Post-translational regulation of COX2 activity by FYN in prostate cancer cells

Abstract

While increased COX2 expression and prostaglandin levels are elevated in human cancers, the mechanisms of COX2 regulation at the post-translational level are unknown. Initial observation that COX2 forms adduct with non-receptor tyrosine kinase FYN, prompted us to study FYN-mediated post-translational regulation of COX2. We found that FYN increased COX2 activity in prostate cancer cells DU145, independent of changes in COX2 or COX1 protein expression levels. We report that FYN phosphorylates human COX2 on Tyr 446, and while corresponding phospho-mimetic COX2 mutation promotes COX2 activity, the phosphorylation blocking mutation prevents FYN-mediated increase in COX2 activity.

Figure 1. Identification of covalent adducts between COX2 and unknown proteins

HMC cells were either left uninfected (-) or infected with AdCOX2 (+) and were incubated either in the absence (-) or presence (+) of COX2 inhibitor NS398. 48h after cells were lysed, subjected to SDS-PAGE, prior to transfer to PVDF membrane and immunoblotted with COX2 antibodies. Equal loading was confirmed by Ponceau S staining as indicated on the left. The nature of COX2 adducts was further uncovered by LC-MS-MS analysis after COX2 was purified by affinity chromatography. Positions of the COX2 band, representing SDS-PAGE resolved not cross-linked monomer COX2, as well as COX2 in complex with FYN, Fibronectin precursor and ELMO1 are shown.

Figure 2. FYN increases COX2 activity independent of changes in COX gene expression

(A) FYN increases cyclooxygenase activity resulting in elevated production of prostaglandins in DU145 cells. DU145 cells infected with either adenovirus encoding GFP (white bars) or FYN (black bars) were homogenized, incubated with 40 μM AA, in the presence of 100% O₂ for 1h, and levels of prostaglandins measured by LC- ESI-MS as described in experimental procedures. Fold change in PG levels from FYN overexpressing cells relative to GFP infected cells is shown. * indicates statistically significant difference in prostaglandin levels in FYN vs. GFP infected cells (n=4). (B) FYN increases activity of COX2 enzyme in vitro. In vitro kinase assays were performed using recombinant COX2 and recombinant active FYN kinase, followed by measurement of COX2 enzymatic activity by endpoint peroxidase assay as described earlier in experimental procedures. The units of COX2 enzymatic activity is reported as nmol/min/ml (U/ml). Six independent experiments are shown (n=6) each experiment containing triplicate samples. * shows statistically significant difference when comparing COX2 vs. COX2+FYN group. (C) FYN overexpression does not increase endogenous expression of COX genes in DU145 cells. DU145 cells transduced with the indicated adenoviruses (moi of 5 or 10) were lysed and expression of COX1 or COX2 was detected by immunoblotting. β-Actin was used as a loading control, and FYN overexpression was confirmed by immunoblotting with FYN antibodies.

Figure 3. COX2 is a substrate for direct phosphorylation by Src family kinases in vitro. In vitro kinase assays were carried out using recombinant COX2 as a substrate for recombinant active

(A) FYN, (B) LYN or (C) JAK2 kinases as described previously in experimental procedures, followed by detection with non-selective phosphor-tyr specific antibodies. Negative controls included no kinase and no substrate reactions. COX2 phosphorylation, as well as kinase autophosphorylation is shown. Each panel is a representative blot of three experiments and immunoblotting with COX2 antibodies is shown as an equal loading control.

Figure 4. FYN and LYN kinases phosphorylate COX2 on two distinct residues in vitro. In vitro kinase assays were performed using recombinant COX2 as a substrate for recombinant active

(A) FYN or (B) LYN kinases as described previously in experimental conditions, followed by multi-stage fragmentation mass spectrometry as described previously in experimental conditions, to detect exact phosphorylation sites. Representative MS² spectra of the COX2 phosphopeptides pYQSFNEYR and SHLIDSPPTYNADpYGYK.

Figure 5. FYN and LYN phosphorylation sites on COX2, located on catalytic and dimerization domains respectively are evolutionary conserved in COX2 from different species, but are absent in the COX1 isoform

(A) General linear structure of human COX2 protein functional domains and catalytic residues. COX2 has four domains: amino terminal signal peptide, dimerization domain/EGF like domain, membrane binding domain and catalytic domain. Important amino acid residues required for enzyme catalysis or substrate binding are shown. The catalytic domain contains V335 and S516 which is essential for governing the stereochemistry of oxygen attack at carbon 15 in the production of PGG₂, Y371, which forms a tyrosyl radical, abstracts hydrogen from the pro-S side of carbon 13 of AA, H193 which is the proton acceptor, and Y341 which is essential for AA binding. R106 located on the dimerization domain of the enzyme is also essential for substrate binding. Indicated are also the sequences of the peptides and positions of the tyrosine phosphorylation sites Y446 and Y120 identified by mass spectrometry. B) The theoretical crystallographic model of human COX2 structure was taken from Protein Data Bank file 1V0X. Y446 and residues involved in the catalysis of COX2 were projected on the crystal structure of COX2 using pYMOL. Functional domains are shown as well: 1) Signal peptide (orange); 2) Dimerization domain (purple); 3) Membrane binding domain (red); Catalytic domains (COX and POX) (green); Membrane targeting sequence (blue). (C) Multiple sequence analysis of tyrosine phosphorylation residues on COX2 enzyme. Sequences of known different species of COX2 and COX isoforms, aligned around the sequences surrounding Y446 and Y120 residues using Uniprot.

Figure 6. Phosphorylation of Tyrosine Y446 residue on COX2 leads to increased activity of the enzyme

(A) Y446E mutant shows increased enzymatic activity when compared to wt COX2 in HEK 293T cells. HEK 293T cells expressing wt or indicated COX2 phosphorylation mutants were stimulated with 30 μM AA for 15 min in serum-free DMEM media and PGE₂ levels were measured by PGE₂ Elisa Kit as described earlier in experimental procedures. PGE₂ levels are shown as pg/1250 cells on a logarithmic scale after normalization to COX2 expression for the different mutants. * represents COX2 mutants demonstrating statistically significant difference in PGE₂ levels when compared to wt COX2 (n=8). (B) Fyn increases activity of wt COX2 but not Y446F mutant in DU145 cells. PGE₂ levels were measured in DU145 cells expressing wt or COX2 mutants in the presence or absence of FYN as described above. Fold change in PGE₂ levels is shown (n=4). * indicates statistically significant difference when FYN+COX2 wt is compared to Null+COX2 wt group.