Transcriptional induction of cyclooxygenase-2 gene by okadaic acid inhibition of phosphatase activity in human chondrocytes: co-stimulation of AP-1 and CRE nuclear binding proteins.

The involvement of serine/threonine protein phosphatases in signaling pathways that control the expression of the cyclooxygenase-2 (COX-2) gene in human chondrocytes was examined. Okadaic acid (OKA), an inhibitor of protein phosphatases 1 (PP-1) and 2A (PP-2A), induced a delayed, time-dependent increase in the rate of COX-2 gene transcription (runoff assay) resulting in increased steady-state mRNA levels and enzyme synthesis. The latter response was dose dependent over a narrow range of 1-30 nmol/L with declining expression and synthesis of COX-2 at higher concentrations due to cell toxicity. The delayed increase in COX-2 mRNA expression was accompanied by the induction of the proto-oncogenes c-jun, junB, junD, and c-fos (but not FosB or Fra-1). Increased phosphorylation of CREB-1/ATF-1 transcription factors was observed beginning at 4 h and reached a zenith at 8 h. Gel-shift analysis confirmed the up-regulation of AP-1 and CRE nuclear binding proteins, though there was little or no OKA-induced nuclear protein binding to SP-1, AP-2, NF-kappaB or NF-IL-6 regulatory elements. OKA-induced nuclear protein binding to 32P-CRE oligonucleotides was abrogated by a pharmacological inhibitor of protein kinase A (PKA), KT-5720; the latter compound also inhibited OKA-induced COX-2 enzyme synthesis. Calphostin C (CalC), an inhibitor of PKC isoenzymes, had little effect in this regard. Inhibition of 12P-CRE binding was also observed in the presence of an antibody to CREB-binding protein (265-kDa CBP), an integrator and coactivator of cAMP-responsive genes. The binding to 32P-CRE was unaffected in the presence of excess radioinert AP-1 and COX-2 NF-IL-6 oligonucleotides, although a COX-2 CRE-oligo competed very efficiently. 32P-AP-1 consensus sequence binding was unaffected by incubation of chondrocytes with KT-5720 or CalC, but was dramatically diminished by excess radioinert AP-1 and CRE-COX-2 oligos. Supershift analysis in the presence of antibodies to c-Jun, c-Fos, JunD, and JunB suggested that AP-1 complexes were composed of c-Fos, JunB, and possibly c-Jun. OKA has no effect on total cellular PKC activity but caused a delayed time-dependent increase in total PKA activity and synthesis. OKA suppressed the activity of the MAP kinases, ERK1/2 in a time-dependent fashion, suggesting that the Raf-1/MEKK1/MEK1/ERK1,2 cascade was compromised by OKA treatment. By contrast, OKA caused a dramatic increase in SAPK/JNK expression and activity, indicative of an activation of MEKK1/JNKK/SAPK/JNK pathway. OKA stimulated a dose-dependent activation of CAT activity using transfected promoter-CAT constructs harboring the regulatory elements AP-1 (c-jun promoter) and CRE (CRE-tkCAT). We conclude that in primary phenotypically stable human chondrocytes, COX-2 gene expression may be controlled by critical phosphatases that interact with phosphorylation dependent (e.g., MAP kinases:AP-1, PKA:CREB/ATF) signaling pathways. AP-1 and CREB/ATF families of transcription factors may be important substrates for PP-1/PP-2A in human chondrocytes.