, MAPK signaling pathways, and plant hormone signal transduction, suggesting a central role in coordinating immune responses [50].
• Research in cotton has confirmed that silencing an NBS-LRR gene reduces resistance to Verticillium dahlia [3], directly linking the expression of this gene family to functional disease resistance.

Response to Bacterial and Viral Pathogens

The role of NBS-LRR genes extends beyond fungal defense. Heterologous expression of a maize CNL gene in Arabidopsis thaliana improved resistance to Pseudomonas syringae [3]. Similarly, a soybean TNL gene was shown to confer broad-spectrum resistance to viral pathogens when overexpressed in soybean [3]. These findings highlight the potential of engineering NBS genes to enhance resistance across plant species and against diverse pathogen types.

Table 2: Experimentally Validated NBS Genes and Their Responses to Stress

Gene Identifier	Plant Species	Stress Condition	Expression Response	Proposed Function	Experimental Method
Bol007132	Brassica oleracea	Fusarium oxysporum	Up-regulated in resistant line	Fungal disease resistance	RNA-seq, qRT-PCR [90]
Bol016084	Brassica oleracea	Fusarium oxysporum	Up-regulated in resistant line	Fungal disease resistance	RNA-seq, qRT-PCR [90]
Bol030522	Brassica oleracea	Fusarium oxysporum	Up-regulated in resistant line	Fungal disease resistance	RNA-seq, qRT-PCR [90]
Dof020138	Dendrobium officinale	Salicylic Acid treatment	Significantly up-regulated	ETI system, signal transduction	RNA-seq, WGCNA [50]
Dof013264	Dendrobium officinale	Salicylic Acid treatment	Significantly up-regulated	ETI system	RNA-seq [50]
Various NBS genes	Nicotiana tabacum	Black shank, Bacterial wilt	Differential expression	Disease resistance	RNA-seq [3]

Expression Profiling of NBS Genes Under Abiotic Stress

While traditionally associated with biotic stress, compelling evidence now links NBS genes to abiotic stress responses, including heat, drought, and salinity.

The most direct evidence comes from a study on Brassica oleracea, where the same set of TNL genes that responded to Fusarium oxysporum were also significantly up-regulated by heat shock treatment [90]. This suggests that certain NBS genes are involved in a convergent signaling pathway that manages combined stress scenarios, which commonly occur in field conditions. High temperature stress can attenuate plant disease resistance while promoting pathogen growth, and the abundance of some R proteins, like barley MLA1 and MLA6, decreases dramatically at high temperatures [90]. This indicates a molecular point of vulnerability where abiotic stress can compromise biotic defense, further underscoring the importance of crosstalk.

Furthermore, the broader context of plant stress signaling involves extensive hormonal crosstalk. Salicylic acid (SA), jasmonic acid (JA), and ethylene (ET) pathways interact synergistically or antagonistically to fine-tune defense responses [91]. The up-regulation of Dendrobium NBS-LRR genes by SA treatment [50] explicitly connects this hormone to NBS-mediated immunity. Abscisic acid (ABA), a central hormone in abiotic stress adaptation, can modulate JA and SA signaling pathways, creating a complex network that integrates signals from both biotic and abiotic environments [91].

Experimental Protocols for Profiling NBS Gene Expression

A robust methodological pipeline is essential for the accurate identification and expression profiling of NBS genes.

Genome-Wide Identification of NBS Genes

Procedure:

• Data Acquisition: Obtain the complete genome assembly and annotated protein sequence file for the target species from public databases like NCBI.
• HMMER Search: Use HMMER software (e.g., v3.1b2) with the hidden Markov model (HMM) profile for the NB-ARC domain (PF00931) from the Pfam database to perform a genome-wide search. The E-value threshold is typically set to 1.0 [3] [21].
• Domain Verification: Confirm the presence of identified domains in the candidate sequences using the NCBI Conserved Domain Database (CDD) and Pfam. Identify TIR (PF01582), RPW8 (PF05659), LRR (PF08191, PF00560, PF07723, PF07725, PF12779, etc.), and CC domains (using tools like CoiledCoil with a threshold of 0.5) [3] [21].
• Classification and Non-redundancy: Classify genes into subfamilies based on their domain architecture and remove redundant sequences to generate a final non-redundant set of NBS genes.

Transcriptome Analysis using RNA-Seq

Procedure:

• Experimental Design and RNA Extraction: Subject plants to stress treatments (e.g., pathogen inoculation, heat, hormone application) with appropriate controls. Collect tissue samples at multiple time points, immediately freeze in liquid nitrogen, and extract total RNA using a standard method like TRIzol reagent [90].
• Library Preparation and Sequencing: Assess RNA quality, prepare sequencing libraries (e.g., using Illumina kits), and perform high-throughput sequencing on an appropriate platform.
• Bioinformatic Analysis:
  • Quality Control and Mapping: Process raw reads with tools like Trimmomatic to remove adapters and low-quality bases. Map the cleaned reads to the reference genome using a splice-aware aligner like HISAT2 [3].
  • Quantification and Differential Expression: Calculate gene expression levels (e.g., FPKM or TPM) using tools like Cufflinks [3]. Identify differentially expressed genes (DEGs) between treatment and control groups using software like Cuffdiff or DESeq2, applying appropriate statistical thresholds (e.g., FDR-adjusted p-value < 0.05, |log2(Fold Change)| > 1) [50] [90].
  • Co-expression Analysis: Perform Weighted Gene Co-expression Network Analysis (WGCNA) to identify clusters of highly correlated genes (modules) and link them to specific stress treatments or pathways [50].

Validation by Quantitative RT-PCR (qRT-PCR)

Procedure:

• cDNA Synthesis: Digest residual genomic DNA and synthesize first-strand cDNA from total RNA using a reverse transcription kit with oligo(dT) and/or random primers.
• Primer Design: Design gene-specific primers for selected target NBS genes and reference housekeeping genes (e.g., Actin, Ubiquitin) using software like Primer3.
• Amplification and Quantification: Perform qRT-PCR reactions in triplicate on a real-time PCR detection system using a SYBR Green master mix.
• Data Analysis: Calculate relative gene expression levels using the comparative 2^(-ΔΔCt) method, normalizing to the reference genes and the control sample [90].

Diagram Title: Experimental Workflow for Profiling NBS Gene Expression

The Scientist's Toolkit: Key Research Reagents and Solutions

Table 3: Essential Reagents and Resources for NBS Gene Research

Reagent/Resource	Function/Application	Example Details/Specifications
HMM Profile (PF00931)	Bioinformatics identification of NB-ARC domain in protein sequences.	Used with HMMER software; E-value threshold of 1.0 [3] [21].
NCBI CDD & Pfam	Verification and annotation of conserved protein domains (TIR, LRR, CC, RPW8).	Critical for accurate classification of NBS genes into subfamilies [3] [21].
TRIzol Reagent	Monophasic solution of phenol and guanidine isothiocyanate for total RNA isolation.	Maintains RNA integrity from plant tissues; used for RNA-seq and qRT-PCR [90].
RNA-seq Library Prep Kits	Preparation of cDNA libraries for high-throughput sequencing.	Illumina TruSeq is a common choice; compatibility with the sequencer is key.
HISAT2, Cufflinks, DESeq2	Bioinformatics software for read alignment, expression quantification, and differential expression analysis.	Standard tools for RNA-seq data analysis [3].
SYBR Green qRT-PCR Master Mix	Fluorescent dye for quantifying DNA amplification during qRT-PCR.	Enables sensitive and specific validation of RNA-seq results for target genes [90].
Salicylic Acid (SA)	Plant hormone elicitor used to simulate biotic stress and study defense signaling pathways.	Used in treatment experiments to activate NBS-LRR gene expression [50].

Integrated Signaling Pathways in Plant Stress Response

NBS genes function within a complex intracellular signaling network that integrates signals from both biotic and abiotic stresses. The following diagram summarizes this interplay and the central role of NBS-LRR proteins in the plant immune response.

Diagram Title: NBS-LRR Genes in Plant Stress Signaling Pathways

As illustrated, NBS-LRR proteins are central to Effector-Triggered Immunity (ETI). They recognize specific pathogen effectors, leading to a robust defense output that includes the hypersensitive response (HR), systemic acquired resistance (SAR), production of reactive oxygen species (ROS), and modulation of phytohormone signaling. This ETI response is interconnected with the broader hormone signaling network (involving SA, JA, ET, and ABA), which also receives inputs from abiotic stresses, allowing for integrated adaptation to complex environmental challenges [50] [90] [91].

The systematic profiling of NBS gene expression patterns provides critical insights into the molecular basis of plant stress resilience. The domain architecture of NBS proteins is a primary determinant of their function within the plant immune system. Evidence from species like Brassica oleracea and Dendrobium officinale clearly links specific NBS genes to defense against fungal pathogens and highlights their involvement in response to abiotic stresses like heat. The experimental framework—combining genome-wide identification, transcriptomics, and functional validation—enables researchers to pinpoint key candidate genes. As research progresses, the integration of genomic data with advanced molecular techniques will be pivotal for unraveling the intricate crosstalk between stress signaling pathways and for leveraging NBS genes in the development of next-generation stress-resistant crops.

This whitepaper explores the evolutionary dynamics of plant nucleotide-binding site-leucine-rich repeat (NBS-LRR) disease resistance genes, focusing on the interplay between birth-and-death models and diversifying selection within LRR domains. The NBS-LRR gene family represents one of the largest and most dynamic classes of resistance (R) genes, exhibiting remarkable diversification through gene duplication, loss, and selective processes. We examine how birth-and-death evolution drives the expansion and contraction of R-gene clusters, while diversifying selection acts predominantly on LRR regions to generate novel pathogen recognition specificities. Within the broader context of domain architecture and classification of NBS disease resistance genes, this review synthesizes current understanding of evolutionary mechanisms that maintain genetic diversity for plant immunity, highlighting implications for crop improvement and sustainable agriculture.

Plant disease resistance genes (R-genes) encode proteins that detect pathogen effectors and initiate robust immune responses. Among these, nucleotide-binding site-leucine-rich repeat (NBS-LRR) genes constitute the largest and most extensively studied family, with over 300 cloned R-genes belonging to this class [41]. The domain architecture of NBS-LRR genes typically includes a conserved NBS domain (NB-ARC, PF00931) and C-terminal LRR repeats, with variable N-terminal domains such as TIR (Toll/interleukin-1 receptor) or CC (coiled-coil) that define major subfamilies (TNLs and CNLs) [32] [3].

The evolutionary dynamics of this gene family are governed by two primary mechanisms: birth-and-death evolution, which describes the continuous processes of gene duplication and loss, and diversifying selection, which promotes amino acid variation in specific protein regions. These processes are particularly active in the LRR (leucine-rich repeat) domains, which are directly involved in pathogen recognition [92] [93]. Understanding these evolutionary forces is essential for elucidating how plants maintain diverse recognition capacities against rapidly evolving pathogens.

Birth-and-Death Evolution of NBS-LRR Genes

Conceptual Framework and Mechanisms

The birth-and-death model of evolution proposes that new disease resistance genes are created through gene duplication, while defeated or non-functional genes are progressively lost from the genome [94]. This model predicts that R-gene clusters undergo continuous turnover, with new specificities emerging through duplication events and genetic exchange, while maintaining a core set of functional genes.

Key genetic mechanisms driving birth-and-death evolution include:

• Tandem duplications: Localized gene duplications that create clustered arrays of similar genes
• Segmental duplications: Larger-scale chromosomal duplications transferring multiple genes
• Whole-genome duplication (WGD): Polyploidization events that duplicate entire genomes
• Unequal crossing-over: Recombination events between misaligned homologous sequences
• Gene conversion: Non-reciprocal genetic transfer between similar sequences

These mechanisms collectively generate the raw material for evolutionary innovation in plant immune systems, creating genetic novelty that can be selected for improved pathogen recognition.

Genomic Evidence Across Plant Lineages

Comparative genomic analyses across diverse plant species reveal dramatic variation in NBS-LRR gene copy numbers, reflecting lineage-specific birth-and-death dynamics:

Table 1: NBS-LRR Gene Family Size Variation Across Plant Species

Plant Species	Genome Type	NBS-LRR Count	Key Evolutionary Features	Reference
Arabidopsis thaliana	Diploid	~200	Reference genome with well-annotated NLRs	[41]
Oryza sativa (rice)	Diploid	>500	Chromosome 11 enrichment with R-gene clusters	[94]
Nicotiana tabacum	Allotetraploid	603	Combination of parental genomes with retention	[3]
Triticum aestivum (wheat)	Hexaploid	2,151	Massive expansion through polyploidization	[32]
Arachis hypogaea (peanut)	Allotetraploid	713	Genetic exchange between subgenomes	[93]
Physcomitrella patens	Bryophyte	~25	Small repertoire representing ancestral state	[32]

The expansion of NBS-LRR genes in angiosperms is particularly striking when compared to non-vascular plants. While bryophytes like Physcomitrella patens maintain only about 25 NBS-LRR genes, flowering plants often possess hundreds to thousands of these genes [32]. This expansion appears to correlate with increasing pathogen pressure and complexity in terrestrial environments.

In cultivated rice (Oryza sativa), comparative analysis of R-gene clusters on chromosome 11 between cultivated varieties and their wild ancestors revealed that cultivated species contain significantly more NBS-LRR genes (53 in indica cultivar Kasalath) compared to their wild progenitors [94]. This suggests that agricultural selection may have favored the retention of duplicated R-genes, potentially enhancing disease resistance in cultivated environments.

Impact of Polyploidization on Gene Family Expansion

Whole-genome duplication (WGD) events have played a significant role in the expansion of NBS-LRR gene families, particularly in polyploid crops. Allotetraploid species such as Nicotiana tabacum (2n=4x=48) exemplify this phenomenon, with 603 NBS genes representing approximately the combined total of its diploid progenitors (N. sylvestris: 344; N. tomentosiformis: 279) [3]. Similarly, cultivated peanut (Arachis hypogaea) possesses 713 full-length NBS-LRR genes, compared to 278 and 303 in its diploid ancestors A. duranensis and A. ipaensis, respectively [93].

Following polyploidization, these gene families undergo a process of diploidization, including gene loss, subfunctionalization, and the emergence of novel genetic combinations. In A. hypogaea, researchers observed sequences containing both TIR and CC domains—a combination not found in either diploid progenitor—suggesting that genetic exchange or gene rearrangement likely resulted in domain fusion after tetraploidization [93].

Diversifying Selection in LRR Regions

Molecular Signatures of Selection

Diversifying selection (also termed positive selection) describes evolutionary processes that favor novel genetic variants, leading to increased diversity at the molecular level. In NBS-LRR genes, this selection predominantly targets the leucine-rich repeat (LRR) domains, which are directly involved in pathogen recognition [92] [93].

The primary method for detecting diversifying selection involves comparing the rates of non-synonymous (dN) and synonymous (dS) nucleotide substitutions:

• dN/dS (ω) > 1: Indicates positive selection, where amino acid changes are favored
• dN/dS (ω) = 1: Suggests neutral evolution
• dN/dS (ω) < 1: Signifies purifying selection, which removes deleterious mutations

Genome-wide studies have demonstrated that LRR domains consistently show higher dN/dS ratios compared to the more conserved NBS domains, with approximately 50% of lineage-specific expanded LRR-RLK genes showing signatures of positive selection [92]. This pattern reflects the evolutionary arms race between plants and their pathogens, where changing recognition interfaces provides selective advantages.

Structural and Functional Implications

The LRR domain typically consists of multiple repeats of 20-30 amino acids that form solenoid structures ideal for protein-protein interactions. Four specific amino acid positions within these repeats show particularly strong signatures of positive selection, suggesting they constitute critical determinants of recognition specificity [92].

This selective pattern creates a paradox: how do plants maintain integrated signaling function while allowing extensive variability in recognition domains? The solution appears to lie in the modular architecture of NBS-LRR proteins, where the conserved NBS domain provides standardized signaling output, while the variable LRR domains provide customizable recognition inputs.

Table 2: Selection Patterns Across NBS-LRR Protein Domains

Protein Domain	Selection Pattern	Biological Function	Evolutionary Constraint
TIR/CC (N-terminal)	Purifying to neutral	Signaling initiation; oligomerization	Moderate
NBS (NB-ARC)	Strong purifying selection	Nucleotide binding; molecular switch	High
LRR (C-terminal)	Diversifying selection	Pathogen recognition; specificity determination	Low
Linker Regions	Variable	Inter-domain communication; regulation	Variable

In cultivated peanut, researchers observed that although relaxed selection acted on both NBS-LRR proteins and LRR domains, LRR domains were preferentially lost compared to diploid progenitors, potentially explaining the lower disease resistance of cultivated varieties [93]. This suggests that artificial selection during domestication may have differentially impacted different protein domains.

Lineage-Specific Selection Patterns

Comparative analyses reveal that selection pressures vary substantially between plant lineages and ecological contexts. In a comprehensive study of 7,554 LRR-RLK genes from 31 flowering plant genomes, researchers found that lineage-specific expanded (LSE) copies were predominantly found in subgroups involved in environmental interactions and showed significantly more indications of positive selection or relaxed constraint than single-copy genes [92].

This pattern is particularly pronounced in wild species compared to their cultivated relatives. For example, in Arachis species, cultivated peanut showed both LRR domain loss and production of young NBS-LRR genes after tetraploidization, with 113 NBS-LRRs associated with disease resistance quantitative trait loci (QTL) classified as 75 young and 38 old NBS-LRRs [93]. This suggests that recent gene duplicates may be particularly important for adapting to new pathogen pressures.

Integrated Evolutionary Model: Birth-and-Death Under Diversifying Selection

The interplay between birth-and-death evolution and diversifying selection creates a dynamic evolutionary system that maintains diversity in plant immune receptors. Gene duplication through birth-and-death processes generates genetic raw material, while diversifying selection fine-tunes recognition specificities, particularly in LRR domains.

This integrated model helps explain several key observations in R-gene evolution:

• Cluster heterogeneity: R-gene clusters show extreme variation in copy number and sequence diversity between closely related species
• Differential selection pressures: Various protein domains experience distinct selection regimes reflecting their functional roles
• Lineage-specific adaptations: Plant families facing different pathogen pressures develop unique R-gene repertoires
• Balanced polymorphisms: Functional and non-functional alleles may coexist in populations through frequency-dependent selection

The model further predicts that genes involved in environmental interactions will show higher turnover rates and stronger positive selection—a pattern consistently observed in empirical studies [92] [94].

Experimental Approaches and Methodologies

Genome-Wide Identification and Classification

Standardized pipelines for NBS-LRR gene identification enable comparative evolutionary analyses across species:

HMMER-based Domain Identification

Orthogroup Analysis with OrthoFinder

Evolutionary Analyses

Selection Pressure Analysis with CodeML

Gene Duplication and Loss Inference

Table 3: Key Research Reagents and Computational Tools for NBS-LRR Evolutionary Studies

Resource Type	Specific Tool/Resource	Primary Function	Application Context
Domain Databases	PFAM (PF00931)	NB-ARC domain identification	Initial gene family annotation
Genome Browsers	Phytozome, NCBI Genome	Genomic context visualization	Synteny and cluster analysis
Selection Analysis	PAML (CodeML), KaKs_Calculator	dN/dS calculation	Detecting diversifying selection
Orthology Inference	OrthoFinder, MCScanX	Orthogroup assignment	Comparative genomics
Gene Tree Reconciliation	NOTUNG, RANGER-DTL	Duplication-loss dating	Birth-and-death modeling
Specialized R-gene Databases	PRGdb, PlantNLRatlas	Curated R-gene collections	Reference-based annotation
Expression Analysis	Cufflinks, DESeq2	Transcript quantification	Functional validation

The evolutionary dynamics of NBS-LRR genes represent a sophisticated balance between birth-and-death processes that generate genetic novelty and diversifying selection that refines recognition specificities, particularly in LRR domains. This evolutionary framework explains how plants maintain diverse recognition repertoires despite the fitness costs associated with large resistance gene families.

Future research directions should focus on integrating evolutionary models with functional validation, particularly through genome editing approaches that can test evolutionary hypotheses directly. The development of more sophisticated computational models that incorporate population genetic parameters, pathogen pressure fluctuations, and ecological variables will enhance our predictive capabilities in plant immunity evolution.

Understanding these evolutionary dynamics has practical implications for crop improvement, suggesting strategies for engineering durable disease resistance by harnessing natural evolutionary processes. As genomic technologies advance, our ability to decipher the complex interplay between birth-and-death evolution and diversifying selection will continue to improve, offering new insights into one of plant biology's most dynamic gene families.

Functional characterization of cloned plant resistance (R) genes and their corresponding pathogen avirulence (Avr) effectors is a cornerstone of plant immunity research. This process definitively confirms the specific molecular interactions that trigger a plant's defense response. Within the context of NBS-LRR gene domain architecture and classification, these studies illuminate how different protein domains—such as the Toll/Interleukin-1 receptor (TIR), coiled-coil (CC), nucleotide-binding site (NBS), and leucine-rich repeat (LRR)—orchestrate pathogen recognition and immune activation. Advanced molecular techniques have enabled the cloning of over 450 R genes from 42 plant species, with about 72% encoding cell surface or NLR (NBS-LRR) immune receptors [95]. This technical guide details the experimental frameworks and key findings from seminal case studies in this field, providing a roadmap for researchers.

Case Studies of Cloned R Genes and Effector Pairs

The following case studies exemplify the diverse strategies plants employ to recognize pathogens and the direct experimental evidence required to validate these interactions.

Table 1: Key Case Studies of Cloned R Gene and Effector Pairs

R Gene / Locus	Host Species	Pathogen & Effector	R Gene Class	Recognition Mechanism	Experimental Evidence
RGA4 / RGA5 (Pi-CO39 locus) [96]	Rice (Oryza sativa)	Magnaporthe oryzae (Avr1-CO39, Avr-Pia)	NBS-LRR pair	Direct binding of both effectors by RGA5	Yeast two-hybrid, co-immunoprecipitation, FRET-FLIM, mutant analysis
Sr50, Sr27, Sr35 [97]	Wheat	Puccinia graminis f. sp. tritici (AvrSr50, AvrSr27-2, AvrSr35)	NLR	Specific R-Avr pairing triggers cell death	Protoplast cell death assay, pooled library screening
LepR3 / Rlm2 [98]	Brassica napus (Canola)	Leptosphaeria maculans (AvrLm1, AvrLm2)	Receptor-Like Protein (RLP)	Extracellular recognition; requires SOBIR1	Genetic analysis, pathogenicity assays
MaNBS89 [99]	Banana (Musa acuminata)	Fusarium oxysporum f. sp. cubense (Foc)	NBS-LRR	Induced expression in resistant cultivar; role in defense validated via silencing	RNA-seq, spray-induced gene silencing (SIGS) with dsRNA

Direct Recognition of Dissimilar Effectors by a Paired NLR System

• System Overview: The rice Pi-CO39 locus provides resistance to the blast fungus Magnaporthe oryzae. This locus was found to encode two NBS-LRR proteins, RGA4 and RGA5, which function as a pair and are both required for resistance [96].
• Effector Recognition: The RGA4/RGA5 pair recognizes two fungal effectors, AVR1-CO39 and AVR-Pia, which share no sequence similarity. A key finding was that RGA5 directly binds both effector proteins [96].
• Domain Function: The interaction is mediated not by the canonical LRR domain but by a small C-terminal heavy metal-associated domain in RGA5. This demonstrates how novel domains within NLRs can facilitate the recognition of diverse pathogen molecules [96].
• Functional Characterization Evidence:
  • Protein-Protein Interaction: Direct binding was confirmed through yeast two-hybrid assays, co-immunoprecipitation, and FRET-FLIM in plant cells [96].
  • Mutant Analysis: Natural variants of AVR-Pia that fail to interact with RGA5 in yeast two-hybrid assays also fail to trigger disease resistance, confirming the biological relevance of the interaction [96].
  • Genetic Complementation: Transgenic expression of both RGA4 and RGA5 in a susceptible rice line was sufficient to recapitulate resistance, providing definitive proof of their function [96].

Deployment of a High-Throughput Effector Screening Platform

• Technology Gap: A major bottleneck in R-Avr research has been the slow, laborious process of identifying Avr genes that correspond to cloned R genes. A novel platform using pooled effector library screening in plant protoplasts was developed to address this [97].
• Workflow: The platform involves co-delivering a single R gene and a pooled library of dozens to hundreds of putative effector genes into protoplasts. A subpopulation of cells expressing the cognate Avr gene undergoes R-dependent cell death, leading to the depletion of its transcript from the living cell population. Subsequent RNA-seq and differential expression analysis identifies the candidate Avr gene [97].
• Validation and Application: This method was validated using known wheat stem rust R-Avr pairs (Sr50-AvrSr50, Sr27-AvrSr27-2, Sr35-AvrSr35) and successfully identified novel Avr genes from a library of 696 putative effectors from Puccinia graminis f. sp. tritici [97].
• Key Experimental Parameter: A critical parameter for success is the multiplicity of transfection (MOT), or the number of plasmid molecules per protoplast. An MOT of 0.14 million molecules per cell for each effector construct was found to optimize for independent transformation and a detectable cell death response, enabling the screening of complex libraries [97].

RLP-Mediated Effector Triggered Defense in an Apoplastic Pathosystem

• Non-NLR Resistance: Not all cloned R genes are NLRs. In the Brassica napus (canola) and Leptosphaeria maculans (blackleg) pathosystem, two cloned R genes, LepR3 and Rlm2, encode Receptor-Like Proteins (RLPs) [98].
• Cell Surface Recognition: RLPs are transmembrane proteins with an extracellular LRR domain for ligand perception and a short cytoplasmic tail without a known signaling motif. They recognize pathogen effectors in the apoplastic (extracellular) space [98].
• Signaling Mechanism: Unlike intracellular NLRs that typically initiate ETI, RLPs like LepR3 initiate Effector-Triggered Defense (ETD). This process requires the co-receptor SOBIR1 to transduce the defense signal after effector recognition [98].
• Downstream Signaling: The defense response involves a signaling cascade including Reactive Oxygen Species (ROS) production, calcium ion influx, activation of Mitogen-Activated Protein Kinases (MAPKs), and Salicylic Acid (SA) hormonal signaling, leading to a hypersensitive response (HR) [98].

Experimental Protocols for Functional Characterization

A robust functional characterization of an R-Avr interaction requires a combination of genetic, biochemical, and cellular assays.

Establishing a Direct Interaction

• Yeast Two-Hybrid (Y2H) Assay: A classic genetic method to test for direct protein-protein interaction. The R protein (or its recognition domain, like RGA5's heavy-metal associated domain) is fused to the DNA-binding domain of a transcription factor, and the effector is fused to the activation domain. Interaction is reported by the activation of reporter genes in yeast [96].
• Co-immunoprecipitation (Co-IP): This biochemical method confirms interaction in a near-native cellular environment. Proteins are expressed in a system like plant cells (e.g., via agroinfiltration), and one protein is immunoprecipitated using a specific antibody or tag. If the interaction occurs, the partner protein will be co-precipitated and detected via immunoblotting [96].
• Fluorescence Resonance Energy Transfer-Fluorescence Lifetime Imaging (FRET-FLIM): A powerful biophysical technique to visualize direct protein interactions in living plant cells. If two proteins tagged with fluorophores (donor and acceptor) are in very close proximity (<10 nm), energy transfer occurs, shortening the fluorescence lifetime of the donor. This change is measured to confirm direct interaction in situ [96].

Validating Recognition and Defense Activation

• Protoplast Cell Death Assay: A quantitative method to test R-Avr recognition. Protoplasts are transformed with the R gene, a reporter gene (e.g., luciferase or YFP), and the candidate Avr gene. Recognition-induced cell death leads to a significant reduction in the proportion of reporter-positive living cells, quantified by flow cytometry [97].
• Heterologous Expression and Complementation: The gold standard for validating R gene function. The candidate R gene is transformed into a susceptible plant genotype. If functional, the transgenic plant will gain resistance to pathogens carrying the corresponding Avr effector. This was used to confirm the function of the rice RGA4/RGA5 pair [96].
• Gene Silencing for Loss-of-Function Analysis: Tools like Virus-Induced Gene Silencing (VIGS) or Spray-Induced Gene Silencing (SIGS) can knock down the expression of a candidate R gene. Increased susceptibility in the silenced plant, as demonstrated for the banana MaNBS89 gene, confirms its role in resistance [99].

Mechanistic Insights and Signaling Pathways

Understanding the consequences of R-Avr recognition is crucial for a complete functional characterization. The signaling network involves multiple interconnected components.

The Scientist's Toolkit: Essential Research Reagents and Solutions

Table 2: Key Research Reagents and Solutions for R-Avr Characterization

Reagent / Solution	Function / Application	Specific Examples / Notes
HMMER / RGAugury [16] [100]	Bioinformatic pipeline for genome-wide identification of Resistance Gene Analogues (RGAs) including NLRs, RLKs, and RLPs.	Used to identify 97 NBS-LRR genes in Musa acuminata [99] and 4499 RGAs in Brassicaceae species [100].
PEG-Mediated Protoplast Transformation [97]	High-efficiency delivery of plasmid DNA (R genes, effector libraries) into plant cells for transient expression assays.	Critical for the pooled effector library screening platform [97].
Fluorescent Protein Reporters (YFP, RFP) [97]	Visualization and quantification of transformation efficiency and cell death in protoplasts via flow cytometry.	Used in the individual cell scoring assay to quantify the proportion of cells undergoing R-Avr-dependent cell death [97].
Spray-Induced Gene Silencing (SIGS) Reagents [99]	Loss-of-function analysis via topical application of dsRNA or sRNA to silence target genes in plants.	dsRNA targeting MaNBS89 in banana confirmed its role in Fusarium wilt resistance [99].
Gateway or Golden Gate Cloning Systems	Modular assembly of genetic constructs for expression in plants, yeast, or protoplasts. Enables high-throughput cloning of effector libraries.	Essential for building the pooled effector libraries used in high-throughput screens [97].

The functional characterization of cloned R genes and their effectors has profoundly advanced our understanding of plant immunity. From revealing novel recognition mechanisms, such as the paired NLR system of rice RGA4/RGA5, to the development of transformative high-throughput screening platforms, these case studies provide both a historical foundation and a forward-looking technical framework. The integration of bioinformatic predictions with rigorous experimental validation—encompassing genetics, biochemistry, and cell biology—remains paramount. As the repertoire of cloned R genes continues to expand, these foundational principles and innovative methodologies will empower researchers to decode the complex language of plant-pathogen interactions, ultimately informing strategies for engineering durable disease resistance in crops.

Conclusion

The systematic classification of NBS-LRR genes based on their domain architecture is fundamental to unlocking the genetic basis of plant disease resistance. This synthesis of foundational knowledge, advanced methodologies, troubleshooting insights, and validation frameworks provides a powerful roadmap for researchers. The integration of traditional domain analysis with cutting-edge deep learning tools like PRGminer is revolutionizing our capacity to mine plant genomes for valuable resistance traits. Future efforts must focus on the high-quality functional characterization of predicted genes and the application of this knowledge in precision breeding. By bridging genomic discovery with practical crop improvement, research into NBS-LRR genes holds immense promise for developing durable disease resistance and safeguarding global food security.

References

• 1. Plant NBS-LRR proteins: adaptable guards - Genome Biology [https://genomebiology.biomedcentral.com/articles/10.1186/gb-2006-7-4-212]
• 2. Phylogenetic, Structural, and Evolutionary Insights into ... [https://www.mdpi.com/1422-0067/26/5/1828]
• 3. Genome-Wide Identification of NBS-LRR Family in Three ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12192410/]
• 4. Genome-wide identification and characterization of NBS ... [https://www.nature.com/articles/s41598-025-03507-5]
• 5. Plant NBS-LRR proteins in pathogen sensing and host defense [https://pmc.ncbi.nlm.nih.gov/articles/PMC1973153/]
• 6. PRGminer: harnessing deep learning for the prediction of ... [https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2025.1411525/full]
• 7. Genome-Wide Characterization and Expression Pattern ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12469828/]
• 8. Interaction between domains of a plant NBS–LRR protein in ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC126192/]
• 9. The leucine-rich repeat domain in plant innate immunity [https://pmc.ncbi.nlm.nih.gov/articles/PMC2762402/]
• 10. Identification and Characterization of NBS Resistance ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC8585750/]
• 11. Functional characterization of NBS-LRR genes reveals an ... [https://bmcbiol.biomedcentral.com/articles/10.1186/s12915-024-01836-x]
• 12. CRISPR activation: identifying and using novel genes for ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12450905/]
• 13. GL6.1 integrates grain length regulation and disease ... [https://www.sciencedirect.com/science/article/pii/S2214514125001990]
• 14. Structural basis of NLR activation and innate immune ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC8813719/]
• 15. Evolution of NLR Resistance Genes in Magnoliids [https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2021.777157/full]
• 16. Genome-Wide Identification of NBS-LRR Family in Three ... [https://www.mdpi.com/2073-4425/16/6/680]
• 17. Genome-wide characterization of the NLR gene family in ... [https://www.frontiersin.org/journals/genetics/articles/10.3389/fgene.2022.931580/full]
• 18. Genome-wide identification of nucleotide-binding domain ... [https://bmcplantbiol.biomedcentral.com/articles/10.1186/s12870-023-04474-7]
• 19. Diversity, abundance, and domain architecture of plant ... [https://www.sciencedirect.com/science/article/pii/S2405844024105063]
• 20. Deciphering Plant NLR Genomic Evolution: Synteny- ... [https://pubmed.ncbi.nlm.nih.gov/39835721/]
• 21. Identification and Characterization of NBS Resistance ... [https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2021.758559/full]
• 22. Functional characterization of NBS-LRR genes reveals an ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC10898138/]
• 23. Genome-Wide Identification of NBS-Encoding Resistance ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC6115920/]
• 24. A Truncated NLR Protein, TIR-NBS2, Is Required for Activated ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC4305288/]
• 25. Genome-wide comparative analysis of NBS-encoding genes ... [https://bmcgenomics.biomedcentral.com/articles/10.1186/s12864-017-3682-x]
• 26. Distinct Evolutionary Patterns of NBS-Encoding Genes in ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC7365912/]
• 27. An atypical NLR protein modulates the NRC immune receptor ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC9851556/]
• 28. Evolution patterns of NBS genes in the genus Dendrobium ... [https://bmcplantbiol.biomedcentral.com/articles/10.1186/s12870-022-03904-2]
• 29. Identification and distribution of the NBS-LRR gene family in ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC4422547/]
• 30. Genome-wide analysis of NBS-LRR genes in Rosaceae ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC9685399/]
• 31. Species-specific duplications driving the recent expansion of ... [https://bmcgenomics.biomedcentral.com/articles/10.1186/s12864-015-1291-0]
• 32. Comparative analysis, diversification, and functional ... [https://www.nature.com/articles/s41598-024-62876-5]
• 33. Genome-Wide Identification and Evolutionary Analysis of ... [https://www.frontiersin.org/journals/genetics/articles/10.3389/fgene.2020.00484/full]
• 34. Comparative genomics and bioinformatics approaches ... [https://www.frontiersin.org/journals/genetics/articles/10.3389/fgene.2024.1358134/full]
• 35. Genome-Wide Analysis of the NBS-LRR Gene Family and ... [https://www.mdpi.com/2311-7524/10/12/1293]
• 36. DaapNLRSeek, prediction, and evolution of resistance ... [https://www.sciencedirect.com/science/article/pii/S2589004225011307]
• 37. Plant NBS-LRR proteins: adaptable guards - PubMed Central [https://pmc.ncbi.nlm.nih.gov/articles/PMC1557992/]
• 38. Identification and distribution of the NBS-LRR gene family in ... [https://bmcgenomics.biomedcentral.com/articles/10.1186/s12864-015-1554-9]
• 39. Pfam is now hosted by InterPro [http://pfam.xfam.org/]
• 40. AlphaFold 3模板搜索机制：HMMER与结构数据库应用 [https://blog.csdn.net/gitblog_01138/article/details/154510402]
• 41. Computational Strategies for Elucidating Plant Disease ... [https://www.sciencedirect.com/science/article/abs/pii/S0885576525003790]
• 42. PRGminer [https://kaabil.net/prgminer/]
• 43. PRGMiner Tutorial [https://kaabil.net/prgminer/tutorial]
• 44. Machine learning-based prediction of antimicrobial resistance ... [https://bmcgenomdata.biomedcentral.com/articles/10.1186/s12863-025-01338-x]
• 45. Development of Machine Learning Methods for Accurate ... [https://www.sciencedirect.com/science/article/pii/S2095809924002431]
• 46. Genome‐wide identification and evolutionary analysis of ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12326289/]
• 47. Searching a Profile HMM Against a Sequence Database ... [https://medium.com/@sepo7/automated-bioinformatics-pipeline-for-searching-profile-hmms-against-a-sequence-database-with-4734d447b047]
• 48. HMMER: biological sequence analysis using profile HMMs [https://github.com/EddyRivasLab/hmmer]
• 49. Common Methods for Phylogenetic Tree Construction and ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC11117635/]
• 50. Evolution patterns of NBS genes in the genus Dendrobium ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC9661794/]
• 51. Transcriptomics and epigenetic data integration learning ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC11299028/]
• 52. OsLB2.2 negatively regulates rice disease resistance at ... [https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2025.1629283/full]
• 53. Integrative transcriptomic and metabolomic analysis ... [https://www.sciencedirect.com/science/article/pii/S092666902400699X]
• 54. Integration of early disease-resistance phenotyping ... [https://www.nature.com/articles/s41438-021-00543-w]
• 55. Genome-wide identification and characterization of NBS ... [https://pubmed.ncbi.nlm.nih.gov/40447724/]
• 56. 植物免疫团队揭示小麦抗白粉病和条锈病基因新模式 [http://genetics.cas.cn/dtxw/kyjz/202506/t20250610_7800887.html]
• 57. Identification, characterization, and validation of NBS ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC10318170/]
• 58. Extensive Error in the Number of Genes Inferred from Draft ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC4256071/]
• 59. Genome-wide analysis of NBS-encoding disease resistance ... [https://bmcgenomics.biomedcentral.com/articles/10.1186/1471-2164-14-109]
• 60. Repeated sequence (DNA) [https://en.wikipedia.org/wiki/Repeated_sequence_(DNA)]
• 61. Repetitive DNA sequence detection and its role in the human ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC10509279/]
• 62. Bioinformatics Tools: Gene Prediction/ Annotation [https://guides.library.yale.edu/bioinformatics/gene-prediction]
• 63. Transcriptome analysis reveals key defense genes associated ... [https://bmcgenomics.biomedcentral.com/articles/10.1186/s12864-025-12056-0]
• 64. Genome annotation for clinical genomic diagnostics: strengths ... [https://genomemedicine.biomedcentral.com/articles/10.1186/s13073-017-0441-1]
• 65. Comparative analysis of NBS domain sequences ... [https://www.sciencedirect.com/science/article/abs/pii/S1055790303002744]
• 66. In silico prediction method for plant Nucleotide‐binding ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12042882/]
• 67. Deep-learning-based single-domain and multidomain ... [https://www.nature.com/articles/s41587-025-02654-4]
• 68. Improving prediction accuracy in chimeric proteins with ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12328686/]
• 69. An Overview of Duplicated Gene Detection Methods [https://pmc.ncbi.nlm.nih.gov/articles/PMC7565063/]
• 70. Recurrent Tandem Gene Duplication Gave Rise to ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC2878002/]
• 71. Structural polymorphism and diversity of human segmental ... [https://www.nature.com/articles/s41588-024-02051-8]
• 72. The Importance of Gene Duplication and Domain Repeat ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC8830517/]
• 73. Profiling of gene duplication patterns of sequenced teleost ... [https://bmcgenomics.biomedcentral.com/articles/10.1186/1471-2164-13-246]
• 74. Discovery of functional NLRs using expression level, high- ... [https://www.nature.com/articles/s41477-025-02110-w]
• 75. Next-generation sequencing of newborn screening genes [https://www.nature.com/articles/s41525-020-00142-z]
• 76. Avoid Common Obstacles in Protein Expression [https://www.neb.com/en-ca/tools-and-resources/feature-articles/bypassing-common-obstacles-in-protein-expression?srsltid=AfmBOorJ16KzWUauePu5kNIBTjRi5x2xT-H4d7oDi8RjZAzJdwBA00hu]
• 77. Semantic design of functional de novo genes from a ... [https://www.nature.com/articles/s41586-025-09749-7]
• 78. 2025 search tool benchmark: key metrics to evaluate ... [https://www.glean.com/blog/2025-search-tool-benchmark-key-metrics-to-evaluate-accuracy-and-speed]
• 79. A comparison of computational methods for expression ... [https://pubmed.ncbi.nlm.nih.gov/41250104/]
• 80. AI benchmarking: Nine challenges and a way forward - AI Watch [https://ai-watch.ec.europa.eu/news/ai-benchmarking-nine-challenges-and-way-forward-2025-09-10_en]
• 81. The genomic landscape of gene-level structural variations ... [https://www.nature.com/articles/s41588-025-02113-5]
• 82. LLM benchmarks in 2025: What they prove and what your ... [https://www.lxt.ai/blog/llm-benchmarks/]
• 83. -Wide In Silico Analysis of Leucine-Rich Repeat Genome - R in... Genes [https://pmc.ncbi.nlm.nih.gov/articles/PMC11855831/]
• 84. Development of a robust and efficient virus-induced gene ... [https://plantmethods.biomedcentral.com/articles/10.1186/s13007-024-01320-1]
• 85. Efficient Virus-Induced Gene Silencing (VIGS) Method for ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12115190/]
• 86. Bioluminescence-based assays for quantifying ... [https://pubmed.ncbi.nlm.nih.gov/40617353/]
• 87. GENESPACE tracks regions of interest and gene copy ... [https://elifesciences.org/articles/78526]
• 88. Cross-Species Sequence Comparisons: A Review of ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC430969/]
• 89. JCVI: A versatile toolkit for comparative genomics analysis [https://pmc.ncbi.nlm.nih.gov/articles/PMC11316928/]
• 90. Response of NBS encoding resistance genes linked to ... [https://www.sciencedirect.com/science/article/abs/pii/S0981942814003428]
• 91. Molecular Breeding for Abiotic Stress Tolerance in Crops [https://www.mdpi.com/1422-0067/26/18/9164]
• 92. Evolutionary Dynamics of the Leucine-Rich Repeat ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC4775120/]
• 93. Evolutionary balance between LRR domain loss and young ... [https://bmcgenomics.biomedcentral.com/articles/10.1186/s12864-019-6212-1]
• 94. Evolutionary dynamics and impacts of chromosome ... [https://www.nature.com/articles/s41598-020-57729-w]
• 95. R we there yet? Advances in cloning resistance genes for ... [https://www.sciencedirect.com/science/article/pii/S1369526623001541]
• 96. Resistance Genes Team up to Recognize Different Fungal ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC3663258/]
• 97. Pooled effector library screening in protoplasts rapidly ... [https://www.nature.com/articles/s41477-024-01641-y]
• 98. From Recognition to Response: Resistance–Effector Gene ... [https://www.mdpi.com/2223-7747/14/3/390]
• 99. NBS-LRR genes of Musa acuminata is involved in disease ... [https://www.sciencedirect.com/science/article/abs/pii/S0304423824005181]
• 100. Exploring Cloned Disease Resistance Gene Homologues and ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12385795/]