Single-Islet Proteomics Maps Pseudo-Temporal Islet Immune Responses and Dysfunction in Presymptomatic Type 1 Diabetes

Aims/hypothesis: Progressive β-cell dysfunction precedes the onset of type 1 diabetes (T1D), yet the molecular mechanisms driving early pathogenesis remain poorly understood. Functional and multiplexed imaging studies have reported lobular heterogeneity in the pancreas with respect to immune and β-cell dysfunction signatures. Although single-cell transcriptomics has identified cellular changes, it provides limited insight into the heterogeneity of distinct islet microenvironments. In this exploratory study, we employ single-islet proteomics to profile intra-donor islet heterogeneity across three multiple autoantibody-positive donors (mAAb+), representing the transition from Stage 1 to Stage 2 T1D, alongside matched non-diabetic controls, to resolve early T1D pathogenesis and identify cellular processes coupled to the islet immune response. Methods: Laser capture microdissection was used to isolate 439 individual pancreatic islets from presymptomatic mAAb+ (n = 3 donors, ~100 islets/donor) and non-diabetic control (n = 3 donors, ~50 islets/donor) organ donors obtained through the Network for Pancreatic Organ Donors with Diabetes. Islet identification and T-cell infiltration were evaluated using 3 multiplex immunohistochemistry assays for insulin, glucagon, and cell differentiation 3 proteins. Adjacent serial sections were used for islet laser capture microdissection and proteomic analysis using the Nanodroplet Processing in One pot for Trace Samples. Downstream proteomics analysis combined, weighted gene co-expression network analysis, random forest-based feature selection, linear modeling with empirical Bayes moderation, and gene set enrichment accounting for inter-gene correlation. Results: The single-islet proteomics workflow demonstrated high analytical reproducibility, with Pearson correlation coefficients exceeding 0.96 and an average of approximately 5,800 proteins quantified per donor. By combining weighted gene co-expression network analysis with random forest-based feature selection, we identified a 40-protein panel, defined as the Islet Immune Response Signature (IIRS), was identified that tracks activation and reflects a pseudo-temporal progression of the islet immune response. Additionally, functionally clustered protein modules across mAAb+ donors revealed significant intra-donor heterogeneity in β-cell-specific markers, pointing to β-cell dysfunction. Along the same axis, we established a panel of 42 proteins, defined as β-cell profile (BCP), with the highest correlation to insulin and Ectonucleoside triphosphate diphosphohydrolase 3, both β-cell markers. IIRS and BCP were found to correlate only weakly (Pearson r = 0.13), indicating that immune activation and β-cell function follow partially decoupled trajectories. Pathway analysis highlighted extracellular matrix remodeling associated with both the signature panels. Strong dysregulation of extracellular matrix organization in mAAb+ donors relative to their non-diabetic counterparts. Specifically, integrin-mediated signaling, cell-matrix adhesion, and collagen fibril organization show low association, while hyaluronan metabolic process shows strong association with IIRS. In contrast, ECM-modifying and ECM-degrading proteins (QSOX1, FBLN7, FAP, DPP4) consistently correlated negatively with BCP. Further, unique to β-cell function are pathways regarding mRNA processing and splicing, particularly in donors with insulin-depleted islets. Conclusion/interpretation: Our results reveal highly consistent proteomic patterns that reflect pseudo-time progression in the islet immune response and β-cell dysfunction. Pathways, including extracellular matrix remodeling and mRNA processing, were identified at the proteomic level as closely associated with progressive islet immune activation and loss of β-cell function. These findings provide evidence of early islet dysfunction, offer a valuable resource for investigating T1D pathogenesis, including novel candidates for functional studies, and underscore the utility of single-islet spatial proteomics for examining islet heterogeneity in T1D.研究目的与假说：进行性β细胞功能障碍先于1型糖尿病（type 1 diabetes, T1D）发作，但驱动早期发病的分子机制仍不甚明晰。功能性与多重成像研究已报道胰腺在免疫及β细胞功能障碍特征方面存在小叶异质性。尽管单细胞转录组学（single-cell transcriptomics）已鉴定出细胞变化，但对不同胰岛微环境的异质性的解析能力仍有限。本探索性研究采用单胰岛蛋白质组学（single-islet proteomics），对3名多重自身抗体阳性（multiple autoantibody-positive, mAAb+）供体（处于1型糖尿病1期向2期的过渡阶段）的供体内胰岛异质性进行分析，并匹配非糖尿病对照，以解析1型糖尿病早期发病机制，并鉴定与胰岛免疫反应相关的细胞过程。 研究方法：通过糖尿病胰腺器官供体网络（Network for Pancreatic Organ Donors with Diabetes）获取的症状前多重自身抗体阳性供体（n=3，每名供体约100个胰岛）与非糖尿病对照供体（n=3，每名供体约50个胰岛）的胰腺组织中，采用激光捕获显微切割（laser capture microdissection）分离得到439个单个胰岛。针对胰岛素、胰高血糖素及3种细胞分化蛋白，采用3项多重免疫组化分析对胰岛进行鉴定并评估T细胞浸润情况。相邻连续切片用于胰岛激光捕获显微切割及基于微量样本单管纳滴处理（Nanodroplet Processing in One pot for Trace Samples）的蛋白质组学分析。下游蛋白质组学分析结合了加权基因共表达网络分析、基于随机森林的特征筛选、经经验贝叶斯校正的线性建模，以及考虑基因间相关性的基因集富集分析。 研究结果：单胰岛蛋白质组学流程展现出较高的分析重现性，皮尔逊相关系数超过0.96，每名供体平均可定量约5800种蛋白质。通过将加权基因共表达网络分析与基于随机森林的特征筛选相结合，我们鉴定出由40种蛋白质组成的胰岛免疫反应特征谱（Islet Immune Response Signature, IIRS），该特征谱可追踪免疫激活状态，并反映胰岛免疫反应的伪时间进展。此外，对多重自身抗体阳性供体的功能聚类蛋白模块分析显示，β细胞特异性标志物存在显著的供体内异质性，提示β细胞功能障碍。基于此，我们构建了由42种蛋白质组成的β细胞特征谱（β-cell profile, BCP），该特征谱与两种β细胞标志物——胰岛素及胞外核苷三磷酸二磷酸水解酶3（Ectonucleoside triphosphate diphosphohydrolase 3）的相关性最高。研究发现，胰岛免疫反应特征谱与β细胞特征谱仅呈弱相关（皮尔逊r=0.13），表明免疫激活与β细胞功能遵循部分解耦的轨迹。通路分析显示，两种特征谱均与细胞外基质重塑相关。与非糖尿病对照相比，多重自身抗体阳性供体的细胞外基质组织存在显著失调：具体而言，整合素介导的信号通路、细胞-基质黏附及胶原纤维组织与胰岛免疫反应特征谱关联较弱，而透明质酸（hyaluronan）代谢过程则与该特征谱呈强关联。与之相反，细胞外基质修饰及降解蛋白（QSOX1、FBLN7、FAP、DPP4）与β细胞特征谱呈持续负相关。进一步研究发现，mRNA加工与剪接通路为β细胞功能所特有，尤其在胰岛素耗竭的胰岛供体中更为显著。 结论与解读：本研究结果揭示了高度一致的蛋白质组学模式，该模式可反映胰岛免疫反应与β细胞功能障碍的伪时间进展。在蛋白质组水平上，包括细胞外基质重塑与mRNA加工在内的通路被鉴定为与进行性胰岛免疫激活及β细胞功能丧失密切相关。这些发现为早期胰岛功能障碍提供了证据，为探索1型糖尿病发病机制（包括功能研究的新型候选靶点）提供了宝贵资源，并凸显了单胰岛空间蛋白质组学在解析1型糖尿病胰岛异质性中的应用价值。