Author ORCID Identifier

https://orcid.org/0000-0001-9854-5823

Date of Award

Winter 12-15-2023

Document Type

Thesis (Ph.D.)

Department or Program

Quantitative Biomedical Sciences

First Advisor

Lucas Salas

Abstract

DNA methylation is an epigenetic modification that regulates gene expression and is essential to establishing and preserving cellular identity. Genome-wide DNA methylation arrays provide a standardized and cost-effective approach to measuring DNA methylation. When combined with a cell-type reference library, DNA methylation measures allow the assessment of underlying cell-type proportions in heterogeneous mixtures. This approach, known as DNA methylation deconvolution or methylation cytometry, offers a standardized and cost-effective method for evaluating cell-type proportions. While this approach has succeeded in discerning cell types in various human tissues like blood, brain, tumors, skin, breast, and buccal swabs, the existing methods have major limitations in accuracy and number of cell types deconvolved in brain and tumor due to their cellular complexity and heterogeneity in the respective tissue types. Here, I present novel reference-based methods using a hierarchical modeling approach with differential DNA methylation patterns for deconvolving 17 cell types in TME, named Hierarchical Tumor Immune Microenvironment Deconvolution (HiTIMED), and 7 cell types, named Hierarchical Brain Extended Deconvolution (HiBED), in human brain tissues. Furthermore, I demonstrated cell composition alterations in blood, brain, and buccal swab samples in the Down syndrome population compared to the euploidy control population across the lifespan using methylation cytometry. Beyond these advances, I leverage machine-learning algorithms and differential methylation patterns to develop a novel tumor-type-specific classifier named "Hierarchical Tumor Artificial Intelligence Classifier" (HiTAIC). This algorithm accurately identifies tissue of origin and categorizes 27 different types of cancers, holding potentially significant clinical implications for diagnosing cancers of unknown origin. In summary, my thesis significantly advances methylation cytometry by introducing novel deconvolution techniques applicable to TME, brain, and tumor tissue of origin tracing. HiTIMED empowers precise resolution of TME cell composition, promising a better understanding of cell heterogeneity in TME and offers new opportunities to study more complex relationships of the TME with etiologic exposures, patient outcomes, and response to treatment. HiBED discerns brain cell heterogeneity and cellular distribution shifts in aging and brain-related disorders. HiTAIC stands as a pioneering tool with immense clinical potential for cancer diagnosis and management.

Original Citation

  1. Zhang, Z.; Wiencke, J. K.; Kelsey, K. T.; Koestler, D. C.; Christensen, B. C.; Salas, L. A., HiTIMED: hierarchical tumor immune microenvironment epigenetic deconvolution for accurate cell type resolution in the tumor microenvironment using tumor-type-specific DNA methylation data. J Transl Med 2022, 20 (1), 516.
  2. Zhang, Z., Wiencke, J.K., Kelsey, K.T., Koestler, D.C., Molinaro, A.M., Pike, S.C., Karra, P., Christensen, B.C., and Salas, L.A. (2023). Hierarchical deconvolution for extensive cell type resolution in the human brain using DNA methylation. Front Neurosci 17, 1198243.
  3. Zhang, Z.; Stolrow, H.G.; Christensen, B.C.; Salas, L.A. Down Syndrome Altered Cell Composition in Blood, Brain, and Buccal Swab Samples Profiled by DNA-Methylation-Based Cell-Type Deconvolution. Cells 2023, 12, 1168.
  4. Zhang Z, Lu Y, Vosoughi S, Levy JJ, Christensen BC, Salas LA. HiTAIC: hierarchical tumor artificial intelligence classifier traces tissue of origin and tumor type in primary and metastasized tumors using DNA methylation. NAR Cancer 2023;5(2):zcad017.

Download

Share

COinS