Mixed messages: modulation of inflammation and immune responses by prostaglandins and thromboxanes (original) (raw)

Pharmacological agents that inhibit COX activity have been used for over 20 years to identify the role of prostanoids in immune responses. More recently, this approach has been complemented by the development of pharmacological inhibitors with specificity for the individual COX isoenzymes and the generation of mice with targeted disruptions of the genes encoding COX-1 or COX-2. Since these two enzymes act at the first step in the production of all prostanoids, these studies do not identify the specific prostanoid or the receptor pathway involved. While agonists and antagonists for some of the prostanoid receptors and synthase enzymes have been developed, many of these do not have the requisite affinity or specificities necessary for physiological experiments. The development of mice deficient in each of the prostanoid receptors has therefore provided an important approach for pinpointing the contributions of individual prostanoids and their receptors in inflammation and immune responses.

Acute inflammation. Acute inflammation is the earliest response to tissue injury, infection, or immunological challenge. This physiologic process involves a coordinated response between the immune system and the tissue in which injury has occurred. Prostanoids were implicated in these processes in the 1970s when it was demonstrated that the pharmacologic inhibition of COX by aspirin and other nonsteroidal anti-inflammatory drugs (NSAIDs) attenuates acute inflammation and that injection of prostanoids into an organism can potentiate many of the signs of inflammation induced by bradykinin and histamine. However, as conventional NSAIDs inhibit both COX isoforms, it was not possible to identify the relative contribution of each COX isoform in these responses. In this regard, mice deficient in COX-1 and COX-2 have been useful for defining the role of COX isoforms in various inflammatory models (Table 1; reviewed in ref. 30). Both COX-1 and COX-2 have been implicated in models of acute inflammation, and it appears that the degree to which each COX isoform contributes depends upon the inflammatory stimulus, the time point examined, and the tissue or organ in which the insult occurs, among other factors.

Application of AA to the surface of the skin elicits an inflammatory response that is entirely dependent on the production of leukotrienes and prostanoids, mimicking the early events in acute inflammation. Because of the substantial contribution of leukotrienes in the AA model, the use of mice deficient in 5-lipoxygenase (5-LO) that are unable to produce leukotrienes provides a means to examine the contribution of prostanoids in isolation. We have found that edema is reduced substantially when 5-LO–deficient mice are pretreated with the nonselective COX inhibitor indomethacin. Studies in mice with combined 5-LO/COX-1 or 5-LO/COX-2 deficiencies show that COX-1 is responsible for the prostanoid component of edema formation in this model (B.H. Koller, unpublished results). These data suggest that the initial inflammatory response to an insult that causes rapid release of AA is mediated by COX-1. The findings in this functional assay are consistent with the constitutive expression of high levels of COX-1 in the skin and the observation that early production of prostanoids by activated mast cells depends primarily on COX-1 (31). However, the demonstration of reduced inflammation in other models, using both COX-2–deficient mice and COX-2 inhibitors, suggests that the prostanoids produced after the rapid induction of COX-2 also contribute to acute inflammatory responses (Table 1).

The availability of prostanoid receptor–deficient mice has facilitated the dissection of prostanoid-dependent pathways leading to the various components of the acute inflammatory response. Vasodilation and increased permeability of postcapillary venules, early events in the inflammatory response, reflect the significant effects of PGs on vascular tone at sites of inflammation. Both PGE2 and PGI2 are potent vasodilators in animals and humans and are produced in sufficient quantities at inflammatory sites to account for the characteristic erythema (rubor) of acute inflammation. Both PGE2 and the IP receptor agonist cicaprost produce systemic hypotension when infused intravenously into wild-type mice. In mice deficient in the IP receptor, blood pressure does not change during cicaprost infusion, confirming the role of the IP receptor activation in vasodilation and providing evidence that a single IP receptor is responsible for the action of PGI2 on vascular smooth muscle (32). Because of the existence of multiple EP receptors with different coupling mechanisms, identification of the pathways through which PGE2 mediates changes in blood flow has been more difficult. Initial studies using EP-deficient mice suggested a major role for the EP2 receptor, at least in the systemic circulation (33). However, these studies identified additional complexity in these vascular responses since the relative contribution of individual EP receptors differed substantially between males and females. The specific EP receptors that mediate PGE2-induced vasodilation in the microcirculation and at sites of inflammation remain undefined.

Along with prostanoids, several other mediators, including histamine, bradykinin, and leukotrienes, influence vascular permeability in models of acute inflammation. In fact, when administered alone, prostanoids produce only small changes in vascular permeability. However, PGE2 and PGI2 can substantially potentiate the effects of bradykinin and histamine on edema formation. In one model of acute inflammation, Murata and colleagues showed that the prostanoid contribution to carrageenan-induced paw edema was entirely due to PGI2 acting through the IP receptor (32). In contrast, studies with all four EP-deficient mouse lines in the AA model have suggested that these actions are mediated by PGE2 via activation of the EP3 receptor (B.H. Koller, unpublished results). The relevant EP3-expressing target cells have not been identified.

While prostanoids appear to promote acute inflammation in the majority of models, important exceptions have been observed. PGE2 has been reported to attenuate some acute inflammatory responses, in particular those initiated by mast cell degranulation. Raud and colleagues have shown that COX inhibition with indomethacin markedly potentiates antigen-induced plasma protein extravasation and leukocyte accumulation in sensitized hamsters (34). Moreover, PGE2 completely reversed the effects of indomethacin and reduced histamine release and plasma leakage in the absence of indomethacin. These results suggest that PGE2 can inhibit certain mast cell functions and are consistent with other reports showing enhanced histamine release from mast cells by indomethacin or suppression by E-type PGs (35).

There is also evidence to support a role for prostanoids in the resolution of inflammatory responses. Gilroy and colleagues showed in the rat carrageenan-induced pleurisy model that treatment with NSAIDs reduced the number of inflammatory cells and the formation of exudates present at 2 hours. However, by 48 hours, when inflammation had largely resolved in the controls, the number of inflammatory cells had increased in the indomethacin-treated group (1). Similar findings have been observed in the air pouch model, where resolution of inflammation occurs more slowly in COX-2–deficient animals than in COX-1–deficient or wild-type controls (30).

Prostanoids also play a key role in pain associated with acute inflammation, and inhibition of these pathways has been an important therapeutic application for NSAIDs. Experiments with prostanoid receptor–deficient mice have provided new insights into the signaling pathways that mediate inflammatory pain, but these studies have also uncovered unexpected complexity in the mechanisms involved. Studies by Murata and colleagues showed that inflammatory pain responses are significantly reduced in mice lacking prostacyclin (IP) receptors, suggesting that IP receptors play a major role in the prostanoid-dependent component of pain (32). While these findings were somewhat unexpected, they are consistent with the previous demonstration of abundant expression of the IP receptor in neurons of the dorsal root ganglion (36). More recently, Stock and associates reported reduced responses to inflammatory pain in EP1-deficient mice (37). This inhibition in pain was virtually identical to that achieved through pharmacological interruption of PG synthesis in wild-type mice using an NSAID and was very similar to the magnitude of pain reduction seen in the IP-deficient mice. While the reasons for these apparently conflicting results is not clear, the differences in genetic background between the IP- and EP1-deficient mouse lines is one factor that may contribute. Alternatively, it is possible that signals from both the EP1 and IP receptors contribute to inflammatory pain and that the absence of either receptor is sufficient to attenuate the response. The observation that IP and EP1 receptors are coexpressed in dorsal root ganglia is consistent with this possibility (36). These data implicate EP1 and IP receptors as potential therapeutic targets for inflammatory pain.

Fever is another component of acute inflammation in which prostanoids may play a causal role. It is well established that COX inhibition reduces the fever associated with acute inflammation. Recent studies have shown that the febrile response to LPS is ameliorated in COX-2– but not COX-1–deficient mice, findings consistent with studies suggesting that COX-2 plays a dominant role in fever production (38). A number of lines of evidence suggest that the prostanoid responsible for fever production is PGE2, and studies using EP-deficient mice have shown that the febrile response to PGE2, IL-1β, and LPS occurs through the action of PGE2 on the EP3 receptor (39).

Chronic inflammation. High levels of prostanoids are observed in chronic inflammatory lesions, and several experimental approaches have been used to identify a role for these mediators in allergic rhinitis, asthma, rheumatoid arthritis, and inflammatory bowl disease. The mechanisms by which prostanoids can modulate these responses and contribute to disease do not differ in principle from those of an acute response: they can alter the response of both the host tissue and the recruited inflammatory cells. Prostanoids have been proposed to act on immune effector cells at any of several levels, and there is evidence that they can modulate the development, function, and survival of these cells. Again, the cells of the inflamed tissue or organ can also produce and respond to the prostanoids, providing an additional mechanism by which these lipid mediators may affect the course of the chronic inflammatory response. For example, prostanoids can alter the production of cytokines by epithelial cells and alter the expression of class II MHC antigens by antigen-presenting cells, thus modulating the course and resolution of the response (40).

Animal models of disease. Mice deficient in COX-isoenzymes and prostanoid receptors have been evaluated in several animal models of disease (Table 1). In a collagen-induced arthritis model, mice deficient in COX-2 display significant reductions in synovial inflammation and joint destruction, whereas arthritis in COX-1–deficient mice is indistinguishable from controls (41). These data strongly suggest that prostanoids generated by COX-2 promote inflammation in this model and are consistent with the beneficial effects of COX-2 inhibitors in arthritis.

In contrast, in models of inflammatory colitis and allergic airway disease, an overall anti-inflammatory role for prostanoids has been suggested. Morteau and colleagues found that colitis induced by low-dose dextran sodium sulfate (DSS) is more severe in COX-2– and, to a lesser extent, COX-1–deficient animals than in wild-type mice (42). Moreover, high-dose DSS is fatal in 50% of mice lacking either COX isoenzyme, whereas none of the control mice died following treatment. These findings are consistent with the reported exacerbation of inflammatory bowel disease in patients receiving NSAIDs.

Gavett and colleagues studied the inflammatory response of COX-1– and COX-2–deficient mice in the ovalbumin-induced asthma model (43). Lung inflammation, airway reactivity, and IgE levels were significantly enhanced in COX-1– and, to a lesser extent, COX-2–deficient mice compared with wild-type controls. As expected, PGE2 levels were significantly reduced in COX-deficient mice and were substantially lower in COX-1– than in COX-2–deficient animals. Peebles and colleagues obtained similar results by treating mice with indomethacin; mice treated with this drug developed increased lung inflammation and tended to have higher serum IgE levels (44). Moreover, levels of IL-5 and IL-13 were significantly higher in mice treated with indomethacin. These data suggest that prostanoids can inhibit the development of the Th2 response in vivo, which is considered central to the pathogenesis of allergic inflammation. However, since leukotriene levels are higher in COX-deficient and indomethacin-treated mice than in controls, it remains undetermined whether increased inflammation might be secondary to shunting of AA toward the lipoxygenase pathway rather than direct inhibition of prostanoid synthesis.

Defining the contribution of individual prostanoids in complex inflammatory diseases has been aided by the development of mice lacking individual prostanoid receptors. In one such study, Matsuoka and colleagues examined the contribution of the Gs-coupled DP receptor in the ovalbumin-induced asthma model (45). In contrast to COX-deficient animals, mice lacking the DP receptor show reduced disease. This finding is consistent with the well-established proinflammatory role of PGD2 in asthma, but the mechanism by which PGD2 mediates its proinflammatory actions is not clear. Since DP receptor activation in most cells examined to date leads to increased intracellular cAMP levels, PGD2 would be expected to attenuate the function of immune cells. Identification of a second receptor for PGD2 (CRTH2) that can couple to Gi may, in part, explain the proinflammatory actions of this lipid mediator (8). Taken together, the results from COX-deficient and DP-deficient mice support a model in which both pro- and anti-inflammatory prostanoids are produced during chronic inflammation. This single example perhaps best illustrates the challenge awaiting us as we continue to dissect the role of individual lipid mediators in vivo.