Cyclooxygenase 2: protein-protein interactions and posttranslational modifications

Abstract

Numerous studies implicate the cyclooxygenase 2 (COX2) enzyme and COX2-derived prostanoids in various human diseases, and thus, much effort has been made to uncover the regulatory mechanisms of this enzyme. COX2 has been shown to be regulated at both the transcriptional and posttranscriptional levels, leading to the development of nonsteroidal anti-inflammatory drugs (NSAIDs) and selective COX2 inhibitors (COXIBs), which inhibit the COX2 enzyme through direct targeting. Recently, evidence of posttranslational regulation of COX2 enzymatic activity by s-nitrosylation, glycosylation, and phosphorylation has also been presented. Additionally, posttranslational regulators that actively downregulate COX2 expression by facilitating increased proteasome degradation of this enzyme have also been reported. Moreover, recent data identified proteins, located in close proximity to COX2 enzyme, that serve as posttranslational modulators of COX2 function, upregulating its enzymatic activity. While the precise mechanisms of the protein-protein interaction between COX2 and these regulatory proteins still need to be addressed, it is likely these interactions could regulate COX2 activity either as a result of conformational changes of the enzyme or by impacting subcellular localization of COX2 and thus affecting its interactions with regulatory proteins, which further modulate its activity. It is possible that posttranslational regulation of COX2 enzyme by such proteins could contribute to manifestation of different diseases. The uncovering of posttranslational regulation of COX2 enzyme will promote the development of more efficient therapeutic strategies of indirectly targeting the COX2 enzyme, as well as provide the basis for the generation of novel diagnostic tools as biomarkers of disease.

Introduction

Cyclooxygenase 2 (COX2) has been implicated in a number of physiological and pathophysiological processes including inflammation, tumor growth, and renal injuries (31, 78, 95).

Biochemical function of COX2.

Cyclooxygenases (COX) are enzymes that catalyze the first, rate-limiting step in the formation of such prostanoids as prostaglandin and thromboxane eicosanoids from phospholipase A₂-released arachidonic acid (AA) (119). There are two forms of cyclooxygenases, cyclooxygenase 1 (COX1), which is constitutively expressed, and cyclooxygenase 2 (COX2), which is the product of an immediate early gene capable of being upregulated by diverse stimuli. There is a 60–65% sequence identity between COX1 and COX2 from the same species and 85–90% identity between orthologs from different species. COX functions as a homodimer in which each subunit of the dimer consists of the epidermal growth factor-like domain, the membrane binding domain, and the catalytic domain. The catalytic domain contains the COX and peroxidase active sites on either side of the heme prosthetic group. COX enzyme is present at the luminal surface of the endoplasmic reticulum (ER) and at the inner and outer membranes of the nuclear envelope (NE).

Once AA is released from the phospholipid bilayer of the plasma membrane, the biosynthesis of prostanoids by COX enzyme proceeds in a multistep process. First, the COX active site of COX oxygenates AA at the 11 and 13 carbons to generate cyclic endoperoxidase prostaglandin G2 (PGG₂). This is followed by reduction of PGG₂ to prostaglandin H2 (PGH₂) by the peroxidase activity of the enzyme (83). The conversion of PGH₂ into distinct prostanoids such as prostaglandin D2 (PGD₂), prostaglandin E2 (PGE₂), prostaglandin F2 (PGF₂), prostaglandin I2 (PGI₂), and thromboxane A2 (TXA₂) is further carried out by subsequent action of appropriate specific downstream prostaglandin synthases (Fig. 1) (49, 83, 108, 117, 119). Newly synthesized prostaglandins efflux from cells by simple diffusion, driven by pH and membrane potential. Prostaglandin transport can also be facilitated by prostaglandin transporter and by organic anion-transporting polypeptides, such as multidrug resistance-associated proteins, which maintain energy-dependent prostaglandin transport across the plasma membrane (90, 111). Once exported into the extracellular microenvironment PGs exert their effects in an autocrine or paracrine manner. Specifically, PGE₂ effects are mediated by a family of G protein-coupled receptors (GPCRs) namely EP₁, EP₂, EP₃, EP₄. Similarly PGD₂, PGF₂, PGI₂, and TXA₂ exert their effects through interactions with specific G protein-coupled receptors termed DP for PGD₂, FP for PGF₂, IP for PGI₂, and TP for TXA₂ (68, 94).

Fig. 1.

Arachidonic acid metabolism by COX enzymes. Arachidonic acid is oxidized by COX1/2 enzymes into a short-lived hydroperoxyl-containing intermediate known as PGG₂, at the cyclooxygenase active site of the enzyme. PGG₂ is further reduced to PGH₂ at the peroxidase active site of the enzyme. PGH₂ is then metabolized by downstream prostanoid synthases to form the various groups of prostanoids. PGE₂, prostaglandin E₂; PGI₂, prostacyclin synthase; PGD₂, prostaglandin D₂; PGF_2α, prostaglandin F₂; TXA₂, thromboxane A₂; COX, cyclooxygenase; O₂, oxygen.

Biological consequences of COX2 in different organs/pathologies.

Through activation of the corresponding GPCRs and downstream signaling pathways, prostanoids, products of cyclooxygenase activity, have been shown to modulate tumor progression and inflammation and contribute to development of such renal disease as glomerulonephritis through several mechanisms. For example, signaling of these lipid mediators has been shown to directly regulate tumor epithelial cell proliferation, apoptosis, migration, and invasion through activation of downstream mitogenic pathways (141, 142). Signaling of prostaglandins has also been shown to induce epithelial cells to secrete growth factors, proinflammatory mediators, and angiogenic factors, creating a microenvironment that supports tumor growth and spread (141, 142). Prostanoids have also been shown to promote cancer progression through inducing an inflammatory microenvironment (142). Additionally, through direct binding of their receptors on stromal cells, prostanoids have been shown to promote a tumor supportive microenvironment by allowing tumor cells to evade attack by the immune system (141, 142). Moreover, signaling of prostanoids through their receptors on stromal cells promotes a tumor-supportive microenvironment by inducing angiogenesis (141, 142). Consistent with the mitogenic signaling of these lipid mediators, increased expression of COX2, prostanoid receptors, and elevated COX2-mediated production of prostanoids have been reported in various human cancers (43, 46, 88, 107, 140). Beyond elevated levels of COX2 in tumors, in vitro and in vivo studies have further established that COX2 signaling contributes to the development and progression of carcinogenesis (79, 96, 97, 134).

Due to inflammatory signaling of prostanoids, overexpression of COX2 and increased production of an array of prostaglandins have been also detected in arthritis and in inflammatory bowel disease (136). Moreover, prostaglandin signaling has been reported to contribute to renal diseases such as proliferative glomerulonephritis by regulating cellular adhesion and proliferation of glomerular mesangial cells (18, 34, 64, 82). Prostaglandin signaling has been shown to increase free cytosolic calcium levels in glomerular mesangial cells promoting glomerular mesangial cell contraction, which can affect glomerular function in diseases such as glomerulonephritis (89). Additionally, prostaglandin signaling has shown to contribute to glomerulonephritis by progressive accumulation of extracellular matrix (ECM) components, inflammatory changes, and podocytes injury, which further affect the glomerular filtration barrier (35, 106, 127). Consistent with these reports, COX2 has been also shown to be overexpressed in proliferative glomerulonephritis (21, 48). Multiple in vivo animal models have further demonstrated that COX2 signaling contributes to the progression and development of glomerulopathies (23, 36, 48, 144).

Regulatory mechanisms of cellular actions of COX2.

Taken together, these studies suggest that COX2 enzyme and COX2-derived prostanoids are therapeutic targets in various human diseases, and as such much effort has been made to uncover the regulatory mechanisms of this enzyme. To this end, it has been demonstrated that cellular actions of COX2 are regulated at multiple levels. One regulatory factor is the availability of AA, which is dependent on the expression and/or activity of PLA₂ enzyme, causing the release of AA. Another regulatory factor is the expression of prostanoid receptors, which mediate prostanoid signaling by inducing various effects in the cells associated with disease. Expression of downstream synthases and hydrolases involved in generation and inactivation of various prostanoids respectively could be another regulatory factor. Additionally, like most genes COX2 expression is regulated both at the transcriptional and posttranscriptional levels (Fig. 2) (28, 133). The gene encoding COX2 protein, PTGS2, is ~8.3 kb long with 10 exons, and it is transcribed as, 4.6, 4.0, and 2.8 kb mRNA variants. The last exon in the PTGS2 encodes the 3′-untranslated region (3′-UTR), containing 23 copies of the “ATTTA” RNA instability element. Moreover, PTGS2 5′-UTR promoter region contains several potential transcription regulatory elements, including a TATA box, an NF-IL6 motif, two AP-2 sites, three Sp1 sites, two NF-κB sites, a CRE motif, and an E-box (4, 71, 130). Transcriptional and posttranscriptional regulation of COX2 gene has been extensively discussed in many of the previous reviews, and thus the reader should refer to these sources for more information.

Fig. 2.

Multiple levels of regulation of cellular actions of COX2 enzyme. Cellular actions of COX2 enzyme are regulated by the availability of AA, which is dependent on PLA2 expression and/or activity. Cellular actions of COX2 enzyme can also be regulated by availability of prostanoid receptors, which are required for signaling of prostanoids. Expression and activity of different synthases and hydrolases and regulation of COX gene at the transcriptional, posttranscriptional, and posttranslational levels further regulate the synthesis of prostanoids and thus cellular actions of COX2 enzyme. PLA2, phospholipase A2; COX2, cyclooxygenase 2.

Regardless the knowledge of transcriptional and posttranscriptional regulation of COX2 enzyme, currently nonsteroidal anti-inflammatory drugs (NSAIDs) and selective COX2 inhibitors (COXIBs) are designed to inhibit the COX1 and COX2 enzymes directly (1, 5, 13, 24, 46, 47, 57, 66, 105, 126, 129). However, such drugs, while still in use in cancer therapy, have been proven to be not beneficial due to adverse side effects (9, 10, 14, 17, 39, 72, 73) and in some cases were shown to act independently of their effect upon the COX2 enzyme (41).

Recent studies have observed that the kinetics of prostaglandin synthesis in mammalian cells does not always correlate with the level of COX protein expression, suggesting the possibility of posttranslational regulation of COX activity and, ultimately, prostaglandin synthesis. Inhibiting COX2 enzyme indirectly by targeting regulators of its enzymatic activity will provide alternative therapeutic strategies in treatment of diseases where COX2 is implicated. In this review, we will summarize studies that provide evidence of posttranslational regulation of COX2 enzyme, report of specific novel posttranslational modulators of COX2 enzymatic activity, and further consider their proposed mechanisms of action and molecular determinants required for their interaction with the COX2 enzyme.

Posttranslational Regulation of COX2 Enzyme

Posttranslational modifications (PTMs) are chemical modifications that regulate protein activity, folding, conformation, stability, localization, and interaction with other proteins. One such modification, s-nitrosylation, is a reversible reaction that involves reaction of nitric oxide (NO), produced by one of the three isoforms of nitric oxide synthase (NOS) with free cysteine residues to form S-nitrothiols. Protein glycosylation is another PTM that involves addition of sugar molecules to proteins either at an asparagine (N) or serine/threonine residue resulting in N- or O-linked glycosylation, respectively. Ubiquitination is a different PTM that involves attachment of Ubiquitin, an 8 kDa polypeptide consisting of 76 amino acids to the ε-NH2 of lysine in target proteins via the COOH-terminal glycine of ubiquitin. Ubiquinated proteins are further recognized by the 26S proteasome in the cytoplasm that catalyze its degradation. COX2 can be ubiquitinated and degraded by the 26S proteasome in the cytoplasm in such a manner, suggesting that COX2 degradation can involve exit from the ER via the ER-associated degradation (ERAD) system(s) followed by proteolysis by the proteasome. COX2 protein degradation can also be initiated by substrate-dependent suicide inactivation. Suicide-inactivated protein is then degraded. The biochemical steps have not been resolved, but substrate-dependent degradation is not inhibited by proteasome inhibitors or inhibitors of lysosomal proteases. Protein phosphorylation is another reversible PTM that occurs on serine, threonine or tyrosine residues mediated by kinases and phosphatases, which phosphorylate and dephosphorylate substrates, respectively (139). Most of the above mentioned PTMs occur as a result of protein-protein interactions (PPIs), which are physical contacts of high specificity established between two or more protein molecules with molecular associations between chains that occur in a cell (26). COX2 has been shown to interact with various proteins that regulate its activity. Known COX2 PTMs and specific PPIs are further discussed below in more detail.

S-nitrosylation of COX2 enzyme.

A complex interplay of cyclooxygenase and NOS pathways have been observed in multiple systems (125). At the posttranslational level activity of COX2 enzyme has been shown to be regulated by s-nitrosylation. Specifically, studies have demonstrated that inducible nitric oxide synthase (iNOS) binds to COX2 (67). NO generated by iNOS is then available to s-nitrosylate COX2, increasing COX2 catalytic activity (67, 125). Neuronal nitric oxide synthase (nNOS) was also shown to bind COX2 via nNOS PDZ domain and with the generated NO to S-nitrosylate and activate the enzyme (131).

Glycosylation and proteasomal degradation of COX2 enzyme.

Studies have demonstrated that glycosylation of Asn 594 and possibly other amino acid residues unique to COX2 and located in COOH-terminal 19 amino acids (aa) of the enzyme (Asn-594–Lys-612) target the protein for proteolysis by the ERAD pathways (85, 86). It was also shown that glycosylation of COX2 protein at Asn-580 leads to low levels of COX2 protein and decreased COX2 activity, and thus glycosylation of this specific residue could be similarly important for entry of COX2 protein into ERAD pathways (113).

Phosphorylation of COX2 enzyme.

Due to the presence of phosphorylation motifs for several kinases within COX2, studies have also investigated COX2 regulation by phosphorylation (65, 112). It was reported that protein tyrosine phosphatase and kinase inhibitors stimulated or inhibited COX2 activity, respectively, in cerebral endothelial cells (99). It was also shown that nitration of tyrosine residues in the active site of the enzyme abolished its catalytic activity (115). In rabbit articular chondrocytes, the mechanism by which 2-deoxy-D-glucose (2DG) mediates the inflammatory response to ER stress is by inhibiting tyrosine phosphorylation of COX2. However, apart from demonstrating that tyrosine phosphorylation of COX2 can modulate its expression and activity, this study was not able to establish that COX2 itself is a substrate for protein tyrosine phosphorylation by immunoprecipitation using phospho-tyrosine antibodies (150). Due to the existence of protein kinase C (PKC) consensus sequences in COX2 a study investigated whether direct phosphorylation of COX2 by the serine/threonine PKC occurs. However, the study concluded against COX2 being a favorable substrate for PKC phosphorylation (138).

PPIs of COX2 enzyme.

One study reported that caveolin-1 (Cav-1) interaction with COX2 at the ER or NE increases COX2 ERAD proteasome degradation (22). Moreover, studies have further shown that GPCRS, beyond inducing their traditional signaling cascades, can increase COX2 proteasome degradation by a mechanism that does not require receptor activation (42, 122). It was demonstrated that the EP1 receptor forms a complex with COX2 and increases COX2 ubiquitination levels, thereby accelerating its degradation by the proteasome pathway (42). Further studies by the same group have shown that elevation in COX2 protein expression is accompanied by an increase in levels of endogenous EP1 receptor via a mechanism that involves an interaction between the two proteins (121). These combined results suggest that interaction between EP1 and COX2 may be part of a feedback loop, whereby an increase in COX2 elevates EP1 receptor levels, which ultimately acts to downregulate COX2 by expediting its degradation. As another example of GPCR regulation of COX2 expression, the angiotensin II type 1 receptor was shown to interconnect with COX2 through elements in its cytosolic carboxyl tail and enhance COX2 ubiquitination (122). Similarly, β1 adrenergic receptor was shown to downregulate expression of COX2 by accelerating its degradation via the proteasome (16). In all these studies it was verified that COX2 downregulation by GPCRs was not accompanied by a reduction in protein synthesis (16, 22, 42, 121, 122). It is highly likely that the interaction of COX2 and GPCRs most likely involves additional proteins, such as an E3 ligase and other scaffold proteins, and is a current area of investigation.

Published data from our laboratory showed that overexpression of COX2 results in the formation of covalent adducts between COX2 and a number of signaling proteins (3, 148). These covalent adducts are likely mediated by spontaneous decomposition of prostaglandin H2, an immediate product of COX2 activity. The decomposition of PGH₂ results in the production of y-keto aldehydes-levuglandins (LGE₂), which are capable of covalently cross-linking Lys residues of proteins within close proximity to each other (55, 110). While the formation of COX2 adducts is likely the consequence of COX2 overexpression, it provides insight into the types of proteins that are located in close proximity of COX2. Supporting the native formation of these complexes inside the cell, we observed formation of these COX2 adducts with endogenous COX2 (3, 148). Furthermore, our findings that the use of the COX2 inhibitor NS-398 prevented formation of cross-linked products further supports specificity of this interaction (3, 148). Mass spectrometry of COX2 adducts indicated the presence of Engulfment and cell motility protein 1 (ELMO1), a bipartite guanine nucleotide exchange factor for the small GTPase Rac (148). Additionally, we identified FYN, member of Src family cytoplasmic tyrosine kinases and Fibronectin precursor in complex with COX2 (3).

Both ELMO1 and COX2 have been associated with glomerular disease which is characterized by accumulated deposits of ECM proteins such as fibronectin by mesangial cells (34, 75, 102, 114, 124). Consistent with the observation that ELMO1 is localized in close proximity to COX2, we reported an interaction between ELMO1 and COX2 in perinuclear regions of human mesangial cells (HMC) (148). The COX2-mediated increase of fibronectin gene expression was shown to be dependent on the cyclooxygenase activity of COX2 and was enhanced by ELMO1 (148).

In addition to ELMO1, FYN was shown to increase COX activity resulting in elevated production of prostaglandins in prostate cancer (DU145) cells independent in changes of COX2 or COX1 steady-state protein levels (3). Furthermore, it appears that Tyr 446 (Y446) within the catalytic domain of COX2 is a substrate for direct phosphorylation by FYN (3). Substitution of Y446 with glutamate (phosphomimetic mutant) increased COX2 activity, whereas substitution of Y446 with phenylalanine (mutant preventing phosphorylation by FYN) prevented FYN-mediated increases in COX2 activity (3). These data support the conclusion that phosphorylation of Y446 COX2 acts as a regulator of enzyme catalysis. Another member of Src family of kinases, LYN, was also shown to phosphorylate COX2 in these studies, albeit on Tyr120 (Y120). This tyrosine residue is located within the dimerization domain of COX2. As LYN has been reported to similarly contribute to progression of prostate cancer, it is possible that LYN acts to increase COX2 activity in the prostate cancer cells as seen with FYN.

Proposed Mechanisms of PPI between COX2 and Protein Regulators

While the mechanisms regulating COX2 proteasome degradation have been addressed (16, 22, 42, 121, 122), much less is known regarding mechanisms of posttranslational regulation of COX2 that upregulate enzymatic activity.

As discussed above, ELMO1 interaction with COX2 was shown to upregulate COX2 enzymatic activity (148). While the mechanism of ELMO1-mediated regulation of COX2 enzymatic activity remains unknown, it is unlikely that interaction of COX2 with ELMO1 interferes with COX2 degradation pathways as in these studies changes in ELMO1 protein expression did not alter COX2 protein expression levels (148). The mechanism of ELMO1 action in regulation of COX2 activity may involve enzymatic conformational changes resulting from interaction between ELMO1 and COX2.

Additionally, FYN phosphorylation of COX2 enzyme on Y446 residue was shown to increase prostaglandin production in prostate cancer cells DU145 (3). The FYN phosphorylation site of Y446 on COX2 is likely to be important regulator of COX2 enzymatic activity, as alignment of COX2 protein orthologs among mammals and birds shows this residue to be conserved (3). Furthermore, this tyrosine residue is only present in the inducible isoform of COX2 but not the constitutive COX1 isoform of the enzyme (3). Additionally, the finding that this residue is unique to COX2 supports a specific role for COX2 regulation (isoenzyme upregulated in diseases) by FYN.

However, while mechanism by which phosphorylation of Y446 residue on COX2 could regulate enzymatic activity is not clear, 2DG mediates the inflammatory response of chondrocytes to ER stress in chondrocytes by inhibiting protein phosphorylation at tyrosine residues within COX2. This in turn leads to decreased expression and activity of this enzyme (150). While the mechanism by which tyrosine phosphorylation could regulate COX2 expression was not directly addressed in this study, one possibility is that phosphorylation of the COX2 enzyme leads to increased COX2 expression as a result of decreased enzyme degradation, which preserves it for prolonged prostaglandin production. Yet, it is unlikely that FYN phosphorylation of COX2 similarly interferes with COX2 degradation pathways, since FYN was not shown to have any effects on COX2 protein expression (3).

Y446 residue is likely positioned too far away from the active site to directly affect COX2 activity. Therefore, we speculate that phosphorylation of this site regulates COX2 activity and prostanoid biosynthesis allosterically by introducing conformational changes to COX2, increasing enzyme activity. In support of this hypothesis, it has been demonstrated that when COX and peroxidase subunits of the COX enzyme undergo irreversible suicide inactivation in vitro during catalysis it is accompanied by dramatic changes in protein structure (85). This results in slow destruction of the protein-heme complex of the COX enzyme, allowing increased access to histidine residues, which are critical for the catalysis of the COX enzyme and are generally shielded by the heme complex when the enzyme is active (85). Therefore, it is possible that allosteric changes resulting from phosphorylation of Y446 prevent suicide inactivation of COX2, resulting in increased activity of the enzyme. Another possibility is that FYN phosphorylation of COX2 changes the intracellular localization of COX2, resulting in a change of biological activity (29, 32, 128). To this end we speculate that phosphorylation of COX2 by FYN will result in localization of COX2 in the cytoplasm where ELMO1 resides (previously mentioned regulator of COX2 enzyme). Further experiments need to be conducted to study intracellular trafficking of COX2 to test this hypothesis.

Regardless of its mechanism of action, one of the key questions remaining is where the FYN and COX2 interactions take place. FYN is largely recognized as a plasma membrane protein, whereas COX2 is largely localized to the ER and NE (77). It was recently reported that for favorable PPI to occur complex-like colocalization needs to form between two proteins (2). Specifically this study demonstrated that the molecule-molecule interactions between COXs and their downstream synthases are essential in determining the production of specific prostaglandins (2). Such complex-like colocalization between FYN and COX2 is most likely to occur in caveolae structures, since COX2 has been shown to be localized in Cav-1-containing vesicles and further colocalize with Cav-1 protein in cells (77, 101). Similarly, FYN has been reported to interact with Cav-1 to couple integrins to the Ras-ERK pathway (145). The 19 aa segment located at the COOH termini of COX2, which is not responsible for the catalytic activity of the enzyme and which appears to be important for COX2 entry into ERAD pathways (85, 86), could play a role in differential trafficking of COX2 enzyme from ER to the caveolae (Fig. 3). Further immunofluorescence studies need to be conducted to address the COX2/FYN colocalization in caveolar structures.

Fig. 3.

Proposed cellular localization of FYN/COX2 in normal (A) and disease (B) conditions. Cytoplasmic nonreceptor tyrosine kinase FYN is not located within proximity to the ER membrane and COX2 in normal cells (A). FYN and COX2 form a complex-like contact in caveola-like structures in diseased cells (B), as a result of COX2 trafficking from ER to caveolae, possibly mediated by the 19 aa segment located at the COOH termini of COX2 enzyme. COX2, cyclooxygenase 2; FYN, nonreceptor tyrosine kinase; PM, plasma membrane; ER, endoplasmic reticulum; aa, amino acid.

It is interesting that LYN-dependent COX2 phosphorylation occurs at Y120, within the dimerization domain (3), as it has been reported that there could be cross talk between the monomers of COX2 homodimers (152). It was shown that one monomer serves as the allosteric subunit, enabling the other catalytic subunit to catalyze the reaction (30, 151, 152). It is possible that the introduction of a negative charge by phosphorylation of LYN on Y120 residue of COX2 could lead to increased activity of the enzyme in the DU145 cells as a result of increased COX2 dimerization.

Molecular Determinants Required for Interaction between COX2 Enzyme and Proteins that Regulate its Activity

Similar to mechanisms of action of COX2 modulators, not much is known regarding molecular determinants that mediate interaction between COX2 and regulators that increase its activity.

ELMO1 protein contains a number of PPI domains and motifs. Pleckstrin homology domain (residues 550–677) mediates direct interaction with DOCK180 and is critical in Rac signaling (70). Additionally, interaction with SH3 containing proteins, including DOCK180, is controlled by Pro-rich motif (PxxP) located at the COOH-terminal region of ELMO1 (residues 707–714) (81). ELMO domain (residues 318–491) was hypothesized to possess GAP activity for members of the Arf family of small G proteins (15). The NH₂-terminal region of ELMO1 (residues 1–280) was shown to interact with members of highly conserved ezrin/radixin/moesin protein family (40), small GTPase RhoG (62), and one of the Shigella effectors IpgB1 (44). Several armadillo repeats were identified in the NH₂ terminus of ELMO1, which are involved in interaction of ELMO1 with RhoG (81). Deletion and truncation mutants of ELMO1 PPI domains should be generated to identify the molecular determinants that might be mediating interaction between ELMO1 and COX2 (Fig. 4A).

Fig. 4.

Schematic domain organizations of ELMO1, FYN, COX2, and schematic structures of proposed mutants. A: human ELMO1 consists of 727 amino acids. Positions of protein-protein interaction domains and regions responsible for binding to signaling molecules are indicated. B: human FYN consists of 537 amino acids. Positions of protein-protein interaction domains and regions responsible for binding to proline and phosphotyrosine residues are indicated. C: human COX2 consists of 604 amino acids. Positions of EGF-like domain, membrane binding domain (MBD), catalytic domain, and protein instability element are indicated. Four functional N-glycosylation sites (53,130, 396, and 580) are shown. Proposed schematic structures of mutants for all 3 proteins are also depicted. ERM, ezrin/radixin/moesin; PH, pleckstrin homology.

While our previous studies demonstrated that FYN increased activity of COX2 enzyme in DU145 cells, binding was not observed between FYN and COX2 in these cells by coimmunoprecipitation assays, possibly due to a weak and transient interaction between a kinase and its substrate (3). More sensitive assays need to be designed to further address COX2/FYN interaction in the DU145 cells. FYN has a 60 aa NH₂ termini unique domain (residues 1–84), that contains the highest degree variability among different Src family kinases and is thought to mediate protein-protein binding interactions (109). It also contains an SH3(residues 84–140) and SH2 (residues 148–230) domains that are highly conserved regions also responsible for interaction, by binding to proline-rich and phosphorylated tyrosine sequences in the target protein, respectively (109). Point, substitution, and deletion mutants of the PPI domains of FYN can be generated to further determine the crucial elements that mediate FYN interaction with COX2 (Fig. 4B).

Experiments can also be designed to identify COX2 molecular determinants, which mediate COX2 interaction with ELMO1 and FYN. COX2 and COX1 have similar primary structures with an exception of the 19 aa insertion at COOH terminus that appears to be important for COX2 entry into the ERAD pathways (85, 86) (Fig. 4C). Considering that both ELMO1 and FYN in previous studies were shown to interact and specifically regulate COX2 activity but not COX1, COX2 mutants can be generated to test whether COX2 interaction with ELMO1 and FYN requires the presence of 19 aa within the COOH-terminal region or other elements of COX2 molecule are important for this interaction.

Finally, knowledge from identification of the crucial elements that govern binding of FYN and ELMO1 to COX2 and vice versa can be used to construct FYN and ELMO1 constructs, which will act efficiently as a dominant negative molecule in respect to FYN- and ELMO1-COX2 interactions and can be further used to confirm regulation of COX2 enzyme by these proteins.

Physiological Significance of Posttranslational Regulation of COX2 Enzyme

Molecular tools available for studying consequences of COX2 signaling in cells.

While the above-mentioned studies reported of various COX2 posttranslational regulatory mechanisms, such studies did not directly address the biological and physiological significance of posttranslational regulation of COX2 enzyme. This is mainly due to a lack of molecular tools available for conducting COX2 gene transfer experiments in the scientific community. Specifically, the presence of instability regions in the coding regions of human COX2 (11), restrict the generation of adenoviral or lentiviral constructs for robust expression of COX2 in cells. Therefore, currently plasmids are the main method of expressing COX2 in gene transfer experiments, however they result in low percentage of cells expressing human COX2, complicating the outcome of such studies. This is in contrast to rat COX2 adenovirus construct, which possibly does not contain these instability regions compared with human COX2 gene and thus has been generated in the past in our laboratory and successfully used in many studies (20, 52, 53, 100). Given the high level of homology among COX2 within different species (49, 83, 108, 117, 119), rat COX2 adenovirus can alternatively be used for expressing COX2 in cells of different origin to establish the biological and physiological consequences of COX2 regulation. Moreover, a recent study has reported that optimization of human COX2 cDNA sequence (by changing 382 of 1,815 nucleotides in the coding region of COX2 gene) resulted in increased COX2 expression and prostaglandin production when incorporated into a lentiviral vector (11). Therefore, this particular lentiviral COX2 construct containing the optimized COX2 cDNA can be further used for future studies to establish the biological and physiological consequences of posttranslational regulation of COX2 enzyme.

Regardless of method of choice for expressing COX2 in cells, we speculate that posttranslational regulation of COX2 enzyme by various protein modulators could have various biological and thus physiological consequences, which are further discussed below in more details.

Potential regulatory mechanisms of COX2 enzyme in glomerular disease.

ELMO1 regulation of COX2 activity in HMC might be important for contributing to the development of glomerular injury through increased production of ECM proteins such as fibronectin. To this end, animal studies can be utilized to further confirm that COX2/ELMO1 interaction actually contributes to the accelerated decline in renal function in ECM-depositing glomerular diseases. For this purpose Dahl S model of hypertension, a well-characterized model of hypertension-induced glomerular injury associated with mesangial expansion, can be used. Renal injury in these animal models has been shown to be induced by high-percentage salt diet (27) or by injecting anti-Thy 1.1 antibody (58), which leads to mesangioproliferative glomerulonephritis. Moreover, transgenic rat strains were shown to be successfully generated using the Sleeping Beauty (SB) transposon element to catalyze the integration of transgenes into the genome by placing the transgene inside of a transposon vector and co-introducing it into the rat embryo with in vitro-transcribed SB transposase mRNA (19). Transgenes introduced by transposition integrate as single copy units, which are easily cloned and genotyped to avoid insertions, which may disrupt endogenous genes and lead to persistent expression (19). This transgene method can be used to overexpress wild-type ELMO1 (SS/ELMO1wt) and dominant negative ELMO1 (construct that will interfere with COX2/ELMO1 interaction) (SS/ELMO1dn) into the Dahl S rat strain (SS). Next, assessment of renal injury in these animals could be evaluated by matrix deposition, mesangial cell proliferation, renal histological abnormalities, and apoptosis to further establish the physiological significance of COX2/ELMO1 interaction in ECM-depositing glomerular diseases. FYN was shown to be involved in signaling from Thy1 in GMC and was shown to be associated with progression of focal and segmental glomerulosclerosis (51, 74). FYN knockout rats were generated at MCW, using Zinc Finger Nuclease to target the FYN gene (38). Future studies can be designed to alternatively address FYN and FYN-mediated COX2 phosphorylation in ECM-depositing renal glomerular diseases using these genetically modified rats deficient in FYN. However, before such studies are carried out, specific conditions of inducing glomerulonephritis in these FYN knockout animal models need to be carefully addressed.

Potential regulatory mechanisms of COX2 enzyme in cancer/related inflammatory disease.

Our studies have previously shown that COX2-mediated prostaglandin synthesis protects prostate cancer cells and rat pheochromocytoma cells from apoptosis (52, 87, 91). It was further determined that COX2 expression results in transcriptional regulation of a number of genes important for cell survival in prostate cancer cells (20, 100, 123). Additionally, it was shown that COX-mediated prostaglandin synthesis induced prostate tumor cell invasiveness and increased cell growth (7, 8, 132). Also, FYN expression level, similar to COX2, was reported to be increased in prostate cancer (104), and FYN has been shown to possess mitogenic effect in this cancer (59, 76, 104, 109). Therefore, FYN-mediated tyrosine phosphorylation of COX2 could result in enhancing COX2-mediated biological effects in prostate cancer cells. In FYN knockout rat models prostate cancer can be induced hormonally or chemically (116, 135), and assessment of tumor growth can be evaluated by analysis of extracted tumors for markers of apoptosis, cell growth, and migration. Overall, this mechanism of COX2 enzyme regulation by FYN kinase could have implications beyond prostate cancer: in other cancers where activity of COX2 may be regulated by FYN. Given the role of prostanoid signaling in inflammation (142), FYN regulation of COX2 activity could also be implicated in other inflammation-related diseases such as arthritis and inflammatory bowel disease. Moreover, regulation of COX2 enzyme by Cav-1 protein and GPCR could be important in resolving inflammation through downregulation of COX2 enzyme.

Potential regulatory mechanisms of COX2 enzyme in hypertension.

Finally, both FYN and COX2 have been previously shown to be implicated in hypertension (143, 147), thus FYN-mediated regulation of COX2 enzyme could be important for contributing to manifestation of this disease.

COX2 s-nitrosylation contributes to enhanced formation of bioactive host-protective molecules produced in neutrophil-endothelial cocultures, termed 13-series resolvins (RvTs) (25). RvTs increase mouse survival during Escherichia coli infection and also regulate mouse and human phagocyte responses (25). S-nitrosylation and activation of COX2 activity in Sprague-Dawley rats, treated with atorvastatin, were shown to cause preconditioning effects and cardioprotection (6). Since modulation of COX and NOS pathways was shown to be principally involved in a variety of pathological conditions, the dissection of their direct and indirect interregulation needs to be better understood to work out possible novel therapeutic strategies (125).

New Targets in Inhibiting COX2 Enzyme in Human Disease and Development of Novel Diagnostic Tools

NSAIDs and COXIBs: targeting COX enzyme.

Currently, the COX enzyme is the major pharmacological target of therapeutic and chemopreventive agents. NSAIDs are chemically distinct compounds that exert their anti-inflammatory, analgesic, and antipyretic effects mainly by inhibiting both COX enzymes: COX1 and COX2 (33, 117, 118, 137). While all NSAIDs compete with AA for binding to the COX active site, each NSAID exhibits one of three kinetic modes of inhibition: 1) rapid, reversible binding (e.g., ibuprofen); 2) rapid, lower-affinity reversible binding followed by time-dependent, higher-affinity, slowly reversible binding (e.g., flurbiprofen), or 3) rapid, reversible binding followed by covalent modification (acetylation) of Ser530 (e.g., aspirin) (2). There is extensive evidence that regular intake of various NSAIDs, particularly aspirin, can lower the risk of cancer at multiple organ sites (1, 24, 46, 57, 66, 129). Unfortunately, apart from beneficial effects of NSAIDs, COX inhibition has been demonstrated to result in unwanted side effects, particularly in the gastrointestinal (GI) tract (39, 73). Since COX-derived prostanoids are presumably involved in housekeeping functions, NSAID gastrotoxicity is considered to be the consequence of inhibition of COX1 isoenzyme. Selective COXIBs were developed to reduce the side effects of NSAIDs associated with inhibition of COX1. The mechanism for differential inhibition by classical NSAIDs and COX2 inhibitors can be rationalized to some extent based on differences between the COX active sites of COX1 and COX2. Substitution of Ile523 in COX1 with Val523 in COX2 results in the presence of a small side pocket adjacent to the active site channel, appreciably increasing the volume of the COX2 active site. This change is compounded by the substitution of Ile434 in COX1 with Val434 in COX2, within the second shell of amino acids surrounding the COX active site. The Ile-to-Val substitution at position 434 outside the COX2 catalytic center further increases the effective size of the active site channel by enhancing the local mobility of side chains within the side pocket. The combination of these two differences at positions 523 and 434 in COX2 causes a movement of Phe518 that further increases the size of the side pocket. The substitution of these particular residues in the COX2 enzyme increases the volume of the COX2 NSAID binding site by ~20% over that in COX1, which provides access to COX2 selective inhibitors (117, 119). While substantial evidence from clinical trials demonstrates that COXIBs reduced and prevented the incidence of various cancers as well as alleviated the GI complications of NSAIDs (5, 13, 47, 105, 126), long-term intake of these drugs was however also associated with an increased risk of cardiovascular (CV) side effects such as myocardial infarction or stroke (9, 10, 14, 17). Drug-mediated imbalance in the levels of PGI2 and TXA₂ with a bias toward TXA₂ may be the primary reason for these events (12). These findings limited the use of COXIBs in patients with a history of atherosclerotic heart disease (12). Additionally, COX2 is constitutively expressed in the kidney and plays an important role in the regulation of renal function (perfusion, water handling, and renin release) in both normal and pathophysiological conditions (45, 50). Therefore, use of NSAIDs and COXIBS has been further associated with adverse renal effects (45, 50). Moreover, an increasing body of evidence suggested that COX2 selective inhibitors can act independently of their effect upon COX2 (41). Regardless, some COXIBs such as celecoxib are still approved for use in treatment of colon cancer. NSAIDs and COXIBS have similarly been used in management of arthritis; however, use of both of these types of inhibitors has been equally effective in this disease and similarly has been associated with increased risk of GI, renal, and CV side effects (84).

Inhibitors targeting prostanoid signaling.

Alternative approaches, such as utilization of EP receptor antagonists and inhibitors of prostaglandin synthases, have also been considered to attenuate PGE₂ signaling in cancer (37, 54, 56, 60). Use of these compounds could limit the side effects observed with direct inhibition of COX isoenzymes by NSAIDs and COXIBs (37, 54, 56, 60). However, while some of EP receptor antagonists and inhibitors of prostaglandin synthases are under preclinical evaluation and have shown promising effects for preventing and/or inhibiting growth of different types of tumors in animal models (63, 93, 103, 120, 146, 149), testing of these antagonists in clinical use is still lacking.

Generation of specific peptides to accelerate COX2 proteasomal degradation.

Downregulation of COX2 enzyme by short peptides originating from known domains of GPCRS and other proteins that mediate COX2 proteasome degradation may present an alternative basis for novel therapeutic means of eliminating COX2 protein in disease. Further understanding the mechanisms that govern glycosylation of COX2 could provide a basis for development of tools capable to increase degradation of COX2 rather than just inhibit it activity.

ELMO1/FYN inhibitors.

Knowledge obtained from identification of the crucial elements that govern binding of ELMO1 to COX2 can be used to generate ELMO1 inhibitors, which will act efficiently as a dominant negative molecule in respect to ELMO1/COX2 interactions. Such ELMO1 inhibitors can be used in glomerulonephritis, where use of COX inhibitors is limited due to renal complications, as well as other diseases where ELMO1 regulation of COX2 enzyme might be important.

Additionally, given the current knowledge of posttranslational regulation of COX2 enzyme by FYN kinase (3), FYN inhibitors can be used as alternative chemotherapeutic agents to the currently used COXIBs to target COX2 enzyme in prostate cancer and possibly other cancers. Such chemotherapeutic agents could avoid the adverse side effects associated with COXIBs by targeting COX2 in cancer cells without affecting the function of this enzyme in normal cells. However, existing FYN inhibitors such as caffeic acid and delphinidin have been shown to have a broad specificity toward other family members of Src kinases as well (109). This is mainly due to the fact that most of Src inhibitors act by binding the kinase domain noncompetitively with ATP and reducing the binding affinity of ATP and the catalytic activity of the kinase (61). Src kinase inhibitors designed against the unique domain of FYN kinase might be more efficient at achieving more specific inhibition.

However, it can be expected that, similar to COXIBs, FYN inhibitors could have multiple side effects of their own, since the biological effects of FYN are diverse, ranging from neurological to immune functions (69, 80, 92, 98). Therefore, FYN inhibition could lead to learning and memory defects as well as to autoimmune diseases. To overcome or reduce side effects of FYN inhibitors it could be considered whether lower doses of FYN inhibitors can be used in combination with other chemotherapeutic agents. Alternatively, Y446 synthetic peptides can be generated to compete with Y446 site on COX2 and thus specifically prevent FYN phosphorylation of COX2. Such chemotherapeutic agents might prove to be therapeutically more beneficial with less severe side effects than single FYN inhibitors. However, additional studies would need to be conducted to test whether these compounds would be stable enough for their efficient delivery into cancer cells. Furthermore, the proposed FYN inhibitors can be also tested in glomerulonephritis and other inflammation-related diseases where FYN regulation of COX2 enzyme might be important, as an alternative to COXIBs. These novel therapeutic strategies in targeting COX2 enzyme in various disease states are shown in Fig. 5.

Fig. 5.

Model describing proposed new therapeutic strategies in inhibiting COX2 enzyme in human diseases. Inhibition of COX 1/2 enzymes by NSAIDs and COXIBs results in GI, renal, and CV side effects. Prostaglandin synthase inhibitors and EP receptor antagonists inhibiting prostanoid synthesis and signaling are currently being tested. Alternatively, FYN inhibitors such as caffeic acid and dasatanib or Y446 peptides that compete with Y446 site on COX2 and prevent FYN-mediated tyrosine phosphorylation of the enzyme can also be considered. Similarly, ELMO1 dominant negative construct, which will compete with wt ELMO1 and prevent ELMO1/COX2 interactions mediating COX2 proteasome degradation, can be further utilized to eliminate COX2 protein. NSAIDs, nonsteroidal anti-inflammatory drugs; COXIBs, COX2 selective inhibitors; GI, gastrointestinal; CV, cardiovascular; COX, cyclooxygenase; FYN, nonreceptor tyrosine kinase; ELMO1, guanine nucleotide exchange factor for the GTPase Rac; PG, prostaglandin; GPCR, G protein-coupled receptor; Y, tyrosine; pY, phospho tyrosine.

Development of novel diagnostic tools.

Y446 was identified as the phosphorylation site of FYN on COX2 (3). Phospho-tyrosine-specific antibodies directed against COX2 phosphorylation site can be generated to screen human cancer tissue samples in tissue microarrays that contain multiple tissue samples from human disease and normal tissues. Such tissue samples are commercially available for multiple human cancers (Origene Technologies, Rockville, MD). Analysis of COX2 phosphorylation status would establish the significance of posttranslational regulation of COX2 activity in human cancers. Moreover, these COX2 phosphotyrosine antibodies can be used as potential diagnostic tools for tissue biopsies collected from probable cancer patients. This will be the first diagnostic tool to detect activity-related modification of COX2 as all other antibody-based tools were only able to report on the level of COX2 expression.

An overall summary of posttranslational modulators of COX2 enzyme, their proposed mechanisms of action, suggested role in different disease states, and a list of potential therapeutic strategies in inhibiting COX2 enzyme in disease is provided in Table 1.

Table 1.

Summary of posttranslational modulators of COX2 enzyme, their proposed mechanisms of action, suggested role in different disease states, and list of potential therapeutic strategies in inhibiting COX2 enzyme in disease

Posttranslational Modulator	Type of Modification	Effect on COX2 Enzyme	Proposed Mechanism	Type of Disease	Therapeutic Strategies	References
nNOS/iNOS	nitrosylation	+	NK	Cardioprotection	NK	(67, 131)
NK	glycosylation (ASN 594/580)	−	proteasome degradation	NK	NK	(113)
Cav-1/GPCR	ubiquitination	−	proteasome degradation	anti-inflammatory	GPCR/Cav-1 peptides	(22, 42, 121, 122)
ELMO1	none	+	conformational changes	glomerulonephritis, inflammatory, cancer	DN Elmo1	(148)
FYN	Phosphorylation (Y446)	+	change in conformation/intracellular localization	cancer, inflammation, glomerulonephritis, hypertension	FYN inhibitors, Y446 peptide	(3)
LYN	phosphorylation (Y120)	NK	changed dimerization	cancer, inflammation, glomerulonephritis	LYN inhibitors	(3)

nNOS, neuronal nitric oxide synthase; iNOS, inducible nitric oxide synthase; Cav-1, caveolin-1 protein; GPCR, G protein-coupled receptor; ELMO1, guanine nucleotide exchange factor for the GTPase Rac; FYN, LYN, nonreceptor tyrosine kinases; NK, not known; DN, dominant negative; Y, tyrosine; Asn, aspartic acid. +, Increase in COX2 enzymatic activity; −, decrease in COX2 enzymatic activity.

Summary

In summary, recent studies, apart from transcriptional and posttranscriptional regulation of COX2 gene expression (28, 130), have provided new evidence of posttranslational regulation of COX2 enzymatic activity (3, 22, 67, 85, 99, 113, 115, 131, 148, 150). Specifically, such studies demonstrated that Cav-1 protein and GPCR interactions with COX2 enzyme reduce COX2 protein expression levels by increasing COX2 proteasome degradation (16, 22, 42, 121, 122). Moreover, ELMO1 interaction with COX2 enzyme was shown to regulate COX2 activity by enhancing COX2-mediated fibronectin expression in human mesangial cells, possibly as a result of conformational changes in the enzyme (148). Additionally, FYN signaling increases COX2-mediated production of prostanoids in prostate cancer cells, through phosphorylation of Y446 residue, which either allosterically regulates activity of the enzyme or impacts subcellular localization of COX2 and thus PPIs, which further regulate its activity (3). LYN, another Src family kinase member, also phosphorylates COX2 but on Y120 residue, located on the dimerization domain of the enzyme, which could lead to increased dimerization and thus increased activity of the enzyme (3). Binding between COX2 and its protein regulators must be mediated by the specific PPI domains, and exact determinants controlling these interactions must be identified. Nonetheless, regulation of COX2 enzyme by Cav-1 protein and GPCRs could be important in resolving inflammation through downregulation of COX2 enzyme. Moreover, considering that COX2, ELMO1, and FYN are associated with susceptibility to glomerular disease (51, 74, 75, 102, 114, 123–125), ELMO1- and FYN-mediated regulation of COX2 enzyme could be important in contributing to the development of glomerulosclerosis. Since FYN similarly to COX2 is implicated in prostate cancer (7, 8, 20, 52, 59, 87, 104, 109, 132), FYN-mediated tyrosine phosphorylation of COX2 may contribute to prostate cancer as well as possibly other cancers, as well as related inflammatory diseases. Alternatively, ELMO1 regulation of COX2 enzyme could be important in cancer and other inflammation-related disease as well. Finally, both FYN and COX2 have been shown to be implicated in hypertension (143, 147); thus FYN-mediated regulation of COX2 could further be important for the manifestation of this disease. Overall, downregulation of COX2 enzyme by short peptides originating from known domains of GPCR and other proteins that mediate COX2 proteasome degradation may present as alternative therapeutic means of eliminating COX2 protein in disease. Additionally, ELMO1 and FYN inhibitors can be used to target the COX2 enzyme indirectly to further overcome side effects of COX inhibitors in treatment of disease. However, before utilizing such peptides and inhibitors in clinical use, their side effects and efficiency of delivery methods need to be carefully addressed. Finally, this new knowledge of posttranslational regulation of COX2 enzyme may lead to the development of novel diagnostic tools as biomarkers of disease.

GRANTS

Supported by National Institutes of Health Grants R01 DK-098159 and R03 CA-182114 to A. Sorokin.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

A.A. and A.S. drafted manuscript; A.A. and A.S. edited and revised manuscript; A.A. and A.S. approved final version of manuscript.