Advances in Medicinal Chemistry
Complete the form below to unlock access to ALL audio articles.
Medicinal chemists play a crucial role in drug discovery through the selection, synthesis and testing of a myriad of compounds. Recent advances in technologies are set to accelerate their progress.
Most small molecule drugs are the end-product of meticulous work by medicinal chemists – the result of an iterative process of selecting, designing, synthesizing and testing thousands of compounds to predict which possess the most drug-like properties.
“Our work is very early on in the drug discovery process – we find interesting compounds that have biological activity in disease models,” explains Donna Huryn of the School of Pharmacy at the University of Pittsburgh and Department of Chemistry at the University of Pennsylvania. “We will then pass these on to someone else to make them into drugs.”
But most compounds that are made will never reach the clinic, with vast numbers discarded along the way. One of the key skills of a medicinal chemist is to filter out compounds that initially appear to be promising, but that are unlikely to be further improved with a reasonable amount of time and effort.
“You’re always trying to rule out compounds that are never going to make good drugs,” says Huryn. “You need to know ahead of time what is their potential value, and what their liabilities are – and what you might need to do to fix them.”
But the latest technological advances, including artificial intelligence (AI) and computational modeling, offer exciting new opportunities to help expedite the work of the medicinal chemist and speed up progress in drug discovery.
The search for the best drug candidate
After the identification of a promising drug target – such as an enzyme that is involved in a disease-critical pathway - medicinal chemists can then start searching for compounds that specifically interact with the target and exert the desired effect.
“As well as potency against the target, the compound will also need to be safe, soluble, and not highly metabolized in the body,” says Huryn. “All these other things come into play as you need to make sure you can dose it at a concentration that is relatively low and that doesn’t cause too much toxicity – or, in other words, unwanted side effects.”
Researchers will initially evaluate the various properties of different compounds in a series of in vitro assays, many involving testing their effects on cultured cells.
“You may have to modify the properties of the compound, such as to improve its ability to permeate through the cell membrane, while also maintaining the properties that give it the desired biological activity,” says Daniel Flaherty of the Department of Medicinal Chemistry and Molecular Pharmacology at Purdue University.
The most promising compounds will undergo increasingly stringent in vivo models to examine how good the compound is at treating the disease (efficacy) and how the body treats the compound (pharmacokinetics).
“When we then move into living systems there are different things to worry about, such as how it’s metabolized in the body – and how’s it’s absorbed through the gastrointestinal tract to get to where it’s needed,” explains Flaherty.
Finally, only a single compound with the best qualities will remain – the drug candidate that is ready to enter clinical trials.
A slow, iterative process
The process of drug discovery will typically kick off by screening libraries containing several thousands of compounds, often with the help of high-throughput functional assays and computer software. The aim is to identify a group of initial “hit” compounds that have some biological activity against the target.
“Ranking their potency will give you a starting point for making decisions about which ones to take forward,” explains Flaherty. “But your prioritization scheme will also need to include other aspects including key physicochemical properties, such as their solubility.”
Cheminformatics can also provide clues about whether a compound is worth pursuing or not. “You also need to know if it’s possible to isolate or make more of the compound or its analogs,” says Huryn. “That might be a major issue if your initial hit is a complex natural product.”
A typical screen will generate hundreds of hits, which are whittled down to a handful of “lead” compounds. Next begins the process of lead optimization, which involves making small incremental changes to the structure of each lead compound to create a series of analogs that are predicted to improve its characteristics.
“We design what we think are better molecules, and then there’s an iterative process of evaluating these in a variety of assays,” explains Huryn. “At some point, we hope to identify compounds that we think are potent enough, soluble enough, metabolically stable enough to move into a more complex model.”
Structure-based drug design involves solving the 3D structure of the molecule bound to the target, typically through X-ray crystallography.
“That’s much more powerful because you can see how the molecule is binding,” enthuses Flaherty. “There may be a group that’s not contributing that you can lose, or you may spot a place to form a productive bond that’s likely to improve its biological activity.”
Ongoing challenges
Possibly the biggest challenge for medicinal chemists is to create a compound that combines good potency with the other desired qualities needed for a successful drug.
"There’s usually a battle between two forces, as the typical ways to improve its potency will be detrimental to its drug-like properties,” explains Flaherty. “Traditionally, you’re adding molecular weight and functional groups that don’t want to be solubilized in water.”
Computational modeling can help with the prioritization of drug design ideas. The approach involves inputting data generated about the different compounds into software, which then ranks according to various properties. “We do as much of in silico design as we can because there are so many molecules that we could make, and we try to make smart decisions,” says Huryn.
The approach can be applied with or without structural knowledge of the target, but the quality of the model is dependent on the data that is available on the compound.
“Sometimes the models are good, but sometimes they’re less predictive,” says Huryn. “There must be a paper every other week about how to predict whether compounds are going to be soluble, which tells me there’s not a good model yet!”
Although in silico modeling can help identify the most drug-like compounds to take forward, it does have limitations and the expertise of the medicinal chemist is still required to make sense of what is suggested.
“It’s reliable to an extent, but it’s certainly not foolproof,” says Flaherty. “You’ll still want to make some molecules that are predicted not to have that activity because they give you good tests on your model.”
The future of medicinal chemistry
AI has huge potential to change the landscape of medicinal chemistry. “AI is blowing up in the field right now,” enthuses Flaherty. “There is often so much data out there and you’re looking at all these multi-parameters that it’s almost impossible for a human to boil all that down into a good design strategy.”
As well as the potential to inform drug design strategies, AI can also help medicinal chemists with predictions for how to make compounds without any preconceived biases about what is most likely to work.
“For a long time, people said that a computer can’t tell you how to make a molecule,” says Huryn. But she recently changed her mind after taking part in an experiment pitting the ability of computer software at predicting how to synthesize a compound against that of a human (known as the “Turing test").
“They gave us two different approaches for synthesis, one that had been done by a person and another by the computer,” says Huryn. “I was convinced I knew which one was done by a human, but I was shocked when I saw the results!”
In the longer term, improving the throughput of drug discovery through automation offers other tantalizing opportunities for boosting productivity. “There’s a big push for automated synthesis – and there are already machines available, but they do pretty simple stuff,” says Huryn. “But my dream would be the ability to program a computer to make any type of compound.”
Working at the heart of drug discovery
The success of small molecule drug discovery heavily relies on a combination of the knowledge, expertise – and a great deal of perseverance – of medicinal chemists. “What’s exciting about working in this area is the potential to find something that will help patients,” says Huryn. “But it’s always challenging, you’re always making new compounds that have never been made before and there’s always something surprising in the data.”
But they also need to have a good dose of enthusiasm and passion about their work, particularly as many of their efforts will not end up with a successful drug. “Working in this area is super exciting, intricate, and complicated,” enthuses Flaherty. “But when you can successfully manipulate something with a small molecule, it’s really great!”