Deep learning has revolutionized biomolecular modeling, enabling the prediction of diverse structures with atomic accuracy. However, leveraging the atomic-level precision of the structure prediction model for de novo design remains challenging. Here, we present HalluDesign, a general all-atom framework for protein optimization and de novo design, which iteratively update protein structure and sequence. HalluDesign harnesses the inherent hallucination capabilities of AlphaFold3-style structure prediction models and enables fine-tune free, forward-pass only sequence-structure co-optimization. Structure conditioning at different noise level in the structure prediction stage allows precise control over the sampling space, facilitating tasks from local and global protein optimization to de novo design. We demonstrate the versatility of this framework by optimizing suboptimal structures, rescuing previously unsuccessful designs, designing new biomolecular interactions and generating new protein structures from scratch. Experimental characterization of a diverse set of proteins—including protein binders for small molecules, a metal ion and proteins; antibody design of phosphorylation-specific peptide; and monomeric proteins—revealed high design success rates and excellent structural accuracy. Together, our comprehensive computational and experimental results highlight the broad utility of this framework. We anticipate that HalluDesign will further unlock the modeling and design potential of AlphaFold3-like models, enabling the robust creation of complex biomolecules for a wide range of biotechnological applications.

Despite the potential of protein therapeutics, the cytosolic delivery of proteins with high efficiency and bioactivity remains a significant challenge owing to exocytosis and lysosomal degradation after endocytosis. Therefore, it is important to develop a safe and efficient strategy to bypass endocytosis. Inspired by the extraordinary capability of filamentous‐actin (F‐actin) to promote cell membrane fusion, a cyanine dye assembly‐containing nanoplatform mimicking the structure of natural F‐actin is developed. The nanoplatform exhibits fast membrane fusion to cell membrane mimics and thus enters live cells through membrane fusion and bypasses endocytosis. Moreover, it is found to efficiently deliver protein cargos into live cells and quickly release them into the cytosol, leading to high protein cargo transfection efficiency and bioactivity. The nanoplatform also results in the superior inhibition of tumor cells when loaded with anti‐tumor proteins. These results demonstrate that this fusogenic nanoplatform can be valuable for cytosolic protein delivery and tumor treatment.

An increase in the number of infections caused by resistant bacteria worldwide necessitates the development of alternatives to antibiotics. Human defensin (HD) 5 is an innate immune peptide with broad-spectrum antibacterial activity, but its complicated structure makes its preparation difficult. Herein, we truncated the HD5 structure by extracting the highly conserved γ-core motif. A structure-activity study showed that this motif was ineffective in killing bacteria in the absence of specific spatial conformation. Notably, after the introduction of two intramolecular disulfide bonds, its antibacterial activity was markedly improved. Glu and Ser residues were then replaced with Arg to create the derivative RC18, which exhibited stronger potency than HD5, particularly against methicillin-resistant S. aureus (MRSA). Mechanistically, RC18 bound to lipid A and lipoteichoic acid at higher affinities than HD5. Furthermore, RC18 was more efficient than HD5 in penetrating the bacterial membranes. Molecular dynamics simulation revealed that five Arg residues, Arg1, Arg7, Arg9, Arg15, and Arg18, mediated most of the polar interactions of RC18 with the phospholipid head groups during membrane penetration. In vivo experiments indicated that RC18 decreased MRSA colonization and dramatically improved the survival of infected mice, thus demonstrating that RC18 is a promising drug candidate to treat MRSA infections.

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 .