Forest M. White, Ph.D. - US grants

DESCRIPTION (provided by applicant): In this two phase (R21/R33) project we will develop methods to elucidate cellular signaling mechanisms underlying a defect in central tolerance in the NOD mouse, a model for human Type 1 diabetes. We will compare signal response to TCP stimulation in thymocytes derived from FTOCs on the BDC2.5/NOD and BDC2.5/H-2g7 genetic background. Our investigation will focus on several signal trasduction pathways implucated in negative selection. However, we will also monitor proteome-wide protein phosphorylation events, as it is possible that the mechanism underlying defective clonal deletion will lie in an alternate pathway. In the R21 phase of the project we will develop methodologies which will enable analysis of specific signal transduction pathways and global protein phosphorylation in thymocytes extracted from FTOCs (fetal thymic organ cultures), a model system for the thymus. To identify Akt substrates we will immunoprecipitate with an Akt phosphorylation motifspecific antibody, enzymatically digest immunoprecipitated proteins, and enrich phosphorylated peptides with an Fe3+- charged immobilized metal ion affinity chromatography (IMAC) column prior to LC/MS/MS analysis for identification of specific phosphorylation sites. A similar strategy will be employed for tyrosine phosphorylated proteins, although a panspecific anti-phosphotyrosine antibody will be used for immunoprecipitation. In the R21 phase of the project we will also develop a method enabling identification of protein phosphorylation on a global scale, by extracting proteins from thymocytes, prefractionating the sample at the protein or peptide level, and enriching peptides on the IMAC column prior to LC/MS/MS analysis. In the R33 phase will we apply each of these methods to the analysis of signal transduction in thymocytes extracted from FTOCs following peptide stimulation of the TCR. We will sample the system at specific time points following peptide stimulation and measure relative quantification of protein phosphorylation across the time course. In addition to Akt kinase and phospho-tyrosine mediated signaling, in the R33 phase will investigate phosphorylation state of 14-3-3 binding proteins. Several regulators of negative selection (i.e. HDAC7 and Nur 77) are sequestered to the cytosol following phosphorylation and binding to 14-3-3. By comparing signal transduction pathways responding to peptide stimulation in thymocytes derived from BDC2.5/NOD and BDC2.5/H-2g7 we hope to identify signaling events regulating negative selection of self-reactive thymocytes, with the specific aim of determining the mechanism underlying defective central tolerance in the NOD mouse.

Proteomics The proteomics core consists of specialized instrumentation in the lab of Prof. Forest White dedicated to CDP projects. The activities of this core include the development and application of new methods for phospho mass spectrometry and are described in detail in D2.2 and 05.3. The core also provides training free of charge to academics interested in applying methods developed by Forest in their own laboratories. There is no question that CDP's investment in Proteomics has been extremely cost effective, stdoctoral fellows in the summer course;he is also active in CDP retreats and conferences.

DESCRIPTION (provided by applicant): Overexpression and mutation of epidermal growth factor receptor (EGFR) and EGFR family members leads to dysregulated signal transduction and has been correlated with increased risk for cancer and poor prognosis for cancer patients due to development of more aggressive cancers (i.e. higher proliferation and metastasis rates). Here we propose to develop an improved mechanistic model of the EGFR signaling network, from which we will be able to identify key nodes in the signaling network which regulate downstream biological response to activated ErbB receptor tyrosine kinases. In this five-year project we will investigate, model, and manipulate the EGFR signaling network to develop an improved mechanistic understanding of cellular signal transduction. In the first phase, we will apply mass spectrometry to quantify temporal phosphorylation profiles for hundreds of phosphorylation sites downstream of EGFR, under a variety of stimulation conditions. In order to link this signaling data to biological outcome, we will acquire phenotypic (migration, proliferation, apoptosis) data for each condition. In the second phase of the project, we will implement a variety of bioinformatic algorithms (hierarchical clustering, SOMs, PLSR) to characterize the data gathered in the first phase of the project. For instance, hierarchical clustering and self-organizing maps will be used to identify co-regulated phosphorylation sites which may function as dynamic modules within the EGFR signaling network. Identification of module components will facilitate assignment of potential biological function to poorly characterized proteins. PLSR will be used to correlate quantitative phosphorylation profiles with downstream biological response data. The result of this method is a functional relationship between the signaling metrics (phosphorylation sites) and biological outcomes (proliferation, migration, and apoptosis);predictions which will be tested experimentally. In this second phase of the project we will construct a mechanistic model of the EGFR signaling network which may then be used to predict behavior of the system. In the third phase of the project, we will attempt to validate model predictions by measuring the response to biological manipulation of the EGFR signaling network. Perturbations may include disrupting the function of various components in the network with RNA interference (RNAi) or small molecule kinase inhibitors (where available), or overexpressing proteins of interest through stable transfection. The final product of this research project will be a more comprehensive and well calibrated mechanistic model of the ErbB signaling network which will have a profound impact on our understanding of oncogenic signaling networks. PUBLIC HEALTH RELEVANCE: Overexpression and mutation of epidermal growth factor receptor (EGFR) and EGFR family members have been implicated in many different tumor types, yet our understanding of these signaling networks is still very incomplete. Here we propose to use cutting-edge analysis and modeling tools to develop a more comprehensive mechanistic understanding of these signaling networks and their linkage to biological response. We will use these improved models to predict biological outcome to novel therapeutic interventions, with the goal of establishing new paradigms for cancer treatment.

The overall goal of the Mitogenic Signaling Networks project is the development of high level statistical and specific physico-chemical models that describe key features of mitogenic signaling networks activated by ErbB receptors and by oncogenic K-ras. Over the past 4 years we have made significant progress in developing models of ErbB family mitogenic signaling networks in a variety of cell types, including statistical and kinetic models describing the effects of increased expression of various ErbB family members. Over the next five years we will extend these models to include mitogenic signaling networks resulting from mutant isoforms of EGFR and K-Ras that are directly associated with poor prognosis in human cancers of the central nervous and respiratory systems. Models will be developed and tested at a variety of scales, including in vitro cell culture systems, murine xenografts, and mouse cancer models. In addition, due to the success of a pilot project funded from our current ICBP, we will extend these models to integrate transcriptional regulatory networks, providing a more global, quantitative model of cellular regulation in response to oncogenic mutation. Since therapeutic resistance is one of the hallmarks of lung and brain tumors driven by mutant EGFR and mutant Ras, in the next phase of this project we will quantify and model signaling and transcriptional network alterations resulting from treatment with a variety of therapeutics, including classical chemotherapeutics, targeted therapeutics, and radiation. The goal of this project is to understand adaptation mechanisms used by tumor cells in developing therapeutic resistance in order to target these adaptive mechanisms to revert resistance. Quantitative models of mitogenic signaling network responses to therapeutics will be applied to human tumors to test their ability to predict responsiveness of human tumors to selected chemotherapeutic agents. This project will facilitate the integration of mitogenic signaling network models with DNA damage response models developed in Project 2, leading to more integrated models of cellular regulatory networks.

DESCRIPTION (provided by applicant): Cell surface receptors linked to tyrosine kinases control a host of important cellular activities, including proliferation, differentiation, and motility. Disregulated tyrosine kinase signaling is a common feature of many human cancers, thus tyrosine kinases and their downstream effectors are targets for the development of new drugs for the treatment of cancer. In order to take full advantage of such promising new therapies, however, we need an understanding of how tyrosine kinase signaling networks process information on a systems level. While considerable progress has been made in developing quantitative models describing tyrosine kinase signaling networks, these efforts are severely hampered by a lack of quantitative information on how changes in tyrosine phosphorylation are coupled to their downstream effectors containing modular phosphotyrosine binding domains. The goals of this collaborative project are to take advantage of new experimental approaches to address this gap in knowledge directly. Specifically, we will use SH2 profiling, a phosphoproteomic approach that is highly complementary to mass spectrometry-based methods, to quantify dynamic changes in binding sites for specific effector proteins upon receptor tyrosine kinase activation. Responses to different receptors and in different cell types will be compared, allowing systems-level behavior to be correlated with biological outputs. We will also use single-molecule imaging methods to monitor the coupling of specific effectors to receptors in the intracellular environment. These studies will afford unprecedented insight into the interaction dynamics of receptor signaling complexes that will enable much more powerful and accurate models of tyrosine kinase signaling. RELEVANCE (See instructions): Signaling from receptors with tyrosine kinase activity plays an important role in a number of human diseases, in particular cancer. Quantitative computer-based models that accurately describe the signaling mechanism used by these receptors will be very useful in designing new therapies for cancer and in deciding which patients will benefit most from those therapies (individualized medicine). The proposed studies use innovative experimental approaches to reveal new mechanistic insights necessary for building more powerful and accurate models.

OVERALL ? SUMMARY The selection of relevant therapeutic agents with optimal pharmacokinetic and pharmacodynamic properties to adequately suppress the intended target across the entire target cell population will be central to the success of genomics-guided precision medicine strategies. Optimal drug therapy for brain tumors is especially challenging due to multiple physical barriers within the vasculature and tumor microenvironment that can result in highly heterogeneous drug delivery. This results in a significant fraction of tumor cells being exposed to sub- therapeutic drug levels that limit the efficacy of therapy and may lead to compensatory cell signaling and emergence of drug resistance. Thus, a central tenet of this proposal is that failure to understand limitations in the physical delivery and distribution of novel therapeutics into brain tumors is a major reason for the collective failure to extend the exciting treatment advances and survival gains realized in peripheral malignancies to the treatment of brain tumors. In this PS-OC, we will focus on understanding physical factors that influence heterogeneous drug distribution and the resulting biology in a highly integrated analysis of patient and animal tumor models using 3-dimensional MR imaging, stimulated Raman scattering (SRS) microscopy, matrix assisted laser desorption/ionization mass spectrometry imaging (MALDI-MSI), immunohistochemistry (IHC), phosphoproteomics, proximity ligation assays (PLA), and RNAseq. Integration of these data sets across a series of drugs evaluated in multiple tumor models will elaborate critical factors that modulate distribution of these drugs and provide the platform for construction of a multi-scale model that could be used to select a targeted therapeutic with an optimal predicted drug distribution based on MRI features of an individual tumor. In this context, we will directly meet the goal of the Physical Sciences in Oncology Program to integrate physical sciences and cancer research perspectives and approaches to address a complex and challenging question in cancer research. Specifically paraphrased from PAR-14-49, we will address: Physical Dynamics of Cancer: How do physical properties and forces within tumors, disseminating cells, and sites of colonization and metastasis contribute to therapeutic delivery and efficacy? How do these factors affect cancer progression and evolution of therapeutic resistance? Spatio-Temporal Organization and Information Transfer in Cancer: Can the evolutionary dynamics of therapeutic resistance be examined in the context of dynamic spatio-temporal environments to better define mechanisms of progression and resistance and rationally design therapeutic strategies?

Summary Project 2? Defining the relationship between tumor composition, spatial heterogeneity, drug delivery, and drug efficacy The ultimate goal of this project is to determine the physical factors regulating therapeutic distribution and therapeutic efficacy in brain tumors. To this end, we have developed a highly innovative integrated strategy to quantitatively map therapeutic distribution with spatially registered characterization of the tumor architecture and therapeutic efficacy, all within a given tumor specimen. Specifically, in this approach we will combine MALDI-MSI to quantify drug distribution, stimulated Raman scattering imaging for label free analysis of tumor architecture with optical imaging resolution, immunohistochemistry to determine the tumor cell state and target distribution, proximity ligation assays for cellular spatial resolution of signaling response to therapy, and laser- capture microdissection RNASeq to quantify spatially resolved transcriptional response to therapy. All of these approaches will be performed in serial sections from individual tumors, thereby enabling the integration of spatially registered data. Together with mass spectrometry based phosphoproteomics and RNASeq analysis to quantify the dynamic signaling and transcriptional network response to a spectrum of defined drug concentrations in additional tumor specimens, the data generated in this project will (1) map spatially heterogeneous drug distribution and drug efficacy and (2) enable the computational modeling of the physical factors governing distribution and regulating the cellular and molecular response to different local drug concentrations. The novel integration of cutting-edge imaging and systems biology approaches applied to different sections of the same tumor, combined with computational models to identify the physical factors governing drug distribution and tumor cell response, will provide unprecedented insight into the complex dynamic behavior of tumor cells in vivo in response to spatially heterogeneous levels of therapeutics.

Education and Outreach - Summary The scientific goal of the MIT/Mayo PS-OC is to define the physical and molecular characteristics of brain tumors that affect therapeutic delivery and efficacy, with the ultimate goal of generating a predictive model that improves therapeutic selection for patients entering the clinic. Using this theme as a platform, we will engage physical scientists, biologists and oncologists at all levels of training to cultivate improved cross-disciplinary fluency. Ultimately, these efforts will have an enduring impact on the field by educating future scientists and oncologists and by fostering lasting trans-disciplinary collaborations between established scientists. We will maximize the education and outreach impact of the center through a multi-tiered strategy to promote experiential training and communication across a range of educational and career levels: High school and undergraduate students: We will tap into and expand existing internship and research opportunity programs at several of our sites to engage both teachers and students in the exciting trans- disciplinary studies being performed in the MIT/Mayo PS-OC. The central goal for these activities is to enthuse participants about potential STEM careers and provide real-world experiences at the interface of physical sciences and oncology. Graduate student and post-doctoral training: We will establish a cross-training program that will provide funding for visiting scientists within the MIT/Mayo PS-OC or within the larger PS-ON. The vision of this program is to enable cross-disciplinary training: physical scientists into biology/clinical settings, and biologists/oncologists into physical science labs. By providing the funding for weeks-to-months of interaction time for multiple visiting scientists per year, this program should provide solid experiential training and immersion in the corresponding field. Annual symposia and visiting faculty series: An annual conference hosted by the MIT/Mayo PS-OC will rotate among the sites involved in our center. A key goal of these conferences is to establish collaborative interdisciplinary opportunities with our PS-OC and within the larger PS-ON. Targeted selection of speakers will maximize the potential for physical science ? oncology collaborations with the host institution and across the PS-ON. A similar strategy will be employed in developing a rotating visiting faculty seminar series. Web-site and public engagement: To broadcast all of these opportunities and to educate the public with regard to the physical science of brain tumors, we will establish a center website. This website will also serve as the hub for accessing data, publications, and computational models generated by our PS-OC. Each of these outreach components is described in more detail below.

ADMINISTRATIVE CORE ? PROJECT SUMMARY The overarching framework for this MIT/Mayo PS-OC is to understand the physical interactions between drugs, tumor tissue and the microenvironment that influence drug distribution and how potentially heterogeneous drug distribution influences tumor cell signaling and therapy resistance emergence (drug efficacy). The PS-OC will require a strong administrative unit to support and coordinate Center activities across all sites, projects and cores; including day-to-day administrative and financial support, planning and evaluation, ongoing communications, pilot and trans-network projects, CAC activities, participation in PS-ON Steering Committee and other initiatives developed over the life of the project. The MIT/Mayo PS-OC will be administratively based at MIT's Koch Institute of Integrative Cancer Research, an NCI-designated Cancer Center that was established in 1974. As demonstrated with past center grants based at the KI, this will allow us to leverage an existing, well-organized administrative infrastructure that will provide support for all business aspects related to the Center. Primary responsibility for scientific project management and managing the workflow of bio-specimens and related high-content data between the two Research Projects and two Shared Resource Cores will be based at the Mayo Clinic. Jointly the members of both teams will be referred to as the Administrative Unit (AU).

Project 5: Project Summary/Abstract PAHs and NDMA are common environmental pollutants that are known to be carcinogenic and are found in high quantities at superfund sites, including the Mystic River Watershed and the former Loring Air Force Base in Maine. Potential adverse health effects of these compounds are concerning to the people in these communities, as NDMA has been detected in well water in Wilmington MA, and the Maine Department of Human Services established a fish advisory stating that weekly consumption of fish from water contaminated by the Loring Air Force Base will lead to an increased risk of cancer. Unfortunately, beyond genotoxicity, the mechanisms underlying potential adverse health effects associated with either acute exposure or chronic, low-dose exposures to these compounds are poorly characterized; yet it is known that PAHs, for instance, have widespread effects on a variety of different cell types and tissues. To determine the systemic, molecular network and cellular effects of exposure to these compounds in this Project we will utilize a systems toxicology approach comprising cutting-edge mass spectrometry for protein phosphorylation profiling, next-gen sequencing for transcript expression profiling, and computational modeling to integrate molecular network data with cell phenotypic data. In collaboration with Projects 3 and 4, we will assess the effects of acute and chronic exposure on the lungs and liver of infant and juvenile mice, connecting molecular network effects with DNA mutation signatures and downstream biological effects while assessing genetic susceptibility. As Projects 1 and 2 define the concentrations and compositions of N-Nitrosodimethylamine (NDMA) and PAHs at these Superfund sites, we will perform in vitro and in vivo studies to assess the combined effects arising from these real-world mixtures, assessing the additivity and synergy of these mixtures compared to the individual compounds. This innovative, integrative strategy will provide new information regarding the health risks and mechanisms underlying exposure to the chemical contaminants present at these sites. Moreover, integrating this information into a predictive quantitative computational model that couples exposure to network response and resulting phenotype will (a) define biomarker signatures of exposure that can be used as an initial starting point for an eventual blood test for exposure signatures, (b) define network nodes governing sensitivity to exposure and therefore potential therapeutic intervention points to abrogate adverse health responses to exposure. Together, the results of this project will not only help to define the health risks for communities at risk, but may also provide potential therapeutic strategies to minimize adverse outcomes from exposures at these sites. These deliverables will have direct relevance to SRP stakeholders, including the EPA and the Massachusetts Department of Public Health.

SUMMARY Over the past decade the accumulation of large-scale systems level data sets has occurred at an accelerating pace. Unfortunately, to date this massive accumulation of biological and medical information has rarely translated into truly efficacious therapies that dramatically alter the course of disease. Clearly new informatics approaches are needed that will enable the identification of transformative therapeutics. The central goal of this proposal is to develop an experimental-theoretical approach that defines, with high accuracy, the altered protein network structures present in each cancer malignancy. We propose to integrate quantitative mass spectrometry- based protein and protein phosphorylation measurements with surprisal analysis, a thermodynamic-based information theory approach, to resolve altered protein network structure in each malignancy. An altered network in each patients' tumor may comprise several distinct, sometimes rewired, protein subnetworks that drive the molecular imbalance in cancer tissue. Identification of unbalanced subnetworks will highlight molecular nodes that will be targeted in each patient to either restore the basal, non-transformed state or to decrease tumor cell viability. To demonstrate the ability of this approach to define unbalanced subnetworks and their associated therapeutic targets, the proposal is divided into three phases with increasing complexity and physiological relevance. In the first phase, RTK networks in breast cancer cell lines representing different subtypes will be stimulated with natural ligands to induce well characterized unbalanced processes to validate the ability of surprisal analysis to identify these networks. In the second phase, unbalanced processes present in the basal, unstimulated state of each cell line will be defined. Therapeutic targeting of these processes, alone or in combination, at high and low dose, will be performed to assess the effect of complete vs. incomplete inhibition. Unbalanced processes mediating the development of therapeutic resistance during long-term low- dose treatment will be quantified at various time points to predict combination therapies to abrogate resistance. Finally, surprisal analysis will be used to identify unbalanced processes associated with chemotherapeutic resistance in vivo in triple negative breast cancer patient derived xenograft tumors. Nodes in these imbalanced networks will be targeted to decrease tumor viability. Combination with chemotherapy may further sensitize tumor cells to treatment. Through these efforts we aim to demonstrate the ability of this combined proteomic- surprisal analysis strategy to rationally design, with high-precision, patient-specific drug cocktails that prevent drug resistance development.

OVERALL ? SUMMARY The selection of relevant therapeutic agents with optimal pharmacokinetic and pharmacodynamic properties to adequately suppress the intended target across the entire target cell population will be central to the success of genomics-guided precision medicine strategies. Optimal drug therapy for brain tumors is especially challenging due to multiple physical barriers within the vasculature and tumor microenvironment that can result in highly heterogeneous drug delivery. This results in a significant fraction of tumor cells being exposed to sub- therapeutic drug levels that limit the efficacy of therapy and may lead to compensatory cell signaling and emergence of drug resistance. Thus, a central tenet of this proposal is that failure to understand limitations in the physical delivery and distribution of novel therapeutics into brain tumors is a major reason for the collective failure to extend the exciting treatment advances and survival gains realized in peripheral malignancies to the treatment of brain tumors. In this PS-OC, we will focus on understanding physical factors that influence heterogeneous drug distribution and the resulting biology in a highly integrated analysis of patient and animal tumor models using 3-dimensional MR imaging, stimulated Raman scattering (SRS) microscopy, matrix assisted laser desorption/ionization mass spectrometry imaging (MALDI-MSI), immunohistochemistry (IHC), phosphoproteomics, proximity ligation assays (PLA), and RNAseq. Integration of these data sets across a series of drugs evaluated in multiple tumor models will elaborate critical factors that modulate distribution of these drugs and provide the platform for construction of a multi-scale model that could be used to select a targeted therapeutic with an optimal predicted drug distribution based on MRI features of an individual tumor. In this context, we will directly meet the goal of the Physical Sciences in Oncology Program to integrate physical sciences and cancer research perspectives and approaches to address a complex and challenging question in cancer research. Specifically paraphrased from PAR-14-49, we will address: Physical Dynamics of Cancer: How do physical properties and forces within tumors, disseminating cells, and sites of colonization and metastasis contribute to therapeutic delivery and efficacy? How do these factors affect cancer progression and evolution of therapeutic resistance? Spatio-Temporal Organization and Information Transfer in Cancer: Can the evolutionary dynamics of therapeutic resistance be examined in the context of dynamic spatio-temporal environments to better define mechanisms of progression and resistance and rationally design therapeutic strategies?

Genomics-guided precision medicine promises to identify the key therapeutic target(s) in an individual patient to enable selection of the most efficacious therapeutic strategy. Central to the success of this strategy needs to be the selection of relevant therapeutic agents with optimal pharmacokinetic and pharmacodynamic properties to adequately suppress the intended target across the entire target cell population. While relevant for all cancers, the selection of appropriate pharmacotherapies is especially challenging in brain tumors, in which the blood-brain barrier in normal and diseased regions can significantly limit drug distribution and efficacy for these tumors. In fact, over 95% of FDA-approved drugs have limited accumulation in the brain, and current predictive algorithms for drug distribution into the brain, based on physico-chemical features of the therapeutic agent, are poorly predictive. In the MIT/Mayo PS-OC, we will develop a platform for modeling drug distribution in brain tumors across scales from organism and tissue down to sub-cellular distribution and signaling and transcript network effect to support magnetic resonance imaging (MRI)-based modeling to enable clinical translation. Integrated with a genomics-guided delineation of therapeutic vulnerabilities, the proposed multi-scale model of drug distribution and efficacy could be used to select a targeted therapeutic with an optimal predicted drug distribution based on MRI features of an individual tumor. A key principle in oncology is that cure is only possible if a potentially curative treatment effectively targets 100% of the tumor cell population. The invasive nature of many brain tumors, with isolated tumor cells invading into regions of normal brain, has made these tumors especially challenging to treat, and despite exciting advances in neurosurgery, radiation therapy, cancer genomics (target identification), and cancer pharmacology (targeted therapeutics), the prognosis for most patients with primary or metastatic brain tumors has not significantly changed over the course of several decades. One of the central tenets of this proposal is that failure to understand limitations in physical delivery and distribution of novel therapeutics into brain tumors is a major reason for this lack of progress. In most brain tumors, the integrity of the vasculature and associated BBB is heterogeneous and critically limits drug delivery to at least some parts of the tumor. Beyond vasculature and the BBB, other physical features regulating therapeutic delivery into tumors are poorly understood, and all of these factors ultimately result in a spatially heterogeneous range of therapeutic drug exposure across a tumor cell population. Further, the dynamic molecular and cellular responses in a heterogeneous tumor to temporally regulated and spatially heterogeneous molecularly targeted therapeutics are poorly understood. Also unknown is the extent that optimizing size, affinity, and/or chemical properties of the therapeutic agent may overcome these physical limitations. Thus, there is a huge unmet medical need to improve our understanding of these physical factors influencing drug distribution and to use this knowledge to develop more effective therapeutic strategies for these devastating tumors. In this PSOC, we will focus on understanding physical factors that influence heterogeneous drug distribution and the resulting biology in a highly integrated analysis of patient and animal tumor models using 3-dimensional MR imaging, stimulated Raman scattering (SRS) microscopy, matrix assisted laser desorption/ionization mass spectrometry imaging (MALDI-MSI), immunohistochemistry (IHC), phosphoproteomics, proximity ligation assays (PLA), and RNAseq. Integration of these data sets across a series of drugs evaluated in multiple tumor models will elaborate critical factors that modulate drug distribution and provide a platform for construction of the planned multi-scale model.

Project Summary The purpose of this study is for a team of materials scientists, biomedical engineers, analytical chemists, and neuroscientists at MIT to develop a micro-invasive implantable device for monitoring the biochemical composition of distinct brain regions. This analytical tool for sampling neurochemicals in brain interstitial fluid (ISF) promises to provide valuable insight into the dynamics of neural circuits in physiological and pathological states. We will apply this tool to study the role of neuropeptides in substance use disorder (SUD). The dynorphin family of neuropeptides has long been implicated in addiction, but no current analysis tool has been able to investigate the long-term spatiotemporal dynamics of these neurochemicals in vivo. Our goal is to demonstrate the efficacy of our sampling platform in measuring neuropeptide expression dynamically in a rodent model of SUD. This will lend greater insight into the biochemical basis of addiction and withdrawal, but perhaps more importantly establish our technology as an effective technique for understanding the onset and progression of neural diseases. Our specific goals are summarized as follows: 1) Design a minimally invasive and implantable device for sampling ISF chronically in vivo. The device will consist of a nanofluidic pump (nanopump) coupled to micro- scale probes (microprobes), with fluid flow characteristics optimized in vitro prior to translation to a stand-alone in vivo device. 2) Optimize the storage and processing of small volumes of sampled ISF, withdrawn via nanopump, for analysis via liquid chromatography-tandem mass spectrometry (LC-MS/MS). 3) Determine the detection limits for the dynorphin neuropeptide family in ISF in vitro prior to detection of these neurochemicals in in vivo samples at physiological and pathological concentrations. 4) Perform short-term monitoring of dynorphin at baseline and in acute stress to demonstrate the efficacy of this tool in tracking these large neuropeptides in real-time. 5) Track the dynorphin family of neuropeptides in a rodent model of cocaine SUD, lending greater insight into the biochemical basis of substance withdrawal and relapse. Our aim is to demonstrate the failsafe function of this sampling platform in vivo and establish its ability to monitor neuropeptide dynamics with precise spatiotemporal control. We aim to provide neuroscientists with a new tool for investigating the biochemical basis of neural pathology in well-established animal models, enabling more accurate diagnosis and treatment of neural disorders in humans in the future.

SUMMARY Over the past decade the accumulation of large-scale systems level data sets has occurred at an accelerating pace. Unfortunately, to date this massive accumulation of biological and medical information has rarely translated into truly efficacious therapies that dramatically alter the course of disease. Clearly new informatics approaches are needed that will enable the identification of transformative therapeutics. The central goal of this proposal is to develop an experimental-theoretical approach that defines, with high accuracy, the altered protein network structures present in each cancer malignancy. We propose to integrate quantitative mass spectrometry- based protein and protein phosphorylation measurements with surprisal analysis, a thermodynamic-based information theory approach, to resolve altered protein network structure in each malignancy. An altered network in each patients' tumor may comprise several distinct, sometimes rewired, protein subnetworks that drive the molecular imbalance in cancer tissue. Identification of unbalanced subnetworks will highlight molecular nodes that will be targeted in each patient to either restore the basal, non-transformed state or to decrease tumor cell viability. To demonstrate the ability of this approach to define unbalanced subnetworks and their associated therapeutic targets, the proposal is divided into three phases with increasing complexity and physiological relevance. In the first phase, RTK networks in breast cancer cell lines representing different subtypes will be stimulated with natural ligands to induce well characterized unbalanced processes to validate the ability of surprisal analysis to identify these networks. In the second phase, unbalanced processes present in the basal, unstimulated state of each cell line will be defined. Therapeutic targeting of these processes, alone or in combination, at high and low dose, will be performed to assess the effect of complete vs. incomplete inhibition. Unbalanced processes mediating the development of therapeutic resistance during long-term low- dose treatment will be quantified at various time points to predict combination therapies to abrogate resistance. Finally, surprisal analysis will be used to identify unbalanced processes associated with chemotherapeutic resistance in vivo in triple negative breast cancer patient derived xenograft tumors. Nodes in these imbalanced networks will be targeted to decrease tumor viability. Combination with chemotherapy may further sensitize tumor cells to treatment. Through these efforts we aim to demonstrate the ability of this combined proteomic- surprisal analysis strategy to rationally design, with high-precision, patient-specific drug cocktails that prevent drug resistance development.