Journal Article2026
Advanced 3D patient-derived glioblastoma organoid-immune cell co-cultures reveal preserved MHC-I tumor cell surface expression and tumor cell-driven T-cell dysfunction
Mahsa Rezaeipour, Marta de Lucas Sanz, Hamed Allahverdi, Gaëlle Triballier, Eliane Klein, Ilaria Salvato, Pilar M. Moreno-Sanchez, Bakhtiyor Nosirov, Petr V. Nazarov, Batuhan Kisakol, Heiko Dussmann, Colum Connolly, Annette T. Byrne, Jochen H.M. Prehn, Dirk Brenner, Bassam Janji, Simone P. Niclou, Anna Golebiewska
Journal of Neuroinflammation
Introduction Glioblastoma represents an aggressive, immunologically cold primary brain tumor largely unresponsive to conventional immunotherapies. Dissecting patient-specific immunosuppressive cues is limited by current preclinical models. Understanding the functional cross-talk between tumor and immune cells across distinct microenvironmental niches is crucial for advancing personalized immunotherapy. Methods We assessed the immunogenic features of glioblastoma cells and cross-talk with T-cells using bulk and single-cell RNA-seq, multicolor flow cytometry, immunohistochemistry, and functional assays in patient tumors, patient-derived organoids (PDOs), orthotopic xenografts and cell cultures. We developed an advanced 3D co-culture system combining PDOs with human peripheral blood T-cells isolated from glioblastoma patients or healthy donor blood to functionally investigate T-cell behavior in a patient-specific context. T-cell subsets integrated and non-integrated into the PDOs were examined for T-cell activation, exhaustion, and T regulatory cell development in response to tumor-derived signals and hypoxia. Results Single-cell RNA-seq-based crosstalk predicted direct ligand-receptor and soluble axes between glioblastoma and T-cells. Glioblastoma cells in patient tumors and diverse patient-derived models showed cell surface MHC class I expression across diverse genetic backgrounds and microenvironmental niches, modulable by interferon gamma. Subsets of CD4 + and CD8 + T-cells integrated into 3D PDOs deprived of myeloid component. PDO-integrated T-cells were enriched for T regulatory cells (Treg), and exhibited a heterogeneous spectrum of activation and dysfunction states, with differential expression of PD-1, LAG-3, TIM-3, CD45RO, Granzyme B, TCF1, and TOX. T-cells subjected to soluble factors were largely unaffected and retained capacity to reintegrate PDOs at sequential co-cultures. Hypoxia influenced T-cell phenotype, but did not prevent activation or infiltration within the early stages of co-culture. Conclusion Our work provides insights into tumor-intrinsic cues of immune evasion in glioblastoma and introduces a scalable patient-relevant ex vivo platform for functional studies. Surface MHC-I was detectable in all assessed glioblastoma models; and was retained in contact with T-cells, indicating that MHC-I loss is not their predominant immune-escape mechanism. Immunosuppression occurred in direct contact with glioblastoma cells, involving both TOX-dependent exhaustion and TOX-independent dysfunction. Advanced co-cultures of glioblastoma PDOs may serve as a powerful tool for functional studies ex vivo, predicting clinically-relevant responses to immunotherapeutic treatments in the future.
Journal Article2026
Structural design of diblock glycopolymers modulates galectin affinity
Marina Rodrigues Tavares, Michaela Hovorková, Markéta Frejková, Vladimı́r Křen, Tomáš Etrych, Pavla Bojarová, Petr Chytil
European Polymer Journal
Galectins are carbohydrate-binding proteins that have recently attracted considerable attention due to their vital role in cancer development and progression, and support of cancer cell survival. The two most common galectins, galectin-1 (Gal-1) and galectin-3 (Gal-3), are promising targets for therapeutic inhibition. Their carbohydrate-binding affinities can be enhanced by exploiting the multivalency effect induced by multiple glycosylation binding sites. This study investigates diblock glycopolymers differing in carbohydrate ligand content and distribution along the polymer chain. Developed diblock glycopolymers take advantage of biocompatible and highly water-soluble N -(2-hydroxypropyl) methacrylamide (HPMA) copolymers, which form the entire polymer backbone. The diblock presentation with a pristine HPMA block attached to the ligand-carrying block enabled the presentation of a high number of ligands while maintaining the favorable physicochemical properties of the backbone, which may have otherwise been compromised by a high ratio of ligand to backbone. Using a simple lactose or N -acetyllactosamine ligand, we identified the polymer design and the ratio of both polymer blocks that best enhanced binding affinity to Gal-1 or Gal-3. Our findings demonstrate that a well-designed glycopolymer structure featuring a basic ligand, such as lactose, can achieve performance comparable to that of higher-affinity ligands, emphasizing the crucial role of structural optimization in multivalent interactions. These results highlight the potential of tailored HPMA-based diblock glycopolymers as a platform for designing more effective drug-free nanomedicines that target and inhibit galectins in cancer therapy.
Journal Article2026
Transglycosylation Capabilities of Wild-Type α-l-Fucosidase iso1 from Paenibacillus thiaminolyticus and Its Engineered Mutants: Preparation of Fucosylated Oligosaccharides
Patricie Vodičková, Lucie Klimešová, Pavlína Nekvasilová, Lucie Petrásková, Helena Pelantová, Terézia Kovaĺová, Petra Lipovová, Pavla Bojarová, Eva Benešová
Microbial Biotechnology
To enhance the yield of l-fucosylated molecules synthesized via transfucosylation, we employed α-l-transfucosidases, enzymes engineered from α-l-fucosidases to favour transglycosylation over hydrolysis. This study investigated the transglycosylation potential of mutated variants of α-l-fucosidase iso1 from Paenibacillus thiaminolyticus designed by structural comparison with Thermotoga maritima α-l-transfucosidase. Using site-directed mutagenesis, point mutations (S237A, S237G, S237P, S237V, Y189F) were introduced, and the corresponding recombinant proteins were successfully expressed and purified. Among them, the S237V variant achieved the highest overall yield of fucosylated lactose and increased the transglycosylation/hydrolysis ratio more than 20-fold compared to the wild-type enzyme. This variant enabled regioselective synthesis of 3'-and 6'-fucosyllactose, as well as a functionalized 3'-fucosyllactose bearing a Boc-protected ethylthioureidyl linker (3'-FucLac-tBoc), with all structures confirmed by NMR spectroscopy. The engineered α-l-transfucosidase iso1 from Paenibacillus thiaminolyticus catalysed the regioselective synthesis of structurally defined fucosyllactose derivatives, including C-1 functionalized analogs suitable for immobilization on biosensor surfaces or macromolecular scaffolds, thus expanding the biocatalytic toolbox and demonstrating the potential of semi-rational enzyme design for targeted HMO-based glycoengineering applications.
Journal Article2026
Structural brain changes in the development of essential tremor: toward early diagnosis
Karmele López de Ipiña, Jose Ignacio Sánchez-Méndez, Jordi Solé‐Casals, Rafael Romero-García, Elsa Fernández, Anujan Poologaindran, Maite García‐Sebastián, Pablo Martinez-Lague, Catalina Requejo, J.F. Marti-Masso, Alberto Bergareche, John Suckling
Frontiers in Aging Neuroscience
Background: Essential tremor (ET) is the most common movement disorder. Its etiology and neuropathology remain controversial, but could be clarified with neuroimaging. Objectives: This study investigated brain structures derived from T1-weighted MRI associated with tremor severity in analyses focusing on the motor circuit and whole-brain to test competing hypotheses. Methods: Structural MRI at 3T was acquired from 50 ET participants and 50 healthy controls (HC). Voxel-based morphometry and surface-based processing generated measures of brain structure. Three two-group comparisons were undertaken: 1) HC versus ET participants; 2) HC versus ET participants with low tremor; and 3) HC versus ET participants with high tremor. For ET participants, correlations with tremor severity were also explored. Results: Within the motor circuit and across the whole brain, no significant case-control differences were observed after correction. Likewise, vertex-wise analyses of cortical volume, cortical thickness, and gyrification did not reveal significant differences. Uncorrected voxel-wise analyses suggested subtle reductions in the cerebellum and the motor circuit, particularly in ET participants with greater tremor severity. However, these effects were generally small and not consistently observed across analyses. Conclusion: The present analysis showed that ET severity was associated with weak structural differences, mainly within the motor circuit, although these findings did not survive correction. Consistent with previous literature, structural MRI abnormalities in ET appeared subtle and heterogeneous across studies, potentially becoming more detectable in patients with more severe symptoms. Although not designed to assess disease progression directly and noting that no finding survived correction for multiple comparisons, the direction of uncorrected exploratory trends within the motor circuit is consistent with the hypothesis of progressive structural alterations; however, this interpretation remains speculative and requires replication in larger, longitudinal cohorts.
Journal Article2026
Clinical registry metadata as a hidden bottleneck in AI-driven drug discovery: a computational audit of translational phase data in glioma research
Iryna Oliynyk
Scientific Reports
Clinical translation in glioma and glioblastoma remains inefficient despite advances in computational drug discovery. An underrecognized contributor to this gap is structural degradation of clinical registry metadata. We analyzed the complete set of 2.357 glioma-related clinical trial records available in the WHO ICTRP registry at the time of extraction (3 January 2026). A deterministic Python-based validation pipeline was developed to normalize phase and study-type annotations while distinguishing technical voids, methodological non-applicability, and structurally inconsistent entries. Two quantitative indices were introduced: the Reporting Gap ([Formula: see text]), reflecting phase metadata completeness, and the Maturity Ratio ([Formula: see text]), describing the balance between early and late translational stages. Phase annotation showed substantial structural degradation, with large fractions of records lacking machine-interpretable phase labels, whereas study-type fields demonstrated high completeness but severe terminological fragmentation. This asymmetry indicates a systemic mismatch between real-world research designs and the classical phase ontology. Consequently, clinically relevant evidence remains invisible to algorithms, which may lead to biased AI-based translational evaluation. The results suggest that semantic validation of registry metadata is a critical prerequisite for robust integration of clinical data into computational drug development cycles.
Journal Article2026
Modelling brain tumours with organoids: towards precision medicine in neuro-oncology
Marta de Lucas Sanz, Simone P. Niclou, Anna Golebiewska
Nature Reviews Neurology
Precision medicine in neuro-oncology hinges on our ability to decode the molecular and functional complexity of brain tumours. These tumours are highly heterogeneous and dynamic ecosystems that remain challenging to replicate with traditional experimental models. Organoids have emerged as next-generation tools for capturing the diversity of brain tumours in a clinically relevant context. Here, we review advances in applying organoids as tumour avatars to foster precision medicine. We assess current methodologies for generating organoids from various brain tumour types and illustrate how these models have enhanced mechanistic insight into tumour initiation, progression and resistance to treatment. Furthermore, we explore their translational potential in functional drug screening and treatment stratification, offering a powerful preclinical and co-clinical platform with which to personalize neuro-oncology.
Journal Article2026
From patient to tumor organoid: Culture protocol choice controls glioblastoma tumor architecture and identity
Jana Slováčková, Ondřej Bernatík, Katarína Čimborová, Martin Barák, Michal Hendrych, Karolina Kocourkova, Marie Sulcova, Jaroslav Oľha, Kateřina Amruz Černá, Zdeněk Hodný, Radim Jančálek, Dáša Bohačiaková
Brain Pathology
Patient-derived tumor organoids are widely used in cancer research, yet the biological impact of tissue processing during model generation remains unclear. Fragment-based and dissociation-based (DIS) approaches are commonly assumed to trade fidelity for uniformity, but their molecular consequences remain incompletely defined. To address this gap, we performed a proteome-wide comparison of fragment-based (cut-and-culture [CUT]) and DIS glioblastoma organoid protocols using quantitative mass spectrometry. Organoids from multiple patient tumors were cultured under growth factor-free or growth factor-supplemented conditions and compared with matched primary tissue. Results show that both protocols produced technically robust glioblastoma organoids when maintained in their native media. However, CUT organoids matched the reproducibility of DIS cultures while preserving a broader extracellular matrix (ECM) repertoire and networks linked to collagen assembly, vascular support, and cell-matrix signaling. DIS cultures were biased toward exogenous basement membrane components and proliferative, growth factor-responsive states. Across tumors, CUT organoids consistently showed greater proteomic similarity to matched primary tissue, retaining neural, glial, stromal, and extracellular features largely absent from DIS models. Taken together, fragment-based glioblastoma organoids can be both reproducible and biologically faithful. Tissue dissociation acts as a major perturbation that reshapes ECM organization, cellular states, and tumor identity, making protocol choice a critical determinant of model fidelity and translational relevance.
Journal Article2026
Glioblastoma–natural killer cell crosstalk: insights from dynamic spheroid models reveal the importance of secreted cytokines and the CD155 axis
Anamarija Habič, Tina Kolenc Milavec, Pia Žižek, Špela Kladnik, Bernarda Majc, Emanuela Senjor, Milica Perišić Nanut, Andrej Porčnik, Borut Prestor, Urban Švajger, Metka Novak, Barbara Breznik
Cell Communication and Signaling
Glioblastoma (GB) is an aggressive primary brain cancer with poor patient prognosis. Natural killer (NK) cells can recognise and eliminate a range of malignant cells, including GB stem cells, which drive GB recurrence. NK cell-based immunotherapy has emerged as a promising approach for GB treatment, but a better understanding of the complex crosstalk between GB and NK cells is needed, particularly within the immunosuppressive GB tumour microenvironment. In this study, we established a reproducible protocol for the production and dynamic culture of uniformly sized GB spheroids using the Celvivo Clinostar system. Our spheroids recapitulated the heterogeneous structure of GB and expressed ligands for NK cell receptors at levels distinct from those observed in corresponding GB cell lines in standard culture, implicating altered sensitivity of GB cells to NK cells in dynamic 3D cultures. GB-NK cell crosstalk was GB cell type dependent and the ability of NK cells to infiltrate GB did not necessarily correlate with their cytotoxicity against GB cells. Spheroids derived from differentiated GB cells secreted higher levels of immunomodulatory cytokines compared to spheroids from GB stem-like cells, and a prominent increase in the secretion of immune-attracting factors was observed in their co-cultures with NK cells. Finally, the CD155-DNAM1/TIGIT axis was indicated as an important regulator of NK cell cytotoxicity against GB stem-like cells. Collectively, our results highlight important factors in GB-NK cell communication and provide a groundwork for further targeted research as well as therapeutic evaluation of NK cell-based approaches in the established dynamic 3D cultures.
Journal Article2026
Artificial intelligence for brain cancer management: multimodal digital workflows for diagnosis, treatment planning, and longitudinal monitoring
Ganesan Saibaba, Olawumi Giwa, Thomas Calvert Booth, Terézia Kisková-Šimková, Arban Uka, Florenc Skuka, Bilal Javed, Dário Ferreira, Sandra Saraiva Ferreira, Jewel Sengupta, Muhammet Karaman, Gabriel F. Calvo, Xinzhong Li, Tatjana Lj. Mitrović, Annalisa Occhipinti, Claudio Angione, Nabiha Missaoui, Milana Frenkel‐Morgenstern
European Journal of Cancer
Brain cancers, especially glioblastoma, remain among the deadliest adult cancers, with outcomes largely unchanged despite multimodal treatments. This review summarizes cutting-edge artificial intelligence (AI) and machine learning (ML) advances transforming neuro-oncology in diagnostics, molecular profiling, treatment planning, and monitoring. Key findings show AI-driven radiomics and deep learning (DL) reaching over 90% accuracy in tumour segmentation and grading from MRI and whole-slide images, non-invasive IDH/MGMT prediction through liquid biopsy (LB) analysis, and augmented reality-guided resection that maximizes tumour removal while safeguarding expressive cortex. Treatment planning benefits from hybrid Convolutional Neural Networks (CNN)-Transformer models for immunotherapy stratification and blood-brain barrier penetrant drug repurposing, while real-time progression detection via multimodal integration helps differentiate true progression from pseudoprogression. Despite these advances, significant challenges remain, including data scarcity and imbalance in rare subtypes, domain shift due to imaging variability, black-box model behaviour eroding trust, regulatory requirements for prospective validation, and workflow fragmentation. Emerging solutions include federated and transfer learning for scalable model development, explainable AI (such as SHapley Additive exPlanations (SHAP) and attention interpretation for vision transformers) to foster clinician-AI collaboration, and foundation models pretrained on large-scale neuro-oncology datasets to facilitate personalization. Achieving this potential will depend on harmonized multi-omics registries, strong ethical and regulatory governance, and deliberate human-AI collaboration frameworks to integrate these tools into precision neuro-oncology.
Report2026
Net4Brain Training School on Mathematical Oncology: Final Activity Report
Gabriel F. Calvo
Net4Brain / COST Action CA22103
Final activity report of the Net4Brain Training School on Mathematical Oncology, held 15–18 September 2026 in Almagro, Spain. The School brought together 15 trainees from ten countries and 12 trainers and academic contributors for ten one-hour lectures and five supervised mini-projects on brain-tumour research problems — from glioblastoma spatial transcriptomics and early tumour evolution to treatment resistance, FGFR3–TACC3 sequence and structure, and brain-metastasis growth. The report documents the scientific programme, participation, training outcomes and the follow-up online early-career researcher session proposed for October 2026.
Journal Article2026
Integrating computational fluid dynamics into organ-on-chip systems: a glioblastoma-centred design and validation framework
Hooman Taleban, Xinzhong Li, Zulfiqur Ali, Karunakaran Kalesh, Jai Prakash, Tugba Bagci-Onder, Barbara Breznik
Frontiers in Bioengineering and Biotechnology
Glioblastoma (GBM) remains one of the most lethal and treatment-resistant brain cancers, driven in part by the complexity of its tumour microenvironment (TME). While organ-on-chip (OoC) platforms offer more physiologically relevant models than traditional 2D or static 3D systems, their design remains largely empirical, lacking predictive control over flow conditions, biochemical gradients, and mechanical cues. Computational Fluid Dynamics (CFD) has emerged as a powerful tool to enhance the design, precision, and biological fidelity of OoC platforms. This comprehensive review highlights current limitations in replicating GBM's biological complexity and technical constraints in device fabrication and maintenance, mapping them to specific CFD strategies. It synthesises current strategies into a structured workflow for integrating CFD into the design, optimisation, and validation of microfluidic tumour models. Finally, this review positions CFD as a core component of GBM-on-chip development and explores how its integration with AI-based optimisation can advance the creation of more predictive, scalable, and biologically relevant in vitro tumour models.
Journal Article2026
Case Report: Rare presentation of spinal meningioma in Kosovo
Flamur Hasimi, Fjolla Ymeri
Open Research Europe
This case report outlines the diagnosis and management of a spinal meningioma in an 18-year-old female presenting with progressive weakness in both lower limbs. Magnetic resonance imaging revealed an intradural extramedullary tumour compressing the spinal cord, subsequently confirmed as a meningioma. Urgent neurosurgical intervention was performed, involving total laminectomy at the TH6/7 vertebral level and meticulous tumour resection. Postoperatively, the patient demonstrated significant neurological improvement, with muscle strength increasing from 2/5 to 4/5. This case highlights the importance of prompt diagnosis, multidisciplinary management, and close postoperative monitoring in achieving favourable outcomes for spinal meningiomas.
Journal Article2026
Bridging modalities with AI: a review of AI advances in multimodal biomedical imaging
Le Minh Thao Doan, Kaveh Shahhosseini, Suraj Verma, Abdolreza Marefat, Giorgio Locicero, Sneha Verma, Claudio Angione, Annalisa Occhipinti
communications engineering
The rapid evolution of AI has facilitated innovative solutions in analysing different biomedical imaging modalities. By leveraging the complementary information from each modality, multimodal AI solutions have shown a huge potential to go beyond human capabilities and offer advances in bioimaging. At the same time, new foundation models and transformer-based architectures are now poised to address unsolved challenges in this field. This review aims to explore and discuss the state-of-the-art AI techniques applied in multimodal biomedical imaging, presenting the key challenges and future directions. We discuss several integration strategies to combine multiple biomedical imaging data types. We also focus on methods to overcome the open challenges related to data quality, model interpretability, and ethical implications.
Journal Article2026
COST action CA22103 Net4Brain: Pan-European network advancing translational research in brain cancer
Terezia Kiskova-Simkova, Patrik Simko, Milica Pešić, Annalisa Occhipinti, Barbara Breznik, Anna Golebiewska, Alkinoos Athanasiou, Theodora Gazea, Aida Hajdarpasic, Thomas Booth, Tugba Bagci Onder, Yasin Kaymaz, Claudio Angione, Uchenna Ngameduru, Radim Jancalek, Xinzhong Li
ELSEVIER
Brain and central nervous system cancers remain the deadliest forms of cancer, with limited improvement in survival following treatment. The Net4Brain COST Action (CA22103) addresses the urgent need for coordinated and multidisciplinary approaches to understand the pathogenesis, optimize early diagnosis, and personalize treatment strategies. Net4Brain unites a pan-European network of experts in basic and clinical neuroscience, oncology, molecular biology, data science and bioengineering. Through five thematic Working Groups (WGs), innovative Young Researchers Group and one dissemination WG, supported by Creative Group, the Action focuses on non-invasive biomarker discovery, preclinical modeling, tumor profiling, computational modeling, and treatment optimization. This initiative supports collaborative research, capacity building, and the development of a shared pan-European brain cancer centralized virtual database and biobank. Progress has been made in standardizing biomarker discovery, building innovative advanced preclinical models, and fostering integration of omics and imaging data for clinical applications. Net4Brain overcomes existing fragmentation in brain cancer research by providing a transdisciplinary platform for research and development, innovation, dissemination, and training, ultimately aiming to improve diagnostic accuracy, therapeutic outcomes, and patient quality of life. It also builds critical mass of knowledge/experts in the field of brain cancer and neuro-oncology.
Journal Article2026
Exploring glioblastoma microenvironment using organoids: opportunities, limitations, and emerging concepts
Anuja Lipsa, Gaëlle Triballier, Anna Golebiewska
ELSEVIER
Glioblastoma (GBM) remains the most lethal primary brain tumor in adults. These aggressive tumors evolve as dynamic, spatially organized ecosystems in which tumor cells continuously interact with the surrounding brain parenchyma and systemic environment. These reciprocal interactions actively drive invasion, therapeutic resistance, and ultimately, inevitable recurrence. Modelling this level of complexity has long required significant compromise. GBM organoids have emerged as a promising intermediate platform, bridging the gap between costly, low-throughput animal models and overly simplistic two-dimensional in vitro cultures. In this review, we summarize the diverse protocols currently used for GBM organoid derivation and long-term maintenance, focusing on the recapitulation of microenvironmental traits. We further discuss how these systems enable the investigation of tumor niche architecture and dynamic crosstalk with key components of the tumor microenvironment, including neural and immune elements, vascular-associated signals, and extracellular matrix cues. Although the inherent limitations of ex vivo systems must be carefully considered, increasingly advanced and well-designed protocols will enable robust interrogation of interactions within the GBM ecosystem and provide powerful platforms for therapeutic testing.
Journal Article2026
Brain Cancer: Molecular Alterations and Emerging Trends in Neuropharmacology
Beata Leskova, Ilaria D'Agostino, Simona Mattova, Nicol Urbanska, Alzbeta Blicharova, Patrik Simko, Aylin Toplu, Muhammet Karaman, Terezia Kiskova-Simkova
International Journal of Molecular Sciences
Central nervous system (CNS) tumors represent a heterogeneous group of neoplasms associated with significant morbidity and mortality despite their relatively low incidence. Advances in the fifth edition of the World Health Organization (WHO) classification have emphasized the integration of histopathological, immunohistochemical, and molecular features, fundamentally transforming diagnostic and prognostic frameworks in neuro-oncology. This manuscript aims to provide an overview of CNS tumor biology, focusing on key diagnostic markers, genetic and epigenetic alterations, and emerging therapeutic strategies. It further describes recent advances in multi-omics approaches and artificial intelligence, which enable deeper characterization of tumor heterogeneity and support the development of precision medicine strategies. Finally, current and emerging therapeutic modalities, including combination therapies, targeted treatments, and novel molecular targets, are examined with emphasis on overcoming resistance mechanisms and improving clinical outcomes. Overall, the integration of molecular biology, advanced diagnostics, and innovative therapeutic approaches represents a critical step toward personalized management of CNS tumors and improved patient survival.
Journal Article2025
Targeting galectin-4 with glycoconjugates of varying architectures: a multivalency étude with accent on anti-tumor effect and protection against apoptosis
Andrea Vopálenská, Soňa Balogová, Michaela Hovorková, David Vrbata, Jakub Červený, Lucie Petrásková, Helena Pelantová, Pavlína Nekvasilová, Kristýna Slámová, Josef Cvačka, Anna Shaliutina-Loginova, Vladimı́r Křen, Pavla Bojarová
European Journal of Medicinal Chemistry
Galectin-4 (Gal-4) is a protumorigenic protein that strongly participates especially in gastrointestinal cancer and is increasingly recognized as a therapeutic target in gastrointestinal malignancies, where its dysregulated expression contributes to tumor progression and immune modulation. Its abundance and the extent of its involvement in these pathologies are comparable with its much better-studied counterparts, galectin-1 and -3. However, to date, no systematic effort has been made to design efficient defined Gal-4 inhibitors with a potential therapeutic effect. In this work, we present a library of biocompatible multivalent Gal-4 inhibitors with diverse architectures to investigate how different modes of multivalent presentation of a common ligand, lactose, influence the binding affinity to Gal-4. The most efficient scaffold was then loaded with a high-affinity tetrasaccharide ligand of Gal-4, derived from lacto- N -tetraose, to further enhance Gal-4 binding. The resulting glycopolymer exhibited outstanding performance in inhibiting Gal-4-induced apoptosis of T lymphocytes, and, in addition, it efficiently scavenged Gal-4 from the surface of cancer cells. Both of these processes are relevant for tumor immune evasion. Given the high biocompatibility, and low toxicity of the POx carrier, this glycopolymer has a strong potential as a candidate for further development of glycotherapeutics against Gal-4-associated diseases such as gastrointestinal cancer. • New defined multivalent inhibitors of Gal-4 were synthesized • Gal-4 prefers statistical to clustered ligand presentation on polyoxazoline carrier • Glycopolymer inhibits Gal-4-induced apoptosis of T cells in nanomolar concentration • Glycopolymer scavenges exogenous Gal-4 from cancer cell microenvironment
Journal Article2025
Deep Learning Model-Based Architectures for Lung Tumor Mutation Profiling: A Systematic Review
Samanta Ortuño‐Miquel, Reyes Roca, Cristina Alenda, Ignacio Aranda, Natividad Martínez-Banaclocha, Sandra Amador, David Gil
Cancers
Background/Objectives: Lung cancer (NSCLC), which accounts for approximately 85% of lung cancers, exhibits marked heterogeneity that complicates molecular characterization and treatment selection. Recent advances in deep learning (DL) have enabled the extraction of genomic-related morphological features directly from routine Hematoxylin and Eosin (H&E) histopathology, offering a potential complement to Next-Generation Sequencing (NGS) for precision oncology. This review aimed to evaluate how DL models have been applied to predict molecular alterations in NSCLC using H&E-stained slides. Methods: A systematic search following PRISMA 2020 guidelines was conducted across PubMed, Scopus, and Web of Science to identify studies published up to March 2025 that used DL models for mutation prediction in NSCLC. Eligible studies were screened, and data on model architectures, datasets, and performance metrics were extracted. Results: Sixteen studies met inclusion criteria. Most employed convolutional neural networks trained on publicly available datasets such as The Cancer Genome Atlas (TCGA) to infer key mutations including EGFR, KRAS, and TP53. Reported areas under the curve ranged from 0.65 to 0.95, demonstrating variable but promising predictive capability. Conclusions: DL-based histopathology shows strong potential for linking tissue morphology to molecular signatures in NSCLC. However, methodological heterogeneity, small sample sizes, and limited external validation constrain reproducibility and generalizability. Standardized protocols, larger multicenter cohorts, and transparent validation are needed before these models can be translated into clinical practice.
Journal Article2025
Novel Adamantane–Sclareol Hybrids Exploit ROS Vulnerability to Overcome Multidrug-Resistance in Glioblastoma Cells
Ema Lupšić, Pavle Stojković, Marija Grozdanić, Nataša Terzić-Jovanović, Milica Pajović, Fani Koutsougianni, Dimitra Alexopoulou, Igor M. Opsenica, Milica Pešić, Ana Podolski-Renić
Molecules
Multidrug resistance (MDR) presents a significant challenge in the treatment of glioblastoma. We evaluated six novel adamantane–sclareol hybrids that integrate a natural labdane diterpene scaffold with an adamantane moiety to address this issue. Compounds 2, 5, and 6 demonstrated the ability to bypass P-glycoprotein (P-gp)-mediated resistance in resistant U87-TxR cells and induced collateral sensitivity, with compound 2 exhibiting the highest selectivity for glioblastoma compared to normal glial cells. Mechanistic studies revealed that compounds 2 and 5 selectively triggered early apoptosis in MDR cells, significantly elevated levels of H2O2 and peroxynitrite, and disrupted mitochondrial membrane potential. Additionally, these compounds altered the expression of key genes involved in glutathione (GSH) and thioredoxin (Trx) antioxidant defense systems and increased ASK1 protein levels, indicating the activation of ROS-driven apoptotic signaling. A preliminary Rag1 xenograft study demonstrated that compound 5 effectively suppressed tumor growth without causing significant weight loss.
Journal Article2024
Integration of 3D-QSAR, molecular docking, and machine learning techniques for rational design of nicotinamide-based SIRT2 inhibitors
Aleksandra Ilić, Nemanja Djoković, Teodora Djikić, Katarina M. Nikolic
Computational Biology and Chemistry
Selective inhibitors of sirtuin-2 (SIRT2) are increasingly recognized as potential therapeutics for cancer and neurodegenerative diseases. Derivatives of 5-((3-amidobenzyl)oxy)nicotinamides have been identified as some of the most potent and selective SIRT2 inhibitors reported to date (Ai et al., 2016; Ai et al., 2023, Baroni et al., 2007). In this study, a 3D-QSAR (3D-Quantitative Structure-Activity Relationship) model was developed using a dataset of 86 nicotinamide-based SIRT2 inhibitors from the literature, along with GRIND-derived pharmacophore models for selected inhibitors. External validation parameters emphasized the reliability of the 3D-QSAR model in predicting SIRT2 inhibition within the defined applicability domain. The interpretation of the 3D-QSAR model facilitated the generation of GRIND-derived pharmacophore models, which in turn enabled the design of novel SIRT2 inhibitors. Furthermore, based on molecular docking results for the SIRT1-3 isoforms, two classification models were developed: a SIRT1/2 model using the Naive Bayes algorithm and a SIRT2/3 model using the k-nearest neighbors algorithm, to predict the selectivity of inhibitors for SIRT1/2 and SIRT2/3. External validation parameters of the selectivity models confirmed their predictive power. Ultimately, the integration of 3D-QSAR, selectivity models and prediction of ADMET properties facilitated the identification of the most promising selective SIRT2 inhibitors for further development.
Report2024
Sabbatical Report: Hands-On Experience in 3D Cell Culture and Glioma-on-Chip Research
Hooman Taleban
Twente University Sabbatical Report
This sabbatical report details Hooman Taleban's one-month research visit to Twente University, Netherlands, from 19 August to 23 September 2024. The goal was to gain hands-on experience in 3D cell culture, organoids, and spheroids to support ongoing PhD research on glioma-on-chip. Hosted in Prof. Jai Prakash's laboratory, the sabbatical focused on learning fundamental biological techniques such as cell handling, scaffolding, and spheroid formation. Despite the absence of active glioma-on-chip projects, work on breast cancer cell lines provided critical transferable skills. Collaborations with researchers including Prof. Andries van der Meer, Dr. Liliana Moreira Teixeira, and Prof. Marcel Karperien enriched the experience, fostering interdisciplinary perspectives on organ-on-chip systems. The sabbatical strengthened both technical proficiency and conceptual understanding, bridging the gap between engineering and biological experimentation.