Frankfurt Cancer Institute

Optical calcium imaging is a powerful tool for recording neural activity across a wide range of spatial scales, from dendrites and spines to whole-brain imaging through two-photon and widefield microscopy. Traditional methods for analyzing functional calcium imaging data rely heavily on spatial features, such as the compact shapes of somas, to extract regions of interest and their associated temporal traces. This spatial dependency can introduce biases in time trace estimation and limit the applicability of these methods across different neuronal morphologies and imaging scales. To address these limitations, the Graph Filtered Temporal Dictionary Learning (GraFT) uses a graph-based approach to identify neural components based on shared temporal activity rather than spatial proximity, enhancing generalizability across diverse datasets. Here we present significant advancements to the GraFT algorithm, including the integration of a more efficient solver for the L1 least absolute shrinkage and selection operator (LASSO) problem and the application of compressive sensing techniques to reduce computational complexity. By employing random projections to reduce data dimensionality, we achieve substantial speedups while maintaining analytical accuracy. These advancements significantly accelerate the GraFT algorithm, making it more scalable for larger and more complex datasets. Moreover, to increase accessibility, we developed a graphical user interface to facilitate running and analyzing the outputs of GraFT. Finally, we demonstrate the utility of GraFT to imaging data beyond meso-scale imaging, including vascular and axonal imaging.

There are compelling reasons to opt for primary human natural killer (NK) cells when validating Ca2+ indicators. 1.) NK cells exhibit a high degree of vulnerability to stressors such as indicator loading or light exposure. 2.) The lack of research on NK Ca2+ signaling underscores the necessity for developing reliable assays. 3.) The increased utilization of NK cell therapies necessitates a more profound comprehension of Ca2+ dependent signal transduction. Consequently, an assay was developed to monitor cytosolic Ca2+ signals in individual NK cells simultaneously with their cytotoxic function against cancer cells. We used this assay to assess the suitability of fura-2, fura-PE3, fura-8, fura-10 or fura-red for quantifying Ca2+ signals in NK cells without compromising their cytotoxic function. In contrast to the widely used fura-2, its red-shifted derivative fura-10 did not interfere with NK cytotoxicity over several hours. It exhibited a superior signal-to-noise ratio and good dynamic range, accompanied by minimal bleaching or leakage. Fura-8 and fura-red also preserved NK cell cytotoxicity, but had other disadvantages compared to fura-10. We successfully used fura-10 to report Ca2+ signals in NK cells from blood donors and patients diagnosed with lymphoma and leukemia over several hours at 37 °C during apoptotic or necrotic killing of different cancer cells (K562, THP1, OCI-AML2, and TMD8). Additionally, we show that fura-10 is well suited to report Ca2+ signals in intact murine pancreatic islets, another stress-sensitive cell preparation. Consequently, fura-10 is an optimal choice for measuring Ca²⁺ in primary human NK cells and other primary cell preparations.

Human tissue samples are a crucial resource for cancer research, offering key insights into physiological and pathological processes while enabling comprehensive characterization of molecular signatures across cancer subtypes. Recent advances in genetic analysis techniques lead to a substantiall expansion of specific molecular pathways knowledge. The application of these technologies to fresh tissue samples from patients undergoing surgical resection for metastatic disease represents a promising approach to gain a deeper understanding of the biology of brain metastasis. Brain metastases remain particularly challenging due to their poor prognosis and the complex mechanisms underlying central nervous system invasion. This study sought to identify preoperative factors influencing the research utility of fresh brain metastasis tissue samples. A pipeline was established to transfer fresh, surplus tissue from surgical resections to research laboratories with histological quality assessment. Of the fifty-five fresh specimens collected, thirty-eight (69%) were classified as suitable for further research applications. Statistical analysis revealed that only two factors significantly affected sample quality. First, the extent of MRI-derived necrosis was significantly higher in unsuitable samples (mean 18.0%) than in suitable samples (mean 8.3%) (p=0.0273). Second, prior treatment with target-specific therapeutics (TST) was associated with a lower proportion of suitable samples (47%) compared to no prior TST (79%) (p=0.018). Logistic regression confirmed these variables as significant predictors, with MRI-derived necrosis (odds ratio 1.049) and target-specific therapy exposure (odds ratio 4.486) independently increasing the likelihood of obtaining suboptimal samples. Other parameters, including age, gender, metastasis volume, localization, primary cancer site, and other therapeutic interventions, showed no significant impact on sample quality. Based on these findings, the collection pipeline was modified to include evaluation by board-certified neuropathologists before samples are used for research purposes, improving the efficiency of translational research utilizing brain metastasis tissue.

Abstract BACKGROUND Glioblastoma are hallmarked by clonal evolution and transcriptional heterogeneity linked to genetic and environmental diversity resulting in therapy resistance and disease progression. In this study, we integrate heritable molecular barcoding and spatial transcriptomics to enable high-resolution cell lineage tracing in the spatial context of the glioblastoma TME. MATERIAL AND METHODS Patient-derived glioblastoma cells from two individual cases were isolated from anatomically and pathologically distinct regions: the primary tumor core, the tumor periphery, and the recurrent lesion. Each subline was transduced with predefined FlexTrack lentiviral plasmids encoding GFP, resulting in uniquely barcoded (9N) cell lines. Immediately before injection, the barcoded glioblastoma populations were pooled and introduced into ex vivo human cortical slices. Live-imaging was used to trace GFP-positive, barcoded cell populations. After incubation, the fresh-frozen slices were processed for spatial transcriptomics and the line-specific barcodes were mapped according to the 3′ poly(A) tails of captured transcripts. Graph neural networks based on single-cell deconvolution were employed to spatially analyze the cellular ecosystems. RESULTS We successfully deconvoluted and reconstructed the trajectories of injected glioblastoma cells linking their transcriptome phenotypes along with clonal identity. Multi-omic integration of spatial copy-number alterations and barcoding based pseudo-trajectories enabled reconstruction of the temporal transcriptional dynamics linked to the heterogeneity of tumor ecosystems. CONCLUSION This study serves as a proof-of-concept for integrating molecular barcoding and spatial transcriptomics in cortical slice culture. Our approach to molecular barcoding in complex model systems opens new avenues to investigate the contextual dynamics of clonal evolution in glioblastoma.

Chimeric antigen receptor (CAR)-T cell therapy has transformed the outcomes of patients with haematological malignancies, yet its use is limited by labour-intensive manufacturing, constrained production capacity and variable clinical performance. In vivo CAR-T cell engineering, in which CAR-T cells are generated directly inside the patient's body, seeks to overcome these challenges by eliminating the need for ex vivo cell processing and complex logistics, as well as improve clinical performance. Recent advances in virology, RNA medicines and nanotechnology have catalysed a radical overhaul of this approach, which uses targeted delivery systems such as lentiviral vectors and lipid nanoparticles to introduce CAR-encoding genetic material into endogenous T cells. Early clinical studies have shown efficient transduction, sustained CAR expression and initial signs of antitumour activity, establishing proof of concept. This Review explores the underlying technologies - including RNA delivered by lipid nanoparticles and engineered viral vectors - and discusses how they are being adapted to develop more broadly applicable, scalable, safe and effective CAR-T cell therapies. By removing the need for ex vivo manipulation and chemotherapeutic conditioning, this strategy could enable the wider application of CAR-T cell therapies not just to blood cancers but to autoimmune diseases for which ex vivo CAR-T cell therapies have shown strong promise, such as systemic lupus erythematosus.