Date of Award

January 2023

Document Type


Degree Name

Medical Doctor (MD)



First Advisor

Aaron D. Tward

Second Advisor

Smita Krishnaswamy


Introduction: Improving the therapeutic landscape for hearing loss necessitates an improved understanding of inner ear biology. Historically, our molecular understanding of inner ear physiology has lagged behind our knowledge of mechanistic understanding of signal transduction due to the rare, diverse, and fragile cell types of the inner ear. However, recent advances in high-throughput sequencing have allowed us to decode the molecular underpinnings of physiologic changes in the murine inner ear. Herein, we single-cell RNA sequence fetal tissue and present a comprehensive atlas of the developing human inner ear. Harnessing newer approaches in computational analysis, we delineate diversity of cell types and highlight molecular changes underlying developmental phenomena.

Materials and Methods: First, donated fetal inner tissue at various gestational timepoints — 15, 17, 18, 23, 24 weeks, and adult — underwent microdissection to separate cochlea and vestibule from surrounding structures. Tissue underwent preparation, including library preparation, for 10X sequencing. Part of the each tissue underwent fluorescence activated cell sorting to enrich for cells expressing EpCAM before preparation. Lastly, the tissue underwent microfluidics-based single-cell RNA sequencing with barcodes as unique molecular identifiers for each cell. After all data was sequenced, the expression matrices were merged using CellRanger. The combined expression matrix first underwent quality analysis and filtering to remove poor quality cells and doublets. Then, the data was inputted into CellFindPy, an unbiased hierarchical clustering approach utilizing biologically relevant parameters to cluster cells to a high resolution. The highly dimensional expression matrix underwent advanced dimension reduction algorithms, UMAP and PHATE, for visualization of the intercellular relationships. Cells were separated for subgroup analysis, in which trajectory analysis and differential gene expression was used.

Results: We present an atlas with 140,007 transcriptomes of 26,429 genes remaining after quality analysis and filtering. CellFindPy clustering yielded 9 groups of cells at the highest level originating from 5 tissues. Analysis of epithelial cell subclusters reveals enrichment of rare epithelial cells across the sequenced timepoints, further evidenced by gene markers of cochlear and vestibular hair cell subtypes. Furthermore, rare cell types of stria vascularis are also found among 3 different tissues of origin. Velocity analysis reveals molecular hallmarks of vestibular hair cell maturation and differentiation of Schwann cells from neural crest progenitors.

Conclusions: The present work is a proof-of-concept for creating a pan-tissue atlas of the developing human inner ear. However, single-cell sequencing is limited by dropout, batch variation, and challenges in analysis across multiple axes of change. Future work should focus on validating findings, evaluating intercellular interactions driving development, and analyzing dynamic change over time.


This thesis is restricted to Yale network users only. It will be made publicly available on 05/22/2026