Modeling the Pancreatic Cancer Microenvironment in Search of Control Targets

Abstract

Pancreatic Ductal Adenocarcinoma is among the leading causes of cancer related deaths globally due to its extreme difficulty to detect and treat. Recently, research focus has shifted to analyzing the microenvironment of pancreatic cancer to better understand its key molecular mechanisms. This microenvironment can be represented with a multi-scale model consisting of pancreatic cancer cells (PCCs), pancreatic stellate cells (PSCs), as well as cytokines and growth factors which are responsible for intercellular communication between the PCCs and PSCs. We have built a stochastic Boolean network (BN) model, validated by literature and clinical data, in which we probed for intervention strategies that force this gene regulatory network (GRN) from a diseased state to a healthy state. To do so, we implemented methods from phenotype control theory to determine a procedure for regulating specific genes within the microenvironment. We identify target genes and molecules such that the application of their control drives the GRN to the desired state by suppression (or expression) and disruption of specific signaling pathways that will eventually lead to the eradication of the cancer cells. After applying well studied control methods such as stable motifs, feedback vertex set, and computational algebra, we discovered that each produces a different set of control targets that are not necessarily minimal nor unique. Yet, we were able to gain more insight about the performance of each process and the overlap of targets discovered. Nearly every control set contains cytokines, KRas, and HER2/neu which suggests they are key players in the system’s dynamics. To that end, this model can be used to produce further insight into the complex biological system of pancreatic cancer with hopes of finding new potential targets.

Uncovering potential interventions for pancreatic cancer patients via mathematical modeling

“…When such a situation arises, there are reduction techniques available to reduce the computational load, while maintaining important system dynamics. Notably, methods of stable motifs and feedback vertex set (FVS) require apoptotic attractors to determine control targets [27]. However, we were able to avoid magnitude issues because inducing KRAS, TP53/KRAS, TP53/SMAD4/KRAS, and TP53/CDKN2A/KRAS mutations created strictly diseased attractors (autophagy and apoptosis).…”

“…In this work, we used a previously developed multicellular model of pancreatic cancer based on a Boolean network approximation (Fig 1), and we used control strategies that direct the system from a diseased state to a healthy state by suppression (or expression) and disruption of specific signaling pathways. The models consists of pancreatic cancer cells, pancreatic stellate cells, cytokine molecules of diffusing in the local microenvironment, and internal gene May 20, 2022 2/24 regulations for both cell types [27]. Although the model includes several components of the tumor microenvironment, its Boolean nature allows the application of recent control techniques.…”

“…Although the model includes several components of the tumor microenvironment, its Boolean nature allows the application of recent control techniques. Further, we used techniques from phenotype control theory [24,27,32,36], which is primarily concerned with identifying key markers of the system that aid in understanding the various functions of cells and their molecular mechanisms. In a biological sense, phenotypes represent the observable traits of an organism.…”

“…This means that a standard doctor’s exam will likely not reveal a tumor, in conjunction with an absence of detecting and imaging techniques for early stage tumors. Treatments are often unsuccessful because diagnoses are frequently given at late stages of the disease, leaving little time to successfully eradicate and reverse its impact [27, 34].…”

“…Black barbed arrows indicate signal expression, while red bar arrows indicate suppression. Grey nodes located in the PCC indicate prevalent mutant genes [27].…”