Samuel Aparicio

Cancer is an ecosystem of genetically diverse evolving clones, which emerge in time and space as a result of genomic and non-genomic instability. Consequently, clonal evolution provides a basis for fluctuating cell fitness, which impacts etiology and drug resistance. However, progress in defining clonal fitness by copy number alterations (CNA), has been impeded by a lack of perturbation experiments with timeseries sampling. This dissertation is focused on understanding and measuring clonal fitness determined by copy number genomic instability in breast cancer to single cell resolution. First, I established a series of transplantable human breast cancers in immunodeficient mice strains that were tolerant to DNA damaging chemotherapy. I selected four PDXs (patient derived xenografts) for detailed analysis of clonal fitness under no treatment and after treatment with DNA damaging agents, cisplatin and CX5461. Subsequently, I optimized methods of tumor dissociations, and determined the effects of digestion time and temperature on single cell gene expression. Next, I established that single cell whole genome sequencing of CNA in PDX passaged over time resulted in positive fitness over clones that gradually sweep the population. I established a mixture re-transplant paradigm to demonstrate that the fitness predictions of a Wright-Fisher population genetics model applied to the data, were reproducible experimentally. I described that fitness landscapes of all TNBC tumors were inverted under drug because low fitness clones under no treatment gave rise to drug resistant clones. Drug holiday experiments showed that cisplatin resistance had a fitness cost. I demonstrated using CX5461, that drug resistance could arise in a common background, suggesting common drug resistant states. Finally, I examined the impact of CNA on transcriptional phenotypes. By making assignments of single cells RNA-seq, to CNA defined clones, I observed that in cis, CNA mediated clonal gene expression impacts between 5-50% of the transcriptome. Time series analysis of expression revealed reversible, but time dependent expression reprogramming, thus defining the non-CNA dependence of clonal transcriptomes. Pathway analysis revealed common sets of pathways associated with late drug resistance to platinum in TNBC. These comprehensive measurements and analysis of clonal structure will advance interpretation of polyclonal resistance to therapy.

View record

Mutational processes, including DNA repair defects, have been extensively documented across multiple cancer types. Targeting such defects has yielded much promise in the field. Previous studies have shown great therapeutic potential in targeting G-quadruplex (G4) structures in HR (homologous recombination) and NHEJ (non-homologous end-joining)-deficient tumour cells with the G4 stabilizer, CX-5461. In this thesis, we hypothesized that additional genome maintenance pathways and genes may also be relevant in repairing lesions from G4 stabilization. First, we performed an in vitro subgenome-wide CRISPR/Cas9 screen targeting 480 genome stability-associated genes, in HCT116 colorectal carcinoma cells treated with CX-5461, pyridostatin (PDX; a known G4 binder), and BMH-21 (a non-G4 binder). We discovered novel G4-associated genes and pathways including nucleoplasm, DNA secondary structure binding, and ubiquitin signaling. In addition, we identified DNA polymerase theta (POLQ), known to have roles in the microhomology-mediated end-joining (MMEJ) pathway which is commonly thought to act as backup repair to HR and NHEJ, as a top depleted gene. G4 stabilizers sensitized POLQ deficient cells in multiple cell lines. Next, we studied in vivo the competitive dynamics of cell populations harbouring different sgRNA-induced DNA repair defects in both cell line- and patient-derived xenografts (PDX). Particularly in cell line xenografts, CX-5461 led to the specific lethality of HR (BRCA2) and MMEJ (POLQ) deficient subpopulations, underscoring the importance of both pathways for G4 stabilization in vivo. Finally, we assessed the mutations and changes in DNA repair patterns caused by G4 stabilization. From whole genome sequencing and targeted probe capture sequencing, we observed an increase in mutations in drug-treated cells. Furthermore, using a sequencing based assay to measure different DNA repair pathways, we discovered that in NHEJ-depleted cells, cells preferentially used POLQ-mediated MMEJ, rather than HR, to repair DNA lesions caused by G4 stabilization. Altogether, this study discovered a spectrum of additional genome maintenance-associated vulnerabilities that could be targeted with G4 stabilizer drugs for cancer treatment. In addition, this study identified the novel role of POLQ in repairing DNA damage induced by G4 stabilizers both in vitro and in vivo.

View record

Tumours are heterogeneous populations of cells that evolve dynamically in response to selective pressures. Bulk genome sequencing has been used successfully to characterize constituent populations in tumours. However, it is difficult to detect very low prevalence mutations in bulk genomes, and even more challenging to determine which variants co-occur in the same cells. These challenges can be overcome by sequencing the genomes of individual cells, but single-cell genome sequencing methods present their own obstacles. Sample contamination, biases introduced through amplification, and high cost have thus far precluded large-scale studies of single-cell genomes. We present, in three chapters, computational and experimental approaches to the study of tumour evolution at single-cell resolution. The first chapter examines clonal dynamics in breast cancer patient-derived xenograft models. Using a Bayesian probabilistic method, we infer the dynamics of sub-clonal populations, and validate our predictions with targeted single-cell measurements. In the second chapter, we present a new method for preparing single-cell libraries for whole-genome sequencing. Using a direct tagmentation and indexing strategy without whole-genome preamplification, we enable cost-effective generation of low-depth single-cell libraries with uniform coverage. We show that by merging single-cell genomes in silico, we can produce clonal and pseudo-bulk genomes with uniformity equivalent to standard bulk genomes. As such, in a single experiment, our direct library preparation (DLP) approach permits detailed characterization of copy number heterogeneity at the single-cell level, and inference of single-nucleotide variants at the clone or population level. The final chapter presents a new probabilistic model for improving inference of copy number variants from low-depth single-cell genomes. Using a mixture of hidden Markov models, we assign each cell to a ploidy sub-population, and borrow strength across cells to jointly estimate the model parameters. This research paves the way for large-scale studies of genomic heterogeneity at single-cell resolution, and presents new avenues of investigation into the reproducibility and predictability of tumour evolution in response to selective pressures.

View record

Acute myeloid leukemia (AML) spans a wide array of distinct clinical entities and likely molecular determinants. Despite early treatment success, many aspects of leukemogenesis remain poorly understood, including determinants of leukemic phenotype and identity, and genes and pathways critical to leukemic stem cell (LSC) function.Meningioma 1 (MN1) is a transcriptional co-factor that is an independent prognostic marker for normal karyotype AML, with high expression linked to poor survival and resistance to treatment by ATRA-induced differentiation. MN1 is also a potent and sufficient oncogene in murine leukemia, able to block differentiation and promote LSC self-renewal through transformation of cells at the common myeloid progenitor level. Using this single-hit oncogenic model, MN1 overexpression was exploited to gain further insight into the leukemic process.The objective of this thesis work was to identify and better understand key regulators in LSC function. Sixteen MN1 structural variants were generated to investigate if the leukemic properties of increased proliferation and self-renewal, arrested hematopoietic differentiation, in vivo leukemogenic activity, and resistance to all-trans retinoic acid-induced differentiation could be localised to specific protein regions. Functional assays revealed that the MN1 C-terminus is critical for blocking myeloid and lymphoid differentiation and ATRA resistance while the N-terminus is essential for leukemogenicity, proliferation and self-renewal, and arrested erythro-megakaryocyte differentiation, demonstrating that these leukemic properties can be attributed to specific and largely distinct regions.To identify key genes and pathways underlying leukemic activity, the phenotypic heterogeneity of MN1 leukemic cells was functionally assessed, revealing leukemic and non-leukemic subsets. Gene expression profiling of these subsets was combined with previously-published datasets comparing wildtype leukemic MN1 and mutant versions with varying leukemogenic activity to identify candidate genes critical to leukemia. Through functional analysis of leukemic properties, Hlf and HoxA9 were identified as critical to in vitro proliferation, self-renewal, and impaired myeloid differentiation in MN1 leukemia. Furthermore, this work identifies Meis2 as a novel player in MN1-induced leukemia, with essential roles in proliferation, self-renewal, differentiation, and apoptosis.Together, these models provide a platform to unravel the basis for dysregulated gene expression associated with leukemia and to probe the cellular and molecular determinants of leukemogenesis.

View record

Traditional classifications and treatment of human cancers have operated with limitations surrounding tumor homogeneity and mutational stasis. Clinical metrics of malignant tumors focused on descriptive and behavioral properties such as tissue of origin, cellular morphologic features and extent of spread. Missing has been an understanding of the dynamics of cellular subpopulations that underpin divergent functional properties in space and time. This dissertation is focused on the development and application of methods, including next generation DNA sequencing, computational modeling, and single-cell genotyping protocols to elucidate breast tumor heterogeneity and clonal evolution at single nucleotide and single-cell resolution. First, I present advances in our knowledge of the mutational spectrum that may occur and evolve in an individual epithelial cancer, namely a lobular breast cancer metastases and matched primary tumor separated by a nine year interval. This seminal study demonstrated clonal evolution in a patient’s breast cancer and the successful application of targeted deep sequencing for determining digital allelic prevalences and clonal genotypes in bulk tumors. Second, I describe the diversity of genomic sequence and clonal heterogeneity in tumors of the triple-negative breast cancer subtype. The study uncovered wide clonal diversity in these primary tumors at first diagnosis. Third, I demonstrate via genotyping single tumor cells, that computational inferences of tumor clonal architecture can be made reliably from bulk tissue-derived data sets. This was performed using both somatic point mutations and loss of heterozygosity loci as clonal marks. And fourth, I applied single-cell analysis to study the clonal evolution in breast tumor murine xenografts following engraftment and serial passaging. This research uncovered a range of outcomes in tumor clonal composition upon initial engraftment and serial passaging. The same clonal groups were found to arise independently in separate xenografts derived from the same primary tumor, suggesting selection of functionally significant genotypes. Comprehensive capabilities in the measurement and analysis of clonal structure in cancers offers improved classification and combinatorial treatments of subpopulations in heterogeneous tumors and better use of murine xenograft models. Functionally relevant subpopulations of tumor cells, irrespective of numerical abundance or spatiotemporal persistence, can thereby be targeted using clonally informative genomic profiles.

View record

PPP2R2A is a regulatory subunit of protein phosphatase 2A (PP2A). Genomic analysis based on 2000 breast cancer cases identified PPP2R2A as one of the deletion hotspots in breast cancer genomes and 63% of PPP2R2A deletions were found in estrogen receptor positive (ER+) breast cancers. We hypothesized that the high frequency deletion of PPP2R2A, as well as its correlation with ER expression status, implies its functional roles in breast cancer development. In this thesis, I investigated the functional impacts of PPP2R2A deletion in breast cancer development from the aspects of ER signaling, PP2A complex composition, and cell morphological changes. First, I studied ER signaling activity and mechanism in T47D cells with the analysis of transcriptome, ER binding specificity and ER co-factor recruitment, etc. Second, I studied PP2A complex composition in three breast cell models with targeted quantitative mass spectrometry (MRM). Last, I studied morphological changes in mammary breast cell models, 184-hTERT and MCF10A, in terms of cell proliferation, surface protein localization, and cell mobility based on transcriptome and signaling network analysis. As a result, reduced expression of PPP2R2A led to differential expression of ER response genes, alterations in ER binding specificity, and changes in ER co-factor composition. SPDEF was particularly up regulated and recruited in ER transcriptional machinery. Analysis in clinical samples also confirmed the correlation of SPDEF up regulation with ER expression status and PPP2R2A copy number loss. In addition, MRM analysis revealed changes in PP2A complex composition after PPP2R2A knockdown with increased relative abundance of STRN in PP2A complexes in ER positive cell models. Finally, my morphological studies demonstrated mislocalization of cell surface proteins and enhanced cell mobility in breast mammary epithelial cell models with reduced PPP2R2A expression. In conclusion, PPP2R2A plays an important role in regulating cell signaling activity and cellular morphology and its genomic deletion would significantly contribute to the development of breast cancer.

View record

Genomic aberrations such as copy number alterations (CNA) and loss of heterozygosity (LOH) are hallmarks of human malignancies. These genomic abnormalities can have a measurable effect on the structure and dosage of chromosomal regions. Tumour suppressors and oncogenes altered by CNAs often contribute to a tumourigenic phenotype of increased proliferation. CNA and LOH can accrue through the process of branched evolution, resulting in the emergence of divergent clones with distinct aberrations present at diagnosis. Therefore, measuring and modeling how CNA/LOH distribute in cell populations can elucidate the abundance of specific clones and, ultimately, enable the study of clonal evolution. CNA/LOH events in tumours can be profiled using SNP genotyping arrays and whole genome sequencing (WGS). However, to maximize biological interpretability from these data, accurate and statistically robust computational methods for inferring CNA/LOH are necessary.I present three novel probabilistic approaches that apply hidden Markov models (HMM) to analyze CNA/LOH in tumour genomes. The first method is HMM-Dosage, which distinguishes somatic and germline copy number events. This tool was used to profile 2000 breast cancers, the largest study of this kind in the world. The second method is APOLLOH, which was one of the earliest methods developed to profile LOH in tumour WGS data. Its application to WGS of 23 triple negative breast cancers (TNBC) represents the first time that LOH and its effects on allelic expression were jointly analyzed from sequencing data. The third method is TITAN, which simultaneously infers CNA/LOH and the clonal population dynamics from tumour WGS data. This method provides an analytical route to studying the degree of clonal evolution driven by CNA/LOH. I applied TITAN to a novel set of primary breast tumours and corresponding mouse xenografts, presenting the results of distinct modes of temporal clonal selection patterns. In conclusion, this dissertation presents a suite of novel approaches and their application to real-world cancer datasets, contributing to significant discoveries in breast and ovarian cancers. Future applications of these approaches will further facilitate the elucidation of cancer evolution, the genetic basis of metastatic potential, and therapeutic response and resistance.

View record

The processes involved in mammary gland development are intimately linked withthose that drive breast oncogenesis. Regulation of growth, tissue polarity and genomestability are a few of the factors the maintain homeostasis in breast epithelium and preventmalignant progression. In this work, a series of clonal 184-hTERT cell lines were generatedthat modeled the in vitro growth characteristics of bi-potent mammary progenitor cells; theywere dependent upon fibroblasts for low-density growth and formed dual-lineage acini in 3Dculture. These lines were subsequently used in a genome-wide siRNA screen to identify thefactors that regulate fibroblast-driven epithelial cell growth. Fibroblasts constitute themajority of cells within stroma, which plays a major role in supporting mammary progenitorcell growth. From this screen, 49 surface and secreted factors were identified that putativelytransduce the signals emanating from the fibroblasts that are required for epithelial cellgrowth. These factors were more potent than any of the previously described growth factorreceptors. When assessed in primary tissue, Gpr39, Scarb2, Ntn1, Efna4, Nptx1, and Ctnna1were found to have the greatest effect on overall progenitor cell growth, while SerpinH1differentially suppressed luminal progenitor cells, and Nkain4 and Kcnj5 differentiallysuppressed bi-potent progenitor cells. Further profiling of these lines identified the planarcell polarity protein Celsr1 as differentially regulated under fibroblast-dependent conditions.Silencing of Celsr1 increased the number of bi-potent progenitor cells detected in the colonyformingassay. Furthermore, it induced branching morphogenesis within normally sphericalacini and disrupted the apical polarity of these structures in 3D culture. Within this system,Celsr1 is suspected of signaling through Shisa4. This is the first description of a noncanonicalCelsr1 interactor. Finally, a curious variant line was identified amongst thecollection of 184-hTERT cells generated for this work. This line harbours mitotic spindleand cell cycle checkpoint defects, and rapidly gains chromosome 20 during passaging.Through an elimination process, de novo promoter hypomethylation and subsequentoverexpression of CENPI was identified as likely being responsible for this phenotype. Thisis the first description of CENPI deregulation and one of a few descriptions of gene promoterhypomethylation resulting in genome instability.

View record

Kisspeptins and their receptor, GPR54, mediate sex hormone release through stimulation of the hypothalamic-pituitary-gonadal axis and have been implicated as metastasis suppressors. Expression of kisspeptin and GPR54 has been associated with less invasive cancers as determined by RNA expression, and a multitude of in vitro studies has consistently shown that overexpression of either ligand or receptor in malignant cell lines results in a less invasive phenotype. We hypothesized that expression of GPR54/kisspeptin in epithelial malignancies is predictive of disease outcome and altering endogenous GPR54 signalling in malignant breast and ovarian epithelial cells could alter their metastatic properties. We have determined by immunohistochemistry that kisspeptin and GPR54 are independent favourable prognostic markers for ovarian carcinoma and are specific for the clear cell cancer subtype; the least characterized of the subtypes. Additionally, loss of GPR54 is associated with poor prognosis in node positive breast cancer patients and is also lost in prostate cancer and testicular germ cell nonseminomas as compared to more benign disease. Moreover, secreted kisspeptin is elevated above physiological levels in the plasma of women with gynaecological cancers, including ovarian cancer. We evaluated GPR54 expression across a panel of breast and ovarian cancer cell lines to create an in vitro model system with which to knockdown GPR54 expression using RNA interference. However, we discovered that endogenous GPR54 was internalized rather than localized to the plasma membrane of these cancer cell lines. Consequently, internal GPR54 was unable to signal through its canonical Gαq pathway. To discover novel genes involved in kisspeptin-GPR54 signalling, we assessed gene expression differences between the Gpr54 and Kiss1 knockout mice as compared to wildtype mice. Our novel candidate list provides insight into physiological signalling in the hypothalamus that can then be applied to epithelial anti-metastatic signalling. Our results also support the sex hormone negative feedback effect on kisspeptin expression as reported in the current literature.In summary, we have confirmed kisspeptin and GPR54 as favourable prognostic markers, are the first to report the intracellular localization of GPR54 in endogenously expressing cancer cell lines, and we have introduced a list of novel genes involved in signalling.

View record