Post-transcriptional regulation of connexin43 in H-Ras-transformed cells

Abstract

Connexin43 (Cx43) expression is lost in cancer cells and many studies have reported that Cx43 is a tumor suppressor gene. Paradoxically, in a cellular NIH3T3 model, we have previously shown that Ha-Ras-mediated oncogenic transformation results in increased Cx43 expression. Although the examination of transcriptional regulation revealed essential regulatory elements, it could not solve this paradox. Here we studied post-transcriptional regulation of Cx43 expression in cancer using the same model in search of novel gene regulatory elements. Upon Ras transformation, both Cx43 mRNA stability and translation efficiency were increased. We investigated the role of Cx43 mRNA 3' and 5'Untranslated regions (UTRs) and found an opposing effect; a 5'UTR-driven positive regulation is observed in Ras-transformed cells (NIH-3T3(Ras)), while the 3'UTR is active only in normal NIH-3T3(Neo) cells and completely silenced in NIH-3T3(Ras) cells. Most importantly, we identified a previously unknown regulatory element within the 3'UTR, named S1516, which accounts for this 3'UTR-mediated regulation. We also examined the effect of other oncogenes and found that Ras- and Src-transformed cells show a different Cx43 UTRs post-transcriptional regulation than ErbB2-transformed cells, suggesting distinct regulatory pathways. Next, we detected different patterns of S1516 RNA-protein complexes in NIH-3T3(Neo) compared to NIH-3T3(Ras) cells. A proteomic approach identified most of the S1516-binding proteins as factors involved in post-transcriptional regulation. Building on our new findings, we propose a model to explain the discrepancy between the Cx43 expression in Ras-transformed NIH3T3 cells and the data in clinical specimens.

Figure 1. H-Ras regulates Connexin43 expression at post-transcriptional levels.

a) RNA stability measured by the rate of Cx43 mRNA decay following inhibition of de novo transcription by actinomycin D. NIH-3T3^Neo and NIH-3T3^Ras cells were treated with actinomycin D (15 µg/ml), harvested at various time points and Cx43 and GAPDH mRNA levels were determined by RT−PCR. The values of Cx43/GAPDH mRNAs ratios are reported over the duration of treatment. The lines are based on a linear regression and indicate that there is a significant slope deviation from the zero for Neo (r = 0.84; P  = 0.0095) versus Ras (r  = 0.0012; P = 0.94). b) Translation efficiency of the Cx43 mRNA. Polysomal and subpolysomal fractions were extracted from NIH-3T3^Neo (Neo) and NIH-3T3^Ras (Ras) cells using a sucrose density gradient and the relative levels of Cx43 and GAPDH mRNAs in these fractions were determined by RT-PCR. The results are reported as a Cx43/GAPDH mRNA ratio in different fractions. The lines are based on a linear regression and indicate that there is a significant slope deviation from the zero for Ras (r  = 0.014; P  = 0.84) versus Neo (r  = 0.71; P  = 0.07).

Figure 2. 3′ and 5′UTRs-driven regulation in control versus Ras-overexpressing cells.

a) Description of the different Cx43 mRNA 3′UTR and 5′UTR constructs used in this experiment. All constructs, except the pGL3 basic vector, contain the SV40 promoter (SV40-P) and the Luciferase coding region (pGL3-pr) in addition to the 3′UTR (pGL3-3′UTR) and/or the 5′UTR (pGL3-5′UTR) full length sequences. b) Luciferase assay using Cx43 mRNA 3′UTR and 5′UTR constructs in NIH3T3^Neo cells. The firefly luciferase activities were reported to the Renilla luciferase control values as explained in “Materials and Methods”. The experiments were performed at least three times in quadruplicates (*p<0.05).

Figure 3. Characterization of the Cx43 mRNA 3′UTR regulatory regions in NIH3T3^Neo and NIH3T3^Ras by luciferase assay.

a) Localization of the different Cx43 3′UTR constructs used in this work. All constructs contain the SV40 promoter (SV40-P) and the Luciferase coding region in addition to the pGL3 polyadenylation signal after the 3′UTR segments (not shown in graphics). The position of a putative polyadenylation signal is indicated. The S1516 region is shown in light grey. b, c) Luciferase assay of the first set of 3′UTR constructs used to transfect NIH3T3^Neo and NIH3T3^Ras cells respectively. d, e) Luciferase assay of the second set of 3′UTR constructs used to transfect NIH3T3^Neo and NIH3T3^Ras cells respectively. The firefly luciferase activities were reported to the Renilla luciferase control values as explained in “Materials and Methods”. The experiments were performed at least three times in quadruplicates (*p<0.05, reported to the pGL3-pr construct).

Figure 4. 3′UTR and 5′UTR-driven effects in MCF7 breast cancer cells.

Luciferase assay in MCF7 cells transfected with different constructs including the pGL3 control, the pGL3-Pr, in addition to the full length 3′untranslated region (pGL3-Pr-3′UTR) or the S1516 regulatory element (pGL3-Pr-S1516). The experiments were performed at least three times in quadruplicates (*p<0.05).

Figure 5. 3′UTR and 5′UTR-driven effects in Src and ErbB2-transformed cells. a)

Western blot of the Cx43 expression in protein extracts from NIH3T3^Neo, NIH3T3^Ras, NIH3T3^Src and NIH3T3^ErbB2 cells. An antibody to β-Actin is used as a control for equal loading. b) Luciferase assay using the NIH3T3^Neo, NIH3T3^Src and NIH3T3^ErbB2 cells transfected with various Cx43 mRNA 3′ and 5′UTR constructs. The experiments were performed at least three times in quadruplicates (*p<0.05).

Figure 6. Differential binding of trans-acting factors to regions of the 3′UTR S1516.

a) REMSA using three different riboprobes (R1, R2 and R3) spanning the S1516 region. Binding reactions were performed with the riboprobes as explained in “Materials & Methods” in the presence of protein extracts from NIH3T3^Neo (Neo) and NIH3T3^Ras (Ras) cells. RNA/proteins complexes are indicated by stars. N.S.: non-specific binding. b) Luciferase assay using the NIH3T3^Neo and NIH3T3^Ras cells, transfected with Cx43 mRNA 3′UTR constructs corresponding to the three riboprobes R1, R2 and R3 (i.e. full length S1516 element, R1/2/3), riboprobes 1 and 2 (R1/2) or riboprobe 3 (R3). The experiments were performed at least three times in quadruplicates (*p<0.05). The experiments were performed at least three times in quadruplicates (*p<0.05).

Figure 7. Model of the post-transcriptional regulation of Cx43 expression in normal and transformed cells.

a) In normal cells, the 3′UTR exerts a positive regulation on Cx43 expression. b) Ras-mediated transformation exerts a blockade on a regulatory element (S1516) embedded within the 3′UTR, but releases a positive regulation by the 5′UTR, with a final result of induction of Cx43 expression. c) In human cancer cells, additional oncogenic events inhibit the 5′UTR-mediated regulation, which combined with the S1516-mediated blockade, results in the silencing of Cx43 expression in cancer cells.