Human Digital Twin: Automated Cell Type Distance Computation and 3D Atlas Construction in Multiplexed Skin Biopsies

Abstract

AbstractMapping the human body at single cell resolution in three-dimensions (3D) is an important step toward a “digital twin” model that captures important structure and dynamics of cell-cell interactions. Current 3D imaging methods suffer from low resolution and are limited in their ability to distinguish cell types and their spatial relationships. We present a novel 3D workflow: MATRICS-A (Multiplexed Image Three-D Reconstruction and Integrated Cell Spatial - Analysis) that generates a 3D map of cells from multiplexed images and calculates cell type distance from endothelial cells and other features of interest. We applied this workflow to multiplexed data from sequential skin sections from younger and older donors (n=10; 33-72 years) with biopsies from ten anatomical regions with different sun exposure effects (mild, moderate-marked). Up to 26 sequential sections from each sample underwent multiplexed imaging with 18 biomarkers covering 12 cell types (keratinocytes (granular, spinous, basal), epithelial and myoepithelial cells, fibroblasts, macrophages, T helpers, T killers, T regs, neurons and endothelial cells, markers of DNA damage and repair (p53, DDB2) and cell proliferation (Ki67). Following cell classification, the tissue and classified cells were reconstructed into 3D volumes. A significant inverse correlation between DDB2 positive cells and age was found (corr= -0.78, adj. p=0.047). This suggests reduced capacity for repair in non-cancer older sun-exposed individuals. While absolute immune cell count did not differ by age or sun exposure, the ratio of T Helper/T Killer cells was positively correlated with age (corr=0.82, adj. p=0.048) This is the first such 3D study in skin and paves the way for cataloging more cell types and spatial relationships in aging and disease in skin and other organs.