2.1 Ethical Approval

The present research, which was authorized by the Research Ethics Committee of Birjand University of Medical Sciences (IR.BUM.REC1400.4.9), required written informed consent from all probands and their family members. If the family members were minors, consent was obtained from their guardians.

2.2 Clinical Data and Sample Collection

We recruited four pedigrees with a confirmed diagnosis of RP and a clear family history of at least two affected members in Nehbandan city (South Khorasan province, Iran) between January 2022 and May 2022.

RP within families was determined based on symptoms reported by probands and their relatives, such as vision problems and difficulty seeing at night. These reports were then verified through clinical assessments. To confirm retinal pathology, we employed digital fundus photography, focusing on conditions such as bone spicule pigmentation and optic disc pallor.

All patients underwent comprehensive ophthalmic examinations, and genetic counseling sessions were conducted with the probands' parents, incorporating the expertise of a senior human geneticist. We meticulously documented family history, including any notable occurrence of this condition in historical records (Table 1). Figure 1 displays the pedigree chart generated using Evagene software (https://www.evagene.com/). Venous blood samples were collected from both affected and unaffected family members using EDTA vacutainers.

2.3 Whole-Exome Sequencing (WES)

In this study, we employed WES to uncover genetic variants associated with RP. genomic DNA was isolated from the patient's blood using the QIAamp DNA Blood Mini-Kit (QIAGEN, Germany) and evaluated the quality and quantity of the extracted DNA using a NanoDrop spectrophotometer (Epoch, Bio-Tek Inc., VT, USA). The methods employed for our genetic analysis were previously outlined [16]. Briefly, we selectively captured the protein-coding regions of the genome using the SureSelect Human All Exon V6 kit from Agilent. The captured DNA fragments underwent sequencing using the HiSeq2500 platform from Illumina (San Diego, USA), generating paired-end reads of 150 base pairs. Standard bioinformatics pipelines were used to process the resulting sequence data. We aligned sequences to the reference genome using Burrows-Wheeler Aligner (BWA) software. Duplicate reads were remove using Picard tools, and base quality scores were adjusted using the Genome Analysis Toolkit (GATK). Variant calling was performed using GATK's Haplotype Caller tool [17]. Identified variants were annotated using ANNOVAR software [17]. Variants with a minor allele frequency (MAF) greater than 1% in genetic databases (such as ExAC_EAS, 1000 Genomes Project, and Iranome) (http://iranome.ir/) were filtered out. The remaining variants were further filtered based on functional annotation, predicted pathogenicity, and known association with RP. We evaluated the clinical significance of the variant using various genetic mutation databases, including ClinVar, Online Mendelian Inheritance in Man (OMIM), and the Human Gene Mutation Database (HGMD). The variant was then classified and interpreted according to the guidelines provided by the American College of Medical Genetics (ACMG) [18]. Sanger sequencing confirm the identified variants in affected individuals and tested for segregation in family members. The PCR primers are designed for a 377 base-pair fragment: Forward (5′-CATTCTCGGGTCCCAGGACAGCA-3′) and Reverse (5′-ACTCAGTTACCTGGCTTGAAGAGG-3′). Sequencing was conducted on an ABI 3730XL sequence analyzer (Applied Biosystems, USA).

2.4 Protein Modeling

The amino acid sequences for NR2E3 and NR1D1 were retrieved from the NCBI protein database, and the LBD of NR2E3, containing the identified mutations, was modeled using homology modeling techniques. In our study, we utilized Iterative Threading Assembly Refinement (I-TASSER) as a computational tool for predicting the three-dimensional structure of the mutant protein. This approach provided valuable insights into the structural features and potential functional implications of the mutated variant, deepening our understanding of its molecular properties. The quality of the I-TASSER model was assessed using the confidence score (C-score), which ranges from −5 to 2. The C-score serves as a measure of the accuracy and reliability of the predicted protein structures. Subsequently, the model underwent assessment using Molprobity. This allowed us to assess the geometric and energetic properties of both the mutant and normal protein structures. Furthermore, we constructed Ramachandran plots using MolProbity to compare the quality of the structures.

2.5 Protein–Protein Docking

Computational tools have become invaluable for detecting pathogenic variants and assessing their impact on protein function. In this study, our primary objective was to explore the potential effects of mutations occurring in the LBD of NR2E3 on its interaction with NR1D1. To achieve this, we employed protein–protein docking simulations to investigate the dynamics between these two proteins. The docking process was initiated using the ClusPro 2.0 protein–protein docking server (https://cluspro.bu.edu/) to dock the crystal structure of NR1D1 and NR2E3 onto the LBD of NR1D1. This approach generated various complex structures representing potential binding configurations between NR2E3 and NR1D1. To assess and gain insights into these docked complexes, we used the PyMOL v2.4 (Schrödinger, USA) molecular graphics system. Leveraging its powerful visualization capabilities, we meticulously examined and analyzed feasible interactions and structural models resulting from the protein–protein docking simulations. This step provided valuable information on potential binding modes and the stability of the NR2E3-NR1D1 complexes. By combining these techniques, our study aimed to illuminate the interplay between mutations in the LBD of NR2E3 and their interaction with NR1D1, offering insights into the molecular mechanisms underlying these interactions and their potential implications.

2.6 Molecular Dynamics Simulation

In our investigation, we employed GROMACS 2021.4 for molecular dynamics (MD) simulations, utilizing CHARMM27 force field parameters to accurately capture the behavior of both mutant and normal proteins. To ensure system fidelity, we carefully selected simulation boxes with appropriate dimensions, applying periodic boundary conditions. Placing the protein complexes approximately 1 nm away from the carriers' surface within each box effectively mitigated undesired interactions between neighboring components, preserving system integrity.

After placing the protein complexes, we solvated the simulation boxes meticulously using explicitly treated water molecules, adopting the widely-accepted TIP3P model. This choice allowed us to accurately mimic the behavior of water molecules, creating a realistic environment for the proteins within the simulation. To maintain system neutrality and establish a biologically relevant context, we strategically introduced Na and Cl ions by replacing water molecules. This crucial step balanced any charge imbalances, ensuring system stability and physiological accuracy throughout the simulation. Subsequently, we conducted energy minimization to relax the system, followed by a series of equilibration steps to gradually bring the system to the desired temperature and pressure. These equilibration procedures allowed the system to reach equilibrium, enabling the proteins to adjust and settle into their normal conformations. Executing these methodologies within GROMACS, our simulations aimed to unravel the intricate dynamic behavior, conformational changes, and intermolecular interactions exhibited by the mutant and normal protein complexes. This comprehensive exploration provides valuable insights into the structural stability, flexibility, and functional implications of the proteins under investigation, advancing our understanding of their behavior within a simulated biological context. The simulations allowed us to calculate several important parameters, such as Root Mean Square Deviation (RMSD), Root Mean Square Fluctuation (RMSF), and hydrogen bond (HB) interactions. RMSD provides information about the structural stability and conformational changes of proteins during the simulation. It quantifies the overall structural similarity or dissimilarity over time. RMSF allows us to examine the flexibility and fluctuations of individual residues within the proteins. By analyzing RMSF, we gain insights into specific regions that experience significant conformational changes or exhibit higher mobility during the simulation. HBs play a crucial role in protein stability and function. Studying HB interactions during MD simulations helps us understand how these bonds contribute to maintaining the protein's structure and stability.

Analyzing these parameters obtained from the MD simulations deepens our understanding of the dynamic behavior and structural properties of both mutant and normal proteins.

3.1 Clinical Finding

Among the twenty-five members of a large family with definite or suspected clinical diagnoses of RP, fourteen individuals (II4, III4, III6, III11, III15, III19, III25, III30, III31, III41, IV7, IV13, IV15, and IV21) underwent clinical examinations performed by geneticists and ophthalmologists. Within this specific subgroup, DNA samples from five individuals (III4, III15, III30, IV7, and IV21) diagnosed with RP were analyzed using WES. In the affected family members with RP, a variety of symptoms were observed, including tunnel vision, photophobia, color vision abnormalities, and visual disturbances. Additionally, they exhibited accompanying ocular conditions such as cataracts, strabismus, and myopia. Figure 2 provides fundus images from four different patients, illustrating the diverse range of retinal findings. These valuable observations are summarized in Table 1.

3.2 Genetic and Bioinformatics Findings

Using WES, we identified a pathogenic missense variant (c.932G>A, p.R311Q; rs28937873) in exon 6 of the NR2E3 gene (GenBank NM_014249.4). This variant leads to the substitution of arginine (codon: CGG) with glutamine (codon: CAG) at position 311 of the NR2E3 protein. The NR2E3 transcript consists of eight exons and is characterized by seven domains and features. Notably, this variant is located within the NR_LBD. Following the standards and guidelines of the American College of Medical Genetics and Genomics (ACMG), we rigorously assessed the pathogenicity of the identified variant. Among the sequencing results, only the NR2E3 gene was associated with RP, and no other genes related to eye disorders were identified. To understand the variant's frequency in various human populations, we consulted databases such as TOPMED, GnomAD, ALFA, 1000 Genomes (1000G), and ExAC. The observed frequencies were as follows: 0.000468, 0.000321, 0.000338, 0.0008, and 0.000342, respectively. Based on the information extracted from HGMD and ClinVar, this mutation is firmly established as a pathogenic variant. Direct sequencing confirmed the presence of the c.932G>A, p.R311Q variant in a homozygous state in all five patients included in the WES analysis, consistent with autosomal recessive inheritance (Figure 3). Taken together, these results strongly support the notion that the NR2E3 variant fully segregates with RP in this family, suggesting it is likely the causal mutation underlying their condition.

3.3 Molecular Docking Findings

To gain a more comprehensive understanding of the molecular mechanics and structural implications arising from the p.R311Q mutation in NR2E3, we utilized online I-TASSER and SWISS MODEL servers to generate models of the mutant protein. The computational models allow us to visualize how the mutation might affect the protein's three-dimensional structure. The quality of the modeled protein was rigorously assessed using multiple methods, including analysis of the Ramachandran plot, ERRAT, utilization of the MolProbity tools, and examination with Bioluminate software (Release 2023-1, Schrödinger, USA) based on the protein reliability report (Figure 4).

To examine how the mutation impacts the binding site of NR2E3, we conducted molecular docking analyses using BioLuminate (Release 2023-1, Schrödinger, USA). Our target was the structural complex formed by NR2E3-NR1D1. To accomplish this, we generated combined individual crystal structures of NR2E3 and NR1D1 to create potential models. These models represent the interaction between the two proteins. We then compared these models with the docking complex formed by the mutant NR2E3 and NR1D1. This allowed us to pinpoint any differences caused by the p.R311Q mutation. Figure 5 illustrates the docking of both the wild-type and mutant NR2E3 with NR1D1. Notably, the docking results revealed significant changes in the interaction between the mutated NR2E3 and NR1D1 when compared to the wild type. Furthermore, our structure refinement analysis indicated a substantial total energy change of 141.85 kcal/mol upon substituting the arginine residue with glutamine in the NR2E3 amino acid sequence (Table 2).

3.4 Molecular Dynamics Simulation Findings

Following our docking analysis results, we delved into the dynamic characteristics of both the normal and mutant complexes. To initiate the MD simulation, we selected the docking conformation with the most negative binding energy, as obtained from BioLuminate software. The RMSD plots (Figure 6) revealed that the normal complex achieved a stable RMSD value of approximately 0.6 nm after the initial 75 ns, indicating a conserved overall structure throughout the simulation. On the other hand, the mutant complex exhibited slightly increased fluctuations, with an RMSD reaching 0.7 nm. This implies a certain level of conformational flexibility in the mutant variant. To assess residue-specific fluctuations, we examined the RMSF profiles (Figure 6). In the normal complex, residues 50–130 and 125–250 exhibited increased RMSF values, indicating higher flexibility in these regions. Interestingly, in the mutant complex, residues 120–150 displayed even higher RMSF values, suggesting a potential impact of the mutation on local dynamics. Furthermore, we investigated the HB interactions in both complexes. This indicates that the mutation introduced in the protein did not have a substantial impact on the HB network within and between the complexes. The HB interactions remained relatively stable and consistent between the normal and mutant variants, suggesting that the mutation did not disrupt the crucial HB interactions involved in the complex formation. This finding implies that the mutation may not directly affect the binding affinity or stability of the complex through alterations in HB interactions.