Abstract:
Solid tumors consist predominantly of malignant cells as well as stromal and immune cells resulting in a complex tumor microenvironment. Here, I will present our recent data on understanding the cross-talk between malignant and immune cells in tumor and adjancet normal regions of cancer. I will also demonstrate the power of spatial transcriptomics to understand the heterogeneity of this cross-talk within tumor and its evolution as cancer progress.

Abstract:
The recombination of T-cell receptor (TCR) loci enables the immune system to respond dynamically to seemingly innumerable antigens, arising from pathogens or mutations in neoplastic cells. Diversity of the TCR repertoire is an important biomarker because the composition of a T-cell population encodes information on the presence and number of antigens recognized by an active immune response. Furthermore, in the context of a tumor, the physical distribution of T cells within the microenvironment can give further information about the status of the ongoing immune response and potential to respond to immunotherapy. However, the location of expanded T-cell clones in tumors is currently unknown.

We have developed an approach to estimating T-cell repertoires with spatial context on the GeoMx digital spatial profiling platform by utilizing targeted probes against the constant and variable regions of alpha, beta, gamma, and delta TCRs. First, we show proof-of-principle data on the sensitivity and specificity of our approach using a series of cell pellet arrays containing defined ratios of embedded cell lines with known TCRs to confirm the viability of this approach. To demonstrate the application of this approach in a biologically relevant context, we have characterized TCR repertoires of melanoma patients primarily from lymph node biopsies via RNA sequencing (RNA-seq), TCR capture-sequencing, and RAGE-seq.

In this experiment, we profiled the matched tumor samples using the GeoMx Cancer Transcriptome Atlas plus additional probes covering the TCR regions of expanded clones. We demonstrate detection of multiple TCR variable regions in a series of melanoma tumors, sampling from multiple regions within the tissues yielding insights not only about the TCR repertoires and arrangement but also deep profiling of the tumor cells and the surrounding immune milieu. Furthermore, by combining TCR expression with endogenous expression of immune and tumor-intrinsic genes, we can infer key facets of immunosurveillance happening in situ.

In summary, we demonstrate TCR profiling capable of spatially resolving the antigen-responsive clones in the context of the tumor, yielding valuable insights for immune-oncology treatments aimed at enhancing anti-tumor T-cell responses.

Abstract:
SMARCA4 (BRG1) is one of two mutually exclusive ATPases of SWItch/Sucrose Non-Fermentable (SWI/SNF) chromatin remodeling complexes and is among the most frequently mutated genes in lung adenocarcinoma. Yet, the functional consequences of its alterations on tumor initiation and progression, the chromatin landscape, and gene expression in lung cancer are unknown. Here, we address these unknown functional consequences using a combination of genetically engineered mouse models, and epigenomic and transcriptomic profiling. By inactivating SMARCA4 in an autochthonous mouse model of lung adenocarcinoma, we show that loss of SMARCA4 sensitizes a distinct population of cells within the lung to malignant transformation and results in highly advanced tumors and increased metastatic incidence. Consistent with this phenotype, SMARCA4-deficient tumors are absent of lung lineage transcription factor activities, similar to a metastatic cell state. We further show that loss of chromatin accessibility upon SMARCA4 inactivation is a direct effect of the inability of SWI/SNF complexes to bind and open chromatin at regulatory regions in the absence of SMARCA4. Finally, by spatially profiling with the GeoMx Digital Spatial Profiler using the mouse Whole Transcriptome Atlas, we identify gene expression programs that characterize highly advanced and metastatic SMARCA4-deficient lung cancers. Collectively, this work provides key insights into SMARCA4-mediated tumor suppression and SWI/SNF function in the lung.

Abstract:
Retrospective analysis of GC patients treated with Immune checkpoint inhibitors (ICI) showed that there are subsets of gastric cancer patients (e.g., MSI-H GC tumor) who have significant response to ICI. However, underlying molecular mechanisms of MSI-H GC tumors responded to ICI is largely unknown. Recent studies indicated that spatial analysis of tumor immune microenvironment (TIME) within a patient’s tumor could improve our understanding of how immune environment have on influence on response to ICI. In this work, we are developing and applying novel machine learning and artificial intelligence (AI) algorithms utilizing digital histopathological whole slide images (WSIs) with matched spatial transcriptome data (NanoString GeoMx) to 1) detect and analyze spatial organization of tumor infiltrated lymphocytes (TILs) and MSI-H tumor regions within the patient’s tumor and 2) perform spatial transcriptome anlaysis of TILs and MSI-H tumor regions for better understanding of ICI response in GC tumors. Our preliminary analysis demonstrates that spatial organization and spatial and cellular heterogeneity of TILs and MSI-H tumor cells within TIME could provide a novel biomarker to predict ICI response. Taken together, our work provides new foundation of how spatial cellular and image analysis of TIME could be used to predict and guide immunotherapeutic responsiveness as well as the potential to develop new therapeutic targets.of the run and analysis data generated after the run.

Abstract:
Chronic kidney disease (CKD) is a global health threat, affecting over 10% of the world population, including an estimated 37 million Americans. Importantly, glomerular diseases account for 70% of all the CKD. To date, no effective treatment for CKD exists, except that of dialysis or transplantation, which are associated with comorbidities, complications, and poor quality of life. Both of these treatment options present an enormous economic burden on the healthcare system, not to mention that transplantation is not even readily available due to a shortage of donor organs. Therefore, new breakthrough knowledge to generate more potent therapies for CKD is needed.

In the glomerulus, podocytes and endothelial cells together with the glomerular basement membrane form the glomerular filtration apparatus, and damage to any one of its components leads to kidney disease. Many different theories of disease progression have emerged over time, ranging from podocyte depletion hypothesis to recent evidence for metabolic dysfunction. However, none of these claims have been able to clearly elucidate the exact path of progression, which has ultimately resulted in failed attempts for treatment.

We use Alport syndrome (AS) as a well-established model to study CKD, a hereditary disease, caused by mutations in one of the α-chains of the collagen IV heterotrimer (ColIVα3α4α5), the major constituent of the glomerular basement membrane. In the absence of one of the α-chains, ColIVα3α4α5 network fails to assemble properly and is substituted by a much weaker ColIVα1α2α1 network. Like all CKDs, AS progresses to end-stage renal disease and ultimately to kidney failure. Despite the fact that the origin of the disease is well-established, pathogenic mechanisms of progression are not well understood.

To better understand disease mechanisms and contribution of the glomerular cell types, in this proposal we seek to apply spatial transcriptome analysis tools to investigate patterns of gene expression using kidney biopsies from healthy and AS patients. Analysis of this kind has never been performed on the kidney, especially on kidneys from AS. The use of this novel technique not only will increase our knowledge of glomerular cell gene expression, but it will define their histological setting (morphological context), which is critical to understand their integrated function and their role in disease progression, thus providing new information about the morphological and genetic changes of all the entire glomerulus in a human setting. Therefore, data generated from this study will be exceptionally valuable in understanding the molecular pathways that are altered in AS and responsible for the pathogenesis of this disease, providing important data on potential role and involvement of individual cell types or regions within the glomerulus. Combined with physiologic studies and other data from our laboratory we will uncover important new targets and mechanisms in the setting of AS.

In conclusion, this strategy will enable us to answer unresolved questions about glomerular damage and repair mechanisms and will help in the development of new therapies to halt glomerular disease progression. The knowledge gained from studies of disease triggering mechanisms in AS might be applicable to other forms of CKD.

Abstract:
Bulk or single-cell sequencing has been highly successful at investigating cell populations but loses key spatial information present in an undissociated tissue sample. To understand the effect of localized transcriptomic changes, high-plex spatial technologies are needed to profile gene expression while maintaining information on tissue architecture. GeoMx® Digital Spatial Profiling (DSP) measures analytes within areas of interest (AOIs) defined by tissue morphology or gene expression, and the recent launch of digital spatial profiling next-generation sequencing (DSP-NGS) readout enables tens of thousands of targets to be profiled simultaneously. The Human Whole Transcriptome Atlas (HuWTA) for GeoMx DSP leverages this high-plex potential to target over 99% of human protein-coding genes, with curation dropping 10 extreme high expressors to optimize readout efficiency. Here, we demonstrate the capabilities of HuWTA by benchmarking against other bulk and spatial technologies and by interrogating multiple healthy and disease tissue types.

We first profiled cell pellet arrays (CPAs) with known RNA sequencing (RNA-seq) expression profiles and found that HuWTA has comparable sensitivity and specificity to the previously developed Cancer Transcriptome Atlas (CTA). We also found good concordance between HuWTA and RNAscope, CTA, and bulk RNA-seq (for all HuWTA comparisons Rho >0.70, Spearman). We further leveraged single-target quantification from RNAscope to determine an absolute number of transcripts required per cell and per AOI to detect a target with HuWTA. Finally, we confirmed these cell line results by examining HuWTA and CTA performance on tissue microarrays (TMAs) and assessing results relative to published datasets. Across multiple-size regions of interest (ROI), approximately 10,000 expressed genes were observed.

We next examined unique morphological structures in multiple tissues to demonstrate the HuWTA application across diverse tissue settings. To characterize sequencing read depth required to maximize biological insight, we performed in silico subsampling on sequencing results. We found that different analyses require distinct minimum read depths, and smaller AOI sizes are more sensitive to lower sequencing depth, but all metrics are optimized by 100 reads/μm2. We then used these guidelines in experiments profiling directed AOIs across healthy and diseased kidney samples to uncover region- and cell-specific changes in gene expression in diabetic kidney disease.

Together, our results demonstrate that HuWTA successfully integrates transcriptome-scale spatial biology with sophisticated ROI design to enable flexible discovery with spatial context.

Abstract:
Bladder cancer is the fourth most common cancer in men and the fifth most common malignancy overall. There were over 80,000 new cases of bladder cancer and 17,000 deaths in 2018 in the United States alone. At presentation, 75% are non-invasive and 25% are muscle-invasive or metastatic. The aggressiveness of the cancer is determined by the types of genetic mutations and the tumor microenvironment, and each stage has a different behavior and response to therapy. Treatment is based on stage and grade, but new opportunities are emerging to categorize based on cellular composition. A deeper understanding of the cellular and molecular composition of bladder tumor subtypes could provide much-needed biomarkers to aid the clinician in treatment choices as well as in the development of new therapies.

We used an unbiased approach by single-cell RNA sequencing (scRNA-seq) to identify cell types of the normal human bladder and to develop signatures of various cell populations. To identify the spatial location and relationships of cells within the bladder, we used NanoString’s GeoMx Digital Spatial Profiler and the novel Whole Transcriptome Atlas. Using common structural markers and cell type-specific markers identified by scRNA-seq, we analyzed the whole transcriptomes of anatomical regions in highly structured normal bladder, invasive tumors, and surrounding tumor-invaded tissue in formalin-fixed, paraffin-embedded (FFPE) tissue sections. Spatial expression profiles were mapped back to each region of interest and integrated with scRNA-seq data to build spatially resolved and cell type-specific profiles of gene expression in normal bladder and to identify dysregulation in cancer. These data will help guide the identification of biomarkers and cell types altered in cancer, as well as provide an atlas of gene expression and cell composition of the normal human bladder.