Volume Electron Microscopy Workflows for the study of Large-Scale Neural Connectomics
Volume Electron Microscopy Workflows for the study of Large-Scale Neural Connectomics
复制标题
用于研究大规模神经连接组学的体积电子显微镜工作流程
DOI：
10.1093/micmic/ozad067.622
复制
发表时间：
2023
影响因子：
2.8
通讯作者：
Berger, D
中科院分区：
文献类型：
--
作者：
Schalek, R L;Petkova, M;Boulanger-Weill, J;Karlupia, N;Wang, S;Wang, X;Dhanyasi, N;Berger, D
在过去的15年里，Lichtman实验室已经开发了样品制备，成像和计算工具，以及相关的工作流程，用于采集神经组织体积，范围从。002 mm3（c. elegans，1.1 TB）[1]到1 mm 3（人类皮质，> 1.3 PB）[2]。为SEM开发的三种不同的工作流程（Zeiss Sigma、Zeiss multiSEM和Thermofisher Magellan）都依赖于磁带到SEM自动磁带超薄切片机（ATUM）技术。此外，在开发相关光学和电子显微镜（CLEM）工作流程[3]和使用X射线uCT（Xradia 510）作为中间分辨率相关工具方面已经做出了相当大的努力。通过合作努力和资源共享U24赠款[4]，该实验室为扩展体积电子显微镜社区提供了连续切片成像服务，该社区由三个用户组组成：在实验室中，外部和远程用户。磁带到SEM ATUM技术通过允许二次电子或背散射电子成像策略和工作流程暂停来提供工作流程灵活性，这在其他连续切片中不容易获得成像技术。ATUM技术的优势包括大样本量、多分辨率成像、多区域成像、多模式成像、切片重新成像策略（非破坏性）和失败惩罚小。此外，这种切片技术的选择为组织染色[5]和后染色以及切片存档提供了更大的范围。样本量和类型从人脑活检样本到小鼠脑组织再到整个动物（斑马鱼、水螅等）。通常，项目需要30-65 nm厚的连续切片，这些切片收集在碳涂层Kapton带上，切片错误率为每500个切片1个切片。连续切片的数量从几百到几千不等（迄今为止切割和收集的最大系列切片为33 000个）。ATUM方法的缺点是多个步骤、复杂的3D对齐、褶皱形成（如果收集不当）以及易受灰尘颗粒和划痕的影响。本演示文稿概述了基于ATUM的通用工作流（参见图1）。这里描绘了从步骤1到步骤7的整个流水线。由于小块（<1 mm x 1 mm x 1 mm）连续成像可以用FIB/SEM或连续块面方法进行，因此本讨论将主要集中在样品制备、切片切割和收集以及较大样品的成像上。图2中的数据显示了从新生（P0）小鼠脑（107 mm x 10 mm）、人视网膜（102.3 mm x 3 mm）和斑马鱼全脑获得的连续切片。大多数项目使用4 nm的图像像素尺寸;然而，可以使用更大和更小的像素尺寸。当按比例放大到更大的样品尺寸时遇到的主要问题分为两类：（1）由金刚石刀和超薄切片机的切割作用施加的机械限制，以及（2）试剂扩散和化学反应的物理限制。将介绍克服这些限制的方法示例[6]。
For the past fifteen years, the Lichtman laboratory has developed sample preparation, imaging, and computational tools–along with the associated workflows–for the acquisition of neural tissue volumes ranging from. 002 mm3 (c. elegans,∼ 1 TB)[1] to 1 mm3 (human cortex,> 1.3 PB)[2]. Three distinct workflows developed for SEMs (Zeiss Sigma, Zeiss multiSEM, and Thermofisher Magellan) all rely upon the tape-to-SEM automated tape ultramicrotome (ATUM) technique. In addition, considerable effort has been made in developing correlative light and electron microscopy (CLEM) workflows [3] and using x-ray uCT (Xradia 510) as an intermediate resolution correlation tool. Through collaborative efforts and a resource-sharing U24 grant [4] the lab has provided serial section imaging services for the extended volume electron microscopy community consisting of three user groups: in-lab, external and remote users.The tape-to-SEM ATUM technology provides workflow flexibility by allowing either secondary electron or backscattered electron imaging strategies and workflow pauses which are not easily available in other serial section imaging technologies. The advantages of the ATUM technology include large sample size, multi-resolution imaging, multi-region imaging, multi-modal imaging, section reimaging strategies (non-destructive), and small penalties for failure. Furthermore, this choice of sectioning technology provides a greater range for tissue staining [5] and post-staining, in addition to section archiving. Sample sizes and types range from human brain biopsy samples to mouse brain tissue to whole animals (zebrafish, hydra, etc.). Typically, projects require 30–65 nm thick serial sections which are collected on carbon-coated Kapton tape with sectioning error rates of 1 section per 500 sections. The number of serial sections range from several hundred to many thousands (largest series cut and collected so far is 33,000 sections). Disadvantages to the ATUM approach, multiple steps, complicated 3D alignment, wrinkle formation–if not collected properly–, and susceptibility to dust particles and scratches. This presentation gives an overview of the generalized ATUM-based workflow (see Figure 1). Here the entire pipeline from step 1 to step 7 is depicted. Since small block (< 1mm x 1mm x 1 mm) serial imaging can be performed with FIB/SEM or serial block face approaches, this discussion will focus primarily on the sample preparation, section cutting and collecting and the imaging of larger samples. The data in Figure 2 shows serial sections acquired from a newborn (P0) mouse brain (∼ 7 mm x 10 mm), a human retina (∼ 2.3 mm x 3 mm), and the whole brain of a zebrafish. The majority of projects use an image pixel size of 4 nm; however, both larger and smaller pixel sizes can be used. The major problems encountered when scaling to larger sample sizes fall into two categories:(1) mechanical limitations imposed by the diamond knife and by the cutting action of the ultramicrotome, and (2) the physical limitations of reagent diffusion and chemical reactions. Examples of approaches to overcome these limitations will be presented [6].