By Jonathan D. Grinstein, PhD

Human geneticist David Goldstein, PhD, wants to make economically viable medicines for rare diseases with unmet needs.

“We start out not knowing what a patient has, find out what they have, and tell them this is what it is, but unfortunately, we can’t do very much about it,” Goldstein, the founding director of the Institute for Genomic Medicine at the Columbia University Medical Center, told GEN Edge. “The more common rare ones are getting lots and lots of attention [but] the rarer rare ones are not getting any attention at all. If you’re going after very rare diseases, the market may not be sufficient to make an economic case.”

So, Goldstein, author of several books, including “The End of Genetics: Designing Humanity’s DNA Hardcover,” began to concentrate on bringing rare diseases up to the same playing field as common diseases. “I started to think that you might be able to identify rare disease targets that would represent an opportunity to develop an effective treatment for rare diseases but also learn something that would connect the target to a more common indication,” said Goldstein.

Behind a $55-million Series A round, Goldstein, along with co-founder John McHutchison, AO, MD, today launched Actio Biosciences to identify effective treatments for terrible rare diseases and new knowledge connecting their target to other indications.

Canaan, Deerfield Management, DROIA Ventures, EcoR1, and Euclidean Capital led the funding round. The money will be used to move their lead small molecule program targeting TRPV4 mutations closer to the clinic so it can be used to treat Charcot-Marie-Tooth disease type 2C and other serious bone diseases.

Man vs. machine

At the heart of Actio’s drug identification and derisking pipeline is the Rare Disease Target Atlas, a large proprietary target database. With it, they are identifying rare diseases with mutations that all have the same molecular phenotype so that a bespoke medicine does not have to be tailored to each patient.

“We know that we have chemical matter that can normalize the activity of all pathogenic mutations,” said Goldstein. “So, we take all of the [mutations] and ask, what does chemical matter do to them? Then, we do all of those biological de-risking events so that the clinical trials are as de-risked as possible.”

The Rare Disease Target Atlas has a catalog of the roughly 4,000 genes that cause the 7,000-plus Mendelian diseases, as well as informatics tools to catch genes where the protein can be modulated. Other criteria include a high unmet need for the disease presentation and indicators that you can build a mouse model for the disorder.

“We have a variety of informatics tools that can identify [a target] out of 4,000 Mendelian disease genes, and for the ones that look like they might match, we dive in and just read paper after paper and decide what we really believe,” said Goldstein. “That’s a big part of what we do because if you dive into any of these rare disease genes, you find the literature points in every imaginable direction. You have to decide, ‘Well, I don’t believe all those papers. I believe these ones, and this is the experiment I have to do in the lab.’ So it’s informatics filtering first, followed by expert review, a bit of large-scale association analysis from all the publicly available data transcript analysis, and that kind of stuff.”

This approach may be surprising given the growing trends in the use of AI in drug discovery. But Goldstein thinks that AI won’t solve everything. He thinks that, in some cases, it’s humorous how much people depend on AI.

“It’s clear that AI works really well when you have enough data that you can identify the patterns. But what we’re trying to do is very clear—the data aren’t out there,” said Goldstein. “So, the whole idea is that you can skip the expert review part and just have the fanciest possible pattern recognition to say, ‘This is the target, and this is the indication.’ But it’s a pendulum, and the pendulum will swing against this too. There are very targeted uses of AI that work really well. But this general black box—here’s your target; your indication is frankly comical. So, we dive in, and we find everyone knows that what’s in the literature is just flat wrong. So you have to discard all that, and you have to make a judgment about it.”

Keep it simple

While Actio is theoretically modality-agnostic in practice, the first program was chosen to be as straightforward as possible. That means selecting targets where activating mutations cause disease and where they can be inhibited with small molecules. While Goldstein said that Actio will not stick to small molecules forever, right now, the company employs just 20 people with a small molecule chemistry team. But the strategy, of course, identifies lots of targets that don’t look amenable to modulation with small molecules, which they will eventually go after.

The TRPV4 program was selected as the lead because the rare sensorimotor neuropathic disease has a high unmet need, and pre-clinically, it looks tractable, according to Goldstein. That means that the Actio team is working specifically with mutations that have activation effects, which, theoretically, need to be countered to have a therapeutic effect.

Accordingly, for the lead program, if the same compound did not all normalize the different activating mutations, different drugs would be required for patients with different mutations, and that would be an extremely difficult proposition.

“That would rule it out for us,” said Goldstein. “We don’t test a few mutations and hope for the best. We test every single mutation to know whether we can treat the entire set of patients. We do everything that can be done preclinically to know whether or not [a treatment] will work, and then have the first clinical effort be in a completely homogeneous patient population. For every single patient in the first trial, we’ll have an activated channel that needs to be normalized in its activity. That maximizes the chances that it will work in the first effort for the company.”

More than that, this approach also creates a completely homogeneous patient population for biomarker discovery that might connect the target to other indications, which Goldstein said is a critical part of the story. Goldstein wants Actio to be part of the history of drug development by trying to identify rare, relatively homogeneous patient populations first and then expanding out from there.

Back to biology

Goldstein doesn’t mind opting out of the herd moving toward the use of human preclinical models, such as iPSC models and organoids.

“I think they’re extremely challenging to work with,” said Goldstein. They’re very noisy systems, and I don’t think things like organoids add a great deal to the picture right now.”

Golstein’s view is that they first need to show a consistent effect of the pathogenic mutations at the molecular level across the different pathogenic mutations—not one or two of them but consistent across the entire set. Then, they must have multiple valid mouse models covering these pathogenic mutations, which they are doing in partnership with the Jackson Laboratory.

“The best position you can be in is when you’re lined up from top to bottom,” said Goldstein. “The underlying cause of the disease is the same—not similar, but precisely the same. Our models start with an actual mutation that causes disease, and we have to know that the mutation does something similar so the channel gets activated similarly in the mouse. In addition to that, we require that the mouse look like a human presentation. There’s no way to do better than that.”

With the peripheral neuropathy presentation for the lead program, the mice are very severely affected, and the introduction of an inhibitor results in a complete reversal of the severe symptoms over 12 hours.

The final piece of the puzzle for Actio is that to make this inhibitor applicable to other indications, they must first understand the underlying biology. In the case of their lead program, an overactive channel is what drives peripheral neuropathy, which compromises the blood-brain barrier between the vascular system and the central nervous system. And, indeed, according to Goldstein, when they use the inhibitor, the blood-brain barrier gets shored up, and the mouse recovers.

All in all, Goldstein does not seem to be a person who suffers from Shiny Object Syndrome. He doesn’t fear missing out on the latest and most complex trend and, instead, believes that Actio’s deep understanding of biology is what will help them take on the development of drugs that people haven’t touched with a ten-foot pole because they don’t know how to make a profit. Each of Actio’s drug programs has the potential to be an initial drop that can send ripples throughout not only the rare disease world but also that of the most common. It won’t be until Actio’s work in 2024 to begin to evaluate whether to believe in this process as much as Goldstein and the investors do in Actio.

Previous articleNew Swine Flu Strains Raise Concerns about Pandemic Risks
Next articleGenotoxic Effects by Base and Prime Editors in Human HSPCs Aren’t Surprising, Says Pioneer