| | |  | | NEWSLETTER ::: NO. 22 ::: NOV 2021 | | THE VIRTUAL KIDNEY | | The idea of virtual organs —digital representations of organ physiology— promises nearly boundless research opportunities by means of virtual experiments. Imagine investigating kidney function without worrying about ethics committees, long breeding times, or compliance of test subjects. Imagine studying the effect of gene mutations by the stroke of a button. How far away are we from replacing in vivo models by in silico representations? And how will such models change the role of kidney researchers? |  | | Image by Diego Rossinelli and Willy Kuo, provided by Vartan Kurtcuoglu. | The field of computational physiology has its beginnings in the 1950s, with the first in silico models addressing macroscopic physical phenomena. The cardiovascular system lent itself particularly well to such modeling, leading to a head start compared to other systems that is still evident today. Therefore, it is not surprising that the most famous early model including elements of kidney function was established by Arthur Guyton and co-workers to investigate the regulation of blood pressure and cardiac output. The model, built on a combination of basic physical principles and empirical data, was instrumental in demonstrating the role of the kidney in the long-term control of blood pressure.
In theory, fundamental principles of physics could be used to model kidney function by starting at the atomic level and calculating the interactions of atoms in a process known as atomistic molecular dynamics simulation. In practice, while even a smartphone is powerful enough to solve the corresponding equations for the short-term interaction of a few atoms, not even all of the computers on our planet combined could calculate the entirety of molecular interactions in a renal cell using this approach, let alone in an organ. Therefore, molecular dynamics simulations are used for small systems, for example to study the functional movement of membrane transporters.
As long as we lack the computer power to expand atomistic modeling to the organ level, we cannot produce a one-to-one virtual representation of the kidney. This means that, in the foreseeable future, there will not be one universal virtual kidney, but a number of them, each made for a specific set of research questions. This will be familiar to those using transgenic animals in their research: While it would be ideal to have a universal mouse model in which the expression of any gene can be controlled at will for any application, in reality, it is the research question that determines which mouse model is required.
The same goes for virtual kidney models. One family of models derives its principles and mathematical formulations from empirical data, to study, for example, tubular transport. Empirically derived kinetics of individual transporters can be linked to capture transporter interactions and net transport at the cellular level. Typically, the renal anatomy is not considered explicitly in a geometrically accurate manner, but rather simplified to functional compartments such as intracellular spaces, interstitium, or tubular and vascular lumina. This approach has been applied to study solute transport along nephrons, tubuloglomerular feedback, as well as oxygen transport and consumption. An attractive feature of these virtual systems is the possibility to modulate the activity of each transporter individually allowing for highly specific virtual knock-out experiments. For example, Layton, Vallon and Edwards developed a model of solute transport, accounting, among other factors, for glucose and sodium cotransport along a superficial nephron of a rat kidney to study the effects of diabetes and SGLT2 inhibition on sodium reabsorption and renal oxygen consumption.
Another family of models builds on fundamental laws of physics at the macroscopic level. A typical example are simulations of fluid dynamics, be it of renal blood flow or of the flow of primary urine through nephrons. When the rheology of the fluid, the geometry of its conduit and the driving forces are known, equations representing mass and momentum conservation can be used to calculate velocity and pressure at any given location in an artery or tubule. This is relevant, for example, for calculating the shear forces that mechanosensors of the proximal tubular epithelium are exposed to. When also the properties of solutes of interest are available, their distribution in the fluid can be predicted by additionally considering the advection-diffusion equation. One application is in the interpretation of tracer studies: Do the physical mechanisms taken into account in the model suffice to explain the experimentally observed tracer distribution, or is there a clear discrepancy, suggesting that other mechanisms, such as a yet unknown active transporter, must be at play?
Members of the two families of models can also be merged to expand their scopes of application. For instance, to investigate the fate of oxygen in the kidney, the transport of oxygen with blood, its dissociation from hemoglobin, diffusion through tissue, as well as the chemical reactions involved in oxygen consumption must be considered. While blood flow and diffusion can be captured by physical models, the chemical steps are described empirically, including with the well-known oxygen–hemoglobin dissociation curve. We have used such a hybrid model to assess the plausibility of pre-glomerular arterial-tovenous oxygen shunting.
It should be evident now that the first question asked in the introduction needs to be rephrased: How far have we come in using in vivo and in silico models synergistically? Given the astounding progress in computer technology over the last decades, the expectations may be high. However, unless we are considering the ultimate virtual kidney built on atomistic modeling, the main bottleneck is not computer power, but the schism between experimental and computational kidney research. This answers the second question: Computational models will not replace experimental kidney research anytime soon, but those scientists who can design and integrate experimental studies and computational models will be best positioned to make groundbreaking contribution
to renal research. | | Vartan Kurtcuoglu and Dianne de Zélicourt are members of the NCCR Kidney.CH.
Vartan Kurtcuoglu is an Associate Professor of Computational and Experimental Physiology at the Institute of Physiology of the University of Zurich. His research focuses on the study of fluid flow and associated solute transport processes in the kidney, the brain and the cardiovascular system.
Diane de Zélicourt is a junior group leader within the Interface Group at the Institute of Physiology of the University of Zurich, with a long standing interest in computational physiology and fluid mechanics applied to cardiac and renal pathologies. |
|
| | | | Kidney - Control of Homeostasis | | is a Swiss research initiative, headquartered at University of Zurich, which brings together leading specialists in experimental and clinical nephrology and physiology from the universities of Bern, Fribourg, Geneva, Lausanne, and Zurich, and corresponding university hospitals. |
| | |
| |
|