Regulation of cyclooxygenase-2 expression by the translational silencer TIA-1

Abstract

The cyclooxygenase-2 (COX-2) enzyme catalyzes the rate-limiting step of prostaglandin formation in inflammatory states, and COX-2 overexpression plays a key role in carcinogenesis. To understand the mechanisms regulating COX-2 expression, we examined its posttranscriptional regulation mediated through the AU-rich element (ARE) within the COX-2 mRNA 3'-untranslated region (3'UTR). RNA binding studies, performed to identify ARE-binding regulatory factors, demonstrated binding of the translational repressor protein TIA-1 to COX-2 mRNA. The significance of TIA-1-mediated regulation of COX-2 expression was observed in TIA-1 null fibroblasts that produced significantly more COX-2 protein than wild-type fibroblasts. However, TIA-1 deficiency did not alter COX-2 transcription or mRNA turnover. Colon cancer cells demonstrated to overexpress COX-2 through increased polysome association with COX-2 mRNA also showed defective TIA-1 binding both in vitro and in vivo. These findings implicate that TIA-1 functions as a translational silencer of COX-2 expression and support the hypothesis that dysregulated RNA-binding of TIA-1 promotes COX-2 expression in neoplasia.

Figure 1.

TIA-1 binds the COX-2 ARE. (A) Biotin-labeled RNA containing the COX-2 ARE, GM-CSF ARE, or control CAT RNA were incubated with in vitro translated ³⁵S-labeled TIA-1 protein. TIA-1 bound to immobilized RNA was detected by electrophoresis on SDS-PAGE. Total TIA-1 protein used per reaction is shown in the first lane and nonspecific binding is shown in the second lane. (B) Cytoplasmic HeLa lysate was incubated with ³²P-labeled COX-2 ARE and bound proteins were cross-linked to RNA by UV-light irradiation. Reactions were immunoprecipitated using anti-TIA-1 antibody (α-TIA-1) or IgG. The -Ab lane shows the bound proteins in 1/10 of the cross-linking reaction before immunoprecipitation. Molecular weight standards (kD) are listed on the left.

Figure 2.

TIA-1 inhibits COX-2 expression. (A) MEFs derived from wild-type or TIA-1^−/− mice were examined for COX-2 protein expression by Western blot. Molecular weight standards (kD) are listed on the left. (B) COX activity measured by PGE₂ production in wild-type (open bars) and TIA-1^−/− (filled bars) MEFs treated with carrier or NS-398. (C) Western blot of COX-2 protein in TIA-1^−/− MEFs stably transfected with a TIA-1 expression construct or vector.

Figure 3.

TIA-1 does not alter rapid COX-2 mRNA decay. (A) Steady-state COX-2 mRNA in wild-type and TIA-1^−/− MEFs levels were detected by RNase protection assay. (B) COX-2 transcription was determined using a luciferase reporter construct containing either the COX-2 promoter (filled bars) or vector (open bars) cotransfected with pSV-βgal. (C) COX-2 and c-myc mRNA half-life experiments were initiated by adding 5 μg/ml ActD to MEFs for the indicated times and decay was analyzed by RNase protection assay. 28S RNA is shown as a control for RNA loading.

Figure 4.

COX-2 polysome profile in colon cancer cells. HT29 and LoVo cytoplasmic lysates were fractionated on a continuous 15–40% sucrose gradient. Polysome profiles were obtained by running RNA samples from each fraction on an agarose gel to identify the 40S and 60S ribosomal subunits along with 80S and polysomes (bottom panel). The distribution of COX-2 mRNA from HT29 cells (filled circles) and LoVo cells (open triangles) was detected by RNase protection assay of gradient fractions.

Figure 5.

Deficient binding of TIA-1 occurs in HT29 cells. (A) Cytoplasmic protein complexes from control HeLa, HT29, and LoVo cells bound to the ³²P-labeled COX-2 ARE were UV-cross-linked and immunoprecipitated using TIA-1 antibody. (B) RT-PCR (left panel) and Western blot (right panel) analysis of TIA-1 and β-actin from control HeLa, HT29, and LoVo cells. (C) In vivo cross-linking of TIA-1 protein to poly(A) mRNA was examined in HT29 and LoVo cells.