Introduction

Neurodevelopmental delay (NDD) is a term used to describe problems in skill development in infants or children by comparison to typical patterns.[1]Brain volume is one of the important indexes used to evaluate NDD. Significant brain volume loss and gain (microcephaly and macrocephaly, respectively) usually indicate severe NDD with poor prognoses. Along with trauma, brain tumors, and inflammation of the central nervous system, genetic predisposition is one of the dominant etiologies of NDD. The Online Mendelian Inheritance in Man (OMIM) database lists 637 genes that can cause a phenotype of microcephaly and 178 genes that can cause macrocephaly. Researchers also found that risk genes or single nucleotide polymorphisms (SNPs) may be related to changes in the volume of specific brain regions in some neurological diseases.[2,3] The complex genetic background of NDD patients has not been thoroughly studied, especially for those with indeterminate etiologies.[4]

Head circumference is a convenient and routine method to measure brain volume in clinical practice. Children with a head circumference more than two standard deviations (SDs) above or below the norm are recommended for further neurological evaluation and necessary treatment. New advances in radiology have provided methods to estimate brain volume with greater sensitivity and precision than what previous methods could achieve.[5,6] Studies have found that magnetic resonance imaging (MRI)-based brain volume calculation may be the better imaging-based technique to describe the development of the brain in both neonates and children.[5,6] It can also provide real-time analyses of brain function. For NDD patients without unknown etiologies, brain volume calculated by MRI may be a potentially vital method to describe the clinical phenotype spectrum or prognoses in detail. These patients may have unique genetic backgrounds and novel molecular mechanisms. Therefore, in this study, we first established a reference curve for whole-brain volume based on the Han Chinese population. Then, we explored the clinical and genetic characteristics of NDD patients with significant whole-brain volume deviation (WBDV) by comparing them to the reference curve.

Methods Ethical approval

The study protocol was approved by the Ethics Committee of the Children's Hospital of Fudan University (Aproval No. 2017-161) before the study began. The samples used in this study were collected with the appropriate informed consent of the patients’ parents.

Study population

From January 2018 to December 2019, we included patients diagnosed with NDD disorders according to the Bayley-III scale. Clinical diagnoses were evaluated by two clinical specialists in succession. Patients with possible acquired etiologies, such as metabolic disease, congenital infection, meningitis, hemorrhage, ischemic stroke, hemorrhagic stroke, or trauma, were excluded. Patients with positive genetic diagnoses were also excluded.

MRI data acquisition and volume calculation

From January 2018 to December 2019, we recruited subjects who underwent cranial MRI in the Children's Hospital of Fudan University. They underwent an MRI test to exclude neurological problems. The inclusion criteria were as follows: (1) no diagnosed neurological diseases and (2) no positive findings in the MRI report. The exclusion criteria were as follows: (1) brain injury, meningitis, cranial hemorrhage, stroke, brain tumor, brain abnormalities, or hydrocephalus; (2) long-term ventilation therapy; (3) congenital heart disease requiring surgical therapy; and (4) low-quality MRI data. We divided those subjects by sex and divided subjects of each sex into subgroups according to the standard curve of head circumference published by the World Health Organization (https://www.who.int/tools/child-growth-standards/standards/head-circumference-for-age). Each subgroup contained at least three subjects. After MRI quality control and screening, 4222 subjects ranging in age from 1 day to 18 years were included. We defined ± 2 SDs of the reference brain volume as the boundary points; subjects outside this range were considered to have significant WBDV.

Raw images in the Digital Imaging and Communications in Medicine format were extracted from the Picture Archiving and Communication Systems databases of Children's Hospital of Fudan University by entering the patients’ identification codes as queries. Images taken in the same modality on the same scanners were merged into NIfTI-1 (https://nifti.nimh.nih.gov/nifti-1) files by dcm2niix (https://github.com/rordenlab/dcm2niix).[7] Due to the low contrast within the T1 modality for neonates and the high contrast for older children, we chose the T2 modality for further image analyses. The T2-weighted sequence was set as follows: time of repetition = 5000 ms, time of echo = 89 ms, and field of view = 220 mm × 220 mm. NIfTI files of the T2 data were processed by denoising the images and stripping away the skull to determine the brain regions. The volumes of the brain regions were quantified by multiplying the voxel numbers by the image resolution. The final output of the volume was shown in cm3. The volume computation was obtained using the Advanced Normalization Tools in R (ANTsR) package within the R environment (https://github.com/ANTsX/ANTsR) and Functional Magnetic Resonance Imaging of the Brain Software Library (FSL) (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki).[8] This is a rough measurement of brain regions that includes parts of the skull and cerebrospinal fluid but is sufficient to represent the head circumference. Data from the 1.5 T (Siemens MAGNETOM Avanto, Berlin, Germany) and 3 T (GE Discovery MR750, Atlanta, United States) systems were included. All data were changed into dcm2niix to unite the format.

Genetic sequencing and variants curation

The genetics department cooperated with the patients’ families to decide which type of sequencing to perform. A clinical exome sequencing (CES) panel of 2742 genes was generated using the Agilent ClearSeq Inherited Disease Kit (Agilent Technologies, Santa Clara, CA) and the Illumina Cluster and SBS Kits (Illumina Inc., San Diego, CA, USA). Sequencing was performed on an Illumina HiSeq 2000/2500 platform (Illumina Inc., San Diego, CA, USA). Clean reads were aligned to the reference human genome (UCSC hg19) by the Burrows–Wheeler Aligner (BWA, v.0.5.9-r16, http://bio-bwa.sourceforge.net/). GATK4 Best Practices (https://gatk.broadinstitute.org/hc/en-us/articles/360036194592-Getting-started-with-GATK4) were employed for variant calling. Annotation of variants was performed using a previously published pipeline.[9] The mutation type was annotated by the Ensemble Variant Effect Predictor [10] and ANNOtate VARiation.[11] Missense variants were assessed within silico prediction programs Sorting Intolerant From Tolerant (SIFT) (https://sift.bii.a-star.edu.sg/),[12] PolyPhen-2 (http://genetics.bwh.harvard.edu/pph2/)[13] and MutationTaster (https://www.mutationtaster.org/).[14] The pathogenicity of variants was curated according to the American College of Medical Genetics and Genomics criteria.[15] For copy number variant (CNV) detection, we performed and CNVs with an Arbitrary Number of Exome Samples (CANOES) (https://github.com/ShenLab/CANOES).[16] For small CNVs (<1 Mb), we ordered real-time quantitative polymerase chain reaction or multiplex ligation-dependent probe amplification for verification. We considered large CNVs (>1 Mb) to be highly credible and did not perform further verification. Each case was evaluated by two geneticists in succession.

Genetic burden analysis

For mutation burden analysis, we grouped low-allele-frequency variants (minor allele frequency <2.5%) into four types: protein-truncating variants (PTVs), missense or non-synonymous variants (MISs), synonymous variants (SYNs), and non-coding variants (NONs) [Supplementary Table 1, https://links.lww.com/CM9/B292]. Next, we summarized the number of variants within each type of mutation in each gene of each sample. Then, Fisher's exact test was performed to test whether the number of variants from each mutation type was significantly higher in subjects with WBDV than in those without WBDV. The PTV and MIS variant tests were used to identify brain volume-related burden genes. Genes with significant differences found at the SYN or NON-level were filtered out. The threshold of significance was P < 0.05. All analyses were performed using R software (version 3.6.0, http://cran.r-project.org).

Literature review analyses

We performed the literature review analyses based on PubMed database. First, we searched the gene name and “brain” as key words in the PubMed database using the easyPubMed R package (e.g., COL4A2[Title/Abstract] AND brain[Title/Abstract]) and recorded the PubMed IDs for each gene. Second, we downloaded the gene2pubmed files from ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2pubmed.g, which contain the relationship between species, PubMed ID and gene. We restricted the species to Homo sapiens (taxid: 9606) so that, for each gene, we could obtain a list of PubMed IDs wherein the gene was reported in humans. By finding the intersection of the two lists of PubMed IDs, we obtained the list of papers for each gene that was reported in humans and related to the brain. Finally, we identified the genes with more than five studies as reliable candidate genes.

BrainSpan analysis based on the human brain developmental transcriptome

We downloaded human RNA sequencing (RNA-seq) data from BrainSpan, the largest genomic dataset of standard postmortem brain samples (http://brainspan.org/).[17] This database includes 524 brains with 16 anatomical brain region tissue samples. The donors’ ages ranged from eight postconceptional weeks to 40 postnatal years. We constructed a brain development-specific signaling interactome by SJARACNe (https://github.com/jyyulab/SJARACNe)[18] and performed the BrainSpan analysis. A functional interaction network [Supplementary Figure 1, https://links.lww.com/CM9/B292] of genes was constructed by the ReactomeFI plugin in Cytoscape (https://cytoscape.org/). The dataset from GSE104276 and related meta-information was downloaded from GEO database. R package Seurat (https://cran.r-project.org/web/packages/Seurat/index.html) and MySeuratWrappers (https://github.com/satijalab/seurat-wrappers) were used to draw the volcano plot for genes of interest in different cell types, grouped by weeks of age.

Statistical analysis

We established a reference curve of whole-brain volume from the first postnatal day to 18 years of age and then calculated the 95% reference range. We set boundaries at 2 SDs above and below the reference curve for brain volume; individuals outside theses boundaries were considered to have significant WBDV. Fisher's exact test was used to detect differences in categorical variables between different groups. P < 0.05 was considered to be statistically significant. Statistical analysis was conducted by R 4.0.3 (https://cran.r-project.org/bin/windows/base/old/).

Results Study population

In total, 253 patients diagnosed with NDD were collected [Figures 1 and 2]. These patients aged from 6 months to 15 years, including 177 males and 76 females. We established both male and female reference curves of brain volume tested by MRI with an age span from first postnatal day to 18 years. By comparing subjects to the reference curve, we identified 213 cases without significant WBDV, 26 cases with significant negative WBDV (<−2 SDs), and 14 cases with significant positive WBDV (>+2 SDs). Among them, two cases had a severe positive WBDV (>+3 SDs). In the study sample, there were two patients with microcephaly, both of whom were evaluated as having significant negative WBDV; there was also one individual with macrocephaly, and that patient was found to have significant positive WBDV. This indicates the consistency of brain volume calculations with clinical diagnoses.

Figure 1:

Workflow of this study. The blue box is the disease-causing gene analysis of the test population. The pink box is the method to detect burden genes related to changes of brain volume. NDD: Neurodevelopmental delay; SDs: Standard deviations; WBDV: Whole-brain deviation volume.

Figure 2:

Reference curve of brain volume based on MRI. One blue dot means one sample. We established male (A) and female (B) brain volume separately. MRI: Magnetic resonance imaging.

Clinical features of NDD patients with significant WBDV

To identify the differences in clinical features between patients with and without significant WBDV, we acquired and compared the neurological features of these groups; we found that patients with WBDV had higher rates of motor development delay (49.8% [106/213] vs. 75.0% [30/40], P = 0.003) [Supplementary Table 2, https://links.lww.com/CM9/B292]. In contrast, the rate of speech/language development delay, a dominant feature in this population, was not significantly different between the two groups (70% [150/213] vs. 70% [28/40], P = 1.000). Additionally, no significant difference was detected in the rate of intellectual disability (8% [17/213] vs. 18% [7/40], P = 0.076). Next, we compared three neurological phenotypes between the positive WBDV group and the negative WBDV group; no significant differences were detected [Table 1].

Table 1 - Neurodevelopment of NDD patients with WBDV. Brain volume of WBDV patients NDD characteristics <−2 SDs (n = 26) >+2 SDs (n = 14) P value Motor development delay 21 (79) 9 (50) 0.278 Speech/language development delay 17 (66) 11 (78) 0.484 Intellectual disability 5 (21) 2 (22) 1.000

Data are presented as n (%). NDD: Neurodevelopmental delay; SDs: Standard deviations; WBDV: Whole-brain deviation volume.

Genetic burden analyses

To investigate the potential reason for different phenotype spectrums between patients with or without WBDV, we performed a genetic burden test to compare the allele frequencies (AFs) of rare variants. As missense and truncating mutations have different effects on the resulting protein, we calculated their AFs separately. The results showed that 30 burden genes had elevated AF in cases with significant WBDV (P < 0.05) [Supplementary Table 3, https://links.lww.com/CM9/B292]. Among these genes, three genes had increased AFs of PTV variants in the WBDV group, and 27 had increased AFs of MIS variants in that group. COL4A2 was the top significant gene for PTV variants, and PDE8B was the top significant gene for MIS variants. These two genes have been reported to regulate brain development.

To test whether burden genes could regulate human brain volume, we performed a literature review [Supplementary Figure 1, https://links.lww.com/CM9/B292]. We obtained lists of microcephaly-related (n = 637) and macrocephaly-related (n = 178) genes from the OMIM database as a positive control gene set [Supplementary Table 4, https://links.lww.com/CM9/B292]. The full set of test genes in the CES panel (3203 genes) constituted the background gene set. By Fisher's exact test, we found that those positive control genes were identified in more brain-related studies than background genes (P = 8.89e–10), and burden genes were also more related to the brain than background genes (P = 1.656e–9). No significant difference was detected between the positive control genes and the burden genes (P = 0.674). These results showed that the burden genes were not randomly selected, and had been linked to the human brain by previous studies, suggesting their potential role in brain development.

Function of burden gene in motor development

As we found more patients with significant WBDV presented motor development delay, we investigated the expression pattern of these burden genes in motor development-related brain regions during different prenatal and postnatal stages. Based on BrainSpan database, we included primary motor cortex, primary somatosensory cortex, dorsolateral prefrontal cortex, ventrolateral prefrontal cortex, striatum, hippocampus, and cerebellar cortex. Most of these burden genes had higher expression level before 37-week gestational age than postnatal stages in these brain regions [Supplementary Figure 2, https://links.lww.com/CM9/B292]. This observation may indicate the vital function of these genes in early brain development. To identify the function of these burden genes, we performed a gene functional network analysis and found that those burden genes were enriched in embryonic brain development, positive regulation of synaptic growth at the neuromuscular junction, positive regulation of DNA-templated transcription, and response to hormone [Figure 3]. These results indicated the potential function of burden genes in early neuroevelopment. To identify the function of the gene set related to embryonic brain development and positive regulation of synaptic growth at the neuromuscular junction in neural progenitor cells and neurons, we analyzed the single cell RNA-seq data from human prefrontal cortex in the embryonic stage. In embryonic brain development, the TUBB2B gene had elevated expression levels in neural progenitor cells, interneuron, and excitatory neuron [Figure 4A]. TUBB2B mutation is known to cause microcephaly.[19] By analyzing the genes associated with positive regulation of synaptic growth at the neuromuscular junction, we found that five genes (CNNM4, MOGS, RFX5, SOX15, and ARGN) have elevated expression levels in stem cells. Among them, SOX15 also had high expression in interneuron and excitatory neuron [Figure 4B]. These results indicate the potential function of burden genes in the early stages of brain development.

Figure 3:

The functional network analysis of burden genes. Gene functional network analysis found burden genes were enriched in embryonic brain development, positive regulation of synaptic growth at the neuromuscular junction, positive regulation of DNA-templated transcription, and response to hormone. Gene in the circle were burden genes. Gene in red were linker genes. P value of each term was showed in box.

Figure 4:

The relative expression level of burden genes in embryonic neuronal progenitor cell, excitatory neuron, and interneuron. Relative expression level of genes of embryonic brain development (A) and positive regulation of synaptic growth (B) at the neuromuscular junction terms in three kinds of cells (neural progenitor, excitatory neuron, and interneuron) in human embryonic stage from gestational weeks 8 to 26.

Discussion

This study described the genetic background of patients with idiopathic NDD. We found that patients with brain volume changes are more likely to have motor development delay. This may indicate that those patients had different characteristics compared to NDD patients without whole-brain volume change. In addition, the whole brain volume calculated by MRI may be be a crucial predictor of the severity and phenotype of NDD patients. Physicians could pay more attention to the motor development milestones of those patients with significant WBDV. Early diagnoses and treatment could help to improve the patients’ prognosis.

Several functional brain structures, including the motor cortex and cerebellum participate in motor regulation. Motor learning is a complex process subserved by numerous regions, circuits, and networks, including the frontal lobe, temporal lobe, hippocampus, and white matter association tracts.[20,21] We analyzed the expression of burden genes in several brain regions reported to regulate motor behavior (primary motor cortex, primary somatosensory cortex, cerebellar cortex), motor learning (hippocampus), and motor planning (striatum and ventrolateral prefrontal cortex).[22,23] We also considered the dorsolateral prefrontal cortex, which connects to the primary motor cortex to regulate muscle movement.[24] We found that most burden genes had elevated expression in the embryonic stage. They were also enriched in embryonic brain development and positive regulation of synaptic growth at the neuromuscular junctions. These results indicate the important function of burden genes in early brain development and their potential to affect its outcome significantly. However, the motor-related brain regions considered in this study are not a comprehensive set; there are other brain regions that participate in motor function but are not included in the BrainSpan database. Therefore, the potential functions of risk genes still need further investigation. For further analyses, functional MRI would be helpful to map the regulation of motor function in detail. Subregion volume analyses in patients could also help to identify the candidate brain regions responsible for gross vs. fine motor development.

When we analyzed the functions of burden genes in different cell types, we found that the TUBB2B and SOX15 genes had higher expression levels in neural progenitor cells than other burden genes. As reported in the OMIM database, TUBB2B mutation can cause cortical dysplasia, complex, with other brain malformations type 7 [OMIM: 610031], which is inherited in an autosomal dominant manner. Patients with pathogenic variants of TUBB2B may present microcephaly, abnormality of the basal ganglia, dysgenesis of the corpus callosum, and dysplasia of the cerebellar vermis.[19,25] Pathogenic variants of TUBB2B can impair the migration of neurons in the mouse cortex.[26] The function of SOX15 in the central nervous system is unknown. Our results provide further evidence that TUBB2B and SOX15 are essential for the volumetric growth of the brain during development. More molecular investigation is necessary.

The top two burden genes were also related to motor development. COL4A2 encodes the collagen type IV alpha 2 chain, which is essential for the formation of sheet-like basement membranes. As reported in the OMIM database, COL4A2 mutation is responsible for brain small vessel disease 2 [OMIM: 614483]. This inherited disease is characterized by disturbed vascular supply and cerebral degeneration; the phenotype can appear as early as childhood.[27,28] Patients can present mental retardation and reduced white matter volume.[29] Variable movement abnormalities are the dominant feature of COL4A2-related hereditary disease.[30] In addition to pathogenic variants, intragenic duplication of COL4A2 can lead to delayed motor development.[31] In COL4A2 heterozygous mouse models, the mouse presents congenital cortical lamination defects and reduced thickness of the cortical layers.[32]PDE8B is another top gene associated with significant WBDV. PDE8B encodes a cyclic nucleotide phosphodiesterase that catalyzes the hydrolysis of the second messenger cAMP. Additionally, OMIM notes that mutation of this gene can cause striatal degeneration [OMIM: 609161]. SNPs of PDE8B are associated with serum thyroid-stimulating hormone (TSH) levels.[33] TSH is known to be essential for intellectual development. This information could help explain the relatively high rates of motor impairment and intellectual disability in patients with WBDV. In addition, PDE8B is expressed in the hippocampus, ventral striatum, and cerebellum in mice. Mice with PDE8B gene inactivation (PDE8B knockout) presented an age-induced decline in motor coordination.[34] Gene burden analysis can help clarify the underlying genetic risk factors leading to disease susceptibility. This method can overcome several shortcomings, including large multiplex families, locus heterogenicity, and incomplete penetrance, to identify complex genetic causes of disease.[35,36] Gene burden analysis can also provide clues to inform further mechanistic investigation.

In this study, we chose MRI-based calculations as a measure of brain volume. Although head circumference measurement is an efficient way to detect brain volume in clinical practice, some mild head circumference changes are inconspicuous enough to go unnoticed by physicians. It is difficult for physicians and radiologists to detect abnormalities in the brain if the quantity of brain tissue matches the volume of the skull or or if there is no dramatic change in the volume of cerebrospinal fluid. MRI examination can provide detailed information about brain structure and function. It is suitable for pediatric patients and fetuses suspected of detecting a neurological phenotype. However, brain volume calculations based on MRI have not been widely used in clinical practice, especially in children's hospitals. This may be due to the lack of age-dependent standard curves and automated algorithms. Herein, we attempted to establish a reference curve from 4222 samples representing a range of ages from the neonatal period to 18 years. A more accurate and reliable standard curve needs to be constructed from a well-established population, and the algorithm also needs to be updated for rapid calculation. Additionally, data from more NDD patients are needed to further the clinical application of brain volume calculations in diagnosis.

As exome sequencing is currently more cost-effective than whole-genome sequencing in clinical practice, the sequencing data applied in this study are all exome-based. However, exome sequencing does not cover intron regions and misses a large portion of information about regulatory regions. Therefore, the results of gene burden analysis would be more accurate if genome-wide sequencing data were applied.

In summary, we uncovered the clinical features and genetic backgrounds of NDD patients without diagnosed etiologies. The results indicated differences in genetic characteristics and burden genes between patients with and without significant genetic WBDV, which may help to explain the difference in the phenotype spectrum.

Acknowledgements

We appreciate the patients and the families’ willingness and cooperation in the study. Thanks for the efforts of many colleagues and staff.

Funding

This work was supported by grants from the Science and Technology Commission of Shanghai Municipal (No. 19411964400), Shanghai Municipal Science and Technology Major Project (No. 2018SHZDZX01), and ZJLab.

Conflicts of interest

None.

References 1. Villagomez AN, Muñoz FM, Peterson RL, Colbert AM, Gladstone M, MacDonald B, et al. Neurodevelopmental delay: case definition & guidelines for data collection, analysis, and presentation of immunization safety data. Vaccine 2019; 37:7623–7641. doi: 10.1016/j.vaccine.2019.05.027. 2. Sinopoli VM, Erdman L, Burton CL, Easter P, Rajendram R, Baldwin G, et al. Serotonin system gene variants and regional brain volume differences in pediatric OCD. Brain Imaging Behav 2020; 14:1612–1625. doi: 10.1007/s11682-019-00092-w. 3. Wu K, Hanna GL, Easter P, Kennedy JL, Rosenberg DR, Arnold PD. Glutamate system genes and brain volume alterations in pediatric obsessive-compulsive disorder: a preliminary study. Psychiatry Res 2013; 211:214–220. doi: 10.1016/j.pscychresns.2012.07.003. 4. De Felice A, Ricceri L, Venerosi A, Chiarotti F, Calamandrei G. Multifactorial origin of neurodevelopmental disorders: approaches to understanding complex etiologies. Toxics 2015; 3:89–129. doi: 10.3390/toxics3010089. 5. Bartholomeusz HH, Courchesne E, Karns CM. Relationship between head circumference and brain volume in healthy normal toddlers, children, and adults. Neuropediatrics 2002; 33:239–241. doi: 10.1055/s-2002-36735. 6. Kawasaki Y, Yoshida T, Matsui M, Hiraiwa A, Inomata S, Tamura K, et al. Clinical factors that affect the relationship between head circumference and brain volume in very-low-birth-weight infants. J Neuroimaging 2019; 29:104–110. doi: 10.1111/jon.12558. 7. Li X, Morgan PS, Ashburner J, Smith J, Rorden C. The first step for neuroimaging data analysis: DICOM to NIfTI conversion. J Neurosci Methods 2016; 264:47–56. doi: 10.1016/j.jneumeth.2016.03.001. 8. Jenkinson M, Beckmann CF, Behrens TEJ, Woolrich MW, Smith SM. Functional Magnetic Resonance Imaging of the Brain Software Library. Neuroimage 2012; 62:782–790. doi: 10.1016/j.neuroimage.2011.09.015. 9. Yang L, Kong Y, Dong X, Hu L, Lin Y, Chen X, et al. Clinical and genetic spectrum of a large cohort of children with epilepsy in China. Genet Med 2019; 21:564–571. doi: 10.1038/s41436-018-0091-8. 10. McLaren W, Gil L, Hunt SE, Riat HS, Ritchie GRS, Thormann A, et al. The ensembl variant effect predictor. Genome Biol 2016; 17:1–14. doi: 10.1186/s13059-016-0974-4. 11. Wang K, Li M, Hakonarson H. ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res 2010; 38:e164doi: 10.1093/nar/gkq603. 12. Kumar P, Henikoff S, Ng PC. Predicting the effects of coding non-synonymous variants on protein function using the SIFT algorithm. Nat Protoc 2009; 4:1073–1081. doi: 10.1038/nprot.2009.86. 13. Adzhubei IA, Schmidt S, Peshkin L, Ramensky VE, Gerasimova A, Bork P, et al. A method and server for predicting damaging missense mutations. Nat Methods 2010; 7:248–249. doi: 10.1038/nmeth0410-248. 14. Schwarz JM, Rödelsperger C, Schuelke M, Seelow D. MutationTaster evaluates disease-causing potential of sequence alterations. Nat Methods 2010; 7:575–576. doi: 10.1038/nmeth0810-575. 15. Richards S, Aziz N, Bale S, Bick D, Das S, Gastier-Foster J, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med 2015; 17:405–424. doi: 10.1038/gim.2015.30. 16. Backenroth D, Homsy J, Murillo LR, Glessner J, Lin E, Brueckner M, et al. CANOES: detecting rare copy number variants from whole exome sequencing data. Nucleic Acids Res 2014; 42:e97doi: 10.1093/nar/gku345. 17. Li M, Santpere G, Imamura Kawasawa Y, Evgrafov OV, Gulden FO, Pochareddy S, et al. Integrative functional genomic analysis of human brain development and neuropsychiatric risks. Science 2018; 362:1–31. doi: 10.1126/science.aat7615. 18. Khatamian A, Paull EO, Califano A, Yu J. SJARACNe: a scalable software tool for gene network reverse engineering from big data. Bioinformatics 2018; 35:2165–2166. doi: 10.1093/bioinformatics/bty907. 19. Jimenez J, Herrera DA, Vargas SA, Montoya J, Castillo M. (-Tubulinopathy caused by a mutation of the TUBB2B gene: magnetic resonance imaging findings of the brain. Neuroradiol J 2019; 32:148–150. doi: 10.1177/1971400919828142. 20. Yoon YB, Shin WG, Lee TY, Hur JW, Cho KIK, Sohn WS, et al. Brain structural networks associated with intelligence and visuomotor ability. Sci Rep 2017; 7:1–9. doi: 10.1038/s41598-017-02304-z. 21. Yuzaki M. Cerebellar LTD vs. motor learning-lessons learned from studying GluD2. Neural Netw 2013; 47:36–41. doi: 10.1016/j.neunet.2012.07.001. 22. Vink M, Kahn RS, Raemaekers M, van den Heuvel M, Boersma M, Ramsey NF. Function of striatum beyond inhibition and execution of motor responses. Hum Brain Mapp 2005; 25:336–344. doi: 10.1002/hbm.20111. 23. Sakagami M, Pan X. Functional role of the ventrolateral prefrontal cortex in decision making. Curr Opin Neurobiol 2007; 17:228–233. doi: 10.1016/j.conb.2007.02.008. 24. Hasan A, Galea JM, Casula EP, Falkai P, Bestmann S, Rothwell JC. Muscle and timing-specific functional connectivity between the dorsolateral prefrontal cortex and the primary motor cortex. J Cogn Neurosci 2013; 25:558–570. doi: 10.1162/jocn_a_00338. 25. Romaniello R, Tonelli A, Arrigoni F, Baschirotto C, Triulzi F, Bresolin N, et al. A novel mutation in the β-tubulin gene TUBB2B associated with complex malformation of cortical development and deficits in axonal guidance. Dev Med Child Neurol 2012; 54:765–769. doi: 10.1111/j.1469-8749.2012.04316.x. 26. Breuss M, Heng JI, Poirier K, Tian G, Jaglin XH, Qu Z, et al. Mutations in the β-tubulin gene TUBB5 cause microcephaly with structural brain abnormalities. Cell Rep 2012; 2:1554–1562. doi: 10.1016/j.celrep.2012.11.017. 27. Verbeek E, Meuwissen ME, Verheijen FW, Govaert PP, Licht DJ, Kuo DS, et al. COL4A2 mutation associated with familial porencephaly and small-vessel disease. Eur J Hum Genet 2012; 20:844–851. doi: 10.1038/ejhg.2012.20. 28. Meuwissen ME, Halley DJ, Smit LS, Lequin MH, Cobben JM, de Coo R, et al. The expanding phenotype of COL4A1 and COL4A2 mutations: clinical data on 13 newly identified families and a review of the literature. Genet Med 2015; 17:843–853. doi: 10.1038/gim.2014.210. 29. Hausman-Kedem M, Ben-Sira L, Kidron D, Ben-Shachar S, Straussberg R, Marom D, et al. Deletion in COL4A2 is associated with a three-generation variable phenotype: from fetal to adult manifestations. Eur J Hum Genet 2021; 29:1654–1662. doi: 10.1038/s41431-021-00880-3. 30. Cavallin M, Mine M, Philbert M, Boddaert N, Lepage JM, Coste T, et al. Further refinement of COL4A1 and COL4A2 related cortical malformations. Eur J Med Genet 2018; 61:765–772. doi: 10.1016/j.ejmg.2018.10.004. 31. Koene S, Peeters-Scholte CMPCD, Knijnenburg J, de Vries LS, van Scheltema PNA, Meuwissen ME, et al. Intracerebral hemorrhage in a neonate with an intragenic COL4A2 duplication. Am J Med Genet A 2021; 185:571–574. doi: 10.1002/ajmg.a.61988. 32. Kuo DS, Labelle-Dumais C, Mao M, Jeanne M, Kauffman WB, Allen J, et al. Allelic heterogeneity contributes to variability in ocular dysgenesis, myopathy and brain malformations caused by Col4a1 and Col4a2 mutations. Hum Mol Genet 2014; 23:1709–1722. doi: 10.1093/hmg/ddt560. 33. Horvath A, Faucz F, Finkielstain GP, Nikita ME, Rothenbuhler A, Almeida M, et al. Haplotype analysis of the promoter region of phosphodiesterase type 8B (PDE8B) in correlation with inactivating PDE8B mutation and the serum thyroid-stimulating hormone levels. Thyroid 2010; 20:363–367. doi: 10.1089/thy.2009.0260. 34. Tsai LC, Chan GC, Nangle SN, Shimizu-Albergine M, Jones GL, Storm DR, et al. Inactivation of Pde8b enhances memory, motor performance, and protects against age-induced motor coordination decay. Genes Brain Behav 2012; 11:837–847. doi: 10.1111/j.1601-183X.2012.00836.x. 35. Guo MH, Dauber A, Lippincott MF, Chan YM, Salem RM, Hirschhorn JN. Determinants of power in gene-based burden testing for monogenic disorders. Am J Hum Genet 2016; 99:527–539. doi: 10.1016/j.ajhg.2016.06.031. 36. Lee S, Abecasis GR, Boehnke M, Lin X. Rare-variant association analysis: study designs and statistical tests. Am J Hum Genet 2014; 95:5–23. doi: 10.1016/j.ajhg.2014.06.009.