In vivo base editing of post-mitotic sensory cells

In vivo base editing of post-mitotic sensory cells ARTICLE DOI: 10.1038/s41467-018-04580-3 OPEN 1,2,3,4,5 4,6 1,2,3 4,5,6,7 1,2,3 Wei-Hsi Yeh , Hao Chiang , Holly A. Rees , Albert S.B. Edge & David R. Liu Programmable nucleases can introduce precise changes to genomic DNA through homology- directed repair (HDR). Unfortunately, HDR is largely restricted to mitotic cells, and is typically accompanied by an excess of stochastic insertions and deletions (indels). Here we present an in vivo base editing strategy that addresses these limitations. We use nuclease-free base editing to install a S33F mutation in β-catenin that blocks β-catenin phosphorylation, impedes β-catenin degradation, and upregulates Wnt signaling. In vitro, base editing installs the S33F mutation with a 200-fold higher editing:indel ratio than HDR. In post-mitotic cells in mouse inner ear, injection of base editor protein:RNA:lipid installs this mutation, resulting in Wnt activation that induces mitosis of cochlear supporting cells and cellular reprogramming. In contrast, injection of HDR agents does not induce Wnt upregulation. These results establish a strategy for modifying posttranslational states in signaling pathways, and an approach to precision editing in post-mitotic tissues. 1 2 Department of Chemistry and Chemical Biology, Harvard University, Cambridge, MA 02138, USA. Howard Hughes Medical Institute, Harvard University, Cambridge, MA 02138, USA. Merkin Institute of Transformative Technologies in Healthcare, Broad Institute of MIT and Harvard, Cambridge, MA 02142, 4 5 USA. Eaton-Peabody Laboratory, Massachusetts Eye and Ear, Boston, MA 02114, USA. Program in Speech and Hearing Bioscience and Technology, 6 7 Harvard Medical School, Boston, MA 02115, USA. Department of Otolaryngology, Harvard Medical School, Boston, MA 02115, USA. Harvard Stem Cell Institute, Cambridge, MA 02138, USA. These authors contributed equally: Wei-Hsi Yeh, Hao Chiang. Correspondence and requests for materials should be addressed to A.S.B.E. (email: albert_edge@meei.harvard.edu) or to D.R.L. (email: drliu@fas.harvard.edu) NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications 1 | | | 1234567890():,; ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 30,31 tandard genome editing agents such as ZFNs, TALENs, or oncogenesis from widespread upregulation of Wnt activity Cas9 are programmable nucleases that induce a double- limits the use of small-molecule GSK-3β inhibitors in vivo. 1–4 Sstranded DNA break (DSB) at the target locus . While To test the ability of in vivo base editing in post-mitotic cells to such agents can efficiently disrupt genes by inducing non- induce a physiological outcome, we hypothesize that we could homologous end joining (NHEJ) and other processes that result upregulate Wnt activity in the inner ear by stabilizing β-catenin in stochastic insertions and deletions (indels) and translocations through base editing. β-catenin lacking phosphorylation at Ser 33 32–34 at the site of interest, the introduction of precise changes such as or Ser 37 is stabilized since it is not ubiquitinated by β-TrCP . point mutations in genomic DNA using homology-directed Deletion of β-catenin exon 3, which contains the sites of GSK-3β- repair (HDR) is difficult. Recutting of edited DNA containing a mediated phosphorylation, causes supporting cell proliferation single point mutation can substantially erode yields of desired and hair cell transdifferentiation in the postnatal sensory 5 21,22,35 product . In addition, HDR is thought to be restricted primarily epithelium of transgenic mice . However, unless exon 3 to the S and G2 phases of the cell cycle, when homologous deletion is limited to the inner ear, the resulting mice develop 6 36 recombination between sister chromatids naturally takes place . tumors in a variety of organs, including liver, prostate, and skin . Since most post-mitotic cells poorly express the cellular Local delivery of a base editor to prevent β-catenin phosphor- machinery required for this process, HDR in post-mitotic cells is ylation in the cochlea, which shows a striking resistance to 1,7,8 37 typically very inefficient . oncogenesis , may enable supporting cell renewal and hair cell We recently developed base editing, an alternative genome generation. editing strategy that directly converts one base pair to another In this study we develop a base editing strategy to alter protein base pair at a target locus without reliance on HDR and without posttranslational modification and stability in post-mitotic cells introducing double-stranded DNA breaks that lead to an abun- in vivo. We use BE3 to recode a single residue in the gene 3,9–11 dance of indels .The most widely used base editors are encoding β-catenin, preventing phosphorylation and degradation fusions of a catalytically disabled form of Cas9, a cytidine dea- of its protein product. In cell culture, base editing of β-catenin minase such as APOBEC1, and a DNA glycosylase inhibitor such increases its abundance and elevated Wnt signaling. In the as uracil glycosylase inhibitor (UGI) . Third-generation base cochlea of postnatal mice, lipid-mediated delivery of the β- editors (BE3 and its variants) convert C� G base pairs to T� A base catenin-targeting base editor results in the proliferation of post- pairs at programmable target loci within a window of ~1–5 mitotic supporting cells and the differentiation of supporting cells nucleotides and are compatible with a wide variety of into cells expressing the hair cell marker Myo7a. Our findings protospacer-adjacent motif (PAM) sequences . A new class of establish that base editing can be used to change the post- adenine base editors using a laboratory-evolved deaminase translational modification state, abundance, and signaling domain convert A� TtoG� C base pairs with minimal bypro- potential of a protein in post-mitotic cells in vivo, and also sug- ducts . Base editing has proven to be a robust approach to gest an approach to localized Wnt signaling activation. achieving efficient, permanent conversion of individual base pairs with minimal indel formation in fungi, plants, mammalian cells, 10,12–19 zebrafish, mice, frogs, and even human embryos . Results The steps involved in base editing are not thought to rely on Base editing prevents phosphorylation of Ser 33 in β-catenin. 3,9 cellular recombination machinery , raising the possibility that We hypothesized that base editing the β-catenin gene (CTNNB1 the process might take place efficiently in non-dividing cells in humans) to replace an amino acid that is phosphorylated by in vivo. We sought to test the ability of base editing, compared GSK-3β with a residue that cannot be phosphorylated would with a current HDR method, to generate precise point mutations impede β-catenin degradation and increase activation of T-cell in terminally differentiated cells in vivo efficiently enough to factor/lymphoid enhancer factor (TCF/LEF) transcription factors. result in a physiological outcome. In the mammalian inner ear, Previous work suggests that mutation of β-catenin Ser 33 to Tyr, sensory cells such as cochlear supporting cells and hair cells are Pro, or Cys can abolish recognition by β-TrCP, preventing 20 33 post-mitotic . The apparent lack of sensory cell regeneration in degradation of β-catenin . We designed a single-guide RNA the mammalian cochlea contributes to progressive, permanent (sgRNA) that when complexed with BE3, should introduce a C- hearing loss after damage. Recent studies in transgenic mice to-T mutation at the DNA nucleotide encoding Ser 33 in β- suggest that stabilization of β-catenin protein can facilitate the catenin, resulting in the conversion of a TCT (Ser) codon to a regeneration of sensory hair cells by increasing signaling through TTT (Phe) codon (Fig. 1b). Murine S33-targeting sgRNA and 21,22 the canonical Wnt pathway . Activation of Wnt signaling human S33-targeting sgRNA share very similar protospacers and stimulates the proliferation of supporting cells and can induce the both target the same codon. Since the phenylalanine side chain development of hair cells from supporting cells , suggesting that cannot be phosphorylated, this change should increase the cyto- stabilization of β-catenin in the cochlea might trigger similar solic lifetime of β-catenin, increasing activation of TCF/LEF cellular reprogramming events, even though additional steps are transcription factors and downstream signaling (Fig. 1a). 24,25 likely needed for these cells to become functional hair cells . Next we tested if base editing is capable of installing the desired Wnt activation induces β-catenin accumulation in the cytoplasm point mutation into β-catenin in cultured human cells. We and translocation into the nucleus, resulting in the activation of transfected plasmids expressing BE3 and S33-targeting sgRNA or Wnt target genes. In the absence of Wnt activation (Fig. 1a), an unrelated control sgRNA into HEK293T cells. After a 3-day cytosolic β-catenin is phosphorylated at specificserineand threo- incubation, cells were harvested and the C-to-T conversion nine residues by glycogen synthase kinase 3β (GSK-3β) .Phos- efficiency at nucleotide C8 of the β-catenin Ser 33 codon (counting phorylated β-catenin is recognized by β-transducin repeat- the PAM as nucleotide positions 21–23) was measured by high- containing proteins (β-TrCP), resulting in the ubiquitination and throughput DNA sequencing (Fig. 1b). We observed efficient degradation of β-catenin (Fig. 1a) . Previously a small-molecule mutation of the target codon from TCT to TTT (31% ± 0.9%) GSK-3β inhibitor and histone deacetylase inhibitor were used to together with the efficient conversion of another cytosine within the upregulate Wnt-responsive genes, resulting in substantial expansion base editing window, C6–T6 (28% ± 0.7%) (all efficiencies listed are of supporting cells and differentiation into hair cells in vitro . mean ± S.E.M. for three biological replicates with no enrichment for However, toxicity arising from inhibition of protein kinases that transfected cells). Editing at C6 was expected given the 5-base 29 3,10 share homology with GSK-3β as well as the potential for editing window of BE3 (C4–C8) .In addition,C1–T1 (1.7% ± 2 NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 ARTICLE a b WNT WNT Fz Fz 40 BE3: S33-targeting sgRNA BE3: unrelated sgRNA BE3 GSK-3β GSK-3β β-TrCP S33-targeting β-TrCP F33 sgRNA 20 β-catenin β-TrCP β-TrCP β-TrCP β-TrCP S33 β-catenin β-TrCP β-TrCP Proteasome Proteasome C C C C C C 1 6 8 15 16 20 F33 β-catenin CTNNB1: C TGGAC TC TGGAATC C ATTC TGG TCF TCF 1 6 8 15 16 20 S33 Wnt-responsive gene Wnt-responsive gene c d % C to T % Indel S33F mutation Ratio of S33F: % Indels efficiency indels CORRECT 5.4 ± 0.59 49 ± 6.8 0.11 HDR BE3 31 ± 0.88 2.0 ± 0.27 15.5 Fig. 1 Base editing strategy and comparison of HDR and base editing following plasmid delivery. a Schematic representation of the canonical Wnt pathway and a base editing strategy to stabilize β-catenin. In the absence of Wnt signaling, β-catenin is phosphorylated at Ser 33 by GSK-3β and degraded in a phosphorylation-dependent manner. Base editing with BE3 precisely mutates the Ser 33 codon to instead encode Phe, which cannot be phosphorylated. The resulting S33F β-catenin has an extended half-life and can activate target gene transcription by binding with TCF/LEF transcription factors. b HEK293T cells were transfected with plasmids expressing BE3 and S33-targeting sgRNA, or BE3 and an unrelated sgRNA. The percentage of total sequencing reads (with no enrichment for transfected cells) with C8 converted to T8 (resulting in the S33F mutation) was measured with high-throughput sequencing (HTS). c Plasmid delivery of Cas9 and BE3 (750 ng) with S33-targeting sgRNA (250 ng) into HEK293T cells using 1.5 µL of Lipofectamine 2000 per well of a 48-well plate. C-to-T conversion efficiency and d product selectivity ratio (desired S33F mutation: undesired indel ratio) resulting from the best-performing ratio of Cas9:sgRNA:ssDNA template and BE3. Values and error bars reflect mean ± S.E.M. of three biological replicates performed on separate days 0.1%) and C15–T15 (0.90% ± 0.05%) conversions were also non-phosphorylated β-catenin in cells, we measured the amount observed at much lower frequencies, consistent with our previous of total and non-phosphorylated β-catenin using cell fractiona- 3,10 studies . Importantly, unlike editing of the target C8 to T8, tion followed by Western blotting (Fig. 2b). We observed 6-fold conversion of C1, C6, or C15 to T does not change the predicted greater levels of non-phosphorylated β-catenin in nuclear extracts amino acid sequence of the resulting protein, as all three of the of cells transfected with plasmids encoding BE3 and S33-targeting other C-to-T mutations are silent. sgRNA compared to control cells transfected with plasmids We observed low (2.0% ± 0.3%) indel frequencies from BE3 encoding BE3 and an unrelated sgRNA (Fig. 2b and Supple- and S33-targeting sgRNA plasmid transfection. Control samples mentary Fig. 3a). The total amount of β-catenin, including both treated with plasmids encoding BE3 and an unrelated sgRNA phosphorylated and non-phosphorylated forms, was 7-fold were also analyzed, resulting in no C-to-T mutation at the target higher in nuclear protein extracts from cells treated with BE3 locus above our limit of detection (~0.025% mutation; see and the S33-targeting sgRNA than in control cells treated with Methods). Taken together, these observations validate a base BE3 and an unrelated-sgRNA (Fig. 2b and Supplementary editing strategy that converts the wild-type β-catenin gene to the Fig. 3a), consistent with stabilization and enhanced translocation S33F mutant in mammalian cells efficiently and with a high of β-catenin into the nucleus. degree of product selectivity. To assay the effects of installing the β-catenin S33F mutation on Wnt signaling, we used an established Wnt reporter system in HEK293T cells . This assay requires co-transfection with three Effect of S33F β-catenin mutation on Wnt signaling in vitro. plasmids: (i) Topflash, which contains TCF/LEF operators that To test if the S33F mutation in β-catenin increases the amount of NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications 3 | | | % of total sequencing reads with target C-to-T or indel Cas9:sgRNA: 0 μg ssDNA Cas9:sgRNA: 0.7 μg ssDNA Cas9:sgRNA: 1.3 μg ssDNA Cas9:sgRNA: 2.7 μg ssDNA BE3:sgRNA No treatment % of total sequencing reads with target C converted to T ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 ab **** Cytosol Nucleus Total β-catenin Non-phospho β-catenin S33F Wild-type GAPDH β-catenin β-catenin CTNNB1 cDNA (500 ng) c d BE3: S33-targeting sgRNA BE3: S33-targeting sgRNA BE3: unrelated sgRNA BE3: unrelated sgRNA * * *** * **** 50 *** ** **** **** **** Total BE3:sgRNA plasmid (ng) Total BE3:sgRNA plasmid (ng) Topflash plasmid: + + + + + + – Fopflash plasmid: – – – – – – + Fig. 2 Biological outcomes associated with base editing S33F in β-catenin in human cells. a HEK293T cells were transfected with plasmids encoding Topflash (β-catenin-responsive firefly luciferase reporter) or Fopflash (mutant form of Topflash that cannot be activated by β-catenin) and mutant S33F β- catenin or wild-type (Ser 33) β-catenin. Wnt signaling was measured by the ratio of Topflash:Fopflash luciferase activity. b Cytosolic and nuclear extracts of HEK293T cells treated with base editor and S33-targeting sgRNA or unrelated sgRNA were subjected to western blot analysis for total β-catenin or non- phosphorylated β-catenin (Ser33/Ser37/Thr41). Each blot represents one antibody. GAPDH was used as a loading control. c HEK293T cells were transfected with plasmids encoding the Topflash or Fopflash reporters, base editor, and S33F or unrelated control sgRNA. The Topflash:Fopflash luciferase ratio for BE3 + S33-targeting sgRNA (blue) and BE3+ unrelated sgRNA (red) are shown. d Percent C-to-T conversion at the target Ser 33 codon, which results in the S33F mutation in β-catenin, assayed by HTS. Values and error bars reflect mean ± S.E.M. of three biological replicates performed on different days. * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, and **** p ≤ 0.0001 (Student’s two tailed t-test) activate expression of firefly luciferase when bound by β-catenin, expressing the reporters and BE3+ S33-targeting sgRNA exhibited or Fopflash, a negative control plasmid that contains mutated a time-dependent increase in Topflash:Fopflash luminescence ratio TCF/LEF sites that cannot be activated by β-catenin; (ii) Renilla, (Supplementary Fig. 1b). After 3 days, cells treated with the highest which expresses renilla luciferase to enable normalization for dose of base editor (750 ng BE3 plasmid and 250 ng sgRNA transfection efficiency; and (iii) a β-catenin cDNA plasmid that plasmid per well of a 48-well plate) exhibited a much higher expresses either the wild-type (Ser 33) or mutated (Phe 33) Topflash:Fopflash ratio (310 ± 31) than cells transfected with BE3 β-catenin. 3 days post-transfection, the level of TCF/LEF- and an unrelated sgRNA (1.0 ± 0.02) (Fig. 2c). The C-to-T mediated Wnt signaling was quantified by the luminescence ratio conversion efficiency at the target Ser codon was analyzed by of the Topflash and Fopflash reporters (Fig. 2a). HEK293T cells high-throughput DNA sequencing (Fig. 2d). At the highest BE3 transfected with a plasmid expressing mutated S33F β-catenin dose, the C-to-T conversion efficiency at position 33 was 16 ± 0.1% showed a time-dependent increase in the Topflash:Fopflash (Fig. 2d). Lower doses of BE3 and sgRNA resulted in markedly luminescence ratio (Supplementary Fig. 1c), reaching a maximum lower base editing efficiency and lower Wnt signaling levels ratio of 180 ± 1.2 3 days post-transfection, 4.3-fold higher than the (Figs. 2cand 3d). Wnt signaling levels were strongly correlated Topflash:Fopflash luminescence ratio in cells transfected with a with S33F base editing efficiency (R = 0.97, p <0.0001 for non- plasmid expressing wild-type β-catenin (42 ± 0.7) (Fig. 2a). These zero slope, linear regression analysis, Supplementary Fig. 1a) results support a model in which mutating Ser 33 to Phe in suggesting that treatment with BE3 and the S33-targeting sgRNA β-catenin increases Wnt signaling activity in mammalian cells. strongly enhances Wnt signaling levels in a base editing-dependent Next, we used base editing to install the β-catenin S33F mutation manner. in human cells and assayed the effect on Wnt signaling. We co- Finally, we assayed the ability of β-catenin base editing to transfected HEK293T cells with four plasmids: a BE3-expression increase the expression of known endogenous Wnt-responsive plasmid, an sgRNA-expression plasmid (either targeting β-catenin genes in HEK293T cells, including Axin-related protein (AXIN2), Ser 33, or encoding an unrelated control sgRNA), the transfection cyclin D1 (CCND1), cyclin dependent kinase inhibitor 1A efficiency reporter Renilla, and either the Topflash or Fopflash (CDKN1A), and fibroblast growth factor 20 (FGF20). Transfec- reporter plasmid. HEK293T cells transfected with plasmids tion of HEK293T cells with BE3 and β-catenin S33-targeting 4 NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications | | | Topflash:Fopflash Topflash:Fopflash luciferase ratio luciferase ratio 62.5 % of total sequencing reads with target C converted to T 62.5 Non-related sgRNA S33F SgRNA Non-related sgRNA S33F SgRNA 1000 NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 ARTICLE a b 1.0 % C to T % Indel 0.5 0 0.0 Molar ratio of Cas9:sgRNA:ssDNA Molar ratio of Cas9:sgRNA:ssDNA c d 15 BE3 (1:1.1) S33F mutation Ratio of CORRECT (1:1.1:0.5) % Indels efficiency S33F: indels Control CORRECT 4.9 ± 0.81 37 ± 1.0 0.13 HDR BE3 13 ± 0.28 0.53 ± 0.03 26 C C C C C C 1 6 8 15 16 20 CTNNB1: C1TGGAC TC TGGAATC C ATTC TGG 6 8 15 16 20 S33 Fig. 3 Comparison of efficiency and product selectivity of HDR vs. base editing following RNP delivery. RNP delivery into HEK293T cells of 200 nM Cas9 or 200 nM BE3 pre-complexed with the S33-targeting sgRNA or an unrelated sgRNA and delivered in a cationic liposome. a Frequency of S33F mutation (blue) and indels (green) from treatment with CORRECT HDR agents (Cas9, sgRNA, and ssDNA donor template) in the ratios shown. b Product selectivity ratio, defined as the ratio of S33F modification to indel modification, resulting from treatment with the CORRECT HDR agents used in a. c Comparison of target C to T conversion efficiency following RNP delivery of BE3 (blue) or CORRECT HDR (purple) and the S33-targeting sgRNA. The control corresponds to cells treated with BE3 protein and unrelated sgRNA. d Product selectivity of base editing and CORRECT HDR. Values and error bars reflect mean ± S.E.M. of three independent biological replicates performed on different days sgRNA resulted in 1.8 ± 0.21-fold higher expression of AXIN2, HDR with a donor DNA template that both installs the desired 1.7 ± 0.17-fold higher expression of CCND1, 1.7 ± 0.7-fold higher mutation and that alters the PAM sequence to prevent re-cutting expression of CDKN1A, and 12 ± 2.2-fold higher expression of of the desired DNA product. Lipid-mediated plasmid transfection FGF20 compared with cells treated with BE3 and an unrelated of Cas9:sgRNA constructs and an optimized amount of ssDNA sgRNA (Supplementary Fig. 3b). donor template into HEK293T cells resulted in levels of precise Taken together, these results demonstrate that treatment of installation of β-catenin S33F (5.4 ± 0.6%) and indels (49 ± 7%) cells with the β-catenin S33F base editor converts the endogenous consistent with previous reports (Fig. 1c). Since plasmid β-catenin gene into the S33F mutant allele, substantially increases transfection of BE3:sgRNA constructs resulted in 31 ± 0.9% levels of non-phosphorylated β-catenin, increases expression of a conversion of β-catenin Ser 33 to Phe and 2.0 ± 0.3% indels reporter gene downstream of TCF/LEF operators in a base (Fig. 1d), the product selectivity ratio (desired S33F mutation: editing-dependent manner, and also increases expression of undesired indel ratio) was 0.11 for CORRECT HDR and 16 for endogenous Wnt-responsive genes. base editing, a 140-fold difference. We recently demonstrated delivery of Cas9 nuclease ribonu- cleoprotein (RNP) complexes to the inner ear to mediate spatially Comparison of base editing and HDR. A commonly used localized genome editing in vivo with enhanced DNA specificity approach to introduce precise modifications into genomic DNA 11,41,42 relative to plasmid transfection . In light of these advan- is to harness HDR using a targeted nuclease and a donor DNA tages, we also compared RNP delivery of base editors with RNP template that contains the desired modification and is homo- delivery of HDR agents. We optimized RNP delivery-mediated logous to the target locus. We compared the outcomes of base CORRECT HDR by treating cells with different stoichiometries of editing to Cas9 nuclease-mediated HDR to install the S33F Cas9 protein, S33-targeting sgRNA, and a donor ssDNA mutation in β-catenin in HEK293T cells. For this comparison we template. Optimized cationic lipid-mediated delivery of Cas9, used the recently described CORRECT HDR method, which guide RNA, and donor DNA template resulted in an average introduces desired mutations into target loci more efficiently than HDR efficiency of 4.9 ± 0.8% S33F mutation using a 1:1.1:0.5 39, 40 previous HDR methods . First we performed CORRECT NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications 5 | | | % of total sequencing reads with % of total sequencing reads with target C converted to T target C-to-T or indel 1:1.1:0 1:1.1:0.25 1:1.1:0.5 1:1.1:1 1:1.1:2 No treatment Ratio of S33F mutation frequency to indel frequency 1:1.1:0 1:1.1:0.25 1:1.1:0.5 1:1.1:1 1:1.1:2 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 molar ratio of Cas9:S33-targeting sgRNA:donor ssDNA template potently deliver negatively charged proteins or protein:nucleic (Fig. 3a). In contrast, the efficiency of installing the S33F acid complexes, including Cas9:sgRNA ribonucleoprotein (RNP) mutation using BE3, the same guide RNA, and the same cationic complexes, into the organ of Corti . Because activation of Wnt lipid averaged 13 ± 0.3% (Fig. 3c). Consistent with the DNA signaling outside the inner ear can promote oncogenesis , such a transfection results above, the CORRECT HDR method was also local delivery platform that minimizes exposure of other cells to accompanied by a much higher indel frequency than that of the the editing agent is ideally suited for in vivo base editing of base editing approach; we observed 37 ± 1% indels from β-catenin. CORRECT HDR, but only 0.52 ± 0.03% indels from base editing We tested different ratios of base editor:sgRNA:lipid in vivo. (Fig. 3b,d). Therefore, the product selectivity ratio of S33F editing: The optimal ratio combined purified BE3 (57 µM), the β-catenin indels was 200-fold higher for base editing than for CORRECT S33-targeting sgRNA or an unrelated sgRNA (100 µM) and 2.0 µL HDR. Taken together, these results indicate that base editing cationic lipid in a total volume of 12.0 µL. After a 5-min either by DNA delivery or RNP delivery enables more efficient incubation, we injected 1.0 µL of the resulting mixture into the installation of the S33F mutation in β-catenin with fewer indels cochlea of postnatal day 1 (P1) wild-type CD1 mice (Fig. 4a). The and thus much higher product selectivity than the CORRECT intracochlear injection was followed by subcutaneous injection of HDR method. 5-ethynyl-2′-deoxyuridine (EdU) daily throughout the 5 days to enable detection of proliferating cells. Cochlear tissues were harvested at P7 and EdU, Myo7a (a marker for cochlear hair Off-target analysis of β-catenin S33F base editing. Genome cells), and Sox2 (a marker for supporting cells) were detected by editing agents can induce unintended DNA modifications at off- chemical staining and immunofluorescence. target genomic loci that are similar in sequence to the target Confocal microscopy of cochlear tissue harvested from mice 43–45 locus . Recent studies have shown that off-target base editing treated with BE3 and the β-catenin S33-targeting sgRNA revealed mediated by BE3 is generally a subset of off-target loci modified cells positive for both EdU and Sox2, consistent with newly by the corresponding Cas9 nuclease and guide RNA, as expected divided supporting cells (Fig. 4h). The post-mitotic status of the given that the DNA-binding capability of BE3 is derived from postnatal day 7 (P7) cochlear sensory epithelium was confirmed 3,10,11 Cas9 . We investigated the potential off-target genome by the lack of EdU incorporation in the organ of Corti, in contrast editing by Cas9 nuclease programmed by the S33F β-catenin with EdU incorporation in mesenchymal cells (e.g., tympanic sgRNA used herein with two methods. First, we used GUIDE- 22,52,53 border cells), which are known to be mitotic (Supplemen- Seq , an unbiased genome-wide method that has been exten- tary Fig. 2, orange arrowhead). A cochlea treated with BE3 and sively used to identify off-target loci in mammalian cells following the β-catenin S33-targeting sgRNA displayed multiple EdU- Cas9:sgRNA exposure (See Methods). We performed GUIDE-Seq positive supporting cells in the apical turn (n = 3, EdU and Sox2 on murine NIH/3T3 cells treated with Cas9:S33-targeting sgRNA. double-positive cells = 16 ± 5.3, Fig. 4h). We also observed EdU The on-target β-catenin locus was identified with 1108 GUIDE- and Sox2 double-positive cells expressing the hair cell marker Seq reads, corresponding to an on-target modification frequency Myo7a (Fig. 4m,n, blue arrows). Indeed, all EdU and Myo7a of 23%. Despite robust detection of on-target modification, we double-positive cells observed were also Sox2-positive (Fig. 4e,h, observed zero GUIDE-Seq reads corresponding to off-target m,n), consistent with transdifferentiation of supporting cells into modification following treatment with Cas9:S33-targeting sgRNA hair cells . The expansion of supporting cells (EdU and Sox2 (Supplementary Table 1), suggesting little or no Cas9-mediated double-positive cells) was within the inner pillar cell region and off-target modification by this guide RNA in NIH/3T3 cells. likely represented Lgr5-positive cells, consistent with previous As a complementary approach, we used the Cutting Frequency 22,28 reports . Determinant (CFD) algorithm to predict off-target loci in the In contrast, treatment of the cochlea with optimized COR- 42,46,47 mouse genome associated with S33-targeting sgRNA . RECT HDR reagents (1:1.1:0.5 molar ratio of Cas9: sgRNA: Following nucleofection of plasmids encoding Cas9 and S33- donor ssDNA), resulted in no evidence of newly divided targeting sgRNA into NIH/3T3 cells, we performed deep supporting cells (EdU- and Sox2-positive) or newly divided hair sequencing to measure indel frequency at the top ten computa- cells (EdU- and Myo7a-positive) (Fig. 4d,g,k,l), consistent with tionally predicted off-target loci (Supplementary Table 1). Con- the inefficiency of HDR in these post-mitotic cells. A control sistent with the GUIDE-Seq results, we observed no detectable cochlea treated with BE3 and an unrelated sgRNA also showed no indel formation (<0.05%) at any of the ten predicted off-target newly divided supporting cells or hair cells (Fig. 4c,f,i,j). The lack loci (Supplementary Table 1). of EdU-positive cells in the sensory epithelium excludes the These data collectively suggest that the S33-targeting sgRNA possibility of cell division resulting from lipid-mediated BE3 used to modify the β-catenin locus mediates few, if any, off-target protein delivery. Together, these results suggest that base editing editing events by Cas9 in murine cells. Since off-target base of Ser 33 to Phe in β-catenin, in contrast with Cas9 nuclease- editing is typically a subset of Cas9 off-target modification for a mediated HDR, can induce cell division and transdifferentiation 3,9–11 given sgRNA , these findings suggest that the changes in of supporting cells into hair cells in post-mitotic cells in vivo. This β-catenin phosphorylation state and Wnt signaling activity are difference can be attributed to mechanistic differences between unlikely to arise from off-target base editing. base editing and HDR-mediated editing, as the cellular machinery that mediates HDR is inactive or poorly expressed in non- In vivo base editing induces post-mitotic cell reprogramming. dividing cells such as the target cells in the sensory epithelium . Base editing relies on cellular mismatch repair machinery, which To visualize the location and distribution of cationic lipid- 48,49 is expressed in most cells , in contrast to the cellular DSB mediated protein and RNP delivery following intracochlear repair and recombination machinery that mediates HDR, which injection into the mouse inner ear of P1 mice, we performed is poorly expressed in non-mitotic cells . This difference raises analogous intracochlear injections of lipid complexed with the possibility that base editing may be effective in post-mitotic proteins. Cre-mediated recombination in Ai9 tdTomato mice cells in vivo, even though HDR in post-mitotic cells remains a results in tdTomato fluorescence. We injected (–30)GFP–Cre major challenge. complexed with lipid into these mice, and observed tdTomato We previously discovered that local in vivo injection of cationic fluorescence in supporting cells (Supplementary Fig. 4). We then lipid reagents normally used for nucleic acid transfection could performed injections into wild-type CD1 mice of lipid complexed 6 NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 ARTICLE a b Organ of Corti Other hair cells P1 mouse pups Inner hair cells Supporting cells & 10 inner-piller-cell region Cationic lipid BE3 + BE3 + CORRECT BE3:sgRNA nanoparticle unrelated sgRNA S33-targeting sgRNA HDR cd e BE3 + unrelated sgRNA CORRECT HDR BE3 + S33-targeting sgRNA Sox2/Myo7a/EdU f gh i jk l m n Fig. 4 Outcomes associated with in vivo RNP-mediated base editing in post-mitotic cochlea. a The cochlea of a postnatal day 1 (P1) mouse was injected with lipid nanoparticles encapsulating BE3 + sgRNA, CORRECT HDR agents, or controls. The next day, mice received 5-ethynyl-2-deoxyuridine (EdU) by subcutaneous injection once per day for 5 days. At postnatal day 7 (P7), one day after the fifth EdU injection, the cochlea was dissected and organ of Corti was visualized by chemical staining (to visualize EdU) and immunofluorescence (Myo7a and Sox2). In the cochlea, Myo7a (red) is expressed in hair cells and Sox2 (green) is expressed in supporting cells. EdU (white) marks newly divided cells. b Quantification of Sox2 and EdU double-positive cells at the apical region of the organ of Corti from mice treated with BE3 + S33-targeting sgRNA (n= 3), CORRECT HDR agents (n= 3), or BE3+ unrelated sgRNA (n = 3). Values and error bars reflect mean and S.E.M. c–n Images from the organ of Corti tissue of mice treated with BE3 + unrelated sgRNA (c, f, i, j); CORRECT HDR agents (d, g, k, l); or BE3+ S33-targeting sgRNA (e, h, m, n). The blue arrows point to triple-positive cells that had undergone proliferation and reprogramming to cells expressing Myo7a. c–e x–y plane of hair cell layers. f–h x–y plane of supporting cell layers. (i, k, m) x–z plane of samples in f, g, h at the dotted yellow lines. j, k, m x–z plane of samples in f, g, h at the dotted yellow lines, but with the Myo7a and Sox2 channels shown. Scale bar (white)= 25 μm with BE3 and fluorescein-labeled S33-targeting sgRNA and observed no substantial C-to-T conversion (≤0.25%) or indels observed fluorescein localized within the organ of Corti in (≤0.1%) in three regions of the cochlea injected with optimized regions containing supporting cells and hair cells (Supplementary CORRECT HDR agents (Fig. 5), consistent with the known Fig. 4). Injection of lipid complexed with fluorescein-labeled ineffectiveness of HDR in post-mitotic supporting cells . sgRNA without BE3 did not result in fluorescein signal, likely due Collectively, these results confirm that base editing, in contrast to sgRNA degradation. These observations suggest that intraco- to HDR, can mediate local installation of the β-catenin S33F chlear injection of cationic lipid-meditated protein or RNP mutation in post-mitotic sensory cells in vivo. delivery results in localized delivery within the cochlea, including supporting cells and hair cells of interest. Discussion High-throughput DNA sequencing of genomic DNA from This study establishes in vivo base editing of post-mitotic sensory bulk cochlear tissue of treated mice confirmed base editing of β- cells through the local injection of a base editor RNP:lipid com- catenin Ser 33 to Phe in three regions of the cochlea: the organ of plex into the inner ear of mice. Here, the resulting base editing Corti (2.8% of total sequencing reads containing S33F β-catenin), event precisely introduced a S33F mutation into β-catenin, the stria vascularis (3.0%), and the modiolus (0.7%), with low altering its ability to be phosphorylated and decreasing its indel formation (averaging 0.4% across all tissues) (Fig. 5). We degradation rate, thereby enhancing Wnt signaling in vitro and note that samples of cochlear cells from treated mice included in vivo. This single amino acid change was installed in cultured cells that were not exposed to base editor, and thus we expect the cells more efficiently and with far fewer undesired genome percentage of dissected tissue containing the S33F mutation to be modifications using base editing with BE3 than using HDR with substantially less than the frequency of base editing observed in Cas9 nuclease and a donor DNA template. Our observations cultured cells, consistent with previous reports . In contrast, we in vivo also reveal that base editing, but not HDR, can be used to NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications 7 | | | x–z Supporting cell layer x–y Hair cell layer x–y # of Sox EdU positive cells in the apex ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 ab 9 5 Organ of Corti Organ of Corti Stria vascularis Stria vascularis Modiolus Modiolus Control No treatment 0 0 BE3 CORRECT HDR BE3 CORRECT HDR C-to-T conversion in serine 33 Ctnnb1: TTGGATTC TGGAATC C ATTC TGG 8 15 16 20 S33 Fig. 5 In vivo S33F mutation of β-catenin induced by injection of base editor RNPs. a Tissue was harvested from the cochlea of mice injected with either BE3 + S33-targeting sgRNA, or with CORRECT HDR agents. HTS of genomic DNA isolated from tissue samples revealed the frequency of the S33F mutation. Note that because tissue samples contain cells not exposed to editing agents, the observed genome modification frequency in these samples is less than the editing efficiency of treated cells. b Indel frequency at the Ser 33 locus following treatments described in a. Values and error bars reflect mean ± S.E.M. of four mice injected with BE3 and four mice injected with CORRECT HDR agents. Control samples in a, b are organ of Corti from three contralateral uninjected ears effect physiological changes in the post-mitotic mammalian inner Cloning of plasmids. The sgRNA plasmids were generated by USER cloning. Phusion U Hot Start DNA Polymerase (Thermo Fisher) was used to replace desired ear, consistent with the lack of dependence of base editing on protospacers from the sgRNA template plasmid. The cDNA plasmids were gen- homologous recombination machinery. erated by site directed mutagenesis (New England Biolabs). Primers were designed In contrast with the use of base editing to directly restore the with overhang containing the desired point mutation sequence and used to amplify sequence of a mutated gene to that of the corresponding wild- from a previously reported β-catenin cDNA construct . PCR products were car- 3,9,16,17,54 ried out using NEB stable Competent cells (New England Biolabs). See Note S2 for type allele , this study establishes that base editing the a full list of primers used in this study. potential of a protein to undergo post-translational modification, in this case to block its ability to be phosphorylated and degraded, Expression and purification of BE3 protein. BE3 protein was prepared by can also achieve a desired physiological outcome. Such an overexpressing in BL21 Star (DE3)-competent E. coli cells using a plasmid approach offers a potential advantage over simple gene correction encoding the bacterial codon-optimized base editor with a His N-terminal pur- in cases in which a low level of protein alteration can exert an ification tag. Detailed purification steps are described in our previous study , and amplified physiological effect. In this application, the effects of β- the expression plasmid is available on Addgene (Note S1). After protein expression, bacteria cells were lysed by sonication and the lysate was cleared by centrifugation. catenin S33F base editing, amplified through the ability of a The cleared lysate was incubated with His-Pur nickel nitriloacetic acid (nickel- persistent transcription factor to mediate multiple transcriptional NTA) resin. The resin was washed before bound protein was eluted with elution events, greatly augments Wnt signaling, leading to detectable buffer. The resulting protein fraction was further purified on a 5 mL Hi-Trap HP changes in cell proliferation and cell state. SP (GE Healthcare) cation exchange column using an Akta Pure FPLC. Protein- containing fractions were concentrated using a column with a 100,000 kDa cutoff Our work establishes that local delivery of base editor as an (Millipore) centrifuged at 3,000 g and the concentrated solution was sterile filtered RNP complex into the cochlea enables a high degree of specificity through an 0.22 μm PVDF membrane (Millipore). both for the target DNA locus and for the target cells in the cochlea. These features are distinct from the delivery of diffusible In vitro transcription of sgRNA. PCR was performed using Q5 Hot Start High- 20,55,56,57 small molecules , which can perturb the activity of Fidelity DNA Polymerase (New England Biolabs) and primers as listed in the Note homologous protein targets and have a greater potential to affect S2 to linearize DNA fragments containing the T7 RNA polymerase promoter the homeostasis of other tissues in vivo. Another key feature of sequence upstream of the desired 20 bp sgRNA protospacer and the sgRNA backbone. HiScribe T7 Quick High Yield RNA Synthesis Kit (New England Bio- the RNP delivery approach used in this study is that precise labs) was used to transcribe sgRNA at 37 °C for 14–16 h with 1 µg of linear tem- genome alterations are made without exposing cells to exogenous plate per 20 µL reaction. Fluorescein-labeled sgRNA was transcribed by adding DNA or virus, preventing the possibility of random integration of 10% v/v Fluorescein RNA Labeling Mix (Sigma Aldrich) to the transcription sys- DNA into the host cell genome. tem. Purification of sgRNA was performed with MEGAClear Transcription Clean Up Kit (Thermo Fisher), following the manufacturer’s instructions. Purified Although activating the Wnt pathway in the cochlea is likely 57–60 sgRNAs (100 µM) were stored in aliquots at −80 °C. insufficient to restore function in a damaged cochlea , these findings suggest the potential to manipulate complex signaling Protein extraction and western blotting. Total proteins were extracted with pathways by a precise in vivo editing strategy. This approach has radioimmune precipitation assay buffer from whole cells. Cytoplasmic and nuclear potential for in vivo cellular reprogramming, and, in principle, protein extracts were prepared with NE-PER nuclear and cytoplasmic extraction may be applicable to other disorders for which current therapies reagents (Thermo Fisher) according to the manufacturer’s instructions. The pro- 58,59 tein lysates were separated on 4–12% NuPAGE Bis-Tris gels (Invitrogen) and are repeated dosing of small-molecule Wnt agonists . electrotransferred to 0.2 µm PVDF (polyvinylidene difluoride) membranes (Bio- Rad). The membranes were probed with rabbit anti-Non-phospho β-catenin Methods (1:1000, Cell signaling 4270), rabbit anti-β-catenin (1:2000, Sigma C2206), and Animal models. CD1-IGS mice and floxP-tdTomato mice were obtained from mouse anti-GAPDH (1:800, Millipore MAB374) followed by horseradish Charles River Laboratories. All animal experiments were approved by the Insti- peroxidase-conjugated anti-rabbit (Chemicon), or anti-mouse (Chemicon) anti- tutional Animal Care and the Use Committee of Massachusetts Eye and Ear. bodies. The blots were detected with ECL-Plus Western Blotting Substrate (Thermo Fisher) (See Supplementary Fig. 5). As noted in the Millipore data sheet, GAPDH resides in both the cytosol and nucleus, where GAPDH is translocated to Cell line authentication and quality control. HEK293T (American Type Culture the nucleus when cells respond to the initial stages of apoptosis or oxidative stress. Collection, ATCC CRL-3216) and NIH/3T3 (ATCC CRL-1658) were maintained in DMEM plus GlutaMax (Thermo Fisher) supplemented with 10% (v/v) fetal bovine serum, 100 U/mL penicillin/streptomycin (Thermo Fisher) at 37 °C with 5% Real-time quantitative PCR (RT-qPCR). For each biological replicate, total RNA CO . Cells were obtained from ATCC and were authenticated and verified to be was extracted from HEK293T and reverse-transcribed into cDNA with oligo(dT) free of mycoplasma by ATCC upon purchase. primers (Thermo Fisher) and Superscript III reverse transcriptase (Life 8 NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications | | | % of total sequencing reads with S33F mutation % total sequencing reads containing an indel NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 ARTICLE Technologies). RT-qPCR was performed on a LightCycler 96 (Roche) for genes of Tissue dissection for HTS. Three to five days after the BE3: sgRNA delivery, interest with Gapdh as the housekeeping gene. Ct values for genes were averaged cochlea tissues were collected by microdissection for high-throughput sequencing. from three technical replicates. Amplification primers are purchased from Taqman Tissues were dissected into the organ of Corti, stria vascularis and modiolus. Each probe: GAPDH (Hs02758991), AXIN2 (Hs00610344), CCND1 (Hs00765553), tissue was further dissected into between five and ten separate pieces, and DNA CDKN1A (Hs00355782), and FGF20 (Hs00173929). extraction was performed separately for each sample, followed by high-throughput sequencing as described above. The data presented in Fig. 5 shows sequencing data resulting from the extraction of one microdissected sample from each cochlear Plasmid transfection into cell lines. HEK293T cells were seeded on 48-well region. collagen-coated BioCoat plates (Corning) in an antibiotic-free medium. After 12 h, HEK293T cells were transfected at ~70% confluency. For BE3 or HDR-CORRECT Data analysis. Sequencing reads were demultiplexed using MiSeq Reporter plasmid transfection, 750 ng of editing agent plasmid and 250 ng of sgRNA plas- (Illumina). Indel frequencies were assessed using a previously described MATLAB mids, with or without ssDNA (0, 0.7, 1.3, 2.7 µg) were transfected using 1.5 µl of script , which counts indels of ≥1 base occurring in a 30-base window around the Lipofectamine 2000 (Thermo Fisher) per well. For Wnt activity, unless otherwise BE3 nicking site. Indels were defined as detectable if there was a significant dif- noted, 200 ng Topflash, 20 ng Renilla, 750 ng of BE and 250 ng of sgRNA ference (Student’s two-tailed t-test, p < 0.05) between indel formation in the treated expression plasmids were transfected using 1.5 µl of Lipofectamine 2000 (Thermo sample and untreated control. Base editing frequencies were further assessed using Fisher) per well according to the manufacturer’s protocol. Balancing pUC19 a previously described MATLAB script . In brief, reads which did not contain plasmid (New England Biolabs) was transfected to make constant total DNA insertions or deletions were aligned to an appropriate reference sequence via the amount across conditions. For GUIDE-seq, 500 ng of Cas9, 250 ng of sgRNA Smith-Waterman algorithm. Individual bases with an Illumina quality score less encoding plasmids, and 100 pmol dsODN were transfected into NIH/3T3 cells than or equal to 30 were converted to the placeholder nucleotide (N). This quality using LONZA 4D-Nucleofector with the EN-158 program according to the man- threshold results in nucleotide frequencies with an expected theoretical error rate of ufacturer’s protocols. 1 in 1000. This ensures that reads containing both base edits and indels are not counted as successful base-edits, and only analyzes the non-indel containing population of reads. To calculate the number of edited reads as a percentage of the Protein transfection into cell lines. HEK293T cells were seeded on 48-well col- total number of successfully generated sequencing reads, the percentage of non- lagen-coated BioCoat plates (Corning) in 250 µl of an antibiotic-free medium. After indel containing edited reads as measured from the alignment algorithm were 12 h, HEK293T cells were transfected at ~70% confluency. Base editor protein was multiplied by (1- fraction of reads containing an indel). incubated with 1.1 times molar excess of the necessary sgRNA at room temperature for 5 min. In parallel, Cas9 protein was incubated with 1.1 times molar excess of sgRNA and different specified molar of ssDNA at room temperature. We observed Data availability. High-throughput sequencing data that support the findings of that higher sgRNA concentration with constant BE3 concentration did not increase this study have been deposited in the NCBI Sequence Read Archive database under base editing efficiency by HTS in vitro. The complex was then incubated with 1.5 µl Accession number SRP136325. All other data are available upon reasonable Lipofectamine 2000 (Thermo Fisher) and transfected according to the manu- request. facturer’s protocol for plasmid delivery. BE3 and Cas9 protein were added to a final concentration of 200 nM (based on a total well volume of 275 µl). Received: 31 January 2018 Accepted: 8 May 2018 High-throughput sequencing. Genomic DNA was isolated using the Agencourt DNAdvance Genomic DNA Isolation Kit (Beckman Coulter) according to the manufacturer’s instructions. First DNA amplification was performed by quanti- tative PCR with Phusion U Hot Start and SYBR Gold Nucleic Acid Stain (Thermo Fisher) to the top of the linear range. PCR products were purified using RapidTips References (Diffinity Genomics). The second PCR was performed to attach Indexing Adapters 1. Cox, D. B. T., Platt, R. J. & Zhang, F. Therapeutic genome editing: prospects (Illumina). The products were gel-purified and quantified using the KAPA Library and challenges. Nat. Med. 21, 121–131 (2015). Quantification Kit-Illumina (KAPA Biosystems). Samples were sequenced using a 2. Doudna, J. A. & Charpentier, E. The new frontier of genome engineering with single-end read from 200–250 bases (depending on the amplicon size) on the CRISPR-Cas9. Science 346, 1258096 (2014). MiSeq (Illumina) according to the manufacturer’s instructions. 3. Komor, A. C., Kim, Y. B., Packer, M. S., Zuris, J. A. & Liu, D. R. Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage. Nature 533, 420–424 (2016). Intracochlear delivery of ribonucleoprotein. Intracochlear delivery was per- 4. Sander, J. D. & Joung, J. K. CRISPR-Cas systems for editing, regulating and formed in post-natal day one (P1) CD1 mice or floxP-tdTomato mice as described targeting genomes. Nat. Biotechnol. 32, 347–355 (2014). previously . Briefly, mice were anaesthetized by lowering the body temperature 5. Biot-Pelletier, D. & Martin, V. J. J. Seamless site-directed mutagenesis of the before the surgical procedure. A postauricular incision was made near the right ear, Saccharomyces cerevisiae genome using CRISPR-Cas9. J. Biol. Eng. 10,6 and the bulla was lifted to expose the cochlea. BE3 protein (57.7 μM stock con- (2016). centration) was pre-complexed with the sgRNA or fluorescein–sgRNA (100 μM 6. Lin, S., Staahl, B. T., Alla, R. K. & Doudna, J. A. Enhanced homology-directed stock concentration) in a 1:1.1 molar ratio and then mixed with Lipofectamine 2000 (Thermo Fisher) in a 1:1 volumetric ratio. For delivery of (−30) GFP-Cre in human genome engineering by controlled timing of CRISPR/Cas9 delivery. eLife 3, e04766 (2014). floxP-tdTomato mice, 3 µL of 45 μM protein was mixed with 3 µL of Lipofectamine 2000. The resulting solution (1.0 µL) was injected with a glass pipette (end dia- 7. Jeggo, P. A. & Lobrich, M. DNA double-strand breaks: their cellular and meter, 5 μm) attached to a nanolitre micropump (WPI, UMP3 + Micro4 + clinical impact? Oncogene 26, 7717–7719 (2007). −1 NanoFil) at the rate of 150–200 nL min through the cochlear capsule into scala 8. Ran, F. A. et al. Genome engineering using the CRISPR-Cas9 system. Nat. media at the cochlear basal turn. Controls included the contralateral side of the Protoc. 8, 2281–2308 (2013). uninjected cochlea, pups that received BE3 protein pre-mixed with an unrelated 9. Gaudelli, N. M. et al. Programmable base editing of A� TtoG� C in genomic sgRNA, or pups that received sgRNA-only. After injection, the incision was closed DNA without DNA cleavage. Nature 551, 464 (2017). and the mice were brought onto a heating pad to recover. 10. Kim, Y. B. et al. Increasing the genome-targeting scope and precision of base editing with engineered Cas9-cytidine deaminase fusions. Nat. Biotechnol. 35, 371–376 (2017). EdU incorporation for cell proliferation. Pups received 10 µl of 5-ethynyl-2- 11. Rees, H. A. et al. Improving the DNA specificity and applicability of base −1 deoxyuridine (EdU, 10 mg ml ) by subcutaneous injection once daily for 5 days. editing through protein engineering and protein delivery. Nat. Commun. 8, EdU incorporation was detected with Click-iT EdU Imaging Kit (Invitrogen) 15790 (2017). according to the manufacturer’s instructions. 12. Chadwick, A. C., Wang, X. & Musunuru, K. In vivo base editing of PCSK9 (proprotein convertase subtilisin/kexin Type 9) as a therapeutic alternative to genome editing. Arterioscler. Thromb. Vasc. Biol. 37, 1741–1747 (2017). Immunohistochemistry. The temporal bones of P7 pups with BE3 and EdU 13. Hess, G. T. et al. Directed evolution using dCas9-targeted somatic delivery were fixed with 4% paraformaldehyde at 4 °C overnight. Cochlear tissues hypermutation in mammalian cells. Nat. Methods 13, 1036–1042 (2016). were either dissected into apical, mid, and basal turn for whole-mounts or 14. Kim, K. et al. Highly efficient RNA-guided base editing in mouse embryos. embedded in optimal cutting temperature medium (OCT) for frozen serial sections Nat. Biotechnol. 35, 435–437 (2017). with cryostat (LEICA CM3050). After EdU detection, tissues were incubated with 15. Li, G. et al. Highly efficient and precise base editing in discarded human rabbit anti-MYO7a (1:500, Proteus Biomedical 25-6790) and goat anti-Sox2 (1:100, tripronuclear embryos. Protein Cell. 8, 776–779 (2017). Santa Cruz Biotechnology sc-17320) antibodies followed with secondary antibodies 16. Liang, P. et al. Correction of β-thalassemia mutant by base editor in human conjugated with Alexa Fluor 488, and 568 (Invitrogen A11055, A10042). Images embryos. Protein Cell.8, 811–822 (2017). were taken with a confocal microscopy (Leica TCS SP8) and processed with Image J (NIH). NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications 9 | | | ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 17. Nishida, K. et al. Targeted nucleotide editing using hybrid prokaryotic and 49. Pluciennik, A. et al. PCNA function in the activation and strand direction of vertebrate adaptive immune systems. Science 353, aaf8729 (2016). MutLα endonuclease in mismatch repair. Proc. Natl. Acad. Sci. 107, 18. Zong, Y. et al. Precise base editing in rice, wheat and maize with a Cas9- 16066–16071 (2010). cytidine deaminase fusion. Nat. Biotechnol. 35, 438–440 (2017). 50. Heyer, W.-D., Ehmsen, K. T. & Liu, J. Regulation of homologous 19. Park, D.-S. et al. Targeted base editing via RNA-guided cytidine deaminases in recombination in eukaryotes. Annu. Rev. Genet. 44, 113–139 (2010). xenopus laevis embryos. Mol. Cells 40, 823–827 (2017). 51. Klaus, A. & Birchmeier, W. Wnt signaling and its impact on development and 20. Roccio, M., Hahnewald, S., Perny, M. & Senn, P. Cell cycle reactivation of cancer. Nat. Rev. Cancer 8, 387–398 (2008). cochlear progenitor cells in neonatal FUCCI mice by a GSK3 small molecule 52. Jan, T. A. et al. Tympanic border cells are Wnt-responsive and can act as inhibitor. Sci. Rep. 5, 17886 (2015). progenitors for postnatal mouse cochlear cells. Development 140, 1196–1206 21. Chai, R. et al. Wnt signaling induces proliferation of sensory precursors in the (2013). postnatal mouse cochlea. Proc. Natl. Acad. Sci. 109, 8167–8172 (2012). 53. Taniguchi, M., Yamamoto, N., Nakagawa, T., Ogino, E. & Ito, J. Identification 22. Shi, F., Hu, L. & Edge, A. S. B. Generation of hair cells in neonatal mice by β- of tympanic border cells as slow-cycling cells in the cochlea. PLoS ONE 7, catenin overexpression in Lgr5-positive cochlear progenitors. Proc. Natl. Acad. e48544 (2012). Sci. 110, 13851–13856 (2013). 54. Chadwick, A. C., Wang, X. & Musunuru, K. In vivo base editing of PCSK9 23. Chai, R. et al. Dynamic expression of Lgr5, a Wnt target gene, in the (proprotein convertase subtilisin/Kexin Type 9) as a therapeutic alternative to developing and mature mouse cochlea. JARO J. Assoc. Res. Otolaryngol. 12, genome editing. Arterioscler. Thromb. Vasc. Biol. 37, 1741 (2017). 455–469 (2011). 55. Wang, M. et al. Efficient delivery of genome-editing proteins using 24. Waqas, M., Zhang, S., He, Z., Tang, M. & Chai, R. Role of Wnt and Notch bioreducible lipid nanoparticles. Proc. Natl Acad. Sci. 113, 2868–2873 (2016). signaling in regulating hair cell regeneration in the cochlea. Front. Med. 10, 56. DeJonge, R. E. et al. Modulation of Wnt signaling enhances inner ear organoid 237–249 (2016). development in 3D culture. PLoS ONE 11, e0162508 (2016). 25. Geng, R. et al. Comprehensive expression of Wnt signaling pathway genes 57. Jacques, B. E. et al. A dual function for canonical Wnt/β-catenin signaling in during development and maturation of the mouse cochlea. PLoS ONE 11, the developing mammalian cochlea. Development 139, 4395 (2012). e0148339 (2016). 58. Leeds, P. R. et al. A new avenue for lithium: intervention in traumatic brain 26. MacDonald, B. T., Tamai, K. & He, X. Wnt/β-catenin signaling: components, injury. ACS Chem. Neurosci. 5, 422–433 (2014). mechanisms, and diseases. Dev. Cell 17,9–26 (2009). 59. Medina, M., Garrido, J. J. & Wandosell, F. Modulation of GSK-3 as a 27. Liu, C. et al. β-Trcp couples β-catenin phosphorylation-degradation and therapeutic strategy on Tau pathologies. Front. Mol. Neurosci. 4, 24 (2011). regulates Xenopus axis formation. Proc. Natl. Acad. Sci. 96, 6273–6278 (1999). 60. Shi, F., Cheng, Y.-F., Wang, X. L. & Edge, A. S. B. β-catenin up-regulates 28. McLean, W. J. et al. Clonal expansion of Lgr5-positive cells from mammalian Atoh1 expression in neural progenitor cells by interaction with an Atoh1 3′ cochlea and high-purity generation of sensory hair cells. Cell Rep. 18, enhancer. J. Biol. Chem. 285, 392–400 (2010). 1917–1929 (2017). 61. Hu, L. et al. Diphtheria toxin-induced cell death triggers Wnt-dependent hair 29. Eldar-Finkelman, H. & Martinez, A. GSK-3 inhibitors: preclinical and clinical cell regeneration in neonatal mice. J. Neurosci. 36, 9479 (2016). focus on CNS. Front. Mol. Neurosci. 4, 32 (2011). 30. Polakis, P. Wnt signaling and cancer. Genes Dev. 14, 1837–1851 (2000). Acknowledgements 31. Ring, A., Kim, Y.-M. & Kahn, M. Wnt/catenin signaling in adult stem cell D.R.L. gratefully acknowledges support from DARPA HR0011-17-2-0049; NIH RM1 physiology and disease. Stem Cell Rev. 10, 512–525 (2014). HG009490, R01 EB022376, and R35 GM118062; and HHMI. H.A.R. is supported by the 32. Li, VivianS. W. et al. Wnt signaling through inhibition of β-catenin Kilpatrick Educational fund from the Chemistry and Chemical Biology Department, degradation in an intact Axin1 complex. Cell 149, 1245–1256 (2012). Harvard University. A.S.B.E. is grateful for the support from NIH R01 DC014089, NIH 33. Liu, C. et al. Control of β-catenin phosphorylation/degradation by a dual- R01 DC007174, and the Shulsky Foundation. kinase mechanism. Cell 108, 837–847 (2002). 34. Verheyen, E. M. & Gottardi, C. J. Regulation of Wnt/β-catenin signaling by protein kinases. Dev. Dyn. Off. Publ. Am. Assoc. Anat. 239,34–44 (2010). Author contributions 35. Shi, F., Kempfle, J. S. & Edge, A. S. B. Wnt-responsive Lgr5-expressing stem W-H.Y., A.S.B.E., and D.R.L. designed the research. W-H.Y., H.C., and H.A.R. conducted cells are hair cell progenitors in the cochlea. J. Neurosci. 32, 9639 (2012). the experiments. W-H.Y., A.S.B.E., and D.R.L. analyzed the data. A.S.B.E. and D.R.L. 36. Moon, R. T., Kohn, A. D., Ferrari, G. V. D. & Kaykas, A. WNT and β-catenin supervised the research. All authors contributed to writing the manuscript. signaling: diseases and therapies. Nat. Rev. Genet. 5, 691 (2004). 37. Sulg, M. et al. Differential sensitivity of the inner ear sensory cell populations to forced cell cycle re-entry and p53 induction. J. Neurochem. 112, 1513–1526 Additional information (2010). Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- 38. Veeman, M. T., Slusarski, D. C., Kaykas, A., Louie, S. H. & Moon, R. T. 018-04580-3. Zebrafish prickle, a modulator of noncanonical Wnt/Fz signaling, regulates gastrulation movements. Curr. Biol. 13, 680–685 (2003). Competing interests: D.R.L. is a consultant and co-founder of Editas Medicine, Beam 39. Kwart, D., Paquet, D., Teo, S. & Tessier-Lavigne, M. Precise and efficient Therapeutics, and Pairwise Plants, companies that are using genome editing. A.S.B.E. is a scarless genome editing in stem cells using CORRECT. Nat. Protoc. 12, consultant and co-founder of Decibel Therapeutics. The remaining authors declare no 329–354 (2017). competing interests. 40. Paquet, D. et al. Efficient introduction of specific homozygous and heterozygous mutations using CRISPR/Cas9. Nature 533, 125–129 (2016). Reprints and permission information is available online at http://npg.nature.com/ 41. Gao, X. et al. Treatment of autosomal dominant hearing loss by in vivo reprintsandpermissions/ delivery of genome editing agents. Nature. 553, 217–221 (2017). 42. Zuris, J. A. et al. Cationic lipid-mediated delivery of proteins enables efficient Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in protein-based genome editing in vitro and in vivo. Nat. Biotechnol. 33,73–80 published maps and institutional affiliations. (2015). 43. Fu, Y. et al. High-frequency off-target mutagenesis induced by CRISPR-Cas nucleases in human cells. Nat. Biotechnol. 31, 822–826 (2013). 44. Pattanayak, V. et al. High-throughput profiling of off-target DNA cleavage Open Access This article is licensed under a Creative Commons reveals RNA-programmed Cas9 nuclease specificity. Nat. Biotechnol. 31, Attribution 4.0 International License, which permits use, sharing, 839–843 (2013). adaptation, distribution and reproduction in any medium or format, as long as you give 45. Tsai, S. Q. et al. GUIDE-seq enables genome-wide profiling of off-target appropriate credit to the original author(s) and the source, provide a link to the Creative cleavage by CRISPR-Cas nucleases. Nat. Biotechnol. 33, 187–197 (2015). Commons license, and indicate if changes were made. The images or other third party 46. Davis, K. M., Pattanayak, V., Thompson, D. B., Zuris, J. A. & Liu, D. R. Small material in this article are included in the article’s Creative Commons license, unless molecule-triggered Cas9 protein with improved genome-editing specificity. indicated otherwise in a credit line to the material. If material is not included in the Nat. Chem. Biol. 11, 316–318 (2015). article’s Creative Commons license and your intended use is not permitted by statutory 47. Haeussler, M. et al. Evaluation of off-target and on-target scoring algorithms regulation or exceeds the permitted use, you will need to obtain permission directly from and integration into the guide RNA selection tool CRISPOR. Genome Biol. 17, the copyright holder. To view a copy of this license, visit http://creativecommons.org/ 148 (2016). licenses/by/4.0/. 48. Heller, R. C. & Marians, K. J. Replisome assembly and the direct restart of stalled replication forks. Nat. Rev. Mol. Cell Biol. 7, 932–943 (2006). © The Author(s) 2018 10 NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications | | | http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nature Communications Springer Journals

In vivo base editing of post-mitotic sensory cells

Free
10 pages

Loading next page...
 
/lp/springer_journal/in-vivo-base-editing-of-post-mitotic-sensory-cells-FgCBtu1kiB
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s)
Subject
Science, Humanities and Social Sciences, multidisciplinary; Science, Humanities and Social Sciences, multidisciplinary; Science, multidisciplinary
eISSN
2041-1723
D.O.I.
10.1038/s41467-018-04580-3
Publisher site
See Article on Publisher Site

Abstract

ARTICLE DOI: 10.1038/s41467-018-04580-3 OPEN 1,2,3,4,5 4,6 1,2,3 4,5,6,7 1,2,3 Wei-Hsi Yeh , Hao Chiang , Holly A. Rees , Albert S.B. Edge & David R. Liu Programmable nucleases can introduce precise changes to genomic DNA through homology- directed repair (HDR). Unfortunately, HDR is largely restricted to mitotic cells, and is typically accompanied by an excess of stochastic insertions and deletions (indels). Here we present an in vivo base editing strategy that addresses these limitations. We use nuclease-free base editing to install a S33F mutation in β-catenin that blocks β-catenin phosphorylation, impedes β-catenin degradation, and upregulates Wnt signaling. In vitro, base editing installs the S33F mutation with a 200-fold higher editing:indel ratio than HDR. In post-mitotic cells in mouse inner ear, injection of base editor protein:RNA:lipid installs this mutation, resulting in Wnt activation that induces mitosis of cochlear supporting cells and cellular reprogramming. In contrast, injection of HDR agents does not induce Wnt upregulation. These results establish a strategy for modifying posttranslational states in signaling pathways, and an approach to precision editing in post-mitotic tissues. 1 2 Department of Chemistry and Chemical Biology, Harvard University, Cambridge, MA 02138, USA. Howard Hughes Medical Institute, Harvard University, Cambridge, MA 02138, USA. Merkin Institute of Transformative Technologies in Healthcare, Broad Institute of MIT and Harvard, Cambridge, MA 02142, 4 5 USA. Eaton-Peabody Laboratory, Massachusetts Eye and Ear, Boston, MA 02114, USA. Program in Speech and Hearing Bioscience and Technology, 6 7 Harvard Medical School, Boston, MA 02115, USA. Department of Otolaryngology, Harvard Medical School, Boston, MA 02115, USA. Harvard Stem Cell Institute, Cambridge, MA 02138, USA. These authors contributed equally: Wei-Hsi Yeh, Hao Chiang. Correspondence and requests for materials should be addressed to A.S.B.E. (email: albert_edge@meei.harvard.edu) or to D.R.L. (email: drliu@fas.harvard.edu) NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications 1 | | | 1234567890():,; ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 30,31 tandard genome editing agents such as ZFNs, TALENs, or oncogenesis from widespread upregulation of Wnt activity Cas9 are programmable nucleases that induce a double- limits the use of small-molecule GSK-3β inhibitors in vivo. 1–4 Sstranded DNA break (DSB) at the target locus . While To test the ability of in vivo base editing in post-mitotic cells to such agents can efficiently disrupt genes by inducing non- induce a physiological outcome, we hypothesize that we could homologous end joining (NHEJ) and other processes that result upregulate Wnt activity in the inner ear by stabilizing β-catenin in stochastic insertions and deletions (indels) and translocations through base editing. β-catenin lacking phosphorylation at Ser 33 32–34 at the site of interest, the introduction of precise changes such as or Ser 37 is stabilized since it is not ubiquitinated by β-TrCP . point mutations in genomic DNA using homology-directed Deletion of β-catenin exon 3, which contains the sites of GSK-3β- repair (HDR) is difficult. Recutting of edited DNA containing a mediated phosphorylation, causes supporting cell proliferation single point mutation can substantially erode yields of desired and hair cell transdifferentiation in the postnatal sensory 5 21,22,35 product . In addition, HDR is thought to be restricted primarily epithelium of transgenic mice . However, unless exon 3 to the S and G2 phases of the cell cycle, when homologous deletion is limited to the inner ear, the resulting mice develop 6 36 recombination between sister chromatids naturally takes place . tumors in a variety of organs, including liver, prostate, and skin . Since most post-mitotic cells poorly express the cellular Local delivery of a base editor to prevent β-catenin phosphor- machinery required for this process, HDR in post-mitotic cells is ylation in the cochlea, which shows a striking resistance to 1,7,8 37 typically very inefficient . oncogenesis , may enable supporting cell renewal and hair cell We recently developed base editing, an alternative genome generation. editing strategy that directly converts one base pair to another In this study we develop a base editing strategy to alter protein base pair at a target locus without reliance on HDR and without posttranslational modification and stability in post-mitotic cells introducing double-stranded DNA breaks that lead to an abun- in vivo. We use BE3 to recode a single residue in the gene 3,9–11 dance of indels .The most widely used base editors are encoding β-catenin, preventing phosphorylation and degradation fusions of a catalytically disabled form of Cas9, a cytidine dea- of its protein product. In cell culture, base editing of β-catenin minase such as APOBEC1, and a DNA glycosylase inhibitor such increases its abundance and elevated Wnt signaling. In the as uracil glycosylase inhibitor (UGI) . Third-generation base cochlea of postnatal mice, lipid-mediated delivery of the β- editors (BE3 and its variants) convert C� G base pairs to T� A base catenin-targeting base editor results in the proliferation of post- pairs at programmable target loci within a window of ~1–5 mitotic supporting cells and the differentiation of supporting cells nucleotides and are compatible with a wide variety of into cells expressing the hair cell marker Myo7a. Our findings protospacer-adjacent motif (PAM) sequences . A new class of establish that base editing can be used to change the post- adenine base editors using a laboratory-evolved deaminase translational modification state, abundance, and signaling domain convert A� TtoG� C base pairs with minimal bypro- potential of a protein in post-mitotic cells in vivo, and also sug- ducts . Base editing has proven to be a robust approach to gest an approach to localized Wnt signaling activation. achieving efficient, permanent conversion of individual base pairs with minimal indel formation in fungi, plants, mammalian cells, 10,12–19 zebrafish, mice, frogs, and even human embryos . Results The steps involved in base editing are not thought to rely on Base editing prevents phosphorylation of Ser 33 in β-catenin. 3,9 cellular recombination machinery , raising the possibility that We hypothesized that base editing the β-catenin gene (CTNNB1 the process might take place efficiently in non-dividing cells in humans) to replace an amino acid that is phosphorylated by in vivo. We sought to test the ability of base editing, compared GSK-3β with a residue that cannot be phosphorylated would with a current HDR method, to generate precise point mutations impede β-catenin degradation and increase activation of T-cell in terminally differentiated cells in vivo efficiently enough to factor/lymphoid enhancer factor (TCF/LEF) transcription factors. result in a physiological outcome. In the mammalian inner ear, Previous work suggests that mutation of β-catenin Ser 33 to Tyr, sensory cells such as cochlear supporting cells and hair cells are Pro, or Cys can abolish recognition by β-TrCP, preventing 20 33 post-mitotic . The apparent lack of sensory cell regeneration in degradation of β-catenin . We designed a single-guide RNA the mammalian cochlea contributes to progressive, permanent (sgRNA) that when complexed with BE3, should introduce a C- hearing loss after damage. Recent studies in transgenic mice to-T mutation at the DNA nucleotide encoding Ser 33 in β- suggest that stabilization of β-catenin protein can facilitate the catenin, resulting in the conversion of a TCT (Ser) codon to a regeneration of sensory hair cells by increasing signaling through TTT (Phe) codon (Fig. 1b). Murine S33-targeting sgRNA and 21,22 the canonical Wnt pathway . Activation of Wnt signaling human S33-targeting sgRNA share very similar protospacers and stimulates the proliferation of supporting cells and can induce the both target the same codon. Since the phenylalanine side chain development of hair cells from supporting cells , suggesting that cannot be phosphorylated, this change should increase the cyto- stabilization of β-catenin in the cochlea might trigger similar solic lifetime of β-catenin, increasing activation of TCF/LEF cellular reprogramming events, even though additional steps are transcription factors and downstream signaling (Fig. 1a). 24,25 likely needed for these cells to become functional hair cells . Next we tested if base editing is capable of installing the desired Wnt activation induces β-catenin accumulation in the cytoplasm point mutation into β-catenin in cultured human cells. We and translocation into the nucleus, resulting in the activation of transfected plasmids expressing BE3 and S33-targeting sgRNA or Wnt target genes. In the absence of Wnt activation (Fig. 1a), an unrelated control sgRNA into HEK293T cells. After a 3-day cytosolic β-catenin is phosphorylated at specificserineand threo- incubation, cells were harvested and the C-to-T conversion nine residues by glycogen synthase kinase 3β (GSK-3β) .Phos- efficiency at nucleotide C8 of the β-catenin Ser 33 codon (counting phorylated β-catenin is recognized by β-transducin repeat- the PAM as nucleotide positions 21–23) was measured by high- containing proteins (β-TrCP), resulting in the ubiquitination and throughput DNA sequencing (Fig. 1b). We observed efficient degradation of β-catenin (Fig. 1a) . Previously a small-molecule mutation of the target codon from TCT to TTT (31% ± 0.9%) GSK-3β inhibitor and histone deacetylase inhibitor were used to together with the efficient conversion of another cytosine within the upregulate Wnt-responsive genes, resulting in substantial expansion base editing window, C6–T6 (28% ± 0.7%) (all efficiencies listed are of supporting cells and differentiation into hair cells in vitro . mean ± S.E.M. for three biological replicates with no enrichment for However, toxicity arising from inhibition of protein kinases that transfected cells). Editing at C6 was expected given the 5-base 29 3,10 share homology with GSK-3β as well as the potential for editing window of BE3 (C4–C8) .In addition,C1–T1 (1.7% ± 2 NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 ARTICLE a b WNT WNT Fz Fz 40 BE3: S33-targeting sgRNA BE3: unrelated sgRNA BE3 GSK-3β GSK-3β β-TrCP S33-targeting β-TrCP F33 sgRNA 20 β-catenin β-TrCP β-TrCP β-TrCP β-TrCP S33 β-catenin β-TrCP β-TrCP Proteasome Proteasome C C C C C C 1 6 8 15 16 20 F33 β-catenin CTNNB1: C TGGAC TC TGGAATC C ATTC TGG TCF TCF 1 6 8 15 16 20 S33 Wnt-responsive gene Wnt-responsive gene c d % C to T % Indel S33F mutation Ratio of S33F: % Indels efficiency indels CORRECT 5.4 ± 0.59 49 ± 6.8 0.11 HDR BE3 31 ± 0.88 2.0 ± 0.27 15.5 Fig. 1 Base editing strategy and comparison of HDR and base editing following plasmid delivery. a Schematic representation of the canonical Wnt pathway and a base editing strategy to stabilize β-catenin. In the absence of Wnt signaling, β-catenin is phosphorylated at Ser 33 by GSK-3β and degraded in a phosphorylation-dependent manner. Base editing with BE3 precisely mutates the Ser 33 codon to instead encode Phe, which cannot be phosphorylated. The resulting S33F β-catenin has an extended half-life and can activate target gene transcription by binding with TCF/LEF transcription factors. b HEK293T cells were transfected with plasmids expressing BE3 and S33-targeting sgRNA, or BE3 and an unrelated sgRNA. The percentage of total sequencing reads (with no enrichment for transfected cells) with C8 converted to T8 (resulting in the S33F mutation) was measured with high-throughput sequencing (HTS). c Plasmid delivery of Cas9 and BE3 (750 ng) with S33-targeting sgRNA (250 ng) into HEK293T cells using 1.5 µL of Lipofectamine 2000 per well of a 48-well plate. C-to-T conversion efficiency and d product selectivity ratio (desired S33F mutation: undesired indel ratio) resulting from the best-performing ratio of Cas9:sgRNA:ssDNA template and BE3. Values and error bars reflect mean ± S.E.M. of three biological replicates performed on separate days 0.1%) and C15–T15 (0.90% ± 0.05%) conversions were also non-phosphorylated β-catenin in cells, we measured the amount observed at much lower frequencies, consistent with our previous of total and non-phosphorylated β-catenin using cell fractiona- 3,10 studies . Importantly, unlike editing of the target C8 to T8, tion followed by Western blotting (Fig. 2b). We observed 6-fold conversion of C1, C6, or C15 to T does not change the predicted greater levels of non-phosphorylated β-catenin in nuclear extracts amino acid sequence of the resulting protein, as all three of the of cells transfected with plasmids encoding BE3 and S33-targeting other C-to-T mutations are silent. sgRNA compared to control cells transfected with plasmids We observed low (2.0% ± 0.3%) indel frequencies from BE3 encoding BE3 and an unrelated sgRNA (Fig. 2b and Supple- and S33-targeting sgRNA plasmid transfection. Control samples mentary Fig. 3a). The total amount of β-catenin, including both treated with plasmids encoding BE3 and an unrelated sgRNA phosphorylated and non-phosphorylated forms, was 7-fold were also analyzed, resulting in no C-to-T mutation at the target higher in nuclear protein extracts from cells treated with BE3 locus above our limit of detection (~0.025% mutation; see and the S33-targeting sgRNA than in control cells treated with Methods). Taken together, these observations validate a base BE3 and an unrelated-sgRNA (Fig. 2b and Supplementary editing strategy that converts the wild-type β-catenin gene to the Fig. 3a), consistent with stabilization and enhanced translocation S33F mutant in mammalian cells efficiently and with a high of β-catenin into the nucleus. degree of product selectivity. To assay the effects of installing the β-catenin S33F mutation on Wnt signaling, we used an established Wnt reporter system in HEK293T cells . This assay requires co-transfection with three Effect of S33F β-catenin mutation on Wnt signaling in vitro. plasmids: (i) Topflash, which contains TCF/LEF operators that To test if the S33F mutation in β-catenin increases the amount of NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications 3 | | | % of total sequencing reads with target C-to-T or indel Cas9:sgRNA: 0 μg ssDNA Cas9:sgRNA: 0.7 μg ssDNA Cas9:sgRNA: 1.3 μg ssDNA Cas9:sgRNA: 2.7 μg ssDNA BE3:sgRNA No treatment % of total sequencing reads with target C converted to T ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 ab **** Cytosol Nucleus Total β-catenin Non-phospho β-catenin S33F Wild-type GAPDH β-catenin β-catenin CTNNB1 cDNA (500 ng) c d BE3: S33-targeting sgRNA BE3: S33-targeting sgRNA BE3: unrelated sgRNA BE3: unrelated sgRNA * * *** * **** 50 *** ** **** **** **** Total BE3:sgRNA plasmid (ng) Total BE3:sgRNA plasmid (ng) Topflash plasmid: + + + + + + – Fopflash plasmid: – – – – – – + Fig. 2 Biological outcomes associated with base editing S33F in β-catenin in human cells. a HEK293T cells were transfected with plasmids encoding Topflash (β-catenin-responsive firefly luciferase reporter) or Fopflash (mutant form of Topflash that cannot be activated by β-catenin) and mutant S33F β- catenin or wild-type (Ser 33) β-catenin. Wnt signaling was measured by the ratio of Topflash:Fopflash luciferase activity. b Cytosolic and nuclear extracts of HEK293T cells treated with base editor and S33-targeting sgRNA or unrelated sgRNA were subjected to western blot analysis for total β-catenin or non- phosphorylated β-catenin (Ser33/Ser37/Thr41). Each blot represents one antibody. GAPDH was used as a loading control. c HEK293T cells were transfected with plasmids encoding the Topflash or Fopflash reporters, base editor, and S33F or unrelated control sgRNA. The Topflash:Fopflash luciferase ratio for BE3 + S33-targeting sgRNA (blue) and BE3+ unrelated sgRNA (red) are shown. d Percent C-to-T conversion at the target Ser 33 codon, which results in the S33F mutation in β-catenin, assayed by HTS. Values and error bars reflect mean ± S.E.M. of three biological replicates performed on different days. * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, and **** p ≤ 0.0001 (Student’s two tailed t-test) activate expression of firefly luciferase when bound by β-catenin, expressing the reporters and BE3+ S33-targeting sgRNA exhibited or Fopflash, a negative control plasmid that contains mutated a time-dependent increase in Topflash:Fopflash luminescence ratio TCF/LEF sites that cannot be activated by β-catenin; (ii) Renilla, (Supplementary Fig. 1b). After 3 days, cells treated with the highest which expresses renilla luciferase to enable normalization for dose of base editor (750 ng BE3 plasmid and 250 ng sgRNA transfection efficiency; and (iii) a β-catenin cDNA plasmid that plasmid per well of a 48-well plate) exhibited a much higher expresses either the wild-type (Ser 33) or mutated (Phe 33) Topflash:Fopflash ratio (310 ± 31) than cells transfected with BE3 β-catenin. 3 days post-transfection, the level of TCF/LEF- and an unrelated sgRNA (1.0 ± 0.02) (Fig. 2c). The C-to-T mediated Wnt signaling was quantified by the luminescence ratio conversion efficiency at the target Ser codon was analyzed by of the Topflash and Fopflash reporters (Fig. 2a). HEK293T cells high-throughput DNA sequencing (Fig. 2d). At the highest BE3 transfected with a plasmid expressing mutated S33F β-catenin dose, the C-to-T conversion efficiency at position 33 was 16 ± 0.1% showed a time-dependent increase in the Topflash:Fopflash (Fig. 2d). Lower doses of BE3 and sgRNA resulted in markedly luminescence ratio (Supplementary Fig. 1c), reaching a maximum lower base editing efficiency and lower Wnt signaling levels ratio of 180 ± 1.2 3 days post-transfection, 4.3-fold higher than the (Figs. 2cand 3d). Wnt signaling levels were strongly correlated Topflash:Fopflash luminescence ratio in cells transfected with a with S33F base editing efficiency (R = 0.97, p <0.0001 for non- plasmid expressing wild-type β-catenin (42 ± 0.7) (Fig. 2a). These zero slope, linear regression analysis, Supplementary Fig. 1a) results support a model in which mutating Ser 33 to Phe in suggesting that treatment with BE3 and the S33-targeting sgRNA β-catenin increases Wnt signaling activity in mammalian cells. strongly enhances Wnt signaling levels in a base editing-dependent Next, we used base editing to install the β-catenin S33F mutation manner. in human cells and assayed the effect on Wnt signaling. We co- Finally, we assayed the ability of β-catenin base editing to transfected HEK293T cells with four plasmids: a BE3-expression increase the expression of known endogenous Wnt-responsive plasmid, an sgRNA-expression plasmid (either targeting β-catenin genes in HEK293T cells, including Axin-related protein (AXIN2), Ser 33, or encoding an unrelated control sgRNA), the transfection cyclin D1 (CCND1), cyclin dependent kinase inhibitor 1A efficiency reporter Renilla, and either the Topflash or Fopflash (CDKN1A), and fibroblast growth factor 20 (FGF20). Transfec- reporter plasmid. HEK293T cells transfected with plasmids tion of HEK293T cells with BE3 and β-catenin S33-targeting 4 NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications | | | Topflash:Fopflash Topflash:Fopflash luciferase ratio luciferase ratio 62.5 % of total sequencing reads with target C converted to T 62.5 Non-related sgRNA S33F SgRNA Non-related sgRNA S33F SgRNA 1000 NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 ARTICLE a b 1.0 % C to T % Indel 0.5 0 0.0 Molar ratio of Cas9:sgRNA:ssDNA Molar ratio of Cas9:sgRNA:ssDNA c d 15 BE3 (1:1.1) S33F mutation Ratio of CORRECT (1:1.1:0.5) % Indels efficiency S33F: indels Control CORRECT 4.9 ± 0.81 37 ± 1.0 0.13 HDR BE3 13 ± 0.28 0.53 ± 0.03 26 C C C C C C 1 6 8 15 16 20 CTNNB1: C1TGGAC TC TGGAATC C ATTC TGG 6 8 15 16 20 S33 Fig. 3 Comparison of efficiency and product selectivity of HDR vs. base editing following RNP delivery. RNP delivery into HEK293T cells of 200 nM Cas9 or 200 nM BE3 pre-complexed with the S33-targeting sgRNA or an unrelated sgRNA and delivered in a cationic liposome. a Frequency of S33F mutation (blue) and indels (green) from treatment with CORRECT HDR agents (Cas9, sgRNA, and ssDNA donor template) in the ratios shown. b Product selectivity ratio, defined as the ratio of S33F modification to indel modification, resulting from treatment with the CORRECT HDR agents used in a. c Comparison of target C to T conversion efficiency following RNP delivery of BE3 (blue) or CORRECT HDR (purple) and the S33-targeting sgRNA. The control corresponds to cells treated with BE3 protein and unrelated sgRNA. d Product selectivity of base editing and CORRECT HDR. Values and error bars reflect mean ± S.E.M. of three independent biological replicates performed on different days sgRNA resulted in 1.8 ± 0.21-fold higher expression of AXIN2, HDR with a donor DNA template that both installs the desired 1.7 ± 0.17-fold higher expression of CCND1, 1.7 ± 0.7-fold higher mutation and that alters the PAM sequence to prevent re-cutting expression of CDKN1A, and 12 ± 2.2-fold higher expression of of the desired DNA product. Lipid-mediated plasmid transfection FGF20 compared with cells treated with BE3 and an unrelated of Cas9:sgRNA constructs and an optimized amount of ssDNA sgRNA (Supplementary Fig. 3b). donor template into HEK293T cells resulted in levels of precise Taken together, these results demonstrate that treatment of installation of β-catenin S33F (5.4 ± 0.6%) and indels (49 ± 7%) cells with the β-catenin S33F base editor converts the endogenous consistent with previous reports (Fig. 1c). Since plasmid β-catenin gene into the S33F mutant allele, substantially increases transfection of BE3:sgRNA constructs resulted in 31 ± 0.9% levels of non-phosphorylated β-catenin, increases expression of a conversion of β-catenin Ser 33 to Phe and 2.0 ± 0.3% indels reporter gene downstream of TCF/LEF operators in a base (Fig. 1d), the product selectivity ratio (desired S33F mutation: editing-dependent manner, and also increases expression of undesired indel ratio) was 0.11 for CORRECT HDR and 16 for endogenous Wnt-responsive genes. base editing, a 140-fold difference. We recently demonstrated delivery of Cas9 nuclease ribonu- cleoprotein (RNP) complexes to the inner ear to mediate spatially Comparison of base editing and HDR. A commonly used localized genome editing in vivo with enhanced DNA specificity approach to introduce precise modifications into genomic DNA 11,41,42 relative to plasmid transfection . In light of these advan- is to harness HDR using a targeted nuclease and a donor DNA tages, we also compared RNP delivery of base editors with RNP template that contains the desired modification and is homo- delivery of HDR agents. We optimized RNP delivery-mediated logous to the target locus. We compared the outcomes of base CORRECT HDR by treating cells with different stoichiometries of editing to Cas9 nuclease-mediated HDR to install the S33F Cas9 protein, S33-targeting sgRNA, and a donor ssDNA mutation in β-catenin in HEK293T cells. For this comparison we template. Optimized cationic lipid-mediated delivery of Cas9, used the recently described CORRECT HDR method, which guide RNA, and donor DNA template resulted in an average introduces desired mutations into target loci more efficiently than HDR efficiency of 4.9 ± 0.8% S33F mutation using a 1:1.1:0.5 39, 40 previous HDR methods . First we performed CORRECT NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications 5 | | | % of total sequencing reads with % of total sequencing reads with target C converted to T target C-to-T or indel 1:1.1:0 1:1.1:0.25 1:1.1:0.5 1:1.1:1 1:1.1:2 No treatment Ratio of S33F mutation frequency to indel frequency 1:1.1:0 1:1.1:0.25 1:1.1:0.5 1:1.1:1 1:1.1:2 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 molar ratio of Cas9:S33-targeting sgRNA:donor ssDNA template potently deliver negatively charged proteins or protein:nucleic (Fig. 3a). In contrast, the efficiency of installing the S33F acid complexes, including Cas9:sgRNA ribonucleoprotein (RNP) mutation using BE3, the same guide RNA, and the same cationic complexes, into the organ of Corti . Because activation of Wnt lipid averaged 13 ± 0.3% (Fig. 3c). Consistent with the DNA signaling outside the inner ear can promote oncogenesis , such a transfection results above, the CORRECT HDR method was also local delivery platform that minimizes exposure of other cells to accompanied by a much higher indel frequency than that of the the editing agent is ideally suited for in vivo base editing of base editing approach; we observed 37 ± 1% indels from β-catenin. CORRECT HDR, but only 0.52 ± 0.03% indels from base editing We tested different ratios of base editor:sgRNA:lipid in vivo. (Fig. 3b,d). Therefore, the product selectivity ratio of S33F editing: The optimal ratio combined purified BE3 (57 µM), the β-catenin indels was 200-fold higher for base editing than for CORRECT S33-targeting sgRNA or an unrelated sgRNA (100 µM) and 2.0 µL HDR. Taken together, these results indicate that base editing cationic lipid in a total volume of 12.0 µL. After a 5-min either by DNA delivery or RNP delivery enables more efficient incubation, we injected 1.0 µL of the resulting mixture into the installation of the S33F mutation in β-catenin with fewer indels cochlea of postnatal day 1 (P1) wild-type CD1 mice (Fig. 4a). The and thus much higher product selectivity than the CORRECT intracochlear injection was followed by subcutaneous injection of HDR method. 5-ethynyl-2′-deoxyuridine (EdU) daily throughout the 5 days to enable detection of proliferating cells. Cochlear tissues were harvested at P7 and EdU, Myo7a (a marker for cochlear hair Off-target analysis of β-catenin S33F base editing. Genome cells), and Sox2 (a marker for supporting cells) were detected by editing agents can induce unintended DNA modifications at off- chemical staining and immunofluorescence. target genomic loci that are similar in sequence to the target Confocal microscopy of cochlear tissue harvested from mice 43–45 locus . Recent studies have shown that off-target base editing treated with BE3 and the β-catenin S33-targeting sgRNA revealed mediated by BE3 is generally a subset of off-target loci modified cells positive for both EdU and Sox2, consistent with newly by the corresponding Cas9 nuclease and guide RNA, as expected divided supporting cells (Fig. 4h). The post-mitotic status of the given that the DNA-binding capability of BE3 is derived from postnatal day 7 (P7) cochlear sensory epithelium was confirmed 3,10,11 Cas9 . We investigated the potential off-target genome by the lack of EdU incorporation in the organ of Corti, in contrast editing by Cas9 nuclease programmed by the S33F β-catenin with EdU incorporation in mesenchymal cells (e.g., tympanic sgRNA used herein with two methods. First, we used GUIDE- 22,52,53 border cells), which are known to be mitotic (Supplemen- Seq , an unbiased genome-wide method that has been exten- tary Fig. 2, orange arrowhead). A cochlea treated with BE3 and sively used to identify off-target loci in mammalian cells following the β-catenin S33-targeting sgRNA displayed multiple EdU- Cas9:sgRNA exposure (See Methods). We performed GUIDE-Seq positive supporting cells in the apical turn (n = 3, EdU and Sox2 on murine NIH/3T3 cells treated with Cas9:S33-targeting sgRNA. double-positive cells = 16 ± 5.3, Fig. 4h). We also observed EdU The on-target β-catenin locus was identified with 1108 GUIDE- and Sox2 double-positive cells expressing the hair cell marker Seq reads, corresponding to an on-target modification frequency Myo7a (Fig. 4m,n, blue arrows). Indeed, all EdU and Myo7a of 23%. Despite robust detection of on-target modification, we double-positive cells observed were also Sox2-positive (Fig. 4e,h, observed zero GUIDE-Seq reads corresponding to off-target m,n), consistent with transdifferentiation of supporting cells into modification following treatment with Cas9:S33-targeting sgRNA hair cells . The expansion of supporting cells (EdU and Sox2 (Supplementary Table 1), suggesting little or no Cas9-mediated double-positive cells) was within the inner pillar cell region and off-target modification by this guide RNA in NIH/3T3 cells. likely represented Lgr5-positive cells, consistent with previous As a complementary approach, we used the Cutting Frequency 22,28 reports . Determinant (CFD) algorithm to predict off-target loci in the In contrast, treatment of the cochlea with optimized COR- 42,46,47 mouse genome associated with S33-targeting sgRNA . RECT HDR reagents (1:1.1:0.5 molar ratio of Cas9: sgRNA: Following nucleofection of plasmids encoding Cas9 and S33- donor ssDNA), resulted in no evidence of newly divided targeting sgRNA into NIH/3T3 cells, we performed deep supporting cells (EdU- and Sox2-positive) or newly divided hair sequencing to measure indel frequency at the top ten computa- cells (EdU- and Myo7a-positive) (Fig. 4d,g,k,l), consistent with tionally predicted off-target loci (Supplementary Table 1). Con- the inefficiency of HDR in these post-mitotic cells. A control sistent with the GUIDE-Seq results, we observed no detectable cochlea treated with BE3 and an unrelated sgRNA also showed no indel formation (<0.05%) at any of the ten predicted off-target newly divided supporting cells or hair cells (Fig. 4c,f,i,j). The lack loci (Supplementary Table 1). of EdU-positive cells in the sensory epithelium excludes the These data collectively suggest that the S33-targeting sgRNA possibility of cell division resulting from lipid-mediated BE3 used to modify the β-catenin locus mediates few, if any, off-target protein delivery. Together, these results suggest that base editing editing events by Cas9 in murine cells. Since off-target base of Ser 33 to Phe in β-catenin, in contrast with Cas9 nuclease- editing is typically a subset of Cas9 off-target modification for a mediated HDR, can induce cell division and transdifferentiation 3,9–11 given sgRNA , these findings suggest that the changes in of supporting cells into hair cells in post-mitotic cells in vivo. This β-catenin phosphorylation state and Wnt signaling activity are difference can be attributed to mechanistic differences between unlikely to arise from off-target base editing. base editing and HDR-mediated editing, as the cellular machinery that mediates HDR is inactive or poorly expressed in non- In vivo base editing induces post-mitotic cell reprogramming. dividing cells such as the target cells in the sensory epithelium . Base editing relies on cellular mismatch repair machinery, which To visualize the location and distribution of cationic lipid- 48,49 is expressed in most cells , in contrast to the cellular DSB mediated protein and RNP delivery following intracochlear repair and recombination machinery that mediates HDR, which injection into the mouse inner ear of P1 mice, we performed is poorly expressed in non-mitotic cells . This difference raises analogous intracochlear injections of lipid complexed with the possibility that base editing may be effective in post-mitotic proteins. Cre-mediated recombination in Ai9 tdTomato mice cells in vivo, even though HDR in post-mitotic cells remains a results in tdTomato fluorescence. We injected (–30)GFP–Cre major challenge. complexed with lipid into these mice, and observed tdTomato We previously discovered that local in vivo injection of cationic fluorescence in supporting cells (Supplementary Fig. 4). We then lipid reagents normally used for nucleic acid transfection could performed injections into wild-type CD1 mice of lipid complexed 6 NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 ARTICLE a b Organ of Corti Other hair cells P1 mouse pups Inner hair cells Supporting cells & 10 inner-piller-cell region Cationic lipid BE3 + BE3 + CORRECT BE3:sgRNA nanoparticle unrelated sgRNA S33-targeting sgRNA HDR cd e BE3 + unrelated sgRNA CORRECT HDR BE3 + S33-targeting sgRNA Sox2/Myo7a/EdU f gh i jk l m n Fig. 4 Outcomes associated with in vivo RNP-mediated base editing in post-mitotic cochlea. a The cochlea of a postnatal day 1 (P1) mouse was injected with lipid nanoparticles encapsulating BE3 + sgRNA, CORRECT HDR agents, or controls. The next day, mice received 5-ethynyl-2-deoxyuridine (EdU) by subcutaneous injection once per day for 5 days. At postnatal day 7 (P7), one day after the fifth EdU injection, the cochlea was dissected and organ of Corti was visualized by chemical staining (to visualize EdU) and immunofluorescence (Myo7a and Sox2). In the cochlea, Myo7a (red) is expressed in hair cells and Sox2 (green) is expressed in supporting cells. EdU (white) marks newly divided cells. b Quantification of Sox2 and EdU double-positive cells at the apical region of the organ of Corti from mice treated with BE3 + S33-targeting sgRNA (n= 3), CORRECT HDR agents (n= 3), or BE3+ unrelated sgRNA (n = 3). Values and error bars reflect mean and S.E.M. c–n Images from the organ of Corti tissue of mice treated with BE3 + unrelated sgRNA (c, f, i, j); CORRECT HDR agents (d, g, k, l); or BE3+ S33-targeting sgRNA (e, h, m, n). The blue arrows point to triple-positive cells that had undergone proliferation and reprogramming to cells expressing Myo7a. c–e x–y plane of hair cell layers. f–h x–y plane of supporting cell layers. (i, k, m) x–z plane of samples in f, g, h at the dotted yellow lines. j, k, m x–z plane of samples in f, g, h at the dotted yellow lines, but with the Myo7a and Sox2 channels shown. Scale bar (white)= 25 μm with BE3 and fluorescein-labeled S33-targeting sgRNA and observed no substantial C-to-T conversion (≤0.25%) or indels observed fluorescein localized within the organ of Corti in (≤0.1%) in three regions of the cochlea injected with optimized regions containing supporting cells and hair cells (Supplementary CORRECT HDR agents (Fig. 5), consistent with the known Fig. 4). Injection of lipid complexed with fluorescein-labeled ineffectiveness of HDR in post-mitotic supporting cells . sgRNA without BE3 did not result in fluorescein signal, likely due Collectively, these results confirm that base editing, in contrast to sgRNA degradation. These observations suggest that intraco- to HDR, can mediate local installation of the β-catenin S33F chlear injection of cationic lipid-meditated protein or RNP mutation in post-mitotic sensory cells in vivo. delivery results in localized delivery within the cochlea, including supporting cells and hair cells of interest. Discussion High-throughput DNA sequencing of genomic DNA from This study establishes in vivo base editing of post-mitotic sensory bulk cochlear tissue of treated mice confirmed base editing of β- cells through the local injection of a base editor RNP:lipid com- catenin Ser 33 to Phe in three regions of the cochlea: the organ of plex into the inner ear of mice. Here, the resulting base editing Corti (2.8% of total sequencing reads containing S33F β-catenin), event precisely introduced a S33F mutation into β-catenin, the stria vascularis (3.0%), and the modiolus (0.7%), with low altering its ability to be phosphorylated and decreasing its indel formation (averaging 0.4% across all tissues) (Fig. 5). We degradation rate, thereby enhancing Wnt signaling in vitro and note that samples of cochlear cells from treated mice included in vivo. This single amino acid change was installed in cultured cells that were not exposed to base editor, and thus we expect the cells more efficiently and with far fewer undesired genome percentage of dissected tissue containing the S33F mutation to be modifications using base editing with BE3 than using HDR with substantially less than the frequency of base editing observed in Cas9 nuclease and a donor DNA template. Our observations cultured cells, consistent with previous reports . In contrast, we in vivo also reveal that base editing, but not HDR, can be used to NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications 7 | | | x–z Supporting cell layer x–y Hair cell layer x–y # of Sox EdU positive cells in the apex ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 ab 9 5 Organ of Corti Organ of Corti Stria vascularis Stria vascularis Modiolus Modiolus Control No treatment 0 0 BE3 CORRECT HDR BE3 CORRECT HDR C-to-T conversion in serine 33 Ctnnb1: TTGGATTC TGGAATC C ATTC TGG 8 15 16 20 S33 Fig. 5 In vivo S33F mutation of β-catenin induced by injection of base editor RNPs. a Tissue was harvested from the cochlea of mice injected with either BE3 + S33-targeting sgRNA, or with CORRECT HDR agents. HTS of genomic DNA isolated from tissue samples revealed the frequency of the S33F mutation. Note that because tissue samples contain cells not exposed to editing agents, the observed genome modification frequency in these samples is less than the editing efficiency of treated cells. b Indel frequency at the Ser 33 locus following treatments described in a. Values and error bars reflect mean ± S.E.M. of four mice injected with BE3 and four mice injected with CORRECT HDR agents. Control samples in a, b are organ of Corti from three contralateral uninjected ears effect physiological changes in the post-mitotic mammalian inner Cloning of plasmids. The sgRNA plasmids were generated by USER cloning. Phusion U Hot Start DNA Polymerase (Thermo Fisher) was used to replace desired ear, consistent with the lack of dependence of base editing on protospacers from the sgRNA template plasmid. The cDNA plasmids were gen- homologous recombination machinery. erated by site directed mutagenesis (New England Biolabs). Primers were designed In contrast with the use of base editing to directly restore the with overhang containing the desired point mutation sequence and used to amplify sequence of a mutated gene to that of the corresponding wild- from a previously reported β-catenin cDNA construct . PCR products were car- 3,9,16,17,54 ried out using NEB stable Competent cells (New England Biolabs). See Note S2 for type allele , this study establishes that base editing the a full list of primers used in this study. potential of a protein to undergo post-translational modification, in this case to block its ability to be phosphorylated and degraded, Expression and purification of BE3 protein. BE3 protein was prepared by can also achieve a desired physiological outcome. Such an overexpressing in BL21 Star (DE3)-competent E. coli cells using a plasmid approach offers a potential advantage over simple gene correction encoding the bacterial codon-optimized base editor with a His N-terminal pur- in cases in which a low level of protein alteration can exert an ification tag. Detailed purification steps are described in our previous study , and amplified physiological effect. In this application, the effects of β- the expression plasmid is available on Addgene (Note S1). After protein expression, bacteria cells were lysed by sonication and the lysate was cleared by centrifugation. catenin S33F base editing, amplified through the ability of a The cleared lysate was incubated with His-Pur nickel nitriloacetic acid (nickel- persistent transcription factor to mediate multiple transcriptional NTA) resin. The resin was washed before bound protein was eluted with elution events, greatly augments Wnt signaling, leading to detectable buffer. The resulting protein fraction was further purified on a 5 mL Hi-Trap HP changes in cell proliferation and cell state. SP (GE Healthcare) cation exchange column using an Akta Pure FPLC. Protein- containing fractions were concentrated using a column with a 100,000 kDa cutoff Our work establishes that local delivery of base editor as an (Millipore) centrifuged at 3,000 g and the concentrated solution was sterile filtered RNP complex into the cochlea enables a high degree of specificity through an 0.22 μm PVDF membrane (Millipore). both for the target DNA locus and for the target cells in the cochlea. These features are distinct from the delivery of diffusible In vitro transcription of sgRNA. PCR was performed using Q5 Hot Start High- 20,55,56,57 small molecules , which can perturb the activity of Fidelity DNA Polymerase (New England Biolabs) and primers as listed in the Note homologous protein targets and have a greater potential to affect S2 to linearize DNA fragments containing the T7 RNA polymerase promoter the homeostasis of other tissues in vivo. Another key feature of sequence upstream of the desired 20 bp sgRNA protospacer and the sgRNA backbone. HiScribe T7 Quick High Yield RNA Synthesis Kit (New England Bio- the RNP delivery approach used in this study is that precise labs) was used to transcribe sgRNA at 37 °C for 14–16 h with 1 µg of linear tem- genome alterations are made without exposing cells to exogenous plate per 20 µL reaction. Fluorescein-labeled sgRNA was transcribed by adding DNA or virus, preventing the possibility of random integration of 10% v/v Fluorescein RNA Labeling Mix (Sigma Aldrich) to the transcription sys- DNA into the host cell genome. tem. Purification of sgRNA was performed with MEGAClear Transcription Clean Up Kit (Thermo Fisher), following the manufacturer’s instructions. Purified Although activating the Wnt pathway in the cochlea is likely 57–60 sgRNAs (100 µM) were stored in aliquots at −80 °C. insufficient to restore function in a damaged cochlea , these findings suggest the potential to manipulate complex signaling Protein extraction and western blotting. Total proteins were extracted with pathways by a precise in vivo editing strategy. This approach has radioimmune precipitation assay buffer from whole cells. Cytoplasmic and nuclear potential for in vivo cellular reprogramming, and, in principle, protein extracts were prepared with NE-PER nuclear and cytoplasmic extraction may be applicable to other disorders for which current therapies reagents (Thermo Fisher) according to the manufacturer’s instructions. The pro- 58,59 tein lysates were separated on 4–12% NuPAGE Bis-Tris gels (Invitrogen) and are repeated dosing of small-molecule Wnt agonists . electrotransferred to 0.2 µm PVDF (polyvinylidene difluoride) membranes (Bio- Rad). The membranes were probed with rabbit anti-Non-phospho β-catenin Methods (1:1000, Cell signaling 4270), rabbit anti-β-catenin (1:2000, Sigma C2206), and Animal models. CD1-IGS mice and floxP-tdTomato mice were obtained from mouse anti-GAPDH (1:800, Millipore MAB374) followed by horseradish Charles River Laboratories. All animal experiments were approved by the Insti- peroxidase-conjugated anti-rabbit (Chemicon), or anti-mouse (Chemicon) anti- tutional Animal Care and the Use Committee of Massachusetts Eye and Ear. bodies. The blots were detected with ECL-Plus Western Blotting Substrate (Thermo Fisher) (See Supplementary Fig. 5). As noted in the Millipore data sheet, GAPDH resides in both the cytosol and nucleus, where GAPDH is translocated to Cell line authentication and quality control. HEK293T (American Type Culture the nucleus when cells respond to the initial stages of apoptosis or oxidative stress. Collection, ATCC CRL-3216) and NIH/3T3 (ATCC CRL-1658) were maintained in DMEM plus GlutaMax (Thermo Fisher) supplemented with 10% (v/v) fetal bovine serum, 100 U/mL penicillin/streptomycin (Thermo Fisher) at 37 °C with 5% Real-time quantitative PCR (RT-qPCR). For each biological replicate, total RNA CO . Cells were obtained from ATCC and were authenticated and verified to be was extracted from HEK293T and reverse-transcribed into cDNA with oligo(dT) free of mycoplasma by ATCC upon purchase. primers (Thermo Fisher) and Superscript III reverse transcriptase (Life 8 NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications | | | % of total sequencing reads with S33F mutation % total sequencing reads containing an indel NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 ARTICLE Technologies). RT-qPCR was performed on a LightCycler 96 (Roche) for genes of Tissue dissection for HTS. Three to five days after the BE3: sgRNA delivery, interest with Gapdh as the housekeeping gene. Ct values for genes were averaged cochlea tissues were collected by microdissection for high-throughput sequencing. from three technical replicates. Amplification primers are purchased from Taqman Tissues were dissected into the organ of Corti, stria vascularis and modiolus. Each probe: GAPDH (Hs02758991), AXIN2 (Hs00610344), CCND1 (Hs00765553), tissue was further dissected into between five and ten separate pieces, and DNA CDKN1A (Hs00355782), and FGF20 (Hs00173929). extraction was performed separately for each sample, followed by high-throughput sequencing as described above. The data presented in Fig. 5 shows sequencing data resulting from the extraction of one microdissected sample from each cochlear Plasmid transfection into cell lines. HEK293T cells were seeded on 48-well region. collagen-coated BioCoat plates (Corning) in an antibiotic-free medium. After 12 h, HEK293T cells were transfected at ~70% confluency. For BE3 or HDR-CORRECT Data analysis. Sequencing reads were demultiplexed using MiSeq Reporter plasmid transfection, 750 ng of editing agent plasmid and 250 ng of sgRNA plas- (Illumina). Indel frequencies were assessed using a previously described MATLAB mids, with or without ssDNA (0, 0.7, 1.3, 2.7 µg) were transfected using 1.5 µl of script , which counts indels of ≥1 base occurring in a 30-base window around the Lipofectamine 2000 (Thermo Fisher) per well. For Wnt activity, unless otherwise BE3 nicking site. Indels were defined as detectable if there was a significant dif- noted, 200 ng Topflash, 20 ng Renilla, 750 ng of BE and 250 ng of sgRNA ference (Student’s two-tailed t-test, p < 0.05) between indel formation in the treated expression plasmids were transfected using 1.5 µl of Lipofectamine 2000 (Thermo sample and untreated control. Base editing frequencies were further assessed using Fisher) per well according to the manufacturer’s protocol. Balancing pUC19 a previously described MATLAB script . In brief, reads which did not contain plasmid (New England Biolabs) was transfected to make constant total DNA insertions or deletions were aligned to an appropriate reference sequence via the amount across conditions. For GUIDE-seq, 500 ng of Cas9, 250 ng of sgRNA Smith-Waterman algorithm. Individual bases with an Illumina quality score less encoding plasmids, and 100 pmol dsODN were transfected into NIH/3T3 cells than or equal to 30 were converted to the placeholder nucleotide (N). This quality using LONZA 4D-Nucleofector with the EN-158 program according to the man- threshold results in nucleotide frequencies with an expected theoretical error rate of ufacturer’s protocols. 1 in 1000. This ensures that reads containing both base edits and indels are not counted as successful base-edits, and only analyzes the non-indel containing population of reads. To calculate the number of edited reads as a percentage of the Protein transfection into cell lines. HEK293T cells were seeded on 48-well col- total number of successfully generated sequencing reads, the percentage of non- lagen-coated BioCoat plates (Corning) in 250 µl of an antibiotic-free medium. After indel containing edited reads as measured from the alignment algorithm were 12 h, HEK293T cells were transfected at ~70% confluency. Base editor protein was multiplied by (1- fraction of reads containing an indel). incubated with 1.1 times molar excess of the necessary sgRNA at room temperature for 5 min. In parallel, Cas9 protein was incubated with 1.1 times molar excess of sgRNA and different specified molar of ssDNA at room temperature. We observed Data availability. High-throughput sequencing data that support the findings of that higher sgRNA concentration with constant BE3 concentration did not increase this study have been deposited in the NCBI Sequence Read Archive database under base editing efficiency by HTS in vitro. The complex was then incubated with 1.5 µl Accession number SRP136325. All other data are available upon reasonable Lipofectamine 2000 (Thermo Fisher) and transfected according to the manu- request. facturer’s protocol for plasmid delivery. BE3 and Cas9 protein were added to a final concentration of 200 nM (based on a total well volume of 275 µl). Received: 31 January 2018 Accepted: 8 May 2018 High-throughput sequencing. Genomic DNA was isolated using the Agencourt DNAdvance Genomic DNA Isolation Kit (Beckman Coulter) according to the manufacturer’s instructions. First DNA amplification was performed by quanti- tative PCR with Phusion U Hot Start and SYBR Gold Nucleic Acid Stain (Thermo Fisher) to the top of the linear range. PCR products were purified using RapidTips References (Diffinity Genomics). The second PCR was performed to attach Indexing Adapters 1. Cox, D. B. T., Platt, R. J. & Zhang, F. Therapeutic genome editing: prospects (Illumina). The products were gel-purified and quantified using the KAPA Library and challenges. Nat. Med. 21, 121–131 (2015). Quantification Kit-Illumina (KAPA Biosystems). Samples were sequenced using a 2. Doudna, J. A. & Charpentier, E. The new frontier of genome engineering with single-end read from 200–250 bases (depending on the amplicon size) on the CRISPR-Cas9. Science 346, 1258096 (2014). MiSeq (Illumina) according to the manufacturer’s instructions. 3. Komor, A. C., Kim, Y. B., Packer, M. S., Zuris, J. A. & Liu, D. R. Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage. Nature 533, 420–424 (2016). Intracochlear delivery of ribonucleoprotein. Intracochlear delivery was per- 4. Sander, J. D. & Joung, J. K. CRISPR-Cas systems for editing, regulating and formed in post-natal day one (P1) CD1 mice or floxP-tdTomato mice as described targeting genomes. Nat. Biotechnol. 32, 347–355 (2014). previously . Briefly, mice were anaesthetized by lowering the body temperature 5. Biot-Pelletier, D. & Martin, V. J. J. Seamless site-directed mutagenesis of the before the surgical procedure. A postauricular incision was made near the right ear, Saccharomyces cerevisiae genome using CRISPR-Cas9. J. Biol. Eng. 10,6 and the bulla was lifted to expose the cochlea. BE3 protein (57.7 μM stock con- (2016). centration) was pre-complexed with the sgRNA or fluorescein–sgRNA (100 μM 6. Lin, S., Staahl, B. T., Alla, R. K. & Doudna, J. A. Enhanced homology-directed stock concentration) in a 1:1.1 molar ratio and then mixed with Lipofectamine 2000 (Thermo Fisher) in a 1:1 volumetric ratio. For delivery of (−30) GFP-Cre in human genome engineering by controlled timing of CRISPR/Cas9 delivery. eLife 3, e04766 (2014). floxP-tdTomato mice, 3 µL of 45 μM protein was mixed with 3 µL of Lipofectamine 2000. The resulting solution (1.0 µL) was injected with a glass pipette (end dia- 7. Jeggo, P. A. & Lobrich, M. DNA double-strand breaks: their cellular and meter, 5 μm) attached to a nanolitre micropump (WPI, UMP3 + Micro4 + clinical impact? Oncogene 26, 7717–7719 (2007). −1 NanoFil) at the rate of 150–200 nL min through the cochlear capsule into scala 8. Ran, F. A. et al. Genome engineering using the CRISPR-Cas9 system. Nat. media at the cochlear basal turn. Controls included the contralateral side of the Protoc. 8, 2281–2308 (2013). uninjected cochlea, pups that received BE3 protein pre-mixed with an unrelated 9. Gaudelli, N. M. et al. Programmable base editing of A� TtoG� C in genomic sgRNA, or pups that received sgRNA-only. After injection, the incision was closed DNA without DNA cleavage. Nature 551, 464 (2017). and the mice were brought onto a heating pad to recover. 10. Kim, Y. B. et al. Increasing the genome-targeting scope and precision of base editing with engineered Cas9-cytidine deaminase fusions. Nat. Biotechnol. 35, 371–376 (2017). EdU incorporation for cell proliferation. Pups received 10 µl of 5-ethynyl-2- 11. Rees, H. A. et al. Improving the DNA specificity and applicability of base −1 deoxyuridine (EdU, 10 mg ml ) by subcutaneous injection once daily for 5 days. editing through protein engineering and protein delivery. Nat. Commun. 8, EdU incorporation was detected with Click-iT EdU Imaging Kit (Invitrogen) 15790 (2017). according to the manufacturer’s instructions. 12. Chadwick, A. C., Wang, X. & Musunuru, K. In vivo base editing of PCSK9 (proprotein convertase subtilisin/kexin Type 9) as a therapeutic alternative to genome editing. Arterioscler. Thromb. Vasc. Biol. 37, 1741–1747 (2017). Immunohistochemistry. The temporal bones of P7 pups with BE3 and EdU 13. Hess, G. T. et al. Directed evolution using dCas9-targeted somatic delivery were fixed with 4% paraformaldehyde at 4 °C overnight. Cochlear tissues hypermutation in mammalian cells. Nat. Methods 13, 1036–1042 (2016). were either dissected into apical, mid, and basal turn for whole-mounts or 14. Kim, K. et al. Highly efficient RNA-guided base editing in mouse embryos. embedded in optimal cutting temperature medium (OCT) for frozen serial sections Nat. Biotechnol. 35, 435–437 (2017). with cryostat (LEICA CM3050). After EdU detection, tissues were incubated with 15. Li, G. et al. Highly efficient and precise base editing in discarded human rabbit anti-MYO7a (1:500, Proteus Biomedical 25-6790) and goat anti-Sox2 (1:100, tripronuclear embryos. Protein Cell. 8, 776–779 (2017). Santa Cruz Biotechnology sc-17320) antibodies followed with secondary antibodies 16. Liang, P. et al. Correction of β-thalassemia mutant by base editor in human conjugated with Alexa Fluor 488, and 568 (Invitrogen A11055, A10042). Images embryos. Protein Cell.8, 811–822 (2017). were taken with a confocal microscopy (Leica TCS SP8) and processed with Image J (NIH). NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications 9 | | | ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04580-3 17. Nishida, K. et al. Targeted nucleotide editing using hybrid prokaryotic and 49. Pluciennik, A. et al. PCNA function in the activation and strand direction of vertebrate adaptive immune systems. Science 353, aaf8729 (2016). MutLα endonuclease in mismatch repair. Proc. Natl. Acad. Sci. 107, 18. Zong, Y. et al. Precise base editing in rice, wheat and maize with a Cas9- 16066–16071 (2010). cytidine deaminase fusion. Nat. Biotechnol. 35, 438–440 (2017). 50. Heyer, W.-D., Ehmsen, K. T. & Liu, J. Regulation of homologous 19. Park, D.-S. et al. Targeted base editing via RNA-guided cytidine deaminases in recombination in eukaryotes. Annu. Rev. Genet. 44, 113–139 (2010). xenopus laevis embryos. Mol. Cells 40, 823–827 (2017). 51. Klaus, A. & Birchmeier, W. Wnt signaling and its impact on development and 20. Roccio, M., Hahnewald, S., Perny, M. & Senn, P. Cell cycle reactivation of cancer. Nat. Rev. Cancer 8, 387–398 (2008). cochlear progenitor cells in neonatal FUCCI mice by a GSK3 small molecule 52. Jan, T. A. et al. Tympanic border cells are Wnt-responsive and can act as inhibitor. Sci. Rep. 5, 17886 (2015). progenitors for postnatal mouse cochlear cells. Development 140, 1196–1206 21. Chai, R. et al. Wnt signaling induces proliferation of sensory precursors in the (2013). postnatal mouse cochlea. Proc. Natl. Acad. Sci. 109, 8167–8172 (2012). 53. Taniguchi, M., Yamamoto, N., Nakagawa, T., Ogino, E. & Ito, J. Identification 22. Shi, F., Hu, L. & Edge, A. S. B. Generation of hair cells in neonatal mice by β- of tympanic border cells as slow-cycling cells in the cochlea. PLoS ONE 7, catenin overexpression in Lgr5-positive cochlear progenitors. Proc. Natl. Acad. e48544 (2012). Sci. 110, 13851–13856 (2013). 54. Chadwick, A. C., Wang, X. & Musunuru, K. In vivo base editing of PCSK9 23. Chai, R. et al. Dynamic expression of Lgr5, a Wnt target gene, in the (proprotein convertase subtilisin/Kexin Type 9) as a therapeutic alternative to developing and mature mouse cochlea. JARO J. Assoc. Res. Otolaryngol. 12, genome editing. Arterioscler. Thromb. Vasc. Biol. 37, 1741 (2017). 455–469 (2011). 55. Wang, M. et al. Efficient delivery of genome-editing proteins using 24. Waqas, M., Zhang, S., He, Z., Tang, M. & Chai, R. Role of Wnt and Notch bioreducible lipid nanoparticles. Proc. Natl Acad. Sci. 113, 2868–2873 (2016). signaling in regulating hair cell regeneration in the cochlea. Front. Med. 10, 56. DeJonge, R. E. et al. Modulation of Wnt signaling enhances inner ear organoid 237–249 (2016). development in 3D culture. PLoS ONE 11, e0162508 (2016). 25. Geng, R. et al. Comprehensive expression of Wnt signaling pathway genes 57. Jacques, B. E. et al. A dual function for canonical Wnt/β-catenin signaling in during development and maturation of the mouse cochlea. PLoS ONE 11, the developing mammalian cochlea. Development 139, 4395 (2012). e0148339 (2016). 58. Leeds, P. R. et al. A new avenue for lithium: intervention in traumatic brain 26. MacDonald, B. T., Tamai, K. & He, X. Wnt/β-catenin signaling: components, injury. ACS Chem. Neurosci. 5, 422–433 (2014). mechanisms, and diseases. Dev. Cell 17,9–26 (2009). 59. Medina, M., Garrido, J. J. & Wandosell, F. Modulation of GSK-3 as a 27. Liu, C. et al. β-Trcp couples β-catenin phosphorylation-degradation and therapeutic strategy on Tau pathologies. Front. Mol. Neurosci. 4, 24 (2011). regulates Xenopus axis formation. Proc. Natl. Acad. Sci. 96, 6273–6278 (1999). 60. Shi, F., Cheng, Y.-F., Wang, X. L. & Edge, A. S. B. β-catenin up-regulates 28. McLean, W. J. et al. Clonal expansion of Lgr5-positive cells from mammalian Atoh1 expression in neural progenitor cells by interaction with an Atoh1 3′ cochlea and high-purity generation of sensory hair cells. Cell Rep. 18, enhancer. J. Biol. Chem. 285, 392–400 (2010). 1917–1929 (2017). 61. Hu, L. et al. Diphtheria toxin-induced cell death triggers Wnt-dependent hair 29. Eldar-Finkelman, H. & Martinez, A. GSK-3 inhibitors: preclinical and clinical cell regeneration in neonatal mice. J. Neurosci. 36, 9479 (2016). focus on CNS. Front. Mol. Neurosci. 4, 32 (2011). 30. Polakis, P. Wnt signaling and cancer. Genes Dev. 14, 1837–1851 (2000). Acknowledgements 31. Ring, A., Kim, Y.-M. & Kahn, M. Wnt/catenin signaling in adult stem cell D.R.L. gratefully acknowledges support from DARPA HR0011-17-2-0049; NIH RM1 physiology and disease. Stem Cell Rev. 10, 512–525 (2014). HG009490, R01 EB022376, and R35 GM118062; and HHMI. H.A.R. is supported by the 32. Li, VivianS. W. et al. Wnt signaling through inhibition of β-catenin Kilpatrick Educational fund from the Chemistry and Chemical Biology Department, degradation in an intact Axin1 complex. Cell 149, 1245–1256 (2012). Harvard University. A.S.B.E. is grateful for the support from NIH R01 DC014089, NIH 33. Liu, C. et al. Control of β-catenin phosphorylation/degradation by a dual- R01 DC007174, and the Shulsky Foundation. kinase mechanism. Cell 108, 837–847 (2002). 34. Verheyen, E. M. & Gottardi, C. J. Regulation of Wnt/β-catenin signaling by protein kinases. Dev. Dyn. Off. Publ. Am. Assoc. Anat. 239,34–44 (2010). Author contributions 35. Shi, F., Kempfle, J. S. & Edge, A. S. B. Wnt-responsive Lgr5-expressing stem W-H.Y., A.S.B.E., and D.R.L. designed the research. W-H.Y., H.C., and H.A.R. conducted cells are hair cell progenitors in the cochlea. J. Neurosci. 32, 9639 (2012). the experiments. W-H.Y., A.S.B.E., and D.R.L. analyzed the data. A.S.B.E. and D.R.L. 36. Moon, R. T., Kohn, A. D., Ferrari, G. V. D. & Kaykas, A. WNT and β-catenin supervised the research. All authors contributed to writing the manuscript. signaling: diseases and therapies. Nat. Rev. Genet. 5, 691 (2004). 37. Sulg, M. et al. Differential sensitivity of the inner ear sensory cell populations to forced cell cycle re-entry and p53 induction. J. Neurochem. 112, 1513–1526 Additional information (2010). Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- 38. Veeman, M. T., Slusarski, D. C., Kaykas, A., Louie, S. H. & Moon, R. T. 018-04580-3. Zebrafish prickle, a modulator of noncanonical Wnt/Fz signaling, regulates gastrulation movements. Curr. Biol. 13, 680–685 (2003). Competing interests: D.R.L. is a consultant and co-founder of Editas Medicine, Beam 39. Kwart, D., Paquet, D., Teo, S. & Tessier-Lavigne, M. Precise and efficient Therapeutics, and Pairwise Plants, companies that are using genome editing. A.S.B.E. is a scarless genome editing in stem cells using CORRECT. Nat. Protoc. 12, consultant and co-founder of Decibel Therapeutics. The remaining authors declare no 329–354 (2017). competing interests. 40. Paquet, D. et al. Efficient introduction of specific homozygous and heterozygous mutations using CRISPR/Cas9. Nature 533, 125–129 (2016). Reprints and permission information is available online at http://npg.nature.com/ 41. Gao, X. et al. Treatment of autosomal dominant hearing loss by in vivo reprintsandpermissions/ delivery of genome editing agents. Nature. 553, 217–221 (2017). 42. Zuris, J. A. et al. Cationic lipid-mediated delivery of proteins enables efficient Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in protein-based genome editing in vitro and in vivo. Nat. Biotechnol. 33,73–80 published maps and institutional affiliations. (2015). 43. Fu, Y. et al. High-frequency off-target mutagenesis induced by CRISPR-Cas nucleases in human cells. Nat. Biotechnol. 31, 822–826 (2013). 44. Pattanayak, V. et al. High-throughput profiling of off-target DNA cleavage Open Access This article is licensed under a Creative Commons reveals RNA-programmed Cas9 nuclease specificity. Nat. Biotechnol. 31, Attribution 4.0 International License, which permits use, sharing, 839–843 (2013). adaptation, distribution and reproduction in any medium or format, as long as you give 45. Tsai, S. Q. et al. GUIDE-seq enables genome-wide profiling of off-target appropriate credit to the original author(s) and the source, provide a link to the Creative cleavage by CRISPR-Cas nucleases. Nat. Biotechnol. 33, 187–197 (2015). Commons license, and indicate if changes were made. The images or other third party 46. Davis, K. M., Pattanayak, V., Thompson, D. B., Zuris, J. A. & Liu, D. R. Small material in this article are included in the article’s Creative Commons license, unless molecule-triggered Cas9 protein with improved genome-editing specificity. indicated otherwise in a credit line to the material. If material is not included in the Nat. Chem. Biol. 11, 316–318 (2015). article’s Creative Commons license and your intended use is not permitted by statutory 47. Haeussler, M. et al. Evaluation of off-target and on-target scoring algorithms regulation or exceeds the permitted use, you will need to obtain permission directly from and integration into the guide RNA selection tool CRISPOR. Genome Biol. 17, the copyright holder. To view a copy of this license, visit http://creativecommons.org/ 148 (2016). licenses/by/4.0/. 48. Heller, R. C. & Marians, K. J. Replisome assembly and the direct restart of stalled replication forks. Nat. Rev. Mol. Cell Biol. 7, 932–943 (2006). © The Author(s) 2018 10 NATURE COMMUNICATIONS (2018) 9:2184 DOI: 10.1038/s41467-018-04580-3 www.nature.com/naturecommunications | | |

Journal

Nature CommunicationsSpringer Journals

Published: Jun 5, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off