Genetic and functional correction of argininosuccinate lyase deficiency using CRISPR adenine base editors

Abstract

Argininosuccinate lyase deficiency (ASLD) is a recessive metabolic disorder caused by variants in ASL. In an essential step in urea synthesis, ASL breaks down argininosuccinate (ASA), a pathognomonic ASLD biomarker. The severe disease forms lead to hyperammonemia, neurological injury, and even early death. The current treatments are unsatisfactory, involving a strict low-protein diet, arginine supplementation, nitrogen scavenging, and in some cases, liver transplantation. An unmet need exists for improved, efficient therapies. Here, we show the potential of a lipid nanoparticle-mediated CRISPR approach using adenine base editors (ABEs) for ASLD treatment. To model ASLD, we first generated human-induced pluripotent stem cells (hiPSCs) from biopsies of individuals homozygous for the Finnish founder variant (c.1153C>T [p.Arg385Cys]) and edited this variant using the ABE. We then differentiated the hiPSCs into hepatocyte-like cells that showed a 1,000-fold decrease in ASA levels compared to those of isogenic non-edited cells. Lastly, we tested three different FDA-approved lipid nanoparticle formulations to deliver the ABE-encoding RNA and the sgRNA targeting the ASL variant. This approach efficiently edited the ASL variant in fibroblasts with no apparent cell toxicity and minimal off-target effects. Further, the treatment resulted in a significant decrease in ASA, to levels of healthy donors, indicating restoration of the urea cycle. Our work describes a highly efficient approach to editing the disease-causing ASL variant and restoring the function of the urea cycle. This method relies on RNA delivered by lipid nanoparticles, which is compatible with clinical applications, improves its safety profile, and allows for scalable production.

Keywords: ABE; ASL; CRISPR; LNP; argininosuccinic aciduria; genetic defect; genome editing; lipid nanoparticles; liver; mRNA.

Graphical abstract

Figure 1

hiPSC differentiation into hepatocyte-like cells (A) Timeline for the 18-day differentiation protocol. The basic media employed for each stage is depicted in bold letters (BASAL1, CDM3, CDM4, and CDM5). The supplements incorporated in each stage are listed below the basic media. (B) mRNA levels of essential hepatocyte marker genes. Representative mRNA samples from different individuals, differentiation batches, stages (day 0, 2.5, 7, 13, and 18), and with different genotypes (control = HEL24.3, edited, and not edited) were analyzed by qPCR. The mRNA levels are expressed in fold change and normalized to the HepG2 commercial hepatocarcinoma cell line (illustrated with a yellow line at the fold change 1 on the y axis when the scale allows it). Each point represents an independent hiPSC line, which we consider a biological replicate. Data are represented as the mean ± SEM. Statistical significance based on Tukey test; p > 0.05 (ns, not significant), p < 0.05 (^∗), p < 0.01 (^∗∗), p < 0.001 (^∗∗∗), p < 0.0001 (^∗∗∗∗). (C) Representative immunocytochemistry images of day 18 hepatocyte-like cells. Hoechst is depicted in blue, AFP in red, HNF4α, and albumin in green. All the images were acquired and processed with the same settings to allow comparison. The white bar represents 100 μm. (D) Western blot for ASL, albumin, and actin. Representative protein samples from different individuals, stages (day 0, 2.5, 7, 13, and 18), and with different genotypes (control = HEL24.3, edited, and not edited) were imaged at the same time and using fluorescent antibodies for western blot.

Figure 2

Rescue of the metabolic phenotype after ABE-mediated edition of the ASLD c.1153C>T variant (A) Urea cycle diagram. (B) mRNA levels of ASL. Representative mRNA samples from different individuals, differentiation batches, stages (day 0, 7, 13, and 18), and with different genotypes (control = HEL24.3, edited, and not edited) were analyzed by qPCR. The mRNA levels are expressed in fold change and normalized to the HepG2 commercial hepatocarcinoma cell line (illustrated with a yellow line at the fold change 1 on the y axis). Each point represents an independent hiPSC line, which we consider a biological replicate. Data are represented as the mean ± SEM. Statistical significance based on Tukey test; p > 0.05 (ns, not significant), p < 0.05 (^∗), p < 0.01 (^∗∗), p < 0.001 (^∗∗∗), p < 0.0001 (^∗∗∗∗). (C and D) The sum of the absolute abundance of all the metabolites detected by LC-MS in the cell lysate (C) and the media (D). Each shape represents independent differentiation batches (circle, square, diamond, triangle). We employed day-18 hiPSC-derived hepatocyte-like cells from two different individuals. We analyzed two independently edited hiPSC lines per proband (four biological replicates), two not edited independent hiPSC lines per proband (four biological replicates), and HEL24.3 as a control (two biological replicates). We processed five technical replicates of each sample in the LC-MS. The sum of the absolute abundance of all the metabolites in each sample was employed as a normalization to calculate the relative abundance of individual metabolites. Data are represented as the mean ± SEM. Statistical significance based on Tukey test; p > 0.05 (ns, not significant), p < 0.05 (^∗), p < 0.01 (^∗∗), p < 0.001 (^∗∗∗), p < 0.0001 (^∗∗∗∗). (E–H) Relative abundance of intracellular and media ASA (E and F) and citrulline (G and H) detected by LC-MS in the same samples described in the previous graph. Each shape represents independent differentiation batches (circle, square, diamond, triangle). As expected, in some of the control and edited samples, the ASA levels were below the detection limit of the mass spectrometer. We did not consider these values for the graph bar or the statistical analysis, but we illustrated these cases with a shape containing a black center. Relative abundance is the absolute abundance value normalized to the sum of all metabolites. Data are represented as the mean ± SEM. Statistical significance based on Tukey test; p > 0.05 (ns, not significant), p < 0.05 (^∗), p < 0.01 (^∗∗), p < 0.001 (^∗∗∗), p < 0.0001 (^∗∗∗∗).

Figure 3

Editing efficiency, toxicity profile, and rescue of the metabolic phenotype in primary fibroblasts after lipid nanoparticles ABE8e treatment (A) Diagram of the lipid nanoparticle contents: an sgRNA targeting the ASL c.1153C>T variant plus an RNA cassette for ABE8e expression. The spacer section of the sgRNA, which targets the DNA, is written from base 1 to base 20. The PAM, not included in the sgRNA oligo, corresponds to bases 21–23 (GGG). (B) On target A-to-G editing efficiency. We employed primary fibroblasts from two different individuals. We independently treated these fibroblasts in duplicates (four biological replicates) with eight different doses (0–5,100 ng RNA) of three types of lipid nanoparticle ABE8e (mRNA-1273, BNT162b2, and ALN-18328). One week after the treatment, we estimated the on-target A-to-G editing efficiency by analyzing the Sanger sequence data through EditR. The solid lines represent the mean of each treatment, and each data point is individually represented. (C) Sequencing data from the experiment in point B is shown here in detail, considering the on-target (green) and bystander (orange) A-to-G editing efficiency. We found bystander editing just in the adenine position 11 but not in position 12. Data are represented as the mean ± SEM. A mid-high dose of ABEmax with no replicates was added as a comparison showing the editing efficiency of the previous generation of ABE. (D and E) We independently treated primary fibroblasts from two different individuals in duplicates (four biological replicates) with three different doses (17, 850, and 5,100 ng RNA) of three types of lipid nanoparticle ABE8e (mRNA-1273, BNT162b2, and ALN-18328). We followed the fibroblast populations for 70 h, taking pictures in the Incucyte every 2 h. The lipid nanoparticle ABE8e treatment was applied at the 2 h time point. We assessed the confluency (D) and the number of dead cells (E) estimated by the CytoTox dye. The solid lines represent the mean of each treatment. For simplicity, the error bars (SD) are shown only for the non-treated samples (in black). For more detailed data, please check Figure S5. (F) Direct ASL enzyme activity in fibroblast cell lysates. We employed cell lysates from two fibroblast populations derived from each of the two healthy donors (control, four biological replicates), three ASLD fibroblast populations independently treated with ABE8e mRNA-1273 LNPs targeting the ASL variant (edited, three biological replicates), and two not-treated fibroblast populations per each of the two individuals (not edited, four biological replicates). Each sample was processed in two technical replicates. Within each treatment, the data points illustrated with the same shape represent a technical replicate of the enzyme activity assay using the same fibroblast lysate. For the edited group, the triangle and the circle correspond to a lipid nanoparticle dose of 85 ng of RNA, whereas the square corresponds to a dose of 1,700 ng of RNA. Data are represented as the mean ± SEM. Statistical significance based on Tukey test; p > 0.05 (ns, not significant), p < 0.05 (^∗), p < 0.01 (^∗∗), p < 0.001 (^∗∗∗), p < 0.0001 (^∗∗∗∗). (G and H) Relative abundance of ASA in the cell lysate (G) and the media (H) detected by LC-MS. We independently treated primary fibroblasts from two different individuals in duplicates (four biological replicates) with 85 ng RNA of the mRNA-1273 lipid nanoparticle ABE8e plus the variant-targeting sgRNA (LNP-ABE8e), or with vehicle (not treated), or with mRNA-1273 lipid nanoparticle ABE8e containing the sgRNA site_16 targeting an unrelated locus (non-targeting sgRNA). As a control, we used fibroblasts coming from three healthy donors of different genders and ages (healthy donors, three biological replicates). Two weeks after the treatment, we analyzed the metabolite content of each condition, processing five technical replicates of each sample in the LC-MS. As expected, in some of the control and edited samples the ASA levels were below the detection limit of the mass spectrometer. We did not consider these values for the graph bar or the statistical analysis, but we illustrated these cases with a shape containing a black center. Relative abundance is the absolute abundance value normalized to the sum of all metabolites. Data are represented as the mean ± SEM. Statistical significance based on Tukey test; p > 0.05 (ns, not significant), p < 0.05 (^∗), p < 0.01 (^∗∗), p < 0.001 (^∗∗∗), p < 0.0001 (^∗∗∗∗).