Beta-Catenin stabilizes cyclooxygenase-2 mRNA by interacting with AU-rich elements of 3'-UTR

Abstract

Cyclooxygenase-2 (COX-2) mRNA is induced in the majority of human colorectal carcinomas. Transcriptional regulation plays a key role in COX-2 expression in human colon carcinoma cells, but post-transcriptional regulation of its mRNA is also critical for tumorigenesis. Expression of COX-2 mRNA is regulated by various cytokines, growth factors and other signals. beta-Catenin, a key transcription factor in the Wnt signal pathway, activates transcription of COX-2. Here we found that COX-2 mRNA was also substantially stabilized by activating beta-catenin in NIH3T3 and 293T cells. We identified the beta-catenin-responsive element in the proximal region of the COX-2 3'-untranslated region (3'-UTR) and showed that beta-catenin interacted with AU-rich elements (ARE) of 3'-UTR in vitro and in vivo. Interestingly, beta-catenin induced the cytoplasmic localization of the RNA stabilizing factor, HuR, which may bind to beta-catenin in an RNA-mediated complex and facilitate beta-catenin-dependent stabilization of COX-2 mRNA. Taken together, we provided evidences for beta-catenin as an RNA-binding factor and a regulator of stabilization of COX-2 mRNA.

Figure 1

Stabilization of COX-2 mRNA by β-catenin. (A) Real-time PCR analysis of the COX-2 UTR. Cells were co-transfected with the pBBB vector containing β-globin fused to the 3′-UTR of COX-2 mRNA (pBBB^{COX-2 UTR}) and either empty vector or the S37A β-catenin expression vector. Gene expression was induced by 20% serum, and total cytoplasmic RNA was isolated at different times after serum addition. mRNA stability was determined by measuring β-globin mRNA levels by real-time PCR. (B) Western blot analysis. Cells were treated with KCl, LiCl (20 mM each) and MG132 (20 μM), respectively, and the expression of endogenous β-catenin and HuR was determined by western blotting. Alpha-tubulin served as a loading control. (C) The stabilities of pBBB^ΔUTR and pBBB^{COX-2 UTR} in cells as (B) were examined by real-time PCR. The data shown are means ± SE (n = 3), and are representative of three independent experiments. (D) RT–PCR analysis of the endogenous COX-2 mRNA. 293T cells were co-transfected with empty vector (Control), the S37A β-catenin expression vector or LiCl treatment. Actinomycin D (ActD, 10 μg/ml) was added to the culture medium at time 0, and total RNA was isolated at the indicated times, and the amount of COX-2 mRNA was analyzed by RT–PCR. GAPDH served as a loading control.

Figure 2

Requirement of the proximal region of COX-2 3′-UTR for β-catenin-dependent stabilization. (A) The reporter construct containing 1455 nt of the COX-2 3′-UTR (connecting line) together with the luciferase coding region (filled box). AU-rich elements are indicated as vertical lines. (B) Luciferase assay. Cells were transfected with various luciferase reporters. Luciferase activities were measured after treatment with LiCl or KCl for 24 h. Three independent experiments were performed. (C) Schematic representation of the locations of the F1, F2 and F3 regions. Vertical lines represent AREs. (D) In vitro RNA degradation assays to identify the β-catenin-responsive element in the COX-2 3′-UTR. [α-³²P]-labeled RNA substrates were incubated with cytoplasmic extracts from either vector or β-catenin-expressing NIH3T3 cells, and the reactions were stopped by adding stop buffer at the indicated times. Processed RNA was resolved on a 7 M urea/5% acrylamide gel and visualized by autoradiography. (E) In vitro analysis of mRNA decay upon MG132 treatment. Labeled F1 mRNA was incubated with cytoplasmic extracts from either DMSO (Control) or MG132-treated NIH3T3 cells, and decay was analyzed as in (D). (F) Stabilization of the F1 UTR by LiCl treatment. Labeled F1 mRNA was incubated with cytoplasmic extracts from either KCl (Control) or LiCl-treated NIH3T3 cells, and degradation was examined as in (D).

Figure 3

Mapping of the β-catenin-dependent stabilizing element in the F1 UTR. (A) Schematic representation of the different F1 UTR probes. ARE sequences are indicated as boxes and mutant sequences are shown by underlining and in italics. (B) RNA substrates were incubated with cytoplasmic extracts from NIH3T3 cells, and the reactions were stopped by adding stop buffer at the indicated times. Processed RNA was resolved on a 7 M urea/5% acrylamide gel and visualized by autoradiography. (C) RNA substrates were incubated with cytoplasmic extracts from either vector (control) or β-catenin-expressing NIH3T3 cells. mRNA degradation was examined as in (B).

Figure 4

Specific association of β-catenin with the F1 region of COX-2 mRNA. (A) Specificity of β-catenin binding to RNA. [α-³²P]-labeled F1 RNA (80 pM) was co-incubated with recombinant β-catenin supplemented with unlabeled F1 RNA (F1) or non-specific NC RNA (10-, 50-, 200-fold each). Complexes are indicated as the arrow. (B) RNA-EMSAs. [α-³²P]-labeled F1 RNA was incubated with C-terminal domain of β-catenin (CTD), HuR, Arm 1-12, T-cell factor (TCF), or GST (200 nM each). (C) The RNA-EMSAs were performed with recombinant Arm 1-12 protein (2.5, 5, 10, 25, 50, 75, 100 and 200 nM), and the complexes are labeled 1–3. The HuR–F1 complex as control is indicated by an open arrowhead. (D) Specificity of Arm 1-12 binding to RNA. Labeled F1 RNA was co-incubated with Arm 1-12 supplemented with unlabeled F1 RNA (F1), aptamer (Apt; 10-, 50-, 200-fold each) or F2 RNA (200-fold). Following incubation, RNA–protein complexes were resolved on 5% native gels and visualized by autoradiography. Input indicates RNA only.

Figure 5

In vivo interaction of COX-2 mRNA with β-catenin. (A) RNA immunoprecipitation assay. NIH3T3 cells co-transfected with various reporters (ΔUTR, Full, F1, ΔF1 as in Figure 2A and Antisense of F1, F1 AS). After formaldehyde fixation, immunoprecipitations were performed with normal IgG or anti-β-catenin antibody. Bound RNA was extracted from the immune complexes and analyzed by RT–PCR. (B) Supershift assay. Labeled F1 RNA was incubated with cytoplasmic extracts of NIH3T3 cells containing either empty vector (control) or β-catenin expression in the presence of normal IgG, anti-HuR, or anti-β-catenin antibodies and analyzed by 5% native gels. The supershifted bands of β-catenin (arrow) as well as of HuR (open arrowhead) are indicated. (C) Cytoplasmic (CE) and nuclear (NE) extracts were prepared from either normal (Mock) or β-catenin-overexpressing (β-cat) NIH3T3 cells and analyzed by western blotting.

Figure 6

RNA-mediated interaction between β-catenin and HuR. (A) Whole cell (WCE), cytoplasmic (CE) and nuclear extracts (NE) were prepared from HT-29 colon cancer cells and the distributions of β-catenin and HuR were examined by western blotting. (B) Immunoprecipitation assays were carried out using cytoplasmic and nuclear extracts of HT-29 cells, and either normal IgG, or anti-HuR and anti-β-catenin antibodies, followed by western blotting with the indicated antibodies. (C) IP reactions on extracts of NIH3T3 cells were performed without further treatment (control), or in the presence of heparin or RNases. IP complexes were identified by western blotting.