How DNA markers influence gene expression in different cells and tissues

A new study, led by Dr Sarah Marzi (UK DRI at King’s) and Alan Murphy (UK DRI at Imperial), reveals key insight into how gene expression is regulated in health and disease. The findings, published in the journal Nucleic Acids Research (NAR), could be applied to identify influences on gene expression in neurodegenerative conditions, helping identify new drug targets.

What was the challenge? 

Our genes, sections of DNA that code for different proteins, have a set of switches and levers that tell them how to express themselves, known as epigenetics. As part of this system, there are markers on our DNA called ‘histone marks’, which are known to influence gene activity and how different genes are expressed. 

To understand the complex relationship between histone marks and gene expression, recent studies have used advanced data models to predict their interactions. However, these methods have missed important contributing factors such as cell states, the function of histone marks, and how they act at a distance from the genes they affect. 

What did the team do and what did they find? 

Dr Marzi and her team carried out the most comprehensive study of histone marks to date. They examined seven different types of histone marks across 11 types of cells, using advanced tools, including deep learning models, to predict how histone marks influence gene activity in different parts of the genome. 

The researchers found that histone mark function, how far they are from genes, and the condition of cells, shape the impact of histone marks on gene expression. Importantly, no single histone mark could consistently predict gene expression, these contributing factors were all shown to play a role.

What is the impact? 

This research could be used to help identify key regions and mechanisms in the genome that influence gene expression in brain cells, including those affected in neurodegenerative conditions. By pinpointing histone mark functions and their roles in gene regulation, researchers may uncover the earliest changes in disease or targets for drug development, improving our understanding of disease progression.

First author, PhD student Alan Murphy, said: 

“Our study represents a major step forward in understanding the relationship between histone marks, key regulators of gene activity, and gene expression in different cell states. The use of computational techniques to simulate changes in histone activity opens the door to identifying disease-related genetic regions and biological insights, showcasing the power of AI in advancing research.”

Dr Marzi added: 

“By using deep learning models, we’ve shown that histone mark functions, their genomic context, and cell states together shape their impact on transcription. In the long term, insights gained from this research can guide the development of precision medicine approaches for neurodegenerative diseases. By leveraging these deep learning models, scientists can better understand how cells regulate expression, which will help us design cell type-specific treatments that modulate disease-causing signatures to counteract neurodegeneration.”

Reference: Alan E Murphy, Aydan Askarova, Boris Lenhard, Nathan G Skene, Sarah J Marzi, Predicting gene expression from histone marks using chromatin deep learning models depends on histone mark function, regulatory distance and cellular states, Nucleic Acids Research, 2024;, gkae1212, https://doi.org/10.1093/nar/gkae1212