NATURE BIOTECHNOLOGY VOLUME 36 NUMBER 9 SEPTEMBER 2018 839 BRIEF COMMUNICATIONS 1 Horae Gene Therapy Center, University of Massachusetts Medical School, Worcester, Massachusetts, USA. 2 Li Weibo Institute for Rare Diseases Research, University of Massachusetts Medical School, Worcester, Massachusetts, USA. 3 Department of Microbiology and Physiological Systems, University of Massachusetts Medical School, Worcester, Massachusetts, USA. 4 RNA Therapeutics Institute, University of Massachusetts Medical School, Worcester, Massachusetts, USA. 5 Howard Hughes Medical Institute, University of Massachusetts Medical School, Worcester, Massachusetts, USA. 6 Viral Vector Core, University of Massachusetts Medical School, Worcester, Massachusetts, USA. 7 Program in Molecular Medicine and Department of Molecular, Cell and Cancer Biology, University of Massachusetts Medical School, Worcester, Massachusetts, USA. 8 State Key Laboratory of Biotherapy, West China Hospital, Sichuan University, Chengdu, Sichuan, China. Correspondence should be addressed to G.G. (guangping.gao@umassmed.edu), W.X. (wen.xue@umassmed.edu), or P.D.Z. (phillip.zamore@umassmed.edu). Received 12 June 2017; accepted 1 June 2018; published online 13 August 2018; doi:10.1038/nbt.4219 Cas9-mediated gene editing promises to correct DNA mutations underlying human diseases. In principle, many mutations can be indi- vidually corrected by homology-directed repair (HDR) using an exog- enous DNA template. However, monogenic recessive genetic diseases typically involve many distinct mutations present across the gene. For example, more than 95 known mutations in the FAH gene can cause hereditary tyrosinemia type 1 (HT1) 1,2 . Mutation-independent gene repair strategies that avoid the need to deliver an exogenous DNA tem- plates would substantially simplify the development of therapies. In compound heterozygotes, each allele of the mutant gene har- bors a different genetic lesion. A sizable fraction of human autosomal recessive genetic disorders—including HT1 and lysosomal diseases— are caused by such compound heterozygous mutations 1,3 . We hypoth- esized that reconstituting the correct genetic information present on both compound heterozygous mutant alleles into one allele could create a functional copy of the disease gene (Fig. 1a). Targeted nucle- ases 4–8 have been used to induce reciprocal translocation between non-homologous chromosomes in vitro and in vivo, aiming to study the mechanisms of chromosome translocation or model certain types of cancer. Here, we used Cas9 to create double-stranded DNA breaks Cas9-mediated allelic exchange repairs compound heterozygous recessive mutations in mice Dan Wang 1–3 , Jia Li 1,2 , Chun-Qing Song 4 , Karen Tran 1,2 , Haiwei Mou 4 , Pei-Hsuan Wu 4,5 , Phillip WL Tai 1–3 , Craig A Mendonca 1,2 , Lingzhi Ren 1,2 , Blake Y Wang 1,2 , Qin Su 6 , Dominic J Gessler 1–3 , Phillip D Zamore 4,5 , Wen Xue 4,7 & Guangping Gao 1–3,8 We report a genome-editing strategy to correct compound heterozygous mutations, a common genotype in patients with recessive genetic disorders. Adeno-associated viral vector delivery of Cas9 and guide RNA induces allelic exchange and rescues the disease phenotype in mouse models of hereditary tyrosinemia type I and mucopolysaccharidosis type I. This approach recombines non-mutated genetic information present in two heterozygous alleles into one functional allele without using donor DNA templates. in both chromosomal homologs in a compound heterozygous mouse model of HT1, thereby inducing allelic exchange between two differ- ent mutant alleles and rescuing the disease phenotype. HT1 is caused by loss-of-function mutations in the FAH gene, which encodes fumarylacetoacetate hydrolase (FAH), an enzyme required for tyrosine catabolism. Loss of FAH causes accumula- tion of toxic tyrosine metabolites in the liver and other organs. The toxicity can be relieved by oral administration of 2-(2-nitro-4- trifluoromethylbenzoyl)-1,3-cyclohexanedione (NTBC), which blocks tyrosine catabolism at an early step, and thus prevents the formation of toxic intermediates 9 . To test our strategy, we used two HT1 mouse models that carry different mutations in the Fah gene. The Fah neo/neo mouse carries an insertion of a neomycin-resistance gene (neo) in exon 5, while the Fah PM/PM mouse carries a G-to-A transition that causes exon 8 skipping during splicing. Neither Fah mutant produces FAH protein detectable by immunohistochemistry (IHC). By crossing homozygotes of each mutation, we generated compound heterozygous Fah neo/PM mice (Fig. 1b). To generate genomic DNA breaks promoting allelic exchange, newborn Fah neo/PM mice were systematically treated with a pair of liver-tropic recombinant adeno-associated virus (rAAV) vectors: one expressing Streptococcus pyogenes Cas9 (rAAV9-SpCas9) and the other producing a single guide RNA (sgRNA) targeting Fah intron 7 (scAAV8-sgFah; Supplementary Fig. 1a–c). As a control, we replaced sgFah with an sgRNA targeting intron 2 of the Aspa gene (scAAV8- sgAspa). To allow sufficient time for genome editing to occur, mice were maintained on drinking water containing NTBC until 5 weeks old (Fig. 1c). We euthanized two mice from each group at 5 weeks old, and confirmed that both the Fah and Aspa sgRNAs had induced insertions or deletions (indels) at the targeted genomic sites in the liver (Supplementary Fig. 1d–f). FAH IHC of liver tissue sections detected clusters of FAH-positive hepatocytes in the sgFah-treated mice but not the sgAspa control mice (Fig. 1d). Reverse-transcription PCR (RT-PCR) of total liver RNA revealed the presence of wild-type Fah mRNA spanning exon 5 to exon 9 in the sgFah but not the sgAspa control mice (Fig. 1e). To determine whether sufficient functional FAH protein was pro- duced by the Cas9/sgFah-treated mice, we withdrew NTBC from a subset of treated mice 5 weeks after birth. All control mice lost >20% of body weight within 1 month after NTBC withdrawal owing to liver failure and were euthanized (Fig. 2a). Naive compound hetero- zygous mice receiving no rAAV treatment suffered similar weight loss within 1 month (Supplementary Fig. 2). In contrast, the sgFah- treated mice initially lost weight, but eventually achieved normal,