If we want to skip animal testing in future, good in vitro models alone are not enough. We need the power of computer models to convert laboratory data into clinical prediction outcomes. Collaborators from RMU principal investigator Prof. Masereeuw have set a nice example.
A matter of complexity
The use of animals for scientific research is a constant matter of debate. Unfortunately, there is often no way around it; we depend on the complexity of the body for accurate results. However, an increasing number of scientists are working hard on the development of alternatives, so called in vitro models, with increasing complexity. Instead of using a flat layer of human cells in a petri dish, cells are cultured on bioengineered materials, in co-culture with other cell types, under fluid flow, or in a three-dimensional environment. By mimicking the environment as found in the body, cellular function in vitro can often be improved.
An example is the bioengineered kidney tubule, or kidney-on-a-chip, developed in the lab of Prof. Roos Masereeuw at Utrecht University. By culturing kidney cells on bio-functionalized hollow fibers, the cells can grow into a tube structure – just like in the body, where they function as drainage system for waste products. They can also be exposed to fluid flow – just like in the body, where blood and urine are passing by the cells.
Besides mimicry, the tubule set-up also brings another advantage: the researchers can study how the cells transport waste products and drugs from one compartment to the other – just like in the body from blood to urine. For instance, the excretion of indoxyl sulfate can be studied; a waste product that contributes to the progression of kidney disease and cardiovascular complications when blood concentrations rise too high.
However, extravagant in vitro models alone are not sufficient to translate the improved laboratory results into the clinics. For the next step, we need the help of computers. As biology and informatics are both advancing very fast, they have the potential to form a bridge from petri dish to patient bed; a bridge that might by-pass animal testing in the future. How? Computer simulations, also called in silico models, can be based on in vivo information and fed with in vitro information. The data obtained can then be converted into risk evaluations and clinical outcome predictions.
Kidney cells and computer calculations
Recently, researchers at the University of Manchester have used the bioengineered kidney tubule to take this next step. Van der Made and colleagues developed an in silico model and collected experimental in vitro data from the kidney-on-a-chip to quantitatively predict the excretion of indoxyl sulfate in vivo. However, their goal was not only to predict indoxyl sulfate clearance. At the same time, they used the model to validate the hypothesis of albumin-facilitated uptake.
Albumin as helping hand
In our blood, indoxyl sulfate is greatly bound to albumin, a very abundant plasma protein. In the liver, it has been shown that drugs are taken up more easily when bound to albumin, and similar mechanisms have been proposed for the kidney. Thus, the researchers from Manchester hypothesized that indoxyl sulfate is more excreted when bound to albumin. Using the in vitro model, they proved that the metabolite is indeed taken up to a greater extent in presence of the protein. And with their in silico model, they could accurately predict to what extent.
The fact that the in vitro / in silico predictions were in line with what is happening in real life suggests that the model works well. Moreover, the researchers could validate their hypothesis and concluded that future clearance studies should include albumin (especially when the compound of interest is highly protein-bound).
Revealing knowledge gaps
Since albumin slightly changes shape in kidney patients, van der Made and colleagues also investigated the impact of albumin modification on indoxyl sulfate clearance. They found that albumin modifications lead to lower binding and hence to less clearance – which is indeed observed in kidney patients! This is the first time that quantitative translation of in vitro data has been applied in a clinically relevant setting, which has led to the identification of a knowledge gap for accurate prediction.
This study nicely exemplifies how in vitro data can be brought closer to the clinic, with the help of in silico modeling. And at the same time, this can bring us one step closer to a world without animal experiments.