Research Resource: The Endometrium Database Resource (EDR)

Abstract

In order to understand the biology of the endometrium and potentially develop new diagnostic tools and treatments for endometrial diseases, the highly orchestrated gene expression/regulation that occurs within the uterus must first be understood. Even though a wealth of information on endometrial gene expression/regulation is available, this information is scattered across several different resources in formats that can be difficult for the average bench scientist to query, integrate, and utilize. The Endometrium Database Resource (EDR) was created as a single evolving resource for protein- and micro-RNA-encoding genes that have been shown by gene expression microarray, Northern blot, or other experiments in the literature to have their expression regulated in the uterus of humans, mice, rats, cows, domestic pigs, guinea pigs, and sheep. Genes are annotated in EDR with basic gene information (eg, gene symbol and chromosome), gene orthologs, and gene ontologies. Links are also provided to external resources for publication/s, nucleic and amino acid sequence, gene product function, and Gene Expression Omnibus (GEO) phase expression graph information. The resource also allows for direct comparison of relative gene expression in different microarray experiments for genes shown in the literature to be differentially expressed in the uterus. It is available via a user-friendly, web-based interface and is available without charge or restriction to the entire scientific community. The EDR can be accessed at http://edr.research.bcm.edu.

The uterus functions to support fetal growth and is essential for the perpetuation of mammalian species. In order to accomplish this function, expression of thousands of genes must be tightly regulated. This requirement is particularly striking in the dynamic endometrial layer of the uterus, which undergoes significant changes in preparation for embryo implantation. In humans the endometrium undergoes cyclical phases of proliferation and shedding or resorption during the menstrual cycle in response to estrogen, progesterone, other hormones, and additional autocrine and paracrine factors (1). A variety of conditions ranging from uterine cancer to endometriosis can occur from alterations in mRNA- and micro-RNA-encoding (miRNA) gene expression and genetic mutations. Endometriosis alone was estimated to have cost $22 billion in US dollars in 2002 (2) and today affects an estimated 176 million women of reproductive age worldwide (www.endometriosis.org). Likewise, it is estimated that between 4% and 8% of women worldwide suffer from polycystic ovarian syndrome (3–5) with diagnostics and treatments costing billions of US dollars (6). Endometrial cancer causes approximately 6% of all cancers in women and is the most common gynecological cancer in US women (www.cancer.gov). Not only is the incidence of endometrial cancer growing, but minimal progress has been made in reducing the mortality of this disease. In order to understand the biology of and potentially develop new diagnostics and treatments for these diseases, the molecular changes that occur within the uterus must first be understood.

One can search the scientific literature for information on these changes by using data aggregation services (eg, PubMed), but such searches typically lack specificity and structure. Thus, it is common for researchers to spend hours reviewing articles, extracting the desired data, and synthesizing hypotheses based upon the context of the articles. Although a large number of biological databases [eg, BioGPS, formerly SymAtlas (7) or Online Mendelian Inheritance in Man, OMIM (8)] provide information on genes, salient uterine-specific gene annotations are scattered across different resources. Even when such data are found, it can be complicated to integrate and trace them back to the publications from which the data in these resources originated or to determine how these data were collected and by whom.

Like the Ovarian Kaleidoscope Database [OKdb; (9)], a sister resource that provides information on genes expressed and regulated in the ovary, the Endometrium Database Resource (EDR) was developed to address the need for a single evolving resource that compiles data on a variety of protein- and miRNA-encoding genes that have been shown experimentally to be expressed in the uterus. The EDR includes gene expression data for salient human, mouse, rat cow, domestic pig, sheep, and guinea pig genes. To assist end users in understanding the greater context of the gene expression data in the organisms with the largest number of data points, the EDR also contains core gene annotations as described below for all human, mouse, and rat genes, although only a subset of these have curated expression data. To ensure that the resource accurately reflects the literature, expert curators review the literature and identify relevant scientific publications that include gene expression data. The genes that were found by the authors of these publications to have significantly altered relative expression are annotated with information from the article, basic gene information (eg, Entrez gene symbol or chromosome), gene orthologs, gene ontologies, and Gene Expression Omnibus (GEO) Profiles (Figure 1). External links to related publications, nucleic and amino acid sequence, gene product function, and relative expression in different human tissues (GEO Phase Graphs) are provided for each record. The EDR also supplies tools that allow scientists to easily compare relative RNA expression from experiments in different curated publications to visualize the effects of different experimental conditions on gene expression, to determine how gene expression differs in various layers of the uterus, and to generate new scientific hypotheses. Overall, the EDR will serve as an evolving resource that benefits the global reproductive biology research community. The EDR can be accessed at http://edr.research.bcm.edu.

Figure 1.

A sample EDR gene record. The top of each gene record contains Gene Detail Information with information such as gene identifier and symbol at left and External Links to other web resources' information about the gene at right. At the bottom of each record are three tabs, each of which contains additional information for the following three categories: Gene Annotations, PubMed Gene Search, and Microarray Papers. The Gene Annotations tab contains links to Phase Graphs, gene ontology (GO) terms for the gene, and Orthologs in sections that can be expanded, as shown for Phase Graphs, or collapsed as shown for GO terms. The PubMed Gene Search tab has links to endometrium and reproductive-relevant publications with data on the gene of interest that are indexed within PubMed. Microarray Papers provides additional information on EDR-curated publications.

Materials and Methods

The EDR version 2.4 is a modular web-based application built around a gene-centered, Oracle 11g database. The resource is a Java Enterprise Edition 5 system that uses the model-view-controller architecture with the JBoss Seam 2 framework to manage its various components. EDR uterine biology experts regularly reviewed the literature and identified scientific articles that contain uterine gene expression data (from microarrays, Northern blot, or quantitative PCR experiments) from one of the nine organisms. Other EDR biological domain experts then verified that the data from each identified publication were publicly available. For genes the expression of which was found by the authors to be significantly altered, the curators created lists of the gene symbols and whether the expression was significantly increased or decreased. Because the resource reports authors' interpretation of the data, not all of the genes had the same cutoffs for significance. For example, one publication might have identified significant P values as those corresponding with genes having ± 1.5-fold change in expression, whereas a second only those having ± 2.0-fold change as significant. All of these genes from both publications would be included in the EDR. If the authors did not list genes that were significantly differentially expressed, but instead provided a table of the genes and their relative fold-change in expression, curators extracted only those genes with ± 2.0-fold change. These data and annotations were entered into EDR for each appropriate experiment from the identified publications. If an experiment was performed over a time course, data from each time point were treated by the curators as “separate” experiments. For example, if endometrial gene expression was determined at 2, 4, 8, and 16 hours after pharmacological agent exposure, each of the four microarray data sets would be entered as a separate experiment for the publication in the EDR.

Using the EDR Editor data entry feature, the curators further annotated the extracted genes with the experimental conditions, the phase during which the gene was predominantly expressed in the menstrual cycle (when stated by the author), and the PubMed identifier for the publication. The Editor functionality consumed an National Center for Biotechnology Information service that utilizes the PubMed identifier and provided the publications author names, title, and other additional information. National Center for Biotechnology Information gene annotations, orthologs, OMIM, Mouse Genome Informatics (MGI), and some other external data that list orthologs of a given gene in different species, were pulled directly from their source databases and integrated within the database of the EDR.

Results

Currently, the EDR contains 199 total experiments using tissues or cell lines from 121 scientific publications. Most the EDR experiments, 172, are gene expression microarrays, but 27 other experiments include combinations of quantitative RT-PCR, expressed sequence tag Northern blot, and in situ hybridization. More than 142 000 gene records are available from various species including human (Homo sapiens), house mouse (Mus musculus), brown rat (Rattus norvegicus), domestic sheep (Ovis aries), cow (Bos taurus), guinea pig (Cavia porcellus), and domestic pig (Sus scrofa [domesticus]).

Searching the EDR for gene annotations

End users may find annotations of individual genes either by searching for a particular gene or browsing gene lists under the EDR Genes link. End users can use the Search Gene function when searching for a particular gene. Under the Search Gene function, end users can search by gene symbol, gene name, or Entrez gene identifier. The EDR search feature will autosuggest gene names or symbols that contain the text that has been entered into the field and allows users to limit the search by species. Because for many genes there is no single consensus as to the “proper” gene symbol and to disambiguate symbols that could refer to more than one gene, the autosuggestions also contain synonyms with the official Entrez gene symbol in parentheses. Search results for a particular gene symbol, gene name, or Entrez GeneID will be returned in a table that includes the Entrez GeneID, gene symbol, gene name, chromosome, and species. The gene name and symbol links to the EDR gene record, whereas the Entrez GeneID link connects to the gene's Entrez web page. When using the Browse Gene function, end users can scroll through a list of genes for a given organism or use alphabetical links to jump to genes beginning with a particular letter. By default the genes are sorted alphabetically by symbol, and 15 genes can be viewed per page. In order to make the browse function more manageable, the columns for gene identification, symbol, name, and chromosome are sortable, and the user can select to view 15, 25, 50, or 100 genes per page. By clicking on the gene name, the end user will be directed to that particular EDR gene record.

Each gene record contains a variety of biological annotations and external links that are organized in tables by content: Gene Detail Information, External Links, and Gene Annotations (Figure 1). The Gene Detail Information section includes the gene's Entrez GeneID with a link to its Entrez webpage, Entrez Gene Symbol, Entrez Gene Name, the chromosome on which the gene is located, and the species name (Figure 1). For each gene record, external links are provided for Unigene, GenBank, OMIM (for human genes only), the Human Protein Atlas (for human genes only), the MGI database (for mouse genes only) (10, 11), University of California Santa Clara (UCSC) Genome Browser (12), Homologene Search (11), BioGPS, and the OKdb (13)].

Under the Gene Annotations tab, the Phase Graphs button links to all reproduction-relevant GEO Profiles for a given gene (Figure 1). These profiles are species-specific endometrial cycle phase graphs that show the relative level of expression of a given gene in endometrial tissue samples that were collected at various phases in the cycle. Gene Ontology (GO) terminologies, identifiers, categories, and PubMed links for that particular GO term and gene are provided in the GO table. Orthologs are provided in a table listing the species and Entrez GeneID, which links to the EDR gene record for that ortholog. The PubMed Gene Search tab contains links for a PubMed search, which returns any publication in PubMed for the given gene symbol, and a uterine-specific search, which returns any publication in PubMed with the gene symbol, organism name, and any of the following terms: uterus, endometrium, reproductive tract, myometrium, oviduct, cervix or vagina as keywords [eg, ((Mapk10[Gene Symbol]) AND Mus musculus[Organism]) AND ((((endometrium OR myometrium) OR uterus) OR reproductive) OR vagina)]. The Microarray Papers tab contains a list of publications with differentially expressed genes that have been curated in EDR and a link to PubMed for each publication. Additionally, a shopping cart icon, which is located next to each paper, can be clicked to add that paper to the My Papers cart for relative gene expression comparison.

Searching the EDR for gene expression papers

In the Microarray Papers workflow, one may search for scientific publications that contain uterine gene expression data and even cross compare the RNA expression data from these papers. The EDR allows end users to browse papers by species or search papers by author name or title, which utilizes the autosuggest functionality. Individual papers contain author's names, title, species, a PubMed link to each publication, and a hyperlink that connects to additional annotations within EDR. These EDR annotations include general publication information and a list of uterine gene expression experiment data, which has a List Genes link that connects to a microarray chip platform, where applicable, and the list of genes that were found by the authors to be expressed at significantly increased or decreased levels in the experiment.

Through the My Papers shopping cart feature, the EDR allows scientists to compare gene expression results from different experiments. Publications with the desired experimental data can be added to the My Papers cart by checking the box beside any paper (see Microarray Papers workflow) or by clicking on the shopping cart icon located next to each paper in the gene record (see section on gene pages). Clicking on the shopping cart icon at the top of the web page allows end users to view the selected papers and to compare gene expression data by clicking the Compare Genes button. A table listing all genes from the selected papers contains columns for the Entrez gene identifier, gene symbol, gene name, and PubMed Identifier (PMID) number for each paper (Figure 2). By default, the genes are sorted by those with the greatest number of hits. However, the columns for gene identification, symbol, and name can be sorted. Up and down arrows for each gene in each paper show whether the gene's relative expression was significantly increased or decreased for a given experiment (Figure 2). By hovering over the PMID number, a pop-up pane will appear and provide general publication information (eg, author names, article title, where the article was published, and PubMed identifier).

Figure 2.

My Papers and microarray gene comparison features. Clicking the My Papers shopping cart icon (top right of any EDR page) shows the list of papers that have been saved to the cart. By clicking the Compare Genes button, a table showing a comparison of which genes were significantly overexpressed (green arrows) or underexpressed (red arrows) in each selected experiment can be viewed. Each publication included in the table is represented by its PMID number, with additional data available as end users hover their mouse over the PMID. If a publication has more than one gene expression experiment, each experiment will be shown as a unique entry in the publication list.

Use cases

We next demonstrate the utility of EDR as a research tool in a series of two experimental use cases that represent how EDR can be used to visualize the effects of different experimental conditions on gene expression, determine how gene expression differs in various layers of the uterus, and generate new scientific hypotheses.

Use case 1

In this use case, the user has a specific gene in which he is interested, phosphatase and tensin homolog (PTEN), which is the most frequently mutated gene in estrogen-dependent endometrioid endometrial carcinomas (14). PTEN acts as a tumor suppressor gene and has both lipid and protein phosphatase activities with each serving different functions. PTEN's protein phosphatase activity has been shown to inhibit focal adhesion formation, cell spread and migration, and growth factor-stimulated MAPK signaling (15). After performing a gene symbol search for PTEN, an end user can then click on the link to the PTEN gene record and then the Phase Graphs link to view 41 reproduction-relevant GEO Profiles for PTEN. These profiles range from a time course of progesterone effects on the uterus to a model of decidualization and menstruation. Also, a researcher can use the uterine-specific PubMed gene search to find 193 papers relevant to PTEN expression in the uterus. Lastly, three microarray papers have been identified by expert curators as uterine-relevant scientific publications. By adding these papers to the shopping cart, a researcher can compare gene expression results from these publications. This comparison indicates that PTEN gene expression was up-regulated in the placentas of women who underwent vaginal deliveries and also in tamoxifen-treated human endometrium (16, 17). This comparison also shows that PTEN gene expression was down-regulated in the eutopic endometrium of women with deep endometriosis and in tamoxifen- and estrogen-treated endometrial tissue (17, 18). Likewise, cyclin A2 was up-regulated in the eutopic endometrium of women with deep endometriosis and in tamoxifen- and estrogen-treated endometrial tissue and down-regulated in tamoxifen-treated endometrial tissue.

Use case 2

In this use case, a user wants to compare the gene expression of K-ras and B-raf, which have been shown to be involved in endometrial oncogenesis. Approximately 10–30% of endometrioid carcinomas exhibit K-ras proto-oncogene mutations (19). Studies have shown that the K-ras gain of function may represent an early event in endometrioid-type tumorigenesis (20, 21) and is typically associated with increased proliferation, transformation, and cell survival. In contrast, only a few studies have reported B-raf mutations in patients with endometrial cancer. After performing a gene symbol search for K-ras, an end user can then click on the link to the mouse K-ras gene record and find that three microarray papers have been identified by expert curators as uterine-relevant scientific publications, and these papers can be added to the shopping cart for gene expression comparison. Similarly, an end user can perform a search for mouse B-raf and add those microarray papers to the shopping cart. The end user can then compare the relative gene expression from all the papers added to the cart with experiments in which K-ras and B-raf were differentially expressed. The most common genes with altered gene expression were CCAAT/enhancer-binding protein, δ (Cebpd), cytidine 5′-triphosphate synthase (Ctps), dual specificity phosphatase 1 (Dusp1), early growth response 1 (Egr1), Hexokinase 2 (Hk2), jun-B oncogene (Junb), kruppel-like factor 2 (Klf2), suppressor of cytokine signaling 3 (Socs3), tubulin, β 6 (Tubb6), and wingless-related mouse mammary tumor virus integration site 4 (Wnt4). Interestingly, when B-raf was up-regulated, Cebpd, Dusp1, Hk2, Klf2, and Socs3 were down-regulated in cervical and skin tissues from K14E6 transgenic mice (22). However Cebpd, Dusp1, Hk2, Klf2, and Socs3 were always up-regulated in the estrogen-treated mouse uterus and blastocyst activation in mice when K-ras was up-regulated (23, 24). Hk2 was also up-regulated when K-ras was down-regulated in uterine receptivity and embryo implantation of mice (25). Taken together, these two Use Cases illustrate how an end user can extract new information about gene expression in various systems to generate testable hypotheses on dysregulation of genes and, by extension, cellular pathways that may underlie one of more distinct experimental conditions or phenotypes.

Discussion

The EDR is a centralized source for a diverse amount of information about genes that are regulated in the uterus and is also a complement to the other Specialized Cooperative Centers Program In Reproduction (SCCPIR) resources, such as the OKdb, which provides services for the ovarian research community (9) and www.sccpir.org. It also serves as a complement to the Endometrial Data Base (EDB), where one can find thousands of data points from hundreds of endometrial-relevant publications that have been categorized by Natural Cycles, Contraception, Endometriosis, Endometrial Cancer, Preeclamsia, and multiple other functional classifications. Whereas the current version of EDR focuses mainly on gene expression analyses, an important next step will be to identify all potential linkages with these EDB category data points and provide links within the EDR to data on uterine pathology information on endometriosis, endometrial cancer, in vitro models, etc., Currently, links to SCCPIR, OKdb, and EDB are located under Links on the EDR home page. Altogether, the EDR is as a continuously evolving one-stop shop for data on all genes in humans, mice, and rats. It also aggregates rich gene annotations from BioGPS, GenBank, Homologene, MGI, OMIM, Human Protein Atlas, UCSC Genome Browser, and Unigene to provide a more complete picture of gene expression/regulation that occurs within the uterus. In fact, within the last year, the EDR site has had more than 900 visits, more than 7300 page views, and an average visit duration of 5.5 min.

Future versions of the resource will be enhanced by the integration of endometrium and uterine-specific cistromic studies from chromatin immunoprecitation-chip and chromatin immunoprecitation-seq data. By including these cistromic studies, one can explore pathways and the genes associated with the proteins, miRNAs, and other biological entities within the pathways that are expressed in the uterus. Additional features will also include the incorporation of complex network analysis and visualization tools such as Cytoscape, a bioinformatics software for visualizing molecular interaction networks and biological pathways and integrating these networks with pathway annotations, heat maps, etc. Annotations on clinical trials and animal models will also be included. The EDR has long been a rich resource for basic research data; however, it had not previously incorporated information from clinical trials. Recently, work has begun to extract uterus and reproduction-relevant clinical trials data from ClinicalTrials.gov. Key data such as study title, principal investigator name, phase, intervention, etc., with links back to the national repository website will be incorporated into the EDR and, where appropriate, incorporated into human gene EDR pages. Integration of clinical trials information will be key for linking the advancements in molecular biology with clinical trials. Also the addition of an Animal Models section will allow scientists to review phenotypes of uterine gene knockout, knock-in, conditional knockout, or other genetically altered animal models. These phenotype records will be linked with gene annotation and microarray publication records within the EDR. Lastly, analyses of GEO datasets by users can be burdensome, and, in order to alleviate these challenges, an important next step for EDR will be the incorporation of automated analyses of GEO array datasets. A pipeline tool such as Gene Pattern or Galaxy could be used to process the data using standard parameters.