Cell Reports Methods最新文献

Investigating microglial phagocytosis is essential for understanding the mechanisms underlying brain health and disease. Dysregulation of phagocytosis is implicated in various neurological disorders, necessitating accurate analysis. Leveraging advances in deep learning, this study explores the application of a U-Net-based neural network for image cytometry to enhance the analysis of microglial phagocytosis. Murine microglia were imaged using the Olympus ScanR system, generating a substantial dataset for training a U-Net. The U-Net (AIstain) demonstrated superior performance in cell detection compared to live cell staining and the established segmentation tools SAM2 and Cellpose 3. Additionally, the model's applicability can be extended to other cell types, including leukemia and breast cancer cells, highlighting its versatility. AIstain provides a straightforward approach for the analysis of live cell images and microglial phagocytosis. This method enhances the precision of the results while simultaneously reducing the complexity of the experiment, thus facilitating substantial progress in the domain of neurobiological research.

Single-cell multiomics technologies generate paired measurements of different cellular modalities, such as gene expression and chromatin accessibility. However, multiomics technologies are more expensive than their unimodal counterparts, resulting in smaller and fewer available multiomics datasets. Here, we present scPairing, a deep learning model inspired by contrastive language-image pre-training (CLIP), which embeds different modalities from the same single cells onto a common embedding space. We leverage the common embedding space to generate novel multiomics data following bridge integration, a method that uses an existing multiomics bridge to link unimodal data. Through extensive benchmarking, we show that scPairing constructs an embedding space that fully captures both coarse and fine biological structures. We then use scPairing to generate new multiomics data from retina, immune, and renal cells. Furthermore, we extend scPairing to generate trimodal data. The generated multiomics datasets can facilitate the discovery of novel cross-modality relationships and the validation of existing biological hypotheses.

Mitochondria are central to cardiomyocyte function, and their spatial organization regulates nuclear signaling and gene transcription, holding potential for novel cardioprotective interventions. We developed a transmission electron microscopy platform optimized for resolving mitochondrial subpopulations and nuclear architecture in adult cardiomyocytes. This approach reliably captures longitudinal sections containing the center of the nucleus and perinuclear regions, enabling consistent imaging of subcellular nanostructures, assessment of pharmacological effects within the same organism, and visualization of extracellular vesicles carrying dysfunctional mitochondria. Integrated with an analysis workflow employing machine learning-based segmentation for annotation, the method allows automated quantification of mitochondrial and nuclear architecture and positioning. Using Drp1-deficient mice with impaired mitochondrial fission, we demonstrate this tool's ability to uncover nanoscale remodeling of mitochondria and nuclei under stress. Our platform overcomes challenges in electron microscopy analysis, providing a powerful resource to interrogate mitochondrial-nuclear dynamics in cardiac (patho)physiology. These insights will inform therapeutic targeting of bioenergetic failure.

Calcium fluorescence imaging enables us to investigate how individual neurons of live animals encode sensory input or drive specific behaviors. Extracting and interpreting large-scale neuronal activity from imaging data are crucial steps in harnessing this information. A significant challenge arises from uncorrectable tissue deformation, which disrupts the effectiveness of existing neuron segmentation methods. Here, we propose an open-source software, DynamicNeuronTracker (DyNT), which generates dynamic neuron masks for deforming and/or incompletely registered 3D calcium imaging data using patch-matching iterations. We demonstrate that DyNT accurately tracks densely populated neurons under positional jitters. DyNT also includes automated statistical analyses for interpreting neuronal responses to multiple sequential stimuli. We applied DyNT to analyze the responses of pheromone-sensing neurons in mice to controlled stimulation. We found that four bile acids and four sulfated steroids activated 15 subpopulations of sensory neurons with distinct combinatorial response profiles, revealing a strong bias toward detecting sulfated estrogen and pregnanolone.

To detect precise DNA methylation patterns in long-read DNA sequencing analysis, an efficient target enrichment method is needed. In this study, we established t-nanoEM, a practical method that integrates a hybridization-based capture step into a long-read enzymatic methyl (EM)-seq library for nanopore sequencing. We achieved a high sequencing coverage of up to ×570 at 5 kb N50 in length. We applied this method to the long-read methylation analysis of cancers. Using breast cancer as an example, we demonstrated that the signature changes in DNA methylation occurring in local cell populations could be displayed in a haplotype-aware manner. In lung cancer, the spatial diversity in gene expression as detected by the spatial expression profiling analysis may be associated with changes in DNA methylation.

Rapid, spatially controlled methods are needed to investigate RNA localization in bacterial cells. APEX2 proximity labeling was shown to be adaptable to rapid RNA labeling in eukaryotic cells and, through the fusion of APEX2 to different proteins targeted to diverse subcellular locations, has been useful to identify RNA localization in these cells. Therefore, we adapted APEX2 proximity labeling of RNA to bacterial cells by generating an APEX2 fusion to the ribonuclease (RNase) E gene, which is necessary and sufficient for bacterial ribonucleoprotein (BR)-body formation. APEX2 fusion is minimally perturbative, and RNA can be rapidly labeled on the sub-minute timescale with alkyne-phenol, outpacing the rapid speed of mRNA decay in bacteria. Alkyne-phenol provides flexibility in the overall application with copper-catalyzed click chemistry for downstream processes, such as fluorescent dye azides or biotin-azides for purification. Altogether, APEX2 proximity labeling of RNA provides a useful method for studying RNA localization in bacteria.

Traumatic brain injury (TBI) is the leading environmental risk factor for neurodegenerative diseases, yet its molecular link to chronic neurodegeneration is unclear. While animal models of TBI are commonly used, emerging research suggests that induced pluripotent stem cell (iPSC)-derived brain organoids offer a promising human-specific alternative, particularly for studying processes like cryptic exon splicing. However, widespread use has been limited by methodological variability and the need for expensive and specialized equipment. To address these challenges, we developed a tabletop blast device capable of delivering highly reproducible pressure waves via a gravity-based pressure chamber. We validated the applicability of our approach by assessing the short- and long-term consequences of mechanical stress on brain organoids after pressure wave exposure. Our approach provides a controllable and reproducible method to apply complex pressure cycles on brain organoids, enabling broader accessibility for studying the mechanistic links between TBI and neurodegeneration in a human-relevant context.

Spatial transcriptomics (ST) enables in situ analysis of gene expression patterns and spatial microenvironments. However, current ST technologies are limited by detection sensitivity and gene coverage, posing significant challenges for precise cell type annotation at the single-cell level. To address this, we present stTransfer, a method that integrates reference single-cell RNA sequencing (scRNA-seq) data with ST context using a graph autoencoder and transfer learning. This approach minimizes information transfer loss between scRNA-seq and ST datasets. Benchmark analyses on publicly available spatial transcriptomic datasets demonstrate that stTransfer outperforms existing methods in both accuracy and robustness for cell type annotation. Lastly, we apply stTransfer to annotate neuronal populations in a high-precision Stereo-seq dataset of the zebra finch optic tectum.

Hematoxylin and eosin (H&E) staining has been a standard in clinical histopathology for many decades but lacks molecular detail. Advances in multiplexed spatial proteomics imaging allow cell types and tissues to be annotated by their expression patterns as well as their morphological features. However, these technologies are at present unavailable in most clinical settings. In this work, we present a machine learning framework that leverages histopathology foundation models and paired H&E and spatial proteomic imaging data to enable enhanced cell type annotation on H&E-only datasets. We trained and evaluated our method on kidney datasets with paired H&E and spatial proteomic imaging data and found that models trained using our methods outperform models trained directly on the imaging data. We also show how our framework can be used to study biological differences between two major kidney diseases.