Biology subjects

Paavolainen, L.

Publications and source records attributed to Paavolainen, L..

11 recordsLinked to original sources

bioRxiv2025

Explainable machine learning-assisted exploration of chromatin dynamics reveals chromosome-specific response to serum starvation

Chromatin is dynamic at all length scales, influencing chromatin-based processes, such as gene expression. Even large-scale reorganization of whole chromosome territories has been reported upon specific signals, but lack of suitable methods has prevented analysis of the underlying dynamic processes. Here we have used CRISPR-Sirius for time-lapse imaging of chromatin loci dynamics during serum starvation. We show that chromosome 1 loci move towards the nuclear envelope during the first hour of serum starvation in a chromosome-specific manner. Machine learning-assisted exploration of acquired multiparametric data combined with the Shapley values-based explanation approach allowed us to uncover the critical features that characterize chromatin dynamics during serum starvation. This analysis reveals that although serum starvation affects overall nuclear morphology and chromatin dynamics, chromosome 1 loci display a specific response that is characterized by maintenance of dynamics in constrained environment, and long "jumps" at the nuclear periphery. Interestingly, the two homologous chromosomes display differential behaviors, with the more peripheral homolog being more responsive to the signal than the internal one. Overall, the presented machine learning-assisted dataset exploration helps us navigate the multidimensional data to understand the underlying dynamic processes and can be applied to a wide variety of research questions in imaging and cell biology in general.

cell biology↗

bioRxiv2025

Self-supervised learning enables unbiased patient characterization from multiplexed microscopy images

Multiplexed immunofluorescence microscopy offers detailed insights into the spatial architecture of cancer tissue. However, classical single-cell analysis approaches are limited by segmentation accuracy, reliance on predefined features, and the inability to capture spatial interrelationships among cells. We developed a hierarchical self-supervised deep learning framework that learns spatial protein marker patterns from multiplexed microscopy images by encoding the tissue at both local (cellular) and global (tissue architecture) levels. Applied to lung, prostate, and renal cancer tissue microarray cohorts, our method stratified patients into prognostically distinct groups with significantly different survival outcomes. These groupings were consistent with prior expert-driven single-cell analyses, demonstrating the validity of our approach. Furthermore, attention maps extracted from these models highlighted biologically relevant tissue regions associated with specific marker patterns. Overall, our framework effectively profiles complex multiplexed microscopy images and offers a scalable, interpretable tool for improved biomarker discovery, with potential to support more informed cancer treatment decisions.

bioinformatics↗

bioRxiv2024

Fibroblast activation protein defines aggressive EMT-associated cancer cell state and prognosis in localized ccRCC

Clear cell renal cell carcinoma (ccRCC) is among the most immune-infiltrated cancers, where high inflammation correlates with poor clinical outcomes. While epithelial-to-mesenchymal transition (EMT), immunosuppression, and cancer-associated fibroblasts (CAFs) have been linked with poor prognosis in metastatic ccRCC, their relevance and interplay in localized disease remain uncertain. To address this gap, we performed single-cell spatial profiling of highly inflamed localized tumors to characterize interactions between EMT-associated cancer cells and immune subsets, aiming to identify tumor microenvironment cell subsets and states responsible for the poor prognosis in highly inflamed ccRCC. Multiplexed immunofluorescence imaging with a 33-marker panel was applied to 1,728 tissue cores collected from tumor centers, invasive borders, and adjacent benign tissue from 435 localized ccRCC patients. Independent discovery (n=196) and validation (n=239) cohorts were used to confirm the significance of identified prognostic markers. Hierarchical clustering revealed TME subsets enriched for CD45 immune cells, CD31 endothelial cells, or fibroblasts. High CD45 cell density significantly correlated with poor recurrence-free survival across tumor regions, including tumor center, invasive border, and adjacent benign tissue. Within tumors exhibiting high CD45 infiltration, EMT-associated cancer cells strongly expressed fibroblast activation protein (FAP) at invasive borders, independently predicting worse prognosis and increased risk of liver metastasis. Furthermore, CD45high tumors with elevated FAP expression featured immunosuppressive FAP CAFs, M2-like macrophages, exhausted T cells, and regulatory T cells. Tumor-cell-specific FAP expression was validated as an independent prognostic biomarker in both early-stage localized ccRCC (pT1-2) and patients subsequently developing metastases who received sunitinib, highlighting its potential for improved patient stratification. SignificanceFAP expression identifies an aggressive, immunosuppressive EMT-associated subtype of localized ccRCC, uncovering critical interactions between cancer cells and the TME and revealing novel biological insights and therapeutic vulnerabilities driving early-stage ccRCC progression. Graphical AbstractO_ST_ABSTumor-cell FAP Expression Defines an Aggressive, Immunosuppressive EMT Subtype in Highly Inflamed Localized ccRCCC_ST_ABS O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=197 SRC="FIGDIR/small/620479v2_ufig1.gif" ALT="Figure 1"> View larger version (55K): org.highwire.dtl.DTLVardef@d25831org.highwire.dtl.DTLVardef@19c0de7org.highwire.dtl.DTLVardef@54bd9eorg.highwire.dtl.DTLVardef@16d0387_HPS_FORMAT_FIGEXP M_FIG C_FIG Localized ccRCC tumors were stratified by CD45 immune infiltration, revealing that CD45high tumors were associated with significantly poorer recurrence-free survival (5-year RFS rate: 39%). Among these, tumor cell-specific expression of fibroblast activation protein (FAP) identified a particularly aggressive subtype marked by epithelial-to-mesenchymal transition (EMT), reduced endothelial content, and an immunosuppressive tumor microenvironment enriched with M2-like macrophages, regulatory T cells, exhausted T cells, and FAP cancer-associated fibroblasts (CAFs). FAP tumors had an even lower 5-year RFS rate of 26%, compared to 50% in FAP- CD45high tumors. Tumor-cell FAP thus provides an additional layer of risk stratification within the CD45high subgroup and reflects a biologically distinct, immune-evasive tumor state. These findings highlight FAP as both a potential therapeutic target and imaging biomarker in localized ccRCC.

cancer biology↗

bioRxiv2024

Achieving high-resolution single-cell segmentation in convoluted cancer spheroids via Bayesian optimization and deep-learning

Accurate quantification of drug responses in 3D tumor-immune co-cultures remains challenging because complex spatial architecture and cellular heterogeneity limit the interpretability of bulk viability assays. Here, we present MorphoNavigator-3D ('Morphological Navigator in 3D';MoNa-3D), an automated framework for high-resolution, annotation-free single-cell analysis in complex 3D co-cultures. The approach integrates optimized live-cell staining, deep learning-based segmentation, and Bayesian optimization (BO) to adapt end-to-end image-analysis workflows across diverse experimental conditions. MoNa-3D was applied to clear cell renal cell carcinoma (ccRCC)-immune cell 3D-spheroid co-cultures, exposed to PI3K/mTOR pathway inhibitors and immunomodulatory compounds in a high-content imaging-based drug screen. The pipeline was used to extract multiscale phenotypic features encompassing ATP-based cell viability, morphology, nuclear remodeling, spatial dispersion, and immune infiltration. This analysis resolved distinct drug-induced phenotypes: PI3K/mTOR inhibitors promoted spheroid disintegration, nuclear enlargement, and immune exclusion, whereas immunomodulators preserved spheroid architecture and T-cell engagement. Multivariate phenotypic integration distinguished drug classes and revealed intra-class variation, including divergent spatial responses to dual PI3K/mTOR versus mTORC1 inhibition. These phenotypes were consistent with known drug mechanisms, supporting the biological interpretability of the framework. Together, these findings establish MoNa-3D as a generalizable platform for multidimensional phenotypic profiling across complex 3D multicellular systems, supporting applications in drug discovery, tumor-immune interaction studies, and precision oncology.

cancer biology↗

bioRxiv2024

Piximi - An Images to Discovery web tool for bioimages and beyond

Deep learning has greatly accelerated research in biological image analysis yet it often requires programming skills and specialized tool installation. Here we present Piximi, a modern, no-programming image analysis tool leveraging deep learning. Implemented as a web application at Piximi.app, Piximi requires no installation and can be accessed by any modern web browser. Its client-only architecture preserves the security of researcher data by running all computation locally. Piximi offers four core modules: a deep learning classifier, an image annotator, measurement modules, and pre-trained deep learning segmentation modules. Piximi is interoperable with existing tools and workflows by supporting import and export of common data and model formats. The intuitive researcher interface and easy access to Piximi allows biological researchers to obtain insights into images within just a few minutes. Piximi aims to bring deep learning-powered image analysis to a broader community by eliminating barriers to entry.

cell biology↗

bioRxiv2024

Morphological single-cell analysis of peripheral blood mononuclear cells from 390 healthy blood donors with Blood Cell Painting

The morphological diversity of blood immune cells of healthy individuals, critical for recognizing disease-related phenotypes, remains largely uncharacterized. To address this gap, we developed Blood Cell Painting (BCP): a high content, high throughput fluorescence imaging assay for peripheral blood mononuclear cells. We generated a BCP Atlas with images of 50 million cells from 390 healthy blood donors, identifying 18 distinct immune cell morphology clusters. A genome-wide association study of BCP-derived imaging-based cellular features revealed 93 significant associations across 30 genetic loci. These loci include genes linked to mast cell function, inflammation, immune signaling, mitochondrial maintenance and circadian immune modulation. We also observed correlations between immune cell morphological features and clinical traits, such as respiratory conditions and healthcare visits related to contraceptive management, potentially reflecting hormonal influences on immune cell phenotypes. As a proof of concept for clinical application, acute myeloid leukemia subtypes were distinguished by BCP. Our study establishes BCP as a versatile method for immune cell profiling to uncover genetic, phenotypic and clinical determinants of immune cell morphology in health and disease.

cell biology↗

bioRxiv2021

Multiparametric assessment of cellular lipid metabolism in hypercholesterolemia

Systematic insight into cellular dysfunctions can improve understanding of disease etiology, risk assessment and patient stratification. We present a multiparametric high-content imaging platform enabling quantification of low-density lipoprotein (LDL) uptake and lipid storage in cytoplasmic droplets of primary leukocyte subpopulations. We validated this platform with samples from 65 individuals with variable blood LDL-cholesterol (LDL-c) levels, including familial hypercholesterolemia (FH) and non-FH subjects. We integrated lipid storage data into a novel readout, lipid mobilization, measuring the efficiency with which cells deplete lipid reservoirs. Lipid mobilization correlated positively with LDL uptake and negatively with hypercholesterolemia and age, improving differentiation of individuals with normal and elevated LDL-c. Moreover, combination of cell-based readouts with a polygenic risk score for LDL-c explained hypercholesterolemia better than the genetic risk score alone. This platform provides functional insights into cellular lipid trafficking from a few mls of blood and is applicable to dissect metabolic disorders, such as hypercholesterolemia. MotivationWe have limited information on how cellular lipid uptake and processing differ between individuals and influence the development of metabolic diseases, such as hypercholesterolemia. Available assays are labor intensive, require skilled personnel and are difficult to scale to higher throughput, making it challenging to obtain systematic functional cell-based data from individuals. To overcome this problem, we established a scalable automated analysis pipeline enabling reliable quantification of multiple cellular readouts, including lipid uptake, storage and mobilization, from different white blood cell populations. This approach provides new personalized insights into the cellular basis of hypercholesterolemia and obesity. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=200 SRC="FIGDIR/small/440471v4_ufig1.gif" ALT="Figure 1"> View larger version (84K): org.highwire.dtl.DTLVardef@9222c1org.highwire.dtl.DTLVardef@27e289org.highwire.dtl.DTLVardef@89ba82org.highwire.dtl.DTLVardef@33c9cd_HPS_FORMAT_FIGEXP M_FIG C_FIG

cell biology↗

bioRxiv2020

Regression plane concept: analysing continuous cellular processes with machine learning

Biological processes are inherently continuous, and the chance of phenotypic discovery is significantly restricted by discretising them. Using multi-parametric active regression we introduce a novel concept to describe and explore biological data in a continuous manner. We have implemented Regression Plane (RP), the first user-friendly discovery tool enabling class-free phenotypic supervised machine learning.

bioinformatics↗

bioRxiv2019

BIAFLOWS: A collaborative framework to benchmark bioimage analysis workflows

Automated image analysis has become key to extract quantitative information from scientific microscopy bioimages, but the methods involved are now often so refined that they can no longer be unambiguously described using written protocols. We introduce BIAFLOWS, a software tool with web services and a user interface specifically designed to document, interface, reproducibly deploy, and benchmark image analysis workflows. BIAFLOWS allows image analysis workflows to be compared fairly and shared in a reproducible manner, safeguarding research results and promoting the highest quality standards in bioimage analysis. A curated instance of BIAFLOWS is available online; it is currently populated with 34 workflows that can be triggered to process image datasets illustrating 15 common bioimage analysis problems organized in 9 major classes. As a complete case study, the open benchmarking of 7 nuclei segmentation workflows, including classical and deep learning techniques, was performed on this online instance. All the results presented can be reproduced online.

bioinformatics↗

bioRxiv2019

A deep learning framework for nucleus segmentation using image style transfer

Single cell segmentation is typically one of the first and most crucial tasks of image-based cellular analysis. We present a deep learning approach aiming towards a truly general method for localizing nuclei across a diverse range of assays and light microscopy modalities. We outperform the 739 methods submitted to the 2018 Data Science Bowl on images representing a variety of realistic conditions, some of which were not represented in the training data. The key to our approach is to adapt our model to unseen and unlabeled data using image style transfer to generate augmented training samples. This allows the model to recognize nuclei in new and different experiments without requiring expert annotations.

bioinformatics↗

bioRxiv2018

Association of tamoxifen resistance and lipid reprogramming in breast cancer.

BackgroundTamoxifen treatment of estrogen receptor (ER)-positive breast cancer reduces mortality by 31%. However, over half of advanced ER-positive breast cancers are intrinsically resistant to tamoxifen and about 40% will acquire the resistance during the treatment.\n\nMethodsIn order to explore mechanisms underlying endocrine therapy resistance in breast cancer and to identify new therapeutic opportunities, we created tamoxifen-resistant breast cancer cell lines that represent the luminal A or the luminal B. Gene expression patterns revealed by RNA-sequencing in seven tamoxifen-resistant variants were compared with their isogenic parental cells. We further examined those transcriptomic alterations in a publicly available patient cohort.\n\nResultsWe show that tamoxifen resistance cannot simply be explained by altered expression of individual genes, common mechanism across all resistant variants, or the appearance of new fusion genes. Instead, the resistant cell lines shared altered gene expression patterns associated with cell cycle, protein modification and metabolism, especially with the cholesterol pathway. In the tamoxifen-resistant T-47D cell variants we observed a striking increase of neutral lipids in lipid droplets as well as an accumulation of free cholesterol in the lysosomes. Tamoxifen-resistant cells were also less prone to lysosomal membrane permeabilization (LMP) and not vulnerable to compounds targeting the lipid metabolism. However, the cells were sensitive to disulfiram, LCS-1, and dasatinib.\n\nConclusionAltogether, our findings highlight a major role of LMP prevention in tamoxifen resistance, and suggest novel drug vulnerabilities associated with this phenotype.

cancer biology↗