Delineation of functionally essential protein regions for 242 neurodevelopmental genes

in Brain (2022), awac381

Neurodevelopmental disorders (NDDs), including severe pediatric epilepsy, autism, and intellectual disabilities are heterogeneous conditions in which clinical genetic testing can often identify a ... [more ▼]

Neurodevelopmental disorders (NDDs), including severe pediatric epilepsy, autism, and intellectual disabilities are heterogeneous conditions in which clinical genetic testing can often identify a pathogenic variant. For many of them, genetic therapies will be tested in this or the coming years in clinical trials. In contrast to first-generation symptomatic treatments, the new disease-modifying precision medicines require a genetic test-informed diagnosis before a patient can be enrolled in a clinical trial. However, even in 2022, most identified genetic variants in NDD genes are ‘Variants of Uncertain Significance’. To safely enroll patients in precision medicine clinical trials, it is important to increase our knowledge about which regions in NDD-associated proteins can ‘tolerate’ missense variants and which ones are ‘essential’ and will cause a NDD when mutated. In addition, knowledge about functionally indispensable regions in the three-dimensional (3D) structure context of proteins can also provide insights into the molecular mechanisms of disease variants. We developed a novel consensus approach that overlays evolutionary, and population based genomic scores to identify 3D essential sites (Essential3D) on protein structures. After extensive benchmarking of AlphaFold predicted and experimentally solved protein structures, we generated the currently largest expert curated protein structure set for 242 NDDs and identified 14,377 Essential3D sites across 189 gene disorders associated proteins. We demonstrate that the consensus annotation of Essential3D sites improves prioritization of disease mutations over single annotations. The identified Essential3D sites were enriched for functional features such as intermembrane regions or active sites and discovered key inter-molecule interactions in protein complexes that were otherwise not annotated. Using the currently largest autism, developmental disorders, and epilepsies exome sequencing studies including >360,000 NDD patients and population controls, we found that missense variants at Essential3D sites are 8-fold enriched in patients. In summary, we developed a comprehensive protein structure set for 242 neurodevelopmental disorders and identified 14,377 Essential3D sites in these. All data are available at https://es-ndd.broadinstitute.org for interactive visual inspection to enhance variant interpretation and development of mechanistic hypotheses for 242 NDDs genes. The provided resources will enhance clinical variant interpretation and in silico drug target development for NDD-associated genes and encoded proteins. [less ▲]

Predicting functional effects of missense variants in voltage-gated sodium and calcium channels

in Science Translational Medicine (2020), 12(556), 6848

Malfunctions of voltage-gated sodium and calcium channels (encoded by SCNxA and CACNA1x family genes, respectively) have been associated with severe neurologic, psychiatric, cardiac, and other diseases ... [more ▼]

Malfunctions of voltage-gated sodium and calcium channels (encoded by SCNxA and CACNA1x family genes, respectively) have been associated with severe neurologic, psychiatric, cardiac, and other diseases. Altered channel activity is frequently grouped into gain or loss of ion channel function (GOF or LOF, respectively) that often corresponds not only to clinical disease manifestations but also to differences in drug response. Experimental studies of channel function are therefore important, but laborious and usually focus only on a few variants at a time. On the basis of known gene-disease mechanisms of 19 different diseases, we inferred LOF (n = 518) and GOF (n = 309) likely pathogenic variants from the disease phenotypes of variant carriers. By training a machine learning model on sequence- and structure-based features, we predicted LOF or GOF effects [area under the receiver operating characteristics curve (ROC) = 0.85] of likely pathogenic missense variants. Our LOF versus GOF prediction corresponded to molecular LOF versus GOF effects for 87 functionally tested variants in SCN1/2/8A and CACNA1I (ROC = 0.73) and was validated in exome-wide data from 21,703 cases and 128,957 controls. We showed respective regional clustering of inferred LOF and GOF nucleotide variants across the alignment of the entire gene family, suggesting shared pathomechanisms in the SCNxA/CACNA1x family genes. [less ▲]

Identification of pathogenic variant enriched regions across genes and gene families

in Genome Research (2020), 30(1), 62-71

Missense variant interpretation is challenging. Essential regions for protein function are conserved among gene family members, and genetic variants within these regions are potentially more likely to ... [more ▼]

Missense variant interpretation is challenging. Essential regions for protein function are conserved among gene family members, and genetic variants within these regions are potentially more likely to confer risk to disease. Here, we generated 2,871 gene family protein sequence alignments involving 9,990 genes and performed missense variant burden analyses to identify novel essential protein regions. We mapped 2,219,811 variants from the general population into these alignments and compared their distribution with 76,153 missense variants from patients. With this gene family approach, we identified 465 regions enriched for patient variants spanning 41,463 amino acids in 1,252 genes. As a comparison, testing the same genes individually we identified less patient variant enriched regions involving only 2,639 amino acids and 215 genes. Next, we selected de novo variants from 6,753 patients with neurodevelopmental disorders and 1,911 unaffected siblings, and observed an 8.33-fold enrichment of patient variants in our identified regions (95% C.I.=3.90-Inf, p-value = 2.72x10-11). Using the complete ClinVar variant set, we found that missense variants inside the identified regions are 106-fold more likely to be classified as pathogenic in comparison to benign classification (OR = 106.15, 95% C.I = 70.66-Inf, p-value < 2.2 x 10-16). All pathogenic variant enriched regions (PERs) identified are available online through the “PER viewer” a user-friendly online platform for interactive data mining, visualization and download. In summary, our gene family burden analysis approach identified novel pathogenic variant enriched regions in protein sequences. This annotation can empower variant interpretation. [less ▲]

Identification of pathogenic variant enriched regions across genes and gene families

E-print/Working paper (2019)

Missense variant interpretation is challenging. Essential regions for protein function are conserved among gene family members, and genetic variants within these regions are potentially more likely to ... [more ▼]

Missense variant interpretation is challenging. Essential regions for protein function are conserved among gene family members, and genetic variants within these regions are potentially more likely to confer risk to disease. Here, we generated 2,871 gene family protein sequence alignments involving 9,990 genes and performed missense variant burden analyses to identify novel essential protein regions. We mapped 2,219,811 variants from the general population into these alignments and compared their distribution with 65,034 missense variants from patients. With this gene family approach, we identified 398 regions enriched for patient variants spanning 33,887 amino acids in 1,058 genes. As a comparison, testing the same genes individually we identified less patient variant enriched regions involving only 2,167 amino acids and 180 genes. Next, we selected de novo variants from 6,753 patients with neurodevelopmental disorders and 1,911 unaffected siblings, and observed a 5.56-fold enrichment of patient variants in our identified regions (95% C.I. =2.76-Inf, p-value = 6.66×10−8). Using an independent ClinVar variant set, we found missense variants inside the identified regions are 111-fold more likely to be classified as pathogenic in comparison to benign classification (OR = 111.48, 95% C.I = 68.09-195.58, p-value < 2.2e−16). All patient variant enriched regions identified (PERs) are available online through a user-friendly platform for interactive data mining, visualization and download at http://per.broadinstitute.org. In summary, our gene family burden analysis approach identified novel patient variant enriched regions in protein sequences. This annotation can empower variant interpretation. [less ▲]

Functional Interpretation of Single Amino Acid Substitutions in 1,330 Disease-Associated Genes

in Biophysical Journal (2019, February 15), 116(3), 420-421

Elucidating molecular consequences of amino-acid-altering missense variants at scale is challenging. In this work, we explored whether features derived from three-dimensional (3D) protein structures can ... [more ▼]

Elucidating molecular consequences of amino-acid-altering missense variants at scale is challenging. In this work, we explored whether features derived from three-dimensional (3D) protein structures can characterize patient missense variants across different protein classes with similar molecular level activities. The identified disease-associated features can advance our understanding of how a single amino acid substitution can lead to the etiology of monogenic disorders. For 1,330 disease-associated genes (>80%, 1,077/1,330 implicated in Mendelian disorders), we collected missense variants from the general population (gnomAD database, N=164,915) and patients (ClinVar and HGMD databases, N=32,923). We in silico mapped the variant positions onto >14k human protein 3D structures. We annotated the protein positions of variants with 40 structural, physiochemical, and functional features. We then grouped the genes into 24 protein classes based on their molecular functions and performed statistical association analyses with the features of population and patient variants. We identified 18 (out of 40) features that are associated with patient variants in general. Specifically, patient variants are less exposed to solvent (p<1.0e-100), enriched on b-sheets (p<2.37e-39), frequently mutate aromatic residues (p<1.0e-100), occur in ligand binding sites (p<1.0e-100) and are spatially close to phosphorylation sites (p<1.0e-100). We also observed differential protein-class-specific features. For three protein classes (signaling molecules, proteases and hydrolases), patient variants significantly perturb the disulfide bonds (p<1.0e-100). Only in immunity proteins, patient variants are enriched in flexible coils (p<1.65e-06). Kinases and cell junction proteins exhibit enrichment of patient variants around SUMOylation (p<1.0e-100) and methylation sites (p<9.29e-11), respectively. In summary, we studied shared and unique features associated with patient variants on protein structure across 24 protein classes, providing novel mechanistic insights. We generated an online resource that contains amino-acid-wise feature annotation-track for 1,330 genes, summarizes the patient-variant-associated features on residue level, and can guide variant interpretation. [less ▲]

Identification and Characterization of Variant Intolerant Sites across Human Protein 3-Dimensional Structures

in Biophysical Journal (2018, February 02), 114(3, Suppl. 1), 664

The functional interpretation of genetic variation in disease-associated genes is far outpaced by data generation. Existing algorithms for prediction of variant consequences do not adequately distinguish ... [more ▼]

The functional interpretation of genetic variation in disease-associated genes is far outpaced by data generation. Existing algorithms for prediction of variant consequences do not adequately distinguish pathogenic variants from benign rare variants. This lack of statistical and bioinformatics analyses, accompanied by an ever-increasing number of identified variants in biomedical research and clinical applications, has become a major challenge. Established methods to predict the functional effect of genetic variation use the degree of amino acid conservation across species in linear protein sequence alignment. More recent methods include the spatial distribution pattern of known patient and control variants. Here, we propose to combine the linear conservation and spatial constrained based scores to devise a novel score that incorporates 3-dimensional structural properties of amino acid residues, such as the solvent-accessible surface area, degree of flexibility, secondary structure propensity and binding tendency, to quantify the effect of amino acid substitutions. For this study, we develop a framework for large-scale mapping of established linear sequence-based paralog and ortholog conservation scores onto the tertiary structures of human proteins. This framework can be utilized to map the spatial distribution of mutations on solved protein structures as well as homology models. As a proof of concept, using a homology model of the human Nav1.2 voltage-gated sodium channel structure, we observe spatial clustering in distinct domains of mutations, associated with Autism Spectrum Disorder (>20 variants) and Epilepsy (>100 variants), that exert opposing effects on channel function. We are currently characterizing all variants (>300k individuals) found in ClinVar, the largest disease variant database, as well as variants identified in >140k individuals from general population. The variant mapping framework and our score, informed with structural information, will be useful in identifying structural motifs of proteins associated with disease risk. [less ▲]