“消未起之患、治未病之疾，医之于无事之前”。对于传染性疾病，预防永远优先于治疗。在当前全球新发突发传染病频发的背景下，疫苗研发的时效性与效力提升已成为公共卫生领域的核心课题。为实现疫苗的快速响应和高效免疫，本研究围绕病毒抗原免疫逃逸、mRNA疫苗优化设计、宿主免疫应答协同机制，通过多学科交叉融合，发展了基于人工智能预测、密码子对序列优化、免疫激活剂应用的创新性疫苗增效策略。突变预测模型指导了疫苗的抗原选择，对流行病防控具有重要意义。针对病毒高频变异导致的抗原免疫逃逸、感染力增强等问题，我们开发了一种突变进化预测语义模型，以自然语言处理框架解析病毒序列的“生物语法”规律。以新冠病毒为例，我们构建了S1蛋白序列的“语法框架”，通过引入突变频率并分析突变位点的组合规律，在XBB.1.16、EG.5及JN.1等毒株出现之前即实现了对完整序列或关键突变的预测，并经实验验证其增强的ACE2受体结合能力与免疫逃逸特性。该模型突破了传统系谱树和深度突变扫描的限制，为前瞻性疫苗抗原设计提供了新工具。在选择合适的抗原后，我们希望对疫苗抗原序列做进一步的优化。mRNA疫苗因具有可快速合成的突出优势，在传染病应急防控中可发挥重要作用。目前已上市mRNA疫苗广泛使用假尿苷修饰策略，然而其翻译时的核糖体移码风险不容忽视。本研究首次提出基于密码子对偏好（CPB）的三维优化框架，同步调控密码子适应性指数（CAI）、二级结构稳定性（MFE）及密码子对组合规律。以新冠病毒S蛋白为模式抗原，通过CAI-CPB-MFE联合优化，以一种替代假尿苷化学修饰的方式，在动物模型中实现了中和抗体反应和CD8+ T细胞应答的显著提升。对于应急疫苗研发，为进一步在疫苗设计以外“以强胜强”，免疫增强剂在其中能够发挥关键作用。本研究揭示花生四烯酸（ARA）作为膳食佐剂，其代谢产物前列环素（PGI2）能够上调B细胞CD86，激活诱导胞苷脱氨酶（AID）以促进抗体亲和力成熟，显著提升模式疫苗中和抗体滴度并增强免疫保护作用。人体试验初步证实口服ARA可加速疫苗接种后的抗体生成，使其在初次免疫后仅一周即可达到足以抵抗病毒的保护水平，为临床转化奠定基础。综上，本研究针对传染病疫苗研发的现实需求，从高频突变病毒靶标抗原预测、mRNA疫苗序列优化、新型免疫激活剂三个方面，提出了多维创新的疫苗免疫增效策略，对传染病防控关键技术储备具有重要意义。
Preventing potential diseases before they arise is considered the highest pursuit of ancient Chinese physicians. For infectious diseases, prevention is always a better choice over treatment. In the global context of the frequent emergence of new infectious diseases, the speed and efficacy of vaccine development have become central issues for public health. Focusing on viral immune escape, mRNA vaccine optimization bottleneck, and the synergistic mechanisms of host immune responses, our study developed an innovative vaccine enhancement strategy based on artificial intelligence predictive model, codon pair optimization, and immune activator application through multidisciplinary integration. The mutation prediction model guides antigen selection for vaccines, which is of significant importance for epidemic prevention and control. Addressing the issues of increased antigenic immune escape and infectivity caused by high-frequency viral mutations, we developed a lightweight language model for evolution prediction, using a natural language processing (NLP) framework to decode the "biological grammar" of viral sequences. Taking SARS-CoV-2 as an example, we constructed a "grammatical framework" of the S1 sequence for dimensionality reduction and semantic representation to capture the underlying patterns of the model. The mutational profile, defined as the frequency of mutations, was introduced to incorporate randomness. By analyzing the combinatorial patterns of mutation sites, we detected the complete sequences or key mutations of prevailing strains such as XBB.1.16, EG.5, and JN.1 before their emergence. We further validated their enhanced ACE2 binding affinity and antibody escape capacity via pseudovirus assay. This model breaks through the limitations of traditional strategies such as phylogenetic trees and deep mutational scanning, providing a new tool for prospective vaccine design.After selecting appropriate antigens, we aimed to optimize the vaccine sequence, especially that of the novel mRNA vaccine. In order to overcome the potential risk of frameshift mutation and inflammation caused by the canonical pseudouridine modification strategies and improve translation efficiency, this study proposed for the first time a three-dimensional optimization framework based on codon pair bias (CPB), codon adaptation index (CAI), and secondary structure stability (MFE). Using the SARS-CoV-2 S protein and the Chikungunya virus E protein as model antigens, CAI-CPB-MFE optimization significantly enhanced mRNA stability and protein translation efficiency. In vivo experiments showed that the three-dimensional optimized vaccine induced a higher level of neutralizing antibody and CD8+ T cell immune responses than those modified by traditional optimization strategies. Further research found that codon pair modification significantly reduced the levels of inflammatory cytokines induced by mRNA vaccines, outperforming pseudouridine modification. Studies on different variants of the same virus validated the generalizability and universality of this strategy.In addition to antigen sequence optimization, novel immune activators can also serve as an independent and effective vaccine optimization strategy. To further enhance the vaccine-induced host's immune responses, this study revealed that reports that arachidonic acid (ARA), an Omega-6 polyunsaturated fatty acid, promotes germinal center (GC) B cell response and humoral immunity. Dietary administration of ARA significantly boosts rabies vaccine-induced production of neutralizing antibodies and protection against lethal rabies virus (RABV) infection in mice. Mechanistically, ARA is enriched in lymph nodes and metabolized into immune modulators there. One of the ARA metabolites, prostaglandin I2 (PGI2), via the cyclic adenosine monophosphate (cAMP)-protein kinase A (PKA) axis, upregulates the expression of a costimulatory molecule CD86, and activates activation-induced cytidine deaminase (AID) in B cells. Preliminary human trials confirmed that oral ARA accelerated antibody production after vaccination, reaching levels sufficient to protect against the virus as early as one week after primary immunization, laying the groundwork for clinical translation.
