Date of Award

Winter 12-15-2022

Author's School

Graduate School of Arts and Sciences

Author's Department

Biology & Biomedical Sciences (Computational & Systems Biology)

Degree Name

Doctor of Philosophy (PhD)

Degree Type



Worldwide incidence rates of cutaneous melanoma are increasing, and while survival rates for early stages of melanoma are high, rates drop precipitously for metastatic melanomas or those that are unable to be targeted by currently available treatments. As melanomas have a propensity to quickly metastasize, understanding the contributions of melanoma initiation remains critical for early intervention. Onset of melanoma is characterized most by mutations that stimulate mitogen-activated protein kinase (MAPK) signaling, disrupt DNA damage checkpoints, and trigger mechanisms to bypass senescence through elongation of telomeres. Additionally, in zebrafish melanoma models, the earliest cluster of melanoma-initiating cells activate expression of a neural crest reporter which remains on during the melanoma lifespan. Neural crest cells are highly multipotent and migratory stem cells that arise in early development. Bulk and single-cell RNA-sequencing have confirmed the prevalence of cells that are transcriptionally similar to neural crest cells in human and mouse melanomas and established the importance of this lineage in initiation, metastasis, immune response, and drug evasion.

Cutaneous melanomas have one of the highest mutational loads of any cancer types. The most common protein-coding mutations occur in BRAF or NRAS, which activate MAPK signaling, and CDKN2A, PTEN, or TP53, which inactivate tumor suppressors. Even more common are non-coding mutations in the promoter of TERT. These non-coding variants, present in 80% of melanomas, create a novel GABPA binding site leading elevated TERT transcription. Emerging evidence suggests that up-regulation of TERT elongates telomeres and assists in bypassing senescence brought on by excessive MAPK signaling. Since the discovery of the TERT promoter mutation, several other functional non-coding variants have been identified in not only in melanoma but across other cancer types and diseases. As cutaneous melanomas have some of the highest mutation rates, the number of functional non-coding variants in melanoma likely remains largely uncharacterized.

Non-coding variants typically occur in or create cell-type specific enhancers. Under this assumption, we created a pipeline to identify recurrent variants in putative melanoma regulatory regions. Within these regions, we identified 140 statistically significant recurrently mutated regions, i.e. hotspots, that harbored ~2000 putative cis-regulatory variants. As we started with almost 21 million variants, the almost 10,000-fold reduction in the number of variants led to a pool of high-confidence variants for which to validate. Statistically significant variants were almost exclusively identified in promoters and more specifically at ETS transcription factor binding sites. Our pipeline identified the TERT promoter mutations as the 13th highest scoring hotspot. Through several rounds of validation by luciferase assays and massively parallel reporter assays, we narrowed in on mutations in the promoter of CDC20 which is mutated in 39 out of 140 cutaneous melanoma samples, spanning different stages of the melanoma lifespan.

Almost all the promoter mutations in CDC20 reduced reporter activity significantly across seven melanoma cell lines, one primary melanocyte cell line, and a human embryonic kidney cell line, likely though the disruption of an ETS motif. As high CDC20 is prognostic for worse overall survival, we hypothesized that low levels of CDC20 were important for early stages of melanoma but disadvantageous at later stages. Inspecting variant allele frequencies across different stages of melanoma supported this hypothesis, demonstrating variant allele frequencies like that of BRAF and TERT, known early events, in primary melanomas. However, many of the CDC20 promoter variants were not detected in distant metastases, supporting the notion that CDC20 promoter variants are selected against in metastatic populations as high levels of CDC20 appear to be beneficial for migration. Furthermore, in samples with low levels of CDC20, we identified high levels of key neural crest transcription factors and known melanoma oncogenes including SOX10, an important neural crest and melanocyte lineage specifier. Previous studies have shown that increased expression of SOX10 is observed in melanoma and leads to faster melanoma formation while knock-down of SOX10 slows down melanoma onset.

We engineered a small indel in the promoter of CDC20 in a human melanoma cell line and observed decreased migration capabilities and increased expression of SOX10, among other key neural crest genes. Ongoing and future work will establish the mechanisms by which high and low CDC20 expression leads to metastasis and neural crest identity re-emergence, respectively. One promising hypothesis is that changes in the length of the specific cell cycle stages could lead to changes in gene expression specific to differentiation pathways. For example, prolonging of the G2 phase in human embryonic stem cells upregulated pluripotency maintenance factors. Overall, we identify a non-coding variant that abolishes an ETS motif, leads to down-regulation of CDC20, and establishes a transcriptional program more reminiscent of early stages of melanoma as opposed to the metastatic stage. Therefore, we propose that CDC20 has a dosage-dependent effect on melanoma and that throughout the melanoma lifespan, CDC20 promoter variants confer different advantages at different stages of cancer progression.


English (en)

Chair and Committee

Charles K. Kaufman

Committee Members

Ting Wang, Tychele Turner, George Souroullas, Christopher Maher,