What problem does this paper attempt to address?

The main problem that this paper attempts to solve is to screen out deaminases that can be used for gene - editing tool proteins by using AlphaFold2 to predict protein structures. Specifically, the research aims to discover new single - stranded DNA (ssDNA) deaminases, which can be used as the basis for cytosine base editors (CBEs) to achieve efficient gene editing. In addition, the research also focuses on how to design small CBEs suitable for single adeno - associated virus (AAV) packaging through artificial intelligence - assisted protein engineering methods, in order to improve their editing efficiency in plant and mammalian cells and reduce off - target effects. ### Main Objectives: 1. **Discover New Deaminases**: Predict the structure by AlphaFold2 and perform cluster analysis to find new deaminases with ssDNA deamination activity. 2. **Optimize the Design of CBEs**: Use AlphaFold2 to assist in the design of miniaturized CBEs so that they can be effectively packaged by a single AAV vector, thereby improving their feasibility and efficiency in practical applications. 3. **Verify the Editing Effects of CBEs**: Evaluate the editing efficiency and specificity of newly discovered deaminases in plant and mammalian cells through the fluorescence reporter system and high - throughput sequencing technology. ### Background: - **Functions of Deaminases**: Deaminases play a crucial role in host immunity, mutation, and DNA and RNA metabolism, and have been used for DNA and RNA base editing in recent years. - **Development of CBEs**: CBEs are mainly based on the CRISPR - SpCas9 system and are used to convert the target cytosine (C) into thymine (T), but the currently available ssDNA and dsDNA - targeting deaminases are limited. - **Importance of Structure Prediction**: Traditional protein sequences cannot fully explain their functions, and the three - dimensional structure is crucial for understanding protein functions. ### Methods: 1. **Structure Prediction and Clustering**: Select more than 200 protein sequences containing deaminase domains from the InterPro database, use AlphaFold2 to predict their structures, generate a similarity matrix through multi - structure alignment, and finally cluster these proteins into 20 different structural branches. 2. **Function Verification**: Integrate the candidate deaminase domains into the CRISPR - CBE system and evaluate their editing activity in rice protoplasts through the fluorescence reporter system. 3. **Miniaturization Design**: Use AlphaFold2 to assist in the design of miniaturized CBEs so that they are suitable for packaging by a single AAV vector, and verify their editing efficiency and specificity in rice and HEK293T cells. ### Results: - **Discovery of New Deaminases**: The research has discovered multiple new deaminases with ssDNA deamination activity, especially Sdd7 in the SCP1.201 branch, which performs particularly well. - **Efficient Editing**: Sdd7 shows high - efficiency editing ability in rice protoplasts and HEK293T cells, and has a low off - target effect. - **Successful Miniaturization**: Through AlphaFold2 - assisted design, small CBEs suitable for single AAV packaging have been successfully created, further enhancing their potential in practical applications. ### Significance: - **Expand Gene - Editing Tools**: By discovering new deaminases, the library of tool proteins available for gene editing has been expanded. - **Improve Editing Efficiency**: The designed small CBEs show high efficiency and high specificity in plant and mammalian cells, providing new possibilities for future gene therapy and agricultural breeding. - **Application of AI in Protein Engineering**: It demonstrates the great potential of AI in protein structure prediction and design, providing a new direction for future research.

Discovering deaminases using AlphaFold2: a strategy to search for tool proteins for gene editing

Discovery of deaminase functions by structure-based protein clustering

Structure-guided discovery of highly efficient cytidine deaminases with sequence-context independence

AlphaFold2 and Deep Learning for Elucidating Enzyme Conformational Flexibility and Its Application for Design

AlphaFold2 in biomedical research: facilitating the development of diagnostic strategies for disease

Active Finetuning Protein Language Model: A Budget-Friendly Method for Directed Evolution

Benchmarking Refined and Unrefined AlphaFold2 Structures for Hit Discovery

Empowering AlphaFold2 for protein conformation selective drug discovery with AlphaFold2-RAVE

Genomics Refined: AI-powered Perspectives on Structural Analysis.

Designing novel protein structures using sequence generator and AlphaFold2

AlphaDesign: A de novo protein design framework based on AlphaFold

Guiding discovery of protein sequence-structure-function modeling

Cracking AlphaFold2: Leveraging the power of artificial intelligence in undergraduate biochemistry curriculums

AlphaFold2 for protein structure prediction: Best practices and critical analyses

Discovery of a Potentially New Subfamily of ELFV Dehydrogenases Effective for L-Arginine Deamination by Enzyme Mining

Active Learning-Assisted Directed Evolution

AlphaFold2 structures guide prospective ligand discovery

AF2-Mutation: Adversarial Sequence Mutations against AlphaFold2 on Protein Tertiary Structure Prediction

Interpretable Atomistic Prediction and Functional Analysis of Conformational Ensembles and Allosteric States in Protein Kinases Using AlphaFold2 Adaptation with Randomized Sequence Scanning and Local Frustration Profiling