HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Mollace, V. Articles by Salvemini, D. Search for Related Content PubMed PubMed Citation Articles by Mollace, V. Articles by Salvemini, D.

Article

Modulation of Prostaglandin Biosynthesis by Nitric Oxide and Nitric Oxide Donors

Faculty of Pharmacy, University of Catanzaro "Magna Graecia", Roccelletta di Borgia, Catanzaro, Italy (V.M., C.M.); Department of Preclinical and Clinical Pharmacology, University of Florence, Florence, Italy (E.M.); Institute of Pharmacology, University of Messina, Messina, Italy (S.C.); and Department of Pharmacological and Physiological Science, School of Medicine, St. Louis University, St. Louis, Missouri (D.S.)

Abstract

I. Introduction

II. The Prostaglandin Biosynthetic Pathway

A. The Contribution of Prostaglandin Biosynthesis in Disease States

1. Inflammation and Pain.

2. Cancer.

3. Neuronal Excitotoxicity.

4. Neuroinflammatory Processes.

5. Alzheimer's Disease.

6. Brain Ischemia.

B. Selective Nonsteroidal Anti-Inflammatory Drugs

III. Nitric Oxide and Nitric Oxide Donors

A. Biosynthesis of Nitric Oxide

1. Endothelial Nitric-Oxide Synthase.

2. Neuronal Nitric-Oxide Synthase.

3. Inducible Nitric-Oxide Synthase.

B. Contribution of Nitric Oxide Biosynthesis and Release in Disease States

C. Nitric-Oxide Synthase Inhibitors

D. Nitric Oxide Donors

IV. Interaction between Nitric Oxide and Prostaglandin Biosynthesis

A. Nitric Oxide/Cyclooxygenase Reciprocal Interactions

B. Molecular Basis of Nitric Oxide/Cyclooxygenase Reciprocal Modulation

C. Drugs Acting Simultaneously on Nitric Oxide and Cyclooxygenase

V. Perspectives and Concluding Remarks

The biosynthesis and release of nitric oxide (NO) and prostaglandins (PGs) share a number of similarities. Two major forms of nitric-oxide synthase (NOS) and cyclooxygenase (COX) enzymes have been identified to date. Under normal circumstances, the constitutive isoforms of these enzymes (constitutive NOS and COX-1) are found in virtually all organs. Their presence accounts for the regulation of several important physiological effects (e.g. antiplatelet activity, vasodilation, and cytoprotection). On the other hand, in inflammatory setting, the inducible isoforms of these enzymes (inducible NOS and COX-2) are detected in a variety of cells, resulting in the production of large amounts of proinflammatory and cytotoxic NO and PGs. The release of NO and PGs by the inducible isoforms of NOS and COX has been associated with the pathological roles of these mediators in disease states as evidenced by the use of selective inhibitors. An important link between the NOS and COX pathways was made in 1993 by Salvemini and coworkers when they demonstrated that the enhanced release of PGs, which follows inflammatory mechanisms, was nearly entirely driven by NO. Such studies raised the possibility that COX enzymes represent important endogenous "receptor" targets for modulating the multifaceted roles of NO. Since then, numerous papers have been published extending the observation across various cellular systems and animal models of disease. Furthermore, other studies have highlighted the importance of such interaction in physiology as well as in the mechanism of action of drugs such as organic nitrates. More importantly, mechanisticstudies of how NO switches on/off the PG/COX pathway have been undertaken and additional pathways through which NO modulates prostaglandin production unraveled. On the other hand, NO donors conjugated with COX inhibitors have recently found new interest in the understanding of NO/COX reciprocal interaction and potential clinical use. The purpose of this article is to cover the advances which have occurred over the years, and in particular, to summarize experimental data that outline how the discovery that NO modulates prostaglandin production has impacted and extended our understanding of these two systems in physiopathological events.

I. Introduction

Prostaglandins (PGs¹) and nitric oxide (NO) represent some of the most relevant local mediators that participate, under basal conditions, in the modulation of many cellular functions. This occurs as a consequence of constitutive biosynthesis and release of both mediators via well identified bioenzymatic complexes [cyclooxygenases (COX) for PGs and nitric-oxide synthases (NOS) for NO]. These highly regulated biosynthetic pathways act on specific substrates [arachidonic acid (AA) and L-arginine, respectively], leading to pulsed release of nanomolar concentrations of both mediators. The basal release of NO and PGs has been shown to exert a protective role in many physiopathological conditions, such as vascular diseases (via enhanced vasodilatation and antiplatelet activity), gastric lesions (via activation of gastroprotective processes), erectile dysfunction, and learning and memory processes (via potentiation of neuronal plasticity). Therefore, the use of PG derivatives and NO donors has been proposed in the treatment of such disorders due to their ability of restoring basal levels of PGs and NO (e.g., organic nitrate esters, which release NO after enzymatic bioconversion, have been used for more than a century in the treatment of myocardial ischemia).

Under inflammatory states or in early stages of many diseases characterized by the occurrence of inflammatory processes, NO and PGs are released simultaneously in large amounts; this effect is mainly due to the activation of inducible enzymes, which release NO and PGs in micromolar concentrations. The release of PGs and NO in large amounts and the evidence that both PGs and NO overproduction may be detrimental for cell survival, suggested experimental work over the last two decades to identify the possible participation of both mediators in the pathogenesis of many disease states. In particular, overt production of PGs and NO has been shown to occur in the damaged tissue accompanying the inflammatory processes involved in rheumatic diseases, chronic degenerative disorders, and, in the brain, neurodegenerative processes associated with brain ischemia as well as in neuroinflammatory diseases (such as multiple sclerosis, demyelinization, HIV-related brain disorders, Alzheimer's disease). In addition, the development of novel and more selective COX and NOS inhibitors suggested their potential use for the protection of inflamed tissues.

Recently, evidence has been accumulated indicating that there is a constant cross talk between NO and PG release that occurs at many levels, but having its central feature in the modulation of molecular mechanisms that regulate PG-generating pathway. The final effect of this modulatory activity is not univocal, since endogenous NO as well as NO donors have been found to switch on/off the COX pathway, depending on the basal levels of NO released, by the cell type in which PG biosynthesis is generated and by the intensity of the stimulus employed for PG release. In addition, NO conjugates with superoxide anions, leading to the formation of peroxynitrite, which modulates COX enzymes. Finally, recent evidence has been accumulated that NO donors conjugated with COX inhibitors (e.g., nitroaspirin and nitroflurbiprofen) may have a potential use in the treatment of inflammatory as well as noninflammatory disorders, suggesting that NO/COX reciprocal interaction are relevant in the pathophysiological mechanisms underlying many disease states. Taken together, these findings seem to give credence to an active reciprocal modulation of NO and PG-generating systems which represent the basis for a rationale use of traditional NO donors and for the development of novel and more suitable NO donors and COX inhibitors.

II. The Prostaglandin Biosynthetic Pathway

PGs and related compounds are some of the most prevalent autacoids detected in every tissue and body fluid except for the red blood cells. As local hormones, they produce in minute concentrations an incredibly broad spectrum of effects that modulate almost every biological function. They mostly derive from the 20-carbon fatty acid, arachidonic acid, by the action of the PG synthase in a two-step conversion (Fig. 1). First, arachidonic acid is converted into a cyclic and unstable endoperoxide (PGG₂) by PG endoperoxide synthase (PGHS) or COX, which is then followed by a peroxidase that cleaves the peroxide to yield the endoperoxide (PGH₂). These unstable intermediate products of arachidonic acid metabolism are then rapidly converted to the prostanoids (PGD₂, PGF₂ PGE₂, thromboxane A₂, PGI₂)by specific isomerase enzymes (Flower and Vane, 1972; Smith and Lands, 1972; Needleman et al., 1986; Smith et al., 1991) (Fig. 1). In 1976, COX was first purified from sheep seminal vesicles, a prodigious source of the protein (Hemler and Lands, 1976; Miyamoto et al., 1976; Van der Ouderaa et al., 1977), as a homodimer with a molecular mass of 71 kDa and subsequently cloned from the same tissue (DeWitt and Smith, 1988; Merlie et al., 1988) (Fig. 2). The enzyme exhibits both COX and hydroperoxidase activities. With the availability of the cDNA encoding the protein and specific antibodies, numerous studies were performed to evaluate the distribution, expression, and regulation of COX both in vitro and in vivo.

View larger version (25K): [in this window] [in a new window]   FIG. 1. The membrane phospholipids—arachidonic acid cascade.

View larger version (21K): [in this window] [in a new window]   FIG. 2. Representation of COX isoforms and pattern of molecular modulation of both enzymes.

Over the years, there were various suggestions that there was a second COX enzyme. As early as 1972, Smith and Lands (1972) and Flower and Vane (1972) speculated on the existence of isoenzymes. Later, Lin et al. (1989) hypothesized that a different COX enzyme could be induced by platelet-derived growth factor, and as often happens before a new discovery is properly defined, many pharmacologists and biochemists reported an inducible COX without knowing that they were working with a different enzyme. Needleman and his colleagues (Morrison et al., 1978) demonstrated a progressive increase in PG release dependent on de novo synthesis of COX enzyme in a inflammatory model of ureter-obstructed kidney.

In the following years, a number of studies have illustrated that COX activity is increased in certain inflammatory states and is induced in cells by proinflammatory cytokines and growth factors in vitro (Bailey et al., 1985; Sano et al., 1992). Needleman and his group continued their earlier work and reported that bacterial lipopolysaccharide (LPS) increased the synthesis of PGs in human monocytes in vitro and in mouse peritoneal macrophages in vivo. This increase, but not the basal level of enzyme, was inhibited by dexamethasone and associated with de novo synthesis of a new COX protein. They reinforced the concept of "multiple COX pools, constitutive and stimulated, possibly under different regulatory controls" (Seibert et al., 1991). The break-through discovery of a different COX came from molecular biologists outside the field of PGs. Simmons and his colleagues (Simmons et al., 1989; Xie et al., 1991) were studying early response genes and discovered an inducible form of COX in chicken embryo cells. It was encoded by a 4.1-kb mRNA similar in size to that reported by other authors (Rosen et al., 1989). They cloned the gene, deduced the protein structure, and found it homologous to COX, but to no other known protein (Fig. 2). Independently, other groups found similar results in different animal species (O'Banion et al., 1991; Sirois and Richards, 1992). Thus, there are two distinct enzymes, COX-1, the constitutive isoform, and COX-2 (the inducible one). Both enzymes have a molecular weight of 70 to 71 kDa and are almost identical in length, with just over 600 amino acids, of which 63% are in an identical sequence. However, the human COX-2 gene of 8.3 kb is a small immediate early gene, whereas human COX-1 originates from a much longer 22-kb gene. The gene products also differ, with the mRNA for the inducible enzyme being approximately 4.5 kb and that of the constitutive enzyme being 2.8 kb (Otto and Smith, 1995; Herschman, 1996). Garavito and his colleagues (Picot et al., 1994) have determined the three-dimensional structure of COX-1, which consists of three independent folding units: an epidermal growth factor-like domain, a membrane binding section, and an enzymatic domain. The three-dimensional X-ray crystal structure of human or murine COX-2 can be superimposed on that of COX-1; the residues that form the substrate binding channel, the catalytic sites, and the residues immediately adjacent are all identical except for two small variations. In these two positions, the same substitutions occur: isoleucine in COX-1 is exchanged for valine in COX-2 at position 434 and 523 (Fig. 2). Most nonsteroidal anti-inflammatory drugs (NSAID) compete with arachidonic acid for binding to the active site; uniquely, aspirin irreversibly inhibits COX-1 by acetylation of serine 530, thereby, excluding access to the endogenous substrate (Figs. 2 and 3) (Roth et al., 1975). COX-1 is ubiquitous and has clear physiological functions. Its activation leads, for instance, to the production of prostacyclin (PGI₂) which, when released into blood vessels, produces vasodilation and antithrombogenic activity and is cytoprotective when released by the gastric mucosa (Moncada et al., 1976; Whittle et al., 1978). Maintenance of kidney function both in animal models of diseases and in patients with congestive heart failure, liver cirrhosis, and renal insufficiency is dependent on vasodilator PGs (PGE₂ and PGI₂) mainly by COX-1, although low levels of mRNA for COX-2 have been reported (Harris et al., 1994), and up-regulation of COX-2 expression has been observed in the macula densa following salt deprivation (Harris et al., 1994). In platelets, the only isoform detectable is COX-1, and loss of arachidonic acid-induced platelet aggregation is not only a well established side effect of NSAID treatment, but also the therapeutic aim of the "half an aspirin a day" prophylaxis against thromboembolic disease. This prophylaxis is achieved through inhibition of COX-1, which leads to decreased production of thromboxane A₂.

View larger version (14K): [in this window] [in a new window]   FIG. 3. Most NSAIDs compete with arachidonic acid for binding to the active site. COX-2 isoform structure has been targeted to design a new class of a more selective compound.

An intense biosynthesis of AA metabolites has also been shown to occur in the brain in almost all periods of life, participating in many physiological changes and being strongly involved in many disease states. High levels of PGs, especially PGE₂, are detected in the blood and brain of the neonate (Mitchell et al., 1978; Jones et al., 1993; Li et al., 1995). In the retina, both increased COX-1 and COX-2 activities contribute to the augmented production of neonatal PGs (Hardy et al., 1998). In the brain, however, this mostly arises from increased expression and activity of the COX-2 pathway in brain vasculature, as opposed to adult brain, where prostanoid formation is catalyzed mainly by COX-1 (Peri et al., 1995). The rapid drop in PG levels in brain within the first 48 to 72 h after birth (Jones et al., 1993) is associated with a relative decrease in COX-2 expression, which seems to increase again thereafter (Parfenova et al., 2000).

COX-1 and COX-2 are expressed in discrete brain regions under basal conditions (for a review, see Yermakova and O'Banion, 2000; Hurley et al., 2002). No evidence has been collected yet on the possible localization of COX-3 in the human brain, unless the putative target for the antipyretic activity of the COX-3 inhibitor acetaminophen is represented by the thermoregulatory regions of the hypothalamus (for a review, see Warner and Mitchell, 2002). In particular, evidence has been accumulated indicating that the central nervous system (CNS) represents one of the few tissues that contain detectable concentrations of COX-2 mRNA expressed constitutively (Seibert et al., 1994), although this point still remains to be elucidated. Indeed, according to the variability in the experimental procedures used by different groups, the constitutive expression of COX-2 isoenzyme within CNS still represents a controversial point. Basically, COX-2 immunoreactivity was found within the hippocampus (CA3 and CA4 areas in the rat's brain), piriform cortex, amygdala, and in layers II and III of the neocortex. The distribution of COX-2 immunoreactive neurons in the CNS suggests that this isoform may be involved in the processing of visceral and special sensory input and in elaboration of the autonomic, endocrine, and behavioral responses (Breder et al., 1995). In particular, COX-2 localization seems to strictly correlate with the glutamatergic neurotransmission. Indeed, COX-2 is mainly expressed by glutamatergic neurons that are immunolocalized in the cell body, proximal and distal dendrites, and dendritic spines (Kaufmann et al., 1996). Both basal expression of COX-2 and seizure-induced COX-2 overexpression are blocked by MK-801, a noncompetitive antagonist acting at the N-methyl-D-aspartate (NMDA) receptors for the excitatory amino acid L-glutamate, thus suggesting the possible involvement of constitutive COX-2 in glutamatergic mechanisms, mainly within the hippocampus (Yamagata et al., 1993). In addition, the possible link between COX-2 expression and excitotoxic mechanisms driven by exaggerated release of glutamate has also been described in many injury states (see below). Thus, the neuronal localization of COX-2 seems to correlate with the critical period of activity-dependent synaptic remodeling (Kaufmann et al., 1996) and strongly correlates with areas of CNS (such as the hippocampus) where synaptic plasticity may occur. COX is also expressed by cell types other than neurons including astrocytes, oligodendrocytes, and microglia (see Minghetti et al., 1998). In particular, COX-1 positive neurons are detectable in pyramidal neurons of CA3 and CA4 of human hippocampus, unless the concentrations of COX-2 seem to be predominant in these areas. However, COX-1 immunoreactivity prevails on COX-2 within the midbrain, pons, and medulla.

Although the central localization of COX-2, even controversial, seems to correlate with restricted populations of neurons, COX-1 is mainly located on microglial cells where the most relevant immunohistochemical labeling in the CNS is detectable (Yermakova et al., 1999). The possible role of COX-1 is still controversial. However, in contrast to the intense induction of COX-2 by different stimuli applied to microglial cells, the total COX-1 concentration seems to remain unchanged. Until recently, intense proliferation of COX-1 positive microglial cells has been described in many injury states including perilesional areas of cerebral infarct, allergic encephalomyelitis, and experimental glioma (Deininger and Schluesener, 1999; Schwab et al., 2000a,b).

Expression of COX-1 is much greater than COX-2 in fetal hearts, kidneys, lungs, and brains as well as in the decidual lining of the uterus (Gibb and Sun, 1996). Constitutive COX-1 activity in the amnion could also contribute in the maintenance of a healthy pregnancy (Trautman et al., 1996). Furthermore, the so-called "cytoprotective" action of PGs in preventing gastric erosions and ulcerations has mainly brought about endogenously produced prostacyclin and PGE₂, which reduce gastric acid secretion and exert direct vasodilator action on the vessels of the gastric mucosa; prostanoids also stimulate the secretion of gastric mucus and duodenal bicarbonate. Knockout mice, in which the COX-1 gene was deleted, do not develop gastric ulcers spontaneously and show a decreased sensitivity to the damaging effects of indomethacin (Langenbach et al., 1995). In inflammatory processes, COX-2 is expressed in many cells (e.g., macrophages, monocytes, fibroblasts, and synoviocytes) and accounts for the synthesis of prostanoids involved in pathological processes, such as acute and chronic inflammatory states (Wu, 1995).

A. The Contribution of Prostaglandin Biosynthesis in Disease States

1. Inflammation and Pain. PGs are important mediators in spinal nociceptive processing. Basal release of PGD₂, PGE₂, PGF₂, and PGI₂ occurs in the spinal cord and dorsal root ganglia. PGs then bind to G-proteincoupled receptors located in intrinsic spinal neurons (receptor types DP and EP₂) and primary afferent neurons (receptors EP₁,EP₃,EP₄, and IP) (Oida et al., 1995; Kawamura et al., 1997; Beiche et al., 1998; Rowlands et al., 2001; Vanegas and Schaible, 2001). Acute and chronic peripheral inflammation, interleukins, and spinal cord injury increase the expression of COX-2 and the release of PGE₂ and PGI₂. By activating the cAMP and protein kinase A pathway, PGs enhance tetrodotoxin-resistant sodium currents, inhibit voltage-dependent potassium currents, and increase voltage-dependent calcium inflow in nociceptive afferents. These events decrease firing threshold, increase firing rate, and induce release of excitatory amino acids, substance P, calcitonin gene-related peptide (CGRP), and NO. Conversely, glutamate, substance P, and CGRP increase PG release. PGs also facilitate membrane currents and release of substance P and CGRP induced by low pH, bradykinin, and capsaicin. All this should enhance elicitation and synaptic transfer of pain signals in the spinal cord. Direct administration of PGs to the spinal cord causes hyperalgesia and allodynia, and some studies have shown an association between induction of COX-2, increased PG release, and enhanced nociception by facilitating transmission of the nociceptive input (Yamamoto and Nozaki-Taguchi, 1996). The localization of PG biosynthetic pathway in the spinal cord is relevant to understand the contribution of COX enzymes in inflammatory pain. In particular, immunohistochemical studies detailed, in the last few years, the localization and regulation of COX-1 and COX-2 and neuronal NOS in lumbar spinal cord under basal conditions as well as following induction of painful peripheral stimuli. COX-1 immunoreactivity was found in glial cells of the dorsal and ventral horns, but not in neurons (Maihofner et al., 2001). In unstimulated mice, COX-2 immunoreactivity was found in the motorneurons of the ventral horns and in lamina X, but not in dorsal horn neurons. Inflammation causes an increased synthesis of COX-2-dependent PGs, which sensitize peripheral nociceptor terminals and produce localized pain hypersensitivity (Beiche et al., 1996). In addition, after induction of a paw inflammation with zymosan, COX-2 immunoreactivity increased dramatically in dorsal horn neurons of laminae I–VI and X, paralleled by a significant increase in PGE₂ release from lumbar spinal cord (Maihofner et al., 2001). In particular, COX-2 was colocalized with neuronal NOS immunoreactivity in several neurons in superficial laminae of the dorsal horns and in the area surrounding the central canal. Under the same experimental conditions, NOS was distributed in the cytoplasm and extended to processes of some neurons. In contrast, electron microscopy revealed that COX-2 immunoreactivity was restricted to the nuclear membrane and rough endoplasmic reticulum, suggesting that inflammatory processes modulate NOS and COX enzymes in different subcellular compartments of spinal cord neurons involved in pain sensitivity (Maihofner et al., 2001).

Another important role of inflammatory PGs is represented by the induction of swelling. In this case, PGs are thought to cause plasma exudation in a synergistic fashion with other mediators such as complement factor 5a (Williams and Peck, 1977).

In addition, in animal models of arthritis, COX-2 is induced and thought to be responsible for the associated increase in PG production (Anderson et al., 1996). In particular, COX-2 expression has been identified in human osteoarthritis-affected cartilage (Amin et al., 1997) as well as in synovial tissue taken from patients with rheumatoid arthritis (Kang et al., 1996); moreover, PGE₂ appears to "sensitize" peripheral sensory nerve endings located at the site of inflammation (Bley et al., 1998). However, despite a clear role for COX-2 in causing inflammatory swelling in animal models (Chan and Rodger, 1997), the relative role of the two isoforms in pain is more complex. Clearly the perception of acute pain is more likely to be modulated by COX-1 because time for induction must elapse for COX-2. However, a number of studies have strongly implicated a role for COX-2 in inflammatory pain. For instance, highly selective COX-2 inhibitors inhibit hyperalgesia in rats (Riendeau et al., 1997), and the process of sensing pain may lead to the induction of COX-2 in spinal cord (Beiche et al., 1996). Importantly, selective inhibitors of COX-2, e.g., rofecoxib, have been shown to be analgesic in humans when used for postdental surgery pain (Morrison et al., 1999).

Prostanoid-related neurochemical mechanisms other than inflammatory response are also involved in pain sensitivity. Indeed, bicuculline, an antagonist acting at GABA-A receptors when applied to the dorsal surface of the spinal cord, induces highly localized allodynia, an effect sustained with repeated bicuculline application and evoked by NMDA-dependent afferent input. This response is related to spinal PGs overproduction via constitutive COX-2 and depends on spinal PGs contribution to the abnormal processing of tactile input via spinal EP₁ receptors.

On the basis of these evidences, spinal application of specific COX inhibitors sometimes diminishes behavioral responses to persistent nociception. In particular, evidence has recently shown that some COX-2 inhibitors may be useful in many pathological conditions in which an altered pain sensitivity should occur. In particular, it has been shown that meloxicam, a selective COX-2 inhibitor, given peripherally, reduces the prolonged stimulation-evoked afterdischarge of dorsal horn neurons, suggesting that COX-2 may be involved in mediating and/or modulating excitatory effects of nociceptive inputs to dorsal horn neurons (Pitcher and Henry, 2002). In addition, acute administration of a selective COX-2 inhibitor attenuated both ongoing and movement-evoked bone cancer pain, whereas chronic inhibition of COX-2 significantly reduced ongoing and movement-evoked pain behaviors and reduced tumor burden, osteoclastogenesis, and bone destruction by >50% (Sabino et al., 2003). On the other hand, it has been reported that the specific COX-2 inhibitor celecoxib suppressed inflammation-induced PG levels in cerebrospinal fluid, whereas the selective COX-1 inhibitor SC560 was inactive in this regard (Smith et al., 1997). Moreover, the intraspinal administration of a COX-2 inhibitor was accompanied by a decrease in central PGE₂ levels and mechanical hyperalgesia (Samad et al., 2001).

The effect of COX-1 and COX-2 inhibitors depends upon different pathophysiological conditions and by noxious modalities by which nociception is activated. Indeed, it has been shown that the selective COX-1 inhibitor SC560 significantly reduced the formalin-evoked nociceptive response and completely abolished the formalin-evoked PGE₂ raise. In contrast, celecoxib, a selective COX-2 inhibitor, was ineffective in both regards. Indeed, the flinching behavior was largely unaltered, and the formalin-induced PGE₂ raise as assessed using microdialysis was only slightly, not significantly, reduced (Tegeder et al., 2001). This suggests that the formalin-evoked rapid PG release was primarily caused by COX-1 and was independent on COX-2. COX-2 mRNA and protein expression in the spinal cord were not affected by microdialysis alone, but the mRNA rapidly increased following formalin injection and reached a maximum at 2 h. COX-2 protein was unaltered up to 4 h after formalin injection. The time course of COX-2 up-regulation suggests that the formalin-induced nociceptive response precedes COX-2 protein de novo synthesis and may therefore be unresponsive to COX-2 inhibition. Thus, it may be hypothesized that the efficacy of celecoxib in early injury-evoked pain may be less than that of unselective NSAIDs (Tegeder et al., 2001). Thus, on the basis of these evidences, it may be argued that unless clear evidence exists in favor of the benefit of selective and nonselective NSAIDs in acute pain, the relative contribution of COX-1 and COX-2 in chronic pain (i.e., associated with rheumatoid or osteoarthritis) remains to be established.

2. Cancer. One of the exciting observations correlated with the use of NSAIDs is the association with a reduction in the incidence of colon cancer. Indeed, a retrospective study revealed the findings that patients taking relatively low doses of aspirin (a maximum effect being seen at four to six tablets per week for long periods of time) had substantially reduced risks of developing colon cancers (Thun et al., 1991; Giovannucci et al., 1995). The NSAID sulindac can reduce the number and size of polyps in patients with familiar adenomatous polyposis (Giardiello et al., 1993; Giovannucci et al., 1994). It is not entirely clear how this protective effect of NSAIDs is exerted. However, adenocarcinomas in human subjects appear associated with marked increases in COX-2 expression (Smalley and Dubois, 1997), and evidence from studies with isolated cells in culture (Dubois et al., 1998) or animal models (Williams et al., 1997) similarly points to COX-2 being the level at which the beneficial effects of NSAIDs are exerted. Several hypotheses have been postulated to clarify the mechanisms by which the overexpression of COX-2 may contribute to colorectal carcinogenesis. The overexpression of COX-2 in rat intestinal epithelial cells and the addition of PGE₂ to human colon cells are associated with increased levels of Bcl-2 and thus, with resistance to apoptosis (Tsujii and Dubois, 1995; Sheng et al., 1998). Moreover, the growth of tumors is associated with immune suppression; PGE₂ inhibits, in vitro, the production of tumor necrosis factor- (TNF-) and induces the production of interleukin-10 (Kambayashi et al., 1995), and this response is reduced by exposure to NSAIDs of the colon cancer cells (Dubois et al., 1998). These results suggest that PGE₂ can suppress the cell-mediated anti-tumor immune response. COX-2 activity is also associated with the promotion of tumor angiogenesis in colon cancer cell lines (Tsujii et al., 1998) and in human colorectal cancer (Cianchi et al., 2001). The effect of selective COX-2 inhibitors on colon cancer in man is currently under investigation. However, Tsujii and colleagues (1998) have also shown that COX-1 in endothelial cells plays an important role in the modulation of angiogenesis. PGs produced by COX-1 in endothelial cells could be important in regulating genes required for endothelial tube formation, and it could be a relevant target for cancer prevention or treatment in tumors lacking COX-2 expression. As such, NSAID could inhibit angiogenesis by inhibition of COX-2 activity in colon carcinoma cells and down-regulating production of angiogenic factors by induction of apoptosis and by inhibiting COX-1 activity in endothelial cells. Recent studies have further indicated COX-2 overexpression in various other malignancies, such as lung (Hida et al., 1998), breast (Ristimaki et al., 2002), esophageal (Zimmermann et al., 1999), gastric (Murata et al., 1999), pancreatic (Tucker et al., 1999), prostate (Klotz et al., 1998), head and neck (Gallo et al., 2000; Gupta et al., 2000), and hepatocellular carcinoma (Bae et al., 2001; Fantappiè et al., 2002). On the basis of these data, it is conceivable that specific COX-2 inhibitors might be used as adjuvants in the treatment of tumors, as well as in cancer prevention.

3. Neuronal Excitotoxicity. The contribution of PG biosynthesis and release in the pathophysiological mechanisms underlying brain disorders has been investigated in the last three decades in almost all brain injury experimental models. In particular, excitotoxic mechanisms associated with acute exposure of brain tissue to noxious endogenous as well as exogenous neurotoxins and chronic neuroinflammatory/neurodegenerative disorders have widely been explored to evaluate the potential neuroprotective activity of compounds known to modulate PG biosynthesis. Indeed, direct administration in brain tissues of excitatory stimuli such as NMDA and kainate (which stimulate ionotropic channels reactive to the endogenous excitatory amino acid L-glutamate) has been associated to prostanoid-mediated neuronal cell death (Chen et al., 2002). This effect is associated with NMDA-related activation of COX-2, which as an immediate-early gene, is dramatically and transiently induced in these neurons. In addition, administration of kainic acid, an analog of the excitotoxin glutamate, results in hippocampal cell death and seizures in mice. The hippocampal lesions were associated with a high level of COX-2 production as well as astrogliosis (Chen et al., 2002).

Besides these evidences, the possible effect of COX inhibitors in such mechanisms leading to excitotoxic brain damage is still to be clarified. Indeed, in primary cortical neurons, both indomethacin (COX-1/-2 nonselective inhibitor) and aspirin (COX-1 preferential inhibitor) significantly reduced basal and kainic acid-induced PGE₂ production and prevented neuronal cell death after kainic acid treatment. In contrast, NS-398 (COX-2 selective inhibitor) had no effect on kainic acid-induced neuronal cell death. In hippocampal neurons, however, COX-2 inhibitors prevented both kainic acid-induced neuronal death and PGE₂ production (Kim et al., 2001). COX-2 expression was remarkably up-regulated by kainic acid in hippocampal neurons, whereas in cortical neurons, COX-2 expression was comparatively less significant. Astrocytes were unresponsive to kainic acid in terms of PGE₂ production and cell death. Thus, the release of PGE₂ induced by kainic acid seems to occur through COX-1 activity rather than COX-2 in cortical neurons (Kim et al., 2001). The inhibition of PGE₂ release by COX-1 inhibitors prevented kainic acid-induced cortical neuronal death, whereas in the hippocampal neurons, COX-2 inhibitors prevented kainic acid-induced PGE₂ release and hippocampal neuronal death, thus suggesting that excitotoxic mechanisms may involve differentially COX-1 and COX-2 according to the area of the brain undergoing excitotoxic lesion (Kim et al., 2001).

The reasons of such controversial effects of COX inhibitors in excitatory neurotransmission is unknown. However, recent studies indicate that AA metabolites interfere in discrete brain regions with L-glutamate turnover acting at many levels, including the reuptake and storage of excessive L-glutamate by astrocytes, the modulation of Ca²⁺-linked glutamate receptors, and the release of free radicals (including NO and peroxynitrite) by postsynaptic neurons which, in turn, results in neuronal and astroglial damage (Volterra et al., 1992, 1994). In particular, it has been shown that AA itself is able to inhibit glutamate reuptake by glial cells, a key event in the regulation of glutamate concentration in the synaptic wall. Since the neurotoxic activity of glutamate is closely related to its relative concentration across L-glutamate receptors, it is likely that COX inhibitors, by regulating back AA concentrations, may directly or indirectly interfere with L-glutamate neurotoxicity (Volterra et al., 1992, 1994).

Mechanisms other than excitotoxicity have also been taken into account for detecting the possible role of prostanoid biosynthesis in brain injury. In particular, apoptotic cell death, occurring mainly via the chronic activation of cytokine/free radical-mediated mechanisms, strongly involves PG release. In particular, evidence exists that most of the inflammatory processes occurring in the early stages of some neuroimmune diseases such as allergic encephalomyelitis, multiple sclerosis, Alzheimer's disease (AD), and HIV-related brain disorders are associated with COX-1 and mainly COX-2 overexpression, an effect which suggested the potential benefit when using COX inhibitors in the treatment of such diseases (see below). The exact mechanisms involved in prostanoid-mediated activation of apoptotic machinery is unclear. However, evidence exists that apoptosis is mainly mediated by the release of PGE₂ which would activate the programmed death of neurons via an EP₂-like receptor (Takadera et al., 2002).

Besides these evidences, the possible use of COX-2 inhibitors in the treatment of such and other diseases characterized by apoptotic neuronal cell death is still controversial. Indeed, treatment of male Sprague-Dawley rats with kainic acid, which triggers limbic seizures in 60% of the animals and induces COX-2 mRNA expression in the pyramidal cells of the hippocampus, induces cell loss via apoptotic mechanisms in the amygdala and the piriform cortex. Treatment with rofecoxib selectively attenuated the number of apoptotic cells in the hippocampus, whereas the cells of the thalamus, amygdala, and piriform cortex were not protected, thus suggesting that COX-2 inhibitors may exert an antiapoptotic effect against excitotoxic stimuli only in very selected areas of the brain (Kunz and Oliw, 2001).

The role of COX-2 expression in apoptotic phenomena seem to be associated with intense cooperation with neurotrophin-mediated mechanisms occurring in insulted brain cells. Indeed, it has been demonstrated that activation of COX-2 inhibits nerve growth factor withdrawal apoptosis in differentiated PC12 cells (Chang et al., 2000). The inhibition of apoptosis by COX-2 was concomitant with prevention of caspase 3 activation and was associated with increased expression of prosurvival genes coupled to inhibition of NO- and superoxide-mediated apoptosis. This indicates that an active cooperation exists between the COX pathway and neurotrophin/free radical-mediated mechanisms which regulate both necrosis and apoptotic cell death according to the cell type and the intensity of the injuring stimuli applied to brain tissues.

4. Neuroinflammatory Processes. An altered biosynthesis of prostanoids is clearly involved in neuroinflammatory processes (Genis et al., 1992; Cao et al., 1996). Indeed, acute inflammation following spinal cord injury results in secondary injury and pathological reorganization of the CNS architecture associated with substantial changes in COX expression (Schwab et al., 2000a,b). In particular, in injured spinal cord, COX-1 positive microglia/macrophages accumulated significantly at perilesional areas and in the developing necrotic core early after injury, being the number of COX-1 positive cells persistently elevated up to 4 weeks following injury. Furthermore, COX-1 positive cells were located in perivascular spaces, between spared axons, and in areas of Wallerian degeneration (Schwab et al., 2000a,b).

On the other hand, overexpression of both COX-1 and COX-2 and subsequent abnormal release of prostanoids has been described after exposure of brain cells to many inflammatory agents such as endotoxin and cytokines. Astroglial cells and, in a most relevant extent, microglial cells release large amounts of PGE₂ and PGD₂ after exposure to Escherichia coli LPS (see Minghetti and Levi, 1998). This effect seems to correlate directly with the expression of COX-2 enzyme in microglial cell culture. A nearly similar effect has been described when cytokines were incubated with glial cells. In particular, IL-1 alone or in combination with interferon- (IFN-) has been found to release large amounts of PGE₂ (Mollace et al., 1995, 1998; Janabi et al., 1996; O'Banion et al., 1996; Slepko and Levi, 1996) from cultured astrocytes as well as microglial cells, an effect mainly due to COX-2 expression.

Many substances able to produce inflammatory responses in brain tissues other than LPS or cytokines have been found to also act via prostanoid formation. In particular, evidence exists that several components of HIV envelope, such as gp120 glycoprotein, may account for some inflammatory aspects of neurodegenerative disorders accompanying AIDS-dementia complex by activating both NO and prostanoid formation (Genis et al., 1992; Mollace et al., 1993, 1994; Bagetta et al., 1998; Pereira et al., 2000). This effect occurred via expression of COX-2 isoenzyme in HIV-infected monocyte-derived macrophages, human brain microvascular endothelial cells in vitro (Pereira et al., 2000), and in neocortical neurons in vivo in which COX-2 overexpression was accompanied by apoptotic phenomena (Bagetta et al., 1998).

COX-2 protein is also up-regulated in macrophages causing active demyelination. Indeed, by means of in situ hybridization it has been described that COX-2 mRNA signals were strongly expressed on macrophages adhering to the demyelinating nerve fibers at the endoneurium. This observation has suggested a rationale for the application of neuroprotective strategies employing COX-2 inhibitors in inflammatory demyelinating neuropathies (Kawasaki et al., 2001). This is further demonstrated by the evidence that progression of myelopathies, which involve spinal motorneurons such as amyotrophic lateral sclerosis, is driven by inflammatory-related events associated with altered prostanoid biosynthesis. Indeed, in both early symptomatic and end-stage transgenic mice undergoing amyotrophic lateral sclerosis, neurons and, to a lesser extent, glial cells in the anterior horn of the spinal cord exhibit robust COX-2 immunoreactivity (Almer et al., 2001).

On the other hand, other aspects of inflammation of brain tissues are related to COX-2 up-regulation and the subsequent prostanoid formation. Indeed, it has recently been shown that in mice expressing transgenic COX-2 in anterior hypothalamus, the febrile response was significantly potentiated in transgenic compared with non-transgenic mice, with an accelerated onset of fever by 1 to 2 h after LPS administration, suggesting a role for neuronally derived COX-2 in the fever response (Vidensky et al., 2003). In addition, overexpression of COX-2 in the brain of a transgenic mouse line leads to selective induction of endogenous complement component C1qB expression in neurons, suggesting that neuronal COX-2 may influence inflammatory responses in the brain, in part, through the modulation of complement gene expression (Spielman et al., 2002). Finally, it has recently been shown that COX-2, in part, through TNF--related mechanisms, contributes to LPS-induced neuronal death, since this effect was antagonized by selective COX-2 inhibitors. COX-2, in addition to its role in glutamate excitotoxicity, participates in the cytotoxicity associated with neuroinflammatory processes (Araki et al., 2001).

5. Alzheimer's Disease. Evidence exists that neuroinflammatory mechanisms associated with abnormal modulation of COX pathway represent the central feature of AD. In AD, signs of an inflammatory activation of microglia and astroglia are present inside and outside amyloid deposits. In addition, studies carried out be means of cell culture and animal models of AD suggest an interactive relationship between inflammatory activation, reduced neuronal functioning, and deposition of amyloid (see Pasinetti, 2001).

In particular, it has been shown that COX-2 is up-regulated in the brain with AD, an effect which has been shown to be injurious to neurons (Pasinetti, 2001). The immunointensity of COX-2 signal in the CA3 and CA2 but not CA1 subdivisions of the pyramidal layers of the hippocampal formation of the AD brain increased as the disease progressed. COX-2 signal was increased in all three regions examined among cases characterized by severe dementia, indicating that neuronal COX-2 content in subsets of hippocampal pyramidal neurons may be an indicator of progression of dementia in early AD. This has also been shown in transgenic mice overexpressing constitutively COX-2 in neurons and producing elevated levels of PGs in brain. Those animals developed an age-dependent deficit in spatial memory at 12 and 20 months but not at 7 months and a deficit in aversive behavior at 20 months of age. These behavioral changes were associated with a parallel age-dependent increase in neuronal apoptosis occurring at 14 and 22 months but not at 8 months of age and astrocytic activation at 24 months of age. These findings suggest that neuronal COX-2 may contribute to the pathophysiology of age-related diseases such as AD by promoting memory dysfunction, neuronal apoptosis, and astrocytic activation in an age-dependent manner (Andreasson et al., 2001).

Many factors have been identified in AD brain which are known to promote and sustain inflammatory responses and subsequently altered biosynthesis and release of PGs. They include -amyloid protein, the pentraxins C-reactive protein and amyloid P, complement proteins, the inflammatory cytokines interleukin-1, interleukin-6, and TNF-, the protease inhibitors -2-macroglobulin, and -1-antichymotrypsin (for a review, see McGeer and McGeer, 2001). In particular, in AD brains, COX-1 positive microglial cells were primarily associated with amyloid plaques, while the number of COX-2 positive neurons was increased compared with that in control brains. The different distribution patterns of COX-1 and COX-2 in AD could implicate that these enzymes are involved in different cellular processes in the pathogenesis of AD (Hoozemans et al., 2001).

The possible interactions between prostanoid formation and inflammatory effects of -amyloid in AD has been better assessed in the past few years as a potential target for the treatment of AD with COX inhibitors. Recent studies demonstrated that COX-2 expression is closely related to the expression of high levels of mRNA for the amyloid precursor protein (APP). Amyloid -peptide and a secreted form of APP, both derived from APP by proteolysis, were also increased, since both effects were inhibited by a selective COX-2 inhibitor (JTE-522) and by nonselective COX inhibition using indomethacin, thus suggesting that COX pathways may play important roles in the -amyloid-related neurodegenerative processes of AD (Kadoyama et al., 2001). In addition, injection into the nucleus basalis of the rat of preaggregated amyloid- [A(1–42)] segments produced a congophylic deposit and microglial and astrocyte activation and infiltration and caused a strong inflammatory reaction characterized by IL-1 production, increased COX-2, and inducible NOS (iNOS) expression. Rofecoxib, a COX-2 inhibitor, reduced microglia and astrocyte activation, iNOS induction, and p38 mitogen-activated protein kinase (MAPK) activation to control levels showing that COX-2 overexpression by -amyloid is a crucial event in the inflammatory cascade occurring in AD (Giovannini et al., 2002). Moreover, human COX-2 expression in APP/COX-2-expressing mutant mice induces potentiation of brain parenchymal amyloid plaque formation and a greater than 2-fold increase in PGE₂ production suggesting that COX-2 influences APP processing and promotes amyloidosis in the brain (Xiang et al., 2002). On the other hand, it has been shown that both the synthesis of the APP and its processing (i.e., to amyloidogenic A peptides, soluble nonamyloidogenic APPs, and other APP fragments) are regulated by mediators, including prostanoids, able to elevate cAMP levels into brain cells. In addition, evidence exists that the neurotoxic and proinflammatory actions of the Alzheimer peptide A are dependent on its aggregation and -sheet conformation. Chronic use of NSAIDs, such as aspirin for arthritis, decreases the risk of developing AD by unknown mechanisms (Pasinetti 2001). Recently, it has been found that aspirin prevented enhanced A aggregation by aluminum, an environmental risk factor for AD. This antiaggregatory effect was restricted to NSAIDs and was not exhibited by other drugs used in AD therapy (Thomas et al., 2001). Furthermore, S-2474, a selective COX-2 inhibitor, showed a protective effect on A-induced cell death in primary cultures of rat cortical neurons. In particular, S-2474 ameliorated A-induced apoptotic features, such as the condensation of chromatin and the fragmentation of DNA completely, indicating that COX-2 inhibitors may possess therapeutic potential for AD via ameliorating degeneration in neurons as well as by suppressing chronic inflammation in non-neuronal cells (Yagami et al., 2001). Finally, it has been demonstrated that neuroimmunophilin ligands [like cyclosporin A or FK-506 (tacrolimus)] and NSAIDs, including COX-2 inhibitors, can also prevent APP overexpression and the overproduction of amyloidogenic peptides. Indeed, APP overexpression by PGE₂ is inhibited by neuroimmunophilin ligands like cyclosporin A or FK-506. In addition, NSAIDs, which reduce PG synthesis by inhibiting COX-1 and -2 enzymes, might also be expected to lower APP levels and increased levels of soluble APPs in the media of cultured astrocytes and neurons, perhaps acting by inhibition of PG production. Since APP haloprotein can be amyloidogenic, whereas APPs may be neurotrophic, it can be suggested that some neuroimmunophilin ligands, NSAIDs, and COX-2 inhibitors might suppress amyloid formation and enhance neuronal regeneration in Alzheimer's disease (Lee and Wurtman, 2000).

The expression of COX-2 in the brain of patients with AD, however, is strictly related to the inflammatory injury which characterizes the early stages of the disease. Indeed, COX-2 occurs simultaneously to the activation of inflammation-associated enzymes such as p38 MAPK and is not restricted to glial cells, but may also be found in neurons and may contribute to intraneuronal damage (Hull et al., 2002). In contrast, neuronal COX-2 expression is decreased in AD subjects with dementia compared with nondemented subjects in different hippocampal subfields (Yermakova and O'Banion, 2001). These changes also occurred in subjects with other dementia and thus may not be disease specific. The proportion of COX-2 positive neurons decreased in subjects with clinical dementia rating 5 but not clinical dementia rating 4, suggesting that this was a late event in the course of the disease. Furthermore, COX-2 was not preferentially associated with paired helical filament immunoreactivity, a marker of neuronal pathology. COX-2 immunoreactivity was also observed in astrocytes and cerebrovasculature. Indeed, the density of COX-2 immunopositive astrocytes was increased in AD temporal cortex. Thus, it is unlikely that neuronal COX-2 contributes to pathology in end-stage AD (Yermakova and O'Banion, 2001).

6. Brain Ischemia. PGs and other arachidonic acid metabolites are subjected to altered synthesis or relocation after an ischemic insult. This effect is mainly due to changes occurring in many steps of prostanoid formation, including phospholipase A2 (PLA2) and COX isoforms. In addition, many COX inhibitors have shown to possess protective effects when used in experimental models of brain ischemia.

In particular, expression of group IIA secretory PLA2 (sPLA2-IIA) is documented in the cerebral cortex after ischemia, suggesting that sPLA2-IIA is associated with neurodegeneration. Indeed, after mean cerebral artery occlusion, sPLA2 activity was increased in the cortex and associated to a neurodegenerative effect, since both responses were prevented by the sPLA2 inhibitor, indoxam (Yagami et al., 2002). The neuroprotective effect of indoxam was observed even when it was administered after occlusion. In addition, in primary cultures, sPLA2-IIA caused marked neuronal cell death. Morphologic and ultrastructural characteristics of neuronal cell death by sPLA2-IIA were apoptotic, as evidenced by condensed chromatin and fragmented DNA. Before apoptosis, sPLA2-IIA liberated AA and generated PGD2 from neurons. Indoxam significantly suppressed not only AA release, but also PGD2 generation. Indoxam also protected neurons from sPLA2-IIA-induced cell death, since this effect was found even when it was administered after sPLA2-IIA treatment. Furthermore, the use of COX-2 inhibitors significantly prevented neurons from sPLA2-IIA-induced PGD2 generation and subsequent cell death (Yagami et al., 2002). This suggests that sPLA2-IIA induces neuronal cell death via apoptosis, which might be associated with overproduction of AA metabolites, especially PGD2 in ischemic brain tissues.

An increase of both COX-1 and COX-2 have also been found in brain ischemia. In particular, focal ischemia induced in the frontoparietal region of rat brain is accompanied by formation of PGD2 peaking 60 to 90 min postischemia and declining thereafter. This effect is due to increased COX-2 and is characterized by morphological alterations with necrosis of neurons, glial cells, and blood vessels surrounded by a halo with pyknotic cells with cytoplasm swelling and vacuolization (Govoni et al., 2001). Many cell types within ischemic brain tissue have been shown to overexpress COX enzymes; however, in focal ischemic damage, the elevation of COX-2 was restricted to microglial cells, suggesting that the isozymes of COX are differentially regulated depending on the cellular source and the types of ischemic damage (Tomimoto et al., 2002).

The administration of COX inhibitors before as well as soon after the ischemic insult reduces the extension of cerebral damage in rats. In particular, either selective inhibition of COX-2 with rofecoxib or inhibition of COX-1 with valeryl salicylate significantly increased the number of healthy neurons in the hippocampal CA1 sector even when the treatment began 6 h after global ischemia, suggesting that both COX isoforms are involved in the progression of neuronal damage following global cerebral ischemia and have important implications for the potential therapeutic use of COX inhibitors in cerebral ischemia (Candelario-Jalil et al., 2003). In the same way, SC-58236, a selective COX-2 inhibitor, dose-dependently prevented ischemia-induced eicosanoid formation and caused significant reduction of the damaged area, suggesting that selective inhibitors of COX-2 are neuroprotective (Govoni et al., 2001). Moreover, anoxic stress attenuates NMDA-induced pial arteriolar dilation via a mechanism involving actions of COX-derived reactive oxygen species, an effect protected by the selective COX-2 inhibitor NS-398 (Domoki et al., 2001). Finally, the specific long-acting COX-2 inhibitor SC-58236 also showed a protective effect in a reversible rabbit spinal cord ischemia model (Lapchak et al., 2001).

The effects of COX antagonism are relevant not only for the protection of ischemic tissue but also in the area surrounding the ischemic core. Indeed, the marginal area surrounding a region of ischemic brain tissue, designated as the penumbra, is of interest as a potential area for the rescue of neurons from cell death. In the penumbra and surrounding cortex, it has been shown that up-regulation of c-Fos, brain-derived neurotrophic factor, and COX-2 mRNAs was observed, whereas expression of HSP70 mRNA was restricted to the penumbra. This spatial discrepancy of mRNA expression suggests that different mechanisms are involved in the regulation of c-Fos/brain-derived neurotrophic factor/COX-2 and HSP70 expression (Kinoshita et al., 2001).

In addition to these evidences, inhibition of COX does not seem to represent the only mechanism involved in the neuroprotection that follows the treatment of ischemic injury by means of NSAIDs. Indeed, both piroxicam and NS-398 protect neurons against hypoxia/reperfusion. However, evidence exists that their protective effect is independent of COX activity and not solely explained by modulation of NF-B and AP-1 binding activity. Instead, piroxicam- and NS-398-induced phosphorylation through extracellular signal-regulated kinase pathway may contribute to the increased neuronal survival (Vartiainen et al., 2002). Furthermore, COX-2 is an important modulator in the enhancement of proliferation of neural progenitor cells after ischemia (Sasaki et al., 2003). Due to these latter evidences, the use of COX-2 inhibitors is restricted to the early stages of ischemic insult (when the neurochemical changes of injured tissues relevant to neuronal and glial dysfunction are mainly driven by inflammatory processes). These have to be further investigated in the phases of repair mechanisms which occur later in the course of the disease.

B. Selective Nonsteroidal Anti-Inflammatory Drugs

NSAIDs is a "catch all" name for a large number of chemically distinct drugs. Together they represent the single most important group of self-prescribed pharmaceuticals and the most widely used drug class. Sales of NSAIDs are now estimated at 5.8 billion dollars a year with the United States accounting for 1.8 billion dollars (Vane and Botting, 1996). The therapeutic benefits of all NSAIDs include inhibition of swelling and/or pain at the site of inflammation. In addition, aspirin also offers protection against stroke and thrombosis, Alzheimer's disease, and cancer (see above). There are, however, side effects of NSAIDs that limit their use in some patients. Most common among these side effects is irritation and damage to the gastrointestinal mucosa, particularly at the gastric level. Each member of the NSAIDs family has some individual effects, however, the unifying common mechanism of action of all is inhibition of COX (Vane, 1971). Together with the identification of two distinct isoforms of COX, a new hypothesis has been formulated to explain the effects of this class of drugs; COX-2 inhibition accounts for the therapeutic benefits and inhibition of COX-1 for the side effects of NSAIDs (Mitchell et al., 1994). The following explains why this hypothesis was presented and how its validity was proven. As a class, the NSAIDs represent a major risk for morbidity and mortality from gastrointestinal damage, perforation, ulcers, and bleeding. As reported by Mitchell and Warner (1999), in the United States the number of deaths per year due to NSAIDs is approximately 16,500 with 107,000 hospitalizations in the same period (Fries, 1998). In the United Kingdom, it is estimated that 12,000 ulcer complications and 1200 deaths per year are directly linked to NSAID intake (Hawkey, 1996). However, some NSAIDs cause more gastrointestinal side effects than others (Henry et al., 1996).

A number of different experimental assays have been used to compare the potencies of NSAIDs on COX-1 and COX-2. The results from these assays are used to calculate a measure of COX-2 selectivity, and then NSAIDs are compared with each other by ranking their COX-2 selectivity. The many test systems developed have resulted in different COX-1/COX-2 ratios, sometimes for the same drug, thus confusing comparisons. However, on the whole the in vitro studies have shown that the ability of a given NSAID to inhibit COX-1 correlates with the degree of side effects it causes. Moreover, the use of COX screens in vitro has been remarkably successful and has led to the development of a number of COX-2 selective compounds. These include from Merck Frosst (Kirkland, QC, Canada) L-745,337 (Chan et al., 1995), DFU [5,5-dimethyl-3-(3-fluorophenyl)-4-(4-methylsulphonyl)phenyl-2-(5H)-furanone] (Riendeau et al., 1997), and DFP [3-(2-propyloxy)-4-(4-methylsulphonylphenyl)-5,5-dimethylfuranone] (Black et al., 1999) and from Monsanto Searle (St. Louis, MO) SC-58125 (Guo et al., 1996). Furthermore, this approach has led to the development of two COX-2 selective NSAIDs with the approval of the Food and Drug Administration that are currently available in the United States and Europe. These are rofecoxib (Vioxx; Chan et al., 1999) from Merck Frosst and celecoxib (Celebrex; Geis, 1999) from Pfizer (St. Louis, MO). In addition to the development of new compounds by these screens, it has been possible to identify currently prescribed NSAIDs that are COX-2 selective inhibitors. For example, using whole blood assay compounds displaying more than 5-fold selectivity for COX-2 would include etodolac, meloxicam, and nimesulide (Warner et al., 1999).

However, two other important questions are still ongoing: are COX-2 selective inhibitors really safe for the gastrointestinal tract, and is the selective inhibition of cyclooxygenase-2 sufficient for a full anti-inflammatory efficacy? A large amount of studies have been published showing different responses to the above questions. Recently, Wallace (1999) tried to comment on these two different aspects in a very interesting article published in Trends in Pharmacological Science. There is large evidence to support the claim that selective inhibitors of COX-2 produce less gastric damage than standard NSAIDs when administered acutely to healthy animals (Seibert et al., 1994; Chan et al., 1995; Futaki et al., 1993). However, it has also been demonstrated that most patients taking standard NSAIDs do not develop clinically significant gastric injury; rather, it occurs in a subset of patients who are more susceptible to the gastric-damaging actions of these drugs (Wolfe et al., 1999). Patients infected with Helicobacter pylori are more likely than those not infected to develop endoscopic ulcers when taking NSAIDs. Other influences on endoscopically detected ulcers are less well established. Potential factors that have been reported as associated with NSAID-induced ulceration are smoking, male sex for duodenal lesions, apparently independent of H. pylori status, age, and dose of drug assumed, particularly for gastric ulcers (Hawkey et al., 1998). We still have much to learn about the potential risks of the inhibition of COX-2 in the gastrointestinal tract; for example, COX-2 inhibitors impair tolerance of dietary antigen (Newberry et al., 1999) and exacerbate experimental colitis in rodents (Morteau et al., 2000).

COX-2 also appears to play an important role in promoting the healing of ulcers in the stomach. It has been recognized for many years that NSAIDs interfere with the healing of peptic ulcers in humans (Armstrong and Blower, 1987; Stadler et al., 1991), whereas administration of PGs accelerate their healing (Jaszewski et al., 1992). Mizuno and coworkers (1997) recently showed that COX-2 mRNA and protein were strongly induced in the mouse stomach in which an ulcer had been induced, and a parallel increase in mucosal PG synthesis was found. Although COX-2 expression in the healthy stomach is low and expression around a site of ulceration is considerably higher, it should be noted that rapid induction of COX-2 can occur, even in response to quite subtle mucosal irritation. A further functional role for COX-2 in mediating gastric epithelial proliferation has been demonstrated by Sawaoka et al. (1997), whereas Nakatsugi et a1. (1996) have provided evidence that COX-2 contributes to mucosal defense in a rat model of stress ulcer. Central to the "selective COX-2 inhibitor" thesis is the assumption that COX-2 is the major form responsible for the production of PGs at sites of inflammation. However, if COX-1 contributes significantly to the production of PGs at such sites, selective block of COX-2 will not produce anti-inflammatory effects to the same extent as drugs that inhibit both isoforms. There is excellent evidence for marked up-regulation of COX-2 at sites of inflammations. In some models, there is also evidence for anti-inflammatory effects of COX-2 inhibitors at doses that do not significantly affect gastrointestinal PG synthesis or COX-1 activity. COX-2 has been constitutively localized in the renal vasculature, the cortical macula densa, and the medullary interstitial cells of the kidney, and its content in these areas increases with age. COX-1 is found in the vasculature, the collecting ducts, and in the thin loops of Henle (Nantel et al., 1999). The presence of both isoforms in the vasculature raises the question of which is the predominant source of the increased production of vasodilator PGs that are critical to the preservation of renal blood flow in the presence of volume depletion. Inhibition of this homeostatic response accounts for the most common renal side effects associated with nonselective NSAID therapy (Nantel et al., 1999). Little information on the renal pharmacology of COX-2 inhibitors in humans is available at this moment. An analysis of the postmarketing data for celecoxib revealed that edema occurred in 2.1% of patients, hypertension in 0.8%, and exacerbation of preexisting hypertension in 0.6%, a profile similar to those of nonselective NSAIDs (Whelton et al., 2001); similarly, post hoc analysis of the rofecoxib database revealed that peripheral edema occurred in 3.8% of patients treated daily with 25 mg of rofecoxib (Whelton, 2001). Controlled studies are necessary to assess the risk of hypertension. However, there are also examples of studies in which anti-inflammatory efficacy was not observed unless doses of the COX-2 inhibitors were used that were well above those necessary for COX-2 inhibition. Indeed, in this particular case, suppression of PG synthesis at the site of inflammation correlated significantly with the suppression of COX-1, but not with the suppression of COX-2 (Wallace et al., 1998).

III. Nitric Oxide and Nitric Oxide Donors

NO is a free radical molecule, which was discovered to be a potent vasodilator (Vallance et al., 1989) as well as a novel type of retrograde neurotransmitter (Snyder and Bredt, 1992). At the vascular level, NO appears to mediate endothelium-dependent relaxation of vascular smooth muscle (Ignarro et al., 1987; Palmer et al., 1987) and to be chemically equivalent to endothelium-derived relaxing factor (EDRF) discovered by Furchgott and Zawadzki (1980). NO is a double-edged sword, serving as a key signal molecule in both physiological and pathological processes. NO and its reaction products, reactive nitrogen oxide species, has been found to modulate all facets of physiology and pathology in species as evolutionarily distant as plants and humans. Additionally, the chemical literature regarding the role of NO, especially as a pollutant, has been with us for over a century. The naming of NO as "Molecule of the Year" by Science magazine in 1992 and the award in 1998 of the Nobel Prize in Medicine for the role of NO in the cardiovascular system reflects the importance given to this molecule by the general scientific community.

NO was described as an effector product of activated macrophages (Tayeh and Marletta, 1989). As is often the tendency in science, the ability of NO to cause cytostasis and cytotoxicity in tumor cells and certain pathogens resulted in its initial perception as a beneficial mechanism to the host. Furthermore, the finding that NO derived from the endothelial NOS (eNOS) could inhibit platelet aggregation and the adhesion of platelets (Radomski et al., 1987; Sneddon and Vane, 1988) and of activated neutrophils (Kubes et al., 1991) suggested a beneficial role for NO in ischemia/reperfusion injury. Under physiological conditions, in all parts of the body, the concentrations of NO are believed to be fluctuating continuously at rather low levels. These levels are controlled by constitutively expressed neuronal and endothelial types of NO synthase [nNOS (NOS-I) and eNOS (NOS-III), respectively], which are widely distributed in the body. However, realization that concentration of NO can rapidly increase by the massive expression of the iNOS (NOS-II) in sepsis (Szabò and Thiemermann, 1995) or by hyperactivation of nNOS in glutamate-mediated neurodegenerative processes in the central nervous system, led to a change in this perception (Dawson, 1995). Rapidly, data were published which implicated the aberrant expression of iNOS in numerous inflammatory conditions, such as inflammatory bowel disease, Crohn's disease, Alzheimer's disease, and hemorrhagic shock, to name a few (Boughton-Smith et al., 1993; Clancy and Abramson, 1995). These studies and numerous others began to paint a darker picture of NO as a toxic byproduct of inflammation that should be inhibited to restore homeostasis. Thus, the great effort of several groups and in particular of the Moncada group was to develop selective inhibitors of the inducible form of the enzyme as novel therapeutics for disease (Moncada and Higgs, 1995; Ho and Pasinetti, 2001). These tools provided not only invaluable information as to the proinflammatory roles of NO but started to highlight its potential role as an anti-inflammatory molecule. Thus, it is now becoming more and more evident that NO can be considered a double-edged sword. On one hand, it can exert beneficial effects on the body by acting as an antibacterial, antiparasital, antiviral agent, or as a tumoricidal agent; on the other hand, high levels of NO, if uncontrolled, can be detrimental. Such negative ef