Stephen TC Wong, PhD - US grants

DESCRIPTION (provided by applicant): Studies of quantitative image processing for longitudinal neuroimaging followup have been reported for more than two decades. Many algorithms and methods of brain image processing and analysis have been reported, validated and are freely available, but very few have been engineered for practical clinical and research use or integrated into clinical information systems. This has seriously limited the impact that computational algorithms and methods developed by the medical imaging research community have made on clinical neuroimaging practice and neuroimaging based clinical research. We propose to integrate available validated neuroimage analysis tools into an image informatics system AFINITI (Assisted Followup in Neuroimaging of Therapeutic Intervention), designed to enhance diagnostic accuracy and reduce error in the interpretation of neuroimaging followup studies for longitudinal followup of therapeutic intervention in several common conditions, by providing the interpreting and referring physicians with automated tools and augmented information needed for precise, reproducible, quantitative assessment of longitudinal change in brain images. The primary goal is to translate tools and techniques produced by biomedical informatics research into an actual clinical setting where it can be used in diagnosis, therapeutic decision making, treatment response monitoring, clinical education and research in neurological disease. The AFINITI system will be developed, evaluated, and implemented into the clinical workflow and validated in Brigham and Women's Hospital-Harvard Medical School (BWH-HMS) clinical neuroradiology section and the BWH-HMS Longwood MRI Research Center (LMRC). We will package the validated software into a software toolkit and distribute it in an open source environment, compatible with popular medical imaging platforms, such as VTK (Visualization Toolkit), ITK (Insight Toolkit), and 3D-Slicer.The data integration method and database schemas for this neuroimage database will also be made available to public. AFINITI system architecture and components will form a prototype for the development of future clinical imaging followup systems in other imaging centers. This will be a substantial contribution to the public health.

DESCRIPTION (provided by applicant): We propose to develop an integrated computational system to track and analyze neuronal dendrites and spines observed in 3D time-lapse optical microscopy and manage all the data. The proposed work will overcome one barrier in neuroscience research, i.e., lack of computational method to quantitatively analyze large-scale time-lapse image data. Current manual analysis is confined to small datasets and qualitatively interpretation of 3D images. Detailed neurological changes may be missed by manual analysis. Also results of manual analysis are not immediately ready for data management and analysis because of long hours it takes to store manual analysis results to a database. Thus searches for treatment of neurodegenerative conditions and so on may be hindered. Our plan to develop the integrated computational system can increase the throughput of neuronal image analysis in neuroscience. The system will automatically track and analyze neuronal dendrites and spines. Dendrites and spines are two structures of neuronal cell that manifest changes in neurodegenerative conditions. For example, researches have shown that spines change over time, may appear and disappear entirely. The newest technology is to perform time-lapse imaging of the dendrites and spines to obtain their temporal behavior. The integrated computational system can track multiple spines over the time course and extract important features about them. The features include their length, width, volume, etc. All these features are then saved in common file format and can readily be ported into the built-in database for statistical analysis. Therefore, the proposed work enables neuroscientists to conduct large-scale time-lapse study and track neuronal changes at more time points. All can be critical to finding a cure to neurological conditions. Today's microscopy can image neuronal cells in three-dimensions over a period of time. However use of data is limited because there is no computer-based tool to analyze and systematically manage the data. We plan to develop a computed-based tool to increase the efficiency and effectiveness of the data. Public Health Relevance: This project enables researchers to quickly screen hundreds of neuron images to study diseases and identify possible new drugs. Overall, the project has the impact of facilitating therapy development for neurological conditions such as Alzheimer's disease.

[unreadable] DESCRIPTION (provided by applicant): The aim of this proposal is to develop NCELLIQ, Neuron and Cellular Imaging Quantitator, in order to assess quantitatively neurite loss and outgrowth screened by high throughput, automated fluorescent microscopy imaging. Such a tool will be essential in high content screening (HCS) of neuro-based assays. Loss of neuronal projections in Alzheimer's disease (AD) can be modeled in vitro in primary mouse cortical neuron cultures treated with the amyloid beta peptide, which has been shown to be a major cause of neurodegeneration in AD patients. As is the case in vivo, neurite loss precedes neuronal death in this disease model, which can be assessed visually either in live neurons through bright field microscopy, or through immunofluorescence following fixation and staining with neuronal marker class III tubulin beta antibody. Given recent advances in automated microscopy, the later visualization technique could be adapted for use in HCS of chemical libraries in order to identify compounds that can specifically suppress amyloid-induced neurite damage and loss. The hypothesis of this proposal is that the HCS informatics system developed, NCELLIQ, will be an important image processing and analytic tool to help identify possible drugs in treating Alzheimer's disease. Using multi-channel image data obtained from hippocampal neurons, we integrate and develop techniques to screen for potential drug leads in AD treatment. To test the hypothesis, we aim to define the high throughput image processing pipeline of NCELLIQ, develop automated algorithms for neurite centerline extraction and cellular image analysis, implement computational modeling tools, and evaluate the utility of the NCELLIQ with established, novel, and well-defined biology-driven experiments. NCELLIQ provides three key technical contributions. First, NCELLIQ will provide an integrated neural image processing pipeline using advanced computational algorithms to extract image contents of neuron- based screening assay automatically. Second, it will develop an innovative, effective, and fully automatic neurite centerline extraction methods using detector of curvilinear structures and dynamic programming. Third, it will develop mathematical representation of compound vector, as well as an innovative and effective scoring method to allow intuitive comprehension of the HCS results. The success of NCELLIQ will lead to a new class of bioinformatics tools for identifying quality hits in AD drug development and for determining cyptological profile of neurites. Upon the successful completion of the project, we will set up a website to disseminate the NCELLIQ software and sample image datasets. [unreadable] [unreadable]

DESCRIPTION (provided by applicant): High-content screening (HCS) is defined as the integration of sample preparation, automatic microscopic imaging, and bioinformatics tools that permit experimentation and discovery with high-throughput cell images. It has potential to make large-scale cell biology a tractable approach by generating functional information through the automated measurements of the temporal and spatial activities of genes and proteins in living cells. However, there are significant computational challenges, such as accurate segmentation of the large population of cells and classification of cellular phenotypes, in high-content screening, and image informatics has become the rate-limiting factor in realizing its full potential. Therefore, we propose to develop a new generation of computational tools to fill that gap. We emphasize three key technical contributions of G-CELLIQ. First, G-CELLIQ will provide an integrated cell image processing pipeline using advanced computational algorithms to extract contents of RNAi screening images, reducing the time required in processing by manual analysis and the variability in manual analysis. Second, we will develop novel classification-controlled feedback systems to refine cell boundaries and to increase the accuracy of the scoring method that reflect the mixture of different cell phenotypes in the screening. Third, we will develop an innovative and effective scoring method based on the fuzzy set-theoretic approach. The succinct score will allow researchers to easily comprehend the significance of the results and identify the genes of interest. The hypothesis of this application is that the proposed image informatics system, G-CELLIQ (Genomic CELLular Imaging Quantitator), is critical for large scale RNAi genome screening to identify novel effectors of Rho proteins. The Rho family of small GTPases is essential for cell shape changes during normal cell migration and cancer metastasis. The goal of genome-wide RNAi screening is to identify novel effectors of Rho proteins using a cell-based assay for Rho activities. To test our hypothesis, we will evaluate the utility of the G-CELLIQ with a set of well defined, biological-driven experiments. Upon completion of the proposed project, we plan to make this package freely available to biomedical research community through a public website. More importantly, the completion of this screening project will help to answer some critical questions related to cancer metastasis. Such understanding will in turn advance our knowledge in tumor biology and open up the possibility of novel treatments in the future. This project will be a substantial contribution to the public health by understanding Rho family of small GTPases which is of fundamental relevance to developmental and cancer biology. More importantly, the completion of this screening project will help to answer some critical questions related to cancer metastasis. Such understanding will in turn advance our knowledge in tumor biology and open up the possibility of novel treatments in the future.

The overall goal ofthis proposal is to build a multi-scale modeling platform for investigation ofthe breast cancer, with special emphasis on the roles ofthe tumor-initiating cells (TIC). This modeling platform will mainly consist of two closely related components: biological experiments and mathematical computational modeling. For the experiment component, we seek to use newly developed experimental and imaging methodologies to identify, localize, purify and characterize TIC. Further experiments will be designed to discover the spatial localization and movement, and specific changes in gene expression and cellular signaling of breast cancer TIC. Combined functional genomics and data mining strategies will allow us to characterize novel growth regulators. Further, our combined experimental and systems biology approach will allow us to evaluate responses to experimental therapeutics that may inhibit or kill TIC specifically in a manner not possible before. For the mathematical modeling component, we will develop bioinformatics and bio-imaging models to integrate and analyze the data generated from biological experiments, and make use ofthe information obtained from data analysis, biological knowledge to build in silico models to model TIC behavior, cancer cell apoptosis, cell migration, cycle and drug treatment response. Besides providing a basic framework for understanding the mechanism underlying breast cancer stem cell evolution, the models can also give birth to hypotheses or experimental design. More important, these models will allow one to predict the biological state under investigation and predict how the natural process will behave in various circumstances. Iterative feedback between these two components will refine our proposed platform further. The ultimate goal is an integrated modeling platform of breast cancer biology that can mimic in vivo processes faithfully enough to serve as a h5TDOthesis-generation and screening tool, and in the distant future, as a tool for evaluating clinical procedures and their expected outcomes.

A highly effective administrative resource is critically important in the successful establishment of this CSMCaD. This core will be the administrative center of the Methodist-Baylor-UTHSC team. The administrative core resource will consist of the Pl. the Core Pis, a Research Project Coordinator and an Administrative Assistant. The CSMCaD advisory committee will communicate directly with this resource in terms of monitoring progress, providing evaluation and counseling the Pl in issues arising during the administration of the Center. The administrative core resource will coordinate the administration of the CSMCaD. organize the steering and advisory committee meetings, and track milestones and project progress. Retreats, symposia, seminars, and meetings will be coordinated and organized through this resource. Monitoring and reconciliation of the various budgets, and facilitation of the purchase of supplies will be also be provided by this resource. The CSMCaD project team consists of seventeen investigators spread over three institutions. Efficient communication and a high level of interaction will be achieved through the administrative resource which will include the maintenance of an interactive Wiki project web site and annual all-hands meetings. Furthermore, email listings where daily postings on day-to-day activities and information will be provided. The administration core will also have support from TMHRI administration resources. Under the direction of Edward Jones, M.B.A., Vice-President in charge of administration, TMHRI has a fully developed and staffed research administration and support infrastructure. Among its innovations is web-based management of the document flow for the Institutional Review Board and other research administration functions. TMHRI is fully compliant with all NIH Grants Policies and OHRP policies regarding human research subjects. The administration core will ensure synergies and resource sharing of Component 1 & Component 2. The weekly meeting between these two components will take place as usual. The core will also make sure management of component 3 -the Education & Training is going well, see details in Section N5.

Mathematical modeling involves the use of mathematical equations and relationships to represent biological phenomena. Complementary to this type of modeling is the use of computer simulations to represent these modeling approaches in multiple dimensions. These approaches serve two purposes. First, they provide a basic framework for the interrogation and integration of data, often providing insight into the type and quality needed for addressing a hypothesis or experimental design. This feature is especially useful when trying to integrate or analyze the large datasets generally associated with systems biology. Second, and more importantly, these models or simulations should allow one to predict the biological state under investigation and predict how the natural process will behave in various circumstances. These problems center on the understanding of the behavior of biological systems whose function is governed by the spatial and temporal ordering of multiple interacting components at the molecular, cellular, and tissue levels. We will also develop bioinformatics and bioimaging models to integrate and analyze the data generated from Component 1, and make use of the information obtained from data analysis, biological knowledge to build in silico models to model TIC behavior, cancer cell apoptosis, cell migration, cell cycle changes and drug treatment response. The goal of this component is to take advantage of our combined expertise in cell biology and computational modeling to develop coherent experimental protocols and construct biomathematical models for understanding the mechanism underiying breast cancer stem cell evolution, i.e., how one stem cell evolves into breast tumor with various sizes and compositions in cell microenvironment. Our hypothesis is that that TIC behavior during tumor development can be simulated using a robust, multiscale mathematical/computational model of TIC behavior during breast cancer development. Further, that these models can be built to reflect not only the molecular, cellular, and tissue-level dynamics, but also to allow prediction of the response of TIC to experimental therapeutics.

? DESCRIPTION (provided by applicant): There is an unmet need for molecularly targeted therapies for the treatment of non-small cell lung cancer (NSCLC). Taking into account the emerging paradigm that the reprogrammed intratumoral stromal cells contribute to carcinogenesis, we have employed integrated experimental and computational approaches to identify tumor-stroma crosstalk pathways that drive NSCLC progression. To explore paracrine/autocrine crosstalk, we performed RNA deep sequencing analysis of specific cellular myeloid and epithelial compartments isolated from freshly harvested lungs of NSCLC patients, and a genetically engineered mouse model of NSCLC. We compared transcriptomes of intratumoral myeloid cells (monocytic, neutrophils and macrophages) and tumor epithelial cells with their counterparts within matched adjacent non-neoplastic tissue. In this application, we will develop a multi-cellular crosstalk signaling network modeling and visualization software tool (Aim 1) and apply this model to multi-cellular RNA-seq data to identify tumor-stroma crosstalk pathways; genes involved in these signaling mechanisms will be considered potential candidates that mediate NSCLC tumor progression and will undergo rapid validation using in vitro assays (Aim 2). Finally, we will determine the function of selected crosstalk pathways in NSCLC progression and in mediating therapeutic resistance (Aim 3). In summary, this study explores the relatively understudied tumor-stroma crosstalk pathways as a largely untapped source of drug targets and has tremendous potential for the development of novel therapeutic strategies that target tumor-stroma interactions and may complement existing treatments that target cancer cells.

PROJECT SUMMARY Given the complexity of Alzheimer's Disease (AD) pathogenesis and the associated co-morbid conditions, both the ?depth? and the ?width? of currently available drug repurposing solutions need to be improved in order to deliver effective AD therapeutic solutions. The depth of a drug-repurposing project refers to the level of understanding of disease mechanism and drug-target interactions across a wide searching space for the combination of dosage and treatment time. Achieving depth requires a reliable AD model system that comprehensively recapitulates AD pathogenesis in a human brain-like environment, and sophisticated transcriptomic profiles, which can reveal molecular-level changes underlying disease-reversing phenotypes across multiple treatment conditions. The width of a therapy search relies on the efficacy of predicting and validating effects of candidate compounds from an enormous search space. Width can be achieved from novel computational algorithms connecting ?omics changes with phenotypic changes, thus guiding the search with improved knowledge on mechanisms and avoiding exhaustive testing of every available drug. Integrating the systems medicine and drug repositioning expertise of the Wong Lab at the Houston Methodist Research Institute of Houston Methodist Hospital with the Alzheimer's biology expertise of the Kim and Tanzi labs at Massachusetts General Hospital, we propose a SysteMatic Alzheimer's disease drug ReposiTioning (SMART) framework based on bioinformatics-guided phenotype screening. Reformatting a novel three- dimensional human neural stem cell culture model of AD (a.k.a. Alzheimer's in a dish) developed in the Kim and Tanzi labs for high content screening, the Wong lab screened 2,640 known drugs and bioactive compounds and obtained a panel of 38 primary hits that strongly inhibit ?-amyloid-driven p-tau accumulation. We hypothesize that iteratively running relatively small screens with our novel 3D cell model and applying systematic artificial intelligence modeling to the transcriptomic profiles of the screening hits will allow us to: 1) quickly obtain a panel of robust novel drug candidates for AD, and 2) gain an in-depth understanding of disease mechanisms from those repositioned drug candidates, which will subsequently improve the success rate of predicting novel hits. Using the primary 38 hits as a starting point, the SMART computational modules will update the existing NeuriteIQ software package to quantify the image data from high content screening; it will also incorporate publicly available big data transcriptomic profiles to predict candidate compounds inducing similar pathway changes as those original compounds, effectively expanding the search width to tens of thousands of compounds while only requiring functional validation of less than 100 drug candidates. The validated predictions will, in turn, add to the panel of known hits that will launch the next round of computational predictions and experimental validations, efficiently generating candidates for novel AD therapies (Aim 1). SMART's iterative prediction-validation scheme effectively connects more transcriptomic profiles to desirable phenotypic changes. Thus, we will apply systematic image-omics modeling to uncover novel mechanisms driving such phenotypes. For all the validated hits, dose-responses for the phenotype of pTau inhibition will be obtained using the 3D culture model; while the dose-responses for individual genes and pathways will be modeled through public and in-house generated transcriptomic profiles. We will use Partial Least Square Regression models to identify gene modules with matching dose-response curves as the phenotypes, thus allowing us to go beyond the confinement of canonical pathway maps and identify novel functional modules specifically related to phenotypes of interest (Aim 2). Selected compounds derived from the previous two aims will be evaluated in human neurons directly derived from AD patients and in animal models (Aim 3). Success of this work will lead to new AD therapeutic compounds ready for translation into clinical trials, as well as a deeper understanding of the molecular mechanisms of AD pathophysiology. In addition, the SMART framework for drug repositioning will be generalizable to other big data and disease platforms.

ABSTRACT Breast cancer continues to be one of the leading causes of cancer death among women in the United States, despite the advances made in the identification of prognostic and predictive markers for breast cancer treatment. Mammographic reporting is the first step in the screening and diagnosis of breast cancer. Abnormal mammographic findings such as a mass, abnormal calcifications, architectural distortion, and asymmetric density can lead to a cancer diagnosis. The American College of Radiology developed the Breast Imaging Reporting and Data System (BI-RADS) lexicon to standardize mammographic reporting to facilitate biopsy decision-making. However, application of the BI-RADS lexicon has resulted in substantial inter-observer variability, including inappropriate term usage and missing data. This observer variability has lead in part to a considerable variation in the rate of biopsy across the US, with a majority of breast biopsies ultimately found to be benign lesions. Hence, there is the need for a system that can better stratify the risk of cancer and define a more optimum threshold for biopsy. To address this need, we propose to develop an intelligent-augmented risk assessment system for breast cancer management based on multimodality image and clinical information with deep learning and data mining techniques. This study aims to develop a well-defined, novel risk assessment system incorporating multi-modality datasets with a novel predictive model that outputs a probability measure of cancer that is more clinically relevant and informative than the six discrete BI-RADS scores. Using mammographic or breast ultrasound BI- RADS reporting signatures and radiomics features, a predictive model that is more precise and clinically relevant may be developed to target well-characterized and defined specific biopsy patient subgroups rather than a broad heterogeneous biopsy group. Our proposed technique entails a novel strategy using Natural Language Processing to extract pertinent clinical risk factors related to breast cancer from vast amounts of patient charts automatically and integrate them with corresponding image-omics data and radiologist- generated reports. We will extract and quantitate image features from both large amounts of mammography and breast ultrasound images and combine them with the radiology reports and pertinent clinical risk profile and other patient characteristics to generate a risk assessment score to aid radiologists and oncologists in breast cancer risk assessment and biopsy decisions. Such a web-based application tool will be the first breast cancer risk assessment system based on integrative radiomics data augmented by AI methods. The iBRISK tool will enhance engagement between the patient and clinician for making an informed decision on whether or not to biopsy. Our hypothesis is that BI-RADS reports and the imaging metrics contain significant features for the breast cancer risk assessment and biopsy decision-making. By using BI-RADS reports and the imaging metrics, we will be able to develop new metrics to better breast cancer risk assessment. The novelty of the breast cancer risk assessment system is that it will incorporate a new predictive model that deploys deep learning and AI technology to provide a more reliable stratification of the BI-RADS subtypes for breast cancer risk assessment and reduce unnecessary breast biopsies and patients? anxiety.

As treatment outcomes of primary or systemic cancer sites improve, the clinical importance of brain metastasis (BM) is growing. Twenty-four to 45 percent of all cancer patients develop BM, the majority from lung, breast or melanoma primary cancers, but few patients with BM live longer than a year, and BM constitutes 20% of annual cancer deaths. Ironically, recent advancement in chemotherapy has further increased the incidence of BM because most therapeutic agents cannot effectively penetrate the blood-brain barrier (BBB) and tumor cells find the brain as a sanctuary. Therefore, it is of paramount importance to have a deeper understanding of mechanisms that promote BM growth, which could be specifically leveraged to overcome current limitations in therapy. As opposed to the molecular mechanisms involving cancer cell?host interactions shared by multiple cancer types that result in organ specific metastasis, a highly distinct set of structural, anatomic, physiologic and molecular factors regulate metastasis to the brain. Astrocytes, the most common glial cell comprising ~ 50% of all human brain cells, are a well characterized perilesional component of BM and recent discoveries, including ours, provide compelling evidence that molecular crosstalk between astrocytes and cancer cells is integral to BM development. Although seminal findings indicate that interactions with astrocytes occur at both early and late stages of tumor colonization process, our understanding of the reciprocal astrocyte-cancer cell crosstalk is limited. In preliminary studies, we have employed our Cell-Cell Communication Explorer (CCCExplorer), a unique computational modeling tool, in identifying the novel PCDH7-EGFR, IL6-IL6R, and CCL5-CCR5 astrocyte-tumor crosstalk signaling in regulating BM. Based on these observations and in view of the secretory nature of glial cells, we propose here to test the hypothesis that crosstalk with astrocyte-derived secreted factors is critical for tumor cell colonization in the brain. Given that an even more complicated paracrine signaling network may dynamically evolve at different stages of BM development, and the interactions could provide both anti- and pro-metastatic stimuli to cancer cells, we will test our hypothesis through the following aims: 1) to assess therapeutic potential of the PCDH7- EGFR, IL6-IL6R and CCL5-CCR5 paracrine signaling in BM mouse models employing gain and loss of function and pharmacologic approaches in syngeneic mouse and human cancer xenografts; 2) to assess the astrocyte secreted proteins in modifying the function of BBB and microglia/macrophage in early BM; 3) to further characterize the temporally evolved astrocyte-BM cell crosstalks in a cancer type specific fashion. Our study is highly innovative in that (i) this study integrates knowledge and methods from both neuroscience and cancer to identify and characterize pro- and anti-metastatic astrocyte molecular mechanisms, their evolution during disease progression, and their manipulation in order to provide a valuable means of targeting astrocyte-cancer cell interactions. (ii) This study leverages powerful predictive modeling of cell-cell communications (CCCExplorer) to investigate and delineate the complex network of tumor-astrocyte interactions holistically in an unbiased manner. (iii) This study will address whether there is any specific therapeutic window as to which time point during BM might represent the most effective point of modulating and targeting the vicious astrocyte-tumor crosstalk. (iv) Given the strong response of astrocytes to BM during the course of brain colonization, the identification of secreted molecules may represent putative biomarkers of early diagnosis or response to therapy. (v) Data generated in this study would form an extraordinary repository for comparative analyses between different brain disorders to interrogate common and different aspects of astrocyte biology in different scenarios as well as to evaluate the potential new therapeutic strategies such as drug repurposing and combinations. The outcome of our study will provide a paradigm shift in current understanding of the pathology of BM, while achieving a significant impact on future treatments for this devastating disease.

ABSTRACT About 20-40% of breast cancer patients develop metastasis to the bone, years to even decades after surgical removal of primary tumors. Little is known about the biology of the latent, microscopic bone metastases before they outgrow to overt osteolytic macrometastases. This represents a significant gap in our understanding of bone metastasis. Targeting cancer cells that have not fully adapted to the bone microenvironment might provide therapeutic benefit and prevent the occurrence of overt metastases. Bone and bone marrow comprise of several highly distinctive microenvironment niches. Dormant, single disseminated tumor cells (DTCs) reside in the perivascular niche, whereas proliferative, multi-cell bone micrometastases (BMMs) are found in the osteogenic niche that exhibits features of active osteogenesis. Mechanisms through which the transition of different niches occurs to switch fates of metastatic seeds remain elusive. The overall objectives of this project are to investigate the spatiotemporal dynamics, the molecular crosstalk, and the therapeutic targets underlying the interaction between breast cancer cells and different microenvironment niches in bone. We will pursue three specific aims. First, we will dissect the spatiotemporal dynamics of the perivascular and osteogenic niches and the cancer-niche interactions in bone micrometastasis models. We will use high-resolution, whole- bone, multi-photon microscopy and laser-captured microdissection (LCM) followed by transcriptome profiling (LCM-seq) to obtain relative localization and mutual impacts between cancer cells and niche cells in situ. Second, we will integrate transcriptomic and imaging data and develop computational models for discovery of new mechanisms and therapies toward blockade of cancer-niche interactions. Established and new algorithms will be used to uncover the microenvironment molecules, and autocrine and paracrine signaling pathways mediating niche-tumor interactions. Drug-repurposing analyses will be carried out to identify potential therapies that have already been used for other diseases. We will achieve a systematic understanding of early-stage bone colonization and generate testable mechanistic and therapeutic hypotheses. Third, we will validate the discovered mechanisms and predicted drug efficacies in animal models. The Zhang laboratory has adopted and established a series of genetically engineered mouse models and bone metastasis assays, which will be utilized to validate computational predictions generated by computational modeling by the Wong group. Both metastatic burden and frequency/distribution of DTCs and BMMs will be examined as endpoints. This study will unbiasedly profile the molecular process of early stage metastasis progression in the bone from DTCs to BMMs at single-to-few cell resolutions. This knowledge is unprecedented and critical for the ultimate understanding of metastasis latency, a long-standing clinical challenge. The modeling tool developed through this study will likely be applicable to other biological contexts involving highly spatiotemporally specific cancer- niche interaction. The computer-aided drug repurposing will likely lead to fast clinical translation.

Abstract Alzheimer?s disease (AD) affects more than 50 million people worldwide but there is no clear therapeutic option for the patients. For last two decades, AD research has been focusing on a neuron-centric biochemical process that leads to synaptic deficits and neuronal degeneration. However, recent failures in clinical trials clearly demonstrate a gap in knowledge in our current understanding of AD pathogenesis and call for studies that lead to unbiased and holistic understanding of disease pathways in different types of brain cells. This project aims to tackle this important and urgent issue by combining a computational systems biology platform Single-Cell Resolution Brain Interactome (SCRBI) Explorer, 3D human Alzheimer?s-in-a-dish models, and the publicly available multiple-omics AD databases through NIH-funded AMP-AD portal. We will expand the knowledge base of SCRBI Explorer to handle single cell transcriptomic and multiple omics profiles from 3D cell models and human brain tissues, which can detect on multiple layers of neuron-glia and glia-glia crosstalk pathways via ligand-receptor interactions, cytokine/chemokine signaling, intracellular signaling activities, and transcriptional activation. The central hypothesis is that the combined use of multi-cellular systems biology modeling and 3D human AD cellular models will identify AD-specific neuron-glia and glia-glia crosstalk pathways, which would provide novel therapeutic targets for drug repositioning. We will test this hypothesis by pursuing three specific aims: 1) Develop a multi-cellular crosstalk model to uncover altered neuron-glia and glia-glia crosstalk pathways in AD, 2) identify and validate AD-specific neuron-glia and glia-glia crosstalk pathways that are enriched in 3D human AD cellular models and human AD brain cells, and 3) evaluate the therapeutic potential of neuron-glia and glia-glia crosstalk using 3D human neural cell culture models of AD. The potential impact of this proposal is high because the proposed study, if successful, will provide a unique integrated bioinformatics tool to unbiasedly identify neuron-glia and glia-glia crosstalk pathways in AD and even other neurodegenerative diseases. More importantly, it will provide novel therapeutic targets based on altered neuron-glia interaction pathways in AD and open up a new vista for drug repositioning targeting cell-cell interactions in the brain of AD patients.