Z-Flipons: How Specific DNA Regions Help Regulate Gene Function

© iStock

Researchers at HSE University and InsideOutBio have applied machine learning to identify the location and functions of mirror-twisted DNA structures, known as Z-flipons, in human and mouse genomes. The scientists discovered which Z-DNA regions were conserved in both species throughout evolution and demonstrated for the first time that Z-DNA accelerates the process of creating RNA copies of genes. The findings will contribute to the development of new treatments for genetic diseases. The study has been published in Scientific Reports.

The structure of a DNA molecule—a double helix resembling a spiral staircase—might be familiar to many from their high school days. The 'steps' of this staircase are made up of pairs of nitrogenous bases, while the 'railings' consist of alternating chains of sugar and phosphate groups. Typically, the DNA strand twists to the right, but certain regions can temporarily twist to the left, playing a role in regulating gene activity. Due to their resemblance to the letter Z, these regions are referred to as Z-flipons. 

A team of researchers from the International Laboratory of Bioinformatics at the AI and Digital Science Institute of the HSE Faculty of Computer Science and InsideOutBio analysed the human and mouse genomes to predict the locations of Z-flipons and determine their functions. To achieve this, the scientists examined whether the Z-DNA segment is conserved across different species throughout evolution; if the segment remains unchanged, it indicates its importance to the organism's function and survival.

The researchers used the previously developed DeepZ deep learning system that considered not only information from the linear DNA sequence but also data from tens of thousands of Omix experiments. This included, for example, information about epigenetic tags—chemical markers on DNA or proteins that help regulate gene activity without altering the DNA structure itself. Additionally, the scientists incorporated data on the transition energy required for a DNA region to alter its structure. Two machine learning models were developed using this data: one for humans and one for mice. Afterward, the trained model 'scanned' the entire genome, identifying areas with a high probability of Z-DNA regions. The model predictions were compared, and then the regions conserved in both the human and mouse genomes were identified. 

The researchers were able to structure data on the locations of Z-flipons in the mouse and human genomes and identify the genes in which they are found. The scientists have demonstrated that Z-flipons are conserved elements which are shared by different organisms and persist throughout evolution. By applying clustering, the researchers discovered that Z-flipons are grouped by function: some are involved in transcription regulation, while others play a role in the formation of chromatin—the 'packaging' of DNA within the cell. This confirms that Omix features do indeed determine the functional class of Z-flipons, which is crucial for understanding their role.

Also, for the first time, the scientists discovered and statistically demonstrated that Z-flipons accelerate the initiation of transcription, the process of creating RNA copies of genes. This feature enables cells to adapt more quickly to changes, which is particularly important for genes involved in the development of the nervous system and in other critical processes.  

Maria Poptsova

'To create a copy of a gene, RNA polymerase must bind to a specific DNA region and produce an RNA copy. If many copies are needed, multiple "photocopying machines" attach to the region simultaneously. However, the mechanism is slightly different: instead of repeatedly copying one page, there is a single "book"—a DNA sequence—to be copied. Small "photocopying machines" are deployed along the "book," each moving along the DNA and creating a copy. To produce more copies, it is essential for each new "copying machine" to attach immediately after the previous one finishes its work. The frequency at which new copies are initiated is called the rate of transcription initiation,' explains Maria Poptsova, co-author of the paper and Head of the International Laboratory of Bioinformatics of the HSE Faculty of Computer Science. 

The team at the International Laboratory of Bioinformatics has developed a website that hosts machine learning-based algorithms for data analysis, as well as whole-genome annotations—detailed information about the functional elements of the genome.