Automated STED microscopy for cell-biological high-throughput assays

by Sebastian Bierbaum
Doctoral thesis
Date of Examination:2022-10-17
Date of issue:2022-11-28
Advisor:Prof. Dr. Stefan Jakobs
Referee:Prof. Dr. Stefan Jakobs
Referee:Prof. Dr. Andreas Janshoff
Referee:Prof. Dr. Stefan Hell
Referee:Prof. Dr. Sarah Köster
Referee: Prof. Dr. Apl. Alexander Apl. Egner
Referee:Dr. Helmut Berg
Persistent Address: http://dx.doi.org/10.53846/goediss-9583

 

 

Files in this item

Name:Dissertation_SebastianBierbaum.pdf
Size:6.33Mb
Format:PDF

The following license files are associated with this item:

Abstract

English

In the past decade, super-resolution fluorescence microscopy has revolutionized biological research by enabling the study of biological functions down to the molecular scale. Despite profound technological developments, the image acquisition process is, however, still a mainly manual task. Robust automated microscopy techniques and streamlined acquisition schemes will be key to increase the throughput of super-resolution methods to meet the requirements for industrial applications, e.g., in pharmaceutical drug screenings. In this thesis, I established an automated stimulated emission depletion (STED) nanoscopy platform with special attention to the attainable imaging speeds. By implementing camera-based widefield imaging for both cell and feature detection, the throughput of the imaging platform could be considerably increased compared to an approach based solely on laser scanning. A use case from current pharmaceutical research addressing centrosomal clusters in cancer cells was successfully optimized for STED microscopy. Mitotic cells of interest and the centrosomes they contain were identified and localized by means of machine learning and classical image analyses. Extracted widefield feature coordinates were then reproducibly transformed into the confocal image coordinate system for subsequent STED image acquisitions. The feasibility of the proposed acquisition scheme was demonstrated in first automated proof of concept measurements with ten potentially active ingredients. STED images of individual centrosomes revealed the centriole orientation and protein localization with superior accuracy. Automated STED microscopy thus offers the possibility to observe drug-mediated responses with unprecedented detail and can be a valuable additional decision tool for refined mode of action elucidation and target validation in future drug screening campaigns.
Keywords: automated microscopy; super-resolution fluorescence microscopy; STED nanoscopy; cell-based assay; drug screening; centrosomal clustering; machine learning; deep learning; image analysis
 
Statistik