Visualizing gene expression in single cells is one of the newest and fastest-moving areas in cell biology. The field has exploded from the single-cell emulsion droplet barcoding technique published in 2015 to new techniques to profile the heterogeneity of tumors, whole organs, and even whole organisms.

“Malignant features have natural length scales that span an enormous range, from nanometers to meters. It’s important to preserve the spatial information in cancer because so many processes are incredibly difficult to recapitulate in vitro,” said H. Courtney Hodges, PhD, Baylor College of Medicine. “If we want to understand the processes that contribute to malignancy and the plasticity of tumor cell biology, we need to be able to explore the genomic dependencies of these processes in a way that preserves their spatial contexts.”

Hodges was chair of the Educational Session on Saturday, April 9, that provided an update on the latest techniques that enable dissection and characterization of tumors with unprecedented precision. The session, Single Cell Biology in Scale and Spatial Genomics, can be viewed on the virtual platform by registered meeting participants through July 13, 2022.

The problem, Hodges said, is that tumors are heterogeneous and highly adaptable ecosystems with a natural hierarchy of cell identities and states that are specific to certain cell types and tissue architectures. Techniques that rely on sum or average scores can lead to inappropriate conclusions, which led to single cell RNA sequencing (scRNA-seq).

Single-cell tools work with dissociated cells. FACS (fluorescence-activated cell sorting) and MACS (magnetic-activated cell sorting) allow sorting of cells and identification of subpopulations. CyTOF (cytometry by time-of-flight mass spectrometry or mass cytometry) provides high-dimensional expression at the single-cell level.

Other techniques use tissue sections. Spatially barcoded mRNA capture is suitable for tissue-level features and single-cell resolution is still being optimized. MERFISH (multiplexed error-robust fluorescence in situ hybridization) focuses on one or two cells and has a very high sensitivity. This technique is well suited for subcellular localization of transcripts. Other methods are used for mapping proteins: imaging mass spectrometry visualizes high-dimensional cell states and is suitable for analyzing small, highly abundant proteins such as histons; multiplexed immunofluorescence (CyCIF) allows for deep exploration of a sample by probing it with a large set of antibodies.

“There is no singular best technique,” Hodges said. “There are tradeoffs between cost, specimen size, biased versus unbiased methods, the need for capital equipment or specialized expertise, and more. There are new approaches to visualize spatial architecture of more than RNA, with the challenge to integrate these techniques with expertise in animal models and human specimens.”

There is also the challenge of moving beyond genomics to epigenomics.

“Cancer is not only a disease of the genome but a disease of the epigenome—the molecules and mechanisms that can initiate and perpetuate alternative gene activity states in the context of the same DNA sequence,” said Bing Ren, PhD, University of California, San Diego.

Epigenomic information is typically determined by methylation, chromatin accessibility, histone modifications, higher order chromatin organization, and non-coding RNAs that modulate and regulate DNA expression. These epigenetic writers, readers, and erasers are frequently mutated in cancers and are increasingly being explored as therapeutic targets.

Improved sequencing technologies are driving epigenomic studies to identify and visualize promoters and enhancers affecting both expressed and repressed genes. Chromatin access, DNA methylation patterns, variable histone expression, chromatin loops, and topologically associated domains (TADs) are all marks of epigenetic regulation at work, Ren explained.

ATAC-seq (assay for transposase-accessible chromatic sequencing) uses barcoding to map open chromatin at the single-cell level. SnapATAC (single nucleus analysis pipeline for ATAC-seq) uses a computational approach to visualize gene clustering and predict interactions to better identify transcription regulators.

Ren described a single-cell atlas of chromatin accessibility from human adult and fetal tissues that allowed the researchers to identify 1.2 million candidate regulatory elements across 222 cell types, Ren noted. Diving deeper into brain tissue revealed 123 distinct brain cell types, a useful reference for better understanding tumorigenesis.

Single-cells epigenome technologies are useful to gain a deeper understanding of the molecular basis of cancer. Glioblastoma, for example, includes tumor cells in multiple cell states with frequent and continuous transitions between states. This epigenetic plasticity and heterogeneity may explain the limited efficacy of treatments targeting specific signaling pathways. Ren and collaborators applied single-cell technologies to study the transcriptional and epigenetic changes that underlie the plasticity in glioblastoma. Open chromatin analysis revealed a common NF1 motif. Knocking down NF1A/B in xenograft mouse models of glioblastoma reduces tumor growth and increases survival, Ren said.

Tumor topology also affects cancer biology. Just as plants and animals express different subtypes, or ecotypes in response to different physical environments, tumors express different ecotypes based on their environment. The tumor microenvironment contains distinct cellular neighborhoods that are linked to prognosis and outcome.

“Tumors are not monolithic collections of cells, but complex ecosystems comprised of cancer cells and a variety of immune cell subsets,” said Aaron M. Newman, PhD, Stanford University School of Medicine. “The spatial arrangement of cells can be important to tumor development and clinical outcomes.”

Multiplexed imaging, bulk transcriptomics, scRNA-seq, and spatial transcriptomics all offer different advantages and disadvantages in visualizing these tumor ecotypes, Newman explained. A new digital approach, CIBERSORTx, can combine multiple approaches to maximize the advantages of each to visualize clinically distinct clusters of co-associated cell subsets, or ecotypes.

Applying this computational cytology approach to 6,475 bulk carcinoma tumors and adjacent normal tissues enabled identifying 69 distinct cell states that sort into 10 clinically relevant carcinoma ecotypes (CEs) that have strong prognostic associations, Newman said. Each CE has a unique spatial structure. Seven ECs are associated with shorter survival and three are associated with longer survival, he said.

“The CytoSPACE framework allows us to spatially resolve tumors at the single-cell to whole-tissue level for potential diagnostic and prognostic use,” Newman said. “The approach has been successful in nearly all cancer types analyzed. A preprint is in the works.”

[sub-post-content]