Biology subjects

Fleming, J.

Publications and source records attributed to Fleming, J..

18 recordsLinked to original sources

bioRxiv2026

Cancer cells hijack mitochondria from Bhlhe40+ T cells for immune evasion

Cancer cells hijack mitochondria from T cells via tunneling nanotubes (TNTs) to promote immune evasion, yet the underlying mechanisms of mitochondrial-loss remain poorly understood. Here, through targeted screening of relevant genes, we demonstrated that CD38 is a potent regulator of TNT formation through interaction with its cognate ligand, CD31. Conditional knockdown of Cd38 in murine models leads to CD8+ T cell-dependent tumor shrinkage. Single-cell RNA sequencing of tumor-infiltrating T cells revealed that cancer cells preferentially target the Bhlhe40+ effector subset. Bhlhe40 is a transcription factor that upregulates the expression of S100a4, a reported chemoattractant of nanotubes. Bhlhe40 upregulation creates mitochondrial-loss vulnerability for tumor neoantigen-specific T cells. Dual mitochondria transfer blockade with Cd38 knockdown and S100a4 inhibitor trifluoperazine markedly suppressed tumor growth. Collectively, our findings identified the Bhlhe40-S100a4 cascade as a critical regulator of cancer/T cell mitochondrial transfer, providing insights for the development of new cancer immunotherapies.

immunology↗

bioRxiv2026

Antigen-aware TCR embeddings reveal conserved immune patterns for personalized immune monitoring

T cell receptor (TCR) repertoires are highly individualized, while technical variation further complicates comparisons across individuals and cohorts. Here, we present TCRmeta, an antigen-aware framework that integrates population-scale repertoire relationships with experimentally defined antigen specificity to establish a shared space for repertoire analysis. TCRmeta predicted mutation-induced changes and captured antigen-related relationships in TCR specificity. Despite extensive sequence diversity, TCRmeta revealed a conserved repertoire landscape, enabling construction of a healthy reference for quantifying repertoire dispersal. Across cohorts, aging was associated with conserved V gene changes and progressive dispersal, with greater dispersal in CMV-positive individuals. Longitudinal analysis showed that repertoires progressively shifted toward the healthy reference following bone marrow transplantation or antiretroviral therapy. To directly compare repertoires in TCRmeta space, we developed Energy Shift, a pairwise measure robust to technical variation and HLA background while remaining sensitive to biological differences. Together, TCRmeta enables quantitative cross-individual and longitudinal immune monitoring.

bioinformatics↗

bioRxiv2026

The Human Bindome: A Proteome-scale Atlas of Designed Binder Candidates

Affinity reagents such as antibodies are indispensable for interrogating proteins biological function. Yet they are costly and frequently unreliable, with unknown sequences, posing challenges to reproducible experimental research. Deep learning-based protein design can now in silico generate affinity reagents achieving reliable experimental success rates, but has remained largely confined to specialist laboratories. Here we present the Human Bindome, a proteome-scale atlas of high-confidence in silico protein binder candidates. By embedding the experimentally benchmarked BindCraft method in an accelerated, parallelized framework with automated domain-level target selection, we generated 306,146 binder candidates covering 8,296 human proteins (40.9% of the full proteome). Every candidate carries a defined sequence, a predicted binder-target structure model, and in silico confidence metrics. We characterize proteome-wide coverage and show that binder epitopes frequently overlap functional sites. This positions the Bindome as a resource of genetically encodable perturbagens for site-specific, modular control of protein function. The Bindome is freely available through a web interface (https://bindome.epfl.ch), with agentic, natural-language querying and as data splits for machine-learning model development. We anticipate that the Bindome will be valuable for the scientific community by providing affinity and perturbation reagents with broad applications in dissecting biological mechanisms as well as in drug and target discovery.

synthetic biology↗

bioRxiv2026

Benchmarking Boltz-2 for Screening of Therapeutic Antibody-Antigen Interactions

Protein structure prediction models (such as AlphaFold, Chai, Boltz) have transformed structural biology and are increasingly explored for drug discovery; however, their utility for large-scale screening of antibody-antigen (AB-AG) interactions remains unclear, particularly for distinguishing true binding from non-binding pairs at scale. To our knowledge, there has not been an exhaustive exploration of Boltz-2 inference settings on this high impact problem, and in this paper we set out to describe and implement a novel benchmarking framework that can accelerate progress in the field. We evaluated Boltz-2 (NVIDIA NIM implementation) on 519 therapeutic monoclonal antibodies from Thera-SAbDab, pairing each antibody with its cognate target and a randomly assigned non-cognate antigen. We developed a novel evaluation framework that systematically captures variability across stochastic seeds while benchmarking different inference settings, including datasets with and without crystallographically resolved antibody structures. Across settings, Boltz-2-derived confidence metrics showed weak, though above-chance, discrimination (0.5 < ROC-AUC < 0.60). Among evaluated metrics, the minimum value of the interface predicted TM-score (ipTM-min) across seed-samples, captured the strongest signal. Interestingly, additional feature aggregation and multivariate modelling provided little to no improvement. Increasing the number of stochastic predictions yielded front-loaded gains, with diminishing returns beyond [~]15-20 seed-samples, suggesting limited value of extensive sampling in practical workflows. Notably, inference without multiple sequence alignments (MSAs) slightly improved performance on non-crystallized antibodies ({Delta}AUROC {approx} +0.027) while reducing runtime by [~]8 seconds per prediction compared to shallow MSA settings. Overall, these results indicate that off-the-shelf confidence metrics from general-purpose structure prediction models may be insufficient for reliable target-antibody screening and highlight the need for task-specific optimization, while confirming that modest amounts of sampling can be helpful, but not in itself sufficient to improve performance significantly as gains plateau relatively quickly.

bioinformatics↗

bioRxiv2026

AI-enabled virtual immunopeptidomics links quantitative neoantigen presentation to immunogenicity

Effective anti-tumor T cell response depends on both neoantigen quality (non-selfness) and quantity (abundance). However, existing methods for neoantigen prioritization largely overlook peptide abundance because it is difficult to measure and model. To bridge this gap, we developed epiVIP, a deep learning framework that predicts the abundance of individual HLA-I peptides using widely available (sc)RNA-seq data. Trained on 1.7 million immune peptides paired with gene expression profiles, epiVIP demonstrated strong generalizability across unseen samples. Analyzing 33,711 neoantigens from clinical datasets revealed a compensatory relationship between abundance and non-selfness in determining antigenicity, providing quantitative support for the TCR avidity theory. Importantly, abundance independently predicted tumor reactivity and patient survival across multiple neoantigen vaccine cohorts and immune checkpoint blockade cohorts. Mechanistic interpretation of epiVIP further identified directional regulation of MAGEA3 epitope presentation by PSME4, which was validated experimentally using T cell functional assays. Together, these findings established AI-enabled virtual immunopeptidomics as a powerful strategy to improve cancer immunotherapy.

bioinformatics↗

bioRxiv2026

AlphaFold Database expands to proteome-scale quaternary structures

Protein function is governed by molecular interactions, yet structural coverage of these interactions remains sparse. The AlphaFold Protein Structure Database (AFDB) transformed access to accurate monomeric protein structures at scale. Here, we expand the AFDB to quaternary structures by predicting 31M candidate homo- and heterodimeric protein complexes, compiled from 4,777 proteomes, including model- and global health organisms. We established confidence criteria via analysis of experimentally determined structures, resulting in 1.81M high-confidence predictions. These models enabled the discovery of emergent structures and topologies not present in monomeric predictions. Additionally, the top 1% of structural clusters accounted for [~]44% of all complexes, and [~]8.3% of clusters were conserved across multiple domains of life, pointing to a substantial fraction of ancient, universally retained assemblies. Structural search highlighted that high-confidence predictions are anchored in experimental multimer space, yet 31.3% extend beyond detectable PDB coverage. These freely accessible proteome-scale predictions facilitate functional and mechanistic hypothesis generation across biology.

synthetic biology↗

bioRxiv2025

Quantitative and large-scale investigation of human TCR-HLA Cross-Reactivity

The interaction between Human leukocyte antigens (HLA) and T cell receptor (TCR) is essential for adaptive immune recognition. While it is known that one TCR can map to multiple HLA alleles, the extent of this cross-reactivity remains poorly understood. Here, we introduce THNet, a TCR-based HLA similarity inference method, and performed a comprehensive analysis of HLA-TCR cross-reactivity. This method is built upon clustering over 47 million TCRs to identify over 9 million significant HLA-TCR pairs. We created similarity networks for both class I and class II HLA alleles, illustrating how peptide cross-presentation contributes to HLA-TCR cross-reactivity. This analysis revealed novel disease-susceptibilities missed by single-HLA enrichment analyses, especially in the Black populations. Finally, we demonstrated that THNet prioritized optimal HLA mismatch candidates for organ transplantation, thereby improving patient survival rates. Our investigation of HLA-TCR cross-reactive network might provide useful insights for autoimmune risk prediction and better transplantation outcomes. One Sentence SummaryWe introduced THNet, a large-scale TCR-based HLA similarity mapping network that uncovers previously unrecognized cross-reactivity patterns across HLA alleles and provides valuable insights into their influence on disease susceptibility and graft rejection.

bioinformatics↗

bioRxiv2024

Finding Salient Multi-Omic Interactomes Relevant to Multiple Biomedical Outcomes using Graph Ensemble Neural Networks

Although multi-omics integration relevant to patient outcome is typically characterized by an analyte interactome, current multi-omic integration methods either (1) model outcome without directly including associations between analytes, (2) model the interactome without directly evaluating the saliency of the model in the context of outcome, or (3) model outcome in a high-dimensional parameter space not suitable for small sample sizes (which are common in multi-omics studies). We introduce Graph Ensemble Neural Network (GENN), a methodology that learns the interactome most predictive of outcome in a low-dimensional parameter space built on complementary attributes for all possible analyte associations (metafeatures). We show that GENN is robust to noise in measurements using a theoretical model, outperforms the predictive performance of existing methods when evaluated on Tegafur drug response in NCI-60 cancer cell line data, and uncovers potentially novel multi-omic mechanisms driving total serum IgE levels in pediatric asthma and patient survival in glioblastomas.

bioinformatics↗

bioRxiv2024

Indel-driven evolution of the canavanine tRNA-editing deacetylase enzyme CtdA

AbstractProteins are heteropolymers composed of twenty standard amino acids. However, over 500 non-proteogenic amino acids exist in nature that can get misincorporated into proteins. Canavanine is an antimetabolite of L-arginine, with which it shares high chemical similarity. It can be utilized by bacteria such as Pseudomonas canavaninivorans in the legume rhizome as a sole source of carbon and nitrogen. However, canavanine is also incorporated in proteins of this bacterium as its arginyl-tRNA synthetase loads tRNAArg with both canavanine and arginine. The recently discovered canavanyl-tRNAArg deacetylase (CtdA) removes canavanine from misloaded tRNAArg and thereby prevents its incorporation in proteins. CtdA is the first enzyme known to edit tRNA mischarged with a non-proteinogenic amino acid. We have elucidated its crystal structure to 1.5 [A] resolution and studied its active site using site-directed mutagenesis. We found that CtdA is a small monomeric enzyme that presents a central, deep cavity that predictably constitutes the canavanine binding site and a positively charged surface area that likely coordinates the CCA-3 tRNA attachment sequence. The stand-alone, trans-editing CtdA is distantly related to the B3/B4 cis-editing domains of the large multi-subunit enzyme Phenylalanine tRNA synthetase (PheRS). Our comparative study reveals that CdtA and B3/B4 domains from bacterial and archeal/eukaryotic origin are three subclasses of a same conserved 3D-fold that differ in type-specific indels, which distinctly shape the substrate binding cleft of these proteins. We propose a unifying nomenclature of secondary structure elements for this 3D-fold. In CtdA, residues E191, Y104, N105 and E118 prove to be relevant for catalysis, of which N105 is conserved in bacterial B3/B4 domains. No other shared residues of catalytic relevance could be identified across enzymes of this class, so that a shared mechanism of catalysis appears unlikely in these editing enzymes.

biochemistry↗

bioRxiv2023

Development of a cellular reporter assay to measure activity of MutSβ, a therapeutic target for Huntington's disease

Genetic modifiers of age of onset in Huntingtons disease (HD) provide compelling evidence that somatic expansion of the CAG repeats is a critical driver of pathogenesis and demonstrate that repeat instability is modulated by DNA mismatch repair (MMR). A component of this pathway, MutS{beta}, a heterodimer comprised of MSH2 and MSH3, has emerged as a potential target for small-molecule therapeutic intervention. However, a robust cellular assay to interrogate genetic and pharmacological modifiers of MutS{beta} has not been reported. We have repurposed and optimized a tetranucleotide reporter assay to measure MutS{beta} activity in MMR-competent cells. We show that repeat instability is modulated by MSH3 protein levels and by its ATPase activity. In addition, we show that an inhibitor of HDAC3 modulates repeat instability, demonstrating the utility of the assay for pharmacological studies.

neuroscience↗

bioRxiv2023

Utilizing Aggregated Molecular Phenotype (AMP) Scores to Visualize Simultaneous Molecular Changes in Mass Spectrometry Imaging Data

Mass spectrometry imaging (MSI) has gained increasing popularity for tissue-based diagnostics due to its ability to identify and visualize molecular characteristics unique to different phenotypes within heterogeneous samples. Data from MSI experiments are often visualized using single ion images and further analyzed using machine learning and multivariate statistics to identify m/z features of interest and create predictive models for phenotypic classification. However, often only a single molecule or m/z feature is visualized per ion image, and mainly categorical classifications are provided from the predictive models. As an alternative approach, we developed an aggregated molecular phenotype (AMP) scoring system. AMP scores are generated using an ensemble machine learning approach to first select features differentiating phenotypes, weight the features using logistic regression, and combine the weights and feature abundances. AMP scores are then scaled between 0 and 1, with lower values generally corresponding to class 1 phenotypes (typically control) and higher scores relating to class 2 phenotypes. AMP scores therefore allow the evaluation of multiple features simultaneously and showcase the degree to which these features correlate with various phenotypes, leading to high diagnostic accuracy and interpretability of predictive models. Here, AMP score performance was evaluated using metabolomic data collected from desorption electrospray ionization (DESI) MSI. Initial comparisons of cancerous human tissues to normal or benign counterparts illustrated that AMP scores distinguished phenotypes with high accuracy, sensitivity, and specificity. Furthermore, when combined with spatial coordinates, AMP scores allow visualization of tissue sections in one map with distinguished phenotypic borders, highlighting their diagnostic utility.

bioinformatics↗

bioRxiv2023

Bioelectronic Zeitgebers: targeted neuromodulation to re-establish circadian rhythms.

Existing neurostimulation systems implanted for the treatment of neurodegenerative disorders generally deliver invariable therapy parameters, regardless of phase of the sleep/wake cycle. However, there is considerable evidence that brain activity in these conditions varies according to this cycle, with discrete patterns of dysfunction linked to loss of circadian rhythmicity, worse clinical outcomes and impaired patient quality of life. We present a targeted concept of circadian neuromodulation using a novel device platform. This system utilises stimulation of circuits important in sleep and wake regulation, delivering bioelectronic cues (Zeitgebers) aimed at entraining rhythms to more physiological patterns in a personalised and fully configurable manner. Preliminary evidence from its first use in a clinical trial setting, with brainstem arousal circuits as a surgical target, further supports its promising impact on sleep/wake pathology. Data included in this paper highlight its versatility and effectiveness on two different patient phenotypes. In addition to exploring acute and long-term electrophysiological and behavioural effects, we also discuss current caveats and future feature improvements of our proposed system, as well as its potential applicability in modifying disease progression in future therapies.

bioengineering↗

bioRxiv2022

Deep learning in human neurons predicts mechanistic subtypes of Parkinson's

Parkinsons disease (PD) is a common, devastating, and incurable neurodegenerative disorder. Several molecular mechanisms have been proposed to drive PD, with genetic and pathological evidence pointing towards aberrant protein homeostasis and mitochondrial dysfunction. PD is clinically highly heterogeneous, it is likely that different mechanisms underlie the pathology in different individuals, each requiring a specific targeted treatment. Recent advances in stem cell technology and fluorescent live-cell imaging have enabled the generation of patient-derived neurons with different mechanistic subtypes of PD. Here, we performed multi-dimensional fluorescent labelling of organelles in iPSC-derived neurons, in healthy control cells, and in four different disease subclasses. We generated a machine learning-based model that can simultaneously predict the presence of disease, and its primary mechanistic subtype. We independently trained a series of classifiers using both quantitative single-cell fluorescence variables and images to build deep neural networks. Quantitative cellular profile-based classifiers achieve an accuracy of 82%, whilst image based deep neural networks predict control, and four distinct disease subtypes with an accuracy of 95%. The classifiers achieve their accuracy across all subtypes primarily utilizing the organellar features of the mitochondria, with additional contribution of the lysosomes, confirming their biological importance in PD. Taken together, we show that machine learning approaches applied to patient-derived cells are able to predict disease subtypes, demonstrating that this approach may be used to guide personalized treatment approaches in the future.

cell biology↗

bioRxiv2021

Convergent evolution of polyploid genomes from across the eukaryotic tree of life

By modeling the homoeologous gene losses that occurred in fifty genomes deriving from ten distinct polyploidy events, we show that the evolutionary forces acting on polyploids are remarkably similar, regardless of whether they occur in flowering plants, ciliates, fishes or yeasts. The models suggest these events were nearly all allopolyploidies, with two distinct progenitors contributing to the modern species. We show that many of the events show a relative rate of duplicate gene loss prior to the first post-polyploidy speciation that is significantly higher than in later phases of their evolution. The relatively low selective constraint seen for the single-copy genes these losses produced lead us to suggest that most of the purely selectively neutral duplicate gene losses occur in the immediate post-polyploid period. We also find ongoing and extensive reciprocal gene losses (RGL; alternative losses of duplicated ancestral genes) between these genomes. With the exception of a handful of closely related taxa, all of these polyploid organisms are separated from each other by tens to thousands of reciprocal gene losses. As a result, it is very unlikely that viable diploid hybrid species could form between these taxa, since matings between such hybrids would tend to produce offspring lacking essential genes. It is therefore possible that the relatively high frequency of recurrent polyploidies in some lineages may be due to the ability of new polyploidies to bypass RGL barriers.

evolutionary biology↗

bioRxiv2021

Utilizing Pine Needles to Temporally and Spatially Profile Per- and Polyfluoroalkyl Substances

As concerns continue to mount over exposure to per- and polyfluoroalkyl substances (PFAS), novel methods of profiling their presence and modifications are greatly needed as some have known toxic and bioaccumulative characteristics while others have unknown effects. This task however is not simple as over 5000 PFAS of interest have been named by the Environmental Protection Agency and this list continues to grow daily. In this work, we utilized widely available archived and field-sampled pine needles and a novel non-targeted analytical method to evaluate the temporal and spatial presence of numerous PFAS. Over 70 PFAS were detected in the pine needles from this study, providing information from the last six decades related to PFAS exposure, contamination, and reduction.

pharmacology and toxicology↗

bioRxiv2021

Cyclocreatine suppresses prostate tumorigenesis through dual effects on SAM and creatine metabolism

Prostate cancer is highly prevalent, being the second most common cause of cancer mortality in men worldwide. Applying a novel genetically engineered mouse model (GEMM) of aggressive prostate cancer driven by deficiency of PTEN and SPRY2 (Sprouty 2) tumour suppressors, we identified enhanced creatine metabolism within the phosphagen system in progressive disease. Altered creatine metabolism was validated in in vitro and in vivo prostate cancer models and in clinical cases. Upregulated creatine levels were due to increased uptake through the SLC6A8 creatine transporter and de novo synthesis, resulting in enhanced cellular basal respiration. Treatment with cyclocreatine (a creatine analogue that potently and specifically blocks the phosphagen system) dramatically reduces creatine and phosphocreatine levels. Blockade of creatine biosynthesis by cyclocreatine leads to cellular accumulation of S-adenosyl methionine (SAM), an intermediary of creatine biosynthesis, and suppresses prostate cancer growth in vitro. Furthermore, cyclocreatine treatment impairs cancer progression in our GEMM and in a xenograft liver metastasis model. Hence, by targeting the phosphagen system, cyclocreatine results in anti-tumourigenic effects from both SAM accumulation and suppressed phosphagen system.

cancer biology↗

bioRxiv2021

Targeting glucose metabolism sensitizes pancreatic cancer to MEK inhibition

Pancreatic ductal adenocarcinoma (PDAC) is almost universally lethal. A critical unmet need exists to explore essential susceptibilities in PDAC and identify druggable targets for tumor maintenance. This is especially challenging in the context of PDAC, in which activating mutations of KRAS oncogene (KRAS*) dominate the genetic landscape. By using an inducible KrasG12D-driven p53 deficient PDAC mouse model (iKras model), we demonstrate that RAF-MEK-MAPK signaling is the major effector for oncogenic Kras-mediated tumor maintenance. However, MEK inhibition has minimal therapeutic effect as single agent for PDAC both in vitro and in vivo. Although MEK inhibition partially downregulates the transcription of glycolysis genes, it surprisingly fails to suppress the glycolysis flux in PDAC cell, which is a major metabolism effector of oncogenic KRAS. Accordingly, In vivo genetic screen identified multiple glycolysis genes as potential targets that may sensitize tumor cells to MAPK inhibition. Furthermore, inhibition of glucose metabolism with low dose 2-deoxyglucose (2DG) in combination with MEK inhibitor dramatically induces apoptosis in KrasG12D-driven PDAC cell in vitro, inhibits xenograft tumor growth and prolongs the overall survival of genetically engineered mouse PDAC model. Molecular and metabolism analyses indicate that co-targeting glycolysis and MAPK signaling results in apoptosis via induction of lethal ER stress. Together, our work suggests that combinatory inhibition of glycolysis and MAPK pathway may serve as an alternative approach to target KRAS-driven PDAC.

cancer biology↗

bioRxiv2018

Mitochondrial Fusion Suppresses Pancreatic Cancer Growth via Reduced Oxidative Metabolism

Pancreatic cancer is a highly lethal disease whose aggressive biology that is driven by mitochondrial oxidative metabolism. Mitochondria normally form a network of fused organelles, but we find that patient-derived and genetically engineered murine pancreatic cancer cells exhibit highly fragmented mitochondria with robust oxygen consumption rates (OCR). When mitochondrial fusion was activated by the genetic or pharmacological inhibition Drp1, the morphology and metabolism of human and murine pancreatic cancer cells more closely resembled that of normal pancreatic epithelial cells. This reduced metabolism was correlated with slower tumor growth, fewer metastases, and enhanced survival in a syngeneic orthotopic model. Similarly, directly activating mitochondrial fusion by overexpression of Mfn2 also reduced tumor growth and metastases. Mitochondrial fusion in pancreatic cancer cells was associated with reduced mitochondrial mass and Complex I expression and function. Thus, these data suggest that enhancing mitochondrial fusion through Drp1 inhibition or enhanced Mfn2 expression or function has strong tumor suppressive activity against pancreatic cancer and may thus represent a highly novel and efficacious therapeutic target.

cancer biology↗