Receptor exocytosis imaged with high temporal resolution for diverse receptor cargos

Cells perceive and interact with their environment in part through the expression, activation, and regulation of receptors on their plasma membrane. These receptors are dynamically trafficked from the plasma membrane in a process called endocytosis and delivered to the plasma membrane via exocytosis. Different receptors take diverse routes through the cell before being delivered via exocytosis. The data in this project focuses on 3 prototypical plasma membrane receptors - the B2 adrenergic receptor, the µ opioid receptor, and the transferrin receptor. Using a pH-sensitive green fluorescent protein variant, we visualized these receptors in cells as they recycled to the plasma membrane. We subsequently hand-labeled a subset of the data in order to build an automated image analysis method that could be used to detect receptor exocytosis across diverse imaging conditions. This repository contains our primary microscopy data from these studies as well as the labeling for use in supervised machine learning. These data support Evans et al 2021 and subsequent publications. <strong>Data Collection</strong><br> TIFF image stacks were collected using a Nikon Eclipse TiE Inverted Microscope using TIRF illumination with a solid state 488nm laser through a Nikon 60x/1.49NA TIRF objective and captured using an Andor iXon 897+ EMCCD camera. The camera was windowed to a 300x300 pixel view and images were collected with a 18.5ms exposures (~54Hz). Images were collected across two days, with two coverslips of each condition collected on day 1, and one coverslip collected on day 2. <strong>DNA Constructs</strong><br> The 3 cargos imaged in these data are the transferrin receptor (TfR), the B2-adrenergic receptor (B2AR, B2), and the µ opioid receptor (MOR). Constructs encoding these receptors, tagged extracellularly with the ph-sensitive GFP variant Superecliptic pHluorin (SpH, Sankaranarayanan et al. 2000, have been previously described in Yudowski et al. 2006 for B2AR, Yu et al. 2010 for MOR, and Yudowski et al. 2009 for TfR. <strong>Cell Culture</strong><br> HEK293 cells were cultured in DMEM High Glucose (Hyclone) supplemented with 10% Heat Inactivated FBS (Gibco). Cells expressing B2 and MOR were stably selected from transient transfection using G418. Cells expressing TfR were transfected 3 days before the experiments presented here using Effectene following manufacturers' instructions. Before imaging, cells were transferred to 25mm diameter #1.5 glass coverslips (Electron Microscopy Sciences). Two days after plating, experiments began. <strong>Imaging conditions</strong><br> Cells were imaged in L-15 minimal media supplemented with 1% FBS. For MOR and B2, cells were imaged for 1 minute at ~0.16Hz without perturbation. Then agonist was added (10µM DAMGO for MOR, 10µM isoproterenol for B2) to the media and cells were imaged for 5 minutes to ensure that receptors clustered and internalized. After internalization, cells were bleached with 100% laser power for 1 minute and then imaged at 54Hz to visualize exocytic events. exocytosis was captured for up to 20 minutes after initial treatment, one cell at a time. For TfR, a single frame was taken before bleaching to show receptor expression levels and then cells were bleached and imaged as described above. <strong>Data blinding</strong><br> After collection, files were renamed as described in <em>map.md</em>. All metadata files and internalization imaging were separated into the 2 "extras" folders. The exocytosis movies were 'scrambled' to hide cargo identity using the included <em>scrambler.py</em> file. <em>OPP_scramble.log</em> described the mapping of scrambled filenames to the original imaging. <strong>Human labeling</strong><br> A subset of the images (22, with roughly equal representation across cargos) were hand labeled for exocytic events. Images were viewed in FIJI Schindelin et al. 2012 nad played back at 0.5x. When exocytic events were identified by eye, the playback was paused and the appearance of an event was found through manual advancing of the frames of the movie. The event was labeled using the Cell Counter plugin. Each movie was watched twice to identify as many events as possible. Labeled events are saved a <em>&lt;movie-name&gt;-ZYW-1.xml</em> in this dataset. <strong>Data organization</strong><br> All exocytic event movies and any matching human labeling are included in this base directory. All internalization movies and all metadata for all movies are included in the Extras folder for the day that movie was recorded. Coverslip and cargo identity are listed in <em>map.md</em> and the ground truth for cargo identity is in <em>OPP_scramble.log</em>细胞感知并与外界环境发生相互作用的途径之一，是通过其质膜（plasma membrane）上受体的表达、激活与调控来实现的。此类受体可从质膜经胞吞作用（endocytosis）被动态转运至细胞内，再通过胞吐作用（exocytosis）被运送回质膜。不同受体在经胞吐作用被运送至质膜前，会在细胞内经由多条不同的转运路径。本项目的数据集聚焦于3种典型的质膜受体——B2肾上腺素能受体（B2 adrenergic receptor）、μ阿片受体（µ opioid receptor）与转铁蛋白受体（transferrin receptor）。本研究使用pH敏感型绿色荧光蛋白（GFP）变体，对细胞内循环返回质膜的上述受体进行了可视化成像。随后，我们对部分数据集开展人工标注，以构建可在多种成像条件下检测受体胞吐事件的自动化图像分析方法。本数据集包含本研究的核心显微成像数据，以及可用于监督机器学习的标注文件，为Evans等人2021年的研究及后续相关发表成果提供了数据支撑。<strong>数据采集</strong><br> 图像栈（TIFF image stacks）采用尼康Eclipse TiE倒置显微镜采集：搭配固态488nm激光器进行全内反射荧光（TIRF）照明，通过尼康60x/1.49NA TIRF物镜成像，并使用Andor iXon 897+电子倍增电荷耦合器件（EMCCD）相机捕获图像。相机视窗设置为300×300像素，图像采集曝光时长为18.5ms（约54Hz帧率）。数据采集共持续两天：第1天采集每种实验条件下的2个盖玻片样本，第2天采集1个盖玻片样本。<strong>DNA构建体</strong><br> 本数据集中成像的3种靶标受体分别为转铁蛋白受体（transferrin receptor, TfR）、B2肾上腺素能受体（B2-adrenergic receptor, B2AR, B2）以及μ阿片受体（µ opioid receptor, MOR）。编码上述受体的DNA构建体均在其胞外区域标记了pH敏感型GFP变体超折叠pHluorin（Superecliptic pHluorin, SpH，Sankaranarayanan等人2000年首次报道）；其中B2AR的相关构建体参见Yudowski等人2006年的研究，MOR的构建体参见Yu等人2010年的研究，TfR的构建体参见Yudowski等人2009年的研究。<strong>细胞培养</strong><br> HEK293细胞培养于高糖DMEM培养基（Hyclone品牌），并添加10%热灭活胎牛血清（FBS，Gibco品牌）。表达B2和MOR的细胞通过瞬时转染后，使用G418进行稳定株筛选。表达TfR的细胞则在本实验开展前3天，使用Effectene转染试剂按照厂商说明书进行转染。成像前，将细胞接种于直径25mm的#1.5规格玻璃盖玻片（Electron Microscopy Sciences品牌）；接种两天后，启动成像实验。<strong>成像条件</strong><br> 成像时，细胞培养于添加了1% FBS的L-15基础培养基中。针对MOR和B2受体的样本：先在无干扰条件下以约0.16Hz的帧率成像1分钟；随后向培养基中加入激动剂（MOR对应10μM DAMGO，B2对应10μM 异丙肾上腺素），继续成像5分钟以确保受体发生聚集并内吞。内吞完成后，以100%激光功率对细胞进行光漂白处理1分钟，随后以54Hz帧率成像以观测胞吐事件；从初始给药起，最多连续采集20分钟的成像数据，每次仅对单个细胞进行拍摄。针对TfR受体的样本：光漂白前先采集单帧图像以记录受体表达水平，后续光漂白及成像流程与上述一致。<strong>数据盲化处理</strong><br> 数据采集完成后，按照<em>map.md</em>中的说明对文件进行重命名。所有元数据文件及内吞成像数据被分别存入两个“extras”文件夹。本数据集附带的<em>scrambler.py</em>脚本可对胞吐成像视频进行“打乱重命名”处理，以隐藏样本对应的受体类型；<em>OPP_scramble.log</em>文件记录了打乱后的文件名与原始成像数据的对应关系。<strong>人工标注</strong><br> 我们对部分成像数据（共22个视频，各受体类型的样本量大致相当）进行了胞吐事件的人工标注。使用FIJI软件（Schindelin等人2012年报道）查看图像，并以0.5倍速播放视频。当通过目视识别到胞吐事件时，暂停播放并逐帧手动推进视频以精确定位事件出现的时刻，随后使用Cell Counter插件对该事件进行标注。每个视频需观看两遍，以尽可能识别所有胞吐事件。本数据集中的标注事件以<em>&lt;movie-name&gt;-ZYW-1.xml</em>格式保存。<strong>数据组织形式</strong><br> 所有胞吐事件成像视频及对应的人工标注文件均存放于当前根目录。所有内吞成像视频及所有视频的元数据，均按照成像日期存入对应的Extras文件夹。盖玻片编号及对应受体类型的信息详见<em>map.md</em>；受体类型的真实标签（ground truth）则记录于<em>OPP_scramble.log</em>文件中。