Eric Kelsic: Origins of Dyno and the Future of Gene Therapy (Nucleate Insights #1)

Published in
8 min readApr 6, 2022


Welcome to the Nucleate Insights program.

Nucleate is a student-led nonprofit agency for future biotech leaders. Among our educational initiatives is Insights, which invites life sciences trainees to engage in deep-dive discussions with company founders who transformed academic projects into thriving biotechnology companies.

In these journal club-style discussions, we uncover what helped founders fully realize and grow the commercial potential of their original academic projects.


Eric Kelsic, CEO of Dyno Therapeutics

Insights’ first guest was Eric Kelsic, CEO and co-founder of Dyno Therapeutics, a company at the frontier of gene therapy using artificial intelligence. Prior to founding Dyno, Eric led a team that developed the technology underlying Dyno’s AI-powered capsid engineering platform in George Church’s lab. In 2019, he and his colleagues published Comprehensive AAV capsid fitness landscape reveals a viral gene and enables machine-guided design in Science, forming the basis for Dyno Therapeutics.

Eric Kelsic, Ph.D., CEO at Dyno Therapeutics

Under Eric’s leadership, Dyno has raised $109M in VC financing, including a 2021 Series A led by a16z, grown to 75 employees, and signed partnerships with world-leading gene therapy companies (Novartis, Sarepta, Roche, Spark and Astellas). We joined Eric to discuss the landmark Science paper and the basis for Dyno, summarized below.

Scientific Context

Gene therapy holds a grand promise in a simple premise: cure patients with genetic diseases by delivering a functional version of their faulty or absent gene. The simple premise hides a scientific and technological obstacle course requiring major innovative leaps across the last half-century. One such innovative leap is known as Adeno-associated virus (AAV), a virus-like particle discovered by Bob Atchinson, M. David Hoggan, and Wallace Rowe in 1965.

In the 60 years since its discovery, AAV has emerged as one of the major platforms studied and optimized for in vivo gene therapy. A strength of AAV for this application is its structure: AAV is a non-enveloped virus composed of a protein capsid and surrounding a programmable single-stranded DNA genome payload. Subtle differences in capsid proteins can alter surface interactions to dictate which cell types the AAV will target. This is immensely important for a gene therapy, since not only must the correct gene sequence be introduced, but that it must also do so to the correct target tissue.

AAV-based clinical trials now span disease indications across ophthalmology, neurology, hematology, musculoskeletal and metabolic targets. Each of these requires unique AAV specifications. Trials are aiming to join the short list of two currently FDA approved AAV-based gene therapies (both from Novartis’ program: Luxturna, approved in 2017; and Zolgensma, in 2019).

Around the same time of these FDA approvals, a group in George Church’s Lab was working to gain a deeper understanding of AAV biology. The group, led by Dr. Eric Kelsic, focused in on the AAV2* viral capsid. They exhaustively characterized all single codon mutations, mapping the capsid fitness landscape with respect to multiple in vivo gene delivery-relevant parameters. The result was an algorithmically designed and experimentally validated capsid library. Remarkably, they exceeded viability when compared to libraries generated by random mutagenesis approaches. The group also revealed the importance of spatially defined capsid residues for determining tissue biodistribution.

Their work culminated in the group’s 2019 Science paper, which formed the conceptual basis for Dyno Therapeutics.

*(AAV2 is a particular serotype of AAV with broad tissue tropism, infecting many different tissue types rather than a single tissue)

Key Insights

1. Concept to Company

Proof: Before company formation, Dr. Kelsic wanted to prove, first to himself, and then to the broader gene therapy community, that machine-guided AAV design was a worthwhile idea. The experiments and logical flow of the paper were an extension of the points that he felt needed to be validated before progress could be made. The experiments were designed to stress-test the robustness of a comprehensive platform technology for AAV design. Specifically, he aimed to demonstrate improvements over other mutagenesis approaches. Thoughtful controls and designs gave him the proof of principle that he (and others) needed to see before attempting further experiments.

Intuition: The publication was well-timed with Novartis’ FDA-approvals. Dyno’s technology seemed the perfect response to what the gene therapy community was hoping for: better, safer, more precise engineered vectors. The timing and storytelling involved in getting Dyno off the ground was key to its success. The field’s reaction to the publication gave intuitive momentum to company formation.

Support: It is not coincidental that Dr. Kelsic found himself in the Church Lab, with its strong record of spinning out companies from academic projects. This mentorship, paired with tech transfer infrastructure and seed funding from Harvard and the Wyss Institute, propelled Dyno into existence.

2. The Differentiating Value

Know Your Niche: Early on, Dyno knew their unique differentiating value was “machine-guided design.” Whereas competitors were tied to a limited, or even singular, parameter of AAV design such as phylogeny or structure, Dyno leveraged their data science expertise to carve out a unique approach within a competitive space. Even acknowledging the imperfection of their approach, Dyno still highlights their unique value. According to Dr. Kelsic, “No model is going to be perfect, but by bringing it all into the data layer, now you can pay attention to the data that is most meaningful to translation.” He notes that, “You don’t have to worry about how to build the perfect physical model, you just need to know how to analyze data across multiple models.”

In summary, Dyno’s unique value is in their ability to work across many parameters at once, optimizing AAV design based on clinical need.

3. Leading the narrative

The Message: For Dyno, getting the word out about their work ahead of publication was a strategic plan. While this raised the specter of being “scooped,” they stood to gain from establishing their leadership in a new conversation about machine-guided AAV design. The team secured a provisional patent ahead of publication and spoke freely about their work publicly at conferences.

“When talking to investors, their saying no isn’t a bad thing. It might just mean that you aren’t telling the story in a way that makes sense to them or speaks to what is different.”

The Audience: In leading the narrative about their technology, Dyno was able to 1) dictate many of the talking points about their unique value, and 2) drum up momentum for their fundraise. When pitching to investors, Dyno was in a unique position within the gene therapy space — pitching a platform rather than a product. Dr. Kelsic and his team took investors’ skepticism in stride, explaining, “when talking to investors, their saying no is not a bad thing. It might just mean that you aren’t telling the story in a way that makes sense to them or speaks to what is different.” The key was helping investors understand Dyno’s differentiating value.

4. Multiplexed philosophy

Science: At its core, Dyno conceptualizes AAV design as a multiplexed problem, perfectly suited to machine-guided solutions. The future of protein engineering, to Dr. Kelsic, is a landscape of multiplex measurements. This core philosophy drives the company’s innovations.

Partnerships: Dyno approaches their business model with a similar philosophy. Instead of focusing their platform on a narrow disease indication, Dyno has formed partnerships with multiple clinical companies to multiplex their patient reach. Dyno partnered with Novartis, Roche, Sarepta Tx, Spark Tx, and Astellas, aiming their platform across a breadth of disease applications.

5. Smart partnerships

Choose Wisely: In choosing early partners, Dyno focused on three criteria.

  1. Groups that enhance the credibility of the platform.
  2. Those that demonstrate commitment to patients.
  3. Those with a proven track record of experience.

Dr. Kelsic mentioned the importance of making wise decisions about internal partnerships as well. As CEO, Dr. Kelsic no longer has a daily involvement at the heart of the science. He emphasizes that putting together an excellent team of scientists allowed him to place trust in his team and focus on what he sees as his role now: facilitating the success of his team, steering the ship, and building strong partnerships.

6. Looking forward

Staying Ahead: Dr. Kelsic acknowledges that AAV may not be a premier gene therapy technology “forever,” but he does believe that AAV is in its technology window and that that window will be rather long. In sticking with the original goal of building a machine-guided design platform, Dyno is, in a sense, also well positioned to pivot toward different applications of their platform.

“Multiplexed measurement is the future of protein engineering.”

At the forefront of synthetic biology, Dyno is optimizing AAV design across multiple parameters simultaneously. These parameters include: tissue targeting, immune evasion, packaging size, and manufacturing. As he looks ahead, Dr. Kelsic is convinced that “multiplexed measurement is the future of protein engineering, and the majority of information will be coming through data itself rather than mechanistic insights.”

Learn more

Thank you to all who participated in this vibrant discussion! To dig deeper into the details, check out the full recording of this Insights session.

To participate in a future session of Insights, please apply here:

Ogden PJ, Kelsic ED, Sinai S, Church GM (2019) Comprehensive AAV capsid fitness landscape reveals a viral gene and enables machine-guided design. Science 366: 1139–1143 DOI: 10.1126/science.aaw2900


Jessica Roginsky is a PhD student in the lab of Dr. Susan Kaech at the Salk Institute / UCSD, where she studies immune cell interactions in the brain and its barriers.



About Nucleate
Nucleate is a nonprofit organization dedicated to empowering the next generation of biotech leaders, with chapters spanning 10 regions and participation from over 60 academic institutions. Nucleate identifies future biotech entrepreneurs, removes barriers, and helps founders concentrate on building transformational technologies.

Nucleate’s programs are made possible thanks to our generous sponsors: (platinum) Alnylam, Genentech, Pillar; (gold) Alexandria LaunchLabs, Benchling, Emerald Cloud Labs, Morrison & Foerster; (silver) Latch Bio, Watershed; and philanthropic support from Schmidt Futures. Visit for our regional sponsors and more information.

About Dyno Therapeutics
Dyno is a pioneer in applying artificial intelligence to gene therapy. The company’s powerful and proprietary genetic engineering platform is designed to rapidly and systematically develop improved AAV capsids that redefine the gene therapy landscape. Dyno was founded by experienced biotech entrepreneurs and leading scientists in the fields of synthetic biology, gene therapy, and machine learning. Dyno is located in Watertown, Massachusetts and is actively hiring for local and distributed roles. Visit for additional information.

Eric Kelsic, Ph.D. is CEO & co-founder of Dyno Therapeutics. Under Eric’s leadership, Dyno has raised $109M in VC financing, including a 2021 Series A led by a16z, grown to 75 employees, and signed partnerships with world-leading gene therapy companies (Novartis, Sarepta, Roche, Spark and Astellas). Dyno was named Xconomy’s 2020 Startup of the Year and Eric was recognized as one of Endpoint’s 20 under 40 next-gen biotech leaders in 2021. Prior to founding Dyno, Eric led a team to develop the technology underlying Dyno’s artificial intelligence powered capsid engineering platform in George Church’s lab at the Wyss Institute of Harvard Medical School. There he measured the first comprehensive fitness landscape of the adeno-associated virus (AAV) capsid protein and co-discovered the AAV MAAP gene. He earned a PhD in Systems Biology from Harvard University and a BS in Physics from Caltech.