Methods for characterizing mechanobiology of the tumor microenvironment landscape

Cells within the tissue microenvironment sense, process and respond to mechanical cues from their environment. This mechanosensing is essential for tissue homeostasis, and its dysregulation drives tumor development, growth, and metastasis. Therefore, characterizing its mechanobiology – the interplay between

mechanical cell-microenvironment interaction and cell signaling – can play an important role in helping (a) cancer biologists gain mechanistic insights for developing improved drug treatments for cancer; and (b) provide pathologists and clinicians the ability to help develop improved markers for predicting risk of cancer development

and relapse, its metastatic potential, and response to therapy in individual patients. However, despite its importance in both preclinical and clinical settings no computational methods currently exist to readily incorporate mechanobiological properties of tissue microenvironments in cancer research and its translation. To overcome

this gap in our knowledge, we aim to develop a novel algorithm that combines high-resolution Hematoxylin and Eosin (H&E) digital pathology (DP) imaging and highly multiplexed immunofluorescence (HxIF) microscopy with physical optics-based principles of light-matter interaction to characterize mechanobiology properties of three-

dimensional tumor microenvironments (TME) across whole slide tissue sections at sub-cellular resolution. In our published work we have shown that light-matter interaction can capture structural alterations with nanoscale sensitivity within the specimen. Here, we hypothesize that computationally implementing this principle on tumor

microenvironments imaged using DP and HxIF microscopy can quantitatively capture intrinsic mechanobiological properties of the cellular and acellular components of the microenvironment that go beyond image analysis and machine learning based feature extraction. Combining this computational imaging method

with information theoretic principles we also aim to provide researchers with the ability to quantify the major cellular interactions driving these mechanical properties. We note that in many scenarios – for example, in pathology – it is not possible to access tissue samples at a temporal resolution that faithfully captures the

complexity of an evolving tumor. Therefore, the ability of our method to capture interaction from a single tissue microenvironment will be very valuable for pathologists to better predict future outcomes. It will also be very useful for cancer and developmental biologists using animal models and organoids to study mechanisms driving

cell fate decisions. We aim to develop our algorithm with successful completion of two aims. (1) Computational imaging algorithm for mechanobiological characterization of 3D microenvironment. (2) Information theory-based algorithm for characterizing directed mechanobiological interactions. By providing mechanobiological

characterization of TMEs for the first time, our methods will have multifactorial impact on understanding the spatial systems biology of TMEs from development of new mechanobiology based diagnostic, prognostic, and surveillance biomarkers to new targets for overcoming mechanobiology mediated patient resistance to therapy.