Biology subjects

Fragoza, R.

Publications and source records attributed to Fragoza, R..

6 recordsLinked to original sources

bioRxiv2025

Integrating structural homology with deep learning to achieve highly accurate protein-protein interface prediction for the human interactome

A significant portion of disease-causing mutations occur at protein-protein interfaces however, the number of structurally resolved multi-protein complexes is extremely small. Here we present a computational pipeline, PIONEER2, that integrates 3D structural similarity with geometric deep learning to accurately predict protein binding partner-specific interfacial residues. We compare the performance of PIONEER2 to that of AlphaFold3 and found, using a test set of PDB structures, that their performance is quite similar. However, about 20% of AlphaFold3 predictions for protein-protein complexes in the PDB have AlphaFold3 ranking scores below 0.5, which indicates an uncertain model. For these structures, PIONEER2 outperforms AlphaFold3 at discriminating interfacial from non-interfacial residues. Further, about half of the AlphaFold3 ranking scores on high confidence protein-protein interactions (PPIs) not associated with a PDB structure are below 0.5 indicating that PIONEER2 offers superior interface prediction for a large number of PPIs for which structures are not available. We created a comprehensive 3D structurally informed interactome encompassing all 352,124 experimentally detected binary human PPIs in the current literature and made PIONEER2 interface predictions for each. We experimentally validated these predictions by generating 1,866 mutations and testing their disruptive impact on 5,010 mutation-interaction pairs. PIONEER2-predicted interfaces are found to be comparable to PDB structures in their ability to predict disruptive mutations while AlphaFold3 performance is reduced. Similarly, PIONEER2-predicted interfaces outperform AlphaFold3 in accounting for the depletion of non-deleterious common population variants and the enrichment of disease-related mutations on protein surfaces. Overall, our results suggest that PIONEER2-predicted interfaces provide a valuable tool for studying disease etiology, advancing personalized medicine and for fundamental research. We further implemented PIONEER2 as a user-friendly web server (https://pioneer2.yulab.org) platform for users to explore our 3D interactome models and conduct genome-wide functional genomics studies.

bioinformatics↗

bioRxiv2024

A Novel Engineered U7 Small Nuclear RNA Scaffold Greatly Increases in vitro and in vivo ADAR-Mediated Programmable RNA Base Editing

Custom RNA base editing using the endogenous human Adenosine Deaminase Acting on RNA (ADAR) enzyme presents a promising approach for precision therapeutics, alleviating concerns of permanent DNA damage or immunogenicity from1 foreign bacterial proteins such as CRISPR/Cas. ADAR can be directed to act on therapeutic RNA targets by antisense guide RNAs (gRNAs) that create a substrate for ADARs adenosine-to-inosine (effectively A-to-G) deamination activity. Delivery of gRNAs via a DNA expression construct provided by Adeno-Associated Virus (AAV) might allow life-long duration of the therapy. However, a major challenge for RNA editing using gene-encoded gRNAs and endogenous levels of ADAR is achieving sufficient gRNA activity inside cells, especially in therapeutic situations where AAV delivery may provide as low as one viral genome per cell. Here we show that embedding antisense gRNAs into a U7 small nuclear RNA (snRNA) framework and adding hnRNP A1 binding domains greatly increases the efficiency of custom RNA editing. This increased editing efficiency allows for detectable RNA editing from a single genomic insertion of gRNA construct per cell, which enabled a pooled library screen of 750+ gRNA variations to further improve the SmOPT U7 hairpin system. The screen revealed critical residues responsible for RNA editing and generated new SmOPT and U7 hairpin variants that further boosted RNA editing. The final design, combined with an improved synthetic U7 promoter, resulted in up to 76% targeted editing with a single integrated copy of construct per cell, representing a 10- to 100-fold increase over existing circular gRNA approaches. Using systemic in vivo AAV delivery, we achieved an unprecedented 75% RNA editing in the total brain of a mouse model of Hurler syndrome. Our novel SmOPT U7 system also improved published antisense oligos for DMD exon skipping, currently in clinical trials, by up to 25-fold in differentiated myoblasts, and therefore represents a universal scaffold for ADAR-based RNA editing as well as other antisense RNA therapies.

bioengineering↗

bioRxiv2021

In vivo versus in silico assessment of potentially pathogenic missense variants in human reproductive genes

Infertility is a heterogeneous condition, with genetic causes estimated to be involved in approximately half of the cases. High-throughput sequencing (HTS) is becoming an increasingly important tool for genetic diagnosis of diseases including idiopathic infertility, however, most rare or minor alleles revealed by HTS are variants of uncertain significance (VUS). Interpreting the functional impacts of VUS is challenging but profoundly important for clinical management and genetic counseling. To determine the consequences of population polymorphisms in key fertility genes, we functionally evaluated 11 missense variants in the genes ANKRD31, BRDT, DMC1, EXOI, FKBP6, MCM9, M1AP, MEI1, MSH4 and SEPT12 by generating genome-edited mouse models. Nine variants were classified as deleterious by most functional prediction algorithms, and two disrupted a protein-protein interaction in the yeast 2 hybrid assay. Even though these genes are known to be essential for normal meiosis or spermiogenesis in mice, only one of the tested human variants (rs1460351219, encoding p.R581H in MCM9), which was observed in a male infertility patient, compromised fertility or gametogenesis in the mouse models. To explore the disconnect between predictions and outcomes, we compared pathogenicity calls of missense variants made by ten widely-used algorithms to: 1) those present in ClinVar, and 2) those which have been evaluated in mice. We found that all the algorithms performed poorly in terms of predicting the effects of human missense variants that have been modeled in mice. These studies emphasize caution in the genetic diagnoses of infertile patients based primarily on pathogenicity prediction algorithms, and emphasize the need for alternative and efficient in vitro or vivo functional validation models for more effective and accurate VUS delineation to either pathogenic or benign categories. SignificanceAlthough infertility is a substantial medical problem that affects up to 15% of couples, the potential genetic causes of idiopathic infertility have been difficult to decipher. This problem is complicated by the large number of genes that can cause infertility when perturbed, coupled with the large number of VUS that are present in the genomes of affected patients. Here, we present and analyze mouse modeling data of missense variants that are classified as deleterious by commonly-used pathogenicity prediction algorithms but which caused no detectible phenotype when introduced into mice by genome editing. We find that augmenting pathogenicity predictions with preliminary screens for biochemical defects substantially enhanced the proportion of prioritized variants that caused phenotypes in mice. The results emphasize that, in the absence of substantial improvements of in silico prediction tools or other compelling pre-existing evidence, in vivo analysis is crucial for confident attribution of infertility alleles.

genetics↗

bioRxiv2021

MLH1/3 variants causing aneuploidy, pregnancy loss, and premature reproductive aging

Most spontaneous pregnancy losses are a result of embryonic aneuploidy stemming from mis-segregation of chromosomes during meiosis. Proper disjunction of homologous chromosomes is dependent upon precise control of crossing-over, a process requiring the mismatch repair (MMR) genes MLH1 and MLH3. Both are required for fertility and completion of meiosis in mice. People inheriting variants in these genes are often at high risk for colorectal cancer and Lynch syndrome, yet the potential impacts of variants upon reproduction are unclear. To determine if MLH1/3 variants (namely single nucleotide polymorphisms, or SNPs) in human populations can cause reproductive abnormalities, we used a combination of computational predictions, yeast two-hybrid assays, and assays of MMR and recombination in yeast to select nine MLH1 and MLH3 variants for modeling in mice via genome editing. We identified 7 alleles that caused reproductive defects in mice including subfertility in females, male infertility, reduced sperm counts, and increased spermatocyte apoptosis. Remarkably, these alleles in females caused age-dependent decreases in litter size, and increased resorption of embryos during pregnancy. These outcomes were likely a consequence of reduced meiotic chiasmata, in turn causing an increase in misaligned chromosomes and univalents in meiotic metaphase I (MI). Our data indicate that segregating hypomorphic alleles of meiotic recombination genes in populations can predispose females to increased incidence of pregnancy loss from gamete aneuploidy.

genetics↗

bioRxiv2020

Variants in RABL2A causing male infertility and ciliopathy

PurposeApproximately 7% of men suffer from infertility worldwide and sperm abnormalities are the major cause. Though genetic defects are thought to underlie a substantial fraction of all male infertility cases, the actual molecular bases are usually undetermined. Because the consequences of most genetic variants in populations are unknown, this complicates genetic diagnosis even after genome sequencing of patients. Some patients with ciliopathies, including primary ciliary dyskinesia (PCD) and Bardet-Biedl syndrome (BBS), also suffer from infertility because sperm flagella, which share several characteristics with cilia, are also affected in these patients. MethodsTo identify infertility-causing genetic variants in human populations, we used in silico predictions to identify potentially deleterious SNP (single nucleotide polymorphism) alleles of RABL2A, a gene essential for normal cilia and flagella function. Candidate variants were assayed for protein stability in vitro, and the destabilizing variants were modeled in mice using CRISPR/Cas9-mediated genome editing. The resulting mice were characterized phenotypically for reproductive and developmental defects. ResultsTwo of the SNP alleles, Rabl2L119F (rs80006029) and Rabl2V158F (rs200121688), destabilized the protein. Mice bearing these alleles exhibited PCD- and BBS-associated disorders including male infertility, early growth retardation, excessive weight gain in adulthood, heterotaxia, pre-axial polydactyly, neural tube defects (NTD) and hydrocephalus. ConclusionOur study identified and validated pathogenicity of two variants causing ciliopathies and male infertility in human populations, and identified phenotypes not previously described for null alleles of Rabl2.

genetics↗

bioRxiv2017

Interactome INSIDER: A Multi-Scale Structural Interactome Browser For Genomic Studies

Protein interactions underlie nearly all known cellular function, making knowledge of their binding conformations paramount to understanding the physical workings of the cell. Studying binding conformations has allowed scientists to explore some of the mechanistic underpinnings of disease caused by disruption of protein interactions. However, since experimentally determined interaction structures are only available for a small fraction of the known interactome such inquiry has largely excluded functional genomic studies of the human interactome and broad observations of the inner workings of disease. Here we present Interactome INSIDER, an information center for genomic studies using the first full-interactome map of human interaction interfaces. We applied a new, unified framework to predict protein interaction interfaces for 184,605 protein interactions with previously unresolved interfaces in human and 7 model organisms, including the entire experimentally determined human binary interactome. We find that predicted interfaces share several known functional properties of interfaces, including an enrichment for disease mutations and recurrent cancer mutations, suggesting their applicability to functional genomic studies. We also performed 2,164 de novo mutagenesis experiments and show that mutations of predicted interface residues disrupt interactions at a similar rate to known interface residues and at a much higher rate than mutations outside of predicted interfaces. To spur functional genomic studies in the human interactome, Interactome INSIDER (http://interactomeinsider.yulab.org) allows users to explore known population variants, disease mutations, and somatic cancer mutations, or upload their own set of mutations to find enrichment at the level of protein domains, residues, and 3D atomic clustering in known and predicted interaction interfaces.

genomics↗