Repair of the TGFBI gene in human corneal keratocytes derived from a granular corneal dystrophy patient via CRISPR/Cas9-induced homology-directed repair

Granular corneal dystrophy (GCD) is an autosomal dominant hereditary disease in which multiple discrete and irregularly shaped granular opacities are deposited in the corneal stroma. GCD is caused by a point mutation in the transforming growth factor-β-induced (TGFBI) gene, located on chromosome 5q31. Here, we report the first successful application of CRISPR-Cas9-mediated genome editing for the correction of a TGFBI mutation in GCD patient-derived primary corneal keratocytes via homology-directed repair (HDR). To correct genetic defects in GCD patient cells, we designed a disease-specific guide RNA (gRNA) targeting the R124H mutation of TGFBI, which causes GCD type 2 (GCD2). An R124H mutation in primary human corneal keratocytes derived from a GCD2 patient was corrected by delivering a CRISPR plasmid expressing Cas9/gRNA and a single-stranded oligodeoxynucleotide HDR donor template in vitro. The gene correction efficiency was 20.6% in heterozygous cells and 41.3% in homozygous cells. No off-target effects were detected. These results reveal a new therapeutic strategy for GCD2; this method may also be applicable to other heredity corneal diseases.

opacities usually recur within several years 8 . Compared with PTK, corneal clarity can be retained for longer durations using keratoplasty, but opacity eventually occurs via the gradual invasion of host corneal cells, especially in homozygous patients 9 . Thus, the development of a radical treatment is needed. GCD2 is typically associated with an R124H (histidine replacing arginine) point mutation in the TGFBI gene; accordingly, a gene therapy approach may be effective.
CRISPR/Cas9 (clustered, regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein)-mediated genome editing has been increasingly applied to repair mutated genome sequences 10 . This versatile tool for genome engineering enables the induction of site-specific double-strand breaks (DSBs) using guide RNAs (gRNAs) [11][12][13][14][15] . DSBs can be repaired by two major pathways, non-homologous end joining and homology-directed repair (HDR). In the presence of exogenous donor DNA as a repair template, DSBs can be repaired precisely via the HDR pathway. This technique is useful for codon replacements or reporter insertions 16,17 . For small genetic modifications, such as point mutations, the application of single-stranded oligodeoxynucleotides (ssODNs) as HDR templates shows higher editing efficiency than that of plasmid donors 18 . Here, we report the first CRISPR-mediated HDR using cultured corneal keratocytes derived from an R124H GCD2 patient. The results of this study have important clinical implications given the lack of effective treatment options for GCD2.

Results
Gene targeting strategy and construction for CRISPR/Cas9-mediated HDR of an R124H mutation. To develop an efficient strategy to repair the genetic mutation in GCD using CRISPR/Cas9, we used human cultured corneal keratocytes derived from an R124H GCD2 patient as a model system. The TGFBI R124H mutant keratocytes have a monoallelic point mutation at Arg124 (GCA→ACA) in Exon 4 of TGFBI (Fig. 1a). To repair mutant R124H cells, we designed an R124H mutation-specific gRNA based on a public algorithm (Fig. 1b). Then, the designed gRNAs were computationally evaluated for potential off-target effects using the E-CRISP algorithm. The gRNA with the lowest off-target risk was selected for subsequent analyses.
For the HDR repair template, we synthesized a 100-nucleotide (nt) donor repair template ssODN with a novel BsiWI restriction site (Fig. 1b). The substitutions ensured that the sequence of the wild-type donor template was resistant to CRISPR/Cas9 cleavage by the R124H mutation-specific gRNA, and the BsiWI restriction site allowed the tracking of HDR by restriction fragment length polymorphism (RFLP) (Fig. 1b). A pair of annealed oligos encoding a target sequence of R124H mutation-specific gRNA was cloned into the px458 vector, which enabled bicistronic expression of Streptococcus pyogenes Cas9 (spCas9) and green fluorescence protein (GFP) (Fig. 1c).

CRISPR/Cas9-mediated HDR of an R124H mutation in human corneal keratinocytes. The
CRISPR plasmid expressing spCas9/gRNA was co-transfected into primary R124H mutant human corneal keratinocytes with the ssODN as a donor template. After 7 days, single GFP-expressing cells were harvested, added to individual wells of a 96-well plate, and clonally expanded. Then, the presence of a novel BsiWI restriction site was examined by RFLP-based genotyping. Genomic PCR products for wild-type alleles were not cleaved by BsiWI (Fig. 2a). However, genomic PCR products for several transfected colonies were cleaved by BsiWI, suggesting target site alterations by HDR (Fig. 2a). We also confirmed the genomic sequences of the PCR products (Fig. 2b).
The sequence of wild-type cells had CGC, specifying arginine, at the 124 th amino acid, and R124H mutant cells had CAC at this position. Neither is expected to be cleaved by BsiWI; however, gene-edited cells have CGT, which is expected to be cut at CGTAC. In an RFLP assay, we detected cells with heterozygous and homozygous editing, as shown in Fig. 2c.
Efficiency of Cas9-mediated genome editing of the TGFBI R124H mutant gene. To examine the editing efficiency of the R124H mutant TGFBI gene, genomic DNA was extracted from the clonally expanded cells in 96-well plates and examined by RFLP-based genotyping. Owing to the low growth rate and viability of flow cytometry-sorted primary keratinocytes, not all cells were sufficiently expanded by single-cell cloning in 96-well plates. Cell growth and gene editing efficiency are summarised in Fig. 2c. Thirty-eight out of 192 clones were sufficiently expanded and examined by RFLP. Among all examined clones, 20.6% exhibited monoallelic TGFBI correction and 41.3% showed biallelic correction. Accordingly, 62% showed clear TGFBI R124H allele correction derived from the HDR template.

Analysis of off-target cleavage by R124H mutation-specific gRNA.
To evaluate off-target effects mediated by the gRNA, a T7 endonuclease (T7EN1) cleavage assay was used to assess off-target cleavage. Since we rigorously designed and selected a TGFBI-specific gRNA to reduce the risk of off-target effects, only 3 potential off-target sites were found for the gRNA (Fig. 3a,b). We could not find any potential off-target sites (OTS) with mismatches of less than 3 nt. The 3 potential OTS had mismatches of more than 4 nt with the TGFBI gRNA (Fig. 3b). In the T7EN1 cleavage assay, we did not detect any off-target effects at the 3 OTS (Fig. 3c).

Discussion
Current therapeutic modalities for GCD, i.e. PTK and keratoplasty, are invasive and are associated with frequent recurrence. The correction of TGFBI mutations in the local cornea may be a radical treatment for GCD patients, minimizing progression and the recurrence of corneal opacities. In this study, we successfully repaired point mutations in R124H mutant cells using CRISPR/Cas9 and HDR in vitro, without detectable off-target effects.
The CRISPR/Cas 9 system is an efficient tool for genome engineering and disease treatment. Kaminski et al. successfully eliminated HIV genomes in human T cells ex vivo 19 ; they reported a low editing efficiency in primary culture, even using a lentivirus delivery system 19 . Similarly, the low transfection and growth efficiency in this study (Fig. 2c) may be attributed to the use of primary culture cells. However, in general, plasmid transfection may be safer than viral transfection in vivo. Fortunately, despite the low growth rate, our results reveal that the efficiency of CRISPR/Cas9 in gene correction was higher compared with those of previous studies 15 . It is reported that the efficacy of HDR is generally not high, however, the efficiency of HDR using asymmetric donor DNA is much higher (maximum 60%) than that of conventional HDR 20 . In this study, the efficacy of HDR using ssODN was greater than 60%. The reason for the high efficiency of HDR in our study is unclear, but may be explained by the unique characteristics of the DNA repair ability of corneal epithelial cells. Previously, Mallet et al. 21 demonstrated that DNA damage in human corneal epithelial cells by ultraviolet radiation could be repaired faster than that in epidermal keratinocytes. This suggests that there are corneal-specific mechanisms in DSB repair. This issue should be evaluated in future studies.
CRISPR/Cas9 itself has some probability of causing off-target mutations 22,23 . CRISPR RNA-guide endonucleases tolerate single and double mismatches in their sequences at the gRNA interface in bacterial cells 22 and human cells 23 . Wu et al. 24 reported that only 2 out of 12 samples had off-target mutations when they co-injected Cas9 mRNA and a single gRNA into mouse zygotes with dominant mutations in Crygc that cause cataracts. Additionally, off-target mutations were detected at 1 of 10 potential OTS in the two samples. Thus, although off-target effects are an important safety issue for clinical use, they can be greatly reduced by a cautious gRNA sequence design 25,26 . Moreover, according to previous studies, gRNA does not cleave nonspecific targets with mismatches of 3 nt or more 23,27 . In our study, based on these findings, we made highly specific gRNAs using an off-target prediction tool. A T7 Endonuclease 1 cleavage assay was performed to examine off-targets effects, but the three predicted OTS were not detected in any sample (Fig. 3c).
In ocular tissues, several reports have demonstrated successful gene editing using the CRISPR/Cas9 system 28-31 . Wu et al. corrected a genetic disease in mice that show early-onset cataracts using non-homologous end joining and HDR 24 . The gene correction was conducted at the embryonic stage and cataracts occurred in 10 out of 12 mice. Wang et al. 32 and Bakondi et al. 30 successfully edited retinal genes by electroporation, and Hung et al. 28 also successfully edited retinal genes using a virus delivery system. In the cornea, Courtney et al. reported the effectiveness of DNA cleavage by CRISPR/Cas9 for the treatment of cornea dystrophy caused by a KRT12 mutation 29 . To our knowledge, this study is the first to demonstrate in vitro gene correction in mutant human primary corneal cells using CRISPR/Cas9 and HDR. The cornea is an excellent tissue for the application of genome editing therapy owing to its accessibility and high amenability to naked plasmid DNA transfection via intrastromal injection 33 . Thus, gene editing is a radical GCD treatment and the in vivo application of this system is ideal for clinical settings in which conventional treatments are limited. In the future, it is necessary to develop safer and more efficient methods to modify local corneal genes in vivo.
In conclusion, we used CRISPR-Cas9-mediated HDR to correct the R124H mutation. Our data suggest that the approach is highly specific, with no observed off-target effects. Given the lack of effective treatment options for GCD2, this gene editing system is a potentially radical treatment for TGFBI-related corneal dystrophy and can be used to protect corneal opacities. The in vivo application of this system is an important future challenge.

Methods
Cell culture. Primary human corneal keratocytes of a GCD2 patient with a heterozygous TGFBI mutation (R124H) were isolated from a surgical specimen during deep anterior lamellar keratoplasty. Ethics approval for this work was obtained from the Institutional Review Board of the Inoue Eye Hospital and informed consent was obtained from the patient. All tissues were provided form Inoue Eye Hospital and no tissues were procured from prisoners. All the experimental methods were carried out in accordance with the guidelines verified and approved by the Ethics Committee of The University of Tokyo.

G A C C C T G G G A G T C G T T G G A T C C A C C A C C A C T C A G C T G T A C A C G G AC C G T A C G G A GA A G CT GA GG CC TGAGATGGAGGGGCCCGGCAGCTTCACCATCT-3′.
Transfection and cloning. CRISPR-Cas9 constructs (2.5 µg per well) and ssODN (1 µg per well) were transfected into R124H primary cells using FuGENE (Promega, Madison, WI, USA) according to the manufacturer's instructions and the cells were incubated for an additional 48 h. Images were obtained by fluorescence microscopy (BZ-9000; Keyence, Osaka, Japan). The cells expressing GFP were single-cell-sorted by FACS (Aria III, Becton-Dickinson, Franklin Lakes, NJ, USA) at 1 week after transfection. The sorted cells were then clonally expanded and analysed as described below.

Indel analysis by restriction fragment length polymorphism (RFLP).
Total DNA was extracted from cells using the Nucleospin Kit (Takara Bio Inc.). Polymerase chain reaction (PCR) using specific primer sets (Forward: 5′-GTTGAGTTCACGTAGACAGGC-3′, Reverse: 5′-GACTCCCATTCATCATGCCCA-3′) was performed to amplify the DNA using the KOD FX Kit (KOD FX; Toyobo, Osaka, Japan) with the following temperature profile: 94 °C for 2 min, followed by 40 cycles of 98 °C for 10 s and 55 °C for 30 s, and 72 °C for 2 min. The PCR products were treated with the restriction enzyme BsiWI (New England Biolabs, Ipswich, MA, USA). One microgram of DNA was treated with 1 unit of enzyme and NE Buffer 2.1 at 55 °C for 15 min. The samples were analysed by electrophoresis on a 5% polyacrylamide TBE gel.