ArticlePublished: 18 March 2026Rong Zheng1,2,Zhike Lu ORCID: orcid.org/0000-0002-9238-74891,2,3,Rongwei Wei3,Young-Cheul Shin ORCID: orcid.org/0000-0002-5989-50784,5,Jiang Du1,2,Qingfeng Zhang3,Jianbo Li2,Xiaoqi Wang3,Yi Wei3,Botao Liu4,5,Yang Chen4,5,Lihong Ding3,Heng Zhang3,Hui Chen3,Jing Huang ORCID: orcid.org/0000-0001-9639-29071,2 &…Lijia Ma ORCID: orcid.org/0000-0001-8592-81391,2,3 Nature Structural & Molecular Biology (2026)Cite this articleSubjectsBiological techniquesCryoelectron microscopyDNA metabolismMolecular biologyNucleasesAbstractEngineering CRISPR enzymes for high fidelity often impairs cleavage activity. Meanwhile, a mechanistic understanding of why high-fidelity mutations reduce Cas9’s cleavage activity remains unclear, presenting a challenge in balancing nuclease specificity and efficiency for clinical applications. In this study, we show that extending the spacer region to 21 or 22 nucleotides restores the impaired cleavage activity of SuperFi-Cas9, a high-fidelity Cas9 variant with 7 mutations in the RuvC domain at the protospacer adjacent motif (PAM)-distal region. Cryo-electron microscopy structures and mutational analyses reveal that the negatively charged mutations in a protruding loop of the RuvC domain create repulsive forces that destabilize the nuclease–single guide (sg)RNA–DNA complex. Spacer extension enhances interactions in the PAM-distal region, effectively restoring cleavage activity and balancing editing efficiency with specificity. In addition, we develop a deep learning model, AIdit-SuperFi, to predict optimal sgRNA length for high-fidelity genome editing. Our findings introduce a straightforward strategy to enhance CRISPR complex stability and provide mechanistic insights into the impaired cleavage activity of engineered high-fidelity Cas9, presenting a pathway toward precise and efficient genome editing and clinical translation of CRISPR technologies.This is a preview of subscription content, access via your institutionAccess optionsAccess Nature and 54 other Nature Portfolio journalsGet Nature+, our best-value online-access subscription27,99 € / 30 dayscancel any timeLearn moreSubscribe to this journalReceive 12 print issues and online access269,00 € per yearonly 22,42 € per issueLearn moreBuy this articlePurchase on SpringerLinkInstant access to the full article PDF.39,95 €Prices may be subject to local taxes which are calculated during checkoutFig. 1: Nucleotide preferences of SuperFi-Cas9.Fig. 2: Spacer extension enhances the cleavage activity of SuperFi-Cas9.Fig. 3: Editing specificity of SuperFi-Cas9.Fig. 4: High-throughput analysis of cleavage activity of SuperFi-Cas9 with different sgRNA lengths.Fig. 5: AIdit-SuperFi predicts the activity and specificity of SuperFi-Cas9.Fig. 6: Cryo-EM structures of SuperFi-Cas9.Data availabilityNGS data are available from Gene Expression Omnibus, accession no. GSE279806. Cryo-EM maps and models are available from Electron Microscopy Data Bank (EMDB) and Protein Data Bank (PDB) with the following accession nos.: EMD-62055 and PDB 9K4D for the 20-nt main map, EMD-65730 and PDB 9W7Q for 20-nt class A, EMD-65732 and PDB 9W7T for 20-nt class B, EMD-65733 and PDB 9W7U for 20-nt class C, EMD-65734 and PDB 9W7V for 20-nt class D, EMD-65827 and PDB 9WAW for 20-nt class E, EMD-62054 and PDB 9K4C for the 22-nt main map, EMD-65771 and PDB 9W9D for 22-nt class A, EMD-65809 and PDB 9WA9 for 22-nt class B, EMD-65810 and PDB 9WAA for 22-nt class C, EMD-62056 and PDB 9K4E for mis22-nt class A, EMD-62057 and PDB 9K4F for mis22-nt class B and EMD-62059 and PDB 9K4H for mis22-nt class C. Plasmid data are available from WeKwikGene, accession numbers 002156 and 002157. Source data are provided with this paper.Code availabilityCode is available at https://github.com/LijiaMALab/SuperFi-AIdit/tree/main.ReferencesWiedenheft, B., Sternberg, S. H. & Doudna, J. A. RNA-guided genetic silencing systems in bacteria and archaea. Nature 482, 331–338 (2012).Article CAS PubMed Google Scholar Jinek, M. et al. A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science 337, 816–821 (2012).Article CAS PubMed PubMed Central Google Scholar Kleinstiver, B. P. et al. High-fidelity CRISPR-Cas9 nucleases with no detectable genome-wide off-target effects. Nature 529, 490–495 (2016).Article CAS PubMed PubMed Central Google Scholar Slaymaker, I. M. et al. Rationally engineered Cas9 nucleases with improved specificity. Science 351, 84–88 (2016).Article CAS PubMed Google Scholar Chen, J. S. et al. Enhanced proofreading governs CRISPR–Cas9 targeting accuracy. Nature 550, 407–410 (2017).Article CAS PubMed PubMed Central Google Scholar Bravo, J. P. K. et al. Structural basis for mismatch surveillance by CRISPR–Cas9. Nature 603, 343–347 (2022).Article CAS PubMed PubMed Central Google Scholar Dagdas, Y. S., Chen, J. S., Sternberg, S. H., Doudna, J. A. & Yildiz, A. A conformational checkpoint between DNA binding and cleavage by CRISPR-Cas9. Sci. Adv. 3, eaao0027 (2017).Article PubMed PubMed Central Google Scholar Fu, Y., Sander, J. D., Reyon, D., Cascio, V. M. & Joung, J. K. Improving CRISPR-Cas nuclease specificity using truncated guide RNAs. Nat. Biotechnol. 32, 279–284 (2014).Article CAS PubMed PubMed Central Google Scholar Kulcsar, P. I. et al. Blackjack mutations improve the on-target activities of increased fidelity variants of SpCas9 with 5ʹG-extended sgRNAs. Nat. Commun. 11, 1223 (2020).Article CAS PubMed PubMed Central Google Scholar Kocak, D. D. et al. Increasing the specificity of CRISPR systems with engineered RNA secondary structures. Nat. Biotechnol. 37, 657–666 (2019).Article CAS PubMed PubMed Central Google Scholar Kulcsar, P. I., Talas, A., Ligeti, Z., Krausz, S. L. & Welker, E. SuperFi-Cas9 exhibits remarkable fidelity but severely reduced activity yet works effectively with ABE8e. Nat. Commun. 13, 6858 (2022).Article CAS PubMed PubMed Central Google Scholar Zhang, H. et al. Deep sampling of gRNA in the human genome and deep-learning-informed prediction of gRNA activities. Cell Discov. 9, 48 (2023).Article CAS PubMed PubMed Central Google Scholar Kim, H. K. et al. High-throughput analysis of the activities of xCas9, SpCas9-NG and SpCas9 at matched and mismatched target sequences in human cells. Nat. Biomed. Eng 4, 111–124 (2020).Article CAS PubMed Google Scholar Frangoul, H. et al. CRISPR-Cas9 gene editing for sickle cell disease and beta-thalassemia. N. Engl. J. Med. 384, 252–260 (2021).Article CAS PubMed Google Scholar Xu, L. et al. CRISPR-edited stem cells in a patient with HIV and acute lymphocytic leukemia. N. Engl. J. Med. 381, 1240–1247 (2019).Article CAS PubMed Google Scholar Musunuru, K. et al. In vivo CRISPR base editing of PCSK9 durably lowers cholesterol in primates. Nature 593, 429–434 (2021).Article CAS PubMed Google Scholar Gillmore, J. D. et al. CRISPR-Cas9 in vivo gene editing for transthyretin amyloidosis. N. Engl. J. Med. 385, 493–502 (2021).Article CAS PubMed Google Scholar Longhurst, H. J. et al. CRISPR-Cas9 in vivo gene editing of KLKB1 for hereditary angioedema. N. Engl. J. Med. 390, 432–441 (2024).Article CAS PubMed Google Scholar NTLA-3001. IntelliaTx https://www.intelliatx.com/pipeline/ (2025).RGX-121. RegenXbio https://www.regenxbio.com/therapeutic-programs/ (2025).Lundberg, S. M. et al. From local explanations to global understanding with explainable AI for trees. Nat. Machine Intell. 2, 56–67 (2020).Article Google Scholar Chen, T. & Guestrin, C. XGBoost: a scalable tree boosting system. In Proc. 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining 785–794 (Association for Computing Machinery, 2016).Ke, G. L. et al. LightGBM: a highly efficient gradient boosting decision tree. In Proc. 31st International Conference on Neural Information Processing Systems 3149–3157 (Curran Associates, 2017).Banerjee, D. et al. Improved nearest-neighbor parameters for the stability of RNA/DNA hybrids under a physiological condition. Nucleic Acids Res. 48, 12042–12054 (2020).Article CAS PubMed PubMed Central Google Scholar Vaswani, A. et al. Attention is all you need. In Advances in Neural Information Processing Systems 30 5998–6008 (NeurIPS, 2017).Zhu, X. et al. Cryo-EM structures reveal coordinated domain motions that govern DNA cleavage byCas9. Nat. Struct. Mol. Biol. 26, 679–685 (2019).Article CAS PubMed PubMed Central Google Scholar Pacesa, M. et al. R-loop formation and conformational activation mechanisms of Cas9. Nature 609, 191–196 (2022).Article CAS PubMed PubMed Central Google Scholar Wang, X. et al. Negatively charged, intrinsically disordered regions can accelerate target search by DNA-binding proteins. Nucleic Acids Res. 51, 4701–4712 (2023).Article CAS PubMed PubMed Central Google Scholar Lorenz, R. et al. ViennaRNA Package 2.0. Algorithms Mol. Biol. 6, 26 (2011).SantaLucia, J., Allawi, H. T. & Seneviratne, A. Improved nearest-neighbor parameters for predicting DNA duplex stability. Biochemistry 35, 3555–3562 (1996).Article CAS PubMed Google Scholar Su, J. L. et al. RoFormer: enhanced transformer with rotary position embedding. Neurocomputing 568, 127063 (2023).Shazeer, N. Glu variants improve transformer. Preprint at https://arXiv.org/abs/2002.05202 (2020).Yang, Y. Z., Zha, K. W., Chen, Y. C., Wang, H. & Katabi, D. Delving into deep imbalanced regression. International Conference on Machine Learning (eds Meila, M. & Zhang, T.) Vol. 139, 11842–11851 (PMLR, 2021).Dalla-Torre, H. et al. Nucleotide transformer: building and evaluating robust foundation models for human genomics. Nat. Methods 22, 287–297 (2025).Zhou, Z. et al. DNABERT-2: efficient foundation model and benchmark for multi-species genomes. In The Twelfth International Conference on Learning Representations (ICLR, 2024).Schiff, Y. et al. Caduceus: bi-directional equivariant long-range DNA sequence modeling. In Proc. 41st International Conference on Machine Learning (eds Salakhutdinov, R. et al.) Vol. 235, 43632–43648 (PMLR, 2024).Breiman, L. Random forests. Mach. Learn. 45, 5–32 (2001).Article Google Scholar Zheng, S. Q. et al. MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopy. Nat. Methods 14, 331–332 (2017).Article CAS PubMed PubMed Central Google Scholar Rohou, A. & Grigorieff, N. CTFFIND4: fast and accurate defocus estimation from electron micrographs. J. Struct. Biol. 192, 216–221 (2015).Article PubMed PubMed Central Google Scholar Ru, H. et al. Molecular mechanism of V(D)J recombination from synaptic RAG1-RAG2 complex structures. Cell 163, 1138–1152 (2015).Article CAS PubMed PubMed Central Google Scholar Kimanius, D., Forsberg, B. O., Scheres, S. H. & Lindahl, E. Accelerated cryo-EM structure determination with parallelisation using GPUs in RELION-2. eLife 5, e18722 (2016).Article PubMed PubMed Central Google Scholar Zivanov, J. New tools for automated high-resolution cryo-EM structure determination in RELION-3. eLife 7, e42166 (2018).Article PubMed PubMed Central Google Scholar Punjani, A., Rubinstein, J. L., Fleet, D. J. & Brubaker, M. A. cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nat. Methods 14, 290–296 (2017).Article CAS PubMed Google Scholar Abramson, J. et al. Accurate structure prediction of biomolecular interactions with AlphaFold 3. Nature 630, 493–500 (2024).Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta Crystallograph. D 66, 486–501 (2010).Article CAS Google Scholar Liebschner, D. et al. Macromolecular structure determination using X-rays, neutrons and electrons: recent developments in Phenix. Acta Crystallograph. D 75, 861–877 (2019).Article CAS Google Scholar Pettersen, E. F. et al. UCSF ChimeraX: structure visualization for researchers, educators, and developers. Protein Sci. 30, 70–82 (2021).Article CAS PubMed Google Scholar Eastman, P. et al. OpenMM 7: rapid development of high performance algorithms for molecular dynamics. PLoS Comput. Biol. 13, e1005659 (2017).Article PubMed PubMed Central Google Scholar Huang, J. et al. CHARMM36m: an improved force field for folded and intrinsically disordered proteins. Nat. Methods 14, 71–73 (2017).Article CAS PubMed Google Scholar Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W. & Klein, M. L. Comparison of simple potential functions for simulating liquid water. J. Chem. Phys. 79, 926–935 (1983).Article CAS Google Scholar Download referencesAcknowledgementsThis research was supported by the Key R&D Program of Zhejiang (grant nos. 2024SSY0033 and 2023C03109) and the National Science Foundation of China (grant no. 32171247 to J.H.). We thank the TORCHES Project team and M. Liao for their invaluable assistance with cryo-EM sample preparation and data collection. We thank the Biomedical Research Core Facilities, Laboratory Animal Resource Center and High-Performance Computer Center of Westlake University for their excellent technical assistance. We thank H. Shen, D. Ma, R. Wang and H. Gao, from Westlake University and Y. Wang from the Institute of Biophysics (Chinese Academy of Sciences), for the inspiring suggestions on analyzing and interpreting the cryo-EM data. We also appreciate the feedback provided by D. Li during the development of this study. Molecular graphics and analyses were performed with UCSF ChimeraX, developed by the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco, with support from the National Institutes of Health (grant no. R01-GM129325) and the Office of Cyber Infrastructure and Computational Biology, National Institute of Allergy and Infectious Diseases.Author informationAuthors and AffiliationsWestlake Laboratory, Hangzhou, ChinaRong Zheng, Zhike Lu, Jiang Du, Jing Huang & Lijia MaSchool of Life Sciences, Westlake University, Hangzhou, ChinaRong Zheng, Zhike Lu, Jiang Du, Jianbo Li, Jing Huang & Lijia MaWestlake Genetech. LTD., Hangzhou, ChinaZhike Lu, Rongwei Wei, Qingfeng Zhang, Xiaoqi Wang, Yi Wei, Lihong Ding, Heng Zhang, Hui Chen & Lijia MaDepartment of Chemical Biology, School of Life Sciences, Southern University of Science and Technology, Shenzhen, ChinaYoung-Cheul Shin, Botao Liu & Yang ChenInstitute for Biological Electron Microscopy, Southern University of Science and Technology, Shenzhen, ChinaYoung-Cheul Shin, Botao Liu & Yang ChenAuthorsRong ZhengView author publicationsSearch author on:PubMed Google ScholarZhike LuView author publicationsSearch author on:PubMed Google ScholarRongwei WeiView author publicationsSearch author on:PubMed Google ScholarYoung-Cheul ShinView author publicationsSearch author on:PubMed Google ScholarJiang DuView author publicationsSearch author on:PubMed Google ScholarQingfeng ZhangView author publicationsSearch author on:PubMed Google ScholarJianbo LiView author publicationsSearch author on:PubMed Google ScholarXiaoqi WangView author publicationsSearch author on:PubMed Google ScholarYi WeiView author publicationsSearch author on:PubMed Google ScholarBotao LiuView author publicationsSearch author on:PubMed Google ScholarYang ChenView author publicationsSearch author on:PubMed Google ScholarLihong DingView author publicationsSearch author on:PubMed Google ScholarHeng ZhangView author publicationsSearch author on:PubMed Google ScholarHui ChenView author publicationsSearch author on:PubMed Google ScholarJing HuangView author publicationsSearch author on:PubMed Google ScholarLijia MaView author publicationsSearch author on:PubMed Google ScholarContributionsL.M. and Z.L. conceived the project and designed the experiments. Z.L. designed the profiling libraries. R.Z. conducted the bioinformatics analysis and built the DL models with help from H.C., Z.L. and Q.Z. R.W. conducted the experiments with help from J.L., X.W., Y.W., L.D. and H.Z. Z.L., Q.Z. and J.D. conducted the MD analysis supervised by J.H. Through SUSTech Open Research Collaboration via High-resolution EM Service (TORCHES), Y.C. purified proteins, B.L. prepared cryo-EM grids and collected cryo-EM data together with Y.-C.S. and the latter performed all steps of EM image processing and interpreted cryo-EM density maps. R.Z. built atomic models. Q.Z. and H.C. built the graphical user interface of the AIdit-SuperFi prediction model. L.M. and R.Z. wrote the paper with input from all co-authors.Corresponding authorsCorrespondence to Jing Huang or Lijia Ma.Ethics declarationsCompeting interestsWestlake Genetech and Westlake University share intellectual property based on the findings of this study. Z.L. and L.M. are co-founders of Westlake Genetech. R.W., Q.Z., X.W., Y.W., L.D., H.Z. and H.C. are full-time employees of Westlake Genetech when participating in this project. The other authors declare no competing interests.Peer reviewPeer review informationNature Structural & Molecular Biology thanks Ailong Ke and David Taylor for their contribution to the peer review of this work. Peer reviewer reports are available. Primary Handling Editor: Dimitris Typas, in collaboration with the Nature Structural & Molecular Biology team.Additional informationPublisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Extended dataExtended Data Fig. 1 Cryo-EM structure of SuperFi-Cas9/22-sgRNA (19 MM).Cryo-EM density maps, overall structural models and sgRNA-TS duplex models of SuperFi-Cas9/22-sgRNA (19 MM) class A (A), class B (B), and class C (C). The sgRNA spacer lengths and the TS protospacer lengths of well-resolved structures are indicated by orange and red labels, respectively.Extended Data Fig. 2 Structural alignment across catalytic states.A. Schematic diagrams of SuperFi-Cas9 or SpCas9 structures across different catalytic states. The states I, II, and IV in this study correspond to the states I, II, and III of a previous study from Subramaniam’s group. And we also added a state III to represent a transitional state between states II and IV. In state I (pre-catalytic), the HNH domain was inactive, and both HNH and REC2 were well resolved. In state II (post-catalytic I), HNH rotated toward the cleavage site of TS and became activated, while REC2 became disordered. In state III (post-catalytic II), HNH dissociates and adopted a disordered conformation, whereas REC2 began to regain structural order but remains at low resolution. In state IV (product), REC2 was again well resolved. B. Distinct conformational structures of SuperFi-Cas9/20-sgRNA (top; SF/20), SuperFi-Cas9/22-sgRNA (middle; SF/22), and SpCas9/20-sgRNA (bottom; SP/20) across different catalytic states. Panel a created in BioRender: Ma, L. https://BioRender.com/jgd7qga (2026); Ma, L. https://BioRender.com/ouicsii (2026); Ma, L. https://BioRender.com/dbr4wyv (2026); Ma, L. https://BioRender.com/redjn6t (2026).Extended Data Fig. 3 Local structures of SuperFi-Cas9.A. Overall structural alignment and local structural alignment of HNH domain between SF/22 and SP/20 in state I. B. Density map of SF/20 and SF/20 in the interaction region between RuvC domain and PAM-distal duplex. C. Repulsive forces between the mutated RuvC domain (7 aspartic acids mutations) of SuperFi-Cas9 and the PAM-distal sgRNA-TS duplex. D. Repulsive forces between the RuvC domain of SpCas9 (6O0Y) and the PAM-distal sgRNA-TS duplex. TS and NTS structures that conflicted with the RuvC domain were hidden.Extended Data Fig. 4 The sgRNA–TS duplex structures across catalytic states.Overall structural models and sgRNA-TS duplex conformations of state I (pre-catalytic), state II (post-catalytic I), state III (post-catalytic II), and state IV (product) for SF/20 (A), SF/22 (B), and SP/20 (C). The orange numbers indicate the number of base pairs of well-resolved sgRNA spacers within each structure.Extended Data Fig. 5 Proposed working models.A. Without the 7D mutations in the RuvC domain of SpCas9, the stability at the PAM-distal region does not rely on the interactions between the sgRNA-TS duplex and residues K929, K948, and R951. B. The 7D mutations in the RuvC domain of SuperFi-Cas9 cause repulsive forces with the sgRNA-TS duplex, which weaken the stability at the PAM-distal region and make the interactions provided by K929, K948, and R951 important. C. The spacer extension extended the sgRNA-TS duplex and enhanced the electrostatic interactions with K929, K947, and R951, which compensated the repulsive forces from the 7D. Panels a–c created in BioRender; Ma, L. https://BioRender.com/lgg5a36 (2025).Supplementary informationSupplementary Information (download PDF )Supplementary Figs. 1–17.Reporting Summary (download PDF )Peer Review File (download PDF )Supplementary Tables (download XLSX )Supplementary Tables 1–10.Source dataSource Data Figs. 1–6 (download XLSX )Statistical source data.Rights and permissionsSpringer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.Reprints and permissionsAbout this article