In vivo adenine base editing rescues adrenoleukodystrophy in a humanized mouse model

Abstract

X-linked adrenoleukodystrophy (ALD), an inherited neurometabolic disorder caused by mutations in ABCD1, which encodes the peroxisomal ABC transporter, mainly affects the brain, spinal cord, adrenal glands, and testes. In ALD patients, very-long-chain fatty acids (VLCFAs) fail to enter the peroxisome and undergo subsequent β-oxidation, resulting in their accumulation in the body. It has not been tested whether in vivo base editing or prime editing can be harnessed to ameliorate ALD. We developed a humanized mouse model of ALD by inserting a human cDNA containing the pathogenic variant into the mouse Abcd1 locus. The humanized ALD model showed increased levels of VLCFAs. To correct the mutation, we tested both base editing and prime editing and found that base editing using ABE8e(V106W) could correct the mutation in patient-derived fibroblasts at an efficiency of 7.4%. Adeno-associated virus (AAV)-mediated systemic delivery of NG-ABE8e(V106W) enabled robust correction of the pathogenic variant in the mouse brain (correction efficiency: ∼5.5%), spinal cord (∼5.1%), and adrenal gland (∼2%), leading to a significant reduction in the plasma levels of C26:0/C22:0. This established humanized mouse model and the successful correction of the pathogenic variant using a base editor serve as a significant step toward treating human ALD disease.

Keywords: ABCD1; CRISPR; adenine base editing; adrenoleukodystrophy; genome editing; humanized mouse model; very-long-chain fatty acid.

Graphical abstract

Figure 1

Characterization of the humanized ALD mouse model (A) Schematic overview of the process used to KI the human cDNA at the mouse Abcd1 locus (5′ UTR) with TILD-CRISPR. A human ABCD1 (hABCD1) cDNA (2,778 bp) donor sequence harboring the c.1534G>A mutation (red triangle) flanked with mouse left and right homology arms is depicted at the top. The blue arrows indicate the forward (F) and reverse (R) primers used in qRT-PCR to analyze hABCD1 mRNA expression; the dark red and orange sets of arrows indicate the primers used to analyze the 5′ and 3′ junction sites, respectively, by PCR. LHA, left homology arm; UTR, untranslated region; RHA, right homology arm; mutABCD1, mutated ABCD1; TILD, target integration with linearized dsDNA; KI, knockin. (B) PCR analysis of hALD pup (F1) genotypes with different sets of primers (labeled “In-In,” “Out-In,” and “In-Out”; see details in Figure S1A) used to confirm the hABCD1 cDNA KI in the mouse Abcd1 locus. Red numbers indicate pups containing the KI; diagnostic bands of the expected size are marked with red arrows. Pups 2 and 3 (marked with red rectangles) were used for further breeding and subsequent experiments. Genomic DNAs from WT mice and F0 pup 20 were used as negative and positive controls, respectively, for this analysis (see also Figures S1C and S1D). (C) Expression of the human ABCD1 gene in the brain and testis of WT and hALD mice was evaluated with qRT-PCR. The values, normalized to beta-actin and Gapdh expression in brain and testis, respectively, are shown as fold changes compared with the expression in hALD mice, presented as the average ± SEM (n = 2/genotype; ∗p < 0.05, two-tailed unpaired Student t test). (D) Relative levels of VLCFAs in different tissues of 9-month-old WT and hALD mice. C22:0, C24:0, C24:0/C22:0, C26:0, and C26:0/C22:0 abundance in the brain, liver, and testis were measured by LC-MS/MS; abundance in hALD mice is expressed relative to that in WT. Data are shown as average ± SEM (n = 4/genotype; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001).

Figure 2

Analysis of prime editing and adenine base editing for correcting the ABCD1 c.1534G>A mutation in vitro (A) Schematic representation of the high-throughput screening of pegRNA activities. A lentivirus library was generated from oligonucleotides containing pairs of pegRNA-encoding sequences and corresponding target sequences with oligo-specific barcode sequences, and then transduced into PE2-expressing HEK293T cells to generate a cell library. The cell library underwent puromycin selection for 7 days, after which genomic DNA was extracted. Target sequences with barcodes were then PCR amplified and deep sequenced to measure the editing efficiency associated with each pegRNA (see also Figure S4). (B) Prime editing efficiencies were analyzed in two independent experimental replicates. Among the 97 tested pegRNAs, 25 were associated with more than 100 barcode reads and were included in the final analysis; the three pegRNAs with the highest average editing efficiencies are indicated with red arrows and labeled with their IDs. (C) pegRNA and sgRNA sequences. The target sequence for correction of the ALD mutation is shown in bold letters, and the hABCD1 c.1534G>A mutation is highlighted in red. The spacers for three sgRNAs described in the text are indicated by the green arrows. “TGTTC” PAM sequence for the ABE are underlined. The three pegRNAs from the library that exhibited the highest efficiencies are indicated with arrows (5′–3′) and labeled with their IDs; their common spacer sequence is indicated by the blue arrow. (D) Schematic overview of the split-NG-ABE8e(V106W) expression plasmids (AAV-N-intein_NG-ABE8e(V106W) and AAV-C-intein_NG-ABE8e(V106W) with sgRNA) used for AAV production. Both the N- and C-terminal halves of ABE8e(V106W) were expressed under the control of the CMV promoter, whereas the sgRNA was expressed using the U6 promoter. CMV, cytomegalovirus enhancer; chicken-beta, chicken beta-actin promoter; NLS, nuclear localization signal; TadA-8e v106w, tRNA adenine deaminase-8e v106w; NT-nCas9, N terminus of Cas9 nickase; WPRE, woodchuck hepatitis virus posttranscriptional regulatory element; bGH, bovine growth hormone polyadenylation signal; CT-nCas9, C terminus of Cas9 nickase; sgRNA, single guide RNA; U6, human U6 promoter. (E) The on-target A-to-G editing efficiencies of NG-ABE8e(V106W) with the indicated sgRNA variants and PE2max with the indicated pegRNAs were determined in patient-derived fibroblasts by deep sequencing (n = 2 for each group). Error bars represent the mean ± SEM. (F) The activity of the selected gx19 sgRNA was evaluated at eight potential off-target sites in patient-derived fibroblasts using deep sequencing (n = 2 for each group). Error bars represent the mean ± SEM.

Figure 3

Therapeutic adenine base correction in hALD mice after delivery of NG-ABE8e_V106W with AAV (A) Schematic of the experimental workflow for AAV-mediated delivery of variable doses of ABE8e(V106W) and sgRNA to determine the gene-editing efficiency in 6-week-old humanized model mice. Six weeks after the injection, the mice were euthanized and different organs were analyzed. (B) To determine the adenine base-editing efficiency in specific areas of AAV-injected mice, the brain and spinal cord were respectively divided into five (A to E) and two (upper and lower) different regions. The regions of the brain are indicated with gold rectangles labeled A–E and are referred to as “Brain A” to “Brain E”. (C and D) Efficiencies of the intended base correction (A to G) in different organs from AAV- (high and low dose) or phosphate-buffered saline-treated (uninjected, negative control) mice were analyzed in genomic DNA (C) and cDNA (D). The percentage of edited reads was determined by targeted deep sequencing. Error bars represent the mean ± SEM. n = 3 or 2 per group.

Figure 4

Therapeutic effect of AAV-PHP.B-mediated adenine base editing with NG-ABE8e_V106W on plasma VLCFA levels in hALD mice (A and B) LC-MS/MS was utilized to quantify the relative levels of C26:0/C22:0 (A) and C26:0 (B) in the plasma of 12-week-old male mice. The samples analyzed from the AAV-injected mice are identical to those used for the targeted deep sequencing in Figures 3C and 3D. hALD-AAVL, hALD-AAV Low Dose; hALD-AAVH, hALD-AAV High Dose. VLCFA C26:0 levels were normalized to levels of the long-chain fatty acid C22:0 and are represented as the C26:0/C22:0 ratio. The data are presented as fold changes compared with levels in WT mice. Mean ± SEM is used to depict the data. Each group was analyzed using n = 3 mice. Statistical analysis was performed through a one-way ANOVA followed by Dunnett’s multiple comparisons test, with a single pooled variance (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001).