Machine builds druglike organic molecules in a snap

Organic chemistry, one of science’s most grueling disciplines, is poised to get a whole lot easier. Six years ago, a team of chemists created a robotic system that could construct a wide variety of organic molecules including potential pharmaceuticals, dyes, and perfumes. But the setup had limited abilities: It could build flat molecules—chains or 2D rings—but it couldn’t make anything 3D, which many medicines and materials require.

Now, 3D has arrived. The team reports this week it has reworked its design to synthesize 3D organic molecules, enabling automated setups to build the majority of molecules now painstakingly assembled by organic chemists in the lab.

“It’s a landmark achievement,” says Timothy Cernak, a medicinal chemist at the University of Michigan, Ann Arbor, who was not involved with the work. “This has been a holy grail for chemistry for a long time.”

Robots have previously revolutionized the making of DNA, RNA, and short proteins. Progress was swift because in each case, these so-called biopolymers come together from a relatively small number of building blocks that can all be linked by making the same chemical bond.

That’s not the case with small organic molecules—those needed for medicines and many other applications—which use an array of reaction conditions, catalysts, reagents, and so on to forge bonds between different atoms and at different angles and orientations. Instead of behaving like boxcars on a train all linked by a similar connection, organic molecules are more like making craftsman furniture with an array of joinery techniques to secure irregularly shaped pieces of wood together in the desired architecture.

As a result, synthetic organic chemistry has remained open only to those with years of experience with the craft. “We’re trying to change that,” says Martin Burke, a chemist at the University of Illinois, Urbana-Champaign.

He and colleagues started on this path in 2015 by creating a machine capable of synthesizing a vast array of organic compounds. To do so, they began with a wide assortment of chemical building blocks of different shapes—each containing two chemical linkers, one called a MIDA that’s attached to one carbon atom, the other called a boronate that’s attached to a different carbon. The machine then piped in reactants sequentially so that the MIDA on the first building block comes together with the boronate on the second. The MIDA and boronate linkers wash away as the carbon atoms they were attached to bond to each other. The second building block’s MIDA then hooks up with boronate on the third, joining their carbon atoms, and so on, until the desired molecule was complete.

The setup allowed the creation of many druglike organic compounds. Since that initial paper, more than 250 academic and industrial labs have used MIDA-boronate chemistry to synthesize new molecules that have spawned more than 750 publications and 200 patent applications, Burke says.

But the approach hasn’t worked for everything. That’s because the link between the MIDA on one building block to the boronate on the other almost always creates an “sp²-sp²” bond, which links two carbon atoms and the atoms attached to them into a single plane. “Nature is not flat,” Cernak says. “It lives in 3D.” Innumerable organic molecules, from antibiotics to perfumes, include bonds, called sp³ bonds, in which some of the two carbons’ attachments jut out of the plane the carbons are in, much as a stool’s legs descend in three directions out of the seat.

Burke and his colleagues tried to use MIDA-boronate connectors to forge sp³ bonds. But making such a link requires harsher reagents, which wound up ripping the MIDA groups off the building blocks, causing a cascade of uncontrolled chemical reactions that created a broad mix of byproducts rather than just the targeted compound.

The team has now solved this problem by discovering a MIDA relative, known as TIDA. TIDA linkers are up to 1000 times more stable than MIDA linkers, which allows them to withstand the harsher reagents needed to construct sp³ bonds, Burke and colleagues report this week in Nature.

TIDA-boronate linkers in hand, the team used its molecular synthesizer to create two complex natural products, an antibiotic called ieodomycin C, and an antifungal compound called sch725674. Both contain what are known as chiral centers, in which elements are arranged in just one of two distinct mirror-image orientations. Such chiral centers are key to the function of many drugs and other organic compounds.

Now, Burke says the challenge is to come up with libraries of building blocks that contain TIDA and boronate linkers so that moleculemakers can assemble them at will. “We want nonspecialists to be able to make molecules at the push of the button,” he says.

Burke is now in discussions with chemical companies to supply such libraries, and he is working to commercialize his automated setup, so that labs around the globe can use it. He says he’s also working with colleagues to create artificial intelligence software to dream up novel organic compounds as potential new medicines. “If it becomes child’s play to make new molecules, the bottleneck becomes imagination.”