Platelet-Activating Factor Peter M. Henson* Program in Cell Biology, National Jewish Medical Research Center, 1400 Jackson Street, Denver, CO 80206, USA *corresponding author tel: 303 398 1380, fax: 303 398 1381, e-mail: [email protected] DOI: 10.1006/rwcy.2000.12005. SUMMARY StafforiniandTjoelker,1999)andanumberofbooks (Handley,1990;O’FlahertyandRamwell,1990;Bailey, Platelet-activating factor (PAF) is a phospholipid 1991;Braquet,1991;Shukla,1992;Anonymous,1994; mediator with a very broad spectrum of potential Cunningham, 1994; Hanahan, 1996; Nigam et al., biologic activities. Although early studies focused on 1996) published in the 1990s specifically on the itslikelyroleinacute,IgE-mediatedallergicreactions subject of PAF, as well as proceedings from the six or as a naturally occurring antihypertensive agent, it triennial PAF and Related Ether Lipid International has become apparent that its production and effects Meetings. Herein, the focus will be predominantly on extend way beyond these arenas, from inflammation major points and concepts within the PAF field. to parturition and development to actions in the References will not be exhaustive but rather selected nervous system. In actuality a family of related phos- to give access to appropriate information and leads pholipids with varying potency, or molecules with for more indepth consideration. For this I must similar activity, may be produced by enzymatic or apologize in advance to my friends and colleagues nonenzymatic means. As with all potent mediators whose efforts in this field may not have been there exist equally potent control mechanisms which extensively described. includehererapidreceptordesensitizationandspecific Essentially it will be suggested that, as mediators, inactivating enzymes. PAF reacts with a seven trans- PAF and related molecules most likely play a role in membrane, heterotrimeric G protein-linked mem- many pathophysiologic processes, but largely as brane receptor for which a knockout mouse has cofactors; as but one strand in a complex mediator recently been created. A huge number of receptor web. Antagonism or abrogation of its effects will antagonistshavebeensynthesizedandmanyarebeing therefore diminish, alter or bias these processes but exploredinvitro,inanimalmodelsandinavarietyof seldom result in their complete elimination. A PAF- clinical trials. In addition, the complexity of the PAF blockingdrugmaybeusefulinspecialcircumstances, story is increased by the observation that in the would probably be really good as a component of a majority of cells that synthesize it, most of the PAF multiple therapy approach but will not itself be the remains within the cell, leading to speculation that it magic bullet. In addition, it is becoming increasingly plays an intracellular role, perhaps even on an intra- clearthatthesemoleculescannolongerbeconsidered cellularreceptor,inadditiontoitsroleasextracellular solely as mediators acting through classical G communication molecule. protein-linked receptors on responsive target cells. Theliteratureonthesubjectisconsiderableandthe ThePAFfieldmustalsobeconsideredinconjunction interested reader is referred to recent reviews (Boulay withcellularresponsestooxidativestress;inparticular et al., 1997; Feuerstein et al., 1997; Howarth, 1997; to the generation, removal, and effects of oxidized Kulikov and Muzya, 1997, 1998; Loucks et al., 1997; phospholipids.Itseemsquiteplausibletosuggestthat Mathiak et al., 1997; Muguruma et al., 1997; Negro in evolution, extensive mechanisms (e.g. specific Alvarez et al., 1997; Prescott, 1997; Shimizu and remodeling and inactivating enzymes) developed ini- Mutoh,1997;Silversteinetal.,1997;Tettaetal.,1997; tially to deal with oxidized phospholipids and that Yamashita et al., 1997; Bazan, 1998; Derewenda and receptor-mediated responses to these molecules Derewenda, 1998; Heller et al., 1998; Itabe, 1998; appeared secondarily, along with elements of the 1336 Peter M. Henson inflammatory response, as an amplifying pathway to Figure 1 Structure of PAF. recognize and respond to tissue injury. BACKGROUND Discovery PAFwasfirstidentifiedinthelate1960sasanactivity derived from leukocytes that led to platelet stimula- tion. A separate renal antihypertensive agent was de- scribed in rats – the two were eventually shown to be identical as their structures were elucidated in 1979. Alternative names Over the years the activity has been called platelet- activatingfactor,PAF,paf-acetherorAGEPC(acetyl glycerol ether phosphoryl choline) and antihyperten- sive renal phospholipid. At this point, in spite of its action on a wide variety of cells, PAF or platelet- activating factor seems to be the name in most common usage. Structure glycerol backbone have indicated that all three of these positions contribute to the biologic activity Classical PAF, i.e. that produced enzymatically upon (Vargaftig et al., 1989) and led to models of receptor stimulation of leukocytes, is mostly a mixture of binding involving a need to fit optimal structures at hexadecyl- and octadecyl-acetylphosphocholine (i.e. each position, even before the receptor was cloned 1-O-alkyl-2-acetyl sn-glycero-3-phosphocholines) (Dive et al., 1989). (Figure 1). Two major routes for enzymatic synthesis have Each portion of the molecule is structurally critical been described (Snyder, 1994, 1995) – the so-called for its overall biologic activity. All the PAFs and remodeling and de novo pathways (Figure 2). related molecules require choline as the polar head In the former, and most important within inflam- group. Optimal biologic activity is seen with an ether matory cells, the sn2 fatty acid (usually arachidonyl) linkage in the sn1 position of the phospholipid and of membrane alkylacylglycerophosphocholine is with a 16- or 18-carbon alkyl chain. On the other removed by phospholipase A (PLA ), thereby 2 2 hand, diacyl phosphatidylcholine is much more liberating arachidonic acid for reincorporation or abundant in the cell than the alkyl, acyl species and, metabolism down the various pathways to generate even though PAF with an sn1 acyl linkage is less eicosanoids. The remaining lysophosphatide, alkyl activebiologically,insomecircumstancesitspotential lyso GPC, also called lyso-PAF, is the substrate for abundance may make it relevant. The native phos- PAF synthesis. The most important PLA for PAF 2 phatidylcholine from which most of the PAF is syn- synthesis is likely the cytosolic, arachidonyl-specific, thesized usually has a long-chain, unsaturated, calcium-requiring cPLA but other PLA s may 2 2 ester-linked fatty acid in the sn2 position. The critical participate. Markedly diminished PAF synthesis is step in this synthesis pathway, then, is a chain seen in cells from cPLA knockout mice (Uozumi 2 shortening of the group in the sn2 position. Optimal et al., 1997), thus supporting an important role for biological activity is seen with a two-carbon chain this enzyme. The next step in synthesis is the transfer (acetyl) moiety and, as discussed below, the key of an acetyl group from acetyl-CoA to the sn2 enzyme for PAF synthesis is an acetyltransferase position of the lyso-PAF by action of a specific which generates just this structure. Increasing the enzyme lyso-PAF acetyltransferase (Lee et al., 1992; sn2 chain length decreases the activity. In fact, Kume et al., 1997). This is found in the endoplasmic structural modifications at each position of the reticulum, which in many inflammatory cells is Platelet-Activating Factor 1337 Figure 2 Pathways for PAF synthesis. Figure 3 Hypothetical scheme for mechanisms of PAF release. CoA-independent transacylase cPLA2 viaPLA2 and transacylase PAF acetylhydrolase Lyso-PAF acetyltransferase Acetyltransferase Phosphohydrolase The de novo pathway involves transfer of CDP- Cholinephosphotransferase choline to alkylacetylglycerophosphate, previously formed from acetylation of alkyl lyso-glyceropho- sphate by an alkylglycerophosphate (AGP) acetyl- transferase that is different from that involved in the remodelingpathway(Snyder,1994).Currentthinking continuous with the nuclear membrane (Baker and suggests its involvement in the endogenous PAF Chang, 1997). Importantly, the cPLA has a similar found in many cell types, for example, spermatozoa 2 cellular distribution suggesting a contiguous relation- (Muguruma and Johnston, 1997). The rapid stimu- ship between the two steps in PAF synthesis lated production of PAF necessary for it to act as a analogous with a similar co-localization between the mediator is believed to result from activation of the translocated cPLA and cyclooxygenase of the pro- remodeling pathway. 2 stanoid pathway or 5-LO of the leukotriene path- Intracellular localization of the biosynthetic enzy- ways. Whether the enzymes are physically associated mes suggests that most PAF is produced at the endo- with each other or merely present in the same plasmic reticulum or nuclear membranes (Bratton membrane is not yet clear. Like cPLA , the acetyl- et al., 1994a). For it to act as a mediator, then, it 2 transferase requires calcium (Ninio et al., 1983) and must transit the cytoplasm and plasma membrane phosphorylation (Domenech et al., 1987; Holland (Figure 3). et al., 1992) for activation, suggesting multiple pot- Acarrierproteinforintracellularmobilityhasbeen ential regulatory points in PAF biosynthesis. An suggested (Lumb et al., 1983; Banks et al., 1988) but alternative remodeling process may involve transfer not yet completely characterized. On the other hand, of acetate to lyso-PAF by action of a membrane its shortened sn2 group renders PAF more hydro- transacetylase that may be particularly important in philic than its precursor, so diffusion might also the generation of acyl-PAF (Karasawa et al., 1999). contribute to this step. Passage across the plasma Finally, an additional supply of lyso-PAF can come membrane appears to be mediated by the action of a from the sequential action of PLA on phosphatidyl phospholipid scramblase (Bratton et al., 1994b). A 2 ethanolamine and action of CoA-independent candidate scramblase molecule (or family of mole- transacylase exchanging sn2 arachidonyl groups cules)hasrecentlybeencloned(Zhouetal.,1997)and between choline and ethanolamine phospholipids seems to require calcium and PKC-dependent phos- (Snyder, 1994). phorylationforoptimalactivity(Brattonetal.,1997). 1338 Peter M. Henson The scramblase mediates general bidirectional trans- although in any cell the proportion released is likely locationofphospholipids(irrespectiveofheadgroup) to vary on the basis of many factors, including those across the membrane. The potential importance for described above. Importantly, even if only a small PAF is that, while there are other mechanisms for percentage of the total synthesized PAF is released, translocation of, for example, aminophospholipids, the high potency of the molecule as a mediator may phosphatidylcholines are not as readily passed meanthatthissmallamountis,nevertheless,plentyto across the membrane, even when the sn2 position is activate nearby cells. shortened. Addition of PAF to the outside of cells leads to uptake by the same nonreceptor-driven pro- Main activities and cess and, interestingly, results in accumulation at the nuclear membrane (Chilton, Murphy and Henson, pathophysiological roles unpublished observations), suggesting an exact rever- sal of the presumed secretory pathway. Generation PAF has a very broad spectrum of potential biologic of extracellular PAF for cell stimulation requires at activities. It was first implicated in acute IgE-medi- least three overall steps involving cPLA , acetyl- 2 ated allergic reactions or as a naturally occurring transferase,andphospholipidscramblase;eachstepis antihypertensiveagent,butnowhasbeensuggestedto activatable, but by different pathways. participate in multiple systems from inflammation to A further issue for action as a mediator, i.e. pre- parturition and development to actions on blood sentation to other cells, is that even after transloca- vessel function and development and in the nervous tion from inner to outer membrane leaflet during system. secretion, PAF, as a phospholipid, partitions preferentially into the outer membrane rather than into the extracellular medium unless there is a carrier present. The molecule binds readily to albumin (Clay CELLULAR SOURCES AND et al., 1990) which probably represents the major TISSUE EXPRESSION extracellular carrier. Thus, albumin was early shown to be necessary for optimal release of PAF from Cellular sources that produce generatingcells(Benvenisteetal.,1972).Infact,ithas been argued (although by no means generally accepted: Grigoriadis and Stewart, 1992) that PAF Many different mammalian cell types have been can only be effectively presented to responding cell shown to be capable of synthesizing PAF via the receptors on a carrier such as albumin (Clay et al., remodeling pathway in vitro (Bratton et al., 1994a). 1990).It is also possible that PAFin the extracellular Not included in that review are cells in the neuronal membrane leaflet of a synthesizing cell can in some system (Bazan et al., 1997) or epithelial cells of the fashion be recognized by a receptor on a closely skin (Mallet and Cunningham, 1985; Travers et al., adherent target cell. Thus, in a potentially important 1997).Thecriticalquestions,therefore,arewhenthey role in inflammation, endothelial cell PAF contri- do so in vivo, is the PAF released and, if so, in butes to activation and increased adherence of concentrations to be biologically active, does it play leukocytes (Zimmerman et al., 1990). Given the an intracellular role in these cells? An additional need for all three portions of the PAF molecule for synthesis pathway may involve cooperative action of high-affinity binding to PAF receptors, it is hard to more than one cell type, i.e. a transcellular biosyn- seehowthetarget cellreceptor canbindefficientlyto thetic pathway akin to that described in eicosanoid PAFwithinamembrane.PossiblePAFcanbind,and production. Here, one cell makes the lyso-PAF for be presented by, proteins on the endothelial cell export and subsequent uptake and acetylation by surface, although no evidence of such has yet been another (Bratton et al., 1994a). Whether this too found. Alternatively, the molecule may be picked up occurs in vivo is also unknown. nonspecifically by the target cell membrane with its PAF and related lipids have been detected in bio- receptorcapableof‘seeing’aPAFmoleculeinitsown logic fluids from all parts of the body, including membrane (see section on Receptor utilization). plasma,amnioticfluid,surfactant,cerebrospinalfluid, Possibly as a consequence ofregulation at the level urine, and saliva. In most cases the cells of origin are of the secretory pathway, in most cells a large unknown. In addition, the assay techniques used proportion of the newly synthesized PAF is not for detection may prejudice the interpretation. The released to the outside (Bratton et al., 1994a). marked ability of inflammatory cells to synthesize Exceptions to this may include the cell from which PAFandpresentitforreleasewouldcertainlyleadto the molecule was originally described, the basophil, expectations of PAF presence at the site of any Platelet-Activating Factor 1339 inflammatory reaction, whether or not it was playing (Lehr et al., 1997). (For a general review of oxidized animportantroleintheongoingprocess.Anexample phospholipids, see Itabe, 1998.) here is the observation that PAF is present in saliva All in all, and as indicated above, PAF is more an (Noguchi et al., 1989; Woodard et al., 1995). In fact, operational term for a family of phospholipids with it appears in this fluid because of the constant emi- underlying similarities in structure, generated by gration of neutrophils through the gingival crevice. enzymatic or nonenzymatic pathways, acting on a Whether it plays a role in neutrophil function in this common receptor and varying in efficacy as a normal process is unclear; presumably it may contri- function of structural differences presumably reflect- bute to the enhanced responses in gingivitis. ing different affinities for this receptor. Alkylacylglycerophosphocholines are found widely in nature from protozoa to mammals and even yeast, Eliciting and inhibitory stimuli, other fungi, and plants (Kulikov and Muzya, 1997; Nakayama et al., 1997). In many of these PAF syn- including exogenous and thesis has been detected by mass spectrometry endogenous modulators (Nakayama et al., 1994; Sugiura et al., 1995; Calligerou et al., 1996; Tellis et al., 1996) and in some cases, in addition, biologic activity for deli- As with most potent mediators whose activity is berately added PAF. Whether this activity represents limited in time or scope in vivo, mechanisms exist for a likely occurrence in real life and whether the PAF inactivation or inhibition. The most potent of organisms express specific receptors is, for the most these are the PAF acetylhydrolases. These enzymes part, unclear. While eubacteria do not contain the exhibit PLA -like activity leadingtohydrolysisofthe 2 requisite precursors for PAF synthesis, they do seem sn2esterbond,therebyreversingtheacetyltransferase to contain acyltransferases that can synthesize the synthetic step and generating lyso-PAF and acetate. molecule if supplied with lyso-PAF (Kulikov and Twooftheenzymesalsoremoveothershortchainsn2 Muzya, 1997). The observation may have nothing to acyl groups, including those generated by phospho- do with bacterial physiology but may reflect the lipid oxidation. Since lyso-PAF has no activity importance of acyltransferase-induced acetylation of against the receptor, the result is potent inactivation, phospholipids as a critical housekeeping function of although the product could still be available for cells in general (see section on Intracellular PAF and reuptakeandreacetylationbacktoPAF.Intriguingly, oxidized phospholipids). ongoing study of the acetylhydrolases is generating Oxidized LDL or deliberately oxidized phosphati- new suggestions for in vivo activities and importance dylcholine have been shown to exhibit biologic activ- for PAF and related lipids. The three-dimensional ities similar to those of PAF (Lehr et al., 1993; structure and function of these enzymes are reviewed Liapikosetal.,1994;Heeryetal.,1995;Tanakaetal., in Derewenda and Derewenda (1998). 1995;Frostegardet al.,1997; Matsuzawaet al., 1997; Plasma PAF acetylhydrolase is a monomeric, Tsoukatos et al., 1997; Yamada and Yokota, 1998). 43kDa enzyme with significant specificity for short Recent structural analysis suggests that this results (no longer than nine methylene groups) chain sn2 because of nonenzymatic oxidant-induced chain fatty acids (Stafforini and Tjoelker, 1999) which is shorteningofthesn2unsaturatedfattyacid(Marathe also capable of inactivating the oxidatively generated et al., 1999) generating a family of molecules with material (Stremler et al., 1991). It is synthesized different sn2 groups but capable of acting like PAF andsecretedbymacrophages(andtoalesserextentin because of recognition by the PAF receptor. The sn1 theliver)and,asaconsequence,canlimittheamount group of the oxidized phospholipids may be ether or of PAF produced by these cells (Palmantier et al., ester-linked, with presumed differences in biologic 1989). In plasma it is found associated with lipopro- activity reflecting the differences between these seen teins. The molecule has been cloned (Tjoelker et al., with enzymatically derived PAF. Not surprisingly, 1995) and is transcriptionally upregulated as mono- such lipids are found in oxidized lipoproteins cytes mature into macrophages or stimulation by (Marathe et al., 1999). Many of these studies were proinflammatory mediators or PAF itself (Cao initiallycarriedoutinrelativelyartificialsystems,but et al., 1998). The macrophage enzyme may also be arenowbeingextendedintomorephysiologicoreven activated by PKC(cid:14)-mediated phosphorylation in in vivo conditions. For example, oxidized phospho- response to macrophage stimulation with oxidized lipids have been found in human atherosclerosis lipids, further suggesting important, self-limiting (Mallat et al., 1999) and smoking has been reported control processes (Claus et al., 1999). The enzyme is to generate biologically active, PAF-like phospho- inhibited by serine esterase inhibitors (e.g. DFP or lipids by a process that is blocked by antioxidants PMSF) and by PAF receptor antagonists with PAF 1340 Peter M. Henson analog structures (Adachi et al., 1997). A missense cellularresponsesthatareseenare,ingeneral,typical mutation in the plasma acetylhydrolase gene near the ofrapidmediatoractionviaseventransmembrane,G enzyme active site is seen in 4% of the Japanese protein-coupled receptors and include calcium mobi- population and, importantly, is associated with an lization, shape change (morphologic polarization), increasedincidenceofasthma(Stafforiniet al., 1999); actin polymerization, chemokinesis, and chemotaxis, another shows association with stroke (Hiramoto priming for NADPH oxidase activation and eicosa- et al., 1997). Experimentally, addition of this enzyme noid production, adhesion, and integrin regulation. has been shown to block PAF-induced paw edema Platelet responses are also relatively standard, for and pleurisy in vivo (Tjoelker et al., 1995) or the example, potent aggregation and some release induction of diabetes in rats (Lee et al., 1999). reaction. Interestingly, phosphatidylserine expression Two intracellular acetylhydrolases have been de- is not observed (Henson and Landes, 1976) and, as a scribed and cloned. Type Ib is a heterotrimeric consequence,procoagulanteffectsareminimal.There enzyme with similarities to G proteins (Ho et al., has been some suggestion that PAF is relatively 1997) and a specificity for acetyl in the sn2 position selective as a proadhesive and chemotactic agent for (Hattori et al., 1995). It may be involved in brain eosinophils(Wardlawetal.,1986;Stewart,1994),but developmentsinceamutationinoneofitssubunitsis it certainly has similar activity against other leuko- associated with Miller–Dieker lissencephaly (Hattori cytes in vitro and its administration to animals does et al., 1994; Hirashima et al., 1999). It should be not result in an exclusively eosinophilic response. On noted in passing that the PAF receptor knockout the other hand, the addition of IL-5 increases the mousedoesnotexhibitthislesion.Thus,ifPAFisthe eosinophilresponseselectively(Warringaetal.,1992), cause of this developmental abnormality, it is thus emphasizing the importance of mediator presumably not acting via this receptor. The type II cooperation in selecting specific cells for activation. enzyme is more widely distributed, a 40kDa Longer term cellular responses in inflammatory monomer with significant homology to the plasma cellshavebeenlessofteninvestigatedexceptforthose enzyme (Hattori et al., 1996). It shows a broader classified under the rubric of priming responses, i.e. specificitythanthetypeIbenzymeacting,inaddition, an enhancing function in cooperation with other on oxidized phospholipids (Matsuzawa et al., 1997). stimuli. Even here there is much room for further It seems reasonable to suggest that an important investigation, for example in the de novo synthesis of activity, and probably the early evolutionary signi- new mediators, regulatory enzymes of the eicosanoid ficance, of these enzymes is to control unwanted and pathways, etc. A similar confusing story is seen when potentially toxic oxidized phospholipids in situations questions of PAF action in cell proliferation and of oxidative stress (Matsuzawa et al., 1997; Stafforini maturation are raised. Reported effects include and Tjoelker, 1999). The presence of both intracel- growth inhibition (Shimada et al., 1998; Huang lular and secreted forms with activity against free or et al., 1999) and downregulation of myelopoiesis LDL-associated oxidized phospholipids is in keeping (Dupuis et al., 1998) or, on the other hand, evidence with this suggestion. The additional activity of PAF of growth stimulation (Bennett and Birnboim, 1997; via action on a specific receptor adds to their impor- Rougier et al., 1997). It is presumed that responses tanceasinhibitorsofitsmediatorfunctionsininflam- will vary between cell types and, for the most part, mation and other tissue reactions. will be mediated via secondary production or inhibition of regulatory molecules. RECEPTOR UTILIZATION Regulatory molecules: Inhibitors A PAF receptor has been identified, cloned, and and enhancers shown to be a member of the seven transmembrane, heterotrimeric G protein-linked receptor family. A common statement is that PAF is a mediator of inflammation. What is specifically meant by this is not always clear. The presence of PAF receptors on IN VITRO ACTIVITIES inflammatory cells (including platelets) as well as on the endothelium provides for an action on these cells In vitro findings viathecellresponseslistedabove.Inartificialsystems PAF can be shown to induce chemotaxis, transen- When added to isolated cells with functioning PAF dothelial migration (Casale et al., 1993), and, in con- receptors(e.g.neutrophilsoreosinophils)thetypesof junction with other stimuli, production and secretion Platelet-Activating Factor 1341 of oxidants, other lipid mediators and proinflam- radioimmunoassayshavebeenemployed(Baldoetal., matory mediator proteins. In a significant number of 1991; Karasawa et al., 1991; Tsukioka et al., 1996; circumstances in vivo, the observed PAF effect is Mathiaketal.,1997;Akisuetal.,1998;Ribaldietal., actually mediated by one of these downstream 1998). The problem here is potential crossreactivity products, often an eicosanoid, i.e. PAF is an inter- since the family of related glycerophospholipids is mediary, perhaps contributing an amplification step huge and not all show significant reactivity with the to the overall proinflammatory signal. PAF receptor. The most reliable approach is that A potential theme that might be considered for based on structure. A variety of mass spectrometric PAF action is that it is, for the most part, a cofactor. assays have been described and are increasingly Its role as a priming agent for neutrophil oxidant proving essential to the detection and quantitation of productioniswellestablished.Takingthisasamodel PAF and related lipids in vitro and, even more then, the ability of signaling pathways initiated importantly, in biological samples (Christman and through the PAF receptor to integrate with signaling Blair,1989;HaroldsenandGaskell,1989;Clay,1990; from other, often more cell-specific, receptors could Pickett and Ramesha, 1990; Travers et al., 1997; serve to explain many of the proposed PAF actions. Calleaetal.,1999).Theproblemswiththesemethods For example, its role as an eosinophil chemoattrac- are the requirement for expensive equipment and the tant may involve interaction with the relatively nontrivial sample preparation. Nevertheless, it is eosinophil-selective agent IL-5: similarly with mono- becoming apparent that real understanding of the cytes in the presence of endothelial P-selectin roles for PAF in pathophysiological processes will (Zimmerman et al., 1996). The idea of selective require significant attention to methodology, even at recognition resulting from integratednetworksrather the expense ofincreased effort (Mathiaket al., 1997). than exclusively cell- and ligand-specific receptors is Because of the rapid metabolism of PAF, an gaining general ground (see, for example, the latest additional advantage of the structural approach is concepts of selectivity in odorant recognition; theconcurrentanalysisoflyso-PAFlevelswhichmay Mombaerts, 1999). Similar processes could apply in help indicate the levels of PAF that were originally inflammation where the complex network of media- present (and/or the availability of substrate for tors may drive specific, directed, and sequential cell further PAF generation). responses by the integrated action of numerous, intrinsicallygenerallyactingmediatorsonagivencell type.AcofactorroleforPAFcouldreadilymeanthat IN VIVO BIOLOGICAL its antagonism could dampen down many specific responses(asseeninthestudieswithantagonists)but ACTIVITIES OF LIGANDS IN alsothatspecificessentialactionsareveryhardtopin ANIMAL MODELS down. It would also explain the relative absence of phenotype in the receptor knockout mice, in which Normal physiological roles compensatory mechanisms within the network would be particularly prone to take over. The enormous heterogeneity of the possible effects of PAF in vitro and ultimately in vivo makes it difficult Bioassays used first, to summarize these in simple terms in one chapter and second, to determine which of these Three general methods are available for detection of possibilities is actually operative, or even more PAF and related molecules. The majority of studies critical,important,inthewholeanimal.Todetermine have utilized bioassays, usually platelet stimulation a specific role for a potential mediator, such as PAF with either aggregation or release of previously (and its analogs), in any given process, it would need incorporated serotonin as the readout. Such assays to be present at the site and shown able to induce the are sensitive and will include the PAF-like oxidized responseifintroduced;theeffectshouldbeabrogated phospholipids. Their major disadvantage is the need if the mediator is removed, blocked or absent and to distinguish the stimulation from the many other return when reintroduced. As indicated above, the molecules that can activate platelets, especially when availability of a number of PAF receptor antagonists analyzing biological fluids. Partial purification (lipid has allowed various combinations of these criteria to extraction) helps narrow the field but the assay be applied and has suggested the many potential specificitymustbeenhancedbydemonstratingblock- actions. The presence and wide distribution of potent adewithspecificreceptorantagonistsand/orpretreat- inactivatingenzymeshavealsoledtospeculativeroles ment with acetylhydrolase or PLA . More recently, forPAFinvivoand,insomecases,geneticvariantsof 2 1342 Peter M. Henson these enzymes have been shown to have disease Finally,therearemanyareasofpathophysiologyin associations. which PAF has been invoked that have not yet been Thefieldwasreviewedin1994byStewartbutmost testedrigorouslyinthePAFR(cid:255)/(cid:255)mice,inparticular, of the more recent reviews have addressed specific acute and chronic inflammation, platelet effects, areas of potential action. In this section we will neurophysiologicalprocesses,andresponsesmediated discuss some of the likely areas of PAF effect, not through putative intracellular PAF effects. To date, with any expectation of completeness but rather to then, the receptor knockout animals show alterations emphasize some specific points, concepts, and areas only in IgE-mediated responses – anaphylaxis and for future development. As a general example, the bronchoconstriction. potential importance of oxidized phospholipids in atherosclerosis and the recognition that some of these have PAF-like activity opens up a number of PATHOPHYSIOLOGICAL ROLES important questions that are crying out for more IN NORMAL HUMANS AND detailed investigation. DISEASE STATES AND A receptor knockout mouse has recently been created with a phenotype that calls into question DIAGNOSTIC UTILITY some of these wide-ranging effects – at least as far as those mediated through this receptor. The conse- Role in experiments of nature quences of this deletion are discussed below. and disease states Knockout mouse phenotypes Inflammation PAF is found in inflammatory reactions; it exhibits Such mice developed normally and, when adult, many effects that could potentiate the response and showed no obvious physical or physiologic abnorm- antagonists generally show some mitigating effect on alities (Ishii et al., 1998). The animals reproduced inflammatory processes in which they are examined. normally despite a suggested role for PAF in In such lesions PAF and PAF-like lipids may be reproduction. Interestingly, the mean arterial blood synthesized in the tissues themselves as a response to pressure was unchanged in the PAFR(cid:255)/(cid:255) mice, injury, or generated by nonenzymatic oxidative which is not easily consistent with a role for PAF in reactions. The latter might be endogenous (as, for normal vasomotor tone, although confirmation of example, in ischemia–reperfusion or in oxidized this would require more detailed investigation. There lipoproteins) or could be exogenous via inhaled or is a long-standing suggestion that PAF is involved in ingestedtoxins(e.g.tobaccosmoke;FitzGerald,1997; anaphylactic shock and endotoxic shock. In keeping Lehr et al., 1997). Two key elements in early with the former, intravascular antigen administration inflammation are local alterations in blood flow to sensitized PAFR(cid:255)/(cid:255) animals did indeed result in leading to enhanced opportunity for and increased markedly diminishedshockand also the concomitant granulocyteadhesion tothe endothelium. PAF seems bronchial constriction and airways resistance. By to contribute to the former, for the most part by contrast, endotoxin effects in knockout mice were induction of prostanoid synthesis (Hellewell et al., indistinguishable from the control PAFR+/+ 1991). PAF acts directly on circulating inflammatory animals. Since a wide variety of PAF receptor anta- cells to increase adhesiveness for the endothelium. It gonists have been shown capable of reducing or also induces actin assembly (Omann et al., 1995), abolishing endotoxic shock, this finding was indeed a which, in turn, reduces cell deformability and results surprise. Even more so since PAF receptor over- in trapping within pulmonary capillaries (Downey expressing mice showed increased susceptibility to et al., 1993) or in low-pressure systems during shock. LPS (Ishii et al., 1997). Admittedly, many of the One such example of PAF involvement has been pharmacologicalstudieswithPAFinendotoxinshock describedinsomedetail.Exposureofendothelialcells were performed in rats and there are marked species to stimuli such as thrombin results in a rapidly differences in LPS responses. Additionally, possible increasedneutrophilmediatedadhesivenesscausedby compensatory mechanisms are well known to have synthesisof PAFinthe endothelial cells, exposureon the ability to confound conclusions drawn from the surface, interaction with the neutrophil PAF knockoutmice.Despitethecaveats,atfirstglancewe receptor and inside-out signaling for (cid:12) -integrin 2 mayhavetoreassessourconceptsofPAFactionand/ upregulationandactivation(Zimmermanetal.,1990). or reconsider the issue of alternative receptors or PAF acetylhydrolase blocked the process by remov- modes of action for this group of molecules. ingthePAFfromtheendothelialsurface.Thistypeof Platelet-Activating Factor 1343 activation has been termed juxtacrine signaling, and controversial and difficult to interpret. One current may involve a number of different molecules (e.g. suggestion to explain these discrepancies is that the P-selectin) each cooperating to optimize leukocyte– early antagonists were just not potent enough. The endothelial cell interaction (Zimmerman et al., 1997). results of ongoing trials with more effective antago- A similar involvement of P-selectin and PAF is nists are eagerly awaited. An alternative explanation, reportedtomediatetheleukocyteadhesionseenwhen inkeepingwiththeoverallthemeofthisdiscussion,is the endothelium is activated by the membrane attack that PAF is again a bit player (although perhaps an complex of the complement system (Kilgore et al., important one) in the local allergic and asthmatic 1998). responses but that overproduction or underinactiva- Whether the PAF activity as a true chemoattrac- tion as in the acetylhydrolase abnormalities increases tant in vitro is ever actually evident in vivo is not at its overall participation in mediator network control- this point clear. It would require the generation of lingtheinflammatoryresponses.Athirdpossibilityis PAF(orrelatedmolecules)tocreateastablegradient that PAF contributes to the allergic responses by in the tissue allowing responsive cells to recognize activities in addition to those mediated through the and migrate up the gradient towards its source. G protein-linked receptor, perhaps within the cell. In Inactivation by acetylhydrolases or uptake into cells this circumstance an acetylhydrolase effect would be would act against this. Perhaps more realistically expected, but not one from deletion of the receptor. PAF may trigger increased migratory action (chemo- kinesis)intherespondingcellsasoneinasequenceof Anaphylactic Shock stimulithateventuallyleadtoaccumulationatthesite of injury or insult. As indicated above, an even more Perhaps here is the condition most obviously asso- important function in inflammation may be the ciatedwithPAFeffectsinvivo.Thereceptorknockout priming effect, in which exposure to PAF at some mice are protected from IgE-mediated anaphylaxis; point in the accumulation process leads to enhanced PAF injections can produce an almost identical responsiveness to other stimuli in the production of response; PAF has been detected in the blood during oxidants, eicosanoids, and other stimuli. For exam- anaphylaxis (Feuerstein and Hallenbeck, 1987) and, ple, this is likely the role played by PAF in the injury followingrecovery,thefingerprint ofPAFactioncan seen during immune complex-induced inflammatory be detected in specific desensitization of circulating responses (Warren et al., 1990; Ward et al., 1991). platelets (Henson and Pinckard, 1977). The potency of PAF–receptor interaction, the rapidity of PAF synthesis, and secretion from IgE-responsive cells Asthma and Allergic Reactions (mast cells and basophils), as well as breadth of There has long been a suggested role for PAF in respondingcells,inparticularinthiscasethoseofthe allergy and asthma deriving originally from the IgE- vasculature, all make this type of immediate and induced release of PAF from basophils and only short-livedresponseaprimecandidatetobemediated enhanced by the association of PAF with eosinophil byPAF.Ontheotherhand,evenhere,PAFisclearly chemoattraction and activation. Overexpression of not the only mediator involved. What the observa- the PAF receptor in mice led to increased bronchial tions do suggest is that during allergic responses inflammation (Ishii et al., 1997). Interestingly, despite PAFisproduced,hasmanyactivitiesandtargets,and theeffectonIgE-anaphylacticshock,thereisasyetno is rapidly inactivated. In tissues these participate in report of bronchial responsiveness in the receptor theoverallreaction.Inthecirculationthroughoutthe knockoutmice.PAFexposuretohumanairwaysmay bodytheseeffectscanbemomentarilyoverwhelming. induce bronchoconstriction and hyperresponsiveness (Cuss et al., 1986); leukocytes from asthmatics have Sepsis and Endotoxin Shock been reported to show altered responses to PAF (Bialasiewiczetal.,1997)andincreasedPAFhasbeen PAF antagonists have long been known to block detected in asthmatics (Tsukioka et al., 1996). A endotoxin shock (particularly in rats; Mathiak et al., PAF receptor antagonist (SR27417A) has been 1997)aswellasdamagetospecificorganssuchasthe reported to show some protection against antigen- lung (Chang et al., 1990). PAF has been detected in induced hyperresponsiveness (Evans et al., 1997). the blood in experimental animals and humans and As indicated above, an acetylhydrolase mutation levels in clinical sepsis have been correlated with correlates well with asthma incidence in a Japanese severity. Increased endotoxin lethality was seen in population. On the other hand, early trials of PAF PAF receptor overexpressing mice (Ishii et al., 1997) receptor antagonists were disappointing and the but this observation is at odds with the lack of direct effects of PAF administration are somewhat protectioninthereceptorknockoutanimals.Possibly 1344 Peter M. Henson the PAF receptor does not play such a critical role in fashion (DeFily et al., 1996) even though in vivo its the normal animal but can contribute if present in effects may be much more complicated. Other effects larger amounts. More likely, compensatory mecha- areclearlymediatedviaformedelementsintheblood nisms have developed in the knockout mouse to (Voelkel et al., 1992). reduce the dependence on PAF. There may also be The effects of PAF and related lipids on the heart significant species differences in the predominance of are also very complicated and multifactorial PAFeffectsinendotoxemia(Spapenetal.,1997).The (Feuerstein et al., 1997). Here, too, the actions are recent review by Mathiak addresses important issues both direct on myocardial cells and indirect via andconcernswithregardtothepossibleroleforPAF effects on endothelium or inflammatory cells and in endotoxic shock in humans and animals (Mathiak release of eicosanoids or mediators such as TNF(cid:11). et al., 1997). The problem is that, despite exciting The molecules induce coronary artery constriction, possibilitiesderivedfromexperimentalmodels,exten- modulate contractility, and may contribute to some sion into the clinic has been extremely disappointing arrhythmias. In addition, because of their general and frustrating (Fink, 1998). A call is made for more involvement in the ischemia–reperfusion processes, accuratemeansofdetectingPAFandthe timecourse they are also implicated in such responses within the of its generation, better animal models, and more heart. Here too is an arena in which the judicious appropriately designed clinical trials. Mechanistically investigation of antagonists may have clinical value. it has been suggested that PAF exhibits direct effects on the systemic vasculature and heart as well as Ischemia–Reperfusion Reactions enhancing the production of additional mediators suchasTNF(cid:11)andIL-1(Davies and Hagen,1997)or PAF is widely associated with episodes of ischemia– interacting with them to produce a more profound reperfusion in most areas of the body, e.g. brain injury (Rabinovici et al., 1993). (Silverstein et al., 1997, Jean et al., 1998), gut (Muguruma et al., 1997), and heart (Loucks et al., 1997). This may be a major circumstance for pro- Vascular and Cardiac Effects duction and action. It is synthesized at such lesions PAF has long been known as an antihypertensive and appears to contribute to inflammatory cell agent (Blank et al., 1982). It exhibits negatively ino- accumulation and activation. Antagonists can often tropic effects on the heart and lowers systemic blood be shown to mitigate the injurious effects. However, pressure, inducing shock. Receptor antagonists have here, as elsewhere in examination of PAF as been reported to reduce pulmonary hypertension in mediator, it only acts in conjunction with many chronically hypoxic rats, probably at the level of other factors. Importantly, in most cases, a role for vascular remodeling (Ono et al., 1992). Despite early PAF is suggested only on the basis of antagonismor, suggestions that PAF might play a role in normal if the molecule was detected in the lesions, the assay vasculartoneandreactivity,itnowseemsmorelikely was based on its activity. It seems reasonable to that it acts more as a vasomodulator in injurious or suggest that ischemia–reperfusion is, par excellence, inflammatory situations (Chen et al., 1992). an opportunity for the generation of oxidized phos- The actions of PAF on the vasculature are pholipids, that it is important for the inflammatory extremely complex. It can induce both vasoconstric- system to recognize and be activated in such cir- tion and dilation, in some cases at different doses cumstances and that these oxidized phospholipids, as (Juncos et al., 1993) or times or on different vessels well as the acetyltransferase-generated PAF, are (Gaoet al.,1995).PAF-inducedvasodilationappears important contributors to the response. to be primarily mediated via endothelium-derived nitric oxide (Arima et al., 1996; Kamata et al., 1996) Gastrointestinal Tract whereasvasoconstrictionmaybecausedbysecondary production of thromboxane (Hellewell et al., 1991; PAF has been implicated as a mediator in various Uhlig et al., 1996) or leukotrienes (Voelkel et al., toxic and inflammatory conditions of the gastro- 1983) and may act synergistically in this regard with intestinal tract (Muguruma et al., 1997). It acts as a other mediators such as TNF(cid:11) (Chang, 1994). secretogogue in the colon (Thielman et al., 1997; One issue for the effects of PAF in the vasculature Rocha et al., 1998) and may be associated with the iswhetheritisactingdirectlyonthevessel(evenifby in vivo actions of toxins associated with clostridial the secondary generation of eicosanoids) or whether infection and cholera. It may also play a role in the itseffectisviaactivationofcirculatingneutrophilsor actionsofvariouslaxatives(Izzoetal.,1998)andhas platelets. Thus, PAF was shown to prevent coronary been reported to stimulate secretion of gastric acid artery vasodilation in a non-neutrophil-dependent (Sobhani et al., 1995).