Ilya Kupershmidt

Ontology-Based Meta-Analysis of Global Collections of High-Throughput Public Data

Background The investigation of the interconnections between the molecular and genetic events that govern biological systems is essential if we are to understand the development of disease and design effective novel treatments. Microarray and next-generation sequencing technologies have the potential to provide this information. However, taking full advantage of these approaches requires that biological connections be made across large quantities of highly heterogeneous genomic datasets. Leveraging the increasingly huge quantities of genomic data in the public domain is fast becoming one of the key challenges in the research community today. Methodology/Results We have developed a novel data mining framework that enables researchers to use this growing collection of public high-throughput data to investigate any set of genes or proteins. The connectivity between molecular states across thousands of heterogeneous datasets from microarrays and other genomic platforms is determined through a combination of rank-based enrichment statistics, meta-analyses, and biomedical ontologies. We address data quality concerns through dataset replication and meta-analysis and ensure that the majority of the findings are derived using multiple lines of evidence. As an example of our strategy and the utility of this framework, we apply our data mining approach to explore the biology of brown fat within the context of the thousands of publicly available gene expression datasets. Conclusions Our work presents a practical strategy for organizing, mining, and correlating global collections of large-scale genomic data to explore normal and disease biology. Using a hypothesis-free approach, we demonstrate how a data-driven analysis across very large collections of genomic data can reveal novel discoveries and evidence to support existing hypothesis.

Slow Genetic Divergence of Helicobacter pylori Strains during Long-Term Colonization

ABSTRACT The genetic variability of Helicobacter pylori is known to be high compared to that of many other bacterial species. H. pylori is adapted to the human stomach, where it persists for decades, and adaptation to each host results in every individual harboring a distinctive bacterial population. Although clonal variants may exist within such a population, all isolates are generally genetically related and thus derived from a common ancestor. We sought to determine the rate of genetic change of H. pylori over 9 years in two asymptomatic adult patients. Arbitrary primed PCR confirmed the relatedness of individual subclones within a patient. Furthermore, sequencing of 10 loci (∼6,000 bp) in three subclones per time and patient revealed only two base pair changes among the subclones from patient I. All sequences were identical among the patient II subclones. However, PCR amplification of the highly divergent gene amiA revealed great variation in the size of the gene between the subclones within each patient. Thus, both patients harbored a single strain with clonal variants at both times. We also studied genetic changes in culture- and mouse-passaged strains, and under both conditions no genetic divergence was found. These results suggest that previous estimates of the rate of genetic change in H. pylori within an individual might be overestimates.

Global gene expression analysis in time series following N-acetyl L-cysteine induced epithelial differentiation of human normal and cancer cells in vitro.

BackgroundCancer prevention trials using different types of antioxidant supplements have been carried out at several occasions and one of the investigated compounds has been the antioxidant N-acetyl-L-cysteine (NAC). Studies at the cellular level have previously demonstrated that a single supplementation of NAC induces a ten-fold more rapid differentiation in normal primary human keratinocytes as well as a reversion of a colon carcinoma cell line from neoplastic proliferation to apical-basolateral differentiation [1]. The investigated cells showed an early change in the organization of the cytoskeleton, several newly established adherens junctions with E-cadherin/β-catenin complexes and increased focal adhesions, all features characterizing the differentiation process.MethodsIn order to investigate the molecular mechanisms underlying the proliferation arrest and accelerated differentiation induced by NAC treatment of NHEK and Caco-2 cells in vitro, we performed global gene expression analysis of NAC treated cells in a time series (1, 12 and 24 hours post NAC treatment) using the Affymetrix GeneChip™ Human Genome U95Av2 chip, which contains approximately 12,000 previously characterized sequences. The treated samples were compared to the corresponding untreated culture at the same time point.ResultsMicroarray data analysis revealed an increasing number of differentially expressed transcripts over time upon NAC treatment. The early response (1 hour) was transient, while a constitutive trend was commonly found among genes differentially regulated at later time points (12 and 24 hours). Connections to the induction of differentiation and inhibition of growth were identified for a majority of up- and down-regulated genes. All of the observed transcriptional changes, except for seven genes, were unique to either cell line. Only one gene, ID-1, was mutually regulated at 1 hour post treatment and might represent a common mediator of early NAC action.The detection of several genes that previously have been identified as stimulated or repressed during the differentiation of NHEK and Caco-2 provided validation of results. In addition, real-time kinetic PCR analysis of selected genes also verified the differential regulation as identified by the microarray platform.ConclusionNAC induces a limited and transient early response followed by a more consistent and extensively different expression at later time points in both the normal and cancer cell lines investigated. The responses are largely related to inhibition of proliferation and stimulation of differentiation in both cell types but are almost completely lineage specific. ID-1 is indicated as an early mediator of NAC action.

Gene expression analysis of human epidermal keratinocytes after N-acetyl L-cysteine treatment demonstrates cell cycle arrest and increased differentiation.

Objectives: Several cancer prevention programmes have previously been executed using treatment of antioxidant compounds. The antioxidant N-acetyl L-cysteine (NAC), a membrane-permeable aminothiol, is a sulfhydryl reductant reducing oxidised glutathione, as well as being a precursor of intracellular cysteine and glutathione. A previous report based on the cellular response to NAC treatment showed that NAC induced a 10-fold more rapid differentiation in normal primary keratinocytes as well as a reversion of a colon carcinoma cell line from neoplastic proliferation to apical-basolateral differentiation. In order to investigate molecular events underlying thechanges in proliferation and differentiation induced by NAC treatment, we performed global gene expression analysis of normal human epidermal keratinocytes in a time series. Methods: Treated samples were compared to untreated samples through a reference design using a spotted cDNA array comprising approximately 30,000 features. B statistics was used to identify differentially expressed genes, and RT-PCR of a selected set of genes was performed to verify differential expression. Results: The number of differentially expressed genes increased over time, starting with 0 at 30 min, 73 at 3 h and increasing to 952 genes at 48 h. Results of the expression analysis showed arrest of the cell cycle and an upregulation of cytoskeletal reorganisation, implicating increased differentiation. A comparison to gene ontology groups indicated downregulation of a large number of genes involved in cell proliferation and regulation of the cell cycle. Conclusions: A significant fraction of the differentially expressed genes could be classified according to their role in the differentiation process, demonstrating that NAC regulates the conversion from proliferation to differentiation at a transcriptional level.

Identification of candidate SNPs for drug induced toxicity from differentially expressed genes in associated tissues

The growing collection of publicly available high-throughput data provides an invaluable resource for generating preliminary in silico data in support of novel hypotheses. In this study we used a cross-dataset meta-analysis strategy to identify novel candidate genes and genetic variations relevant to paclitaxel/carboplatin-induced myelosuppression and neuropathy. We identified genes affected by drug exposure and present in tissues associated with toxicity. From ten top-ranked genes 42 non-synonymous single nucleotide polymorphisms (SNPs) were identified in silico and genotyped in 94 cancer patients treated with carboplatin/paclitaxel. We observed variations in 11 SNPs, of which seven were present in a sufficient frequency for statistical evaluation. Of these seven SNPs, three were present in ABCA1 and ATM, and showed significant or borderline significant association with either myelosuppression or neuropathy. The strikingly high number of associations between genotype and clinically observed toxicity provides support for our data-driven computations strategy to identify biomarkers for drug toxicity.

Abstract 5151: Integrative genomic analysis identifies cancer-testis antigen LEMD1 as a prognostic biomarker.

The ability to molecularly distinguish disease states can improve prognostic resolution and therapeutic options. Cancer-testis antigens (CTA), expressed in germ cells of the testis and in a variety of cancers, are of particular interest for their potential use as targets for anti-cancer vaccines. We highlight the application of the NextBio platform, which integrates multiple datatypes such as RNA/miRNA expression, copy number variation, epigenetic changes, and next-generation sequencing with clinical attributes to identify and characterize novel disease associations in multiple myeloma. Genomic and clinical data of 2,546 multiple myeloma patients obtained from public data sources were investigated in the NextBio platform. We initially analyzed the gene expression of various clinical subgroups to the median expression of all groups to identify genes significantly correlated to patient survival. We filtered for patients with advanced disease, defined as those with beta-2 microglobulin blood levels greater than 5.5mg/L, and identified LEMD1, a CTA, as positively correlated with survival (p=2.3E-6). We then analyzed the gene expression of LEMD1 in patients of all disease stages, and found those with increased expression of LEMD1 had improved survival compared to those who did not (p=9.4E-3). Patients with increased expression of LEMD1, however, had concomitant decreased expression in 14/20 CTAs of the MAGE family (p A preliminary exploration of LEMD1 expression in other cancers found increased expression of LEMD1 to be significantly associated with improved survival in ovarian cancer (p=5.4E-3) but reduced survival in renal cancer (p=0.01). Further analysis of LEMD1 expression in normal tissues showed high expression of LEMD1 in spermatozoa and limited expression in 315 other tissue and cell types. As tissue specific antigens behind the blood-testis barrier, the expression of CTAs in cancer can potentially serve as immunogenic targets, and further investigation is warranted to explore their use in tumor immunotherapy. Citation Format: Jenny Park, Beatrice Chiu, Joe Delaney, Cate Paschal, Daniel Roche, James Shima, Anita Umesh, Robert Wisotzkey, Mamatha Shekar, Ilya Kupershmidt. Integrative genomic analysis identifies cancer-testis antigen LEMD1 as a prognostic biomarker. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 5151. doi:10.1158/1538-7445.AM2013-5151

Identification of AGR3 as a potential biomarker though public genomic data analysis of triple-negative (TN) versus triple-positive (TP) breast cancer (BC).

31 Background: TNBCs have an aggressive clinical phenotype with worse prognosis than TPBC. Here we sought to identify other genomic changes in patients with TN vs. TPBC, by mining TCGA breast cancer genomic data using the NextBio Clinical platform. METHODS Breast cancer patients with decreased (n=51) and increased (n=46) ER/PR/HER2 mRNA levels were selected with the molecular filter in NB Clinical. Top-ranked differentially expressed genes were identified then analyzed via NB Research, which comprises a library of individually curated publically available research and clinical grade genomic data. RESULTS Whole transcriptome analysis of TNBC and TPBC patients identified anterior gradient-3 (AGR3), a protein disulfide isomerase, as most downregulated (TNBC vs. TPBC, fold change = -178.2, p=2.7E-32). AGR3 mRNA was reduced in 94% of TNBC and increased in 96% of TPBC patients. AGR3 was hypomethylated in 50% of TPBC vs. 4% of TNBC patients, suggesting regulation by methylation. When the entire breast cancer cohort was stratified by AGR3 expression, underexpression was prevalent in younger patients (<50 yr, 39%) compared to those with overexpression (<50 yr, 23%). Further, 63% of AGR3 underexpressing patients had severe TP53 mutations vs. 19% with AGR3 overexpression. Breast cancer was most correlated to AGR3 in the NextBio Disease Atlas, with supporting data from 110 studies. AGR3 expression was decreased in ER- vs. ER+, PR- vs. PR+, HER2- vs. HER2+, and G3 vs. G1 tumors, and AGR3 promoter was hypermethylated in ER- vs. ER+, consistent with the TCGA results. Body Atlas showed strongest AGR3 expression in mucosa, fallopian tube, breast, and epithelial cells, in line with its proposed role in epithelial barrier function. Knockdown Atlas results showed TP53 mutation associated with reduced AGR3 mRNA, and the PharmacoAtlas showed positive regulation of AGR3 by estrogen analogues. CONCLUSIONS Using the NextBio Clinical and Research platforms, we identified AGR3 as a marker of TNBC vs. TPBC that correlates with TP53 mutation status. AGR3 may provide an alternate route of TNBC treatment. Its potential in regulating epithelial barrier function may provide insights into the mechanism of disease progression.

Abstract 2989: Identification of a potential tumor suppressor function for SNX9 by mining public genomic data

Proceedings: AACR 103rd Annual Meeting 2012‐‐ Mar 31‐Apr 4, 2012; Chicago, IL Members of the sorting nexin (SNX) family of proteins are involved in intracellular trafficking and endocytosis processes, with functions essential to regulation of cell surface receptor levels and subsequent signaling by receptor-mediated endocytosis. Detrimental mutations and altered expression of SNX genes have been found in several cancers and tumor cell lines. A recent publication (Mao et al. AJCR 2011) identified a chromosomal translocation that created a SNX9:UNC5C fusion gene in prostate cancer cells. UNC5C is a known tumor suppressor gene. Although SNX9 is known to regulate degradation of EGFR, a well-established prognostic marker for lung cancer, no tumorigenic role has been identified for SNX9. We investigated associations of SNX9 to cancer phenotypes by mining curated public genomic data available in NextBio. Integration of diverse datatypes such as gene/miRNA expression, methylation, and copy-number variation in NextBio allowed us to uncover several lines of evidence which indicated that SNX9 activity is altered in many types of cancer. i) Loss of the SNX9 gene was observed in many different cancers including multiple myeloma, leukemia, lymphoma, melanoma, glioblastoma, renal clear cell carcinoma, and breast cancer. ii) Hypermethylation of SNX9, indicative of overall gene silencing, occurred in colorectal cancer, breast cancer and glioblastoma. iii) The miRNA predicted to target SNX9, miR-493-5p, was up-regulated in prostate, testicular, liver, and lung cancers. iv) SNX9 gene expression was down regulated relative to normal tissues in multiple myeloma, leukemia, lymphoma, cancers of the skin, breast, prostate and more. Correlations within the data suggest an association between the loss of SNX9 and more aggressive or recurrent disease; SNX9 is reduced in higher grade, metastatic clear cell carcinoma relative to lower grade, non-metastatic clear cell carcinoma. In myeloma and leukemia patients, lower SNX9 expression in primary tumors correlated with earlier relapsing disease. SNX9 expression was also lower when comparing recurring with primary acute myeloid leukemia. Querying our PharmacoAtlas application revealed that SNX9 gene expression was altered by a number of anti-neoplastic therapies. For example, paclitaxel, bortezomib, and letrozole treatments were each found to increase SNX9 gene expression in breast cancer cells whereas doxorubicin treatment resulted in decreased SNX9 expression. The integrative analysis of curated public genomic data drawn from thousands of studies strongly supports a hypothesis that SNX9 is a putative tumor suppressor and biomarker for more aggressive forms of cancer. As genomic technologies are applied to patient evaluation and therapeutic guidance, integrating multiple lines of evidence from research and clinical data to identify biomarkers such as SNX9 will increasingly inform clinical decision making. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 2989. doi:1538-7445.AM2012-2989