Robust activation of microhomology-mediated end joining for precision gene editing applications

Abstract

One key problem in precision genome editing is the unpredictable plurality of sequence outcomes at the site of targeted DNA double stranded breaks (DSBs). This is due to the typical activation of the versatile Non-homologous End Joining (NHEJ) pathway. Such unpredictability limits the utility of somatic gene editing for applications including gene therapy and functional genomics. For germline editing work, the accurate reproduction of the identical alleles using NHEJ is a labor intensive process. In this study, we propose Microhomology-mediated End Joining (MMEJ) as a viable solution for improving somatic sequence homogeneity in vivo, capable of generating a single predictable allele at high rates (56% ~ 86% of the entire mutant allele pool). Using a combined dataset from zebrafish (Danio rerio) in vivo and human HeLa cell in vitro, we identified specific contextual sequence determinants surrounding genomic DSBs for robust MMEJ pathway activation. We then applied our observation to prospectively design MMEJ-inducing sgRNAs against a variety of proof-of-principle genes and demonstrated high levels of mutant allele homogeneity. MMEJ-based DNA repair at these target loci successfully generated F0 mutant zebrafish embryos and larvae that faithfully recapitulated previously reported, recessive, loss-of-function phenotypes. We also tested the generalizability of our approach in cultured human cells. Finally, we provide a novel algorithm, MENTHU (http://genesculpt.org/menthu/), for improved and facile prediction of candidate MMEJ loci. We believe that this MMEJ-centric approach will have a broader impact on genome engineering and its applications. For example, whereas somatic mosaicism hinders efficient recreation of knockout mutant allele at base pair resolution via the standard NHEJ-based approach, we demonstrate that F0 founders transmitted the identical MMEJ allele of interest at high rates. Most importantly, the ability to directly dictate the reading frame of an endogenous target will have important implications for gene therapy applications in human genetic diseases.

Fig 1. MMEJ is a unique DSB repair pathway that results in highly efficient and highly stereotyped mutagenesis.

DSB by conventionally designed Programmable Nucleases typically proceeds through a versatile yet unpredictable classical non-homologous end joining (NHEJ) pathway. As a result, a rather diverse cohort of mutant alleles are generated, making the subsequent selection process labor intensive to enrich for the allele of interest. The resulting genetic composition of the specific loci are often complex, requiring careful molecular characterization of each allele. Efficient activation of microhomology-mediated end joining (MMEJ) pathway, on the other hand, can greatly limit allelic diversity and enable the intentional generation of a particular deletion allele of interest at a high rate. Consequently, the downstream applications become more streamlined with facile generation of homozygous frameshift allele in diploid cells.

Fig 2. PreMA TALEN reagent can be used to recapitulate previously reported loss-of-chrd-function phenotype in 1 dpf F0, injected larvae.

A. Top–Wildtype chrd sequence with TALEN binding sites annotated in teal. The dotted red boxes are MH arms predicted to be used most frequently. Raw sequence alignment of the whole PCR amplicon demonstrates that the majority of reads are the expected 7 bp deletion allele. Bottom–summary data from subcloning analyses. 50% of the mutant allele recovered were of the predicted MH allele. B. Previously reported chrd loss-of-function phenotype was successfully recapitulated using this TALEN pair. Phenotype severity was graded by the degree of Intermediate-Cell-Mass expansion in the tail and by the reduced head size by 1 dpf. Box plot demonstrating phenotypic penetrance is provided with each experiment denoted by a unique marker shape. N = 3 biological and technical replicates. At least 29 injected animals were scored in each experiment.

Fig 3. PreMA sgRNA against tyr can be used to recapitulate loss-of-melanophore phenotype in 2 dpf, injected F0 larvae.

A. Top–Wildtype tyr sequence with the #2 sgRNA target site annotated in green. The dotted red boxes are MH arms predicted to be used most frequently. Raw sequence alignment of the whole PCR amplicon demonstrates that the majority of reads are the expected 4 bp deletion allele. Bottom–summary data from subcloning analyses. 88% of the mutant allele recovered were of the predicted MH allele. B. Previously reported tyr loss-of-function phenotype was successfully recapitulated using this CRISPR-Cas9. Phenotype severity was graded by the loss of retinal pigmentation. Partial loss of retinal pigmentation was considered a Weak phenotype, whereas complete loss of pigmentation in one or both eyes were considered Moderate and Severe phenotypes, respectively. Box plot demonstrating phenotypic penetrance is provided with each experiment denoted by a unique marker shape. N = 3 biological and technical replicates. At least 12 injected animals were scored in each experiment.

Fig 4. Prospectively designed PreMA reagent against tdgf1 can be used to reproduce gross developmental defect in 1 dpf, injected F0 larvae.

A. Top–Wildtype tdgf1 sequence with sgRNA target site annotated in orange. The dotted red boxes are MH arms predicted to be used most frequently. Raw sequence alignment of the whole PCR amplicon demonstrates that the majority of reads are the expected 4 bp deletion allele. Bottom–summary data from subcloning analyses. 72% of the mutant allele recovered were of the predicted MH allele. B. Previously reported tdgf1 loss-of-function phenotype was successfully recapitulated using this CRISPR-Cas9. Phenotype severity was graded by the “pinhead” morphology and cyclopia. Pinhead morphology alone was classified as Weak, whereas Moderate and Severe phenotypes also presented with varying degrees of cyclopia judged by the distance of forebrain protrusion. In the Severe class, the forebrain does not separate the eyes, and they are fused together. Box plot demonstrating phenotypic penetrance is provided with each experiment denoted by a unique marker shape. N = 4 with 3 biological and 4 technical replicates. At least 42 injected animals were scored in each experiment.

Fig 5. PreMA reagent against ttn.2 N2B results in specific reduction of shortening fraction in 2 dpf F0 zebrafish.

A. Top–Wildtype ttn.2 sequence at the N2B exon with sgRNA target site annotated in red. The dotted red boxes are MH arms predicted to be used most frequently. Raw sequence alignment of the whole PCR amplicon demonstrates that the majority of reads are the expected 5 bp deletion allele. Bottom–summary data from subcloning analyses. 86% of the mutant allele recovered were of the predicted MH allele. B. Previously reported pickwick phenotype was successfully recapitulated using this CRISPR-Cas9. 2 dpf zebrafish were immobilized in 3% methylcellulose for live recording of cardiac functions. Whereas injections with Cas9 only (660 pg), N2B #1 sgRNA only (300 pg), or tyr #2 sgRNA RNP (300 pg sgRNA + 660 pg Cas9) did not result in changes in shortening fraction at this age, MMEJ-inducing RNP injection targeting N2B #1 (300 pg sgRNA + 660 pg Cas9) resulted in a specific reduction in shortening fraction by 78.4%. In contrast, NHEJ-inducing RNP injection targeting N2B #2 (300 pg sgRNA + 660 pg Cas9) resulted in attenuated effects on shortening fraction (53.3% reduction), despite similarly high edit efficiency. Each data point represents an individual animal scored with the shape of the marker denoting unique experiment. N ≥ 3 biological and technical replicates, except for N2B #2 where N = 2. At least 5 injected animals were scored in each experiment. P-values calculated by Wilcoxon’s Each Pair Calculation (adjusted for multiple comparisons).

Fig 6. PreMA reagent can be used for in-frame gene alteration.

A. Top–Wildtype ttn.2 sequence with sgRNA target site annotated in red. The dotted red boxes are MH arms predicted to be used most frequently. Raw sequence alignment of the whole PCR amplicon demonstrates that the majority of reads are the expected 12 bp deletion allele. Bottom–summary data from subcloning analyses. 73% of the mutant allele recovered were of the predicted MH allele. B. 2 dpf zebrafish larvae injected with ttn.2 #2 sgRNA RNP (300 pg sgRNA + 660 pg Cas9) grossly appear normal with the exception of mild cardiac edema. Median penetrance was 50%. N = 3 biological and technical replicates. At least 9 injected animals were scored in each experiment.

Fig 7. Competition hypothesis version 2.

A. Outlier plot summarizing repair outcomes from 47 genomic targets using TALEN and CRISPR-Cas9. Close proximity of top predicted MH arms (Groups 3 and 4) appears to be the primary determinant for generating PreMA type outcomes as no target from Groups 1 and 2 had Top MH Fraction exceeding 0.5. When the top predicted allele had at least 50% higher Pattern Score than the second predicted allele (Groups 2 and 4), it was a strong indicator for inducing MMEJ-class repairs. B. Top Definition for each of the 4 groups used in Panel A. Each and every zebrafish genomic locus was segmented into these categories. Pattern scores were derived using RGEN online tool. Bottom P-values calculated by Wilcoxon’s Each Pair Calculation (adjusted for multiple comparisons). C. Graphical representation of each group detailed in Panel A. Groups 1 and 2 are prone to activate NHEJ-type outcomes, presumably because the yet-unidentified MMEJ factor fails to localize to suitable microhomology arm pairs, limited by how far apart these arms are. Group 4 is most suitable for strong MMEJ activation because it satisfies the proximity requirement AND the relative strength requirement. The latter may aid in the kinetics of the yet-unidentified MMEJ factor binding to the microhomology arms. Our data suggest that Group 3 is an intermediate group in terms of MMEJ activation. Perhaps extragenetic factors, such as cell cycle and epigenetic status may determine how favorable the loci are for MMEJ inductions.

Fig 8. Competition hypothesis V2 targets trigger primary repair by MMEJ in HEK293T cells.

A & B. Top–Wildtype human GJB2 sequences with sgRNA target sites annotated. The dotted red boxes denote the top predicted MH arms. Summary TIDE analysis outcomes are also presented showing ~ 45% Top MH Fractions for GJB2 #1 and #2 sgRNA. Red bar indicates the predicted deletion allele. Calculations for Adjusted Prevalence conform to calculations for Top MH Fractions detailed in S3 Note. Bottom–summary data from subcloning analyses for GJB2 #1 sgRNA (A) and #2 sgRNA (B). C & D. Top–Wildtype human AAVS1 and MYO7A sequences with sgRNA target sites annotated. The dotted red boxes denote the top predicted MH arms. Bottom–summary data from subcloning analyses for AAVS1 #2 sgRNA (A) and MYO7A #3 sgRNA (B).