Denoising sparse microbial signals from single-cell sequencing of mammalian host tissues.

Existing genomic sequencing data can be used to study host-microbiome ecosystems, however distinguishing signals originating from truly present microbes versus contaminating species and artifacts is a substantial and often prohibitive challenge. Here we show that emerging sequencing technologies definitely capture reads from present microbes. We developed SAHMI, a computational resource to identify truly present microbial nucleic acids and filter contaminants and spurious false-positive taxonomic assignments from standard transcriptomic sequencing of mammalian tissues. In benchmark studies, SAHMI correctly identifies known microbial infections present in diverse tissues, and we validate SAHMI's enrichment for correctly classified, truly present species using multiple orthogonal computational experiments. The application of SAHMI to single-cell and spatial genomic data thus enables co-detection of somatic cells and microorganisms and joint analysis of host-microbiome ecosystems.

Hierarchical and automated cell-type annotation and inference of cancer cell of origin with Census.

MOTIVATION: Cell-type annotation is a time-consuming yet critical first step in the analysis of single-cell RNA-seq data, especially when multiple similar cell subtypes with overlapping marker genes are present. Existing automated annotation methods have a number of limitations, including requiring large reference datasets, high computation time, shallow annotation resolution, and difficulty in identifying cancer cells or their most likely cell of origin. RESULTS: We developed Census, a biologically intuitive and fully automated cell-type identification method for single-cell RNA-seq data that can deeply annotate normal cells in mammalian tissues and identify malignant cells and their likely cell of origin. Motivated by the inherently stratified developmental programs of cellular differentiation, Census infers hierarchical cell-type relationships and uses gradient-boosted \decision trees that capitalize on nodal cell-type relationships to achieve high prediction speed and accuracy. When benchmarked on 44 atlas-scale normal and cancer, human and mouse tissues, Census significantly outperforms state-of-the-art methods across multiple metrics and naturally predicts the cell-of-origin of different cancers. Census is pretrained on the Tabula Sapiens to classify 175 cell-types from 24 organs; however, users can seamlessly train their own models for customized applications. AVAILABILITY AND IMPLEMENTATION: Census is available at Zenodo https://zenodo.org/records/7017103 and on our Github https://github.com/sjdlabgroup/Census.

Tumor microbiome links cellular programs and immunity in pancreatic cancer.

Microorganisms are detected in multiple cancer types, including in putatively sterile organs, but the contexts in which they influence oncogenesis or anti-tumor responses in humans remain unclear. We recently developed single-cell analysis of host-microbiome interactions (SAHMI), a computational pipeline to recover and denoise microbial signals from single-cell sequencing of host tissues. Here we use SAHMI to interrogate tumor-microbiome interactions in two human pancreatic cancer cohorts. We identify somatic-cell-associated bacteria in a subset of tumors and their near absence in nonmalignant tissues. These bacteria predominantly pair with tumor cells, and their presence is associated with cell-type-specific gene expression and pathway activities, including cell motility and immune signaling. Modeling results indicate that tumor-infiltrating lymphocytes closely resemble T cells from infected tissue. Finally, using multiple independent datasets, a signature of cell-associated bacteria predicts clinical prognosis. Tumor-microbiome crosstalk may modulate tumorigenesis in pancreatic cancer with implications for clinical management.

Inference on spatial heterogeneity in tumor microenvironment using spatial transcriptomics data.

In the tumor microenvironment (TME), functional interactions among tumor, immune, and stromal cells and the extracellular matrix play key roles in tumor progression, invasion, immune modulation, and response to treatment. Intratumor heterogeneity is ubiquitous not only at the genetic and transcriptomic levels but also in the composition and characteristics of TME. However, quantitative inference on spatial heterogeneity in the TME is still limited. Here, we propose a framework to use network graph-based spatial statistical models on spatially annotated molecular data to gain insights into modularity and spatial heterogeneity in the TME. Applying the framework to spatial transcriptomics data from pancreatic ductal adenocarcinoma samples, we observed significant global and local spatially correlated patterns in the abundance score of tumor cells; in contrast, immune cell types showed dispersed patterns in the TME. Hypoxia, EMT, and inflammation signatures contributed to intra-tumor spatial variations. Spatial patterns in cell type abundance and pathway signatures in the TME potentially impact tumor growth dynamics and cancer hallmarks. Tumor biopsies are integral to the diagnosis and clinical management of cancer patients; our data suggest that owing to intra-tumor non-genetic spatial heterogeneity, individual biopsies may underappreciate the extent of clinically relevant, functional variations across geographic regions within tumors.

Reconstructing physical cell interaction networks from single-cell data using Neighbor-seq.

Cell-cell interactions are the fundamental building blocks of tissue organization and multicellular life. We developed Neighbor-seq, a method to identify and annotate the architecture of direct cell-cell interactions and relevant ligand-receptor signaling from the undissociated cell fractions in massively parallel single cell sequencing data. Neighbor-seq accurately identifies microanatomical features of diverse tissue types such as the small intestinal epithelium, terminal respiratory tract, and splenic white pulp. It also captures the differing topologies of cancer-immune-stromal cell communications in pancreatic and skin tumors, which are consistent with the patterns observed in spatial transcriptomic data. Neighbor-seq is fast and scalable. It draws inferences from routine single-cell data and does not require prior knowledge about sample cell-types or multiplets. Neighbor-seq provides a framework to study the organ-level cellular interactome in health and disease, bridging the gap between single-cell and spatial transcriptomics.

Multiplexed Profiling of Single Extracellular Vesicles.

Extracellular vesicles (EV) are a family of cell-originating, membrane-enveloped nanoparticles with diverse biological function, diagnostic potential, and therapeutic applications. While EV can be abundant in circulation, their small size (∼4 order of magnitude smaller than cells) has necessitated bulk analyses, making many more nuanced biological explorations, cell of origin questions, or heterogeneity investigations impossible. Here we describe a single EV analysis (SEA) technique which is simple, sensitive, multiplexable, and practical. We profiled glioblastoma EV and discovered surprising variations in putative pan-EV as well as tumor cell markers on EV. These analyses shed light on the heterogeneous biomarker profiles of EV. The SEA technology has the potential to address fundamental questions in vesicle biology and clinical applications.

Analysis of transcription factor network underlying 3T3-L1 adipocyte differentiation.

Lipid accumulation in adipocytes reflects a balance between enzymatic pathways leading to the formation and breakdown of esterified lipids, primarily triglycerides. This balance is extremely important, as both high and low lipid levels in adipocytes can have deleterious consequences. The enzymes responsible for lipid synthesis and breakdown (lipogenesis and lipolysis, respectively) are regulated through the coordinated actions of several transcription factors (TFs). In this study, we examined the dynamics of several key transcription factors (TFs) - PPARγ, C/EBPß, CREB, NFAT, FoxO1, and SREBP-1c - during adipogenic differentiation (week 1) and ensuing lipid accumulation. The activation profiles of these TFs at different times following induction of adipogenic differentiation were quantified using 3T3-L1 reporter cell lines constructed to secrete the Gaussia luciferase enzyme upon binding of a TF to its DNA binding element. The dynamics of the TFs was also modeled using a combination of logical gates and ordinary differential equations, where the logical gates were used to explore different combinations of activating inputs for PPARγ, C/EBPß, and SREBP-1c. Comparisons of the experimental profiles and model simulations suggest that SREBP-1c could be independently activated by either insulin or PPARγ, whereas PPARγ activation required both C/EBPß as well as a putative ligand. Parameter estimation and sensitivity analysis indicate that feedback activation of SREBP-1c by PPARγ is negligible in comparison to activation of SREBP-1c by insulin. On the other hand, the production of an activating ligand could quantitatively contribute to a sustained elevation in PPARγ activity.