The Advances Behind The World's First Personalized CRISPR Treatment

It was huge news earlier this year: the first patient in the world, an infant, was successfully treated with a CRISPR gene editing therapy personalized to his genetic mutation. “Baby KJ” was diagnosed with a rare and dangerous metabolic disease shortly after his birth. Within a seven-month whirlwind of developing cell and mouse model systems, testing base editing variants, and assessing the safety of their intervention, the team led by Kiran Musunuru and Rebecca Ahrens-Nicklas administered the first dose of his personalized genetic medicine (Musunuru et al., 2025).

Much has been made of the speed with which KJ’s successful base editing intervention was created. Only about half of babies with KJ’s disorder survive long enough to receive a liver transplant — creating intense time pressure for the research team. Ultimately, the lifesaving effort was only possible because of a massive collaboration between research hospitals, universities, private companies, and federal regulators.

But the speed is impressive on a larger scale, too. The first paper reporting genome editing by CRISPR in mammalian cells appeared in 2013 (Cong et al., 2013; reviewed in Lander, 2016). The first base editor was reported three years later in 2016 (Komor et al., 2016). Advancing from this invention to KJ’s clinical application in under a decade is an astounding feat of public investment and collaborative scientific effort.

As we like to say at Addgene: sharing speeds science. Here, we’d like to highlight the advances in base editing that made this treatment possible — and may contribute to more in the future.

Figure 1: A timeline of CRISPR base editing advances that led to KJ’s treatment. Created with BioRender.com.

Invention of base editing

Base editing was invented in David Liu’s lab by combining Cas9 DNA-targeting technology with a fused cytidine deaminase to create C→T base edits at a target locus. A year later, an A→G base editor (ABE) followed based on similar principles (Gaudelli et al., 2017). These editors did not create double-stranded breaks in the DNA backbone, making them significantly less prone to indel mutations that plagued CRISPR editing at the time.

KJ was born with a C→T mutation on one allele of his CPS1 gene, which could be repaired with an A→G edit on the complementary strand. But the invention of base editing doesn’t necessarily mean researchers can fix any problematic C or A mutations in the genome. Off-target effects, editing efficiency, and target access were all challenges for these early base editors. Further innovations were needed!

Minimizing off-target editing

ABEs were originally developed from RNA deaminases, using directed evolution to get them to work on DNA instead. However, they retained a small amount of RNA-editing activity, which can cause off-target effects in a variety of RNA transcripts independent of Cas9 targeting (Rees et al., 2019). The Liu lab characterized this RNA-editing activity and developed a deaminase variant with a V106W mutation, which reduced RNA editing to background levels while retaining acceptable on-target DNA editing activity. In fact, it also reduced the (very low) levels of off-target DNA editing and on-target indel formation.

For the safety of KJ’s experimental treatment, it was critical to reduce off-target editing as much as possible, even at the cost of reducing editing efficiency. The V106W mutation was an important addition.

Improved activity of adenine base editors

The first ABE was about 50% efficient in human cells on average, but its effectiveness varied depending on the chromosomal location, sequence context, and cell type. Multiple labs worked to improve its effectiveness through directed evolution, introducing mutations to the deaminase subunit to produce a series of ABE8s that were faster, more efficient, and more flexible in their target preferences (Gaudelli et al., 2020; Richter et al., 2020). 

Five years later, three of these variants (ABE8.8, ABE8.20, and ABE8e) would be tested for their base-editing efficacy in cellular models of KJ’s CPS1 mutation. Every genetic target is slightly different, so more options mean better chances of identifying a successful editor. For this application, ABE8e turned out to be most effective.

Targeting the right PAM

Expanding PAM preferences

Base editors can only access a short sequence, between four and eight bases, close to the PAM recognized by Cas9. If no PAM is available nearby, a base editor may not be able to access an otherwise fixable mutation. To tackle this problem, Benjamin Kleinstiver’s lab engineered new variants of S. pyogenes Cas9. While the wild-type enzyme requires a PAM sequence of NGG (where N represents any base), their engineered variants are more permissive: SpG can target NGN PAMs, while SpRY can target both NAN and NGN PAMs effectively and NCN and NTN PAMs with lower efficiency — essentially allowing nearly any sequence to be targeted (Walton et al., 2020).

The initial screen for base editors that could target KJ’s mutation in cellular models tested both the SpG and SpRY variants of Cas9. With many PAMs available, a large number of gRNAs could be screened for the most efficient editing. In these initial experiments, SpG Cas9 came out on top.

Narrowing PAM preferences

The Kleinstiver lab also worked on the opposite project: engineering Cas9 variants with a preference for specific PAMs. Though near-PAMless Cas9s can broadly target many sites in the genome, they also have a higher risk of off-target editing and longer genome search times. Using a machine learning approach, the lab proposed multiple sets of Cas9 mutations that conferred new PAM preferences (Silverstein et al., 2025).

In cellular models of KJ’s mutation, SpG Cas9 + ABE8e performed the best when paired with the highest-efficiency sgRNA. The PAM sequence near this site is AGC, so in a secondary screen, the team tested Cas9 mutations that imposed a preference for an NGC PAM. Although all the tested mutations produced efficient editors, the Kleinstiver set performed best when paired with the V106W mutation (described earlier) to minimize off-target editing.

Base editing in the present—and future

The final base editor that was administered to KJ was designated NGC-ABE8e-V106W. The name is hardly flashy, but that mouthful reflects the number of different advances that were all combined to create a single safe and effective treatment.

And base editing advances won’t stop here. By coincidence, KJ’s genetic sequence was able to tolerate bystander edits—unintended changes to nearby target bases. Up to 5 bystander adenine bases were edited in cellular tests, in addition to the target base. However, all bystander edits produced synonymous changes, meaning the new codon corresponded to the same amino acid. Thus, bystander editing was deemed to pose low risk to KJ. This approach won’t be acceptable in every case, however, and recent work has already been published on improving base editing precision.

Similar treatments may one day have a smoother path to approval, as well. If the editor and delivery components could be rigorously safety tested and reused, only the RNA portion targeting base editors to problematic mutations would need to change to target different mutations in different patients. While these gRNAs would still need to be optimized and assessed for off-target risks, such a modular platform could accelerate the development of genetic therapies even further.

In the meantime, base editing will undergo even more improvements in the lab. We’re so happy to be a part of a collaborative community focused on CRISPR research, and we’re grateful for every scientist who uses Addgene to share their tools. We can’t wait to see what you create next!

---

References and Resources

References

Cong, L., Ran, F. A., Cox, D., Lin, S., Barretto, R., Habib, N., Hsu, P. D., Wu, X., Jiang, W., Marraffini, L. A., & Zhang, F. (2013). Multiplex genome engineering using CRISPR/Cas systems. Science (New York, N.Y.), 339(6121), 819–823. https://doi.org/10.1126/science.1231143 

Gaudelli, N. M., Komor, A. C., Rees, H. A., Packer, M. S., Badran, A. H., Bryson, D. I., & Liu, D. R. (2017). Programmable base editing of A•T to G•C in genomic DNA without DNA cleavage. Nature, 551(7681), 464–471. https://doi.org/10.1038/nature24644 

Gaudelli, N. M., Lam, D. K., Rees, H. A., Solá-Esteves, N. M., Barrera, L. A., Born, D. A., Edwards, A., Gehrke, J. M., Lee, S.-J., Liquori, A. J., Murray, R., Packer, M. S., Rinaldi, C., Slaymaker, I. M., Yen, J., Young, L. E., & Ciaramella, G. (2020). Directed evolution of adenine base editors with increased activity and therapeutic application. Nature Biotechnology, 38(7), 892–900. https://doi.org/10.1038/s41587-020-0491-6 

Komor, A. C., Kim, Y. B., Packer, M. S., Zuris, J. A., & Liu, D. R. (2016). Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage. Nature, 533(7603), 420–424. https://doi.org/10.1038/nature17946 

Lander, E. S. (2016). The Heroes of CRISPR. Cell, 164(1), 18–28. https://doi.org/10.1016/j.cell.2015.12.041 

Musunuru, K., Grandinette, S. A., Wang, X., Hudson, T. R., Briseno, K., Berry, A. M., Hacker, J. L., Hsu, A., Silverstein, R. A., Hille, L. T., Ogul, A. N., Robinson-Garvin, N. A., Small, J. C., McCague, S., Burke, S. M., Wright, C. M., Bick, S., Indurthi, V., Sharma, S., … Ahrens-Nicklas, R. C. (2025). Patient-Specific In Vivo Gene Editing to Treat a Rare Genetic Disease. New England Journal of Medicine, 392(22), 2235–2243. https://doi.org/10.1056/NEJMoa2504747 

Rees, H. A., Wilson, C., Doman, J. L., & Liu, D. R. (2019). Analysis and minimization of cellular RNA editing by DNA adenine base editors. Science Advances, 5(5), eaax5717. https://doi.org/10.1126/sciadv.aax5717 

Richter, M. F., Zhao, K. T., Eton, E., Lapinaite, A., Newby, G. A., Thuronyi, B. W., Wilson, C., Koblan, L. W., Zeng, J., Bauer, D. E., Doudna, J. A., & Liu, D. R. (2020). Phage-assisted evolution of an adenine base editor with improved Cas domain compatibility and activity. Nature Biotechnology, 38(7), 883–891. https://doi.org/10.1038/s41587-020-0453-z 

Silverstein, R. A., Kim, N., Kroell, A.-S., Walton, R. T., Delano, J., Butcher, R. M., Pacesa, M., Smith, B. K., Christie, K. A., Ha, L. L., Meis, R. J., Clark, A. B., Spinner, A. D., Lazzarotto, C. R., Li, Y., Matsubara, A., Urbina, E. O., Dahl, G. A., Correia, B. E., … Kleinstiver, B. P. (2025). Custom CRISPR-Cas9 PAM variants via scalable engineering and machine learning. Nature, 643(8071), 539–550. https://doi.org/10.1038/s41586-025-09021-y 

Walton, R. T., Christie, K. A., Whittaker, M. N., & Kleinstiver, B. P. (2020). Unconstrained genome targeting with near-PAMless engineered CRISPR-Cas9 variants. Science (New York, N.Y.), 368(6488), 290–296. https://doi.org/10.1126/science.aba8853

Additional resources on the Addgene Blog

• CRISPR 101: Cytosine and Adenine Base Editors
• Progress Towards a PAM-Free CRISPR
• CRISPR in the Clinic

Resources on Addgene.org

• Browse CRISPR plasmids for base editing
• Read our CRISPR 101 eBook
• Brush up with our CRISPR Guide