Potent therapy candidate for fatal prion diseases

That is why Vallabh was excited to discover, when she entered into a collaboration with Whitehead Institute Member Jonathan Weissman, that Weissman's group likes to work at full throttle. In less than two years, Weissman, Vallabh, and their collaborators have developed a set of molecular tools called CHARMs that can turn off disease-causing genes such as the prion protein gene -- as well as, potentially, genes coding for many other proteins implicated in neurodegenerative and other diseases -- and they are refining those tools to be good candidates for use in human patients. Although the tools still have many hurdles to pass before the researchers will know if they work as therapeutics, the team is encouraged by the speed with which they have developed the technology thus far.

"The spirit of the collaboration since the beginning has been that there was no waiting on formality," Vallabh says. "As soon as we realized our mutual excitement to do this, everything was off to the races."

Co-corresponding authors Weissman and Vallabh and co-first authors Edwin Neumann, a graduate student in Weissman's lab, and Tessa Bertozzi, a postdoc in Weissman's lab, describe CHARM -- which stand for Coupled Histone tail for Autoinhibition Release of Methyltransferase -- in a paper published in the journal Science on June 27.

"With the Whitehead and Broad Institutes right next door to each other, I don't think there's any better place than this for a group of motivated people to move quickly and flexibly in the pursuit of academic science and medical technology," says Weissman, who is also a professor of biology at the Massachusetts Institute of Technology and an HHMI Investigator. "CHARMs are an elegant solution to the problem of silencing disease genes, and they have the potential to have an important position in the future of genetic medicines."

To treat a genetic disease, target the gene

Prion disease, which leads to swift neurodegeneration and death, is caused by the presence of misshapen versions of the prion protein. These cause a cascade effect in the brain: the faulty prion proteins deform other proteins, and together these proteins not only stop functioning properly but also form toxic aggregates that kill neurons. The most famous type of prion disease, known colloquially as mad cow disease, is infectious, but other forms of prion disease can occur spontaneously or be caused by faulty prion protein genes.

Most conventional drugs work by targeting a protein. CHARMs, however, work further upstream, turning off the gene that codes for the faulty protein so that the protein never gets made in the first place. CHARMs do this by epigenetic editing, in which a chemical tag gets added to DNA in order to turn off or silence a target gene. Unlike gene editing, epigenetic editing does not modify the underlying DNA -- the gene itself remains intact. However, like gene editing, epigenetic editing is stable, meaning that a gene switched off by CHARM should remain off. This would mean patients would only have to take CHARM once, as opposed to protein-targeting medications that must be taken regularly as the cells' protein levels replenish.

Research in animals suggests that the prion protein isn't necessary in a healthy adult, and that in cases of disease, removing the protein improves or even eliminates disease symptoms. In a person who hasn't yet developed symptoms, removing the protein should prevent disease altogether. In other words, epigenetic editing could be an effective approach for treating genetic diseases such as inherited prion diseases. The challenge is creating a new type of therapy.

Fortunately, the team had a good template for CHARM: a research tool called CRISPRoff that Weissman's group previously developed for silencing genes. CRISPRoff uses building blocks from CRISPR gene editing technology, including the guide protein Cas9 that directs the tool to the target gene. CRISPRoff silences the targeted gene by adding methyl groups, chemical tags that prevent the gene from being transcribed or read into RNA and so from being expressed as protein. When the researchers tested CRISPRoff's ability to silence the prion protein gene, they found that it was effective and stable.

Several of its properties, though, prevented CRISPRoff from being a good candidate for a therapy. The researchers' goal was to create a tool based on CRISPRoff that was just as potent but also safe for use in humans, small enough to deliver to the brain, and designed to minimize the risk of silencing the wrong genes or causing side effects.

From research tool to drug candidate

Led by Neumann and Bertozzi, the researchers began engineering and applying their new epigenome editor. The first problem that they had to tackle was size, because the editor needs to be small enough to be packaged and delivered to specific cells in the body. Delivering genes into the human brain is challenging; many clinical trials have used adeno-associated viruses (AAVs) as gene-delivery vehicles, but these are small and can only contain a small amount of genetic code. CRISPRoff is way too big; the code for Cas9 alone takes up most of the available space.

The Weissman lab researchers decided to replace Cas9 with a much smaller zinc finger protein (ZFP). Like Cas9, ZFPs can serve as guide proteins to direct the tool to a target site in DNA. ZFPs are also common in human cells, meaning they are less likely to trigger an immune response against themselves than the bacterial Cas9.

Next, the researchers had to design the part of the tool that would silence the prion protein gene. At first, they used part of a methyltransferase, a molecule that adds methyl groups to DNA, called DNMT3A. However, in the particular configuration needed for the tool, the molecule was toxic to the cell. The researchers focused on a different solution: instead of delivering outside DNMT3A as part of the therapy, the tool is able to recruit the cell's own DNMT3A to the prion protein gene. This freed up precious space inside of the AAV vector and prevented toxicity.

The researchers also needed to activate DNMT3A. In the cell, DNMT3A is usually inactive until it interacts with certain partner molecules. This default inactivity prevents accidental methylation of genes that need to remain turned on. Neumann came up with an ingenious way around this by combining sections of DNMT3A's partner molecules and connecting these to ZFPs that bring them to the prion protein gene. When the cell's DNMT3A comes across this combination of parts, it activates, silencing the gene.

"From the perspectives of both toxicity and size, it made sense to recruit the machinery that the cell already has; it was a much simpler, more elegant solution," Neumann says. "Cells are already using methyltransferases all of the time, and we're essentially just tricking them into turning off a gene that they would normally leave turned on."

Testing in mice showed that ZFP-guided CHARMs could eliminate more than 80% of the prion protein in the brain, while previous research has shown that as little as 21% elimination can improve symptoms.

Once the researchers knew that they had a potent gene silencer, they turned to the problem of off-target effects. The genetic code for a CHARM that gets delivered to a cell will keep producing copies of the CHARM indefinitely. However, after the prion protein gene is switched off, there is no benefit to this, only more time for side effects to develop, so they tweaked the tool so that after it turns off the prion protein gene, it then turns itself off.

Meanwhile, a complementary project from Broad Institute scientist and collaborator Benjamin Deverman's lab, focused on brain-wide gene delivery and published in Science on May 17, has brought the CHARM technology one step closer to being ready for clinical trials. Although naturally occurring types of AAV have been used for gene therapy in humans before, they do not enter the adult brain efficiently, making it impossible to treat a whole-brain disease like prion disease. Tackling the delivery problem, Deverman's group has designed an AAV vector that can get into the brain more efficiently by leveraging a pathway that naturally shuttles iron into the brain. Engineered vectors like this one make a therapy like CHARM one step closer to reality.

Thanks to these creative solutions, the researchers now have a highly effective epigenetic editor that is small enough to deliver to the brain, and that appears in cell culture and animal testing to have low toxicity and limited off-target effects.

"It's been a privilege to be part of this; it's pretty rare to go from basic research to therapeutic application in such a short amount of time," Bertozzi says. "I think the key was forming a collaboration that took advantage of the Weissman lab's tool building experience, the Vallabh and Minikel lab's deep knowledge of the disease, and the Deverman lab's expertise in gene delivery."

Looking ahead

With the major elements of the CHARM technology solved, the team is now fine-tuning their tool to make it more effective, safer, and easier to produce at scale as will be necessary for clinical trials. They have already made the tool modular, so that its various pieces can be swapped out and future CHARMs won't have to be programmed from scratch. CHARMs are also currently being tested as therapeutics in mice.

The path from basic research to clinical trials is a long and winding one, and the researchers know that CHARMs still have a way to go before they might become a viable medical option for people with prion diseases, including Vallabh, or other diseases with similar genetic components. However, with a strong therapy design and promising laboratory results in hand, the researchers have good reason to be hopeful. They continue to work at full throttle, intent on developing their technology so that it can save patients' lives not someday, but as soon as possible.