Skip to main content
NCBI home page
BMC Biotechnology logoLink to BMC Biotechnology
. 2026 May 9;26:83. doi: 10.1186/s12896-026-01164-8

PAM-flexible SpCas9 variants expand the targeting scope for porcine genome editing and cellular disease modeling

Zhiwei Peng 1, Wenxin Duan 1, Yuhang Fan 1, Qiang Yang 1, Yu Ye 1, Yuyun Xing 1,✉
PMCID: PMC13326552  PMID: 42106677

Abstract

Background

CRISPR-Cas-mediated gene editing has revolutionized life sciences, yet the targeting scope of the widely used SpCas9 is limited by its strict requirement for the NGG protospacer adjacent motif (PAM). To overcome this limitation, PAM-flexible SpCas9 variants have been developed and characterized in multiple species; however, their potential in pigs (an important biomedical model for humans) remains unexplored. Here, we systematically evaluated the editing performance of three PAM-flexible SpCas9 variants (SpRY, SpG, and SpCas9-NG) and their derived base editors in porcine fetal fibroblasts (PFFs).

Results

Profiling across 228 target sites revealed that SpRY exhibits nearly PAM-less activity, with significantly higher editing efficiency at NRN (15.82%, R = A/G) than at NYN PAMs (5.75%, Y = C/T). SpG and SpCas9-NG preferentially targeted NGN PAMs, achieving mean efficiencies of 14.81% and 16.33%, respectively. PAM‑flexible cytosine base editors (CBEs) mediated efficient C:G‑to‑T:A conversion, with mean efficiencies of 12.01% for SpRY‑BE4max (NNN PAMs), 15.43% for SpG‑BE4max (NGN PAMs), and 18.39% for SpCas9‑NG‑BE4max (NGN PAMs). Similarly, PAM‑flexible adenine base editors (ABEs) mediated efficient A:T‑to‑G:C conversion, with mean efficiencies of 15.66% for SpRY‑ABE8e (NNN PAMs), 24.16% for SpG‑ABE8e (NGN PAMs), and 20.50% for SpCas9‑NG‑ABE8e (NGN PAMs). By exploiting this expanded targeting scope, we successfully introduced 16 pathogenic single‑nucleotide variants (SNVs) at NRN PAM sites in the porcine genome, with editing efficiencies reaching up to 40.68% for CBEs and 61.76% for ABEs.

Conclusions

PAM-flexible SpCas9 variants and their derived base editors greatly expand the targeting scope for porcine genome engineering, thereby substantially broadening the applicability potential of CRISPR-Cas-mediated genome editing tools in porcine genetic improvement and disease model generation.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12896-026-01164-8.

Keywords: CRISPR-Cas9, Base editing, PAM-flexible, SpRY, SpG, SpCas9-NG, Porcine fetal fibroblasts

Background

CRISPR-Cas-mediated gene editing has emerged as a revolutionary technology that has reshaped life sciences research [1–3]. Among the diverse CRISPR-Cas systems identified to date, the type II system derived from Streptococcus pyogenes (SpCas9) has been extensively studied and widely applied in genome engineering [4–7]. The SpCas9 nuclease induces site-specific DNA double-strand breaks (DSBs), which activate multiple endogenous DNA repair pathways, including non-homologous end joining (NHEJ), microhomology-mediated end joining (MMEJ), and homologous recombination (HR). Harnessing these pathways enables a wide range of genomic modifications, including gene deletion, insertion, and sequence replacement [8]. However, DSB-based approaches can lead to undesired translocations and large deletions, as well as complex rearrangements [9–11]. To circumvent the limitations inherent to DSB-based editing, researchers have fused either SpCas9(D10A) nickase to deaminases to form base editing systems (BEs) [12–14], or SpCas9(H840A) nickase to reverse transcriptase to form prime editing systems [15, 16]. BEs primarily include cytosine base editors (CBEs), which convert C:G-to-T:A, and adenine base editors (ABEs), which convert A:T-to-G:C. Together, they can theoretically correct >60% of pathogenic single-nucleotide variants (SNVs) and provide an efficient strategy for modeling disease-associated SNVs [17–19].

A key determinant of CRISPR-SpCas9 activity is its requirement for a short DNA sequence adjacent to the target site, known as the protospacer adjacent motif (PAM) [14, 20, 21]. Wild-type (WT) SpCas9 preferentially recognizes the NGG PAM; however, the limited genomic distribution of NGG PAM sites substantially restricts the targetable scope [22, 23]. To overcome this limitation, directed evolution and structure-guided protein engineering have generated a series of PAM-flexible SpCas9 variants, including the NGN-preferring SpG and SpCas9-NG, and the nearly PAM-less SpRY [24, 25]. These variants have dramatically expanded the editable genomic space and enabled efficient genome editing in human cells and plants [25–27].

Pork is the most commonly consumed meat worldwide. In addition, pigs serve as indispensable large animal models due to their close physiological, anatomical, and metabolic similarities to humans, making them particularly valuable for genetic disease modeling and xenotransplantation [28–30]. Clinical evidence indicates that a large proportion of inherited diseases are caused by SNVs that result in single amino acid changes [31, 32]. However, conventional SpCas9‑based base editors are constrained by the NGG PAM requirement, which restricts targeting to merely 8.27% of the pig genome [33], leaving many candidate pathogenic SNVs or causative mutations affecting economically important traits inaccessible. Thus, PAM-flexible SpCas9 variants represent a promising solution to overcome this bottleneck, with applications in both porcine disease model generation and genetic improvement. However, to date, their performance has not been systematically assessed in pigs.

In this study, we systematically evaluated the editing performance of three PAM-flexible SpCas9 variants (SpRY, SpG, and SpCas9-NG) and their derived CBEs and ABEs in porcine fetal fibroblasts (PFFs), which are primarily used as donor cells for somatic cell nuclear transfer [34]. All SpCas9 variants and their derived CBEs and ABEs exhibited substantial editing efficiencies, albeit with large variation across targeting loci. The nearly PAM-less SpRY variant and its derived CBE and ABE showed remarkably higher editing efficiencies at NRN PAMs than at NYN PAMs. Using PAM-flexible CBEs and ABEs, we successfully introduced 16 pathogenic SNVs (eight per tool) into the porcine genome, achieving average editing efficiencies of 7.50% for CBEs and 28.13% for ABEs, respectively. Together, our findings establish a versatile platform for porcine genome engineering that substantially broadens the targetable genomic space, thereby advancing genetic improvement and disease modeling capabilities.

Methods

Experimental animals

The Large White sows used in this study were obtained from Taihe Aomu Breeding Co., Ltd. (Jiangxi, China). All animals were housed under standard environmental conditions with free access to feed and water. Animals were deeply anesthetized via intramuscular injection of Zoletil™ 50 (5 mg/kg; Virbac). After confirming deep anesthesia (loss of consciousness and absence of pedal withdrawal reflex), pigs were euthanized by intravenous injection of potassium chloride (100 mg/kg) to induce cardiac arrest.

Plasmid construction

DNA oligonucleotides for constructing sgRNA expression vectors were synthesized, annealed, and cloned into the BsaI-digested pGL3-U6-sgRNA-PGK-Puro plasmid (Addgene, #51133). The target sgRNA sequences are listed in Supplementary Table S1. The SpRY, SpG, SpCas9-NG, and PAMmla‑optimized SpCas9 variants, carrying distinct mutations relative to WT SpCas9 (Figs. 1A and S4A) [24, 25], were constructed using pCMV-T7-SpCas9-BPNLS-3×FLAG-P2A-EGFP (Addgene, #139987) as the backbone vector. SpRY-BE4max, SpG-BE4max, and SpCas9-NG-BE4max plasmids were constructed by cloning SpRY, SpG, and SpCas9-NG variants into pCAG-BE4max-P2A-EGFP (Addgene, #140083), whereas SpRY-ABE8e, SpG-ABE8e, and SpCas9-NG-ABE8e plasmids were constructed by cloning SpRY, SpG, and SpCas9-NG variants into pCMV-TadA8e-P2A-Puro (Addgene, #138489). All these PAM-flexible Cas9 or base editor plasmids were synthesized by GenScript (Nanjing, China).

Fig. 1.

Fig. 1

Open in a new tab

Cleavage activity of PAM-flexible SpCas9 variants in PFFs. (A) Schematic diagrams of SpRY, SpG, and SpCas9-NG variants engineered from SpCas9. CMV: cytomegalovirus promoter; P2A: porcine teschovirus-1 2A peptide; EGFP: enhanced green fluorescent protein; PolyA: polyadenylation signal. (B) Mean nuclease activity plots for SpRY at 228 target sites with NRN (R = A/G) or NYN (Y = C/T) PAMs; for SpG and SpCas9-NG at 57 target sites with NGN PAMs. (C) Mean nuclease activity plots for SpRY with NAN, NCN, and NTN PAMs. (D) Mean nuclease activity plots for SpRY, SpG and SpCas9-NG with NGN PAMs, for SpCas9 with NGG PAMs. Data points represent the mean of three biological replicates for each site; horizontal black lines represent the mean of tested sites for each PAM class. Statistical significance was determined by two-tailed Student’s t-test (*p < 0.05; **p < 0.01; ***p < 0.001)

Cell culture and transfection

PFFs were isolated from Large White pig fetuses at day 30 of gestation using 200 U/mL collagenase type IV (Sigma) [10]. Isolated cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Gibco) supplemented with 12% fetal bovine serum (FBS, Excell Bio) and 1% penicillin-streptomycin (Gibco), and maintained at 37 °C in a humidified atmosphere containing 5% CO₂.

Electroporation of PFFs (passage 4) was conducted using the Lonza 4D-Nucleofector™ X Unit with 16-well Nucleocuvette™ Strips, following a customized dual program (CA137 + CA137) [35]. For each reaction, 1 µg of Cas9 plasmid or base editor plasmid and 1 µg of sgRNA plasmid were mixed in 20 µL of Entranster-E buffer (Engreen, Beijing, China) before electroporation into approximately 4×10⁵ cells. Immediately after electroporation, cells were seeded into 6-well plates, and culture medium was replaced at 6 h post-electroporation. At 48 h post-electroporation, the medium was replaced with fresh medium containing 2 µg/mL puromycin. After 48 h of selection, surviving cells were harvested and lysed in buffer (containing 0.45% NP40 and 6 mg/mL Proteinase K), followed by incubation at 56 °C for 90 min and 85 °C for 15 min.

Generation of PAMmla‑optimized SpCas9 variants

PAMmla‑optimized SpCas9 variants were generated using the online web tool (https://pammla.streamlit.app/) via computational directed evolution. Briefly, three pre‑trained models (model1, model2, model3) were employed in parallel as an ensemble. Starting from 1,000 random seed sequences (each containing six amino acid substitutions), iterative optimization was performed to maximize predicted nuclease activity. In each round, 1,000 variants were generated, and the top 20 variants with the highest predicted activity were selected as parents for the next round. The evolution process was continued for two additional rounds after the activity score reached a plateau to ensure convergence.

Amplicon deep sequencing and editing efficiency analysis

Two rounds of PCR with Ex Premier DNA Polymerase (Takara) were performed to generate amplicons for sequencing. The first round of PCR was performed to amplify the target region using site-specific primers (Supplementary Table S1) and the lysate as template. For the second round of PCR, specific inner primers (Supplementary Table S1) incorporating Illumina forward and reverse indices were employed, using the products from the first round of PCR as the template. The products from the second round of PCR were sequenced commercially (Tsingke Biotech) using the NovaSeq 6000 or MiSeq platform, with at least 5,000 reads generated per sample. All sequencing data were analyzed using CRISPResso2 [36] in pooled mode with custom input parameters. The settings were as follows: nucleases, --min_reads_to_use_region 100; CBEs, --min_reads_to_use_region 100 -w 20 --cleavage_offset −10 --base_editor_output --conversion_nuc_from C --conversion_nuc_to T; ABEs, --min_reads_to_use_region 100 -w 20 --cleavage_offset −10 --base_editor_output --conversion_nuc_from A --conversion_nuc_to G.

GUIDE-seq assay for off-target analysis

The GUIDE-seq assay was performed as previously described [37]. Briefly, 1 µg of Cas9 plasmid, 1 µg of sgRNA plasmid, and 200 ng of double-stranded oligodeoxynucleotide (dsODN) were mixed in 20 µL of Entranster-E buffer (Engreen, Beijing, China) before electroporation into approximately 4×10⁵ cells. Electroporation was performed using the Lonza 4D-Nucleofector™ X Unit with 16-well Nucleocuvette™ Strips and a customized CA137 + CA137 program. A control group was electroporated with 200 ng of dsODN only. After electroporation, cells were cultured for 48 h, then the medium was replaced with fresh medium containing 2 µg/mL puromycin, and selection was maintained for 48 h. Genomic DNA was extracted using a Cell/Tissue Genomic DNA Extraction Kit (Generay Biotech, China) and sheared to an average fragment size of 500 bp using a sonicator. The fragmented gDNA was subjected to end repair, adenylation, and adapter ligation. Target sequences were enriched through two rounds of PCR amplification using oligodeoxynucleotide (ODN)-specific primers. The resulting library was quantified with a Qubit fluorometer and sequenced on an MGI 2000 platform. Sequencing data were analyzed using GUIDE-seq software with the Sscrofa11.1 genome (GenBank: GCA_000003025.6) as the reference [38]. Off-target sites identified in the control group were subtracted as background to filter false-positive sites.

Data processing and statistical analysis

The experimental data were statistically analyzed via GraphPad Prism 9.0 software. All quantitative data are presented as the mean ± SD (standard deviation) from three independent biological replicates. Comparisons between two groups were determined using a two‑tailed Student’s t-test. For comparisons among three groups, one‑way analysis of variance (ANOVA) was performed, followed by Tukey’s post‑hoc test for multiple comparisons. Statistical significance was defined as *p < 0.05, **p < 0.01, and ***p < 0.001; ns indicates non-significance (p ≥ 0.05).

Results

Systematic evaluation of the editing performance of PAM‑flexible SpCas9 variants in PFFs

To systematically evaluate the nuclease activity and PAM preference of SpRY in porcine genome editing, we used the SpRY variant which contains 11 amino acid substitutions relative to SpCas9 (Fig. 1A) [25]. To cover all NNN PAMs, we randomly selected 228 target sites spanning 64 distinct PAM sequences across four porcine genes, including B4GALNT2, CMAH, GGTA1, and TP53 (Supplementary Table S1). Indel frequencies were assessed by next-generation sequencing (NGS). The results demonstrated that SpRY is nearly PAM-less, with significantly higher cleavage activity at NRN (R = A/G) PAMs (15.82%) than at NYN (Y = C/T) PAMs (5.75%, p < 0.001; Figs. 1B and S1), which is consistent with previous observations in human cells and plant systems [25–27]. Among 56 NAN PAM sites, the mean editing efficiency was 16.64% (NAA: 10.94%, NAC: 19.46%, NAG: 22.55%, NAT: 13.62%) (Fig. 1B and C). For 57 NGN PAM sites, the mean editing efficiency was 15.02% (NGA: 14.09%, NGC: 17.32%, NGG: 13.25%, NGT: 15.56%) (Fig. 1B and D). In contrast, the mean editing efficiency across 57 NCN PAM sites was 5.46% (NCA: 9.16%, NCC: 4.54%, NCG: 1.89%, NCT: 6.20%), and across 58 NTN PAM sites it was 6.03% (NTA: 3.72%, NTC: 6.21%, NTG: 10.81%, NTT: 3.53%) (Fig. 1B and C).

Meanwhile, we assessed two additional variants (SpG and SpCas9-NG) at the same 57 NGN PAM sites tested with SpRY. The SpG variant harbors six substitutions (D1135L, S1136W, G1218K, E1219Q, R1335Q, and T1337R) [25], whereas SpCas9-NG harbors seven substitutions (L1111R, D1135V, G1218R, E1219F, A1322R, R1335V, and T1337R) [24] (Fig. 1A). Analysis by PAM subtype revealed distinct performance profiles among these three variants. At NGA PAM sites, both SpG (12.30%) and SpCas9-NG (11.74%) exhibited slightly lower editing efficiencies than SpRY (14.09%) (Figs. 1D and S2A). A similar trend was observed at NGC PAM sites, where the mean editing efficiencies of SpG (15.81%) and SpCas9-NG (13.12%) were marginally lower than that of SpRY (17.32%) (Figs. 1D and S2B). In contrast, at canonical NGG PAM sites, both SpG (16.71%) and SpCas9-NG (17.92%) outperformed SpRY (13.25%) (Figs. 1D and S2C). The most notable difference was observed at NGT PAM sites, where SpCas9-NG achieved a markedly higher editing efficiency (22.43%) compared to SpG (14.28%) and SpRY (15.56%) (Figs. 1D and S2D). Notably, WT SpCas9 achieved a mean editing efficiency of 32.61% across the 15 NGG PAM target sites, which was significantly higher than that of SpRY (13.25%; p < 0.01) and SpG (16.71%; p < 0.05), and marginally higher than that of SpCas9-NG (17.92%; p = 0.063) (Figs. 1D and S2C).

Furthermore, we performed unbiased genome-wide off-target detection for these SpCas9 variants using GUIDE-seq at six randomly selected target sites with diverse PAMs. The analysis demonstrated that SpRY induced off-target effects at all six sites: high-frequency off-target events were observed at the four non-NGG PAM sites, whereas only low-frequency events were detected at the two NGG PAM sites (Fig. S3). In contrast, at the three NGN PAM sites tested, SpG and SpCas9-NG exhibited a lower overall off-target propensity than SpRY, with no off-target events detected for either variant at the AGG-1 site (Figs. S3D-F). For comparison, WT SpCas9 exhibited extremely low-frequency off-target effects at the two NGG PAM sites (Figs. S3E-F).

Notably, despite its nearly PAM-less targeting capability, SpRY exhibited relatively low editing efficiencies at certain YYN PAM sites (Fig. S1). Given that the PAMmla model has been reported to facilitate the design of PAM-specific SpCas9 variants [39], we employed it to engineer eight SpCas9 variants tailored for eight distinct YYN PAMs (Fig. S4A). We then assessed the editing efficiencies of these variants across 31 YYN PAM target sites in PFFs. Compared with SpRY, the PAMmla-optimized SpCas9 variants exhibited significantly reduced editing efficiencies at 29 of the 31 target sites (p < 0.05) (Fig. S4B). For example, at the TTT-1 site, SpRY achieved an editing efficiency of 9.16%, whereas the PAMmla-optimized VSFVEM variant exhibited an editing efficiency of only 0.16%, representing a 57.25-fold reduction. These results indicate that PAMmla-based optimization does not enhance SpCas9-mediated genome editing at specific YYN PAM sites in PFFs.

PAM-flexible base editors enable efficient C:G-to-T:A and A:T-to-G:C editing in PFFs

The restrictive PAM requirement of SpCas9 nucleases and the narrow editing window conferred by deaminase positioning pose key limitations to base editing. To address the PAM constraint, we assembled CBEs and ABEs by fusing PAM-flexible SpCas9 variants with deaminases and systematically evaluated their editing profiles in PFFs. For cytosine base editing, we prepared SpRY-BE4max by introducing the corresponding mutations into BE4max (Fig. 2A) and assessed its editing activity at 32 target sites (eight each for NAN, NGN, NCN, and NTN PAMs). The data showed that SpRY-BE4max mediated C:G-to-T:A conversions at all tested sites, albeit with varying efficiencies across different PAM subtypes (Figs. 2B and S5A-B). Consistent with the previous NHEJ-based editing results, SpRY-BE4max exhibited a significantly higher mean editing efficiency at NRN (18.90%) than at NYN PAM sites (5.11%, p < 0.001). Specifically, the mean editing efficiencies were 18.71% at NAN and 19.08% at NGN PAM sites, compared with 6.28% at NCN and 3.95% at NTN PAM sites (Fig. 2C).

Fig. 2.

Fig. 2

Open in a new tab

Editing efficiency of CBEs using PAM-flexible SpCas9 variants in PFFs. (A) Schematic diagrams of SpRY-BE4max, SpG-BE4max, and SpCas9-NG-BE4max. CAG: cytomegalovirus early enhancer/chicken β-Actin hybrid promoter; rAPOBEC1: rat apolipoprotein B mRNA-editing enzyme catalytic polypeptide 1; P2A: porcine teschovirus-1 2A peptide; EGFP: enhanced green fluorescent protein; PolyA: polyadenylation signal. (B) C-to-T editing efficiencies of all cytosines within the protospacer for SpRY-BE4max at 32 target sites (eight sites each for NAN, NCN, NGN, and NTN PAMs), for SpG-BE4max, and SpCas9-NG-BE4max at eight sites with NGN PAMs. Data points represent the percentage of aligned reads with any C‑to‑T conversion within the protospacer, shown as mean ± SD of n = 3 biological replicates, with individual replicate values overlaid. (C–D) Mean C-to-T editing efficiencies (C) and indel frequencies (D) for SpRY-BE4max (NRN, R = A/G; NYN, Y = C/T), SpG-BE4max (NGN), and SpCas9-NG-BE4max (NGN) at target sites from panel B. Data points represent the mean of three biological replicates for each site; horizontal black lines represent the mean of tested sites for each PAM class. (E) Combined scatter-and-line plot showing C-to-T editing efficiencies of SpRY-BE4max at each cytosine position within the protospacer, using the sites from panel B to define the editing window. Data points represent the percentage of aligned reads with C‑to‑T conversion at specific position, shown as the mean of three biological replicates for each site. Black squares indicate the grand mean across all sites at each spacer position, with error bars showing the SD across sites, and a modified Bezier curve is fitted to these values. Statistical significance was determined by two-tailed Student’s t-test (***p < 0.001)

Meanwhile, we employed the same strategy to assemble SpG-BE4max and SpCas9-NG-BE4max (Fig. 2A), and assessed their editing activity at eight NGN PAM sites. Both editors efficiently mediated C:G-to-T:A editing at all tested NGN PAM sites (Figs. 2B and S5B). SpG-BE4max showed a mean editing efficiency of 15.43%, slightly lower than that of SpRY-BE4max (19.08%), whereas SpCas9-NG-BE4max achieved a comparable mean editing efficiency of 18.39% (Fig. 2C). Further analysis revealed that all three base editors exhibited low levels of indels. SpRY-BE4max averaged 1.05% at NRN PAM sites and 0.42% at NYN PAM sites, whereas SpG-BE4max and SpCas9-NG-BE4max showed averages of 0.98% and 0.88%, respectively, at NGN PAM sites (Fig. 2D). Editing window analysis revealed that SpRY-BE4max maintained high activity within positions C4 to C8 (Fig. 2E), confirming that this editor retains an editing window comparable to that of BE4max while effectively addressing PAM restrictions [17]. Likewise, SpG-BE4max and SpCas9-NG-BE4max displayed an identical editing window (Fig. S5C).

For adenine base editing, we prepared SpRY-ABE8e by introducing the corresponding mutations into ABE8e (Fig. 3A), then evaluated its editing activity at 16 target sites (four sites each for NAN, NGN, NCN, and NTN). The data demonstrated that SpRY-ABE8e catalyzed A:T-to-G:C conversions at all tested sites (Figs. 3B and S6A-B). Consistent with our previous results in this study, SpRY-ABE8e exhibited a markedly higher mean editing efficiency at NRN PAM sites (19.37%) than at NYN PAM sites (11.94%). Among these, NGN PAM sites showed the highest mean editing efficiency of 20.11%, followed by NAN (18.63%), NCN (12.39%), and NTN (11.49%) PAM sites (Fig. 3C). Furthermore, SpRY-ABE8e showed very low indel frequencies, averaging 0.36% at NRN and 0.31% at NYN PAM sites (Fig. 3D). Analysis of the editing window showed that SpRY-ABE8e mediated efficient A-to-G editing primarily at positions A3 to A8 (Figs. 3E), consistent with that previously reported for ABE8e [18].

Fig. 3.

Fig. 3

Open in a new tab

Editing efficiency of ABEs using PAM-flexible SpCas9 variants in PFFs. (A) Schematic diagrams of SpRY-ABE8e, SpG-ABE8e, and SpCas9-NG-ABE8e. CMV: cytomegalovirus promoter; TadA-8e: E. coli tRNA-specific adenosine deaminase variant 8e; PolyA: polyadenylation signal. (B) A-to-G base editing efficiencies of all adenines within the protospacer for SpRY-ABE8e at 16 target sites (four sites each for NAN, NCN, NGN, and NTN PAMs), for SpG-ABE8e, and SpCas9-NG-ABE8e at four sites with NGN PAMs. Data points represent the percentage of aligned reads with any A‑to‑G conversion within the protospacer, shown as mean ± SD of n = 3 biological replicates, with individual replicate values overlaid. (C–D) Mean A-to-G base editing efficiencies (C) and indel frequencies (D) for SpRY-ABE8e (NRN, R = A/G; NYN, Y = C/T), SpG-ABE8e (NGN), and SpCas9-NG-ABE8e (NGN) at target sites from panel B. Data points represent the mean of three biological replicates for each site; horizontal black lines represent the mean of tested sites for each PAM class. (E) Combined scatter-and-line plot showing A-to-G editing efficiencies of SpRY-ABE8e at each adenine position within the protospacer, using the sites from panel B to define the editing window. Data points represent the percentage of aligned reads with A‑to‑G conversion at specific position, shown as the mean of three biological replicates for each site. Black squares indicate the grand mean across all sites at each spacer position, with error bars showing the SD across sites, and a modified Bezier curve is fitted to these values. Statistical significance was determined by two-tailed Student’s t-test (p = 0.172)

Similarly, we assessed the editing activity of SpG-ABE8e and SpCas9-NG-ABE8e at four NGN PAM sites. Both editors mediated A:T-to-G:C conversions at all target sites and exhibited editing windows comparable to that of SpRY-ABE8e (Figs. 3B and S6C). SpG-ABE8e exhibited a mean editing efficiency of 24.16%, slightly higher than that of SpRY-ABE8e (20.11%), whereas SpCas9-NG-ABE8e achieved a comparable efficiency of 20.50% (Fig. 3C). Moreover, SpG-ABE8e and SpCas9-NG-ABE8e exhibited low indel frequencies at NGN PAM sites, averaging 0.52% and 0.48%, respectively (Fig. 3D). Collectively, these results demonstrate that SpRY, SpG, and SpCas9-NG significantly expand the targetable range of both CBEs and ABEs while preserving their canonical editing windows.

PAM-flexible base editing facilitates the introduction of pathogenic human SNVs into the porcine genome

Base editing enables precise single-nucleotide substitutions, making it well-suited for modeling disease-associated SNVs. To evaluate the potential of PAM-flexible base editors for targeting human pathogenic SNVs lacking accessible NGG PAMs, we selected six disease-associated genes: CFTR (cystic fibrosis) [40], GJB2 (hereditary hearing loss) [41], LMNA and MYH7 (cardiomyopathy) [42], PAH (phenylketonuria) [43], and RHO (retinitis pigmentosa) [44]. From the ClinVar database [45], we identified 16 human pathogenic SNVs that could be orthologously mapped to the porcine genome, comprising eight C:G-to-T:A substitutions for CBEs and eight A:T-to-G:C substitutions for ABEs (Figs. S7 and S8). Notably, none of these pathogenic loci harbored targetable sgRNAs compatible with the NGG PAM. Based on the higher editing efficiencies observed at NRN PAMs in our previous experiments, we designed sgRNAs targeting NAN or NGN PAMs for each SNV, ensuring they fell within the canonical editing windows (positions C4 to C8 for CBEs and A3 to A8 for ABEs) and assessed their editing efficiency in PFFs.

For the eight C:G-to-T:A SNVs, SpRY-BE4max achieved mean editing efficiencies of 7.28% at NAN PAMs and 8.35% at NGN PAMs, with no significant difference between PAM types but significant site-specific variation (Figs. 4 and S9A). For example, at the PAH-T266I site, the NAN PAM yielded a significantly higher efficiency (9.16%) than NGN PAM (1.89%, p < 0.001) (Fig. 4G), whereas at the LMNA-S143F site, the NAN PAM yielded a significantly lower efficiency (28.91%) than NGN PAM (40.68%, p < 0.001) (Fig. 4D). At NGN PAM sites, SpG-BE4max and SpCas9-NG-BE4max exhibited mean editing efficiencies (6.24% and 8.12%, respectively) comparable to that of SpRY-BE4max, but with site-dependent differences (Fig. S9A). For instance, SpRY-BE4max exhibited the highest efficiency at GJB2-R75W (20.90%) (Fig. 4B), whereas SpG-BE4max did so at PAH-A259V (1.20%) (Fig. 4F), and SpCas9-NG-BE4max at PAH-T266I (9.34%) (Fig. 4G).

Fig. 4.

Fig. 4

Open in a new tab

Efficiency of pathogenic SNV introduction (C-to-T) by CBEs using PAM-flexible SpCas9 variants in PFFs. (A–H) Editing efficiencies of target cytosines for pathogenic SNV introduction: (A) CFTR-R1067C, (B) GJB2-R75W, (C) LMNA-R41C, (D) LMNA-S143F, (E) MYH7-P838L, (F) PAH-A259V, (G) PAH-T266I, (H) RHO-R135W. In each panel, the underlined triplet denotes the codon encoding the target amino acid with the target cytosines in red and the PAM sequence in yellow. Data points represent the percentage of aligned reads with C‑to‑T conversion at target cytosines, shown as mean ± SD of n = 3 biological replicates, with individual replicate values overlaid. Comparisons between two groups were performed using two-tailed Student’s t-test; comparisons among three groups were performed using one‑way ANOVA with Tukey’s post‑hoc test. Statistical significance was defined as follows: *p < 0.05; **p < 0.01; ***p < 0.001; ns, not significant

For the eight A:T-to-G:C SNVs, SpRY-ABE8e mediated mean editing efficiencies of 28.53% (NAN PAMs) and 27.85% (NGN PAMs), again with no overall significant difference but marked site-specific variation (Figs. 5 and S10A). At five of these sites, targeting NAN PAMs resulted in significantly higher efficiencies, whereas the other three sites favored NGN PAMs (Fig. 5). At NGN PAM sites, SpG-ABE8e and SpCas9-NG-ABE8e exhibited mean editing efficiencies of 34.74% and 21.40%, respectively, with no statistically significant difference compared with SpRY-ABE8e (Fig. S10A). Notably, SpRY-ABE8e achieved the highest efficiency at LMNA-Q130R (61.76%) (Fig. 5C), whereas SpG-ABE8e did so at LMNA-E203G (52.02%) (Fig. 5D).

Fig. 5.

Fig. 5

Open in a new tab

Efficiency of pathogenic SNV introduction (A-to-G) by ABEs using PAM-flexible SpCas9 variants in PFFs. (A–H) Editing efficiencies of target adenines for pathogenic SNV introduction: (A) CFTR-E60G, (B) CFTR-Y517C, (C) LMNA-Q103R, (D) LMNA-E203G, (E) MYH7-E328G, (F) MYH7-D1208G, (G) PAH-I269V, (H) PAH-Y277C. In each panel, the underlined triplet denotes the codon encoding the target amino acid with the target adenines in red and the PAM sequence in yellow. Data points represent the percentage of aligned reads with A‑to‑G conversion at target adenines, shown as mean ± SD of n = 3 biological replicates, with individual replicate values overlaid. Comparisons between two groups were performed using two-tailed Student’s t-test; comparisons among three groups were performed using one‑way ANOVA with Tukey’s post‑hoc test. Statistical significance was defined as follows: *p < 0.05; **p < 0.01; ***p < 0.001; ns, not significant

In addition, all PAM-flexible base editors (both CBEs and ABEs) exhibited mean indel frequencies of less than 1% at target sites (Figs. S9B and S10B). Thus, when pathogenic SNVs lack targetable NGG PAMs, these editors enable the successful introduction of desired pathogenic SNVs into PFFs by targeting alternative NRN PAMs.

Discussion

The canonical NGG PAM preference of SpCas9 poses a key constraint for its application in porcine genome editing. PAM-flexible SpCas9 variants largely overcome this limitation, consistent with previous findings in human cells and plant systems [25–27]. However, at canonical NGG PAM sites, WT SpCas9 outperformed all three PAM-flexible SpCas9 variants, exhibiting higher editing efficiencies across all 15 tested sites (Fig. S2C). This finding confirms that WT SpCas9 remains the preferred nuclease whenever a targetable NGG PAM is available [46, 47]. By contrast, at NGN PAMs, the three PAM-flexible SpCas9 variants showed comparable mean efficiencies, albeit with pronounced locus-specific variation (Fig. 1B and S2). Although SpRY achieved higher mean efficiency at NGA and NGC sites, SpCas9-NG at NGT sites, no consistent PAM preference emerged across individual sites (Figs. 1D and S2). Therefore, as a practical guideline, these locus-specific variations require empirical evaluation of all three variants to identify the most suitable editor for each target site with NGA, NGC, or NGT PAMs.

Notably, our data revealed that SpRY exhibits residual PAM selectivity, with extremely low cleavage activity at certain YYN PAM sequences (Fig. S1). To improve cleavage activity at these recalcitrant sites, we used the PAMmla model to engineer SpCas9 variants based on SpRY, which has been shown to enhance editing efficiency at specific PAMs in human cells and mice [39]. Unexpectedly, the eight PAMmla‑optimized variants showed significantly lower activity than SpRY in PFFs. This discrepancy likely stems from the limited training scope of the PAMmla model, which was trained on only six sites within SpRY (Fig. S4A), whereas SpRY itself harbors 11 mutations relative to SpCas9 (Fig. 1A). Thus, optimizing only these six positions may not alleviate the intrinsically low editing efficiency at YYN PAMs. Indeed, previous studies have demonstrated that editing efficiency is constrained not only by PAM preference but also by multiple additional factors, including sgRNA sequence, secondary structure, and chromatin accessibility [48–51]. These additional factors likely explain the suboptimal performance of the PAMmla‑optimized variants.

PAM-flexible base editors will substantially facilitate the generation of porcine models of human genetic diseases, many of which are caused by pathogenic SNVs [31, 32]. Our results demonstrate that for such SNVs lacking accessible NGG PAMs, PAM-flexible CBEs and ABEs achieved substantial editing efficiencies of 0.60%−40.68% and 17.89%−61.76%, respectively (Figs. 4 and 5). These efficiencies are unattainable with conventional SpCas9-based base editors, confirming the utility of PAM-flexible tools for previously inaccessible pathogenic loci. Notably, our results further revealed that the introduction of pathogenic SNVs using base editors frequently resulted in unintended bystander editing, likely due to their broad editing windows (Figs. S9C and S10C). Meanwhile, introducing SNVs at the same position, the editing efficiencies vary significantly when targeting distinct PAM sites with the same SpCas9 variant or the same PAM site with different SpCas9 variants. Therefore, targeting alternative PAMs with different variants enables optimal sgRNA positioning, thereby improving on‑target accuracy while minimizing bystander effects [12, 52]. Beyond Cas9‑dependent bystander editing, base editors can also induce Cas9‑independent DNA off‑target edits, which are caused by deaminases randomly targeting DNA in the genome and are independent of Cas9 binding to mismatched protospacers [53, 54]. Such off‑target effects may compromise the fidelity of porcine genetic engineering, particularly in disease models requiring precise pathogenic SNVs. Consequently, comprehensive risk assessment is warranted in the future generation of live porcine models.

While PAM-flexible SpCas9 variants significantly broaden the targetable genomic space in pigs, the targeting specificity of these tools remains a key consideration for their broader application. GUIDE-seq analysis revealed that all three variants induced detectable off-target effects, with SpRY showing markedly higher off-target activity than SpG and SpCas9-NG (Fig. S3). These findings suggest that increased targeting flexibility may come at the cost of reduced DNA recognition specificity, thereby elevating the risk of off-target effects [55, 56]. This trade-off presents a critical challenge that must be addressed for the broader application of PAM-flexible SpCas9 variants in porcine genome engineering. Previous studies have shown that engineered high-fidelity (HF) SpCas9 variants can greatly reduce off-target activity without compromising on-target efficiency [25, 57, 58]. Meanwhile, AI-driven protein engineering approaches are increasingly being used to optimize nucleases [59–61]. Building on these advances, we anticipate that similar strategies could be applied to further engineer PAM-flexible SpCas9 variants, potentially generating next-generation editors that combine broader targeting scope with improved specificity.

Conclusions

In summary, this study systematically evaluated the cleavage activity, PAM preference, and specificity of three PAM‑flexible SpCas9 variants in PFFs. SpRY exhibited nearly PAM-less activity, whereas SpG and SpCas9‑NG preferentially targeted NGN PAM sequences, together considerably expanding the targeting scope for porcine genome editing. Meanwhile, PAM‑flexible CBEs and ABEs mediated efficient C:G‑to‑T:A and A:T‑to‑G:C conversions at non-NGG PAM sites. Harnessing this capability, we successfully introduced 16 pathogenic SNVs into the porcine genome at sites lacking accessible NGG PAMs. Collectively, these tools substantially broaden the applicability of CRISPR-Cas-mediated genome editing in pigs, not only providing a versatile platform for porcine genetic improvement and disease modeling, but also offering a valuable paradigm for other farm animals.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1 (24.4KB, xlsx)
Supplementary Material 2 (3.6MB, docx)

Acknowledgements

We sincerely thank Prof. Lusheng Huang from Jiangxi Agricultural University for his valuable suggestions on this research.

Abbreviations

ABE

Adenine base editor

CBE

Cytosine base editor

CRISPR

Clustered regularly interspaced short palindromic repeats

PAM

Protospacer adjacent motif

PFF

Porcine fetal fibroblast

SNV

Single-nucleotide variant

SpCas9

Streptococcus pyogenes Cas9

SpCas9-NG

PAM-flexible SpCas9 variant

SpG

PAM-flexible SpCas9 variant

SpRY

PAM-flexible SpCas9 variant

Author contributions

Y.Y.X. designed the experiments and revised the manuscript. Z.W.P. performed most of the experiments, analyzed the data, and drafted the manuscript. W.X.D. and Y.H.F. contributed to cell culture and PCR detection. Q.Y. and Y.Y. contributed to sgRNA plasmid construction. All authors read and approved the final manuscript.

Funding

This study was supported by grants from the National Key Research and Development Program of China (Grant number 2023YFC3404302) and National Natural Science Foundation of China (Grant number 32260825).

Data availability

All data needed to evaluate the conclusions of this study are provided in the main text and/or the supplemental information online. The deep sequencing data have been deposited in the Genome Sequence Archive (GSA: CRA040106) under the BioProject number PRJCA060141, and are publicly accessible at https://ngdc.cncb.ac.cn/gsa/browse/CRA040106.

Declarations

Ethics approval and consent to participate

The study protocol was reviewed and approved by the Ethics Committee of Jiangxi Agricultural University (ethics approval number: JXAULL-2024-01-019). Pigs used in this study were obtained from Taihe Aomu Breeding Co., Ltd. (Jiangxi, China), with permission from the company for use in research. All animal procedures were performed in accordance with the Guidelines for the Care and Use of Experimental Animals issued by the Ministry of Agriculture and Rural Affairs of the People’s Republic of China, and followed the AVMA Guidelines for the Euthanasia of Animals.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  • 1.Pacesa M, Pelea O, Jinek M. Past, present, and future of CRISPR genome editing technologies. Cell. 2024;187(5):1076–100. [DOI] [PubMed] [Google Scholar]
  • 2.Tuncel A, Pan C, Clem JS, Liu D, Qi Y. CRISPR-Cas applications in agriculture and plant research. Nat Rev Mol Cell Biol. 2025;26(6):419–41. [DOI] [PubMed] [Google Scholar]
  • 3.Pickar-Oliver A, Gersbach CA. The next generation of CRISPR-Cas technologies and applications. Nat Rev Mol Cell Biol. 2019;20(8):490–507. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Xue C, Greene EC. DNA repair pathway choices in CRISPR-Cas9-mediated genome editing. Trends Genet. 2021;37(7):639–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Chang HHY, Pannunzio NR, Adachi N, Lieber MR. Non-homologous DNA end joining and alternative pathways to double-strand break repair. Nat Rev Mol Cell Biol. 2017;18(8):495–506. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Sung P, Klein H. Mechanism of homologous recombination: mediators and helicases take on regulatory functions. Nat Rev Mol Cell Biol. 2006;7(10):739–50. [DOI] [PubMed] [Google Scholar]
  • 7.Sfeir A, Symington LS. Microhomology-mediated end joining: a back-up survival mechanism or dedicated pathway? Trends Biochem Sci. 2015;40(11):701–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Ran FA, Hsu PD, Wright J, Agarwala V, Scott DA, Zhang F. Genome engineering using the CRISPR-Cas9 system. Nat Protoc. 2013;8(11):2281–308. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Mou H, Smith JL, Peng L, Yin H, Moore J, Zhang XO, et al. CRISPR/Cas9-mediated genome editing induces exon skipping by alternative splicing or exon deletion. Genome Biol. 2017;18(1):108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Yang Q, Qiao CM, Liu WW, Jiang HY, Jing QQ, Liao YY, et al. BMPR-IB gene disruption causes severe limb deformities in pigs. Zool Res. 2022;43(3):391–403. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Kosicki M, Tomberg K, Bradley A. Repair of double-strand breaks induced by CRISPR-Cas9 leads to large deletions and complex rearrangements. Nat Biotechnol. 2018;36(8):765–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Rees HA, Liu DR. Base editing: precision chemistry on the genome and transcriptome of living cells. Nat Rev Genet. 2018;19(12):770–88. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Komor AC, Kim YB, Packer MS, Zuris JA, Liu DR. Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage. Nature. 2016;533(7603):420–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Jinek M, Chylinski K, Fonfara I, Hauer M, Doudna JA, Charpentier E. A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science. 2012;337(6096):816–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Anzalone AV, Randolph PB, Davis JR, Sousa AA, Koblan LW, Levy JM, et al. Search-and-replace genome editing without double-strand breaks or donor DNA. Nature. 2019;576(7785):149–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Liu W, Duan W, Peng Z, Liao Y, Wang X, Liu R, et al. Highly efficient prime editors for mammalian genome editing based on porcine retrovirus reverse transcriptase. Trends Biotechnol. 2025;43(12):3253–78. [DOI] [PubMed] [Google Scholar]
  • 17.Koblan LW, Doman JL, Wilson C, Levy JM, Tay T, Newby GA, et al. Improving cytidine and adenine base editors by expression optimization and ancestral reconstruction. Nat Biotechnol. 2018;36(9):843–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Richter MF, Zhao KT, Eton E, Lapinaite A, Newby GA, Thuronyi BW, et al. Phage-assisted evolution of an adenine base editor with improved Cas domain compatibility and activity. Nat Biotechnol. 2020;38(7):883–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Porto EM, Komor AC, Slaymaker IM, Yeo GW. Base editing: advances and therapeutic opportunities. Nat Rev Drug Discov. 2020;19(12):839–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Sternberg SH, Redding S, Jinek M, Greene EC, Doudna JA. DNA interrogation by the CRISPR RNA-guided endonuclease Cas9. Nature. 2014;507(7490):62–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Walton RT, Hsu JY, Joung JK, Kleinstiver BP. Scalable characterization of the PAM requirements of CRISPR–Cas enzymes using HT-PAMDA. Nat Protoc. 2021;16(3):1511–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Anders C, Niewoehner O, Duerst A, Jinek M. Structural basis of PAM-dependent target DNA recognition by the Cas9 endonuclease. Nature. 2014;513(7519):569–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Jiang W, Bikard D, Cox D, Zhang F, Marraffini LA. RNA-guided editing of bacterial genomes using CRISPR-Cas systems. Nat Biotechnol. 2013;31(3):233–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Nishimasu H, Shi X, Ishiguro S, Gao L, Hirano S. Engineered CRISPR-Cas9 nuclease with expanded targeting space. Science. 2018;361(6048):1259–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Walton RT, Christie KA, Whittaker MN, Kleinstiver BP. Unconstrained genome targeting with near-PAMless engineered CRISPR-Cas9 variants. Science. 2020;368(6488):290–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Ren Q, Sretenovic S, Liu S, Tang X, Huang L, He Y, et al. PAM-less plant genome editing using a CRISPR–SpRY toolbox. Nat Plants. 2021;7(1):25–33. [DOI] [PubMed] [Google Scholar]
  • 27.Xu Z, Kuang Y, Ren B, Yan D, Yan F, Spetz C, et al. SpRY greatly expands the genome editing scope in rice with highly flexible PAM recognition. Genome Biol. 2021;22(1):6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Lunney JK, Goor AV, Walker KE, Hailstock T, Franklin J, Dai C. Importance of the pig as a human biomedical model. Sci Transl Med. 2021;13(621):eabd5758. [DOI] [PubMed] [Google Scholar]
  • 29.Griffith BP, Goerlich CE, Singh AK, Rothblatt M, Lau CL, Shah A, et al. Genetically modified porcine-to-human cardiac xenotransplantation. N Engl J Med. 2022;387(1):35–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Yan S, Tu Z, Liu Z, Fan N, Yang H, Yang S, et al. A huntingtin knockin pig model recapitulates features of selective neurodegeneration in Huntington’s disease. Cell. 2018;173(4):989–e100213. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Auton A, Abecasis GR, Altshuler DM, Durbin RM, Bentley DR, et al. A global reference for human genetic variation. Nature. 2015;526(7571):68–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Wu Z, Jiang Z, Li T, Xie C, Zhao L, Yang J, et al. Structural variants in the Chinese population and their impact on phenotypes, diseases and population adaptation. Nat Commun. 2021;12(1):6501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Wang Y, Bi D, Qin G, Song R, Yao J, Cao C, et al. Cytosine base editor (hA3A-BE3-NG)-mediated multiple gene editing for pyramid breeding in pigs. Front Genet. 2020;11:592623. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Zhang YT, Yao W, Chai MJ, Liu WJ, Liu Y, Liu ZH, et al. Evaluation of porcine urine-derived cells as nuclei donor for somatic cell nuclear transfer. J Vet Sci. 2022;23(2):e40. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Liu W, Wang X, Liu R, Liao Y, Peng Z, Jiang H, et al. Efficient delivery of a large-size Cas9-EGFP vector in porcine fetal fibroblasts using a Lonza 4D-Nucleofector system. BMC Biotechnol. 2023;23(1):29. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Clement K, Rees H, Canver MC, Gehrke JM, Farouni R, Hsu JY, et al. CRISPResso2 provides accurate and rapid genome editing sequence analysis. Nat Biotechnol. 2019;37(3):224–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Tsai SQ, Zheng Z, Nguyen NT, Liebers M, Topkar VV, Thapar V, et al. GUIDE-seq enables genome-wide profiling of off-target cleavage by CRISPR-Cas nucleases. Nat Biotechnol. 2014;33(2):187–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Tsai SQ, Topkar VV, Joung JK, Aryee MJ. Open-source guideseq software for analysis of GUIDE-seq data. Nat Biotechnol. 2016;34(5):483. [DOI] [PubMed] [Google Scholar]
  • 39.Silverstein RA, Kim N, Kroell A-S, Walton RT, Delano J, Butcher RM, et al. Custom CRISPR–Cas9 PAM variants via scalable engineering and machine learning. Nature. 2025;643(8071):539–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Nicosia L, Harrison PT. CRISPR for cystic fibrosis: advances and insights from a systematic review. Mol Ther. 2025;33(9):4091–112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Toft M, Pang X, Chai Y, Sun L, Chen D, Chen Y, et al. Characterization of spectrum, de novo rate and genotype-phenotype correlation of dominant GJB2 mutations in Chinese Hans. PLoS ONE. 2014;9(6):e100483. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Parikh VN, Day SM, Lakdawala NK, Adler ED, Olivotto I, Seidman CE, et al. Advances in the study and treatment of genetic cardiomyopathies. Cell. 2025;188(4):901–18. [DOI] [PubMed] [Google Scholar]
  • 43.Hillert A, Anikster Y, Belanger-Quintana A, Burlina A, Burton BK, Carducci C, et al. The genetic landscape and epidemiology of phenylketonuria. Am J Hum Genet. 2020;107(2):234–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Hang C, Zhao T, Chen Z, Wang H, Chen L, Zhang T, et al. Geographic and phenotypic variability in RHO-associated retinopathy: evidence from a Chinese cohort and global literature. Am J Ophthalmol. 2025;159:671–3. [DOI] [PubMed] [Google Scholar]
  • 45.Landrum MJ, Lee JM, Riley GR, Jang W, Rubinstein WS, Church DM, et al. ClinVar: public archive of relationships among sequence variation and human phenotype. Nucleic Acids Res. 2014;42(Database issue):D980–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Goldberg GW, Spencer JM, Giganti DO, Camellato BR, Agmon N, Ichikawa DM, et al. Engineered dual selection for directed evolution of SpCas9 PAM specificity. Nat Commun. 2021;12(1):349. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Liu J, Wang Y, Wei J, Wang S, Li M, Huang Z, et al. Enhanced genome editing with a Streptococcus equinus Cas9. Commun Biol. 2025;8(1):196. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Verkuijl SA, Rots MG. The influence of eukaryotic chromatin state on CRISPR-Cas9 editing efficiencies. Curr Opin Biotechnol. 2019;55:68–73. [DOI] [PubMed] [Google Scholar]
  • 49.Briner AE, Donohoue PD, Gomaa AA, Selle K, Slorach EM, Nye CH, et al. Guide RNA functional modules direct Cas9 activity and orthogonality. Mol Cell. 2014;56(2):333–9. [DOI] [PubMed] [Google Scholar]
  • 50.Riesenberg S, Helmbrecht N, Kanis P, Maricic T, Paabo S. Improved gRNA secondary structures allow editing of target sites resistant to CRISPR-Cas9 cleavage. Nat Commun. 2022;13(1):489. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Liu G, Yin K, Zhang Q, Gao C, Qiu JL. Modulating chromatin accessibility by transactivation and targeting proximal dsgRNAs enhances Cas9 editing efficiency in vivo. Genome Biol. 2019;20(1):145. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Molla KA, Yang Y. CRISPR/Cas-mediated base editing: technical considerations and practical applications. Trends Biotechnol. 2019;37(10):1121–42. [DOI] [PubMed] [Google Scholar]
  • 53.Zuo E, Sun Y, Wei W, Yuan T, Ying W, Sun H, et al. Cytosine base editor generates substantial off-target single-nucleotide variants in mouse embryos. Science. 2019;364(6437):289–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Doman JL, Raguram A, Newby GA, Liu DR. Evaluation and minimization of Cas9-independent off-target DNA editing by cytosine base editors. Nat Biotechnol. 2020;38(5):620–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Hibshman GN, Bravo JPK, Hooper MM, Dangerfield TL, Zhang H, Finkelstein IJ, et al. Unraveling the mechanisms of PAMless DNA interrogation by SpRY-Cas9. Nat Commun. 2024;15(1):3663. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Zhang W, Yin J, Zhang-Ding Z, Xin C, Liu M, Wang Y, et al. In-depth assessment of the PAM compatibility and editing activities of Cas9 variants. Nucleic Acids Res. 2021;49(15):8785–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Kleinstiver BP, Pattanayak V, Prew MS, Tsai SQ, Nguyen NT, Zheng Z, et al. High-fidelity CRISPR-Cas9 nucleases with no detectable genome-wide off-target effects. Nature. 2016;529(7587):490–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Slaymaker IM, Gao L, Zetsche B, Scott DA, Yan WX, Zhang F. Rationally engineered Cas9 nucleases with improved specificity. Science. 2016;351(6268):84–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Park JC, Uhm H, Kim YW, Oh YE, Lee JH, Yang J, et al. AI-generated MLH1 small binder improves prime editing efficiency. Cell. 2025;188(21):5831–e4621. [DOI] [PubMed] [Google Scholar]
  • 60.Fei H, Li Y, Liu Y, Wei J, Chen A, Gao C. Advancing protein evolution with inverse folding models integrating structural and evolutionary constraints. Cell. 2025;188(17):4674–e9219. [DOI] [PubMed] [Google Scholar]
  • 61.Xu K, Hua G, Wu M, Zhang H, Liu J, Feng H, et al. AlphaCD: a machine learning model capable of highly accurate characterization for 21,335 cytidine deaminases. Cell Res. 2025;35(10):750–61. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1 (24.4KB, xlsx)
Supplementary Material 2 (3.6MB, docx)

Data Availability Statement

All data needed to evaluate the conclusions of this study are provided in the main text and/or the supplemental information online. The deep sequencing data have been deposited in the Genome Sequence Archive (GSA: CRA040106) under the BioProject number PRJCA060141, and are publicly accessible at https://ngdc.cncb.ac.cn/gsa/browse/CRA040106.

Articles from BMC Biotechnology are provided here courtesy of BMC

RESOURCES