Embedded Files
Background
Biological aging is defined as the gradual accumulation of damage to various cells and tissues within the body, ultimately resulting in a loss in cellular/tissue function and death. Aging is a leading risk factor for major diseases like cancer, metabolic syndrome, cardiovascular disease, and neurodegenerative diseases. A key hallmark of cellular aging is a gradual decline in cellular proteostasis (protein homeostasis), or cellular processes that maintain proper protein folding, localization, and degradation. Age-related losses in proteostasis often result in the accumulation of mis-folded proteins leading to the formation of insoluble fibrils and protein aggregates which ultimately manifest in neurodegeneration as observed in Alzheimer’s, Parkinson’s, and Huntington’s disease. Current research seeks to identify how and why proteostasis declines with age to develop effective treatments against neurodegeneration and aging.
Project Overview:
Using cellular image data of young and old cells to identify drugs that restore proteostasis in aged-cells and delay aging.
High-Content Image Analysis of Drug Perturbations
With the rise of automated microscopes, image-analysis software, and machine learning, high-content phenomics (studying the traits that make up an organisms phenotype) has served as a powerful research tool for analyzing cell health readouts and biological responses to small-molecule treatments (drugs). We extracted relevant cellular morphological features from images of different aged cells so that we can build predictive models of cellular aging and proteostasis decline. We then used these models to screen a small molecule library of proteostasis-regulator compounds to identify drugs that delay aging. My research was focused on building custom processing and analysis pipelines for high-content imaging and drug discovery platforms.
Key Findings
We applied image-based profiling using a Cell Painting assay to evaluate the effects of a diverse small-molecule library on proteostasis in human dermal fibroblasts. Over 4,000 morphological features were extracted per cell and analyzed using cosine similarity and mean average precision to identify compounds with significant phenotypic activity.
We trained two machine learning models: one to predict cellular age based on morphology, and another to predict protein misfolding using fluorescent readouts from NTPAN-MI, a reporter of unfolded proteins. These models linked both aging and neurodegenerative diseases with increased protein misfolding and distinct morphological signatures. Several compounds were identified that reduce predicted cellular age or proteome instability, highlighting potential avenues for therapeutic intervention.
Together, these findings demonstrate that image-based profiling can effectively assess proteostasis and aging phenotypes, enabling the identification of small molecules that may reverse age-associated cellular decline. These insights open new avenues for promoting healthy aging and improving quality of life for older adults by addressing age-related neurological decline.
Learn more by reading my final research paper and going through my colloquium presentation below!
My Final Paper and Presentation:
My Mentor
I worked with Dr. Brian Hodge, a research specialist in the Gestwicki Lab at UCSF’s Institute for Neurodegenerative Diseases. His work focuses on investigating the interplay of environmental and genetic factors in aging and longevity.
Page updated
Report abuse