Simon Flyvbjerg Nørrelykke

BACKGROUND: Reverse cholesterol transport by HDLs (high-density lipoproteins) is considered an antiatherogenic metabolic pathway. Hepatocytes are the main contributors to the efficacy of this pathway by the production of apoA-I (apolipoprotein A1) and its lipidation by ABCA1 (ATP-binding cassette transporter A1), selective uptake of cholesterol via SR-BI (scavenger receptor class B type 1), and uptake of entire HDL particles. The molecular determinants of the latter step are not well understood. METHODS: We performed a genome-wide RNA interference screen for genes limiting the uptake of HDL fluorescently labeled at its protein moiety into Huh-7 hepatocarcinoma cells. Top hit genes were validated by targeted in vitro experiments and the analysis of associations between their variants and HDL-C (HDL-cholesterol) levels in the databases of the Global Lipids Genetics Consortium and the UK Biobank, as well as inborn errors of metabolism and their respective mouse models. RESULTS: The knockdown of 128 genes significantly inhibited HDL uptake. Six of them encode components of the COPI (coat protein I) coatomer, namely, COPA , COPB1 , COPB2 , COPG1 , ARCN1 , and COPZ1 . Knocking down any of them decreased the uptake of both fluorescently labeled proteins and lipids of HDL, the cell surface abundance of SR-BI, and APOA1 expression and apoA-I secretion but increased the cell surface abundance of ABCA1. Common single-nucleotide polymorphisms of ARCN1 and COPB1 were associated with significantly higher HDL-C levels in the population, while rare COPA and COPG1 variants causing immunopathies were associated with rather lower levels of HDL-C in both affected patients and the corresponding genetically modified mice. CONCLUSIONS: In hepatocytes, the COPI coatomer regulates HDL holoparticle uptake, selective lipid uptake, apoA-I secretion, and ABCA1 expression, and thereby, it influences plasma levels of HDL-C.

Pancreatic ductal adenocarcinomas (PDACs) are aggressive, stroma-rich tumours. They are unresponsive to treatments, and patients relapse quickly on DNA-damaging chemotherapies. PDAC stroma consists of extracellular matrix proteins (ECM), secreted by cancer-associated fibroblasts (CAFs). Here we show an unexpected link between CAF-secreted ECM proteins and enhanced DNA repair. We identify NDRG1 (N-myc downstream-regulated gene 1) as a key mediator that senses signals from the ECM via adhesion receptors and serum and glucocorticoid-activated kinase. We establish NDRG1 as a DNA repair factor that physically associates with replication forks, maintains DNA replication, resolves stalled forks caused by chemotherapies and is involved in reducing R-loops, RNA-DNA hybrids known to cause genomic instability. NDRG1 is highly expressed in PDAC tumours and its high expression correlates with poor disease-specific survival and poor response to chemotherapy. In conclusion, our data reveal an unexpected role for CAF-secreted ECM proteins in promoting DNA repair via NDRG1, mechanistically linking tumour stroma to replication fork homeostasis and R-loop regulation.

Normalized Raman Imaging (NoRI) enables high-resolution, label-free quantification of protein and lipid concentrations in biological tissues. Because NoRI provides rich molecular information, the analysis of its large, multi-channel datasets becomes a significant computational bottleneck. In this work, we introduce a novel, modular computational pipeline for the automated segmentation and quantification of kidney tissue structures imaged with NoRI. The pipeline integrates classical image processing with state-of-the-art machine learning tools, including the Segment Anything Model (SAM) and ilastik, to segment key anatomical and biochemical features–such as tubules, nuclei, brush borders, and lumens. A custom contrast-enhancement strategy was developed to create a third SAM input channel from NoRI data, leading to a substantial improvement in segmentation performance (F1 score: 0.9226). Our framework enables accurate cytoplasm-resolved quantification of protein and lipid concentrations and reveals distinct biochemical signatures across renal tubule subtypes and experimental conditions. Core segmentation relies solely on the Raman channels; immunofluorescence markers are used optionally for tubule classification and brush border detection. This method offers a robust, scalable foundation for quantitative tissue analysis and enhances the utility of NoRI imaging for biomedical research.

Conventional histological relies on fixation, embedding, sectioning, and staining methods that distort cellular architecture, extract lipids, and introduces variability, limiting reproducibility. We present Normalized Stimulated Raman Imaging (NoRI) that enables quantitative, label-free protein and lipid measurements at high-spatial resolution. NoRI computationally corrects protein, lipid, and water Raman signals, allowing quantitative biomass measurements while preserving tissue architecture, facilitating analysis by convolutional neural networks. Applied to mouse kidney, NoRI accurately classified tubule types (F1-[harmonic mean of precision and recall against manual annotation]=0.93), anatomical regions (F1=0.91), and biological sex (F1=0.97), revealing greater cytoplasmic lipid (+6.9mg/mL; p=0.028) and nuclear protein (+26.3mg/mL; p<0.001) in female tubules. In acute kidney injury, NoRI captured dynamic lipid and protein organization and quantified brush border remodelling and lipid droplet changes over 25 days. Spatial lipid quantification was central for feature classification in AKI (F1=1.0). These results establish NoRI as a reproducible, quantitative method for diagnostics, histopathology, and feature discovery.

Normalized Raman Imaging (NoRI) enables high-resolution, label-free quantification of protein and lipid concentrations in biological tissues. Because NoRI provides rich molecular information, the analysis of its large, multi-channel datasets turns into a significant computational bottleneck. In this work, we introduce a novel, modular computational pipeline for automated segmentation and quantification of kidney tissue structures imaged with NoRI. The pipeline integrates classical image processing with state-of-the-art machine learning tools, including the Segment Anything Model (SAM) and ilastik, to segment key anatomical and biochemical features—such as tubules, nuclei, brush borders, and lumens. A custom contrast-enhancement strategy was developed to create a third SAM input channel from NoRI data, leading to a substantial improvement in segmentation performance (F1 score: 0.9226). Our framework enables accurate cytoplasm, resolved quantification of protein and lipid concentrations and reveals distinct biochemical signatures across renal tubule subtypes and experimental conditions. This method offers a robust, scalable foundation for quantitative tissue analysis and enhances the utility of NoRI imaging for biomedical research.

Aggressive cancers, such as pancreatic ductal adenocarcinoma (PDAC), are often characterized by a complex and desmoplastic tumor microenvironment, a stroma rich supportive connective tissue composed primarily of extracellular matrix (ECM) and non-cancerous cells. Desmoplasia, a dense deposition of stroma, is a major reason for therapy resistance, acting both as a physical barrier that interferes with drug penetration and as a supportive niche that protects cancer cells through diverse mechanisms. Precise understanding of spatial cell interactions in stroma-rich tumors is essential for optimizing therapeutic responses. It enables detailed mapping of stromal-tumor interfaces, comprehensive cell phenotyping, and insights into changes in tissue architecture, improving assessment of drug responses. Recent advances in multiplexed immunofluorescence imaging have enabled the acquisition of large batches of whole-slide tumor images, but scalable and reproducible methods to analyze the spatial distribution of cell states relative to stromal regions remain limited. To address this gap, we developed an open-source computational pipeline that integrates QuPath, StarDist, and custom Python scripts to quantify biomarker expression at a single- and sub-cellular resolution across entire tumor sections. Our workflow includes: (i) automated nuclei segmentation using StarDist, (ii) machine learning-based cell classification using multiplexed marker expression, (iii) modeling of stromal regions based on fibronectin staining, (iv) sensitivity analyses on classification thresholds to ensure robustness across heterogeneous datasets, and (v) distance-based quantification of the proximity of each cell to the stromal border. To improve consistency across slides with variable staining intensities, we introduce a statistical strategy that translates classification thresholds by propagating a chosen reference percentile across the distribution of marker-related cell measurement in each image. We apply this approach to quantify spatial patterns of distribution of the phosphorylated form of the N-Myc downregulated gene 1 (NDRG1), a novel DNA repair protein that conveys signals from the ECM to the nucleus to maintain replication fork homeostasis, and a known cell proliferation marker Ki67 in fibronectin-defined stromal regions in PDAC xenografts. The pipeline is applicable for the analysis of markers of interest in stroma-rich tissues and is publicly available.

1. Aggressive cancers, such as pancreatic ductal adenocarcinoma (PDAC), are often characterized by a complex and desmoplastic tumor microenvironment rich in stroma, a supportive connective tissue composed primarily of extracellular matrix (ECM) and non-cancerous cells. Desmoplasia, which is a dense deposition of stroma, is a major reason for therapy resistance, acting both as a physical barrier that interferes with drug penetration and as a supportive niche that protects cancer cells through diverse mechanisms. A precise understanding of spatial cell interactions within the tumor microenvironment in stroma-rich cancers is essential for optimizing therapeutic responses. It allows detailed mapping of stromal-tumor interfaces, comprehensive phenotyping of diverse cell types and their functional states, and insights into changes in cellular distribution and tissue architecture, thus leading to an improved assessment of drug responses. Recent advances in multiplexed immunofluorescence imaging have enabled the acquisition of large batches of whole-slide tumor images, but scalable and reproducible methods to analyze the spatial distribution of cell states relative to stromal regions remain limited. To address this gap, we developed an open-source computational pipeline that integrates QuPath (Bankhead et al. 2017), StarDist (Schmidt et al. 2018), and custom Python scripts to quantify biomarker expression at a single- and sub-cellular resolution across entire tumor sections. Our workflow includes: (i) automated nuclei segmentation using StarDist, (ii) machine learning-based cell classification using multiplexed marker expression, (iii) modeling of stromal regions based on fibronectin staining, (iv) sensitivity analyses on classification thresholds to ensure robustness across heterogeneous datasets, and (v) distance-based quantification of the proximity of each cell to the stromal border. To improve consistency across slides with variable staining intensities, we introduce a statistical strategy that translates classification thresholds by propagating a chosen reference percentile across the distribution of marker-related cell measurement in each image. We apply this approach to quantify spatial patterns of distribution of the phosphorylated form of the N-Myc downregulated gene 1 (NDRG1), a novel DNA repair protein that conveys signals from the ECM to the nucleus to maintain replication fork homeostasis, and a known cell proliferation marker Ki67 in fibronectin-defined stromal regions in PDAC xenografts. The pipeline is applicable for the analysis of various stroma-rich tissues and is publicly available: https://github.com/HMS-IAC/stroma-spatial-analysis-web . 2. Summary paragraph: Our study introduces a scalable and reproducible image analysis pipeline that quantifies spatial biomarker distributions relative to the stroma in tumor tissues using open-source tools. By modeling cell-level intensity distributions and calibrating classification thresholds across heterogeneous images, we uncover spatially organized patterns of stroma sensing, DNA damage, and proliferative response in pancreatic tumors. This approach enables robust, quantitative analysis of tumor-stroma interactions and is readily adaptable to other tumor types and biomarker panels, providing a valuable resource for spatial pathology and tumor microenvironment research.

Bioimage analysis (BIA), a crucial discipline in biological research, overcomes the limitations of subjective analysis in microscopy through the creation and application of quantitative and reproducible methods. The establishment of dedicated BIA support within academic institutions is vital to improving research quality and efficiency and can significantly advance scientific discovery. However, a lack of training resources, limited career paths and insufficient recognition of the contributions made by bioimage analysts prevent the full realization of this potential. This Perspective – the result of the recent The Company of Biologists Workshop ‘Effectively Communicating Bioimage Analysis’, which aimed to summarize the global BIA landscape, categorize obstacles and offer possible solutions – proposes strategies to bring about a cultural shift towards recognizing the value of BIA by standardizing tools, improving training and encouraging formal credit for contributions. We also advocate for increased funding, standardized practices and enhanced collaboration, and we conclude with a call to action for all stakeholders to join efforts in advancing BIA.

Human CtIP is best known for its role in DNA end resection to initiate DNA double-strand break repair by homologous recombination. Recently, CtIP has also been shown to protect reversed replication forks from nucleolytic degradation upon DNA replication stress. However, still little is known about the DNA damage response (DDR) networks that preserve genome integrity and sustain cell survival in the context of CtIP insufficiency. Here, to reveal such potential buffering relationships, we screened a DDR siRNA library in CtIP-deficient cells to identify candidate genes that induce synthetic sickness/lethality (SSL). Our analyses unveil a negative genetic interaction between CtIP and BARD1, the heterodimeric binding partner of BRCA1. We found that simultaneous disruption of CtIP and BARD1 triggers enhanced apoptosis due to persistent replication stress-induced DNA lesions giving rise to chromosomal abnormalities. Moreover, we observed that the genetic interaction between CtIP and BARD1 occurs independently of the BRCA1-BARD1 complex formation and might be, therefore, therapeutical relevant for the treatment of BRCA-defective tumors.