Precise human genome editing using circular donor DNA and a cutting enzyme

The fibrosarcoma cell line HT1080 was used to evaluate circular donor DNA cleavage-induced targeted duplication and natural replacement. Other cell lines used in this study were derived from HT1080 cells. The cells were cultured in an atmosphere of 37 °C/5% CO₂ in plastic tissue culture dishes and plates. Dulbecco’s modified Eagle’s medium (DMEM), supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin, was used for routine passages and expansion of cells. Hygromycin (100 μg/mL) was used for the selection of donor DNA-integration clones, and 7.5 μg/mL 6-thioguanine was added for the selection of targeted duplication clones into the HPRT locus. Ganciclovir (1 μM; InvivoGen) or HAT supplement (50×) (Gibco) was used for the selection of natural replacement clones derived from a targeted duplication clone. For several experiments, advanced Dulbecco’s modified Eagle’s medium (advanced DMEM), supplemented with 1× GlutaMAX, 2% fetal bovine serum, and 1% penicillin/streptomycin, was used for routine passage and expansion of the cells. When an advanced DMEM-based culture medium was used, 50 μg/mL hygromycin, 7.5 μg/mL 6-thioguanine, and 0.1 μM ganciclovir were added for the selections. HT1080 and its derivatives were stored at −80 °C as follows: cultured cells were treated with a mixture of phosphate-buffered saline (PBS) and a dissociation reagent, 0.25 w/v% trypsin–1 mM EDTA (FUJIFILM Wako Pure Chemical Corporation), for 5 min at room temperature (trypsinization); the dissociation reaction was terminated by adding a 5- to 10-fold volume of DMEM-based culture medium, followed by pipetting and centrifugation, and the cells were resuspended in CELLBANKER 1 (TaKaRa Bio) solution, followed by storing several tubes (0.5–2 × 10⁶ cells/tube) at −80 °C.

pMB1KmHygTK: This plasmid is the vector to transfect a donor DNA. First, pSelect-zeo-HSV1tk (InvivoGen), cleaved by EcoRI and PstI to remove the ZEO gene, was ligated with the fragment including the Km^R gene constructed using KOD-PLUS-DNA polymerase (TOYOBO)-driving PCR on the pCR-BluntII-TOPO plasmid DNA (Invitrogen) with the primer pair 5′-CTT​AAT​TAA​CCT​GCA​GCC​GGA​ATT​GCC​AGC​TG-3′ and 5′-ATG​TGG​TAT​GGA​ATT​CGG​TGG​CCC​TCC​TCA​CGT​GC-3′ using the In-Fusion kit (TOYOBO) to obtain pSelect-Km-HSV1tk (1). On the other hand, pSelect-zeo-HSV1tk plasmid DNA (InvivoGen), cleaved by NcoI and SphI followed by blunt-ended treatment with T4 DNA polymerase with dNTPs at 11 °C, was ligated with the fragment including the Hyg gene constructed using KOD-PLUS-DNA polymerase (TOYOBO)-driving PCR on pcDNA3.1/Hygro plasmid with the primer pair 5′-TCA​CCG​GTC​ACC​ATG​AAA​AAG​CCT​GAA​CTC​ACC​GCG-3′ and 5′-TCA​AAG​GCA​GAA​GCA​ACT​TCT​ACA​CAG​CCA​TCG​GTC​C-3′ using the In-Fusion kit (TOYOBO) to obtain pSelect-zeo-HygTK (28–10), which contains the fusion gene of Hyg and HSV1tk genes. Finally, pSelect-Km-HSV1tk(1), cleaved by NotI and NheI to remove the HSV1tk gene, was ligated with the fragment including the HygTK gene constructed using KOD-PLUS-DNA polymerase (TOYOBO)-driving PCR on pSelect-zeo-HygTK(28-10) with the primer pair 5′-ATT​TAA​ATC​AGC​GGC​CGC​GGA​TCT​GCG​ATC​GCT​CCG-3′ (GT84) and 5′- TGT​CTG​GCC​AGC​TAG​CTC​AGG​TTT​AGT​TGG​CC-3′ using the In-Fusion kit (TOYOBO) to obtain pMB1KmHygTK (1).

pBS-HPRTEx2: This plasmid contains 5.4-kb of DNA covering the region containing intron 1, exon 2, intron 2, exon 3, and intron 3 of the HPRT locus. First, the 5′-2.8-kb fragment was constructed using KOD-PLUS-DNA polymerase (TOYOBO)-driving PCR on genomic DNA from the fibrosarcoma-derived cell line HT1080 with the primer pair 5′-AGC​CTG​GGC​AAC​ATA​GCG​AGA​CTT​C-3′ and 5′-TCT​GGT​CCC​TAC​AGA​GTC​CCA​CTA​TAC​C-3′, and the 3′-3.1-kb fragment was constructed using KOD-PLUS-DNA polymerase (TOYOBO)-driving PCR on genomic DNA from the fibrosarcoma-derived cell line HT1080 with the primer pair 5′-GCT​GGG​ATT​ACA​CGT​GTG​AAC​CAA​CC-3′ and 5′-TGG​CTG​CCC​AAT​CAC​CTA​CAG​GAT​TG-3’. There is a shared 0.5-kb sequence in the terminal end and the primed end of the 5′-2.8-kb fragment and the 3′-3.1-kb fragment, respectively. Finally, pBluescript SK+ (pBS) cleaved by NotI was ligated with the 5.4-kb fragment constructed using KOD-PLUS-DNA polymerase (TOYOBO)-driving PCR on both the 5′-2.8-kb fragment and the 3′-3.1-kb fragment with the primer pair 5′- TAG​TTC​TAG​AGC​GGC​CGC​AGC​CTG​GGC​AAC​ATA​GCG​AGA​CTT​C-3′ and 5′-CAC​CGC​GGT​GGC​GGC​CGC​TGG​CTG​CCC​AAT​CAC​CTA​CAG​GAT​TG-3′ using the In-Fusion kit (TOYOBO) to obtain pBS-HPRTEx2 (18-7). This plasmid was used as the template DNA for site-directed mutagenesis to introduce a designed sequence.

pBS-HPRTEx2ISCEI: This plasmid contains 5.4-kb DNA covering the region containing intron 1 and the ISCEI recognition sequence-included exon 2, intron 2, exon 3, and intron 3 of the HPRT locus. First, the 5′-2.5-kb fragment was constructed using KOD-PLUS-DNA polymerase (TOYOBO)-driving PCR on genomic DNA from the fibrosarcoma-derived cell line HT1080 with the primer pair 5′-TAG​TTC​TAG​AGC​GGC​CGC​AGC​CTG​GGC​AAC​ATA​GCG​AGA​CTT​C-3′ and the ISCEI recognition sequence (underlined)-tagged 5′-ATT​ACC​CTG​TTA​TCC​CTA​ACC​TGG​TTC​ATC​ATC​ACT​AAT​CTG-3′, and the 3′-2.9-kb fragment was constructed using KOD-PLUS-DNA polymerase (TOYOBO)-driving PCR on genomic DNA from the fibrosarcoma-derived cell line HT1080 with the primer pair ISCEI recognition sequence (underlined)-tagged 5′-TAG​GGA​TAA​CAG​GGT​AAT​TAT​GAC​CTT​GAT​TTA​TTT​TGC​ATA​CC-3′ and 5′-CAC​CGC​GGT​GGC​GGC​CGC​TGG​CTG​CCC​AAT​CAC​CTA​CAG​GAT​TG-3’. There is a shared ISCEI recognition sequence in the terminal end and the primed end of the 5′-2.8-kb and 3′-3.1-kb fragments, respectively. Finally, pBluescript SK + cleaved by NotI was ligated with the 5′-2.5-kb fragment and the 3′-2.9-kb fragment using the In-Fusion kit (TOYOBO) to obtain pBS-HPRTEx2ISCEI(21-1). This plasmid was used as the original plasmid to replace the fragment containing a designed sequence made by site-directed mutagenesis.

The sub-plasmid pBS-HPRTEx2Syn: This plasmid contains 5.4-kb DNA covering the region containing intron 1 and the synonymous sequence-included exon 2, intron 2, exon 3, and intron 3 of the HPRT locus. First, site-directed mutagenesis was performed using the KOD-PLUS-Inverse PCR mutagenesis kit (TOYOBO) with inverse primer pairs containing the synonymous sequence (underlined), 5′-GGC​TAC​GAT​CTC​GAC​CTC​TTT​TGC​ATA​CCT​AAT​CAT​TAT​GC-3′ and 5′-TGG​TTC​ATC​ATC​ACT​AAT​CTG-3′, and the template DNA pBS-HPRTEx2(18-7) plasmid to obtain pBS-HPRTEx2Syn(Inv15). Finally, pBS-HPRTEx2ISCEI(21-1), cleaved by BglII and SphI, was ligated with the synonymous sequence-included fragment constructed by digesting pBS-HPRTEx2Syn(Inv15) with BglII and SphI, followed by gel extraction to obtain pBS-HPRTEx2Syn (Inv15-2).

The sub-plasmid pBS-HPRT(Ex2Syn(1195)ISCEI): This plasmid contains 5.4-kb DNA covering the region containing intron 1, the synonymous sequence-incorporated exon 2, and the ISCEI recognition sequence-incorporated intron 2, exon 3, and intron 3 of the HPRT locus. First, site-directed mutagenesis was performed using the KOD-PLUS-Inverse PCR mutagenesis kit (TOYOBO) with inverse primer pairs ISCEI recognition-sequence (underlined)-tagged 5′-TAG​GGA​TAA​CAG​GGT​AAT​ATT​TTG​TAG​AAA​CAG​GGT​TCG​C-3′ and 5′-AAA​AAT​ATT​AGC​TGG​GAG​TGG-3′ and the template DNA pBS-HPRTEx2Syn (Inv15-2) plasmid to obtain pBS-HPRT(Ex2Syn(1195)ISCEI) (1).

The designed donor pMB1KmHygTK-HPRT(Ex2Syn(1195)ISCEI): The donor vector pMB1KmHygTK, cleaved by NotI, was ligated with the fragment with the synonymous sequence and the ISCEI recognition sequence, constructed by digesting pBS-HPRT(Ex2Syn(1195)ISCEI) (1) with NotI to obtain pMB1KmHygTK-HPRT(Ex2Syn(1195)ISCEI) (5). It was verified whether the direction of the HPRT sequence is identical to that of the HygTK gene.

The sub-plasmid pBS-HPRT(ISCEI(1195)Ex2Syn): This plasmid contains 5.4-kb DNA covering the region containing the ISCEI recognition sequence-incorporated intron 1 and the synonymous sequence-incorporated exon 2, intron 2, exon 3, and intron 3 of the HPRT locus. First, site-directed mutagenesis was performed using the KOD-PLUS-Inverse PCR mutagenesis kit (TOYOBO) with inverse primer pairs containing ISCEI recognition-sequence (underlined)-tagged 5′-TAG​GGA​TAA​CAG​GGT​AAT​CAA​AGC​ACT​GGG​ATT​ACA​AGT​G-3′ and 5′-GGA​GGC​TGA​GAC​AGG​AGA​GTT​GC-3′ and the template DNA pBS-HPRTEx2Syn (Inv15-2) plasmid to obtain pBS-HPRT(ISCEI(1195)Ex2Syn) (3).

The designed donor plasmid pMB1KmHygTK-HPRT(ISCEI(1195)Ex2Syn): The donor vector pMB1KmHygTK, cleaved by NotI, was ligated with the fragment with the synonymous sequence and the ISCEI recognition sequence, constructed by digesting pBS-HPRT(ISCEI(1195)Ex2Syn) (3) with NotI to obtain pMB1KmHygTK-HPRT(ISCEI(1195)Ex2Syn) (1). It is verified whether the direction of the HPRT sequence was identical to that of the HygTK fusion gene.

The sub-plasmid pBS-HPRT(ISCEI(545)Ex2Syn): This plasmid contains 5.4-kb DNA covering the region containing the ISCEI recognition sequence-incorporated intron 1 and the synonymous sequence-incorporated exon 2, intron 2, exon 3, and intron 3 of the HPRT locus. First, site-directed mutagenesis was performed using the KOD-PLUS-Inverse PCR mutagenesis kit (TOYOBO) with inverse primer pairs containing the ISCEI recognition-sequence (underlined)-tagged 5′-TAG​GGA​TAA​CAG​GGT​AAT​CAA​AGC​ACT​GGG​ATT​ACA​AGT​G-3′ and 5′-GGA​GGC​CGA​GGC​GGG​TGG​ATC​A-3′ and the template DNA pBS-HPRTEx2Syn (Inv15-2) plasmid to obtain pBS-HPRT(ISCEI(545)Ex2Syn) (4).

The designed donor plasmid pMB1KmHygTK-HPRT(Ex2Syn(545)ISCEI): The donor vector pMB1KmHygTK, cleaved by NotI, was ligated with the fragment containing the synonymous sequence and the ISCEI recognition sequence, constructed by digesting pBS-HPRT(ISCEI(545)Ex2Syn) (4) with NotI to obtain pMB1KmHygTK-HPRT(ISCEI(545)Ex2Syn) (1). It was verified whether the direction of the HPRT sequence is identical to that of the HygTK fusion gene.

The sub-plasmid pBS-HPRT(ISCEI (318)Ex2Syn): This plasmid contains 5.4-kb DNA covering the region containing the ISCEI recognition sequence-incorporated intron 1 and the synonymous sequence-incorporated exon 2, intron 2, exon 3, and intron 3 of the HPRT locus. First, site-directed mutagenesis was performed using the KOD-PLUS-Inverse PCR mutagenesis kit (TOYOBO) with inverse primer pairs containing the ISCEI recognition-sequence (underlined)-tagged 5′-TAG​GGA​TAA​CAG​GGT​AAT​TGT​ATT​TTT​AGT​AGA​GAC​GGG-3′ and 5′-AAA​AAA​TTA​GCC​GGG​TGT​GG-3′ and the template DNA pBS-HPRTEx2Syn (Inv15-2) plasmid to obtain pBS-HPRT(ISCEI (318)Ex2Syn) (2).

The designed donor plasmid pMB1KmHygTK-HPRT(Ex2Syn (318)ISCEI): The donor vector pMB1KmHygTK, cleaved by NotI, was ligated with the fragment containing the synonymous sequence and the ISCEI recognition sequence, constructed by digesting pBS-HPRTISCEI (318)Ex2Syn (2) with NotI to obtain pMB1KmHygTK-HPRT(ISCEI(318)Ex2Syn) (1). It was verified whether the direction of the HPRT sequence is identical to that of the HygTK fusion gene.

pCI-neo-nls-I-SceI: First, the I-SceI meganuclease gene was cloned from Saccharomyces cerevisiae W303 strain using KOD-PLUS-DNA polymerase (TOYOBO)-driving PCR on its genomic DNA with the primer pair nuclear localization signal sequence (underlined)-included 5′-GGA​TCC​TGC​AAA​GAT​GGA​TAA​AGC​GGA​ATT​AAT​TCC​CGA​GCC​TCC​AAA​AAA​GAA​GAG​AAA​GGT​CGA​ATT​GGG​TAC​CAT​GAA​AAA​TAT​TAA​AAA​AAA​TCA​AGT​AAT​GAA​TCT​GGG​TCC-3′ and I-SceI gene’s termination codon (underlined)-included 5′-ATG​CAT​TTA​TTT​TAA​AAA​AGT​TTC​GGA​TGA​AAT​AGT​ATT​AGG​C-3′. Next, the pCI-neo plasmid (Promega), cleaved by NotI, was ligated with the fragment including the nls-I-SceI gene constructed using the KOD-PLUS-DNA polymerase (TOYOBO)-driving PCR on its DNA with the primer 5′-GTC​GAC​CCG​GGC​GGC​CGC​CAT​GGA​TAA​AGC​GGA​ATT​AAT​TCC​CG-3′ and I-SceI gene’s termination codon (underlined)-included 5′-CTA​AAG​GGA​AGC​GGC​CGC​TTA​TTT​TAA​AAA​AGT​TTC​GG-3′ using the In-Fusion kit (TOYOBO) to obtain pCI-neo-nls-I-SceI (29-2) (DDBJ accession number to nls-I-SceI: LC920516↗).

nls-I-SceI-expression Sendai virus (SeV) vector: The RNA-dependent RNA polymerase of the SeV vector tends to make errors in the incorporation of ribonucleotide at A-rich sequences. To avoid this, primers for PCR were designed to target an A-rich site with a synonymous conversion from A/T to G/C. To construct the SeV-compatible I-SceI gene, the first PCR product was constructed using KOD-PLUS-Ver. 2-DNA polymerase (TOYOBO)-driving PCR on pCI-neo-nls-I-SceI (29-2) DNA with the primer pair 5′-CGA​GCC​TCC​AAA​gAA​GAA​GAG​AAA​GGT​CGA​ATT​GGG​TAC​CAT​GAA​AAA​TAT​TAA​gAA​gAA​TCA​AGT​AAT-3′ (NLS-I-SceIN_A36G_A78G_A81G_N) and 5′-GTT​CGG​GAT​GGT​TTT​cTT​GTT​GTT​AAC​G-3′ (NLS-I-SceI_A426G_C), and the second PCR product was constructed using KOD-PLUS-Ver. 2-DNA polymerase (TOYOBO)-driving PCR on pCI-neo-nls-I-SceI (29-2) DNA with the primer pair 5′-CGT​TAA​CAA​CAA​gAA​AAC​CAT​CCC​GAA​C-3′ (NLS-I-SceIN_A426G_N) and 5′-ATA​TGC​GGC​CGC​GAT​GAA​CTT​TCA​CCC​TAA​GTT​TTT​CTT​ACT​ACG​GTT​ATT​TTA​AAA​AAG​TTT​CGG​ATG-3′ (NLS-I-SceI_EIS_Not1_C). On these PCR products, the last PCR product was constructed using the same enzyme-driving PCR with the primer pairs NotI site-included 5′-ATA​TGC​GGC​CGC​GAC​GCC​ACC​ATG​GAT​AAA​GCG​GAA​TTA​ATT​CCCGA​GCC​TCC​AAA​GAA​GAA​GAG​AAA​GGT​CG-3′ (Not1_NLS-I-SceIN_A36G_N) and the same PCR as follows (NLS-I-SceI_EIS_Not1_C) to obtain the full length of nls-I-SceI DNA. Finally, a SeV vector plasmid DNA pSeV18+/TS15ΔF (Ban et al., 2011) cleaved by NotI, followed by BAP treatment, was ligated with the full DNA of nls-I-SceI cleaved by NotI to obtain pSeV18+nls-I-SceI/TS15ΔF. The SeV vector SeV18+nls-I-SceI/TS15ΔF was produced by reconstituting with this plasmid DNA as a template for RNA polymerization, followed by amplification.

Preparation of exponentially growing cells: A cell stock (1 × 10⁶ cells) of HT1080 cells was seeded in a T75 flask with DMEM-based culture medium. Two or three days after seeding, the second culture was succeeded by the passage procedure as follows: the HT1080 cells were trypsinized, the dissociation reaction was terminated by adding a 10-fold volume of DMEM-based culture medium, followed by pipetting and centrifugation, and the cells were resuspended with a fresh medium. For the second culture, the recovered cells were plated at a cell density of 0.1–1.0 × 10⁴ cells/cm² in T75 or T225 flasks with a fresh medium. When cells reached a cell density of 1.0–10 × 10⁴ cells/cm² (2–3 days after plating), the exponentially growing cells were used for transfection by electroporation.

Electroporation: The exponentially growing HT1080 cells were trypsinized, the dissociation reaction was terminated by adding a 10-fold volume of DMEM-based culture medium, followed by pipetting and centrifugation, and the cells were resuspended in Opti-MEM for washing and finally resuspended in Opti-MEM at a concentration of 0.5–1 × 10⁷ cells/mL or 2 × 10⁷ cells/mL. An aliquot of 0.8 mL of the cell suspension was mixed with 10 μg of a designed circular donor plasmid DNA, which was purified with an adsorption column (QIAGEN) and dissolved in endotoxin-free TE, and 12.6 μg of an nls-I-SceI expression plasmid (pCI-neo-nls-I-SceI (29–2)) DNA or 13.3 μg of another nls-I-SceI expression plasmid (pCAG.I-SceI) DNA (Rouet et al., 1994a; Holkers et al., 2012), which was purified with an adsorption column (QIAGEN) and dissolved in endotoxin-free TE, or 1, 2, or 5 μg of the messenger RNA expressing nls-I-SceI enzyme, and was transferred into an electroporation cuvette (4 mm in width), followed by cooling on ice for 5 min. Otherwise, 10 μg of a linearized donor DNA alone was transferred into an electroporation cuvette (4 mm in width), followed by cooling on ice for 5 min. Each cuvette was subjected to three pulses of 140 volts/70 ms with 200 ms intervals using Electro Square Porator BTX T820 (Artisan Technology Group), followed by cooling on ice for 5 min.

Plasmid-type nls-I-SceI gene expression vector: An intra-cellular cleaver expression plasmid, pCI-neo-nls-I-SceI (29-2), constructed in this study, or another nls-I-SceI expression plasmid, pCAG.I-SceI (Rouet et al., 1994a; Holkers et al., 2012) (bottom of lane v in Figure 2A; lane ii in Figure 3C), was used to cleave a designed circular donor DNA after transfection.

Messenger RNA expressing nls-I-SceI enzyme: This was prepared by T7 RNA polymerase-mediated in vitro transcription using the pCI-neo-nls-I-SceI (29-2) plasmid DNA as a template.

RNA virus-type nls-I-SceI enzyme expression vector: SeV18+nls-I-SceI/TS15ΔF was used to cleave a designed circular donor DNA in host cells. To prepare the freezing stocks of the HT1080 cells infected with the SeV vector, the second culture of HT1080 cells was spread onto T75 flasks at 4 × 10⁶ cells/T75 flask and incubated for approximately 24 h at 37 °C; then, the SeV vector was adsorbed with Opti-MEM onto the adherent cells at a multiplicity of infection (moi) of 3 at 32 °C [permissive temperature for replication of the TS15-type SeV vector (Ban et al., 2011)] by shaking the T75 flasks forth/back and left/right every 15 min for 2 hours. After viral adsorption, the SeV/Opti-MEM solution was removed, fresh Opti-MEM buffer was added to wash them, and then a fresh medium was added, followed by incubation at 32 °C/5% CO₂ for 24 h to amplify the SeV vector genome and allow nls-I-SceI enzyme expression. HT1080/SeV cells were trypsinized, washed, and resuspended in Opti-MEM to allow recovery and finally resuspended in CELLBANKER 1 solution at a concentration of 2 × 10⁶ cells/mL; several tubes (2 × 10⁶ cells/tube) were then stored at −80 °C.

For transfection of a circular donor DNA, the stocks of the HT1080/SeV cells were plated and incubated at 37 °C (non-permissive temperature for replication of the TS15-type SeV vector) during the next different period (days): (1) plating onto four T225 flasks at 0.5 × 10⁶/T225, followed by incubation for approximately 96 h at 37 °C; (2) plating onto two T225 flasks at 1.0 × 10⁶/T225, followed by incubation for approximately 72 h at 37 °C; (3) plating onto two T225 flasks at 2.0 × 10⁶/T225, followed by incubation for approximately 48 h at 37 °C. Each of them reached 0.5–1.7 × 10⁴ cells/cm² on T225 in cell density, and 4–13 × 10⁶ cells were recovered. These recovered cells were used to transfect a circular donor DNA by electroporation.

ISCEI site-linearized donor DNA: Each of the three types of ISCEI recognition site-containing designed donor plasmid DNA was digested with 1 unit of the I-SceI enzyme (NEB) per μg of DNA for 2 h at 37 °C and purified using an adsorption column (QIAGEN), followed by dissolution in endotoxin-free TE. The quality of ISCEI-linearized DNA was confirmed by agarose gel electrophoresis to verify linearization, and the quantity was confirmed compared with that of the circular donor DNA.

Selection of donor-targeted duplication clones: After transfection by electroporation with a circular donor plasmid DNA and an nls-I-SceI expression plasmid vector DNA, the cells were dispersed into 165 mL of pre-warmed DMEM and then plated at 5–10 × 10⁶ cells per 9-cm dish onto 16 dishes. Three days after electroporation, the medium was replaced with fresh medium containing hygromycin to select targeted duplication clones or random integration clones, which was repeated every 2 or 3 days. At 14–17 days after electroporation, hygromycin-resistant (Hyg^R) colonies of two or four dishes out of 16 were counted to calculate the total number of Hyg^R colonies obtained, and then the medium was replaced with fresh medium containing hygromycin and 6-thioguanine to select targeted duplication clones showing an HPRT-deficient phenotype, which was repeated every 2 or 3 days. Twenty-four days after electroporation, hygromycin- and 6-thioguanine-resistant colonies from all dishes were counted. They were collected as HTG clones and allowed to grow in fresh medium containing the two drugs, followed by the preparation of cell-banker stocks and cell pellets for genomic DNA analysis, which were stored at −80 °C.

Analysis for genomic structure of donor-targeted duplication: Genomic DNA was extracted from each of the cell pellets of HTG clones, and then two PCR tests (duplication test; Figure 1D) were performed using the KOD-FX-DNA polymerase (TOYOBO) with one primer pair, plasmid-5′-inside forward primer 5′-TTG​CAA​GCA​GCA​GAT​TAC​GC-3′ (GT112), target-3′-outside reverse primer 5′-GCC​ACT​GCA​CCC​AGC​CGT​ATG​T-3′ (GT69), and the other primer pair, target-5′-outside forward primer 5’-ACG​ACC​TCG​CCC​GGC​CTT​GTA​TT-3′ (GT68), plasmid-3′-inside reverse primer 5′-AGC​TTC​AGC​TGT​GTT​CTG​GC-3′ (GT124), on the genomic DNA to verify the structure of targeted duplication into exon 2 at the HPRT locus, which consists of duplication of the donor DNA with the designed sequence and the recipient DNA with the original sequence, separated by the plasmid backbone containing the positive and negative selection marker pair. These primer pairs do not work in unexpected HTG clones, which are neither double-positive nor single-positive in the duplication test. For HTG clones with donor-targeted duplication, sequence analysis was performed by PCR using the KOD-FX-DNA polymerase with the primer pair, Ex2-5′-outside forward primer 5′-GCT​GGG​ATT​ACA​CGT​GTG​AAC​CAA​CC-3′ (GT19), Ex2-3′-outside reverse primer 5′-TCT​GGT​CCC​TAC​AGA​GTC​CCA​CTA​TAC​C-3′ (GT22), followed by a sequencing reaction on the PCR products (allele sequence test; Figure 1D). Duplication of the donor DNA with the designed sequence and the recipient DNA with the original sequence shows an overlapping sequencing chart (doublet chart) of the designed sequence and the original sequence. If an HTG clone shows a singlet chart of the original sequence, it indicates that the designed sequence has been excluded and, therefore, is not used to isolate designed donor replacement clones.

Dish screening: After transfection with a circular donor plasmid DNA and an nls-I-SceI gene expression plasmid vector, the cells were dispersed into 165 mL of pre-warmed DMEM and then plated at 5∼10 × 10⁶ cells per 9-cm dish onto 16 dishes. Three days after electroporation, the medium was replaced with the fresh medium containing 100 μg/mL hygromycin to select targeted duplication clones or random integration clones, which was repeated every 2 or 3 days. At 14–17 days after electroporation, hygromycin-resistant (Hyg^R) colonies of two dishes out of 16 were counted to calculate the total number of Hyg^R colonies obtained. The mixed cells from each dish were collected, followed by the preparation of cell-banker stocks and cell pellets for genomic DNA analysis, which were stored at −80 °C. One PCR test (TD screening; Figure 5B) was performed using the KOD-FX-DNA polymerase with a primer pair, plasmid-5′-inside forward primer 5′-TTG​CAA​GCA​GCA​GAT​TAC​GC-3′ (GT112) and target-3′-outside reverse primer 2 5′-TGG​GGA​AGG​AGG​CTA​TAG​AAG​ACC-3′ (GT330) on the genomic DNA to identify TD-containing dishes. This primer pair does not work in random integration clones. The other PCR test (Syn-TD screening; Figure 5B) was performed using the KOD-FX-DNA polymerase with a primer pair, target-5′-outside forward primer 2 5′- GCA​CAA​ACC​TCA​AAT​TTC​TCA​GCA​CTG​G-3′ (GT336) and Syn sequence (underlined)-included reverse primer 5′- GCA​AAA​GAG​GTC​GAG​ATC​GTA​GCC-3′ (GT340) on the genomic DNA to determine which of the TD dishes retain the Syn allele.

Well screening: The identified Syn-TD dish stock was plated and cultured for a few days. The pre-cultured Syn-TD dish cells were resuspended in DMEM-based culture medium and seeded at 2 cells/well onto three 96-well plates containing an HT1080 feeder or plated at 2 cells/well onto three fresh 96-well plates without a feeder. The HT1080 feeder plates were prepared before plating Syn-retaining TD dish-derived cells to improve the low clonal capacity of HT1080-derivative cells as follows: HT1080 pre-cultured cells were plated onto three 96-well plates at 64 cells/well in DMEM-based culture medium, and after 1-day incubation, they were used as HT1080-feeder plates. Six days after plating Syn-TD dish-derived cells onto 96-well plates, the medium was exchanged for DMEM-based culture medium with 100 μg/mL hygromycin; the resultant growing wells under hygromycin selection were transferred to 24-well plates and propagated, a small portion of which was used for genomic DNA extraction and stored at −80 as several cell stocks. PCR analysis was performed using the primer pairs for TD screening and Syn-TD screening to identify TD-containing wells and determine which of the TD wells retain the Syn allele, respectively. Each of the identified Syn-TD well stocks was plated onto a well in 6-well plates and cultured for a few days, followed by the preparation of cell stocks and cell pellet samples for genomic DNA analysis, which were stored at −80 °C.

Selection of naturally generated donor-replacement clones: HTG clones with donor-targeted duplication containing a designed sequence were re-grown in the presence of hygromycin and 6-thioguanine from cell-banker stocks, pre-cultured without any drug for a few days, and then plated at 1 × 10⁴ cells per 15-cm diameter dish onto six dishes with DMEM-based culture medium or at 500 cells per 9-cm diameter dish onto 8 or 16 dishes with advanced DMEM-based culture medium. On day 5 or 6, the medium was replaced with fresh medium containing 1 μM ganciclovir with DMEM-based culture medium or 0.1 μM ganciclovir with advanced DMEM-based culture medium to select donor replacement clones three times per week by at most day 35. Ganciclovir-resistant (GCV^R) colonies were collected as GCV clones and allowed to grow in fresh medium with the drug, followed by the preparation of cell-banker stocks and cell pellet samples for genomic DNA analysis, which were stored at −80 °C. In the case of HAT selection, 1× HAT supplement was added instead of GCV.

Analysis for genomic structure of designed donor replacement: Genomic DNA was extracted from each of the cell pellets of GCV clones, and then, a PCR test (Hyg test; Figure 1E) was performed with the primer pair 5′-Hyg forward primer 5′-TCA​CCG​GTC​ACC​ATG​AAA​AAG​CCT​GAA​CTC​ACC​GCG-3′ (GT38) and 3′-Hyg reverse primer 5′-TCA​AAG​GCA​GAA​GCA​ACT​TCT​ACA​CAG​CCA​TCG​GTC​C-3′ (GT39) on the Hyg structural gene and the genomic DNA to verify whether the plasmid backbone with the Hyg^R marker gene is removed from the target duplication structure. For the GCV clones, another PCR test (pop-out test; Figure 1E) was performed with the primer pair, target-5′-outside forward primer 2 5′-GCA​CAA​ACC​TCA​AAT​TTC​TCA​GCA​CTG​G-3′ (GT336) and target-3′-outside reverse primer 2 5′-TGG​GGA​AGG​AGG​CTA​TAG​AAG​ACC-3′ (GT330), targeting the sites outside the counter region of the 5.4-kb donor DNA and the genomic DNA to verify whether the plasmid backbone, with either single unit of the doublet form, was popped out from the target duplication structure. For the GCV clones, sequence analysis was performed by PCR with the primer pair, Ex2-5′-outside forward primer 5′-GCT​GGG​ATT​ACA​CGT​GTG​AAC​CAA​CC-3′ (GT19) and Ex2-3′-outside reverse primer 5′-TCT​GGT​CCC​TAC​AGA​GTC​CCA​CTA​TAC​C-3′ (GT22), across exon 2 of the HPRT gene, followed by a sequencing reaction on the PCR products (allele sequence test; Figure 1E), to determine whether it is the designed sequence or the target sequence. The duplication test was also performed to determine whether any duplicative structure remained within each of the GCV clones. The TK-deficient TD clones found as the GCV-resistant clones that are still doubly positive in the duplication test were analyzed by the PCR with the primer pair, 5′-ATT​TAA​ATC​AGC​GGC​CGC​GGA​TCT​GCG​ATC​GCT​CCG-3’ (GT84) and 5′-GCC​ACC​GAA​TTC​CAT​ACC​AC-3′, amplifying the region including the overall codes of the HygTK fusion gene, followed by sequencing, to identify TK mutations. When a GCV^R or HAT^R cell stock showed that the pop-out test was positive, but the Hyg test was weakly positive, or the allele sequence test was the doublet type of the Syn sequence and the SEXA1 sequence, cell cloning was performed by plating at 1 or 2 cells/well onto 96-well plates, followed by propagation. Approximately 10–20 or fewer clones were analyzed by PCR and sequencing and classified into Syn-, SEXA1-, and TK-deficient TD clones.

The new concept of this precision strategy for human genome correction is to introduce a DSB within a circular donor plasmid to prime homologous recombination between the donor DNA and the target locus, whereas targeted genomic breakage, such as ZFN, TALEN, or CRISPR–Cas9, introduces a DSB at the target locus. This is the best solution to circumvent on-target mutagenesis due to DSB-induced non-homologous end-joining at the target locus, which is an issue of targeted genomic breakage-mediated genome editing. As the cleaver of the circular donor plasmid, we selected the I-SceI endonuclease (Colleaux et al., 1988), which recognizes a specific 18-bp sequence. This sequence is not in the databases of the human genome. The high specificity of this enzyme for the recognition sequence has been demonstrated as follows: human cells transfected with I-SceI showed background levels of DSB signals (Lin et al., 2015); Mammalian cells showed no signals of I-SceI-induced DSB in the genome (Soutoglou et al., 2007); cellular toxicity with I-SceI expression was not observed in mammalian cells (Rouet et al., 1994a). Thus, this enzyme is a suitable tool to circumvent off-target mutagenesis, which is another issue associated with targeted genomic breakage-mediated genome editing.

To construct a model system to evaluate the frequency and accuracy of such InCDC-mediated genome editing, we used the HT1080 cell line derived from a male individual as human cells and constructed the model system comprising a target site for human genome correction, the region around exons 2 and 3 at the HPRT locus (middle in Figure 1A) in the X chromosome (upper in Figure 1A), and a designed donor DNA, which covers the 5.4-kb section across introns 1–3 (lower in Figure 1A).

The designed sequence consists of six continuous synonymous codons (Syn allele sequence: upper in Figure 1B), in which the six original codons (SEXA1 allele sequence: lower in Figure 1B) are replaced. The unique double-strand break site ISCEI, which is the I-SceI meganuclease recognition sequence (right in Figure 1B), is positioned at 318 bp, 545 bp, or 1195 bp from the designed sequence Syn in a designed donor DNA (Figure 2A). When nls-I-SceI enzymes, to which a nuclear localization signal was added (Materials and Methods), are expressed in a transfected cell, a circular donor plasmid is cleaved at its ISCEI site (right in Figure 1B).

It is well known that targeted duplication inserts the whole donor plasmid DNA into the target site through crossover-type homologous recombination (Hasty et al., 1991; Brenneman et al., 1996). As homology-directed DNA repair is known to occur in the late S and G2 phases of the cell cycle (Smirnikhina et al., 2022), the exponentially growing HT1080 cells were used for transfection with donor plasmid DNA (Materials and Methods). The cleavage of circular donor plasmid mediates the integration into the target region around exons 2–3 at the HPRT locus (blue-lined chromosome in TD1; Figure 1C) and forms the configuration of a duplication of the donor DNA with the designed sequence Syn and the target DNA with the original sequence SEXA1, intervened by the plasmid vector bearing the positive and negative selection fusion gene (HygTK) (targeted duplication product: TD2). The targeted duplication clones were isolated from the positively selected clones [hygromycin-resistant (Hyg^R) colonies]. HPRT-defective clones exhibit resistance to a guanine analog 6-thioguanine. Taking advantage of this HPRT^– phenotype, hygromycin/6-thioguanine-double-resistant (Hyg^RTg^R) colonies, as candidates for HPRT exon 2-targeted duplication clones, were analyzed by PCR with a certain primer pair (Figure 1D). Finally, out of targeted duplication clones (TD2), natural replacement clones can be generated as the negatively selected clones (natural replacement product: TD3-1 or -2 in Figure 1C), which are verified by PCR and sequencing for the presence of the singlet allele Syn out of the negatively selected clones (GCV^R colonies). The natural replacement reaction is associated with popping-out of the circular plasmid DNA containing the HygTK gene and either the designed or the original allele. Thus, the natural replacement clones (TD3-1 or -2) do not contain the HygTK gene.

Several reports on random integration of foreign DNA indicate the involvement of microhomology-mediated conjugation (Yan et al., 2010; Yan et al., 2013). Circular or cleaved donor DNA integrates at sites on chromosomes (brown-lined chromosome: RI1 in Figure 1C), which is mostly generated through microhomology-mediated conjugation. A positively selected clone (Hyg^R) has the resulting typical random integration structure (random integration product: RI2), comprising the plasmid DNA bearing the selection marker gene (HygTK) and the Syn-designed donor sequence but was never isolated by PCR using a certain primer pair for the targeted duplication structure (Figure 1D).

To verify a targeted duplication structure (TD2 in Figure 1C) of Hyg^RTg^R clones, the duplication test (Figure 1D) is performed by PCR with one primer pair, which is positioned at the 5′-outside genomic region of donor homologous DNA (a forward primer) and the 3′-inside plasmid region of donor homologous DNA (a reverse primer), and with the other primer pair, which is positioned at the 5′-inside plasmid region of donor homologous DNA (a forward primer) and the 3′-outside genomic region of donor homologous DNA (a reverse primer). Next, to verify whether a duplicative structure is hetero-allelic (Syn and SEXA1) (TD2), the allele sequence test (Figure 1D) is performed by PCR and sequencing. A typical result of this test shows a doublet chart, including double-peaks of the six synonymous and original codons, which are identical in height (lower in Figure 1D).

To verify a natural replacement structure (TD3-1 or -2) of GCV^R colonies, the pop-out test (Figure 1E) is performed by PCR using the PCR primer pair, which is positioned at the 5′-outside genomic region of donor homologous DNA (a forward primer) and the 3′-outside genomic region of donor homologous DNA (a reverse primer), followed by the allele sequence test to verify whether a natural replacement clone is Syn (TD3-1) or SEXA1 (TD3-2) by PCR and sequencing. Concurrently, we verified the absence of plasmid DNA generated by a popping-out reaction during a natural replacement event, as indicated by the absence of PCR products using a primer pair targeting the Hyg structural gene sequence. If 1.0-kb PCR products on the Hyg gene are detected and 6.3-kb PCR products across the target region are not detected in a GCV^R colony, it is supposed to be TK-deficient derivatives from the targeted duplication clone (TD2).

Conventional ends-in gene targeting methods using linearized donor plasmid (Hasty et al., 1991) were reported as the procedure leading to the replacement of the target sequence with a desired sequence in mES cells (Wu et al., 1994). However, the report suggests caution that post-replication non-targeted integration (apparent targeted integration) events occur in mES cells (Mangerich et al., 2009). Regarding it as a critical issue, we examined whether in human cells, ends-in gene targeting-based methods produce such apparent targeted integration using the assay system (Figure 1).

When a circular donor plasmid pMB1KmHygTK-HPRT(ISCEI(545)Ex2Syn) was transfected with nls-I-SceI gene expression plasmid, the TD clones appear to be obtained at a frequency of 3.6 × 10⁻³ (23/6348 Hyg^R) per overall-integration clone (lane iii in Figure 2A), based on the criteria to be doubly positive in the duplication test (Figure 1D) (upper of Figure 2B).

These candidates for TD clones were examined using the random integration (RI) test (lower in Figure 2B), which is a PCR assay with target-5′-outside forward primer 2 (closed blue arrow) and target-3′-outside reverse primer 2 (open blue arrow), being the same primer pair as in the pop-out test (Figure 1E). If a TD candidate clone is negative in the RI test, it is considered a targeted duplication clone. All of the 23 TD candidate clones were shown to be negative in the PCR assay (Figure 2B). These results indicate that all of these TD structures (doubly positive in the duplication test) are at the target site. The abortion-type (singly positive in the duplication test) clone (HTG-A9 in Figure 2B) was positive in the RI test. This clone appears to be the outcome of a post-replication non-targeted integration, as is the case of mES cells (Mangerich et al., 2009). We conclude that, in HT1080 cells, post-replication non-targeted integration events are rare, and their frequency is far lower than that of TD events.

These 23 TD clones were examined using the allele sequence test (Figure 1D) to identify the TD clones retaining the designed sequence Syn. Out of 23, 7 clones were associated with the hetero-duplication, which consisted of the designed sequence Syn and the target sequence SEXA1, and one clone was associated with the homo-duplication, which consisted of only the designed sequence Syn (Figure 2C). The Syn retention frequency per TD clone was 35% (8/23) for the circular donor cleavage at the ISCEI site, located 545 bp away from the Syn allele sequence (lane iii in Figure 2A).

It is later discussed how each of the four types of allelic duplications [15 clones out of 23 TD clones: 5′SEXA1/SEXA13’ (Figure 7C); 5 clones: 5′Syn/SEXA13’ (Figure 7D); 2 clones: 5′SEXA1/Syn3’ (Figure 7E); 1 clone: 5′Syn/Syn3’ (Figure 7F)] is generated in InCDC-mediated crossover-type HR. These four types were determined using 5'/3′Syn primer PCR analysis (Figures 7A,B).

When a circular donor plasmid pMB1KmHygTK-HPRT(ISCEI(318)Ex2Syn) was transfected without any expression tool of the nls-I-SceI enzyme, the targeted duplication clones were obtained at a frequency of 2.0 × 10⁻⁴ (1/5068 Hyg^R) per overall-integration clone (lane i in Figure 2A). When the same circular donor plasmid pMB1KmHygTK-HPRT(ISCEI(318)Ex2Syn) was transfected with the plasmid expressing nls-I-SceI enzyme, the targeted duplication clones were obtained at a frequency of 2.4 × 10⁻³ (39/16184 Hyg^R) per overall-integration clone, which significantly increased (12-fold) compared with that without expression of the nls-I-SceI gene (*: lane i vs. lane ii in Figure 2A). These results indicate that circular donor cleavage at the ISCEI site is crucial for priming targeted duplication events.

Out of 39 targeted duplication clones generated using the circular donor cleavage of pMB1KmHygTK-HPRT(ISCEI (318)Ex2Syn), 8 clones were associated with hetero-duplication, which consisted of the donor DNA with the designed sequence Syn and the target DNA with the original sequence SEXA1 (5′Syn/SEXA13′); however, the other 31 clones were associated with the homo-duplication (5′SEXA1/SEXA13′) (Figure 2A). The Syn retention frequency per TD clone was 21% (8/39) for the case of the circular donor cleavage at the ISCEI site, located 318 bp away from Syn (lane ii in Figure 2A).

When the circular donor plasmid pMB1KmHygTK-HPRT(Ex2Syn(1195)ISCEI) was transfected with the plasmid expressing nls-I-SceI enzyme, the targeted duplication clones were obtained at a frequency of 1.2 × 10⁻³ (13/10556 Hyg^R) per overall-integration clone (lane v in Figure 2A). Furthermore, 9 clones out of 13 clones were associated with the hetero-duplication, which consisted of the donor DNA with the designed sequence Syn and the original sequence SEXA1. The Syn retention frequency per TD clone was 69% (9/13) for the circular donor cleavage at the ISCEI site, located 1,195 bp away from Syn (lane v in Figure 2A).

Based on Syn-retention frequencies per TD clone (lanes ii, iii, and v in Figure 2A), Syn deletion frequencies per TD clone were 79% [(39-8)/39], 65% [(23–8)/23], and 31% [(13-9)/13] in each of the circular donor-plasmids with HPRT[ISCEI (318)Ex2Syn], HPRT[ISCEI (545)Ex2Syn], and HPRT[Ex2Syn (1195)ISCEI], respectively. The Syn-retention frequency per TD clone in the circular donor-plasmid with HPRT[Ex2Syn (1195)ISCEI] is significantly higher than that with HPRT[ISCEI (318)Ex2Syn] or HPRT[ISCEI (545)Ex2Syn] (^※: lane ii vs. lane v; ^#: lane iii vs. lane v in Figure 2A). These results imply that the circular donor plasmid is indeed cleaved at the ISCEI site to prime targeted duplication events, and a putative exonuclease attack occurs immediately after intra-cellular cleavage against the circular donor plasmid to cause the removal of the designed sequence. Thus, the concept of “Safety Distance”, in which the designed sequence is placed several hundred base pairs away from the ISCEI site in a circular donor plasmid to protect it from a putative exonuclease attack, is crucial to obtain designed sequence-retaining TD clones.

The donor plasmid linearized with the commercial I-SceI enzyme was electroporated, but this technique is less successful in obtaining the targeted duplication clones (lanes iv and vi in Figure 2A) compared with the InCDC technique (lanes iii and v in Figure 2A). We approach why linearized donor DNA is less successful in targeted duplication by quantifying the overall integration frequency per transfected viable cell. In this approach, an advanced DMEM-based culture medium was used to obtain a more corrected value of transfected viable cells as the clonal capacity of the advanced DMEM-based culture medium is greater than that of the DMEM-based culture medium (Materials and Methods).

When a circular donor plasmid pMB1KmHygTK-HPRT(ISCEI(318)Ex2Syn) was transfected with the plasmid expressing nls-I-SceI enzyme, the overall integration (Hyg^R) clones were obtained at a frequency of 0.95 × 10⁻³ per transfected viable cell (lane i in Figure 3A). Alternatively, when the linearized donor plasmid pMB1KmHygTK-HPRT(ISCEI(318)Ex2Syn) was transfected alone, the overall integration (Hyg^R) clones were obtained at a frequency of 0.35 × 10⁻³ per transfected viable cell (lane ii in Figure 3A), which significantly decreased (0.37-fold) compared with the co-transfection of the circular donor plasmid and the plasmid DNA expressing nls-I-SceI enzyme (lanes i and ii in Figure 3A).

When another donor plasmid, pMB1KmHygTK-HPRT(ISCEI(545)Ex2Syn) or pMB1KmHygTK-HPRT(Ex2Syn(1195)ISCEI), was used, the donor linearization technique significantly decreased (0.31–0.40-fold) compared with that using the InCDC technique in overall integration frequency per transfected viable cell (lanes i and ii in Figure 3B; lanes i and ii in Figure 3C).

Taking together these overall integration frequencies, the linearized donor DNA appears to be more susceptible to digestion in a cell than the circular donor plasmid co-transfected with nls-I-SceI expression plasmid DNA.

When the circular donor plasmid pMB1KmHygTK-HPRT(ISCEI(318)Ex2Syn) was transfected, the targeted duplication clones were obtained at a frequency of 3.3 × 10⁻⁴ (5/14820 Hyg^R: lane i in Figure 3A) per overall integration clone using the InCDC technique but were not obtained using the donor linearization technique (0/5816 Hyg^R: lane ii in Figure 3A).

When the circular donor plasmid pMB1KmHygTK-HPRT(ISCEI(545)Ex2Syn) was transfected, the targeted duplication clones were obtained at a frequency of 1.3 × 10⁻³ (11/8448 Hyg^R: lane i in Figure 3B) per overall integration clone using the InCDC technique but were not obtained using the donor linearization technique (0/3044 Hyg^R: lane ii in Figure 3B), for which statistical analysis showed that the TD frequency per overall integration clone with InCDC significantly increased compared with that using the donor linearization technique (*: lane i vs. lane ii in Figure 3B).

When the circular donor plasmid pMB1KmHygTK-HPRT(Ex2Syn(1195)ISCEI) was transfected, the targeted duplication clones were obtained at the frequencies of 3.2 × 10⁻⁴ and 3.1 × 10⁻⁴ (5/15636 Hyg^R; 5/16276 Hyg^R: lanes i and ii in Figure 3C) per overall integration clone using the InCDC technique but were not obtained using the donor linearization technique (0/7396 Hyg^R: lane iii in Figure 3C).

Taking together these TD frequencies, the InCDC technique is more efficient and successful in obtaining the targeted duplication clones than the donor linearization technique.

We confirmed that Syn-retaining TD clones were identified among the TD clones derived from each of the three types of circular donor plasmid using the InCDC techniques (2 of 5: lane i in Figure 3A; 4 of 11: lane i in Figure 3B; 3 of 5 and 3 of 5: lanes i and ii in Figure 3C).

We also analyzed using the RI test (Figure 2B) and the duplication test (Figure 1D) for the Hyg^RTg^R clones generated by each of the InCDC and donor linearization techniques to determine their final forms (targeted duplication, abortion-types, targeted deletion, and random integration), as summarized in Figure 3D. The forms except TD (non-TD forms) derived from InCDC were generated at the frequencies of 1/6, 4/15, and 4/14 per Hyg^RTg^R clone (lane v in Figure 3D); the non-TD forms derived from donor linearization were generated at the frequencies of 2/2, 3/3, and 3/3 per Hyg^RTg^R clone (lane x in Figure 3D). We statistically analyzed them to evaluate InCDC vs. linearization in gene editing accuracy and demonstrated that InCDC was more accurate than linearization (*: lane v vs. lane x; ^※: lane v vs. lane x; ^#: lane v vs. lane x in Figure 3D).

We found that the donor linearization technique resulted in lower overall integration frequencies and higher non-TD frequencies and was inefficient for generating targeted duplication. We also found that InCDC genome editing generated non-TD clones and TD clones, but the TD (HR) clones were more frequent than non-TD (non-HR) clones. This contrasts with the fact that CRISPR–Cas9 generated far more Indel (non-HR) clones than the HR clones (Maruyama et al., 2015).

When a plasmid expression vector is transfected to express the nls-I-SceI enzyme gene for intra-cellular cleavage, a DNA molecule of the plasmid vector can integrate at sites on host chromosomes, potentially causing disruption of the genomic structure in a patient receiving InCDC-mediated precision genome editing for therapy. Therefore, we attempted to use messenger RNA molecules expressing the nls-I-SceI enzyme gene.

When a circular donor plasmid pMB1KmHygTK-HPRT(ISCEI(545)Ex2Syn) was co-transfected into the recovered cells with 1, 2, or 5 μg of messenger RNA molecules expressing the nls-I-SceI gene, the targeted duplication clones were obtained at a frequency of 1.7–7.4 × 10⁻⁴ (1/5728 Hyg^R, 6/8100 Hyg^R, and 3/6520 Hyg^R: lanes i, ii, and iii in Figure 4A) per overall integration clone, which was similar to that with the pCI-nls-I-SceI plasmid (8/7228: lane iv in Figure 4A). We confirmed that the frequencies of Syn retention per TD clone were 1/1, 2/6, and 1/3 (lanes i, ii, and iii in Figure 4A), which were similar to that with the pCI-nls-I-SceI plasmid (2/8: lane iv in Figure 4A).

We also attempted to use an RNA virus vector, a temperature-sensitive SeV vector (Ban et al., 2011), expressing the nls-I-SceI enzyme (Figure 4B). The HT1080 cells were infected with SeV18+nls-I-SceI/TS15ΔF, amplified, and stored at −80 °C (Materials and Methods). When a circular donor plasmid pMB1KmHygTK-HPRT(ISCEI(545)Ex2Syn) was transfected into the recovered cells with the SeV vector expressing nls-I-SceI cleaver, the targeted duplication clones were obtained at a frequency of 3.8–6.8 × 10⁻⁴ (3/7932 Hyg^R, 5/8140 Hyg^R, and 7/10272 Hyg^R: lanes i, ii, and iii in Figure 4B) per overall integration clone, which were similar to those with messenger RNA molecules expressing the nls-I-SceI gene (lanes i, ii, and iii in Figure 4A), and the frequencies of Syn retention per TD clone were 0/3, 1/5, and 2/7 (lanes i, ii, and iii in Figure 4B), which were similar to those with messenger RNA molecules expressing the nls-I-SceI gene (lanes i, ii, and iii in Figure 4A).

These results indicate that messenger RNA or an SeV vector expressing the nls-I-SceI enzyme is useful for InCDC-mediated precision genome editing in clinical cells for cell therapy.

It is expected that a natural replacement clone (HPRT⁺) will generate as a GCV^R colony among TD cells (HPRT⁻; GCV^S background cells) incubated on a plated ware (Figure 1C); however, it is supposed that such desired GCV^R colonies will be decreased by intercellular bystander effects of GCV triphosphate molecules (Yang et al., 1998). The natural replacement clone should exhibit a HAT (hypoxanthine, aminopterin, and thymidine)-resistant phenotype (HPRT⁺ = HAT^R), whereas the TD background cells should exhibit a HAT-sensitive phenotype (HPRT⁻ = HAT^S). To verify whether GCV selects for natural replacement clones derived from targeted duplication cells, we first compared the frequency of GCV^R clones generated per viable cell with that of HAT^R clones to determine the accuracy of GCV selection for natural replacement (pop-out) clones, expressed as the ratio of GCV^R/HAT^R.

We used the HTG786 clone, a hetero-allelic (5′SEXA1/Syn3′) TD clone, which was obtained from the circular donor with HPRT(Ex2Syn(1195)ISCEI) (lane v in Figure 2A).

When 1.9 × 10⁴ viable cells of HTG786 were plated at 21 cells/cm² as a seeding viable cell density and selected by GCV medium or HAT medium from the fifth day after plating, four GCV^R colonies were observed; the generation of 39 HAT^R colonies was based on the data observed (lane i in Figure 5A). The frequency of each of the GCV^R clones and HAT^R clones per viable HTG786 cell plated is 0.21 × 10⁻³ and 2.0 × 10⁻³, respectively (GCV^R/HAT^R: 0.10). The resultant four GCV^R clones were analyzed using the pop-out test, Hyg test (Figure 1E), and duplication test (Figure 1D) so that three clones out of them were verified as the natural replacement (pop-out) type structure (GCV-A2, A3, and A4 in Figure 5B). To verify the generation of pop-out clones remaining the designed allele Syn, we analyzed these four GCV^R clones derived from HTG786 with the allele sequence test (Figure 1E). Out of four GCV^R clones, one clone had the Syn allele (upper of Figure 6D) and two clones had the SEXA1 allele (lane i in Figure 5C). The remaining clone indicated the presence of the Hyg gene and the doublet configuration (GCV-A1 in Figure 5B), suggesting that it is a TK-deficient TD clone (lane i in Figure 5C). The accuracy of GCV selection for pop-out clones (0.075) is finally calculated by multiplying 0.1 (GCV^R/HAT^R) by 3/4 (pop-out/GCV^R) when the seeding viable cell density is 21 cells/cm².

Second, the cell suspension of HTG786 was plated at 4.9 cells/cm² as a lower density of viable cells, considering intercellular bystander effects of GCV triphosphate molecules (Yang et al., 1998). GCV selection or HAT selection from day 5 after plating isolated two GCV^R colonies and eight HAT^R colonies (lane ii in Figure 5A). The frequency of each of the GCV^R clones and HAT^R clones per viable HTG786 cell plated is 0.90 × 10⁻³ and 3.6 × 10⁻³, respectively (GCV^R/HAT^R: 0.25). The resultant two GCV^R clones were analyzed using the pop-out test, Hyg test (Figure 1E), and duplication test (Figure 1D), and one of the clones was verified to have the natural replacement (pop-out) type structure (GCV-B1 in Figure 5B). To verify the generation of pop-out clones retaining the designed allele Syn, we analyzed the GCV^R clones using the allele sequence test (Figure 1E). Out of two GCV^R clones, one clone had the Syn allele (lane ii in Figure 5C; upper of Figure 6D). The remaining clone indicated the presence of the Hyg gene and the doublet configuration (GCV-B2 in Figure 5B), suggesting that it is a TK-deficient TD clone (lane ii in Figure 5C). We confirmed that the resultant eight HAT^R clones consisted of five Syn alleles and three SEXA1 alleles and no TK-deficient TD clones were included (lane iii in Figure 5C). The accuracy of GCV selection for pop-out clones (0.13) is finally calculated by multiplying 0.25 (GCV^R/HAT^R) by 1/2 (pop-out/GCV^R) when the seeding viable cell density is 4.9 cells/cm².

These results indicate that several pop-out clones are possible to obtain through the GCV selection against more than 2,000 viable cells of a hetero-allelic TD clone and that the plating at densities as low as approximately 5 seeding viable cells/cm² is useful to alleviate intercellular bystander effects of GCV triphosphate molecules (Yang et al., 1998).

We suppose that human genome correction mediated by the InCDC technology of circular donor cleavage-mediated targeted duplication and natural replacement requires the following process:1. Transfection of designed circular donor plasmid with an nls-I-SceI enzyme gene expression vector into human cells (TD1 and RI1 in Figure 1C)2. Hygromycin selection of the targeted duplication clones (TD2 in Figure 1C) or the random integration clones (RI2 in Figure 1C)3. PCR screening of TD clone-containing dishes for all of the dishes4. PCR screening of TD clone-containing wells for all of the wells containing the cells derived from a TD clone-containing dish5. Ganciclovir selection for natural replacement clones (TD3-1 and -2 in Figure 1C) derived from the TD clone (TD2 in Figure 1C)6. Isolation of natural replacement (pop-out) clones with the designed sequence (TD3-1 in Figure 1C)

To verify that the above process leads to natural replacement clones with the designed sequence, we transfected with a circular donor plasmid, pMB1KmHygTK-HPRT(ISCEI(545)Ex2Syn) (Syn allele), and an nls-I-SceI expression plasmid into HT1080 cells (SEXA1 allele) as the first step (lanes i and ii in Figure 6A) so that 2,136 Hyg^R colonies (overall-integration clones) were obtained on 16 dishes as the second step (lane iii in Figure 6A).

In the third step, to identify the TD dishes that contain targeted duplication clones are included, the Hyg^R colonies from each dish are created, and 16 frozen stocks are made. The genomic DNA is extracted from the cells derived from every dish and analyzed using PCR with the primer pair, plasmid-5′-inside forward primer and target-3′-outside reverse primer (TD screening in Figure 6B), to verify the structure of targeted duplication. Four TD-containing dishes (#3, 9, 12, and 14) out of 16 dishes were obtained (Figures 6A,B). To determine which of the four TD dishes contains Syn allele-retaining TD clones, the genomic DNA is also analyzed by PCR using the primer pair, target-5′-outside forward primer and Syn reverse primer (Syn-TD screening in Figure 6B). The #9 TD dish contained only Syn-TD clones (Figures 6A,B). Out of 16 dishes containing 2,136 overall integration clones as Hyg^R colonies, four dishes contained TD. Among these, only one dish contained Syn-TD. These results are in harmony with the frequencies of Syn-TD clones per overall integration clone [8/6,348 (0.1%): lane iii in Figure 2A; 4/8,448 (0.05%): lane i in Figure 3B; 2/7,228 (0.03%): lane iv in Figure 4A]. These frequencies are similar to those (0.02%) of HR clones per transfected cell obtained through CRISPR–Cas9 with oligonucleotide donor DNA in mammalian cells (Maruyama et al., 2015), implying that the screening sizes required to isolate desired HR clones are similar between InCDC and CRISPR–Cas9.

In the fourth step, to identify Syn-TD clones from TD dish stock #9 using 96-well plates, fresh cells derived from #9 TD dish stock were serially diluted, seeded at two cells per well into two sets of three 96-well plates, and then incubated in the presence of hygromycin. The resultant growing cells were observed in 128 and 127 wells, respectively (Figure 6A), and then analyzed by PCR using the primer pairs for TD screening and Syn-TD screening (Figure 6B) to verify Syn-retaining TD wells. Four wells (Syn-TD well ID: d9p1H10, d9p2B9, d9p2C8, and d9p3G12) out of 128 cell-growing wells and two wells (Syn-TD well ID: d9p5F12 and d9p6E8) out of 127 cell-growing wells were identified as Syn-retaining TD clones (Figures 6A,B). In this study, we quantified the frequencies of InCDC-mediated donor-DNA overall integration events (mostly random integration events) per viable cell and obtained values of approximately 1 × 10⁻³ (Figures 3, 4). This represents the risk level that each screened Syn-retaining TD well clone contains a random integration fragment somewhere in the chromosomes.

In the fifth step, the cell suspension of a Syn-TD well clone d9p5F12 was plated at 4.6 cells/cm² as a low density of viable cells. GCV selection from the fifth day after plating isolated five GCV^R colonies (lane iii in Figure 5A). The frequency of GCV^R clones per viable d9p5F12 cell plated is 1.2 × 10⁻³ (lane iii in Figure 5A). The resulting five GCV^R clones were analyzed using the pop-out test, Hyg test (Figure 1E), and duplication test (Figure 1D), and two of the clones were verified to have the natural replacement (pop-out)-type structure (GCV-C1 and C3 in Figure 5B).

To verify the generation of the pop-out clones retaining the designed allele Syn, we analyzed the GCV^R clones derived from d9p5F12 for the allele sequence test (Figure 1E). Out of five GCV^R clones, one clone had the Syn allele (lower of Figure 6D) and one clone had the SEXA1 allele (lane i in Figure 6C). The remaining three clones indicated the presence of the Hyg gene and the doublet configuration (GCV-C2, C4, and C5 in Figure 5B), suggesting that it is a TK-deficient TD clone (lane ii in Figure 6C). The other experiment with the same Syn-TD well clone d9p5F12 (lane iv in Figure 5A and GCV-D1–D9 in Figure 5B) showed similar results [one clone: pop-out with the Syn allele (lower of Figure 6D); two clones: pop-out with the SEXA1 allele; three clones: TK-deficient TD (lane ii in Figure 6C)] as the first experiment, except for the generation of three of TD-derived deletion clones (GCV-D7, D8, and D9 in Figure 5B and lane ii in Figure 6C).

A GCV^R colony-derived cell stock, which was obtained from each of these two GCV selection experiments against d9p5F12, contained clonally different cells consisting of a Syn clone and a TK-deficient TD clone or of a Syn clone and a SEXA1 clone. Thus, additional 96-well cell cloning for the GCV^R cell stock (Materials and Methods) was needed, as described in Supplementary Tables 6, 7.

These results demonstrate that the positive selection and the two-step (dish-to-96-well) screening are successful in obtaining a single Syn-TD clone, and the subsequent negative selection with additional cell cloning leads to the isolation of a single desired clone through InCDC-primed precision genome editing.

Targeted genomic breakage-related techniques have shown that the precise end-joining pathway, which is simple rejoining, was a rare event, and the microhomology-mediated end-joining pathway, in which the double-strand break has been filled with unexpected sequences or rejoined with a small deletion, was a frequent event in human cells (Maruyama et al., 2015). Targeted genomic breakage repeatedly occurs after the precise end-joining (Lin et al., 2013) so that the imprecise end-joining product is fixed to survive the host cell. That is the reason why these techniques mainly produce the microhomology-mediated end-joining products. A technique to regulate or attenuate CRISPR–Cas9 activity is useful to prevent repeated targeted breakage, leading to a decrease in imprecise end joining and genome toxicity (Kawamata et al., 2023). The current challenge of these techniques is to facilitate the desired pathway, such as the synthesis-dependent strand-annealing pathway, to achieve precise genome editing (genome correction).

The InCDC techniques have been programed to facilitate ends-in gene targeting crossover-type HR in host cells to produce the targeted duplication structures (Figures 1C,D). Examples of such a crossover-type HR-mediated targeted duplication had been reported as follows: (1) ends-in gene targeting into murine ES cells (Hasty et al., 1991) or human HT1080 cells (Brenneman et al., 1996) of linearized plasmid (neo) donor DNA; (2) ends-in gene targeting into Drosophila melanogaster of cleavable extrachromosomal donor DNA (Rong and Golic, 2000; Maggert et al., 2008); (3) ends-in gene targeting into Saccharomyces cerevisiae of linearized plasmid donor DNA (Hashida-Okado et al., 1998; Yamana et al., 2005; Miura et al., 2013). In this study, we verified the applicability to ex vivo precision genome editing (Figure 6) of InCDC-mediated crossover-type HR (targeted duplication) (Figures 7C–F). The current challenge of the InCDC techniques is to enhance a double-Holliday junction-mediated pathway for more efficient production of precision editing (genome correction). It is supposed that a putative exonuclease-attack occurs soon after a double-strand break and facilitates the formation of a broad-type double-Holliday junction, which stably connects the InCDC ends and its targeted locus (Figures 7C,D). Such a broad-type intermediate structure can protect against dissolution reactions of double Holliday structures, which produce non-crossover products (Wu and Hickson, 2003), thereby facilitating targeted duplication events. Thus, the safety distance from the ISCEI cleavage site to the designed sequence Syn protects the Syn sequence from a putative exonuclease attack and also serves to keep it outside a broad-type intermediate structure (third lane in Figure 7D), which facilitates Syn-retaining targeted duplication events (Figure 7D). The HPRT assay system, using the Syn allele and the SEXA1 allele for InCDC-mediated human genome editing, helps identify enzymes involved in crossover-type homologous recombination in humans.

A previous study (Brenneman et al., 1996) demonstrated meganuclease-induced pop-out events between the designed meganuclease-recognition sequence and the target DNA on the doublet configuration, which was intervened by plasmid DNA containing a positive selection gene (neo) in human cells. Other previous studies (Rong and Golic, 2000; Maggert et al., 2008) demonstrated the break-induced replacement events between the designed donor DNA and the target DNA in Drosophila. In mice, the natural replacement events of a duplication structure had been verified (Wu et al., 1994). In human HT1080 cells, we demonstrated that natural replacement clones could be isolated by negative selection without any artificial breakage in the doublet configuration, which was intervened by plasmid DNA bearing a positive–negative selectable fusion gene (HygTK). This natural property is supposed to be related to the popping out of the circular DNA, consisting of the plasmid DNA and either donor or target DNA, via DNA replication slippage (Viguera et al., 2001). In this study, we quantified the frequency of popping-out events per viable cell of a TD clone and obtained the values of more or less 3 × 10⁻³ in the 5.4-kb duplicative DNA regions (Figure 1E) based on the frequencies of HAT^R clones generated per viable TD cell (lanes i and ii in Figure 5A), which may indicate frequencies of DNA replication slippage.