Constance Jean Eaves

Human breast cancers are well recognized for their genetic and biologic heterogeneity but are, nevertheless, usefully classified into subgroups that guide their treatment. Unfortunately, those now identified as highly proliferative ER-, PR- and HER2-negative cancers (i.e., “triple-negative”, TNBCs) remain generally incurable and hence are in particular need of new strategies to design and test targeted treatments. The ultimate goal of this project was to create a reproducible human model of TNBCs by building on a previous de novo experimental design that consistently produces variably ER-, PR- and HER2-positive human invasive ductal breast carcinomas, from either of two types of EGF-responsive normal human mammary cells, transduced with a KRASᴳ¹²ᴰ(K)-encoding vector prior to being transplanted into adult immunodeficient mice. However, although self-sustaining, these tumours do not expand after the first 4-6 weeks post-transplant. Interestingly, experiments designed to determine whether two candidate extrinsic modifiers, human hepatocyte growth factor and a diet-induced obese environment, might enhance the growth of the K-induced tumours, failed to reveal any evidence of such an effect. In contrast, the concurrent overexpression of BMI1, MYC and TP53ᴿ²⁷³ᶜ with K (KBMT) in human mammary cells showed that within a week, transplants of these cells already contained an increased frequency of Ki67+ cells and subsequently produced still YB-1-dependent, but now continuously expanding TNBCs. Transduction with KRASᴳ¹²ᴰ was required for any tumours to form from this collection of potential oncogenic drivers, although variable outputs were obtained when any of the BMI1, MYC, or TP53ᴿ²⁷³ᶜ oncogenes were omitted. FACS analyses showed the surface marker phenotypes of the cells present in these rapidly growing tumours to be remarkably homogeneous, with features of normal basal cells. In addition, despite any evidence of metastatic capability detected in vivo, cells isolated from all KBMT-induced tumours tested could be expanded in vitro and generated fast growing tumours in secondary mice. These findings demonstrate the feasibility of rapidly and reproducibly creating a model of human TNBC de novo from initially normal mammary cells, thus offering a new and powerful platform for potentially developing novel prevention, diagnostic, as well as treatment approaches applicable to this disease.

View record

The full abstract for this thesis is available in the body of the thesis, and will be available when the embargo expires.

View record

Identification of phenotypes of human hematopoietic cells that display long-term mature cell outputs in vitro and repopulating capability in immunodeficient mice has been important to anticipating the therapeutic potential of fresh harvests of bone marrow or cord blood before or after their physical or genetic manipulation. Characterizing additional properties of these cells and elucidating the mechanisms that regulate their ability to sustain mature blood cell production has thus remained a major focus of interest. Previous studies have shown that fetal and adult human cells with long-term blood cell output potential are highly enriched in their respective GPI80+ and CD49f+ subsets of a developmentally preserved CD45+CD34+CD38-CD45RA-CD90+ population. The so-called “GPI80” hematopoietic cells found in first trimester human fetal liver are of particular interest because of their very high regenerative capability compared to their adult or even neonatal (cord blood) “CD49f” counterparts. However, in vitro conditions that enable the extensive innate self-renewal capability of either of these cell types to be maintained have remained elusive. The goals of this project were first to test the hypothesis that retention of the high regenerative activity of the GPI80+ cells in vitro would be optimized under different conditions than those that regulate their viability or mitogenesis; and second, that the GPI80+ cells of interest would co-express CD49f. The results show first, that the defining long-term cell output function of human hematopoietic stem cells (HSCs) in fetal liver is best maintained in vitro by FLT3L alone despite its poor ability to support their concomitant survival and proliferation. Secondly, co-expression of CD49f within the GPI80+ population identifies a subset with reduced short-term myeloid colony-forming activity in semi-solid medium, and greater progeny outputs in both 12-week growth factor-supplemented stromal co-cultures, and in transplanted immunodeficient mice. These findings demonstrate CD49f is a pervasive marker of human HSCs throughout ontogeny and aging. They also reinforce an increasing body of evidence that the molecular maintenance of potent regenerative potential, even in the first HSCs to appear during human development, is adversely affected by exclusive exposure to factors that most strongly stimulate their mitogenesis and even their survival.

View record

Mechanisms that regulate cell survival and proliferation are important for the both the development and homeostasis of normal tissue. Elements of the pathways that mediate these roles are also variably deregulated both during the establishment and perpetuation of malignant states. CASPASE-3 (CASP3) is well known for its key function in apoptosis, but an involvement in other cellular processes has also been documented. The investigations carried out in this thesis focused on understanding if and how CASP3 might affect the proliferation, survival, and cell cycle progression of normal and malignant human mammary cells. A series of lentiviral-mediated CASP3 knockdown (KD) experiments ± vectors encoding the wild-type CASP3 cDNA or a cDNA lacking the catalytic domain required for its apoptotic function showed that loss of CASP3 diminished both the growth and survival of normal and malignant human mammary cells in vitro and the growth of malignant human mammary cells in vivo. Moreover, the “rescue” experiments showed these activities of CASP3 did not require its known catalytic function. Initial mass spectrometry-based proteomic comparisons of control and CASP3 KD cells were then used to identify molecular targets and pathways dependent on CASP3. This revealed a number of candidates with known roles in controlling the clearance of protein aggregates by the ubiquitin–proteasome system (UPS) and components of the autophagy process. Additional intracellular flow cytometric analyses of these candidate molecular targets in both normal and malignant human mammary cells showed that the levels of protein aggregates were consistently highest in the CASP3 KD cells and, conversely, CASP3 overexpressing cells harboured reduced levels of protein aggregates further implicating an important role of CASP3 in the clearance of protein aggregates. These observations suggest that the toxic consequences of an accumulation of protein aggregates might be a driving factor in causing the impaired survival and proliferation of CASP3 KD human mammary cells. The results of functional assays of the effect of inhibiting these targets added strength to the idea of their involvement in the dependency function of CASP3 in normal mammary cells that may be exploited to enhance the treatment of highly aggressive types of human breast cancer cells.

View record

Hematopoietic stem cells (HSCs) comprise a functionally and molecularly heterogeneous population of cells that collectively maintain the lifelong production of mature blood cells. Known developmental changes in their properties include an early postnatal switch from a rapidly cycling high self-renewal state to a quiescent state with an overall reduced self-renewal potential. Additional age-associated alterations in mouse HSC properties have been predicted but these have remained poorly explored in human HSCs. Of recent interest has been the finding that most healthy humans of advancing age possess enlarged clones of normal blood cells marked by somatic mutations in genes commonly involved in hematopoietic malignancies. Together, these findings suggest that altered regulation of HSC cycling may be a feature that develops with advancing age in humans. To examine this possibility, an investigation of the HSC and progenitor compartments in healthy human donors aged 0-69 years was initiated. A first characterization of cells within the HSC-enriched subset characterized by a CD34+CD38-CD45RA-CD90+CD49f+ (CD49f+) phenotype showed that these cells are consistently present from birth to old age at similarly low frequencies. Functional assays showed their long-term and lympho-myeloid differentiation capacity in vitro and in vivo (in transplanted immunodeficient mice) to be similar but indicated a possible increased in vivo supportive requirement with age. Kinetic analyses of individually tracked CD49f+ cells further revealed a strong and progressive aging-related delay in completing their first and subsequent divisions in vitro that was exacerbated when the growth factor stimulus was reduced. Development of a method for simultaneous cell-cycle staging and multiplexed molecular analysis of single CD49f+ cells traced this delay to a mitogen-sensitive G1 elongation which was also evident at slightly later stages of hematopoietic cell differentiation. This delay appeared related to a reduced immediate growth factor-induced activation of AKT and β-catenin obtained in adult cells. These findings point to a newly identified intrinsic and pervasive, aging-related alteration in specific early signaling intermediates that are required to drive G1 progression in HSCs (and their early progeny) and lay the foundation for further analyses of how this regulatory change may impact the acquisition of other aging-related phenotypes in the hematopoietic system.

View record

Human acute myeloid leukemias (AMLs) are a genetically and biologically diverse group of clonal hematologic malignancies in which abnormally differentiated hematopoietic cells expand excessively and are often characterized by MYC transcription factor overexpression. Current evidence suggests that the acquisition of genetic and epigenetic abnormalities by hematopoietic cells drives the disease. However, it is not known whether these genetic insults alone are sufficient for AML initiation, or from which cell types within the hematopoietic hierarchy the expansion of a leukemic clone can be initiated. To investigate these questions, I first undertook experiments to determine if it would be possible to initiate a leukemia from normal human CD34+ hematopoietic cells forced to overexpress MYC. Initial results showed their transplantation into immunodeficient mice genetically engineered to express human IL3, GM-CSF and SCF growth factors (GFs) resulted in the production of a rapidly fatal AML population, but this did not occur in hosts that did not express these human GFs. In fact, in mice not producing these GFs, the same MYC-transduced cells produced a normal spectrum of differentiating hematopoietic cell types for months. However, upon transfer into secondary GF-producing hosts, they rapidly initiated AML, demonstrating a critical role of these GFs in forcing the activation of a leukemogenic program long after expression of the oncogenic driver was initiated. Transplantation of MYC-transduced cells co-transduced with IL3 or GM-CSF or SCF individually into non-GF mice revealed IL3 and GM-CSF to have equivalent and exclusive AML-activating activity. Transplantation of GF-producing mice with different subsets of MYC-transduced CD34+ cells showed serially transplantable AML populations could be obtained at high efficiency from both primitive (CD34+CD38-) and mature granulopoietic progenitor cells (GMPs). Preliminary analysis of their properties further showed their cell surface phenotypes, transcriptomes, GF-dependence and content of in vitro clonogenic cells to be indistinguishable, and most like normal GMPs amongst different subsets of normal cord blood CD34+ cells. These results indicate that extrinsically-derived immunomodulatory signals can have an important role in eliciting a leukemogenic potential in human cells, and thus lay a foundation for improved understanding of the mechanisms involved in the establishment of human AML.

View record

Intra-tumoural biological, transcriptional and genomic heterogeneity are hallmarks of human breast cancers. However, tumour propagating activity appears confined to subsets of cells within each tumour. This finding is thought to indicate a persistence of mechanisms that maintain a hierarchical growth and differentiation structure in the normal mammary gland. Because little is known about the responses of either primary normal or malignant human mammary cells to existing therapies, this thesis sought to examine the intrinsic sensitivity of different purified human mammary colony-forming cell (CFC) types and tumours derived from them to ionizing radiation.Luminal progenitor (LP) CFCs were found to be ~1.5-fold more radioresistant than basal cell (BC) CFCs and LPs also showed evidence of checkpoint adaptation, slower repair activity and greater predisposition of γH2AX foci accumulation. Two human breast cancer cell lines (MDA-MB231 and SUM149) and a non-tumorigenic human mammary cell line (MCF-10A) were all found to be more radioresistant than the normal LP-CFCs. CFCs isolated from 8-week tumours generated in mice transplanted with normal human BCs or LPs transduced with KRASG¹²D showed even greater radioresistance and this was further increased in serially passaged derivative lines with more aggressive growth properties. To examine the responsiveness of malignant cells with tumor-initiating cell (TIC) activity in vivo, a dose-response analysis was first undertaken of their frequencies in the MDA-MB231 and SUM149 cells. Limiting dilution analysis (LDA) and single-cell transplants showed the frequency of TICs in both to be very high (
View record

The normal mammary gland contains “stem cells” with extensive in vivo growth and bi-lineage differentiation potential and a surface phenotype of basal cells (BCs). BCs also contain cells with more limited growth and differentiation activity in vitro. An analogous luminal-restricted progenitor (LPs) subset has surface characteristics of both basal and luminal cells. I hypothesized that the growth and differentiation activity displayed by individual mammary epithelial cells from both subsets would be highly diverse, and that the properties of tumours produced from these cells would be affected by their cell of origin. To address this hypothesis, I first developed a lentiviral-mediated barcoding strategy that involves transducing each cell with a unique 27-base pair non-coding DNA sequence so that the number of its clonal progeny can be inferred from high-throughput sequencing data obtained on the progeny of bulk-transduced populations. The use of “spiked-in” control cells carrying a known barcode provided an internal calibration for clone size calculations and allowed clones of ≥100 cells to be reliably detected. Application of this strategy to normal mouse and human mammary cells identified expected bi-lineage clones but an unanticipated predominance of lineage-restricted clones produced in primary transplants. These experiments also revealed that many clones apparent in secondary hosts were not detected in the primary hosts, indicating their origin from cells with very delayed growth activity. Application of the barcoding strategy to normal human BCs and LPs transduced with lentiviruses encoding KRASG¹²D ± PI3KCAH¹⁰⁴⁷R ± TP53R²⁷³C showed tumour formation in subsequently transplanted immunodeficient mice was rapid (within 8 weeks) and efficient from both cell types (8-12/18 donors, 1/200-1/4,000 transduced cells). However, tumours generated from LPs contained larger clones than tumours generated from BCs. Surprisingly, none of the LP-derived tumours were ERα⁺ (typical of luminal-like breast cancers) whereas 60% of the BC-derived tumours were. Earlier analysis of xenografts of similarly transduced cells revealed changes in both the number and phenotype of the cells present. Taken together, these findings underscore the diverse regenerative activity of normal mammary cells and provide definitive evidence that the cell of origin can affect the properties of human breast tumours generated using identical oncogenes.

View record

Transplantation of hematopoietic cells is a critical component of treatments used to cure a range of malignancies and congenital disorders. Cells with rapid but short-term repopulating ability (STRCs) are the main source of neutrophils and platelets early post-transplant, but methods to increase their availability are lacking. In this study the use of NOD/SCID-IL-2Receptor-γchain-null mice engineered to produce human interleukin-3, granulocyte-macrophage colony-stimulating factor and Steel factor as hosts of CD34⁺ cells was shown to improve the quantification of human STRCs. This mouse was found to support a 5-fold higher human myeloid cell output at early time-points post-transplant, sufficient to assess total engraftment from an analysis of circulating human cells. This strategy was then used to determine the optimal protocol for mobilizing STRCs in normal adult human donors, comparing administration of granulocyte colony stimulating factor (G-CSF) plus plerixafor (P) to administration of P alone. Blood and marrow samples were obtained from 10 normal adult donors before, during, and after treatment and then evaluated for their content of CD34⁺ cells, colony-forming cells (CFCs), long-term culture-initiating cell activity, and cells with in vivo STRC activity. The results show all activities were maximally increased in the blood 4 hours after administration of P, with or without pre-treatment with G-CSF. In vivo assessment showed that administration of P led to a 30-fold increase in STRC activity over baseline levels, with further enhancement (90-fold over baseline) by prior G-CSF treatment. The ability of currently available growth-factor (GF)-containing HSC expansion protocols to support the expansion of STRCs was then assessed. The results revealed a significant (10-fold) loss of STRC activity after being cultured for 7 to 10 days, in spite of an extensive increase in CD34⁺ cells and CFCs. This GF-induced loss of in vivo STRC activity occurred within 24 hours, and was paralleled in time and amount by a loss of CFC homing to the bone marrow. Together these findings provide a further understanding of STRC biology and provide a foundation for developing improved yields of manipulated STRCs for clinical use.

View record

The concept of stem cell self-renewal was developed from clonal tracking of hematopoietic stem cell (HSC) divisions in vivo 50 years ago. However, protocols to expand these cells in vitro without loss of their stem cell properties have remained elusive. A number of factors contribute to this inability. Key among these is a lack of knowledge of the critical molecular characteristics that distinguish HSCs from hematopoietic progenitors as well as how the control of the fundamental biological programs of survival, division and differentiation are integrated in HSCs. Using a combination of single-cell tracking, transcriptomics, and in vivo readouts applied to highly enriched mouse HSCs, we now show that their survival, proliferation, and maintenance of stem cell properties are mechanistically dissociable. Discovery of a protocol that allows input numbers of functionally intact human HSC numbers to be maintained for 3 weeks in vitro using defined growth factors, was then leveraged to design single human HSC cell tracking and functional analyses. The results of these showed that for human HSC, as in the mouse model, survival, proliferation, and maintenance of stem cell status are mechanistically dissociable, and controlled in a combinatorial manner. We then developed a panel of mass cytometry detectors to enable >40 surface and intracellular proteins to be simultaneously measured at single cell resolution. Using this panel, we identified some of the signaling intermediates activated by growth factors that differentially control human HSC biological responses assessed in high-throughput assays. Correlation of the molecular properties, surface phenotypes and functional activities of CD34+ subsets have further revealed a surprising degree both of heterogeneity within each phenotype and overlap between phenotypes. In some cases, the results suggest a given phenotype contains distinct subsets and a broader scheme of differentiation pathways than suggested by current models of human hematopoietic cell differentiation. Finally, we identify CD33+ as a novel marker which demarcates the most potent human HSC within the current best phenotypic enrichment strategy. These results lay a foundation on which future HSC expansion strategies can be constructed, and have implications for the development of leukemia.

View record

Mouse hematopoietic stem cells (HSCs) undergo a post-natal transition in several properties, including a marked reduction in their self-renewal activity. To investigate the molecular basis of this difference, we devised a single strategy to isolate fetal and adult HSCs at similarly high frequencies. This strategy, involving fluorescence-activated cell sorting of cells with a CD45⁺EPCR⁺CD48-CD150⁺ (ESLAM) phenotype, allows isolation of HSCs at a frequency of ~1 in 2 from all developmental time points tested (mouse embryonic day (E) 14.5 to adult). Comparison of differentially expressed genes in primitive populations of fetal and adult hematopoietic cells showed that heightened expression of Hmga2 was a feature of fetal as compared to adult HSCs. We also identified let-7 microRNAs (miRNAs) and a negative regulator of their biogenesis, Lin28b, to be expressed in an opposite and similar pattern to Hmga2, respectively. Since Hmga2 is a well-established target of let-7 miRNAs, we hypothesized that the Lin28b-let-7-Hmga2 axis plays a central role in the determination of fetal versus adult HSC self-renewal identity. We also found that Lin28 overexpression in adult HSCs restores a higher, fetal-like, self-renewal potential in them, and this effect is phenocopied by direct overexpression of Hmga2. Conversely, HSCs from fetal Hmga2-/- mice display a prematurely acquired adult-like self-renewal activity. Importantly, we show that Lin28-mediated activation of Hmga2 expression, which is responsible for the activation of a fetal-like self-renewal potential in adult HSCs, is not the mechanism by which Lin28 reprograms adult HSCs to undergo fetal-like B-cell differentiation. Together, these findings suggest a model of development in which Lin28b acts as a master regulator and Hmga2 serves as a more specific downstream modulator of HSC self-renewal. These findings may help inform strategies to improve the therapeutic use of HSCs. Furthermore, since Lin28b and Hmga2 are oncogenes, we speculate that the fetal/neonatal specific pattern of expression of these genes may contribute to the pathogenesis of pediatric leukemias.

View record

No abstract available.

No abstract available.

High-level expansion of hematopoietic stem cells (HSCs) in vitro will have an important clinical impact in addition to enabling elucidation of their regulation. Recently, it has been demonstrated that engineered NUP98-HOXA10hd expression stimulates >1,000-fold net expansions of murine HSCs in 10-day cultures initiated with bulk lin⁻Sca-1⁺c-kit⁺ cells. In this thesis I coupled such ability of engineered NUP98-HOXA10hd expression, with strategies to purify fetal and adult HSCs and analyze their expansion clonally. I discovered that NUP98-HOXA10hd stimulates comparable expansions of HSCs from both sources at near unit efficiency in cultures initiated with single cells. The clonally expanded HSCs showed preservation of normal proliferation kinetics in vitro and consistent balanced contributions long-term to the lymphoid and myeloid lineages in vivo without evidence of leukemogenic transformation. Preservation of a normal proliferating HSC phenotype allowed their re-isolation in large numbers at 25% purity. These findings point to the effects of NUP98-HOXA10hd on HSCs in vitro being mediated by promoting self-renewal and set the stage for further dissection of this process. Although there is growing excitement about the prospect of in vitro expansion of HSCs and their use to enhance the safety and application of transplant-based therapies, deleterious consequences of such manipulations remain unknown. Thus, I further examined the impact of HSC self-renewal divisions in vitro and in vivo on their subsequent regenerative and continuing ability to sustain blood cell production in the absence of telomerase. HSC expansion in vitro was obtained using NUP98-HOXA10hd transduction strategy and, in vivo, using a serial transplant protocol. I observed ~10kb telomere loss in leukocytes produced in secondary mice transplanted with HSCs regenerated in primary recipients of NUP98-HOXA10hd-transduced and in vitro-expanded Tert⁻/⁻ HSCs 6 months before. The second generation leukocytes also showed elevated expression of γH2AX (relative to control) indicative of greater accumulating DNA damage. In contrast, significant telomere shortening was not detected in leukocytes produced from freshly isolated, serially transplanted wild-type or Tert⁻/⁻ HSCs, suggesting that HSC replication post-transplant is not limited by telomere shortening in the mouse. These findings document a role of telomerase in telomere homeostasis, and in preserving HSC functional integrity upon prolonged self-renewal stimulation.

View record

No abstract available.

No abstract available.