Biology subjects

Franz, S.

Publications and source records attributed to Franz, S..

10 recordsLinked to original sources

bioRxiv2026

DeePEn - A Depth sensitive benchmark for Protein Engineering

Recent progress in modeling techniques and high-throughput screening has significantly enhanced the accessibility of protein engineering. Nevertheless, further progress gets hindered by the lack of robust benchmarks that capture the practical challenges for real-world protein engineering. Here, we introduced DeePEn, a Depth-sensitive benchmark for Protein Engineering that quantifies a models generalization capabilities when predicting protein fitness at increasing mutational distance from the wildtype or training data. We defined distance as the number of simultaneous point mutations, i.e., single amino acid variants (SAVs), moving from wild-type to mutant (edit distance in computer science jargon). Specifically selecting four deep mutational scanning (DMS) datasets with sufficient multi-mutation data points from ProteinGym, we assessed recent predictive models, including general and biophysics-informed protein Language Models (pLMs), and a non-transformer neural network. Our results highlight how the performance of all models deteriorates with increasing mutational distance and that no single metric sufficiently captures the diverse requirements of protein engineering. To overcome these shortcomings, DeePEn provides a readily available resource for multi-metric benchmarking that focuses on the prediction of distant variants.

bioinformatics↗

bioRxiv2026

Pseudoperplexity Probes Memorization in Protein Language Models

Protein Language Models (pLMs) have significantly advanced computational biology. Yet their scale and reliance on redundant training data raise a fundamental question: do pLMs generalize the statistical grammar of proteins, or do they simply memorize their training data? To investigate this, we used pseudoperplexity as a probe for sequence-level memorization, comparing ProtT5's pseudoperplexity on a pre-training proxy dataset against a post-training holdout of genuinely novel sequences. To ensure a valid comparison, we matched the datasets by sequence length, cluster size, and taxonomic family. As a statistical baseline, we trained n-gram language models; analysis of higher-order n-gram composition and a statistically significant divergence in perplexity confirmed that the post-training sequences were genuinely novel at the local sequence level. ProtT5 showed a statistically significant difference in pseudoperplexity between seen and unseen sequences, though further analysis revealed this memorization signal to be modest. These findings suggest that ProtT5 exhibits detectable but limited memorization of its training data as measured by a pseudoperplexity-based probe.

bioinformatics↗

bioRxiv2025

biocentral: embedding-based protein predictions

AO_SCPLOWBSTRACTC_SCPLOWThe rise of protein Language Models (pLMs) is reshaping the landscape of protein prediction. Embeddings are powerful protein representations provided by pLMs, but they come at a cost: their generation requires expensive hardware, and leveraging models often requires expert knowledge. To some extent, these hurdles limit the ease of use and benefits of those methods both for experimental and computational biologists. Biocentral aims at providing a free and open embedding-based service which addresses these challenges. We support standardized access to most pLMs currently in use, enabling researchers to generate embeddings, get embedding-based protein feature predictions, and train embedding-based models. Here, we showcase biocentral in a large-scale analysis of the BFVD virus database through biocentrals predict module. We also show how readily biocentrals training module reproduces an existing embedding-based prediction method. The server is accessible through a graphical user interface and a programmatic Application Programming Interface (API) at: https://biocentral.rostlab.org

bioinformatics↗

bioRxiv2025

A STAT5B-driven mouse model of hepatosplenic γδ T-cell lymphoma reveals therapeutic efficacy of JAK inhibition

Hepatosplenic T-cell lymphoma (HSTCL) is a rare and aggressive neoplasm associated with poor responses to standard chemotherapy regimens and low survival rates. No targeted therapies are available for HSTCL, and preclinical models to test new treatment options have not been established. The JAK-STAT signaling cascade is a key dysregulated pathway in HSTCL, and STAT5BN642His the most frequent somatic mutation in the disease. Here, we report on newly established clonal, murine {gamma}{delta} T-cell lymphoma cell lines initiated and driven by oncogenic STAT5BN642H, which recapitulate key immunophenotypic features, gene expression profiles and typically low cytolytic activity of patient-derived human HSTCL cells. CRISPR-Cas9 mediated knockout demonstrated growth dependence on STAT5BN642H. Murine C15 cells were allo-engrafted intravenously into both immunodeficient and immunocompetent mice to model an aggressive HSTCL-like disease at high penetrance, with recipient mice displaying hepatosplenomegaly and destructive {gamma}{delta} T cell organ infiltration, including bone marrow and blood involvement. We identified the potential of JAK inhibition as a targeted treatment strategy for HSTCL, and found the clinically approved JAK inhibitor upadacitinib to display selective anti-tumor efficacy against STAT5B-mutated HSTCL cell lines in vitro, in vivo, and in primary HSTCL patient samples. Overall, we describe the first robust STAT5B-driven preclinical model resembling features of HSTCL in an immune competent setting. This tool is expected to accelerate the study of HSTCL disease mechanisms and the testing of novel therapies. Our data further present the JAK inhibitor upadacitinib as a promising targeted treatment option for STAT5B-mutated HSTCL.

cancer biology↗

bioRxiv2025

Genetic admixture between East and West European Gravettian-associated populations in Western Europe before the Last Glacial Maximum

Modern humans first settled in Europe at least 45,000 years ago. However, limited genomic data from individuals dating between 45,000 and 20,000 years ago still restricts our understanding of population dynamics and admixture during the Upper Palaeolithic. Before the Last Glacial Maximum (LGM, 26.5-19 cal kya), Gravettian culture-associated populations were widespread and genetically diverse, comprising at least two distinct genetic groups, referred to as the Fournol and V[e]stonice clusters. We present genome-wide data from three Gravettian-associated individuals: two from cave sites in the Franco-Cantabrian region (Chufin and Isturitz) and one from Italy (Ostuni1b). These data reveal previously undetected gene flow linking the ancestry of 34,000-year-old individuals from Sungir (Russia) to Gravettian individuals from Western Europe, challenging the prevailing model of population continuity from the Aurignacian to the Solutrean. As osseous remains are scarce for this time period, DNA from sediments deposited by ancient humans opens a new possibility to obtain genomic data. We thus examine sedimentary DNA from Solutrean Layer 122 at El Miron Cave (Cantabria, [~]22,000 cal BP), recovering approximately 16,000 human SNPs, among the highest yields reported from a Palaeolithic context. Generating these data required over 1.15 billion sequencing reads, illustrating both the potential of sediment DNA for autosomal analysis and the technical challenges of the approach.

genomics↗

bioRxiv2025

Disc-Toroid Hybrid Lipid Nanoparticles for Efficient Drug Encapsulation and Subcutaneous Delivery

The development of an effective for subcutaneous or intradermal injection drug delivery requires systems with improved bioavailability and biocompatibility. Systematic physicochemical and biological interrogation of carnauba-wax/red-palm-oil lipid nanoparticles (LNPs) stabilised with d--tocopheryl-PEG-1000-succinate and polysorbate-40 shows that purposeful matrix engineering yields a robust sub-50 nm carrier for under-skin delivery. Cryo-TEM and SAXS uncover a disc-toroid hybrid morphology dominated by 30-40 nm particles with toroidal/disc shape - an advantageous biconcave geometry to enhance surface-to-volume ratio and is expected to accelerate enzymatic erosion after injection. Orthogonal analytics (AF4-MD, DLS, MALS, WAXS) confirmed that loading with quinine or dihydroartemisinin leaves size and crystallinity unchanged while delivering encapsulation efficiencies of approximately 90 % and long-term particle stability up to 18 months at 4 {degrees}C. Red-palm oil and the dual-surfactant corona act synergistically to suppress bimodality and narrow size distribution compared with single-component controls. Short-term viability assays in keratinocytes, fibroblasts and macrophages showed no cytotoxicity even at [≥]1 % (w/v) lipid, underscoring excellent biocompatibility. Fluorescein-labelled LNPs injected into ex vivo human skin traversed the dermis and hypodermis, while only nanomolar lipid concentrations appeared in the receiver medium, indicating a sustained local depot. Collectively, these insights link composition, structure and performance, positioning wax-based disc-toroid LNPs as a flexible platform for high-load delivery of small-molecule or biopharmaceutical therapeutics via minimally invasive under-skin administration.

biochemistry↗

bioRxiv2023

Maximizing Efficiency in SedaDNA Analysis: A Novel Extract Pooling Approach

In recent years, the field of ancient DNA (aDNA) has taken a new direction toward studying human population dynamics through sedimentary DNA (sedaDNA), enabling the study of past ecosystems. However, the screening of numerous sediment samples from archaeological sites remains a time-consuming and costly endeavor, particularly when targeting hominin DNA. Here, we present a novel high-throughput method that facilitates the fast and efficient analysis of sediment samples by applying a pooled testing method. Our approach involves combining multiple extracts, allowing users to parallelize laboratory procedures early in the sample preparation pipeline while effectively screening for the presence of aDNA. Pooled samples that exhibit aDNA signals can then undergo detailed analysis, while empty pools are discarded. We have successfully applied our extract pooling method to various sediment samples from Middle and Upper Paleolithic sites in Europe, Asia, and Africa. Notably, our results reveal that an aDNA signal remains discernible even when pooled with four negative samples. We also demonstrate that the DNA yield of double-stranded libraries increases significantly when reducing the extract input, potentially mitigating the effects of inhibition. By embracing this innovative approach, researchers can analyze large numbers of sediment samples for aDNA preservation, achieving significant cost reductions of up to 70% and reducing hands-on laboratory time to one-fifth.

genomics↗

bioRxiv2022

PRE-INJURY MECHANORECEPTOR ABLATION REDUCES NOCICEPTOR-DRIVEN SPINAL CORD INJURY-INDUCED NEUROPATHIC PAIN

Evidence from previous studies supports the concept that spinal cord injury (SCI) induced neuropathic pain (NP) has its neural roots in the peripheral nervous system. There is uncertainty about how and to which degree nociceptors and mechanoreceptors contribute. Sensorimotor activation-based interventions (e.g. treadmill training) have been shown to reduce NP following experimental SCI, suggesting transmission of pain-alleviating signals through mechanoreceptors. At the same time, nociceptors have been shown to become hyperexcitable early after SCI and peptidergic axons sprout into deeper laminae of the below injury level dorsal horn. The aim of the present study is to comprehensively understand the relative contribution of each pathway in respect to NP presentation in a moderate mouse contusion SCI model. After genetic ablation of tropomyosin receptor kinase B (TrkB) expressing mechanoreceptors before SCI mechanical allodynia was reduced. The identical genetic ablation after SCI did not yield any change in pain behavior. CGRP sprouting into lamina III/IV below injury level as a consequence of SCI was not altered by either mechanoreceptor ablation. Moreover, detection of hyperexcitability in nociceptors, not in mechanoreceptors, in skin-nerve preparations of contusion SCI mice 7 days after injury makes a substantial direct contribution of mechanoreceptors to NP maintenance unlikely. SNS reporter mice allowing specific visualization of the entire nociceptor population confirmed significant sprouting of respective neurons into laminae III/IV as early as 5 days post-injury. Genetic ablation of SNS-Cre mice severely affected their overall health condition, which precluded them to undergo experimental SCI and subsequent further analysis. Complementing animal data, quantitative sensory testing in human SCI subjects indicated reduced mechanical pain thresholds, whereas the mechanical detection threshold was not altered. Taken together, early mechanoreceptor ablation modulates pain behavior, most likely through indirect mechanisms. Hyperexcitable nociceptors with consecutive peptidergic fiber sprouting in the dorsal horn are confirmed as the likely main driver of SCI-induced NP. Future studies need to focus on injury-derived factors triggering early onset nociceptor hyperexcitability, which could serve as targets for more effective therapeutic interventions.

neuroscience↗

bioRxiv2020

Binding of SARS-CoV-2 spike protein to ACE2 is disabled by thiol-based drugs; evidence from in vitro SARS-CoV-2 infection studies.

Neutrophil-induced oxidative stress is a mechanism of lung injury in COVID-19, and drugs with a functional thiol group ("thiol drugs"), especially cysteamine, have anti-oxidant and anti-inflammatory properties that could limit this injury. Thiol drugs may also alter the redox status of the cysteine-rich SARS-CoV-2 spike glycoprotein (SARS-2-S) and thereby disrupt ACE2 binding. Using ACE2 binding assay, reporter virus pseudotyped with SARS-CoV-2 spikes (ancestral and variants) and authentic SARS-CoV-2 (Wuhan-1), we find that multiple thiol drugs inhibit SARS-2-S binding to ACE2 and virus entry into cells. Pseudoviruses carrying variant spikes were less efficiently inhibited as compared to pseudotypes bearing an ancestral spike, but the most potent drugs still inhibited the Delta variant in the low millimolar range. IC50 values followed the order of their cystine cleavage rates and lower thiol pKa values. In hamsters infected with SARS-CoV-2, intraperitoneal (IP) cysteamine decreased neutrophilic inflammation and alveolar hemorrhage in the lungs but did not decrease viral infection, most likely because IP delivery could not achieve millimolar concentrations in the airways. These data show that thiol drugs inhibit SARS-CoV-2 infection in vitro and reduce SARS-CoV-2-related lung injury in vivo and provide strong rationale for trials of systemically delivered thiol drugs as COVID-19 treatments. We propose that antiviral effects of thiol drugs in vivo will require delivery directly to the airways to ensure millimolar drug concentrations and that thiol drugs with lower thiol pKa values are most likely to be effective. One Sentence SummaryThe effect of cysteamine to decrease SARS-CoV-2 pneumonia in vivo and of multiple thiol drugs to inhibit SARS-CoV-2 infection in vitro provides rationale for clinical trials of thiol drugs in COVID-19.

microbiology↗

bioRxiv2020

Human IFITM3 restricts Chikungunya virus and Mayaro virus infection and is susceptible to virus-mediated counteraction

Interferon-induced transmembrane (IFITM) proteins restrict infection by enveloped viruses through interfering with membrane fusion and virion internalisation. The role of IFITM proteins during alphaviral infection of human cells and viral counteraction strategies remain largely unexplored. Here, we characterized the impact of IFITM proteins and variants on entry and spread of Chikungunya virus (CHIKV) and Mayaro virus (MAYV) in human cells, and provide first evidence for a CHIKV-mediated antagonism of IFITM proteins. IFITM1, 2 and 3 restricted infection at the level of alphavirus glycoprotein-mediated entry, both in the context of direct infection and during cell-to-cell transmission. Relocalization of normally endosomal IFITM3 to the plasma membrane resulted in the loss of its antiviral activity. rs12252-C, a naturally occurring variant of IFITM3 that has been proposed to associate with severe influenza in humans, restricted CHIKV, MAYV and influenza A virus infection as efficiently as wild-type IFITM3. Finally, all antivirally active IFITM variants displayed reduced cell surface levels in CHIKV-infected cells involving a posttranscriptional process mediated by one or several non-structural protein(s) of CHIKV.

microbiology↗