Carl Herrmann

Neuroendocrine differentiation (NED) promotes lineage plasticity and therapy resistance in prostate and lung cancers. Here we identify NED as a route of treatment escape in breast cancer: neuroendocrine features emerged or increased in 22% of progressive cases, predominantly, but not exclusively, in HER2-positive subtype. Using patient-derived, epithelial/neuroendocrine organoid and xenograft models, we map the epigenetic and transcriptional divergence between epithelial and neuroendocrine states arising from a shared mutational background and define a breast cancer neuroendocrine signature. NED involves epigenetic silencing of EpCAM and amplified HER2, while NEUROD1 functions as a master transcriptional regulator of the neuroendocrine fate. NCAM-positive putative circulating tumor cells were detected in patients in whom EpCAM-based detection had failed. Neuroendocrine tumor cells exhibit an FGFR1 dependency associated with NCAM–FGFR1 interaction and mesenchymal FGFR1 splicing. Together, these findings establish NED as a mechanism of escape from HER2-targeted therapy in breast cancer, with implications for disease monitoring and intervention.

Understanding how transcriptional heterogeneity is organized across tumors, patients, and treatment conditions remains a central challenge in cancer biology. Here, we present kaleidoCell, a GPU-accelerated Python framework for consensus non-negative matrix factorization that identifies reproducible meta-programs across independent samples. When benchmarked against its principal counterpart, the geneNMF R package, kaleidoCell achieves a twofold speed improvement on large datasets. In addition, it includes an integrated analysis module that generates a comprehensive HTML report containing key results and visualizations—including marker genes corresponding to the meta-programs, gene set enrichment analysis, UMAP projections and violin plots—without requiring additional user code. Using glioblastoma as a case study, we applied kaleidoCell to two published datasets. In a panobinostat-treated cohort, kaleidoCell resolves the cellular landscape of the tumor microenvironment and delineates how HDAC inhibition reshapes malignant cell states at single-cell resolution. We extend prior descriptions of the metallothionein-associated stress program in treatment response and identify co induction of IER3 as a candidate component of the associated survival signalling. In addition, we uncover novel transcriptional signatures associated with HDAC inhibition. Beyond confirming suppression of a neural progenitor cell-/oligodendrocyte progenitor cell-like program which is consistent with prior reports, kaleidoCell identifies loss of an astrocyte-like identity program as a previously unrecognized candidate mechanism of panobinostat action in glioblastoma. Together, these results establish kaleidoCell as a fast, user-friendly framework that enables robust discovery of biologically meaningful transcriptional programs in large, heterogeneous single-cell datasets.

Background: The accurate prediction of cellular responses to perturbations, such as drug treatments, remains a pivotal challenge in single-cell transcriptomics. While numerous deep learning tools have been developed for this task, their systematic benchmarking across diverse datasets and performance metrics has been limited. Results: Here, we present scArchon, a reproducible, modular benchmarking platform built on Snakemake. It is designed to evaluate perturbation response prediction tools in an unbiased and extensible manner. Employing six representative single-cell RNA-seq datasets, we compare leading methods such as scGen, CPA, trVAE, scPRAM, scVIDR, scDisInFact, SCREEN, scPreGAN, and CellOT against baselines. We assess model performance using a composite of statistical and biological metrics. Our analysis reveals heterogeneous performance. While methods like trVAE, scGen, scPRAM, and scVIDR achieve robust results across multiple datasets, other tools occasionally underperform even compared to linear or control baselines. Notably, models with favorable quantitative scores may fail to retain key biological perturbation signatures, underscoring the need for gene-level evaluation. Conclusions: scArchon provides a unified, extensible foundation for large-scale, standardized benchmarking of perturbation prediction tools, facilitating methodological transparency and accelerating development in this rapidly evolving field. We encourage adoption of scArchon and sharing of containerized tools to drive progress in single-cell perturbation modeling.

The intestinal epithelium maintains tissue homeostasis through a dynamic balance of stem cell proliferation and differentiation. This process is spatially regulated along the crypt–villus axis, with intestinal stem cells in the crypt regions proliferating and progenitor cells differentiating as they migrate toward the villus tips. Because the lumen of the gut contains very low levels of oxygen (i.e., hypoxia), an oxygen gradient is established within the crypt–villus axis, placing the crypt regions under normoxic conditions while the villus tips reside under hypoxic conditions. Hence, intestinal epithelial cells encounter distinct oxygen microenvironments throughout their life span as they migrate along the crypt–villus structures during their proliferation and differentiation process. To investigate how oxygen availability influences intestinal stem cell proliferation and differentiation, we cultured patient-derived human ileum organoids (i.e., enteroids) under normoxic (20% oxygen) or hypoxic (1% oxygen) conditions. Under hypoxia, enteroid growth was reduced, and expression of several stem cell markers, such as OLFM4 and LGR5, was decreased. Bulk and single-cell RNA sequencing revealed that hypoxia suppressed Wnt signaling pathways and reduced stem cell activity. Importantly, pharmacological stabilization of HIF-1α under normoxic conditions recapitulated the hypoxia-induced loss of stemness, demonstrating that HIF-1α is a key mediator of oxygen-dependent stem cell regulation in enteroids. These findings establish that physiological hypoxia in the intestinal epithelium directly regulates stem cell fate through HIF-1α stabilization, providing mechanistic insight into how oxygen availability along the crypt–villus structures controls intestinal homeostasis.

Hepatitis E virus (HEV) is a significant human pathogen causing both acute and chronic infections worldwide. The cell-intrinsic antiviral response serves as the initial defense against viruses and has been shown to be activated upon HEV infection. HEV can replicate in the presence of this response, but the underlying mechanisms remain poorly understood. Here, we investigated the roles of the structural proteins ORF2 and ORF3 in the cell-intrinsic antiviral response to HEV infection. Mechanistically, we validated that ectopic ORF2, but not ORF3, interfered with antiviral and inflammatory signaling downstream of pattern recognition receptors, in part through interaction with the central adaptor protein TANK binding kinase 1. In the full-length viral context, ORF2 contributed to a reduced antiviral response and consequently, more efficient viral replication. In addition, we discovered a protective mechanism mediated by ORF2 that shielded viral replication from antiviral effectors. Using single-cell RNA-sequencing, we confirmed that the presence of ORF2 in infected cells dampened antiviral responses in both actively infected cells and bystanders. As a consequence, we found that early in the infection process, the progression of authentic HEV infection relied on the presence of ORF2, facilitating a balance between viral replication and the antiviral response. Altogether, our findings shed new light on the multifaceted role of ORF2 in the HEV life cycle and improve our understanding of the determinants that contribute to persistent HEV replication in cell culture.

Background: DNA methylation plays a crucial role in cancer development and progression and has been linked to genetically and clinically distinct tumor classes, including IDH-mutated and IDH-wildtype adult-type diffuse gliomas. Here, we identify a CpG-island methylator phenotype (CIMP) that characterizes the receptor tyrosine kinase 2 (RTK2) subtype of IDH-wildtype glioblastoma. Results: This RTK2-CIMP affects genomic locations and cell functions distinct from those of IDH mutation-associated IDH-CIMP and suppresses the expression of its target genes. The RTK2-CIMP-region chromatin is characterized by a combination of repressive and activating marks, including polycomb-associated H3K27me3 and enhancer-associated H3K4me1, consistent with DNA methylation-mediated silencing of genes with bivalent-state promoters in neural progenitor cells. Functionally, RTK2-CIMP affects neuronal lineage genes and is significantly associated with astrocyte-like glioblastoma, suggesting that RTK2-CIMP is an epigenetic signature of the astrocyte-like cell state. Furthermore, we demonstrate that RTK2-CIMP can be induced by genetic manipulation in glioblastoma cells. Conclusions: Our results suggest that RTK2-CIMP is a key contributor to cell-state plasticity in glioblastoma.

The accurate prediction of cellular responses to perturbations, such as drug treatments, remains a pivotal challenge in single-cell transcriptomics. While numerous deep learning tools have been developed for this task, their systematic benchmarking across diverse datasets and performance metrics has been limited. Here, we present scArchon, a reproducible, modular benchmarking platform built on Snakemake, designed to evaluate perturbation response prediction tools in an unbiased and extensible manner. Employing four representative single-cell RNA-seq datasets, we compare leading methods such as scGen, CPA, trVAE, scPRAM, scVIDR, scDisInFact, SCREEN, scPreGAN, and CellOT against baselines. We assess model performance using a composite of statistical and biological metrics. Our analysis reveals heterogeneous performance: while methods like trVAE, scGen, scPRAM, and scVIDR achieve robust results across multiple datasets, other tools occasionally underperform even compared to linear or control baselines. Notably, models with favorable quantitative scores may fail to retain key biological perturbation signatures, underscoring the need for gene-level evaluation. scArchon provides a unified, extensible foundation for large-scale, standardized benchmarking of perturbation prediction tools, facilitating methodological transparency and accelerating development in this rapidly evolving field. We encourage adoption of scArchon and sharing of containerized tools to drive progress in single-cell perturbation modeling.

Molecular changes underlying the persistent health effects after SARS-CoV-2 infection remain poorly understood. To discern the gene regulatory landscape in the upper respiratory tract of COVID-19 patients, we performed enzymatic DNA methylome and single-cell RNA sequencing in nasal cells of COVID-19 patients ( n = 19, scRNA-seq n = 14) and controls ( n = 14, scRNA-seq n = 10). In addition, we resampled a subset of these patients for transcriptome analyses at 3 ( n = 7) and 12 months ( n = 5) post infection and followed the expression of differentially regulated genes over time. Genome-wide DNA methylation analysis revealed 3112 differentially methylated regions between COVID-19 patients and controls. Hypomethylated regions affected immune regulatory genes, while hypermethylated regions were associated with genes governing ciliary function. These genes were not only downregulated in the acute phase of the disease but sustained repressed up to 12 months post infection in ciliated cells. Validation in an independent cohort collected 6 months post infection ( n = 15) indicated symptom-dependent transcriptional repression of ciliary genes. We therefore propose that hypermethylation observed in the acute phase may exert a long-term effect on gene expression, possibly contributing to post-acute COVID-19 sequelae.

Circulating tumor cells (CTCs) drive metastasis, the leading cause of death in individuals with breast cancer. Due to their low abundance in the circulation, robust CTC expansion protocols are urgently needed to effectively study disease progression and therapy responses. Here we present the establishment of long-term CTC-derived organoids from female individuals with metastatic breast cancer. Multiomics analysis of CTC-derived organoids along with preclinical modeling with xenografts identified neuregulin 1 (NRG1)–ERBB2 receptor tyrosine kinase 3 (ERBB3/HER3) signaling as a key pathway required for CTC survival, growth and dissemination. Genome-wide CRISPR activation screens revealed that fibroblast growth factor receptor 1 (FGFR1) signaling serves a compensatory function to the NRG1–HER3 axis and rescues NRG1 deficiency in CTCs. Conversely, NRG1–HER3 activation induced resistance to FGFR1 inhibition, whereas combinatorial blockade impaired CTC growth. The dynamic interplay between NRG1–HER3 and FGFR1 signaling reveals the molecular basis of cancer cell plasticity and clinically relevant strategies to target it. Our CTC organoid platform enables the identification and validation of patient-specific vulnerabilities and represents an innovative tool for precision medicine.