Cyclooxygenase (COX) is the enzyme catalyzing the rate-limiting step in prostaglandin synthesis, converting arachidonic acid into prostaglandin H2. Two isoforms of this enzyme have been identified. Cyclooxygenase-1 (COX-1) is constitutively expressed in many tissues and plays roles in tissue homeostasis. In most cells and tissues, cyclooxygenase-2 (COX-2) is an inducible isoform whose expression is stimulated by growth factors, cytokines, and tumor promoters (Kutchera et al., 1996; Smith et al., 1996). Despite the structural similarity between the two isoforms, COX-1 and COX-2 differ substantially in the regulation of their expression, and their roles in tissue biology and disease (Smith et al., 1996; Dubois et al., 1998). In the past decade, tremendous progress has been made in understanding the functional roles of COX-2 in cell growth, cell death, cell motility, and in cancer. In addition, studies with inhibitors of this enzyme have generated encouraging results. Non-specific COX inhibitors, the non-steroidal anti-inflammatory drugs (NSAIDs), and recently developed COX-2 specific inhibitors have shown significant effects in reducing the incidence and progression of tumors in both animal models and in treatment of cancer patients (Subbaramaiah et al., 1997; Taketo, 1998; Gupta and DuBois, 2000). This review is intended to summarize the major advances in this promising field. We apologize that some work has not been cited directly but this proved impossible as a result of the extensive recent advances in this field.
Cyclooxygenase-2 (COX-2) is the inducible isoform of cyclooxygenase, the enzyme that catalyzes the rate-limiting step in prostaglandin synthesis from arachidonic acid. Various prostaglandins are produced in a cell type-specific manner, and they elicit cellular functions via signaling through G-protein coupled membrane receptors, and in some cases, through the nuclear receptor PPAR. COX-2 utilization of arachidonic acid also perturbs the level of intracellular free arachidonic acid and subsequently affects cellular functions. In a number of cell and animal models, induction of COX-2 has been shown to promote cell growth, inhibit apoptosis and enhance cell motility and adhesion. The mechanisms behind these multiple actions of COX-2 are largely unknown. Compelling evidence from genetic and clinical studies indicates that COX-2 upregulation is a key step in carcinogenesis. Overexpression of COX-2 is sufficient to cause tumorigenesis in animal models and inhibition of the COX-2 pathway results in reduction in tumor incidence and progression. Therefore, the potential for application of non-steroidal anti-inflammatory drugs as well as the recently developed COX-2 specific inhibitors in cancer clinical practice has drawn tremendous attention in the past few years. Inhibition of COX-2 promises to be an effective approach in the prevention and treatment of cancer, especially colorectal cancer. J. Cell. Physiol. 190: 279–286, 2002. © 2002 Wiley-Liss, Inc.
COX-2 IN CONTROL OF CELL GROWTH
An expanding body of evidence indicates that the prostaglandin products of the COX-2 (or COX-1) pathway enhance cell proliferation and growth in both normal and tumor cells. Several species of prostaglandins, including prostaglandin E2 (PGE2) and prostaglandin I2 (PGI2), have been observed to increase DNA synthesis and cell proliferation in rat hepatocytes (Kimura et al., 2000). This action of prostaglandins may play an important role during liver regeneration. A recent report showed that COX inhibitors abolished liver regeneration, suggesting prostaglandin production and the signaling elicited from it are required during recovery from liver injury (Rudnick et al., 2001). In a mouse model of colonic injury, PGE2 has been shown to prevent chemically induced attenuation of epithelial cell proliferation (Tessner et al., 1998). These data suggest that the proliferation-promoting effect of prostaglandins may be critical during tissue injury.
COX-2 is upregulated in cancers of the colon, breast, lung, pancreas, and esophagus as well as squamous cell carcinoma of the head and neck (Eberhart et al., 1994; Wolff et al., 1998; Chan et al., 1999; Ratnasinghe et al., 1999; Tucker et al., 1999). The excessive production of prostaglandins in cancerous tissues, mainly PGE2, enhances tumor cell growth. COX-2 specific inhibitors have been shown to inhibit cell growth in a number of tumors including colon, skin epidermal, gall bladder, esophageal adenocarcinoma, and pancreatic cancer cells (Sheng et al., 1997; Molina et al., 1999; Grossman et al., 2000; Higashi et al., 2000; Souza et al., 2000). The action of these COX inhibitors depends on the COX-2 pathway in cultured colon cancer cell lines: these inhibitors significantly suppressed proliferation of cells that have a high level of COX-2 expression, but exerted minimal effects on proliferation in cells with a significantly lower level of COX-2 (Tsuji et al., 1996; Sheng et al., 1997). In contrast to the growth-promoting effect of PGE2 in many types of cells and tissues, this prostaglandin appeared to exert an opposite effect in a few other cell types. For example, induction of COX-2 inhibited cell growth in both human airway and arterial smooth muscle cells (Bornfeldt et al., 1997; Belvisi et al., 1998). These observations argue for a cell-type specific growth response to PGE2.
Several studies suggest that the COX-2 pathway may be involved in the signaling of other mitogens. It has been demonstrated that PGE2 release is a critical event in mediating the growth-promoting effect of growth factors and oncogenes (Coffey et al., 1997; Dermott et al., 1999). In addition, COX inhibitors blocked the cell proliferation effect of EGF in 3T3 cells, and this effect could be reversed by adding back exogenous PGE2 (Handler et al., 1990). Moreover, synergistic effects on cell proliferation and expression of the oncogene c-myc were observed when PGE2 treatment was combined with EGF. A recent study in the colon cancer cell line HCA-7 showed that blocking both the COX-2 and EGF-like HER-2/neu pathways synergistically reduced cancer cell growth (Mann et al., 2001). Interestingly, although prostaglandins are important for signaling of the EGF and EGF-related pathways, they have a minimal effect on mitogenesis in response to PDGF, implying that there is a specific coupling between EGF signaling and prostaglandins (Handler et al., 1990).
Prostaglandins exert their effects locally in both autocrine and paracrine patterns. PGE2 effects are mediated by a family of G-protein-coupled receptors, namely, EP1, EP2, EP3, and EP4 (Fig. 1). In some cell types, PPAR nuclear receptors are also involved in mediating the prostaglandin effects. The genes for the EP receptors have been cloned. Multiple alternative splicing variants in EP1 and EP3 were found, adding to the diversity of the EP receptor family (for a detailed discussion of EP receptors, see Breyer et al., 2001). The mechanism by which PGE2 interacts with a specific EP receptor subtype is not clear, but depends on the differential expression of each subtype of EPs in tissues and cells, binding affinity to PGE2, and differential activation of the receptors (Belley and Chadee, 1999; Takafuji et al., 2000; Stock et al., 2001). Functional studies on the EP receptors relied on two major strategies: use of subtype-specific agonists and antigonists and construction of EP knockout mice. EP1−/−, EP2−/−, and EP3−/− homologous knockout mice have been generated and they are viable (Ushikubi et al., 1998). EP4−/− mice have a neonatal lethal phenotype; EP4+/− heterozygous mice are viable (Segi et al., 1998). Signaling downstream from the different EP receptors varies, probably because each receptor interacts with different G proteins. The EP1 receptor usually signals through an increased level of intracellular Ca2+. Activation of the majority of EP3 variants results in a reduced level of cAMP, whereas EP2 and EP4 activation leads to an increased level of cAMP. Growth-promoting effects are signaled via EP receptors in several cell types. In primary keratinocytes, growth promotion was signaled via an EP2 and/or EP4 receptor-cAMP coupled response (Konger et al., 1998). In a recent study in colorectal carcinoma cells, PGE2 promoted cell growth and motility via the EP4 receptor by activation of the phosphatidylinositol 3-kinase (PI3K)/protein kinase B (Akt/PKB) pathway (Sheng et al., 2001). Interestingly, EP2 and EP3 were also expressed in the colorectal carcinoma cells and their binding affinities to PGE2 are similar to EP4. It is currently not understood how growth signaling was selectively directed to the EP4 receptor in this case. In addition, the signaling cascade connecting the EP4 receptor to activation of PI3K has not been established.
COX-2 IN PREVENTION OF APOPTOSIS
Another important action of COX-2 on cellular processes is inhibition of apoptosis, which in many cases constitutes another mechanism to promote tumor cell growth. The initial observation was that NSAIDs induced apoptosis in chicken embryo fibroblasts (Lu et al., 1995). Later studies revealed that COX-2-specific and non-specific NSAIDs caused apoptosis in a variety of cancer cells, including colon, gastric, glioma, pancreatic, and lung cancer cells (Ding et al., 2000; Hida et al., 2000; Joki et al., 2000; Uefuji et al., 2000; Li et al., 2001). A study with a panel of 15 colon cancer cell lines showed that the induction of apoptosis by NSAIDs was more significant in the colon cancer cells that express COX-2 (Li et al., 2001). These findings suggest that the COX-2 pathway might play a key role in preventing apoptosis in many cell types and NSAIDs reverse the response by inhibiting the COX-2 pathway. However, the interpretation of NSAID studies is complicated by possible COX-2-independent mechanisms. In several cases, NSAIDs only induced apoptosis at much higher concentrations than their IC50s. In addition, NSAIDs have been shown to trigger apoptosis in COX-2 deficient cells (Zhang et al., 1999), and NSAID metabolites that do not inhibit COX, such as sulindac sulfone, can also cause apoptosis (Piazza et al., 1995). Therefore, selective targeting of the COX-2 pathway via overexpression was used to address the preventive role of COX-2 in apoptosis more conclusively. It is found that overexpression of COX-2 significantly reduced the response of intestinal epithelial cells to certain apoptotic stimuli, and this effect could be reversed by sulindac sulfide, a COX inhibitor (Tsujii and Dubois, 1995). The reduction of apoptosis in cells highly expressing COX-2 was associated with an elevated expression of Bcl-2 and a reduced expression of transforming growth factor β2 receptor.
The preventive role of COX-2 in apoptosis is attributable to two possible mechanisms: It could be mediated by generation of prostaglandin products or mediated by removal of the substrate, arachidonic acid, via COX-2 catalytic activity. It is documented that the COX product PGE2 contributes to the preventive action of COX-2 in apoptosis. In human colon cancer cells, PGE2 inhibited apoptosis and also induced Bcl-2 expression (Sheng et al., 1998). There is also experimental support for the other mechanism—a high intracellular concentration of arachidonic acid induced apoptosis. Perturbation of the level of free arachidonic acid can be achieved via the COX-2 pathway, or other pathways that use or generate arachidonic acid (Surette et al., 1996, 1999; Chan et al., 1998; Fonteh et al., 2001; Chen et al., 2001a). NSAID treatment has been shown to lead to an accumulation of intracellular arachidonic acid at the same concentration that induced apoptosis in colon cancer cells (Chan et al., 1998). Apoptosis induced by arachidonic acid involves activation of caspase 3 and a transition of mitochondrial permeability (Scorrano et al., 2001). In an inducible cell system stably transfected with COX-2, we have demonstrated that induction of COX-2 expression or overexpression of another enzyme involved in arachidonic acid utilization, fatty acid CoA ligase 4 (FACL4) (Cao et al., 1998), prevented apoptosis induced by arachidonic acid. Strikingly, simultaneous induction of COX-2 and FACL4 exerted a more pronounced effect in inhibition of apoptosis (Cao et al., 2000). These findings strongly suggest that metabolic removal of arachidonic acid inhibits apoptosis. Therefore, COX-2 prevents apoptosis by generating an anti-apoptotic product, prostaglandin(s), as well as by removing a pro-apoptotic substrate, arachidonic acid.
What are the downstream signaling pathways for the preventive effect of COX-2 in apoptosis? The complete picture is not yet understood, but at least three pathways have been shown to be involved; the Bcl-2 mediated pathway, the nitric oxide pathway, and production of ceramide (Fig. 2). Bcl-2 is a key anti-apoptotic and anti-oxidant protein, and one of its cellular functions is to suppress lipid peroxidation (Hockenbery et al., 1993). Accumulating evidence indicates that induction of COX-2 correlates with increased Bcl-2 expression, and treatment with COX inhibitors leads to reduced Bcl-2 expression (Tsujii and Dubois, 1995; Liu et al., 1998; Sheng et al., 1998; Cao et al., 2000). The Bcl-2 mediated pathway is likely to be downstream of both arachidonic acid and prostaglandins. It acts to prevent peroxidation of certain intermediates, and therefore, blocks cell death that would be otherwise caused by these oxidized lipids. It is known that molecular cross-talk between the nitric oxide and the COX-2 pathways contributes to tissue homeostasis and plays key roles in pathophysiological processes. In a recent study, induction of COX-2 expression was found to prevent apoptosis induced by withdrawal of nerve growth factor in pheochromocytoma cells (McGinty et al., 2000). Using a differential expression array, the same group showed that apoptosis inhibition by COX-2 involved stimulation of expression of the dynein light chain followed by inhibition of neuronal nitric oxide synthase activity (Chang et al., 2000). The modulation of the nitric oxide pathway is downstream of PGE2, as the effect can be reproduced by addition of exogenous PGE2. With a cDNA microarray analysis in an epithelial cell model, we identified that COX-2 induction regulated the expression of another signaling molecule in the nitric oxide pathway and that this regulation occurred via generation of the PGE2 product (Cao et al., unpublished communications). Therefore, PGE2 production prevents programmed cell death by inhibiting nitric oxide signaling in certain types of cells. Intracellular accumulation of free arachidonic acid has been shown to elevate the level of a small lipid messenger, ceramide (Chan et al., 1998). Since ceramide has been demonstrated to be an effective inducer of apoptosis, removal of arachidonic acid by COX-2 might prevent apoptosis by reducing the concentration of intracellular ceramide. However, the timing with which ceramide level responded to apoptosis stimuli varied dramatically from one cell type to another (Dbaibo et al., 1993; Chan et al., 1998). We only observed a moderate increase (two-fold) in ceramide level in arachidonic acid-exposed epithelial cells (Cao et al., 2000). Therefore, whether ceramide production is an early signaling step or a late event in apoptosis awaits further investigation.
In addition to the above mechanisms, theoretically COX-2 could prevent programmed cell death by enhancing the pro-proliferative pathways, such as the Akt kinase-mediated pathway or the NFκB pathway. The COX inhibitors aspirin and sulindac can inhibit the activation of the NFκB pathway (Kopp and Ghosh, 1994; Grilli et al., 1996; Yin et al., 1998; Yamamoto et al., 1999), and the COX-2 specific inhibitor celecoxib induced apoptosis by inhibiting Akt activation (Hsu et al., 2000). However, since other COX inhibitors such as indomethacin do not have similar effects, it is probable that these agent-specific actions of COX-2 inhibitors are due to COX-independent mechanisms (Kopp and Ghosh, 1994; Grilli et al., 1996; Yin et al., 1998). Again, a better approach to this question would be overexpression or deletion of COX-2. We did not detect an activation of NFκB pathway when COX-2 expression was induced (Cao et al., unpublished communications). Whether induction of COX-2 activates the Akt pathway awaits further investigation. It has been reported that cyclopentenone prostaglandins can inhibit NF-κB activation via inhibition of the IκB kinase (Rossi et al., 2000; Straus et al., 2000). Recent work demonstrated that PGE2 promoted the transcriptional activity of the p65/Rel A subunit of NFκB (Poligone and Baldwin, 2001). These data seem contradictory, but suggest that there is positive and negative regulation of NFκB by different products of the COX-2 pathway.
COX-2 IN CELL MOTILITY AND ADHESION
COX inhibitors have been found to reduce tumor cell migration, cell adhesion, and tumor invasiveness in in vivo and in vitro experimental systems. These effects were observed in various cancer cells including liver, prostate, colon, and breast (Attiga et al., 2000; Jiang et al., 2001; Rozic et al., 2001; Chen et al., 2001b). The actions of COX inhibitors on these cellular processes are at least partially due to inhibition of the COX-2 pathway: in colon cancer Caco-2 cells stably transfected with a COX-2 cDNA construct, constitutive COX-2 expression leads to enhanced cell invasiveness compared to the parental Caco-2 cells (Tsujii et al., 1997). In addition, overexpression of COX-2 in rat intestinal epithelial cells increased cell adhesion to the extracellular proteins laminin and Matrigel (Tsujii and Dubois, 1995). The biochemical changes associated with these altered cellular dynamics include increased expression and activation of metalloproteinase-2 and reduced expression of E-cadherin (Tsujii and Dubois, 1995; Tsujii et al., 1997). However, whether these expression changes are critical mediators of the effects of COX-2 in cell motility needs to be further tested. A recent study in non-small cell lung cancer cell lines concluded that cell invasiveness promoted by COX-2 is mediated by the cell surface receptor for hyaluronate, CD44 (Dohadwala et al., 2001). Overexpression of COX-2 caused cell invasiveness and also increased CD44 expression. Moreover, cell invasion was significantly reduced in cells overexpressing COX-2, but where CD44 was specifically blocked. In colorectal carcinoma cells, the phosphatidylinositol-3-kinase/Akt kinase/protein kinase B pathway was recently found to mediate PGE2-promoted cell motility (Sheng et al., 2001). Based on these results, there are likely to be multiple, cell-type dependent mediating pathways for COX-2-promoted cell invasion. The multiple actions of COX-2 on cell motility and cell adhesion explain the key roles of COX-2 in tumor invasiveness and angiogenesis.
COX-2 AND CANCER
Substantial information from genetic and pharmacological studies indicates that COX-2 induction is an early step in tumorigenesis, particularly in colon cancer. Oshima et al. (1996) reported that in APCΔ716 knockout mice, a model of human familial adenomatous polyposis (FAP), a COX-2 null mutation dramatically reduced the number and size of the intestinal polyps. This result suggests that COX-2 is a rate-limiting step in the formation of intestinal polyps. In APC+/−Δ716 and min mutant mice, adenomas at early stages showed loss of heterozygosity at the normal APC allele and COX-2 overexpression (Oshima et al., 1995). Moreover, COX-2 expression was not detected in polyps smaller than 2 mm (Oshima et al., 1996), suggesting that COX-2 induction occurs after the loss of the second APC allele. This implies that COX-2 is not involved in tumor initiation, but in the promotion of colorectal adenomas (Prescott and White, 1996).
Genetic evidence supporting a key role of COX-2 in tumor development also comes from studies with transgenic mice that tissue-specifically express COX-2. A recent report examined the effect of overexpression of COX-2 under the control of a promoter that is specifically activated in basal keratinocytes of interfollicular epidermis and basal cells of the pilosebaceous unit. The results revealed that transgenic COX-2 expression in skin caused epidermal hyperplasia and dysplasia at discrete body sites (Neufang et al., 2001). In addition, the causal role of COX-2 in tumor development was demonstrated in mammary tissues: COX-2 transgenic mice that specifically express COX-2 in the mammary gland were established using the murine mammary tumor virus promoter. Transgenic multiparous females had a much higher incidence of focal mammary gland hyperplasia, dysplasia, and metastatic tumors (Liu et al., 2001). These data strongly suggest that upregulation of COX-2 expression is sufficient to induce tumorigenesis in certain types of tissues.
The tumor-promoting effect of COX-2 is thought to be a consequence of the multiple actions of COX-2 on cells as described above, i.e., promotion of cell growth and prevention of apoptosis by induction of COX-2 leads to increased size and number of tumors. In addition, enhancement of cell adhesion and motility are mechanisms responsible for the roles of COX-2 in tissue invasion and angiogenesis (see discussion below). The COX-2 tumor-promoting effect is partly mediated through its prostaglandin products, since deficiency prostaglandin receptor EP2 gene also reduced the size and number of intestinal tumors in APCΔ716 mice, similar to deficiency of the COX-2 gene (Sonoshita et al., 2001). How is COX-2 expression induced at a certain stage of adenoma formation? The answer is not clear, but several hypotheses have been formed based on available data: one hypothesis proposes that APC or β-catenin binds to or regulates the transcription factor LEF-1, and LEF-1 may activate COX-2 transcription by itself or by forming a transcriptional complex with other factors (Prescott and White, 1996). This hypothesis is supported by a study in human colorectal cancer showing that APC mutations were associated with enhanced COX-2 expression and an elevated nuclear β-catenin level (Dimberg et al., 2001). Although no data yet indicate that LEF-1 directly binds to the COX-2 promoter, a recent study demonstrated that the transcription factor PEA3 activated the COX-2 promoter in response to Wnt1. This implies that PEA3 might also upregulate COX-2 in response to APC mutation (Howe et al., 2001). Other hypotheses suggest that tyrosine kinases that respond to growth factors or protein kinase C are upstream pathways that activate COX-2 transcription (Subbaramaiah et al., 1997). Supportive data for this hypothesis includes the facts that mitogen-activated protein kinase (MAPK) and extracellular signal-regulated kinase (ERK) mediated COX-2 upregulation by transforming growth factor-alpha (Matsuura et al., 1999). It also includes that both transcriptional and post-transcriptional regulation of COX-2 involved the MAPK pathway in colon carcinoma cells (Shao et al., 2000).
Classically, the COX-1 isoform is thought to be constitutively and universally expressed. It is believed to exert important functions in tissue homeostasis, but to have minimal effects on inflammation and carcinogenesis. However, a recent study with COX-1 deficient mice showed that lack of COX-1 significantly reduced intestinal tumorigenesis in min mice, a phenotype similar to that of COX-2−/− mice (Chulada et al., 2000). This genetic evidence strongly suggests that COX-1 plays a role in tumorigenesis and that the capacity of prostaglandin production in cells is a key regulatory point during adenoma development.
COX-2 inhibitors in animal experiments
The effects of non-specific COX-2 inhibitors in tumorigenesis have been extensively examined in three types of animal models: mice with intestinal neoplasia, e.g., min or APCΔ716 mice, rats treated with the chemical mutagen azoxymethane (AOM), and nude mice subjected to tumor xenografts. The results indicate that NSAIDs are very effective in reducing the number and size of tumors in all three models. After COX-2-specific inhibitors became available, their effectiveness in inhibiting tumorigenesis, especially in colon cancer, was evaluated in all the three animal models. Celecoxib decreased tumor number and size significantly in the min mouse model of adenomatous polyposis (Jacoby et al., 2000). Rofecoxib was shown to be a potent chemopreventive agent in the ApcΔ716 mouse (Oshima et al., 2001). Moreover, administration of celecoxib during both initiation/post-initiation stages and promotion/progression stages significantly suppressed the incidence and multiplicity of AOM-induced adenocarcinomas in rats (Reddy et al., 2000). This strongly suggests that the COX-2-specific inhibitor celecoxib may potentially be an effective chemopreventive agent for primary as well as secondary prevention of colon cancer.
The tumor inhibitory effects of COX-2 inhibitors involve COX-2-dependent and independent pathways. There is evidence that some of the anti-proliferative and anti-neoplastic effects of NSAIDs are independent of COX inhibition. NSAIDs reduced tumorigenesis in COX-null cells (Zhang et al., 1999) and sulindac sulfone, a metabolite of sulindac that does not inhibit COX activity, induced apoptosis by inhibiting cGMP-dependent phosphodiesterase (Piazza et al., 1995).
COX-2 inhibitors in cancer clinical practice
Sulindac, a COX-2 non-specific inhibitor, was often used in the early clinical trials of COX-2 inhibitors in patients with FAP. The therapeutic effects of sulindac on FAP patients were summarized in a recent review (Taketo, 1998). In all the clinical studies, sulindac significantly reduced the number and size of the polyps, however, severe side effects, e.g., gastric bleeding, were present in some FAP patients under treatment. Primarily for this reason, great effort has been employed to develop COX-2 specific inhibitors. These new inhibitors showed great effectiveness in tumor inhibition in animal models, and a recent study describing the use of celecoxib in FAP patients is encouraging: celecoxib caused a significant reduction in the number of colorectal polyps (Steinbach et al., 2000). More importantly, unpublished data indicates that the incidence of ulcers was reduced about 50% in patients treated with rofecoxib or celecoxib, compared to those treated with traditional NSAIDs (Sundy, 2001). These results suggest that COX-2 specific inhibitors may define a new generation of chemopreventive and chemotherapeutic agents. However, it will take more extensive clinical studies to determine whether COX-2 specific inhibitors are superior to NSAIDs in cancer chemoprevention and, if so, what the cost implications will be.
COX-2 and angiogenesis
In addition to a causal role of COX-2 in tumorigenesis, compelling evidence also indicates that it plays an important role in tumor angiogenesis, a process involved in later stages of tumor development and migration. The evidence comes from three types of experiments. First, both non-specific and specific COX-2 inhibitors significantly inhibited angiogenesis (Jones et al., 1999; Masferrer et al., 2000), and this action of the inhibitors involved inactivation of MAPK (Jones et al., 1999). Second, COX-2 expression significantly induced the production of vascular endothelial growth factor (VEGF), a proangiogenic growth factor, and this induction could be blocked by a COX-2 specific inhibitor, NS-398 (Tsujii et al., 1998; Masferrer et al., 2000). In a transmembrane system, Tsujii et al. (1998) demonstrated that increased COX-2 expression in colon cancer cells stimulated these epithelial cells to promote endothelial cell mobility and formation of tubular structure, and that COX-2 inhibitors reversed this effect. Third, genetic studies with COX-2−/− APCΔ716 and EP receptor−/− APCΔ716 double deficient mice strongly suggest that polyp stromal expression of COX-2 plays a key role in promoting tumor angiogenesis in intestinal tumors (Sonoshita et al., 2001). In this report, the authors found that disruption of the gene for EP2, but not other PGE2 receptors, caused a decrease in the number and size of intestinal polyps in APCΔ716 mice, similar to deficiency of the COX-2 gene. Moreover, in polyps of EP2−/− APCΔ716 as well as COX-2−/− APCΔ716 mice, induction of VEGF and angiopoietin-2 were completely inhibited.
Based on current data, we can propose a mechanism for the COX-2 promoted tumor angiogenesis. When COX-2 expression is induced in polyp stromal cells, increased PGE2 is generated and signals through the EP2 receptor located on the membrane surface. An elevated level of cAMP, a known downstream event of EP2 activation, stimulates VEGF production, and thereby promotes neovascularization and tumor angiogenesis. Other studies suggest that COX-1 is also involved in endothelial cell migration (Tsujii et al., 1998; Jones et al., 1999), and that other products of the COX pathway, thromboxane A2 and PGI2, also play roles in promotion of angiogenesis (Daniel et al., 1999; Gately, 2000). Nevertheless, COX-2 inhibitors and EP2 antagonists might prove to be effective chemotherapeutic agents to inhibit tumor angiogenesis, which could have application in multiple tumors as adjunctive therapy.
Significant advances have been made in the past five years in understanding the COX-2 pathway. Establishment of genetic models, delineation of downstream signaling events, and characterization of the EP prostaglandin receptors are hallmarks of research advances in this field. The mechanism of COX-2 action is complicated since this pathway exerts diverse cellular effects and often these effects are cell type-specific. In the near future, we expect to see studies address questions that are crucial to understanding the role of the COX-2 pathway in cellular functions and tumorigenesis. First, what are the downstream signaling pathways for the preventive role of COX-2 in apoptosis? These will include signaling routes downstream of the COX-2 prostaglandin products as well as those governing the level of free arachidonic acid. Both linear signaling cascades and cross-talk between pathways need further investigation. Second, how is PGE2 coupled to a specific EP receptor considering the fact that multiple EP receptors are often available on the same cell surface? What is the biological significance of the splicing variants of EP receptors? Third, how is COX-2 induced in cancer cells? If the mechanism is transcriptional, what transcription factors are involved in regulating the COX-2 promoter? It will also be of great interest to see if COX-2 is sufficient for colon carcinogenesis. Fourth, what specific role does COX-1 play during tumorigenesis? Is the key modulator for adenoma development the total prostaglandin production or the different prostaglandin products derived from the COX-1 and COX-2 pathways? Do COX-1 and COX-2 contribute to adenoma development at the same or different stages of tumorigenesis? Finally, what are the COX-2-dependent vs. COX-2-independent effects of the COX-2-specific inhibitors? Searching for answers to the above questions will reveal the cellular mechanisms of COX-2 action and will provide a molecular basis for understanding the effectiveness of the use of COX-2 non-specific and specific inhibitors in different stages of cancer treatment.