In Silico ascription of gene expression differences to tumor and stromal cells in a model to study impact on breast cancer outcome
Ontology highlight
ABSTRACT: We present a method for identifying and ascribing differentially expressed genes to tumor epithelial and/or stromal cells, by utilizing pathologic information and weighted t-statistics Sets of differentially expressed gene-probes were identified in tumors from patients who developed distant metastasis compared with those who did not, by weighing the contribution from each tumor with the relative content of stromal and tumor epithelial cells in their individual tumor specimen. The analyses were performed under various assumptions of mRNA transcription level from tumor epithelial cells compared with stromal cells. Supplementary files: Human_AB1700_Annotations_09_06.txt: This file includes control probes and obsolete probes not listed in GPL2986. Raw_data.txt: This file includes control probes not listed in GPL2986.
Project description:Background: Epithelial-stromal crosstalk plays a critical role in invasive breast cancer (IBC) pathogenesis; however, little is known on a systems level about how epithelial-stromal interactions evolve during carcinogenesis. Results: We develop a framework for building genome-wide epithelial-stromal co-expression networks composed of pairwise co-expression relationships between mRNA levels of genes expressed in the epithelium and stroma across a population of patients. We apply this method to laser capture micro-dissection expression profiling datasets in the setting of breast carcinogenesis. Our analysis shows that epithelial-stromal co-expression networks undergo extensive re-wiring during carcinogenesis, with the emergence of distinct network hubs in normal breast, ER-positive IBC, and ER-negative IBC, and the emergence of distinct patterns of functional network enrichment. In contrast to normal breast, the strongest epithelial-stromal co-expression relationships in IBC mostly represent self-loops, in which the same gene is co-expressed in epithelial and stromal regions. We validate this observation using an independent laser capture micro-dissection dataset and confirm that self-loop interactions are significantly increased in cancer by performing computational image analysis of epithelial and stromal protein expression using images from the Human Protein Atlas. Conclusions: Epithelial-stromal co-expression network analysis represents a new approach for systems-level analyses of spatially-localized transcriptomic data. The analysis provides new biological insights into the re-wiring of epithelial-stromal co-expression networks and the emergence of epithelial-stromal co-expression self-loops in breast cancer. The approach may facilitate the development of new diagnostics and therapeutics targeting epithelial-stromal interactions in cancer. 36 flash-frozen human primary breast cancer samples were subjected to laser capture microdissection to separately isolate matched tumor epithelial and tumor-associated stromal components. RNA was isolated, subjected to 2 rounds of amplification, and hybridized on Agilent 4x44K microarrays along with a common reference (single round-amplified commercially obtained Universal Human Reference RNA) in a dyeswap design. For two samples of tumor-associated stroma, a second technical replicate was performed. Samples were labelled as ER-positive based on ESR1 gene expression levels in the tumor epithelium, using univariate Gaussian mixture model-based clustering via the mclust package in R.
Project description:Transcriptomic profiling Background Cholangiocarcinoma accounts for 5-10% of primary hepatic cancers. The etiology is unclear and patients are often diagnosed without risk factors. Resection is the only curative treatment although patients frequently remain undiagnosed until advanced stage of disease. Methods To construct molecular classification of cholangiocarcinoma, we profiled the transcriptomes of 104 freshly-frozen tumors and 59 matched non-cancerous livers obtained from Australia, Europe and the United States. We also performed mutational analysis of KRAS, EGFR and BRAF, and used laser-capture microdissection to obtain independent gene expression profiles for epithelial and stromal compartments in a subset of tumors. The selected target genes were validated by western blotting and immunohistochemistry. Results Transcriptomic profiling classified cholangiocarcinoma into two distinct subclasses defined by survival (P<0.0007) and early recurrence (P<0.001). Applying leave-one-out cross-validation, we optimized the prognostic classifier to 238 genes which were positively enriched in the epithelial tumor compartment. A deregulated HER2 network was associated with the epithelial compartment which also showed a frequent overexpression of Ki67, EGFR, MET and pRPS6 whereas inflammatory cytokines were enriched in tumor stroma specifically in patients with poor prognosis. KRAS mutations were found in 24.6% of patients with poor disease outcome. Conclusion Our study presents new insights into pathogenesis of cholangiocarcinoma and stratification of the patients according to survival and recurrence. Identification of a subgroup of patients among the poor prognostic cohort characterized by KRAS mutations and oncogenic-addiction may provide a novel therapeutic opportunity for this treatment-refractory malignancy. Profiling of individual cholangiocarcinomas and non-cancerous matched surrounding livers using normal bile ducts as reference
Project description:Progressive fibrosis of the skin and internal organs accounts for the intractable nature and the high mortality of scleroderma. As the principal effector cells responsible for fibrosis, stromal fibroblasts and myofibroblasts contribute to excessive deposition of collagens and other extracellular matrix proteins. Transforming growth factor β (TGF-β), which stimulates collagen synthesis, myofibroblast differentiation and epithelial-mesenchymal transition (EMT), is implicated as a key initiating factor in both physiological and pathological tissue remodeling. However, the mechanism responsible for the persistence of the fibrotic process associated with pathological repair remains poorly understood. In this study, we analyzed the gene expression in dermal fibroblasts using different treatments (Poly I:C, IFN-beta, Egr-3 overexpression and other conditions). Primary fibroblast cultures were established by explantation from neonatal foreskin. Biopsy protocols were approved by the Institutional Review Board at Northwestern University. Fibroblasts were maintained in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (Lonza, Basel, Switzerland), 50 µg/ml penicillin, and 50 µg/ml streptomycin in a humidified atmosphere of 5% CO2 at 37°C, and studied between passages 2-8. At confluence, serum-free media supplemented with 0.1% bovine serum albumin (BSA) were added to the cultures for 24 h prior to the treatment of Poly I:C for 2 hours or 24 hours (10 ug/ml), or IFN-β (100 U/ml) or other reagents for 24 hours, or infected with Ad-Egr-3 or Ad-GFP at 50 MOI for 48 hours. Total RNA was isolated using RNeasy Mini Plus Kits (Qiagen, Valencia, CA). The integrity of RNA was determined using Agilent Bioanalyzer (Santa Clara, CA). Fluorescently labeled cDNA was prepared (Ambion, Austin, TX), and was hybridized to Illumina Human HT-12 Version 4 microarray chips containing 44,000 probes (Illumina, San Diego, CA). Raw signal intensities for each probe were obtained using Illumina BeadStudio data analysis software and imported to the Bioconductor lumi package for transformation and normalization. The data were preprocessed using a variance stabilization transformation method followed by quantile normalization. Data from probes that produced signals near or below background levels (estimated based on Illumina negative control probes) with all samples were discarded.
Project description:Previously we have demonstrated that inactivation of retinoic acid receptor beta (Rarb) in the mouse results in a protective effect against ErbB2-induced mammary gland tumorigenesis although Rarb has been reported as a tumor suppressor before. In the current study, we further confirmed that ablation of Rarb has a very similar impact on Wnt1-induced mammary gland tumorigenesis as those on ErbB2-induced mammary gland tumorigenesis. Nevertheless, the mechanisms by which Rarb confers its effects on tumor progression is quite different although both involving in tumor microenvironment (TME) remodeling. In the Wnt1 tumors, ectopic wnt1 produced by malignant luminal cells activates nearby stromal cells by a paracrine manner. In return, the stromal cells secreted IGF1 to regulate the growth of tumor cells. There is a need of Rarb expression in this interaction. Deletion of Rarb inhibits both wnt1/β-catenin signaling and IGF1/Akt axis in the myoepithelial tumor cells which results in the suppression of epithelial-mesenchymal transition (EMT) in these tumors. Since wnt1 tumors resemble basal-like breast cancer with a poor clinical prognosis in which EMT is one of the most important way for tumor cells to survive against standard treatment and to go to metastasis, we propose that (1) the stromal gene expression signature of Rarb ablation in wnt1 tumors could have some clinical value in predicting the breast cancer outcome; and (2) Rarb antagonist might be a potential therapeutic strategy in EMT-driven aggressive cancers such as basal-like breast cancer. Laser capture microdissection (LCM) was performed to separate the mammary tumor samples into epithelial cell compartment and stromal cell compartment. Transcriptional profiling of the two compartments were investigated by microarray analysis.
Project description:Using a large representative sample of postmenopausal women in the Norwegian Women and Cancer (NOWAC) postgenome study, we investigated blood gene expression changes due to intra-technical variability, normal inter-individuality (age, body mass index, fasting status), and exposure variables (smoking, hormone therapy and medication use) at proportion and level of real life situation revealing mechanistic insights of these effects mirrored in blood. We used a representative sample of postmenopausal women (N=286) in the NOWAC postgenome study. We investigated blood gene expression changes due to intra-technical variability, normal inter-individuality (age, body mass index, fasting status), and exposure variables (smoking, hormone therapy and medication use). A total of 304 arrays, including 18 technical replicates, were analyzed. We filtered out samples which had less than 40% probes with a signal to noise ratio (S/N) greater than or equal to 3. When a technical replicate was conducted, the array with the least number of probes with S/N greater than or equal to 3 was excluded. After samples filtration, a total of 286 arrays were analyzed. The following samples were filtered out; they were not used in the normalization processing, and none of the study conclusions are based on these samples. NWbAB_ 1 NWbAB_ 2 NWbAB_ 3 NWbAB_ 4 NWbAB_ 5 NWbAB_ 7 NWbAB_ 8 NWbAB_ 61 NWbAB_ 63 NWbAB_ 116 NWbAB_ 120 NWbAB_ 121 NWbAB_ 180 NWbAB_ 183 NWbAB_ 214 NWbAB_ 249 NWbAB_ 254 NWbAB_ 299 Sample records (GSMxxxx) are not provided for the filtered-out samples. However, the metadata for the filtered-out samples is included in the Series supplementary file GSE15289_filtered_metadata.txt. The raw data for both the filtered-out samples and non-filtered-out samples (GSM381832-GSM382117) are included in the Series supplementary file GSE15289_raw.txt.
Project description:Tumor infiltrating neutrophils (TAN) have been shown to exert both pro- and anti-tumoral activities and their recruitment and polarization are triggered by tumor-derived signals. Resident mesenchymal stromal cells (MSC) could contribute to tumor-supportive cell niche and have been shown to display tumor-specific transcriptomic, phenotypic, and functional features compared to normal tissue. In our study, we investigate whether these two cell subsets establish a bidirectional crosstalk in the context of B-cell lymphoma. We used microarrays to explore how neutrophils could trigger the polarization of tumor-supportive stromal cells. Gene expression analysis were performed on stromal cells (MSC) derived from bone marrow (BM) or tonsil (Resto) of healthy donors. These BM-MSC (n=3) or Resto (n=3) were primed or not with neutrophils for 1 day to induce stromal modification.
Project description:The goal of the study was to identify molecular subgroups that might predict clinical outcomes in serous epithelial ovarian cancer (EOC) patients. A second objective was to identify potential therapeutic targets for serous EOC based on improved understanding of the molecular diversity of the disease. Ovarian tissues and matched peripheral blood samples were prospectively obtained from sequential patients undergoing planned gynecologic surgery at Cedars-Sinai Medical Center between 1989 and 2005. All patients underwent surgery and received adjuvant chemotherapy with a contemporaneous standard-of-care regimen. Ovarian tissue samples (n=172) were compared to a reference pool of 106 ovarian samples. Mixed reference includes normal, benign, borderline, and malignant samples.
Project description:We performed poly(A) enriched RNA sequencing of HeLa, U-2 OS and mES cells. This series includes two re-analyzed samples from mouse: GSM1032506 and GSM1032518 (denoted as iPS.A and iPS.B, respectively). The re-analyzed data for these samples are included in the file INCLUSION_LEVELS_FULL-Mm25.tab.gz. This series includes three re-analyzed samples from human: GSM808734, GSM1023087 and GSM1023070 (denoted as iPS.A, iPS.B and iPS.C, respectively). The re-analyzed data for these samples are included in the file INCLUSION_LEVELS_FULL-Hs23.tab.gz
Project description:Pancreatic ductal adenocarcinoma (PDAC) remains a lethal disease with a 5-year survival of 4%. A key hallmark of PDAC is extensive stromal involvement, which makes capturing precise tumor-specific molecular information difficult. Here, we have overcome this problem by applying blind source separation to a diverse collection of PDAC gene expression microarray data, which includes primary, metastatic, and normal samples. By digitally separating tumor, stroma, and normal gene expression, we have identified and validated two tumor-specific subtypes including a “basal-like†subtype which has worse outcome, and is molecularly similar to basal tumors in bladder and breast cancer. Furthermore, we define 'normal' and 'activated' stromal subtypes which are independently prognostic. Our results provide new insight into the molecular composition of PDAC which may be used to tailor therapies or provide decision support in a clinical setting where the choice and timing of therapies is critical. Analysis of the landscape of gene expression in pancreatic adenocarcinoma. Data include 145 primary and 61 metastatic PDAC tumors, 17 cell lines, 46 pancreas and 88 distant site adjacent normal samples. Arrays represent distinct samples. The SPOT column in the raw data file (linked to each sample record) contains Agilent feature extraction numbers (included in the 'GPL4133-20424.txt' linked to the platform records).
Project description:Tissue of the breast is heterogeneous, consisting of a variety of cell types and connective tissue. This heterogeneity is also present in breast tumors and will complicate proteomic analysis, as it is not always clear whether a signal originates from the stromal environment, from normal epithelial or tumor cells. Here we microdissected a variety of cell types and stroma from benign and malignant breast tissues. We compared proteomic differences between these tissues, both from cells of epithelial origin and the stromal environment. Differences in protein abundances corresponded with several hallmarks of cancer, including loss of cell adhesion, transformation to a migratory phenotype, and enhanced energy metabolism. Furthermore, despite enriching for (tumor) epithelial cells, many changes to the extracellular matrix were detected in microdissected cells of epithelial origin. The stromal compartment was heterogeneous and richer in the number of fibroblast and immune cells in malignant sections, compared to benign tissue sections. Although this heterogeneity complicated detection of differentially abundant proteins, several markers were exclusively detected in stroma. However, as heterogeneity in the stroma is more difficult to be reduced through microdissection, comparative analysis was most informative in microdissected cells of epithelial origin, and provided a relatively complete picture of malignant transformations.