Longxing Cao
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Longxing Cao has a total of 150 co-authors. Here are the top co-authors by number of shared publications:
Chentong Wang
School of Life Sciences, Westlake University, Hangzhou, China
8 shared publications- David Baker
University of Washington, Howard Hughes Medical Institute
3 shared publications Jinyang Liao
State Key Laboratory of Gene Expression, School of Life Sciences, Westlake University, Hangzhou 310024, China
3 shared publicationsQihan Jin
State Key Laboratory of Gene Expression, School of Life Sciences, Westlake University, Hangzhou 310024, China
3 shared publications- Alexandra C. Walls
Department of Biochemistry, University of Washington, Seattle, WA 98195, USA
2 shared publications - An-Ping Zeng
School of Engineering, Westlake University, Hangzhou, China
2 shared publications
Latest Publications
- 21 October 2026
- Published by Springer Nature in Nature
- p. 1-10
Programmable design of functional proteins from natural language
Programming biological function—designing bespoke proteins to perform specified molecular tasks—is a foundational goal of molecular engineering. However, current design models remain fundamentally limited1,2,3: they typically require either natural proteins as starting points for optimization or manual reformulation of functional goals as geometric and sequence-level constraints to guide candidate generation. Here we introduce Pinal, a 16-billion-parameter foundation model that designs candidate proteins from natural-language descriptions of desired function. Trained on 1.7 billion synthetically annotated protein–text pairs, Pinal links functional intent to protein sequence and structure. In computational evaluations, generated candidates combined high predicted foldability with functional-description alignment and sequence diversity, providing a basis for prioritizing experimentally testable designs. We applied Pinal to four distinct design tasks—a fluorescent protein, a polyethylene terephthalate hydrolase, an alcohol dehydrogenase and a metabolic H-protein—and experimentally observed the intended function in each case, including catalytic activity for both designed enzymes. Without iterative experimental optimization, a Pinal-designed H-protein increased product titre by 1.7-fold relative to the corresponding native Escherichia coli H-protein in a multi-enzyme CO2 fixation pathway. These findings support natural language as a high-level interface for candidate generation in protein design, enabling programmable exploration with reduced reliance on manually specified structural or sequence constraints.
- 21 October 2026
- Published by Springer Nature in Nature
- p. 1-10
Programmable design of functional proteins from natural language
Programming biological function—designing bespoke proteins to perform specified molecular tasks—is a foundational goal of molecular engineering. However, current design models remain fundamentally limited1,2,3: they typically require either natural proteins as starting points for optimization or manual reformulation of functional goals as geometric and sequence-level constraints to guide candidate generation. Here we introduce Pinal, a 16-billion-parameter foundation model that designs candidate proteins from natural-language descriptions of desired function. Trained on 1.7 billion synthetically annotated protein–text pairs, Pinal links functional intent to protein sequence and structure. In computational evaluations, generated candidates combined high predicted foldability with functional-description alignment and sequence diversity, providing a basis for prioritizing experimentally testable designs. We applied Pinal to four distinct design tasks—a fluorescent protein, a polyethylene terephthalate hydrolase, an alcohol dehydrogenase and a metabolic H-protein—and experimentally observed the intended function in each case, including catalytic activity for both designed enzymes. Without iterative experimental optimization, a Pinal-designed H-protein increased product titre by 1.7-fold relative to the corresponding native Escherichia coli H-protein in a multi-enzyme CO2 fixation pathway. These findings support natural language as a high-level interface for candidate generation in protein design, enabling programmable exploration with reduced reliance on manually specified structural or sequence constraints.
- Open Access
De novo design of ligand binding and sensing with a physics based generative approach
The de novo design of ligand-binding proteins has tremendous potential to revolutionize biosensor technology, yet converting these designs into functional sensors remains a major challenge due to the need for ligand-induced conformational changes or modulation of protein–protein interactions. Here, we introduce a physics-based generative approach for the de novo creation of proteins that bind small molecules and metal ions. Our method achieves customizable ligand-binding pocket formation in parallel with simulated protein folding, allowing for precise architectural control of the protein–ligand complex and facilitating the development of biosensors based on either ligand-triggered protein reassociation via split-protein reassembly or ligand-induced protein folding. We demonstrate the versatility of our computational method through successful designs targeting five small molecules, including the very small neurotransmitters serotonin and dopamine, and two metal ions. Biophysical characterization confirmed correct ligand binding, and crystal structures closely matched computational models. We demonstrated the biosensor engineering potential of these designs by constructing serotonin and dopamine sensors using a split protein strategy and explored several approaches to enhance sensor activity. Additionally, we developed a zinc sensor through a zinc-induced protein folding mechanism. Overall, our physics-based generative approach provides a robust framework for the de novo design of ligand-binding proteins, opening new avenues for the development of ligand-responsive biosensors.
Open AccessDe novo design of ligand binding and sensing with a physics based generative approach
The de novo design of ligand-binding proteins has tremendous potential to revolutionize biosensor technology, yet converting these designs into functional sensors remains a major challenge due to the need for ligand-induced conformational changes or modulation of protein–protein interactions. Here, we introduce a physics-based generative approach for the de novo creation of proteins that bind small molecules and metal ions. Our method achieves customizable ligand-binding pocket formation in parallel with simulated protein folding, allowing for precise architectural control of the protein–ligand complex and facilitating the development of biosensors based on either ligand-triggered protein reassociation via split-protein reassembly or ligand-induced protein folding. We demonstrate the versatility of our computational method through successful designs targeting five small molecules, including the very small neurotransmitters serotonin and dopamine, and two metal ions. Biophysical characterization confirmed correct ligand binding, and crystal structures closely matched computational models. We demonstrated the biosensor engineering potential of these designs by constructing serotonin and dopamine sensors using a split protein strategy and explored several approaches to enhance sensor activity. Additionally, we developed a zinc sensor through a zinc-induced protein folding mechanism. Overall, our physics-based generative approach provides a robust framework for the de novo design of ligand-binding proteins, opening new avenues for the development of ligand-responsive biosensors.
- Open Access
HalluDesign-NA: Extending HalluDesign for De Novo Nucleic Acid Design
AlphaFold3 has revolutionized the prediction of biomolecular structures and interactions, including atomic-level modeling of nucleic acids. However, the de novo design of structured and functional nucleic acids remains a significant challenge. Here, we extend our HalluDesign framework to nucleic acid design by integrating NA-MPNN for nucleic acid sequence optimization and design. This new framework, HalluDesign-NA, enables iterative sequence-structure co-optimization, facilitating the de novo design of nucleic acids. Computational benchmarking across ssDNA, ssRNA, and aptamer design tasks demonstrates consistent improvements in confidence scores (pLDDT, ipTM), supporting the feasibility of de novo nucleic acid design under various constraints, such as sequence length, symmetry, and protein structure context. We anticipate that HalluDesign-NA will accelerate the de novo design of functional nucleic acids for applications in biotechnology and medicine. The source code for HalluDesign-NA is available at https://github.com/MinchaoFang/HalluDesign_NA .
Open AccessHalluDesign-NA: Extending HalluDesign for De Novo Nucleic Acid Design
AlphaFold3 has revolutionized the prediction of biomolecular structures and interactions, including atomic-level modeling of nucleic acids. However, the de novo design of structured and functional nucleic acids remains a significant challenge. Here, we extend our HalluDesign framework to nucleic acid design by integrating NA-MPNN for nucleic acid sequence optimization and design. This new framework, HalluDesign-NA, enables iterative sequence-structure co-optimization, facilitating the de novo design of nucleic acids. Computational benchmarking across ssDNA, ssRNA, and aptamer design tasks demonstrates consistent improvements in confidence scores (pLDDT, ipTM), supporting the feasibility of de novo nucleic acid design under various constraints, such as sequence length, symmetry, and protein structure context. We anticipate that HalluDesign-NA will accelerate the de novo design of functional nucleic acids for applications in biotechnology and medicine. The source code for HalluDesign-NA is available at https://github.com/MinchaoFang/HalluDesign_NA .
- 25 March 2026
- Published by Elsevier in Current Opinion in Structural Biology
- Vol. 98, 103256
Frontiers and challenges in the design of binders for intrinsically disordered proteins
- 25 March 2026
- Published by Elsevier in Current Opinion in Structural Biology
- Vol. 98, 103256
Frontiers and challenges in the design of binders for intrinsically disordered proteins
- Open Access
- 12 March 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Design of miniprotein inhibitors targeting complement C9 to block membrane attack complex assembly
The abnormal formation of the membrane attack complex (MAC) is intrinsically linked to a range of acute and chronic immune diseases. The insertion of complement C9 into the membrane is the final step and kinetic bottleneck of MAC formation. However, research on blocking the MAC formation of C9 is currently limited. Given its broad, flat, and polar functional interface, complement C9 is a challenging target for rational design. Here, we utilize deep learning-based methods for protein scaffold generation, sequence design, and complex structure prediction to de novo design mini-protein inhibitors that specifically block the membrane insertion of soluble complement C9. The binding affinity of the mini-protein inhibitor is further optimized to 700 pM via partial diffusion. Design accuracy and binding specificity are verified through X-ray crystallography and biochemical studies. An in vivo acute hemolysis inhibition assay reveals that the C9 mini-protein inhibitors remain effective against hemolysis even 8 minutes after complement activation, outperforming the complement C5 inhibitor eculizumab. The de novo designed C9 mini-protein inhibitors can offer an optional therapeutic approach for the prevention and treatment of acute or chronic immune diseases associated with abnormal complement activation.
- 12 March 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Open AccessDesign of miniprotein inhibitors targeting complement C9 to block membrane attack complex assembly
The abnormal formation of the membrane attack complex (MAC) is intrinsically linked to a range of acute and chronic immune diseases. The insertion of complement C9 into the membrane is the final step and kinetic bottleneck of MAC formation. However, research on blocking the MAC formation of C9 is currently limited. Given its broad, flat, and polar functional interface, complement C9 is a challenging target for rational design. Here, we utilize deep learning-based methods for protein scaffold generation, sequence design, and complex structure prediction to de novo design mini-protein inhibitors that specifically block the membrane insertion of soluble complement C9. The binding affinity of the mini-protein inhibitor is further optimized to 700 pM via partial diffusion. Design accuracy and binding specificity are verified through X-ray crystallography and biochemical studies. An in vivo acute hemolysis inhibition assay reveals that the C9 mini-protein inhibitors remain effective against hemolysis even 8 minutes after complement activation, outperforming the complement C5 inhibitor eculizumab. The de novo designed C9 mini-protein inhibitors can offer an optional therapeutic approach for the prevention and treatment of acute or chronic immune diseases associated with abnormal complement activation.
- Open Access
- 1 January 2026
- Published by China Engineering Science Press Co. Ltd. in Vita
De novo design of metalloproteases for targeted amyloid-β cleavage
De novo protein design has not yet achieved the creation of proteases capable of selectively cleaving any desired peptide bond within a native protein with high precision. Here, we report the use of the flow-based generative model ProteusFlow to design metalloproteases by generating enzyme–substrate complexes conditioned on a target peptide sequence and a predefined catalytic motif. Our approach employs a two-step encapsulation strategy to create clamp-like metalloproteases that bind the target peptide in a manner that maximizes substrate sequence recognition. This generative process simultaneously optimizes the selective positioning of the target peptide bond within a catalytically competent configuration and accurately scaffolds the transition state catalytic residues. Using this strategy, we designed zinc metalloproteases targeting three distinct cleavage sites within the aggregation-prone regions of amyloid-β (Aβ), a key pathogenic factor in Alzheimer’s disease. Experimental validation confirmed that the designed metalloproteases possess substrate cleavage activity and are capable of degrading Aβ peptides into smaller fragments. Cryo-EM structures of three designed enzymes in complex with the Aβ peptide, each targeting a distinct cleavage site, matched closely the computational design models. Structure-guided optimization further enhanced catalytic activity and substrate affinity. Together, these results demonstrate the potential of sequence-guided generative approaches for achieving programmable, sequence-selective proteolysis and lay the groundwork for future applications in basic research and therapeutic development.
- 1 January 2026
- Published by China Engineering Science Press Co. Ltd. in Vita
Open AccessDe novo design of metalloproteases for targeted amyloid-β cleavage
De novo protein design has not yet achieved the creation of proteases capable of selectively cleaving any desired peptide bond within a native protein with high precision. Here, we report the use of the flow-based generative model ProteusFlow to design metalloproteases by generating enzyme–substrate complexes conditioned on a target peptide sequence and a predefined catalytic motif. Our approach employs a two-step encapsulation strategy to create clamp-like metalloproteases that bind the target peptide in a manner that maximizes substrate sequence recognition. This generative process simultaneously optimizes the selective positioning of the target peptide bond within a catalytically competent configuration and accurately scaffolds the transition state catalytic residues. Using this strategy, we designed zinc metalloproteases targeting three distinct cleavage sites within the aggregation-prone regions of amyloid-β (Aβ), a key pathogenic factor in Alzheimer’s disease. Experimental validation confirmed that the designed metalloproteases possess substrate cleavage activity and are capable of degrading Aβ peptides into smaller fragments. Cryo-EM structures of three designed enzymes in complex with the Aβ peptide, each targeting a distinct cleavage site, matched closely the computational design models. Structure-guided optimization further enhanced catalytic activity and substrate affinity. Together, these results demonstrate the potential of sequence-guided generative approaches for achieving programmable, sequence-selective proteolysis and lay the groundwork for future applications in basic research and therapeutic development.
- Top Cited Papers
- 1 January 2026
- Published by American Association for the Advancement of Science (AAAS) in Science
- Vol. 391 (6780) , eady6017
De novo design of small molecule–regulated protein oligomers
Small molecule–regulated protein oligomerization provides a powerful mechanism for manipulating biological processes by controlling protein proximity with high temporal precision. However, such systems only rarely exist in nature and remain a substantial challenge for de novo design. In this work, we describe a computational method for designing protein homooligomers whose assembly is regulated by small-molecule ligands with matching symmetry. We designed protein homotrimers regulated by the Food and Drug Administration (FDA)–approved drug amantadine and further designed amantadine-responsive heterodimers and heterotrimers. Biophysical characterization confirmed their amantadine-dependent assembly, and their crystal structures closely matched the design models. We demonstrated their broad applicability in controlling protein localization, membraneless condensate formation, and gene expression. Our approach opens new avenues for designing small molecule–responsive proteins and expands the chemogenetic toolkit for manipulating complex biological processes.
- 1 January 2026
- Published by American Association for the Advancement of Science (AAAS) in Science
- Vol. 391 (6780) , eady6017
Top Cited PapersDe novo design of small molecule–regulated protein oligomers
Small molecule–regulated protein oligomerization provides a powerful mechanism for manipulating biological processes by controlling protein proximity with high temporal precision. However, such systems only rarely exist in nature and remain a substantial challenge for de novo design. In this work, we describe a computational method for designing protein homooligomers whose assembly is regulated by small-molecule ligands with matching symmetry. We designed protein homotrimers regulated by the Food and Drug Administration (FDA)–approved drug amantadine and further designed amantadine-responsive heterodimers and heterotrimers. Biophysical characterization confirmed their amantadine-dependent assembly, and their crystal structures closely matched the design models. We demonstrated their broad applicability in controlling protein localization, membraneless condensate formation, and gene expression. Our approach opens new avenues for designing small molecule–responsive proteins and expands the chemogenetic toolkit for manipulating complex biological processes.
- 28 August 2025
- Published by Springer Nature in Nature Chemistry
- Vol. 17 (12) , 1910-1919
De novo design of light-responsive protein–protein interactions enables reversible formation of protein assemblies
Light-responsive proteins play an essential role in all domains of life by sensing and responding to environmental light signals. However, the de novo design of light-responsive proteins with precisely defined structures and reversible responsive behaviours is an unmet challenge. Here we describe a computational approach to design protein–protein interactions regulated by non-canonical amino acids, focusing on the light-responsive phenylalanine-4′-azobenzene (AzoF). Using this approach, we designed light-responsive cyclic homo-oligomers and heterodimers, which only assemble in AzoF’s trans configuration and disassemble when AzoF photoisomerizes to the cis configuration. Biophysical characterization confirms the light-responsive assembly and disassembly of these complexes, and the crystal structures match the design models with atomic accuracy. We demonstrate the applicability of these light-responsive proteins in constructing light-responsive hydrogels and engineering synthetic ligand receptors to optocontrol cell signalling in mammalian cells. Our approach opens avenues for designing environmentally responsive protein structures and broadens the toolkit for optogenetics and optochemistry.
- 28 August 2025
- Published by Springer Nature in Nature Chemistry
- Vol. 17 (12) , 1910-1919
De novo design of light-responsive protein–protein interactions enables reversible formation of protein assemblies
Light-responsive proteins play an essential role in all domains of life by sensing and responding to environmental light signals. However, the de novo design of light-responsive proteins with precisely defined structures and reversible responsive behaviours is an unmet challenge. Here we describe a computational approach to design protein–protein interactions regulated by non-canonical amino acids, focusing on the light-responsive phenylalanine-4′-azobenzene (AzoF). Using this approach, we designed light-responsive cyclic homo-oligomers and heterodimers, which only assemble in AzoF’s trans configuration and disassemble when AzoF photoisomerizes to the cis configuration. Biophysical characterization confirms the light-responsive assembly and disassembly of these complexes, and the crystal structures match the design models with atomic accuracy. We demonstrate the applicability of these light-responsive proteins in constructing light-responsive hydrogels and engineering synthetic ligand receptors to optocontrol cell signalling in mammalian cells. Our approach opens avenues for designing environmentally responsive protein structures and broadens the toolkit for optogenetics and optochemistry.
- Open Access
- 11 December 2024
- Published by Springer Nature in Nature Biomedical Engineering
- Vol. 8 (12) , 1744
Publisher Correction: Targeting overexpressed antigens in glioblastoma via CAR T cells with computationally designed high-affinity protein binders
- 11 December 2024
- Published by Springer Nature in Nature Biomedical Engineering
- Vol. 8 (12) , 1744
Open AccessPublisher Correction: Targeting overexpressed antigens in glioblastoma via CAR T cells with computationally designed high-affinity protein binders
- 17 October 2024
- Published by Springer Nature in Nature Biomedical Engineering
- Vol. 8 (12) , 1634-1650
Targeting overexpressed antigens in glioblastoma via CAR T cells with computationally designed high-affinity protein binders
Chimeric antigen receptor (CAR) T cells targeting receptors on tumour cells have had limited success in patients with glioblastoma. Here we report the development and therapeutic performance of CAR constructs leveraging protein binders computationally designed de novo to have high affinity for the epidermal growth factor receptor (EGFR) or the tumour-associated antigen CD276, which are overexpressed in glioblastoma. With respect to T cells with a CAR using an antibody-derived single-chain variable fragment as antigen-binding domain, the designed binders on CAR T cells promoted the proliferation of the cells, the secretion of cytotoxic cytokines and their resistance to cell exhaustion, and improved antitumour performance in vitro and in vivo. Moreover, CARs with the binders exhibited higher surface expression and greater resistance to degradation, as indicated by bulk and single-cell transcriptional profiling of the cells. The de novo design of binding domains for specific tumour antigens may potentiate the antitumour efficacy of CAR T cell therapies for other solid cancers.
- 17 October 2024
- Published by Springer Nature in Nature Biomedical Engineering
- Vol. 8 (12) , 1634-1650
Targeting overexpressed antigens in glioblastoma via CAR T cells with computationally designed high-affinity protein binders
Chimeric antigen receptor (CAR) T cells targeting receptors on tumour cells have had limited success in patients with glioblastoma. Here we report the development and therapeutic performance of CAR constructs leveraging protein binders computationally designed de novo to have high affinity for the epidermal growth factor receptor (EGFR) or the tumour-associated antigen CD276, which are overexpressed in glioblastoma. With respect to T cells with a CAR using an antibody-derived single-chain variable fragment as antigen-binding domain, the designed binders on CAR T cells promoted the proliferation of the cells, the secretion of cytotoxic cytokines and their resistance to cell exhaustion, and improved antitumour performance in vitro and in vivo. Moreover, CARs with the binders exhibited higher surface expression and greater resistance to degradation, as indicated by bulk and single-cell transcriptional profiling of the cells. The de novo design of binding domains for specific tumour antigens may potentiate the antitumour efficacy of CAR T cell therapies for other solid cancers.