Gene ONC TSG orUND br e
Gene (ONC, TSG orUND)
e One of the polyadenylation sites in the longest variant transcript derived from MAX gene (NM_145113.2) does not have a close PAS (upstream region). d Analysis considering alternative CS mapped in all transcript variants (for each gene) from samples of 22 normal human tissues (You et al., 2015, reference Liu et al., 2016). c Analysis using this database included only cleavage/polyadenylation sites identified from human peripheral blood samples (Müller et al., 2014, reference Bertero et al., 2013). b Distance (in bp) between the PAS and CS calculated according their positions indicated in the NCBI interface. a Constitutive PAS and CS indicated in NCBI database for the longest variant transcript (in bp) derived from each cancer predisposition gene. Alternative Polyadenylation Sites Database; NI, not identified. ONC, oncogene; TSG, tumor suppressor gene; UND, “undetermined function” gene; PAS, polyadenylation signals; CS, cleavage sites; bp, number of APTSTAT3-9R pairs; APADB, Alternative Polyadenylation Database; APASdb,
Comparison between the frequencies of human polyadenylation signal hex-amers identified by previous genomic studies and in cancer predisposition genes analyzed in the current study.
Hexamer Frequency, % (ranking)
Genomic Genomic Cancer Oncogenes; Tumor
data; data; predisposition N = 12d suppressor
reported reported by genes; N = 96c
by Tian Beaudoing
NI, not identified.
a Frequency of hexamer/hexanucleotide types identified as polyadenylation signals (PAS) in the human genome as previously reported by Tian et al. (2005). The method developed by Beaudoing et al. (2000) was applied to the human genome sequences located 1 to 40 nucleotides upstream of polyadenylation sites to detect hexamers that may function as PAS. b Frequency of hexamer types identified as PAS in the human genome using a method described by Beaudoing et al. (2000).
c Frequency considering PAS indicated in NCBI for 96 of 117 genes included in this study (the remaining genes have no description of PAS in this database). Each hexamer was counted as a unit, including the case of transcripts with more than one PAS sequence and transcripts derived from RET gene that have dif-ferent hexamers (AAUAAA and AUUAAA) in their specific isoforms.
d Frequency considering PAS indicated in NCBI for 12 of 17 oncogenes in-cluded in this study. Two putative oncogenes were included in this group (PALLD and SHOC2).
e Frequency considering PAS indicated in NCBI for 69 of 81 tumor suppressor genes (TSG) included in this study. Eight putative TSG were included in this group (see Materials and Methods).
f Frequency comparison of most common hexamers functioning as PAS (AAUAAA and AUUAAA) between oncogenes and tumor suppressor genes: P = 0.78 (Fisher's exact test).
experimentally validated and predicted interactions, respectively, were identified as regulators of both gene groups, and the number of inter-sections among these sets of miRNAs was defined (Fig. S1). Considering the filtering criteria employed in these in silico analyses, only the FANCB gene had no miRNA regulation reported in our data. We then performed an over-representation analysis to identify potentially on-cogenic or tumor suppressor miRNAs. Fig. 2 shows the regulatory networks obtained by combining analysis of both validated and pre-dicted data, indicating miRNAs and miRNA families significantly overrepresented (P < 0.01) as regulators of tumor suppressor genes and oncogenes. As seen in these interactomes, statistically significant data were derived mainly from experimental sources (Fig. 2A and B). Interestingly, experimental data also represented the majority of col-lected miRNA-target gene interactions compared to the data only pre-dicted by computational tools (Fig. 2C and D).
Lastly, we observed that the mir-192 family was the most sig-nificantly overrepresented among tumor suppressor genes (P = 0.002), which could suggest an oncogenic function. This miRNA family has experimentally validated binding sites in the 3′UTR of 20 tumor sup-pressor genes included in this study (Fig. 2B). Among them, there are central genes in human DNA repair pathways such as BRCA1, BRCA2,