Cyclo-oxygenase-2 Plays a Critical Role in UV-induced Skin Carcinogenesis


  • This paper is part of a special issue dedicated to Professor Hasan Mukhtar on the occasion of his 60th birthday.

*Corresponding author email: (Susan M. Fischer)


Besides induction of DNA damage and p53 mutations, chronic exposure to UV irradiation leads to the constitutive up-regulation of cyclo-oxygenase-2 (COX-2) expression and to increased production of its primary product in skin, prostaglandin E2 (PGE2). COX-2 has also been shown to be constitutively overexpressed in mouse, as well as human, UV-induced skin cancers and premalignant lesions. UV exposure results in ligand-independent activation of the epidermal growth factor receptor and subsequent activation of mitogen-activated protein kinase and phosphatidylinositol 3-kinase/Akt pathways leading to transcriptional activation of the COX-2 gene. Use of COX-2-specific inhibitors and genetic manipulation of COX-2 expression have demonstrated that UV induction of COX-2 in the skin contributes to the induction of epidermal hyperplasia, edema, inflammation, and counters the induction of apoptosis after UV exposure. Likewise, inhibition of COX-2 activity or reduced expression in COX-2 knockout mice resulted in significantly reduced UV-induced tumorigenesis, while overexpression of COX-2 in transgenic mice enhanced UV-induced tumor development. A combination of signaling from the PGE2 EP1, EP2 and/or EP4 receptors mediates the effects of COX-2 overexpression. These studies demonstrate the crucial role of COX-2 in the development of UV-related nonmelanoma skin cancers.


Cyclo-oxygenases (COXs) are enzymes that catalyze the first and rate-limiting step in the conversion of arachidonic acid into prostaglandins (PGs), prostacyclins and thromboxanes, all of which are bioactive lipids. Arachidonic acid is hydrolyzed from membrane phospholipids by phospholipase A2 in response to various stimuli. COXs contain separate COX and peroxidase activities, with the COX activity inserting two molecules of O2 into arachidonic acid to form PGG2 and the peroxidase activity then reducing PGG2 to PGH2 with a two-electron transfer (1). The unstable PGH2 intermediate is then converted into the biologically active endproducts by specific PG and thromboxane synthases, which are differentially expressed in different tissues (2).

There are two major isoforms of COX, COX-1 and COX-2, which are encoded by separate genes. COX-1 is generally constitutively expressed in most tissues and its expression usually does not vary greatly in the adult animal (1). The PG products of COX-1 are involved in normal physiologic functions, such as maintenance of the gastric mucosa and regulation of renal blood flow (3). On the other hand, COX-2 expression is generally undetectable in most unperturbed adult epithelial tissues, but can be highly induced by multiple mitogenic and inflammatory stimuli, including growth factors, cytokines, hormones, serum, hypoxia, bacterial endotoxins, tumor promoters and UV light (1,4,5). PGs produced by COX-2 are involved in pathophysiological functions such as inflammation, pain, fever, wound repair, angiogenesis, vasodilation and increased vascular permeability (1,6). More recently, a third isoform, COX-3, has been identified (7,8). COX-3 is a splice variant of COX-1 (sometimes called COX-1b), in which intron 1 is retained resulting in the insertion of 30–34 amino acids and a shift in the reading frame in mice and humans (7,8). COX-3 is most highly expressed in the brain, kidney and heart, but because of the frame shift, this splice variant encodes a completely unrelated protein that lacks COX activity (7).

UV and Skin Cancer

Skin cancer is one of the most common forms of cancer in the world and is especially prevalent among Caucasians and other light-skinned races. UV radiation (in particular UVB wavelengths, 280–320 nm) from sun exposure is the major environmental causative agent in the induction of nonmelanoma skin cancer, which includes squamous cell carcinomas (SCCs) and basal cell carcinomas (BCCs) (9,10).

A mouse model frequently used for UV carcinogenesis studies is the SKH-1 hairless mouse. Chronic exposure of these mice to UVB irradiation leads to the development of benign epidermal tumors, most of which become SCCs (11). This model has been extremely useful in elucidating the mechanisms of UV-induced carcinogenesis. Acute exposure of SKH-1 mice to UVB leads to the production of DNA damage in epidermal cells in the form of strand breaks, cyclobutane thymidine dimers and pyrimidine (6-4) pyrimidone photoproducts (9,10). In addition, absorption of UV light by molecules in the cells results in the generation of reactive oxygen species (ROS), which can cause oxidative DNA damage, often in the form of 8-oxo- or 8-hydroxy-deoxyguanosine adducts (12,13). This DNA damage leads to an increase in the number of epidermal cells with wild-type p53 expression (peaking 8–12 h after exposure), which is paralleled by increases in p21 expression, appearance of apoptotic sunburn cells and a reduction in proliferation (bromodeoxyuridine [BrdU] labeling) (10). Recruitment of inflammatory cells into the dermis begins about 4 h after UVB exposure and neutrophil infiltration remains high for several days (10). After either repair of the damaged DNA or induction of apoptosis, epidermal proliferation returns with marked increases in BrdU labeling by 48 h (10). Chronic or multiple UVB exposures lead to an increase in epidermal proliferation and thickness (14) and can also lead to p53 mutations and/or allelic loss (9,11). Increased clusters of p53-positive cells and distinctive UV-induced p53 mutations and/or allelic loss of p53 have been detected after chronic UV exposure in early premalignant skin lesions and SCCs in SKH-1 mice (11,15–17), as well as in human premalignant actinic keratosis lesions and skin SCCs (9,18). These data demonstrate that UV-induced mutations and alterations in p53 expression are early events in UV-induced skin carcinogenesis and are pivotal in the carcinogenic process.

COX-2 Expression in Skin Cancer

In addition to p53 alterations, other UV-induced events are also considered crucial to the process of skin cancer development. One of those events is the up-regulation of COX-2 expression leading to increased levels of its primary PG product in skin, PGE2. While COX-2 is usually undetectable in most normal adult tissues including skin, it has been shown to be highly overexpressed in a number of different human cancers, as well as in numerous animal models of carcinogenesis (2,3). In human skin, COX-2 expression has been shown to be elevated in actinic keratoses, SCCs and BCCs (5,19). Similarly, in the SKH-1 mouse model, COX-2 is overexpressed in hyperplastic skin, benign papillomas and in SCCs resulting from chronic UV exposure (19,20). Like p53 alterations, constitutive overexpression of COX-2 occurs early in the process of UV-induced carcinogenesis.

Indeed, COX-2 has been shown to be transiently induced in human and mouse skin after a single acute exposure to UV radiation (19,21). Coincident with COX-2 up-regulation, primarily in the basal layer, is an increase in PGE2 production and an increase in keratinocyte proliferation and apoptosis (21–23). COX-2 overexpression as well as treatment with exogenous PGE2 have been shown to induce proliferation in vitro in a number of different cell types, including human keratinocytes (4). Likewise, multiple studies have shown that COX-2/PGE2 have antiapoptotic properties in various cell types (4). In addition, PGE2 mediates signals involved in the induction of inflammation, angiogenesis, vasodilation, and vascular permeability (6). All of these downstream effects of COX-2 overexpression and PGE2 signaling promote the development of UV-induced skin carcinogenesis.

UV Induction of COX-2 Expression

The molecular mechanisms by which UV irradiation leads to the induction of COX-2 expression have been elucidated with a fair amount of detail using both cultured keratinocytes and SKH-1 mice skin in vivo and this is summarized in part in Fig. 1. The electromagnetic energy of UV light is absorbed by molecules within the cell, and this energy is then transferred to molecular oxygen generating ROS. Among its other actions, ROS readily reacts with conserved cysteines in the active site of protein tyrosine phosphatases. In particular, ROS leads to the oxidative inhibition of receptor-type protein tyrosine phosphatase-κ, which normally keeps the epidermal growth factor receptor (EGFR) in an unphosphorylated, inactive state (24). Thus, inactivation of receptor-type protein tyrosine phosphatase-κ by ROS leads to ligand-independent activation of EGFR, which in turn activates numerous downstream signaling pathways including Ras/Rac1/p38 mitogen-activated protein kinase (MAPK) and phosphatidylinositol 3-kinase (PI3K)/Akt (25,26). Besides activation of the p38 MAPK pathway, Rac1 activation also induces additional ROS generation through the activation of NADPH oxidase (27).

Figure 1.

 Molecular mechanisms by which UVB light induces cyclo-oxygenase-2 (COX-2) expression. (a) Absorption of UV electromagnetic energy by molecules within a cell is transferred to molecular oxygen-generating ROS. These lead to the oxidative inactivation of receptor-type protein tyrosine phosphatase-κ (RTPTPκ), which normally keeps the epidermal growth factor receptor (EGFR) in a dephosphorylated, inactive state. With RTPTPκ inactivated, EGFR becomes activated in a ligand-independent manner, which then results in activation of the phosphatidylinositol 3-kinase (PI3K)/Akt and Ras/Rac1/p38 mitogen-activated protein kinase (MAPK) pathways. In addition to activating the p38 MAPK pathway, activation of Rac1 also leads to the generation of more ROS by activating NADPH oxidase. p38 activation leads to the phosphorylation of the cyclic AMP response element (CRE) binding protein (CREB) at Ser-133, while activation of Akt phosphorylates, and so inactivates, glycogen synthase kinase-3β (GSK3β), which normally phosphorylates CREB on Ser-129. Phosphorylation of CREB at Ser-129 is inhibitory, while phosphorylation of CREB at Ser-133 activates its transcriptional activity. Thus, the combined actions of GSK3β inactivation and p38 activation allow CREB to bind to the CRE site in the COX-2 promoter and activate its transcription. (b) Absorption of UVB energy by the chromophore tryptophan (Trp) results in the formation of 6-formylindolo[2,3-b]carbazole (FICZ), a ligand for the arylhydrocarbon receptor (AhR). Normally sequestered in the cytosol in an inactive complex, which includes heat shock protein 90 (HSP) and Src, AhR is released after binding FICZ, allowing it to translocate to the nucleus and transactivate gene transcription. Src released from the complex can then activate EGFR and its downstream signaling pathway Ras/Raf/MEK/Erk1/2 leading to transcriptional up-regulation of the COX-2 gene.

p38 activation results in the phosphorylation of cyclic AMP response element (CRE) binding protein (CREB) and activating transcription factor-1, which then bind to the CRE site in the COX-2 gene promoter, activating transcription (28). PI3K activation by EGFR leads to phosphorylation (and thus activation) of Akt at both Thr-308 and Ser-473, which then in turn phosphorylates glycogen synthase kinase-3β (GSK-3β) at Ser-9, inactivating GSK-3β (27,29). GSK-3β normally phosphorylates CREB at Ser-129, which is an inhibitory phosphorylation site, and thus inactivation of GSK-3β by Akt leads to dephosphorylation of CREB at Ser-129, relieving CREB inhibition (27). This together with phosphorylation via p38 signaling of CREB at Ser-133, which is activating, allows CREB to bind to the CRE site on the COX-2 promoter, to recruit its coactivator, CREB binding protein, and to interact with the basal transcriptional machinery (27). By using various signaling pathway inhibitors and overexpressing wild-type or dominant negative specific kinases, both the PI3K/Akt and the p38 MAPK pathways have been shown to be necessary for the maximal induction of COX-2 expression by UVB (27–29). This pathway for UVB-induction of COX-2 is summarized in Fig. 1a.

Recently, another mechanism for the activation of EGFR and induction of COX-2 by UVB irradiation has been identified in which the arylhydrocarbon receptor (AhR) mediates UV responses (Fig. 1b) (30). Absorption of UVB energy by intracellular tryptophan converts it to the photoproduct, 6-formylindolo[2,3-b]carbazole (FICZ), which is a ligand for AhR. Binding of FICZ to AhR releases AhR from an inactive complex, which includes heat shock protein 90 and associated proteins such as Src. Src can then translocate to the cell membrane and through yet-to-be identified mechanisms activate EGFR. Stimulation of EGFR and subsequent internalization results in activation of the Ras/Raf/MEK/extracellular signal-regulated kinase (ERK)1/2 pathway leading to up-regulation of COX-2 gene transcription (30,31). Treatment of keratinocytes with FICZ was able to induce COX-2 expression, while inhibition or knockdown of AhR, tryptophan starvation or Src inhibition was able to block (at least in part) UVB-induced COX-2 expression (30).

Another group has also shown that the upstream stimulatory factor (USF)-1 and USF-2 transcription factors, which recognize Ebox elements, bind to the overlapping CRE-Ebox site in the COX-2 promoter in UVB-irradiated keratinocytes and that overexpression of USF-2 enhances COX-2 promoter activity (32). However, the mechanism by which UV enhances USF DNA binding and/or transcriptional activity has not been elucidated. We also found that USF-1 and -2 binding to the CRE-Ebox site in the COX-2 promoter was important for the constitutive overexpression of COX-2 in a mouse SCC cell line and that ectopic expression of USF-1 and -2 enhanced COX-2 promoter activity (33).

In addition, COX-2 expression can be induced by a p53-mediated activation of the Ras/Raf/ERK pathway. Genotoxic stress, such as UVB-induced DNA damage, induces p53 expression, which leads to induction of heparin-binding EGF (HB-EGF) expression (34). HB-EGF then activates the Ras/Raf/ERK pathway via binding to and activation of EGFR (34). Dominant negative forms of Ras or Raf, as well as a MEK inhibitor all blocked p53-dependent up-regulation of COX-2 expression (34). The antiapoptotic activity of COX-2/PGE2 prevents in part the induction of apoptosis by DNA damage/p53, as treatment of cells with a COX-2 inhibitor enhanced genotoxic stress-induced apoptosis (34). Thus, p53 induction of COX-2 may be a mechanism to mitigate the cellular stresses associated with DNA damage/p53 expression.

In all likelihood, a combination of UVB-elicited events including ROS generation, AhR activation and HB-EGF induction leads to the activation of EGFR and its downstream signaling pathways (PI3K/Akt and MAPK) that result in the induction of COX-2 gene expression by the transcription factors CREB and USF-1 and -2.

Although UVB wavelengths are primarily responsible for UV-induced DNA damage and can induce COX-2 expression, UVA wavelengths (320–400 nm) have also been shown to induce COX-2 (35). In fact, it has been estimated that approximately 70% of solar light induction of COX-2 expression and PGE2 production is the result of wavelengths in the UVA-2 region (320–350 nm) (35). UVA initiates signaling at the cell membrane in raft microdomains, which are rich in cholesterol, sphingomyelin and caveolin-1 (36). UVA induces production of ROS causing lipid peroxidation in the cell membranes, as well as activation of phospholipase A2 (37). In addition, singlet oxygen generated by UVA leads to the transient activation of p38 MAPK, which is responsible for increased COX-2 protein levels through the stabilization of COX-2 mRNA (37). Activation of c-Jun N-terminal kinase (JNK) also appears to contribute to UVA-induced COX-2 expression, since a JNK inhibitor was able to block COX-2 induction by UVA in artificial human epidermis (35). Thus, UV irradiation of the skin leads to activation of multiple pathways that result in up-regulation of COX-2 expression and PGE2 production.

COX Inhibitors Suppress UV-Induced Skin Inflammation and Carcinogenesis

To demonstrate that UV induction of COX-2/PGE2 expression is important for UV-induced skin responses and carcinogenesis, various nonsteroidal anti-inflammatory drugs (NSAIDs), which are inhibitors of COXs, have been used in the SKH-1 and other mouse models. Both indomethacin, a general COX inhibitor, and celecoxib, a COX-2-selective inhibitor, when fed to mice in the diet, were found to significantly inhibit UV-induced tumorigenesis in terms of number of tumors/mouse and tumor size (38,39). Similarly, topical treatment of mice with celecoxib immediately following each UV irradiation reduced the number of tumors/mouse and tumor size (13). Papillomas that did develop on mice treated with UV plus topical celecoxib had reduced levels of PGE2 and reduced the number of proliferating cells and p53-positive cells compared with papillomas from mice not treated with celecoxib (13). Even when celecoxib treatment (dietary or topical) was initiated after tumors had developed and UV exposures were stopped, celecoxib was able to inhibit new tumor formation, although there was little or no regression of established tumors (39–41). Thus, COX-2 inhibition appears to have better chemopreventive than therapeutic activity, even though UV-induced tumors have high COX-2 expression.

Other studies have shown that COX-2 inhibition prevents much of the skin damage induced by acute UV exposure. Topical treatment of SKH-1 mice with celecoxib immediately after a single exposure or after each of three exposures to UVB irradiation was shown to significantly reduce UV-induced skin edema, dermal neutrophil infiltration and activation (induction of myeloperoxidase activity), number of sunburn cells, number of p53-positive cells, epidermal proliferation (number of proliferating cell nuclear antigen-positive cells) and oxidative DNA damage (number of 8-oxo-deoxyguanosine-positive cells) (13,42). These celecoxib effects all correlated with reduced UV-induced PGE2 production in the skin, although celecoxib had no effect on UV-induced COX-2 expression, as expected (42). On the other hand, 1 week of dietary celecoxib prior to a single acute UV exposure reduced UV-induced skin PGE2 levels, but did not affect UV-induced epidermal hyperplasia, proliferation (BrdU labeling), edema or inflammatory cell infiltration (38). The discrepancies in these results may be due to the route of administration of celecoxib and the effective dose of celecoxib achieved in the skin, although both routes reduced UV-induced epidermal PGE2 production. In another study, a different COX-2-selective inhibitor, SC-791, fed in the diet reduced UV-induced epidermal proliferation (BrdU labeling) and hyperplasia, and enhanced UV-induced apoptosis, demonstrating the proliferative and antiapoptotic effects of COX-2/PGE2 (21).

Together the above studies demonstrate that inhibition of COX-2 activity strongly suppresses UV-induced skin damage, inflammation and carcinogenesis. This is relevant to the development of human skin cancer as it has been demonstrated that regular users of NSAIDs have a reduced risk of developing actinic keratoses and cutaneous SCCs (43). However, the use of COX-2 inhibitors is not definitive for elucidating the contribution of COX-2 expression/activity on carcinogenesis as these inhibitors have been shown to have COX-2-independent effects (44). Therefore, genetic approaches have also been used to specifically modify expression of the COXs.

COX-2 Gene Expression Levels Affect UV Carcinogenesis

Stable overexpression of COX-2 in a human BCC cell line with low basal expression of COX-2 resulted in elevated PGE2 production and a dramatic resistance to UV-induced apoptosis compared to vector-only transfected cells (45). Conditioned media from COX-2 overexpressing BCC cells contained higher levels of the angiogenic factors, vascular endothelial growth factor and basic fibroblast growth factor, and exhibited higher angiogenic activity than media from vector-only cells. The COX-2 clones also had greater anchorage-independent growth in soft agar and tumor growth and angiogenesis when inoculated into immunodeficient mice (45). These results demonstrate that overexpression of COX-2 enhances tumorigenesis of BCC cells by its antiapoptotic and angiogenic effects.

COX-1 and COX-2 knockout mice have been generated and used to define each COX’s contribution to UV-induced skin changes and carcinogenesis. One limitation of these models is that homozygous COX-2 knockout mice do not survive beyond 2–6 months of age (6–8 weeks of age on the SKH-1 background) due to renal failure (23,46,47), and thus are not useful for long-term UV studies. Studies using COX-1 knockout mice that had been backcrossed onto the SKH-1 background showed that COX-1 null mice had reduced UV-induced epidermal PGE2 production and enhanced UV-induced apoptosis (48). However, there was no difference in UV-induced tumorigenesis compared with wild-type mice (48). Likewise, we found no differences in UV-induced tumor incidence (percent mice with tumors) or multiplicity (tumors/mouse) in heterozygous COX-1 knockout mice on an SKH-1 background compared with wild-type SKH-1 mice (23). Although UV-induced PGE2 production was reduced in the COX-1 (+/−) mice, UV-induced apoptosis was similar to that in wild-type mice (23). Another group using homozygous COX-1 knockout mice on their original 129/BL6 mixed background also found that UV-induced apoptosis was not enhanced in COX-1 knockout mice, even though PGE2 production (with or without UV) was reduced compared to wild-type mice (22).

To examine the effect of COX-2 deficiency on UV carcinogenesis we used heterozygous COX-2 knockout mice backcrossed to SKH-1. Latency to the appearance of the first tumor was increased, tumor multiplicity was decreased by 50–65%, and tumors were smaller in the COX-2 (+/−) mice compared with wild-type mice (23). Short-term UV experiments showed that COX-2 (+/−) mice had reduced UV-induced PGE2 production and epidermal proliferation (Ki67 immunostaining), as well as enhanced UV-induced apoptosis (23). Short-term experiments with homozygous COX-2 knockout mice (mixed 129/BL6 background) also demonstrated that UV-induced apoptosis was increased and epidermal thickness and proliferation were decreased in the knockout animals relative to wild-type mice (22).

To complement the UV studies using COX knockout mice, we also generated COX-2 overexpressing transgenic mice. COX-2 expression was directed to the epidermis using the keratin 14 promoter and resulted in about a two-fold increase in basal epidermal PGE2 levels (49). These transgenic mice, generated on an FVB background, were backcrossed onto the SKH-1 background for UV studies. Tumor latency was decreased and tumor multiplicity increased in the COX-2 overexpressing transgenic mice compared to wild-type mice (23).

Together these studies clearly demonstrate that susceptibility to UV-induced skin carcinogenesis is correlated with COX-2 gene copy number and expression levels. On the other hand, COX-1 gene/expression has no effect on UV-induced tumorigenesis, even though it contributes to UV-induced PGE2 production. Reduced COX-2 expression in COX-2 heterozygous knockout mice resulted in reduced UV-induced tumorigenesis, while overexpression of COX-2 in transgenic mice resulted in enhanced tumorigenesis. Overexpression of COX-2 in BCC cells enhanced tumorigenesis by its antiapoptotic and proangiogenic activities. Reduced UV tumorigenesis in COX-2 deficient mice correlated with reduced proliferation and enhanced apoptosis in response to UV.

Role of PGE2 EP Receptors in UV Carcinogenesis

Now that it is well established that UV induction of COX-2 expression and its primary product, PGE2, are important mediators of UV-induced carcinogenesis, attention has turned to the downstream effectors of PGE2. PGE2 can bind and activate four G-protein-coupled receptors called EP1–EP4 (50). Each EP receptor is coupled to different G-proteins leading to activation of various downstream mediators. EP1 activation leads to increased intracellular calcium and activation of phospholipase C; EP2 and EP4 are both coupled to adenylate cyclase leading to increased cyclic AMP levels; and EP3, which has several splice variants, can activate or inhibit adenylate cyclase or increase intracellular calcium levels depending on the splice variant expressed (50,51).

In SKH-1 mouse skin, immunohistochemical staining has shown that EP1 receptor expression is low, EP2 is patchy, EP3 is moderate and EP4 is undetectable (52). In human skin, EP1, EP2 and EP3 expression has been demonstrated throughout the epidermis (53,54). UV irradiation strongly induced expression of EP1 primarily in the suprabasal layers, while EP3 expression was reduced to undetectable levels in SKH-1 mice (52,55). Similarly, in UV-induced mouse skin tumors (papillomas and SCCs), EP1 mRNA and protein expression was elevated compared to normal skin, while EP3 expression was reduced (52). In human skin SCCs, all four EP receptors were shown to be expressed by immunostaining, with expression of EP1, EP2 and EP4 mRNAs up-regulated compared to normal skin (52). On the other hand, in both mouse and human BCCs, there was little or no detectable immunostaining of any of the EP receptors (52).

Genetic and pharmacologic manipulation of EP receptor expression and activity, respectively, has been used to determine which EP receptors are responsible for mediating UV-induced skin changes and tumorigenesis. In one study, UV-induced ear swelling (edema), inflammatory cell infiltration and blood flow (all markers of inflammation) were all reduced in EP2 and EP4 knockout mice or wild-type mice treated with an EP4 antagonist (56). Another study with EP2 knockout mice showed that UV-induced proliferation (BrdU labeling) and increase in epidermal thickness were reduced in the knockout mice (57). Chun et al. showed that protein expression of EP2 and EP4, in addition to EP1, were induced by UV exposure, but that induction of EP2 and EP4 was reduced in COX-2 knockout mice (58). Treatment with either an EP2 or EP4 agonist reduced the excessive UV induction of apoptosis seen in COX-2 knockout mice (58). In other experiments, it was demonstrated that PGE2via the EP2 receptor is responsible for UV-induced down-regulation of E-cadherin-mediated cell–cell contacts that may contribute to UV-induced tumorigenesis (59). Together these results suggest that the EP2 and/or EP4 receptors are important for UV-mediated inflammation, epidermal hyperplasia, reduced cell–cell contacts and for the antiapoptotic effects of COX-2/PGE2. However, others have suggested that EP1 can mediate UV-induced inflammation and tumorigenesis. Treatment of SKH-1 mice with an EP1 antagonist reduced UV-induced increases in skin thickness/edema, neutrophil infiltration and activation, and increase in number of p53-positive cells same as treatment with the COX-2 inhibitor, celecoxib (55). In UV tumorigenesis experiments, the EP1 antagonist was able to reduce the number of tumors/mouse, although not tumor incidence (55). EP2 knockout mice on an SKH-1 background also showed reduced tumor multiplicity in UV carcinogenesis protocol; however, more large tumors and less well-differentiated, more aggressive tumors developed on the EP2 knockout mice compared to wild-type mice (57).

Thus it is still not certain which of the EP receptors mediate the downstream effects of UV induction of COX-2 and PGE2 in the skin. As EP3 is down-regulated after UV exposure and in UV-induced papillomas and SCCs, it probably plays little role in mediating the effects of UV. On the other hand, it is likely that the combination of EP1, EP2 and/or EP4 signaling together is important for activating pathways that are responsible for UV induction of inflammation, proliferation and tumorigenesis.

Future Directions

Understanding the molecular and signaling mechanisms by which COX-2/PGE2 mediate UV promotion of skin cancer offers novel approaches for chemoprevention. In order to be useful for chemoprevention, agents should be effective, easily administered, have no or very low side effects and be low-cost (60). As traditional NSAIDs inhibit both COX-1 and COX-2, and therefore inhibit the homeostatic functions of COX-1-derived PGs, COX-2-selective inhibitors were developed to prevent the harmful side effects (e.g. gastrointestinal complications) from chronic use of NSAIDs such as aspirin (60). However, long-term use of several COX-2-selective inhibitors (e.g. rofecoxib or Vioxx® and valdecoxib) has recently been shown to be associated with an increased risk of cardiovascular events, possibly by reducing endothelial prostacyclin production (60). Thus, the development of specific inhibitors/antagonists of PGE2/EP signaling may offer effective chemopreventive activity with minimal side effects. A more complete understanding of the EP signaling pathways mediating UV-induced PGE2 effects is still needed before specific agents can be proposed for human trials.


Figure 2 summarizes the overall consequences of UV induction of COX-2 expression that contribute to skin cancer development. The induction of COX-2 expression by acute UV exposure and constitutive up-regulation of COX-2 in UV-induced benign and malignant tumors leads to increased PGE2 production and activation of EP receptor signaling resulting in increased epidermal proliferation, induction of inflammation, angiogenesis and vascular permeability. In addition, COX-2/PGE2 have strong antiapoptotic activity and are able to counteract the induction of apoptosis by UV-mediated DNA damage and p53 induction. Also, by virtue of the oxidative nature of COX enzyme activity, COX-2 can lead to the production of ROS and contribute to a pro-oxidant state in combination with the release of ROS from activated inflammatory cells (13). All these consequences of COX-2 up-regulation contribute to and, along with UV-induced p53 mutations, are likely the driving forces behind the carcinogenic process initiated and promoted by chronic UV exposures.

Figure 2.

 Consequences of UV induction of cyclo-oxygenase-2 (COX-2) expression. Induction of COX-2 expression by UV irradiation results in increased production of prostaglandin E2 (PGE2). PGE2 binds to and activates four G protein-coupled receptors, of which EP1, EP2 and/or EP4 appear to be important for the effect of PGE2 in UV-induced skin damage and carcinogenesis. Signaling through the EP receptors results in the induction of various genes, including oncogenes, cytokines and growth factors, which then lead to inflammation with infiltration and activation of inflammatory cells in the dermis, induction of angiogenesis, vasodilation and vascular permeability leading to edema, increased keratinocyte proliferation resulting in epidermal hyperplasia, and a reduction in apoptosis otherwise mediated by UV induction of oxidative stress, DNA damage and p53. Chronic UV exposure leads to p53 mutations and other DNA damage, chronic inflammation and chronic up-regulation of COX-2 expression and its downstream effects, which then can result in skin carcinogenesis.


Acknowledgement— This work was supported by NIH grants CA100140, CA105345, CA16672 and ES07784.