Biology subjects

Unger, M.

Publications and source records attributed to Unger, M..

13 recordsLinked to original sources

bioRxiv2026

Ex vivo activation unmasks a sex-convergent, exhaustion-associated CD8+ T cell expansion in Parkinsons disease

Parkinsons disease (PD) affects an estimated 6.4 million men and 5.3 million women worldwide, and still lacks a validated peripheral biomarker. Brain tissue is inaccessible in living patients, making peripheral blood an attractive alternative, but existing studies have profiled immune cells almost exclusively as static, resting-state snapshots that cannot reveal how those cells function under challenge. A longitudinal design, following patients over time, or applying a controlled stimulus to reveal functional differences invisible at rest, offers a more sensitive window onto disease-associated dysfunction, and sex, despite differential PD incidence and progression, is rarely treated as a primary variable. Here, using ex vivo PMA/ionomycin stimulation as a controlled functional challenge, we profiled 195k PBMCs from 84 samples of 14 PD patients and 14 controls by single-cell RNA sequencing across an activation time course (0h, 2h, 4h), stratified by sex throughout. Sex explained more transcriptional variance than disease status, and male and female PD patients showed largely divergent responses at rest that converged, by peak activation, on a discrete CD8+ effector memory T cell subpopulation (Tem-CD8). This subpopulation showed an exhaustion-consistent programme, coinciding with a failure to resolve AP-1 induction and a reduction in inferred intercellular communication. The PD peripheral immune phenotype is therefore better characterised as a activation-dependent response than a fixed resting-state signature, identifying Tem-CD8 exhaustion as a disease-associated, sex-convergent candidate for further study.

neuroscience↗

bioRxiv2026

Buoyancy-driven sorting of synthetic cells for nanopore activity

The development of sorting strategies that directly report on functional activity remains a bottleneck in synthetic cell research. Current methodologies typically rely on sequential label-dependent probing, which limits throughput. Here, we introduce a label-free, buoyancy-driven selection strategy in which the mode of separation and the mode of decision-making are intrinsically linked, coupling pore activity directly to the synthetic vesicles internal density in a one-pot assay. In this system, sorting emerges intrinsically: Giant unilamellar vesicles (GUVs) that contain a dense medium sediment by default, while only those with functional transmembrane pores undergo solute exchange, leading to density equilibration and flotation. We exploit this principle to separate pore-active from non-functional GUVs without external markers or imaging-based readouts. Using protein pores and DNA origami and DNA tile nanopores, we demonstrate that buoyancy-driven separation enables parallel functional assessment of heterogeneous populations and supports flow-based enrichment of highly active synthetic cells. By directly linking molecular transport performance to GUV buoyancy, this approach collapses decision-making into the physical separation process itself, providing a scalable platform for screening, sorting, and evolving membrane pores in synthetic cell systems.

biophysics↗

bioRxiv2025

Cell-in-Water: measurements using Optical Photothermal Infrared spectroscopy

Conventional Fourier transform infrared absorption spectroscopy (FT-IR) allows for measurements of spatial resolution comparable to a whole cell size, which are performed on dried cells to avoid water interference. However, such conditions can disrupt the native morphology and chemical composition of cells, whilst not providing for crucial sub-cellular biochemical detail. Pump-probe Optical Photothermal Infrared (O-PTIR) Microscopy offers high-resolution measurements of cells, but the challenge of measuring cells without disturbing their morphology remains. To overcome this, we developed a protocol enabling O-PTIR measurements of cells in aqueous environments using a silicone-sealed microchamber, which allows imaging of live cells in 2 {micro}l of buffer for up to 4 hours. Analysis of O-PTIR data obtained from hydrated and dehydrated cells confirms that hydrated conditions better preserve subcellular organization, as evidenced by secondary structure changes of proteins. We also propose a step-by-step algorithm that can improve image quality through ratiometric analysis, histogram assessment, and the application of RGB overlays. This approach allows for visualization of multiple subcellular components, such as the nucleus or lipid droplets. Trough the combination of "sandwich" sample configuration and O-PTIR, we offer a promising strategy to advance O-PTIR measurements of cells in water, which have long been considered a major challenge. A direct comparison of O-PTIR with Raman microscopy highlights the complementarity of these two techniques and similar spatial resolution.

biophysics↗

bioRxiv2025

Deep Learning for Biomarker Discovery in Cancer Genomes

BackgroundAccurate determination of genomic biomarkers from tumor sequencing is fundamental to precision oncology, informing disease classification and treatment decisions. In practice, biomarker inference relies on computational pipelines that often compress high-dimensional mutation data into predefined summaries such as mutational signatures or composite genomic features. While robust and widely adopted, these representations may not fully capture the complexity of cancer genomes. Deep learning (DL) offers an end-to-end alternative by learning features directly from raw genomic data. However, clinical translation remains challenging due to limited empirical validation of new DL models and a lack of systematic comparisons with established machine learning (ML) baselines, particularly when transitioning from information-rich genome or exome data to real-world targeted sequencing profiles. Here, we compare state-of-the-art DL architectures with classical ML models across variant-level, copy-number (CNV), and multimodal inputs, using microsatellite instability (MSI) and homologous recombination deficiency (HRD) prediction as oncologically relevant tasks. We aim to derive practical guidance on modelling strategies across different data modalities and clinical sequencing contexts. MethodsFor MSI and HRD prediction, we trained multiple DL models, including supervised and self-supervised encoders, alongside feature-based ML approaches using tumor mutation data, copy-number alterations, and their multimodal combinations. Analyses were conducted on 5,647 patients in The Cancer Genome Atlas (TCGA), the Clinical Proteomic Tumor Analysis Consortium (CPTAC), and two targeted sequencing panel cohorts. Model performance was evaluated on both whole-exome and panel-based datasets, and explainability analysis were performed for both DL and ML models. ResultsFor MSI, DL demonstrated stronger generalization than ML on external validation data (F1 0.97 vs 0.76) and maintained comparatively high performance under pseudo-panels conditions, whereas ML performance dropped. In a real-world targeted panel cohort, DL again showed more robust generalization than ML, with performance partly affected by cross-assay variability. For HRD, incorporation of CNV data was the primary determinant of predictive performance. Once CNVs were included, DL and ML achieved similar accuracy on external datasets (F1 0.61 vs 0.58). In panel-based settings, DL retained an advantage over ML (F1 0.78 vs 0.62). Model interpretation analyses indicated that both DL and ML relied on mutation and chromosomal patterns consistent with established MSI and HRD biology. ConclusionOverall, predictive performance depended strongly on data availability and clinical sequencing context. When information-rich inputs were available, both DL and classical ML achieved robust biomarker prediction, with DL generally matching or exceeding ML performance. The most pronounced advantages of DL emerged in cross-assay evaluations and data-sparse settings, where generalization was more reliable. Notably, the best-performing DL models were lightweight and interpretable, supporting practical deployment. In clinical genomics workflows, such models may complement established pipelines by leveraging patient sequencing data to provide additional evidence for treatment-relevant biomarker assessment.

bioinformatics↗

bioRxiv2024

Counterfactual Diffusion Models for Mechanistic Explainability of Artificial Intelligence Models in Pathology

Deep learning can extract predictive and prognostic biomarkers from histopathology whole slide images. However, explainable artificial intelligence approaches widely used in digital pathology, such as attention heatmaps and class activation mapping, offer only limited interpretability regarding the features captured by classifiers. Here, we present MoPaDi (Morphing histoPathology Diffusion), a framework for generating counterfactual explanations for histopathology images that reveal which morphological or style features drive classifier predictions. MoPaDi combines diffusion autoencoders with task-specific multiple instance learning classifiers to manipulate images and flip predictions by modifying relevant features. We evaluated the framework on multiple datasets spanning colorectal, breast, liver, and lung cancers, including tissue type, cancer subtype, and biomarker (microsatellite instability) classification tasks. We assessed counterfactual explanations through quantitative analyses, pathologists evaluations, and independent foundation model-based classifiers. We found that MoPaDi was able to generate realistic counterfactual histopathology images, enabling pathologists to identify morphological features associated with the change in model predictions. Unlike conventional reviews of highly attended regions typical in digital pathology, MoPaDi explanations enabled pathologists to directly identify morphological features driving the classifiers predictions from a limited number of top-contributing tiles. Consistent with the literature, our biomarker classifier associated high microsatellite instability with mucinous differentiation, glandular patterns, and lymphocytic infiltration. Furthermore, MoPaDi revealed that changes in classifier predictions were mainly driven by morphological alterations rather than staining differences. Overall, MoPaDi is a practical framework for counterfactual explanations in computational pathology that reveals model-specific drivers of classification and increases trust in deep learning models.

bioinformatics↗

bioRxiv2023

A Small Tau Fragment Specifically Templates Four Repeat Tau Aggregates Through Multiple Generations

Prion-like spread of disease-specific tau conformers is a hallmark of all tauopathies. A 19-residue probe peptide containing a P301L mutation and spanning the R2/R3 splice junction of tau, folds and stacks into seeding-competent fibrils and induces aggregation of 4R, but not 3R tau. These tau peptide fibrils propagate aggregated intracellular tau over multiple generations, have a high {beta}- sheet content, a colocalized lipid signal, and adopt a well-defined U-shaped fold found in 4R tauopathy brain-derived fibrils. Fully atomistic replica exchange molecular dynamics (MD) simulations were used to compute the free energy landscapes of the conformational ensemble of the peptide monomers. These identified an aggregation-prohibiting {beta}-hairpin structure and an aggregation-competent U-fold unique to 4R tauopathy fibrils. Guided by MD simulations, we identified that the N-terminal-flanking residues to PHF6, which slightly vary between 4R and 3R isoforms, modulate seeding. Strikingly, when a single amino acid switch at position 305 replaced the serine of 4R tau with a lysine from the corresponding position in the first repeat of 3R tau, the seeding induced by the 19-residue peptide was markedly reduced. Conversely, a 4R tau mimic with three repeats, prepared by replacing those amino acids in the first repeat with those amino acids uniquely present in the second repeat, recovered aggregation when exposed to the 19- residue peptide. These peptide fibrils function as partial prions to recruit naive 4R tau--ten times the length of the peptide--and serve as a critical template for 4R tauopathy propagation. These results hint at opportunities for tau isoform-specific therapeutic interventions. Significance StatementA structural motif corresponding to a short junction sequence spanning R2 and R3 forms fibrils that adopt a fold characteristic of 4R tauopathy fibrils and induces misfolding of the larger tau protein with loss of microtubule binding and a prion-like specificity for 4R tau. Simulations, validated experimentally, pinpointed the specific amino acids in the peptide that can toggle its properties between aggregation competent and incompetent. The modifications suggest design principles for a therapeutic intervention potentially capable of disaggregating tau or preventing its aggregation in the 4R tauopathies.

molecular biology↗

bioRxiv2023

Piezo regulates epithelial topology and promotes precision in organ size control

Mechanosensitive Piezo channels regulate cell division through calcium-mediated activation of ERK signaling or activate Rho signaling to mediate cell extrusion and cell death. However, systems-level functions of Piezo in regulating organogenesis remain poorly understood. Here, we demonstrate that Piezo controls epithelial cell topology to ensure precise organ growth through the integration of live imaging experiments with pharmacological and genetic perturbations and computational modeling. Notably, knockout or knockdown of Piezo led to bilateral asymmetry in wing phenotypes. While pharmacological activation of Piezo stimulated an increase in the frequency of spikes in cytosolic Ca2+, we discovered that Piezo overexpression counterintuitively reduces Ca2+ signaling dynamics. Knockdown of Piezo inhibited proliferation and decreased apoptosis, resulting in an overall increase in epithelial overcrowding. In contrast, either genetic overexpression or pharmacological activation of Piezo increased cell proliferation and cell removal through basal extrusion. Surprisingly, Piezo overexpression increased the hexagonality of cellular topology. To test whether Piezo regulates cell topology, we formulated computational simulations to investigate how expression levels of Piezo protein regulate cell proliferation and apoptosis through modulation of the cut-off tension required for Piezo channel activation. Quantitative analysis validated computational simulation predictions of how perturbations to Piezo impacted epithelial topology. Overall, our findings demonstrate that Piezo promotes robustness in regulating epithelial topology and is necessary for precise organ size control.

developmental biology↗

bioRxiv2023

Expression of ALS-PFN1 impairs vesicular degradation in iPSC-derived microglia

Microglia play a pivotal role in neurodegenerative disease pathogenesis, but the mechanisms underlying microglia dysfunction and toxicity remain to be fully elucidated. To investigate the effect of neurodegenerative disease-linked genes on the intrinsic properties of microglia, we studied microglia-like cells derived from human induced pluripotent stem cells (iPSCs), termed iMGs, harboring mutations in profilin-1 (PFN1) that are causative for amyotrophic lateral sclerosis (ALS). ALS-PFN1 iMGs exhibited lipid dysmetabolism and deficits in phagocytosis, a critical microglia function. Our cumulative data implicate an effect of ALS-linked PFN1 on the autophagy pathway, including enhanced binding of mutant PFN1 to the autophagy signaling molecule PI3P, as an underlying cause of defective phagocytosis in ALS-PFN1 iMGs. Indeed, phagocytic processing was restored in ALS-PFN1 iMGs with Rapamycin, an inducer of autophagic flux. These outcomes demonstrate the utility of iMGs for neurodegenerative disease research and highlight microglia vesicular degradation pathways as potential therapeutic targets for these disorders.

neuroscience↗

bioRxiv2022

G protein-coupled Receptor Contributions to Wing Growth and Morphogenesis in Drosophila melanogaster

The development of multicellular organisms relies on a symphony of spatiotemporally coordinated signals that regulate gene expression. G protein-coupled receptors (GPCRs) are the largest group of transmembrane receptors that play a pivotal role in transducing extracellular signals into physiological outcomes. Emerging research has implicated neurotransmitter GPCRs, classically associated with communication in neuronal tissues, as regulators of pattern formation and morphogenesis. However, how these receptors interact amongst themselves and signaling pathways to regulate organogenesis is still poorly understood. To address this gap, we performed a systematic RNA interference (RNAi)-based screening of 111 GPCRs along with 8 G, 3 G{beta}, and 2 G{gamma} protein subunits in Drosophila melanogaster. We performed a coupled, machine learning-based quantitative and qualitative analysis to identify both severe and more subtle phenotypes. Of the genes screened, 25 demonstrated at least 60% penetrance of severe phenotypes with several of the most severe phenotypes resulting from the knockdown of neuropeptide and neurotransmitter GPCRs that were not known previously to regulate epithelial morphogenesis. Phenotypes observed in positive hits mimic phenotypic manifestations of diseases caused by dysregulation of orthologous human genes. Quantitative reverse transcription polymerase chain reaction and meta-analysis of RNA expression validated positive hits. Overall, the combined qualitative and quantitative characterization of GPCRs and G proteins identifies an extensive set of GPCRs involved in regulating epithelial morphogenesis and relevant to the study of a broad range of human diseases.

developmental biology↗

bioRxiv2022

The pharmacokinetics and pharmacodynamics of 4-methylumbelliferone and its glucuronide metabolite in mice

Hyaluronan (HA) is an extracellular matrix glycosaminoglycan, with important roles in chronic inflammation, cancer and autoimmunity. 4-methylumbelliferone (4-MU), a small molecule inhibitor of HA synthases, is widely used to study HAs interactions with the surrounding tissues and the immune cells. There is substantial experimental and therapeutic interest in using oral 4-MU to inhibit HA synthesis, but pharmacokinetic and pharmacodynamic data on treatment routes have been lacking. Moreover, it recently became clear that the main metabolite of 4-MU, 4-methlyumbelliferyl glucuronide (4-MUG), is bioactive. We therefore sought to define the pharmacokinetics and pharmacodynamics of 4-MU and its active metabolite 4-MUG in mice. Single dose mouse studies showed that 4-MU administered intravenously (i.v.) resulted in 100-fold higher 4-MU exposure compared to oral (p.o.) administration. The 4-MU ratio AUC i.v./AUC p.o. was 96/1. 4-MUG exposures were much higher than 4-MU exposures after both 4-MU i.v. and p.o. administration, but only small differences in 4-MUG exposure were seen after 4-MU i.v. versus p.o. administration. The 4-MUG metabolite was also administered as a single dose both i.v. and p.o. and showed a 25.9% bioavailability. Compared to 4-MUG p.o. dosing, 1.14 higher 4-MUG exposures were seen after 4-MU p.o. dosing. 4-MU exposure after 4-MUG p.o. administration was minimal but similar to 4-MU exposure after 4-MU p.o. administration. In mice treated for several weeks with 4-MU in chow, the 4-MU concentration immediately drops after treatment was stopped, whereas the 4-MUG concentration showed a peak 1 hour after treatment stop. In a build-up study, 4-MU and 4-MUG treatment in mice lead to a plateau of 4-MU concentration starting at 4 days post treatment start. These 4-MU and 4-MUG concentration findings in vivo will inform future clinical studies and experimental work with 4-MU.

pharmacology and toxicology↗

bioRxiv2020

MAPPER: A new image analysis pipeline unmasks differential regulation of Drosophila wing features

Phenomics requires quantification of large volumes of image data, necessitating high throughput image processing approaches. Existing image processing pipelines for Drosophila wings, a powerful model for studying morphogenesis, are limited in speed, versatility, and precision. To overcome these limitations, we developed MAPPER, a fully-automated machine learning-based pipeline that quantifies high dimensional phenotypic signatures, with each dimension representing a unique morphological feature. MAPPER magnifies the power of Drosophila genetics by rapidly identifying subtle phenotypic differences in sample populations. To demonstrate its widespread utility, we used MAPPER to reveal new insights connecting patterning and growth across Drosophila genotypes and species. The morphological features extracted using MAPPER identified the presence of a uniform scaling of proximal-distal axis length across four different species of Drosophila. Observation of morphological features extracted by MAPPER from Drosophila wings by modulating insulin signaling pathway activity revealed the presence of a scaling gradient across the anterior-posterior axis. Additionally, batch processing of samples with MAPPER revealed a key function for the mechanosensitive calcium channel, Piezo, in regulating bilateral symmetry and robust organ growth. MAPPER is an open source tool for rapid analysis of large volumes of imaging data. Overall, MAPPER provides new capabilities to rigorously and systematically identify genotype-to-phenotype relationships in an automated, high throughput fashion. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=102 SRC="FIGDIR/small/422888v1_ufig1.gif" ALT="Figure 1"> View larger version (32K): org.highwire.dtl.DTLVardef@bdbecdorg.highwire.dtl.DTLVardef@139a7fdorg.highwire.dtl.DTLVardef@43d5b9org.highwire.dtl.DTLVardef@1846af9_HPS_FORMAT_FIGEXP M_FIG C_FIG

developmental biology↗

bioRxiv2020

Functional and molecular early enteric biomarkers for Parkinson's disease in mice and men

Parkinsons disease (PD) usually has a late clinical onset. The lack of early biomarkers for the disease represents a major challenge for developing timely treatment interventions. Here, we use an -synuclein-overexpressing transgenic (Th-1-SNCA-A30P) mouse model of PD to identify appropriate candidate markers in the gut for early stages of PD before hallmark symptoms begin to manifest. A30P mice did not show alterations in gait parameters at 2 months of age, and these mice were therefore defined as pre-symptomatic A30P mice (psA30P). We discovered early functional motility changes in the gut and early molecular dysregulations in the myenteric plexus of psA30P mice by comparative protein and miRNA profiling and cell culture experiments. We found that the proteins neurofilament light chain, vesicle-associated membrane protein 2 and calbindin 2, together with the miRNAs that regulate them, are potential biomarkers of early PD that may facilitate timely treatment and/or prevention of PD in men.

neuroscience↗

bioRxiv2019

Positional isomers of a non-nucleoside substrate differentially affect myosin function

Molecular motors have evolved to transduce chemical energy from adenosine triphosphate into mechanical work to drive essential cellular processes, from muscle contraction to vesicular transport. Dysfunction of these motors is a root cause of many pathologies necessitating the need for intrinsic control over molecular motor function. Herein, we demonstrate that positional isomerism can be used as a simple and powerful tool to control the molecular motor of muscle, myosin. Using three isomers of a synthetic non-nucleoside triphosphate we demonstrate that myosins force and motion generating capacity can be dramatically altered at both the ensemble and single molecule levels. By correlating our experimental results with computation, we show that each isomer exerts intrinsic control by affecting distinct steps in myosins mechano-chemical cycle. Our studies demonstrate that subtle variations in the structure of an abiotic energy source can be used to control the force and motility of myosin without altering myosins structure. Statement of SignificanceMolecular motors transduce chemical energy from ATP into the mechanical work inside a cell, powering everything from muscle contraction to vesicular transport. While ATP is the preferred source of energy, there is growing interest in developing alternative sources of energy to gain control over molecular motors. We synthesized a series of synthetic compounds to serve as alternative energy sources for muscle myosin. Myosin was able to use this energy source to generate force and velocity. And by using different isomers of this compound we were able to modulate, and even inhibit, the activity of myosin. This suggests that changing the isomer of the substrate could provide a simple, yet powerful, approach to gain control over molecular motor function.

biophysics↗