Towards somatic cell genome editing therapies
Quantitative design rules are needed to inform the translation of next-generation genome editors. A new platform - combining human in vitro and in silico models - offers novel insights into the design of efficacious therapeutic strategies for both monogenic and polygenic diseases.
Bioengineering CRISPR-based genome editors could produce several newly-approved, therapeutic products in the next few years. Indeed, promising results from clinical trials of CRISPR approaches are steadily being reported, with one for Sickle Cell Disease and β-Thalassemia reported just a few days ago1. Despite this impressive progress, nearly all therapeutic strategies thus far have focused on editing one gene with a single injection or infusion. Even strategies to treat Sickle Cell Disease and other recessive disorders target a single DNA sequence inherited from both parents. What if a patient has inherited different variants from their parents, with each variant contributing to their disease? Could multiple doses over several months provide additional efficacy, or alternatively lead to adverse events arising from cumulative unintended genomic damage? What if a more precise editor can increase safety but is difficult to deliver into the cell's nucleus?
For these more complex scenarios, my lab set out to explore these and related questions using both experimental and computational approaches2 (Figure 1). Our journey started in attempting to design genome editors for a compound heterozygous recessive disorder, Pompe disease. Mutations in Pompe patients are distinct in most cases, separated by thousands of bases, and therefore multiple editors would need to be delivered to edit both mutations. Using patient-derived pluripotent stem cell models, we systematically corrected both mutations in a single cell using CRISPR-Cas9 based nanoparticles, previously developed by my lab3. These editors tether the donor repair template to the nuclease to achieve higher precision. Careful and quantitative evaluation of Pompe diseased cellular dysfunction revealed potent rescue of phenotypes by gene correction. However, this nanoparticle strategy more than doubles the size of the editor that needs to be delivered into the body for somatic cell editing. Plus, there are additional risks of unintended genomic consequences of having two different editors within the same cell. We next wanted to further evaluate these strategies in vivo within animal models.
Figure 1. A combined in vitro and in silico platform for evaluating genome editing strategies2. Schematic indicating the modeling approach in which samples from patients are collected ex vivo and then genome-edited in vitro. Genotypes and phenotype outcomes from the in vitro studies are inputs for an in silico model that simulates the delivery of the therapeutic in vivo as well as tissue morphogenesis. The results of the in silico model can ultimately guide dosing and formulation decisions.
Here, we hit a bottleneck: testing our strategies in relevant animal models was not feasible with existing resources. While several Pompe disease models are available and well-characterized, nearly all have been produced by large disruptions to the mouse Gaa gene which do not provide relevant human genomic targets to edit. Producing new animal models with human genomic targets was daunting: recent work to produce a compound heterozygous Pompe disease model in mice4 did not fully recapitulate disease processes observed in Pompe patients. Generating a new animal model for each of the different human mutations observed in patients would take years, if not decades. Many rare diseases have a wide spectrum of mutations scattered across one or more associated genes, and the dearth of animal models frequently stifles the development of new therapeutic strategies.
Therefore, we pivoted to building an in silico model of a human infant to evaluate the delivery of various genome editors. Our computational model is incredibly flexible, as we can simulate, within hours, the delivery and dosing of many different editors to developing infants. This work quantitatively identified tradeoffs for precision, delivery, and progenitor cell targeting for any editor in theory - including traditional Cas9 nucleases and newly-developed base editors. As more empirical data with new editors emerge in the field, these in silico models could be trained further to build on this framework and make more accurate predictions with the model.
Overall, this platform could create "digital twins" of the patient (e.g., in cardiology5) to enable rapid simulation of many different dosing and editing strategies. This information could complement the data gleaned from animal studies on the immune response and adverse physiological responses from the delivery of an editor. Altogether, it is likely that a wide variety of tools will be required to generate data for filings to regulatory agencies involving first-in-human trials with editors, as recently demonstrated for the EDIT-101 trial6. Data from human stem cell models have already been used to provide critical data for moving forward with gene therapy approaches7, and similar approaches are likely fruitful for genome editing strategies. As collective efforts in academia and industry mature in the field (e.g., the NIH SCGE and NIST Genome Editing Consortium), there is a bright future ahead with a greater ecology of tools that could move somatic cell editing strategies towards therapeutic applications.
For further reading: Carlson-Stevemer, Jared, Amritava Das, Amr Abdeen, David Fiflis, Benjamin Grindel, Tugce Akcan, Tausif Alam, et al. 2020. “Design of Efficacious Somatic Cell Genome Editing Strategies for Recessive and Polygenic Diseases.” Nature Communications 11, 6277. https://www.nature.com/articles/s41467-020-20065-8; PDF file; press release
1. Frangoul, Haydar, David Altshuler, M. Domenica Cappellini, Yi-Shan Chen, Jennifer Domm, Brenda K. Eustace, Juergen Foell, et al. 2020. “CRISPR-Cas9 Gene Editing for Sickle Cell Disease and β-Thalassemia.” The New England Journal of Medicine, no. NEJMoa2031054 (December). https://doi.org/10.1056/nejmoa2031054
2. Carlson-Stevemer, Jared, Amritava Das, Amr Abdeen, David Fiflis, Benjamin Grindel, Tugce Akcan, Tausif Alam, et al. 2020. “Design of Efficacious Somatic Cell Genome Editing Strategies for Recessive and Polygenic Diseases.” Nature Communications 11, 6277. https://www.nature.com/articles/s41467-020-20065-8
3. Carlson-Stevermer, Jared, Amr A. Abdeen, Kaivalya Molugu, Krishanu Saha, Lucille Kohlenberg, Madelyn Goedland, Meng Lou, and Krishanu Saha. 2017. “Assembly of CRISPR Ribonucleoproteins with Biotinylated Oligonucleotides via an RNA Aptamer for Precise Gene Editing.” Nature Communications 8 (1): 1711. https://doi.org/10.1038/s41467-017-01875-9
4. Huang, Jeffrey Y., Shih-Hsin Kan, Emilie K. Sandfeld, Nancy D. Dalton, Anthony D. Rangel, Yunghang Chan, Jeremy Davis-Turak, Jon Neumann, and Raymond Y. Wang. 2020. “CRISPR-Cas9 Generated Pompe Knock-in Murine Model Exhibits Early-Onset Hypertrophic Cardiomyopathy and Skeletal Muscle Weakness.” Scientific Reports 10 (1): 10321. https://doi.org/10.1038/s41598-020-65259-8
5. Corral-Acero, Jorge, Francesca Margara, Maciej Marciniak, Cristobal Rodero, Filip Loncaric, Yingjing Feng, Andrew Gilbert, et al. 2020. “The ‘Digital Twin’ to Enable the Vision of Precision Cardiology.” European Heart Journal, March. https://doi.org/10.1093/eurheartj/ehaa159
6. Maeder, Morgan L., Michael Stefanidakis, Christopher J. Wilson, Reshica Baral, Luis Alberto Barrera, George S. Bounoutas, David Bumcrot, et al. 02 2019. “Development of a Gene-Editing Approach to Restore Vision Loss in Leber Congenital Amaurosis Type 10.” Nature Medicine 25 (2): 229–33. https://www.nature.com/articles/s41591-018-0327-9
7. Duong, Thu T., Vidyullatha Vasireddy, Pavitra Ramachandran, Pamela S. Herrera, Lanfranco Leo, Carrie Merkel, Jean Bennett, and Jason A. Mills. 2018. “Use of Induced Pluripotent Stem Cell Models to Probe the Pathogenesis of Choroideremia and to Develop a Potential Treatment.” Stem Cell Research 27 (March): 140–50. https://doi.org/10.1016/j.scr.2018.01.009