FARE-CAFE case studies
Case study 1: Impact of TFs on FG induced cancer progressionIn this study, we collected FGs with the following features, (i) the 5’ gene transcribed by TFs (TF information from the PAZAR database and activity information from Metacore database) (Figure 1a), (ii) the 3’ gene is oncogenic (from NCG 4.0 (1)), (iii) the 3’ gene expression is controlled by transcription repressor (from Metacore database), and (iv) the 3’ gene is not targeted by miRNAs (Figure 1b). Figure 1c illustrates the FG structure. A total of 14 FGs satisfy those constrains (Table 1).
Figure 1. (a) The wild-type 5’ gene is activated by experimentally reported transcription enhancers (Exp. TEs) at the promoter region and with experimentally reported targeted miRNAs at the 3’UTR region, (b) The wild-type 3’ oncogene inhibited by experimentally reported transcription repressor (Exp. TR) at the promoter region and not targeted by miRNA at the 3’ UTR region, (c) The FG composed of 5’ gene transcribed by TEs at the promoter region, and coding region forms from the fusion of the 5’ gene and 3’ oncogene which is not targeted by miRNA at the 3’ UTR region. Bold-faced and italic fonts denote the wild-type 5’ gene and 3’ gene respectively.
Observations and analysis
- The FG is transcribed by several TFs, which bind the 5’ gene promoter region.
- After translocation, the 3’ oncogene lost its TF binding sites for the transcription repressor and fused with the 5’ gene.
- The 3’ partner gene of the FG is not targeted by miRNA, so the 3’ gene-specific transcription repressor or miRNA does not suppress this kind of FG.
- We found that the ETS1 (from Metacore) acts as a transcription repressor for the wild-type 3’ gene but a transcription enhancer for two FGs (ETS1 activates the 5’ gene); i.e. PAX5-JAK2 (2; 3) and RUNX1-CBFA2T3 (4; 5). Another TF EBF is found acting on the FG, PAX5-ZNF521 (2). Here, the EBF activates the wild-type 5’ gene PAX5, but represses the wild-type 3’ gene ZNF521 (from Metacore).
Case Study 2: Impact of miRNAs on FG induced cancer progressionFor this study, we collected FGs to satisfy the following two features from miRTarbase and NCG 4.0: (i) the 5’ gene is oncogenic with experimentally verified miRNAs targeted at its 3’ UTR (Figure 2a) and (ii) the 3’ gene is not targeted by any miRNA (Figure 2b). To address feature (ii), we utilized three miRNA target prediction algorithms, i.e. miRDB (6), MiRanda (7) and TargetScan (8) to determine whether the 3’ UTR of the 3’ gene is targeted by any miRNA or not. Then, we compared predicted miRNAs with the 5’ gene’s miRNAs. If any one of the algorithm’s prediction matched with the 5’ gene miRNA, we assigned a score value (S) of one to such prediction, so the maximum score is three. In order to enhance the reliability of the prediction, we queried the predicted miRNAs against the miRCancer database (9) to determine if those miRNAs are recorded or not. MiRCancer is a database, which collected cancer-related miRNAs. There are four possible outcomes, which are summarized in below Table.
Table . The four possible outcomes based on the results of score S and miRCancer
|Score, S||miRNA found in miRCancer||Hypothesis|
|S ≥ 1||No||miRNA has a low possibility of regulating the FG|
|S ≥ 1||Yes||miRNA has high possibility of regulating the FG|
|S = 0||No||miRNA has a very low possibility of regulating the FG|
|S = 0||Yes||miRNA has a low possibility of regulating the FG|
A total of 33 FGs are found which satisfy these constraints (Figure 2c), their detail information are summarized in the Table 2. It is noted that if the 3’ gene is not targeted by any miRNA, and its host FG is activated; then, the FG has a higher chance of exhibiting its oncogenic role.
Observations and analysis
- After chromosomal translocation, all of the collected 33 FGs lost their 3’ UTR regions of the 5’ oncogenes, therefore, they are not regulated by any miRNA.
- The 5’ oncogenes fused with 3’ genes, where the 3’ genes are not targeted by any miRNA.
- A majority (more than 95%) of the 3’ genes’ predicted miRNAs do not have common miRNAs with the 5’ genes. The remainder of the 5% are overlapping events.
- Our predictions are further supported by the findings that more than 90% of the 3’ genes’ predicted miRNAs do not have any record of miRCancer. The rest of 10% events can be found in mirCancer.
Case study 3: Impact of functional elements on FG induced cancer progressionWe studied the impact of functional elements, i.e. domains, DDIs and PPIs, of the fusion proteins in acute myeloid leukemia (AML), chronic myeloid leukemia (CML) and Ewing’s sarcoma cancer. A total of 38 fusion proteins with their 197 isoforms are associated with the three selected cancer types. The functional elements for those fusion proteins and their wild-type 5’ and 3’ partner proteins are summarized in Table 3.
Observations and analysis
- After translocation, fusion protein is formed which is composed of a smaller group of domains compared to the union of its 5’ and 3’ wild-type genes.
- Since the fusion protein composes of a smaller set of domains, certain biologically significant DDIs and DDI-mediated PPIs are disrupted.
- For example, the BCR-ABL fusion protein (causing AML and CML) interacts with the GAB2 (contains PH domain) and H-Ras (contains Ras domain) proteins through DDI with the RhoGEF domain (belongs to BCR). These PPIs play a key role in the ‘CML pathway’ (KEGG ID: hsa05220). Also, BCR-ABL interacts with the CRK and STAT5A proteins (with SH2 domain in CRK and STAT5A) through DDI with the C2 domain (belongs to BCR). These PPIs also play a role in the ‘CML pathway’. However, certain BCR-ABL fusion isoforms (10; 11) lost the RhoGEF and C2 domains; hence, disrupting the ‘CML pathway’.
- For Ewing’s sarcoma cancer, the presence of the RRM_6 domain in protein EWSR1 interacts with the SF3B4 and SRSF5 proteins; this interaction is listed in the ‘Spliceosome pathway’ (hsa03040). The PPI is mediated by the DDI between domains RRM_6 and RRM_1 (belongs to SF3B4 and SRSF5). Also, the presence of the zf-RanBP domain in the EWSR1 interacts with the RAD23A and HERPUD1 proteins. This interaction is recorded in the ‘Protein processing in endoplasmic reticulum pathway’ (hsa04141).
- The PPI is mediated by the DDI between domains zf-RanBP and ubiquitin (belongs to RAD23A and HERPUD1). After translocation, EWSR1 lost both RRM_6 and zf-RanBP domains and fused with FLI1 to form the oncogenic fusion protein EWS-FLI1; hence, causing Ewing’s sarcoma (for more details, see ‘Sarcomas’ under the ‘Transcriptional Mis-regulation in Cancer’ (hsa05202)).
- An, O., Pendino, V., D'Antonio, M. et al. (2014) NCG 4.0: the network of cancer genes in the era of massive mutational screenings of cancer genomes. Database. 2014: article ID bau015.
- Wheat,W., Fitzsimmons, D., Lennox, H. et al. (1999) The highly conserved beta-hairpin of the paired DNA-binding domain is required for assembly of Pax-Ets ternary complexes. Mol Cell Biol.;19(3):2231-41.
- Hollenhorst, P.C., Chandler,K.J., Poulsen, R.L. et al. (2009) DNA specificity determinants associate with distinct transcription factor functions. PLoS Genet. 5(12):1-12.
- Mao, S., Frank,R.C, Zhang,J. et al. (1999) Functional and physical interactions between AML1 proteins and an ETS protein, MEF: implications for the pathogenesis of t(8;21)-positive leukemias. Mol Cell Biol.19 (5): 3635-44.
- Ajore,R., Kumar,P., Dhanda,R.S. et al. (2012) The leukemia associated nuclear corepressor ETO homologue genes MTG16 and MTGR1 are regulated differently in hematopoietic cells. BMC Mol Biol.13(11):1-16.
- Wong,N., and Wang, X. (2015) miRDB: an online resource for microRNA target prediction and functional annotations. Nucleic Acids Research. 43(D1):D146-152.
- Betel, D., Wilson, M., Gabow, A. et al, (2008) The microRNA.org resource: targets and expression. Nucleic Acids Research. 36(Database Issue): D149-53.
- García, D. M., Baek, D., Shin,C. et al. (2011) Weak Seed-Pairing Stability and High Target-Site Abundance Decrease the Proficiency of lsy-6 and Other miRNAs. Nat Struct Mol Biol., 18:1139-1146
- Xie, B., Ding, Q., Han, H. and Wu,D. (2013) miRCancer: a microRNA-cancer association database constructed by text mining on literature. Bioinformatics. 29(5):638-644.
- Iwata, S., Mizutani, S., Nakazawa, S. and Yata, J. (1994) Heterogeneity of the breakpoint in the ABL gene in cases with BCR/ABL transcript lacking ABL exon a2. Leukemia. 8(10):1696-702.
- Van der Velden, V.H., Beverloo, H. B., Hoogeveen, P. G. et al. (2007) A novel BCR-ABL fusion transcript (e18a2) in a child with chronic myeloid leukemia. Leukemia. 21: 833–835.