Searchable abstracts of presentations at key conferences in endocrinology
Endocrine Abstracts (2023) 90 P549 | DOI: 10.1530/endoabs.90.P549

ECE2023 Poster Presentations Adrenal and Cardiovascular Endocrinology (72 abstracts)

Differentiation of Pluripotent Stem Cells into Steroidogenic Cells with the Application of Artificial Intelligence

Ioannis Oikonomakos 1,2,3 , Melissa Sanabria 1,4 , Yasmine Neirijnck 3 , Stefan Bornstein 1,2 , Andreas Schedl 3 & Charlotte Steenblock 1,2


1University of Technology, Dresden, Germany, 2University Hospital Carl Gustav Carus, Dresden, Germany, 3Université Côte d’Azur – Campus Valrose, IBV, Nice, France, 4Center for Molecular and Cellular Bioengineering, Biotechnology Center, Dresden, Germany


Adrenal insufficiency is a life-threatening condition in which the adrenal glands fail to produce adequate amounts of steroid hormones thus leading to severe disturbances of body homeostasis. Today’s treatment options are limited to hormone replacement therapies that are, however, hampered by serious side effects. Cell replacement therapies with adrenocortical stem cells could present a potential cure, but the culture of such stem cells has proven difficult. We have developed an in vitro protocol for the differentiation of mESCs into adrenocortical steroidogenic cells. Our protocol recapitulates key steps in embryonic development by fating the cells towards the adrenal primordium. Molecular analysis demonstrated upregulation of fetal adrenal markers and the expression of genes coding for adrenal cortex specific steroidogenic enzymes (e.g. Cyp21A1). Indeed, steroidomics profiling revealed the secretion of adrenal cortex specific steroid hormones into the culture medium. The proportion of steroidogenic cells produced in our differentiation protocol remained, however, relatively low for therapeutic purposes. Machine Learning (ML) has been shown to be a powerful tool for improving the interpretation of biological data (1,2). To improve culture conditions and increase the yield of properly differentiated cells, we aim to apply ML methods to our in vitro differentiation protocol. In pilot studies, a VGG-16 convolutional neural network followed by random forest classification was employed to distinguish phase contrast images from different stages of differentiating cells based on their morphology. Indeed, we were able to distinguish cells based on the treatment they had received and to predict their differentiation. We are currently planning to optimize our in vitro differentiation protocol by taking into consideration the input from the ML model. In the future, we will incorporate more molecular data to increase our prediction accuracy (3). Once high yields of differentiated adrenocortical cells have been obtained, they will be tested for long term engraftment and hormone production in vivo (transplantation experiments). Taken together, we have generated glucocorticoid producing adrenocortical cells from pluripotent stem cells in vitro. In the long run, our protocols may pave the way for the development of cell replacement therapies for patients suffering from adrenal insufficiencies.

Reference

(1) Guo et al., Stem Cell Reports. 2021;16(5):1331–46. (2) Rostam et al., Sci Rep. 2017;7(1):1–11. (3) Cheng et al., Nat Commun. 2021;12(1):1–15.

Volume 90

25th European Congress of Endocrinology

Istanbul, Turkey
13 May 2023 - 16 May 2023

European Society of Endocrinology 

Browse other volumes

Article tools

My recent searches

No recent searches.