Cyclooxygenase Inhibitors: Basic Aspects

Inhibition of Cyclooxygenases

Inhibition of prostaglandin production through blockade of cyclooxygenases (COXs) was identified in the early 1970s as the major mechanism of action of aspirin/acetylsalicylic acid and the pharmacologically related non-steroidal anti-inflammatory drugs (NSAIDs) ( ). A variety of stimuli induce the formation of arachidonic acid from phospholipids of the cell membrane, mainly through cytosolic phospholipase A ₂ . Arachidonic acid then serves as the substrate for prostaglandin G/H synthases, colloquially also called COXs, which convert arachidonic acid into prostaglandin G ₂ (PGG ₂ ) and PGH ₂ in a two-step process consisting of an initial COX reaction and a second hydroperoxidase reaction ( Fig. 32-1 ). In the late 1980s and early 1990s, the first cDNA of COX enzymes from seminal vesicles was sequenced ( , ). Molecular biologists subsequently discovered a second COX isoform whose expression was regulated by cytokines and glucocorticoids ( , ). Both isoforms were found to be encoded by separate genes and were called prostaglandin G/H synthase-1 and -2, or COX-1 and COX-2 ( Fig. 32-2 ). They differ in their regional expression and temporal inducibility, which forms the basis of their different functions in health and disease ( ). COX-1 is constitutively expressed in most tissues and provides the tonic supply of prostanoids required for homeostasis in many organs, organ systems, and cells, including the upper gastrointestinal (GI) tract, platelets, and kidney. COX-2, on the other hand, is under the transcriptional control of pro- and anti-inflammatory cytokines. In many cells such as macrophages it becomes expressed only in response to inflammatory stimuli ( ), whereas immunosuppressants such as glucocorticoids reduce its expression ( , ). Such concepts fostered the idea of selective inhibition of COX-2 being sufficient for analgesic/anti-inflammatory effects but sparing unwanted effects in the GI tract and possibly in other organ systems. This dichotomous picture of constitutive COX-1 and inducible COX-2 is, however, not consistently followed in all organs. Constitutive or inflammation-independent expression of COX-2 has, for example, been observed in endothelial cells and in the kidney. Inhibition of constitutive COX-2 expression at these sites may contribute to the undesired actions of NSAIDs and coxibs, such as arterial hypertension ( ) and increased cardiovascular risk ( , ).

Cloning and subsequent heterologous expression of both enzymes allowed high-throughput screening of large compound libraries in the quest for selective inhibitors. Subsequent testing of established COX inhibitors on recombinant enzymes, as well as in isoform-specific ex vivo assays, revealed that most existing drugs were non-specific inhibitors of both enzymes ( ), but some experimental compounds proved to be selective ( , ).

Heterologous expression of both enzymes also paved the path for crystallographic analysis of both COXs. These studies provided important insight into the structural basis of their enzymatic activity ( ) and their interaction with inhibitors ( , ). It was found that COX-1 and COX-2 insert as homodimers into lipid membranes ( ) and form a hydrophobic channel that allows arachidonic acid to reach the catalytic center of the enzymes ( ). Most known COX inhibitors inhibit the catalytic site or the channel in a competitive manner. Only aspirin/acetylsalicylic acid blocks this access irreversibly through covalent acetylation of a serine residue (S530) close to the catalytic center ( ). This irreversible action forms the basis for its unique long-lasting inhibitory action on platelet aggregation. The availability of structural information also permitted structure-based rational drug design approaches. It was discovered that differences in the structure of COX-1 and COX-2 were sufficiently large to enable selective inhibition of COX-2 ( ). A side pocket present in the channel formed by the COX-2 enzyme but absent from COX-1 turned out to be particularly relevant for the design of subtype-selective blockers. A third COX isoform, called COX-3 ( ), is not encoded by a separate gene but results from alternative splicing of the COX-1 gene and retention of intron 1. COX-3 mRNA has, however, been detected only in dogs and is apparently not present in humans ( ).

Although the role of COX inhibition and diminished prostanoid formation is generally accepted as the predominant mechanism of action of NSAIDs and coxibs, their role is much less clear in acetaminophen/paracetamol, dipyrone/metamizole, and related compounds. These drugs show only limited inhibition of COXs in vitro and largely lack anti-inflammatory action. They may, however, still inhibit COXs in vivo (e.g., through active metabolites). Recent evidence from ex vivo studies indeed suggests that acetaminophen/paracetamol possesses inhibitory actions on COX-2 in humans comparable even to those of coxibs ( ). The combination of antipyretic and analgesic actions of acetaminophen/paracetamol is clearly consistent with an inhibitory effect of acetaminophen/paracetamol on prostaglandin formation. Its non-acidic (chemically neutral) nature may facilitate permeation of acetaminophen/paracetamol across the blood–brain barrier into the central nervous system (CNS), where an inhibitory action on COX-2–dependent production of prostaglandins would exert analgesic and antipyretic effects. The apparent lack of a clear anti-inflammatory action of acetaminophen/paracetamol may, on the other hand, be ascribed to the absence of their enrichment in inflamed tissue (see also below) and to the presence of higher arachidonic acid or peroxide concentrations in inflamed tissue, which would render competitive inhibition less effective.

Cyclooxygenase Products as Mediators of Hyperalgesia

Prostaglandins, in particular PGE ₂ and PGI ₂ (also called prostacyclin), increase the sensitivity of peripheral nociceptor endings to noxious stimuli ( Fig. 32-3 A). They facilitate activation of the receptors and ion channels involved in the nociceptive transduction process, such as the transient potential channel TRPV1 ( ) and certain voltage-gated Na ⁺ channels, in particular, Nav1.8 ( ). These actions occur through G protein–coupled receptors specific for the different subtypes of prostaglandins and through the subsequent activation of different protein kinases.

Application of PGE ₂ to isolated dorsal root ganglion (DRG) cells ( ) or to HEK 293 cells transfected with TRPV1 ( ) dramatically increases current responses to capsaicin and heat, two prototypical activators of TRPV1. This phenomenon is likely to underlie the thermal hyperalgesia in inflammatory disease states ( , ). It is tempting to speculate that a leftward shift in TRPV1 activation by several degrees could render TRPV1 susceptible to activation by physiological temperatures. Such a process could contribute to the generation of spontaneous pain.

Tetrodotoxin-resistant Na ⁺ channels, especially Na _v 1.8, represent another potential substrate for peripheral nociceptive sensitization by prostaglandins. These channels become more readily activated in the presence of a number of inflammatory mediators, including PGE ₂ ( , ). In particular, Na _v 1.8 is selectively expressed in nociceptive small- and medium-sized DRG neurons ( ). Modulation of these Na ⁺ channels involves activation of adenylyl cyclase and increases in cyclic adenosine monophosphate, which possibly leads to protein kinase A–dependent phosphorylation of the channels. Via this mechanism, the prostaglandins produced during inflammatory responses may significantly increase the excitability of nociceptive nerve fibers and also contribute to the recruitment of nociceptors. A significant proportion of nociceptors in healthy (uninflamed and uninjured) tissue are mechanically insensitive and are not activated even by strong stimuli ( , ). Following tissue trauma and the release of prostaglandins, these silent nociceptors become excitable to pressure, changes in temperature, and tissue acidosis ( ), which contributes to the generation and maintenance of hyperalgesia ( ).

The sensitizing actions of prostaglandins are not restricted to peripheral nociceptor endings but also occur at spinal sites ( Fig. 32-3 B). Intrathecal injection of PGE ₂ causes sensitization to thermal and mechanical stimuli. Expression of COX-2 ( , ) and inducible prostaglandin E synthase (mPGES1) ( , ) increases in the CNS in response to peripheral inflammation. In rodents, hyperalgesia can be partially prevented or reversed by intrathecally injected NSAIDs ( ) or coxibs ( ), an action that is accompanied by a decrease in spinal PGE ₂ concentrations ( ). Different cellular mechanisms of action have been proposed for the pro-nociceptive effects of PGE ₂ in the spinal cord. have shown that PGE ₂ can directly depolarize deep dorsal horn neurons, and demonstrated that PGE ₂ reduces inhibitory glycinergic neurotransmission through a post-synaptic mechanism. This latter action is triggered by activation of PGE ₂ receptors of the EP2 subtype ( ). Activation of these receptors leads to the protein kinase A–dependent phosphorylation and inhibition of a certain isoform of glycine receptors that contain the α3 subunit (GlyRα3) ( ). This subunit is expressed in the superficial layers of the spinal dorsal horn, where nociceptive afferents terminate and where glycinergic neurotransmission is inhibited by PGE ₂ . Evidence of the in vivo relevance of this centrally mediated form of inflammatory pain sensitization comes from several studies using mice carrying targeted mutations in genes critically involved in these pathways. Mice lacking neuronal protein kinase A showed less thermal hyperalgesia after injection of PGE ₂ into the spinal canal ( ). Specific disruption of COX-2 in neural cells (through the use of nestin-driven cre expression) protects mice from mechanical hyperalgesia after peripheral inflammation ( ). Mice deficient in either GlyRα3 or the EP2 receptor not only lack PGE ₂ -mediated inhibition of glycinergic neurotransmission in the spinal cord dorsal horn but also largely lack the pro-nociceptive effects of spinal PGE ₂ ( , ). Subsequent work has studied the phenotypes of GlyRα3-deficient mice in more detail. These studies showed that disruption of the GlyRα3 gene did not affect nociceptive responses in the formalin test and after peripheral nerve lesions ( ), which is consistent with the specific involvement of prostaglandins in inflammatory pain. Similar unchanged nociceptive responses were also reported for carrageenan-induced cutaneous inflammation and kaolin/carrageenan-induced arthritis or iodoacetate osteoarthritis ( ).

The contribution of a spinal site of action of COX inhibitors in humans has recently been addressed in two studies that tested the action of intrathecally injected ketorolac against experimental pain in human volunteers ( ) and postoperative and chronic pain in patients ( ). Intrathecal ketorolac was well tolerated, but its analgesic effects were much less than expected based on preclinical data from animal experiments. It showed some efficacy against sunburn-induced hyperalgesia in volunteers but did not induce significant analgesia against postoperative pain. Only in a subset of chronic pain patients whose spinal prostaglandin levels were increased most significantly did intrathecal ketorolac reduce spinal PGE ₂ levels and show analgesic efficacy.

Diminished Prostaglandin Production As a Source of Other Desired Effects of Cyclooxygenase Inhibitors

Diminished formation of prostanoids can also explain most of the anti-inflammatory and antipyretic actions of antipyretic analgesics. Edema formation and plasma extravasation are promoted in particular by PGI ₂ ( ). Antipyresis originates from reduced activation of EP3 receptors in the hypothalamus ( ). There is good evidence that the pro-pyretic PGE ₂ does not originate from neural sources as the pro-nociceptive PGE ₂ does but instead originates from endothelial cells that reside in the walls of cerebral blood vessels and produce PGE ₂ through COX-2 ( ). Inhibition of platelet aggregation may also be considered a desired effect of antipyretic analgesics. It occurs through reduced formation of COX-1–dependent thromboxane in platelets. Inhibition of platelet aggregation to a degree relevant for the prevention of arterial thrombosis is primarily seen with low doses of aspirin/acetylsalicylic acid. Inhibition by other reversible COX inhibitors is probably too transient to exert clinically relevant therapeutic effects ( ). With respect to cardiovascular risks, increases in blood pressure and the reduced anti-aggregatory and vasodilating effects of PGI ₂ probably counteract the transient beneficial effects from transient thromboxane inhibition by classic NSAIDs ( ). The transient nature of inhibition of platelet aggregation does, however, not exclude increased bleeding risk in predisposed patients receiving NSAIDs or dipyrone/metamizole.

Undesired Effects of Cyclooxygenase Inhibitors

Classic side effects of COX inhibitors include ulcers of the upper GI tract, asthma, increases in blood pressure, kidney damage, and bleeding. Inhibition of constitutive prostaglandin formation in the GI tract and in the kidneys is a major cause of the undesired effects of the classic NSAIDs, in particular, GI tract toxicity and renal impairment. PGE ₂ protects the gastric mucosa by reducing acid and promoting mucous secretion via activation of EP1 and EP3 receptors ( , , ). The GI tract toxicity of NSAIDs is probably aggravated by accumulation of these acidic compounds in mucosal cells through a process called ion trapping (see also below), and part of the mucosal damage evoked by orally administered NSAIDs may also be derived from COX-independent toxicity, which is supported by the observation that gastric ulcers still develop in COX-1–deficient mice after oral administration of indomethacin ( ). Asthma attacks induced by aspirin/acetylsalicylic acid and by NSAIDs have for a long time been attributed to the so-called leukotriene shift—a shift of arachidonic acid metabolism from COX-mediated production of prostaglandins to increased lipoxygenase-dependent leukotriene production. More recent evidence suggests that under conditions of airway inflammation, certain prostaglandins, in particular, PGD ₂ and PGE ₂ , exert a protective function by down-modulating the immune responses ( ) or by dilating the airways ( ). Blockade of their formation would then cause acute bronchoconstriction. In the kidney, prostaglandins regulate perfusion, filtration, and salt reuptake, actions that contribute to the significant increases in arterial blood pressure that are already seen after short-term use of classic NSAIDs. Long-term use or abuse of NSAIDs has been the leading cause of terminal renal failure for several decades.

In addition to these classic side effects, the potential cardiovascular toxicity of coxibs and NSAIDs has gained significant attention after the withdrawal of rofecoxib from the market because of its association with increased cardiovascular risk ( , ). Initially, it was thought that the increased cardiovascular toxicity would be a problem specific to coxibs, most likely caused by a coxib-induced imbalance of protective (PGI ₂ ) and potentially harmful (thromboxane) mediators. There is now widespread consensus that the increase in cardiovascular risk results from inhibition of PGI ₂ formation and occurs independent of the presence or absence of additional blockade of thromboxane formation ( ).