Repair Drive improves gene editing in the liver

Main text

The liver is a vital metabolic hub for the entire body, performing essential tasks in amino acid and glucose metabolism, drug detoxification, protein secretion, thrombosis, fibrinolysis, and lipid homeostasis. There are at least 700 monogenic diseases that could be treated with gene therapy in the liver. However, despite recent advances in genome editing, precise gene correction remains highly inefficient. In our recent work published in Science Translational Medicine,¹ we introduce a new strategy to address this challenge. Our approach harnesses the regenerative capacity of the liver to promote selective expansion of gene-corrected hepatocytes through the transient inhibition of an essential gene.

Targeted transgene insertion using adeno-associated virus (AAV) vectors is gaining traction for liver-directed gene therapies. Recent results indicate that episomal AAV vectors are prone to silencing and degradation in non-human primates.² Interestingly, some of the persistent transgene expression in liver gene therapy may actually derive from the chromosomal insertion of AAV genomes rather than episomal maintenance. By purposefully coupling AAV delivery with targeted transgene insertion via gene editing, therapies could become mutation independent, making them applicable to a broader range of patients with a particular disease. This approach would also ensure the transgene is inherited during cell division, addressing a major challenge in the treatment of newborns, pediatric patients, and individuals with recurring liver injury. Several attempts have been made to insert transgenes at the albumin (Alb) locus in humans using AAV, but these efforts have been hindered by inefficient co-delivery of editing enzymes and inherently low rates of homology-directed repair (HDR) in the mostly quiescent adult liver.^3,4

In certain genetic liver conditions, hepatocytes that acquire an integrated copy of a therapeutic transgene gain a natural selective advantage, allowing them to outcompete diseased cells, survive, and expand. This selective advantage has been reported in animal models of hereditary tyrosinemia type I, methylmalonic acidemia, Z-allele anti-alpha trypsin deficiency, Wilson disease, and a few others.⁵ In these cases, corrected cells can expand over time, sometimes even replacing the majority of the liver parenchyma. This selective advantage can dramatically lower the threshold of initial gene editing efficiency needed to achieve disease correction. Attempts have been made to artificially confer a selective advantage to hepatocytes for gene therapy or hepatocyte transplantation. One approach involves AAV insertion at the Alb locus for co-expressing therapeutic transgenes in cis with a short hairpin RNA against 4-OH phenylpyruvate dioxygenase (Hpd), which blocks an early step in tyrosine catabolism. This prevents the accumulation of toxic metabolites in edited hepatocytes following the chemical inhibition of fumarylacetoacetate hydrolase (Fah)—a downstream enzyme in the pathway.⁶ Li et al. used diphtheria toxin (DT) treatment to enrich cells carrying gene edits for inactivating the DT receptor and simultaneously inserting a therapeutic transgene.⁷ Vonada et al. used CRISPR-Cas9 to disrupt the cytochrome p450 reductase (Cypor), which protects edited cells from acetaminophen injury, enabling their selective expansion and repopulation of the liver.⁸

In our recent work, we developed “Repair Drive,” a platform to selectively expand gene-corrected hepatocytes through transient inhibition of an essential gene.¹ We used AAV delivery of CRISPR-Cas9 to insert a transgene cassette into the final coding exon of Apoa1. This cassette encodes the essential enzyme FAH alongside a tdTomato reporter, serving as a stand-in for a therapeutic protein. The Apoa1 gene, which encodes the main protein component of high-density lipoprotein, was chosen for its high promoter activity in hepatocytes.⁹ Cells that successfully underwent transgene insertion through HDR expressed both human FAH and tdTomato. Following gene editing, the mice received monthly doses of a GalNac-modified small interfering RNA (siRNA) against the mRNA of mouse Fah, which is essential for tyrosine catabolism and cell survival. Homology-directed-repaired cells, which express the human FAH enzyme non-targetable by the siRNA, not only survived but also proliferated to regenerate the liver. Liver conditioning with siRNA eliminated many unedited cells, evident by modest transaminase elevations, which returned to normal within 12 weeks. During this time, the proportion of homology-directed-repaired hepatocytes increased from 0.3% to 9%, demonstrating a clear selective advantage. Furthermore, we found that increasing selective pressure with a high protein diet drove hepatocyte expansion even further, reaching ∼25% of all liver cells (Figure 1).

Figure 1.

Repair Drive: A platform for selective expansion of gene-targeted hepatocytes

Top: a therapeutic transgene is inserted into the genome in tandem with a recoded version of an essential gene at a safe harbor site. Next, siRNA is used to knock down the essential gene in the entire liver, eliminating untargeted or incorrectly edited hepatocytes. Over time, cells with the therapeutic transgene divide and regenerate the liver. Bottom: liver sections from mice with HDR insertion of a tdTomato reporter at the Apoa1 locus, either with insertion alone (unselected) or following selection on a high-protein diet for 12 weeks (Repair Drive). The scale bar is 100 μm.

To evaluate the long-term safety of Repair Drive, we inserted a human factor IX transgene (FIX), which encodes the coagulation factor deficient in patients with hemophilia B. Mice undergoing Repair Drive for FIX were monitored for 1 year after AAV delivery. Repair Drive increased FIX secretion by the liver 5-fold relative to mice with transgene insertion alone, with stable expression levels throughout the study. In-depth pathological analysis of the livers revealed no signs of fibrosis or hepatocellular carcinoma in mice treated with Repair Drive. Likewise, tyrosine levels were only modestly elevated relative to germline Fah-knockout controls and returned to baseline before the end of the experiment. Importantly, Fah protein levels were fully restored across the entire liver after siRNA washout, not just in the gene-corrected cells.

We also performed a comprehensive examination of both on-target and off-target editing in mice undergoing Repair Drive, including detailed characterization of HDR-mediated insertion events using long-read sequencing. At the Apoa1 locus, CRISPR-Cas9 cutting resulted in heterogeneous insertion events, including correct HDR-mediated insertions, truncated HDR products, and whole AAV genomes by non-homologous end joining. We found that correct HDR at the 5′ end was critical for selective expansion of targeted Apoa1 alleles, and the proportion of cells with these edits was dramatically increased with Repair Drive. We also identified some off-target AAV integrations elsewhere in the genome, including previously described sites. Importantly, siRNA conditioning eliminated many incorrectly edited cells, as evident from the lower proportion of off-target AAV integrations in Repair Drive-treated mice.

Repair Drive offers several important advantages over previous approaches. First, it uses an endogenous essential gene for selection, avoiding the use of non-human drug resistance markers or toxic pharmacological agents. Second, the siRNA agent allows for temporary and controlled liver conditioning through recoding of the essential gene in the targeting cassette. Third, after conditioning, the expression of the essential gene (Fah) is restored across the entire liver. Lastly, normal liver physiology and function are preserved without permanent disruption of other metabolic pathways in the edited cells.

Future work on Repair Drive will be needed to translate this technology into new therapies. We used AAV as proof of concept to deliver both CRISPR-Cas9 and the transgene cassette. However, for clinical applications, transient delivery of Cas9 and guide RNA through non-viral systems, such as lipid nanoparticles, may be a safer and more scalable alternative. Beyond HDR, which is generally limited to the ∼1% of actively dividing hepatocytes in the liver, this approach could also be applied to other editing approaches. Repair Drive could conceivably be coupled with emerging RNA-templated insertion strategies. In addition, Repair Drive could be extended to other essential genes beyond Fah and optimized for smaller transgene configurations, broadening its therapeutic potential. An important next step will be demonstrating safety and efficacy in large animal models with human relevance. Ultimately, we envision that the Repair Drive platform will improve both liver-directed gene therapy and hepatocyte transplantation. By increasing the proportion of precisely repaired cells in the liver, we hope to significantly improve the effectiveness of gene editing therapies.

Acknowledgments

This work was supported by National Institutes of Health grants U42OD026645, U42 OD035581, DK124477, and HL132840 to W.R.L.; UG3 HL151545 to G.B. and W.R.L.; and R01HL169761 to G.B. This work was supported in part by grant #PNC24-260845 to M.D.G. from the American Association for the Study of Liver Diseases Foundation.

Declaration of interests

M.D.G. and W.R.L. have filed a patent application for the Repair Drive technology and its application to gene therapy: WO2021108269 (“Selective expansion of gene-targeted cells”).