(130e) Multiscale Computational Modeling of Renal Intercellular Cross-Talk at the Onset of Diabetic Kidney Disease

Diabetic kidney disease (DKD) is one of the most severe complications of diabetes and is the primary cause for kidney failure. In the early stages of DKD, kidney tissue damage is primarily focused in glomeruli, which are bundles of capillaries within the filtration units (nephrons) of the kidney where blood is filtered to produce urine. Significant damage to glomeruli occurs before the leakage of proteins through glomeruli into urine. This damage is detectable by non-invasive measurements in a condition called proteinuria, which is a clinical marker of DKD. Continued damage ultimately leads to kidney failure. Some of these renal tissues including podocyte cells are terminally differentiated and cannot repair themselves. Thus, it is critical to slow the rate of progression of DKD before irreversible injury and subsequent nephron loss occurs.

Hyperglycemia-induced metabolic dysregulation is known to initiate DKD through complex biochemical signaling pathways. The connections between the various molecular mechanisms that lead to glomerular injury in DKD are not completely understood. A glomerulus consists of a bundle of capillaries and a scaffold of cells and extracellular domains that hold the bundle together. The scaffold includes the mesangium, podocytes, and endothelial cells. The mesangium forms the interior of the capillary cluster of a glomerulus. Podocyte cells surround the glomerular capillaries with an interlocking network of long projections or foot processes with small slits between them for filtration. The endothelial cells line the capillaries and are covered by a cell-surface layer called the glycocalyx. Additionally, the glomerular basement membrane (GBM) is a specialized extracellular matrix that exists between the endothelial cells and podocytes. Experimental studies have shown different biochemical pathways are triggered in hyperglycemic or diabetic conditions that alter these cells leading to glomerular injury mainly in four ways: podocyte cell loss, mesangial expansion, GBM thickening, and endothelial glycocalyx thinning. The roles in glomerular damage of cross-talk between these cell types have been suggested in the literature; however, there is as yet no way to predict the systemic impacts of the cross-talk throughout entire glomeruli in DKD. The objective of this work is to create a computational model to simulate the biochemical and physical interactions between glucose, cell signaling pathways, and various cell types within glomeruli at the onset of DKD in advance of detection of proteinuria. The hypothesis is that simultaneous concerted interactions between glucose-stimulated signaling pathways in multiple cell types within the glomeruli and the cross-talk between cell types are needed in order for hyperglycemia to progress to diabetic proteinuria. Such a computational model can provide insight into understanding complex networks of the physiological and biochemical processes that may lead to emergent behavior in DKD.

Podocytes express a local renin-angiotensin system (RAS) that is altered in diabetes. RAS is a network of biochemical reactions involving multiple enzymes and hormones. Angiotensin II (ANG II) is one of the crucial hormones in the RAS and is responsible for regulating blood pressure. Studies have shown that ANG II is upregulated in diabetic conditions and triggers irreversible glomerular injury by causing mesangium expansion, GBM thickening, shrinking of the glycocalyx, podocyte apoptosis and detachment. This deterioration of the glomerular filtration barrier leads to leakage of proteins into the urine and eventually, kidney failure. Angiotensin converting enzyme (ACE) is responsible for conversion of ANG I to ANG II. ACE inhibitors are widely used pharmaceuticals that inhibit the production of ANG II. However, these drugs are not well characterized for use in CKD. The progression of CKD could be slowed by controlling the ANG II levels to prevent irreversible podocyte loss. We have developed an ODE-based PK/PD model of glucose-stimulated RAS dynamics in podocytes using MATLAB. The model predicts the effects of different dosages of pharmaceutical drugs on the concentrations of ANG II in local podocyte RAS. Glucose dependency is added to the model through enzymatic parameters that can help predict ANG II for different glucose levels. The model is parametrized for two pharmaceutical drugs in the cases of normal and impaired renal function. The model can also take patient specific glucose dynamics data as a model input to predict ANG II levels and required drug dosage to suppress elevated ANG II levels inside of podocytes.

Another scale of the computational model is developed for the convection and diffusion of chemical species through the porous mesangium matrix. This model can be applied to hyperglycemic conditions to study the effects of glucose-stimulated elevated ANG II levels and transport rates on the different pathophysiological processes including GBM thickening and mesangial matrix density.

The biochemical mechanisms and their effects in the podocytes and mesangium can be incorporated together in a hybrid continuous and discrete agent-based model to connect the biochemical models to the physical effects on the glomerular cell types. This model is developed to produce a anatomically-scaled geometry of the cross section of the glomerular filtration barrier composed of the podocytes, mesangium, capillarie,s and GBM in Compucell3D to simulate the interactions among the kidney cells. Compucell3D uses a cellular Potts model to model the tissue environment and computes an energy associated with cell properties. User-created Python scripts are also used to augment the behavior of the cells and add functionality to the simulation. While Compucell3D does have an embedded partial differential equation (PDE) solver, the method is forward Euler explicit method, which is not suited for simulating convection. The present work has tailored Compucell3D to allow for the arbitrary execution of any Python code, including the package FiPy, which is an extremely flexible PDE solver library. In our model, FiPy will be used to solve the diffusion and convection of biochemical reaction networks in different cell types (podocytes and mesangium) and predict the overall effect of hyperglycemia-stimulated cellular cross-talk on glomerular injury. The agent-based part of the model includes features related to cell migration, growth, death, and adhesion.

The overall objective of this work is to connect these models focused individually on podocytes and on the mesangium to build a multiscale computational model that can improve the understanding of the mechanisms governing physiological effects of various chemicals, e.g., metabolites (glucose), hormones (ANG II), and pharmaceutical drugs, on kidney damage as a complication of diabetes. By detecting DKD at the onset through conditions identified by the model, treatments could be administered to slow the progression before significant damage occurs.