Efficacy and long-term safety of CRISPR/Cas9 genome editing in the SOD1-linked mouse models of ALS

Abstract

CRISPR/Cas9-mediated genome editing provides potential for therapeutic development. Efficacy and long-term safety represent major concerns that remain to be adequately addressed in preclinical studies. Here we show that CRISPR/Cas9-mediated genome editing in two distinct SOD1-amyotrophic lateral sclerosis (ALS) transgenic mouse models prevented the development of ALS-like disease and pathology. The disease-linked transgene was effectively edited, with rare off-target editing events. We observed frequent large DNA deletions, ranging from a few hundred to several thousand base pairs. We determined that these large deletions were mediated by proximate identical sequences in Alu elements. No evidence of other diseases was observed beyond 2 years of age in these genome edited mice. Our data provide preclinical evidence of the efficacy and long-term safety of the CRISPR/Cas9 therapeutic approach. Moreover, the molecular mechanism of proximate identical sequences-mediated recombination provides mechanistic information to optimize therapeutic targeting design, and to avoid or minimize unintended and potentially deleterious recombination events.

Conflict of interest statement

The authors declare no competing interests.

Figures

Fig. 1. Schematic representation of the human SOD1-SpCas9 transgene.

a Human SOD1 structure. b A 20 bp DNA sequence (in red) of hSOD1-exon2 was used to express guide RNA, and this 20 bp sequence was inserted into the BbSI site of a plasmid vector, pSpCas9 (BB)-2A-GFP (PX458) (c). The plasmid DNA was digested with SacII and NotI. The 7.5 kb transgene was used for microinjection into mouse fertilized eggs. The protospacer adjacent motif (PAM) is shown in green (b). U6 U6 promoter, sgRNA single guide RNA containing a targeting sequence (crRNA sequence), and a Cas9 nuclease-recruiting sequence (tracrRNA), CAG a hybrid promoter , SV40NLS simian virus 40 nuclear localization signal, hSpCas9 a human codon optimized SpCas9, npNLS nucleoplasmin nuclear localization signal, T2A a viral 2A self-cleaving peptide, EGFP enhanced green fluorescent protein, bGH poly (A) bovine growth hormone poly (A) signal. Diagrams are not drawn to scale for clarity. d Kaplan–Meier plots showing the age of onset (black line, 109.5 ± 6.3 days) and cumulative survival (end stage, blue line, 141.5 ± 8.1 days) of G93A high expresser line (G1H, n = 14) and G1H/Cas9 double transgenic mice (green, n = 15). The G1H/Cas9 mice did not develop an ALS-like phenotype.

Fig. 2. ALS-free pathology in G1H/Cas9 mice.

The spinal cord sections were stained and analyzed using immunohistochemistry (a, b, d, e) or confocal microscopy (c, f) using the indicated antibodies. Representative images from an end stage G1H (#3702, 148 days) and hSOD1-targetd G1H/Cas9 mice (#7450, 196 days) are shown. Immunoreactive aggregates are indicated by arrows (a, b). c, f Microglial activation and astrocytosis are shown by Aif1 (green) and Gfap (red) staining, respectively. g, j H&E staining of the spinal cord sections showing motor neuron shrinkage (arrows) in the anterior horn of G1H mice. h, k Loss of large-caliber axons in an anterior root in G1H mice. Arrow indicates SOD1 aggregates in a motor axon. i, l H&E staining of the gastrocnemius muscles showing muscle fiber atrophy (green arrows) in the G1H mice. Scale bar, 50 µm. These pathological changes are absent in the G1H/Cas9 mice.

Fig. 3. Efficient targeting of hSOD1 in the G1H/Cas9 mice.

a Targeting events identified in a G1H/Cas9 mouse (#8190, 585 days). Among 112 individual clones analyzed, 19 different targeting events were identified. PAM sequence (TGG) is labeled in green. Red arrowheads indicate the Cas9 cleavage site. The deleted nucleotides are shown by red dashed lines. Red letters represent the inserted nucleotides. The number of clones harboring the indicated mutation is shown on the left. Individual mutations are on the right. For deletions exceeding six nucleotides, the deleted nucleotides are represented by numbers for clarity. b Efficient removal of hSOD1 in the G1H/Cas9 mice. Immunoblotting of the spinal cord homogenates from mice was performed with antibodies indicated on the right. β-actin and α-tubulin were used as internal loading controls.

Fig. 4. Off-target editing events in the G1H/Cas9 mice.

a, bhSOD1 sequence for Cas9 gRNA is on the top. DNA sequence, chromosomal position, number of nucleotide mismatch and bulge (Mi/Bu), and number of the identified mutants among the total number of the sequenced clones (Mut/clone) in 15 genome-wide potential off-target sites are shown. PAM sequences are shown in green. Nucleotide mismatches are shown in red. Nucleotide in blue indicates DNA bulge and (−) in blue indicates RNA bulge. Two G1H/Cas9 mice were analyzed (#8190, 585 days: a; #7449, 196 days: b). c Editing events identified in off-target #1. Arrowheads indicate the Cas9 cleavage site in off-target #1. Individual edited events are labeled on the top of sequencing chromatograms. Among 132 clones, three mutant clones were identified (delG, delAGinsGG, and delGTinsGAGTGGTCA) in mouse #8190, and among 156 clones, two mutant clones were identified (delG and delAAGG) in mouse #7449. The sites of mutations are indicated by arrows.

Fig. 5. PIS-mediated large deletions.

a Schematic representation of the cloning and enzymatic analysis of a 6.6 kb fragment from hSOD1 in G1H/Cas9 mice. The restriction enzyme sites used for analysis are labeled. Red arrowhead indicates the Cas9 cleavage site. Red arrows indicate plasmid vector DNA. Six types of clones were identified. Type 1 was Wt-like, the other five types showed smaller inserts. b–m Twelve deletions were identified, ranging from 325 to 4608 bp in size. Sequencing chromatogram of the DNA sequence around the junction is shown on the top. The PIS is color coded in red. The 5’ junction site is indicated by an arrow. Description of the deletions is based on the NCBI reference sequence (NC_000021.9) for human SOD1. n Schematic representation of deletions showing the sizes and locations (diagrams are not drawn to scale for clarity). Six Alu elements are shown by filled boxes, and the deleted regions are shown by lines. Deletions mediated by different PIS in the same Alu elements are shown on the top of the lines. The total number of deletions are summarized on the right, and detailed in Table 1.

Fig. 6. Loss of hSOD1 transgene copy number in G1H/Cas9.

The hSOD1 transgene copy number was determined by using multiplex ligation-dependent probe amplification (MLPA). a Five mouse-specific MLPA probes were designed to hybridize to two sex chromosome-specific (Nr0b1 and SRY) and three autosomal control (Ext1, Ep300, and Crebbp) genes. Two distinct MLPA probes were designed to hybridize to intron 1 (probe 1) and intron 2 (probe 2) of hSOD1 beyond the known deleted regions. The PCR amplification products generated by each probe differ in size by at least four nucleotides (nt). Nontransgenic mouse DNA (NTg mDNA) in the b absence and c presence of human DNA (1:1) was used to illustrate hSOD1 probe specificity. hSOD1 transgene copy number in the d G1H (n = 4) and e G1H/Cas9 (n = 5) mouse lines were estimated by dividing the average peak height of hSOD1 (probe 1) by the average peak height of the three mouse control genes (e, f). This ratio was then normalized to the ratio of mouse nontransgenic DNA spiked with human DNA. Three independent MLPA reactions were performed for each condition and the standard deviation (SD) was plotted to illustrate variation. hSOD1 transgene copies in G1H were estimated to be ~30 (29.7 ± 2.31). The lowest and the highest hSOD1 transgene copies remained in G1H/Cas9 mice were estimated to be ~9 (8.6 ± 0.25) and 23 (23.3 ± 1.28), respectively.