Fragment-Based Drug Discovery Enters the Mainstream

Drug hunters are increasingly turning to fragment-based screening to search for high-quality chemical starting points for early-stage drug discovery.

Small molecule drug discovery is plagued by high attrition rates along the development pathway. Many promising drug candidates identified during initial screening programs are discarded at a later stage and in many cases, these failures can be traced back to the quality of the initial chemical leads. Employing alternative approaches, such as fragment-based drug discovery (FBDD), could help to increase the chances of success as well as open new opportunities to tackle previously intractable biological targets.

“You use very small compounds that are promiscuous, which allows you to find some chemical starting points from which you can then go on and assemble properly potent molecules,” says Frank von Delft, professor of structural chemical biology at the University of Oxford.

FBDD typically employs very sensitive biophysical methods to rapidly screen a small library of low molecular weight fragments to identify initial “hits” that bind – albeit weakly – to a target protein. As these chemical starting points are small and typically soluble, they are more likely to have better drug-like properties, improving the likelihood of eventually producing a viable drug.

“Because the fragments are so small, there are so many ways you could then build them up to a full-sized drug molecule, so screening a library of a few thousand compounds is representative of a vast amount of chemical space,” explains Andy Merritt, associate director of chemistry at LifeArc, a UK-based medical research charity that supports translational research.

In 2011, the targeted cancer treatment vemurafenib was the first fragment-derived drug to be approved by the US Federal Drug Administration (FDA). Since then, dozens more drug candidates derived from fragments have entered the clinic – with further approvals and many currently in late-stage trials.

Traditional drug discovery

The process of drug discovery usually kicks off with the identification of a biologically important target, such as a protein that plays a critical role in disease. Medicinal chemists will then begin the hunt for a compound that can bind to the target and modify its activity.

Traditionally, the search will often start with high-throughput screening (HTS), which involves testing libraries containing hundreds of thousands – or even millions – of drug-sized chemical compounds (< 500 Daltons [Da]) to narrow down a list of “hits" with promising activity against the target. These initial candidates can then be further refined through subsequent rounds of development to improve potency and refine their “drug-like” properties – such as non-toxicity, solubility and stability – which are also essential for clinical success.

“These screens take a lot of effort and are expensive, and while they can test a lot of compounds, it’s a relatively small number compared to the vastness of the potential chemical space,” says Merritt.

An early estimate put the number of possible small drug-like molecules at 10⁶³, making even the largest compound libraries look tiny in comparison. And so, finding the rare ones that are a good fit for the target will extrapolate to exploring a staggeringly wide range of molecules – especially due to the probable high number of “near misses”.

“Intuitively, you might think by having more complicated and bigger molecules will improve the odds of finding a good candidate in your initial screening experiment,” says Merritt. “But if just one of those interactions is wrong, it won’t bind to the target at all.”

A smarter approach

Rather than physically screening increasingly high numbers of drug-sized compounds, FBDD instead begins by testing much smaller collections containing just hundreds or thousands of fragments. While individual fragments are too small (< 300 Da) to have strong interactions with the target, they are still capable of binding into lots of tiny crevices in its structure.

“Due to their small size, these fragments have much less decoration than drug-sized molecules,” describes Merritt. “So you’re much less likely to hit detrimental interactions – increasing your chances of finding something that binds to the target.”

Since each fragment has so few interactions with the target protein, the binding affinities are typically in the high micromolar to the millimolar region – which is much lower than for larger molecules.

“A big disadvantage is you have to start from very low potency, which can be very disconcerting for your typical medicinal chemist,” says von Delft. “As the signal of binding is weak, fragment-based screening requires sensitive and robust assays so you can differentiate appropriate fragments from false positives.”

Biophysical methods – including nuclear magnetic resonance (NMR) spectroscopy, surface plasmon resonance (SPR) and X-ray crystallography – are the most commonly used techniques for FBDD. Of these, X-ray crystallography offers the advantage that the 3D structures show exactly how each fragment binds.

“You can see the position of the atoms,” explains von Delft. “That’s enormously powerful because it cuts out a lot of the guesswork.”

Medicinal chemists can then take clues from these structural data, which will help them to piece together larger, more potent drug-like molecules for onward development.

Boosting efficiency

While the throughput of FBDD is generally lower than other screening methods, the introduction of automation in crystal handling and data analysis has improved this issue.

At the Diamond Light Source, the UK national synchrotron, the XChem facility has streamlined the process of fragment-based screening using X-ray crystallography, enabling academic and industrial researchers to screen up to 1,000 compounds in less than a week.

“We’ve made it possible to run the X-ray crystallography many hundreds of times in 24 hours without interference,” enthuses von Delft. “So users can come and perform this gold-standard experiment, and by the end of it, they will have some strong data to take away and build on.”

A powerful example of how fragment-based screening can help accelerate drug discovery comes from the COVID Moonshot, a spontaneous global collaboration that came together in early 2020 to urgently design an effective new antiviral treatment. The search was kicked off with an XChem experiment of over 1,250 unique fragments to probe a key protein of the SARS-CoV-2 virus, identifying 74 initial hits.

“It was quite mind-blowing at the time – we were all taken aback by it,” reflects von Delft. “We had very interesting data, and we decided to put it out to the world to ask medicinal chemists for ideas of how to take these fragments forward.”

This crowdsourcing effort had attracted more than 4,000 suggestions for improved, more potent compounds within two weeks – and all contributions were submitted with no intellectual property attached.

“We found a company in Ukraine who committed to synthesizing some of the most promising compounds at cost,” says von Delft. “If everything goes well, we’re expecting to have drugs that are ready for Phase I clinical testing by early next year.”

A changing tide

After decades of development, fragment-based screening is now becoming one of the mainstream approaches in small molecule drug discovery. Due to its lower initial costs, it has opened the possibility of lead discovery to smaller companies and academic researchers.

“In the past, it was very difficult to get traction as people didn’t start drug programs based on millimolar binding affinities,” says Merritt. “But we now have the right tools as well as computational and structural support to drive these projects forward.”

Because of its higher hit rates, it has also opened new opportunities for drug discoverers to crack more challenging targets such as protein–protein interactions.

“It’s become part of the arsenal now,” concludes Merritt. “It’s a technique that everybody will use at the early stages of small molecule drug discovery.”