During a call with Alltrna co-founder and CIO Theonie Anastassiadis and CEO Michelle C. Werner, I’m immediately called upon to dust off my Biology 101 notes to grasp the significance and scope of their transfer RNA (tRNA) therapeutic platform.
“Humans only need 47 different tRNAs to decode mRNA into protein,” said Anastassiadis. “There are over 600 different tRNAs genes in the human genome and tRNAs are the most abundant RNA in the cell. They’re the most modified and most diversely modified with over 120 [potential] modifications. That creates a foundation for this really interesting programmable tRNA.”
Even if I could remember all that, the part that has me stumped is this: how in the world would a tRNA therapeutic work?
Anastassiadis and Werner offer several interesting characteristics of tRNA therapeutics. First, tRNA therapeutics can, theoretically, correct mutations that are found in the genetic makeup of people’s cells without actually touching the genetics. Second, they can also govern translation dynamics by changing how much of a particular protein is expressed. And third, tRNAs break off into smaller fragments that have novel roles beyond the translation of proteins (a topic for another day).
Enter at your own risk
The inception of Alltrna dates back to 2017–18 when Anastassiadis stumbled across tRNA biology during “an exploration” at Flagship Pioneering—the first stage of the venture creation process. “We wondered why no one else had done this before,” said Anastassiadis. “We quickly realized that it’s actually a pretty tough field to break into.”
At the time, there were few technologies to probe tRNA biology, including expressing tRNAs, chemical synthesis with modifications, and simple quantification. The first thing that Anastassiadis, with the help of Lovisa Afzelius, an origination partner at Flagship and a co-founder of Alltrna, embarked on was a mission to build proprietary tools that became the foundation of their tRNA platform to be able to do basic, proof-of-concept experiments.
“We did that in 2019, and then in 2020 we were series A-funded, which allowed us to build out the fantastic team that we’ve assembled today as well as grow out the platform and our capabilities,” said Anastassiadis.
In 2022, Alltrna filled multiple leadership roles, including Werner’s appointment to a dual role at Flagship and Alltrna. Werner was previously an executive at AstraZeneca, including stints as global franchise head in hematology and head of U.S. oncology.
Last August, Joanne Protano, formerly senior vice president of finance and operations at Rubius Therapeutics, was named CFO. Alltrna also announced the appointment of Caroline Köhrer, PhD, as vice president of discovery platform in October. Köhrer spent 20 years at MIT studying protein synthesis across all kingdoms of life. Today, of Alltrna’s eight members of the leadership team, seven are women.
Agnostic to a t(RNA)
When thinking about the initial application, Alltrna has been looking at patients in a very different way. Rather than looking at the diagnosis of a specific disease, the company considers a specific mutation that spans multiple diseases. By pooling patients with these shared common mutations that exist from disease to disease, it opens up a new opportunity for Alltrna to identify and treat these patients.
“The real power of our platform is to design and deliver these tRNAs as medicines to restore a full-length functional protein. We can do this to be able to address what we’re calling ‘stop codon disease,’ which represents approximately 30 million patients around the world,” said Werner. Many genetic diseases including cystic fibrosis, Duchenne muscular dystrophy, β-thalassemia, and many types of cancers, can be caused by the presence of premature stop mutations in RNA transcripts. Werner declined to reveal specific indications that Alltrna is or plans on pursuing but talked more generally about Alltrna’s approach.
“Instead of having a full-length functional protein, you would have a truncated or mutated protein, which is really what causes [much] disease at the end of the day. Our plan is to engineer modified tRNAs that are specifically designed to read through these premature termination codon mutations. So, instead of stopping prematurely, the translation process continues all the way to the end, when you would expect it to stop, resulting in a full-length functional protein.”
As Anastassiadis and Werner assert that any tRNA therapeutic could be used to treat patients with different indications, I’m curious to understand how this technology achieves any specificity. Wouldn’t the same tRNA target premature stop codons on different mRNAs? How would this work without creating various unintended translation problems?
“The context for a natural termination code (NTC) versus a premature termination code (PTC) is quite different,” Anastassiadis explained. “An NTC has evolved to be a very strong stop, as opposed to a PTC, which is a random mutation in a spot that should not be terminated.” Recall that translation of polypeptides doesn’t cease at a stop codon; binding proteins in the polyA tail recruit release factors that terminate translation. So, there is preferential recruitment of release factors at that site, and tRNAs are not a good substrate for that termination reaction.
In contrast, ribosomes moving through the elongation step are constantly sampling the space with tRNAs. That’s where Alltrna’s modified tRNAs will be preferentially incorporated to continue translation through that PTC as opposed to release factors. This naturally creates a huge preferential difference in tRNA incorporation at the PTC versus NTC and a natural therapeutic window.
The reveal at ASGCT
Alltrna isn’t the first company to wade into this space. After Alltrna’s series A in 2020, three startups—ReCode Therapeutics, Shape Therapeutics, and Tevard Biosciences—collectively raised $240 million to develop tRNA-based therapeutics. Another company, hC Bioscience, subsequently announced that it has raised $24 million in Series A financing to develop tRNA-based therapies, backed by Takeda. Against the backdrop, PTC Therapeutics, which was founded in 1998 (and stands for post-transcriptional control and not premature terminal codon), continues to look at nonsense mutations.
But what Anastassiadis and Werner think sets Alltrna apart is that they are focused on learning the design rules of tRNAs to be able to program certain features into them. Last week at the American Society of Gene and Cell Therapy (ASGCT) conference in Los Angeles, Alltrna presented some preliminary data about the company’s platform, which is enabled by machine learning (ML) that can design, modify, produce, and deliver engineered tRNA oligonucleotides with dramatically increased potency and activity.
“We’re making tremendous progress with this unique machine learning-driven platform,” said Werner. “We’re optimizing for sequence and modifications of our tRNAs, which is really important because this is the secret sauce to turning our molecules into medicines, which is really going to be important when we move into that clinical development plan.”
“This is the first time that a modified engineered tRNA oligonucleotide is showing the ability to demonstrate in vivo this universal readthrough, and we’ve done this with a mouse model,” continued Werner. “We’re demonstrating that we can read through premature termination codons independent of the gene and location of the mutation, which speaks to the ability for us to be able to use this as a modality to have a single medicine that could treat many different diseases.”
Alltrna has data showing that their machine learning can sequence-optimize tRNAs, significantly increasing engineered tRNA activity. “One of the other things that we’re learning is how we can further expand out that therapeutic window using sequence and chemical modifications in our tRNAs,” said Anastassiadis. “We’ve also built a very thorough molecular profiling platform that allows us to use many tools, including proprietary tools we’ve developed in-house, to be able to monitor any kind of NTC read-through or any extra protein product that would be generated as well.”
“Not all tRNAs are created equal, and in fact, not all engineered tRNAs are created equal,” said Werner. “The power of optimizing using this machine learning engine for both sequence and modification is a crucial component to ensuring that we have the best development candidates to move into the clinic.”
Alltrna has built cellular models of disease reporters, testing the same tRNA across 24 different models, including 14 different genes. Then, using a combination of several different locations across the gene, they show that the same tRNA can actually read through PTCs, agnostic to the gene and location within a gene.
“This is exciting because the other novel modalities that are coming forward today—whether gene therapies or gene editing technologies, which are major advancements in the field—each one of those has to take a very specific gene-by-gene, disease-by-disease, and often mutation-by-mutation approach to be able to address the patient population,” said Anastassiadis.
Werner wants to leverage Alltrna’s engineered tRNAs across different diseases, independent of the gene, and independent of the location and the mutation. “We have the opportunity to unify the populations that we have access to,” said Werner. “We’re not just limited to looking at common diseases, we can be selecting diseases that are both rare and common in nature because we don’t have to worry about small patient populations and whether or not there’s an opportunity that is meaningful and relevant to be working towards.”