Biology subjects

Marsh, J. A.

Publications and source records attributed to Marsh, J. A..

At least 19 recordsLinked to original sources

bioRxiv2026

Variant effect predictors are shaping the ClinVar classifications used to train and evaluate them

Variant effect predictors (VEPs) are playing an increasingly influential role in clinical variant classification. Following recent calibration efforts and the development of points-based classification standards, computational predictions can now carry enough weight to move variants out of uncertain significance with little additional evidence. While clinically valuable, this increases the potential for classification-level circularity, where VEP predictions help determine the clinical classifications later used to train, benchmark and calibrate VEPs. Here, we show that computational evidence is already leaving a detectable signature in ClinVar. Among rare missense variants from a major submitter, likely benign classifications shifted markedly towards more benign scores and show a sharp step change at the calibrated moderate-evidence threshold of BayesDel_noAF. This signal extends to nearly all 52 predictors examined, including population-free models unlikely to have been used in classification, probably reflecting shared evolutionary information. Classifications informed by computational evidence therefore cannot be treated as independent ground truth, and temporal holdouts cannot eliminate this circularity. To prevent a self-reinforcing feedback loop between prediction and classification, we recommend that ClinVar submissions routinely record the evidence criteria and computational tools supporting each classification in a standardised format, alongside greater use of classification-independent benchmarks.

genetics↗

bioRxiv2026

Mechanism-selective deep mutational scanning distinguishes ERCC2 disease phenotypes

Pathogenic ERCC2 variants cause xeroderma pigmentosum (XP), trichothiodystrophy (TTD) or both, yet variant effect scores are usually interpreted only as measures of pathogenicity rather than of which disease mechanism is disrupted. XPD, the ERCC2-encoded TFIIH subunit, functions in both nucleotide excision repair and transcription. Using yeast complementation deep mutational scanning, we measured the effects of nearly all XPD amino acid substitutions. The assay was mechanism-selective: it preferentially reported transcription-associated function, with pronounced intolerance at the p44 interface, whereas many substitutions affecting DNA binding and helicase activity retained near-wild-type fitness. Accordingly, TTD variants had much lower fitness than XP variants. Computational predictors discriminated pathogenic from benign variants similarly across phenotypes, but the DMS distinguished XP from TTD variants better than all 73 predictors tested. Phenotype-specific ACMG/AMP calibration provided evidence in both directions for TTD but mainly pathogenic evidence for XP. Thus, the selectivity of functional assays, often viewed as a limitation, can reveal disease mechanisms and support phenotype-aware variant interpretation.

genetics↗

bioRxiv2026

MAPT regulates autophagic-lysosomal function and phagocytosis in human microglia

Tauopathies are characterized by the accumulation and spread of pathogenic tau aggregates throughout the brain, a process that is increasingly recognized to involve not only neurons but also microglia. However, whether pathogenic MAPT directly alters microglial degradative capacity remains poorly understood. Here, using isogenic human induced pluripotent stem cell-derived microglia carrying the pathogenic MAPT IVS10+16 mutation, we identify tau as a regulator of microglial lysosomal function. MAPT IVS10+16 microglia exhibited coordinated suppression of lysosomal and autophagic pathways, reduced lysosomal protease abundance and activity, and impaired autophagosome-lysosome fusion. Mutant microglia also showed reduced uptake of extracellular tau aggregates, reduced tau accumulation in acidic compartments, and a blunted lysosomal response to proteopathic stress. Conversely, genetic loss of MAPT increased lysosomal degradative capacity and accumulation of extracellular tau aggregates within acidic compartments, supporting a cell-intrinsic role for endogenous tau in regulating microglial degradative function. Pharmacologic enhancement of the autophagy lysosome pathway in MAPT IVS10+16 microglia increased proteolytic activity and improved tau handling. Together, these findings reveal a reciprocal relationship between tau and microglial lysosome function and identify degradative capacity as a modifiable component of the microglial response to tau pathology.

neuroscience↗

bioRxiv2026

Ms4a4a loss reprograms amyloid-associated microglia and limits dense-core plaque-associated tau spreading

INTRODUCTIONMicroglia regulate amyloid plaque-associated microenvironments that contribute to downstream tau pathology in Alzheimers disease (AD). Variants within the MS4A locus are strongly associated with AD risk and resilience and are linked to microglial biology; however, the functional role of MS4A4A in plaque-associated tau pathology remains poorly understood. METHODSSingle-nucleus RNA sequencing (snRNA-seq) was performed on hippocampi from non-transgenic, Ms4a4a knockout (4A-KO), 5xFAD, and 5xFAD 4A-KO mice at 6 months of age. To assess plaque-associated tau pathology, AD-derived tau aggregates were injected into the hippocampus of 5xFAD and 5xFAD 4A-KO mice at 6 months, and histological analyses were performed 3 months later. RESULTSAmyloid pathology was the dominant driver of microglial state transitions, while Ms4a4a loss selectively remodeled activated microglial transcriptional programs enriched for interferon, lysosomal, autophagic, and proteostatic pathways. Activated microglia from 5xFAD 4A-KO mice exhibited altered expression of genes linked to immune signaling and protein handling. Following AD-tau inoculation, Ms4a4a loss did not significantly alter overall phospho-tau burden but selectively reduced dense-core plaque-associated neuritic plaque tau (NP-tau), particularly in the contralateral hemisphere. This phenotype was strongest surrounding X-34-positive fibrillar plaques and occurred without major changes in plaque-associated microgliosis. DISCUSSIONThese findings identify Ms4a4a as a regulator of plaque-associated microglial programs linked to NP-tau accumulation in the amyloid-bearing brain. More broadly, this work supports a model in which AD resilience-associated microglial pathways selectively shape plaque-associated microenvironments that promote downstream tau pathology.

neuroscience↗

bioRxiv2026

Clinical evidence yield as a framework for evaluating computational predictors and multiplexed assays of variant effect

Interpreting the clinical significance of missense variants of uncertain significance (VUS) remains a major challenge in clinical genetics. Although computational variant effect predictors (VEPs) and multiplexed assays of variant effect (MAVEs) can generate large-scale functional scores, their value is typically assessed using discrimination metrics such as AUROC rather than by the strength of evidence they provide under ACMG/AMP guidelines. Here, we introduce mean evidence strength (MES), a quantitative metric that summarises the pathogenic and benign evidence assigned across missense variants following gene-level Bayesian calibration. Using the acmgscaler framework, we calibrated 12 population-free VEPs across 367 disease genes and analysed 15 MAVE datasets with sufficient clinical data. MES revealed important discrepancies with AUROC, including cases where methods with similar discrimination differed substantially in evidence yield. MAVEs achieved high average MES despite lower AUROC, while several VEPs showed strong discrimination but more limited calibrated evidence. Among predictors, CPT-1 achieved the highest MES and provided moderate or stronger evidence for the largest fraction of ClinVar VUS. MES therefore provides a practical framework for evaluating computational and experimental variant effect datasets in terms of calibrated clinical evidence yield.

bioinformatics↗

bioRxiv2026

Disagreement among variant effect predictors guides experimental prioritization of target proteins

Interpreting the functional consequences of genetic variation, especially rare missense variants, remains a significant challenge in human genetics. Computational variant effect predictors (VEPs) and multiplexed assays of variant effects (MAVEs) provide complementary approaches, with VEPs offering scalable predictions and MAVEs delivering detailed empirical measurements. However, MAVEs are resource intensive and cannot yet be applied broadly across the proteome, making it important to identify proteins where experimental mapping will be most informative. We hypothesised that MAVEs should be particularly valuable for proteins where computational predictors disagree, as such disagreement may highlight mechanistic blind spots. To test this, we analysed predictions from ten distinct VEPs across more than 13,000 human proteins and quantified inter-predictor concordance. We observed substantial variability across proteins in the degree of agreement across predictors and investigated structural, functional and gene-level features associated with this variation. We find that inter-VEP concordance showed no relationship with agreement to experimental MAVE data. If predictor agreement reflected how intrinsically predictable a protein is, these quantities would be expected to correlate. Their decoupling instead suggests that MAVEs may provide orthogonal information to VEPs, supporting the use of inter-VEP disagreement to prioritise proteins where experimental data will be most informative. We therefore propose using inter-VEP disagreement as a practical strategy to prioritise proteins for experimental characterization. Focusing on proteins with low predictor concordance should maximise the informational value of new MAVEs, and improve variant interpretation in both research and clinical contexts.

bioinformatics↗

bioRxiv2026

Replicable generation of rhesus macaque iPSCs for in vitro modeling of genetic frontotemporal dementia

At the Wisconsin National Primate Research Center, we have identified a family of rhesus carrying the microtubule-associated protein tau (MAPT) R406W mutation linked to frontotemporal dementia (FTD). Rhesus induced pluripotent stem cells (RhiPSCs) derived from these monkeys present a unique opportunity for in vitro modeling and comparison with cells derived from MAPT R406W human carriers. Here, we report the development of a reproducible method to generate RhiPSCs compliant with the standards of the International Society for Stem Cell Research (ISSCR) to support in vitro modeling of FTD-MAPT R406W. Our stepwise approach identified efficient methods for fibroblast derivation, fibroblast reprogramming to RhiPSC, and RhiPSC maintenance over continued culture. To derive fibroblasts from MAPT wild type (WT) and R406W monkeys, a combination of manual processing and overnight enzymatic digestion was required to maximize the number of low passage fibroblasts available for reprogramming. Fibroblast reprogramming to RhiPSC using Sendai viral vectors versus oriP/EBNA1 episomal plasmids revealed the latter as most efficient. Electroporation conditions for oriP/EBNA1 reprogramming were optimized to maximize plasmid uptake and cell survival. Ultimately, eight RhiPSC lines were derived from 4 donor rhesus monkeys (n=2 WT, n=2 R406W; two clonal lines per donor) and fully characterized according to ISSCR standards. RhiPSC stemness and genetic stability was best maintained on mouse embryonic fibroblast feeders in Universal Primate Pluripotency Stem Cell medium, as opposed to Essential 12 medium supplemented with IWR1, which produced cytogenetic abnormalities. Rhesus neural progenitor cells were generated using a monolayer protocol and expressed PAX6 and NESTIN after 21 days of differentiation. Our reliable method will be useful to labs seeking to derive RhiPSCs for preclinical studies. Overall, the RhiPSCs generated from MAPT R406W carriers will be a critical resource for evaluating the molecular underpinnings of tau-related neurodegeneration across primate species.

cell biology↗

bioRxiv2026

BAF complex-independent gene activation by SS18::SSX

In synovial sarcoma, the BAF subunit SS18 is fused to SSX, a transcriptional repressor, generating the oncogenic SS18::SSX fusion protein. Incorporation of SS18::SSX into BAF complexes leads to their aberrant retargeting to Polycomb-repressed genes via SSX, while simultaneously altering their composition and activity. The presence of BAF at Polycomb target sites is widely assumed to be essential for gene activation. Here, we directly tested the requirement for BAF activity in synovial sarcoma cell survival and SS18::SSX-driven transcription. Using targeted degradation of BAF ATPase subunits and deletion of core components, we show that BAF loss has modest effects on sarcoma cell viability and does not impede SS18::SSX target gene expression. Consistently, deletion of the BAF ATPase subunit Smarca4 does not impair SS18::SSX-driven tumor formation in vivo. Using domain-specific SS18::SSX mutants, we further demonstrate that the fusion can activate oncogenic transcription independently of BAF interaction, and that this activity depends on the C-terminal QPGY-rich domain of SS18. Mechanistically, SS18::SSX promotes transcription by engaging the histone acetyltransferase EP300, independently of BAF. Accordingly, pharmacologic degradation of EP300/CREBBP suppresses SS18::SSX-driven transcriptional programs and impairs synovial sarcoma cell survival. Together, these findings challenge the view that BAF activity is required for SS18::SSX-mediated transcriptional activation and demonstrate that aberrant Polycomb target gene expression is sustained through recruitment of transcriptional coactivators in the absence of BAF. Our work reveals new therapeutic vulnerabilities in synovial sarcoma and suggests broader relevance for targeting coactivator-dependent transcription in fusion-driven cancers. HighlightsO_LIBAF degradation does not alter SS18::SSX-activated transcriptional programs C_LIO_LIDirect SS18::SSX transcriptional activation is independent of BAF interaction C_LIO_LIThe SS18 C-terminus engages the co-activator EP300 to promote gene expression C_LIO_LISmall molecule degraders of EP300/CREBBP abolish SS18::SSX-mediated transcription C_LI

cancer biology↗

bioRxiv2025

Integrative Genomic and Functional Analyses Reveal NINL as a Modulator of Tau Aggregation

IntroductionProteostasis dysfunction is a hallmark of frontotemporal dementia (FTD) and Alzheimers disease (AD), yet the genetic and molecular pathways that disrupt protein homeostasis remain poorly understood. MethodsWe integrated human genetics, transcriptomics, and functional studies to identify proteostasis network components involved in tauopathy. ResultsWe identified 18 proteostasis network genes harboring 75 rare, damaging variants enriched in FTD and/or AD. These genes, spanning multiple proteostasis pathways, were differentially expressed in MAPT mutant neurons and dysregulated in FTD and AD brains. NINL, which encodes Nlp, emerged as the only gene consistently upregulated across all datasets. NINL overexpression reduced tau seeding and enhanced lysosomal proteolytic activity, whereas two FTD-enriched NINL frameshift variants impaired Nlp expression and abolished these protective effects. DiscussionOur findings identify a set of proteostasis genes with genetic and transcriptional links to neurodegeneration and reveal NINL as a novel regulator of tau aggregation, potentially upregulated as an adaptive response to proteotoxic stress.

neuroscience↗

bioRxiv2025

ATRX Deficiency Drives Aberrant Type I Interferon Signalling Through cGAS-Dependent Transcriptional Dysregulation

The X-linked -thalassaemia intellectual disability syndrome (ATRX) protein is a chromatin remodeller involved in transcriptional regulation and genome stability. While the importance of ATRX in development and malignancy is well recognised, its role in innate immunity is less well defined. In two unrelated patients demonstrating cerebral white matter disease, learning difficulties and a persistent upregulation of interferon stimulated gene expression in whole blood, we identified the same Y1758C missense substitution in ATRX. Using patient-derived cells, engineered fibroblasts and neuronal models, we show that this substitution, and other loss of function mutations in ATRX, result in enhanced type I interferon signalling through a cGAS-dependent mechanism uncoupled from the DNA sensing activity of cGAS. Loss of ATRX function leads to alterations in the chromatin distribution of DAXX and H3.3, with cGAS essential for the changes in nucleosome composition and gene expression mediated by ATRX deficiency. Thus, our study highlights a previously unrecognized link between ATRX dysfunction and inflammation involving a non-canonical role of cGAS.

immunology↗

bioRxiv2025

ER-liquid condensate contacts sequester FAM134B/C and RhoA to govern cell morphology

ER-phagy receptors have elusive physiological functions beyond ER remodelling. Here, we use proximity biotinylation to identify their cytoplasmic interactomes. Secondary CRISPR/Cas9 screening reveals regulators of the prototypical FAM134B/C receptors, which include PRKAR1A, canonically known as a subunit of PKA. PRKAR1A directly binds an amphipathic helix in the otherwise disordered cytoplasmic domain of FAM134B. Super-resolution, FRAP and CLEM imaging reveal novel interorganellar contacts between liquid-like condensates of cAMP-bound PRKAR1A and the ER. Condensates promote clustering of ER-embedded FAM134B/C with LC3B, independently of regulation of PKA by PRKAR1A. Proteomics reveal that cytoplasmic RhoA interacts with FAM134B/C clusters and that sequestration of both these molecules occurs within lysosomes embedded within the proximal condensates. This results in reduced actomyosin contractility and ER-condensate interactions thusly determine cell morphology and cancer cell invasion modality. In summary, ER-condensate contacts mediated by FAM134B/C are novel cellular degradation hubs that coordinate ER and cellular remodelling.

cell biology↗

bioRxiv2025

Human Mendelian disease and in vivo mutagenesis screening define the molecular architecture of the U8 snoRNA

Over 50 mutations in the vertebrate specific box C/D small nucleolar RNA U8 have been reported to cause leukoencephalopathy with calcifications and cysts (LCC), a progressive cerebral microangiopathy presenting at any age between early infancy and late adulthood. Notably, the majority of these disease-associated mutations are annotated as variants of uncertain significance due to an inability to predict their effect on U8 function. Here, using a zebrafish bioassay, we performed in vivo mutagenesis screening to explore the molecular pathology of LCC. We demonstrate that LCC-associated mutations cluster in function and molecular behaviour according to distinct structural domains within U8, differentially impairing U8 processing, activity and stability. Further, we show that when U8 function is hypomorphic, exogenous human U8 and the endogenous zebrafish U8.3 paralogue increase in stability through a newly discovered post-transcriptional mechanism. This latter response is compromised in patients with LCC due to the presence of a common polymorphism of an N6-methyladenosine modified nucleotide, which we predict to impact disease penetrance.

neuroscience↗

bioRxiv2025

Mechanistic Modelling of Recessive Disease through Allelic Integration of Variant Effects

Interpreting the effects of genetic variants remains a major challenge in recessive diseases, where clinical outcomes often depend on interactions between alleles. Multiplex assays of variant effects (MAVEs) measure variant function at scale, but nonlinear relationships with biochemical activity complicate the interpretation of MAVE scores. Here, we describe a broadly applicable approach to estimate enzymatic activities for thousands of genetic variants using a pair of fitness assays conducted at different expression levels, by modelling the nonlinear relationship between activity and fitness. Activity scores from two alleles are then combined into a single pathogenicity metric that captures their joint effect. We applied this approach to adenylosuccinate lyase (ADSL), a purine biosynthesis enzyme mutated in the autosomal recessive disorder ADSL deficiency. Using a yeast-based MAVE, we quantified the functional impact of over 8,000 coding variants. Our framework distinguished pathogenic from benign alleles based on estimated activity, and the integrated pathogenicity score correlated strongly with biochemical measurements from patient-derived cells, outperforming existing computational predictors. This dual innovation--the mechanistic transformation of MAVE data and allelic integration--offers a generalizable strategy for probing enzyme function and interpreting genetic variation in recessive disorders.

genetics↗

bioRxiv2025

Why variant effect predictors and multiplexed assays agree and disagree

Computational variant effect predictors (VEPs) and multiplexed assays of variant effect (MAVEs) are two key tools for assessing the functional consequences of genetic variants. While their outputs are often concordant, there are also many differences. Here, we analyse missense MAVE data from 37 different human proteins, comparing them to five state-of-the-art VEPs in order to quantify and explain their points of agreement and disagreement. We find that discordance is not random but reflects fundamental differences in how each method infers functional impact. VEPs, which rely heavily on sequence conservation and basic structural features, tend to overcall pathogenicity at buried and hydrophobic residues, while underestimating impact in disordered regions and on charged surface residues. MAVEs, by contrast, capture context-specific mechanisms more accurately, but can miss pathogenic variants when the assay fails to reflect disease biology, or be subject to high levels of experimental noise. By comparing both global patterns and specific clinically relevant variants, we show how protein features, assay design, and variant type shape predictive discordance. Our findings provide a framework for interpreting when and why VEPs and MAVEs diverge and point toward strategies for improving variant interpretation through integrated, mechanism-aware approaches.

bioinformatics↗

bioRxiv2025

A knowledge-based distance metric highlights underperformance of variant effect predictors on gain-of-function missense variants

Most current variant effect predictors (VEPs) are less effective at identifying non-loss-of-function (non-LOF) disease variants, which arise due to a gain of function (GOF) or dominant-negative (DN) mechanism. Previously, we showed that Extent of Disease Clustering (EDC), a structure-derived metric, can be used to prioritize putative non-LOF genes due to the tendency of their pathogenic missense variants to cluster. We also introduced a structure-derived spatial distance metric, Spatial Proximity to Disease Variants (SPDV), which proved useful for scoring RyR1 variants. Here, we apply EDC and SPDV across thousands of human disease genes. In comparison with 72 variant effect predictors (VEPs), SPDV exceeds or is at least as effective as half of the tested models at identifying dominant disease variants. When applied to GOF variants, SPDV performs similarly to top VEPs. We show that SPDV performance improves with increasing target EDC values and identify an EDC threshold after which SPDV consistently surpasses all other predictors. Its use as an analytical metric enables the identification of a subset of putative GOF disease genes where current VEP methods underperform relative to SPDV, indicating they may be improved by incorporating spatial distance information.

bioinformatics↗

bioRxiv2025

Disease-associated genetic variants can cause mutations in tissue-specific protein isoforms

Genetic variants can cause protein-coding mutations that result in disease. Variants are typically interpreted using the reference transcript for a gene. However, most human multi-exon genes encode alternative isoforms. Here, we show that coding exons in alternative isoforms harbour more population variants than exons of reference isoforms, consistent with their reduced evolutionary constraint, and that these variants are more likely to cause nonsynonymous coding mutations. Common and rare disease-associated variants mapping to alternative transcripts can lead to amino acid substitutions predicted to be structurally damaging in the corresponding protein isoform. The alternative transcripts to which disease-associated variants map demonstrate high tissue-specific expression, with many unannotated in reference human genomes, revealed only by long-read RNA-sequencing. As an example, we report an unannotated alternative transcript of the inflammasome regulator DPP9 that is lung epithelium-specific and which harbours a common genetic variant associated with severe COVID-19 and lung fibrosis. The variant causes a p.Leu8Pro missense mutation in an alternative first exon, predicted to disrupt the encoded alpha helix. These findings highlight the importance of considering alternative isoforms, their tissue-specific expression, and full-length transcripts in variant interpretation, with implications for uncovering underappreciated mechanisms of both common and rare disease.

genomics↗

bioRxiv2025

Epitope Effect Prevalence in Affinity-based pQTL studies

Affinity-based proteomics platforms Olink and SomaScan have enabled population-wide human proteogenomic studies, linking genetic variants to protein abundances. However, sequence alterations in the binding site of the detection reagent may introduce platform-specific measurement bias unrelated to protein levels, referred to as the epitope effect. In this study, we investigated the prevalence of epitope effects using cis protein quantitative trait loci (pQTL) discovered in three of the largest proteogenomic studies: UK Biobank, deCODE, and Fenland. Across 5,817 protein targets assayed in these studies, cis-pQTL were identified for 914 proteins by both platforms, 301 (33%) of which were linked to a missense variant. We identified 37 proteins with opposing effect directions in two platforms for the same missense pQTL, and 85 proteins where a missense pQTL was significant in only one platform. We present examples where such discrepancies reflect differences in isoform or proteoform targeting, as well as examples where the discordance appears to result from true platform-specific detection bias. Further structural analyses reveal that missense cis-pQTL are more likely to be detected when they alter residues located on accessible protein surfaces - regions most likely to interfere with reagent binding in affinity-based assays. Overall, our findings suggest that missense-mediated epitope effects influence only a minority (12% or less) of cis-pQTL results. We also highlight the need for detailed assay annotations and structural context to improve result interpretation in proteogenomic studies.

genomics↗

bioRxiv2025

acmgscaler: An R package and Colab for standardised gene-level variant effect score calibration within the ACMG/AMP framework

SummaryA genome-wide variant effect calibration method was recently developed under the guidelines of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG/AMP), following ClinGen recommendations for variant classification. While genome-wide approaches offer clinical utility, emerging evidence highlights the need for gene- and context-specific calibration to improve accuracy. Building on previous work, we have developed an algorithm tailored to converting functional scores from both multiplexed assays of variant effects (MAVEs) and computational variant effect predictors (VEPs) into ACMG/AMP evidence strengths. Our method is designed to deliver consistent performance across different genes and score distributions, with all variables adaptively determined from the input data, preventing selective adjustments or overfitting that could inflate evidence strengths beyond empirical support. To facilitate adoption, we introduce acmgscaler, a lightweight R package and a plug-and-play Google Colab notebook for the calibration of custom datasets. This algorithmic framework bridges the gap between MAVEs/VEPs and clinically actionable variant classification. AvailabilityThe package and the Colab notebook are available on GitHub: https://github.com/badonyi/acmgscaler. Supplementary dataSupplementary data can be downloaded from the OSF repository https://osf.io/7hjnm.

bioinformatics↗