A recurrent CCDC82 frameshift variant associated with syndromic neurodevelopmental disorder in a consanguineous Pakistani family Authors: Tehmeena Akhter, Sami Khan, Ahmed Abdul Mumeen, Muhammad Asif Naeem, Sheikh Riazuddin, Saima Riazuddin Published: 2026 ABSTRACT BackgroundIntellectual disabilities (IDs) are part of neurodevelopmental disorders (NDDs) and are genetically heterogeneous conditions characterized by impairments in cognition, learning, and adaptive functioning. Despite advances in gene discovery, many individuals, particularly those from understudied populations, remain without a molecular diagnosis. Recent reports implicate CCDC82 (HGNC: 26282) as an autosomal recessive ID gene, although the phenotypic spectrum and biological context remain incompletely defined.MethodsExome sequencing (ES) was performed in a consanguineous Pakistani family (PKMR06A) with four affected individuals presenting with moderate to severe ID. Variant segregation was confirmed by Sanger sequencing. In silico analyses, including pathogenicity prediction, protein structural modeling, and domain intolerance assessment, were used to evaluate the functional consequences of the identified variant. Spatiotemporal gene expression patterns were examined using bulk and single-cell human brain transcriptomic datasets.ResultsClinically, affected individuals of family PKMR06A presented with early childhood global developmental delay, speech delay, hypotonia, gait abnormalities, spasticity, and mild facial dysmorphism. Genetic screening revealed a recurrent rare homozygous frameshift variant in CCDC82 (NM_024725.4): c.373del; p.(Asp125Ilefs*6), segregating with disease in all available affected individuals of the family. The identified c.373del variant was absent from the gnomAD database and was classified as pathogenic (PVS1, PM2, and PP1) based on ACMG/AMP criteria. The c.373del variant is predicted to introduce a premature termination codon, p.(Asp125Ilefs*6), leading to deletion of essential coiled-coil domains from the encoded protein, supporting a loss-of-function mechanism. In silico, transcriptomic analyses demonstrated preferential CCDC82 expression during prenatal human brain development, providing developmental context for the neurodevelopmental phenotype associated with the identified truncating variant.ConclusionsThis study expands the mutational landscape of CCDC82 and provides additional clinical and molecular evidence supporting its role in autosomal recessive NDD. The findings reinforce the importance of CCDC82 in human neurodevelopment and highlight the value of genomic investigation in underrepresented populations. Background Intellectual disabilities (IDs) consist of a broad, heterogeneous group of neurodevelopmental conditions characterized by significant impairments in cognitive functioning, learning capacity, and adaptive behavior that arise during early childhood [[1], [2], [3]]. Affecting approximately 1-3% of the global population, these disorders impose substantial social and economic burdens, specifically in regions where diagnostic resources remain limited and specialized services are scarce [[4], [5], [6], [7]]. While environmental and perinatal complications contribute to the prevalence of ID worldwide, a substantial proportion of moderate-to-severe cases have genetic causes [[8], [9], [10], [11]]. Over the past decade, the widespread adoption of massively parallel sequencing, particularly exome sequencing (ES), has transformed the diagnostic field, enabling the discovery of novel monogenic causes and refining our understanding of the genetic architecture underlying IDs [[12], [13], [14], [15]]. Nevertheless, many affected families, especially those in low-resource or understudied populations, still lack molecular diagnoses, highlighting the need for continued research into rare and previously uncharacterized genetic contributors [15,16]. Among the genes recently implicated in neurodevelopmental pathology is CCDC82 (Coiled-Coil Domain Containing Protein 82), located on chromosome 11q21 [17,18]. Although its precise biological role remains incompletely defined, accumulating genomic and transcriptomic data, including developmental expression analyses, suggest that CCDC82 participates in chromatin organization, centrosome dynamics, and multiprotein interaction networks essential for neuronal maturation and maintenance [19]. Recent reports have identified biallelic loss-of-function variants in CCDC82 (NM_024725.4) in individuals with ID, hypotonia, spasticity, and developmental delay; however, because, relatively few cases have been reported, the full clinical spectrum remains incompletely defined [[17], [18], [19], [20]]. Importantly, nearly all reported cases with CCDC82 disease-causing variants originate from consanguineous families in the Middle East and South Asia [18]. Here, we present a consanguineous family (PKMR06A) with four affected individuals from the Southern Punjab province of Pakistan (Fig. 1A). All affected individuals had moderate-to-severe ID (HP:0001249; HP:0002342; HP:0010864), accompanied by early-onset developmental delay (HP:0001263), speech delay (HP:5200240), hypotonia (HP:0001252) and spasticity (HP:0001257), gait abnormalities (HP:0001288), and dysmorphic facial features (HP:0001999). Using exome sequencing followed by Sanger sequencing, we identified a rare homozygous frameshift variant in CCDC82. By integrating clinical observations, genomic data, and computational predictions, this study expands the understanding of CCDC82 loss-of-function as a cause of autosomal recessive IDs. Ultimately, the findings contribute to the broader effort to explain the molecular basis of neurodevelopmental disorders and emphasize the value of comprehensive genetic analysis in improving diagnostic outcomes for affected families.Fig. 1Clinical, genetic, and molecular characterization of family PKMR06A with the CCDC82 (NM_024725.4): c.373del; p.(Asp125Ilefs*6) variant. (A) Pedigree of the consanguineous Pakistani family showing segregation of the rare homozygous CCDC82 frameshift variant. Affected individuals are homozygous for the variant (M/M), parents are heterozygous carriers (W/M), and unaffected siblings are wild-type (W/W). (B) Clinical photographs of affected individuals show shared neurodevelopmental features, including developmental delay, hypotonia, and mild craniofacial dysmorphism. (C) Schematic of the CCDC82 gene showing the location of the c.373del variant identified in this study (red arrow), alongside previously reported pathogenic truncating variants, including c.535C > T; p.(Arg179*) [18], and c.709C > T; p.(Arg237*) [17]; (blue and pink arrows respectively). (D) Protein model illustrating the predicted truncation at p.(Asp125Ilefs*6) and loss of the coiled-coil domain. Clinical, genetic, and molecular characterization of family PKMR06A with the CCDC82 (NM_024725.4): c.373del; p.(Asp125Ilefs*6) variant. (A) Pedigree of the consanguineous Pakistani family showing segregation of the rare homozygous CCDC82 frameshift variant. Affected individuals are homozygous for the variant (M/M), parents are heterozygous carriers (W/M), and unaffected siblings are wild-type (W/W). (B) Clinical photographs of affected individuals show shared neurodevelopmental features, including developmental delay, hypotonia, and mild craniofacial dysmorphism. (C) Schematic of the CCDC82 gene showing the location of the c.373del variant identified in this study (red arrow), alongside previously reported pathogenic truncating variants, including c.535C > T; p.(Arg179*) [18], and c.709C > T; p.(Arg237*) [17]; (blue and pink arrows respectively). (D) Protein model illustrating the predicted truncation at p.(Asp125Ilefs*6) and loss of the coiled-coil domain. Materials and methods Family enrollment and clinical assessment This study was conducted under approval from the Institutional Review Board (IRB) of the University of Maryland, Baltimore (UMB), USA, and the Center of Excellence in Molecular Biology (CEMB), University of the Punjab, Lahore, Pakistan. Written informed consent was obtained from all participating family members or their legal guardians. Family PKMR06A, a consanguineous Pakistani family with four affected individuals, was enrolled from Multan, Pakistan. Comprehensive clinical phenotyping was conducted for all available individuals, including developmental history, motor milestones, hypotonia, speech development, behavioral observations, muscle tone, reflexes, gait analysis, and assessment for dysmorphic facial features. Phenotypic features were cataloged using standardized terminology from the Human Phenotype Ontology [21]. Available photographs of affected individuals were reviewed clinically for shared dysmorphic features. Peripheral blood samples were collected from all the participants, and genomic DNA was extracted following the standard protocol of inorganic extraction [22]. Exome sequencing (ES) and variant filtration ES was performed on the proband (V:1) using the Agilent SureSelect Human All Exon V6 capture kit, followed by sequencing on an Illumina HiSeq platform. Sequencing produced a mean coverage depth exceeding 80x, ensuring high-confidence variant detection. Raw FASTQ data underwent quality assessment using FastQC [23]. Reads were aligned to the hg19 (GRCh37; GCF_000001405.13) human reference genome using BWA-MEM [24]. PCR duplicates were removed using Picard Tools. Base quality score recalibration (BQSR) and variant calling were performed using GATK HaplotypeCaller [25], following best practice guidelines. Sequencing data were filtered and prioritized using the criteria described previously [15,[26], [27], [28]], based on (i) population allele frequency in gnomAD, ExAC, and 1000 Genomes; (ii) predicted functional impact (nonsense, frameshift, splice-site), (iii) homozygosity consistent with consanguinity, and (iv) compatibility with an autosomal recessive inheritance model. Additional filtering excluded synonymous and non-coding variants unlikely to affect protein function. Primer validation and segregation analysis To evaluate the segregation of the candidate variant, Sanger sequencing was performed across all available family members. Primers (forward primer: 5′-TGAGCTCTCTTTCACTTCCTGG-3’; reverse primer: 5′-TCGAGTTGATTGGAGGCGAA-3′) targeting exon 4 of CCDC82 were designed using Primer3.0 [29], and their specificity was validated using NCBI Primer-BLAST [30]. PCR amplification was performed using standard cycling conditions, including an initial denaturation step at 95°C for 10 min, followed by 35 cycles of denaturation (95°C for 30 s), annealing (temperature optimized for each primer pair at 60°C for 30 s), and extension (72°C for 30 s), with a final extension at 72°C for 10 min. Amplified products were purified using the ExoSAP method and sequenced using Sanger sequencing on an ABI platform. Sequence chromatograms were analyzed manually using DNASTAR Navigator v18 [31]. In silico molecular modeling The functional impact of the identified CCDC82 frameshift variant was assessed using in-silico approaches focused on protein domain intolerance and structural modeling. Regional tolerance to variation across the CCDC82 protein was evaluated using MetaDome [32], which integrates population-scale human variation data to identify mutation-intolerant domains. Moreover, three-dimensional (3D) protein structural models were generated using the I-TASSER server [33,34] and visualized with PyMOL v3.1 [35]. Model quality was assessed using the confidence score (C-score), estimated template modeling score (TM-score), estimated RMSD, and template coverage reported by I-TASSER. Because CCDC82 contains predicted coiled-coil and low-complexity regions, the models were interpreted with caution and were primarily used to visualize the predicted truncation and loss of the C-terminal domain, rather than to draw conclusions from experimentally validated structures and mechanisms. CCDC82 interactors were predicted using the STRING v12.0 database of functional protein associations [36]. Expression analysis of CCDC82 interactome in the developing human brain Cell-type-specific expression patterns of CCDC82 and selected interacting genes (LRRC2, GPALPP1, and C11orf54) during human cortical development were examined using dot-plot visualizations generated with the NeuroTri2 VISDOT web interface [37]. Furthermore, spatiotemporal expression data of CCDC82, LRRC2 (HGNC: 14676), GPALPP1 (HGNC: 20298), and C11orf54 (HGNC: 30204) were also collected from the BrainSpan Atlas of the Developing Human Brain [38]. Brain expression datasets were categorized into six biologically defined development stages: first trimester, second trimester, third trimester, early postnatal (<2 years), childhood/adolescence (2–<10 years), and adulthood (≥10 years). Prenatal and postnatal BrainSpan expression levels were compared using the Mann-Whitney U test [39]. The analysis included 237 prenatal and 287 postnatal tissue-region samples, corresponding to 20 prenatal donors and 22 postnatal donors. Expression values were log2-transformed using log2(expression + 1). Because four genes were evaluated, p-values were adjusted using the Benjamini-Hochberg false-discovery-rate method. Effect sizes were calculated using rank-biserial correlation, where positive values indicate higher prenatal expression and negative values indicate higher postnatal expression. Brain regions were ranked by their mean expression across all stages, and the top 25 were selected for visualization. All data processing, statistical analyses, and figure generation were carried out using Python (v3.9), with the pandas and NumPy libraries for data handling and transformation, and seaborn and matplotlib for visualization (Supplementary notes). Single-cell expression analysis of the developing human cortex Single-cell gene expression patterns for CCDC82, LRRC2, GPALPP1, and C11orf54 were examined using publicly available single-cell RNA-sequencing datasets of the developing human cerebral cortex accessed through the UCSC Cell Browser [40]. Uniform Manifold Approximation and Projection (UMAP) embeddings and cell-type annotations were obtained directly from the browser interface. Gene expression levels were visualized as color overlays on the UMAP plots to assess relative expression across major neural and non-neural cell populations during cortical development. Results Clinical manifestations of NDD in family PKMR06A Family PKMR06A represents a large, multigenerational, consanguineous kindred from a Saraiki-speaking community, with four affected individuals in three sibships (Fig. 1A). Clinical evaluation of the affected individuals revealed a strikingly uniform constellation of neurodevelopmental features (Table 1). Global developmental delay was present in these affected children. Sitting and crawling were markedly delayed, typically achieved between 1.4 and 4 years, and independent walking occurred even later, between 2 and 5 years (Table 1). Speech development was profoundly impaired across all affected individuals: none attained functional speech, and vocalizations were limited to babbling or unstructured sounds throughout childhood. ID fell within the moderate-to-severe range, particularly affecting attention, problem-solving, and overall adaptive functioning.Table 1Clinical and neurodevelopmental features of affected individuals in family PKMR06A.Table 1IndividualV:1V:2V:4V:5Ethnicity/originSaraiki/PakistaniHead circumference (cm)53.3448.9048.2653.34Age (years)129714GenderFemaleMaleMaleFemaleHeight (cm)157.48134.60106.68147.32Weight (kg)51281735Motor delaySat & crawled 4y; walked 5ySat & crawled 2y; walked 3ySat & crawled 1.4y; walked 2ySat & crawled 4y; walked 5ySpeech delayNo speechIntellectual/neurological delaymoderate to severeModeratemoderate to severemoderate to severeFacial dysmorphismTriangular foreheadBulbous nasal tipMild dysmorphism; rounded cheeks; Mild hypotelorismBulbous nasal tipHand/foot anomaliesNoneBlue at birthNoneBlue at birthLimb defectsNoneWeak limbs; knees bendLimb defect: weak limbs; knees bend while walkingToe-walking; knees bendBehavioral featuresFrequent laughingNoneNoneNoneVisionNormalNormalNormalNormalHearingNormalNormalNormalNormalHypotoniaNoneMildMildMildMusculoskeletalJoint laxityWeak jointsWeak limbs; knees bend while walkingWeak legs Table 1Clinical and neurodevelopmental features of affected individuals in family PKMR06A.Table 1IndividualV:1V:2V:4V:5Ethnicity/originSaraiki/PakistaniHead circumference (cm)53.3448.9048.2653.34Age (years)129714GenderFemaleMaleMaleFemaleHeight (cm)157.48134.60106.68147.32Weight (kg)51281735Motor delaySat & crawled 4y; walked 5ySat & crawled 2y; walked 3ySat & crawled 1.4y; walked 2ySat & crawled 4y; walked 5ySpeech delayNo speechIntellectual/neurological delaymoderate to severeModeratemoderate to severemoderate to severeFacial dysmorphismTriangular foreheadBulbous nasal tipMild dysmorphism; rounded cheeks; Mild hypotelorismBulbous nasal tipHand/foot anomaliesNoneBlue at birthNoneBlue at birthLimb defectsNoneWeak limbs; knees bendLimb defect: weak limbs; knees bend while walkingToe-walking; knees bendBehavioral featuresFrequent laughingNoneNoneNoneVisionNormalNormalNormalNormalHearingNormalNormalNormalNormalHypotoniaNoneMildMildMildMusculoskeletalJoint laxityWeak jointsWeak limbs; knees bend while walkingWeak legs Clinical and neurodevelopmental features of affected individuals in family PKMR06A. Neurological assessment revealed diffuse hypotonia, impaired balance, and gait abnormalities, including knee bending during ambulation, limb weakness, and, in some cases (V:5), toe-walking. One affected child (V:4) exhibited reduced temperature sensation in the lower limbs, consistent with mild sensory involvement (Table 1). Behavioral abnormalities were minimal, with only one affected (V:1) individual exhibiting frequent, unprovoked laughing. Craniofacial examination identified subtle but recurring dysmorphic features across affected individuals, consistent with their facial images (Fig. 1B). Varying facial dysmorphologies observed in the affected individuals included mild trigonocephaly, a bulbous nasal tip, rounded cheeks, an elongated philtrum, prominent upper incisors, and mild bitemporal narrowing. No visual or hearing deficits were recorded, aside from isolated findings such as mild hypotelorism or intermittent squint (Table 1). Additional systemic findings included joint laxity, particularly in the knees and ankles, and limb weakness, but no fixed contractures, limb malformations, or arthrogryposis. Two affected individuals (V:2 and V:5) experienced cyanosis at birth, presenting as transient blue discoloration of the lips, hands, and feet, with spontaneous resolution and no recurrence later in life. No abnormalities of the skin, hair, or dentition were identified. Skeletal, cardiac, and gastrointestinal examinations were otherwise unremarkable (Table 1). Taken together, the clinical symptoms support a syndromic recessively inherited neurodevelopmental disorder segregating in family PKMR06A. Moreover, brain MRI, EEG, nerve conduction studies, and formal standardized cognitive assessments were not available for the affected individuals. Therefore, ID severity was clinically estimated based on developmental history, speech, and adaptive-function observations, family-reported milestones, and neurological examination. No clinically reported seizure episodes were documented in the available history; however, the absence of EEG data limits assessment of subclinical epileptiform activity. Molecular analysis of NDD segregating in family PKMR06A Genetic analysis of family PKMR06A began with exome sequencing of the proband (V:1). Following quality control, population-frequency filtering, and prioritization of rare protein-altering variants, 77 variants were retained. Further filtering for homozygous variants compatible with the consanguineous pedigree reduced the candidate list to 11 variants. These variants were then evaluated based on predicted functional consequences, gene function, inheritance model, and relevance to the observed neurodevelopmental phenotype. Following phenotype-based prioritization and Sanger segregation analysis in available family members, CCDC82 (NM_024725.4): c.373del; p.(Asp125Ilefs*6) was the only variant that remained consistent with autosomal recessive inheritance and co-segregated with the neurodevelopmental phenotype. No other rare homozygous protein-altering variant with plausible relevance to the phenotype was identified (Fig. 1A). The identified variant consisted of a one-base pair deletion, c.373del, within the coding region of CCDC82 (Fig. 1C). The variant is predicted to result in a reading frameshift and a premature stop codon p.(Asp125Ilefs*6), leading to the removal of the critical coiled-coil domain and the carboxy tail from the resulting protein product (Fig. 1D). The c.373del variant was absent from major population genetic databases, including gnomAD [41], and the 1000 Genomes Project [42], suggesting that this deletion is extremely rare. We did note a single prior listing of the c.373del variant in the ClinVar database [43]; however, the clinical description of the affected individual is not provided. We also previously reported the same variant in another consanguineous Pakistani family [15,[26], [27], [28]] with a neurodevelopmental phenotype closely resembling that observed in PKMR06A, suggesting the possibility of a region-specific recurrent allele. However, the previous report did not provide a detailed clinical characterization of individuals harboring this specific CCDC82 variant. In the present study, we provide expanded phenotypic documentation, segregation analysis, and in silico structural and developmental-expression analyses for affected individuals from family PKMR06A. Thus, the current study should be interpreted as phenotypic refinement and additional support for the pathogenicity of a recurrent CCDC82 loss-of-function allele rather than discovery of a novel variant. Computational pathogenicity assessments were highly concordant with a loss-of-function effect. Frameshift variants often fall outside the optimal range of missense prediction tools; however, Mutation Taster classified the deletion as disease-causing, and CADD [44] yielded a high PHRED-scaled score of 24.2, placing the variant among the most deleterious predicted changes genome-wide (Table 2). Variant interpretation following ACMG/AMP guidelines supports classification of CCDC82 (NM_024725.4): c.373del; p.(Asp125Ilefs*6) as likely pathogenic/pathogenic based on PVS1, PM2, and PP1. PVS1 was applied because the variant is an early frameshift predicted to introduce a premature termination codon and remove the major predicted coiled-coil region of CCDC82. PM2 was applied because the variant is absent from major population databases, including gnomAD and 1000 Genomes. PP1 was applied because the variant segregates with the neurodevelopmental phenotype in the available family members under an autosomal recessive inheritance model.Table 2In silico analysis of identified CCDC82 variant.Table 2FamilyPKMR06AInheritanceHomozygous, autosomal recessiveGenBank accessionCCDC82 (NM_024725.4);GRCh37/hg19 (GCA_000001405.1)HGVS variant11:96117539delCcDNA variantc.373delProtein changep.(Asp125Ilefs*6)Variant typeFrameshift, loss-of-functionClinVar classificationUncertain significance (VUS), due to limited evidenceCADD24.2gnomAD frequency01000 Genomes0ACMG/AMP evidence criteriaPVS1, PM2, PP1Mutation TasterDisease-causing Table 2In silico analysis of identified CCDC82 variant.Table 2FamilyPKMR06AInheritanceHomozygous, autosomal recessiveGenBank accessionCCDC82 (NM_024725.4);GRCh37/hg19 (GCA_000001405.1)HGVS variant11:96117539delCcDNA variantc.373delProtein changep.(Asp125Ilefs*6)Variant typeFrameshift, loss-of-functionClinVar classificationUncertain significance (VUS), due to limited evidenceCADD24.2gnomAD frequency01000 Genomes0ACMG/AMP evidence criteriaPVS1, PM2, PP1Mutation TasterDisease-causing In silico analysis of identified CCDC82 variant. Structural and functional impact of the c.373del variant To explore the possible biological consequences of the homozygous CCDC82 frameshift variant, we performed several in-silico structural and gene-constraint analyses. The CCDC82WT model generated by I-TASSER had a C-score of −1.12, an estimated TM-score of 0.57 ± 0.14, an estimated RMSD of 10.1 ± 4.6 Å, and a template coverage of 0.98%. The predicted truncated p.(Asp125Ilefs*6) model had a C-score of −1.91, estimated TM-score of 0.49 ± 0.15, estimated RMSD of 8.6 ± 4.5 Å, and template coverage of 0.99%. Therefore, the structural analysis was used to illustrate the predicted loss of the C-terminal coiled-coil region caused by the frameshift variant, rather than to make definitive conclusions about altered protein folding or molecular interactions. Together, these computational analyses support the interpretation that the c.373del variant is likely to cause loss of CCDC82 function, although this predicted mechanism requires direct experimental validation (Fig. 2). The MetaDome profile across CCDC82 (Fig. 2A) shows that residue 125 lies within one of the most intolerant regions of the protein (red/orange). This intolerant interval overlaps precisely with the large coiled-coil domain and low-complexity domains essential for higher-order protein-protein interactions, which would be removed by the frameshift variant (Fig. 2B). The loss of the entire carboxy-terminal domain, combined with the disruption of a region under evolutionary constraint, strongly supports a deleterious effect of the c.373del variant. The mRNA harboring the c.373del variant, if it survived through nonsense-mediated mRNA decay, would encode a short polypeptide without a predicted coiled-coil and interaction domains that define CCDC82's functional architecture.Fig. 2Genetic variation, structural consequences, and developmental expression of CCDC82 and interacting genes. (A) MetaDome profile across the CCDC82 coding sequence. The p.(Asp125Ilefs*6) variant lies within a region of increased intolerance (red/orange) overlapping the coiled-coil domain, suggesting functional sensitivity to variation. The coiled-coil domain is marked in red, while the region deleted from the protein as a result of the p.(Asp125Ilefs*6) frameshift variant is indicated in pink. (B) Predicted 3D structural comparison between wild-type CCDC82 (544 aa) and the truncated protein produced by the p.(Asp125Ilefs*6) variant (131 aa). The premature stop codon eliminates the C-terminal coiled-coil domain and alters the local environment surrounding residue 125. (C) STRING-based predicted protein-protein interaction network centered on CCDC82, highlighting high-confidence interactors including LRRC2, GPALPP1, FAM76B, RFX5, SLC45A4, GOLG18, ZNF227, C11orf54, and TBC1D22B. These proteins cluster within transcriptional regulation, chromatin remodeling, and intracellular trafficking pathways. (D) Dot plot illustrating gene expression across key cell types in the developing human cerebral cortex (gestational weeks 14-25). Expression is shown for CCDC82 and its interactors, including LRRC2, GPALPP1, and C11orf54. The dot color shows the scaled average expression, while the dot size reflects the percentage of cells expressing each gene. Cajal-Retzius cells are highlighted, showing increased expression of CCDC82 and its interacting partners during early cortical development. Genetic variation, structural consequences, and developmental expression of CCDC82 and interacting genes. (A) MetaDome profile across the CCDC82 coding sequence. The p.(Asp125Ilefs*6) variant lies within a region of increased intolerance (red/orange) overlapping the coiled-coil domain, suggesting functional sensitivity to variation. The coiled-coil domain is marked in red, while the region deleted from the protein as a result of the p.(Asp125Ilefs*6) frameshift variant is indicated in pink. (B) Predicted 3D structural comparison between wild-type CCDC82 (544 aa) and the truncated protein produced by the p.(Asp125Ilefs*6) variant (131 aa). The premature stop codon eliminates the C-terminal coiled-coil domain and alters the local environment surrounding residue 125. (C) STRING-based predicted protein-protein interaction network centered on CCDC82, highlighting high-confidence interactors including LRRC2, GPALPP1, FAM76B, RFX5, SLC45A4, GOLG18, ZNF227, C11orf54, and TBC1D22B. These proteins cluster within transcriptional regulation, chromatin remodeling, and intracellular trafficking pathways. (D) Dot plot illustrating gene expression across key cell types in the developing human cerebral cortex (gestational weeks 14-25). Expression is shown for CCDC82 and its interactors, including LRRC2, GPALPP1, and C11orf54. The dot color shows the scaled average expression, while the dot size reflects the percentage of cells expressing each gene. Cajal-Retzius cells are highlighted, showing increased expression of CCDC82 and its interacting partners during early cortical development. Protein-protein interaction networks To explore possible functional contexts for CCDC82, we examined predicted protein-protein associations using the STRING database. Because STRING integrates known and predicted associations from multiple sources, this analysis should be interpreted as hypothesis-generating rather than evidence of direct physical interaction [45]. The STRING analysis suggested that CCDC82 may be functionally associated with proteins annotated in chromatin architecture, transcriptional control, intracellular trafficking, and kinase-mediated signaling pathways. These predicted associations provide a framework for future mechanistic studies but do not establish direct interaction or pathway involvement (Fig. 2C). High-confidence interactors included LRRC2, GPALPP1, FAM76B, RFX5, SLC45A4, GOLG18, ZNF227, C11orf54, and TBC1D22B, suggesting broad functional integration across nuclear and cytoskeletal processes. Among these predicted interactors, we selected LRRC2, GPALPP1, and C11orf54 for further analysis due to their highest association network prediction scores (0.597, 0.513, and 0.506, respectively). Cell-type-specific expression of CCDC82 and interacting genes during cortical development Next, we examined the expression of CCDC82 and the predicted interactors in the developing human brain. To explore the cellular context of CCDC82 in human cortical development, we examined single-cell RNA sequencing-based dot-plot visualizations of CCDC82, LRRC2, GPALPP1, and C11orf54 across key cortical cell types (Fig. 2D). CCDC82 exhibited the highest expression levels in early developmental cell populations, especially in Cajal-Retzius cells, radial glial and neuronal progenitor cells, whereas its expression in mature neuronal populations was relatively lower (Fig. 2D). Among the predicted interactors, GPALPP1 and LRRC2 also had overlapping enrichment in progenitor-associated cell types, while C11orf54 displayed a lower and more widely distributed expression pattern (Fig. 2D). The expression patterns of CCDC82, GPALPP1 and LRRC2 in the early-born cortical cell populations, which are primarily involved in neuronal migration and cortical patterning, underscore the role of CCDC82-interactome in early neurodevelopment. CCDC82 expression is enriched prenatally in the developing human brain To further explore CCDC82 expression and the predicted interactome, we analyzed the Human BrainSpan transcriptomic datasets [38]. CCDC82 showed significantly higher prenatal than postnatal expression across the BrainSpan dataset (237 prenatal samples; 287 postnatal samples; Mann-Whitney U test, p = 7.79 × 10−47; FDR-adjusted p = 3.12 × 10−46; rank-biserial correlation = 0.729; Fig. 3). C11orf54 also showed significant prenatal enrichment (FDR-adjusted p = 4.37 × 10−6; rank-biserial correlation = 0.236), while GPALPP1/KIAA1704 showed a smaller but significant prenatal enrichment (FDR-adjusted p = 0.0217; rank-biserial correlation = 0.116). In contrast, LRRC2 showed higher postnatal expression in this dataset (FDR-adjusted p = 6.50 × 10−30; rank-biserial correlation = −0.579). These findings provide a developmental expression context for CCDC82 and selected predicted interactors but do not establish causal involvement in neurodevelopmental mechanisms.Fig. 3Spatiotemporal expression pattern of CCDC82 and its interactors in the developing human brain. The BrainSpan analysis included 524 tissue-region samples: 237 prenatal samples from 20 donors and 287 postnatal samples from 22 donors. Developmental stages were defined as first trimester, 8-13 pcw; second trimester, 16-26 pcw; third trimester, 35-37 pcw; early postnatal, <2 years; childhood/adolescence, 2-<10 years; and adulthood, ≥10 years. Prenatal and postnatal expression levels were compared using the Mann-Whitney U test, with Benjamini-Hochberg correction for multiple testing across the four genes analyzed. Effect sizes were calculated using rank-biserial correlation. Columns represent the developmental stages spanning prenatal to adulthood, and rows depict various brain regions such as cortical, subcortical, and transient developmental areas, including the dorsal thalamus (DTH), mediodorsal thalamic nucleus (MD), striatum (STR), amygdala (AMY), hippocampus (HIP), cerebellum (CB), cerebellar cortex (CBC), medial, lateral, and caudal ganglionic eminences (MGE, LGE, CGE), occipital neocortex (Ocx), primary visual cortex (V1C), primary auditory cortex (A1C), primary somatosensory cortex (S1C), primary motor cortex (M1C), combined motor-somatosensory cortex (M1C-S1C), parietal neocortex (PCx), inferior temporal cortex (ITC), superior temporal cortex (STC), inferior parietal cortex (IPC), dorsolateral prefrontal cortex (DFC), orbital frontal cortex (OFC), medial frontal cortex (MFC), and ventrolateral prefrontal cortex (VFC). Color intensity reflects the mean log2(expression + 1), where darker shades denote higher expression. White cells show region-stage combinations that are absent from the BrainSpan dataset. Spatiotemporal expression pattern of CCDC82 and its interactors in the developing human brain. The BrainSpan analysis included 524 tissue-region samples: 237 prenatal samples from 20 donors and 287 postnatal samples from 22 donors. Developmental stages were defined as first trimester, 8-13 pcw; second trimester, 16-26 pcw; third trimester, 35-37 pcw; early postnatal, <2 years; childhood/adolescence, 2-<10 years; and adulthood, ≥10 years. Prenatal and postnatal expression levels were compared using the Mann-Whitney U test, with Benjamini-Hochberg correction for multiple testing across the four genes analyzed. Effect sizes were calculated using rank-biserial correlation. Columns represent the developmental stages spanning prenatal to adulthood, and rows depict various brain regions such as cortical, subcortical, and transient developmental areas, including the dorsal thalamus (DTH), mediodorsal thalamic nucleus (MD), striatum (STR), amygdala (AMY), hippocampus (HIP), cerebellum (CB), cerebellar cortex (CBC), medial, lateral, and caudal ganglionic eminences (MGE, LGE, CGE), occipital neocortex (Ocx), primary visual cortex (V1C), primary auditory cortex (A1C), primary somatosensory cortex (S1C), primary motor cortex (M1C), combined motor-somatosensory cortex (M1C-S1C), parietal neocortex (PCx), inferior temporal cortex (ITC), superior temporal cortex (STC), inferior parietal cortex (IPC), dorsolateral prefrontal cortex (DFC), orbital frontal cortex (OFC), medial frontal cortex (MFC), and ventrolateral prefrontal cortex (VFC). Color intensity reflects the mean log2(expression + 1), where darker shades denote higher expression. White cells show region-stage combinations that are absent from the BrainSpan dataset. Next, we focused on CCDC82 expression in the developing human cortex, an area primarily involved in cognitive function [46]. The single-cell RNA-seq analysis of the developing human cortex demonstrated that CCDC82 is expressed across multiple progenitor and neuronal lineages, including radial glia, intermediate progenitors, excitatory neurons, inhibitory neurons, oligodendrocyte precursor cells (OPCs), microglia, and endothelial cells (Fig. 4). This widespread expression indicates that CCDC82 is present in several neural and non-neural cell populations during cortical development, providing developmental context for the neurodevelopmental phenotype observed in affected individuals. Comparative expression maps of CCDC82 predicted interactors, LRRC2, GPALPP1, and C11orf54, revealed substantial co-expression across progenitor, excitatory, inhibitory, and glial compartments (Fig. 4). The overlapping developmental expression patterns of CCDC82 and selected predicted interactors suggest possible shared temporal or cellular contexts during cortical development. However, these transcriptomic data do not establish direct functional interaction or causality.Fig. 4Developmental brain expression of CCDC82 and interacting partners. Single-cell RNA-seq UMAP projection of the developing human cortex showing expression of CCDC82 and selected predicted interactors, LRRC2, GPALPP1, and C11orf54, across 49,494 cortical cells. Major cell populations are labeled in the CCDC82 panel: Exc, excitatory neurons; Inh, inhibitory neurons; RG, radial glia; OPC, oligodendrocyte precursor cells; Olig, oligodendrocytes; Astro, astrocytes; Endo, endothelial cells; and Micro, microglia. Color intensity represents relative normalized gene expression, ranging from low to high expression as indicated by the shared expression scale. UMAP axes represent dimension-reduction coordinates and do not correspond to physical anatomical distance. Developmental brain expression of CCDC82 and interacting partners. Single-cell RNA-seq UMAP projection of the developing human cortex showing expression of CCDC82 and selected predicted interactors, LRRC2, GPALPP1, and C11orf54, across 49,494 cortical cells. Major cell populations are labeled in the CCDC82 panel: Exc, excitatory neurons; Inh, inhibitory neurons; RG, radial glia; OPC, oligodendrocyte precursor cells; Olig, oligodendrocytes; Astro, astrocytes; Endo, endothelial cells; and Micro, microglia. Color intensity represents relative normalized gene expression, ranging from low to high expression as indicated by the shared expression scale. UMAP axes represent dimension-reduction coordinates and do not correspond to physical anatomical distance. Discussion Here, we describe a rare homozygous frameshift deletion in CCDC82 (NM_024725.4): c.373del; p.(Asp125Ilefs*6) in a large consanguineous Pakistani family (PKMR06A) segregating a syndromic intellectual developmental disorder characterized by global developmental delay, hypotonia, impaired motor coordination, severe speech impairment, and moderate to severe ID. The clinical picture is accompanied by subtle but reproducible craniofacial dysmorphism and mild joint laxity, with preserved vision and hearing. CCDC82 has previously been proposed as a candidate gene for autosomal recessive NDD; however, our findings provide additional genetic evidence and developmental expression context supporting the association of biallelic CCDC82 loss-of-function with autosomal recessive NDD. To clarify the contribution of the present study, we compared the clinical and molecular features of family PKMR06A with previously reported individuals harboring biallelic CCDC82 variants (Table 3). The c.373del; p.(Asp125Ilefs*6) variant identified in PKMR06A has been previously reported in a Pakistani consanguineous ID cohort, but detailed phenotypic characterization of affected individuals carrying this allele was limited. Other reported CCDC82-associated cases include truncating variants associated with ID, developmental delay, spasticity or spastic paraparesis, hypotonia, seizures/infantile spasms in some individuals, and dysmorphic features. The current study, therefore, does not claim discovery of a novel CCDC82 variant, but provides expanded clinical delineation of a recurrent Pakistani CCDC82 loss-of-function allele and refines the emerging phenotype associated with biallelic CCDC82 disruption.Table 3Comparison of reported individuals/families with biallelic CCDC82 variants.Table 3StudyPopulation/family structureCCDC82 variantVariant type/predicted effectReported phenotypeComparison with current family PKMR06ARiazuddin et al., 2017Large Pakistani consanguineous intellectual disability cohort; one large pedigree reported with affected individuals homozygous for CCDC82 c.373delNM_024725.4:c.373del; p.(Asp125Ilefs*6)Frameshift; premature termination; predicted loss of C-terminal region and coiled-coil domainIntellectual disability, delayed developmental milestones, speech delay, mild hypotonia, and motor weakness were reported briefly in the context of a large exome-sequencing study; detailed phenotypic delineation was limitedSame recurrent Pakistani CCDC82 c.373del allele. The current study provides expanded clinical characterization, segregation analysis, structural/domain interpretation, and developmental brain-expression contextHarripaul et al., 2018Consanguineous family from the Middle East with affected siblingsNM_024725.4:c.535C>T; p.(Arg179*)Nonsense; predicted loss-of-functionNonsyndromic intellectual disability reported in affected siblings; limited additional clinical detail availableSupports biallelic CCDC82 loss-of-function as a cause of recessive intellectual disability; PKMR06A shows a broader syndromic presentation with motor, speech, gait, and mild dysmorphic featuresYahia et al., 2022Sudanese consanguineous family with affected siblingsNM_024725.4:c.535C>T; p.(Arg179*)Nonsense; predicted loss-of-functionGlobal developmental delay, intellectual disability, spasticity, infantile spasms/seizures, and microcephalyDevelopmental delay and spasticity overlap with PKMR06A. Seizures and microcephaly appear more prominent in the Sudanese family, whereas PKMR06A shows severe speech impairment, gait abnormalities, joint laxity, and mild dysmorphismBauer et al., 2022Individual born to consanguineous Pakistani parentsNM_001318736.1:c.183del; p.(Phe61Leufs*27)Frameshift; early premature termination; predicted loss-of-functionIntellectual disability, developmental delay, short stature, spastic paraparesis beginning in early childhood, and dysmorphic features; suspected gelastic seizures were not confirmed on repeated EEGsStrong clinical overlap with PKMR06A, particularly intellectual disability, motor impairment/spasticity, and dysmorphic features. The current family adds an intrafamilial comparison of multiple affected individuals carrying the recurrent c.373del alleleSafarian et al., 2025Consanguineous family with severe early-onset neurodevelopmental phenotypeNM_024725.4:c.709C>T; p.(Arg237*)Nonsense; predicted loss-of-functionSevere neurodevelopmental disorder with infantile spasms and hypotoniaHypotonia and developmental delay overlap with PKMR06A. Infantile spasms represent an additional neurological feature not documented in the current familyCurrent study: PKMR06AConsanguineous Pakistani family from Southern Punjab; four affected individuals reported, andclinically evaluated in detailNM_024725.4:c.373del; p.(Asp125Ilefs*6)Frameshift; premature termination; predicted loss of major C-terminal coiled-coil regionGlobal developmental delay, moderate-to-severe intellectual disability, profound speech impairment, hypotonia, spasticity, gait abnormalities, knee bending during ambulation, toe-walking in one individual, mild dysmorphic features, joint laxity, transient neonatal cyanosis in two individuals, and reduced temperature sensation in one individualProvides detailed phenotypic expansion for the recurrent Pakistani c.373del allele and supports CCDC82 loss-of-function as a cause of syndromic autosomal recessive neurodevelopmental disorder Table 3Comparison of reported individuals/families with biallelic CCDC82 variants.Table 3StudyPopulation/family structureCCDC82 variantVariant type/predicted effectReported phenotypeComparison with current family PKMR06ARiazuddin et al., 2017Large Pakistani consanguineous intellectual disability cohort; one large pedigree reported with affected individuals homozygous for CCDC82 c.373delNM_024725.4:c.373del; p.(Asp125Ilefs*6)Frameshift; premature termination; predicted loss of C-terminal region and coiled-coil domainIntellectual disability, delayed developmental milestones, speech delay, mild hypotonia, and motor weakness were reported briefly in the context of a large exome-sequencing study; detailed phenotypic delineation was limitedSame recurrent Pakistani CCDC82 c.373del allele. The current study provides expanded clinical characterization, segregation analysis, structural/domain interpretation, and developmental brain-expression contextHarripaul et al., 2018Consanguineous family from the Middle East with affected siblingsNM_024725.4:c.535C>T; p.(Arg179*)Nonsense; predicted loss-of-functionNonsyndromic intellectual disability reported in affected siblings; limited additional clinical detail availableSupports biallelic CCDC82 loss-of-function as a cause of recessive intellectual disability; PKMR06A shows a broader syndromic presentation with motor, speech, gait, and mild dysmorphic featuresYahia et al., 2022Sudanese consanguineous family with affected siblingsNM_024725.4:c.535C>T; p.(Arg179*)Nonsense; predicted loss-of-functionGlobal developmental delay, intellectual disability, spasticity, infantile spasms/seizures, and microcephalyDevelopmental delay and spasticity overlap with PKMR06A. Seizures and microcephaly appear more prominent in the Sudanese family, whereas PKMR06A shows severe speech impairment, gait abnormalities, joint laxity, and mild dysmorphismBauer et al., 2022Individual born to consanguineous Pakistani parentsNM_001318736.1:c.183del; p.(Phe61Leufs*27)Frameshift; early premature termination; predicted loss-of-functionIntellectual disability, developmental delay, short stature, spastic paraparesis beginning in early childhood, and dysmorphic features; suspected gelastic seizures were not confirmed on repeated EEGsStrong clinical overlap with PKMR06A, particularly intellectual disability, motor impairment/spasticity, and dysmorphic features. The current family adds an intrafamilial comparison of multiple affected individuals carrying the recurrent c.373del alleleSafarian et al., 2025Consanguineous family with severe early-onset neurodevelopmental phenotypeNM_024725.4:c.709C>T; p.(Arg237*)Nonsense; predicted loss-of-functionSevere neurodevelopmental disorder with infantile spasms and hypotoniaHypotonia and developmental delay overlap with PKMR06A. Infantile spasms represent an additional neurological feature not documented in the current familyCurrent study: PKMR06AConsanguineous Pakistani family from Southern Punjab; four affected individuals reported, andclinically evaluated in detailNM_024725.4:c.373del; p.(Asp125Ilefs*6)Frameshift; premature termination; predicted loss of major C-terminal coiled-coil regionGlobal developmental delay, moderate-to-severe intellectual disability, profound speech impairment, hypotonia, spasticity, gait abnormalities, knee bending during ambulation, toe-walking in one individual, mild dysmorphic features, joint laxity, transient neonatal cyanosis in two individuals, and reduced temperature sensation in one individualProvides detailed phenotypic expansion for the recurrent Pakistani c.373del allele and supports CCDC82 loss-of-function as a cause of syndromic autosomal recessive neurodevelopmental disorder Comparison of reported individuals/families with biallelic CCDC82 variants. The identified variant introduces a premature termination codon only six residues downstream from the deletion, truncating the protein at residue 131 and eliminating more than 60% of the full-length 544 amino acid sequence (Figs. 1D and 2B). This early truncation removes all predicted C-terminal coiled-coil motifs, which together define the structural architecture and interaction capacity of CCDC82. Proteins containing coiled-coil domains are widely recognized for their structural and regulatory roles in the cytoskeleton, intracellular transport, transcriptional modulation, and synaptic development [47]. As a result, disruption of coiled-coil architecture, whether through missense variation, premature truncation, or nonsense-mediated decay, typically produces profound neurodevelopmental consequences [48], similar to the broad neurodevelopmental phenotype observed in PKMR06A. To place our findings in the context of previously reported cases, we compared the genetic and clinical features of individuals harboring pathogenic CCDC82 variants. All pathogenic CCDC82 variants reported to date, including the frameshift variant identified in this study, are truncating alleles predicted to result in loss of function. Across all reported families, affected individuals show overlapping neurodevelopmental features, most prominently developmental delay and hypotonia, with variable additional neurological manifestations such as seizures or infantile spasms [17,18]. The recurrence of truncating variants and the shared clinical features across unrelated families support a loss-of-function disease mechanism for CCDC82-associated NDDs. The potential functional context of CCDC82 in neurodevelopment is further suggested by its predicted interaction landscape and expression pattern. The STRING-derived network identified predicted associations between CCDC82 and LRRC2, GPALPP1, FAM76B, RFX5, SLC45A4, GOLG18, ZNF227, C11orf54, and TBC1D22B (Fig. 2C). These partners include SSRP1 and SUPT16H, core components of the FACT complex required for nucleosome destabilization and transcriptional regulation [[49], [50], [51]], as well as casein kinase II subunits (CSNK2A1, CSNK2A2, CSNK2B) and RALBP1, which have roles in neuronal signaling, vesicle dynamics, and protein turnover [[52], [53], [54], [55]]. These predicted associations suggest that CCDC82 may participate in nuclear, perinuclear, or cytoplasmic regulatory processes relevant to neurodevelopment. However, because these interactions were inferred computationally, they should be considered hypotheses that require experimental confirmation. Our combined single-cell and bulk transcriptomic analyses indicate that CCDC82 is preferentially expressed during early human brain development. Single-cell data demonstrated higher expression of CCDC82 in Cajal-Retzius cells and neural progenitor populations (Fig. 2D), cell types essential for early cortical structuring and neuronal migration. Consistently, BrainSpan bulk RNA-sequencing showed higher prenatal expression of CCDC82 across multiple brain regions, followed by a postnatal decline (Fig. 3). Among the predicted interactors, GPALPP1 shows a similar developmental trajectory, whereas LRRC2 and C11orf54 display lower and more restricted expression patterns. Single-cell RNA-seq data from developing human cortex showed that CCDC82 is broadly expressed across radial glia, intermediate progenitors, excitatory and inhibitory neurons, oligodendrocyte precursor cells, microglia, and endothelial cells (Fig. 4). The predicted interactors also display overlapping expression patterns in neurogenic zones (Fig. 4), reinforcing the idea that CCDC82 participates in shared developmental pathways essential for corticogenesis and neuronal maturation. In addition, the greater overlap in expression between CCDC82 and GPALPP1 or C11orf54 likely reflects shared spatial and/or temporal expression patterns, suggesting coordinated regulation or participation in related cellular processes. Together, these findings provide developmental context for CCDC82-associated disease by showing that CCDC82 is expressed during prenatal brain development and in cell populations relevant to cortical development. However, whether CCDC82 loss directly disrupts corticogenesis, neuronal maturation, or related developmental pathways remains to be tested experimentally. The c.373del; p.(Asp125Ilefs*6) variant identified in family PKMR06A has been reported previously in another consanguineous Pakistani family and is also listed in ClinVar [43]. Therefore, the present study does not represent the discovery of a novel allele but rather provides independent replication of the association between biallelic CCDC82 loss of function and syndromic NDD. Because haplotype analysis was not performed, we cannot determine whether the recurrence of c.373del in Pakistani families reflects a founder allele or an independent mutational event. Nevertheless, the absence of this variant from major population databases and its recurrence in consanguineous Pakistani families raise the possibility of a region-specific founder mutation, which should be tested in future studies using haplotype or genome-wide SNP-based analyses. Several limitations should be acknowledged. First, the proposed loss-of-function mechanism is based on genetic segregation, predicted premature truncation, domain loss, population frequency data, and in silico analyses, but was not directly validated experimentally. Patient-derived RNA, protein lysates, or viable cells were not available; therefore, we could not determine whether the mutant transcript undergoes nonsense-mediated decay or whether CCDC82 protein expression is reduced or absent in affected individuals. Second, the functional consequences of CCDC82 deficiency were not tested in patient-derived cells or animal systems. Future studies should assess mutant transcript stability using RT-PCR or quantitative PCR, evaluate CCDC82 protein abundance and localization by immunoblotting or immunofluorescence, and investigate the neurodevelopmental consequences of CCDC82 loss in cellular or in vivo models. Therefore, while our data strongly support pathogenicity of the c.373del variant, the precise molecular mechanism remains to be experimentally defined. In summary, our data strongly support a pathogenic role for the c.373del frameshift variant in CCDC82 and provide a coherent framework linking early protein truncation, loss of structural domains, altered interaction networks, and syndromic autosomal recessive NDD. CRediT authorship contribution statement Tehmeena Akhter: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Methodology, Formal analysis, Data curation, Conceptualization. Sami Khan: Writing – original draft, Software, Methodology, Formal analysis, Data curation. Ahmed Abdul Mumeen: Writing – original draft, Software, Methodology, Formal analysis, Data curation. Muhammad Asif Naeem: Methodology, Data curation. Sheikh Riazuddin: Writing – review & editing, Validation, Supervision, Investigation, Conceptualization. Saima Riazuddin: Writing – review & editing, Writing – original draft, Validation, Supervision, Resources, Investigation, Funding acquisition, Conceptualization. Ethics declaration Written informed consent, including consent to publish photographs, was obtained from all participating individuals or their legal guardians. This study adheres to the World Medical Association Declaration of Helsinki and was approved by the Institutional Review/Ethic Boards of all participating institutions, including the University of Maryland, Baltimore, USA (IRB: HP-00075913; renewed on 06/19/2017), and the Centre of Excellence in Molecular Biology at the University of the Punjab, Lahore, Pakistan. Funding statement This work was supported by 10.13039/100000065NINDS/NIH grant R01NS107428 (S.R.). Declaration of competing interests The authors declare no conflict of interest.