University of Southern California Dissertations and Theses

Applying multi-omics in cancer liquid biopsy for improved patient monitoring and biomarker discovery

(USC Thesis Other)

Applying multi-omics in cancer liquid biopsy for improved patient monitoring and biomarker discovery

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Copyright 2024 Michael James Schmidt APPLYING MULTI-OMICS IN CANCER LIQUID BIOPSY FOR IMPROVED PATIENT MONITORING AND BIOMARKER DISCOVERY By Michael James Schmidt A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (CANCER BIOLOGY AND GENOMICS) August 2024 ⅱ “When you die, that does not mean that you lose to cancer. You beat cancer by how you live, why you live, and in the manner in which you live. So, live. Live. Fight like hell. And when you get too tired to fight then lay down and rest and let somebody else fight for you.” – Stuart Scott ⅲ To our Forever Sunshine. ⅳ Acknowledgements I did not appreciate what science was until my junior year of college. At that time, the scientific method was a foreign language. I was on track to get a degree in finance until I transferred colleges and decided to take Biology 101 with Suzanne Gollery. It was Suzanne, my first real mentor, who first opened my eyes and introduced me to what science really was about: being curious and finding the truth. I would not be here today without Suzanne. I would like to thank my family, especially my mom and dad, for their unwavering support in whatever I want to do in life. I know if I chose another career path (which I could not imagine) that they would be right by my side, encouraging me to put my best foot forward. I would also like to thank my friends who helped to keep me sane throughout the last few years. I would like to thank my advisors, Jim Hicks and Peter Kuhn, for their constant support and desire to push forward impactful science. When I first rotated in the lab, I sat down with Jim and knew that he was the one I wanted to mentor me. Since that day almost 5 years ago, I could always rely on Jim for any scientific question I had. I learned more than I could ever imagine under Jim, and although we did not accomplish everything we set out to, I think we did a pretty good job. And I am extremely grateful for the opportunity that Peter provided me with top resources and collaborations that made my graduate experience everything I desired it to be. I would like to thank CSI-Cancer as a whole for their support. Special thanks to Rish and Amin, who both have pushed me to become a better scientist and who taught me about bioinformatics. Also, thanks to Rish for being a great critical lab partner during our GeTMoR project and for providing much needed intellectual stimulation. I would also like to thank my collaborators, for Retinoblastoma Liya Xu and Jesse Berry, and for large polyploid cancer cells Sarah Amend and Ken Pienta. I was lucky enough to be ⅴ involved in multiple projects and received guidance from experts in each of these fields. I could not be grateful enough for their support and willingness to help me succeed. I would like to thank my committee advisors Min Yu, David Cobrinik, and Amir Goldkorn for their guidance and advice on this research as it has evolved over the last few years. At every meeting it was refreshing how engaged they were, and they all helped me push this work forward. Lastly, I need to thank my wife, Haley, without whom this would not have been as easily manageable. We made it. Long distance for over 4 years was hard, but worth it. I feel so lucky and overwhelmed with gratitude for having you in my life. You have been my rock, my anchor. You are always there for me, even if it is literally all the way across the continental United States. To talk and listen. To laugh and to cry. I have learned so much from you and I appreciate everything you have sacrificed over the last few years to make our relationship work. I look forward to our next step in our journey with Ada-Pants (the best doggo) by our side to keep the good times, and of course the good sniffs, going. Thank you for being my inspiration. I love you. ⅵ Table of Contents Epigraph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii Dedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ⅳ List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ⅷ List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ⅸ Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ⅹ Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .ⅹi Chapter 1: Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Chapter 2: Simultaneous Copy Number Alteration and Single-Nucleotide Variation Analysis in Matched Aqueous Humor and Tumor Samples in Children with Retinoblastoma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11 2.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.5 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Chapter 3: GeTMoR: Genomic, Transcriptomic, and Morphometric Profiling of Single Rare Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 3.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 3.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 3.5 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 3.6 Full GeTMoR protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Chapter 4: Polyploid cancer cells reveal signatures of chemotherapy relapse . . . . . . 57 4.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 4.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .60 4.3 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .61 ⅵ 4.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 4.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 4.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 Appendix1: Supplementary Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 ⅷ List of Tables Table 2.1. Clinical characteristics of 11 patient RB cohort . . . . . . . . . . . . . . . . . . . . . . . . .. . . 16 Table 4.1. Cohort of 6 advanced prostate cancer patients stained with survival biomarkers . ..76 ⅸ List of Figures Figure 1.1. LBx enables multi-omic profiling from a single blood draw . . . . . . . . . . . . . . . . . . . 4 Figure 1.2. Gallery of rare cells detected across cancer patients . . . . . . . . . . . . . . . . . . . . . . . 7 Figure 1.3. Enumeration of large polyploid cancer cell publications over the last 30 years. . . . 9 Figure 2.1. Schematic of study design for AH versus tumor multi-omic profiling . . . . . . . . . . 15 Figure 2.2. CNA analysis concordance between targeted AH and tumor samples. . . . . . . . . 18 Figure 2.3. SNV landscape of RB patients from targeted sequencing . . . . . . . . . . . . . . . . . . 20 Figure 2.4. Summary of genetic variants observed in the cohort . . . . . . . . . . . . . . . . . . . . . . 22 Figure 2.5. Mutational landscape to CREBBP and BCOR . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Figure 3.1. GeTMoR blood processing schematic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Figure 3.2. GeTMoR separation of RNA and DNA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 Figure 3.3. Recovered transcript quality of GeTMoR processed cells . . . . . . . . . . . . . . . . . . 38 Figure 3.4. GeTMoR retains DNA CNV profiles for MDA-MB-231 and PC3 cells . . . . . . . . . 39 Figure 3.5. GeTMoR profiles for identified MDA-MB-231 and PC3-GFP spiked cells . . . . . 41 Figure 4.1. CTCs with increased genomic content are found in BM aspirate of late-stage prostate cancer patients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .63 Figure 4.2. Large polyploid cancer cells are induced following chemotherapy exposure in MDAMB-231 and PC3 cell lineages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 Figure 4.3. Genomics of large polyploid cancer cells and their progeny . . . . . . . . . . . . . . . . . 68 Figure 4.4. Chemotherapy induced surviving tumor cells share common phenotypes and pathways for survival . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 Figure 4.5. HOMER1, TNFRSF9, and LRP1 are putative markers of resistance . . . . . . . . . . 73 Figure 4.6. HOMER1, TNFRSF9, and LRP1 are positive on CTCs in the BM aspirate of late stage prostate cancer, and are correlated with recurrence in prostate and breast cancers . . 77 ⅹ Abbreviations AH Aqueous humor BM Bone marrow cfDNA Cell-free DNA Chr Chromosome Cis Cisplatin CNV Copy number variation CTC Circulating tumor cell LCTC Large circulating tumor cell with increased genomic content ctDNA Circulating tumor DNA Doc Docetaxel EPI Epithelial cocktail of pan-cytokeratin and EpCAM FRISCR Fixed and recovered intact single-cell RNA GeTMoR Genomic, Transcriptomic, and Morphological profiling or Rare Cells HDSCA High Definition Single Cell Assay IF Immunofluorescence LBx Liquid Biopsy NGS Next-generation sequencing PACC Poly-aneuploid cancer cell PB Peripheral blood PFS Progression free survival PGCC Polyploid giant cancer cell RB Retinoblastoma RB1 Retinoblastoma 1 gene SCNA Somatic copy number alteration SNV Single nucleotide variant TPM Transcripts per million WGS Whole genome sequencing ⅹⅰ Abstract Cancer persists as an enigmatic adversary that the scientific community has yet to fully comprehend. Its daunting presence is marked by both inter-tumoral heterogeneity across patients and intra-tumoral variability within individuals, necessitating advancements to unravel its intricate mechanisms. Enter multi-omics, a transformative approach that, among others, can concurrently explore epigenomic, genomic, transcriptomic, and/or proteomic landscapes. By harnessing this technology, particularly in the realm of liquid biopsy (LBx) for cancer analytics like circulating tumor cells (CTCs) and circulating tumor DNA, new horizons are primed for groundbreaking discoveries that could facilitate a deeper understanding of any given cancer and revolutionize patient care. This dissertation embarks on a journey, beginning with the demonstration of a cuttingedge targeted genomic sequencing technique applied to LBx in pediatric Retinoblastoma patients. By simultaneously profiling copy number and single nucleotide variants through a targeted sequencing approach, this method holds promise for disease monitoring and enhancing patient well-being (Chapter 2). Recognizing the imperative for refined molecular characterization of CTCs, this work proceeds to introduce GeTMoR, an innovative single cell multi-omic approach delving into the Genome, Transcriptome, and Morphology of individual Rare Cells (Chapter 3). Leveraging elements of the GeTMoR pipeline, this study investigates a previously overlooked population of therapy-resistant cells—large polyploid cancer cells— identifying novel biomarkers linked to chemotherapy recurrence (Chapter 4). Cumulatively, this body of work seeks to push the boundaries of single cell multi-omics and LBx, ultimately striving for a deeper comprehension of the complexities of cancer and paving the way for improved clinical outcomes. 1 Chapter 1: Introduction The battle against cancer remains one of the most daunting challenges in modern medicine, characterized by its multifaceted nature and formidable resilience to therapeutic pressures. Despite significant strides in understanding cancer biology and developing promising interventions, the emergence of therapy resistance continues to impede progress, contributing substantially to cancer-related mortality worldwide [1]. Further, cancer diagnosis and disease monitoring represent ongoing challenges in clinical oncology, demanding innovative approaches to improve early detection, to understand disease related mechanisms, and monitor disease progression, all to ultimately patient outcomes. This body of work aims to integrate multi-omic techniques to profile cancer related analytes via liquid biopsies (LBx) (Chapters 2-4), to develop improved multi-omic LBx approaches to better understand cancer (Chapter 2-3), and to delineate biomarkers of chemotherapy resistance and disease recurrence (Chapter 4). The crux of this thesis is to incorporate and improve upon multi-omic techniques and apply them to LBx to aid in the monitoring of disease and discovery of novel biomarkers of recurrence. Multi-omics has emerged as a powerful tool to study cancer Advancements in next-generation sequencing (NGS) have ushered in a new era of accessibility to sequencing methodologies, accelerating the realization of multi-omic studies [2]. Traditionally, multi-omic approaches encompass examination of multiple facets within the central dogma of molecular biology – including genomics, transcriptomics, and proteomics – alongside pivotal factors governing gene expression, such as epigenomics [3]. Each realm within the central dogma presents a myriad of -omic subcategories, such as the analysis of disease driving single nucleotide variants (SNVs) and copy number variations (CNVs) in genomics. Furthermore, the landscape of cancer research has been aided by additional -omic dimensions, such as morphomics (pertaining to cellular morphology), metabolomics, lipidomics, 2 and the incorporation of clinical attributes. Accordingly, numerous multi-omic studies have contributed valuable insights into disease mechanisms and biomarkers [4–6]. The resulting accumulation of data stemming from these diverse -omic modalities has facilitated an unprecedented depth of understanding across various cancer types. Central to this progress are public data repositories like The Cancer Genome Atlas (TCGA) and DepMap, which have emerged as invaluable resources that furnish comprehensive multi-omic datasets for patient cancers and model cancer cell lines, respectively. As techniques continue to improve and multi-omic approaches become more democratized, the amount of data generated will continue to grow exponentially. This wealth of data not only fuels cutting-edge research but also paves the way for the identification of novel biomarkers, as well as the development of targeted therapeutics and personalized treatment strategies. Single cell studies have revolutionized the understanding of tumor plasticity Cancer embodies a complex heterogeneity where malignant cells within a single tumor exhibit striking variations in gene expression profiles, phenotypic traits, and morphological features. This inherent diversity has compelled scientists to delve into the intricacies of individual single cells, recognizing that some cells may have more malignant characteristics than others and therefore respond discordantly to the same therapy [7]. A seminal example of the power of single cell studies lies in the transcriptomic investigation of head and neck cancer, where researchers unveiled a dichotomy within a tumor microenvironment: cells situated at the forefront of the tumor, proximal to the vasculature, exhibited a mesenchymal phenotype indicating enhanced migratory potential and being primed for metastasis. Conversely, cells deeper within the tumor core displayed a distinctive hypoxic signature, which may signify an adaptive response aimed at thriving in nutrient-deprived conditions [8]. This nuanced understanding of intra-tumoral heterogeneity not only underscores the dynamic nature of cancer evolution, progression, and metastasis, but also underscores the need for tailored therapies to 3 target specific cellular states. On top of this study, numerous additional single cell studies across genomics, transcriptomics, and proteomics have identified pathways malignant cells may take to resist therapeutic interventions and disseminate throughout the body [9-11]. Single cell multi-omics has recently become a reality where researchers aim to get the most information possible from a single cell [12-13]. G&T-seq (genome and transcriptome) was released in 2015 as one of the first multi-omic single cell methods to study DNA and RNA [14]. SInce then, numerous methods have been developed that interrogate all aspects of the central dogma and regulators of gene expression. Most single cell multi-omic methods center around transcriptome profiling to understand cellular state and behavior, and then obtain other -omic modalities (i.e., chromatin, histone, DNA, and/or protein) to understand how gene expression is regulated, or how the transcriptome translates into functional protein. For example, a simple single cell study on common cancer cell lines combined single-cell chromatin accessibility (ATAC) and RNA-sequencing to understand heterogeneity within the same cell line. The authors found that different cell lines displayed different levels of expression diversity that were dependent on epigenetic plasticity [15]. This study provides insights that cell line models, commonly used to understand complex mechanisms of disease and/or resistance, display intricate heterogeneity which must be considered for any given research endeavor. As the single cell field progresses, researchers are primed to make seminal discoveries in numerous areas, including oncology. Liquid biopsy (LBx) is an approach to monitor cancer Multi-omic techniques, particularly genomics, are routinely used to molecularly profile cancers to understand mutational drivers and potential therapeutic vulnerabilities. However, to study the tumor, it must be sampled and biopsied. Conventional tumor biopsy procedures, though informative, are restricted by inherent risks and limitations of accessibility. Recognizing the need for more effective monitoring tools, recent advancements have spearheaded the 4 emergence of liquid biopsy (LBx), heralding a transformative tool in cancer research [16]. LBx affords an unprecedented opportunity for longitudinal monitoring of disease dynamics through a minimally invasive methodology [17-19]. This facilitates recurrent assessment of therapeutic responses, disease trajectory, and evolution, thereby providing clinicians with invaluable realtime insights into treatment efficacy and disease progression. LBx offers a non-invasive approach for accessing tumor-derived analytes present in diverse biofluids, thereby reducing the necessity for invasive tissue sampling. Through LBx, researchers can scrutinize an array of analytes, including circulating tumor cells (CTCs), circulating tumor DNA (ctDNA), circulating tumor RNA, extracellular vesicles, and tumorassociated proteins, taken from biofluids such as peripheral blood, bone marrow, urine, cerebral spinal fluid, and ocular fluids (Fig. 1.1). Consequently, the implementation of multi-omic studies through LBx provides a fruitful avenue to explore and obtain as much information about the disease as possible. For example, one study evaluated histone modifications and DNA methylation and identified clinically actionable phenotypes across multiple cancers [20], thereby highlighting the utility of applying multi-omics to understand and target disease states. While promising, applying multi-omics in LBx is in its early stages and more research is needed to understand the unique molecular characteristics from every cancer type. Figure 1.1: LBx enables multi-omic profiling from a single blood draw. 5 Retinoblastoma as a model system for multi-omic LBx Retinoblastoma (RB) is a pediatric cancer that has historically necessitated enucleation for safe tumor profiling given the risk of extraocular tumor spread associated with conventional tissue biopsy methods. However, LBx advancements have opened new avenues for RB characterization by sampling the aqueous humor (AH), which resides in the anterior eye chamber [21] . This novel approach enables the sampling of cell-free tumor-related analytes, notably circulating tumor DNA (ctDNA), exosomes, and proteins, from the AH, presenting a noninvasive and potentially transformative strategy for RB diagnosis and management [22-24]. The utilization of AH holds profound implications for clinical decision-making, particularly concerning the necessity for enucleation. Notably, recent findings have linked chromosomal gains on chromosome 6p – found in both AH ctDNA and corroborated in enucleated tumors – as a disease biomarker with a tenfold increase in the risk of enucleation [25-26]. While previous studies have demonstrated genomic concordance between AH and tumor samples at the CNV and SNV levels [27], a notable gap persists in integrating these findings into a unified assay that could benefit both the patient and clinician. In Chapter 2, we present a targeted multi-omic approach employing CNV and SNV analysis to identify genomic drivers of RB in matched AH and tumor samples [28]. By bridging this gap, our study not only enhances current methodologies but also underscores the clinical utility of a unified assay in facilitating actionable decisions through a LBx approach. This innovative strategy not only has the potential to streamline diagnostic workflows but also holds promise for revolutionizing RB management by offering more confidence in sampling the AH as a means to monitor disease. Improvements are needed in the LBx field to profile single CTC transcriptomes While the AH lacks physical CTCs, many biofluids can have thousands of CTCs/mL of biofluid (depending on the disease stage) that offer valuable information about the tumor. The 6 peripheral blood is one of the most common LBx sample types where CTCs are isolated. In a blood draw, CTCs can be enriched via numerous physical (i.e., density, size) or surface protein markers (i.e., epithelial markers) [29]. While CTCs were first reported in 1859, shedding light into their role in cancer metastasis, only recently has CTC detection technology become reliable and sensitive enough to allow for reproducible molecular-based studies. Typical multi-omic LBx studies will enumerate CTCs, and then assess tumor burden and disease driving mutations (i.e., SNVs and CNVs) through ctDNA. Illustrating the transformative potential of LBx in studying CTCs, a notable case study in a breast cancer patient delineated treatment response and disease evolution through CTC enumeration [30]. The study found that increased CTC counts correlated with disease progression and necessitated clinical treatment interventions. While CTC enumeration is informative and counts have been recognized as prognostic indicators in numerous cancers, there is a pressing need to investigate the molecular features of CTCs themselves rather than simply counting them. Investigations into the heterogeneity of CTC cellular states – encompassing plateletcoated, mesenchymal, CD45+ (white blood cell marker) CTCs, and CTCs with large genomic content (i.e., polyploid) – have shed light on their roles in promoting metastasis, drug resistance, and surviving in circulation (Fig. 1.2) [31-35]. Albeit descriptive, the acquisition and implications of this phenotypic diversity remain elusive, and research into the transcriptomes of CTCs to understand their diverse cellular states is needed. Unfortunately, methods to profile RNA from a single CTC are not robust, in part due to their rarity and the extensive processing steps required for isolation that may damage the cell [36]. Methodological improvements to assaying the transcriptomes from single CTCs are needed to understand their phenotypes and roles in the metastatic process. 7 Figure 1.2: Gallery of rare cells detected across cancer patients indicates diverse cellular phenotypes. Addressing this methodological need, Chapter 3 introduces GeTMoR (Genomic, Transcriptomic, and Morphological profiling of Rare cells), a novel single cell multi-omic approach that builds upon the High Definition Single Cell Assay (HDSCA), which is an enrichment-free technique to enumerate and image CTCs. GeTMoR serves as a comprehensive multi-omic profiling tool, enabling (1) high-resolution imaging of CTCs from the HDSCA, (2) acquisition of CNV profiles for assessing tumor clonality, and (3) extraction of RNA for comprehensive transcriptome analysis. GeTMoR has been validated in cell line models, offering unprecedented insights into the molecular underpinnings of tumor heterogeneity at the genomic and transcriptomic levels. By bridging this methodological gap, GeTMoR represents a significant leap forward in rare cell profiling in LBx, potentially unlocking new avenues for dissecting the intricacies of single cell cancer biology and contributing towards understanding the diverse cellular phenotypes and linking them to genotypes. 8 Single cell multi-omics is needed to understand the significance of large polyploid cancer cells found in LBx samples and cell line models Beyond the need for advancements in LBx molecular profiling, the elucidation of biomarkers indicative of treatment resistance and disease recurrence represents a critical need in cancer biology, and constitutes an area that LBx can address [37]. Many cancers relapse and often come back resistant to therapy. LBx has been successful in monitoring CTCs as a readout of treatment response, but there is an urgency to expand upon the simple enumeration of CTCs. Understanding molecular markers of relapse and recurrence is vital toward improving patient outcomes. Numerous mechanisms of therapy resistance used by cancer cells have been well documented, but some have perplexingly remained unexplored. Notably, the presence of large polyploid cancer cells detected within tumor tissue has recently garnered attention as a potential marker of advanced disease stages. Many patient cancers, such as breast, prostate, colorectal, and ovarian, have reported the presence of large polyploid cancer cells as (1) a response to treatment pressure and (2) being enriched in higher grade, poorly differentiated tumors [38-43]. For example, in a rapid autopsy study on prostate cancer large polyploid cancer cells were detected in 33% of primary lesions but were present in 100% of metastatic castrate resistant samples [44]. Further, via LBx, large polyploid cancer cells have been identified in the urine of prostate cancer patients signifying they may be important in metastasis and are capable of dislodging from the primary cancer tissue [45]. As a response, functional studies in model cell line systems have shed light on survival characteristics of large polyploid cancer cells, including their ability to withstand therapeutic pressure and give rise to viable progeny [46-49]. Despite significant progress, the large polyploid cancer cell field is still in its infancy, denoted by few publications over the last 25 years (Fig. 1.3). This lack of attention, in part, may be due to a lack of consensus on nomenclature; studies will refer to these cells as large polyploid cancer cells, giant polyploid cancer cells, large tumor cells, or poly-aneuploid cancer 9 cells. Further, the field of large polyploid cancer cells lacks single cell resolution, particularly in single cell genomics and transcriptomics. For example, it is unclear if any copy number alterations occur in large polyploid cancer cells that promote their formation and survival, or if they solely increase their genomic content via multiple rounds of whole genome doubling. Further, no single cell transcriptomic study has revealed the molecular heterogeneity and plasticity of polyploid giant cancer cells as they form and recover from therapeutic pressure. Figure 1.3: Enumeration of publications queried on PubMed database queried for large polyploid cancer cells, giant polyploid cancer cells, and poly-aneuploid cancer cells. Chapter 4 sheds light on the mechanisms underlying chemotherapy resistance in prostate and breast cancer by delving into model cell line systems of large polyploid cancer cells. Leveraging the multi-omic techniques pioneered by GeTMoR, this study evaluates alterations in copy number, transcriptomes, and morphological characteristics of chemotherapysurviving large polyploid cancer cells and their progeny. Through this comprehensive analysis, we aim to delineate potential markers of disease recurrence in patients, offering invaluable insights into the molecular determinants of treatment response and disease progression. We then apply the information learned through model cell lines and characterize biomarkers through 10 LBx in the bone marrow of prostate cancer patients. By unraveling the intricate interplay between large polyploid cancer cells and therapeutic resistance, our research opens avenues to develop personalized monitoring strategies, mitigate treatment resistance, and ultimately improve patient outcomes. Overview This work aims to weave together narratives across cancer biology within the context of liquid biopsy. Each individual contribution provides distinct insights and enhancements to the intricate terrain of cancer research, highlighting the transformative capacity of multi-omic methodologies in propelling diagnostic capabilities (Chapters 2-3), refining monitoring techniques (Chapters 2-4), and elucidating disease recurrence markers (Chapter 4). Taken together, the aim of this thesis is to integrate and improve upon multi-omic approaches to improve our understanding of cancer diagnostics and biomarkers. 11 Chapter 2: Simultaneous Copy Number Alteration and SingleNucleotide Variation Analysis in Matched Aqueous Humor and Tumor Samples in Children with Retinoblastoma This chapter is a published scientific article: Int. J. Mol. Sci. 2023, 24(10), 8606; https://doi.org/10.3390/ijms24108606 Authors: Michael J. Schmidt, Rishvanth K. Prabakar, Sarah Pike, Venkata Yellapantula, ChenChing Peng, Peter Kuhn, James Hicks, Liya Xu, and Jesse L. Berry This chapter details the utility of using one targeted sequencing method to determine copy number variants (CNVs) and single nucleotide variants (SNV) in patients with Retinoblastoma. I conducted all bioinformatic analysis with this study, including: compiling and validating the SNV mutect2 pipeline, and establishing concordances in CNV calls between the targeted sequencing approach and our internal low pass WGS technique. I am responsible for making all the figures and for writing the original draft of the manuscript. Liya Xu and Jesse Berry are responsible for collecting and processing the patient samples. Full Author Contributions: Conceptualization, M.J.S., L.X. and J.L.B.; methodology, L.X, J.L.B., M.J.S. and R.K.P.; validation, L.X., J.L.B., V.Y. and M.J.S.; formal analysis, L.X. and M.J.S.; investigation, L.X. and J.L.B.; resources, L.X., J.L.B., V.Y., C.-C.P., J.H. and P.K.; data curation, L.X. and M.J.S.; writing—original draft preparation, M.J.S.; writing—review and editing, M.J.S., R.K.P., S.P., J.H., P.K., L.X. and J.L.B.; visualization, M.J.S.; supervision, J.H., L.X. and J.L.B.; project administration, L.X. and J.L.B.; funding acquisition, P.K., L.X. and J.L.B. All authors have read and agreed to the published version of the manuscript. 12 2.1 Abstract Retinoblastoma (RB) is a childhood cancer that forms in the developing retina of young children; this tumor cannot be biopsied due to the risk of provoking extraocular tumor spread, which dramatically alters the treatment and survival of the patient. Recently, aqueous humor (AH), the clear fluid in the anterior chamber of the eye, has been developed as an organspecific liquid biopsy for investigation of in vivo tumor-derived information found in the cell-free DNA (cfDNA) of the biofluid. However, identifying somatic genomic alterations, including both somatic copy number alterations (SCNAs) and single nucleotide variations (SNVs) of the RB1 gene, typically requires either: (1) two distinct experimental protocols—low-pass whole genome sequencing for SCNAs and targeted sequencing for SNVs—or (2) expensive deep whole genome or exome sequencing. To save time and cost, we applied a one-step targeted sequencing method to identify both SCNAs and RB1 SNVs in children with RB. High concordance (median = 96.2%) was observed in comparing SCNA calls derived from targeted sequencing to the traditional low-pass whole genome sequencing method. We further applied this method to investigate the degree of concordance of genomic alterations between paired tumor and AH samples from 11 RB eyes. We found 11/11 AH samples (100%) had SCNAs, and 10 of them (90.1%) with recurrent RB-SCNAs, while only eight out of 11 tumor samples (72.72%) had positive RB-SCNA signatures were observed in both low-pass and targeted methods. Eight out of the nine (88.9%) detected SNVs were shared between AH and tumor samples. Ultimately, 11/11 cases have somatic alterations identified, including nine RB1 SNVs and 10 recurrent RB-SCNAs with four focal RB1 deletions and one MYCN gain. The results presented show the feasibility of utilizing one sequencing approach to obtain SCNA and targeted SNV data to capture a broad genomic scope of RB disease, which may ultimately expedite clinical intervention and be less expensive than other methods. 13 2.2 Introduction While RB is considered a canonical cancer, having the first molecularly described tumor suppressor gene (RB1), surprisingly, very little is known about the molecular basis underlying the intraocular behavior of this cancer and the varying mechanisms of treatment resistance. This is mainly due to a strict contraindication against invasive biopsy for RB tumors that would risk extraocular cancer spread. Extensive molecular analysis of tumor tissue from advanced enucleated eyes has improved our understanding of RB tumorigenesis. This cancer is predominantly driven by biallelic inactivation of the RB1 gene [50] through either a copy number alteration (CNA) and/or single nucleotide variants (SNVs) that negate RB1 protein function. RB1 function is crucial to RB pathology, and over 1000 RB1 SNVs have been identified that contribute to disease pathogenicity [51]. In almost half the cases, the causal event is inherited, greatly increasing the risk of tumor development in both eyes (bilateral disease) [52]. Further, in approximately 2% of patients, RB develops without any RB1 alterations and is instead driven by oncogenic amplification of MYCN on chromosome 2, which is also associated with aggressive tumor behavior [53]. Apart from loss of function alterations to RB1, deleterious variants of BCOR and CREBBP are the two most common co-occurring events at the SNV level [54–56]. BCOR mutations have been correlated with metastatic potential in RB patients, indicating a more aggressive disease [55], and CREBBP plays a role in regulating cell cycle and differentiation [57]. Although more work is needed to understand the impact of these genes in RB pathogenesis, both BCOR and CREBBP are thought to be tumor suppressors; thus, their inactivation may contribute to disease severity [58]. Despite this body of knowledge of RB1 mutational contributions to RB, none of it is used to improve the care we provide to these young cancer patients. While genomic analyses at the SNV and CNA levels are routinely used to profile the genetic drivers of other tumors, this is not the current clinical practice for children with RB. In the absence of tumor tissue from a biopsy, 14 identifying these alterations for children with RB was previously not possible. This paradigm changed in 2017 when we, and then others, demonstrated that the aqueous humor (AH), the clear intraocular fluid in front of the eye, is an enriched source of tumor-derived cell-free DNA (cfDNA) in RB eyes that can serve as a liquid biopsy [21, 59–63]. We have shown highly concordant somatic CNA (SCNA) profiles with 100 µL AH and matched tumor tissue, and with access to genomic tumor information in eyes undergoing treatment, identified a specific highly recurrent SCNA, chromosome 6p (chr6p) gain, as associated with a 10-fold increased risk of treatment failure requiring surgical removal of the eye [59, 64]. This finding underlies the importance of identifying molecular drivers for intraocular disease and highlights the utility of AH in the study of RB disease without enucleation. However, to assay both SCNAs and SNVs in one clinical sample, two separate experimental protocols are required [64], which increases the overall cost and hinders the utility of expedited AH analysis for clinical use. Methods for systematically profiling genetic disease variants have vastly improved in the next generation sequencing era. Targeted sequencing has been applied to both AH and tumor samples to profile gene regions of interest (primarily the RB1 gene) for SNV variants, along with small insertions and deletions that negate RB1 protein function [54, 64, 65]. With similar intent, SCNA profiling via low-pass whole genome sequencing (WGS) has been utilized to investigate RB1 loss on chromosome 13, along with additional RB-SCNA signatures (i.e., chr6p gains), in both AH and tumor samples. However, utilizing two distinct methods to inform on SNVs and SCNAs is time consuming and expensive, which may ultimately delay, or hasten, life-changing treatment decisions for the patient (i.e., enucleation when it is not needed). We sought to simultaneously profile genomic SNVs and SCNAs through a single targeted method to give clinicians actionable information in less time and with less incurred costs. Here, we present a cohort of 11 patients diagnosed with RB for whom we matched tumor and AH samples. By applying a hybrid capture targeted sequencing approach, we not only systematically profile SNVs along the RB1 gene, but we also observe high intra-patient 15 concordance of SCNAs between respective AH and tumor samples (Fig. 2.1) and confirm all germline RB1 variants in each bioanalyte. The targeted technique also enabled the interrogation of BCOR, CREBBP, and MYCN, from which we reported deleterious variants of BCOR and CREBBP and a focal MYCN amplification. Altogether, we show how the combination of targeted SNV and SCNA detection in a single method can be utilized to profile RB and how this genomic information can be used together to understand the disease. Figure 2.1. Schematic of study design. (A) Eleven patients were present in the cohort. ((A) Top) Sample counts for each biospecimen collected that was processed through the dual-targeted SNV and SCNA methods are displayed. Eleven patients contained matched AH and tumor-targeted sequencing samples. ((A) Bottom) Sample count of lowpass WGS sequencing samples. (B) SCNAs from the SureSelect Targeted Sequencing approach were compared to the gold standard low-pass WGS, and concordance was calculated. The targeted reads were then used to call SNVs, and the 11 matched AH and tumor-targeted reads were directly compared for concordance. This figure was generated with BioRender. 2.3 Results Patient cohort characteristics: This study includes 11 individuals diagnosed with RB at CHLA between 2015 and 2019 (Table 2.1). Cases were de-identified, and tumor stage and 16 seeding class were reported. All patients in the cohort underwent enucleation with tumor tissue available and had matched AH sampled. Further, patient 33 has two AH samples available: one taken at diagnosis (33-dx) and one at enucleation (33-es) following therapy. Demographic and tumor information is available in Table 2.1. The median diagnostic age was 22 months (range, 3–35 months). The presence of a germline RB1 mutation was assayed from routine clinical testing to detect heritable mutations driving each patient’s disease. Three patients harbored germline alterations: cases 9 and 13 harbored inactivating SNVs of the RB1 gene (p.(M148Vfs*8) and p.(R320*), respectively), while case 1 had a hemizygous deletion of chromosome 13q which resulted in a loss of an RB1 allele (Table 2.1). All eyes were enucleated after a median of 95 days (SD = 197.5 days) following therapy. Table 2.1. Clinical characteristics of the 11 patients in the cohort. Case ID Sex Age (Months) Laterality 1 IIRC Group TNM Stage Seed Class Enucleation Laser 2 Sessions Cryo 2 Sessions IVM 3 Germline RB1 Blood Test 1 M 20 U E CT3C None Primary -- -- No 13q8 F 22 U D CT2B Dust Secondary 10 0 Yes Negative 9 F 29 B D CT2B Dust Secondary 2 2 No p.(R320*) 11 F 8 U D CT2B None Secondary 15 4 Yes Negative 13 M 34 U D CT2B Sphere Secondary 8 1 Yes p.(M148Vfs*8) 15 M 10 U D CT2B None Secondary 16 1 Yes Negative 28 F 3 B D CT2B None Secondary 7 1 Yes Negative 33 4 M 22 U D CT2B Sphere Secondary 3 1 Yes Negative 48 M 18 U D CT2B All; cloud Secondary 5 -- -- No Negative 49 M 35 U D CT1B All; cloud Primary -- -- No Negative 50 F 24 U D CT2B None Primary -- -- No Negative 17 Copy number status is concordant in targeted sequencing samples to low-pass WGS samples: From standard low-pass sequencing, which is the gold standard for SCNA determination, all 11 (100%) AH and eight out of 11 (72.7%) tumor samples were identified with one or more SCNAs (Fig. 2.S1). The AH of case 28 has a chr20q gain but with no other recurrent RB-SCNA signature (Fig. 2.S1). To have confidence in SCNA calls from targeted sequencing reads, we evaluated concordance between SCNAs from targeted reads and SCNAs from low-pass WGS reads of tumor samples (see Section 4). When comparing copy number profiles, we found that the mean values were comparable between methods, with a median concordance of 96.2% (SD = 10.7%; Fig. 2.2A). Further, the median concordance for targeted AH to low-pass WGS AH samples was also high at 97.7% (SD = 3%; n = 8; Fig. 2.S2); however, only eight samples were analyzed due to differences in sequencing methodologies between the low-pass WGS samples. 18 Figure 2.2. Copy number alteration analysis shows high concordance between targeted AH and tumor samples. (A) Concordance of derived tumor SCNAs between targeted sequencing reads and their matched low-pass WGS reads. (B) Concordance of SCNAs between targeted AH sequencing samples and their respective targeted tumor sequencing samples. (C) Copy number representation for case 48’s tumor sample overlaying targeted sequencing reads (teal) to low-pass WGS reads (black). A focal MYCN gain was detected on chromosome 2 with both methods. (D) Copy number plot for case 50’s tumor sample overlaying targeted sequencing reads (teal) to low-pass WGS reads (black) showing focal RB1 gene deletion on chromosome 13 and CREBBP loss on chromo-some 16. Since AH and tumor concordance in low-pass WGS samples have been well documented in the literature [59], we sought to determine the concordance between AH tumor samples using our targeted method. For all sample pairs, we found a median concordance of 19 89.9% (SD = 23.4%; Fig. 2.2B). Notably, case 11 had low-quality targeted tumor sequencing reads due to limited biological complexity, which led to low AH/tumor concordance. After removing case 11, the median concordance between targeted AH and targeted tumor reads rose to 90.1% (SD = 5.5%). SCNA profiles from targeted samples recapitulate focal gains and losses: Targeted sequencing has the potential to miss focal chromosomal alterations due to reduced coverage of the genome. However, we verified the sensitivity of our targeted approach by comparing our SCNA results to those of matched samples generated with low-pass WGS. Using our panel, we were able to capture relevant RB SCNAs in our samples that drive disease and serve as prognostic indicators. MYCN is the primary driver of RB in a small fraction of cases without loss of functional RB protein; however, MYCN gains can be both with wildtype RB1 (RB1+/+/MYCNA) and as a secondary driver with a loss of the functional RB protein (RB1-/-). As with other cancers, this gain serves as a poor prognostic indicator in both settings. Case 48 harbored a focal MYCN gain, which was detectable with both low-pass WGS and targeted tumor analyses (Fig. 2.2C). Further, case 50 harbored a focal biallelic loss of the RB1 gene on chromosome 13 that was detected in both low-pass WGS and our targeted approach (Fig. 2.2D). Inactivating SNVs to the RB1 gene: While patients harbored many shared passenger mutations between their respective AH and tumor samples, we were interested in detecting disease driving variants for RB. We found that six out of the 11 cases (54.5%) harbored at least one inactivating RB1 mutation (Fig. 2.3). In total, we detected nine pathogenic variants, of which eight were shared between the AH and tumor (88.9%; Fig. 2.3). One deleterious variant was detected in the tumor from case 28 p.(R255*) but not in the corresponding AH. Visualization of sequencing coverage via IGV confirmed the SNV is present in the tumor but not the AH (Fig. 2.S3). 20 Figure 2.3. SNV landscape of RB patients: (A) Allele frequency plot for deleterious SNVs detected to the RB1 gene. Case 33 had two AH samples (pt33-dx and pt33-es) but had a single tumor sample that was taken at enucleation. Cases that are bold and italicized are germline variants. Cases with asterisk (*) indicate nonsense variants that result in truncated protein. Dark grey boxes indicate the variant was not found in that sample. (B) Read depth for each genomic site in (A). Depth accounts for total reads at each specified site, including both variant reads and normal allele reads. X-axis for depth was cut off at 500 bases for legibility. In general, when the patient harbored a shared AH and tumor mutation, the variant allele frequency (VAF) for the mutation was higher in the AH (Fig. 2.3A). Interestingly, we were able to detect a germline frameshift mutation for case 13 p.(M148Vfs*8) in both the AH and tumor, but the VAF for the tumor was 6.7%, which was significantly lower than 56.5% observed in AH, suggesting a higher normal (i.e., non-RB) cell fraction existing in the tumor sample. (Fig. 2.3A). For case 9, we detected a germline mutation p.(R320*) and a second deleterious mutant p.(R251*); the VAF was higher in the AH for both mutants. Case 33 was a unique patient because we had an AH sample from diagnosis (pt33-dx) and one from secondary enucleation included in the analysis (case 33-es). A splicing variant 21 (c.2325 + 1G>A) was identified in all AH samples with high read depths found in the tumor sample and the AH sample at diagnosis (33-dx; Fig. 2.3A,B). Although the targeted sample for case 33-es failed QC (7 total reads were mapped to the genomic location, which was below our cut-off of 10; Fig. 2.3B), we retained this variant since it was present at diagnosis (case 33-dx) and in the enucleated tumor tissue with high confidence. Similarly, we detected a deleterious variant p.(R455*) on the RB1 gene for case 15 that was only present in the AH but not the tumor (VAF = 90%) that also failed QC (nine reads mapped). Since this variant was only present in the AH and not the tumor, it was discarded; further analysis for case 15 may prove whether this mutation is legitimate. Genomic and genetic alterations for these RB cases are summarized in Figure 4, showing RB signatures, including RB-SCNAs and/or RB1 SNVs that were identified in 10 (90.1%) AH samples and nine tumor samples (81.8%). While our methods detected an MYCN gain for case 48, immunohistochemical staining of the tumor tissue for RB protein showed a lack of nuclear staining. Thus, while an SNV was not detected, there was a loss of functional RB protein suggesting the RB1 gene may have been epigenetically silenced and not that this is a tumor driven primarily by MYCN [66]. Finally, although the AH of case 28 shows neither RBSCNA nor RB1 SNVs, it does have a significant gain on chromosome 20, indicating the existence of somatic alteration (Fig. 2.S1). 22 Figure 2.4. Summary of variants observed in the Cohort. Germline and somatic RB1 SCNAs and SNVs, and other RB-signature genetic variants are displayed. Germline variants detected in blood (separate clinical diagnostic test), AH, and tumor for cases 1, 9, and 13. Cases 33-dx and 33-es were merged; they have the same RB1 somatic SNV detected at diagnosis and enucleation. Variants depicted are found in both targeted and tumor samples. Deleterious variants are detected that impact the BCOR and CREBBP genes: In addition to RB1 variants, we detected deleterious SNVs (n = 3) and SCNAs (n = 2) to the BCOR or CREBBP genes in 5 cases (45.4%). All SNV variants were shared between AH and tumor samples (Fig. 2.5A,B). We also detected a chromosome 16p loss in case 50 (Fig. 2.2D) and case 8 (Fig. 2.5C) that resulted in the loss of a CREBBP allele. No activating MYCN SNVs were detected in the cohort. 23 Figure 2.5. Mutational landscape to CREBBP and BCOR: (A) Allele frequency plot for deleterious mutations to BCOR and CREBBP detected in the cohort. Cases with asterisk (*) indicate nonsense variants that result in truncated protein. (B) Read depth for each genomic site in (A). Depth accounts for total reads at each specified site, including both variant reads and normal allele reads. X-axis for depth was cut off at 500 bases for legibility. (C) SCNA plot for case 8 displaying a loss to chromosome 16 impacting CREBBP. 2.4 Discussion We report the multi-sample genomic analysis of RB-related disease drivers at SCNA and SNV levels by applying a single targeted sequencing method and provide further evidence that the AH can be used as a liquid biopsy approach for the non-invasive evaluation and monitoring of RB disease. This study is the first to examine the SCNA status of both AH and tumor samples while simultaneously analyzing RB1, BCOR, CREBBP, and MYCN for SNV disease drivers. We were able to confidently call RB-signature variants for 10/11 patients in the cohort, including inactivating RB1 gene variants in nine cases and a focal MYCN gain for case 48. Case 15 had no detectable RB1 variants. There are multiple reasons why pathogenic variants are not 24 detected in all cases, including known epigenetic-related silencing events that inactivate RB1 protein function [66]. SCNA profiling of both AH and blood cfDNA has been previously established as an effective liquid biopsy approach by our group and others. In the present study, we found that our target capture approach was highly concordant with the current gold standard of low-pass WGS for calling SCNAs in AH and tumor samples from RB patients (Fig. 2.2A). We were also able to confirm all germline RB1 CNAs that were detected by the clinical blood test for both the AH and tumor in our assay (Table 2.1, Fig. 2.3 and 2.4). Finally, the targeted approach was able to detect a focal MYCN gain in case 48 (Fig. 2.2C), highlighting the sensitivity of the assay. SNV profiling showed strong concordance in the detection of pathogenic RB1 variants between the AH and tumor samples (Fig. 2.3). Taken together, we found eight out of nine (88.9%) disease-driving SNVs in the AH (88.9%), thereby further supporting AH as an alternative liquid biopsy approach for RB. Case 28 was the one outlier where we found a deleterious variant p.(R255*) that was detected in the tumor with a low fraction but not AH (Fig. 2.3 and 2.S3). However, AH SCNA, in this case, harbors clear evidence of the somatic alteration of the disease. Overall, the VAFs in AH samples were found to be higher than those in matched tumor samples. This observation could be attributed to either the presence of normal cells in the tumor biopsy, which could lower the VAFs, or poor sequencing quality leading to noisy data. For instance, in case 49, the copy number profiles show that the ratio of the median of the segmented values in the tumor is much less than that of the AH sample (Fig. 2.S1: Case 49 tumor versus AH). The tumor profile appears to be condensed compared to the AH profile. At the SNV level, the tumor VAF for the p.(E440*) mutation in case 49 was only 12.7%, whereas the AH VAF was much higher at 43.1% (Fig. 2.3). These findings further support the possibility of non-RB cells, such as associated retinal cells contaminating the tumor sample during preparation; this will be especially likely in secondarily enucleated eyes wherein the majority of 25 the tumor has undergone necrosis, and the specimen is highly calcified. We also identified a germline SNV in case 13 (p.(M148Vfs*8)) that was confirmed by our analysis (Table 2.1). However, the VAF of this mutation was found to be significantly lower in the tumor sample (6.7%) than in the AH sample (56.5%; Fig. 2.3). We speculate that this discrepancy could be attributed to the presence of more non-cancerous cells in the tumor sample, as evidenced by the respective copy number profiles of the tumor and AH samples (Fig. 2.S1: Case 13 tumor versus Case 13 AH). Specifically, the tumor samples appeared nearly diploid, while the AH samples displayed clear RB-SCNA signatures. While the blood has been utilized to profile RB1 somatic SNVs [67], the data presented here suggests the AH is superior due to its high tumor fraction. Even with the high quality data observed in the AH, a substantial amount of signal from the disease-driving splicing variant from case 33-es (c.2325 + 1G>A) was lost following treatment, which underlines the challenge and needs for high DNA concentrations for mutational analysis [68]. Though direct AH-blood SCNA comparisons have been done previously [69], further direct comparisons between the blood, AH, and tumor in SNV analysis for RB are needed. While this method provided promising results, improvements can be made that can increase its accuracy. For example, probe optimization would benefit future studies. An alternative probe design, which encompasses boundaries slightly outside the genes of interest, could increase efficiency. Extending probe coverage would also allow for the profiling of potential regulatory sequences in the intronic regions of the genes. Lastly, deep sequencing in an unbiased approach, such as whole exome or whole genome sequencing, would be more expensive but may uncover disease driving mutants outside the scope of the four genes we profiled in this study. Retinoblastoma tumor tissue cannot be biopsied due to the risk of metastatic spread, which means that obtaining tumor tissue for molecular analysis requires surgical removal of the eye. However, the primary objective of therapy is to treat cancer while preserving the eye, which 26 precludes access to tumor tissue. Our analysis further supports the realization that AH samples collected in vivo can serve as a liquid biopsy for RB, and we have demonstrated the feasibility and accuracy of a combined SCNA-SNV analysis from a single sample and single targeted assay in a time-efficient and cost-effective manner [27, 70]. These findings represent a significant clinical advance in the diagnosis and treatment of RB, with the potential to improve outcomes and quality of life for patients and their families. 2.5 Materials and Methods Statement of research ethics: The established biorepository and collection of coded clinical data was approved by the Children’s Hospital Los Angeles Institutional Review Board. This study adhered to the tenets of the Declaration of Helsinki and was in accordance with the Health Insurance Portability and Accountability Act. All patients provided written informed consent via a legal guardian for all procedures performed. Sample collection and processing: Biospecimens were collected as previously reported [64]. Following specimen extraction, samples were stored at −80 °C until processing. All samples underwent cfDNA isolation and sequencing within 1 month of extraction. cfDNA extraction and processing was previously described [64]. Probe design and efficiency: Targeted SNV probes were designed by Agilent SureSelect DNA (Agilent, Santa Clara, CA, USA) and covered portions of the RB1, MYCN, BCOR, and CREBBP genes in accordance with the human genome reference version hg19. Probes summed to cover 55,097 bp in total (Supplementary Materials). Copy number alteration analysis: Copy number variation was performed on the targeted reads with the R package CopywriteR (version 3.16)—which takes advantage of off-target reads in targeted sequencing data—using the suggested parameters [71]. Samples were aligned with 27 BWA-MEM (version 0.7.17). Bins were set to 500 kb for the hg19 genome, and a female gDNA control with no copy number alterations was used as a baseline for all samples (Fig. 2.S1). WGS reads were analyzed as previously described [72] with some modifications. Briefly, samples were 150 base-pairs paired-end sequenced at a depth of 1–2 million reads on Illumina HiSeq 4000. Sequencing reads were aligned via BWA-MEM to the hg19 reference. Reads were mapped to 5000 bins spanning the human genome and were then normalized for GC content. Count data were segmented via the R package DNAcopy (version 1.70.0). SCNA concordance determination: We measured concordance between copy number profiles as previously described [59]. For comparing the targeted versus low-pass sequencing data, we divided the segmented reads from the targeted sample by the WGS sample. Analogously, for comparing tumors to AH samples, segmented means were divided for the tumor by the segmented means for the respective AH. Samples in question were considered concordant if their ratio was between 0.8–1.2 within each bin. Concordant bins were then set to 1; bins whose ratios were either below 0.8 or above 1.2 were considered discordant and set to 0. Each bin was then normalized for size and then summed together to give a final concordance value between 0–1. RB-SCNA signatures—defined here as 1q gain, 2p gain, 6p gain, 13q loss, and/or 16q loss—were manually inspected for concordance between tumor and AH samples in both the shallow and targeted sequencing approaches. Other RB-SCNA (i.e., 17q gain and 7q gain) signatures may exist [63], but we did not evaluate the signature concordance between samples for those alterations. Concordance was not calculated for three pairs of AH-targeted and low-pass WGS samples from cases 11, 49, and 50. These samples were single-end 50 bp sequenced; therefore, concordance metrics would not reflect the rest of the cohort that was paired-end 150 bp sequenced (Fig. 2.S1). 28 Single nucleotide variant calling: Targeted samples were paired-end sequenced at 150 bp on an Illumina HiSeq 4000 (Fulgent, Inc., Temple City, CA, USA). FastQ (version 2.0.1) files were aligned to hg19 with BWA-MEM (version 0.7.17). Reads falling within the probed regions (RB1, BCOR, CREBBP, and MYCN) were considered. SAMtools (version 1.15-1) was used to sort by coordinates [73]. Picard (version 2.27.4) was utilized to mark duplicates. The bam files were then indexed via SAMtools and utilized as the input for the Mutect2 (version 4-4.3) pipeline. For Mutect2 analysis, both the germline resource and panel of normals provided by GATK were utilized [74]. The read orientation (via the learn orientation model), pile-up summaries, and contamination artifacts were all calculated and utilized for the filtering of Mutect2 variant calls. The variant effect was determined through the Annovar pipeline (version 2020-06-08) [75], and specifically, SIFT, Polyphen2, and gnomAD scores were utilized to classify mutation severity (i.e., deleterious or tolerable) [76–78]. Unless otherwise specified, mutations of interest were then filtered to require a read depth greater than 10. Final mutation calls were then viewed through the Integrated Genome Viewer to confirm the calls [79]. For further validation, all Mutect2 calls were cross-checked with an internal bioinformatic pipeline at the CHLA Center for Personalized Medicine. Funding: This study is supported by The Miriam and Sheldon G. Adelson Medical Research Foundation (M.J.S.; R.K.P.). J.L.B has grant support from the following: National Cancer Institute of the National Institute of Health (K08CA232344), The Wright Foundation, Chil-dren’s Oncology, Group/St. Baldrick’s Foundation, The Knights Templar Eye Foundation, Hyundai Hope on Wheels, Danhaki Family Foundation, Childhood Eye Cancer Trust, and Children’s Cancer Research Fund. Other research support comes from The Berle & Lucy Adams Chair in Cancer Re-search, The Larry and Celia Moh Foundation, A. Linn Murphree, MD, Chair in Ocular Oncology, The Institute for Families, Inc., Children’s Hospital Los Angeles, and an unrestricted departmental grant from Research to Prevent Blindness. Drs. Berry and Xu 29 have filed a provisional patent appli-cation entitled Aqueous Humor Cell-Free DNA for Diagnostic and Prognostic Evaluation of Oph-thalmic Disease 62/654,160 (Berry, Xu, Hicks). Institutional Review Board Statement: This study was conducted under the Institutional Review Board approval CHLA-17-0028 at Children’s Hospital Los Angeles and conformed to the requirements of the United States Insurance and Privacy Act and to the tenants of the Declaration of Helsinki. Approval data was 6/23/2017 and has been renewed annually. All patients included in the analysis provided written informed consent for an Institutional Review Board-approved biorepository at Children’s Hospital Los Angeles via a legal guardian. Informed Consent Statement: Written informed consent was obtained from patients included in this study via a legal guardian. 30 Chapter 3: Genomic, Transcriptomic, and Morphometric Profiling of Single Rare Cells (GeTMoR) Authors: Rishvanth K. Prabakar*, Michael J. Schmidt*, Peter Kuhn, and James Hicks *authors contributed equally This chapter is to be submitted to BioRxiv, and the detailed GeTMoR protocol in the methods section is to be submitted to Cell Stars Protocols. I established proof of principle for all wet lab work presented in this Chapter and was the first to get the dT25 method of simultaneous DNA and RNA extraction working in the lab. I am solely responsible for the acquisition and analysis of all data (wet lab and dry lab work) for figures 3-5 presented in this chapter. I did not contribute to the CTC image analysis code presented in the detailed protocol section, but led all other aspects of the protocol. Full Author Contributions: Conceptualization, R.K.P, M.J.S., and J.H.; methodology, R.K.P, M.J.S., and J.H.; validation, R.K.P, M.J.S., and J.H.; formal analysis, R.K.P and M.J.S. ; image analysis and code: R.K.P; investigation, R.K.P, M.J.S., and J.H.; resources, J.H. and P.K.; writing—original draft preparation, M.J.S.; writing—review and editing, R.K.P, M.J.S., and J.H.; visualization, R.K.P and M.J.S.; supervision, J.H.; funding acquisition, P.K. and J.H. 31 3.1 Abstract Circulating tumor cells (CTCs) are extremely rare cells that seed metastatic tumors. A diverse phenotype of CTCs including the classic cytokeratin expressing CTCs, CTC clusters, large polyploid CTCs, and CTCs undergoing epithelial to mesenchymal transition have been observed. Despite their importance and diversity, very little is known about their functionality and their use for diagnostics. A method to simultaneously profile morphology, genome, and transcriptome of CTCs would provide insight into the gene expression and clonal lineage of CTCs, and ultimately reveal their contribution to the metastatic process. The challenge is that CTCs are hidden in the midst of millions of leukocytes, and detecting these cells necessitates extensive processing steps during which the genomic and transcriptomic content needs to be preserved for downstream single cell analysis. We developed GEnomic, Transcriptomic, and MOrphological profiling of Rare cells (GeTMoR), a method that identifies rare cells by imaging millions of cells in a sample to detect ones that differ from the majority, which are extracted for downstream single cell genomic and transcriptomic analysis. GeTMoR is conceptually based on the HDSCA approach of detecting CTCs and is optimized for preserving RNA content of cells. We validated our protocol by spiking in cancer cell lines into whole blood to evaluate the quality of recovered transcripts and DNA quality. The GeTMoR approach provides the ability to link the phenotype of CTCs to their genome and transcriptome, thereby enabling interrogating biology and the diagnostic potential of CTCs. We have provided a detailed description on the GeTMoR protocol and made all the related software freely available. While we have not validated GeTMoR on blood obtained from patient samples to detect and profile CTCs, we believe that our approach will enable researchers to readily apply GeTMoR to patient samples. 32 3.2 Introduction Circulating tumor cells (CTCs) are thought to play a crucial role in the complex process of seeding metastatic tumors from primary tumors [80-81]. Metastatic seeding involves intravasation of tumor cells from the primary tumors, passage of cells through the circulatory system, followed by extravasation at a metastatic site, and seeding the new tumor [82]. Despite the crucial role CTCs play in metastatic formation, little is known about their functionality owing to the rarity of these cells, and the difficulty in capturing them in such a way that they are amenable to the requisite downstream analysis in a single cell. As examples, a few questions that remain unanswered are (1) the gene expression changes in CTCs in relation to the cells in the primary and metastatic tumor that are from the same genomic clone; (2) The relation between the copy number, gene expression, and morphology of CTCs; (3) The survival mechanisms that CTCs employ to remain in the circulatory system. Peripheral blood, on average, consists of 4-6M red blood cells and 4.5M white blood cells (typically 41.5-65% granulocytes, 20-40% lymphocytes, and 2-8% monocytes) per mL. CTCs, if present, are typically in the order of 1-10 cells per mL [80, 83]. Their rarity necessitates using specific properties of CTCs that differentiate them from other blood cells in order to detect CTCs. These include selection techniques via expression of epithelial proteins, such as cytokeratin or EpCAM and the absence of leukocyte markers such as CD45, or through physical properties such as the larger size or deformability of CTCs [84–88]. Given the rarity of CTCs, the sensitivity of a CTC detection method is required to be orders of magnitude higher than that offered by traditional approaches for selecting a specific cell population including flow cytometry. The challenge in performing any molecular analysis on CTCs is that the required analyte needs to be preserved during extensive protocols for isolation. This is especially challenging for RNA which is known to be easily susceptible to RNAse degradation and leakage from permeabilized cells. In spite of these challenges, several CTC detection methods have been used for single-cell RNA-seq analysis. However, these approaches have several limitations. 33 First, they rely on live cells which limits the CTC detection method to using the size of CTCs or using the expression of cell surface markers. Second, they do not perform simultaneous profiling or multiple analyses from CTCs, thereby limiting the information gained from CTCs. For example, simultaneous profiling of the genome and transcriptome would enable understating the relationship between copy number and gene expression changes in CTCs and allow for clonality analysis in relation to primary tumors and other CTCs. Lastly, these approaches often rely on proprietary or expensive instrumentation limiting the availability and making platform customization of the platform for a specific application difficult. Few platforms enable the isolation and capture of CTCs that robustly preserve RNA and do not require customized instrumentation. For example, Hydro-Seq [89], CTC-iChip [90-92], and the MagSweeper [93] have been used to isolate and subsequently profile the transcriptome from single CTCs. While providing pivotal information on understanding CTC heterogeneity, these studies utilize technology that is not standardized nor easily accessible. Further, CellSearch and Parasortix have been used to profile CTC expression signatures based on cell surface marker expression and morphology [94], respectively, but lack high resolution imaging of CTCs. All technologies and techniques listed are limited in only profiling a subset of cells. A standard approach to image all cells and then isolate CTCs for simultaneous copy number profiling and transcript expression analysis would provide new frontiers for cancer researchers. Further, the simultaneous profiling would also aid in more accurately subtyping tumors using CTCs using approaches similar to PAM50 or intClust used for primary tumors. The GeTMoR approach is conceptually based on the High Definition Single Cell Assay (HDSCA) enrichment free approach for detecting tumor related rare cells in circulation [85]. HDSCA depletes RBCs via ammonium chloride lysis and plates all remaining nucleated cells on a specialized adhesion slide, followed by immunofluorescence staining with WBC and epithelial specific markers, imaging the entire slide, identifying potential rare cells, and extracting the rare cells for downstream genomic analysis if needed. This approach has been used to enumerate 34 rare cells found in patients and has determined tumor lineage via single cell copy number analysis on these cells. We re-engineered this approach to perform transcriptomic analysis on the detected cells together with genomic and morphological characterization of rare cells. 3.3 Results GeTMoR for genomic, transcriptomic, and morphological characterization of rare cells The GetMoR approach to detecting and profiling rare cells is conceptually based on the HDSCA approach of plating leukocytes on a microscopic slide, staining the cells with WBC and tumor specific markers, automated imaging and image analysis to detect rare cells, followed by picking these cells with a micro-manipulator for downstream single cell analysis. The rarity of CTCs necessitates these extensive pre-processing steps to detect and isolate CTCs. The HDSCA approach is optimized for CTC detection and for genomic analysis. We developed GetMoR to preserve the transcriptomic content of CTCs while at the same time having a high sensitivity of detection of CTCs. Blood is collected in an EDTA collection tube and can be stored or transported at room temperature and must be processed within 24 hours of collection (Fig. 3.1). The blood sample is processed using a standard FICOLL plaque separation, and the mononuclear cells are extracted and washed in a PBS-based buffer. The cells are counted and plated on an adhesion slide at approximately 3 million cells per slide. Multiple slides can be plated as needed. The cells are fixed with 2% paraformaldehyde, permeabilized with 0.01% Saponin, and stained with the nuclear marker DAPI, epithelial markers pan-cytokeratin and EpCAM, mesenchymal marker Vimentin, and leukocyte marker CD45. The number of markers used is limited by the IF microscope used in the next step. 35 Figure 3.1: GeTMoR blood processing, slide plating, imaging, and single cell isolation schematic. These steps must be performed on the same day blood is received. The slide is loaded onto an IF microscope fitted with an automated stage. The entire slide is imaged generating 2304 frames and up to four channels per frame. The choice of channels to image depends on the downstream application. For detecting CTCs, imaging just on the cytokeratin marker channel is sufficient and saves time. The frame images are segmented to produce images of single cells, with each frame consisting of approximately 1500 cells. The segmented regions are used to obtain a feature vector consisting of the mean intensities of the four channels and the size and shape of the cell. A frame consists mostly of WBCs with at most a few rare cells per frame. These rare cells are detected using a k-nearest neighbor outlier detection method. The coordinates of these rare cells are used to locate them on the slide, which are then picked from the slide into a PCR tube using a micro-manipulator. The isolated single cells can potentially be used for any downstream single cell sequencing approaches. We simultaneously extracted (1) genome for whole genome copy number profiling and (2) transcriptome for full length mRNA profiling using an extended version of FRISCR [95] (Fig. 3.2). GeTMoR could also be used in isolation for only CTC detection, CTC copy number profiling, transcript profiling or any combination of them. 36 Figure 3.2: GeTMoR separation of RNA and DNA uses an adapted FRISCR approach. PFA cross-linked biomolecules are reverse cross-linked via proteinase-K at 56◦C for 1 hour. Poly-adenylated RNA molecules are then hybridized to poly-thymine beads. The supernatant is collected, which contains DNA, while the bound RNA remains in the tube for reverse transcription via SMART-seq2. Transcriptome and genome integrity of spiked in cells We validated GeTMoR by spiking in cell lines into normal blood donor samples, performing all the steps of the GeTMoR protocol to detect and pick the spiked cells. A mix of approximately 1:10 MDA-MD-231 and PC3-GFP cells were spiked into normal blood samples obtained in EDTA tubes. All GeTMoR steps (described in the protocol section) were performed. We performed two replicates of this experiment. 35 cells were picked for genomics and transcriptomics in the first experiment and 11 cells were picked for the second experiment. The integrity of the genome and transcriptome of the spiked cells were determined by comparing them to control cell lines that were plated, fixed, and picked directly without going through the GeTMoR protocol (no blood exposure or FICOLL separation). We first sought to measure the ability of GeTMoR to retain RNA quality of spiked in cell lines. To measure transcript integrity, the percentage of 5’ to 3’ coverage was measured for each condition, with increased 3’ coverage indicating degradation. We found that both spiked replicates displayed no substantial 3’ biases and were comparable with controls (Fig. 3.3A). As another quality metric, we measured the percentage of bases representing mRNA transcripts – denoted as exonic and UTR regions – and found they were comparable for controls and the spiked in samples (Fig. 3.3B). While a slight decrease in regions mapping to mRNA was observed for replicate 2, indicating slight sample degradation (Fig. 3.3A-B), we found at least 37 2,500 protein coding genes were detected for each spike replicate, which provides enough information for informative analysis (Fig. 3.3C). To classify spiked cells as either MDA-MB-231 or PC3, we performed marker analysis on the control samples. We then visualized normalized expression values of the cell type markers and found spiked-in samples clustered with control samples in anticipated ratios (Fig. 3.3D). UMAP visualization confirmed a clear separation of MDA-MB-231 and PC3 cells, as well as spiked-in cells that clustered with their respective cell types (Fig. 3.3E). Taken together, this data shows GeTMoR processing retains transcript quality in spiked-in cells and allows for cell type classification compared to a control ground truth data set. 38 Figure 3.3: Recovered transcript quality of GeTMoR processed cells. (A) Normalized transcript coverage from 5’ to 3’ end for each condition. 3’ accumulation indicates degradation. Color key refers to all panels in the figure. (B) Percentage of bases mapped to the genome. (C) Total expressed protein coding genes for control MDA-MB-231 and PC3 samples, as well as GeTMoR-processed replicates 1 and 2. (D) Clustering analysis for all samples using identified marker genes between PC3 and MDA-MB-231 control samples. (E) UMAP visualization of all samples. As expected from the spiked-in ratios, we recovered more PC3 than MDA-MB-231 cells. This was also confirmed through single cell copy number profiling from the same cells where RNA was extracted from (Fig. 3.4). There was a group of cells that we could not confidently classify as MDA-MB-231 or PC3, likely due to poor sample quality that failed both RNA and 39 DNA single cell sequencing. Despite some anticipated failures in sample preparation and extraction, a majority of cells were confidently identified to a cell lineage showing the ability of GeTMoR to preserve the transcripts of most cells. Figure 3.4: GeTMoR retains DNA copy number for MDA-MB-231 and PC3 spiked cell lines. Single cell segmented ratio copy number profiles from same single cells presented in Figure 3. 40 The complementary components of GeTMoR enable unprecedented analysis of single cells. In addition to the transcriptome, GeTMoR provides morphological and copy number data, which can be used to determine clonal lineage. As a proof of principle, we used PC3-GFP spiked cells to distinguish MDA-MB-231 from PC3 cells through fluorescence imaging. Through image analysis, PC3 cells were identified via GFP positivity, while MDA-MB-231 cells were negative in the GFP channel (Fig. 3.5A). Copy number ratio data further separated MDA-MB231 from PC3 cells and confirmed image analysis (Fig. 3.5B). Given copy number and gene expression data from single cells, we theorize GeTMoR can allow for numerous unprecedented analyses. For example, to visualize how copy number can influence gene expression, we plotted experimentally derived chromosomal copy number values by TPM (Fig. 3.5C). For this data set copy number neutral and gained regions display higher gene expression values than genomic regions with fewer chromosomal copies. 41 Figure 3.5: Examples of GeTMoR identified MDA-MB-231 and PC3-GFP cells. (A) Representative image of MDAMB-231 and PC3 spiked in cells processed with GeTMoR and imaged in TRITC and FITC channels. (B) Copy number profiles from cells isolated in (A). (C) Gene expression (TPM) profiles plotted against chromosomal location and copy number ratio for cells in (A). 3.4 Discussion GeTMoR approach offers several advantages over other methods for isolating CTCs for molecular analysis: (1) We can extract both RNA and DNA from the same CTCs, and our approach can potentially be extended to other assays such as CITE-seq for protein expression analysis or methylation profiling of the extracted DNA. (2) Our assay allows for fixing and permeabilizing cells and thus we can stain for cytoplasmic proteins for CTC detection without the worry of substantial transcript loss; we have stained with cytoplasmic pan-CK and Vimentin. (3) Our approach provides multiples views of the phenotype (IF image and RNA) and genotype 42 (DNA) for each cell, thus enabling joint analysis for these data modalities. (4) Our protocol and all software for image analysis is freely available to enable enhancements and adaptations to the GeTMoR protocol. There are many benefits to GeTMoR, but some limitations exist. First off, to identify and subsequently isolate rare cells GeTMoR is best used with a high throughput imaging apparatus and a micromanipulator for single cell isolation. While available to the public, these may not be affordable or suitable for all lab environments. Further, to ensure transcript quality GeTMoR blood processing and single cell isolation must be performed on the same day the sample is received; it is well known that not every cell passes RNA-QC primarily due to transcript degradation. While vast improvements have been made to GeTMoR to enhance the chances of successfully capturing intact transcripts, there will inevitably be some cells that fail. Further, to expedite the time required before single cells are isolated and to minimize transcript degradation, GeTMoR only images in one or two fluorescent channels, with preferably only one channel being scanned. While one to two channels still provide detailed morphological data, we acknowledge the user may want to visualize more protein markers. Imaging additional channels will require further method optimization for the user specific purposes (i.e., image a smaller area of a slide so the protocol duration is not extended substantially). Lastly, the specified parameters for image analysis expects a few rare cells per image frame and will need to be refined by the user for the number of rare cells they anticipate in each frame. While studying the transcriptome of CTCs and other rare cells is challenging, GeTMoR offers a robust technique to obtain as much information as possible from a single cell. Capturing the genomic content to understand clonal lineage, transcriptome to understand the state of the cell, and morphology to visualize cellular size and texture allows for unprecedented multi-modal analysis that has the potential to advance numerous scientific fields. 43 3.5 Methods Cell line collection and blood processing: PC3 and MDA-MB-231 cell lines were purchased from ATCC and grown in RPMI and DMEM, respectively, with 10% FBS and with 0.5% penicillin / streptavidin. GFP-PC3 was purchased from Angioproteome and grown in RPMI (catalog: cAP-0067GFP). Cells were seeded two days prior to each experiment in a T-25 flask and collected at ~90% confluency. Cells were lifted with 1x versene, spun down at 1000 RPM for 5 minutes, and resuspended in 1X PBS supplemented with RNAse inhibitor. Patient blood was diluted in 1:1 in 1X PBS with RNAse inhibitor. Cell lines were spiked into the diluted blood. The blood mixture was then added to 15mL of FICOLL and spun down at 400G for 20 minutes. Nucleated cells on top of the FICOLL layer were washed again in 1X PBS with RNAse inhibitor and spun down at 1000 RON for 5 minutes. Cells were then plated on specialized Marienfield glass slides (Marienfeld, Lauda, Germany) and incubated at 37◦C for 15 minutes. Immunofluorescence staining: Slides were removed from the incubator and immediately fixed with 2% paraformaldehyde for 10 minutes. Slides were blocked with 2% BSA in PBS and 4′,6-diamidino-2-phenylindole (DAPI; D1306, Thermo Fisher Scientific, Waltham, MA, USA) for 5 minutes. Then, slides were incubated at room temperature with a primary antibody cocktail consisting of mouse IgG1/Ig2a anti-human cytokeratins (CK) 1, 4, 5, 6, 8, 10, 13, 18, and 19 (clones: C-11, PCK-26, CY-90, KS-1A3, M20, A53-B/A2, C2562, Sigma, St. Louis, MO, USA), mouse IgG1 anti-human CK 19 (clone: RCK108, GA61561-2, Dako, Carpinteria, CA, USA), and mouse EpCAM. The CK and EpCAM antibodies make up the EPI channel in patient stains and cell lines. Conjugated mouse anti-human CD45 Alexa Fluor® 647 (clone: F10-89-4, MCA87A647, AbD Serotec, Raleigh, NC, USA), rabbit IgG anti-human vimentin (Vim or V) (clone: D21H3, 9854BC, Cell Signaling, Danvers, MA, USA), and DAPI were also added to the primary incubation buffers. Slides were then washed in 2% BSA in PBS and incubated at room temperature for 10 minutes with Alexa Fluor® 555 goat anti-mouse IgG1 44 antibody (A21127, Invitrogen, Carlsbad, CA, USA), conjugated mouse anti-human CD45 Alexa Fluor® 647, VIM, and DAPI. Slides were washed, coverslipped, then immediately scanned. High content scanning and image analysis: Automated high-throughput fluorescence scanning microscopy was performed in the TRITC (CK) and FITC (GFP) channels with a 10x objective lens, collecting 2304 frames per slide. Images were immediately analyzed with C++ software. Single cell isolation: Rare were identified by X and Y coordinates on the slide and lifted from slides using a robotic micromanipulator system (Eppendorf). Individual cells were deposited into PCR strips containing 1x PBS with 0.2% Triton X-100 and RNAse inhibitor. Cells were then stored on dry ice and then placed in the -80◦C until RNA and DNA extraction. DNA and RNA extraction: RNA and DNA were first reverse cross-linked via a modified FRISCR approach [95]. Briefly, lysis buffer was added to the cell (0.2% Triton X-100 + RNase inhibitor) and incubated at 56◦C for 1 hour. dT25 beads (NEB) were then added to the samples and hybridized at 56◦C for 1 minute. Samples cooled at room temperature for 10 minutes then added to a magnetic PCR rack. DNA was collected in the supernatant while RNA remained attached to the beads. DNA was removed in the supernatant and was subsequently amplified with single cell whole genome amplification (WGA; Sigma-Aldrich; Cat# WGA4). Libraries were constructed using the DNA Ultra FSII Library Prep Kit (New England Biolabs; Cat# E7430). Cells were sequenced paired end by 150 base-pairs on an Illumina HiSeq 4000 (Fulgent) at a depth of 1-2 million reads per sample. RNA was washed once with 2x SSPE buffer and once with 1x PBS, and then eluted in RNAse free water at 80◦C. Purified RNA then underwent a modified Smart-Seq2 protocol [96]. Briefly, dT primer was added to the purified RNA and allowed to hybridize for 3 minutes at 72◦C. Reverse transcription and template switching then took place at 42◦C for 90 minutes, followed by 10 rounds of cycling from 50◦C for 2 minutes and 42◦C for 2 minutes. cDNA was then 45 amplified and sequencing libraries were prepared with Nextera XT (Illumina). Cells were sequenced paired end by 150 base-pairs on an Illumina HiSeq 4000 (Fulgent) at a depth of 1-2 million reads per sample. Single cell RNA-sequencing: Read adapters were trimmed with TrimGalore (version 0.6.7) and aligned with the HiSat2 (version 2.2.1). Picard (version 3.0.0) was used to visualize RNA mapping quality control. HTSeq (version 2.0.2) was used to generate a gene count matrix. Downstream analysis was performed with R (version 4.1.2). The SingleCellExperiment package (version 4.2.2) was utilized for inputting count data into downstream analyses. Seurat (version 4.3.0) FindMarkers tools was used to compare PC3 and MDA-MB-231 controls. Marker genes. Single cell copy number profiling: Copy number profiling from low pass whole genome sequencing samples was conducted as previously described [28, 72]. Sequencing reads were aligned with BWA-MEM to the hg19 reference. Count data was segmented via the R package DNACopy (version 1.70.0), and median segmented ratio values were reported. Copy number concordance: Copy number concordance between control and spiked in samples were conducted as previously described ([28, 72]). Briefly, for comparing spiked cells to control cells, segmented means were divided for the control by the segmented means for the spiked cells. Samples in question were considered concordant if their ratio was between 0.8–1.2 within each bin. Concordant bins were then set to 1; bins whose ratios were either below 0.8 or above 1.2 were considered discordant and set to 0. Each bin was then normalized for size and then summed together to give a final concordance value between 0–1. 46 3.6 Full GeTMoR protocol This protocol describes detecting CTCs that are positive for epithelial markers from a blood draw, isolation of the CTCs, RNA and DNA extraction, modified Nextera XT library preparation for RNA-sequencing (1/5th reactions), and modified WGA4 single cell amplification for copy number profiling. Downstream RNA/DNA steps can be modulated to user preferences. Equipment Required: ● Microwave ● Centrifuge (15mL and 50mL tube racks) ● Automated scanner with GeTMoR application installed on computer ● Picking apparatus for single cell isolation ● Laminar flow hood ● 96-well magnetic plate ● 96-well cooler plate Reagents: ● Nuclease free water (Invitrogen: 10977-015) ● 10x PBS (Invitrogen: 70011-044) ● EDTA (Sigma: 03690) ● Recombinant RNAse inhibitor (NEB) ● RNAse Away (VWR: 7003) ● RNAse free BSA ● Ficoll paque premium (VWR: 17544202) ● Saponin (Sigma: 47036) ● dT25 magnetic beads (NEB: S1419S) ● 20x SSPE hybridization buffer (Sigma: S2015-1L) ● Mouse IgG1/IgG2a anti-human cytokeratin (CK) 1, 4, 5, 6, 8, 10, 13, 18, and 19 (clones: C-11, PCK-26, CY-90, KS-1A3, M20, A53-B/A2, C2562, Sigma) ● Mouse IgG1 anti-human CK 19 (clone: RCK108, GA61561–2, Dako) ● CD326 (EpCAM) Monoclonal Antibody (1B7), eBioscience (Thermo) ● Mouse antihuman CD45:Alexa Fluor 647 (clone: F10–89–4, MCA87A647, AbD Serotec) ● Rabbit IgG antihuman vimentin (VIM): Alexa Fluor 488 (clone: D21H3, 9854BC, Cell Signaling Technology) ● Alexa Fluor 555 goat anti-mouse IgG1 antibody (A21127, Invitrogen) ● 4,6-diamidino-2-phenylindole (DAPI; D1306, Thermo Fisher Scientific) ● Slide coverslips ● Triton-X 100 ● 50% Tween 20 (Life Technologies: 003005) ● SmartScribe Reverse transcriptase (Takara) ● dNTP mix (NEB: N0447S) 47 ● dT primer: (5–AAGCAGTGGTATCAACGCAGAGTACT(30)VN-3) ordered from IDT ● TSO primer: (5-AAGCAGTGGTATCAACGCAGAGTACATrGrG+G-3) ordered from IDT ● PCR primer: (5-AAGCAGTGGTATCAACGCAGAGT-3) ordered from IDT ● KAPA 2x HiFi mix: (Rosche) ● Nextera XT library preparation kit (Illumina) ● Sigma WGA4 kit ● NEB Ultra FS II library preparation kit Preparation: 1. Be sure to use filtered Eppendorf tips throughout the procedure. 2. Wipe down all bench surfaces, pipettes, wash jars, and equipment with RNAse inhibitors. If possible, UV irradiate all wash jars and pipettes. Rinse all wash jars with nuclease free PBS to remove any residual RNAse inhibitors. 3. 2% PFA preparation: Added 3.5mL of 1x PBS in a Falcon tube and add 0.5mL of 16% PFA. This can be made when preparing staining buffers during the 20 minute spin down of FICOLL. Blood processing: 1. Prepare processing buffer: To 50mL of 1x PBS 50mL falcon tube add 100µL of 0.5M EDTA. 2. Add 15mL of FICOLL into a 50mL falcon tube. 3. Dilute blood in the processing buffer. In a new 50mL Falcon tube, add equal amounts of blood to the processing buffer (i.e., to process 5mL of blood, add in 5mL of processing buffer). For every mL of processing buffer add 0.5µL of RNAse inhibitor. Mix by inverting the tube. a. Note: 2mL of blood is used to make 1 Marienfield slide. Therefore, in most cases you should add 2mL of blood, 2mL of 1x PBS, and 1µL of RNAse inhibitor. 4. Very gently pipette the diluted blood on top of the FICOLL layer, being careful not to mix the layers. 5. Centrifuge at 400xG for 20 min at room temperature with acceleration set to 3 and deceleration set to 0. 6. During spin down, go to the cell plating section and prepare slides for plating. Also go to the staining section and prepare the blocking, primary, and secondary solutions. 7. At the end of the centrifugation, add 3mL of processing buffer to a 15mL falcon tube and add 3µL RNAse inhibitor. This will be the washing buffer of FICOLL separated cells. 8. Using the 15mL falcon tube, carefully pipette the entire PBMC layer into the processing buffer, being careful not to pipette any FICOLL and avoiding the plasma layer. Mix by gently inverting the tube. 9. Centrifuge at 1000 RPM for 5 min at room temperature with acceleration and deceleration set to 3. 10. Pipette out and discard the supernatant being careful not to disturb the cell pellet. 11. Resuspend cells in 1mL of processing buffer (with RNAse inhibitor) per slide and gently mix by pipetting. The required volume depends on the plating density of the slides and 48 the cell count. If you are processing 1 slide and using 2mL of blood, then add 1mL of processing buffer. Cell plating: 1. (Steps 1 to 4 can be performed during the FICOLL centrifugation step) Add slide to glass coplin jar. Add RNAse free water to a wash jar and add 4µL of RNAse inhibitor. Microwave on for 30s seconds. 2. Add 1x PBS to a wash jar and add 4µL RNAse inhibitor. 3. Transfer the slides to the wash jar containing PBS. 4. Remove excess PBS from the slide and add 750µL of resuspended cells on the slide. 5. Incubate at 37◦C for 15 min to let the cells adhere to the slide. 6. Remove from incubate and discard excess buffer from the slide. Fixing and staining: 1. Prepare 2% PFA. Add 1mL of 2% PFA on the slide and let the cells fix for 10 min at room temperature. 2. Prepare 2x PBS DAPI: add 5mL of 10x PBS to 20mL of RNAse free water to make 2x PBS. Then add 1µL of DAPI. Mix and use as base buffer for the staining solutions. 3. Blocking solution preparation: 500µL of 2xPBS-DAPI, 100µL of 10% BSA, 396µL of nuclease-free water, 4µL of RNAse inhibitor, and 10µL if 1% Saponin. 4. Primary antibody solution preparation: 500µL of 2xPBS-DAPI, 100µL of 10% BSA, 361µL of nuclease free water, 4µL of RNAse inhibitor, 10µL of primary pan-CK antibody, 5µL of primary CK-19 antibody, 10µL of EpCAM, 10µL of CD45-Alexa-647 conjugated antibody. 5. Secondary antibody solution preparation: 500µL of 2xPBS-DAPI, 100µL of 10% BSA, 372µL of nuclease-free water, 4µL of RNAse inhibitor, 10µL of CD45-Alexa-647 conjugated antibody, 10µL of Vimentin-Alexa-488 conjugated antibody, and 4µL of Mouse-IgG-Alexa-555 antibody. 6. Add 1x PBS to a wash jar and add 4µL RNAse inhibitor. 7. Remove PFA from the slide, transfer the slide to a wash jar, and wait 30s. 8. Add 1mL blocking solution on the slide and wait 5 min. 9. Remove blocking solution from slide. 10. Add 1mL primary antibody solution on slide and wait 10 min. During this step, go turn on BZ6 to allow for the stage to prepare. Be sure to clean the stage carefully. 11. Add 1x PBS to a wash jar and add 4µL RNAse inhibitor. 12. Remove primary antibody solution, transfer slide to wash jar, and wait 30s. 13. Add 1mL secondary antibody solution on the slide and wait 10 min. 14. Mounting media preparation: To 200µL 1x PBS, add 0.4µL EDTA, and 4µL RNAse inhibitor. 15. Add 1x PBS to a wash jar and add 4µL RNAse inhibitor. 16. Remove secondary antibody solution, transfer slide to wash jar, and wait 30s. 17. Remove excess PBS from the slide, add 200µL mounting media. 18. Coverslip with RNAse free HybriWell coverslips. To coverslip, match the end of the slide with the end of the coverslip (the end of the slide is the opposite side of the barcode). 49 19. Prepare picking buffer by taking 10x lysis buffer (2% Triton X-100) and making 1x working solution (1:10 dilution). Generally, making 200 is sufficient (180µL of RNase free water + 20µL of 2% triton X-100). Place on ice until ready for use. 20. Prepare slide buffer by adding 1µL of 50% tween-20 to 10mL of 1x PBS. Vortex vigorously for at least 10 seconds to mix. Place on ice until ready for use. Scanning: 1. (Steps 1 and 2 can be performed during the secondary staining step) Wipe down the stage and scanner with RNAseAway. 2. Open the GeTMoRScan app, make sure that the stage is clear of the objective lens, and click “yes” in the pop-up box to calibrate the stage. 3. Gently place the slide in the leftmost slot of the stage. Be careful not to move the stage, this could lead to the stage going out of calibration. Place the 10x lens over the slide. 4. Select the channels to scan under “Select channel”. 5. Set the exposure and gain for the selected channels. Typical values are 25ms exposure and 4 gain for DAPI, and 50ms exposure and 4 gain for TRITC. 6. Turn on DAPI by selecting “DAPI” under “Channel”. 7. Open the live preview window by selecting “Live preview”. 8. Click “Next” under “Focal points”, adjust the z-axis to make sure that the cells are in focus, click “Set”. Repeat until all the focal points are set. 9. Click “Move to” under “Fiducial point”. 10. Click on “Brightfield” under “channel”. 11. On the microscope, push “RL” to turn it off and then push “TL” to turn it on. 12. Remove protective mat from the light source. 13. Adjust the stage X and y directions to center the fiducial point in the live preview window and then click “Set”. 14. Click on “Off” under “Channels”. 15. Close the live preview window by deselecting “live preview”. 16. On the microscope, push “TL” to turn it off and then push “RL” to turn it on. 17. Cover the light source with the protective mat. 18. Click on “Start scan” to start scanning the slide. Image segmentation and rare cell detection: The parameters for the programs below are based on the assumption that the cells in the cytokeratin channel are “rare” and so at most 3 cells are present in a frame. If this assumption is not met, the parameters need to be changed accordingly to achieve high sensitivity of detection. 1. Run the command below in a cmd prompt window after 20% of the frames are scanned on the TRITC channel. There is a text file in the GeTMoR directory that has the code and path for everything. $ ck_segment <scan_output_dir> <segmentation_dir> <sample_name> \ 1 2304 2305 5 0.995 0.3 3 1 50 where scanoutput dir is the path to the directory containing scanned images, segmentation dir is the output directory to store the segmented images, and samplename is an identifier for the sample. 2. Run the command $ feature_extraction <scan_output_dir> <segmentation_dir> \ <sample_name> 1 2304 <channel_start> where channel start is a comma separated list of channel start offsets to user for feature extraction. This should be set to "1, 2305" when scanned on DAPI and CK channel and to "2305" when scanned only on the CK channel. 3. Run the command to filter cells based on feature values. $ filter_features <sample_name>_feature_vec.txt \ <sample_name>_feature_vec_filt.txt * where * specifies the column in the feature vector file and the minimum and maximum values in that column to retain. To filter events corresponding to typical cell sizes use "5,100,10000". In addition, filters can also be used for DAPI and CK mean intensities. 4. Run the command to visualize and confirm all the segmented events. $ cell_image <scan_output_dir> <segmentation_dir> \ <sample_name>_feature_vec_filt.txt "1,1,0,0" "2305,1,0,0" 150 Cell relocation and single cell picking: Wipe down all equipment with RNAse away before beginning, including where the picking tube will be placed. Turn on UV and place pipettes and picking tubes under UV for at least 15 minutes. 1. Once segmented events are confirmed, choose which events isolate and write down the cell ID and location from the feature vec file. You can also transfer on one of K-labs shared drives and open it on an ix83 computer. 2. Get an ice bucket with picking and slide buffers. Add 4µL of RNAse inhibitor to each tube. Vortex briefly. 3. Get a new coplin jar and fill with 1x PBS. Add 4µL of RNAse inhibitor. Mix. 4. To remove slide cover, place scanned slide in a coplin jar and carefully peel off cover. Do not go too quickly. 5. Place the slide on the left side of the picking scope, next to an empty slide. 6. Add 1000µL of slide buffer to the slide. 7. At the picking scope (IX83), open the image-J Pro application. 8. At the top of the application, click on the macro drop-down menu: Macro management -> Change -> Scripts -> relocate_picking.pm 9. Select FindCellFromFudicial and click RUN. Use the physical limits of the stage. 51 10. Open Notepad++ (or other editing software) on a computer. Load the relocate picking script. 11. Under the cell frame (4th chunk of code) enter your cell frame of interest. The code should subtract that frame by 1. Save the script. 12. Center the fiducial of the slide. The slides barcode should be facing towards the center of the room. If you’re facing the wall, align the fiducial to the lower left of the slide. 13. Open Macro management. Click the FindCellFromFudicial script. Click reload. Click RUN. 14. The stage should move to find the cell. 15. Once the cell is relocated, add 1µL of picking buffer (on ice) to the PCR tube. Add tube to the scope. Pick the cell and deposit the cell in the tube. Take the tube off carefully, spin it down, and immediately place it on dry ice. 16. At the top of the application, click on the macro drop-down menu: Macro management -> select FindCellFromFudicial -> edit macro 17. Be sure to end the macro on the edit script (red stop button) before going to the next cell. 18. Repeat steps 10-17 until you’ve picked all of your cells of interest. 19. Store cells in -80◦C until ready for RNA extractions. Note: This is the stopping point for processing blood, finding and isolating rare cells. Proceed to the next step for RNA extractions. Cells can be stored in -80◦C for at least 1 month. dt25 RNA extraction: The RNA extraction protocol is based on Smart-seq2 [96] and FRISCR [95]. Before conducting the protocol, please review both papers as it is vital to understand and appreciate the importance of each step. Before beginning, wipe down all equipment with RNAse away, including pipettes, counters, ice bucket, tube holders, and thermocycler. Prepare buffers at designated time points. Only add RNAse inhibitor 15 minutes before use and no sooner. 1. Clean everything with RNAse inhibitor. 2. Get a bucket of ice. 3. Place 96-well magnet in -20◦C. 4. Turn on thermocycler and ready method for 56◦C hybridization. This program should incubate at 56◦C for 1h. Set heated top to 75◦C. Start incubation at step 9. 5. Prepare RNA lysis buffer in a clean Eppendorf tube. The example below is for 12 reactions. Adjust accordingly. a. 8µL of 10x lysis buffer (2% triton X-100; 2x final concentration) b. 2.5µL of Proteinase K (0.0625x final concentration) c. 2.5µL of RNase inhibitor (2U final concentration) d. 27µL of RNase free water (to 40µL) 52 6. Transport cells from the freezer in a 96-well cooler. Once in the RNA station, remove tubes from the cooler and put in a PCR strip. 7. Add 3µL of lysis buffer to each cell. 8. Flick or briefly vortex tubes to mix. Spin down then add tubes to the thermocycler. 9. Incubate at 56◦C for 1 hour. 10. Place 96-well cooler plate in 4◦C. Use this again when preparing the final RT reaction. 11. During the hour incubation, prepare buffers, dT25 bead mix, and hybridization mix. Keep buffers and all solutions on ice. 12. Buffers to prepare: a. 2x wash buffer. To prepare 6mL which is sufficient for 12 single cell reactions: i. 1200µL of 20x SSPE (4x final) ii. 12µL of 50% tween-20 (0.1% final) iii. 3µL of RNase inhibitor (0.05% final) iv. 4785µL of RNase free water b. 1x wash buffer: prepare 1x wash buffer from 2x wash buffer (i.e., take 5mL of 2x wash buffer and 5mL of nuclease free water). Leave 1mL of 2x wash buffer for dt25 bead resuspension. c. 10mL of 1x PBS in nuclease free water. d. 5mL of elute buffer – nuclease free water. 13. dt25 bead preparation: remove 8µL of dt25 beads for each sample and add to clean 200µL PCR or 1.5mL Eppendorf tube. a. Add dt25 beads to the magnet. Wait 2 minutes. Remove supernatant and add 200µL of 1x wash buffer. Vortex. Spin down briefly. Be sure to resuspend the beads so they are washed thoroughly. b. Repeat 200µL of 1x wash 2 more times (3 total times). c. Return beads to the magnet. Wait 2 minutes. Remove 200µL. Pipette all liquid out of the tube while beads are on magnet. d. Add 4µL of 2x wash buffer to beads for each sample (half of bead volume you initially took). e. Keep beads on ice or at 4◦C until ready. f. For example, if you have 100 samples, you will use 800µL of beads and then resuspend the washed beads in 400µL of 2x wash buffer. 14. Primer hybridization mix preparation: a. Remove dNTP mix from -20◦C and the dT primer from -80◦C. b. For each reaction, add 0.5µL of dNTP mix and 1µL of dT primer (with about 20% extra solution). For example, for 32 samples, add 40µL of dT primer and 20µL of dNTP mix. c. In clean PCR strips, add 1.5µL of primer hyb mix to each tube. This is where RNA will be eluted from dT25 beads and added to. d. Label and strips in 4◦C until elution. 15. Remove RT reaction mix (red cap) from -20◦C cooler and keep at room temp until it thaws. 16. Prepare tubes for gDNA supernatant in PCR strips if you wish to save for WGA. Set these aside for bead purification. 53 17. Remove samples from the incubator and place in a regular 96-well PCR strip. Keep at room temperature. 18. Take beads and mix so there is no sedimentation at the bottom of the bead tube. 19. Add 4µL of beads to each sample. Flick tubes and briefly spin down. 20. Incubate at 56◦C for 1 minute. 21. Remove from the thermocycler and place tubes at room temperature on the RNA bench for 5 minutes. 22. Add 5µL of RNAse inhibitor to 1x PBS buffer and Elute buffer. a. The 1x wash buffer has RNAse inhibitor already added from the 2x wash buffer. 23. Set thermocycler to elute program (80◦C for 2 minutes). 24. Place samples with beads in them on a room temp 96 strip magnet for 1 minute. 25. Carefully, without disturbing beads, remove supernatant with 10µL multichannel pipette and add supernatant to prepared tubes for gDNA. You should have 5µL of supernatant. Store supernatant for WGA at 4◦C until RNA extraction is complete. a. Alternatively, if you do not wish to profile the genome, throw away supernatant and continue with washing the beads. 26. Keeping beads on magnet, wash beads twice with 100µL of 1x wash buffer. Add in the buffer, let sit for 30 seconds, then remove and add again. Make sure beads are not disturbed. 27. Remove 1x wash buffer and add 100µL of 1x PBS. 28. With pipette set to 150µL, remove 1x PBS. Try to remove all liquid from tubes. 29. Remove tubes from the magnet and place in a regular 96 tube PCR strip. 30. Add 3.5µL of elute buffer (water + RNase inhibitor). Flick tubes to re-suspend beads. Briefly spin down tubes. 31. Place tubes in thermocycler that’s set to 80◦C. Incubate for 2 minutes. 32. During this time, (1) remove primer hybridization mix from 4◦C and (2) remove 96-well magnetic plate from -20◦C. 33. Remove tubes from the thermocycler and place in an ice cold magnetic stand. 34. Set thermocycler to 72◦C hybridization step (72◦C for 3 minutes and 4◦C for 3 minutes). 35. Using pipette set to 7µL, transfer supernatant to tubes containing primer hybridization mix. Vortex and spin down. Add to thermocycler and incubate for 2◦C for 3 minutes and 4◦C for 3 minutes. 36. Prepare RT reaction mixture. Example below is for 1 sample. a. 2µL 5x first strand buffer b. 0.5µL TSO (template switching oligo) c. 0.25µL RNase inhibitor d. 1µL SMARTScribe Reverse Transcriptase 37. Using cooler stored in 4◦C, remove tubes from thermocycler. Add 3.75µL of RT reaction mixture. Flick to vortex and briefly spin down. 38. Set thermocycler to RT reaction program: ● Incubate at 42◦C for 90 minutes to allow for RT and template switching ● Cycle 10 rounds of: ○ 50◦C for 2 minutes to unfold RNA secondary structures. 54 ○ 42◦C for 2 minutes for completion and continuation of RT and template switching. ● Inactivate enzyme at 70◦C for 15 minutes. ● Hold at 4◦C 39. Add tubes to the thermocycler. 40. cDNA can be stored overnight at 4◦C or at -20◦C for at least 1 month, however it is easiest to move forward with cDNA amplification. 41. If you decide to proceed with WGA on the gDNA fraction, proceed to GeTMoR DNA WGA immediately. Since the gDNA fraction of a cell is isolated and lysed, WGA needs to be conducted on the same day as RT extraction or DNA will be lost. cDNA amplification: This step immediately follows RT conversion to cDNA (step above). It’s easiest and suggested to perform immediately after the RT reaction. 1. Remove tubes from incubator or storage. 2. To each tube, add 15µL of the following master mix solution ● 12.5µL of 2x KAPA mix ● 0.25µL of ISPCR ● 2.25µL of nuclease free water 3. Perform PCR using the KAPA program ● Denature for 1 round at 98◦C for 3 minutes ● Cycle 21 rounds of: ○ Denature at 98◦C for 20 seconds – Anneal at 67◦C for 15 seconds ○ Extend at 72◦C for 6 minutes ● Final extension for 5 minutes at 72◦C ● Hold at 4◦C 4. Tubes can be stored overnight at 4◦C or at -20◦C for at least 1 month, but it is suggested to purify with XP PURE beads before storage. cDNA purification with XPure beads: 1. Equilibrate beads at room temp for 30 minutes prior to use. 2. Add 0.8X beads per reaction – add 15µL of room temp XP beads to each amplified cDNA reaction. 3. Incubate at room temperature for 8 minutes. 4. Add beads to the magnetic stand for 5 minutes or until the solution is completely clear of bead mix. 5. Remove supernatant and discard. 6. Add 200µL of freshly prepared 80% ethanol. Incubate for 30 seconds. And remove. 7. Repeat step 6 (above). 8. Remove all liquid from the tube and incubate on the magnet for 3 minutes (DO NOT OVER-DRY BEADS). 9. Add 14µL of RNase free water. Vortex vigorously. Pat tubes on the counter to try and make sure beads settle towards the bottom. 10. Incubate for 5 minutes. Then spin down tubes and place on the magnet for 3 minutes. 55 11. Elute 12µLof purified cDNA into clean PCR tubes. 12. Quantify cDNA yield with Quibit (2µL sample) and cDNA integrity with bioanalyzer (1µL sample). If cells pass QC, proceed to library prep via Nextera XT. Nextera XT cDNA library preparation (1/5 reactions): Nextera XT uses a transpose that is very sensitive to DNA contamination. BE SURE TO CLEAN ALL EQUIPMENT PRIOR TO STARTING THIS PROTOCOL WITH DNA-AWAY. 1. Dilute cDNA to 250 pg (0.250 ng) total into PCR strips with pure water. Make a minimum of 5µL of diluted cDNA. 2. Prepare transposase master mix and 3µL of mix to each tube: a. 2µL transposase buffer b. 1µL transposase 3. Set thermocycler to the following: a. 56◦C hold b. 56◦C for 10 minutes c. 10◦C hold 4. Add 1µL of diluted cDNA to the transposase mix. Vortex briefly, spin down, and place in thermocycler and start 56◦C incubation. 5. Prepare stop NT buffer and get ready to add to samples when they are finished with transposase reaction. 6. Remove tubes and add 1µL of stop buffer. Vortex and spin down. 7. Wait 5 minutes. 8. Add 3µL of NPM pcr mix. 9. Add 1µL of each i5 and i7 PCR primer. 10. Vortex and spin down. 11. Perform PCR using the nextera PCR program ● Gap fill for 1 round for 3 minutes at 72◦C ● Cycle 21 rounds of: ○ Denature at 98◦C for 20 seconds – Anneal at 67◦C for 15 seconds ○ Extend at 72◦C for 6 minutes ● Final extension for 5 minutes at 72◦C ● Hold at 4◦C GeTMoR DNA WGA: WGA is meant to take place on the same day as RNA extractions; the gDNA cannot be stored or else WGA will fail. The RNA lysis step with proteinase K also lyses the nucleus and releases gDNA – this step takes over the DTT-KOH lysis for SIGMA WGA4 protocol. You will begin by adding in the fragmentation mix (10x mix) from the WGA4 protocol but adjust for volume discrepancies. It is assumed that you have roughly 5µL of gDNA in each tube. 1. Prepare 2x fragmentation mix: 1µL 10x fragmentation + 4µLTE buffer. 2. Add 5µL of 2x fragmentation mix to each sample. Mix by vortexing and spin down. 3. Incubate at 99◦C for 4 minutes. 4. Remove samples and place in new PCR tube cooler from genomics -20◦C. 56 5. Make library preparation master mix a. 2µLof 1x single cell library preparation buffer (green cap) b. 1µL of library stabilization solution (yellow cap) 6. Add 3µL of library prep master mix to each sample. Mix and place in thermocycler at 95◦C for 2 minutes. Cool samples on PCR tube cooler. 7. Add 1µL of library preparation enzyme (orange cap), mix by flicking tube (enzyme is fragile), and spin down. 8. Place sample in thermocycler and incubate as follows: a. 16◦C for 20 minutes b. 24◦C for 20 minutes c. 37◦C for 20 minutes d. 75◦C for 5 minutes e. Hold reactions at 4◦C 9. Make amplification master mix a. 7.5µL of 10x amplification master mix (red cap) b. 48.5µL of molecular grade water c. 48.5µL of WGA DNA polymerase (white cap) 10. Add 61µL of master mix to each reaction. Mix by vortexing and spin down. 11. PCR reaction of: a. Initial denaturation at 95◦C for 3 minutes b. 23 cycles of i. 94◦C denature for 30 seconds ii. 65◦C anneal/extend for 5 minutes c. Hold at 4◦C 12. Follow WGA4 quality control metrics: (1) run gel and if gel looks good (2) purify samples with spin columns. 13. Proceed to NEB Ultra FS II library preparation kit for DNA library preparation. 57 Chapter 4: Polyploid cancer cells reveal signatures of chemotherapy relapse Michael J. Schmidt1 , Amin Naghdloo1 , Rishvanth K. Prabakar2 , Mohamed Kamal1 , Peter Kuhn1 , Kenneth J. Pienta3 , Sarah R. Amend3*, James Hicks1* 1) Convergent Science Institute in Cancer, Michelson Center for Convergent Bioscience, University of Southern California, Los Angeles, California, USA 2) Cold Spring Harbor Laboratories 3) Cancer Ecology Center, the Brady Urological Institute, Johns Hopkins University School of Medicine, Baltimore, Maryland *Corresponding Authors: Sarah Amend and James Hicks Full Author Contributions: Conceptualization, M.J.S., and J.H.; methodology, R.K.P, A.N. M.J.S., and J.H.; validation, M.J.S. .; formal analysis, M.J.S. ; image analysis and code: M.J.S., A.N. ; investigation, M.J.S., and J.H.; resources, J.H., K.P., S.A. P.K.; writing—original draft preparation, M.J.S.; writing—review and editing, M.J.S., K.P., S.A., and J.H.; visualization, M.J.S.; supervision, J.H.; funding acquisition, P.K., K.P., S.A., and J.H. 58 4.1 Abstract Background: Therapeutic resistance remains a significant obstacle in the effective management of cancer, contributing substantially to cancer-related mortality. Despite initial treatment responses leading to remission, patients often face the challenge of disease recurrence. The mechanisms underlying this transition remain poorly understood, partly due to limitations in current detection technologies that fail to identify residual malignant cells. Recent investigations have highlighted the presence of therapy resistant large polyploid cancer cells in patient tissue samples, linking their occurrence with advanced disease stages and relapse. Methods: Late-stage prostate cancer patients bone marrow aspirates were analyzed for circulating tumor cells for increased genomic content (CTC-IGC) via liquid biopsy (n=44). Single cell copy number profiling was performed on CTC-IGC and typical CTCs to understand tumor clonality. In vitro induction of polyploid cancer cells was accomplished by treating PC3 and MDA-MB-231 cell lines with sublethal doses of docetaxel or cisplatin. Single cell copy number profiling, single cell RNA-sequencing, and immunofluorescence protein staining was conducted to understand how polyploid cancer cells survive chemotherapy. Patient bone marrow samples were then stained via immunofluorescence to evaluate the presence of markers associated with disease relapse. Results: We discover the presence of CTC-IGC is significantly associated with worse progression free survival. Surprisingly, CTC-IGC and typical CTCs display no distinct copy number breakpoints and are clonal with the tumor. Through cell line models, we identify novel RNA and protein markers associated with survival—HOMER1, TNFRSF9, and LRP1 – and notably, find that surviving progeny from polyploid PC3 cells retain these markers. We then discover a subset of patient CTCs found in the bone marrow are positive for these survival markers. Further, in public gene expression data we uncover a significant association between recurrence and the elevated expression of HOMER1, TNFRSF9, and LRP1. 59 Conclusion: Our findings underscore the active functionality of large polyploid tumor cells, their resilience to chemotherapy, and the expression of novel markers closely linked with cancer relapse. This body of work provides insights into the complex interplay between large polyploid tumor cells and cancer recurrence, offering promising avenues for the development of targeted therapeutic interventions to overcome treatment resistance. Keywords: polyploid cancer cell; circulating tumor cell; relapse; progression-free survival; liquid biopsy; poly-aneuploid cancer cell; chemotherapy resistance; single cell; multi-omics 60 4.2 Introduction Therapy resistance is responsible for over 90% of cancer related deaths [1]. Researchers and clinicians still lack the tools to combat resistance in practice despite outlining numerous detailed intrinsic and extrinsic mechanisms that enable malignant cell survival, resulting in relapse, disease progression, and a worse outcome for the patient [1, 97-98]. Breast and prostate cancer are two of the most diagnosed cancers in women and men, respectively [99]. While initial treatment efficacy is observed in most patients, prostate cancers recur in 24- 48% of cases depending on disease risk [100], and breast cancers relapse in about 30% of the overall patient population [101-102]. In general, late stage metastatic cancers are more difficult to treat and are typically administered chemotherapy; unfortunately, complete response rates from chemotherapy treatments in late stage patients are low and warrant improvements [99, 103]. Large polyploid tumor cells are correlated with late disease stages, poor prognosis, and therapy resistance across virtually every cancer lineage [39, 44, 48, 104-105]. Large polyploid tumor cells are induced through various mechanisms, including common chemotherapies such as Docetaxel and Cisplatin [106-109]. Supporting evidence has shown that whole genome doubling (WGD) events and altered ploidy levels are poor prognostic indicators across cancer types, and are ultimately thought to allow cancer cells the ability to evolve genomically and survive therapy [110-113]. Recent studies have shown that large polyploid tumor cells are capable of giving rise to viable progeny that display more malignant and stem cell characteristics than the parental population they descended from [114]. Importantly, targeting identified pathways, such as AP-1, HIF2a, cholesterol-related, and embryogenic-related pathways reduced the number of surviving large polyploid cancer cells, as well as surviving progeny cells following therapy [49, 114-117]. While significant, these studies lack single-cell molecular resolution and note that not all cells are eliminated. What ultimately matters is that some cancer cells are still capable of survival and 61 disease progresses. Identification of novel biomarkers that can predict patients’ recurrence and resistance to therapy may lead to better treatment outcomes. Utilizing the High Definition Single Cell Assay [85], we find that the presence of circulating tumor cells (CTCs) with increased genomic content in the bone marrow aspirate of late-stage prostate cancer patients are significantly associated with worse progression free survival. This motivated a comprehensive analysis of model systems for large polyploid tumor cells in two cell lines: PC3 (prostate lineage) and MDA-MB-231 (breast lineage). We first treat cells with cisplatin or docetaxel, then functionally characterize the surviving cells through a multi-omic approach, including morphometric, genomic, and transcriptomic profiling at the single cell level. We find that progeny cells, isolated from a single large polyploid PC3 cell treated with docetaxel, differed substantially from the parental population, and most closely resembled the transcriptome of large polyploid tumor cells. We also find novel markers associated with chemotherapy survival – HOMER1, TNFRSF9, and LRP1 – that are upregulated in cells that survive treatment, are retained in the progeny from surviving cells, and are found to be upregulated significantly associated with recurrence in prostate and breast cancer at the RNA level. Further, we find the novel survival biomarkers to be expressed at the protein level in the CTCs of patients who also have recurrent disease. Taken together, our results highlight novel biomarkers of survival and shed light on the functionality of large polyploid tumor cells and their role in disease recurrence. 4.3 Methods Patient samples collection and processing: Liquid biopsy samples were collected and processed as previously described for patients in Figure 1, Table 1 (patients 2 and 4-6), and Figure 6 [31]. Briefly, peripheral blood (PB) and bone marrow aspirate (BM) samples were collected from previously treated participants immediately starting treatment on trial NCT01505868 which evaluated cabazitaxel with or without carboplatin in patients with 62 metastatic castration-resistant prostate cancer. Samples were collected at MDAnderson at baseline prior to clinical trial administration. Two exceptions are patients 1 and 3 (Table 1 and Figure 6) who did not participate in the trial. Patient 1 was acquired from the greater Los Angeles Veterans' Affairs Healthcare System and patient 3 was acquired from MDAnderson. All patients gave written informed consent. Samples were sent to the Convergent Science Institute in Cancer (CSI-Cancer, University of Southern California), for processing as previously described [31]. Following isotonic erythrocyte lysis, the entire nucleated fraction of the peripheral and bone marrow samples were plated onto custom cell adhesion glass slides (Marienfield, Lauda, Germany) and then stored at -80°C until use. Cell culture and drug treatment: PC3 and MDA-MB-231 cell lines were purchased from ATCC and grown in RPMI and DMEM, respectively, with 10% FBS and with 0.5% penicillin / streptavidin. Cells were plated at a density of 625,000 cells in a T-75 flask. To induce the large cell phenotype, PC3 cells were treated with 5 nM docetaxel or 10 µM cisplatin for 72 hours, while MDA-MB-231 cells were treated with 10 nM docetaxel and 10 µM cisplatin for 72 hours. Following treatment, cells were allowed to recover in normal medium for 1 day or 10 days. When indicated in the text, PC3 cells were re-treated at the day 10 recovery time point. At specified time points, cells were lifted from culture and plated on specialized Marienfeld glass slides (Marienfeld, Lauda, Germany) for imaging or single cell isolation. All cell line experiments were conducted in triplicates. Immunofluorescent staining: Baseline High Definition Single Cell Assay Staining: Patient slides depicted in Figure 1 were fixed with paraformaldehyde and stained with a pancytokeratin cocktail mixture imaged in TRITC, multiplexed CD45/CD31 in Cy5 channel, Vimentin imaged in FITC, and DAPI imaged with UV as previously described [31]. Resistance biomarkers: Slides were removed from the -80°C and immediately fixed with 2% paraformaldehyde for 20 minutes. Slides were then blocked with 2% BSA and then 63 incubated overnight at 4°C with a primary antibody cocktail consisting of mouse IgG1/Ig2a antihuman cytokeratins (CK) 1, 4, 5, 6, 8, 10, 13, 18, and 19 (clones: C-11, PCK-26, CY-90, KS1A3, M20, A53-B/A2, C2562, Sigma, St. Louis, MO, USA), mouse IgG1 anti-human CK 19 (clone: RCK108, GA61561-2, Dako, Carpinteria, CA, USA), and mouse EpCAM. The CK and EpCAM antibodies make up the EPI channel in patient stains and cell lines. TNFRSF9 (Thermo, PA5-98296) and HOMER1 (PA5-21487) primary antibodies were incubated on slides overnight at 4°C with the EPI-cocktail of antibodies. Slides were then washed and incubated at room temperature for two hours with Alexa Fluor® 555 goat anti-mouse IgG1 antibody (A21127, Invitrogen, Carlsbad, CA, USA), conjugated mouse anti-human CD45 Alexa Fluor® 647 (clone: F10-89-4, MCA87A647, AbD Serotec, Raleigh, NC, USA), and counter-stained with 4′,6- diamidino-2-phenylindole (DAPI; D1306, Thermo Fisher Scientific, Waltham, MA, USA). LRP1 (Thermo: 377600) was generated in mice and was therefore not compatible with the pan-cytokeratin and EpCAM cocktail. Instead, LRP1 was incubated overnight at 4°C. Slides were then washed and incubated at room temperature for two hours with Alexa Fluor® 555 goat anti-mouse IgG1 antibody. Next, pre-conjugated Alexa Fluor 488 pan-cytokeratin (53-9003-82, Thermo) recognizing CK 10, 14, 15, 16, and 19 was incubated with conjugated mouse antihuman CD45 Alexa Fluor® 647 and DAPI. Slide imaging and analysis: All cell line imaging experiments were conducted in triplicates. Slides were imaged with an automated high throughput microscope equipped with a 10x optical lens, as previously described [85]. Immunofluorescent and bright field images were collected. Image analysis tool, available at https://github.com/aminnaghdloo/if_utils, was developed in python using the OpenCV and scikit-image packages. Briefly, each fluorescent channel was segmented individually using adaptive thresholding and merged into one cell mask. Cell mask and DAPI mask were used to extract features and fluorescent intensity statistics of single cells and their nuclei, respectively. For nucleus size analysis, equivalent diameter was calculated from nucleus area, assuming a circular shape. 64 Progeny collection from single cells: PC3 cells were treated with docetaxel, recovered for 10 days in normal medium, and then lifted with 1x verscene. Cells were then spun down and resuspended in 45 mL of RPMI medium. Biosorter (UnionBio) was used to sort single cells based on size and the largest 15% of cells were sorted into ten 96-well plates (n=960 individual wells) containing RPMI medium and then placed in a 37°C incubator. Media was changed every 2-3 days. Fluorescence in situ Hybridization (FISH): Probes for centromeres of chromosomes 1 (CHR01-10-GR) and 10 (CHR10-10-GR) were purchased from Empire Genomics (Depew, New York) and the hybridization was carried out on Marienfeld glass slides per the manufacturer’s instructions. Slides were then stained with DAPI and then imaged using fluorescence microscopy. Single cell picking: Eppendorf TransferMan NK2 micromanipulator was used to collect the cell of interest in a 20 µM micropipette (for control cells) or a 100 µM micropipette (for treated surviving cells). The cell was transferred to a PCR tube containing 0.2% TritonX-100 and RNAse Inhibitor. Single cells were stored in -80°C for downstream DNA or RNA sequencing. Single cell copy number profiling: Copy number profiling from low pass whole genome sequencing samples was conducted as previously described [28, 72]. Briefly, cells were sequenced at a depth of 1-2 million reads on an Illumina HiSeq 4000. Sequencing reads were aligned with BWA-MEM to the hg38 reference. Count data was segmented via the R package DNACopy (version 1.70.0), and median segmented ratio values were reported. Single cell RNA-sequencing: Single cells were isolated and picked via micromanipulation as previously described. RNA was extracted via a modified Smart-Seq2 approach and library prepped with Nextera XT (Illumina, San Diego, CA). Cells were sequenced paired end by 150 base-pairs on an Illumina HiSeq 4000 (Fulgent). Read adapters were trimmed with TrimGalore (version 0.6.7) and aligned with the HiSat2 (version 2.2.1). Picard (version 3.0.0) 65 was used to visualize RNA mapping quality control. HTSeq (version 2.0.2) was used to generate a gene count matrix. The SingleCellExperiment package (version 4.2.2) was utilized for inputting count data into downstream analyses, such as converting to Seurat (version 4.3.0 and edgeR (version 3.36.0) count matrices [118]. Downstream analysis was performed with R (version 4.1.2). Data visualization was performed with Seurat and ggplot2 (version 3.4.4), and Pheatmap (version 1.0.12) packages. Cell cycling scoring was conducted with reCAT, where markers for each phase were used to categorize each cell in a specific cell cycle phase [119]. Standard batch correction and normalization was performed according to the Seurat batch integration tutorial (version 4). The edgeRQLFDetRate differential expression pipeline was used and sequencing batches were controlled for [120]. Shared genes expressed in surviving large cells were intersected through R. Gene set enrichment data sets were downloaded directly from CHEA3 [121] and MSigDB [122]. Single cell enrichment was conducted through JASMINE, where the code was downloaded directly through the authors github repository [123]. Survival analysis: Survival analysis from patient BM and blood samples was performed with the Survival R package (version 3.5.5) and plotted with ggplot2 (version 3.4.4). Public gene expression survival analysis was analyzed via PanCancSurvPlot [124]. The prostate cancer data set was acquired from GSE116918 [125] and the breast cancer data set was acquired from GSE10893 [126]. Viability: VivaFix cell viability dye (Bio-Rad catalog 1351115) was utilized to evaluate cell permeability. Cells were treated with chemotherapy and allowed to recover. Following recovery cells were lifted with 1x versene (ThermoFisher catalog 15040066), spun down, and resuspended in 1x PBS with VivaFix dye. Cells were then plated on Marienfeld glass slides, incubated at 37°C for 30 minutes, briefly washed in PBS, and then fixed with 2% 66 paraformaldehyde for 20 minutes. Cells were then stained with DAPI, EPI cocktail, and CD45, and imaged via high content scanning. 4.4 Results CTCs with increased genomic content (CTC-IGC) are found in the bone marrow of late-stage prostate cancer patients and are correlated with worse progression free survival Liquid biopsies from the peripheral blood and bone marrow aspirate were acquired from a late-stage prostate cancer cohort that were administered cabazitaxel with or without carboplatin (NCT01505868). Matched bone marrow and peripheral blood samples from 31 patients were analyzed for CTCs and other rare cells. CTCs with increased genomic content (CTC-IGC), identified as having a nuclear diameter at least double the average of the CTC cell population, were found in 9.7% of peripheral blood samples, and CTC-IGC were present in 80.6% of bone marrow samples from the same patients (Fig. 4.1A-B, 4.S1). Survival analysis with 44 bone marrow samples (from the 31 patients with matched samples and 13 patients without matched blood) showed that presence of at least one CTC-IGC detected in the bone marrow was associated with decreased progression free survival (PFS) (Fig. 4.1C). Previous treatment history was available for 33 of the 44 patients and primarily included anti-androgens and other hormonal treatments (i.e., Bicalutamide, Nilutamide, Enzalutamide). Six patients were previously treated with docetaxel, and all six patients were positive for CTCs-IGC in the bone marrow. 67 Figure 4.1: Large tumor cells are found in BM of late-stage prostate cancer patients. (A) Enumeration of patients with matched blood and bone marrow samples with at least 1 CTC-IGC present in liquid biopsy. (B) Representative images of CTC and CTC-IGC found in BM aspirate. Scale bars set to 15 µM. (C) PFS from patients with or without CTC-IGC found in BM samples. (D) Representative image of typical CTC found in BM with merged and DAPI channels (top) and its genomic copy number profile (bottom). (E) Representative image of CTC-IGC found in bone marrow with merged and DAPI images (top) and its genomic copy number profile (bottom). (F) Representative image of mono-nucleated CTC-IGC found in bone marrow with merged and DAPI images (top) and its genomic copy number profile (bottom). Clonal tumor lineage measured via copy number alteration analysis was confirmed in both typical-sized CTCs and CTCs-IGC. No apparent CNAs were identified between the two CTC groups. Further, we found mono- and multi-nucleated CTC-IGC displayed no differences in copy number profiles (Fig. 4.1D-F, 4.S2, 4.S3). These observations show that CTC-IGC can be found in blood and bone marrow aspirate, are tumor derived, and thus may contribute towards relapse in late-stage prostate cancer. Despite the WGD of CTC-IGC, these cells retain the original tumor copy number profile. To understand the importance and behavior of this phenotype, we set out to model the induction of large polyploid tumor cells in vitro to investigate their contribution towards therapeutic resistance. 68 Large polyploid cancer cells form as a response to chemotherapy in prostate and breast cancer cell lineage models PC3 and MDA-MB-231 cells were treated with sublethal doses of docetaxel or cisplatin for 72 hours. Following chemotherapy, cells were allowed to recover for 1 or 10 days in their regular growth medium, lifted from culture, plated on specialized glass slides, stained with cell and nuclear markers, then imaged through high content scanning and evaluated for nuclear size and other morphometric comparisons (Fig. 4.2A). While there was a significant cell death as expected (Fig. 4.S4A), surviving cells increased in both nuclear diameter and cell size as a function of time (Fig. 4.2B-E; Fig. 4.S4B). Figure 4.2: Large polyploid tumor cells are induced following chemotherapy exposure in MDA-MB-231 and PC3 cell lineages. (A) Experimental schematic for in vitro investigation of surviving polyploid cells. (B) Representative bright field (left) and DAPI (right) 40x images of MDA-MB-231 cells and PC3 cells treated with docetaxel and cisplatin (right) 10 days post treatment recovery. (C) Nuclear diameter for all treated conditions for MDA-MB-231 cells. (D) Nuclear diameter for all treated conditions for MDA-MB-231 cell. Arrow indicates the condition where single cell progeny originated from. (E) Nuclear diameter of control parental and progeny-1 cells, and progeny-1 10 days post-treatment cells. 69 To evaluate resistance, we treated cells that initially survived cisplatin treatment (10 days post treatment; 10 DPT) with cisplatin or docetaxel. Compared to the control condition (initially cisplatin treated and then re-treated with DMSO) cell counts and cell viability were not significantly impacted, suggesting that these cells are impervious to additional rounds of chemotherapy (Fig. 4.S4A, 4.S4D, 4.2E). To obtain progeny cells from a single chemotherapy-induced surviving polyploid cell, we isolated and single-cell seeded 10 days post-cisplatin release (n=480) and 10 days postdocetaxel release (n=960), and monitored for colony formation. From these, only 2 polyploid docetaxel-treated PC3 cells gave rise to progeny after 2 months (progeny-1) and 2.5 months (progeny-2). Progeny-2 failed to proliferate following the first passage. Over the course of the three-month experiment, approximately 50% of the polyploid cells treated with either cisplatin or docetaxel remained viable and adherent. Interestingly, the dividing progeny-1 cells displayed a larger nuclear and cellular diameter than the parental PC3 population from which it originated (Fig. 4.2F, 4.S4). We treated progeny-1 with docetaxel or cisplatin and found that the population was sensitive to both chemotherapies. Further, following 10 days of recovery, surviving progeny-1 cells had increased nuclear and cell diameter, similar to what was observed from the original parent population (Fig. 4.2F, 4.S4B). Surviving PC3 progeny and large polyploid cancer cells show no ploidy or copy number ratio alterations compared to parental control cells To understand if any genomic alterations took place in the surviving polyploid cells that were retained in the progeny following their transition from a typical cancer cell to large polyploid cell, and back to a typical cancer cell, we assayed copy number status and cell ploidy. Strikingly, surviving large polyploid PC3 and MDA-MB-231 cells from both docetaxel and cisplatin treatments showed no apparent copy number ratio differences compared to parental DMSO control cells (Fig. 4.3A-C, 4.S5). This result confirms patient data in that copy number status does not differ between CTCs with normal nuclei and CTCs with larger nuclei (Fig. 4.1D- 70 F), suggesting cells are undergoing complete WGD rather than displaying specific copy number breakpoints. While copy number status did display minor differences in the progeny-1 compared to parental DMSO control – for example, progeny-1 displayed an increased 3p gain ratio signature (Fig. 4.3A-B) – no substantial alterations were observed, and the progeny-1 clone can still confidently be classified as a PC3 clone. Conversely, progeny-2, the clone that did not survive the first passage, displayed the most aberrant copy number profile compared to the other conditions (i.e., 6 gain and 4p gain) and clustered separately from the other PC3 cell conditions (Fig. 4.3A). FISH probes for the centromeres of PC3 chromosome 1 (ploidy = 3) and chromosome 10 (ploidy = 1) showed no statistically significant differences when comparing DMSO parental control cells to progeny-1 cells (Fig. 4.3D), suggesting that any apparent scars of ploidy reduction were not present. These results prompted investigation into the cellular behavior of large polyploid cancer cells and progeny through single cell RNA-seq. 71 Figure 4.3: Genomics of PC3 DMSO control, docetaxel treated cells, and docetaxel large cell progeny-1. (A) Segmented copy number ratios for PC3 conditions. Ratio of 1 (white) indicates copy number neutral respective to the entire genome. (B) Representative PC3 DMSO, Doc D10, and Progeny-1 copy number ratio profiles. (C) Representative MDA-MB-231 DMSO control, Doc D10, and Cis D10 copy number ratio profiles. (D) FISH for PC3 DMSO control and progeny-1 cells for centromere of chromosome 1 (ploidy = 3) and chromosome 10 (ploidy = 1). Single cell transcriptomic profiling reveals common genes and pathways upregulated in PC3 and MDA-MB-231 surviving cells To understand the differences in cellular behavior, we isolated individual surviving cells 1- or 10-days post-cisplatin or docetaxel release via micromanipulation and performed singlecell RNA-sequencing. 497 PC3 cells were isolated and sequenced in 5 separate batches (Fig. 4.S6) and included: DMSO control (n=129), 1-day post-cisplatin release (n=78), 10 days post-cisplatin release (n=68), 1-day post-docetaxel release (n=45), 10 days post-docetaxel release (n=118), docetaxel progeny-1 (n=12), docetaxel progeny-2 (n=13). Two batches of 203 total MDA-MB- 72 231 cells included the following conditions: DMSO control (n=43), 1-day post-cisplatin release (n=22), 10 days post-cisplatin release (n=62), 1-day post-docetaxel release (n=24), 10 days post-docetaxel release (n=62) (Fig. 4.S7). Regardless of treatment, a general spatial separation that was dependent on recovery duration was observed in PC3 and MDA-MB-231 cells (Fig. 4.4A, 4.S8). Cell cycle marker analysis showed a decrease of cells in M and G2/M phases, and a subsequent increase in cells in either G1, G1S, and G2 for both cell lines (Fig. 4.S8-9). 73 Figure 4.4: Chemotherapy induced surviving tumor cells share common phenotypes and pathways for survival. (A) UMAP of all conditions for PC3 cells. (B) Comparison of DEGs between MDA-MB-231 large-D10 and PC3 large-D10 cells. Between cisplatin and docetaxel treatments, MDA-MB-231 and PC3 share 309 upregulated genes compared to their respective controls. (B) Genecode annotations for 309 shared genes. (D-E) LFC of shared 309 genes for PC3 vs MDA-MB-231 for docetaxel and cisplatin treatments, respectively. (F) CHEA3 transcription factor enrichment of the shared 309 genes between MDA-MB-231 and PC3 cells. (G) Hallmark enrichment analysis of 309 shared genes. (H) Single cell Hallmark gene set enrichment analysis for all PC3 cells. Treating cells with chemotherapy agents will activate numerous pathways that will in turn cause a dramatic shift in gene expression. To identify convergent phenotypes regardless of tumor type or therapy, we evaluated genes that were upregulated in both PC3 and MDA-MB- 74 231 following either cisplatin or docetaxel treatment. MDA-MB-231 cells 10 days post cisplatin or docetaxel release upregulated 1591 shared genes compared to DMSO control; PC3 cells 10 days post cisplatin or docetaxel treatment upregulated 1178 shared genes compared to DMSO control (LFC > 1.5, FDR < 0.01; Fig. 4.4B). Intersection of the two shared gene sets showed MDA-MB-231 and PC3 cells that survive either cisplatin or docetaxel exposure shared 309 upregulated genes (Fig. 4.4B). The 309 shared genes were considered a survivor cell enrichment data set, which was utilized to understand biology and potential biomarkers for surviving cells. Annotations of the 309 shared genes identified the majority as functional gene products: 77% protein coding and 17% lncRNAs, while the remaining ~6% were pseudogenes or yet to be experimentally confirmed (TEC, not yet tested; Fig. 4.4C). Log-fold change (LFC) values were plotted for PC3-Doc-D10 vs MDA-Doc-D10 (Fig. 4.4D) and PC3-Cis-D10 vs MDA-Cis-D10 (Fig. 4.4E). Within each treatment class, shared DEGs were positively correlated between MDA-MB231 and PC3 cells, indicating the DEGs are upregulated to a similar magnitude. Common transcription factors (TFs) and hallmark pathways upregulated in the survivors were delineated (Fig. 4.4F-G). Interestingly, two top significantly enriched TFs, ZNF697 and NPAS2, were previously reported in cells that transition out of senescence and into a proliferative state [127]. Top enriched hallmark pathways in the 309 gene survivor data set were: epithelial-to-mesenchymal transition (EMT), upregulation of KRAS signaling, coagulation, TNFɑ signaling via NFκB, and hypoxia (Fig. 4.4F). Single cell gene enrichment confirmed the top upregulated hallmark pathways in the shared survivor data set, and as expected displayed heterogeneity of upregulated pathways (Fig. 4.4H, 4.S8-9). Additional pathways identified to be significantly upregulated in the surviving cells at the single cell level were: PI3K AKT mTOR Signaling, Inflammatory Response, and Cholesterol Homeostasis (Fig. 4.4H, 4.S8-9). Identification of HOMER1, TNFRSF9, and LRP1 as putative large cell and chemotherapy RNA survival markers 75 Utilizing data from the shared PC3 and MDA-MB-231 cell survivor gene set, markers were independently evaluated to understand their putative role in chemotherapy survival. All 309 genes were investigated via literature review and queried for terms in September 2023, including: large tumor cell, polyploid giant cancer cell, poly-aneuploid cancer cell, survival pathways, drug resistance, and apoptosis. With prior knowledge that top upregulated genes (MMP-3, SAA1, and C3) functioned in the execution of apoptosis and clearance of apoptotic bodies (Fig. 4.4D-E), and that SAA1 and C3 were correlated with better PFS (Fig. 4.S10), they were not considered novel survival markers. The 309 gene survivor cell enrichment data set was also intersected with genes in the top enriched pathways that modulate survival: TNFɑ via NFκB, PI3K-AKT, and mTOR signaling (Fig. 4.4G-H). Further, we evaluated genes that were upregulated to similar levels in both cell lines (i.e., similar logFC values). We identified TNFRSF9 and LRP1 as genes that were upregulated to similar levels in both cell lines (i.e., similar logFC values) as survival biomarkers (Fig. 4.4D-E, 5). TNFRSF9 and LRP1 are cell surface receptors that promote PI3K activity, which then acts on AKT to promote survival (Fig. 4.S11A) [128–130]. Further, we identified HOMER1 as a PC3-specific survival marker (Fig. 4.5); HOMER1 plays a role in mTOR signaling and plays an important role in neuronal cell survival [131–134]. 76 Figure 4.5: HOMER1, TNFRSF9, and LRP1 are putative markers of chemotherapy resistance. (A) Representative PC3 images of putative marker genes stain in the VAR (4th) channel. DMSO control cells were stained with HOMER1 and were negative. (B) RNA expression for each marker for all PC3 cells. (C) Immunofluorescence quantification for PC3 cells stained with tested markers. (D) Representative MDA-MB-231 images of putative marker genes stained in the VAR (4th) channel. DMSO control cells were stained with HOMER1 and were negative. (E) RNA expression for each marker for all MDA-MB-231 cells. (F) Immunofluorescence quantification for MDA-MB-231 cells stained with tested markers. HOMER1, TNFRSF9, and LRP1 are markers of chemotherapy survival at the protein level and are retained in docetaxel treated PC3 progeny At the protein level, we found chemotherapy surviving PC3 and MDA-MB-231 cells stained positive for HOMER1, TNFRSF9, and LRP1 (Fig. 4.5A, 4.5D). Image quantification revealed all PC3 conditions (except for progeny-1 cisplatin day 10 post-treatment release) were significantly upregulated compared to DMSO controls (Fig. 4.5A, 4.5C). Day 10 survivors showed the highest protein expression levels for each marker tested. Importantly, untreated PC3 progeny-1 displayed significantly higher expression in all three survival markers tested 77 compared to parental DMSO control cells, suggesting these markers were retained following treatment and inherited by progeny (Fig. 4.5A, 4.5C). CD45 is typically utilized as a tumor cell exclusion marker that stains exclusively for white blood cells. At day 10 post-treatment release time points we noted a gain in CD45 protein expression that was also retained in progeny cells in PC3 cells (Fig. 4.5A, 4.S11). MDA-MB-231 cells also showed a significant upregulation of expression for most markers tested, except HOMER1 for MDA-MB-231 docetaxel day 10 posttreatment release and LRP1 for cisplatin day 10 post-treatment release conditions showed no differences at the protein level compared to MDA-MB-231 control cells (Fig. 4.5D, 4.5F). Taken together, these results show that HOMER1, TNFRSF9, and LRP1 in surviving PC3 cells and TNFRSF9 and LRP1 in MDA-MB-231 cells are significantly upregulated at the RNA and protein levels. HOMER1, TNFRSF9, and LRP1 are found at the protein level patient BM samples, and their increased expression is correlated with recurrence in public datasets A subset of bone marrow samples from the prostate cancer patient cohort (Fig. 4.1) were stained with the putative survival markers HOMER1, TNFRSF9, and LRP1 (Fig. 4.6A). All patients profiled had CTCs that were positive for the tested markers, suggesting that CTCs express these genes at the protein level (Fig. 4.6B). While there were CTC-IGC positive for the marker genes in each patient sample (Fig. 4.6A), the tested markers were not selective for CTC-IGC across patient groups (Fig. 4.S12). Interestingly, patient-5 displayed the highest percentage of CTCs positive for putative resistance markers HOMER1 and TNFRSF9 and had the shortest PFS at 1.4 months (Fig. 4.6B, Table 4.1). Additionally, the markers identified cells in the bone marrow that displayed IGC but were negative for epithelial markers (Fig. 4.S13). In publicly available datasets in patients treated with chemotherapy (see methods), high expression of these TNFRSF9 and LRP1 significantly correlated with a shorter progression free survival in patients with prostate cancer; HOMER1 was not statistically significant (p-value = 0.183) but displayed a trend of worse survival (Fig. 4.6C). High gene expression of TNFRSF9, 78 HOMER1, and LRP1 were all significantly correlated with worse relapse free survival in breast cancer (Fig. 4.6D). Taken together, we can conclude the survival genes are associated with recurrence at the RNA level and are present on CTCs-IGC in the bone marrow aspirate of latestage prostate cancer patients. Table 4.1: Characteristics of patient BM samples stained for putative markers. BM-large and Blood-large indicate if a large polyploid cancer cell was detected in the LBx. Patient 1 and patient 3 did not have clinical data available. PT PFS Docetaxel Hormone 1 -- -- -- 2 9.6 no yes 3 -- -- -- 4 6.1 no yes 5 1.4 no yes 6 4.5 yes yes 79 Figure 4.6: HOMER1, TNFRSF9, and LRP1 are positive on CTCs in the BM aspirate of late stage prostate cancer, and are correlated with recurrence in prostate and breast cancers. (A) Representative CTCs from BM of advanced prostate cancer patients that were stained with survival markers HOMER1 (left), TNFRSF9 (TNF; middle), and LRP1 (right). Tested markers appear as white in the merged image. Scale bars are 15 µM. (B) Percentages of CTCs with EPI positivity and cells that were stained with survival markers HOMER1 (left), TNFRSF9 (middle), and LRP1 (right). Cells that are positive for the marker alone (middle bar) cannot be conclusively labeled a tumor derived cell. LRP1 (right) is also a marker of T-cells, so only cells that were EPI positive were included. (C-D) Kaplan-Meyer survival plots for RNA expression of tested markers in prostate (C) and (D) breast cancer patients. 4.5 Discussion Our analysis of bone marrow liquid biopsy samples from previously treated advanced prostate cancer patients reveals that the presence of polyploid cancer cells correlates with poorer progression-free survival. Although clinical reports have frequently observed polyploid 80 cancer cells in later disease stages, a direct link with disease recurrence has not been firmly established. We also found that CTC-IGC have copy number profiles identical to typical CTCs and are predominantly present in the bone marrow rather than in peripheral blood. This aligns with previous research indicating a higher general abundance of CTCs in the bone marrow [31]. Since CTC-IGC include a rare subtype of the CTC population, the likelihood of identifying them in the peripheral blood appears to be drastically lower than the bone marrow. Through single-cell copy number profiling and the isolation of progeny from individual polyploid cells, we demonstrate that the polyploid cancer cell phenomenon represents a change in cell state. Single-cell copy number profiling shows that the copy number ratios in patient CTC-IGC as well as chemotherapy induced polyploid MDA-MB-231 and PC3 cells that survive treatment are identical to those in their paired non-polyploid samples. This indicates that these cells, either identified as patient CTCs or those that survive in the days following therapy release, undergo multiple rounds of whole genome doubling without any additional copy number alterations. These findings provide crucial insights into the dynamics and genetic stability of the polyploid cancer cell state. Obtaining proliferative progeny proved challenging; after three months of culturing single isolated polyploid cells, we successfully derived only one proliferative progeny clone (1/1,440). This outcome is significant for two main reasons: first, it demonstrates that polyploid cancer cells can give rise to progeny, but second, the extremely low success rate underscores why these cells have historically been understudied. To enhance our understanding, future research should employ high-throughput techniques to isolate larger numbers of single cells, such as tens of thousands, which may prove critical in understanding the roles of non-proliferative polyploid cancer cells and assessing their capabilities at reinitiating cell division. Additionally, slight variations in the copy number profiles, such as a 3p gain observed in the progeny-1 clone, hint at genomic evolution. Further studies should explore this genomic evolution in different 81 progeny clones once they are sufficiently collected to understand the dynamics of genomic reorganization in these cells. Through in vitro single cell transcriptomics, we further provide evidence that polyploid cancer cells display a convergent phenotype between MDA-MB-231 (breast cancer) and PC3 (prostate cancer) model systems. Despite being induced with chemotherapies with different mechanisms of action (cisplatin and docetaxel), the different tumor types displayed a shared polyploid signature of upregulating 309 common genes. This convergence reveals significant insights into the biological features of polyploid cancer cells. In our observations, approximately 50% of polyploid cancer cells remained attached to the culture flask in a non-proliferative state during single cell progeny outgrowth experiments. Polyploid cancer cells have been identified to progress through the cell cycle but do not proliferate (i.e., endocycling or cytokinesis failure occur before mitosis) [48,135]. This is hypothesized to be a protective state of the cells that affords protection from therapeutic stressors. This phenomenon aligns with our identification of ZNF697 and NPAS2 as two transcription factors significantly enriched in the convergent polyploid gene set that were previously identified to be upregulated in cells that were in a non-proliferative state and began re-initiating cell division [127]. This suggests that some polyploid cells profiled on day 10 post therapy release may be attempting to re-initiate proliferation since the chemotherapy has been removed. This finding is further supported by a higher percentage of cells at 10-DPT expressing more markers at the M-phase of the cell cycle (Fig. S8-9). Future research should explore the roles of ZNF697 and NPAS2 in polyploid cancer cells and their implications for disease recurrence in progeny cells. Our investigation of polyploid cancer cells confirms the significant upregulation of hypoxia and cholesterol homeostasis pathways. Studies have shown that targeting these pathways in cell line models, including PC3 and MDA-MB-231, reduces the viability of progeny from polyploid cancer cells [115, 117]. Further evidence comes from a study indicating that 82 polyploid cancer cells accumulate lipid droplets in response to chemotherapy [136], underscoring the critical role of lipid balance as cells significantly increase in size. These findings suggest that these pathways are integral to the polyploid cancer cell state and represent promising targets for therapeutic intervention. The convergent surviving cell gene set we identified indicated that pro-survival and antiapoptotic pathways, such as TNFɑ via NFκB, PI3K-AKT, and mTOR signaling, are upregulated in polyploid cancer cells [137-139]. Among the genes identified in these pathways, TNFRSF9, HOMER1, and LRP1 were identified as putative survival genes and were found to be upregulated at the RNA and protein levels [128–134, 140-144]. Notably, these protein markers were retained in the PC3 progeny-1 clone, suggesting their upregulation in cells that survive chemotherapy. Public data sets also support the significance of these genes, showing their upregulation at the RNA level in patient tumors is associated with recurrence. Additionally, a subset of CTCs, including both polyploid and typical CTCs, tested positive for TNFRSF9, HOMER1, and LRP1 at the protein level. Of note, Patient 5, who experienced the worst progression-free survival at 1.4 months, had the highest percentage of CTCs positive for the TNFRSF9 marker, indicating that this gene may play a significant role in cancer cell survival. Further, these markers identified a subset of cells with IGC that were negative in the epithelial channel. These cells may be CTCs that lost epithelial expression (i.e., EMT) and, in combination with the upregulation of the proposed survival markers, could be adept at surviving in the bone marrow. Further studies are needed to evaluate the roles of TNFRSF9, HOMER1, and LRP1 in chemotherapy resistance and as a biomarker to evaluate the emergence of therapeutic resistance. The in vitro environment of cell culture does not always recapitulate the in vivo nature of cancer cell biology. This makes it difficult to speculate how polyploid cancer cells interact with their neighboring malignant cells and the surrounding stroma. Translating the findings of TNFRSF9, HOMER1, and LRP1 as resistance markers in an in vivo model is a critical next step. 83 Future studies should employ mouse models or patient derived xenografts and stain for these biomarkers to understand their prominence in vivo. Further studies should also isolate polyploid cancer cells through nuclear density to further understand their cellular phenotypes in tumor tissue. While patient results are promising, they also have limitations. This study focuses on late-stage patients with disseminated CTCs in the bone marrow and blood. The evaluated cohort comprised advanced-stage patients whose previous treatment regimens had failed. To minimize biases associated with late-stage disease and to better understand initial treatment responses and their role in inducing polyploid cancer cells, future cohorts should include patients undergoing their first rounds of therapy. One concern is that CTCs in peripheral blood are typically found in later disease stages, potentially biasing our patient population towards later stages. Obtaining samples from tissue, blood, and bone marrow could address these concerns and provide valuable insights into the role of polyploid cancer cells in dissemination, initial response to therapy, and disease evolution. 4.6 Conclusions Ultimately, many cancers become resistant to therapy and patients eventually relapse. This study is a small step in the direction of understanding the phenotype of recurrence in cancer cells. We first identify likely polyploid CTCs (CTC-IGC) in the bone marrow of prostate cancer patients. We show that these CTC-IGC are clonally altered with typical CTCs and that there are no apparent copy number discrepancies between the typical CTCs and CTC-IGC. We then model polyploid tumor cells in breast and prostate cancer cell lines and discover (1) there are no copy number changes in the polyploid cells, suggesting complete WGD occurs; (2) PC3 and MDA-MB-231 polyploid tumor cells share 309 upregulated DEGs at the RNA level that are implicated in their survival; and (3) HOMER1, TNFRSF9, and LRP1 are novel markers of chemotherapy resistance that were initially found in polyploid tumor cell lines, are upregulated in 84 public data in patients who recur faster, and are positively expressed on patient CTCs. Many future studies are needed to determine the validity of the biomarkers and the specific pathways they are acting in to promote resistance and recurrence. Determining biomarkers of relapse and resistance may ultimately provide a promising avenue and allow for actionable insights to help switch therapies and provide a better chance for the patient. 85 Chapter 5: Summary, future directions, and limitations For patient outcomes to improve, we must first understand the complexities and heterogeneity in any given cancer. By advancing single cell multi-omic and LBx techniques, this body of work pushes forward the frontier of cancer research and disease monitoring. The application of these techniques aims to identify novel biomarkers associated with treatment resistance and disease recurrence. While each independent chapter pushes forward cancer research in its unique niche, limitations exist that need to be fully understood. Advancing liquid biopsy techniques in Retinoblastoma The second chapter ventures into the realm of LBx and focuses on the unique challenges posed in RB. Traditional biopsy methods are rendered impractical due to the risk of intraocular tumor dissemination, necessitating the exploration of alternative diagnostic modalities. Leveraging the AH as a LBx biofluid, we showed the value and concordance in a one-step targeted sequencing approach to identify somatic genomic alterations at the SNV and CNV levels that are associated with RB disease drivers. This innovative methodology not only has the potential to expedite the diagnostic process but also provides a comprehensive snapshot of the genomic landscape of RB disease. Further, alternative uses of this approach may expand to other cancers where RB, MYCN, BCOR, and CREBBP are expected genetic drivers. These findings not only underscore the utility of affordable techniques via LBx in monitoring RB but may also pave the way for personalized therapeutic interventions tailored to prognostic genomic profiles of individual patients. As a proof-of-concept study, this study was limited to a small cohort of 11 patients. To establish this approach as a viable clinical tool, the next step is to expand the study to include a larger group of patients. This will further validate the utility of sampling the AH to monitor the disease. Once the study achieves sufficient statistical power, these findings could lead to 86 diagnostic LBx techniques that have the potential to preserve the eyes of young children by allowing earlier and more precise detection of disease progression. GeTMoR provides a promising tool to study CTCs The third chapter describes a novel methodology to characterize CTCs. Despite their pivotal role in the metastatic cascade and prognostic indications, the functional attributes of CTCs remain somewhat of mystery. Ultimately, better tools are needed to profile single CTCs to understand their heterogeneity and functionality. The development of the GEnomic, Transcriptomic, and MOrphological profiling of Rare cells (GeTMoR) method represents an advancement in CTC and rare cell profiling, offering a holistic approach that integrates multiomic analysis in a single cell. GeTMoR has been validated in cancer cell lines spiked into the blood and has been optimized to retain RNA transcript integrity but needs to be used in patients. Therefore, we have made all aspects of the method freely available for public use (i.e., image analysis and cell identification code). By meticulously scrutinizing the molecular signatures of individual CTCs, researchers could use GeTMoR to gain unprecedented insights into their heterogeneity, clonal evolution, and morphological attributes that may give novel insights into disease. The major limitation of the GeTMoR methodology is its lack of application in patient samples. While cell lines provide a controlled environment for initial testing and validation, the true utility of GeTMoR will only be realized when it is applied to profiling patient CTCs. The transition from cell lines to patient samples is crucial to assess the real-world relevance and efficacy of GeTMoR. An initial cohort of later stage patients with a plethora of CTCs will be required to ensure sufficient rare cells are available for RNA profiling. Following this initial cohort, GeTMoR may be used to understand different cohorts and ask questions related to mechanisms of treatment resistance, disease evolution, and metastasis. Ultimately, applying GeTMoR to patient samples could yield significant insights into the functionality of CTCs, 87 potentially transforming our understanding of cancer biology, patient disease states, and aiding in the development of more effective treatments. Polyploid cancer cells In the fourth chapter, the intricate relationship between large polyploid cancer cells and therapy resistance is explored through a multi-omic approach, unveiling novel molecular markers that may underlie disease recurrence. This work provides evidence that large polyploid cancer cells are frequented in the BM aspirate of late-stage prostate cancer patients, as opposed to blood, and that their presence is significantly associated with worse PFS. We show that large polyploid cancer cells in patients and in cell line models have no copy number alterations compared to control cells, indicating that the increase in genomic content is due to whole genome doubling events. We further identify a convergent polyploid cancer cell gene set between a prostate and breast cancer cell line that we use to delineate novel biomarkers of resistance: TNFRSF9, HOMER1, and LRP1. We find the expression of these biomarkers to be upregulated in prostate and breast cancer patients who recur faster. Further, TNFRSF9, HOMER1, and LRP1, are all present on CTCs in the BM of prostate cancer patients. These findings not only deepen our understanding of the complex interplay between large polyploid cancer cells and disease recurrence but also offer promising avenues for the development of targeted interventions aimed at circumventing treatment resistance and improving patient outcomes. While this work advances the understanding of polyploid cancer cells, many limitations and open questions remain. Future research needs to focus on several critical areas, including modulating patient cohorts to better understand the variability and prevalence of polyploid cancer cells across different cancers, stages, and sample types. There is also a need to deepen mechanistic insights into how these cells are induced, particularly the differences between mono-nucleated and multi-nucleated phenotypes, and what these differences mean for cancer 88 progression and the viability of progeny. Additionally, improving techniques to capture and study the progeny of polyploid cancer cells is essential, as even a single progeny cell can likely cause cancer recurrence. Finally, understanding the role of biomarkers TNFRSF9, HOMER1, and LRP1 in chemotherapy resistance is crucial for elucidating how these biomarkers influence the survival and proliferation of polyploid cancer cells and their progeny under therapeutic stress. Prior to this work, the presence of polyploid cancer cells in the BM was not established. This study provides concrete evidence that the presence of polyploid CTCs (i.e., CTC-IGC) in BM aspirates is significantly associated with worse PFS in patients with advanced prostate cancer. All patients in this cohort were resistant to various therapies, displaying phenotypes of castrate and chemotherapy resistance. While previous reports have suggested that polyploid cancer cells arise due to therapeutic resistance, not all castrate-resistant patients tested positive for these cells in the BM. Importantly, previous studies focused on tissue-based sampling rather than liquid biopsies. Therefore, it remains unclear what the cancer tissue looks like in these patients and they may all be positive for polyploid cancer cells. The field would significantly benefit from a prospective study that systematically collects tissue, blood, and BM samples from patients with advanced-stage prostate cancer. Such a study should evaluate the presence and characteristics of polyploid CTCs across all these biosamples. By comparing polyploid CTCs across different types of samples, researchers can gain a more comprehensive understanding of their biology and their potential role in metastatic processes. This approach may also reveal how these cells contribute to cancer progression and resistance to therapies, potentially leading to more effective strategies for diagnosis, monitoring, and treatment. In addition to applying this design to late-stage patients, a key question concerns the evolution of polyploid cancer cells and their correlation with disease stage. To better understand initial treatment responses and their role in inducing polyploid cancer cells, future cohorts should also include patients undergoing their first rounds of therapy prior to displaying resistance. This 89 will help delineate the timeline and mechanisms through which polyploid CTCs contribute to cancer dynamics, offering insights that could transform early intervention strategies and improve patient outcomes. Further, it is crucial to understand the differences in polyploid cancer cell induction and progeny production between docetaxel-induced and cisplatin-induced cells. While studies have indicated that the contrasting mechanisms of action between docetaxel and cisplatin lead to predominantly multi-nucleated or mono-nucleated phenotypes respectively, the impact of these differences on progeny production rates remains poorly understood. Some reports suggest that polyploid cancer cells may burst, releasing 5-7 daughter cells at a time, reminiscent of a multinucleated docetaxel-induced phenotype. Additionally, there is evidence that polyploid cancer cells may form via cell fusion rather than through failed cytokinesis and cell division. Future studies on mechanism of formation and how that relates to progeny production both in vitro and in vivo are needed. Future studies into the polyploid cancer cell state should expand the use of various in vitro cell line models, including but not limited to MDA-MB-231 and PC3 lineages, to evaluate the ubiquity and implications of the convergent phenotype observed. To understand the behavior and regulatory elements of this cell state, techniques such as RNA-sequencing (to analyze gene expression), ATAC-sequencing (to assess chromatin accessibility), and Bisulfitesequencing (to examine DNA methylation patterns) should be employed. By delineating commonalities across additional model systems and employing these comprehensive analytical approaches, researchers can gain deeper insights into the fundamental biology of these cells. This approach may also uncover new pathways for therapeutic intervention and enhance our understanding of how these cells contribute to cancer progression and treatment resistance. The study of progeny from polyploid cancer cells remains a provocative and vital topic in cancer research. A significant limitation of current findings, including those from our study, is the lack of detailed experimental replicates. This deficiency restricts our ability to thoroughly 90 investigate the mechanisms by which these cells contribute to cancer progression and recurrence. Future research should focus on robustly isolating and characterizing these progeny cells to better understand their role in the disease process. By doing so, we can uncover how they influence tumor dynamics and therapy resistance, providing potential targets for innovative treatment strategies. To further explore the progeny conundrum of polyploid cancer cells, this study demonstrates that polyploid cancer cells can produce proliferative progeny that retain the survival biomarkers TNFRSF9, HOMER1, and LRP1. Although many polyploid cancer cells either entered a senescent-like state or underwent apoptosis, viable offspring were successfully recovered from one docetaxel-induced polyploid cancer cell (progeny-1 clone). However, the extremely low success rate of recovering progeny—only 1 out of 960 for docetaxel and 0 out of 480 for cisplatin-induced polyploid cancer cells—underscores a significant challenge in this field. These cells are exceedingly difficult to isolate in vitro and maintain from single cells, leading to qualitative rather than quantitative analyses of progeny formation. Despite this challenge, the fact that even a single progeny cell may lead to cancer recurrence in the patient is profoundly significant and merits attention. Studying progeny may provide deeper insights into cancer biology and therapeutic resistance. Future studies should focus on isolating progeny from polyploid cancer cells and elucidating the mechanisms of ploidy reduction and cell division. One potential difficulty is that these cells were isolated individually, preventing cell-to-cell communication that could be crucial for survival and proliferation. To address this, future research should include transwell assays or utilize conditioned mediums to explore the role of paracrine signaling in the re-initiation of the cell cycle in polyploid cancer cells. These approaches could provide deeper insights into the complex biology of polyploid cancer cells and their progeny, paving the way for novel therapeutic strategies. 91 A major limitation of our study is the lack of functional confirmation for the novel survival biomarkers TNFRSF9, HOMER1, and LRP1. These biomarkers are promising due to their upregulation at both the RNA and protein levels in polyploid cancer cells, and their association with chemotherapy resistance in patient tumors. While pathway analyses and existing literature link these genes to critical signaling pathways like PI3K-AKT-mTOR and NF-κB, the detailed mechanisms of their roles in polyploid CTCs and in disease recurrence has yet to be established. To conclusively determine their functional significance, future studies are needed that involve targeted genetic modifications, such as gene knockout or knockdown experiments. These studies will help to clarify the direct roles of TNFRSF9, HOMER1, and LRP1 in promoting the survival of polyploid CTCs and their contribution to therapy resistance and cancer recurrence. Such functional validations are essential to fully leverage these biomarkers in the development of more effective therapeutic strategies. Summary Collectively, these chapters form a tapestry of innovation and discovery, united by the overarching theme of harnessing multi-omics approaches to advance cancer diagnostics and monitoring strategies through LBx. By unraveling the intricate complexities of cancer biology, this work aims to usher in a new era of precision oncology, where personalized monitoring and therapeutic interventions hold the promise of improved patient outcomes and enhanced quality of life. As we embark on this transformative journey, fueled by curiosity, collaboration, and cutting-edge technology, we stand on the precipice of a brighter future in the fight against cancer. 92 References 1. Bukowski K, Kciuk M, Kontek R. Mechanisms of Multidrug Resistance in Cancer Chemotherapy. Int J Mol Sci. 2020;21. 2. Satam H, Joshi K, Mangrolia U, Waghoo S, Zaidi G, Rawool S, et al. Next-Generation Sequencing Technology: Current Trends and Advancements. Biology . 2023;12. 3. Hasin Y, Seldin M, Lusis A. Multi-omics approaches to disease. Genome Biol. 2017;18:83. 4. Demir Karaman E, Işık Z. Multi-Omics Data Analysis Identifies Prognostic Biomarkers across Cancers. Med Sci (Basel). 2023;11. 5. Li R, Zhou W. Multi-omics analysis to screen potential therapeutic biomarkers for anti-cancer compounds. Heliyon. 2022;8:e09616. 6. Dhillon A, Singh A, Bhalla VK. A systematic review on biomarker identification for cancer diagnosis and prognosis in multi-omics: From computational needs to machine learning and deep learning. Arch Comput Methods Eng. 2023;30:917–49. 7. Dagogo-Jack I, Shaw AT. Tumour heterogeneity and resistance to cancer therapies. Nat Rev Clin Oncol. 2018;15:81–94. 8. Puram SV, Tirosh I, Parikh AS, Patel AP, Yizhak K, Gillespie S, et al. Single-Cell Transcriptomic Analysis of Primary and Metastatic Tumor Ecosystems in Head and Neck Cancer. Cell. 2017;171:1611–24.e24. 9. Bassez A, Vos H, Van Dyck L, Floris G, Arijs I, Desmedt C, et al. A single-cell map of intratumoral changes during anti-PD1 treatment of patients with breast cancer. Nat Med. 2021;27:820–32. 10. Duan M, Hao J, Cui S, Worthley DL, Zhang S, Wang Z, et al. Diverse modes of clonal evolution in HBV-related hepatocellular carcinoma revealed by single-cell genome sequencing. Cell Res. 2018;28:359–73. 11. Wu H, Zhang X-Y, Hu Z, Hou Q, Zhang H, Li Y, et al. Evolution and heterogeneity of nonhereditary colorectal cancer revealed by single-cell exome sequencing. Oncogene. 2017;36:2857–67. 12. Liang A, Kong Y, Chen Z, Qiu Y, Wu Y, Zhu X, et al. Advancements and applications of single-cell multi-omics techniques in cancer research: Unveiling heterogeneity and paving the way for precision therapeutics. Biochem Biophys Rep. 2024;37:101589. 13. Baysoy A, Bai Z, Satija R, Fan R. The technological landscape and applications of singlecell multi-omics. Nat Rev Mol Cell Biol. 2023;24:695–713. 14. Macaulay IC, Haerty W, Kumar P, Li YI, Hu TX, Teng MJ, et al. G&T-seq: parallel sequencing of single-cell genomes and transcriptomes. Nat Methods. 2015;12:519–22. 93 15. Zhu Q, Zhao X, Zhang Y, Li Y, Liu S, Han J, et al. Single cell multi-omics reveal intra-cellline heterogeneity across human cancer cell lines. Nat Commun. 2023;14:8170. 16. Alix-Panabières C, Marchetti D, Lang JE. Liquid biopsy: from concept to clinical application. Sci Rep. 2023;13:21685. 17. Keup C, Suryaprakash V, Storbeck M, Hoffmann O, Kimmig R, Kasimir-Bauer S. Longitudinal Multi-Parametric Liquid Biopsy Approach Identifies Unique Features of Circulating Tumor Cell, Extracellular Vesicle, and Cell-Free DNA Characterization for Disease Monitoring in Metastatic Breast Cancer Patients. Cells. 2021;10. 18. Ococks E, Frankell AM, Masque Soler N, Grehan N, Northrop A, Coles H, et al. Longitudinal tracking of 97 esophageal adenocarcinomas using liquid biopsy sampling. Ann Oncol. 2021;32:522–32. 19. Russo A, De Miguel Perez D, Gunasekaran M, Scilla K, Lapidus R, Cooper B, et al. Liquid biopsy tracking of lung tumor evolutions over time. Expert Rev Mol Diagn. 2019;19:1099–108. 20. Baca SC, Seo J-H, Davidsohn MP, Fortunato B, Semaan K, Sotudian S, et al. Liquid biopsy epigenomic profiling for cancer subtyping. Nat Med. 2023;29:2737–41. 21. Berry JL, Xu L, Murphree AL, Krishnan S, Stachelek K, Zolfaghari E, et al. Potential of Aqueous Humor as a Surrogate Tumor Biopsy for Retinoblastoma. JAMA Ophthalmol. 2017;135:1221–30. 22. Liu J, Ottaviani D, Sefta M, Desbrousses C, Chapeaublanc E, Aschero R, et al. A high-risk retinoblastoma subtype with stemness features, dedifferentiated cone states and neuronal/ganglion cell gene expression. Nat Commun. 2021;12:5578. 23. Muniyandi A, Jensen NR, Devanathan N, Dimaras H, Corson TW. The Potential of Aqueous Humor Sampling in Diagnosis, Prognosis, and Treatment of Retinoblastoma. Invest Ophthalmol Vis Sci. 2024;65:18. 24. Pike S, Peng C-C, Neviani P, Berry JL, Xu L. CD63/81 Small Extracellular Vesicles in the Aqueous Humor are Retinoblastoma Associated. Invest Ophthalmol Vis Sci. 2023;64:5. 25. Berry JL, Xu L, Murphree AL, Krishnan S, Stachelek K, Zolfaghari E, et al. Potential of Aqueous Humor as a Surrogate Tumor Biopsy for Retinoblastoma. JAMA Ophthalmol. 2017;135:1221–30. 26. Xu L, Polski A, Prabakar RK, Reid MW, Chevez-Barrios P, Jubran R, et al. Chromosome 6p Amplification in Aqueous Humor Cell-Free DNA Is a Prognostic Biomarker for Retinoblastoma Ocular Survival. Mol Cancer Res. 2020;18:1166–75. 27. Christodoulou E, Yellapantula V, O’Halloran K, Xu L, Berry JL, Cotter JA, et al. Combined low-pass whole genome and targeted sequencing in liquid biopsies for pediatric solid tumors. NPJ Precis Oncol. 2023;7:21. 28. Schmidt MJ, Prabakar RK, Pike S, Yellapantula V, Peng C-C, Kuhn P, et al. Simultaneous Copy Number Alteration and Single-Nucleotide Variation Analysis in Matched Aqueous Humor and Tumor Samples in Children with Retinoblastoma. Int J Mol Sci. 2023;24. 94 29. Lim SB, Di Lee W, Vasudevan J, Lim W-T, Lim CT. Liquid biopsy: one cell at a time. NPJ Precis Oncol. 2019;3:23. 30. Welter L, Xu L, McKinley D, Dago AE, Prabakar RK, Restrepo-Vassalli S, et al. Treatment response and tumor evolution: lessons from an extended series of multianalyte liquid biopsies in a metastatic breast cancer patient. Cold Spring Harb Mol Case Stud. 2020;6. 31. Chai S, Matsumoto N, Storgard R, Peng C-C, Aparicio A, Ormseth B, et al. Platelet-Coated Circulating Tumor Cells Are a Predictive Biomarker in Patients with Metastatic CastrateResistant Prostate Cancer. Mol Cancer Res. 2021;19:2036–45. 32. Jie X-X, Zhang X-Y, Xu C-J. Epithelial-to-mesenchymal transition, circulating tumor cells and cancer metastasis: Mechanisms and clinical applications. Oncotarget. 2017;8:81558–71. 33. Reduzzi C, Gerratana L, Zhang Y, D’Amico P, Shah AN, Davis AA, et al. CK+/CD45+ (dualpositive) circulating cells are associated with prognosis in patients with advanced breast cancer. J Clin Oncol. 2022;40:1093–1093. 34. Sheng W, Ogunwobi OO, Chen T, Zhang J, George TJ, Liu C, et al. Capture, release and culture of circulating tumor cells from pancreatic cancer patients using an enhanced mixing chip. Lab Chip. 2014;14:89–98. 35. Ye Z, Ding Y, Chen Z, Li Z, Ma S, Xu Z, et al. Detecting and phenotyping of aneuploid circulating tumor cells in patients with various malignancies. Cancer Biol Ther. 2019;20:546–51. 36. Batool SM, Yekula A, Khanna P, Hsia T, Gamblin AS, Ekanayake E, et al. The Liquid Biopsy Consortium: Challenges and opportunities for early cancer detection and monitoring. Cell Rep Med. 2023;4:101198. 37. Yee NS. Liquid Biopsy: A Biomarker-Driven Tool towards Precision Oncology. J Clin Med Res. 2020;9. 38. Lv H, Shi Y, Zhang L, Zhang D, Liu G, Yang Z, et al. Polyploid giant cancer cells with budding and the expression of cyclin E, S-phase kinase-associated protein 2, stathmin associated with the grading and metastasis in serous ovarian tumor. BMC Cancer. 2014;14:576. 39. Fei F, Zhang D, Yang Z, Wang S, Wang X, Wu Z, et al. The number of polyploid giant cancer cells and epithelial-mesenchymal transition-related proteins are associated with invasion and metastasis in human breast cancer. J Exp Clin Cancer Res. 2015;34:158. 40. Brooks PJ, Glogauer M, McCulloch CA. An Overview of the Derivation and Function of Multinucleated Giant Cells and Their Role in Pathologic Processes. Am J Pathol. 2019;189:1145–58. 41. Chen J, Niu N, Zhang J, Qi L, Shen W, Donkena KV, et al. Polyploid Giant Cancer Cells (PGCCs): The Evil Roots of Cancer. Curr Cancer Drug Targets. 2019;19:360–7. 42. Mirzayans R, Andrais B, Murray D. Roles of Polyploid/Multinucleated Giant Cancer Cells in Metastasis and Disease Relapse Following Anticancer Treatment. Cancers . 2018;10. 95 43. Xuan B, Ghosh D, Dawson MR. Contributions of the distinct biophysical phenotype of polyploidal giant cancer cells to cancer progression. Semin Cancer Biol. 2022;81:64–72. 44. Trabzonlu L, Pienta KJ, Trock BJ, De Marzo AM, Amend SR. Presence of cells in the polyaneuploid cancer cell (PACC) state predicts the risk of recurrence in prostate cancer. Prostate. 2023;83:277–85. 45. Garrido Castillo LN, Anract J, Delongchamps NB, Huillard O, BenMohamed F, Decina A, et al. Polyploid Giant Cancer Cells Are Frequently Found in the Urine of Prostate Cancer Patients. Cancers . 2023;15. 46. Xuan B, Ghosh D, Cheney EM, Clifton EM, Dawson MR. Dysregulation in Actin Cytoskeletal Organization Drives Increased Stiffness and Migratory Persistence in Polyploidal Giant Cancer Cells. Sci Rep. 2018;8:11935. 47. Song Y, Zhao Y, Deng Z, Zhao R, Huang Q. Stress-Induced Polyploid Giant Cancer Cells: Unique Way of Formation and Non-Negligible Characteristics. Front Oncol. 2021;11:724781. 48. Was H, Borkowska A, Olszewska A, Klemba A, Marciniak M, Synowiec A, et al. Polyploidy formation in cancer cells: How a Trojan horse is born. Semin Cancer Biol. 2022;81:24–36. 49. Zhang X, Yao J, Li X, Niu N, Liu Y, Hajek RA, et al. Targeting polyploid giant cancer cells potentiates a therapeutic response and overcomes resistance to PARP inhibitors in ovarian cancer. Sci Adv. 2023;9:eadf7195. 50. Rao R, Honavar SG. Retinoblastoma. Indian J Pediatr. 2017;84:937–44. 51. Lohmann DR, Gallie BL. Retinoblastoma: revisiting the model prototype of inherited cancer. Am J Med Genet C Semin Med Genet. 2004;129C:23–8. 52. Dimaras H, Kimani K, Dimba EAO, Gronsdahl P, White A, Chan HSL, et al. Retinoblastoma. Lancet. 2012;379:1436–46. 53. Dimaras H, Corson TW. Retinoblastoma, the visible CNS tumor: A review. J Neurosci Res. 2019;97:29–44. 54. Afshar AR, Pekmezci M, Bloomer MM, Cadenas NJ, Stevers M, Banerjee A, et al. NextGeneration Sequencing of Retinoblastoma Identifies Pathogenic Alterations beyond RB1 Inactivation That Correlate with Aggressive Histopathologic Features. Ophthalmology. 2020;127:804–13. 55. Francis JH, Richards AL, Mandelker DL, Berger MF, Walsh MF, Dunkel IJ, et al. Molecular Changes in Retinoblastoma beyond : Findings from Next-Generation Sequencing. Cancers . 2021;13. 56. McEvoy JD, Dyer MA. Genetic and Epigenetic Discoveries in Human Retinoblastoma. Crit Rev Oncog. 2015;20:217–25. 57. Wydner KL, Bhattacharya S, Eckner R, Lawrence JB, Livingston DM. Localization of human CREB-binding protein gene (CREBBP) to 16p13.2-p13.3 by fluorescence in situ hybridization. Genomics. 1995;30:395–6. 96 58. Kooi IE, Mol BM, Massink MPG, Ameziane N, Meijers-Heijboer H, Dommering CJ, et al. Somatic genomic alterations in retinoblastoma beyond RB1 are rare and limited to copy number changes. Sci Rep. 2016;6:25264. 59. Berry JL, Xu L, Kooi I, Murphree AL, Prabakar RK, Reid M, et al. Genomic cfDNA Analysis of Aqueous Humor in Retinoblastoma Predicts Eye Salvage: The Surrogate Tumor Biopsy for Retinoblastoma. Mol Cancer Res. 2018;16:1701–12. 60. Gerrish A, Stone E, Clokie S, Ainsworth JR, Jenkinson H, McCalla M, et al. Non-invasive diagnosis of retinoblastoma using cell-free DNA from aqueous humour. Br J Ophthalmol. 2019;103:721–4. 61. Le Gall J, Dehainault C, Benoist C, Matet A, Lumbroso-Le Rouic L, Aerts I, et al. Highly Sensitive Detection Method of Retinoblastoma Genetic Predisposition and Biomarkers. J Mol Diagn. 2021;23:1714–21. 62. Raval V, Racher H, Wrenn J, Singh AD. Aqueous humor as a surrogate biomarker for retinoblastoma tumor tissue. J AAPOS. 2022;26:137.e1–137.e5. 63. Luo Y, Xu M, Yang L, Yao Y, Berry JL, Xu L, et al. Correlating somatic copy number alteration in aqueous humour cfDNA with chemotherapy history, eye salvage and pathological features in retinoblastoma. Br J Ophthalmol. 2024;108:449–56. 64. Xu L, Shen L, Polski A, Prabakar RK, Shah R, Jubran R, et al. Simultaneous identification of clinically relevant mutations and copy number alterations in aqueous humor of retinoblastoma eyes. Ophthalmic Genet. 2020;41:526–32. 65. Devarajan B, Prakash L, Kannan TR, Abraham AA, Kim U, Muthukkaruppan V, et al. Targeted next generation sequencing of RB1 gene for the molecular diagnosis of Retinoblastoma. BMC Cancer. 2015;15:320. 66. Li H-T, Xu L, Weisenberger DJ, Li M, Zhou W, Peng C-C, et al. Characterizing DNA methylation signatures of retinoblastoma using aqueous humor liquid biopsy. Nat Commun. 2022;13:5523. 67. Kothari P, Marass F, Yang JL, Stewart CM, Stephens D, Patel J, et al. Cell-free DNA profiling in retinoblastoma patients with advanced intraocular disease: An MSKCC experience. Cancer Med. 2020;9:6093–101. 68. Im DH, Pike S, Reid MW, Peng C-C, Sirivolu S, Grossniklaus HE, et al. A Multicenter Analysis of Nucleic Acid Quantification Using Aqueous Humor Liquid Biopsy in Retinoblastoma: Implications for Clinical Testing. Ophthalmol Sci. 2023;3:100289. 69. Berry JL, Xu L, Polski A, Jubran R, Kuhn P, Kim JW, et al. Aqueous Humor Is Superior to Blood as a Liquid Biopsy for Retinoblastoma. Ophthalmology. 2020;127:552–4. 70. Dunn P, Albury CL, Maksemous N, Benton MC, Sutherland HG, Smith RA, et al. Next Generation Sequencing Methods for Diagnosis of Epilepsy Syndromes. Front Genet. 2018;9:20. 71. Kuilman T, Velds A, Kemper K, Ranzani M, Bombardelli L, Hoogstraat M, et al. CopywriteR: DNA copy number detection from off-target sequence data. Genome Biol. 2015;16:49. 97 72. Baslan T, Kendall J, Ward B, Cox H, Leotta A, Rodgers L, et al. Optimizing sparse sequencing of single cells for highly multiplex copy number profiling. Genome Res. 2015;25:714–24. 73. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics. 2009;25:2078–9. 74. Benjamin D, Sato T, Cibulskis K, Getz G, Stewart C, Lichtenstein L. Calling Somatic SNVs and Indels with Mutect2. bioRxiv. 2019. 75. Yang H, Wang K. Genomic variant annotation and prioritization with ANNOVAR and wANNOVAR. Nat Protoc. 2015;10:1556–66. 76. Kumar P, Henikoff S, Ng PC. Predicting the effects of coding non-synonymous variants on protein function using the SIFT algorithm. Nat Protoc. 2009;4:1073–81. 77. Koch L. Exploring human genomic diversity with gnomAD. Nature reviews. Genetics. 2020;21:448. 78. Adzhubei I, Jordan DM, Sunyaev SR. Predicting functional effect of human missense mutations using PolyPhen-2. Curr Protoc Hum Genet. 2013;Chapter 7:Unit7.20. 79. Robinson JT, Thorvaldsdóttir H, Winckler W, Guttman M, Lander ES, Getz G, et al. Integrative genomics viewer. Nat Biotechnol. 2011;29:24–6. 80. Massagué J, Obenauf AC. Metastatic colonization by circulating tumour cells. Nature. 2016;529:298–306. 81. Kim M-Y, Oskarsson T, Acharyya S, Nguyen DX, Zhang XH-F, Norton L, et al. Tumor selfseeding by circulating cancer cells. Cell. 2009;139:1315–26. 82. Lambert AW, Pattabiraman DR, Weinberg RA. Emerging Biological Principles of Metastasis. Cell. 2017;168:670–91. 83. Alvarez Cubero MJ, Lorente JA, Robles-Fernandez I, Rodriguez-Martinez A, Puche JL, Serrano MJ. Circulating Tumor Cells: Markers and Methodologies for Enrichment and Detection. Methods Mol Biol. 2017;1634:283–303. 84. Vona G, Sabile A, Louha M, Sitruk V, Romana S, Schütze K, et al. Isolation by size of epithelial tumor cells : a new method for the immunomorphological and molecular characterization of circulatingtumor cells. Am J Pathol. 2000;156:57–63. 85. Marrinucci D, Bethel K, Kolatkar A, Luttgen MS, Malchiodi M, Baehring F, et al. Fluid biopsy in patients with metastatic prostate, pancreatic and breast cancers. Phys Biol. 2012;9:016003. 86. Rosenberg R, Gertler R, Friederichs J, Fuehrer K, Dahm M, Phelps R, et al. Comparison of two density gradient centrifugation systems for the enrichment of disseminated tumor cells in blood. Cytometry. 2002;49:150–8. 98 87. Park ES, Jin C, Guo Q, Ang RR, Duffy SP, Matthews K, et al. Continuous Flow Deformability-Based Separation of Circulating Tumor Cells Using Microfluidic Ratchets. Small. 2016;12:1909–19. 88. Bozzetti C, Quaini F, Squadrilli A, Tiseo M, Frati C, Lagrasta C, et al. Isolation and Characterization of Circulating Tumor Cells in Squamous Cell Carcinoma of the Lung Using a Non-EpCAM-Based Capture Method. PLoS One. 2015;10:e0142891. 89. Cheng Y-H, Chen Y-C, Lin E, Brien R, Jung S, Chen Y-T, et al. Hydro-Seq enables contamination-free high-throughput single-cell RNA-sequencing for circulating tumor cells. Nat Commun. 2019;10:2163. 90. Aceto N, Bardia A, Miyamoto DT, Donaldson MC, Wittner BS, Spencer JA, et al. Circulating tumor cell clusters are oligoclonal precursors of breast cancer metastasis. Cell. 2014;158:1110– 22. 91. Miyamoto DT, Zheng Y, Wittner BS, Lee RJ, Zhu H, Broderick KT, et al. RNA-Seq of single prostate CTCs implicates noncanonical Wnt signaling in antiandrogen resistance. Science. 2015;349:1351–6. 92. Ting DT, Wittner BS, Ligorio M, Vincent Jordan N, Shah AM, Miyamoto DT, et al. Single-cell RNA sequencing identifies extracellular matrix gene expression by pancreatic circulating tumor cells. Cell Rep. 2014;8:1905–18. 93. Powell AA, Talasaz AH, Zhang H, Coram MA, Reddy A, Deng G, et al. Single cell profiling of circulating tumor cells: transcriptional heterogeneity and diversity from breast cancer cell lines. PLoS One. 2012;7:e33788. 94. Gorges TM, Kuske A, Röck K, Mauermann O, Müller V, Peine S, et al. Accession of Tumor Heterogeneity by Multiplex Transcriptome Profiling of Single Circulating Tumor Cells. Clin Chem. 2016;62:1504–15. 95. Thomsen ER, Mich JK, Yao Z, Hodge RD, Doyle AM, Jang S, et al. Fixed single-cell transcriptomic characterization of human radial glial diversity. Nat Methods. 2016;13:87–93. 96. Picelli S, Faridani OR, Björklund AK, Winberg G, Sagasser S, Sandberg R. Full-length RNAseq from single cells using Smart-seq2. Nat Protoc. 2014;9:171–81. 97. Wang X, Zhang H, Chen X. Drug resistance and combating drug resistance in cancer. Cancer Drug Resist. 2019;2:141–60. 98. Cree IA, Charlton P. Molecular chess? Hallmarks of anti-cancer drug resistance. BMC Cancer. 2017;17:10. 99. Siegel RL, Miller KD, Wagle NS, Jemal A. Cancer statistics, 2023. CA Cancer J Clin. 2023;73:17–48. 100. Kurbegovic S, Berg KD, Thomsen FB, Gruschy L, Iversen P, Brasso K, et al. The risk of biochemical recurrence for intermediate-risk prostate cancer after radical prostatectomy. Scand J Urol. 2017;51:450–6. 99 101. Goss PE, Ingle JN, Pritchard KI, Robert NJ, Muss H, Gralow J, et al. Extending Aromatase-Inhibitor Adjuvant Therapy to 10 Years. N Engl J Med. 2016;375:209–19. 102. Colleoni M, Sun Z, Price KN, Karlsson P, Forbes JF, Thürlimann B, et al. Annual Hazard Rates of Recurrence for Breast Cancer During 24 Years of Follow-Up: Results From the International Breast Cancer Study Group Trials I to V. J Clin Oncol. 2016;34:927–35. 103. Ashdown ML, Robinson AP, Yatomi-Clarke SL, Ashdown ML, Allison A, Abbott D, et al. Chemotherapy for Late-Stage Cancer Patients: Meta-Analysis of Complete Response Rates. F1000Res. 2015;4:232. 104. Amend SR, Torga G, Lin K-C, Kostecka LG, de Marzo A, Austin RH, et al. Polyploid giant cancer cells: Unrecognized actuators of tumorigenesis, metastasis, and resistance. Prostate. 2019;79:1489–97. 105. Lopez-Sánchez LM, Jimenez C, Valverde A, Hernandez V, Peñarando J, Martinez A, et al. CoCl2, a mimic of hypoxia, induces formation of polyploid giant cells with stem characteristics in colon cancer. PLoS One. 2014;9:e99143. 106. Saleh T, Carpenter VJ, Bloukh S, Gewirtz DA. Targeting tumor cell senescence and polyploidy as potential therapeutic strategies. Semin Cancer Biol. 2022;81:37–47. 107. Ogden A, Rida PCG, Knudsen BS, Kucuk O, Aneja R. Docetaxel-induced polyploidization may underlie chemoresistance and disease relapse. Cancer Lett. 2015;367:89–92. 108. Puig P-E, Guilly M-N, Bouchot A, Droin N, Cathelin D, Bouyer F, et al. Tumor cells can escape DNA-damaging cisplatin through DNA endoreduplication and reversible polyploidy. Cell Biol Int. 2008;32:1031–43. 109. Adibi R, Moein S, Gheisari Y. Zoledronic acid targets chemo-resistant polyploid giant cancer cells. Sci Rep. 2023;13:419. 110. Bielski CM, Zehir A, Penson AV, Donoghue MTA, Chatila W, Armenia J, et al. Genome doubling shapes the evolution and prognosis of advanced cancers. Nat Genet. 2018;50:1189– 95. 111. López S, Lim EL, Horswell S, Haase K, Huebner A, Dietzen M, et al. Interplay between whole-genome doubling and the accumulation of deleterious alterations in cancer evolution. Nat Genet. 2020;52:283–93. 112. Pienta KJ, Hammarlund EU, Austin RH, Axelrod R, Brown JS, Amend SR. Cancer cells employ an evolutionarily conserved polyploidization program to resist therapy. Semin Cancer Biol. 2022;81:145–59. 113. Pienta KJ, Hammarlund EU, Axelrod R, Brown JS, Amend SR. Poly-aneuploid cancer cells promote evolvability, generating lethal cancer. Evol Appl. 2020;13:1626–34. 114. Zhou X, Zhou M, Zheng M, Tian S, Yang X, Ning Y, et al. Polyploid giant cancer cells and cancer progression. Front Cell Dev Biol. 2022;10:1017588. 100 115. White-Gilbertson S, Lu P, Esobi I, Echesabal-Chen J, Mulholland PJ, Gooz M, et al. Polyploid giant cancer cells are dependent on cholesterol for progeny formation through amitotic division. Sci Rep. 2022;12:8971. 116. Zhou M, Ma Y, Chiang C-C, Rock EC, Butler SC, Anne R, et al. Single-cell morphological and transcriptome analysis unveil inhibitors of polyploid giant breast cancer cells in vitro. Commun Biol. 2023;6:1301. 117. Carroll C, Manaprasertsak A, Boffelli Castro A, van den Bos H, Spierings DCJ, Wardenaar R, et al. Drug-resilient Cancer Cell Phenotype Is Acquired via Polyploidization Associated with Early Stress Response Coupled to HIF2α Transcriptional Regulation. Cancer Res Commun. 2024;4:691–705. 118. Amezquita RA, Lun ATL, Becht E, Carey VJ, Carpp LN, Geistlinger L, et al. Orchestrating single-cell analysis with Bioconductor. Nat Methods. 2020;17:137–45. 119. Liu Z, Lou H, Xie K, Wang H, Chen N, Aparicio OM, et al. Reconstructing cell cycle pseudo time-series via single-cell transcriptome data. Nat Commun. 2017;8:22. 120. Soneson C, Robinson MD. Bias, robustness and scalability in single-cell differential expression analysis. Nat Methods. 2018;15:255–61. 121. Keenan AB, Torre D, Lachmann A, Leong AK, Wojciechowicz ML, Utti V, et al. ChEA3: transcription factor enrichment analysis by orthogonal omics integration. Nucleic Acids Res. 2019;47:W212–24. 122. Liberzon A, Birger C, Thorvaldsdóttir H, Ghandi M, Mesirov JP, Tamayo P. The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst. 2015;1:417–25. 123. Noureen N, Ye Z, Chen Y, Wang X, Zheng S. Signature-scoring methods developed for bulk samples are not adequate for cancer single-cell RNA sequencing data. Elife. 2022;11. 124. Lin A, Yang H, Shi Y, Cheng Q, Liu Z, Zhang J, et al. PanCanSurvPlot: A large-scale pancancer survival analysis web application. bioRxiv. 2022. 125. Jain S, Lyons CA, Walker SM, McQuaid S, Hynes SO, Mitchell DM, et al. Validation of a Metastatic Assay using biopsies to improve risk stratification in patients with prostate cancer treated with radical radiation therapy. Ann Oncol. 2018;29:215–22. 126. Weigman VJ, Chao H-H, Shabalin AA, He X, Parker JS, Nordgard SH, et al. Basal-like Breast cancer DNA copy number losses identify genes involved in genomic instability, response to therapy, and patient survival. Breast Cancer Res Treat. 2012;133:865–80. 127. Martínez-Zamudio RI, Stefa A, Nabuco Leva Ferreira Freitas JA, Vasilopoulos T, Simpson M, Doré G, et al. Escape from oncogene-induced senescence is controlled by POU2F2 and memorized by chromatin scars. Cell Genom. 2023;3:100293. 128. So T, Croft M. Regulation of PI-3-Kinase and Akt Signaling in T Lymphocytes and Other Cells by TNFR Family Molecules. Front Immunol. 2013;4:139. 101 129. Luo L, Wall AA, Tong SJ, Hung Y, Xiao Z, Tarique AA, et al. TLR Crosstalk Activates LRP1 to Recruit Rab8a and PI3Kγ for Suppression of Inflammatory Responses. Cell Rep. 2018;24:3033–44. 130. He Z, Wang G, Wu J, Tang Z, Luo M. The molecular mechanism of LRP1 in physiological vascular homeostasis and signal transduction pathways. Biomed Pharmacother. 2021;139:111667. 131. Fei F, Li J, Rao W, Liu W, Chen X, Su N, et al. Upregulation of Homer1a Promoted Retinal Ganglion Cell Survival After Retinal Ischemia and Reperfusion via Interacting with Erk Pathway. Cell Mol Neurobiol. 2015;35:1039–48. 132. Luo P, Zhao Y, Li D, Chen T, Li S, Chao X, et al. Protective effect of Homer 1a on tumor necrosis factor-α with cycloheximide-induced apoptosis is mediated by mitogen-activated protein kinase pathways. Apoptosis. 2012;17:975–88. 133. Lv W, Zhang Q, Li Y, Liu D, Wu X, He X, et al. Homer1 ameliorates ischemic stroke by inhibiting necroptosis-induced neuronal damage and neuroinflammation. Inflamm Res. 2024;73:131–44. 134. Reshetnikov VV, Bondar NP. The Role of Stress-Induced Changes of Homer1 Expression in Stress Susceptibility. Biochemistry . 2021;86:613–26. 135. Kim C-J, Gonye AL, Truskowski K, Lee C-F, Cho Y-K, Austin RH, et al. Nuclear morphology predicts cell survival to cisplatin chemotherapy. Neoplasia. 2023;42:100906. 136. Kostecka LG, Pienta KJ, Amend SR. Lipid droplet evolution gives insight into polyaneuploid cancer cell lipid droplet functions. Med Oncol. 2021;38:133. 137. Bhaskar PT, Hay N. The two TORCs and Akt. Dev Cell. 2007;12:487–502. 138. Sharma S, Yao H-P, Zhou Y-Q, Zhou J, Zhang R, Wang M-H. Prevention of BMS-777607- induced polyploidy/senescence by mTOR inhibitor AZD8055 sensitizes breast cancer cells to cytotoxic chemotherapeutics. Mol Oncol. 2014;8:469–82. 139. You B, Xia T, Gu M, Zhang Z, Zhang Q, Shen J, et al. AMPK-mTOR-Mediated Activation of Autophagy Promotes Formation of Dormant Polyploid Giant Cancer Cells. Cancer Res. 2022;82:846–58. 140. Smith SF, Collins SE, Charest PG. Ras, PI3K and mTORC2 - three’s a crowd? J Cell Sci. 2020;133. 141. Dan HC, Antonia RJ, Baldwin AS. PI3K/Akt promotes feedforward mTORC2 activation through IKKα. Oncotarget. 2016;7:21064–75. 142. Gaur U, Aggarwal BB. Regulation of proliferation, survival and apoptosis by members of the TNF superfamily. Biochem Pharmacol. 2003;66:1403–8. 143. Gonias SL, Campana WM. LDL receptor-related protein-1: a regulator of inflammation in atherosclerosis, cancer, and injury to the nervous system. Am J Pathol. 2014;184:18–27. 102 144. Terrand J, Bruban V, Zhou L, Gong W, El Asmar Z, May P, et al. LRP1 controls intracellular cholesterol storage and fatty acid synthesis through modulation of Wnt signaling. J Biol Chem. 2009;284:381–8. 103 Appendix1: Supplementary Information 104 105 106 107 Figure 2.S1: SCNA profiles for low pass WGS and targeted sequencing samples for every AH and tumor sample in the cohort. Low pass WGS plots were generated with an internal pipeline, while targeted sequencing plots were made with CopyWriteR. CopyWriteR utilized a female normal control; therefore, sex chromosomes were not considered for analysis (only chromosomes 1-22). 108 Figure 2.S2: Concordance for targeted AH to low pass WGS AH samples. Samples 11, 49, and 50 were sequenced single-end 50bp (remaining patients sequenced paired-end 150bp) and concordance calculations were deemed to not be comparable for these patients. Figure 2.S3: IGV snapshot for case 28 displaying p.R255* variant in the tumor but not the AH. 109 Figure 4.S1: representative gallery of typical CTCs (far left) and CTC-IGC found in bone marrow aspirate of latestage prostate cancer patients. 110 Figure 4.S2: (A-C) Representative copy number profiles of gCTCs and (D) a non-altered megakaryocyte found in the BM of a prostate cancer patient. 111 Figure 4.S3: Copy number profiles from patient in figure 1 showing clonality of CTCs and L-CTCs. 112 Figure 4.S4: Image analysis supplement. (A) Cell count for DMSO control and recovery conditions for MDA-MB-231 (left) and PC3 (right) cells. (B) Cell diameter calculations for MDA-MB-231 (left), PC3 (middle), and PC3 progeny-1 (right) cells. (C) Nuclear proportion (nuclear diameter / cellular diameter) for MDA-MB-231 (left), PC3 (middle), and PC3 progeny-1 (right) cells. (D) Permeability intensity for MDA-MB-231 (left) and PC3 (right) cells. 113 Figure 4.S5: Segmented copy number ratio data for MDA-MB-231 cells show no differences in ratios from control to larger cells. 114 Figure 4.S6. Batches of RNA from PC3 cells. 115 Figure 4.S7. Batches of RNA from MDA-MB-231 cells. 116 Figure 4.S8: RNA sequencing of MDA-MB-231 cells: (A) UMAP visualization of MDA-MB-231 conditions. (B) Cell cycle classification for each condition. (C) Selected cell cycle hallmark pathway classifications. (D) Full single cell hallmark pathway classifications. 117 FIgure 4.S9: RNA sequencing of PC3 cells. (A) Cell cycle percentage classifications. For each condition. (B) Full single cell hallmark pathway classifications for each condition. 118 Figure S10: SAA1 and C3 are correlated with better PFS in breast and prostate cancer patients. 119 Figure 4.S11: Marker supplement. (A) Hypothesized involvement of TNFRSF9, HOMER1, and LRP1 in survival pathways for cancer cells. (B) Volcano plots for PC3 Doc D10 and MDA-MB-231 Doc D10 highlight top expressed genes and marker genes. (C) PC3 CD45 protein expression increases in D10 recovered cells and is retained in progeny. 120 Figure 4.S12: Biomarker staining in patient BM samples. (A) Total cell counts of CTCs negative for marker (EPI) and positive for marker (EPI | HOMER1, TNFRSF9, or LRP1). Cells positive for only the marker, and not EPI, cannot be confidently classified as tumor derived. (B) Signal intensity split by nuclear size. Black stars indicate gCTCs have significantly higher signal than norm-CTCs, while gray stars indicate norm-CTCs have significantly higher signal (Wilcoxon rank-sum test; p < 0.05). 121 Figure 4.S13: Biomarker staining in patient BM samples of polyploid cells that are positive for marker of interest but negative for EPI channel.

Abstract (if available)

Abstract Cancer persists as an enigmatic adversary that the scientific community has yet to fully comprehend. Its daunting presence is marked by both inter-tumoral heterogeneity across patients and intra-tumoral variability within individuals, necessitating advancements to unravel its intricate mechanisms. Enter multi-omics, a transformative approach that, among others, can concurrently explore epigenomic, genomic, transcriptomic, and/or proteomic landscapes. By harnessing this technology, particularly in the realm of liquid biopsy (LBx) for cancer analytics like circulating tumor cells (CTCs) and circulating tumor DNA, new horizons are primed for groundbreaking discoveries that could facilitate a deeper understanding of any given cancer and revolutionize patient care.

This dissertation embarks on a journey, beginning with the demonstration of a cutting-edge targeted genomic sequencing technique applied to LBx in pediatric Retinoblastoma patients. By simultaneously profiling copy number and single nucleotide variants through a targeted sequencing approach, this method holds promise for disease monitoring and enhancing patient well-being (Chapter 2). Recognizing the imperative for refined molecular characterization of CTCs, this work proceeds to introduce GeTMoR, an innovative single cell multi-omic approach delving into the Genome, Transcriptome, and Morphology of individual Rare Cells (Chapter 3). Leveraging elements of the GeTMoR pipeline, this study investigates a previously overlooked population of therapy-resistant cells—large polyploid cancer cells—identifying novel biomarkers linked to chemotherapy recurrence (Chapter 4). Cumulatively, this body of work seeks to push the boundaries of single cell multi-omics and LBx, ultimately striving for a deeper comprehension of the complexities of cancer and paving the way for improved clinical outcomes.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Heterogeneity and plasticity of malignant and non-malignant circulating analytes in breast carcinomas

PDF

RNA methylation in cancer plasticity and drug resistance

PDF

Deconvolution of circulating tumor cell heterogeneity and implications for aggressive variant prostate cancer

PDF

Molecular signature of aggressive disease and clonal diversity revealed by single-cell copy number analysis of prostate cancer cells across multiple disease states

PDF

Exploring the effects of CXCR4 inhibition on circulating tumor cell populations in metastatic prostate cancer

PDF

Early detection of lung cancer by characterizing circulating rare cells using peripheral blood liquid biopsy

PDF

Potential of aqueous humor as a liquid biopsy for uveal melanoma

PDF

Cisplatin activates mitochondrial oxphos leading to acute treatment resistance in bladder cancer

PDF

Study of bone morphogenetic protein-2 and stromal cell derived factor-1 in prostate cancer

PDF

pRB structures and functions mediating suppression of retinoblastoma initiation

PDF

Optimization of circulating tumor cells isolation for gene expression analysis

PDF

Model development of breast cancer detection and staging via rare event enumeration from a liquid biopsy: a retrospective descriptive clinical research study

PDF

Developing a robust single cell whole genome bisulfite sequencing protocol to analyse circulating tumor cells

PDF

Identification of CBP/FOXM1 as a molecular target in triple negative breast cancer

PDF

The cancer stem-like phenotype: therapeutics, phenotypic plasticity and mechanistic studies

PDF

Multimodal single-cell biology and machine learning to characterize plasma cell neoplasms

PDF

Exploration of the roles of cancer stem cells and survivin in the pathogenesis and progression of prostate cancer

PDF

Integrative analysis of multi-view data with applications in epidemiology

PDF

Circulating tumor cells as biomarkers in lung cancer

PDF

Prostate cancer: genetic susceptibility and lifestyle risk factors

Asset Metadata

Creator Schmidt, Michael James (author)

Core Title Applying multi-omics in cancer liquid biopsy for improved patient monitoring and biomarker discovery

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Cancer Biology and Genomics

Degree Conferral Date 2024-08

Publication Date 06/13/2024

Defense Date 05/09/2024

Publisher Los Angeles, California (original), University of Southern California (original), University of Southern California. Libraries (digital)

Tag cancer,chemotherapy resistance,circulating tumor cell,liquid biopsy,multi-omics,OAI-PMH Harvest,polyaneuploid cancer cell state,polyploid cancer cells,prostate cancer,retinoblastoma,single cell

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Goldkorn, Amir (committee chair), Cobrinik, David (committee member), Hicks, James (committee member), Kuhn, Peter (committee member), Yu, Min (committee member)

Creator Email mjschmid@usc.edu,mjschmidt00@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC1139963AT

Unique identifier UC1139963AT

Identifier etd-SchmidtMic-13105.pdf (filename)

Legacy Identifier etd-SchmidtMic-13105

Document Type Dissertation

Format theses (aat)

Rights Schmidt, Michael James

Internet Media Type application/pdf

Type texts

Source 20240614-usctheses-batch-1169 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu