It’s a bit like building with Lego bricks.
German version/Deutsche Version
Hormones and vitamins are among the fat-soluble molecules that can cross the double membrane of our body cells. Unlike water-soluble messenger substances, they do not depend on membrane-bound proteins to transmit their messages into the cell interior, but instead encounter so-called nuclear receptors after unhindered passage. Strictly speaking, these are dormant transcription factors. Only when, for example, a sex hormone or a thyroid hormone, when cortisone, vitamin A derivatives or vitamin D bind to them do they wake up and, if they are not already attached to it, migrate to the genetic information DNA to cause it to release certain protein building plans. Some agonists of these nuclear receptors are among the most important drugs of all, be it for anti-inflammation, for the treatment of diabetes, for the therapy of hormone-sensitive cancers or for contraception. We humans have 48 nuclear receptors, this has been certain since the decoding of the human genome due to their great structural similarity. However, almost half of them are orphan nuclear receptors, whose function and natural binding partner are unknown. This makes them extremely attractive for drug discovery. “When I started doing my PhD, some nuclear receptors were a big deal as drug targets,” Daniel Merk recalls.
The hype around the farnesoid X receptor
That was in the summer of 2011, after he had completed his pharmacy studies at Ludwig Maximilian University in Munich and moved to Manfred Schubert-Zsilavecz’s group at the Institute of Pharmaceutical Chemistry at Goethe University in Frankfurt for his doctorate. One of the most prominent drug discovery targets at that time was the farnesoid X receptor (FXR). It is involved in the regulation of sugar and fat digestion and is predominantly activated by bile acids. In patients with non-alcoholic fatty liver, which affects about 15 percent of the population in industrialized countries, its additional drug activation promised to become a viable treatment strategy. Just how promising this strategy was considered to be became apparent in the fall of 2014, when the small Ludwigshafen-based company Phenex Pharmaceuticals sold its portfolio of FXR agonists for just under 400 million euros to the U.S. company Gilead, whose coffers were just bulging thanks to the hepatitis C drug sofosbuvir. For Gilead, this was a very risky acquisition. It turned out that FXR agonists also have side effects, which, according to one hypothesis, could be due to excessive activation. While natural agonists of nuclear receptors are often only moderately potent or rapidly degraded, agonists from the laboratory tend to have strong and prolonged activation. “So my PhD topic was to develop partial agonists,” says Daniel Merk.
The art of polypharmacology
However, partial agonism, i.e. targeted weaker activation, also entails a loss of efficacy. Based on his findings with acylanthranilamide derivatives as FXR partial agonists, which had been evaluated summa cum laude, he therefore ventured in the next step to develop a dual modulator for the treatment of fatty liver hepatitis. He fused two pharmacophores in one molecule, i.e., combined those structures of two drugs that are their actual effect carriers, in this case the pharmacophore of an FXR agonist with the pharmacophore of an anti-inflammatory inhibitor of soluble epoxide hydrolase. The dual attack was able to succeed because both targets had lipophilic binding sites. “It’s a bit like building with Lego bricks,” Merk says. “You have to please both pharmacophores without creating clashes with either target when you introduce a structural modification for the other.” Systematic optimization of the fused lead molecule eventually resulted in a dual modulator that delivered high potency with low toxicity and made it to advanced stages of development.[i] Ultimately, however, the indication of fatty liver proved too expensive for further development. To prove an effect of the dual modulator, clinical studies with a very large number of patients over a period of two to three years would have been necessary. No company was willing to do this. Pure FXR agonists, on the other hand, have in some cases failed in the meantime due to their side effects.
Exploratory tools for target validation
Merk is still interested in polypharmacology. “The approach has some potential therapeutically because you can create synergies of action and increase tolerability through lower dosing.” But the time may not be ripe for it yet, he says. “We’d rather make two active ingredients,” they often say. So during his time as a junior research group leader at Goethe University (2015 to 2019), Merk increasingly turned to synthesizing compounds that bind to orphan receptors like tools and can thus help elucidate their function and unlock their pharmacological potential. He did so in the hope of discovering new targets, particularly for the treatment of neurodegenerative diseases such as Alzheimer’s, Parkinson’s and multiple sclerosis (MS). One non-orphan receptor that is the source of this hope is the retinoid X receptor (RXR). It occurs in three subtypes that could not be stimulated independently before Merk discovered a first selective agonist for the beta subtype[ii] . Another subtype of this receptor may have regenerative effects and could trigger the repair of myelin sheaths that encase healthy nerve fibers in an insulating manner. “So this could be an approach not to slow down a disease like MS, but to turn on repair mechanisms against the damage it causes.” He finds such approaches very promising. “Industry has stopped a lot of projects in neurodegeneration because they could hardly produce effects,” Merk says. “I’m not saying we can, but I think it’s important to leave no stone unturned.” Nuclear receptors offer themselves as targets with a future in this field, he says. “To validate these targets, we need potent molecules with defined biological effects as exploratory tools.”
A lot of work goes into the synthesis
The functionally crucial domains of all nuclear receptors are the one with which they bind to DNA and the one to which the ligand binds to regulate the transcription of specific genes. While the DNA-binding domains are confusingly similar, the globular ligand-binding domains, composed of about 250 amino acids, differ significantly. They determine a nuclear receptor’s selectivity for specific ligands and have the binding pockets into which the small molecules designed by Merk and his collaborators must fit and induce conformational changes to cause activation or repression. “A great deal of work by my graduate students goes into synthesis,” says Daniel Merk. “Until we have a molecule that does exactly what it’s supposed to do, we cook up tens of compounds in good cases and hundreds in bad cases.” Tools like the one he is aiming for are a prerequisite for the development of drugs, but their construction requires other qualities in some cases, he says. For example, he said, a drug substance must be bioavailable and safe, while a tool used in cell cultures needs primarily good physicochemical properties and low toxicity. It is a key to unlocking structure-activity relationships, a tool that alters receptor protein activity to reveal its mechanisms and pathomechanisms. “We’re using our tools to try to really understand nuclear receptors in-depth, using molecular design language models for their synthesis in addition to classical systematic medicinal chemistry.”
Design with digital words
Merk learned how to work with such language models at ETH Zurich, where he conducted research as a fellowship scholar from 2017 to 2019 in addition to his work in Frankfurt. These models involve expressing molecules in a specific character encoding like words that can be digitally processed. A prominent example is the SMILES (Simplified Molecular Input Line Entry Specification) code, which Merk became familiar with in Zurich. “These models already work very well. We’re trying to further develop how we can apply them optimally, because the crucial thing is always what data we use to let computers learn from it.” For in silico design of ligands, and in some cases for their optimization, Merk uses such learning language models, but is careful not to drift too far into informatic theory. “We try to adapt the language models to our practical needs, depending on the project.” Because his group works on under-researched targets, he says, it relies on using machine learning to extract the maximum amount of information from a minimum of self-generated structure-activity data. “In one case, we were able to do that with just a single template.”
Dating for two orphans
Two of the orphan receptors that Merk is primarily focusing on to find endogenous ligands and pharmacological tools for them are abbreviated TLX and Nurr1. Preliminary data suggest that both may be attractive targets for treating neurodegenerative diseases. TLX appears to be expressed almost exclusively in neural stem cells in adults and regulates the balance of this cell population, without which severe defects can occur in the central nervous system. Nurr 1 is particularly abundant in neurons that communicate by means of the neurotransmitter dopamine and regulates its metabolism. As a result, the receptor is closely associated with Parkinson’s disease, which particularly affects some dopaminergic neurons. Whether these assumptions are correct and how far-reaching the consequences of activating these nuclear receptors are will only become clear once selective agonists for them have been found. In addition to developing such chemical tools, Merk’s group is trying to develop neuronal in vitro models that can be used to measure the effects of the tool candidates in high throughput.
Enough ideas for the rest of the career
Many of the low-hanging fruits on the tree of nuclear receptor modulators have long since been harvested, says Daniel Merk, who was a group leader in Frankfurt until 2021. What remains, he says, are orphans like TLX, on which some have cut their teeth so far without success. “We remain curious and are looking at the target from all possible angles.” Merk now holds a tenured professorship in pharmaceutical and medicinal chemistry at LMU Munich, helped by support from the Aventis Foundation’s Life Sciences Bridge Award. One of the young professor’s basic scientific interests is to develop small molecules that possess a specific biological activity, thereby enabling the targeted elucidation of the function and therapeutic potential of little-studied proteins. In principle, his research can be transferred from one protein family to another. “Nuclear receptors are one of those families,” he says. “They are my current playground, but it may look very different in 10 years. I have enough ideas for the rest of my career.”
Author: Joachim Pietzsch, Wissenswort
Fotos: © Uwe Dettmar
[i] Schmidt, J.; Rotter, M.; Weiser, T.; Wittmann, S.; Weizel, L.; Kaiser, A.; Heering, J.; Goebel, T.; Angioni, C.; Wurglics, M.; Paulke, A.; Geisslinger, G.; Kahnt, A.; Steinhilber, D.; Proschak, E.; Merk, D. A dual modulator of farnesoid X receptor and soluble epoxide hydrolase to counter non-alcoholic steatohepatitis. J Med Chem 2017, 60(18), 7703-7724.
[ii] Merk, D.; Grisoni, F.; Friedrich, L.; Gelzinyte, E.; Schneider, G. Computer-assisted discovery of retinoid X receptor modulating natural products and isofunctional mimetics. J Med Chem, 2018, 61(12), 5442-5447.