Thesis: Ph. D., Massachusetts Institute of Technology, Department of Biological Engineering, 2018. / Cataloged from PDF version of thesis. / Includes bibliographical references. / Survival rates for ovarian cancer haven't meaningfully improved in thirty years. Ovarian cancer is particularly difficult to treat because it is usually discovered after it has metastasized and it quickly develops resistance to the few drugs that are initially effective at controlling it. Nanomedicine has the potential to change the paradigm for ovarian cancer treatment by delivering complex combinations of conventional drugs plus next-generation therapies like small interfering RNA (siRNA) and immunotherapy. However, nanoparticles must be tailored to the particular drug-delivery challenges and opportunities posed by ovarian cancer. In this thesis, we designed layer-by-layer (LbL) nanoparticles (NPs) to target ovarian cancer using library-based approaches. Using this approach, we identified promising formulations for developing an advanced nanotheranostic that both treats and detects ovarian cancer. In order to develop LbL NPs for treating ovarian cancer, we identified and overcame process engineering and fundamental materials challenges, thereby improving synthesis robustness, throughput and scale. Chapter 2 describes how modern tangential flow filtration significantly improves throughput and scalability in colloidal LbL assembly. Chapter 3 implements this improved synthetic approach to generate a small library of LbL NPs that screen for tumor-targeting properties on ovarian cancer cells, both in vitro and in vivo. Our results demonstrate that ovarian cancer cells have a high affinity to carboxylated LbL NPs, and we report several tumor-targeting formulations with distinct subcellular trafficking patterns. Chapter 4 explores the role of salt in LbL colloidal assembly, and we develop strategies for robustly synthesizing LbL-modified liposomes with high loading of siRNA. Chapter 5 advances a promising formulation identified by our surface chemistry screen, which we developed into an advanced nanotheranostic device that delivers siRNA and mediates urinary-based tumor detection. Future work that continues to improve the synthesis of LbL NPs will be essential to generate larger and more ambitious LbL NP libraries. In turn, these libraries will facilitate systematic studies that further tailor the LbL platform to specific diseases and biomedical applications. / "This material is partly based upon work supported by the National Science Foundation Graduate Research Fellowship under Grant No. 1122374. This material is partly based upon work supported by the National Science Foundation Graduate Research Fellowship under Grant No. 1122374"--Page 187. / by Santiago Correa. / Ph. D.

Thesis: Ph. D., Massachusetts Institute of Technology, Department of Biological Engineering, 2014. / This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. / Cataloged from student-submitted PDF version of thesis. / Includes bibliographical references. / Mass spectrometry has become the tool of choice for proteomics. Its unrivaled coverage and reproducibility has positioned it head and shoulders above competing techniques for analyzing protein expression post-translational modification. With the increased popularity comes a flood of new research applications, each with its own biological motivations and goals. To ensure that mass spectrometry-based proteomics can be useful to as many biological questions as possible, it is of utmost importance to ensure high data quality. This research focuses on two general stages of the typical proteomics workflow and introduces tools to facilitate effective target screening, follow-up analysis, as well as more precise measurements. This new pipeline is then demonstrated in a case study of Epidermal Growth Factor Receptor (EGFR) signaling and phenotype prediction. The quantity of proteomic mass spectrometry data available from a single analysis has increased exponentially as new generations of instruments become quicker and more sensitive. This deluge of data leaves many tempted to forego time-intensive manual validation of database identified targets in favor of global data set quality statistics. Particularly in the realm of post-translational modifications, long lists of putative matches are often reported with little or no scan-specific validation. Such practices no longer provide assurance that any single identified target is indeed correct, leaving researchers vulnerable to expending vast resources chasing false positives. The argument is that manual validation is too time-intensive to be carried out for each and every identification. To remedy this problem we have introduced the Computer Assisted Manual Validation (CAMV) software package to expedite the procedure by preprocessing the database results so as to remove the tedious steps associated with the validation task and only recruit human judgment for the final quality decision. This approach has drastically decreased the time required for manual validation; a task that used to take weeks now is completed in hours. Another focus of this research is the development of a multiplex, multisite absolute quantification method, which has improved the quality of quantitative proteomic mass spectrometry data. Absolute site-specific data allows many more biological hypotheses to be directly tested with a single mass spectrometry experiment, including phosphorylation stoichiometry. This technique has been applied to the EGFR system to better understand signaling downstream of three distinct ligands. These ligands all bind the same receptor yet elicit different phenotypes, suggesting differential information processing. The analysis showed unique patterns of receptor phosphorylation present following sub-saturating ligand treatment. However, at saturating doses the same pattern of phosphorylation is produced regardless of ligand, but the magnitude of that pattern is still ligand-dependent. In this regime, the adaptor proteins were still able to retain ligand-specific phosphorylation patterns presumably responsible for differential phenotypes. The data set also permitted the identification of signals important for the regulation of only one of the two phenotypes examined. / by Timothy Gordon Curran. / Ph. D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Biological Engineering, 2012. / Cataloged from PDF version of thesis. Page 175 blank. / Includes bibliographical references (p. 161-174). / Immunostimulatory therapies that activate immune response pathways are of great interest for overcoming the immunosuppression present in advanced tumors. Agonistic antibodies against the co-stimulatory receptors CD40 and CD137, Toll-Like Receptor (TLR) ligands such as CpG oligonucleotides, and immunostimulatory cytokines such as IL-2 have all previously demonstrated potent, synergistic anti-tumor effects. However, the clinical use of such therapies is significantly hampered by the severe, dose-limiting inflammatory toxicities provoked upon systemic exposure. We hypothesized that by anchoring immunomodulatory agents to lipid nanoparticles we could retain the bio-activity of therapeutics in the local tumor tissue and tumordraining lymph node, but limit systemic exposure to these potent molecules. We first prepared liposomes bearing surface-conjugated anti-CD40 and CpG and assessed their therapeutic efficacy and systemic toxicity compared to soluble versions of the same immuno-agonists, injected intratumorally in established solid tumors in mice. Anti-CD40/CpG-coupled liposomes significantly inhibited primary tumor growth and induced a survival benefit similar to locally injected soluble anti-CD40+CpG. Biodistribution analyses following local delivery showed that the liposomal carriers successfully sequestered anti-CD40 and CpG in vivo, reducing leakage into systemic circulation while allowing draining to the tumor-proximal lymph node. Contrary to locally administered soluble immunotherapy, anti-CD40/CpG liposomes did not elicit significant increases in serum levels of ALT enzyme, systemic inflammatory cytokines, or overall weight loss, confirming that off-target inflammatory effects had been minimized. Thus, these results confirmed the development of a delivery strategy capable of inducing robust antitumor responses concurrent with minimal systemic side effects. We next assessed the dissemination of the tumor-specific immune response that had been primed by locally administered, liposome-conjugated therapy. Since anti-CD40/CpG-coupled liposomes were unable to consistently induce the rejection of a secondary distal tumor challenge, we adapted the strategy of liposome-coupled delivery for the administration of anti-CD 137 and IL-2, two potent T cell-stimulatory agents. Local intra-tumoral therapy using anti-CD137-liposomes + IL-2-liposomes induced the highly potent inhibition of primary treated tumors and achieved a majority of complete cures, while successfully minimizing systemic exposure and eliminating symptoms of inflammatory toxicity, including lethality. In addition, 100% of anti-CD 137 + IL-2 liposome-treated mice were protected against a secondary distal tumor challenge, and demonstrated a significant delay in the progression of simultaneously inoculated, distal untreated tumors. Subsequent analyses confirmed that anti-CD137-liposomes and IL-2-liposomes bound specifically to cytotoxic T cells (CTLs) within the treated tumor, and that the depletion of CTLs abrogated the therapeutic anti-tumor response. Overall, these results indicated the effective local priming of an adaptive tumor-specific response, capable of mediating local, systemic, and memory anti-tumor immunity. The versatility of this liposome conjugation strategy suggests that we have developed a generalizable tool enabling the local delivery of highly potent immunomodulatory agonists in the absence of systemic toxicity, which could substantially improve the clinical applicability of such agents in cancer therapy. / by Brandon Kwong. / Ph.D.

Thesis: Ph. D., Massachusetts Institute of Technology, Department of Biological Engineering, 2016. / Cataloged from PDF version of thesis. / Includes bibliographical references. / The integrity of our DNA is challenged by roughly 100,000 lesions per cell on a daily basis. Failure to cope with DNA damage can lead to cancer, immunodeficiency and degenerative disease. Quantitating and understanding an individual's DNA repair capacity may enable us to predict and prevent disease in a personalized manner. Base Excision Repair (BER) is known for the recognition and repair of endogenous and exogenous mutagenic non-helix-distorting lesions produced by DNA base alkylation, deamination and oxidation. BER is initiated by the action of one of eleven DNA glycosylases known-to-date. Many studies have shown that levels of these glycosylases can vary between individuals, suggesting a basis for inter-individual differences in DNA repair capacity. Moreover, the methods for measuring DNA repair capacity used so far are cumbersome, time consuming, low throughput and only allow for the analysis of one glycosylase at a time. We have taken a fluorescence-based multiplex flow-cytometric host cell reactivation assay wherein the activity of several DNA glycosylases and their immediate downstream endonuclease (APE1) can be tested simultaneously, at single-cell resolution, under physiological conditions. Taking advantage of the transcriptional properties of several DNA lesions we have designed and engineered specific fluorescent reporter plasmids for OGG1, AAG, MUTYH, UNG and APE1. Inter-individual differences in DNA repair capacity of a panel of cell lines derived from healthy individuals have been measured. Regression models that incorporate these measurements have been developed in order to predict cellular sensitivity to the chemotherapeutic and DNA damaging agents 5-FU, H₂O₂ and MMS, with the interest of understanding the contributions that these differences can have on personalized disease prevention and treatment. Finally, we have conducted a pilot population study with 56 healthy subjects where we implemented all the methods developed in order to determine the feasibility of measuring DNA repair capacity variations in a healthy human population. Additionally, we report the discovery of a novel in vivo role of the TC-NER pathway in the repair of the lipid-peroxidation product, 3,N⁴-etheno-cytosine. / by Isaac Alexander Chaim. / Ph. D.

Thesis: Ph. D., Massachusetts Institute of Technology, Department of Biological Engineering, 2015. / Cataloged from PDF version of thesis. / Includes bibliographical references. / Investigating protein location and concentration is critical to understanding function. Reagentless biosensors, in which a reporting fluorophore is conjugated to a binding scaffold, can detect analytes of interest with high temporal and spatial resolution. However, because these biosensors require laborious empirical screening to develop, their adoption has been limited. Hence, we establish design principles that will facilitate development. In this thesis, we first develop a kinetic model for the dynamic performance of a reagentless biosensor. Using a sinusoidal signal for ligand concentration, our findings suggest that it is optimal to use a binding moiety whose equilibrium dissociation constant matches that of the average predicted input signal, while maximizing both the association rate constant and the dissociation rate constant at the necessary ratio to create the desired equilibrium constant. Although practical limitations constrain the attainment of these objectives, the derivation of these design principles provides guidance for improved reagentless biosensor performance and metrics for quality standards in the development of biosensors. Following these guidelines, we use the human tenth type III fibronectin domain to engineer new binders against several ligands of the EGFR receptor. Using these binders and others, we design and characterize biosensors based on various target analytes, scaffolds and fluorophores. We observe that analytes can harbor specific binding pockets for the fluorophore, which sharply increase the fluorescence produced upon binding. Furthermore, we demonstrate that a fluorophore conjugated to locally rigid surfaces possesses lower background fluorescence. Based on these newly identified properties, we design biosensors that produce a 100-fold increase in fluorescence upon binding to analyte, about a 10-fold improvement over the previous best biosensor. In order to improve the methodology of reagentless biosensor design, we establish a method for site-specific labeling of proteins displayed on the surface of yeasts. This procedure allows for the screening of libraries of sensors for binding and fluorescence enhancement simultaneously. Finally, we explore an alternative sensor design, based on competitive inhibition of fluorescence quenching. / by Seymour de Picciotto. / Ph. D.

Thesis: Ph. D., Massachusetts Institute of Technology, Department of Biological Engineering, 2015. / Cataloged from PDF version of thesis. / Includes bibliographical references. / Non-small cell lung carcinoma (NSCLC) is the most common type of lung cancer and is the leading cause of cancer related mortality worldwide. The past decade has seen exciting advances in the development of targeted therapies for the treatment of NSCLC. However, the efficacy of such therapies in prolonging patient survival has been disappointing, and there remains an urgent need for a greater understanding of the molecular events involved in tumor progression and metastases. MET and members of the epidermal growth factor receptor (EGFR)- family are cell-surface receptor tyrosine kinases (RTK) commonly upregulated in NSCLC cells, and their signaling crosstalk can confer therapy resistance. The four EGFR-family members are similar in structure; each contains a ligand binding domain, a transmembrane region, and a phosphotyrosine cytosolic domain. Upon ligand binding, each of the family members dimerizes in a homo- or heterotypic manner to initiate signaling cascades that influence migration and proliferation outcomes. This thesis describes the use of engineered bivalent ligands to systematically manipulate receptor dimerization, perturb downstream signaling processes, and affect phenotypic outcomes. Results showed that biasing away from an EGFR-HER-2 with a bivalent ligand reduces cell motility significantly in a stem-cell model cell line, but does not inhibit cell proliferation or cell survival. Furthermore, reengineering ligands with varying intraligand distance showed that this inhibition of motility is distance-dependent. Lastly, a Neuregulin-Neuregulin (NN) ligand disrupts the crosstalk between MET and Her-3 receptor to reduce hepatocyte growth factor (HGF)-induced cell motility. Based on these results, a mass spectrometry (MS)-based phosphoproteomics approach to quantitatively map cellular signaling events mediated by Her-3 downstream of Met proceeded. The thesis concludes by quantitatively exploring the relationship between the measured signals of phosphorylated proteins and the cellular migration phenotypes, seeking novel potential therapeutic targets. Overall, this work illustrates the complexity of the EGFR signaling pathway and the need for new paradigms to target its signaling pathway to advance our knowledge in developing new therapeutic approaches. / by Edgar Ivan Sanchez Palacios. / Ph. D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Biological Engineering, 2009. / Cataloged from PDF version of thesis. / Includes bibliographical references (p. 153-163). / Cells are constantly bombarded with stimuli that they must sense, process, and interpret to make decisions. This capability is provided by interconnected signaling pathways. Many of the components and interactions within pathways have been identified, and it is becoming clear that the precise dynamics they generate are necessary for proper system function. However, our understanding of how pathways are interconnected to drive decisions is limited. We must overcoming this limitation to develop interventions that can fine tune a cell decision by modulating specific features of its constituent pathway's dynamics. How can we quantatively map a whole cell decision process? Answering this question requires addressing challenges at three scales: the detailed biochemistry of protein-protein interactions, the complex, interlocked feedback loops of transcriptionally regulated signaling pathways, and the multiple mechanisms of connection that link distinct pathways together into a full cell decision process. In this thesis, we address challenges at each level. We develop new computational approaches for identifying the interactions driving dynamics in protein-protein networks. Applied to the cyanobacterial clock, these approaches identify two coupled motifs that together provide independent control over oscillation phase and period. Using the p53 pathway as a model transcriptional network, we experimentally isolate and characterize dynamics from a core feedback loop in individual cells. A quantitative model of this signaling network predicts and rationalizes the distinct effects on dynamics of additional feedback loops and small molecule inhibitors. Finally, we demonstrated the feasibility of combining individual pathway models to map a whole cell decision: cell cycle arrest elicited by the mammalian DNA damage response. By coupling modeling and experiments, we used this combined perspective to uncover some new biology. We found that multiple arrest mechanisms must work together in a proper cell cycle arrest, and identified a new role for p21 in preventing G2 arrest, paradoxically through its action on G1 cyclins. This thesis demonstrates that we can quantitatively map the logic of cellular decisions, affording new insight and revealing points of control. / by Jared E. Toettcher. / Ph.D.

Thesis (S.M.)--Massachusetts Institute of Technology, Dept. of Biological Engineering, 2013. / Cataloged from PDF version of thesis. / Includes bibliographical references (p. 68-76). / The intestinal crypt is a highly dynamic system, as the entire epithelium is constantly turned over and renewed by the proliferative stem cells located at the bottom of the crypt. While this system is crucial for nutrient absorption, any derangements in the proliferative cells can quickly lead to cancer. In this thesis, we sought to better understand the behavioral and genetic characteristics of the different cell types along the intestinal crypt, in a mouse model. We first attempted to quantify the migration velocities and proliferative rates at a single-cell level using 5-ethynol-2'-deoxyuridine (EdU) pulse-chase labeling. While we observed marked differences in the proliferation rates between the absorptive and secretory lineages (the former is faster), our experimental setup was unable to resolve any possible differences in migratory behaviors. We then turned our attention to quantifying the spatial expression patterns of selected transcripts in the intestinal crypt, both in wildtype and in an inducible APC-loss cancer model. We were able to show that the transcript expression profiles of key differentiation and proliferation markers (Creb313, Gob5, Ki-67, cMyc), as well as the ephrin signaling (EphB2 and EphrinBl) were significantly altered in the early stages (7 days) of adenoma formation. A more detailed analysis also separated this derangement in terms of extrinsic factors (e.g. altered cell composition along the crypt) and intrinsic factors (e.g. inherent change of cellular expression profile after APC loss). What is particularly interesting is that even differentiated cells in adenomas can exhibit such derangements. Our hypothesis to explain this observation is that these differentiated cells actually come from transformed stem cells. To test this hypothesis, we needed to verify that these cells indeed have excision at the Apc locus. After several approaches involving both single-molecule fluorescent in situ hybridization and quantitative polymerase chain reaction (PCR), the results were ultimately inconclusive. However, we propose several additional approaches that this hypothesis can be verified, and if verified, some biologically significant questions that can be addressed regarding the early dynamics of stem cells in intestinal cancer development. / by Kevin Su Yau Kung. / S.M.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Biological Engineering, 2013. / Cataloged from PDF version of thesis. / Includes bibliographical references (p. 215-232). / Receptor tyrosine kinases (RTKs) are critical regulators of cellular homeostasis in multicellular organisms. They influence cell proliferation, migration, differentiation, and transcriptional activation, among other processes, and are therefore also relevant to cancer biology. Upon interaction with cognate ligand, RTKs initiate signaling cascades dependent in part on the phosphorylation of proteins. From a computational perspective, this thesis has studied methods for quantifying relationships between measured signals (using Bayesian network inference, correlation, and mutual information-based methods), and between signals and cellular phenotypes (using linear regression, partial least squares regression, and feature selection methods). From a biological perspective, this thesis has studied signaling between RTKs, signaling and cell migration downstream of RTKs in epithelial versus mesenchymal cell states, and comparative signaling across six RTKs. In the latter case, the results show that the six RTKs cluster into three classes based on their inferred signaling networks. Using publicly available transcriptional and pharmacological profiling data from hundreds of cancer cell lines, it was determined that expression of same-class RTK genes or their cognate ligands can correlate with insensitivity to drugs targeting other RTKs in that class. This suggests that resistance to RTK-targeted therapies in cancer may emerge in part because same-class RTKs can compensate for the reduced signaling of the inhibited receptor. The thesis concludes by quantitatively exploring the features of experimental data that improve model accuracy. / by Joel Patrick Wagner. / Ph.D.

Thesis: Ph. D., Massachusetts Institute of Technology, Department of Biological Engineering, 2016. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 129-141). / While technological advances have enabled measurements of thousands of molecules simultaneously, the data from each technology can only show a single-view of biological processes. Capturing a complete picture of these processes requires integrating data of different types, including clinical data, genomics, transcriptomics, proteomics and metabolomics. Here, we have demonstrated novel computational approaches for integrating a variety of biological data and used these methods to study Huntington's disease (HD). First, we established a computational approach for combining transcriptomics with qualitative, ordinal clinical information. Such data are available for a variety of diseases, but are rarely used in conjunction with molecular data. We adapted an ordinal regression model to analyze gene expression data from HD brains in conjunction with their grade of neuronal loss. This approach identified the SGPLl-encoded enzyme (SPL) as a potential therapeutic target for HD. Continuing our data-driven approach, we discovered the dysregulation of pathways associated with SPL and inferred molecular mechanisms by which SPL inhibition exerts protective effects. Then, we demonstrated a novel network-based, machine-learning algorithm for integrative analysis of untargeted metabolomic data. Metabolites are small molecules whose levels directly show cellular phenotypes. Despite their potential, the integrative analysis of metabolomic data has been limited because of challenges in metabolite identification. To address these challenges, we have developed a pioneering method for interpreting the large-scale metabolomic data in the context of other molecules such as proteins. We used our method to infer novel dysregulated pathways in a model of HD and experimentally verified our predictions. These two methods are extremely general and can be applied to a variety of diseases. As the costs of generating high-throughput data decrease, we anticipate that our approaches will have growing relevance to the discovery of therapeutic strategies for precision medicine. / by Leila Pirhaji. / Ph. D.