Spatio-Phenotypic Tumor Profiling via Single-Cell 3-D Image Analysis

Although single-cell techniques have enabled in-depth analyses of diverse cell populations within human tumors, the majority of approaches involve tissue dissociation and loss of spatial context. Recent advances in three-dimensional (3-D) image processing have enabled the study of tumor cells in their native environment, but challenges remain in attaining large-scale data at single-cell resolution. To address these difficulties, van Ineveld, Kleinnijenhuis, and colleagues developed a technique involving multispectral, large-scale, single cell–resolution 3-D (mLSR-3D) tissue imaging. Accommodating staining with up to eight fluorophores, mLSR-3D imaging was performed on a pediatric kidney tumor as well as human fetal kidney tissue (HFK) using antibodies for five markers of early nephrogenic structures in addition to stains and antibodies that indicated nuclei, actin, and cycling cells. Compared with existing methods, mLSR-3D decreased acquisition time, photobleaching, and the requirement for extensive data-storage capacity. The 3-D images were processed using a pipeline called segmentation analysis by parallelization of 3-D datasets (STAPL-3D), which involved single-cell compartment segmentation and compartment-specific feature extraction, and the analysis was further enhanced by the use of deep learning–based segmentation methods. In HFK tissue, cells were assigned to subpopulations based on molecular markers, morphologic features, and location. Compared with HFK as a reference tissue, the pediatric Wilms tumor sample was highly spatially disorganized, although the tumor sample also contained nephrogenic-like structures, as expected given the occurrence of aberrant fetal nephrogenesis in Wilms tumors. Pseudotemporal ordering of nephrogenic cells ranging from early progenitor to late populations revealed that the Wilms tumor harbored a larger subpopulation of renal vesicle and S-shaped body cells, likely explained by increased cycling in addition to increased differentiation from cap mesenchyme-like cells. This analysis also uncovered previously unreported populations in Wilms tumors based on spatial embedding. In summary, this work describes a novel method that enables comprehensive in situ characterization of complex tissues and provides a tool to spatio-phenotypically profile diverse types of human tumors.

van Ineveld RL, Kleinnijenhuis M, Alieva M, de Blank S, Barrera Roman M, van Vliet EJ, et al. Revealing the spatio-phenotypic patterning of cells in healthy and tumor tissues with mLSR-3D and STAPL-3D. Nat Biotechnol 2021 Jun 3 [Epub ahead of print].

Note: Research Watch is written by Cancer Discovery editorial staff. Readers are encouraged to consult the original articles for full details. For more Research Watch, visit Cancer Discovery online at http://cancerdiscovery.aacrjournals.org/CDNews.