The State of Myostatin Inhibition
MSTN, the gene encoding myostatin, is one of the most famous genes in the world. It’s hard to find a popular gene editing article that doesn’t reference myostatin.
History
The gene encoding myostatin was discovered by Se-Jin Lee and Alexandra McPherron in 1997, 22 years ago. Similar to telomeres, we’ve had grand visions of myostatin’s usefulness for decades but no way to successfully affect it. The myostatin knockout mouse model (mighty mouse) created by Se-Jin Lee is one of the most famous animal models ever created.
The Belgian Blue Cow and Bully Whippets often accompany articles on myostatin. These mammals naturally possess mutations that inhibit their myostatin.
Some humans also have mutations that inhibit their natural myostatin. Other than super-human physique, these individuals appear to be completely healthy. This makes myostatin a lucrative target for many disease indications, including muscle disorders, muscle loss due to aging, and even metabolic diseases.
According to Se-Jin Lee, “It’s an evolutionary vestige that we don’t need… that’s the beauty of targeting this pathway for therapeutic uses.”
Hitting the Target
There are several axes to consider for successfully inhibiting myostatin (GDF8).
What tools are we going to use to hit the target?
- small molecule
- antibody
- protein
- In vitro knockout: CRISPR, ZFN, TALENs
- In vivo gene over-expression: AAV
- In vivo knockdown: AAV-CRISPR, siRNA, lentiviral RNAi, etc.
There are other tools, but these are the most common ones.
What mechanism of action are we going to take advantage of to inhibit GDF8?
- Block the activin receptor type IIB (ActRIIB) binding site to prevent GDF8 from binding to it
- Bind to dimerized GDF8 to prevent it from binding to ActRIIB
- Prevent GDF8 from dimerizing and thus becoming activated
- Prevent GDF8 from becoming cleaved and thus becoming activated
- Prevent GDF8 from being produced
Progress
Companies like Pfizer, Novartis, Eli Lilly, and Regeneron have been focusing on antibodies. Wyeth Pharmaceuticals, now a part of Pfizer, started the first GDF8-inhibiting clinical trial in 2005. Novartis started recruiting for a Phase 2 clinical trial for type 2 diabetes using its GDF8 antibody in 2016.
A list of TGF-Beta drugs and their clinical trials can be found here.
Acceleron focuses on making proteins that bind to TGF-beta family members. Their first candidate, ACE-031, also bound to BMP9, another ligand in the same protein family, and caused bleeding. Their more recent candidates avoid targeting BMP9.
One of the primary challenges with protein therapies is dosing and half-life. Follistatin (FST344) is a strong inhibitor of GDF8 with a recombinant half life of 90 minutes. Eli Lilly developed a half-life extended FST344 protein but ultimately couldn’t commercialize it due to manufacturing issues.
Brian Kaspar and Jerry Mendel’s work that lead to Milo Biotechnology uses local intramuscular injections of AAV1 to over-express follistatin (FST344). FST344 binds not only GDF8, but also activin proteins, BMP proteins, and GDF11. They successfully showed local muscle growth but unfortunately the effects were too local to be therapeutically beneficial.
Immusoft is also making use of follistatin. They’re delivering it with their B-cell therapy. The challenge will be producing enough of the protein to see a therapeutic effect in humans. Follistatin when expressed systemically also has undesirable effects like prolapsed anus in mice and potential cancerous effects.
CRISPR, one of the newest tools, has also been used to inhibit GDF8 in vivo. The primary concern with CRISPR for an in vivo knockdown is the potential off-target mutations and cas protein immune responses.
Why Isn’t It Working?
Despite developing antibodies with high GDF8 specificity, no therapy has achieved FDA approval.
The effect sizes observed in human clinical trials have been considerably smaller than those observed in mice. The in vivo inhibition of GDF8 in mice routinely produces 30% increases in muscle mass, whereas 1–3% increases in muscle mass are seen in clinical trials of GDF-8 inhibiting antibodies in humans. Pharma holds the belief that activin A plays a larger role than GDF8 in muscle regulation in humans, but according to Se-Jin Lee, “I don’t buy it.” Mice with inhibited GDF8 see an additional 60% muscle mass growth when activin A is additionally inhibited. This is also potentially in conflict with situations where we see hyper musculature in humans due to naturally occurring GDF8 mutations.
Another complicating factor is the potential for diminishing returns. Some in vivo experiments where GDF8 is inhibited show a 10x increase in blood circulating levels of GDF8. This indicates that inhibiting GDF8 may cause its pathway to be upregulated. The more GDF8 is inhibited, the more inhibition may be necessary to produce a phenotype response.
Dr. Tom Thompson at the University of Cincinnati emphasizes the need to model and test expectations, whereas most companies in the space have been focused on random screens. Crystal structures of GDF8, GDF11, and other TGF-beta complexes have been published only in the past few years, enabling more precise modeling and prediction of these structures.
Choice of indication also plays a large role in the success of these therapies. The results of clinical trials aren’t comparable because they’re matrix of molecules and indications akin to comparing apples and oranges.
Given the field’s setbacks, we see a path forward.
Hypotheses
- Modeling protein structure and binding affinities will greatly speed up the development of lead targets.
- Constant production of GDF8 inhibiting molecules via gene therapy will have a higher efficacy than short half-life therapies like antibodies with high dose dependency.
- Indication is just as important as molecule. A suitable indication will have an intact or up-regulated TGF-beta pathway.
Special thanks to Anukana Bhattacharjee.
Thanks to Se-Jin Lee, Tom Thompson, Lee Sweeney, and George Church.
