Compact Cas12f enables genome editing in avian cells

Abstract

Precise genome editing in avian species has been constrained by the low delivery efficiency of conventional CRISPR nucleases, such as Cas9 and Cas12a, due to their large molecular sizes. Cas12f (also known as Cas14), a compact CRISPR nuclease, has emerged as a potential genome editing system with enhanced delivery efficiency in mammalian systems. However, its effectiveness in avian systems has not been previously validated. Here, Cas12f showed notable transfection efficiency and intracellular expression in chicken Leghorn male hepatoma (LMH) cells and primordial germ cells (PGCs), with no detectable cytotoxicity. Next-generation sequencing (NGS) revealed that Cas12f achieved locus-dependent on-target editing efficiencies, reaching up to 40% at specific loci in LMH cells. Cas12f consistently generated a deletion-dominant indel profile with minimal insertions, distinct from Cas9-mediated patterns. Off-target analysis using Sanger sequencing and Inference of CRISPR Edits (ICE) revealed a few predicted off-target candidates and no detectable off-target mutations above the detection threshold. Consistent with this observation, cross-species in silico analysis showed only a modest increase in predicted Cas12f off-target proportions with increasing genome size. These findings show that Cas12f is a compact genome editing tool in avian cells, serving as a basis for further improvement in genetic engineering and biotechnological research.

Introduction

Throughout history, chickens have been an essential source of protein for humans, delivering diverse benefits in both research and industry. Owing to their oviparous nature, embryos are easily accessible and manipulable, affording distinct advantages as model animals for embryonic development (Lee et al., 2019). Moreover, chickens are increasingly recognized as valuable model animals for studying human health and disease, as their genetic and physiological characteristics enable the investigation of complex biological processes. In this context, precise genome editing in chickens remains scientifically and practically important (Han and Lee, 2017).

The development of genome-editing technologies, including zinc finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs), and clustered regularly interspaced short palindromic repeat (CRISPR)/CRISPR-associated protein (Cas) system, has changed the generation of genome-edited animals (Khalil, 2020). Cas9 (1,368 amino acids (aa)) and Cas12a (1,307 aa) nucleases, in particular, have been widely adopted for targeted gene modification due to their high efficiency and flexibility (Swarts and Jinek, 2018). Nevertheless, their relatively large size often limits delivery efficiency, notably in systems with low transfection capacity, such as chicken cells (Vergara et al., 2015; Molnar et al., 2004). This has driven growing interest in compact Cas variants that retain consistent editing activity while granting enhanced deliverability.

Cas12f (also known as Cas14) is an exceptionally small class-II type-V CRISPR nuclease that consists of 400-700 aa and is capable of cleaving DNA with a T-rich 5′ PAM sequence (5′-TTTR-3′, R = A or G) (Harrington et al., 2018). Cas12f contains an RuvC nuclease domain capable of inducing double-strand DNA (dsDNA) breaks (DSBs), which are predominantly repaired by the host cell through non-homologous end joining (NHEJ) (Karvelis et al., 2020). Its compact size also makes Cas12f particularly suitable for in vivo gene therapy applications or vector systems with restricted cargo capacity, such as adeno-associated virus (AAV) and avian adeno-associated virus (A3V) (Wu et al., 2021; Cui et al., 2024; Robinson et al., 2022). However, native Cas12f often shows very low activity in eukaryotic cells (Kim et al., 2022).

Recent studies have shown that engineered Cas12f variants can induce targeted mutagenesis in both bacterial and mammalian cells. This emphasizes their potential as efficient genome editors. Specifically, Cas12f_ge4.1 is an engineered variant developed to overcome the low cleavage activity of native Un1Cas12f1 by redesigning the guide RNA (gRNA) architecture. Cas12f_ge4.1 exhibited an approximately 867-fold increase in average indel frequency compared with the original system. It demonstrated efficiency comparable to Cas9 while preserving high target specificity (Kim et al., 2022).

Despite recent advances in engineered Cas12f systems in bacterial and mammalian systems, its use in avian systems is still unexplored. Establishing a Cas12f-mediated editing platform in chicken cells could advance avian genetic engineering. Here, we report the first experimental evidence of Cas12f-mediated editing in an avian system. We evaluated the feasibility and efficiency of Cas12f-mediated editing in chicken cells. The focus was on transfection efficiency, on-target editing efficiency, and cell viability. Together, these outcomes establish a foundation for applying compact CRISPR systems in avian genome editing.

Materials and methods

Vector construction

For Cas9-mediated editing, the plasmids pSpCas9(BB)-2A-GFP (PX458) (#48138, Addgene, MA, USA) and pSpCas9(BB)-2A-Puro (PX459) (#48139, Addgene) were used. For Cas12f-mediated editing, the plasmid Cas12f-GE ver4.1(_GFP) (#176544, Addgene) was used, and the Cas12f-GE ver4.1_Puro was generated by replacing the GFP cassette with a puromycin resistance gene through digestion and ligation using EcoRI restriction enzyme (#R0101L, New England Biolabs, MA, USA) and T4 ligase (#2011A, Takara, Shiga, Japan). gRNA sequences (Table 1) were inserted into each vector using Golden Gate assembly with the BpiI (BbsI) restriction enzyme (#ER1011, Thermo Fisher, MA, USA) and T4 ligase. All plasmids were purified using a Plasmid Transfection-grade Prep Kit (#740490, MACHEREY-NAGEL, Dueren, Germany) or an Endo-free plasmid maxi kit (#12362, Qiagen, Hilden, Germany) according to the manufacturer’s instructions.

Table 1.

sgRNA target locus.

Name	5′-gRNA Seq-3′		Chromosome	Locus	Nuclease
COL8A1	CCC ACA GGG CCC TGA GGG CC		CM000093.5	85386690	Cas12f, Cas9
MYLIP	CTG CCA GTG AAC TGC AGC CC		CM000094.5	61480697	Cas12f
	#1	GCA GTT CAC TGG CAG CAA AG		61480703	Cas9
	#2	TGC AGT TCA CTG GCA GCA AA		61480702
NOD1	CTG AAG GTA TAC CGA GAG CT		CM000094.5	41184989	Cas12f
	#1	GAA GGT ATA CCG AGA GCT TC		41184996	Cas9
	#2	AGC AAA GCA ATG CAG GAT GG		41184976
RAG1	TTT TCC TAG GAA CAG CAC GA		CM000097.5	19742909	Cas12f, Cas9
VEGFA	CTG TGC TGT AAG AAG CTC AT		CM000095.5	30782729	Cas12f
	#1	CGC TAT GTG CTG ACT CTG AT		30782757	Cas9
	#2	CTG GAA TGA AAG AAA CAG AG		30782796

Cell culture and transfection

Leghorn male hepatoma (LMH) cells were sub-passaged every 2–4 days in Waymouth's Medium (#11220035, Thermo Fisher) supplemented with fetal bovine serum (FBS) (#SH3091903, Cytiva, MA, USA) and Antibiotic-Antimycotic (ABAM) (#15240062, Thermo Fisher). LMHs were incubated at 37°C in an atmosphere of 5% CO₂ and 60–70% relative humidity.

For LMH cells, 3 µL of Lipofectamine 2000 reagent (#11668027, Thermo Fisher) and 2 µg of the CRISPR plasmid were mixed in 500 µL of Opti-MEM (#31985070, Thermo Fisher), and the mixture was applied to 1 × 10⁵ LMH cells. The medium was replaced with fresh culture medium 6 h post-transfection. Puromycin (1.5 μg/mL) was added to the LMH culture medium 48 h post-transfection, and selection was performed for up to 2 days, by which time the majority of wild-type (WT) cells had died.

Chicken primordial germ cells (PGCs) were isolated from the gonads of embryonic day 6 (E6) Korean Ogye embryos. PGCs were sub-passaged every 3–5 days in knockout DMEM (#10829018, Thermo Fisher) supplemented with 7.5% FBS, 2.5% chicken serum (#D102000500, Rockland, PA, USA), nucleosides (#ES008D, Merck, Darmstadt, Germany), nonessential amino acids (#11140050, Thermo Fisher), ABAM, β-mercaptoethanol (#21985023, Thermo Fisher), 10 mM sodium pyruvate (#11360070, Thermo Fisher), 2 mM L-glutamine (#35050061, Thermo Fisher), and human basic fibroblast growth factor (10 ng/ml; #233FB, R&D Systems, MN, USA). PGCs were incubated at 37°C in an atmosphere of 5% CO₂ and 60–70% relative humidity.

For PGCs, 3 µL of Lipofectamine 2000 reagent, 2 μg of CRISPR plasmid were suspended in 100 µL Opti-MEM, and this mixture was applied to 1 × 10⁵ cultured PGCs with 400 μL PGC culture medium. Gentle pipetting was carried out at 1-1.5 h intervals, and the medium was changed to PGC culture medium 6 h post-transfection. PGCs below passage 30 were used for all experiments.

Flow cytometry analysis

Transfected cells were harvested, resuspended in Flow buffer (2% FBS in PBS), and analyzed with Attune NxT Flow Cytometer (Thermo Fisher). Subsequent analyses were performed using FlowJo software (Treestar, OR, USA). ≥30,000 events/sample were collected, and GFP-positive cells were defined relative to vehicle controls. Viabilities were distinguished through Live/Dead Fixable Near-IR Dead Cell Stain Kit (#L34975, Thermo Fisher). The gating strategy was shown in Fig. S1.

gDNA extraction, PCR, and PCR purification

Genomic DNA (gDNA) was extracted from each cell sample, and DNA concentration was measured using NanoDrop-2000 (Thermo Scientific). The final DNA concentration for every sample was diluted to 100 ng/µL and used for PCR. Primers were designed using Geneious Prime (Biomatters, Auckland, New Zealand) according to the reference chicken genome available on the National Center for Biotechnology Information (NCBI) and synthesized (Cosmogenetech, Seoul, South Korea) (Table S1).

PCR reactions were performed with a total PCR mixture volume of 20 µL containing 0.4 µL of dNTP (10 mM each) (#DN112, Biofact, Seoul, South Korea), 2 µL of 10X Buffer, 0.1 µL of Taq Polymerase (5 units/µL) (#ST111, Biofact), 4 µL of 5X Band Helper (#BB741, Biofact), 1 µL of each primer (10 µM each), 2 µL of extracted gDNA, and 9.5 µL of Ultra-Pure Water (UPW) (#ML01902, Welgene, Namcheon, South Korea). PCR was conducted under the following thermocycling conditions: initial denaturation at 95°C for 5 minutes, followed by 35 cycles of denaturation at 95°C for 30 seconds, annealing at the appropriate temperature (Table S1) for 30 seconds, extension at 72°C for 60 seconds, then a final extension at 72°C for 10 minutes. After confirming successful PCR amplification of the target loci on a 1.5% agarose gel via gel electrophoresis, each amplicon was purified using the Wizard™ SV Gel and PCR Clean-Up System (#A9282, Promega, WI, USA).

Next-generation sequencing (NGS) analysis

NGS was outsourced (Bionics, Seoul, South Korea), using the MiSeq system (Illumina, CA, USA). The resulting FASTQ (.fastq.gz) files were analyzed using CRISPResso2 (Clement et al., 2019). The quantification window center was set to 2 bp downstream and 3 bp upstream from the 3′ end of the gRNA spacer sequence for Cas12f and Cas9, respectively, and the window size was defined as 10 bp on each side of the center. Only indel frequencies were quantified, excluding substitutions.

Amplicons were also analyzed via T7 endonuclease I (T7E1) assay (#M0302L, New England Biolabs). Following denaturation, the amplicons were reannealed to form heteroduplex DNA. Subsequently, the heteroduplex amplicons were treated with T7E1 for 20 min at 37°C and then analyzed by 2% agarose gel electrophoresis. The image was taken and analyzed by the ChemiDoc XRS+ System (Bio-Rad, CA, USA) and Image Lab Software (Bio-Rad).

Prediction of off-target sites and analysis

The CRISPR RGEN tools provided Cas-OFFinder (Bae et al., 2014) (http://www.regenome.net/cas-offinder/) was used to predict putative off-target sites. Potential off-target sites with up to 3-bp mismatches in the chicken reference genome Gallus_gallus-6.0 (GCRc6a) were selected (Table S2). Each site was amplified by target-specific primers (Table S1) and analyzed via Inference of CRISPR Edits (ICE) analysis (Conant et al., 2022) using an ABI Prism 3730XL DNA Analyzer (Thermo Fisher) and available software (https://ice.editco.bio/#/). When no off-targets were detected, the off-target frequency was set to the limit of detection (1%) of the ICE analysis to allow calculation of the specificity ratio.

For the analysis of genome-size–associated trends in predicted off-target proportions, the following reference genomes were used: chicken (Gallus gallus, GRCg6a), pig (Sus scrofa, susScr11), mouse (Mus musculus, mm10), rat (Rattus norvegicus, rn5), and human (Homo sapiens, GRCh38).

Statistical analysis

Statistical analyses were performed using GraphPad Prism (GraphPad Software, CA, USA). Statistical significance between two groups was determined using a two-tailed unpaired Student’s t-test and Welch’s t-test. Differences were considered statistically significant when P < 0.05. All data represent mean ± Standard Error of the Mean (SEM).

Results

Delivery and expression of Cas12f in chicken cells

To assess the delivery efficiency and intracellular expression of Cas12f, LMH cells and PGCs were transfected with GFP-expressing plasmids encoding either Cas12f or Cas9. Cas9 was used as a reference nuclease. Both constructs shared the same promoters and reporter configurations, consisting of CMV-driven nuclease expression followed by a T2A-linked GFP cassette and a U6-driven single-guide RNA (sgRNA) module, thereby allowing direct comparison of nuclease expression levels (Fig. 1A). Bright-field and GFP fluorescence imaging revealed successful transfection in both cell types, with GFP-positive cells readily detected following introduction of either construct (Fig. 1B).

Fig. 1.

Evaluation of cell viability and transfection performance of Cas12f and Cas9 in LMH cells and PGCs (A) Schematic representation of the Cas12f and Cas9 expression plasmids. Both constructs contain a CMV promoter–driven nuclease (Cas12f or Cas9) followed by a T2A-linked GFP reporter, together with a U6 promoter–driven sgRNA cassette. An ampicillin resistance gene (Amp) is included for bacterial selection. (B) Representative bright-field and GFP fluorescence images showing transfection outcomes in LMH cells and PGCs following delivery of Cas12f or Cas9 constructs. Scale bars = 50 μm. (C) Quantification of cell viability 48 h post-transfection across both cell types, showing no significant differences among Cas12f and Cas9. (D) GFP-based transfection efficiency measurements show variable but generally consistent transfection rates for Cas12f across both cell types. (E) Quantification of MFI in LMH cells and PGCs. Cas12f-transfected cells exhibited strong GFP expression, indicating strong expression and delivery efficiency. Data represent mean ± SEM of three independent biological replicates. P-values were determined by a two-tailed unpaired Student’s t-test. Statistical significance is indicated as ns (not significant), *P < 0.05, **P < 0.01, and ***P < 0.001.

To determine whether the expression of the compact Cas12f affects cellular viability, cell survival was assessed 48 h post-transfection, in both LMH and PGCs. No significant differences in viability were observed among Cas12f-, Cas9-treated groups in both LMH cells (90.07 ± 0.96% vs. 92.67 ± 1.46%, n = 3) and PGCs (96.74 ± 0.349% vs. 95.97 ± 0.20%, n = 3) (Fig. 1C). These results suggest that efficient delivery and expression of Cas12f are not associated with increased cellular toxicity.

To evaluate transfection efficiency and intracellular expression, flow cytometric analysis was performed. Cas12f was successfully delivered and expressed in both LMH cells (62.50 ± 0.95% vs. 43.67 ± 1.19%, n = 3, P < 0.001) and PGCs (16.33 ± 0.84% vs. 8.01 ± 2.22%, n = 3, P = 0.025) (Fig. 1D). Likewise, mean GFP fluorescence intensity (MFI), used as a proxy for transgene expression, was markedly increased in Cas12f-expressing cells in both LMH cells (112247 ± 2004.08 vs. 60981 ± 956.70, n = 3, P < 0.001) and PGCs (23739 ± 1539.49 vs. 9880 ± 435.55, n = 3, P < 0.001) (Fig. 1E, Figure S1C). Overall, these data establish Cas12f as a genome editing nuclease with adequate delivery and expression in avian cells.

On-target genome editing by Cas12f in LMH cells

LMH cells and PGCs were transfected with Cas12f- or Cas9-expressing plasmids that enabled puromycin selection, and on-target genome editing efficiencies were evaluated by NGS. Both vectors shared the same promoters and reporter configurations, consisting of CMV-driven nuclease expression followed by a T2A-linked puromycin resistance gene (Puro) cassette and a U6-driven sgRNA module (Fig. 2A). sgRNAs targeting exon 1 of Collagen type VIII alpha 1 (COL8A1), Nucleotide-binding oligomerization domain-containing 1 (NOD1), Recombination activating gene 1 (RAG1), exon 2 of Myosin regulatory light chain interacting protein (MYLIP), and exon 4 of Vascular endothelial growth factor A (VEGFA) were designed, based on TTTR PAMs for Cas12f and the canonical NGG PAM for Cas9 (N = A or T or C or G). To enable direct comparison between Cas12f and Cas9 using an identical spacer sequence, target sites containing a TTTR–N20–NGG motif were required. However, such motifs were not readily available at suitable exon positions in most target genes. Consequently, a shared spacer design was applied only to COL8A1 and RAG1, where appropriate sites were identified. For MYLIP, NOD1, and VEGFA, Cas9 target sites located proximal to the selected Cas12f sites were instead chosen to allow contextual comparison within the same exon (Fig. 2B; Table 1).

Fig. 2.

Genome editing activities of Cas12f in LMH cells (A) Schematic representation of the Cas12f and Cas9 expression plasmids. Both constructs contain a CMV promoter–driven nuclease (Cas12f or Cas9) followed by a T2A-linked puromycin resistance gene (Puro), together with a U6 promoter–driven sgRNA cassette. An ampicillin resistance gene (Amp) is included for bacterial selection. (B) Schematic representation of the target loci. Cas12f PAMs (red, TTTR), target sequences (blue, N20), and Cas9 PAMs (green, NGG) are indicated. sgRNAs target site designed within exon 1 of COL8A1, RAG1, and NOD1, exon 2 of MYLIP, and exon 4 of VEGFA. For MYLIP, NOD1, and VEGFA, two independent Cas9 sgRNAs within the same exon were tested, and editing efficiencies are shown separately. (C) Representative bright-field of Leghorn male hepatoma (LMH) cells transfected with Cas12f or Cas9. Images show overall cell morphology and viability following transfection and puromycin selection. Scale bars = 50 μm. (D) Genome editing efficiencies at COL8A1, MYLIP, NOD1, RAG1, and VEGFA loci. Cas12f- and Cas9-mediated editing efficiencies were quantified by NGS-based deep sequencing. Data represent mean ± SEM of three independent biological replicates. P-values were determined by a two-tailed unpaired Student’s t-test. Statistical significance is indicated as *P < 0.05, **P < 0.01, and ***P < 0.001. (E) Comparison of average genome editing efficiencies by Cas12f and Cas9 across all target loci. Data represent mean ± SEM. P-values were determined by a Welch’s t-test.

LMH cells transfected with either Cas12f or Cas9 displayed comparable morphology and viability 48 h post-transfection and 48 h puromycin selection (Fig. 2C). Genome editing efficiencies at each target locus were subsequently evaluated by NGS-based deep sequencing.

Cas12f induced detectable on-target editing at all tested loci, although locus-dependent variability was observed (Table 2). For contextual reference, Cas9-mediated editing was assessed at selected loci. Overall statistical comparisons indicated that editing efficiencies differed significantly between loci, confirming strong locus- and gRNA-dependent effects. At certain loci, Cas12f exhibited editing efficiencies comparable to Cas9 (e.g., COL8A1), whereas at others Cas9 achieved higher efficiencies (e.g., RAG1, MYLIP, or NOD1) (Fig. 2D; Fig. S2). However, these differences were not uniform across all target sites. When editing efficiencies from five target loci were pooled for integrated analysis, Cas12f-mediated editing efficiency was slightly lower than that of Cas9 (21.60 ± 3.56% vs. 31.93 ± 4.25%, P = 0.069) and did not reach statistical significance compared with Cas9-mediated editing (Fig. 2E). Thus, the apparent advantage of one nuclease over the other was context-dependent rather than absolute. Overall, these results show that Cas12f can achieve measurable, but locus-dependent, on-target genome editing efficiencies in chicken cells.

Table 2.

On-target genome editing efficiency by Cas12f and Cas9 in LMH cells.

Target Gene	Nuclease		Editing Efficiency (%)	SEM	P value (vs. Cas12f)	n
COL8A1	Cas12f		14.62	1.23		3
	Cas9		9.20	0.77	0.020
RAG1	Cas12f		1.92	0.60
	Cas9		30.55	5.68	0.007
MYLIP	Cas12f		26.12	2.78
	Cas9	#1	57.71	5.09	0.006
		#2	14.13	0.85	0.015
NOD1	Cas12f		24.30	3.14
	Cas9	#1	47.10	3.17	0.007
		#2	39.34	3.57	0.034
VEGFA	Cas12f		40.83	2.63
	Cas9	#1	3.18	0.56	< 0.001
		#2	54.20	4.93	0.075

Cas12f induces a characteristic deletion-dominant indel profile

To further characterize Cas12f-mediated genome editing in LMH cells, we analyzed indel profiles surrounding each target locus. For Cas9, maximal editing activity occurred 3 bp upstream of the 3′ end of the target locus, consistent with its cleavage site, whereas for Cas12f, maximal editing was detected 2 bp downstream of the 3′ end. Cas12f generated broad deletion peaks, whereas insertion events were comparatively infrequent and limited in size (Fig. 3A).

Fig. 3.

Characterization of indel profiles generated by Cas12f in LMH cells (A) Representative indel size distribution plots for Cas12f- and Cas9-mediated editing. Red peaks represent deletion events and green peaks indicate insertion events. Vertical dashed lines denote the predicted nuclease cleavage positions, and the grey-shaded region indicates the quantification window used for analysis. (B) Comparison of the average insertion and deletion sizes generated by Cas12f and Cas9 at each locus. Data represent mean ± SEM of three independent biological replicates. P-values were determined by a two-tailed unpaired Student’s t-test. Statistical significance is indicated as ns (not significant), *P < 0.05, **P < 0.01, and ***P < 0.001. (C) Comparison of the average insertion and deletion sizes generated by Cas12f and Cas9 averaged across all target loci. Cas12f induces significantly larger deletions, whereas insertion sizes remain smaller than Cas9. Data represent mean ± SEM. P-values were determined by a Welch’s t-test. Statistical significance is indicated as ***P < 0.001.

Quantitative comparison across shared target loci showed that Cas12f produced significantly larger average deletions distinct from indel patterns typically reported for Cas9 at the COL8A1 locus and a similar trend at the RAG1 locus. In contrast, insertion events induced by Cas12f were minimal relative to those generated by Cas9. Consistent with these outcomes, all target loci showed deletion-dominant indel patterns and larger average deletions in Cas12f-mediated editing, distinct from those in Cas9-mediated editing, although some differences did not reach statistical significance (Fig. 3B; Table 3). When indel sizes from five target loci were pooled for integrated analysis, Cas12f-mediated editing resulted in significantly larger deletions (10.58 ± 0.52 vs. 6.16 ± 0.59, P < 0.001) and minimal insertions (0.06 ± 0.03 vs. 0.52 ± 0.07, P < 0.001) compared with Cas9-mediated editing (Fig. 3C). Collectively, these results show that Cas12f induces a reproducible deletion-dominant indel profile across multiple genomic loci in chicken cells.

Table 3.

Indel patterns generated by Cas12f and Cas9 in LMH cells.

Target Gene	Nuclease		Deletion (bp)	SEM	P value (vs. Cas12f)	Insertion (bp)	SEM	P value (vs. Cas12f)	n
COL8A1	Cas12f		10.10	0.41		0.18	0.09		3
	Cas9		6.14	0.69	0.008	0.69	0.32	0.203
RAG1	Cas12f		9.48	0.56		0.00	0.00
	Cas9		6.92	0.41	0.021	0.75	0.03	< 0.001
MYLIP	Cas12f		11.15	0.20		0.01	0.01
	Cas9	#1	4.87	0.13	< 0.001	0.29	0.01	< 0.001
		#2	8.68	0.18	0.001	0.57	0.02	< 0.001
NOD1	Cas12f		12.11	0.11		0.06	0.01
	Cas9	#1	5.78	0.12	< 0.001	0.19	0.01	0.001
		#2	7.94	0.49	0.001	0.58	0.02	< 0.001
VEGFA	Cas12f		10.55	0.49		0.05	0.01
	Cas9	#1	4.55	2.28	0.062	0.61	0.09	0.003
		#2	4.06	0.20	< 0.001	0.47	0.02	< 0.001

Cas12f exhibits limited detectable off-target editing

To evaluate the specificity of Cas12f and Cas9, potential off-target sites for sgRNAs targeting COL8A1, RAG1, MYLIP, NOD1, and VEGFA were predicted using Cas-OFFinder. The total number of potential off-target sites predicted in silico for each sgRNA was quantified (Fig. 4A). Potential off-target sites were identified using parameters of ≤1 DNA bulge, ≤1 RNA bulge, and ≤3 mismatches relative to the on-target sequence. Across all loci combined, the total number of predicted off-target sites was significantly lower for Cas12f than for Cas9 (166.20 ± 30.04 vs. 940 ± 217.21, P = 0.009) (Fig. 4B). This numerical assessment enabled a comparative evaluation of the theoretical off-target burden for each nuclease and provided a quantitative interpretation of the specificity.

Fig. 4.

In silico prediction and validation of off-target effects of Cas12f (A) Number of potential off-target sites predicted in silico for each sgRNA. Off-target candidates were identified using parameters of ≤1 DNA bulge, ≤1 RNA bulge, and ≤3 mismatches relative to the on-target sequence. (B) Number of potential off-target sites for Cas12f and Cas9 predicted averaged across all target loci. Data represent mean ± SEM. P-values were determined by a Welch’s t-test. Statistical significance is indicated as **P < 0.01.(C) Editing efficiencies at the on-target site and predicted off-target sites of each gRNA were measured. Cas12f-mediated editing resulted in minimal detectable editing within the ICE detection limit across all examined sites.

To experimentally evaluate these predictions, predicted off-target sites, each containing up to three mismatches (Table S2), were examined by Sanger sequencing followed by ICE analysis to deconvolute sequencing traces and quantify indel frequencies. This analysis showed minimal off-target editing above the ICE detection limit across all examined sites (Fig. 4C; Fig. S3, S4).

Genome size–associated trends in predicted off-target proportions across species, including chicken

Chickens possess a relatively small genome compared with most mammals, with a genome size of approximately 1.1 Gb, reflecting evolutionary genome compaction in avian species (Zhang et al., 2014). This suggests that a smaller genomic search space may influence the off-target landscape of CRISPR nucleases. To examine whether Cas12f exhibits reduced off-target propensity in the chicken genome, we performed a cross-species in silico analysis while controlling for guide sequence variability.

Ten evolutionarily conserved genes—ACTB, HOXA4, HOXA5, HOXA7, HOXA13, HOXD8, HOXD13, NLGN1, PROX1, and SOX2—were selected. For each gene, one Cas12f and one Cas9 target site that were 100% conserved across five species (Gallus gallus [1.1 Gb], Sus scrofa [2.5 Gb], Mus musculus [2.7 Gb], Rattus norvegicus [2.9 Gb], Homo sapiens [3.1 Gb]) were identified (Table S3). Off-target prediction was then performed across all genomes for each sequence, thereby isolating the contribution of genomic background while minimizing sequence-dependent bias.

To explore the relationship between genome size and predicted off-target sites, we compared the relative proportions of predicted Cas12f off-target sites (normalized to the total predicted Cas12f and Cas9 sites) across species (Fig. 5A and B). The proportion of predicted Cas12f off-target sites tended to increase with genome size, from Gallus gallus to Homo sapiens. This change remained modest across species.

Fig. 5.

Comparative analysis of predicted off-target sites across species with different genome sizes (A) Stacked bar graph showing the proportion of predicted Cas12f and Cas9 off-target sites for the indicated species: Gallus gallus, Sus scrofa, Mus musculus, Rattus norvegicus, and Homo sapiens. Data represent mean ± SEM. (B) Log₂ fold change of proportion of predicted Cas12f off-target sites relative to Gallus gallus. Each point represents an individual target gene, and horizontal lines represent mean ± SEM. P-values were determined by a two-tailed unpaired Student’s t-test. Statistical significance is indicated as *P < 0.05.

Discussion

In this study, we provided the first experimental evidence that Cas12f can function as a genome editing nuclease in chicken cells. Cas12f has a remarkably small molecular size of approximately 400–700 amino acids (Harrington et al., 2018). This property may contribute to transcriptional and translational efficiency when delivered via plasmids (Fig. 1B-E), consistent with previous reports showing that smaller proteins generally show higher translation efficiency (Fernandes et al., 2017) and improved intracellular folding capacity (Farías-Rico et al., 2018). In addition, the reduced cargo requirement suggests additional benefits for alternative delivery formats, including ribonucleoproteins (RNPs) and messenger RNA (mRNA) delivery.

Importantly, the compact size of Cas12f may provide a practical advantage for in vivo applications using A3V. Although A3V has been reported to enable efficient gene delivery in avian species (Robinson et al., 2022; Matsui et al., 2012), its limited packaging capacity (∼4.7 kb) restricts the use of larger nucleases such as Cas9, often requiring dual-vector strategies that contribute to variability in co-transduction efficiency (Terada et al., 2025). In this context, Cas12f could potentially be packaged with its gRNA in a single A3V vector, which may simplify vector design and delivery. Moreover, the application of AAV-based systems has been shown to exhibit tissue tropism (Terada et al., 2025), suggesting the possibility of tissue-restricted somatic genome editing in vivo. However, these potential advantages remain to be experimentally validated in avian in vivo systems. Therefore, further studies will be required to determine whether Cas12f-based platforms can be successfully applied to cell-type-specific functional studies and targeted genome modification in vivo.

Target genes were comprehensively selected based on their involvement in the major biological axes encompassing vascular function, immune responses, inflammatory regulation, and metabolic homeostasis (Lee et al., 2022; Pérez-Gutiérrez and Ferrara, 2023; Li and Shang, 2024; Weissglas-Volkov et al., 2011; Liu et al., 2024) (Fig. 2B). The relationship between Cas12f expression level and editing efficiency was found to be locus-dependent. Although Cas12f exhibited lower activity than Cas9 at the RAG1 locus, variable editing efficiencies were achieved at most loci examined (Fig. 2D and E). While previous studies have reported relatively low Cas12f activity in vitro (Karvelis et al., 2020), the present study further extends these results by demonstrating locus-dependent yet meaningful editing activity in chicken cells. This observation is consistent with reports from mammalian systems, in which Cas12f activity varies markedly depending on local sequence context and target architecture (Park et al., 2025; Xin et al., 2022). Importantly, these results suggest that high Cas12f expression alone does not uniformly guarantee high editing efficiency. Although Cas12f exhibited relatively high transfection efficiency and expression levels, genome editing outcomes did not increase proportionally across all loci. This difference may reflect the influence of multiple factors, including PAM compatibility, chromatin accessibility, target-dependent variability, and intrinsic nuclease activity, demonstrating the importance of target selection and gRNA optimization (Jensen et al., 2017).

A deletion-dominant indel profile was consistently found across all tested loci, whereas insertion events were relatively rare (Fig. 3B and C). This pattern suggests that Cas12f introduces cleavage outcomes distinct from those of Cas9. Such a bias toward larger deletions may be advantageous for gene knockout applications, where effective disruption of coding sequences is required. This indel pattern may be explained by the fact that Cas12f introduces DSBs at sites distal to the PAM sequence. As a result, the PAM is likely to remain intact after initial NHEJ repair and may enable repeated cycles of cleavage and repair, thereby promoting cumulative mutagenesis over time (Kim et al., 2022). In addition, the asymmetric cleavage pattern of Cas12f can generate extended 5′ overhangs, which may facilitate end resection during classical NHEJ and further bias repair outcomes toward deletion formation (Xiao et al., 2021). These cleavage characteristics may also reduce the frequency of single-base insertions, vector fragment integrations, and chromosomal translocations, which have been associated with Cas9-mediated blunt-end cleavage (Xin et al., 2022). However, whether Cas12f can support homology-directed repair (HDR) remains unclear, and its suitability for precise knock-in strategies will require further investigation.

Another characteristic explored in this study was the off-target profile of Cas12f. Owing to its compact architecture, Cas12f has a simplified PAM recognition interface and a reduced DNA contact surface area (Xiao et al., 2021), which may limit mismatch tolerance. Consistent with this notion, ICE analysis detected minimal off-target signals at the tested loci (Fig. 4C). Furthermore, prior studies (Bae et al., 2014; Tsai et al., 2015; Lazzarotto et al., 2020; Piao et al., 2022) have reported a positive correlation between the number of in silico–predicted off-target sites and experimentally observed off-target editing frequencies. In this context, the relatively low number of predicted off-target sites for Cas12f (Fig. 4A and B) may suggest a potentially reduced likelihood of unintended genome modifications.

In addition to locus-specific specificity analyses, our cross-species in silico comparison suggests that the theoretical off-target landscape of Cas12f may vary across genomic backgrounds, including differences in genome assembly size. When identical conserved target sequences were analyzed across species with increasing genome sizes, the proportion of predicted Cas12f off-target sites increased gradually (Fig. 5A and B). These computational observations raise the possibility that off-target burden is not determined exclusively by nuclease-intrinsic properties or guide sequence composition, but may also be shaped by wider genomic context (Tsai et al., 2015). In relatively compact genomes such as those of avian species (Kapusta et al., 2017), the reduced search space for partially matched sequences may theoretically constrain off-target opportunities. However, because this analysis was based solely on computational prediction, it does not demonstrate experimentally validated differences in specificity across species. Experimental verification will therefore be required to determine whether Cas12f performance varies across distinct genomic backgrounds.

This study has several limitations that should be considered when interpreting the results. Direct head-to-head comparisons between Cas12f and Cas9 were inherently constrained by their different PAM requirements, limiting the availability of fully matched target sites. Further, low-frequency off-target events could not be fully assessed, indicating that high-resolution, unbiased genome-wide assays such as CIRCLE-seq (Tsai et al., 2017) or GUIDE-seq (Tsai et al., 2015) will be required. Because the present off-target analysis relied on Sanger sequencing and ICE analysis, low-frequency editing events below approximately 1% could not be reliably detected. In addition, although inter-target variability was observed, the trend direction was maintained across conserved loci. Because this analysis is based solely on computational prediction, it does not establish a causal relationship between genome size and off-target propensity.

Although LMH cells are widely used for avian genome-editing studies because of their reliable transfection compatibility (Vilela et al., 2020; Xu et al., 2023; Han et al., 2026), they are tumor-derived cells and may not fully recapitulate the editing characteristics of normal primary cells. In addition, while transfection efficiency and cytotoxicity were evaluated in PGCs, direct quantification of genome-editing efficiency in PGCs was not performed in the present study. Because avian PGCs are known to exhibit relatively low editing efficiencies without enrichment or selection procedures due to their specific DNA repair characteristics (Lee et al., 2020; Rengaraj et al., 2022), the PGC experiments were intended primarily to assess delivery feasibility and cellular compatibility of the compact Cas12f platform. Therefore, further evaluation in primary avian cells and PGCs will be necessary to more comprehensively assess the biological applicability and specificity of Cas12f-based genome editing.

In summary, Cas12f demonstrated measurable, reproducible genome-editing activity in chicken cells. These properties support its potential application in disease-resistance breeding, productivity-related trait modification, and fundamental biological research. Further optimization, including codon optimization, refinement of gRNA design, and validation throughout diverse genomic loci, is likely to enhance Cas12f performance and expand its application in avian research.

Data availability

All data generated or analyzed during this study are included in this published article and its supplementary information files.

CRediT authorship contribution statement

Yujin Han: Writing – original draft, Visualization, Methodology, Formal analysis, Data curation, Conceptualization. Seung Je Woo: Writing – review & editing, Methodology, Formal analysis. Hee Jung Choi: Writing – review & editing, Methodology, Formal analysis. Jae Yong Han: Writing – review & editing, Supervision, Project administration, Funding acquisition, Conceptualization.

Disclosures

The authors have no conflicts of interest to declare.