近年来，组学技术的发展极大地推动了人们对生命活动和疾病机理的认识和临床精准诊治。然而，基于组织匀浆或单细胞解离的组学分析技术会破坏组织中细胞和分子所处的空间位置，无法获得高异质性组织（如脑、肿瘤等）中的分子和细胞的分布及相互作用信息。针对上述问题，基于质谱成像（MSI）技术的空间代谢组学（SM）得到快速发展，并在探索发病机制，发现生物标志物方面显示巨大潜力。空间分辨率是MSI的重要参数之一，随着技术的不断发展，微米级分辨率的MSI实现了代谢物在细胞中的精准定位。然而，灵敏度和空间分辨率成反比关系，做好灵敏度和分辨率之间的权衡一直是MSI分析中亟待解决的问题。本研究在课题组研发的空气动力辅助解吸电喷雾电离质谱成像（AFADESI-MSI）的基础上，设计了精细喷雾探针，并对成像的关键参数进行了系统性的优化和改进，研发了一种针对具有复杂微区组织的高清空间分辨代谢组学方法，保证了检测灵敏度的同时，空间分辨率达到30 μm，实现了小鼠脑和临床胃癌组织的高清空间分辨代谢组学分析，构建了空间代谢谱图及代谢网络。

在课题组前期自主研发的空间代谢组学技术的基础上，进一步整合空间脂质组学和空间转录组学技术，在同一肿瘤组织样本的相邻切片上，实现了肿瘤组织微区中代谢组、脂质组和转录组数据的原位精准联合分析。通过鉴别细胞种类，构建代谢物、脂质和上游调控基因的关联网络，实现了肿瘤组织微环境中肿瘤细胞、免疫细胞和基质细胞等代谢调控及交互作用的原位表征。

本论文的研究内容主要包括以下三个部分：

1. 高清空间分辨代谢组学方法研究

针对高空间分辨质谱成像对小分子代谢物检测灵敏度较低的问题，研究开发了一种基于AFADESI-MSI的高分辨、高灵敏、高覆盖的空间代谢组学方法。根据喷雾解吸的原理，设计了毛细管内径为20 μm的精细探针，并对探针位置、喷雾溶剂组成及喷雾溶剂流速等关键参数进行了考察和优化。应用逐点剥蚀的扫描策略，通过完全解吸和减小移动步长有效提高了空间分辨率，并且在获得合理分辨率的同时保证了检测的灵敏度，得到稳定的成像结果。应用该探针和扫描策略，实现了小鼠小脑30 μm的高空间分辨代谢组学分析，小脑的白质、灰质和颗粒层在MSI中清晰可见。从像素点尺寸缩小比率(~6.25倍)来看，离子信号强度降低仅不到1/2，单位面积离子强度有所提高。此外，该方法的动态检测范围覆盖了3-4个数量级，在正、负离子模式下共注释了406个种类丰富的代谢物，包括胆碱类、多胺类、肉碱类、氨基酸类、核苷类、有机酸类、碳水化合物类和脂类等。获得的高质量图像和数据表明该方法稳定可靠，满足代谢组学的分析要求，且对于具有精细结构组织中低含量的小分子代谢物显示出独特的分析优势。

2. 应用高空间分辨代谢组学方法构建小鼠脑的代谢谱图

采用上述方法，对异质性小鼠脑切片进行了全面的高空间分辨代谢组学分析，以绘制小鼠脑的高清空间代谢谱图。MSI获得了代谢物在15个脑功能微区中的特异性分布特征，30 μm的空间分辨率将代谢物定位至海马的亚结构。进一步，将MSI数据与应用LC/GC-MS分析建立的代谢物数据库进行匹配，共在正、负离子模式下注释了333和162个代谢物离子。相关性分析表明大脑各区域的代谢表型基本遵循经典解剖学的结构与功能的划分，相邻或相同类型的结构在代谢分子水平高度相关。将代谢物和脑功能相关联，发现其中大部分代谢物的分布特征与其所在区域功能密切相关。结合代谢通路富集分析，成功绘制了小鼠脑空间代谢网络。相关代谢物分布特征和代谢通路上下游信息，将有助于解释大脑复杂的调控网络，为脑中的区域功能和稳态提供了全面的代谢参考。

3. 采用高空间分辨代谢组学方法表征肿瘤微环境中免疫细胞的代谢改变

应用高空间分辨代谢组学方法对临床胃癌术后样本进行了可视化分析，重点关注肿瘤和正常组织交界处淋巴滤泡的代谢特征，以表征免疫细胞代谢重编程。通过微小区域代谢物信息精确提取和多变量统计分析，在正常组织和淋巴滤泡中共筛选出348个差异代谢物，进一步的聚类分析表明不同微环境中的淋巴滤泡具有不同的代谢特征。对注释的差异代谢物进行代谢通路富集分析，结果显示，淋巴滤泡在能量代谢、氨基酸代谢和脂质代谢中发生了显著的改变。磷酸戊糖途径代谢中，肿瘤附近的免疫细胞更为明显的代谢改变表明，肿瘤细胞与免疫细胞通过能量代谢通路相互作用，免疫细胞在肿瘤侵袭过程中发生了代谢重编程。此外，代谢重编程相关的代谢物如组胺、谷氨酰胺、乳酸和精胺等在免疫细胞中均有不同程度的富集，可能使其免疫功能受到抑制。

4. 空间多组学新技术揭示胃癌相关细胞特异的代谢重编程与互作

肿瘤是由多种不同类型细胞组成的复杂组织，为探究免疫细胞与肿瘤细胞的相互作用和代谢重编程，研究结合了基于质谱成像的空间代谢组学、空间脂质组学以及基于测序的空间转录组学方法，对人胃癌组织进行了空间多组学分析。通过对采样点进行坐标标注对齐了三种方法得到的成像图，将代谢物、脂质和基因表达信息联系在一起，通过数据挖掘从多个层面表征肿瘤代谢重编程。多组学分析的结果显示，胃癌中精氨酸和脯氨酸代谢、谷氨酸和谷氨酰胺代谢、脂质和脂肪酸代谢都发生了代谢重编程，其相关基因的表达也存在明显差异。此外，来自空间多组学方法的综合数据确定了复杂肿瘤微环境中的细胞类型和分布，发现了富含免疫细胞的“肿瘤-正常界面”区域具有明显的转录特征和显著的免疫代谢改变。研究提供了一个包括肿瘤细胞和周围正常细胞在内的代谢物、脂质和基因表达模式的图集，有助于深入了解肿瘤内的生化异质性，并破译新陈代谢在癌症发展中的作用。

In recent years, the development of omics techniques has greatly advanced the understanding of life activities and disease mechanisms and clinical precision diagnosis and treatment. However, omics analysis techniques based on tissue homogenization or single cell dissociation can disrupt the spatial location of cells and molecules in tissues, and cannot obtain information on the distribution and interactions of molecules and cells in highly heterogeneous tissues (e.g., brain, tumor, etc.). In response to these problems, spatial metabolomics (SM) based on mass spectrometry imaging (MSI) technology has been rapidly developed and shows great potential in exploring pathogenesis and discovering biomarkers. Spatial resolution is one of the important parameters of MSI, and with the continuous development of the technology, MSI with micron-level resolution has enabled the precise localization of metabolites in cells. However, sensitivity and spatial resolution are inversely related, and a good trade-off between sensitivity and resolution has been a pressing issue in MSI analysis. In this study, based on the air flow-assisted desorption electrospray ionization mass spectrometry imaging (AFADESI-MSI) developed by the group, we designed a fine spray probe and systematically optimized and improved the key parameters of imaging to develop a high-definition spatially resolved metabolomics method for tissues with complex micro-regions, ensuring the detection sensitivity while achieving a spatial resolution of 30 μm. We have achieved high-definition spatially resolved metabolomics analysis of mouse brain and clinical gastric cancer tissues, and constructed spatial metabolic profiles and metabolic networks.

Based on the spatial metabolomics technology developed by the group, the group further integrated spatial lipidomics and spatial transcriptomics technologies to achieve in situ accurate joint analysis of metabolome, lipidome and transcriptome data in tumor tissue micro-regions on adjacent sections of the same tumor tissue sample. In situ characterization of metabolic regulation and interactions among tumor cells, immune cells and stromal cells in the tumor tissue microenvironment was achieved by identifying cell types and constructing association networks of metabolites, lipids and upstream regulatory genes.

The research of this dissertation consists of three main parts as follows:

1. Study on high-definition spatial-resolution metabolomics method

To address the problem of low sensitivity of high spatial resolution mass spectrometry imaging for small molecule metabolite detection, a high resolution, high sensitivity and high coverage spatial metabolomics method based on AFADESI-MSI was investigated and developed. Based on the principle of spray desorption, a fine probe with capillary inner diameter of 20 μm was designed, and key parameters such as probe position, spray solvent composition and spray solvent flow rate were investigated and optimized. By applying the scanning strategy of point-by-point stripping, the spatial resolution was effectively improved by complete desorption and reduced moving step, and the sensitivity of detection was ensured while obtaining reasonable resolution, and stable imaging results were obtained. Applying this probe and scanning strategy, a high spatial resolution metabolomic analysis of the mouse cerebellum at 30 μm was achieved, and the white matter, gray matter and granular layer of the cerebellum were clearly visible in MSI. The ion signal intensity was reduced by only less than 1/2 in terms of the pixel point size reduction ratio (~6.25 times), and the ion intensity per unit area was improved. In addition, the dynamic detection range of the method covered 3-4 orders of magnitude, and a total of 406 species-rich metabolites, including cholines, polyamines, carnitines, amino acids, nucleosides, organic acids, carbohydrates and lipids, were annotated in both positive and negative ion modes. The high quality images and data obtained indicate that the method is stable and reliable, meets the analytical requirements of metabolomics, and shows unique advantages for the analysis of small molecules metabolites with low content in tissues with fine structure.

2. Application of a high spatial resolution metabolomics approach to construct a metabolic profile of the mouse brain

A comprehensive high spatial resolution metabolomic analysis of heterogeneous mouse brain sections was performed to map the high definition spatial metabolic profile of mouse brain using the above method. MSI obtained metabolite specific distribution characteristics in 15 functional brain microregions, and a spatial resolution of 30 μm localized metabolites to substructures in the hippocampus. Further, the MSI data were matched with the metabolite database established by applying LC/GC-MS analysis, and a total of 333 and 162 metabolite ions were annotated in positive and negative ion modes. Correlation analysis showed that the metabolic phenotypes of brain regions largely followed the classical anatomical division of structure and function, and adjacent or identical types of structures were highly correlated at the metabolic molecular level. Correlating metabolites and brain functions, the distribution characteristics of most of them were found to be closely related to the functions of their regions. Combined with the metabolic pathway enrichment analysis, the spatial metabolic network of mouse brain was successfully mapped. The relevant metabolite distribution characteristics and upstream and downstream information of metabolic pathways will help to explain the complex regulatory network of the brain and provide a comprehensive metabolic reference for regional function and homeostasis in the brain.

3. Application of a high spatial resolution metabolomics approach to characterize the metabolic alterations of immune cells in the tumor microenvironment

A high spatial resolution metabolomics approach was applied to visualize the postoperative samples of clinical gastric cancer, focusing on the metabolic characteristics of lymphoid follicles at the junction of tumor and normal tissues to characterize the metabolic reprogramming of immune cells. A total of 348 differential metabolites were screened in normal tissues and lymphoid follicles by precise extraction of metabolite information from microregions and multivariate statistical analysis, and further clustering analysis showed that lymphoid follicles in different microenvironments had different metabolic profiles. The metabolic pathway enrichment analysis of the annotated differential metabolites showed that lymphoid follicles were significantly altered in energy metabolism, amino acid metabolism and lipid metabolism. The more pronounced metabolic alterations in pentose phosphate pathway metabolism in immune cells near the tumor suggest that tumor cells interact with immune cells through energy metabolic pathways and that immune cells undergo metabolic reprogramming during tumor invasion. In addition, metabolic reprogramming-related metabolites such as histamine, glutamine, lactate and spermine were enriched in immune cells to varying degrees, which may have suppressed their immune function.

4. Spatial multi-omics techniques reveal cell-specific metabolic reprogramming and interactions associated with gastric cancer

Tumors are complex tissues composed of many different cell types. To explore the interaction and metabolic reprogramming between immune cells and tumor cells, the study combined mass spectrometry imaging-based spatial metabolomics, spatial lipidomics, and sequencing-based spatial transcriptomics approaches to perform spatial multi-omics analysis of human gastric cancer tissues. The imaging maps obtained by the three methods were aligned by coordinate annotation of sampling points, linking metabolite, lipid and gene expression information to characterize tumor metabolic reprogramming at multiple levels through data mining. The results of the multi-omics analysis showed that metabolic reprogramming occurred in arginine and proline metabolism, glutamate and glutamine metabolism, lipid and fatty acid metabolism in gastric cancer, and there were significant differences in the expression of their associated genes. In addition, comprehensive data from a spatial multi-omics approach identified cell types and distributions in the complex tumor microenvironment and revealed a distinct transcriptional profile and significant immunometabolic alterations in the "tumor-normal interface" region, which is rich in immune cells. The study provides an atlas of metabolite, lipid and gene expression patterns, including tumor cells and surrounding normal cells, which will help to gain insight into biochemical heterogeneity within tumors and decipher the role of metabolism in cancer development.