Computer modeling may save up to 90 percent of animal testing in detecting active pharmaceutical ingredients

The dual goal of finding a suitable active pharmaceutical ingredient and saving a significant amount of animal testing in the process is the promise of a computer modeling technique developed by Charité scientists in cooperation with the Zuse Institute Berlin. It is the first system world-wide able to simulate the activation of receptors under conditions specific to a disease.

Every approved drug is preceded by an extended process of finding active pharmaceutical ingredients. Thousands and thousands of molecules are typically screened in the process and the most promising candidates are then being checked in extensive animal-testing after chemical refinement. The outcome of these trials is uncertain. What is certain, however, is that most of the candidates do not reach the stage of clinical studies and that much animal-testing was in vain.

For a number of years, pharmaceutical companies and scientific institutions have also been using computer modeling techniques as an alternative in the initial screening phase. But these so-called in silico methods have the disadvantage that potential molecules are being tested under physiological (i.e. healthy) conditions. Therefore, the prognostic value of whether a certain substance will have an effect on a certain disease is low – making it the foremost reason why no approved drugs have yet come from these approaches. The harmful side effects were too great.

Computer simulation of the disease
This is why scientists at the Charité and the Zuse Institute Berlin follow a different strategy.  Their computer modeling technique, which was developed years ago, is the first technique world-wide that allows for the simulation of conditions specific to a disease. Prof. Dr. Christoph Stein from the Institute for Experimental Anaesthesiology at the Charité Campus Benjamin Franklin explains: “When compared to the commonly used in silico methods, the big difference is that we are in a position to model the interaction of the receptors with a certain substance right in the pathological environment. This puts us closer to the human disease and we can find the appropriate molecules more quickly.” For example, with the aid of artificial intelligence the scientists can compute what ligands bind to a certain receptor, whether the target receptor is thus activated, how the forces of the individual chemical residues interact and what concentration a dose should have for the substance to be effective but not toxic.

The novel Berlin technique is generally applicable to all diseases where receptors play a role which are coupled to G proteins (GPCR). These are, for example, heart attacks, cancer, arthritis, high blood pressure, epilepsy, Alzheimer’s and Parkinson’s disease or inflammations of any kind – it is a long list.  Roughly speaking, every third medical drug is targeted towards this receptor group. Opiates also dock on to such a GPC receptor, called the opioid receptor (e.g. the mu-opioid receptor or MOR).

It all began with the quest for a new opiate
In his capacity as anaesthesiologist and pain expert, Prof. Dr. Christoph Stein has always been interested in this prominent receptor; especially on the issue if there is a way to better target inflammatory pain in peripheral tissue than with the opiates of standard practice, which carry harmful side effects. A few years ago, together with the mathematician PD Dr. Marcus Weber of the Zuse Institute Berlin, they developed a method to simulate the inflammatory environment in the computer to develop new kinds of analgesics. Two dozen molecules were checked by this procedure and eventually three appropriate candidates could be identified. These three molecules were then chemically synthesized by an external laboratory and were finally checked in animal testing.

Animal testing significantly reduced
It was a success: the result is a substance called NFEPP, which causes far lesser side effects than morphine and others. “NFEPP selectively activates opioid receptors in injured and inflammatory tissue without triggering the typical opioid side effects mediated by the “normal” opioid receptors in the brain,” Stein reports. Research groups in Europe and the USA could fortunately confirm these promising results.

But this is only the first part of this success story. The second part lies in the fact that the new computer simulations have already led to a significant reduction in animal testing. Where it is often standard in the search for a pharmaceutical ingredient to run hundreds of substances through animal testing, here they only ran three, thanks to the intelligent pre-selection.

Artificial Intelligence (AI) leads to quicker results
“Our method may lead to a reduction of roughly 90 percent of animal testing, because we can dispense with the traditional screening phase,” Stein emphasizes. “What’s more, the computer simulation is much faster, more on target and more cost-effective when it’s done under conditions that are specific to the disease the way we did.”
A fascinating aspect these days is that the AI-aided computer model can be transferred to the pre-eminent research subject of today: COVID-19.  Christoph Stein and his colleague Marcus Weber are already looking at the first projects.

They now hope that this method is used by as many researchers as possible. Everything is published, everything is described in detail and everything is freely accessible on servers. “This can be done by every researcher and it is a blueprint for the goals of 3R.”

(Text:  Beatrice Hamberger)

Links

Prof. Dr. med. Christoph Stein
Director
Department Experimental Anaesthesiology
Campus Benjamin Franklin

Science  03 Mar 2017:Vol. 355, Issue 6328, pp. 966-969. DOI: 10.1126/science.aai8636.
A nontoxic pain killer designed by modeling of pathological receptor conformations.
Spahn V, Del Vecchio G, Labuz D, Rodriguez-Gaztelumendi A, Massaly N, Temp J, Durmaz V, Sabri P, Reidelbach M, Machelska H, Weber M, Stein C.

The Berlin Mathematics Research Center MATH+

Zuse Institut Berlin - PH-dependent Opioids

Contact