Standardized Workflow for High-Content Imaging and Data Processing

Summary

Here, we present a protocol for standardized high-content imaging and data processing of propidium iodide (PI)/Hoechst 33342-stained cells to perform reproducible cell viability analysis using automated image acquisition and analysis workflows.

Abstract

To address the challenges of low reproducibility due to operational variations in scientific research imaging and the complexity of analyzing a large number of samples using conventional methods, we introduce a protocol for staining HT22 cells with PI and Hoechst 33342 and then imaging and analyzing the data using the High-Content Imaging System. This workflow aims to establish robust imaging setting parameters and stringent quality control standards. It demonstrates the instrument's imaging settings and the workflow for three different image processing methods provided by the instrument: Multi-Wavelength Cell Scoring (Multi), Live-Dead (Live), and Custom Module Editor (CME). The image processing methods are compared and validated against the traditional method, ImageJ, to ensure their effectiveness. The study ultimately summarizes a workflow for using the High-Content Imaging System to determine cell viability using the PI and Hoechst double-staining method. By standardizing the entire imaging and data processing workflow, the study significantly enhances the reproducibility and reliability of scientific research imaging, thereby providing solid methodological support for cross-laboratory data comparison and advancing scientific research.

Introduction

High-Content Imaging System serves as a crucial tool in modern scientific research, enabling high-speed imaging and multi-parameter assessment of injured or uninjured cells^1,2,3. It plays an irreplaceable role in various disciplines. In the field of life sciences, it assists researchers in exploring the microscopic world of cells and tissues^4,5. For instance, the precise detection of fluorescently labeled molecules and structures within cells allows clear visualization of cellular fine structures and dynamic changes⁶. In neuroscience research, it can perform three-dimensional imaging of neurons, aiding in the analysis of the complex connections and signal transduction mechanisms of neuronal networks, and providing crucial clues for understanding brain functions and the pathogenesis of neurological diseases^7,8,9.

Despite the significant advantages of High-Content Imaging System technology, several issues remain with its imaging process. In the sample preparation stage, there is a high degree of subjectivity. In biological sample preparation, processes such as cell handling and staining lack relevant standards. In terms of imaging parameters, there are notable differences in exposure times and well plate selections across laboratories and high-content microscope models. In the data processing phase, there is a lack of standardization from noise reduction and contrast adjustment of raw images to the parameter identification and determination of target features. These issues directly lead to biases in microscope imaging results, causing a crisis of reproducibility.

Protocol

This protocol uses a commercial cell line (HT22, ATCC) and does not involve human subjects or primary human tissues. No animals were used in this study.

1. Standardized sample preparation

1. Prepare HT22 cells
   1. When the confluence of HT22 cells reaches 80%^10,11, discard the old medium. Use 0.25% trypsin to dislodge the cells, incubating for 1–2 min at room temperature.
   2. Add Dulbecco’s Modified Eagle Medium (DMEM) at 3x the volume of 0.25% trypsin to stop the dislodgment. Transfer the cell suspension to a 15 mL centrifuge tube.
   3. Centrifuge at 175 x g for 3 min at room temperature, then discard the supernatant. Resuspend the cells in complete DMEM to create a single-cell suspension.
   4. Count the cells using a cell counting chamber. Seed 5,000 cells per well in 100 µL of complete medium into a 96-well plate¹².
   5. Incubate the plate at 37 °C with 5% CO₂ for 24 h. Perform sample grouping, dosing, and staining operations in the 96-well plate according to the experimental design.
      ​NOTE: Use a 96-well plate compatible with the imaging system. Select a plate with the appropriate bottom thickness and geometry, per the manufacturer's specifications. Glass-bottom plates (130–170 µm bottom thickness) are recommended when higher image quality is required^13,14,15,16.
2. Staining procedure
   1. Ensure that cells form a single, well-dispersed monolayer and do not exceed 70% confluency to minimize cell overlap (Figure 1B).
   2. Group the samples according to the experimental requirements and proceed with staining after treatment. Prepare the working dye solutions with a final concentration of 6.684 µg/mL for PI and 10 µg/mL for Hoechst 33342^17,18.
   3. Discard the old medium from the 96-well plate and wash each well twice with 100 µL of PBS. Add the dye working solution and incubate at room temperature in the dark for 30 min for staining.
   4. Wash each well twice with 100 µL of PBS, then add DMEM.
      NOTE: Different experimental designs may have variations in sample loading and staining parameters.

2. Imaging specifications and parameter settings

1. Imaging parameter standards
   ​NOTE: This study demonstrates only the widefield imaging workflow for the instrument after PI and Hoechst 33342 staining.
   1. Power on the instrument and then start the computer. Open the analysis software provided with the device.
   2. Load the 96-well plate into the instrument. Select the field of view with the samples for imaging.
   3. In the Objective and Camera settings, set the Magnification to "20XPh1SPlanFluorELWDADM" (Nikon, S Plan Fluor, ELWD, 20X/0.45, ∞/0-2, WD 8.2-6.9, OFN22, Ph1, AMD, MRH48230), Camera binning to "2", and Acquisition Mode to "Widefield".
   4. In plate mode, select the corresponding parameters of the 96-well plate and adjust the plate parameters according to the product size data.
   5. Under Site Option, typically select 9 fields per well for dense imaging. In Autofocus options, select Enable laser-based focusing.
   6. Adjust the laser autofocus parameters under Autofocus, based on the average plate bottom thickness measured from at least three sample wells. Adjust bottom thickness, Bottom thickness max variation, Adjacent well max variation, Intra-well max variation, and Plate max variation accordingly.
   7. Under Wavelengths, set the Number of wavelengths to 3. For the first Illumination, select DAPI. In Shading Correction, choose "Auto Correction for FL" (Figure 2B).
   8. After finding the image, set the initial Exposure (ms) to 20–30. Observe the captured images and adjust the exposure continuously to achieve a moderate fluorescence intensity without overexposure, and set the bit depth to 16 (Figure 2A).
   9. Choose Laser with z-offset for focusing adjustment. If the focus is not clear, check Use Z stack, click Calculate Offset, and select an appropriate focal distance. Check Digital Confocal (info) and set the adjustment parameters for Increase Sharpness and Reduce Noise to 0.200.
      NOTE: If the image exposure or clarity is not satisfactory, adjust the exposure and recalculate the offset parameters.
   10. For the second Illumination, select Cy3. Follow the same settings as in steps 2.1.7-2.1.9.
   11. For the third Illumination, select TL 25. Follow the same settings as in steps 2.1.7-2.1.9.
       NOTE: Set Shading correction based on the channel, and TL 25 should choose Auto Correction for TL. In Wavelengths, the TL channel must be set for cell positioning and morphological observation.
   12. In the Run tab, name the plate according to the naming convention: plate name-cell type-experiment ID-date-objective magnification-imaging type. Confirm the imaging channel-related parameters under Description.
   13. Click Acquire Plate to start imaging. After imaging is complete, turn off the software, instrument, and computer in sequence.
       ​NOTE: Subsequently, the computer may be opened only for image export, data processing, and result analysis
2. Imaging operation standards and requirements
   1. Center the field of view on the well to avoid edge effects (Figure 1A). During imaging parameter setup, use the positive control with the highest fluorescence intensity as a reference. Additionally, when focusing, choose a field of view with a single layer of cells, clean and free of debris.
   2. For the camera, control the pixel values strictly below 60000 (the maximum value is 65535) to prevent saturation, which can cause permanent loss of information. If saturated bright spots or areas appear, shorten the exposure. At the same time, do not keep the exposure time too low; otherwise, noise will dominate the image.
   3. For weak fluorescence signals, prioritize slightly increasing the exposure rather than forcibly raising the parameters to avoid irreversible damage caused by photobleaching.
      NOTE: For bright-field imaging, these principles can be appropriately relaxed, but the core focus remains on signal integrity and sample protection.
   4. Ensure that the sample's edge contours are clearly distinguishable and make sure that the internal fine structures (such as granules, fibers, organelles) have distinct textures, not blurred, and the background scattered light is evenly distributed without local bright spots or blurred halos. Strictly avoid over-focusing and under-focusing.
   5. Ensure that there are normal, positive, negative, and experimental groups, and that each treatment group has at least 3 parallel wells. It is recommended to arrange the parallel wells vertically.

3. Image processing and analysis

NOTE: This document outlines three image processing methods provided by the instrument’s companion software, all of which are suitable for the staining protocol used in this study. While these methods share the same underlying principle of recognition, they differ in specific processing techniques and types of data results. The same image processing workflow was as follows: Open the file → Load the original image with the corresponding processing method → Set the appropriate channel parameters → Generate a Mask → Set the measurement data type in Measure → Run the analysis → Observe the results → Export the data.

1. Open the analysis software provided with the device → Review Plate → Select Plate.
2. In the Date view, select Well arrangement and choose a field of view from the positive group as a reference.
3. Under Run Analysis, select the corresponding processing program and set the image processing parameters or export the image.
4. After setting and saving the parameters, select Run on all wells/ Run on selections/ Run on displayed site based on the parameter setting object.
5. After the computer finishes processing the data, under the Measurements tab, select the corresponding data processing program in Analysis and choose different measurement results in Measurement to check the data.
6. Click Open Log to start exporting the data. Set the following details in the parameter settings for the three image processing methods:
   1. Image processing method one: Multi-Wavelength Cell Scoring
      1. Select the Multi-Wavelength Cell Scoring program and configure the settings. Set the Number of wavelengths. Choose Standard for the Algorithm.
      2. Set the recognition parameters for all nuclei based on the PI staining results. The default W1 Source image is the DAPI channel.
      3. Set the Approximate min width to 8–10 µm and the Approximate max width to 30–35 µm, based on the size of the individual objects to be recognized.
      4. Set the intensity above local background to 2,000–10,000 gray levels. Use Preview to test and adjust the settings based on the recognition results.
      5. Set the W2 parameters based on Hoechst 33342 staining results. Name this channel. The W2 Source image is Cy3
      6. In the Stained area, select the region stained by this channel as Nucleus. Follow the same settings as in steps 3.6.1.3–3.6.1.4 for subsequent settings. In Configure Summary Log or Configure Data Log (Cells), select the corresponding Parameter configuration based on the desired result parameters.
         NOTE: Typically, Configure Summary Log is used to set the required data types, while Configure Data Log (Cells) is used when special marking of individual cells is needed.
      7. Perform a Test Run to observe the recognition. If satisfied with the recognition results, save the settings.
   2. Image processing method two: Live-Dead
      1. Select the Live-Dead program and begin configuring the settings. Set the Wavelength 1 Parameters and Wavelength 2 Parameters sequentially.
      2. Under Wavelength 1 Parameters, set Wavelength 1 Source to DAPI. The default Algorithm is "Standard."
      3. Set Stained cell type to All Cells and Stained area to Nucleus based on the staining conditions. Set Approximate min width, Approximate max width, and intensity above local background using the same settings as the Multi-Wavelength Cell Scoring method.
      4. Check Split touching objects and use Preview to test.
      5. Under Wavelength 2 Parameters, set Wavelength 2 Source to Cy3.
      6. Set Stained cell type to Dead and Stained area to Nucleus based on the specific staining. Set Approximate min width, Approximate max width, and intensity above local background using the same settings as the Multi-Wavelength Cell Scoring method.
      7. Click on Configure Summary Log and select the Parameter configuration based on the required data types. Perform a Test Run to observe the recognition. If satisfied with the recognition results, save the settings.
   3. Image processing method three: developer mode CME
      1. Select the Custom Module (CME) program and begin configuring the settings. Set up Image Names and channels.
         NOTE: To facilitate subsequent parameter settings, generally set them to the channel names used during imaging.
      2. In Find Objects, open Simple Threshold to set the recognition for the entire set.
      3. Set Source to correspond with Image Name. Based on the 16-bit grayscale value range, fill in 0 at Threshold Low and 65535 at Threshold High.
      4. Name the processing result in Result, apply, and observe the marking situation. In Find Objects, open Find Blobs to generate a Mask result.
      5. Set Source to correspond with Image Name, as in Simple Threshold. Set Approximate min width, Approximate max width, and intensity above local background as per the Multi-Wavelength Cell Scoring settings.
      6. Name the processing result in Result, apply, and observe the marking result. For the Find Blobs recognition result, use the Modify Objects tool to filter, grow, shrink, and invert.
      7. Under Measure, set Modified. In Measurement Inputs, input 1 for Standard Area Value and check Create Object Overlay.
      8. In Objects to Measure, select the source of the marked objects in Mask of Objects and the source image to be measured in Image to Measure. Choose the corresponding Measurement Name based on the required processing result parameters, apply, and observe the parameter recognition result.
      9. After confirming the parameter settings, save the configuration.
   4. General method: ImageJ
      ​NOTE: Analyze the imaging results using the ImageJ method for fluorescent images¹⁹ and then perform a comparative analysis with the three methods of this instrument.
      1. Open the ImageJ software. Image Import: File → Open → Select the original grayscale image to be analyzed.
      2. Convert the image to 8-bit: Image → Type → 8-bit. Grayscale Value Recognition: Image → Adjust → Threshold → Default, Red → Check Dark background → Set → OK.
         NOTE: If the red area recognition is unsatisfactory, replace Default with one of the other 16 built-in ImageJ recognition models or adjust the slider in the dialog box.
      3. Set Parameters: Analyze → Set Measurements. In the pop-up dialog box, check Area, Mean gray value, Integrated density, Limit to threshold, Display label, Redirect to None by default, and input 2 for Decimal places (0-9). Click OK.
      4. Analyze Recognition Results: Analyze → Measure. Save Results: In the pop-up results dialog box, File → Save As, and select the save location.
7. Image export
   1. Select a high-quality positive well field of view as the reference standard.
      NOTE: The basic quality requirements are authenticity, accurate focus, suitable exposure (16-bit), and appropriate contrast (a higher signal-to-noise ratio is better). High-quality requirements include distinct features, contrast, practicality, intuitive morphological changes, and highlighted key points.
   2. Under Run Analysis, select ExportPro in the Analysis section, and start Run Setup for Analysis. In Select output options, choose Individual channel, Color overlay, shrink image (optional), and Scale bar (optional) based on the image requirements.
   3. Select the fluorescent channel images to be exported for each field of view. In Individual Channel, select a color for each channel, such as Blue for the DAPI channel and Monochrome for the TL channel.
   4. In Color overlay, select the channels corresponding to BRIGHTFIELD, RED, BLUE, and GREEN in order (Select the color wavelength for color overlay). Set the Number to 100; observe the preview image and accept the preview result.
   5. In output, select the folder where the exported images will be stored in Export to folder. Start the run.
      NOTE: Images from the same batch must not be exported separately; a batch of images refers to a collection of images from four core samples: negative control, positive control, basic control group, and target experimental group, all obtained under completely consistent experimental conditions.
8. Comparison of image processing methods
   1. Select 9 fields of view and compare the traditional manual counting of dead and live cells with the Multi-Wavelength Cell Scoring, Live-Dead, CME, and ImageJ methods for counting live cells, dead cells, and the ratio of dead to live cells. Use GraphPad Prism version 10.1 to perform a one-way ANOVA, accompanied by Tukey's multiple comparisons test, to analyze the differences.
   2. Select 81 fields of view and compare the differences in cell counting, area, and intensity among the Multi-Wavelength Cell Scoring, Live-Dead, CME, and ImageJ methods. Use GraphPad Prism version 10.1 to perform a one-way ANOVA, accompanied by Tukey's multiple comparisons test, to analyze the differences.
      NOTE: Cell culture and staining experiments must be conducted within a biosafety cabinet. Dispose of waste liquid, tips, and other experimental waste according to laboratory regulations. Ensure full protection throughout the experiment.

Results

Impact of standardized preparation on imaging quality

The plate parameters were configured to ensure that the instrument could reliably locate the cell image layer. Samples prepared as a single layer of cells with moderate confluence facilitated image focusing at different objective magnifications and improved image recognition and quantitative analysis. Standardized sample preparation and optimized plate parameters effectively minimized edge effects, cell overlap, and other factors contributing to poor-quality imaging (Figure 1).

Comparison of the image calculation method functions

The PI/Hoechst 33342 double-staining method was used to distinguish live and dead cells. The Multi-Wavelength Cell Scoring, Live-Dead, and CME methods successfully processed these images by identifying objects based on fluorescence intensity and object size, whereas ImageJ identified objects based on grayscale values. The marking results obtained with these methods were generally comparable (Figure 3A–E).

The Multi-Wavelength Cell Scoring method required nuclear staining to identify all cells (Figure 4A). It can process up to 7 imaging channels and calculate parameters including cell count, area, and optical density.

The Live-Dead method was more flexible than the Multi-Wavelength Cell Scoring method. Both channels provide modules for nuclear staining, cytoplasmic staining or combined nuclear-cytoplasmic staining to identify all cells, live cells, or dead cells. The Split touching objects function separates overlapping objects, allowing calculation of cell count, area, and optical density. However, because it processed only dual-channel data, it was particularly well suited for PI/Hoechst 33342 double-stained cell viability assays (Figure 4B).

The CME method depended on user-defined image-processing logic. The number of processed channels corresponded to the number of acquired channels. The recognition results were refined using functions such as Fill Holes, Invert Objects, Grow Objects, Shrink Objects, Logical Operations, Remove Border Objects, Keep Marked Objects, Grow Objects without Touching, Mark Object Centers, Watershed, Remove Marked Objects, and Filter Mask to eliminate impurities and other sources of interference. This method also enabled the calculation of multiple object parameters for multidimensional analysis.

ImageJ did not process multiple channels simultaneously. For image batches with substantial background variation, manual processing and threshold adjustment were required, increasing subjectivity and reducing reproducibility.

Analysis of differences in processing results

A comparative analysis of the image-processing methods showed no significant differences in total cell counts, dead cell counts, or the dead-to-all-cell ratio among the Multi-Wavelength Cell Scoring, Live-Dead, CME, ImageJ, and traditional manual counting methods across 9 fields of view (Figure 5A–C, Supplementary File 1).

Similarly, no significant difference was observed in the ratio of dead-cell area to total cell area among the Multi-Wavelength Cell Scoring, Live-Dead, CME, and ImageJ methods. However, a significant difference was observed in the ratio of the average fluorescence intensity of dead cells to that of all cells among the Multi-Wavelength Cell Scoring, Live-Dead, and CME methods based on data obtained from 81 fields of view (Figure 5D–F).

Based on these findings, a standardized workflow for the High-Content Imaging System was established (Figure 6).

Figure 1: Low-quality imaging results. (A) The field of view is close to the well edge, revealing shadows from the well edge (BV2 cells in this case). (B) Arrows a and b indicate the overlap of cells in the DAPI channel and TL channel, with excessive cell confluency, and even a second layer or some floating cells. (C) The field of view has no cells; the plate parameters are incorrect; the focus is not on the cell layer, or the cells are not stained, resulting in no image in this field of view. Scale bar = 100 µm. Please click here to view a larger version of this figure.

Figure 2: Imaging parameter settings. (A) Arrow a indicates the bit depth icon. (B) Arrow b indicates the position of Shading Correction. Please click here to view a larger version of this figure.

Figure 3: Results of the same field of view identified by several methods. (A) Multi-Wavelength Cell Scoring mask results: the gray and red areas represent all cell nuclei identified in the DAPI channel, with the red area representing dead-cell nuclei identified in the Cy3 channel. (B) Live-Dead mask results: the green and red areas represent all cell nuclei identified in the DAPI channel, with the red area corresponding to dead-cell nuclei identified in the Cy3 channel. (C) CME mask result: the yellow and blue areas represent all cell nuclei identified in the DAPI channel, with the blue area corresponding to dead-cell nuclei identified in the Cy3 channel. (D) ImageJ mask result in DAPI. (E) ImageJ mask result in Cy3. Scale bar = 100 µm. Please click here to view a larger version of this figure.

Figure 4: Partial parameter setting pages. (A) The location indicated by a shows that the first channel of Multi-Wavelength Cell Scoring can only select the nucleus; (B) b and c indicate that Live-Dead can only process two channels, supporting PI/Hoechst 33342 double staining for dead and live cells. Please click here to view a larger version of this figure.

Figure 5: Differences in fluorescence image processing results. (A) Differences in all cell counts among the five processing methods. (B) Differences in dead cell counts among the five processing methods. (C) Differences in the ratio of dead cells to all cells among the five processing methods. (D) Differences in the ratio of dead cells to all cells among the four processing methods. (E) Differences in the ratio of dead cell area to all cell area among the four processing methods. (F) Differences in the ratio of the average fluorescence intensity of dead cells to that of all cells across the four processing methods. All error bars in the charts represent standard deviations. Please click here to view a larger version of this figure.

Figure 6: Standardized operation process. Workflow summarizing the standardized procedure for sample preparation, fluorescence staining, high-content imaging, image processing, image export, and data analysis. Please click here to view a larger version of this figure.

Supplementary File 1. Raw data for the cell count, area, and fluorescence intensity analysis presented in Figure 5. Please click here to download this file.

Discussion

Using the PI and Hoechst double staining method to determine cell viability, for example, the high-content instrument can quickly and in a high-throughput manner calculate cell viability within the same batch using built-in algorithm modules. This method has several advantages, including low sample volume, low cost, minimal cell damage, weak fluorescence bleaching, resistance to quenching, and the ability to capture multiple images. Additionally, the imaging time for a 96-well plate is generally only 30–45 min²⁰.

In the experimental protocol, to ensure imaging quality, focus on the following steps: First, ensure appropriate plate density to avoid excessive cell density, which can fill the image with cells, or insufficient density, which can wash cells away during staining and washing. Second, thoroughly wash the dye to prevent residual dye from affecting imaging. For cells with low adhesion, handle the addition and removal of liquid gently and slowly before and after staining, avoiding vertical addition. Use a side-wall addition to prevent the center of the field of view from losing cells, paying special attention to the detachment of dead cells during viability assays. Consider centrifuging the sample plate to reduce experimental errors if necessary. Third, choose an appropriate exposure time. Too long an exposure will increase the total imaging time and may miss the detection window, potentially leading to overexposure of some cells or background being recognized. Too short an exposure may result in incomplete image information, losing some characteristic details.

If a clear image cannot be found during imaging, trace the issue back to cell condition, dye residue, parameter settings, and other factors. During the experiment, if there are no cells at the bottom of the well, no image will be found in the field of view. If two or more uneven cell layers form at the bottom of the well, focusing becomes difficult, resulting in a blurry image. Excessive dye residue can lead to a strong background, making the target image less prominent. Large areas of dye residue can obscure cells in the field of view, while small particles of residue, similar in size to cells, can interfere with subsequent image recognition and segmentation. If the plate height and well height are not set according to the manufacturer's specifications for the 96-well plate, no image will be found in the field of view, or the sample position may shift, potentially damaging the objective lens.

Considering the flexibility and simplicity of operation, as well as the good data processing results, the optimal image processing module for the PI and Hoechst double staining method is Live-Dead. This program is specifically optimized for PI and Hoechst dyes and, compared to other methods and software, is simple to operate, calculates quickly, and allows easy batch-loading of images into the database with consistent parameter settings. The results are more comparable than those obtained with the currently common software, ImageJ. Different experimental modes can use corresponding proprietary methods or the Multi-Wavelength Cell Scoring general calculation mode; proprietary methods are more convenient, while the general calculation mode has fewer restrictions on staining methods. Different staining protocols require different optimal image processing modules. Develop computational methods tailored to experimental objectives. The ExportPro method for image export is simple and easy to learn, allowing for previewing reference objects. It is beginner-friendly, with a graphical interface that clearly explains parameter meanings and simplifies measurements, breaking down barriers to multi-software analysis and saving time on tool learning.

This study only validated a basic staining protocol. As a reference, other experimental protocols that differ significantly from this one still need to be validated against the new standard.

High-Content Imaging System serves as a core technology platform that integrates high-resolution imaging, multi-parameter synchronous detection, automated quantitative analysis, and high-throughput screening^21,22,23. The integrated design of the device's imaging and calculation capabilities is expected to enable the development and establishment of corresponding experimental protocols and computational models for more existing imaging experiments^7,24, ultimately improving the consistency and comparability of data obtained by different operators, at different times, and across batches.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
1.5 mL Centrifuge Tube	Labigic Technology Co., Ltd.	MCT-001-150
1000 µL Pipette Tips	Labigic Technology Co., Ltd.	35423003E
15 mL Centrifuge Tube	Labigic Technology Co., Ltd.	CT-002-15A
2.5 µL Pipette Tips	Qingdao Haier Biomedical Co., Ltd.	B10T
200 µL Pipette Tips	Labigic Technology Co., Ltd.	T-001-200
50 mL Centrifuge Tube	Labigic Technology Co., Ltd.	CT-002-50A
96 Well Cell Culture Plate	Corning Incorporated	3599
Adjustable Single-Channel Pipette, 0.1–2.5 μL	Eppendorf SE	3120000216
Adjustable Single-Channel Pipette, 100–1000 μL	Eppendorf SE	3120000267
Adjustable Single-Channel Pipette, 20–200 μL	Eppendorf SE	3120000259
Adjustable Single-Channel Pipette, 2–20 μL	Eppendorf SE	3120000232
Cell Culture Dish	Corning Incorporated	430167
Cryogenic Vials	Labigic Technology Co., Ltd.	BS-20-ST
Dimethyl sulfoxide (DMSO)	Beijing Solarbio Science & Technology Co., Ltd.	D8371
Dulbecco's Modified Eagle Medium (DMEM)	Life Technologies Limited	11966-025	In this study, the complete DMEM was prepared using 10% FBS, 1% Penicillin-Streptomycin Solution, and 89% DMEM.
ESCO CelCulture CO2 Incubator	Esco Micro Pte. Ltd.	CCL-170B-8
Fetal Bovine Serum (FBS,Superfine)	Procell Life Science & Technology Co.,Ltd.	164210-50
Hoechst 33342 Fluorescent Dyes	Beijing Solarbio Science & Technology Co., Ltd.	B8040
HT22 cell	Procell Life Science & Technology Co.,Ltd.	CL-0697
ImageJ	National Institutes of Health		ImageJ is one of the image processing software used in the manuscript.
ImageXpress Micro Confocal High-Content Imaging System	Molecular Devices, LLC	ImageXpress Micro Confocal	In the manuscript, use "High-Content Imaging System" as a substitute for the instrument name.
Low speed freezing centrifuge	Hunan Xiangyi Laboratory Instrument Development Co., Ltd.	L530R
MetaXpress High-Content Image Acquisition and Analysis Software	Molecular Devices, LLC		The software is a companion to the ImageXpress® Micro Confocal High-Content Imaging System. In the manuscript, "The analysis software provided with the device" refers to this software.
OptiMair Vertical Laminar Flow Cabinet	Esco Micro Pte. Ltd.	ACB-4E1
Parafilm M	Pechiney Plastic Packaging Inc.	PM996
PBS	Shanghai Yuchun biology science and technology co., ltd	YC-5013
Penicillin-Streptomycin Solution	Procell Life Science & Technology Co.,Ltd.	PB180120
Propidium Iodide Solution (PI)	Beijing Solarbio Science & Technology Co., Ltd.	C0080
Trypsin-EDTA Solution	BasalMedia Technology Co.,Ltd.	S310KJ