Improving Genome Editing for Therapeutic Purposes

Although several genome editing technologies are available, issues such as random insertion, off-target effects and low efficiency can present challenges for their use in therapeutic applications. Tome Biosciences is working to overcome these limitations and develop a new class of genomic medicines using programmable genomic integration (PGI).

In this interview, John Finn, Tome’s chief scientific officer, tells us more about PGI, how it compares to other genome editing approaches and the benefits it could bring to genomic medicine discovery. Finn also shares Tome’s next steps and his thoughts on how the genomics industry could change in the next few years.

Anna MacDonald (AM): Can you explain what PGI is and how Tome's technology works?

John Finn (JF): PGI refers to the ability to insert any size DNA into a specific location in the genome. While there are many technologies that have been developed for gene insertion over the years, when we say PGI, there are some specific features that we are referring to. First, it has to be fully programmable, meaning that we specifically target where the DNA is inserted. This is in contrast to other integration systems such as lentivirus or most transposons/retrotransposons that randomly insert the cargo (usually with a preference for actively transcribed genes). This is also in contrast to other “safe harbor” approaches using integrases, recombinases or retrotransposons that have natural target sites in the human genome. 

At Tome, we believe that where you insert the DNA matters, and so with our technology, we can essentially decide which sites into which we integrate DNA. Second, the integration event is directional. Some integration approaches using double-stranded breaks (DSBs) can lead to targeted gene insertion, but they are not directional, meaning that half of the inserts go in the wrong direction. Lastly, the efficiency needs to be high enough that it is therapeutically relevant in the target cell. There are some emerging systems, such as CAST (CRISPR-associated transposases), which can efficiently insert DNA in a directional manner in bacteria, but the efficiencies to date in mammalian cells have been extremely low, and I am not aware of these complex systems showing activity in any therapeutically relevant primary human cells to date. 

Tome has a suite of technologies that allow us to do PGI. Our lead system, which is based on groundbreaking work from our co-founders (Omar Abudayyeh and Jonathan Gootenberg at Massachusetts Institute of Technology) takes advantage of the DNA targeting capability of Cas9 and combines it with the large DNA integration capability of integrases. This original technology, PASTE (programmable addition via site-specific targeting elements) uses a Cas9 nickase fused with a writing enzyme (reverse transcriptase) to write a landing site (we call this a “beacon”) at a specific location, and then with a provided DNA template and an integrase, the integrase integrates that large DNA template at the beacon.

One key feature of this system is that they chose an integrase that does NOT have a natural target site in the human genome, which was critical in the development of this PGI technology. This was a very elegant solution to a hard problem, as one of the remarkable features of integrases is that they do not care how big the DNA is. While you will always be limited by delivery (which is a key challenge for the field), the efficiency of integration is independent of the size of insert, and we have shown efficient PGI of over 30kb.

Since the inception of Tome, we have been laser-focused on translating this foundational work into medicine for patients in need. Along the way, we had to significantly optimize every aspect of the system, and have now developed integrase-mediated PGI (I-PGI). The concept is still the same, but now we use novel enzymes and guide architecture/modifications to write the beacon (using reverse transcriptase or ligase) and are using proprietary integrase proteins that have been specifically engineered to increase their activity and specificity. Using this three-enzyme system (Cas9 nickase, writing enzyme, integrase), we can now integrate large pieces of DNA in very specific locations, all without depending on a DSB. I see this as the last tool that we (the field) have needed in the gene editing toolbox.

Our second technology (ligase mediated PGI or L-PGI) is one of the newer tools in our toolbox that still uses the DNA targeting ability of Cas9, but instead of using writing enzymes, such as RNA or DNA polymerases to write the desired edit, we provide the cell with the pre-synthesized piece of DNA we want incorporated, and then use a ligase enzyme to specifically ligate that donor DNA at the target site. This technology is still early in development, but it has high efficiency in dividing and non-dividing cells and is good at smaller genomic integrations, anywhere from 1 to ~100 bases. Having L-PGI expands the type of edits that we can make and allows us to tailor the edit to the patient.

AM: Can I-PGI really target anywhere in the genome, or is it limited to PAM-sites like CRISPR?

JF: Although we still use Cas9 as our targeting system and so are PAM restricted, our strategy is not to target individual mutations (which greatly restricts the PAMs that can be used for editing), but insert healthy copies of genes in early introns of disease genes, providing us with a lot of genomic real estate (often thousands of base pairs) to find good target sites. The clinical advantage is it allows us to create one product to treat most, if not all, patients, independent of their particular mutation, rather than a mutation-specific approach. While as yet we have not found ourselves limited by PAMs given our intronic targeting strategy, there are so many Cas9 variants available with different PAMs, including some near PAM-less versions from the Kleinstiver lab, we really do think that it is possible to target any location in the genome. I think a more complex question is whether a particular sequence is accessible in a particular cell. I think the field is realizing that not every PAM can be targeted in every cell type, and this is likely a function of cell-intrinsic factors (chromatin accessibility, methylation, etc.).  

AM: Does the integrase not generate DSBs as well? Doesn't that mean I-PGI also risks translocations with multiplexing?

JF: This is one of the first questions I asked when I joined Tome. This is something that we have specifically investigated, and we do not see any evidence of the integrase making DSBs. This is likely because the mechanism that the integrase uses is very different from other nuclease or nickase-based systems. Briefly, the integrase binds to each attachment site as a dimer, and then those dimers align to form a tetramer. Each subunit nicks the DNA strand, but that DNA strand is now covalently bound to the subunit. The integrase subunits then rotate 180 degrees with respect to each other and then re-ligate the DNA strands. As the nicked DNA is covalently bound to the integrase, at no point during this process is there ever a DSB or free end. This lack of DSB opens the door for multiplexing without the risk of translocations. As for multiplexing, using different dinucleotide sequences, we have been able to simultaneously multiplex four pieces of DNA into four separate genomic locations.

AM: How does I-PGI compare to other technologies, such as CRISPR, in terms of safety?

JF: The current generation of CRISPR-based medicines are all dependent on DSBs, and while the jury is still out on whether these DSBs are a significant safety concern, I think it is likely best to avoid, or at least minimize, DSBs if possible. One clear theoretical safety advantage is related to CRISPR-related off-targets. As I-PGI and L-PGI use nickases, the risk of off-target mutations is significantly lower than early generation technology using cleavases. Our internal data is very much in line with previously published data using nickases (e.g., Prime editing) showing that it is actually very hard to detect off-target indels using nickases. 

There are also potential safety concerns with respect to how you deliver the editing components, independent of the actual editing modality. At Tome, we are using delivery methods where the editing machinery (nCas9, writing enzyme, integrase) is only expressed transiently, so that we can reduce both the potential risk of off-targets (which are a function of concentration and time), but more importantly, avoid flagging target cells for immune-mediated destruction. All enzymes currently being used for gene editing come from non-human organisms, and as such, are expected to generate robust immune responses if they hang around long enough. Having worked in the gene therapy field for almost 25 years, one lesson the field has learned over and over again is that you ignore the immune system at your peril. Given that it seems like the human immune system is more sensitive than any other animal, I think long term expression of any of these editing components (e.g., AAV-mediated expression) is a non-starter. While there may be some cases where long-term expression is theoretically possible (e.g., immune-privileged sites where the risk-benefit is very high), I see these as the exceptions. After the edit has been made, these enzymes only become a liability. 

AM: What other benefits does I-PGI offer? How can it help to overcome some of the current barriers in genomic medicine discovery?

JF: With respect to cell therapies, our ability to multiplex really opens the door to making very complex cells (e.g., logic gates, multiple transgenes, simultaneous knock out and insertion) in a single step. We have already been using these advantages to accelerate our cell therapy programs. To my knowledge, I-PGI is currently the only system available that can achieve programmable, high efficiency insertion of large DNA sequences, independent of whether the cell is dividing or not. Homology-directed repair (HDR) is the only other technology that can achieve high efficiency, programmable (and directional) inserts, but that technology is dependent on a DSB, and is only efficient in actively dividing cells. This means that it can’t be used to multiplex and can’t be used for in vivo gene editing, where the vast majority of target cells are quiescent. 

It also appears that Cas9-mediated DSBs can affect the viability of certain cells such as hematopoietic stem cells, which further limits the utility of HDR. I think that the ability to put a large piece of DNA in a very specific location will not only have an impact on the next generation of medicine, but will also accelerate the field of biomedical research. Look at how much the CRISPR revolution has advanced our understanding of biology and how cells work, and all of that was done primarily using CRISPR to knock out genes. Now that we can put any piece of DNA in any location, I will be surprised if we don’t see further advances in biomedical research.

AM: Can you tell us about some of the milestones that Tome plans to achieve in 2024?

JF: We are continuing to advance our PGI technologies and expect to begin to present some of our work at scientific conferences throughout 2024.

AM: What is the next phase in developing Tome's platform?

JF: Tome is currently in preclinical testing of both our integrative gene therapy as well as cell therapy platforms. We are also investing in novel modes of delivering our cargo since much of the industry is not focused on delivering gene-sized elements (as prior technologies do not have the capability of making edits on these size scales). Further, our discovery and engineering teams are always looking to enhance and simplify the PGI technologies. While Tome is laser-focused on translating our innovations in PGI into life-changing medicines for patients, we also endeavor to constantly lead and define the field when it comes to what we see as the maturation of genome editing for therapeutic purposes.

AM: Where do you see the genomics industry going in the next three to five years?

JF: I think that we will see many more CRISPR-based medicines being approved and as the impact of these medicines on patients’ lives starts to become evident and we get more safety data under our belt, the interest in gene editing will increase. I think we will start moving towards pediatric gene editing applications, where we can potentially cure children before they are even sick in the first place and will also see some movement towards gene editing for not only very rare monogenic diseases, but for more common diseases that affect a much larger percentage of the population.

John Finn was speaking to Anna MacDonald, Senior Science Editor for Technology Networks.

About the interviewee:

John brings over 20 years of gene therapy experience with a focus on genome editing and delivery technologies. He was most recently vice president of discovery research at Codiak Biosciences, where he led the development of a new class of therapeutics based on engineered exosomes. Prior to Codiak, John was executive director of platform biology and liver discovery at Intellia Therapeutics, where he was responsible for the development of viral and non-viral delivery systems and demonstrated the first in vivo systemic administration of CRISPR-based therapeutics. Previously, John served as director of research at Arthrogen, developing novel AAV and exosome-based therapeutics. He has served as an American Society of Gene and Cell Therapy committee member for multiple committees, including Nanoagents and Synthetic Formulations; Genome Editing; Immune Responses; and Metabolic and Inherited Disease. John trained with Pieter Cullis and Ian MacLachlan and received his PhD in biochemistry and molecular biology from the University of British Columbia and BS in molecular biology and genetics from the University of Guelph. He completed post-doctoral programs in immunology and vaccine research at McMaster University (Jonathan Bramson lab) and hematology and gene therapy at the Children’s Hospital of Philadelphia (Kathy High Lab).