Creyon Bio deal with Lilly brings up to $1 billion in payments for research on RNA-targeting drugs

DNA is not destiny. More and more often, we can rescue patients from diseases caused by mistakes in their genes using medicines made from the same building blocks: nucleic acids. These molecules include sequences designed to latch on to specific stretches of native DNA or RNA to stop them from being expressed as mutant proteins, or to fix splicing errors that marred the proteins these stretches encoded. 

But there’s a major challenge with this realm of drug development. While it’s technically possible to make a huge range of nucleic acid medicines — which can take the form of small interfering RNAs (siRNAs), antisense oligonu­cleotides (ASOs), aptamers, and other types of ​“oligo therapies” — the vast majority of them end up being toxic to cells. Finding safe and effective oligo therapies is typically a matter of contem­plating billions of possi­bil­i­ties and testing each one’s toxicity/activity char­ac­ter­is­tics separately. Creyon Bio, which DCVC Bio has backed since 2020, captured our imagination by showing that they could use internally built, empirically derived data and design AI systems to create molecules that are safe/active by design.

Now phar­ma­ceu­tical giant Lilly has validated Creyon’s work through a licensing and research collab­o­ra­tion worth $13 million up front, plus up to $1 billion in milestone-based development and commer­cial­iza­tion payments. The partnership will help Lilly create new therapies for a broad range of diseases, and will enable Creyon to continue its own efforts to treat devastating neuro­mus­cular conditions and other diseases. All this will occur under the leadership of a new CEO, Serge Messerlian, who is an operating partner at DCVC Bio and has deep experience commer­cial­izing drug programs at Johnson & Johnson and other major drug companies.

Most drugs on the market bind to or interact with proteins. But not every protein of interest is druggable in this way; the promise of RNA-targeting drugs is their ability to block diseases before RNA becomes protein, or even block undesirable RNA activity itself. Initial versions of RNA drugs were ASOs — in fact, the first therapeutic drug, an ASO, was formiversen (Vitravene), approved by the FDA in 1998. The first siRNA drug, patisiran, was approved in 2018. Often confused, ASOs and siRNAs have very different properties. siRNAs are double-stranded RNA molecules with nucleotide sequences comple­men­tary to their target messenger RNAs. siRNAs prevent damaging proteins from being synthesized in the cell at all, by inter­cepting and degrading those mRNA targets. In principle, any gene can be silenced in this way. Conversely, ASOs are single-stranded oligonu­cleotide molecules and are even more versatile, in that they’re active in the nucleus, where they can ​“knock down” or block mRNA while it is being transcribed from genomic DNA. They can also act on non-coding RNA regions; up-regulate certain genes by targeting their regulatory modules; or even fix defects in RNA splicing (which, by some estimates, are responsible for up to 30 percent of genetic diseases).

Importantly, drug companies have developed a multitude of chemical modi­fi­ca­tions that make ASOs, siRNAs, and other RNA-targeted oligo therapies more drug-like by getting them inside the right cells or extending their half-life in the bloodstream — in some cases we can get twice-a-year dosing! But often, candidate ​“active” molecules proved either toxic or overly immune-stimulating — and it wasn’t always easy to say whether these side effects were due to the sequences themselves, the chemical modi­fi­ca­tions, or both. This has presented a ponderous permutation and combination problem for drug developers.

Creyon’s platform was built on a library of survey compounds repre­senting many possible nucleotide sequences and variable chemical modi­fi­ca­tions at different positions in these sequences. Creyon then documented how those compounds affected the livers, kidneys, cardio­vas­cular systems, and nervous systems of rodents, as well as human cells; and then used that purpose-built data to train machine-learning models to predict and engineer tolerable and effective oligo therapies. DNA, RNA, and proteins are such complicated molecules that there’s no first-principles method for predicting all the ways they will interact and create toxic effects — which is why Creyon took a more compu­ta­tional approach, ultimately short-circuiting the trial-and-error screening that previously ruled OBM development and expo­nen­tially accel­er­ating the process.

The company has used its design engine to swiftly develop an in-house pipeline of candidate medicines that could treat genetic neuro­mus­cular, CNS, and immunologic disorders affecting thousands of patients. The company expects to take its first drug candidate into the clinic next year. And now they’ll be offering that same engine to Lilly, under a cash and equity deal granting the multi­na­tional drug developer exclusive rights to lead candidates for target diseases of interest. 

Lilly recognizes that the ability to program­mat­i­cally design knowably safe and effective siRNAs, ASOs, and other oligo therapies could unlock a huge range of therapeutic advances. And Creyon’s new CEO Serge Messerlian brings tremendous experience commer­cial­izing both drug development platforms and drug asset portfolios at orga­ni­za­tions such as Johnson & Johnson and Baxter Inter­na­tional. We think Creyon has found a better way to engineer oligo therapies — and thousands of patients will live better lives as a result.