Although cells of early embryos are essentially identical, as the embryo develops, they can turn into nerve, muscle, skin or other types of cells depending on what genes are turned on and off. Proteins called transcription factors help control which genes are used by a cell. There are many types of these proteins, but even families of closely related transcription factors can have very different functions. To better understand the nuances of how transcription factors in the same family regulate gene expression during development, a team of researchers from Penn State and New York University has received a $3.8 million grant from the National Institutes of Health's National Institute of Neurological Disorders and Stroke.
The team is led by co-principal investigators Shaun Mahony, professor of biochemistry and molecular biology in the Penn State Eberly College of Science; and Esteban Mazzoni, professor of cell biology and of neuroscience, and Timothee Lionnet, associate professor of cell biology, both in the New York University Grossman School of Medicine.
"Our team uniquely brings together expertise in genomics, stem-cell differentiation, super-resolution microscopy, and artificial intelligence and machine learning techniques," Mahony said. "This will allow us to figure out how a closely related set of transcription factors control the development of different types of neurons along the spinal column, which will provide important information about basic biology as well as associated developmental disorders. The new techniques we develop could also be extended to the many other families of transcription factors working in the human body."
Transcription factors are critical to the "cell differentiation" process where stem cells develop into specific cell types. These proteins bind to specific sequences of DNA, usually near genes, where they can then regulate that gene's expression, turning it on and off or ramping up expression.
The team will study a group of transcription factors called homeodomain transcription factors that are coded by Hox genes - highly conserved genes across the animal kingdom that provide a blueprint for an animal's body plan, from head to tail, including the spinal column and where limbs will develop. Specifically, they will investigate transcription factors that pattern the spinal cord and other nerves in mammals.
"Although transcription factors in the same family tend to bind to very similar DNA sequences, they can serve different functions," Mahony said. "For example, some of these homeodomain transcription factors produce spinal motor neuron cells that connect to your limbs, while others produce spinal motor neurons that connect to your body wall. We're interested in how different motor neurons become different motor neurons, and how these closely related transcription factors help make that happen."
Some regions of the genome are easily accessible for transcription factors, while in some regions, the DNA is bundled up in tight, hard-to-access configurations. But some transcription factors, Mahony said, can get into those closed regions and even prefer to bind there. To confirm where different transcription factors are binding, the team will use a technique called chip-seq, which maps specific spots on the genome where a protein of interest binds. They will also use high-resolution imaging techniques to identify exactly which parts of the protein bind to the DNA and explore the biochemistry to understand what is responsible for the protein's preference to bind to open or closed regions of the genome.
"We hope to really drill into what's guiding these transcription factors to bind to different sequences," he said. "We are also looking at how these proteins interact and work together and exploring the consequences of these differences during development."
The team will also explore common mutations in these Hox genes that are associated with developmental defects, such as syndactyly - which produces fused fingers or toes - and hand, foot and genital syndrome - which produces defects in the limbs and genitals. To identify the role of transcription factors in these disorders, the team will explore how these mutations impact the biochemistry and activity of transcription factors.
Additionally, the team will build artificial intelligence (AI) neural networks to better understand the biological processes at play. They will train these models on chip-seq binding data, protein information and other information about the cellular environment so that it can more accurately predict where a transcription factor might bind in different types of cells.
"You can train these models to pretty accurately reproduce the binding signal we would expect when you input a genetic sequence," Mahony said. "So, for example, we could use it tell us how a mutation in a patient's genome might affect binding of a specific transcription factor. Although we have experimental methods to produce these results ourselves - like the training data - these techniques can be expensive or laborious. But we also want to take this a step further, so the machine learning model we are developing is interpretable, meaning we can get at how the model is making that prediction; it's not a black box like some AI models. We hope that understanding how the model works might provide new insights into the underlying biology."
Eventually, this type of information could be used to put together instruction manuals for how stem cells become specific cell types, Mahony explained, which could allow researchers to develop programming techniques to turn one cell type into another. This could support regenerative medicine by allowing researchers to regenerate missing or damaged cells by converting cells into the missing type. Additionally, the new techniques the team develops could be extended to better understand other transcription factors.
"This system can tell us something about why transcription factors bind where they do," Mahony said. "We have 1,600 different transcription factors in our genome, many of them in families with similar binding patterns, so the approaches and techniques we use here could also be adapted to understand gene expression involved in a myriad of other biological processes."
