A Reassessment of the Low Molecular Weight Phospholipase A2 Gene Family in Mammals*

Phospholipase activity was first described in pancreatic juice and cobra venom at about the turn of the century. Phospholipase A2s (PLA2s) 1 are those phospholipases that hydrolyze the sn-2 fatty acid acyl ester bond of phosphoglycerides to free fatty acid and lysophospholipids. PLA2s have been divided into several groups based on molecular weight, amino acid sequence and homology (e.g. position of Cys residues in the low molecular weight enzymes), calcium dependence, and cellular localization (see below). Two groups of ;14-kDa snake venom PLA2s, Group I from cobras and kraits and Group II from rattlesnakes and vipers, are well known. Until 1989, however, the only well characterized mammalian PLA2 was the ;14-kDa enzyme from pancreatic juice, which was classified as a Group IB enzyme (for reviews see Refs. 1 and 2). In 1989 two groups each described the gene for a ;14-kDa enzyme from synovial fluid and platelets, which is distinct from the pancreatic enzyme and was assigned to Group IIA. These publications engendered a relatively large literature on what is thought to be Group IIA PLA2, primarily because it is believed to be elevated in serum and exudates in certain inflammatory diseases, suggesting its involvement in the production of lipid mediators of inflammation (see below). Also, despite conflicting data concerning its physiologic substrate, it was recognized that Group IIA PLA2 has the potential to release arachidonic acid from membranes, which then may serve as the precursor of prostaglandins, thromboxanes, and prostacyclins via the cyclooxygenase pathway or leukotrienes and lipoxins via the lipoxygenase pathway (for review see Ref. 3). In 1994, our group described the cloning of genes and expression of cDNAs for two new mammalian ;14-kDa PLA2s (4–6). This brought the number of well described mammalian low molecular weight PLA2 genes and enzymes to four. This review summarizes what is now known about mammalian ;14-kDa PLA2s, particularly those that have been most recently described. The focus is on possible biological functions of each of the ;14-kDa PLA2s and functional relationships to the structurally unrelated, 85-kDa group IV cytosolic PLA2. I also review new data that force a reevaluation of a significant fraction of the large literature describing tissue distribution and metabolic functions of the mammalian Group IIA PLA2.

Phospholipase activity was first described in pancreatic juice and cobra venom at about the turn of the century. Phospholipase A 2 s (PLA 2 s) 1 are those phospholipases that hydrolyze the sn-2 fatty acid acyl ester bond of phosphoglycerides to free fatty acid and lysophospholipids. PLA 2 s have been divided into several groups based on molecular weight, amino acid sequence and homology (e.g. position of Cys residues in the low molecular weight enzymes), calcium dependence, and cellular localization (see below). Two groups of ϳ14-kDa snake venom PLA 2 s, Group I from cobras and kraits and Group II from rattlesnakes and vipers, are well known. Until 1989, however, the only well characterized mammalian PLA 2 was the ϳ14-kDa enzyme from pancreatic juice, which was classified as a Group IB enzyme (for reviews see Refs. 1 and 2). In 1989 two groups each described the gene for a ϳ14-kDa enzyme from synovial fluid and platelets, which is distinct from the pancreatic enzyme and was assigned to Group IIA. These publications engendered a relatively large literature on what is thought to be Group IIA PLA 2 , primarily because it is believed to be elevated in serum and exudates in certain inflammatory diseases, suggesting its involvement in the production of lipid mediators of inflammation (see below). Also, despite conflicting data concerning its physiologic substrate, it was recognized that Group IIA PLA 2 has the potential to release arachidonic acid from membranes, which then may serve as the precursor of prostaglandins, thromboxanes, and prostacyclins via the cyclooxygenase pathway or leukotrienes and lipoxins via the lipoxygenase pathway (for review see Ref. 3). In 1994, our group described the cloning of genes and expression of cDNAs for two new mammalian ϳ14-kDa PLA 2 s (4 -6). This brought the number of well described mammalian low molecular weight PLA 2 genes and enzymes to four. This review summarizes what is now known about mammalian ϳ14-kDa PLA 2 s, particularly those that have been most recently described. The focus is on possible biological functions of each of the ϳ14-kDa PLA 2 s and functional relationships to the structurally unrelated, 85-kDa group IV cytosolic PLA 2 . I also review new data that force a reevaluation of a significant fraction of the large literature describing tissue distribution and metabolic functions of the mammalian Group IIA PLA 2 .

Classification of Mammalian Low Molecular
Weight PLA 2 The nomenclature of PLA 2 enzyme groups distinguished on the basis of structural and other criteria has been reviewed by Dennis (1, 2) and adopted here. Since the first review of the subject, however, two new ϳ14-kDa mammalian PLA 2 s have been described (see below). Whereas many novel PLA 2 activities and partial amino acid sequences have been reported, we reserve group or subgroup designations for mammalian ϳ14-kDa PLA 2 s with characterized genes and demonstrated expression. Further, we have made an effort to have the gene and enzyme group nomenclature correspond with the designation of genes for both existing and newly discovered PLA 2 s. Therefore, for example, the mouse PLA 2 group IIA, IIC, and V genes have been designated Pla2g2a, Pla2g2c, and Pla2g5, respectively (7).
All of the ϳ14-kDa PLA 2 s contain an even number of Cys at characteristic positions, each of which pairs with another specific Cys to form a disulfide bridge, thus producing a rigid three-dimensional structure. Within group I, x-ray crystallography has demonstrated that enzymes from divergent sources such as snakes and mammals have quite similar crystal structures (1). It is likely that all of the low molecular weight PLA 2 s utilize a specific catalytic His, assisted by an Asp, to polarize a bound H 2 O, which then attacks the carbonyl group of the phospholipid substrate. Ca 2ϩ is required to stabilize the transition state and is bound within the highly evolutionarily conserved "calcium binding loop" observed in all ϳ14-kDa PLA 2 s (1,8).
Both group IA PLA 2 , found only in snakes, and group IB PLA 2 , which appears ubiquitously in mammals, have a disulfide bridge connecting Cys-11 to Cys-77 and a characteristic three-amino acid "elapid loop" composed of residues 54 -56. The mammalian group IB PLA 2 has 14 Cys residues and is secreted predominantly by the pancreas to function extracellularly in digestion (1,3). Group IB PLA 2 is also present in some nondigestive organs, suggesting a possible secondary role (9,10). Group IIA PLA 2 , which has been described for many mammals, and group IIB PLA 2 , which has only been observed in the Gabon viper, also contain 14 Cys residues but, in contrast to the group I enzymes, lack the Cys-11 to Cys-77 disulfide bridge. All group II PLA 2 s have a C-terminal extension of 6 amino acids that terminates in a Cys joined to Cys-50 near the His-48 catalytic site. The mammalian group IIA has been reported to occur in relatively small amounts in mast cells, macrophages, and diverse tissues such as liver and spleen (3). It is also reported to occur in greater amounts in fluid from arthritic synovia and serum from patients with inflammatory diseases such as acute pancreatitis and sepsis (11). The group IIA enzyme is frequently referred to as "secreted PLA 2 ," but this term lacks precision because the enzyme has also been localized within mitochondria (12), and other "secreted" PLA 2 s are now known. As described below, reports of expression of group IIA PLA 2 in various cell types will require detailed reevaluation in light of new data that indicate some methods used for its detection also detect the more recently described group V PLA 2 .
The group IIC PLA 2 gene has been characterized in rat and mouse (5,13). It encodes a mature enzyme, with a calculated molecular mass of 14.8 kDa, which does not contain the Cys-11 to Cys-77 disulfide bridge or elapid loop characteristic of group I but does contain the 6-amino acid C-terminal extension characteristic of group II. The group IIC enzyme is distinguished from group IIA and group IIB enzymes in that it contains 16 Cys residues. Further, the group IIC enzyme from mouse and rat contains only 17 of the 18 amino acids that had been thought to be invariantly conserved in low molecular weight PLA 2 s (14, 15), Ile-9 being replaced by Val. Rodent group IIC PLA 2 is highly expressed in adult but not prepubescent testis (5). In situ hybridization of testis tissue sections indicates that the group IIC gene is expressed mainly in pachytene spermatocytes, secondary spermatocytes, and round spermatids but not in spermatogonia, elongating spermatids, or Sertoli cells (13). The N-terminal portion of exon III is absent in the human group IIC PLA 2 gene, and about 16% of alleles also exhibit a common nonsense mutation in exon II. Significantly, all other parts of the human group IIC gene appear potentially functional and highly homologous to the functional rodent genes, but there is no evidence for group IIC gene expression in human tissues. Thus, we conclude that the group IIC gene has recently evolved into a pseu-dogene in humans (7). It is not known whether there is compensatory activity of one of the other PLA 2 genes in human testis or other tissues.
The group V PLA 2 gene and its product have been characterized in human (4), rat (6), and mouse (16). 2 The mature enzyme, with a calculated molecular mass of 13.6 kDa, contains neither the elapid loop of group I nor the 6-amino acid carboxyl extension of group II. Further, it contains only 12 of the Cys found in group I and II PLA 2 s. Thus, this second new ϳ14-kDa PLA 2 has been placed into a new group known as group V 3 (6). The group V PLA 2 gene is expressed highly in heart, placenta, and, to a lesser extent, lung and liver (4,6). Further, group V, rather than group IIA as was previously believed, appears to be the primary ϳ14-kDa PLA 2 expressed by P388D 1 macrophage-like cells and mast cells (see below). As is the case for the group IIA and IIC genes, the group V gene product is expressed initially as a prepeptide with the first 20 amino acids probably representing a signal peptide that is subsequently cleaved (4).
Some distinguishing properties of the ϳ14-kDa PLA 2 s are summarized in Table I. In addition, there are common features such as pH 7-9 activity optima and a requirement for about 1-10 mM Ca 2ϩ for maximal activity (4 -6). This latter property stands in contrast with the 85-kDa group IV enzyme, which is activated by Ca 2ϩ concentrations in the micromolar range (for review see Ref. 1). However, there is one report that under certain conditions group IIA enzyme, but not group IB, achieves 50% of maximal response with 0.5 M Ca 2ϩ (17). It is also important to recognize that the demonstration of substrate preferences for each of the ϳ14-kDa PLA 2 s (e.g. Refs. 4 -6) was merely intended to distinguish between the various groups of PLA 2 s at a particular assay pH and Ca 2ϩ concentration. Because of inherent variation in the presentation of lipid substrate (1), such assay data should not be generalized to characterize activity in vivo.

Low Molecular Weight PLA 2 Genomics and Evolution
The human group IB PLA 2 gene has been shown to reside on chromosome 12 (18), and the human groups IIA and V genes and group IIC pseudogene are tightly linked on chromosome 1p34-p36.1. Consistent with the human localization, the mouse group IIA, IIC, and V genes are also tightly linked and located on the distal region of chromosome 4, which is known to be syntenic with human 1p34 -36.1. The data from radiation hybrids suggest that the human group IIA and group V genes are very close whereas the group IIC pseudogene is located about 1 centimorgan toward the centromere (7). The clustering of PLA 2 genes invites speculation about possible complex coordinate regulation of expression as is the case for mammalian globin genes.
The amino acid coding regions of each of the ϳ14-kDa PLA 2 s are interrupted by 3 introns, which are almost identically positioned when the enzymes are compared in homologous alignment shown in Fig. 1 (4 -6, 19 -21). In addition, the human and rat group V (4, 6) and the mouse group IIA (22) gene have one upstream noncoding exon, and the rat group IIC gene has three upstream noncoding exons. The first two exons in the rat group IIC gene are alternatively transcribed in testis and brain mRNA (5). This amino acid alignment clearly suggests the common evolutionary origin of these genes. We propose the evolutionary scheme of ϳ14-kDa PLA 2 gene duplication events shown in Fig. 2, which is a modification of Davidson and Dennis (14) in light of the subsequent discovery of groups IIC and V. 4 Group V PLA 2 s exhibit 12 Cys residues that are identical in position to the 12 of 14 Cys residues common to Groups I and II. This suggests that group V is the progenitor to groups I and II. Group III PLA 2 s, which are found only in bees and some lizards, have 10 Cys residues and may have diverged from the common ancestral PLA 2 before the divergence of invertebrates and vertebrates (14). The divergence of groups I and II may have occurred simultaneously or at different times. Group IIC PLA 2 has 16 Cys residues, 14 of which are shared with group IIA and 12 of which are identical to all 12 Cys residues of group IIB (14). Therefore, we suggest that Group IIC diverged from a common group II ancestor with 14 Cys residues and that Groups IIA and IIB diverged at a later time.

Revised View of PLA 2 Gene Expression in Cell Signaling
Although it is well documented that individual cells of different types contain multiple PLA 2 s (23-25), some understanding of how these different enzymes cooperate in receptor-coupled cellular activation has only recently emerged. In certain cell types, agonistinduced PLA 2 activity can be either short term, long term, or biphasic, depending on the agonist or combination of agonists. These patterns are reflected in the spatial and temporal kinetics of arachidonic acid release and the subsequent production of eicosanoids. For example, in P388D 1 macrophage-like cells and mast cells arachidonic acid release in response to certain agonist combinations can be shown to be biphasic and dependent on the activities of both group IV cytosolic PLA 2 and a low molecular weight PLA 2 that was, until recently, believed to be group IIA. Balsinde et al. (25) showed that P388D 1 mouse cells stimulated with bacterial lipopolysaccharide and platelet-activating factor release arachidonic acid in two phases, an initial rapid accumulation inside of the cell within the first few minutes and a subsequent sustained phase of accumulation in the culture medium. It was subsequently shown that group IV PLA 2 acting intracellularly is responsible for the initial, rapid release of arachidonic acid whereas a ϳ14-kDa PLA 2 acting on the outer surface of the cell was responsible for the greater, mostly extracellular, sustained release of arachidonic acid 2 M. V. Winstead and J. A. Tischfield, unpublished results. 3 Dennis (1) numbered the well characterized PLA 2 s in the order of their discovery and clear characterization. Thus, the 85-kDa cytosolic PLA 2 is group IV according to his nomenclature. 4 J. Chen and J. A. Tischfield, unpublished data.  (20,21), and V (4) are from human whereas group IIC (5) is from rat. Yellow indicates four identical amino acids at a particular position, red indicates three identical amino acids, and blue indicates two identical amino acids. Boxed amino acids are specified by codons that are interrupted by introns. The entire encoded sequence including pre/propeptide regions is shown. In this alignment, the mature enzymes begin at position 29. (26). The data also suggested that intracellular and extracellular arachidonic acid arise from different phospholipid pools within the cell. Most recently, Balboa et al. (16) showed that the ϳ14-kDa enzyme in P388D 1 cells associated with the sustained release of arachidonic acid is group V and not group IIA as was previously believed. Whereas there was no detectable mRNA for PLA 2 groups IIA or IIC in either resting or activated cells, the group V mRNA was abundant in both. Antisense oligonucleotides for the highly conserved Ca 2ϩ -binding domain of rat group IIA mRNA, which had previously been used in experiments that were interpreted as implicating group IIA (27), were shown to act in a nonspecific way on group V. However, more specific antisense oligonucleotides against a unique exon region of mouse group V PLA 2 blocked expression by about 60 -70%, whereas the control sense oligonucleotide was without effect. Interestingly, a polyclonal antiserum against human synovial fluid PLA 2 , presumably group IIA, was used to successfully detect expression of the P388D 1 cell surface PLA 2 . This result indicates that this antiserum, and such antisera in general, may not be able to distinguish between PLA 2 groups IIA and V (see below).
The agonist-mediated activation of mast cells includes degranulation and release of ligands such as serotonin and histamine and is similar in several key respects to the biphasic response observed in P388D 1 cells. Bone marrow-derived mouse mast cells stimulated by antigen aggregation of high affinity IgE receptors on the cell surface (28) or with c-kit ligand, IL-10, and IL-1␤ or by priming with c-kit ligand and IL-10 followed by IgE and antigen activation (29) exhibit an immediate phase of arachidonic acid release during the first 10 -20 min followed by a delayed phase from hours 2 to 7. These phases of arachidonic acid release are reflected by early and late phases of prostaglandin D 2 (PGD 2 ) synthesis, which are mediated by the constitutive prostaglandin synthase 1 (PGS1) and the inducible prostaglandin synthase 2 (PGS2), respectively (28,29). However, there is controversy as to which PLA 2 s provide the arachidonic acid for early and late PGD 2 synthesis. The results from one group indicate that a ϳ14-kDa PLA 2 coupled to PGS1 is responsible for early phase PGD 2 synthesis whereas the group IV PLA 2 coupled to PGS2 is responsible for late phase PGD 2 synthesis (30). In contrast, a second group's data suggest the reverse, i.e. a ϳ14-kDa PLA 2 coupled with PGS2 is responsible for late phase PGD 2 production (31). These apparently contradictory conclusions may be a consequence of differences in experimental methodologies. Both groups, however, clearly demonstrate that one phase of PGD 2 synthesis is a consequence of the activity of a low molecular weight PLA 2 that was initially believed to be group IIA. As described below, we now know that Group IIA PLA 2 is not involved in mast cell activation.
The involvement of specific ϳ14-kDa PLA 2 s in mast cell activation has been clarified by the discovery of a naturally occurring mutation in the group IIA PLA 2 gene of many inbred mouse strains. The "murine intestinal neoplasia" or Apc Min gene is the ortholog of the human APC gene, which has been shown to be mutated in a hereditary form of colon cancer known as familial adenomatous polyposis coli. The number of intestinal tumors is increased in mouse strains that also carry the Mom1 (modifier of Min-1) mutation that is likely a frameshift mutation in the gene for group IIA PLA 2 , such that Mom1 homozygotes (Pla2g2a Ϫ/Ϫ genotype) express little or no group IIA mRNA or protein (22,32). Bingham et al. (31) demonstrated that mast cells from Mom1 homozygotes exhibit a normal biphasic response to ligand stimulation and that one phase of this response, previously attributed to group IIA PLA 2 , must therefore be mediated by another PLA 2 that has some properties in common with the group IIA enzyme. Contemporaneously, Reddy et al. (33) demonstrated that mast cells from both Mom1 homozygotes and normal mice exhibit biphasic responses to activation and that both early and delayed PGD 2 production and ϳ14-kDa PLA 2 secretion into the medium are similar in both genotypes. Neither Mom1 homozygotes nor normal cells exhibited any group IIA or group IIC mRNA as determined by Northern blotting and the more sensitive technique of reverse transcriptase/polymerase chain reaction amplification. However, cells of both genotypes exhibited Group V mRNA, and PLA 2 activity was secreted into the medium as determined by assay, binding to monoclonal antibody directed against recombinant human group IIA PLA 2 , and inhibition by a drug (SB203347) developed as an inhibitor of the group IIA enzyme (33). Also, it was shown that the PLA 2 secreted from both Pla2g2a ϩ/ϩ and Pla2g2a Ϫ/Ϫ mast cells could release arachidonic acid in distal mouse Swiss 3T3 cells, which then utilized this arachidonic acid for prostaglandin synthesis (33,34). These data clearly implicate involvement of the group V PLA 2 in one phase of mast cell immune activation.
A biphasic response implicating both the group IV and a ϳ14-kDa PLA 2 is also observed in cytokine-stimulated rat mesangial cells (35,36) and human endothelial cells (37). Clearly, the precise identity of the ϳ14-kDa PLA 2 will have to be reevaluated in light of the above data from P388D 1 and mast cells. High levels of ϳ14-kDa PLA 2 , believed to be group IIA, have been described in serum and tissue exudates from patients with a variety of inflammatory diseases including arthritis, pancreatitis, adult respiratory distress syndrome, and septic shock (20,38,39,(41)(42)(43)(44)(45). The identification of group IIA PLA 2 in these diseases has most frequently been based on enzymatic assays, use of inhibitors developed against group IIA PLA 2 , or immunologic detection (cf. Ref. 46). The data of Balboa et al. (16), Bingham et al. (31), and Reddy et al. (33) clearly indicate that these criteria are inadequate for distinguishing between the group IIA and group V PLA 2 s. Thus, an entire literature will require reevaluation. Furthermore, existing and new PLA 2 inhibitors will require evaluation with regard to their activity against both the group IIA and group V enzymes (47). A recent publication by Tseng et al. (48) suggests that it may be possible to design specific peptide inhibitors for each of the ϳ14-kDa PLA 2 s based on their individual ␤-loop pentapeptide sequences.

Perspectives
Knowledge of at least three (humans) or four (rodents) related genes encoding distinct ϳ14-kDa PLA 2 s that are expressed in a tissue-specific manner begs the question of whether or not there are differences among the PLA 2 s that relate to function in vivo. A first step toward answering this question will be determination of the expression of each PLA 2 in a wide variety of cell types and under different circumstances (e.g. activation). This, in turn, may require technical innovations such as more specific inhibitors or antibodies for immunocytochemistry. Naturally occurring (e.g. Mom1) or targeted gene knockouts could also be useful to this end, especially if, as has been the case to date (33), there is no compensatory activation of another PLA 2 gene. One should be cautious, however, in postulating a molecular basis for a phenotype in knockout mice. For example, cyclooxygenase-2 gene knockout mice or mice treated with specific cyclooxygenase-2 inhibitors exhibit a reduced level of Apc Min -related intestinal polyposis (49). One might have anticipated the opposite result from what is known about Mom1 mice, given that arachidonic acid is the substrate for cyclooxygenase-2. Mom1 mice, which exhibit increased intestinal polyposis, may have reduced levels of arachidonic acid release in intestinal cells, which clearly express the group IIA enzyme in wild-type mice (16). Also, our consideration of phospholipid metabolism should include possible functions of newly discovered PLA 2like genes, such as PLA2L mapped to human chromosome 8q24 (50).
Finally, the gene encoding a receptor for PLA 2 has been cloned from cow (51), rabbit (52), and human (53). These homologous receptors are composed of about 1460 amino acids and are structurally related to the macrophage mannose receptor (51). Whereas the cow and rabbit receptors bind porcine group IB PLA 2 with high affinity and the rabbit receptor binds group IIA PLA 2 with high but somewhat lesser affinity, the human kidney receptor has only weak affinity for group IB and little affinity for group IIA (53). The human PLA 2 receptor is highly expressed in diverse tissues such as kidney, lung, placenta, and skeletal muscle, a pattern that is different from rabbit. Humans also exhibit a soluble form of the receptor not reported for the other species (53), in addition to the membrane-bound form. The initial view based on the research in cow was that the PLA 2 receptor mediated extrapancreatic group IB PLA 2 -induced DNA synthesis, contraction, eicosanoid production, and chemokinetic cell migration (see Ref. 51), but the lack of receptor affinity for group IB PLA 2 in humans does not support this idea (53). The critical region(s) in PLA 2 for binding to the receptor appears to reside within the Ca 2ϩ -binding domain (40), which exhibits relatively high homology among all ϳ14-kDa PLA 2 s (1, 15). There are, as yet, no reports of experiments testing the binding of group IIC or group V PLA 2 to this receptor. It is possible that either one or both of these enzymes will prove to be a natural ligand for the PLA 2 receptor in a particular species, suggesting novel functions for these enzymes. The PLA 2 receptor has also been implicated in the induction of group IIA PLA 2 mRNA transcription by the group IB pancreatic enzyme in rat mesangial cells (36). Perhaps the PLA 2 receptor also mediates cross-talk between additional ϳ14-kDa PLA 2 s in a cell type-or species-specific manner.