INTRODUCTION

TOP ABSTRACT INTRODUCTION PAP1 AND LPP ENZYMATIC... SUBSTRATES FOR LPP CDNAS FOR LPPS RECEPTORS AND OTHER... LPP IN SIGNALING PLATFORMS CELLULAR FUNCTIONS FOR PAP1... FINAL CONSIDERATIONS REFERENCES

SINCE 1899, when Overton (186) reported that the diffusion of chemical compounds into cells varied directly with their solubility in oil compared with water, it has been recognized that lipids are major constituents of plasma membranes. Subsequent studies by Gorter and Grendel (77) and Danielli and Davson (47) in the 1920s and 1930s concluded that the limiting membrane at the cell surface was a bimolecular (phospho)lipid leaflet (see Ref. 286 for further details). Numerous investigations, in particular electron microscopic examinations of cellular and reconstituted membranes (207, 240), led to the conclusion that intracellular and extracellular biological membranes share similar structural properties. Further studies on membrane properties eventually led to the presently accepted fluid-mosaic model (229).

More recently, biochemists and cell biologists have focused considerable attention on mechanisms by which information can be transmitted across the hydrophobic barriers surrounding biological cells. Interestingly, it has become apparent that the phospholipid constituents of cells provide substrates for first and second messengers and that phospholipases, originally thought of only as degradative enzymes, play important roles in signaling. For example, early observations revealed that phospholipase A₂ (PLA₂), by releasing the fatty acids from the sn-2 position of certain phospholipids, provides arachidonic acid for eicosanoid synthesis (51, 72, 133). Hydrolysis of phosphatidylinositol-4,5 bis phosphate (PIP₂) by phosphatidylinositol-specific phospholipase C (PI-PLC) provides both diacylglycerol (DAG) and inositol tris phosphate (IP₃) for protein kinase C (PKC) activation and intracellular calcium mobilization (202). In addition, it has become evident that phospholipids themselves can act as first and second messengers. PLA₂ hydrolysis of 1-alkyl,2-arachidonoyl phosphatidylcholine not only generates substrate for eicosanoid formation but provides the lyso-precursor for 1-alkyl,2-acetyl phosphatidylcholine, also known as platelet-activating factor (195). Also, PIP₂ leads to intracellular messages, not only through degradation by PI-PLC but also via phosphatidylinositol 3-kinase activity, resulting in PI 3,4,5-trisphosphate (PIP₃), an important activator of protein kinase B (21). More germane to the present review, lysophosphatidic acid (LPA) and sphingosine-1-phosphate (S1P) are important serum-derived extracellular first messengers, with potent physiological effects.

The present article focuses on phosphatidic acid (PA) phosphatase (PAP), which hydrolyzes 1,2-diacyl,sn-glycerol-3-phosphate to DAG and inorganic phosphate. Originally considered only for its role in glycerolipid biosynthesis, it has recently been demonstrated that this phosphohydrolase plays an important role as a modulator of cell signaling in relation to cell mobility, differentiation, growth, and survival (24, 25, 31, 120, 260).

Current evidence indicates at least two distinct classes of PAP in mammalian cells. PAP1 is a soluble cytosolic protein, which, under the influence of free fatty acids and perhaps other signals, can translocate to the endoplasmic reticulum (ER) and possibly other intracellular organelles, where it converts newly biosynthesized phosphatidate to DAG. DAG generated by PAP1 is utilized for the formation of phosphatidylcholine (PC), phosphatidylethanolamine (PE), and triacylglycerol (TAG). Although so far intransigent to purification, PAP1 has been shown to be Mg²⁺ dependent and heat and sulphhydryl reagent sensitive.

The second class of phosphatidate phosphatase (PAP2) comprises a family of six-transmembrane-domain glycosylated proteins localized in the plasma membrane. Unlike PAP1, which is highly specific for the Mg²⁺ salts of PA and LPA, PAP2 degrades a number of lipid phosphates, including S1P, ceramide-1-phosphate (C1P), as well as PA and LPA, does not require specific ions, and is heat and sulphhydryl reagent resistant. It has consequently been suggested that PAP2 be designated lipid phosphate phosphatase (LPP) (25, 107, 260). Although not yet accepted by scientific reference bases such as PubMed, which still uses phosphatidic acid phosphatase, this recently suggested nomenclature emphasizes the molecular distinction between PAP1 and PAP2, thereby eliminating past confusion regarding these enzymes. In addition, the new enzyme system (Table 1) greatly simplifies current literature regarding the variously named LPP isoforms that have only recently been discovered. Therefore, this rationalizes nomenclature and will be used here with the previous term included in parentheses as the different isoforms are introduced. The present review will summarize our present understanding of PAP1 and LPP and their physiological functions, with particular emphasis on pulmonary phosphatidate phosphohydrolase. For brevity, liberal use will be made of earlier reviews.

	PAP1 AND LPP ENZYMATIC ACTIVITIES

TOP ABSTRACT INTRODUCTION PAP1 AND LPP ENZYMATIC... SUBSTRATES FOR LPP CDNAS FOR LPPS RECEPTORS AND OTHER... LPP IN SIGNALING PLATFORMS CELLULAR FUNCTIONS FOR PAP1... FINAL CONSIDERATIONS REFERENCES

Phosphatidic acid phosphohydrolase was initially described in plants by Morris Kates in 1955 (123). Subsequently, the dephosphorylation of PA was reported in a number of mammalian tissues (235). Original interest in this activity was related to its role in phospholipid and neutral glycerolipid biosynthesis. PA is situated at an important branch point in the Kennedy de novo pathway (124, 125), leading to the formation of either anionic (acidic) phospholipids or zwitterionic and neutral lipids (Fig. 1). DAG generated by PAP1 acts as a substrate for biogenesis of PC, PE, and TAG.

View larger version (17K): [in this window] [in a new window]   Fig. 1.   The Kennedy pathway for the biosynthesis of phospholipids and triacylglycerol in animal tissues. RCO-S-CoA, fatty acyl CoA. [Modified from Kennedy (124, 125).]

After its discovery in plants (123), phosphatidate phosphohydrolase in mammals was initially characterized in the early 1960s in the brain (2), intestine (111), kidney (41), erythrocytes (92), and liver (225, 273). These investigations, which used relatively high concentrations of phosphatidate as aqueous emulsions in the absence of Mg²⁺, found that this activity (i.e., LPP) was associated with particulate fractions with little or no activity present in the high-speed cytosol. Paradoxically, despite the high phosphohydrolase activity observed with microsomes and mitochondria, radioactivity from labeled glycerol-3-phosphate accumulated in PA (97, 233, 239, 244). Further investigations revealed the cytosolic fraction contained a factor that stimulated production of DAG and TAG from labeled phosphatidate. Although initially controversial, investigations with liver, intestine, adipose tissue, and lung eventually revealed that the cytosolic factor was soluble PAP1 (106, 112, 161, 234, 265; reviewed in Refs. 22, 23, 194).

It is instructive to consider the basis of the initial inability to identify phosphatidate phosphohydrolase in cell supernatants. In retrospect, a primary reason was that the chemical PA substrate used was generated by plant phospholipase D (PLD), which requires Ca²⁺. This divalent cation inhibits Mg²⁺-independent PAP1 but has a much smaller effect on LPP activity. Calcium phosphatidate is very stable, and until the development of the ion-exchange resin Chelex 100 (203), it was virtually impossible to remove this divalent cation completely from PA preparations. Calcium phosphatidate can form hexagonal II (H_II) phase where the phosphate groups would be internalized away from the dispersing buffer (95), thereby limiting access to these groups for enzymatic hydrolysis. A second reason for the initial failure to detect Mg²⁺-dependent PAP1 activity is that LPP possesses higher K_m and V_max. As a result, with most tissues, Mg²⁺-dependent (i.e., PAP1) activity could not be detected, especially under conditions more or less optimized for LPP. In addition, many studies employed detergents such as Tween 20 and Triton X, which stimulate LPP but markedly inhibit the more sensitive PAP1 (263).

PAP1 activity was first recognized using radiolabeled PA biosynthetically generated on microsomal or mitochondrial membranes. The ability of the cytosolic PAP1 to hydrolyze biosynthetically generated, but not aqueously dispersed, PA was attributed to the need for substrate presentation in a membrane-associated form (33). This "natural" form, which persisted in autoclaved microsomes, was eventually found to be related to the presence of Mg²⁺ during generation of the membrane-associated substrate; i.e., the Mg²⁺ salt of PA was formed, and this was the true substrate for PAP1. Subsequently, it was shown that mixing PC with the radiolabeled PA markedly increased the PAP1 activity. Thus the requirement for a natural membrane-associated substrate could be achieved by preparing vesicles of mixed PA/PC and providing optimal Mg²⁺ (23, 194, 242, 264, 265).

Such chemically defined substrates permitted investigators to demonstrate that PAP1 was predominantly localized to the cytosol but translocated to the ER and perhaps other organelles under the influence of free fatty acids, acyl CoAs, and possibly other agents. Washing microsomes with high salt virtually abolished their ability to generate DAG, TAG, or PC from radioactive glycerol-3-phosphate. These biosynthetic capacities were restored by adding back the washings or partially purified PAP1 (265). These investigations showed that PAP1 was involved in glycerolipid biosynthesis. They also demonstrated that the Mg²⁺-independent, heat-stable LPP activity, which was not removed by washing, could not contribute significantly to the de novo pathway shown in Fig. 1. The biological role of the abundant LPP activities retained by microsomal fractions remained unclear until Brindley and colleagues (105, 258, 259) demonstrated localization of this Mg²⁺-independent, N-ethylmaleimide (NEM)-insensitive enzyme in hepatocyte plasma membranes, suggesting a potential role in signal transduction. This group also demonstrated that, in addition to PA and LPA, LPP was also capable of hydrolyzing S1P and C1P, which are potent signaling lipid phosphates.

	SUBSTRATES FOR LPP

TOP ABSTRACT INTRODUCTION PAP1 AND LPP ENZYMATIC... SUBSTRATES FOR LPP CDNAS FOR LPPS RECEPTORS AND OTHER... LPP IN SIGNALING PLATFORMS CELLULAR FUNCTIONS FOR PAP1... FINAL CONSIDERATIONS REFERENCES

This section will briefly discuss, with reference to helpful reviews, the enzymatic mechanisms responsible for producing lipid substrates for LPP. As indicated above, present evidence, albeit circumstantial, indicates LPP does not have access to LPA or PA generated during de novo glycerolipid synthesis via the Kennedy pathway (22, 194, 260) (Fig. 1). Neither S1P nor C1P is considered a biosynthetic intermediate. PA is generated on plasma membranes and other cellular membranes by PLD and DAG kinase in response to cytokines, hormones, neurotransmitters, and related agonists (61, 69, 113, 140, 250, 255). LPA and S1P are present in serum and therefore available to LPP on plasma membranes. C1P can also be generated on plasma membranes in response to certain signals (24, 196).

PLD. PLD was initially discovered in plants by Hanahan and Chaikoff in 1947 (84). Early studies (279) demonstrated that in addition to hydrolyzing phospholipids by introducing H⁺ and OH (i.e., H₂O) across the phosphodiester bond, PLD catalyzes a transphosphatidylation reaction in which a short-chained alcohol replaces water to generate a new phosphatidylalcohol instead of PA and releases the original alcohol, which is choline in the case of PC.

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Schematic representation of the phosphatidylinositol-specific phospholipase C (PI-PLC) and phospholipase D (PLD) signaling pathways. 1) PI-PLC pathway: interaction of an agonist (e.g., ATP) with an appropriate receptor (e.g., P2u purinergic receptor) leads to activation of PI-specific PLC. PI-PLC hydrolyzes phosphatidylinositol-4,5 bis phosphate (PIP2), releasing inositol tris phosphate (IP3) and diacylglycerol (DAG). IP3 releases Ca2+ from intracellular stores, resulting in activation of Ca2+ calmodulin-dependent protein kinase (CM-PK). DAG activates certain PKC isoforms. 2) PLD pathway: DAG-activated PKC can activate PLD. PLD hydrolyzes membrane phosphatidylcholine (PC) to phosphatidic acid (PA). The PA thus formed is degraded by lipid phosphate phosphohydrolase (LPP) to produce DAG and inorganic phosphate. DAG so produced further activates PLD, increasing the pathway, leading to a prolonged increase in PKC activation and resulting in the physiological response.

DAG kinase. In addition to PLD, LPP could act on the PA produced by DAG kinase. A primary role for DAG kinase appears to be the termination of PKC activation by removing DAG generated by PI-PLC (250, 255, 261). The PA produced in this manner is used for formation of CDP-diacylglycerol and PI synthesis similar to the de novo pathway in Fig. 1. The resulting PI is subsequently sequentially phosphorylated to PIP₂ (13-16, 53).

Lipid phosphate growth factors. It has long been apparent that in addition to peptide growth factors such as insulin-like growth factors, serum contains phospholipid growth factors, including LPA, S1P, and related compounds (3, 76, 89, 164, 199). LPA and S1P, present in serum at micromolar concentrations, are released by activated platelets, stimulated leukocytes, and cells in apoptosis that shed microvesicles from their plasma membranes. Calcium influx activates scramblase, a protein that accelerates "flip-flop" of phospholipids in a nonspecific manner that eliminates the membrane asymmetry inherent to healthy cells (17, 171, 200). The loss of membrane phospholipid asymmetry contributes to microvesicle shedding. Calcium also stimulates PLD activity leading to the formation of PA. In addition, conditions promoting platelet and leukocyte activation can also stimulate release of group II secretory PLA₂. This phospholipase, which has a high affinity for acidic phospholipids such as PA (236), generates LPA and other lysophospholipids. The manner in which PA generated by PLD on the inner leaflet of the plasma membrane becomes available for PLA₂ action on the outer leaflet is not known, but scramblase may have a role.

	CDNAS FOR LPPS

TOP ABSTRACT INTRODUCTION PAP1 AND LPP ENZYMATIC... SUBSTRATES FOR LPP CDNAS FOR LPPS RECEPTORS AND OTHER... LPP IN SIGNALING PLATFORMS CELLULAR FUNCTIONS FOR PAP1... FINAL CONSIDERATIONS REFERENCES

Using NH₂-terminal amino acid sequence from a 35-kDa LPP purified from porcine thymus, Kai et al. (118) observed high conservation with internal sequence from Hic 53, a mouse partial cDNA originally identified as a H₂O₂-inducible gene (57). The predicted amino acid sequence of mouse LPP1 demonstrated 34% identity with Drosophila Wunen and 48% identity with rat Dri 42. Wunen appears involved in guiding germ cell migration from the developing gut toward overlying mesenchyme early in Drosophila embryonic development (283). A second LPP cDNA, Dri 42 (LPP3/PAP2b), was identified as a novel gene upregulated during the differentiation of rat intestinal mucosa epithelium (12). Subsequently, Kai et al. (117) cloned the cDNAs for human LPP1 and LPP3 cDNAs from Hep G2 cells, and Hooks et al. (93) reported the cloning of LPP2 (PAP2c) from expressed sequence tags. Roberts et al. (206) cloned human LPP1, LPP2, and LPP3 and showed these clones expressed active enzyme in human embryonic kidney (HEK 293) cells. The human cDNAs showed 40-60% sequence identity to each other but arise from distinct genes, with the human expressing LPP1 being located on chromosome 5 (5q11), the gene for LPP2 on chromosome 19 (19p13), and the gene for LPP3 on chromosome 1 (1pter-p22.1). Additionally, Leung et al. (135) reported an apparent alternate splice variant, LPP1a (PAP2a-1), of human LPP1 from a human lung cDNA library.

Apparent differences in substrate specificity between the Mg²⁺-independent NEM-resistant phosphatidate phosphohydrolase activities in rat lung (174) and liver (259) led Nanjundan and Possmayer (175) to apply reverse transcriptase-polymerase chain reaction (RT-PCR) to clone the pulmonary isoforms from rat lung and type II cell RNA. cDNAs were obtained for LPP1, three apparent alternate splice variants of LPP1 (LPP1a, LPP1b, and LPP1c), and LPP3. These rat LPP cDNAs exhibit ~50% identity at the amino acid level. Attempts to demonstrate LPP2 in rat lung using PCR primers based on the human sequences were unsuccessful. Human LPP2 expression appears to be limited to the brain, pancreas, and placenta. It has recently been found that in the mouse, LPP2 is expressed in the lung, liver, and kidney (282). As a result LPP2 distribution in rat lung must be re-evaluated.

LPP1a, LPP1b, and LPP1c appear to be splice variants of LPP1. An inspection of the cDNAs for LPP1 and these variants suggests four regions: I, IIA, IIB, and III. As illustrated in Fig. 3, all four cDNAs contain regions I and III. LPP1 cDNA, containing regions I, IIB, and III, and LPP1a cDNA, containing regions I, IIA, and III, encode for predicted proteins of 282 and 283 amino acids, respectively. The predicted proteins would contain six transmembrane domains with the NH₂ and COOH termini on the cytoplasmic sides of plasma and other membranes and a consensus N-linked glycosylation site at positions 142 (LPP1) and 143 (LPP1a) of the second extracellular loop. LPP3 cDNA codes for a protein of 308 amino acids with a consensus N-linked glycosylation site at Asn¹⁶⁸. Rat lung LPP1 and LPP3 cDNAs are virtually identical to the previously reported cDNAs for LPP1 in rat liver and LPP3 in rat intestine (Dri 42), and rat lung LPP1a shows 80% amino acid sequence identity to the human isoform. LPP1b and LPP1c are novel isoforms that predict truncated forms of phosphatidate phosphohydrolase.

View larger version (38K): [in this window] [in a new window]   Fig. 3.   Graphical representation of LPP1 and LPP1 variant cDNAs cloned from rat lung. The cDNAs are divided into four regions (I, IIA or IIB, and III). Protein alignments are shown below the cDNAs; the red boxes represent transmembrane domains, and the violet region represents the domains of the active site. All cDNAs have regions I and III. LPP1 has region IIB, whereas LPP1a has region IIA. With the cDNA for LPP1b, regions IIB and IIA are replaced by a single "G," noted with an asterisk. This leads to a frame shift and a truncation predicting a peptide of 30 amino acids. With cDNA LPP1c, regions IIA and IIB are present, except for the deletion of a "G" at the IIA/IIB interface. This results in a frame shift and predicted truncation at 76 amino acids. [From Nanjundan and Possmayer (175).]

LPPs contain a novel conserved phosphatase sequence motif, K(X)₆ RP- (X_12-54)-PSGH- (X_31-54)-SR(X)₅ H(X)₃D, that is shared among several yeast lipid phosphatases, some mammalian phosphatases, including mammalian glucose-6-phosphatase, chloroperoxidases, and some bacterial nonspecific acid phosphatases (Fig. 4). Stukey and Carman (241) have proposed a model that includes 1) nucleophilic attack of the substrate's phosphoryl group by the histidine of active site domain 3 and production of a phosphoenzyme catalytic intermediate; 2) involvement of the conserved arginine residues of active site domains 1 and 3 in hydrogen bonding to equatorial phosphoryl oxygens; and 3) participation of the histidine of active site domain 2 in protonation of the substrate leaving group. Active site domains are indicated such that domains 1 and 2 are located on the second extramembrane loop, whereas the third active site region is present on the third extracellular loop of the protein (Fig. 4). Zhang et al. (284) demonstrated that mutating Lys¹²⁰, Arg¹²⁷, Pro¹²⁸, Ser¹⁶⁹, His¹⁷¹, Arg²¹⁷, or His²²³ from mouse LPP1 results in a 95% or greater loss of activity, whereas altering Gly¹⁷⁰ reduces activity by ~66%. Altering the putative glycosylation site Asn¹⁴² to Gln resulted in a ~4-kDa reduction in molecular mass. The nonglycosylated product remained enzymatically active as occurs with enzymatic removal of the LPP glycan.

View larger version (46K): [in this window] [in a new window]   Fig. 4.   Predicted transmembrane (TM) topology of LPP1 and its active site. Mutation of the amino acids indicated in black results in severe reduction in LPP activity. Asparagine 142, indicated in gray, represents the glycosylation site. [Modified from Zhang et al. (284).]

As indicated above, LPP1b and LPP1c code for truncated proteins. The LPP1b isoform did not contain either cDNA region IIA or IIB (Fig. 3), except for the insertion of a single G nucleotide at the I/III boundary. This insertion resulted in a frame shift leading to early termination of the predicted protein at 30 amino acids. The predicted protein, if formed, would possess an NH₂-terminal extramembrane domain but would possess only part of the first transmembrane domain and differ in 10 of the last 11 amino acids at the predicted COOH terminus. The cDNA for LPP1c contains regions IIA and IIB, except for the deletion of a "G" nucleotide at the IIA-IIB boundary. This deletion also results in a frame shift, leading to early termination at 76 amino acids. Neither LPP1b nor LPP1c would possess an active site.

It must be stressed that whether proteins corresponding to LPP1b or LPP1c nucleotide sequences are produced in the lung or any other tissue remains unknown. LPP1c mRNA is rare, but LPP1b mRNA is relatively abundant. RNAase protection assays reveal that in the rat lung, the mRNA for LPP1b is ~25% of the level of LPP1, whereas LPP1a mRNA is only 10% of LPP1 mRNA levels (L. Zhao and F. Possmayer, unpublished results). RT-PCR studies have shown that LPP1b is relatively high in fetal lung but declines after birth, whereas LPP1a mRNA is low before birth but increases toward adulthood. RT-PCR, using primers designed to distinguish between LPP1, LPP1a, and LPP1b, shows LPP1 was present in rat liver, intestine, kidney, spleen, uterus, heart, and brain, as well as in the lung (175). LPP1b mRNA was particularly high in the lung, brain, intestine, and heart, whereas LPP1a was relatively high in the kidney, intestine, spleen, lung, and liver.

The observed variations in LPP1b expression relative to LPP1 and LPP1a are consistent with biological regulation, but at present physiological relevance remains unclear. It is noteworthy that precedent exists for mRNAs coding for truncated forms of PLD, DAG kinase, and PLC-, which, as in the case of LPP1b, would lack catalytic functions. For example, Steed et al. (238) have reported PLD2c, a variant of active PLD2a, contains a 56-bp insert resulting in premature termination during RNA translation. Expression of the human PLD2c variant mRNA relative to PLD2a is high in the liver and heart but low in the brain and skeletal muscle. In addition, splice variants of human PLD1 and PLD1b were found that possess 114-bp deletions resulting in loss of an essential HKD transphosphatidylation motif, resulting in an inactive enzyme.

A second potentially important example of isoform inactivation through alternate splicing involves DAG kinase. As indicated earlier, DAG kinase functions in the regeneration loop of the PI signaling cycle, where it produces PA for PI synthesis via CDP-DAG but can also downregulate PKC activity by removing DAG. The human retina demonstrates high expression of a phosphatidylserine-dependent DAG kinase- isoform (116). A low level of this full-length active DAG kinase- is also expressed in brain. Most other human tissues express very low levels of the mRNA for the active DAG kinase- but express varying amounts of an mRNA coding for a catalytically inactive form of this enzyme truncated by 25 amino acids within the catalytic domain. Kai et al. (116) have suggested that this represents a biological mechanism for downregulating enzymatic expression of this protein at the mRNA splicing level.

A third example of enzymatic downregulation occurs with an alternate splice form of PKC, designated PKC-III, which is expressed in rat testes (253). PKC-III possesses an 83-nucleotide insert within the caspase-3 recognition domain, which produces premature truncation such that the protein retains regulatory segments but lacks the catalytic domain. The parental PKC-I isoform is cytosolic and translocates to the plasma membrane upon phorbol ester stimulation. The catalytic domain-deficient PKC-III is predominantly localized on plasma membranes and is only slightly more directed from cytosol to peripheral membranes by phorbol esters.

These three precedents all involve phosphatidate metabolism. PA, a substrate for plasma membrane LPP, is generated by DAG kinase and PLD, whereas the LPP product DAG activates conventional and nonconventional PKCs (50). It is possible that the LPP1b isoform could act to control LPP activity by sequestering activation and/or inactivation signals, as has been suggested by Ueyama et al. (253) for PKC-. It is also conceivable that truncation via RNA splicing could act as a physiological mechanism for reducing LPP1 and/or LPP1a isoform mRNA expression in lung and other tissues. Attempts at overexpressing LPP1, LPP1a, LPP1b, and LPP3 in mouse lung epithelial (MLE) cells using the pTracer-CMV2 expression vector were unsuccessful due to cell death, possibly by apoptosis (M. Nanjundan, R. White, A. Brickenden, and F. Possmayer, unpublished results). This may have been related to toxic effects resulting from the particularly strong cytomegalovirus promoter during expression with this vector. Alternatively, transient expression of LPP cDNAs could have led to increases in bioactive sphingolipids implicated in programmed cell death. However, transient expression of rat LPP1, LPP1a, and LPP3 in HEK 293 cells resulted in increased LPP activity assayed with PA without any suggestion of apoptosis (175), as has been observed by others using LPP cDNAs from other species (118, 206). Transient expression of LPP1b in HEK 293 cells does not affect endogenous LPP activity and had no apparent effect on cellular viability (175). It is clear that further studies using lung cells are needed to determine the physiological implications of this LPP1 variant.

	RECEPTORS AND OTHER TARGETS/LIGANDS FOR LIPID PHOSPHATES

TOP ABSTRACT INTRODUCTION PAP1 AND LPP ENZYMATIC... SUBSTRATES FOR LPP CDNAS FOR LPPS RECEPTORS AND OTHER... LPP IN SIGNALING PLATFORMS CELLULAR FUNCTIONS FOR PAP1... FINAL CONSIDERATIONS REFERENCES

As indicated in SUBSTRATES FOR LPP, serum contains a number of lipid mediators, including LPA and S1P. Biological responses to these phospholipid growth factors include 1) calcium mobilization, 2) stimulation of cellular proliferation, 3) inhibition of apoptosis, 4) aggregation of platelets, 5) formation of stress fibers, 6) neurite contraction, 7) contraction of smooth muscle cells, 8) regulation of cell-cell interactions, 9) stimulation of cell mobility and 10) chemotaxis (3, 76, 165).

Endothelial differentiation gene receptors. The obvious physiological importance of the above indicated responses prompted attempts to clone putative receptors linking the extracellular presence of these lipid mediators to intracellular events. A number of difficulties were encountered related, in part, to the lipid nature of these agonists, their tendency to bind avidly to Ca²⁺, and even their variable presence in serum used for cell culture (3, 38, 89). Therefore, the demonstration by Chun et al. (38) that LPA was a natural ligand for ventricular zone-1 (vzg-1) receptor, so named because of its high expression in that particular neural proliferative zone in embryonic cortex, marked an important advance. Because it appeared likely that the LPA receptor vzg-1 was a member of the newly discovered endothelial differentiation gene (EDG) family of G protein-coupled receptors (i.e., EDG-2), this initial discovery facilitated identification of other lipid mediator ligands. The EDG-1 receptor was initially cloned by Hla and Maciag (90) in 1990 as a phorbol ester-inducible mRNA in human umbilical vein endothelial cells, which under the influence of serum, differentiate into capillary-like networks. S1P was identified as a potent mitogen arising from the phosphorylation of sphingosine (88). This lipid phosphate was later recognized as the polar agent in serum that promoted EDG-1-overexpressing HEK 293 cells to form tubular, capillary-like networks (88) and was subsequently found to act as a chemotactic factor in angiogenesis (59, 196). LPA and sphingosylphosphorylcholine were also shown to bind EDG-1 but with low affinity. Expression cloning studies using the serum response element and the jellyfish Ca²⁺-interacting protein apoaequorin confirmed LPA as a ligand for EDG-2 (3). Other members of this seven-transmembrane G protein-coupled family were soon found by various approaches. To date, three LPA high-affinity receptors (EDG-2, EDG-4, and EDG-7) and five S1P-responsive receptors (EDG-1, EDG-3, EDG-5, EDG-6, and EDG-8) have been identified. At a recent conference (Federation of American Societies for Experimental Biology Conference on Lysophospholipids; Tucson, AZ, June 9-14, 2001; Edward J. Goetzl, Timothy Hla, and Gabor Tigyi, organizers), there was a consensus to adopt a more logical nomenclature referring to these receptors in terms of their affinities, such that the receptors for LPA would be designated (with the old nomenclature after the slash) LPA₁/EDG-2, LPA₂/EDG-4, LPA₃/EDG-7, whereas the receptors for S1P would be designated S1P₁/EDG-1, S1P₂/EDG-5, S1P₃/EDG-3, S1P₄/EDG-6, and S1P₅/EDG-8 (Table 1). Adoption of the new nomenclature would conform with the International Union of Pharmacology recommendations and will alert readers to the chemical nature of the ligand. It appears likely that other receptors will be found for lipid phosphates (245).

Intracellular ligands for lipid phosphates. Although sphingosine kinase can be secreted, this enzyme is primarily present within cells (88). Sphingosine kinase activity on the inner aspect of the plasma membrane and possibly other cellular membranes would lead to the intracellular formation of S1P. The manner in which S1P accesses the outer aspect of the cell and interacts with S1P receptors is not known. Although S1P was originally thought of as an intracellular second messenger, a number of the actions attributed to intracellular S1P appear best explained through autocrine interactions with extracellular receptors (88, 196). However, as stressed by Pyne and Pyne (196), the location of S1P-mediated events is not always readily identifiable. The lack of or partial inhibition by pertussis toxin, which inhibits G_i/o but not G_s nor G_q, is inconclusive because neither the S1P₂ nor the S1P₄ receptor is affected by this toxin and their presence is difficult to exclude. The effects of the sphingosine analogs D,L-threo-dihydrosphingosine and N,N-dimethylsphingosine, which inhibit sphingosine kinase, are also ambiguous since these inhibitors can affect certain PKCs (196). Consequently, attributing potential second messenger roles for S1P must await identification of intracellular molecular targets and/or development of specific inhibitors for S1P receptors.

	LPP IN SIGNALING PLATFORMS

TOP ABSTRACT INTRODUCTION PAP1 AND LPP ENZYMATIC... SUBSTRATES FOR LPP CDNAS FOR LPPS RECEPTORS AND OTHER... LPP IN SIGNALING PLATFORMS CELLULAR FUNCTIONS FOR PAP1... FINAL CONSIDERATIONS REFERENCES

The fluid-mosaic membrane model, as suggested by Singer and Nicolson (229), allowed for specialized membrane regions, such as caveolae (literally, "little caves"). Caveolae are nonclathrin-coated vesicular invaginations of the plasma membrane, often with a flask-like shape and possessing diameters of 50-100 nm (4, 139, 220). Large numbers of caveolae have been reported in endothelial cells, adipocytes, fibroblasts, and smooth muscle cells. Caveolae have not been detected in alveolar type II cells but are abundant in type I cells.

Caveolae are attached to the plasma membrane by a short neck, but they can also appear as flat pits. These latter structures could represent early stages of invagination (4). Caveolin, the principal structural component of caveolae, is a ~22-kDa scaffolding protein that acts as a marker for these membrane structures. Mammalian cells possess four forms of caveolin: caveolin-1a, caveolin-1b, caveolin-2, and caveolin-3 (220).

Interest in caveolae increased markedly with evidence that these are not fixed entities but dynamic structures that can bud from the plasma membrane and that such internalization is under the control of the molecular transport machinery responsible for vesicular budding, docking, and fusion (179, 223). It has further become evident that caveolae-like structures are involved in a number of cellular functions, including:

1) Transcellular transport, where caveolar vesicles move between two surfaces of the cell, transporting ions, small molecules, and low-molecular-weight macromolecules. This process is inhibited by NEM and cholesterol-binding agents such as filipin and appears to require GTP (4). Transmembrane transport has been studied extensively in endothelial cells but appears to function in placenta epithelial cells as well and may occur in alveolar type I cells (30, 178).

2) Potocytosis. This involves nonclathrin-coated pit transport of certain ions, low-molecular-weight molecules or macromolecules from the cell surface to the cytoplasm, to internal organelles such as the ER (4), and from one plasma membrane to another plasma membrane site (e.g., apical to basolateral). Folate transport, which involves a glycerolphosphoinositol (GPI)-anchored receptor, is a well-known example of potocytosis (160, 205), but Fe²⁺, Ca²⁺, alkaline phosphatase, and insulin are also transported in this manner (4, 85, 168). Caveolae-based endocytosis can be employed opportunistically by toxins, viruses, bacteria (227), and trypanosomes (63, 232) to gain access to the cell's interior.

3) Cholesterol transport. Caveolae appear involved in directing intracellular cholesterol transport (4, 64, 131, 141, 222). Caveolins bind cholesterol, and the scavenger receptor class B is a fatty acylated glycoprotein directed to caveolae, which mediates selective cholesterol uptake from the high-density lipoproteins. This contrasts with the low-density lipoproteins receptor-mediated transfer known to occur via clathrin-coated pits and vesicles (99).

4) Caveolae participation in signal transduction. The ever-increasing number of signaling molecules found concentrated in caveolar and similar domains has generated a re-evaluation of signaling mechanisms. It has become apparent that a large number of signaling molecules and, consequently, their associated signaling cascades are present in or can be recruited to caveolae. The current list includes receptor [e.g., EGF, platelet-derived growth factor (PDGF), insulin receptor] and nonreceptor (i.e., src family members) tyrosine kinases (32, 85, 144, 268). G protein-coupled receptors and their associated G protein effectors and targets (somatostatin receptor, adenosine receptor, heterotrimeric G proteins, adenylcyclase, PI-3 kinase) (4, 6, 185) and membrane transporters (aquaporin-1, Ca²⁺ ATPase, H⁺ ATPase, and IP₃ receptor) (4, 103, 185) have been localized to caveolae. Recent evidence indicates caveolins can interact directly with and regulate the activity of a number of proteins, including endothelial nitric oxide synthetase, sonic hedgehog receptor, and src tyrosine kinase family members (182, 220, 230).

In addition to caveolin, caveolae are characterized by being highly enriched in cholesterol, sphingomyelin, and glycosphingolipids (70, 184; also reviewed in Ref. 27). As implied by the fluid-mosaic model, most of the lipid bilayer that forms the bulk of the biological membranes is in a fluid state at biological temperatures (229, 286). This is due to the presence of unsaturated (i.e., double bond-containing) fatty acyl groups at the sn-2 position of most cellular glycerophospholipids. Sphingomyelin and glycosphingolipids tend to possess longer, more saturated (i.e., no double bonds) fatty acyl groups, and bilayers composed of these compounds adopt an immobile gel (i.e., solid-like) phase at biological temperatures. Membrane cholesterol tends to dissolve into these regions, creating a liquid-ordered phase with fluidity intermediate between gel-ordered and mobile-fluid phase. Due to their lipid composition, caveolae can be isolated on the basis of buoyant density on suitable gradients (216, 232). The most commonly applied technique involves collecting those microdomains that are insoluble in 1% Triton X-100 at 0-4°C on 0-30% sucrose gradients (35, 163).

Overexpression of caveolin-1 in certain cells leads to an increase in the number of caveolae (68, 231). However, detergent-insoluble cholesterol-sphingomyelin-glycosphingolipid-enriched domains (DIGs) can readily be isolated not only from synthetic liposomes but also from cells such as those from the hematopoietic system and alveolar type II cells, which do not express detectable amounts of caveolins. Sequestering membrane cholesterol with saponin, filipin, or methylcyclodextrin leads to the loss of caveolar invaginations and the detergent-insoluble microdomains (4, 27, 146). It should be apparent that DIGs isolated from cells or tissues could arise either from caveolin-containing invaginations (or vesicles) or from caveolin-lacking cholesterol-sphingomyelin-rich microdomains. Further, both kinds of DIGs could arise from membranes other than the plasma membrane. The noncaveolin-containing microdomains would be flat and lack distinct morphological features by electron microscopy. Such caveolin-deficient cholesterol-sphingomyelin-glycosphingolipid-rich microdomains are often referred to as "rafts" [although readers should be cautious because some literature considers caveolae to be a particular type of raft, so that "rafts" refers to both caveolae (i.e., vesicular) and noncaveolin-containing (i.e., morphologically indistinct) cholesterol-sphingolipid-rich regions]. The concept of membrane rafts is still controversial because DIGs could be generated during detergent processing of the plasma and other membranes. However, considerable evidence has accumulated demonstrating that certain membrane proteins cluster in biological membranes, consistent with the presence of specialized regions (4, 26, 27).

The dynamics between noncaveolin-containing rafts and caveolin-dependent caveolae are still being clarified. Recent studies using model green fluorescent protein (GFP) and GFP-yes chimeric proteins (yes is a src-family member) indicate fatty-acid modifications such as NH₂-terminal acylation contribute to protein sorting into cholesterol-sphingolipid-rich domains, but protein-protein interactions are also important (153, 154). Lipid rafts appear enriched in GPI-anchored proteins, including alkaline phosphatase, 5'-nucleotidase, CD14, folate receptor, and the prion protein (4, 94, 169). The heterotrimeric G proteins G_i and G_s concentrate in lipid rafts, whereas G_q appears to interact specifically with caveolin and is retained in caveolae (181). This agrees with other evidence showing that, although rafts and caveolae share a requirement for sphingolipids and cholesterol, they are biochemically and functionally as well as morphologically distinct.

Evidence for pulmonary rafts/caveolae. DIGs isolated from whole rat lung contain a high proportion of total cholesterol, sphingomyelin plus PC (measured as phosphocholine-containing lipids), and caveolin-1 (176). Rat lung DIGs are also enriched in GPI proteins, 5'-nucleotidase, and alkaline phosphatase. Although PKC-, PKC-µ, and PKC-II were predominantly present in detergent-soluble fractions, a small proportion of PKC-II and PKC-µ colocalized with caveolin-1. Rat lung DIGs account for ~0.25% of total tissue protein.

View larger version (30K): [in this window] [in a new window]   Fig. 5.   Separation of caveolae from lipid rafts in mouse lung epithelial (MLE) 15 cells. Caveolae were purified from other lipid-rich microdomains using cationic colloidal silica particles. A: the specific activity of LPP is shown as PI release in nmol · min1 · mg1 from PA in various membrane fractions: total membranes (TM), plasma membrane (P), caveolae (V), plasma membranes minus caveolae (P  V), and remaining lipid raft microdomains (LR). B: immunoblot analysis with anticaveolin-1 for the fractions in A (1 µg protein/fraction). C: immunoblot analysis with anti-LPP1 (1 µg protein/fraction). Control (ct) represents 50 µg MLE15 total membranes. D: immunoblot analysis with anti-LPP3 (1 µg protein loaded per fraction). Ct represents 10 µg of extracts from SF9 cells overexpressing LPP3. [Modified from Nanjundan and Possmayer (176).]

	CELLULAR FUNCTIONS FOR PAP1 AND LPP

TOP ABSTRACT INTRODUCTION PAP1 AND LPP ENZYMATIC... SUBSTRATES FOR LPP CDNAS FOR LPPS RECEPTORS AND OTHER... LPP IN SIGNALING PLATFORMS CELLULAR FUNCTIONS FOR PAP1... FINAL CONSIDERATIONS REFERENCES

PAP1. This review will close with comments on potential physiological roles for phosphatidate phosphohydrolase activities in the lung, with special emphasis on areas considered important for investigation at the present time. Lung contains two distinct kinds of phosphatidate phosphohydrolase. PAP1 is an Mg²⁺-dependent, cytosolic protein that translocates to the ER and possibly other membranes under the influence of free fatty acids and acyl CoAs. Whether other cellular agents affect PAP1 activity should be determined. PAP1 functions in glycerolipid synthesis by hydrolyzing newly synthesized PA to DAG (Fig. 1). PAP1 activity is essential in lung for the de novo synthesis of phospholipids for cellular membranes and for the production of PC and 1,2-dipalmitoyl-sn-3-phosphatidylcholine for pulmonary surfactant by type II cells (42). As PAP1 has not yet been purified or cloned, it has not been possible to ascertain whether more than one form of this enzyme is present in mammalian tissues, including lung. In addition, little information is available concerning the manner in which this enzyme is regulated.