A robust platform for multiplexed, subcellular proteomic imaging in human tissue

项目摘要 通过飞行时间的多路复用离子束成像（MIBI-TOF）使用二次离子质谱和金属离子质谱。 偶联的一抗在单个细胞中以亚细胞分辨率同时可视化数十种蛋白质， 组织切片该技术与存档福尔马林固定、石蜡包埋组织（FFPE）向后兼容 并已被用于同行评议的工作，同时可视化和定量36种蛋白质在回顾性 人体组织队列。根据HuBMAP联盟的既定目标，开发“高灵敏度， 能够快速提供大面积组织光谱数据的高分辨率成像技术”， 定量成像分析工具，包括自动3D图像分割、特征提取和图像分析。 注释，”这里概述的工作将创建一个标准化的，高通量的，用户友好的工作流程， 在基础和转化研究中使用MIBI-TOF，以深入了解单细胞表型和组织 结构在健康和疾病中是功能性关联。为了实现这一目标，我们将验证100个FFPE抗体 并优化冻干形式的即用型多重染色板， 两年用于蛋白质和mRNA靶标的多重信号扩增的方案和试剂将进一步被公开。 改进，而下一代仪器将增加样本吞吐量，以允许完整的组织切片 在1小时内成像多达40种蛋白质。标准化试剂和更强大的仪器将 伴随着利用标准分割标记集的自动计算流水线， 机器学习，以准确识别任何非神经人体组织中的细胞核和细胞边界。这些数据将 用于根据形态学、蛋白质和/或生物学特性将单细胞事件聚类为功能不同的群体， 表达和组织学分布。这里提出的试剂和计算流水线与 现有的HuBMAP资助的平台，可以很容易地推广到几乎任何高维成像 模态因此，这项工作不仅将提供一个实用的，背部兼容的成像平台，高通量 多路复用成像，但也将加速其他互补成像技术的发展。

Project Summary Multiplexed Ion Beam Imaging by Time of Flight (MIBI-TOF) uses secondary ion mass spectrometry and metal conjugated primary antibodies to simultaneously visualize dozens of proteins at subcellular resolution in a single tissue section. This technology is back compatible with archival formalin fixed, paraffin embedded tissue (FFPE) and has been used in peer-reviewed work to simultaneously visualize and quantify 36 proteins in retrospective human tissue cohorts. In line with the stated goals of the HuBMAP consortium to develop both “High-sensitivity, high-resolution imaging techniques that can rapidly provide spectral data over large areas of tissue” and “Quantitative imaging analysis tools, including automated 3D image segmentation, feature extraction, and image annotation,” the work outlined here will create a standardized, high throughput, and user-friendly workflow for using MIBI-TOF in basic and translational research to gain insight into how single cell phenotype and tissue structure are functionally-linked in health and disease. To achieve this, we will validate 100 FFPE antibodies and optimize ready-to-use multiplexed staining panels in lyophilized format that will permit storage for at least two years. Protocols and reagents for multiplexed signal amplification of protein and mRNA targets will be further refined, while next generation instrumentation will increase sample throughput to permit full tissue section imaging of up to 40 proteins in 1 hour. Standardized reagents and more robust instrumentation will be accompanied by an automated computational pipeline that utilizes a standard set of segmentation markers and machine learning to accurately identify nuclei and cell borders in any non-neural human tissue. This data will be used to cluster single cell events into functionally distinct populations according to morphology, protein expression, and histological distribution. The reagents and computational pipeline proposed here synergize with existing HuBMAP-funded platforms and could be readily generalized to virtually any high dimensional imaging modality. Thus, this work will not only provide a practical, back compatible imaging platform for high throughput multiplexed imaging, but will also accelerate development of other complimentary imaging technologies as well.