Blue Image

Application Note

Automation of iPSC culture, passaging, and expansion with the CellXpress.ai Automated Cell Culture System

  • Automate and scale-up iPSC culture processes with a novel automated cell culture instrument
  • Monitor iPSC cultures while gaining quality control with scheduled imaging and advanced image analysis
  • Enable automated decision-making with machine learning-based algorithms

Download PDF

Introduction

iPSC-derived cell models are a popular tool enabling scientists to generate different cell types, organoids and tissues. Stem cells are also widely used to create disease- specific phenotypes with CRISPR gene editing to discover new drug targets using patient-derived samples. However, the process is often limited by labor-intensive and highly demanding culturing steps. To alleviate the limitations that come with labor-intensive protocols, we have developed the CellXpress.ai™ Automated Cell Culture System. This powerful new solution automates the entire cell culture process with an integrated incubator, liquid handler, and AI-powered, image-based decision-making. This hands- off system manages demanding feeding and passaging schedules by monitoring the development of cell cultures with periodic imaging and analysis, and leverages machine learning to initiate passaging, endpoint assay, or troubleshooting steps. In this application note, we present results from the automation of commonly used iPSC culture protocols, including cluster or single-cell passaging.

Methods

Materials

Automation of iPSC Workflow

The CellXpress.ai Automated Cell Culture System is a workcell for automated 2D and 3D cell culture. The system includes an automated imager, liquid handler, and incubator controlled by a unified AI powered software environment. The cell culture growth was monitored by periodic imaging and analysis. Automated decision- making was utilized to trigger passaging, and to notify the user of assay milestones or required intervention (e.g. refill consumables).

Cell culture protocols

IPSCs (ATCC 201B7) were cultured in Vitronectin (STEMCELL Technologies) in 6-well plates and passaged according to the STEMCELL Technologies protocols

for cluster or single-cell passaging. Here we present a description of cluster passaging method.1

Cell imaging and image analysis

Transmitted light images were acquired on the CellXpress.ai system using 4x or 2x magnifications. The CellXpress.ai system software was used for all analysis. Image analysis was done using pre-trained machine-learning protocols with IN Carta® Image Analysis Software.

Automation of iPSC culture

Automated workflows enabled the following steps:

CellXpress.ai system components and functionality

Figure 1. CellXpress.ai system components and functionality

Schematic diagram of automated iPSC culture

Figure 2. Schematic diagram of automated iPSC culture

Steps of the organoid culture and passaging protocol

Figure 3. Steps of the organoid culture and passaging protocol

Automated procedure for fragment passaging of iPSC

iPSC culture was performed according to the basic STEMCELL Technologies recommended protocol for human iPSC cells. The following steps can be optimized depending on iPSC and specific applications.

Results

iPSC culture and monitoring

The protocol was tested with two commercially available iPSC cell lines and we automated basic iPSC protocols recommended by STEMCELL Technologies. The protocol was set up for media exchange of mTeSR every 24 hours on the liquid handler automated stem iPSCs were monitored by imaging every 12h.

Deep learning-based image segmentation was used to detect iPSC colonies, determine the cell area and confluency, and detection of cell phenotype changes or differentiation. The CellXpress.ai system software includes user-defined “widgets” for data visualization and review. iPSC growth curve, measured as the Area Sum (or confluency) over the time can be plotted in the software (Figure 4). Area Sum is the sum of areas of all colonies in the well. Error bars represent variability between the wells in same plate. Passaging events are marked by arrows.

Images acquired during monitoring were analyzed and measurements such as confluency may be used for automated decision making. Passaging steps can be triggered either by user decision (manual activation of the Passage step) or automatically depending on how the “rules” were configured. Figure 5 shows the user interface where decision parameters can be set up. In this example, 70% of cell confluency for 100% of wells were used as the decision-making point for automated passaging.

Upon decision point, user may select 2 options: notify user (by email) or proceed to the next phase (if configured as a passaging phase). The presence or percentage of differentiated cell can also be quantified during the image analysis step using pre-trained AI models in the software. (Figure 6). We developed two deep learning models for the analysis of iPSC colonies. The first model identifies colonies that shows typical stem cell morphologies while the second model identifies areas of differentiation. Together, these metrics may be used to trigger passaging, or to disregard wells with differentiated cells, or evaluate the extent of differentiation.

The process described here overcomes the challenges of manual handling of iPSC cells and enables automated iPSC culture and iPSC expansion. Cell culture automation offers great potential to reduce labor costs and increase productivity, throughput, and reproducibility.

Stem cell colonies and graph showing total Area Sum or % confluency for iPSC culture in 6-well plate

Figure 4. Representative image of stem cell colonies and graph showing the total Area Sum or % confluency for iPSC culture in 6-well plate for two independent experiments. Passaging steps indicated with arrows. Error bars indicate SDEV between the wells.

Passaging triggered by rules based on selected measurements

Figure 5. Passaging triggered by rules based on selected measurements provided by image analysis. Decision-making software can select one or more criteria, combined by logical operations.

Pre-trained analysis models define stem cell colonies (in blue) and differentiated cell patches (in pink).

Figure 6. Pre-trained analysis models define stem cell colonies (in blue) and differentiated cell patches (in pink).

Summary

Reference

  1. https://cdn.stemcell.com/media/files/manual/10000007757-Maintenance_of_Human_Pluripotent_Stem_Cells_in_mTeSR_Plus.pdf

Download PDF