Elucidating Evolutionary Constraints of Mouse Mammary Cancer Using Adenomatous Polyposis Coli Mutations
Open Access
- Author:
- Keller, Ross Richard
- Graduate Program:
- Biomedical Sciences
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- February 23, 2018
- Committee Members:
- Edward Joseph Gunther, Dissertation Advisor/Co-Advisor
Edward Joseph Gunther, Committee Chair/Co-Chair
Kristin Ann Eckert, Committee Member
Lisa M Shantz, Committee Member
James Riley Broach, Outside Member
Sarah Bronson, Committee Member - Keywords:
- Cancer
Evolution
Genetics - Abstract:
- Cancer is a disease of evolution. Species evolve because mutations are passed to progeny through the germline, but somatic cells acquire mutations as well, and somatic mutation can initiate cancer. In addition, established cancers themselves evolve. Within a tumor, ongoing mutagenesis creates cell-to-cell genetic diversity, allowing subpopulations to undergo natural selection in Darwinian fashion. With a surplus of cells and a staggering number of potential mutations, evolution could, in theory, empower a tumor to overcome a plethora of obstacles. But fortunately, cancer evolution is restricted. Chemical, genetic, cellular, and environmental evolutionary constraints make cancer vulnerable. To uncover novel constraints, we utilized mouse model systems of breast cancer, specifically models driven by mutations in the adenomatous polyposis coli (Apc) gene, which antagonizes oncogenic Wnt signaling. In Chapter 2, we show that chemical and genetic constraints can direct which oncogenes are selected to cooperate in driving cancer. In mice that develop reversible, Wnt pathway-dependent mammary cancers (iWnt mice), exposure to chemical carcinogens that preferentially form adducts with either adenine (7,12 dimethylbenz(a)anthracene, DMBA) or thymine (N-ethyl-N-nitrosourea, ENU) resulted in activating mutations affecting different genes in the oncogenic Ras-Raf pathway (HrasCAA61CTA or BrafGTG636GAG, respectively). Both these mutations occur at A:T base pairs, but differ because in Hras, adenine resides on the sense strand, while in Braf, thymine does. This implicated strand-biased processes, such as strand-specific DNA repair, in mutagenesis. To confirm this, we generated DMBA- and ENU-induced mammary tumors in Apcmin mice, which inherit one nonfunctional Apc allele but routinely acquire second-hit mutations en route to mammary tumorigenesis. Exposure to each chemical led to distinct second-hit Apc mutation patterns, including strand-inverse mutation signatures at A:T sites, which precisely matched the mutation bias observed in Ras-Raf oncogenes. The findings indicate that a mutagen’s chemistry paired with DNA repair processes may have an outsized influence on which cancer driver genes are selected, independent of other pressures. In Chapter 3, we exploited patterns of second-hit Apc driver mutations to demonstrate that ionizing radiation (IR) perturbs mutation spectra in a dose-dependent manner, implying different doses drive cancer evolution via different mechanisms. In Apcmin mice, low-dose IR exposure (1 Gy), whether administered as a one-time dose or fractionated daily doses increased mammary tumor incidence without altering the mutation spectrum compared with IR-naïve tumors. By contrast, high-dose IR (5 Gy) not only increased tumor incidence, but also shifted the mutation spectrum towards microdeletions, a putative signature of DNA double strand break (DSB) repair. Since IR is routinely used to treat breast cancer patients, we went on to determine how IR exposures impact mammary tumor cells. We introduced the Apcmin allele into iWnt mice and determined how IR exposure delivered prior to Wnt withdrawal impacted second-hit Apc mutations responsible for initiating relapse. Again, effects of low- and high-dose IR diverged. Low-dose IR augmented relapse but did not change the mutation spectrum from IR-naïve. By contrast, high-dose IR augmented relapse further and exhibited a microdeletion signature characteristic of DSBs. Thus, high IR doses likely introduce DNA mutations by direct DNA damage, while lower doses promote tumor evolution via indirect mechanisms. These findings have implications for assessing cancer risk and planning radiotherapy regimens. Finally, in Chapter 4 we establish that oncogene overdose is a powerful evolutionary constraint that remains in force throughout tumor lifespan. At tumor initiation, mammary gland cells acquired mutations conferring an elevated, but submaximal oncogenic signal, avoiding oncogene overdose. However, for overdose to render cancer vulnerable, the constraint must remain through stages of tumor progression. Thus, we show that tumor cells responsible for initiating relapse following Wnt withdrawal (i.e. rescue subclones, RSCs) acquired second-hit Apc mutations conferring oncogene overdose avoidance, but we also show RSCs are still susceptible to oncogene overdose. Selection for these mutations during therapy involves an evolutionary trade-off. Rescue mutations, which enhance fitness by restoring oncogenic signaling during therapy, instead reduce fitness by driving oncogene overdose while superimposed on native (i.e., undrugged) signaling. Consistent with this model, RSCs underwent dynamic turnover prior to therapy, and RSCs generated by a timed mutagen exposure decreased in number when therapy was delayed or transiently interrupted. We ruled out neutral drift dynamics by demonstrating RSCs compete poorly against therapy-sensitive parental clones during tumor growth and were lost when the endogenous signal is increased to maximum. Exploiting this vulnerability could further efforts to eliminate cancer cells that drive relapse.