撰文丨王聪 编辑丨王多鱼 排版丨水成文 2025 年 7 月 24 日，国际顶尖学术期刊 Science 撤回了一篇 15 年前发表的重磅研究论文 ，但论文作者对于这一撤稿决定表达了强烈抗议，他们完全拒绝撤稿 声明中" 论文结论存在重大缺陷 "的说法，论文在世的 11 位作者全部不同意此次撤稿。 2010年底，美国国家航空航天局 （NASA） 天体生物学研究所和美国地质调查局的研究人员在国际顶尖学术期刊 Science 上发表了一篇题 为： A Bacterium That Can Grow by Using Arsenic Instead of Phosphorus 的研究论文。 该论文提出了一个令人震惊的发现，研究团队在 莫诺湖 （ Mono Lake ） 这个环境恶劣 （强碱、高盐） 的水域中发现的一种微生物，其与其他已知生命形式都 不一样，能够利用 砷 来生长。 莫诺湖 生命主要由 碳 、 氢 、 氮 、 氧 、 硫 和 磷 这六种元素构成，它们构成了核酸 （DNA 和 RNA） 、蛋白质和脂质，从而构成了生物体的大部分物质，但从理论 上讲，元素周期表中的其他一些元素也可能发挥同样的作用。 这项研究描 ...

Core Viewpoint - The article highlights a significant achievement in artificial intelligence (AI), where large language models (LLMs) have reached gold medal level in the International Mathematical Olympiad (IMO), showcasing their advanced problem-solving capabilities [4][5][6]. Group 1: AI Achievement - Google DeepMind's large language model successfully solved problems equivalent to those in the IMO, achieving a score that surpasses the gold medal threshold of 35 out of 42 [4][5]. - This marks a substantial leap from the previous year's performance, where the model was only at the silver medal level, indicating a qualitative breakthrough in AI's ability to handle complex mathematical reasoning [5][6]. Group 2: Implications of the Achievement - The success of LLMs in the IMO demonstrates their capability to tackle highly complex tasks that require deep logical thinking and abstract reasoning, beyond mere text generation [7]. - Such AI advancements can serve as powerful tools in education and research, assisting students in learning higher mathematics and aiding researchers in exploring new conjectures and theorems [7]. - Achieving gold medal level in mathematics is a significant milestone on the path to artificial general intelligence (AGI), as it requires a combination of various cognitive abilities [7][8]. Group 3: Broader Impact - The breakthroughs by DeepMind and OpenAI not only elevate AI's status in mathematical reasoning but also suggest vast potential for future applications in scientific exploration and technological development [8].

Core Viewpoint - The article discusses the advancements in cancer immunotherapy through the use of generative AI to design artificial T-cell receptors (TCRs) that can specifically bind to pMHC complexes, overcoming the limitations of natural TCRs and enabling more precise targeting of tumor antigens [4][21][23]. Group 1: Traditional TCR Methods - Traditional methods for utilizing TCR in cancer immunotherapy involve isolating T cells from patients, either through tumor-infiltrating lymphocytes (TILs) or expanding T cells from an initial T cell library, which is technically challenging and labor-intensive [3]. - Natural TCRs often have poor affinity for tumor antigens, making it difficult to achieve effective immunotherapy [3]. Group 2: Generative AI Research - On July 24, 2025, three research papers published in the journal Science demonstrated the use of generative AI to design artificial TCRs that can bind with high specificity to pMHC complexes, thus enhancing the precision of tumor antigen targeting [4][5][21]. - The research teams involved include those from the University of Washington, Technical University of Denmark, and Stanford University [5]. Group 3: Protein Design and Functionality - The David Baker/Liu Bingxu team developed a computational method to enhance the immune system's ability to recognize and destroy cells carrying less detectable disease markers by designing proteins that specifically recognize target pMHC complexes [10]. - The designed proteins were tested against 11 different pMHC targets, including fragments from HIV and cancer-related mutations, with 8 of them successfully activating immune cells [13]. - The study confirmed that the designed proteins only bind to their specific targets, achieving atomic-level precision in construction [14]. Group 4: Rapid and Scalable Design Process - The design process demonstrated high adaptability, allowing the creation of new versions of binding proteins for different tumor and viral peptide targets in less than a week [16]. - This digital approach contrasts with traditional drug development methods, significantly shortening the drug development cycle and reducing complexity, while paving the way for more personalized therapies [16]. Group 5: Future Directions and Company Formation - The lead author, Dr. Liu Bingxu, indicated plans to establish a company to translate these research findings into therapies that can benefit patients [18]. - The research highlights the potential for artificial TCRs to revolutionize diagnostic tools and immunotherapies, particularly for diseases that currently lack effective treatments [22][23].

Core Viewpoint - The article discusses the successful application of CAR-T cell therapy in treating refractory autoimmune encephalitis associated with DAGLA antibodies, highlighting its potential as a novel treatment option for patients unresponsive to standard therapies [4][13]. Group 1: Disease Overview - Autoimmune encephalitis is a severe brain condition where the immune system attacks the brain, leading to symptoms such as memory loss, confusion, behavioral changes, and severe seizures [2]. Group 2: CAR-T Cell Therapy - CAR-T cell therapy targets specific immune cells (B cells) and shows promise in treating autoimmune diseases, including those causing brain inflammation [3]. - The therapy involves a single intravenous infusion of fully human second-generation CAR-T cells targeting CD19 [8]. Group 3: Case Study - A 36-year-old male patient with refractory DAGLA antibody-associated autoimmune encephalitis underwent CAR-T cell therapy after failing multiple treatments, including corticosteroids and plasma exchange [9][10]. - Post-treatment, the patient showed significant clinical improvement, with a reduction in harmful antibodies attacking the brain [10][14]. Group 4: Research Findings - The study published in Cell journal demonstrated that CAR-T cell therapy successfully reversed refractory DAGLA antibody-associated encephalitis, with antibodies disappearing from serum and cerebrospinal fluid after treatment [4][14]. - One year after treatment, the patient continued to show sustained clinical improvement [14].

Core Viewpoint - The emergence of immunotherapy has significantly changed the landscape of cancer treatment, but resistance to immunotherapy remains a major obstacle for its broader clinical application. Recent studies indicate that gut microbiota can enhance the efficacy of immunotherapy by modulating anti-tumor immunity [2]. Group 1: Research Findings - A study published by Professor Xu Ruihua's team from Sun Yat-sen University on July 24, 2025, in the journal Cancer Cell, demonstrates that the gut bacterium Alistipes finegoldii can enhance the efficacy of immunotherapy against solid tumors [3][4]. - The research found that a higher abundance of Alistipes finegoldii is associated with improved responses to immunotherapy, particularly enhancing the efficacy of anti-PD-1 monoclonal antibodies in solid tumor models [8]. - Alistipes finegoldii activates the CXCL16-CXCR6 signaling axis to boost anti-tumor immune responses, with lipoproteins derived from Alistipes finegoldii triggering the TLR2-NF-κB-CXCL16 signaling pathway [7][8]. Group 2: Mechanism of Action - The mechanism involves lipoproteins from Alistipes finegoldii binding to Toll-like receptor 2 (TLR2), activating the NF-κB signaling pathway, which enhances the expression of CXCL16 in CCR7+ conventional dendritic cells [7]. - The released CXCL16 aids in recruiting CXCR6+ CD8+ T cells to the tumor microenvironment (TME), effectively inhibiting tumor growth [7][8]. Group 3: Implications for Treatment - Overall, the findings suggest that combining Alistipes finegoldii with immunotherapy could represent a new strategy for treating solid tumors [10].

Core Viewpoint - The article discusses recent research on the biosynthesis of salicylic acid (SA) in plants, highlighting its importance in plant defense mechanisms and its historical significance in human medicine, particularly in the development of aspirin [2][5][13]. Group 1: Research Findings - The study conducted by teams from Zhejiang University and others revealed a three-step enzymatic reaction module (BEBT-BBH-BSE) that facilitates the conversion of benzoyl-CoA to salicylic acid in rice, demonstrating that this pathway is conserved across various crops [3][8]. - The research identified three specific enzymes involved in this pathway: benzoyl-CoA:benzyl transferase (BEBT), benzyl benzoate hydroxylase (BBH), and salicylic acid benzyl ester hydrolase (BSE), which sequentially convert benzoyl-CoA into salicylic acid [6][8]. - The findings fill a significant knowledge gap in the biosynthesis of key plant defense hormones, providing a foundation for developing disease-resistant crops [8][14]. Group 2: Related Studies - Concurrently, two other studies published in Nature also focused on salicylic acid biosynthesis, confirming the conservation of the PAL synthesis pathway in seed plants and its implications for understanding disease resistance mechanisms in various plant groups [9][10]. - The research from Sichuan University and Zhejiang Normal University further elucidated the complete biosynthesis pathway of salicylic acid from phenylalanine, reinforcing the idea of its evolutionary conservation in most seed plants [10][14].

Core Insights - On July 23, 2025, a total of 26 papers were published in the prestigious journal Nature, with 11 authored by Chinese scholars, highlighting the significant contribution of Chinese researchers to global scientific discourse [2][4][6][7][10][12][15][17][19][21][23]. Group 1: Research Contributions - Research by Chen Lingling from the Chinese Academy of Sciences focused on the spatial distribution and functional organization of pre-rRNA in the nucleolus [2]. - Zhang Yuelin from Sichuan University explored a three-step biosynthesis pathway of salicylic acid from benzoyl-CoA in plants [4]. - A study by Pan Ronghui and Fan Pengxiang from Zhejiang University deciphered the biosynthesis of salicylic acid derived from phenylalanine in plants [6]. - Zhang Kewei and collaborators from Zhejiang Normal University detailed the complete biosynthesis of salicylic acid from phenylalanine in plants [7]. - Research by Yile Dai from Yale University examined humoral determinants of checkpoint immunotherapy [10]. - Zepeng Mu from Harvard Medical School investigated the effects of disease variants in CRISPR-edited single cells [12]. - Winston W. Liu from Duke University/Stanford University Medical School studied how gut sensing of microbial patterns regulates feeding [15]. - Ziliang Kang from MIT developed mechanical underwater adhesive devices for soft substrates [17]. - Arthur Zhao from the Howard Hughes Medical Institute researched how eye structure shapes neuron function in Drosophila motion vision [19]. - Zhiqian Li from UC San Diego and Yuemei Dong from Johns Hopkins University focused on driving a protective allele of the mosquito FREP1 gene to combat malaria [21]. - A review by Gao Caixia from the Chinese Academy of Sciences and Li Guotian from Huazhong Agricultural University integrated biotechnological and AI innovations for crop improvement [23].

Core Viewpoint - The article discusses a groundbreaking non-invasive neuromotor interface developed by Meta's Reality Labs, which allows users to interact with computers through wrist-worn devices that translate muscle signals into computer commands, enhancing human-computer interaction, especially in mobile scenarios [2][3][5]. Group 1: Technology Overview - The research presents a wrist-worn device that enables users to interact with computers through hand gestures, converting muscle-generated electrical signals into computer instructions without the need for personalized calibration or invasive procedures [3][5]. - The device utilizes Bluetooth communication to recognize real-time gestures, facilitating various computer interactions, including virtual navigation and text input at a speed of 20.9 words per minute, compared to an average of 36 words per minute on mobile keyboards [6]. Group 2: Research and Development - The Reality Labs team developed a highly sensitive wristband using training data from thousands of subjects, creating a generic decoding model that accurately translates user inputs without individual calibration, demonstrating performance improvements with increased model size and data [5]. - The research indicates that personalized data can further enhance the performance of the decoding model, suggesting a pathway for creating high-performance biosignal decoders with broad applications [5]. Group 3: Accessibility and Applications - This neuromotor interface offers a wearable communication method for individuals with varying physical abilities, making it suitable for further research into accessibility applications for those with mobility impairments, muscle weakness, amputations, or paralysis [8]. - To promote future research on surface electromyography (sEMG) and its applications, the team has publicly released a database containing over 100 hours of sEMG recordings from 300 subjects across three tasks [9].

Core Viewpoint - The development of the Aeneas model by Google's DeepMind aims to assist historians in analyzing and contextualizing ancient Latin inscriptions, addressing the challenges posed by incomplete texts and the vast amount of data available [3][6]. Group 1: Aeneas Model Development - The Aeneas model is a generative AI tool designed to find relationships between Roman-era Latin inscriptions and other texts, helping to determine the context of these inscriptions and predict missing parts [3][6]. - The model was trained on a merged dataset containing 176,861 inscriptions, with 5% accompanied by images, spanning from the 7th century BC to the 8th century AD [6]. Group 2: Model Functionality and Performance - Aeneas consists of three neural networks, each responsible for different tasks: repairing missing text, predicting text origins, and assessing dating [7]. - Historians found that Aeneas provided useful contextual suggestions 90% of the time, improving confidence in key tasks by 44% [7]. - The model's dating accuracy showed a margin of error of less than 13 years, compared to historians' 31 years, indicating a significant improvement in dating inscriptions [7]. Group 3: Practical Applications - Aeneas demonstrated effectiveness in analyzing altars with Latin inscriptions, identifying connections between inscriptions from the same region without prior geographical knowledge [8]. - The model successfully repaired a damaged inscription from a bronze military certificate issued by a Roman emperor, showcasing its ability to handle large datasets that are challenging for humans [10]. - Aeneas provides more logical answers based on its evidence base, contrasting with other popular AI tools that may generate random guesses [10].

编辑丨王多鱼 排版丨水成文 论文首先阐述了 现代组学技术 是实现育种范式变革的基础。基因组学、代谢组学以及单细胞组学等研究方向的兴起，为我们提供了前所未有的能力去深入解析作 物的遗传信息和生命活动规律，从而揭示更多可用于性状精准改良的新位点 。与此同时， 高通量表型鉴定技术 （ HTP ） 利用无人机、传感器和自动化平台， 实现了对海量作物性状数据的快速、精准评估，从而高效地连接了基因型与表型，为筛选优良变异提供了关键支撑 。 在此基础上，论文详细论述了实现作物改良的强大工具。以 CRISPR 为代表的基因组编辑技术，已能够实现对作物基因组开展高效、精准的定向修饰，其跨尺 度、多维度的基因组设计能力将显著缩短育种周期，快速创造和聚合优良性状 。不仅如此， AI 驱动的蛋白质设计技术 正在兴起，它能够从头创造出自然界中不 存在的、具有特定功能的全新蛋白质 。这为开发新型抗病蛋白、实时监测作物健康的生物传感器或降解环境污染物的特制酶提供了可能，从而赋予作物突破性的 新功能 。 通过整合新技术促进潜在作物改良 作物改良中的基因组编辑技术 2025 年 7 月 24 日， 中国科学院遗传与发育生物学研究所 高彩霞 研 ...