Aiming for the perfect fit

Credit: Chris Gash

Macrocyclic peptides could be the perfect size to become therapeutics.

In brief

In the search for new drugs, a series of advances in laboratory technologies and computing power has conspired to make the next generation of therapeutic molecules feel tantalizingly within reach. Known as macrocyclic peptides, these drugs are purported to create the Goldilocks effect—they have the size and shape to provide the advantages of both small- and large-molecule drugs without their shortcomings.

Small-molecule drugs have been instrumental in alleviating a gamut of ailments, from heart disease to the humble headache, but some experts now think we’ve exhausted the number of targets those drugs can reasonably tackle. In contrast, large, protein-based drugs can target protein interactions that small molecules have a hard time influencing, but larger molecules have their own drawbacks.

That’s why scientists have been pursuing macrocyclic peptides. Researchers and investors are betting on building therapeutics from these medium looped molecules.

“Macrocyclic peptides provide the best of both worlds,” says Vikram K. Mulligan, a research scientist at the Center for Computational Biology at the Flatiron Institute and a cofounder of Menten AI, a San Francisco–based company that designs peptide therapeutics.

A macrocyclic peptide is any string of 12 amino acids or more that curls into ring structures, according to Christian Heinis, who develops therapeutics based on cyclic peptides at the Swiss Federal Institute of Technology, Lausanne (EPFL). Heinis has also cofounded two companies to commercialize his discoveries.

But not everyone subscribes to the precise definition that Heinis provides. Some prefer a vaguer characterization. “I wouldn’t even include the 12-or-more component,” Mulligan says. “It’s a smallish chain of amino acids making at least one loop. The ones I’ve worked on ranged from 6 to 60 amino acids. It’s the loop that is important.”

Small droplets fluoresce green yellow in the bottom of a microwell plate.

Credit: Christian Heinis

Nanoliter droplets that show the small volumes in which chemists can synthesize combinatorial libraries of cyclic peptides for high-throughput screening

That cyclic shape provides a greater surface area for peptides to disrupt the deleterious protein-to-protein interactions that drive diseases, and the cyclic structure’s stability helps the molecule stay intact once it encounters the human body. “We have an immune system with proteases evolved to break down peptides and proteins and degrade them. It’s harder for proteases to break down cyclic structures,” Mulligan says.

This molecule is too big

Larger, protein-based drugs, such as antibodies, don’t have the same luxury of stability. “Proteins are more easily detected by proteases. Our body is evolved to eliminate them,” Mulligan says. “Large molecules also tend to be less shelf stable, and they’re prone to unfolding at high temperatures. The other thing is that they’re so big that they can’t get across barriers well.”

For these reasons, large-molecule drugs are used mainly for extracellular targets, and they’re usually injected directly into the bloodstream, which limits the kinds of diseases they can treat.

“A patient is OK with injection if it cures their cancer, but they’re not so keen on the same delivery mechanism for headaches, and we have a lot of diseases we want to treat all the way along that spectrum,” Mulligan says.

The advantage of large molecules lies in their precision. Thanks to their size, they have a larger surface area, which they can exploit to better recognize targets. This recognition, in turn, means that they have a lower risk of mistaking a healthy cell for a diseased cell, which reduces side effects and allows the drugs to be given at higher potencies to thwart the disease more rapidly.

This molecule is too small

Small molecules, by comparison, can struggle with specificity. “Think of small molecules like a small key. A small key can kind of fit into other keyholes whether it’s the right keyhole or not. Small molecules can end up doing this and binding to lots of different receptors, creating side effects,” Mulligan says.

This lack of specificity is what plagues traditional cancer chemotherapies—they kill the cancer cells, but they also end up killing a fair number of healthy cells too, severely harming patients’ quality of life. “It’s challenging for small molecules to impact the large surfaces of protein-protein interactions, which govern a wide range of biological processes in our bodies,” Mulligan says.

For a long time, people thought molecules needed to be smaller to go inside a cell, but we can now show this isn’t the case.

Christian Heinis, biochemist, Swiss Federal Institute of Technology, Lausanne

Small molecules are, however, relatively easy to create through chemical synthesis. They can pass through cell membranes and traverse barriers such as the gut-blood barrier. They are also usually more shelf stable than large molecules. All this means that small-molecule drugs can often be taken as simple pills.

This molecule is just right

Drug developers and clinicians have seen the debate over drug molecule size largely as a never-ending series of trade-offs: which negatives are more palatable for which patients and which conditions? But macrocyclic peptides are beginning to change the discussion.

For example, Heinis has shown that despite macrocyclic peptides’ size, they can pass through cell membranes into cells. “For a long time, people thought molecules needed to be smaller to go inside a cell, but we can now show this isn’t the case,” he says.

Macrocyclic peptides can essentially do what small-molecule drugs do, Heinis says, but they are better at binding to target receptors. “In theory they could be used as drugs for many more targets that we struggle to get with small molecules,” he says. “There is so much enthusiasm for anything cyclic at the moment.”

Earlier this year, Heinis launched his second start-up, Orbis Medicines, to capitalize on his previous work. Novo Holdings, the investment arm of Novo Nordisk, was a key investor in Orbis, and other multinational companies are betting on macrocyclic peptides and their potential to become new drugs.

In January, Merck & Co. invested in Unnatural Products—a California-based company that specializes in macrocyclic molecules—in a deal worth up to $220 million. And the chip giant Nvidia announced this month that it has invested in Vilya—a 2-year-old company that uses artificial intelligence to design macrocyclic drugs—as part of a $71 million funding round for the start-up. Other cyclic peptide–focused firms getting recent funding include California-based Insamo and Curve Therapeutics in the UK.

There is so much enthusiasm for anything cyclic at the moment.

Christian Heinis, biochemist, Swiss Federal Institute of Technology, Lausanne

All this is a long time coming. Scientists have for decades discussed the potential of cyclic peptides to tackle “undruggable” targets, but the US Food and Drug Administration has yet to approve any (besides those that occur in nature). “For a long time, the technology to synthesize and design these macrocyclic peptides for targets of interest wasn’t there, but there have been several breakthroughs that changed the game,” Heinis says.

Mulligan points to three things that are resulting in this crescendo moment—not least a belief that macrocyclic peptide therapeutics can be done. “One is human bias in terms of what can work. Some of it is that people are increasingly frustrated with small molecules and looking for something new.” The two other innovations are more scientific and technical: synthetic advances and the application of artificial intelligence.

Drugging the undruggable

“The ability to make libraries of peptides and new chemistries for making peptides” have hugely helped the search for macrocyclic peptide drugs, Mulligan says. “Relatively recent chemistries for cyclizing the peptides, making the covalent bonds to join the ends, have also helped.”

To look for therapeutic leads, researchers search through huge peptide libraries. The peptide’s corresponding genetic sequence is more useful than the peptide itself at this stage because it’s easier to make a peptide with a DNA-based synthesis method than with traditional chemical synthesis. Using polymerase chain reactions, researchers can take small amounts of DNA and replicate them before implanting the DNA into living cells to produce the macrocyclic peptides. This technique “allowed the field to make libraries of billions of macrocyclic peptides and screen them,” Heinis says.

For that screening, researchers rely on different display technologies. One of the most used is phage displays—in which scientists tinker with a virus’s genome so that the virus’s surface contains peptides for screening against a desired target. Scientists can deploy a similar technique with peptides attached to fragments of messenger RNA instead of viruses. “It’s a laborious process of doing high-throughput experiments that give you some sort of a readout in terms of how well a peptide binds to a target,” Mulligan says.

While libraries have become bigger, the screening technology remains a limiting factor. And there are more possible proteins than can ever be contained within any synthetic library, Mulligan points out. “If I was making a peptide and I was making it from just the 20 natural amino acids and I wanted it to be 10 amino acids long, there are still more possible combinations than exist in the largest of libraries with billions of peptides,” he says. “And that’s just using the natural amino acids. . . . There are thousands of amino acids, so if you look beyond the 20 natural ones, the numbers start to get silly.”


Searching for macrocyclic peptides

The traditional process of finding a candidate macrocyclic drug involves several stages. Some researchers might use artificial intelligence to design candidate peptides and skip the first two steps.

Graphic illustrates the steps from a peptide library to having multiple macrocyclic peptides. The intervening steps are screening the library for matches against a target, identifying a candidate peptide, isolating the DNA code for that candidate, using polymerase chain reaction to create multiple copies of the DNA, and inserting the DNA in living cells.

Credit: Yang H. Ku/C&EN/Shutterstock

An alternative approach is to use artificial intelligence to design macrocyclic peptides, and even use machine learning techniques to predict which molecules from these digital libraries might bind successfully to a desired target.

“This is the work that my company does, so of course I’m going to boast about it,” Mulligan says. “But we can design molecules on the computer so that by the time we get to the wet lab, we’re only testing a few molecules, and that means reduced human labor and lower costs in terms of lab resources too.”

Cyclic peptides have a big opportunity.

Parisa Hosseinzadeh, computational biochemist, University of Oregon

Efforts like this are beginning to yield results, according to a 2022 review by Christina Lamers, a professor at Leipzig University’s Institute for Drug Discovery. These advances “may soon help usher in an age highly shaped by peptide-based therapeutics,” she writes.

“Cyclic peptides have a big opportunity,” agrees Parisa Hosseinzadeh, a computational biochemist at the University of Oregon. She has developed new computational tools that have identified more than 200 designs for cyclic peptides that are predicted to fold into a stable structure.

Many firms are focusing on disrupting the protein-protein interactions that drive different cancers, but the technology isn’t limited to tumors; chemists could in theory design a macrocyclic peptide to target just about any receptor for just about any disease.

“Attention to [macrocyclic peptides] dropped for some time,” Hosseinzadeh says, “but it’s back.”

On your bike

Bicycle Therapeutics is a biotechnology company that’s been around since the first wave of interest in macrocyclic peptide drugs. It was founded in 2009 in Cambridge, England, on the back of research from Heinis and Nobel Prize winner Sir Greg Winter. As the name suggests, the company develops macrocyclic peptides that have two loops in their structures.

Bicycle is recruiting patients to participate in a clinical trial for its lead molecule, known as BT8009, which binds to bladder cancer antigens. “We went for bladder cancer because bladder cancer tumor antigens are expressed very highly, and so it doesn’t require prescreening of patients to make sure they’re expressing the antigen, so it’s a good candidate for a trial run,” Bicycle’s CEO, Kevin Lee, says.

Unlike other macrocyclic peptides, which are designed to treat diseases themselves, BT8009 is used to deliver a toxic payload to bladder cancer tumors. “The bicycle finds the tumor’s postcode, and then it can deliver the toxin,” Lee says. “Some tumor cells are resistant to certain toxins, and so you might want to attach a different payload.” He hopes that this approach will mean BT8009 can be readily adapted according to the genetics of an individual patient’s cancer cells.

Bicycle has two other molecules in development and will likely continue looking for others.

Bumps in the road

Chemists and biotech companies working on macrocyclic peptides are understandably optimistic, bullish even, about these molecules’ future roles. But that’s not to say it’s an easy ride from here to an abundance of approvals from agencies such as the FDA.

Although macrocyclic peptides can passively traverse cell membranes, and indeed this is one of their selling points, researchers have yet to perfect that property. To try to address this, researchers are focusing on macrocyclic peptides’ polarity, Heinis says. But he points to another area that needs more effort: “For me, it’s the library size. We don’t yet get a binding for every target even if we screen the larger libraries.”

And for researchers like Mulligan and Hosseinzadeh, who are exploiting artificial intelligence and machine learning, there is also troubleshooting to be done. “You need data on which to train the AI, and we’re slightly limited here in terms of peptides,” Mulligan says. “If you want to use exotic amino acids, the artificial ones with weird side chains that don’t exist in nature, then we don’t have training examples, and we struggle to use machine learning for that sort of thing.”

Despite these challenges, there’s a positive feeling around the future of macrocyclic peptides. If they live up to their hype under the glare and scrutiny that come with clinical trials and manage to make it to FDA approval, it will be more significant than just a new drug entering the marketplace. Maybe, these researchers hope, macrocyclic peptides will turn out to be the modality that’s just right.

Benjamin Plackett is a freelance science journalist based in London.


Source link

Total
0
Shares
Leave a Reply

Your email address will not be published. Required fields are marked *

Related Posts