Abstract
Recently developed nanobubble ultrasound contrast agents are a promising tool for imaging and drug delivery in tumors. To better understand their unusual kinetics, we implemented a novel pixel clustering analysis, which provides unique information by accounting for spatial heterogeneity. By combining ultrasound results with proteomics of the imaged tumors, we show that this analysis is highly predictive of protein expression and that specific types of nanobubble time-intensity curve are associated with upregulation of different metabolic pathways. We applied this method to study the effects of two proteins, EphB4 and ephrinB2, which control tumor angiogenesis through bidirectional juxtacrine signaling, in mouse models of head and neck cancer. We show that ephrinB2 expression by endothelial cells and EphB4 expression by cancer cells have similar effects on tumor vasculature, despite sometimes opposite effects on tumor growth. This implicates a cancer-cell-intrinsic effect of EphB4 forward signaling and not angiogenesis in EphB4's action as a tumor suppressor.