Using causal aging biomarkers and protein design to develop novel anti-aging interventions

Project Summary In mammals, 5-methylcytosine is the most common form of DNA methylation, and the level of methylation of some specific CpG sites shows a strong correlation with age. These correlations can be used to build machine learning-based models that can accurately predict the age of biological samples. Because these models can quantify age with very high accuracy, researchers have termed them epigenetic aging clocks (e.g., Horvath’s pan-tissue epigenetic clock and Hannum’s blood-based epigenetic clock). However, the reliability of existing epigenetic clocks is limited, as they are built based on pure correlations, and it is unclear whether age- associated methylation changes are causal to aging-related phenotypes. A new generation of epigenetic clocks built on causal information will be more reliable and can enable the possibility of large-scale screening of anti-aging interventions. For the F99 phase of this proposal, I performed epigenome-wide Mendelian Randomization to identify CpGs potentially causal to aging-related traits. This causal information was then incorporated into epigenetic clock models to build causality-informed aging clocks, which are shown to separate age-related damage from adaptation, namely DamAge and AdaptAge. I also built ClockBase, a database that contains over 300,000 experimental samples from GEO with the epigenetic age pre-calculated. I plan to further standardize the sample information using large language models and apply the causality-informed biomarkers to screen for anti-aging interventions. In the K00 phase, I will use the protein language model and protein design tool to expand the universe of anti- aging interventions. Specifically, I will study the protein structural features across mammalian species with various lifespans to understand which features are associated with longevity. Then, I can incorporate this information into protein design and optimize existing proteins to support a longer lifespan. This proposal will advance our understanding of the molecular mechanisms underlying aging by incorporating causality into epigenetic clock models. By distinguishing between age-related damage and adaptation, we can develop more precise and informative aging biomarkers, which will have significant implications for aging research and potential clinical applications. The K00 phase of the project will pioneer the application of protein language models and protein design tools in aging research. Ultimately, it could pave the way for a completely new branch of aging research – treating aging through the gradual redesign of the proteome.