Leveraging Online Resources to Prioritize Candidate Genes for Functional Analyses: Using the Fetal Testis as a Test Case

Abstract

With each new microarray or RNA-seq experiment, massive quantities of transcriptomic information are generated with the purpose to produce a list of candidate genes for functional analyses. Yet an effective strategy remains elusive to prioritize the genes on these candidate lists. In this review, we outline a prioritizing strategy by taking a step back from the bench and leveraging the rich range of public databases. This in silico approach provides an economical, less biased, and more effective solution. We discuss the publicly available online resources that can be used to answer a range of questions about a gene. Is the gene of interest expressed in the system of interest (using expression databases)? Where else is this gene expressed (using added-value transcriptomic resources)? What pathways and processes is the gene involved in (using enriched gene pathway analysis and mouse knockout databases)? Is this gene correlated with human diseases (using human disease variant databases)? Using mouse fetal testis as an example, our strategies identified 298 genes annotated as expressed in the fetal testis. We cross-referenced these genes to existing microarray data and narrowed the list down to cell-type-specific candidates (35 for Sertoli cells, 11 for Leydig cells, and 25 for germ cells). Our strategies can be customized so that they allow researchers to effectively and confidently prioritize genes for functional analysis.

Fig. 1:. Overview and examples of gene entries in the Eurexpress, Gene Expression Atlas, and OMIM databases.

(A) In situ hybridization images of 14.5 dpc embryos are compiled in the Eurexpress Transcriptome Atlas Database for Mouse Embryo (http://www.eurexpress.org). Expression observed in an image is annotated and encoded as text that can be queried. The more general anatomical terms branch into more specific terms. For example, the nested terms that describe the testis are displayed. (B) The Eurexpress viewer scrolls through sections in a “virtual embryo environment” and zoom into regions as similar to a digital microscope. For the gene Star, the testis can be visualized in the 6^th section. The testis region is demarcated by a dashed line; a, adrenal; k, kidney; m, mesonephros. (C, D) The Gene Expression Atlas houses easy to access transcriptomic data for a wide variety of species from all major consortia and sequencing efforts. For the gene Sox9, detailed information are available on expression in the (C) mouse and (D) human. Data is presented in a tabular format that can be sorted using a number of parameters (graphics shown here are part of the full table). Selecting data from a project in the table, such as 16.5 dpc 49 FANTOM (C, in mouse) or 68 FANTOM (D, in human), highlights the organs (in our case, the testis) that are included in that dataset in purple on the interactive body-map. (E) The OMIM database (Online Mendelian Inheritance in Man) catalogues human disease variants. Entries can be either by phenotype or gene. In the example of a search for the gene NR5A1 (also known as SF1), there are many alternative gene symbols that have been used historically; however, the disease information is catalogued under the HUGO approved symbol (NR5A1). Information on the location of the gene, all phenotypes caused by mutations in NR5A1 are listed in a table with ID numbers for each phenotype and additional information on inheritance. The entry number and mapping key encode additional information about the disorder.

Fig. 2:. Sertoli cell-expressed genes identified from “testis” annotated entries in the Eurexpress database.

(A) The list of 295 entries annotated as testis-expressed in the Eurexpress database were compared to 491 Sertoli cell expressed genes from a microarray of sorted cells at 13.5 dpc (Jameson et. al., 2012). 35 genes were represented in both sources (see Supp Table 2). (B) Out of these 35 genes, 5 known Sertoli cell marker genes were identified: Amh, Col9a3, Etv5, Ptgds, and Tyro3. 16 novel Sertoli cell candidate genes with a quality ranking of 1 or 2 were annotated in the Eurexppress database: Adamts16, Atp8b2, Ctnna2, Cxadr, Dsp, Gjb2. Gstm7, Lsr, Mpr1, Pgepep1, Slc20a1, Slc38a1, Tesc, Tmcc3, Tpd52a and Trank1. The testis region is demarcated by a dashed line. a: adrenal; k: kidney; m: mesonephros. Scale bar: 0.5 mm.

Fig. 3:. Germ cell-expressed genes identified from “testis” annotated entries in the Eurexpress database.

(A) The list of 295 entries annotated as testis-expressed in the Eurexpress database were compared to 433 germ cell-expressed genes from a microarray of sorted cells at 13.5 dpc (Jameson et. al., 2012). 25 genes were represented in both sources (see Supp Table 2). (B) These genes included the known germ cell markers Sox2 and Dppa3. The germ cell marker Dazl was also included as a positive control. In the Eurexpress database, there were 10 novel germ cell genes that were also represented in the microarray: Ap3b2, Bub1, Cdkrap1, Irf1, Liph, Mybbp1a, Rcc2, Rcl1, Trip13 and Uchl1. The testis region is demarcated by a dashed line. a: adrenal; k: kidney; m: mesonephros. Scale bar: 0.5 mm.

Fig. 4:. Interstitial and fetal Leydig cell-expressed genes identified from “testis” annotated entries in the Eurexpress database.

(A) The list of 295 entries annotated as testis-expressed in the Eurexpress database were compared to 130 interstitial cell-expressed genes from a microarray of sorted cells at 13.5 dpc (Jameson et. al., 2012). 9 genes were represented in both sources (see Supp Table 2). (C) Among these genes, 4 known interstitial/Leydig cell marker genes were identified: Cyp11a1, Prlr, Star and Tacr3. (C) high quality in situ hybridization images were available for 5 of the 9 overlapping interstitial genes: Clca1, Cxcl12, Itga9, Nrg1 and Sostdc1. (D) 11 genes were represented in both the Leydig cell microarray and the Eurexpress database (see Supp Table 2); (E) high quality in situ hybridization images were available for 8 genes: Fads1, Gramd1b, Hsd17b7, Prkar2b, Ren1, Scarb1, Sct and Stc1. . The testis region is demarcated by a dashed line. a: adrenal; k: kidney; m: mesonephros. Scale bar: 0.5 mm.