Functional links between inflammation, tissue injury and cancer have been noted for a long time. In 1863, Rudolf Virchow suggested that cancer originates at sites of chronic inflammation. In fact, patients with persistent hepatitis B, Helicobacter pylori infections, inflammatory bowel diseases, or chronic pancreatitis were recently observed to carry increased risks for liver, gastrointestinal, or pancreatic cancers, respectively. Chronic inflammation is invariably associated with accumulation of inflammatory cells, growth factors, cytokines, oxidants and proinflammatory lipid mediators, which may cooperate in the development and maintenance of an activated stroma and the proliferation of epithelial cells. Similarly, wound healing is associated with inflammation and cell proliferation during tissue regeneration. Proliferating cells can acquire genetic alterations during excessive inflammation and thus may continue to grow under those conditions and give rise to autonomous tumors after the injurious and inflammatory tissue reaction has been terminated.1 Clearly, hallmarks of inflammation, tissue injury and repair and cancer are increased rates of arachidonic acid and linoleic acid oxidation through both cyclooxygenase (COX) and lipoxygenase (LOX) pathways thereby providing powerful lipid mediators, i.e., eicosanoids.
Substantial evidence supports a functional role for cyclooxygenase- and lipoxygenase-catalyzed arachidonic and linoleic acid metabolism in cancer development. Genetic intervention studies firmly established cause-effect relations for cyclooxygenase-2, but cyclooxygenase-1 may also be involved. In addition, pharmacologic cyclooxygenase inhibition was found to suppress carcinogenesis in both experimental mouse models and several cancers in humans. Arachidonic acid-derived eicosanoid or linoleic acid-derived hydro[peroxy]fatty acid signaling are likely to be involved impacting fundamental biologic phenomina as diverse as cell growth, cell survival, angiogenesis, cell invasion, metastatic potential and immunomodulation. However, long chain unsaturated fatty acid oxidation reactions indicate antipodal functions of distinct lipoxygenase isoforms in carcinogenesis, i.e., the 5- and platelet-type 12-lipoxygenase exhibit procarcinogenic activities, while 15-lipoxygenase-1 and 15-lipoxygenase-2 may suppress carcinogenesis. © 2006 Wiley-Liss, Inc.
Cyclooxygenase- and lipoxygenase-catalyzed arachidonic acid metabolism
Two major metabolic routes, i.e., the COX and LOX pathways, control the biosynthesis of eicosanoids. COX-derived eicosanoids comprise prostaglandins (PG) and thromboxane A2 (TXA2), while hydroperoxyeicosatetraenic acids (HPETE] and leukotrienes (LT) are products of LOX-catalyzed arachidonic acid metabolism (Fig. 1). Two isoforms of COX, i.e., COX-1 and COX-2, have been described: COX-1 is constitutively expressed in most tissues and appears to generate PG that control normal physiological functions such as platelet aggregation, regulation of renal blood flow and maintenance of the gastric mucosa. In contrast, COX-2 is transiently induced by proinflammatory stimuli, growth factors, cytokines and tumor promoters resulting in increased rates of PG formation during tissue injury and repair. The biological significance of a recently described splice variant of COX-1, referred to as “COX-3,” was recently put into question by the observation that mouse, rat and human COX-3 genes are unable to code for functional protein.2 The production of PG is a tightly regulated process that requires free fatty acids as substrates in most cases, the expression and activity of the individual COX isozymes, and the cell type-selective expression and activity of different PG synthases.2, 3 The specificity of action of the different PG is likely to predominantly depend on the cell type-specific expression of the corresponding G protein-coupled seventransmembrane PG receptors.4 COX isozymes, PG synthases and PG receptors are major players of the COX pathway, which control both synthesis and function of PG.
LOX constitute a family of nonheme iron dioxygenases that regio- and stereospecifically insert molecular oxygen into free and/or esterified polyunsaturated fatty acids generating 5-, 8-, 12-, 12R-, or 15-HPETE and, upon reduction, the corresponding HETE with arachidonic acid and 9- or 13-hydroperoxyoctadecadienoic acids (HPODE) and the corresponding hydroxy derivatives (HODE) with linoleic acid as substrates (Fig. 1).5 In addition to 5-LOX, leukocyte-type (l) and platelet-type (p) 12-LOX, and reticulocyte-type 15-LOX-1, the mammalian LOX family has recently expanded to comprise several isozymes that have been found to be preferentially expressed in the epidermis of man and mice. These include the epidermal (e)12-LOX and 12R-LOX, the only mammalian LOX isozyme generating products with R chirality, the mouse 8-LOX and its human ortholog 15-LOX-2, and eLOX-3, a LOX isoform exhibiting hydroperoxide isomerase activity.6, 7 LOX products represent either intermediary products such as HPETE, which are transformed enzymatically into secondary products including LT, hepoxilins, lipoxins and HETE, which may act as discrete signaling molecules5 or give rise to the production of reactive oxygen species (ROS) and oxidative degradation products of HETE and HODE such as 4-hydroxynonenal. The latter electrophilic alkenal derivates are known to covalently modify DNA, generating promutagenic adducts, or proteins, resulting in enzymatic inactivation or otherwise posttranslational modification of protein functions. These events were well demonstrated for thioredoxin reductase, an essential intracellular regulator of the redox potential.8 Signaling of LOX-derived products can occur through G protein-coupled cell surface receptors as for LT and lipoxins9 or through binding and activation of nuclear receptors such as peroxisome proliferator activated receptors (PPAR) as reported for HETE and HODE.10
Evidence for the involvement of COX and LOX in carcinogenesis
The development of tumors in humans and experimental animals is consistently linked to aberrant arachidonic acid metabolism through the COX and LOX pathways.11, 12, 13 With respect to COX, epithelial and nonepithelial tumor development is associated with overexpression of COX-2 in both premalignant and malignant stages, indicating that activation of COX-2 may be an early event during carcinogenesis. In addition, aberrant COX-2 overexpression was shown to be associated with poor prognosis for several human cancer types, e.g., breast, esophageal, gastric, lung, prostate and pancreatic cancers.12, 14, 15, 16, 17, 18, 19 Although constitutively expressed in normal and neoplastic epithelia, as a rule, an upregulation of COX-1 expression has been observed in cervical and ovarian cancers and in cholangiocarcinomas.20, 21, 22
Genetic and pharmacologic studies point to a causal role of COX in cancer development. Genetic ablation of COX-2 strongly reduces tumor formation induced by the two-stage carcinogenesis protocol in mouse skin,23 the development of intestinal polyps in the APCMin mouse model and colonic polyps in APCΔ716 mice.24, 25 On the other hand, keratin 5 (k5) promoter-driven COX-2 overexpression in basal keratinocytes of transgenic mice induces epidermal hyperplasia and dysplasia and sensitizes skin for carcinogenesis in that tumors develop with an initiating dose of the carcinogen DMBA alone.26, 27 Similarly, k5 promoter-driven COX-2 overexpression in myoepithelial and a few ductal cells produces cystic duct dilations and proliferative epithelial lesions in mammary glands.28 When aberrantly expressed in basal epithelial cells of urinary bladder, COX-2 has been shown to induce transitional cell hyperplasia, dysplasia and transitional cell carcinomas.29 In another model of mammary carcinogenesis, overexpression of COX-2 in mammary glands under the control of the murine mammary tumor virus promoter was sufficient to generate mammary gland hyperplasia, dysplasia and the formation of metastatic cancer in multiparous mice.30 These studies provided strong evidence for cause-and-effect relationships between COX-2 and the induction of premalignant alterations in diverse epithelia which occasionally develop into overt malignancy. These relationships are not limited to COX-2 but also apply to COX-1. Genetic ablation of this isozyme shows a comparable efficiency in suppressing colorectal25 and skin tumors23 and forced expression of COX-1 in endothelial cells provoked malignant transformation of these cells.31
The genetic data were corroborated by pharmacologic intervention studies using both nonselective COX inhibitors and COX-2-selective inhibitors. In particular, the latter have been shown to inhibit tumor formation in rodent models, in varying models of colon and skin carcinogenesis, and human cancer cells grafted on nude mice.12 Studies with initiation-promotion models of colon and skin cancer provide evidence that reduced tumor incidence and multiplicity emerge when the COX-2-selective inhibitors celecoxib or rofecoxib were applied after initiation of tumor development during the clonal selection and expansion of preexisting neoplastic cells, i.e., in the tumor promotion process.27, 32 Most importantly, clinical studies revealed a statistically significant inhibition of polyp growth in patients afflicted with familial adenomatous polyposis coli (FAP) at a dose of 400 mg celecoxib bid but not at the twice-daily antiinflammatory dose of 100 mg.33 This result is in line with previous clinical studies with the nonselective COX inhibitors sulindac sulfide and aspirin. The clinical trials underscore numerous epidemiologic studies suggesting an antineoplastic effect of long-time regular intake of these drugs.12 Although nonselective COX inhibitors are effective, their long-term use is limited by gastrointestinal side-effects such as dyspepsia and abdominal pain and, occasionally, gastric or duodenal perforation or bleeding. These side effects were primarily attributed to the inhibition of COX-1. The gastrointestinal intolerance provided an incentive for the development of COX-2-selective inhibitors, which spare the adverse gastrointestinal effects. The COX-2-selective inhibitor trials, however, also revealed a major side effect of long-term treatment in selected patient subgroups. A significantly increased risk of acute myocardial infarction has been observed in patients carrying other risk factors to develop cardiovascular disease. This led to termination of ongoing prevention trials. The adverse reaction is thought to be due to a potential inhibition of endothelial cell-derived COX-2 activity and prostacyclin (PGI2) generation that may shift the delicate PGI2/TXA2 balance towards more TXA2 thereby promoting platelet aggregation and initiating thrombotic effects.34
The role of LOX in carcinogenesis is thought to be more complex because 6 LOX genes have been identified in humans (and 7 in mice) and different profiles of LOX were found in studies on human tumor biopsies and experimentally induced animal tumor models.13, 35, 36 By comparing LOX expression and activity data of various normal and cancerous human and mouse epithelia it became evident that the human 15-LOX-1 or its mouse ortholog l12-LOX and the human 15-LOX-2 or its mouse ortholog 8-LOX are preferentially expressed in normal tissues and benign lesions, but not in carcinomas of bladder, breast, colon, lung, prostate and skin,37, 38, 39, 40, 41 whereas 5-LOX and p12-LOX, as a rule, are absent in normal epithelia, induced by proinflammatory stimuli, and constitutively expressed in various epithelial cancers including colon, esophageal, lung, pancreatic, prostate and skin cancer.42, 43, 44, 45, 46, 47 The data suggests that distinct LOX, the expression of which is lost during progression of cancer, may exhibit antitumorigenic activities, while other isoforms that are preferentially expressed in carcinomas may exert protumorigenic effects (Fig. 2). This view is supported by experimental studies showing that the forced expression of 8-LOX or the induced expression of l12S-LOX reduced DMBA/TPA-induced skin tumor formation in transgenic mice.48, 49 In this context, it is intriguing that high doses of COX inhibitors were shown to induce the expression of 15-LOX-1 in a COX-independent manner in vitro and that this effect may contribute to the antitumor activities of COX inhibitors.50 In the prostate, however, the human ortholog 15-LOX-1 was found to exhibit a protumorigenic activity as its expression was increased in prostate cancer and 15-LOX-1 overexpressing prostate cells showed increased tumor progression when injected in nude mice.51 In the pancreas, 5-LOX expression was already increased in intraepithelial neoplasias, early noninvasive precursor lesions for pancreatic adenocarcinomas, indicating that the 5-LOX pathway may be a target for cancer prevention.44 In addition, the expression level of p12-LOX has been shown to correlate with advanced stage and poor differentiation of human prostate cancer and has together with decreased expression levels of 15-LOX prognostic value in patients with breast cancer.42 On the other hand, genetic ablation of the procarcinogenic p12-LOX led to a significant reduction of skin carcinoma formation.52
Pharmacologic intervention studies with LOX inhibitors aiming at the inhibition or prevention of carcinogenesis are difficult to interpret as the repertoire of LOX isoform-selective inhibitors is still limited.35, 36 Recently, 5-LOX-selective inhibitors were shown to inhibit esophageal carcinogenesis in rats,53 oral squamous carcinomas in hamsters54 and chemically induced lung cancer in mice55 and both 5- and 12-LOX inhibitors suppressed growth of human pancreatic xenografts in mice.56 5-LOX inhibitors, in addition to COX-2 inhibitors, exhibit growth suppressive effects on gliomas and meningiomas in animal models57 and simultaneous inhibition of 5-LOX and COX-2 has been found to suppress cigarette smoke-induced colon cancer.58
Eicosanoid signaling and carcinogenesis
Increased COX-2 mRNA and protein that are observed in premalignant and malignant epithelial and nonepithelial tumors are consistently associated with the accumulation of COX-derived PG. In fact, PGE2 has been found to be the major PG in human colorectal cancer,59 and increased contents of PGE2 and PGF2α were observed in endometrial adenocarcinoma,60 and accumulation of PGE2, PGF2α and PGI2 is a consistent feature of skin tumors.27, 61 In addition, PGE2 was shown to prevent COX inhibitor-induced adenoma regression in APCMin mice and PGF2α reversed the indomethacin-induced inhibition of tumor development in mouse skin.62, 63 This data demonstrates that PG are critically involved in colorectal and skin carcinogenesis in these models. However, that COX-independent effects may also contribute to reduction of tumor formation is indicated by numerous studies that identified COX-2-independent antitumor targets, in particular, at high inhibitor doses.64, 65 The cell types that contribute to aberrant COX-2 expression and PGE2 synthesis are less clear and may depend on the tissue type. In human colorectal cancer, increased COX-2 expression is present in the epithelial tumor parenchyma and in mononuclear cells, vascular endothelium, smooth muscle cells and fibroblasts. Moreover, COX-2 expression levels in the individual cell types may vary with tumor progression.66 In skin, COX-2 expression is restricted to basal keratinocytes and dendritic cells of the papilloma parenchyma and to endothelial cells and macrophages of the tumor stroma. In squamous cell carcinomas, a strong COX-2 expression is observed in keratinocytes and infiltrated inflammatory cells.61
Aberrant expression and activity of the protumorigenic 5- and/or p12-LOX preferentially occur in epithelial cells in human brain, breast, colon, esophagus, lung, oral, pancreatic, prostate and skin cancer and in stromal cells adjacent to the tumor parenchyma.42, 43, 44, 45, 53, 54, 57 In human astrocytomas, 5-LOX was found to be strongly expressed in tumor cells and infiltrating macrophages/microglials cells.57 This indicates that both tumor and stromal cells may contribute to the production of COX- and LOX-derived eicosanoids that participate in tumor development.
Eicosanoids may exert their biological effects in an intracrine manner as activating ligands of transcription factors of the PPAR family (though this is controversially discussed) or through interaction with specific seventransmembrane G protein-coupled cell surface receptors in an autocrine or paracrine manner.4, 5, 9 Besides PGF2α and PGI2, PGE2 and its cognate receptors play a predominant role in the promotion and progression stages of carcinogenesis. PGE2 conveys tumor growth by stimulating prostaglandin E-type (EP) receptors as confirmed by genetic and pharmacologic experiments.67 Thus, deletion of the EP2 receptor reduces papilloma development, whereas ablation of EP3 selectively inhibits carcinoma formation in DMBA/TPA-treated mouse skin.68, 69 In the colon, disruption of EP1 and EP4 decreases the number of chemically induced aberrant crypt foci (ACF), whereas deletion of EP2 decreases the number and size of intestinal polyps in APCΔ716 knockout mice. Accordingly, EP1 and EP4 antagonists were found to reduce the incidence of ACF and the number and size of intestinal polyps in APCMin mice.67 The elevated expression of the prostaglandin F2α (FP) receptor is a consistent property of endometrial adenocarcinomas and PGF2α promotes the growth of neoplastic endometrial epithelial cells via activation of the FP receptor.60
LTB4, a downstream metabolite of the 5-LOX pathway, supports pancreatic tumor growth via activation of its receptors BLT1 and BLT2 that are both upregulated in pancreatic cancer as compared to normal tissue. The receptor antagonist LY293111 suppressed the growth of subcutaneously or orthotopically transplanted human pancreatic cancer cells in athymic mice.70 Other LOX products are thought to induce gene expression by transactivation of transcription factors of the PPAR family. 8-HETE-effected PPARα activation induces the expression of keratin 1 and terminal differentiation in primary keratinocytes suggesting that 8-HETE-activated PPARα may also be involved in the inhibition of papilloma development in the skin of 8-LOX transgenic mice.48 In fact, a PPARα agonist reduces papilloma development in DMBA/TPA-treated mouse skin.71 On the other hand, long-term activation of PPARα by PPARα agonists has been shown to increase the incidence of hepatocellular carcinomas in mice but not in humans, whereas PPARα-null mice are resistant to this effect. Thus, PPARα may mediate both pro- and antitumorigenic activities depending on the cell- or tissue-specific context.10 Potential eicosanoid-type agonists of PPARγ include 15-HETE and 13-HODE. Both 15-LOX metabolites bind and activate PPARγ in vitro and 15-HETE was shown to be an endogenous PPARγ ligand by using a transactivation assay in various epithelial tissues expressing 15-LOX-2.72 Opposing effects of 13-HODE and 15-HETE on PPARγ transactivation have been observed in growth factor-challenged prostate cancer cells. 13-HODE-induced upregulation of EGF-initiated MAP kinase activity, MAP kinase-dependent phosphorylation and subsequent inactivation of PPARγ may be associated with growth stimulation, while 15-HETE-effected down-regulation of MAP kinase activity and activation of PPARγ and PPARγ-dependent gene transcription appear to be linked to growth inhibition. The opposing effects may reflect the contrariwise functions of the anticarcinogenic 15-LOX-2 and 15-HETE in normal and benign prostate and the procarcinogenic 15-LOX-1 and 13-HODE in prostate cancer.51 In line with this data is the observation that activation of PPARγ was associated with anticarcinogenic effects in murine breast, colon and prostate cancer models and the occurrence of PPARγ inactivating mutations in human colon tumors and thyroid follicular carcinomas.10
Downstream targets of eicosanoid signaling in carcinogenesis
Numerous in vitro and in vivo studies indicate that PGE2, PGF2α and LTB4-induced signaling via EP, FP and LTB4 receptors, respectively, and HETE/HODE-initiated signaling cascades are involved in the stimulation of proliferation, inhibition of terminal differentiation or apoptosis of tumor cells, promotion of angiogenesis, stimulation of invasion and motility and immune modulation (Fig. 3).36, 73, 74
Modulation of tumor cell growth
Crosstalk of PGE2 and PGF2α with different growth factor receptor-induced signaling cascades is involved in the stimulation of tumor cell growth. In colorectal cancer cells, PGE2 is known to activate EGF receptor (EGFR) signaling through generation of activating EGFR ligands either by matrix metalloproteinase (MMP)-induced shedding of the ligands from the cell surface or de novo synthesis of EGFR ligands such as amphiregulin. PGE2 also has been found to transactivate EGFR via an intracellular Src-mediated process. In fact, signaling through EGFR is critically involved in various types of cancer such as breast, lung and colorectal cancer.73, 74 Interestingly, aberrant COX-2 expression in breast cancer has been found to be associated with nuclear expression of ErbB-2, a tyrosine kinase receptor that belongs to the EGFR family and acts as a transcriptional activator of several genes including that coding for COX-2.75 In addition, PGE2 was found to stimulate colorectal cancer cell proliferation and COX-2 expression through activation of a Ras-MAPK cascade generating a PGE2-effected feed-forward loop supporting cancer growth.76 Recently, it was shown that in the absence of functional APC, PGE2 is able to stimulate proliferation of colon cancer cells by activating components of the WNT signaling pathway. PGE2-induced activation of the EP2 receptor fosters the association of the activated αs unit of Gs with the regulators of G protein signaling (RGS) domain of axin and subsequently the release of glycogen synthase 3β (GSK-3β) from its complex with axin. The βγ units of the activated Gαs directly stimulate the activation of phosphoinositide-3-kinase (PI-3K) and Akt causing the phosphorylation and inactivation of GSK-3β. The result of the crosstalk between the 2 signaling cascades is the stabilization of β-catenin, its nuclear translocation, and the LEF- and β-catenin-dependent gene expression supporting aberrant growth of colon cancer cells.77 PGE2 also can indirectly stimulate tumor cell proliferation, e.g., via increased expression and activity of aromatase and subsequent enhanced oestrogen synthesis that enhances proliferation of mammary gland epithelial cells.30, 78 PGF2α was found to stimulate growth of endometrial adenocarcinoma explants and cells via FP receptor activation along a pathway involving activation of phospholipase Cβ, mobilization of inositol-1,4,5-trisphosphate, transactivation of the EGF receptor and MAP kinase signaling. In addition, PGF2α-FP receptor activation induced COX-2 expression and PGF2α synthesis provoking a feed-forward loop, which may contribute to endometrial carcinogenesis.73, 79
Two signaling cascades have presently been associated with the stimulation of tumor cell growth by LOX products.35, 36 5-HETE stimulates tumor cell proliferation in various cancer cell lines including lung, pancreatic and prostate cancer cells.80, 81, 82 In pancreatic cancer cells, 5-HETE-induced proliferation was found to be associated with activation of the MEK/ERK1/2 and the PI-3K/Akt cascade and yet undefined tyrosine kinases.82 LTB4-induced and BLT1/2-mediated stimulation of pancreatic cancer cell proliferation is brought about by the simultaneous stimulation of the MEK/ERK1/2 and PI-3K/Akt pathways as indicated by pharmacologic interventions studies.83 ERK1/2 and undefined tyrosine kinases were also found to be involved in 12-HETE-stimulated proliferation of pancreatic cancer cells.82
Tumor cell survival
Tumor growth does not only depend on increased cell proliferation but also on prolonged cell survival by inhibition of terminal differentiation or apoptosis. Thus, the k5 promoter-driven overexpression and activity in the skin of k5 COX-2 transgenic mice induces a delayed onset of epidermal differentiation, while the genetic ablation of COX-2 accelerates keratinocyte differentiation in the skin of COX-2−/−mice.26, 84 PGE2 was shown to inhibit apoptosis induced by both nonselective COX and COX-2-selective inhibitors in colon cancer cells suggesting that COX-2-derived PGE2 induces resistance to programmed cell death. The upregulation of Bcl-2 expression or stimulation of the transcriptional activity of NFκB may be involved in PGE2-induced inhibition of apoptosis.73, 74 Moreover, PGE2 was shown to transactivate PPARδ via activation of the PI-3K/Akt pathway which increases intestinal adenoma formation and cell survival in APCMin mice.85 Another component involved in PGE2-effected suppression of apoptosis is the recently identified nuclear receptor NR4A2, the expression of which is induced by PGE2 in colorectal cancer through a cAMP/protein kinase A-dependent pathway. NR4A2 has been shown to elicit antiapoptotic activity by blocking cleavage of caspase-3.86 Chemotherapeutic agents and radiation were found to induce COX-2 expression and PGE2 synthesis in human cancer cells. The increased levels of PGE2 may counteract the treatment-induced apoptotic cell death thereby increasing resistance to therapy. This suggests that an adjuvant treatment with COX inhibitors may increase the therapeutic efficiency of distinct chemotherapeutic agents and radiation.87
5- or 12-LOX inhibitors were reported to induce apoptosis in human breast, gastric, hepatocellular, pancreatic and prostate cancer cells.35, 36 In human breast cancer cells, both 5-LOX and 12-LOX inhibitors induce apoptosis through cytochrome c release and caspase 9 activation whereas 5-HETE and 12-HETE stimulate proliferation of this cells.88 In pancreatic cancer cells, the triggering of apoptosis is due to a disturbance of the balance between pro- and antiapoptotic proteins promoting a proapoptotic signaling pathway, which includes cytochrome c release from mitochondria and the activation of the caspase cascade. In contrast, 5-HETE, 12-HETE and LTB4 were shown to prevent apoptosis through activation of the MEK/ERK1/2 cascade. In addition, 5- and 12-HETE also activate the PI-3K/Akt pathway, which is thought to contribute to the antiapoptotic effects of the LOX metabolites.89 Intriguingly, tumor cells exhibiting a constitutively active 5-LOX show an impaired apoptotic response to genotoxic anticancer agents and ionizing radiation. Moreover, genotoxic stress by DNA damaging agents was shown to increase 5-LOX expression and activity which selectively suppress p53-induced apoptotic pathways.90
Modulation of angiogenesis
Induction of neovascularization and angiogenesis are prerequisites for tumors to grow beyond 2–3 mm3. To this end, cancer cells, endothelial cells, stroma cells and others produce proangiogenic factors that induce endothelial cell recruitment, proliferation, migration and tube formation. Genetic and pharmacologic studies show that vascular density and the vascularization of the mammary gland are attenuated in COX-2−/− mice24, 91 and that COX inhibitors significantly suppress the generation of proangiogenic factors and inhibit endothelial cell proliferation, migration and assembly.92 Inversely, a strong angiogenic activity is induced in transgenic mice with tissue-specific overexpression of COX-2 in the urinary bladder and mammary gland. This is associated with the production of angiogenic factors such as vascular endothelial growth factor (VEGF) and/or basic fibroblast growth factor (bFGF).29, 93 The proangiogenic effects of aberrant COX-2 expression are mediated by 4 prostanoids, TXA2, PGI2, PGF2α and, in particular, PGE2. The latter was shown to reverse the antiangiogenic effects of COX inhibitors.92 Moreover, homozygous ablation of the EP2 receptor abrogated the induction of VEGF in APCΔ716 mouse polyps.67 PGE2 does not only induce the generation of VEGF and bFGF that are important for growth and survival of endothelial cells but also stimulates αVβ3-Cdc42/Rac-dependent endothelial cell migration and spreading, and mediates VEGF- and bFGF-induced CXCR4-dependent neovessel assembly in vivo.94, 95 In endothelial cells, PGE2-induced expression of VEGF stimulates the expression of COX-2 and PGE2 production in these cells, thereby amplifying the stimulatory effect of PGE2 on the generation of VEGF and bFGF establishing a positive feedback loop.95 It was recently shown that PGF2α contributed to angiogenesis in endometrial adenocarcinomas through FP receptor activation, transactivation of the EGF receptor and induction of VEGF mRNA and protein expression.96
Proangiogenic effects of the LOX metabolism are attributed to 12-HETE that was shown to directly stimulate endothelial cell proliferation, migration and tube formation in vitro. The proangiogenic activity was supported by in vivo data demonstrating that a 12-LOX-selective inhibitor suppressed bFGF-induced angiogenesis.97 5-HETE was recently shown to contribute to angiogenesis through activation of MMP2 and VEGF generation.98 Intriguingly, angiogenesis and tumor formation in 2 xenograft models were found to be inhibited in transgenic mice overexpressing 15-LOX-1 in endothelial cells. These effects may contribute to the known antitumorigenic activity of 15-LOX-1.99
Promotion of invasiveness
Progression of cancer involves loss of cell–cell adhesion within the primary tumor, development of an invasive phenotype and subsequent migration and tumor cell invasion into the matrix and adjacent tissues. Aberrant COX-2 expression has indeed been shown to be correlated with invasive and metastatic growth of breast, colorectal, gastric and lung cancer cells. COX-2 expression induces the expression of MMP1 and MMP2 and metatstatic potential in human colon cancer cells and is associated with increased expression of the hyaluronate receptor CD44 that was found to mediate nonsmall-cell lung cancer invasion.100, 101 PGE2 induces the expression of glycosyltransferases and type I sialyl Lewis antigens favoring the adhesion of colorectal cancer cells to endothelial cells and cell migration through EP4-mediated transactivation of the EGF receptor and PI-3K/Akt signalling.102, 103 The COX-2-selective inhibitor celecoxib inhibited these effects and rofecoxib inhibited the growth and metastatic potential of colorectal carcinomas in mice.104
The ability of tumor cells to generate 12-HETE is positively correlated to their metastatic potential and the increased expression of p12-LOX enhances the metastatic potential of human prostate cancer cells. Moreover, 12-HETE has been found to modulate multiple steps of the metastatic process encompassing tumor cell endothelial cell interactions, tumor cell motility, proteolysis and invasion.105
Modulation of immunosuppression
The important immunomodulatory function of PG, in particular PGE2, is to skew the immune system towards T helper 2 (Th2), which may allow cancer cells to escape the immune defense.73, 106 This shift is achieved by the interaction of PGE2 with various types of immune cells and interference with multiple immune functions. PGE2 was shown to inhibit dendritic cell differentiation, to inhibit proliferation and stimulation of T cells and to restrain the antitumor activity of natural killer cells and macrophages. In addition, PGE2 also impairs the ability of T lymphocytes to generate antitumor TH1 cytokines (TNF-α, IFN-γ, IL-2 and IL-12) and upregulates TH2 cytokines including IL-4 in keratinocytes, IL-10 in peripheral blood lymphocytes and macrophages and IL-6 in T lymphocytes. The overall result is an enhancement of Th2 responses and inhibition of Th1 responses by PGE2.106
Both COX and LOX enzymes are peroxidases that were found to be able to oxidize procarcinogens to carcinogens. Such oxidative reactions are primarily catalyzed by cytochrome P450. However, in tissues with low expression levels of this enzyme family, such as pancreas and skin, significant amounts of xenobiotics may be cooxidized to mutagens by the peroxidase activities of COX.107 Additional mutagens such as malondialdehyde can be generated as by-products of prostaglandin biosynthesis. Lipid peroxy radicals produced by COX or LOX-mediated lipid peroxidation appear to be the source of premutagenic exocyclic DNA adducts.108 In addition to catalyzing the synthesis of mutagens, carcinogens have been shown to induce the expression of COX-2 and 5- and 12-LOX, which in turn cooxidize procarcinogens to carcinogens along a positive feedback loop.89
Elucidation of molecular mechanisms of COX-2, but also of COX-1, and the various LOX and their downstream targets in cancer development has considerably advanced our understanding of cancer biology. Indeed, it may provide a basis for the development of mechanism-based intervention to prevent or suppress carcinogenesis. Numerous preclinical studies established that nonselective and COX-2-selective inhibitors decrease the risk of epithelial and nonepithelial cancer through inhibition of COX-mediated PGE2 biosynthesis. The clinical studies, however, revealed an issue of the long-term safety of selective COX-2 inhibitors. Therefore, the elucidation of downstream pathways of PGE2 biosynthesis and signaling may offer additional targets for cancer chemoprevention and therapy. Although such detailed mechanistic knowledge has not yet been accumulated for the role of LOX, one important aspect is the characterization of distinct LOX isoforms that exhibit either pro- or anticarcinogenic properties and identify the procarcinogenic isoforms as novel targets for preventive or therapeutic measures.