Astellas Pharma plans to develop next-generation adeno-associated virus (AAV) vectors for gene therapies targeting skeletal and cardiac muscle using Dyno Therapeutics’ CapsidMap™ platform, through a collaboration that could generate more than $1.6 billion for the George Church, PhD-co-founded Dyno.

CapsidMap—a delivery platform developed in Church’s lab at Harvard Medical School—uses artificial intelligence (AI) to design novel capsids that confer improved functional properties to AAV vectors.  At the core of CapsidMap, according to Dyno, are advanced search algorithms applying machine learning and the company’s large quantities of experimental data.

“Dyno’s CapsidMap platform is powered by machine learning models trained on our own experimental data, meaning it has the potential to produce optimized AAV capsids for any target tissue or cell type,” Eric Kelsic, PhD, Dyno Therapeutics CEO and co-founder, told GEN. “The Dyno capsids from this collaboration will be designed for improved skeletal muscle and heart targeting relative to existing AAV vectors, along with de-targeting of other organs.”

Kelsic added that the collaboration’s focus on skeletal and cardiac muscle was driven primarily by the Astellas strategic focus and existing expertise.

Muscle disease and gene therapy are two of Astellas’ 10 areas of therapeutic interest. The other eight are oncology, urology, nephrology, immunology, neuroscience, muscle diseases, ophthalmology, and vaccines. Astellas is also interested in digital health and other healthcare businesses it calls “Rx+®”.

“Astellas is on the leading edge of gene therapies for several muscle diseases, and so this collaboration is aligned with its strengths and current pipeline, as well as opportunities for further growth,” added Mathew Pletcher, PhD, SVP, division head of gene therapy research & technical operations, Astellas Gene Therapies. “The novel capsids from this collaboration aim to improve on existing AAV delivery vectors. When available it will be carefully evaluated and considered for all relevant projects.”

Astellas bolstered its gene therapy presence and capabilities when it acquired Audentes Therapeutics for $3 billion, a deal completed in January 2020. In March, Astellas integrated Audentes into its organization as Astellas Gene Therapies and established and Astellas Gene Therapies Center of Excellence, with a primary focus on genetic regulation R&D.

Assessing issues raised by AT132

Among programs overseen by Astellas Gene Therapies is AT132 (resamirigene bilparvovec), an AAV gene therapy candidate being developed to treat X-linked Myotubular Myopathy (XLMTM). In September, Astellas acknowledged that a fourth boy died in the troubled Phase I/II ASPIRO trial (NCT03199469), prompting the FDA to impose a clinical hold on the study.

The three previous patients, who all died last year, were treated with the trial’s higher dose of 3.5×1014 vg/kg before they started to demonstrate signs of liver dysfunction within 3 to 4 weeks after dosing. The fourth—whose age has not been disclosed—received the lower dose of 1.3×1014 vg/kg.

“Our investigation of the underlying mechanism for the SAE [serious adverse event] observed in the AT132 ASPIRO trial is ongoing.  It is too early to know whether this new technology can impact the specific hepatobiliary issues we have seen in the ASPIRO study,” Pletcher said. “Astellas is committed to explore novel capsids and understand the potential for improved targeted delivery of genetic material across a range of products and indications.”

In an October 29 investor presentation of results from its second quarter of Fiscal Year 2021 (three months ending September 30), Astellas disclosed that its planned Biologics License Application (BLA) filing for AT132 would be “delayed beyond FY 2022.”

Astellas said it anticipated additional participant data from a follow-up autopsy and follow-up analysis through December 2021, to be followed early in 2022 by the start of interactions with regulatory agencies “based on consultations with KOLs [key opinion leaders] and results from scientific investigations as well as additional participant data

“Future directions to be confirmed after consultations with regulatory agencies on the path forward for AT132,” Astellas stated, adding that it would offer an update at its R&D meeting scheduled for March 9.

Under its collaboration with Astellas, Dyno has agreed to design novel AAV capsids with improved functional properties for gene therapy using CapsidMap, and to experimentally characterize the properties of these vectors. Astellas in turn has agreed to oversee and conduct preclinical, clinical and commercialization activities for gene therapy product candidates using the novel capsids.

“Astellas is responsible for the design of the genetic payload along with preclinical, clinical and commercialization activities, including the manufacturing of gene therapy product candidates using the Dyno capsids,” Pletcher said.

$235M per product

Astellas has also agreed to pay Dyno $18 million upfront, additional payments of an undisclosed amount during the research phase of the collaboration, as well as clinical and sales milestone payments and royalties for any resulting products. The aggregate potential total value of future milestone and royalty payments to Dyno exceeds $1.6 billion, based on $235 million per product.

Astellas is the fourth biopharma giant to partner with Dyno on collaborations based on CapsidMap.

In October 2020, Roche and its Spark Therapeutics subsidiary agreed to explore CapsidMap-based gene therapies for central nervous system (CNS) and liver disorders—a collaboration that could potentially generate $1.8 billion-plus for Dyno. Five months earlier in May 2020, Novartis and Sarepta Therapeutics launched a combined up-to-$2 billion in collaborations to develop gene therapies for eye and muscle diseases, respectively, based on Dyno’s platform.

Earlier this year, Kelsic, Church, and a third Dyno co-founder, Machine Learning Team Lead Sam Sinai, PhD, joined six colleagues from the company, Church’s lab, Harvard and its Wyss Institute for Biologically Inspired Engineering, Google Research, and University of Cambridge in publishing a study detailing how they successfully used AI to design highly diverse capsid variants from an AAV virus, with the aim of identifying functional variants capable of evading the immune system.

“The key takeaway is, all of a sudden you can screen millions of variants without having to do a single experiment to figure out if it will package or not,” Sinai told GEN Edge in February. “One direction that we are working on in building models is a better ability to fine tune the product to make sure it succeeds in a particular group of patients and, eventually, a single patient.”

According to Dyno, CapsidMap is designed to integrate DNA library synthesis and next generation DNA sequencing to measure in vivo gene delivery properties in high throughput. The capsid platform is designed to expand the universe of diseases treatable via gene therapy by improving upon present-day AAV vectors, which are limited by delivery, immunity, packaging size, and manufacturing challenges.

Driven by delivery challenges

While AAVs have emerged as the safest and most popular delivery vehicle for gene therapy over the past decade, a significant population—one study put it at more than half of humans—has been reported to have pre-existing immunity to naturally occurring forms of those delivery vectors, and thus cannot benefit from AAV gene therapies, according to research that includes studies published in 2013 and last year.

“AAV has become arguably that the workhorse of most gene therapy trials and the clinic, but we do still see reports of adverse events and occasionally some fatalities in clinical trials. So that is an important driver behind this effort,” Church said recently on GEN Edge’s video interview series “Close to the Edge.”

Speaking with GEN Edge in October, Kelsic added: “Our mission is to solve the in vivo gene delivery problem because this is how we open access to new gene therapies and maximize improving patient quality of life.”

Kelsic, Church, four other co-founders established Dyno in 2018. Since then, Dyno has grown to 78 employees and is expanding into a newly built 42,000-square-foot site in Watertown, MA, 5 miles west of Cambridge, MA. The privately-held Cambridge-based company has raised a total $109 million, most of that consisting of the company’s $100 million Series A financing, completed in May.

“Since we signed our first partnerships last year with Novartis, Sarepta and Roche, and thanks to our Series A financing, we’ve made great progress and grown substantially,” Kelsic said. “Thanks to investments in experimental methods and software infrastructure, this year we’ve tripled our productivity. We’re making a billion distinct capsid measurements every two months, focusing on NHP [non-human primate] studies which are the most relevant for informing translation to humans.”