Zebrafish in 96-well plate for AI-enhanced high content microscopy — Sainsbury Laboratory Cambridge University

User Story

AI-Enhanced Microscopy Workflow Expands Possibilities in Shigella Infection Research Using Zebrafish

byZEISS Microscopy 23 May 2024 4 min read

User Story

AI-Enhanced Microscopy Workflow Opens New Possibilities in Shigella Infection Research Using Zebrafish

Using the ZEISS arivis platform, an AI-model automatically identifies and captures images of zebrafish larvae using ZEISS Celldiscoverer 7 for high-throughput and high-content imaging.

Shigella is a widespread human bacterial pathogen for which there is no effective vaccine. Dr. Serge Mostowy heads a research team in the Department of Infection Biology at the London School of Hygiene & Tropical Medicine, United Kingdom. They have successfully developed a zebrafish model to study the cell biology of Shigella infection with the goal of developing therapeutic strategies to combat disease. Using this zebrafish model, Dr. Mostowy and his group have a growing list of published work on Shigella infection; however, much of their microscopy-based experiments were extensive and time consuming, often limiting the range of research questions they could explore.

The team has now developed an automated microscopy acquisition workflow using the ZEISS arivis Cloud to design and train an AI-model which automatically identifies and captures images of zebrafish larvae in 96-well plates with the ZEISS Celldiscoverer 7 high-content imaging system. Larvae whole body measurements and/or specific regions of interest can be automatically analyzed and reported using ZEISS ZEN software. As one example, this new workflow was used in a study to monitor the growth and development of zebrafish mutant lines missing key septin proteins. The septins are a cytoskeletal protein family that Dr. Mostowy and his team believe could be exploited for host-directed therapies and Shigella infection control.

The Mostowy Lab and ZEISS Solutions Lab

Third-party Content Blocked

The video player is blocked due to your tracking preferences. To change the settings and play the video, please click the button below and consent to use of "Functional" tracking technologies.

Although zebrafish are optically accessible for in vivo imaging of Shigella infection, they are still whole animals. This makes studies involving high-throughput and high-resolution microscopy challenging and time-consuming. With Celldiscoverer 7 and the arivis AI-enhanced workflow, we can image and analyze a lot more samples at a much faster pace.

Dr. Serge Mostowy Principal Investigator and Research Team Lead at the Department of Infection Biology, London School of Hygiene & Tropical Medicine, United Kingdom (See Footnote 1)

Overview of the AI-enhanced automated imaging workflow for high content microscopy enabled by ZEISS arivis and ZEISS Celldiscoverer 7 — Overview of the AI-enhanced automated imaging workflow for high-content microscopy enabled by ZEISS arivis and ZEISS Celldiscoverer 7. (A) A subset of the fast overview imaging of zebrafish larvae in 96-well plates. (B) AI model identifies zebrafish larvae and/or regions of interest at different developmental stages (top: 3 days post-fertilization, bottom: 1 day post-fertilization). (C) Further detailed ROI imaging can include multi-channels, Z-stacks, and/or time series.

AI-Enhanced Microscopy Workflow

High-Content & High-Throughput

Manual imaging of samples can be both user- and time-intensive. As a result, experiments from Dr. Mostowy's team were typically limited to 3 to 20 zebrafish larvae.

To efficiently increase their dataset volume, anaesthetized and agarose-embedded zebrafish larvae were imaged in 96-well plates using ZEISS Celldiscoverer 7. Data acquisition was automated using the ZEISS arivis Cloud to design and train an AI model to identify zebrafish larvae. This model was then used within the ZEISS ZEN software to automatically image and outline the larvae and/or regions of interest in the acquired image. Once completed, new time points could be acquired by either using a customized, executable trigger created by the ZEISS Solutions lab or the 96-well plate could be removed and put back into the incubator to allow the larvae to further develop. Additional parameters such as Z-stacks, multi-channels, and/or higher-resolution images could be programmed into the acquisition.

Using this workflow, high-throughput imaging with datasets expanded to 96 larvae along with access to high-content imaging enabled Dr. Mostowy and his team to create more complex experimental designs (e.g. 12 larvae imaged across 8 different conditions).

With these algorithms and workflows in place, I was able to perform quantitative imaging of zebrafish using high-throughput and high-content microscopy. The interface is user-friendly and enables automated, nonbiased data acquisition and analysis.

Dr. Margarida C. Gomes Postdoctoral Researcher, Mostowy Lab, Department of Infection Biology, London School of Hygiene & Tropical Medicine, United Kingdom (See Footnote 1)

Dr. Serge Mostowy and his team with pages with their publication using high throughput microscopy to screen zebrafish null mutants. — Dr. Serge Mostowy and his team (See Footnote 1) with pages from their publication using high-throughput microscopy to screen zebrafish mutants.

High-Content Microscopy Screening for Zebrafish Transgenic Lines

Monitoring Growth & Development

In V. Torraca et al., Dr. Mostowy and his team used the AI-assisted microscopy workflow to monitor zebrafish growth and development.

The research team is studying the septin cytoskeletal protein family as a means of developing host-directed therapies to control Shigella. In this paper, they developed zebrafish transgenic lines that are missing key septin proteins. The workflow enabled them to monitor growth and development of these mutant larvae to test the role of septin proteins in their development as compared to controls.

This was done by imaging anaesthetized larvae in 96-well plates. The AI model had been trained to automatically identify and capture images of zebrafish at different developmental stages. Larvae could be imaged, then placed back into an incubator, and imaged again days later. Body measurements were analyzed automatically to determine growth differences.

Read the full article.

The AI-enhanced workflow for high-content and high-throughput microscopy has transformed our research ambitions using the zebrafish model. Our lab is moving in many exciting new directions because of this technology.

Dr. Serge Mostowy Principal Investigator and Research Team Lead at the Department of Infection Biology, London School of Hygiene & Tropical Medicine, United Kingdom (See Footnote 1)

Aquariums hold zebrafish for high content, high throughput microscopy — See Footnote 1.

New Possibilities Unfold with AI-Enhanced Microscopy Workflow

From In Vivo Single Cell Tracking to Drug and Chemical Compound Screens

Dr. Mostowy currently shares three new areas of exploration in his lab:

Individual Infection Events: Training the microscope to capture infection events at the level of the single cell and tracking the infection in vivo over time to understand its consequence at the whole animal level.
Drug Screens & Infectious Disease Research: Using their workflow to screen drugs, particularly as Shigella is on the WHO list for priority pathogens that are multidrug resistant. Extending their zebrafish model to other bacterial species and moving into new areas of infectious disease research.
Compound Screens & Zebrafish: Testing the role of novel compounds that could affect zebrafish development and/or immunity by expanding their AI-enhanced workflow to identify and classify phenotypes automatically.

¹
All photography of the laboratory and lab personnel are courtesy of James Sykes, London School of Hygiene & Tropical Medicine, United Kingdom.

Share this article

AI-Enhanced Microscopy Workflow Expands Possibilities in Shigella Infection Research Using Zebrafish

AI-Enhanced Microscopy Workflow Opens New Possibilities in Shigella Infection Research Using Zebrafish