featuring Debanik Choudhury and Aditya Sonpal
Wednesday, May 11, 2022
Andreadis Lab Group
Mitochondrial dysfunction, one of the major aging hallmarks, has been associated with the onset of aging phenotypes and age-related diseases including Alzheimer’s, muscular dystrophy and diabetes. In this study, we found that impaired mitochondrial function was accompanied by increased glutamine catabolism in senescent/aged human mesenchymal stem cells (MSC) and myofibroblasts derived from patients, suffering from Hutchison-Gilford progeria syndrome. Increased glutaminolysis induced urea accumulation, which further impaired mitochondrial respiratory function and caused DNA damage. In agreement with our in-vitro findings, tissues isolated from aged and progeria mice (an accelerated aging mice model), displayed increased urea accumulation and glutaminase (GLS1) activity, concomitant with declined mitochondrial function. Interestingly, GLS1 expression and glutaminolysis were regulated through the JNK pathway that was activated in senescent cells. Blocking JNK or GLS1 activity decreased urea accumulation, mitochondrial dysfunction and DNA damage in senescent MSCs. Furthermore, we examined the effects of CB-839 (GLS1 inhibitor) on progeria mice and observed decreased age-associated ROS and urea accumulation in the heart, skin and muscle, resulting in improved mitochondrial function. Hence, inhibition of GLS1 activity rejuvenated mitochondrial function and led to amelioration of aging hallmarks. In conclusion, our data provide novel insight into the mechanism, underlying metabolic reprogramming associated with cellular aging. Targeting glutamine metabolism and associated metabolites and byproducts may be a promising strategy to delay or reverse senescence.
Debanik Choudhury received his B.S. degree from Jadavpur University, India. He is currently a doctoral candidate in Dr. Stelios Andreadis’ Lab in the Department of Chemical and Biological Engineering (CBE) at the University at Buffalo (SUNY). His research focuses on the intersection of aging and metabolism to develop novel drug targets for amelioration of age-related disorders. Specifically, he is working on the role of the amino acids glutamine and proline in regulating mitochondrial function in mammalian age associated pathophysiology. To date, Debanik has co-authored four published journal manuscripts, with several more in review. He has won the AGE Travel Award from the American Aging Association (2022) and best research poster award at the CBE Graduate Research Symposium (2017). He is also currently serving as the Vice-President for CBE Graduate Student Association (GSA).
Hachmann Research Group
Powerful deep learning methods (e.g., deep neural networks or DNNs) are at the heart of data-driven chemical and materials discovery. However, these methods (especially DNNs) are inscrutable. Their opacity has rendered them black-box models and hampered the community’s trust in them. ‘eXplainable Artificial Intelligence’ (XAI) is a new branch of data science concerned with explaining black-box machine learning (ML) and AI models and their results. Therefore, we benchmark three inherently different XAI methods, namely Local Interpretable Model-agnostic Explanations (LIME), deep SHapley Additive feature Explanations (deepSHAP), and Layer-wise Relevance Propagation (LRP). We use these methods to explain the predictions of our DNNs for optical properties of small organic molecules. We validate their results and explore the uses and applications of such XAI methods to improve the case for DNNs in the chemical sciences. We qualitatively and quantitatively compare these methods and have created working implementations of these in our group’s ChemML software. Explaining the results of ML will add scientific value, improve trust in the community, and deepen our understanding of chemical behavior.
Aditya Sonpal is a Ph.D. Candidate in Dr. Johannes Hachmann’s research group at UB CBE. His research lies in the domain of chemical and materials discovery using a combination of molecular modeling, machine learning, and informatics. He believes in constantly pushing the boundaries to achieve higher levels of automation while implementing these methodologies. In addition to his research in deep eutectic solvents and ionic liquids, he has contributed to developing the Hachmann research group’s software ecosystem to democratize the use of these techniques in the molecular sciences. He has also developed explainable artificial intelligence (XAI) methods to open up the black box of deep neural networks and strengthen the case for using AI and ML techniques in the chemistry community. He has received his M.S. In Chemical and Biological Engineering from UB in 2018, and his B.E. in Chemical Engineering from Visveswaraya Technological University (Bengaluru, India) in 2016.