crRNA design and plasmid construction
In the type I-E CRISPR–Cas3 system, CRISPR arrays are transcribed as a long precursor (pre-crRNA) composed of repeat–spacer–repeat units. Each repeat sequence forms a stable stem loop that is specifically recognized and cleaved by the Cas6 endoribonuclease. Cas6 processing generates individual mature crRNAs, each typically consisting of an 8-nt 5′ handle derived from the repeat, a 32-nt spacer that specifies the DNA target and a 21-nt 3′ handle derived from the repeat hairpin. After cleavage, Cas6 remains bound to the 3′ hairpin and the mature crRNA is incorporated into the Cascade surveillance complex. Within Cascade, Cas5 binds the 5′ handle, multiple Cas7 subunits align along the spacer to form the backbone, Cas6 caps the 3′ end and Cas8 and Cas11 complete the complex assembly. This arrangement is essential for target recognition and recruitment of Cas3 for unidirectional DNA degradation50.
The all-in-one plasmid (pRB-EF1a-Cas3-Cascade-U6v2-BbsI-GFP: pCas3) expresses the crRNA and Escherichia coli Cascade–Cas3 complex along with EGFP under the control of the EF1a promoter, ensuring strong and constitutive expression (Extended Data Fig. 2 and Supplementary Data 1). Each Cas protein is linked by 2A self-cleaving peptides, enabling efficient polycistronic expression, and the transcript terminates with a poly(A) signal to enhance mRNA stability and translation. All coding sequences downstream of the EF1α promoter were codon-optimized for mammalian expression. The plasmid was custom-synthesized by VectorBuilder on the basis of our design. Five crRNAs targeting the first or second exons of the mouse Ttr gene (ENSMUST00000075312.5) were designed (Supplementary Table 2). For each crRNA, two complementary 5′-phosphorylated oligonucleotides encompassing the crRNA sequence and BbsI restriction endonuclease site overhangs were synthesized, annealed, subcloned into a pCas3-Cascade-GFP plasmid digested with BbsI-HF (R3539L, New England Biolabs) and purified using NucleoSpin gel and PCR cleanup (740609, Macherey-N-nagel). G211 sgRNA for CRISPR–Cas9 was similarly subcloned into a pSpCas9(BB)-2A-GFP plasmid (pX458; Addgene, 48138). The resultant constructs were subjected to Sanger sequencing to verify proper sub-cloning of the crRNA or sgRNA sequence.
Cell culture
The mouse hepatoma cell line Hepa1-6 (BRCB1638) and the human hepatocellular carcinoma cell line HepG2 (RCB1886) were obtained from the RIKEN Cell Bank. MEFs were collected at E14.5 from the interbreeding of TtrhV30orf/hV30orf mice (Transgenic).
The cells were cultured in DMEM (08459-64, Nacalai Tesque, for Hepa1-6; 11885-084, Thermo Fisher Scientific, for HepG2) supplemented with 10% heat-inactivated FBS (A4736401, Thermo Fisher Scientific) and 0.1 mg ml−1 penicillin and streptomycin (26253-84, Nacalai Tesque) on collagen-I-coated dishes (IWAKI) at 37 °C in a humidified atmosphere with 5% CO2.
Plasmid transfection and FACS isolation of GFP-expressing cells
pCas3 or px458 plasmids were transfected by Lipofectamine 3000 (Thermo Fisher Scientific). Then, 48 h after lipofection, cells were subjected to FACS isolation of GFP+ cells using BD FACSAria Soap (BD Biosciences). Briefly, cells were detached using trypsin–EDTA solution (32777-44, Nacalai Tesque), then passed through a 35-μm cell strainer to achieve a single-cell suspension and sorted for GFP fluorescence, using nontransfected Hepa1-6 cells to define the background fluorescence level. Flow cytometry data were analyzed with FlowJo software (version 10.10.0, BD Biosciences). Gates were set on the basis of forward and side scatter to exclude debris and doublets, while GFP⁺ populations were defined using untransfected cells as negative controls. A representative gating strategy used for the isolation of GFP⁺ cells is shown in Supplementary Fig. 3. After FACS isolation, cells were harvested. gDNA was extracted using NucleoSpin Tissue XS (740901, Macherey-Nagel) and RNA was isolated using an RNeasy mini kit (74104, Qiagen).
Evaluation of genome-editing efficiency
PCR primers were designed around the target site (Supplementary Table 6) and the region was amplified using Quick Taq HS DyeMix (DTM-101, Toyobo), followed by electrophoresis on agarose gel.
To compare the genome-editing efficiency in a quantitative manner, ddPCR was performed. Primer and probe sequences are listed in Supplementary Table 7. TaqMan copy number reference assay, mouse, Tfrc and human RNaseP (Thermo Fisher Scientific) were used as reference. After PCR amplification, plates were transferred into the Bio-Rad QX-200 droplet reader. All assays were analyzed using the QX-200 droplet reader and Quantasoft Analysis Pro (Bio-Rad).
For RT–qPCR, RNA was subjected to RT using ReverTra Ace qPCR RT master mix with gDNA remover (FSQ-301, Toyobo). qPCR was performed using SsoAdvanced Universal SYBR green supermix (Bio-Rad) on an CFX96 deep-well real-time PCR detection system (Bio-Rad). Primers used for RT–qPCR are listed in Supplementary Table 8. The mean cycle threshold (Ct) of target gene (Ttr) expression was normalized to that for the control gene, mouse B2m or human GAPDH.
WGS with short-read and long-read sequencing
gDNA was extracted from pCas3-cr2-transfected Hepa1-6 and HepG2 cells. The TruSeq DNA PCR-free library kit (Illumina) was used to prepare the library for short-read sequencing. Next-generation sequencing (NGS) was conducted on a NovaSeq X Plus platform (2 × 150 bp) at Macrogen. Raw sequencing reads were processed following standard bioinformatics pipelines. Reads were mapped to the reference mouse genome (GRCm39) or human genome (GRCh38) and large deletions were identified using split-read and discordant-read analyses. Candidate off-target sites were filtered on the basis of split-read counts exceeding five and base-coverage ratios above 0.2. The relative split-read count ratio between Cas3-treated and control samples was calculated to identify high-confidence off-target loci, which were visualized using IGV software.
For long-read sequencing, sequencing libraries were prepared using the ligation sequencing kit V14 (Oxford Nanopore Technologies) and sequenced on a PromethION P2 solo sequencer equipped with a FLO-PRO114M flow cell and the SQK-LSK114 library preparation kit. Basecalling was performed with Dorado version 7.3 (super accuracy model version 5.0.0). Quality trimming was conducted using BBDuk (BBtools version 39.13) to remove bases with Q scores < 15 at read ends. Reads with Q scores ≥ 15 and lengths ≥ 100 bp were retained. Sequencing statistics are as follows. Mouse (Hepa1-6): Cas3, 10.1 million reads, 90.7 Gb, median Q score = 25.5; Cas9, 4.8 million reads, 33.8 Gb, Q score = 26.0; control, 4.0 million reads, 27.5 Gb, Q score = 27.1. Human (HepG2): UT05, 6.1 million reads, 86.7 Gb, Q score = 28.1; NTLA (Cas9), 7.7 million reads, 98.9 Gb, Q score = 28.1; control, 4.7 millionM reads, 73.0 Gb, Q score = 27.7. Reads were mapped using Minimap2 version 2.1 to GRCm39 (mouse) or GRCh38 (human). Cas3-mediated deletions (100 bp–1 Mb) were detected using Sniffles version 2.6.2 with the ‘–no-qc’ option. Germline deletions, defined as recurrent identical deletions (n ≥ 3) across all samples, were excluded along with flanking regions ±1 kb. To minimize false positives, we excluded (1) reads with MAPQ < 60; (2) reads with >5 deletions; (3) regions with ≥50 consecutive Ns and surroundings ±10 kb; and (4) regions with >3× median coverage, coverage < 10 or located within ±100 bp of deletions showing a secondary alignment rate ≥ 0.05 or mismatch rate ≥ 0.01. A 100-kb Cas3 target region (10 kb upstream, 90 kb downstream of the recognition site) was defined. Deletions within this region were evaluated after coverage normalization.
CapSeq of off-target analysis at POT of Cas3 and Cas9
Targeted regions were selected as follows. Probes around on-target regions at Ttr covered 90 kb upstream to 10 kb downstream of the PAM. Probes around POT regions of CRISPR–Cas3 covered 4.5 kb upstream to 0.5 kb downstream of the potential PAM. Probes of CRISPR–Cas9 covered 1 kb upstream and downstream of the potential PAM. For NGS, gDNA was extracted from transfected Hepa1-6 cells and sheared with SureSelect enzymatic fragmentation kit (Agilent). After preparation of the DNA library with SureSelectXT-HS2 reagents (Agilent) and custom probe kit designed by SureDesign, genomic sequence analysis was performed using HiSeq X (2 × 150 bp) according to the standard procedure at Macrogen. Discordant reads and split reads were extracted by SAMtools and Lumpy-sv, respectively, and the total number of each read at on-target or POT regions was counted using BEDTools. Mutation patterns at the on-target locus were analyzed from extracted discordant reads mapped at deletion hotspots predicted by split-read counts of every 100 bp of on-target locus. For off-target effects at POT regions, split-read counts were normalized by total read counts of each region and calculated by subtracting read counts of control samples from those of CRISPR-transfected samples. Small indel mutations around PAM sequences, which were undetected using the aforementioned methods, were analyzed by CRISPResso2 (http://crispresso.pinellolab.partners.org/) according to the developers’ protocol73.
RNA transfection
GFP, Cascade (Cas5, Cas6, Cas7, Cas8 and Cas11)–Cas3 and Cas9 mRNAs with modifications including Cap1 and/or N1mΨ triphosphate, 5Me-cytidine triphosphate and 5Mo-uridine triphosphate were synthesized by Elixirgen Scientific. crRNAs containing 2′-O-methyl and phosphorothioate modifications were synthesized by Integrated DNA Technologies. RNAs were introduced into approximately 80% confluent Hepa1-6, HepG2 and MEF cells in six-well plates using Lipofectamine MessengerMAX (LMRNA001, Thermo Fisher Scientific) according to the manufacturer’s instructions. An Attune NxT flow cytometer (Invitrogen) was used to detect GFP signals. Viability of transfected cells was evaluated by propidium iodide staining.
LNP formulation
mRNA was loaded into LNPs using microfluidic mixing methods74. An ethanol solution containing an ionizable lipid, 1,2-distearoyl-sn-glycero-3-phosphocholine, cholesterol and PEG-DMG at fixed molar ratios (50:3:47:1.5) was prepared at a total lipid concentration of 8 mM. Cas9 was loaded into LNPs at a weight ratio of 1:1 for Cas9 mRNA:sgRNA. For Sep-Cas3, Cas3, Cas5, Cas6, Cas7, Cas8, Cas11 and crRNA were incorporated at a weight ratio of 1:1:1:1:1:1:1. For DiT-Cas3, Cas3–Cas5–Cas11, Cas6–Cas7–Cas8 and crRNA were loaded at a weight ratio of 3:3:1, respectively. The RNA cargo was dissolved in 50 mM citrate buffer (pH 4.0), resulting in a final RNA cargo concentration of 81.5 µg ml−1. The lipid ethanol solution and RNA solution were rapidly mixed using a glass-based iLiNP device75 at a total flow rate of 5 ml min−1 and RNA-to-lipid flow rate of 3. The nitrogen-to-phosphate ratio was adjusted to 6. The resulting LNP solution was dialyzed for 2 h or more at 4 °C against 20 mM Tris-HCl buffer (9% sucrose, pH 7.40) using Slide-A-Lyzer G3 dialysis cassettes (molecular weight cutoff (MWCO): 20 kDa; Thermo Fisher Scientific). The LNP solution was concentrated by ultrafiltration using an Amicon Ultra-15 unit (MWCO: 100 kDa; Millipore). The size and polydispersity of LNPs were measured by a Zetasizer Nano ZS ZEN3600 instrument (Malvern Instruments). The encapsulation efficiency and total concentration of mRNA were measured by a Ribogreen assay76.
LNP delivery in vivo
All animal care and experimental procedures were approved by the Institutional Animal Care and Use Committee of the University of Tokyo (approval no. A21-57). Male Jcl:ICR (CLEA), TtrhV30orf/hV30orf (Transgenic) mice were kept under specific-pathogen-free conditions until use. Only male mice were used because serum TTR concentrations were consistently lower in females, which could introduce variability in data interpretation. Using male mice ensured consistency across experimental groups. All mice were housed in standard cages under a 12-h light–dark cycle (lights on from 7:00 a.m. to 7:00 p.m.) at a constant temperature of 22 ± 2 °C and relative humidity of 40–60%. Animals were provided with standard rodent chow and water ad libitum.
LNPs and saline as negative control were dosed through the lateral tail vein. Blood was collected into serum separator tubes (MiniCollect II, Greiner Bio-One) for circulating TTR quantitation and cytokine measurements. Tissue from the liver, spleen and kidney was collected for DNA and/or RNA extraction. Primary hepatocytes were isolated with the gentleMACS perfusion technology (Miltenyi Biotec) according to the manufacturer’s protocol.
TTR ELISA analysis
Total TTR serum levels were determined using a mouse prealbumin (TTR) ELISA kit (Aviva Systems Biology, OKIA00111) for Jcl:ICR mice, and human prealbumin (TTR) AssayMax ELISA kit (AssayPro, EP3010-1) for TtrhV30orf/hV30orf mice. Serum was diluted 1,000–10,000-fold for mouse Ttr and 4,000–8,000-fold for human TTR. Plates were read on a SYNERGY LX multimode reader (Bio Tek) at an absorbance of 450 nm. Serum TTR levels were calculated by Gen5 software (version 3.09). Final serum values were adjusted for assay dilution.
Histological analysis
Mice were deeply anesthetized by isoflurane and then perfused transcardially with 4% paraformaldehyde (PFA). Liver samples were postfixed with 4% PFA. Paraffin-embedded liver sections (4 μm thick) were stained with DFS and anti-mouse Ttr antibody (1:250; LS-C407961, LSBio). Frozen liver sections (4 μm thick) were stained with anti-TTR antibody (1:1,000; 201630-T10, SinoBiological), and anti-CD68 antibody (1:50; ab53444, Abcam)49,63. Images were acquired using a BZ-X700 fluorescence microscope (Keyence) and BZ-X-Analyzer software (Keyence).
Cas3 western blotting
Liver tissue samples of mice were flash-frozen in liquid nitrogen and homogenized in 2× cell lysis buffer (Cell Signaling) supplemented with cOmplete protease inhibitor (Sigma-Aldrich) using a TissueLyser II (Qiagen) according to the manufacturer’s instructions. Homogenized samples were incubated on ice for 30 min and centrifuged at 100g for 30 min at 4 °C. Supernatants were used as lysate samples. Total protein samples (30 μg) were separated on 4–12% Bolt Bis–Tris plus mini protein gels (Invitrogen) and transferred to polyvinylidene difluoride membranes (Invitrogen). Cas3 was detected with rat polyclonal anti-Cas3 (1:1,000, C4U Corporation), followed by anti-rat IgG, horseradish peroxidase (HRP)-linked antibody (1:2,000; 7077, Cell Signaling). α-Tubulin was detected with α-tubulin Antibody (1:2,000; Cell Signaling, 7077 and 2144), followed by anti-rabbit IgG, HRP-linked antibody (1:2,000; Cell Signaling, 7074). Signals were detected using SuperSignal WestPico plus chemiluminescence substrate (Thermo Fisher Scientific) and densitometry analysis was performed using iBright CL 1500 (Thermo Fisher Scientific).
Cytokine analysis
First, 50–100 μl of blood was collected by tail-vein nick for serum cytokine measurements at 4, 24 and 72 h and 1 week after administration. The enhanced chemiluminescence multiplex assay (U-PLEX custom biomarker group 1 (mouse) assays, K15069M-1) was used for cytokine analysis at the Research Center for Immunological Analysis. Serum was diluted threefold using sample diluent 41. Data were analyzed using Discovery Workbench 4.0 (Meso Scale Discovery).
Amplicon sequencing
The ratio of IFMs generated by Cas9 was analyzed by NGS of PCR amplicons. gDNA was extracted from MEFs isolated from TtrhV30orf/hV30orf mice and the 240-bp region including estimated cut site was amplified by PCR. Primer sequences are listed in Supplementary Table 9. Amplicons from PCR were purified by column purification with NucleoSpin gel and PCR cleanup kit (Macherey-Nagel) and sequenced by MiSeq (2 × 250 bp) according to a standard procedure at the NGS Core facility at Osaka University. Raw reads from each sample were analyzed by the ratio of modified to unmodified reads using CRISPResso2 (http://crispresso.pinellolab.partners.org/) according to the developers’ protocol73.
nanoLC–MS/MS analysis
Each serum sample was lysed in 8 M urea, reduced with 1 mM dithiothreitol for 90 min and alkylated with 5.5 mM iodoacetamide for 30 min. After digestion with MS-grade lysyl endopeptidase (Fuji Film Wako Chemicals) at 37 °C for 3 h, the resulting peptide mixtures were diluted with 10 mM Tris-HCl pH 8.2 to achieve a final concentration of <2 M urea and subsequently digested with MS-grade trypsin gold (Promega) at 37 °C for 3 h. An equal amount of trypsin was added for overnight digestion and fragmented peptides were desalted using ZipTip C18 (Millipore). Shotgun proteomic analyses were performed by Orbitrap Eclipse Tribrid MS instrument with FAIMS Pro interface (Thermo Fisher Scientific) connected to a Vanquish Neo ultrahigh-performance LC system (Thermo Fisher Scientific). Peptide samples were separated using a linear gradient of 2–24% mobile phase (0.1% formic acid in acetonitrile) at 300 nl min−1. Full-scan MS spectra were acquired at a resolution of 120,000 and subsequent MS/MS scans were performed in the ion trap using collision-induced dissociation fragmentation with a normalized collision energy of 35% with 10 ms of maximum injection time. Protein identification was conducted by searching against the customized database of UniProt mouse reference proteome (UP000000589) and WT/IFM-derived TTR amino acid sequence data (Fig. 5b and Extended Data Figs. 6 and 7) using the Sequest HT algorithm in Proteome Discoverer Software (version 2.5; Thermo Fisher Scientific).
Recombinant TTR expression and purification
WT and two mutant forms of TTR (V30M and IFM4) were expressed in an E. coli expression system. The overexpressed proteins were purified using a nickel affinity resin (Qiagen), followed by size-exclusion chromatography with a HiLoad 10/300 Superdex 75-pg column (Cytiva) running on an ÄKTA pure 25 system (Cytiva).
Aggregation assay
Under two pH conditions (pH 7.5 and 4.5) at 37 °C, TTR precipitated as large aggregates that can be measured by turbidity using the Prometheus Panta (NanoTemper Technologies GmbH). The turbidity of the protein solution was recorded over a period of 72 h to monitor aggregation kinetics. For TTR tetramer stabilization studies, TTR aggregates were also monitored by SDS–PAGE. Specifically, TTR samples (0.5 mg ml−1) were incubated in acetate buffer (100 mM Tris, 50 mM acetate and 100 mM KCl, pH 4.5–7.0) for 72 h at 37 °C. The TTRs were immediately processed under nonreducing conditions and were loaded on 4–12% Bis–Tris plus mini protein gels (Invitrogen). Protein bands were visualized by Coomassie brilliant blue staining (Nacalai Tesque).
Statistics and reproducibility
Statistical analysis was performed using GraphPad Prism 10 software. All results are presented as the mean ± s.e.m. Comparisons between two groups were performed using an unpaired two-tailed Student’s t-test for parametric data or two-tailed Student’s t-test with Welch’s correction if the s.d. was not equal. Comparisons among multiple groups were analyzed by one-way analysis of variance (ANOVA). All experiments were independently repeated at least three times with similar results. Representative images shown in Fig. 4b and Extended Data Figs. 4c, 6b,i and 8a,e,f are from experiments that yielded comparable outcomes. All relevant experimental conditions, including age, sex, body weight and culture conditions, were controlled and standardized across all experimental groups. Therefore, consideration of additional covariates was not applicable for this study.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
