Lipid phosphate phosphatase (LPP3) and vascular development

Abstract

Lipid phosphate phosphatases (LPP) are integral membrane proteins with broad substrate specificity that dephosphorylate lipid substrates including phosphatidic acid, lysophosphatidic acid, ceramide 1-phosphate, sphingosine 1-phosphate, and diacylglycerol pyrophosphate. Although the three mammalian enzymes (LPP1-3) demonstrate overlapping catalytic activities and substrate preferences in vitro, the phenotypes of mice with targeted inactivation of the Ppap2 genes encoding the LPP enzymes reveal nonredundant functions. A specific role for LPP3 in vascular development has emerged from studies of mice lacking Ppap2b. A meta-analysis of multiple, large genome-wide association studies identified a single nucleotide polymorphism in PPAP2B as a novel predictor of coronary artery disease. In this review, we will discuss the evidence that links LPP3 to vascular development and disease and evaluate potential molecular mechanisms.

The LPP family

Lipid phosphate phosphatases are integral membrane proteins with broad substrate specificity that dephosphorylate lipid substrates including phosphatidic acid (PA), lysophosphatidic acid (LPA), ceramide 1-phosphate (C1P), sphingosine 1-phosphate (S1P), and diacylglycerol pyrophosphate (DGPP) [1]. They belong to a broader class of structurally-unrelated phosphatidic acid-phosphatases (PAP) that comprise both membrane and soluble family members [2]. In humans, three genes, PPAP2A, PPAP2C, and PPAP2B, encode the enzymes LPP1, LPP2 and LPP3, respectively [3, 4]. In this review, we refer to the gene names using the PPAP nomenclature but the corresponding proteins as LPPs.

The predicted topology of the LPPs suggests that they possess six transmembrane domains, an active site comprised from at least 3 regions of the protein that localizes to the extracellular or luminal surface of the membrane, and a glycosylation site on a hydrophilic loop between the first and second active site domains (Figure 1) [2]. Mammalian LPPs form homo- and hetero-oligomers [5]. The Drosophila homolog of mammalian LPP, wunen, forms homodimers via the last C-terminal 35 amino acids, but cannot form heterodimers with wunen2 or mammalian LPP1 or LPP3 [6]. The functional significance of these interactions is not known.

Figure 1. Predicted topology of lipid phosphate phosphatases.

The LPP members are predicted to have six transmembrane spanning regions and an active site composed of regions on the extracellular or abluminal surface of the membrane.

LPPs localize to both the plasma membrane and intracellular membrane organelles, in particular the endoplasmic reticulum and Golgi apparatus [1, 2, 7, 8]. Subcellular localization of these enzymes is both dynamic and cell-specific. LPP1 and LPP3 appear to have distinct subcellular localization [9], between lipid rafts and the apical and basolateral membranes of polarized cells, which could account for their observed differences in biological functions despite their essentially identical catalytic activities. Evidence that LPPs can act on both extracellular and intracellular substrates has come from studies in which these enzymes are over expressed or inactivated in cell culture systems coupled with measurements of their substrates and products using radiolabeling or mass spectrometry based approaches.

Although the three mammalian LPP enzymes demonstrate overlapping catalytic activities and substrate preferences in vitro, the phenotypes of mice with targeted inactivation of the Ppap2 genes indicates that they have nonredundant functions. The Ppap2a gene encoding murine LPP1 has been disrupted using an exon trap insertion strategy. Mice harboring the exon trap inactivated allele appear phenotypically unremarkable [10]. Multiple tissues, including heart, kidney, lung, liver and spleen, isolated from the animals display a reduced ability to dephosphorylate exogenously provided LPA, indicating a role for LPP1 as a widely expressed LPA phosphatase. Decreased dephosphorylation of exogenous LPA by thymocytes from these LPP1 deficient mice indicate that endogenously expressed LPP1 can function as an “ecto” LPA phosphatase, at least in these cells. Mice homozygous for an insertionally inactivated allele of the Ppap2c gene encoding murine LPP2 are phenotypically unremarkable [11]. By contrast, inactivation of Ppap2b results in early embryonic lethality in part to due to failure of extra-embryonic vascular development [12].

The role of LPP3 in blood vascular development

In mice, LPP3 is first expressed in the anterior visceral endoderm, and the extra-embryonic membranes at E7.5 [12, 13]. As gastrulation proceeds, LPP3 appears around the node and the tip of the allantois at E8.0, and allantois, the developing gut, the pericardio-peritoneal canal and somites at E8.5. LPP3 is absolutely required in these tissues, as chorio-allantoic placenta do not form in its absence. By E9.5, LPP3 is present in umbilical cord and the chorionic region, and later in mid-gestation, in the apical ectodermal ridge, mesenchyme of the limb buds, and nervous system. In adult mice, expression of LPP3 is particularly prominent in lung, cerebellum and heart atrium. The dynamic and tissue-specific expression pattern may reflect the importance of LPP3 in specific tissues during development.

A critical role for LPP3 in vasculogenesis is indicated by the phenotype of mice with inherited deficiency in Ppap2b, created by deleting exon 5 that encodes a domain of the protein essential for its catalytic activity [12]. Ppap2b-null embryos die between E7.5 to 9.5 as a consequence of an inability to form extra-embryonic vasculature. The Ppap2b-null embryos fail to form chorio-allantoic placenta, and their abnormal yolk sac vascular network results in accumulation of blood cells in the yolk sac cavity. The embryos also demonstrate abnormalities in embryonic axis formation, with shortening of the anterior-posterior axis, anterior truncation and frequent duplication of axial structures [13]. We have targeted LPP3 in the vasculature by breeding mice containing a Ppap2b-floxed allele with mice expressing Cre recombinase under the control of the Tie2 promoter (Tie2-Cre) to delete exons 3 and 4 of the floxed Ppap2b gene in hematopoeitic and endothelial cells [14]. The excised exons encode the second and third transmembrane domains, the first intracellular and the second outer loop, and 12 amino acids of the fourth transmembrane segment. Global Cre-mediated deletion in mice phenocopies complete Ppap2b deletion [13]. Mice with Tie-Cre mediated deletion, which lack LPP3 in endothelial and some hematopoeitic cells, die embryonically with a milder but similar defect in vasculogenesis as is observed in mice with global lack of Ppap2b [12]. Consistent with the observations that LPP3 is essential for normal vascular development, allantois explants from Ppap2b-null embryos fail to organize endothelial cells into cords. Finally, LPP3 expression is also upregulated as lymphatic endothelial cells organize into capillary-like structures in collagen matrix in vitro. siRNA-targeted knock-down of LPP3 expression enhances capillary formation, suggesting that the protein negatively regulates the process [15]. At the present time, it is not known if LPP3 catalytic activity is required for normal vessel development although in the absence of other well defined non-catalytic functions for the protein this seems likely. In the following sections, we will discuss potential mechanisms by which LPP3 may affect endothelial cell function. We will focus attention on pathways that may be regulated by LPP3 catalytic activity and ways that LPP3 may influence development in phosphatase independent manners.

LPP3 as a critical regulator of lysophospholipid signaling

The bioactive lysophospholipids LPA and S1P elicit cell responses by binding to and activating distinct G-protein coupled receptors, initially classified as Edg (endothelial differentiation gene family) receptors but subsequently rationally re-named as LPA and S1P receptors. LPP3-catalyzed removal of the phosphate group of LPA and S1P renders them inactive at their receptors. Substantial evidence from cell culture experiments indicates that LPP3 can regulate extracellular signaling by lysophospholipids [2, 16, 17]. LPP3 overexpression decreases tumorigenesis and colony forming ability of ovarian cancer cells and the effects of LPP3 on colony-forming activity are substantially reversed by an LPP-resistant LPA analog, O-methylphosphothionate and a series of additional phosphatase resistant LPA analogs which have been developed as synthetic LPA receptor selective ligands [18]. These results imply that the inhibitory effects of LPP3 on tumor growth and survival are mediated at least in part by hydrolysis and inactivation of bioactive LPA. Consistent with these reports, cultures of embryonic fibroblasts derived from LPP3 knockout mice exhibit significantly increased extracellular LPA [12].

LPA in the vasculature

Extracellular, bioactive LPA is generated by the lysophopsholipase D autotaxin, which removes the choline group from lysophosphatidylcholine (LPC) and other lysophopshopholipids (Figure 2) [19]. Plasma LPA levels are <1 μM, but increase with platelet activation, due to phospholipase A1-mediated generation of LPC and association of autotaxin with activated platelet integrins [20]. These mechanisms may account for increases in local concentrations of LPA that could be associated with activation of vascular and blood cells. Given that LPA exerts growth-factor like effects and/or stimulates migration of virtually every cell type studied, enhanced production of LPA could contribute to the pathology of various disorders such as cancer, atherosclerosis and inflammation [21–25].

Figure 2. Proposed role of lipid phosphate phosphatases in regulating lysophospholipid signaling.

Bioactive lysophosphatidic acid (LPA) production is the result of autoxatin mediated hydrolysis of lysophopshatidyl choline (LPC) generated at least in part from phosphatidyl choline (PC). Extracellular LPA can act via its G-protein coupled receptors to elicit intracellular signaling and/or be degraded by surface LPPs to receptor inactive monoacyl glycerol.

A link between LPA signaling and vascular development was suggested by the phenotype of mice lacking Enpp2, the gene encoding autotaxin. Enpp2-null mice die between E9.5 – 10.5 with profound vascular defects in both yolk sac and embryo, and also aberrant neural tube formation [20, 26–28]. Replacing functional autotaxin with a catalytically inactive variant also results in embryonic lethality with the mice displaying apparently similar severe defects in the vascular development, suggesting a requirement for LPA (or other autotaxin-generated products which might be related to the ability of the enzyme to also hydrolyze various nucleotide phosphates) in embryonic vascular formation [29].

The cellular effects of LPA are mediated by at least 8 different G-protein coupled receptors: LPA₁/Edg-2, LPA₂/Edg-4, and LPA₃/Edg-7, LPA₄/GPR23/p2y9, LPA₅/GPR92, and LPA₆/p2y87, LPA₇/p2y5 and LPA₈/p2y10 [30–32]. The LPA_1–3 receptors, members of the original Edg family, share approximately 50% homology in amino acid sequences [33–36], whereas LPA₄ and LPA_6–8 were initially identified as members of the purinergic receptor family based on sequence homology [30–32, 37, 38]. Each LPA receptor couples to specific G proteins including G_12/13, G_i, G_q, and G_s, to activate diverse signaling pathways that involve small GTPase Rho, phosphoinositide 3-kinase, phospholipase C, mitogen-activated protein kinase, and adenylyl cyclase [39, 40]. Via these signaling pathways, LPA alters endothelial cell migration, proliferation and barrier stability. More recently, LPA has been proposed to serve as a ligand for the receptor for advanced glycation endproducts (RAGE) [41].

Insights into the function of LPA in vascular development come from model organisms with targeted deficiencies in specific LPA receptors. None of the LPA receptor knock-out mice phenocopy the absence of autotaxin, suggesting that the LPA receptors may play redundant roles during development and/or indicating the presence of unidentified LPA receptors or that autotaxin has functions that are independent of its ability to generate LPA. Lpar4-deficient mice exhibit impaired blood vessel formation, with up to 30% of the Lpar4-null embryos failing to survive to birth [42]. Embryos lacking Lpar4 develop hemorrhage in many organs, including the heart, skin and lung apparent between E10.5 and E18.5. In addition to subcutaneous hemorrhage, blood vessel dilatation and reduced smooth muscle cell recruitment occurs. Lpar1-deficient mice exhibit variable frontal cephalic hemorrhages, with higher penetrance of the phenotype in mice lacking both LPA₁ and LPA₂ receptors. In zebrafish, combined targeting of Lpar1 and Lpar4, but not either individual receptor, impairs vascular development [43]. Normally in zebrafish embryo development, segmental arteries sprout from the dorsal aorta and extend to the horizontal myoseptum to form longitudinal anastomotic vessels. Zebrafish embryos in which both LPA₁ and LPA₄ receptors have been targeted for knockdown by morpholino antisense oligonucleotides, extension of the segmental arteries stalls and instead of connecting with the myoseptum, abnormal connections are formed between neighboring segmental arteries. A similar defect occurs when autotaxin is targeted by morpholino antisense oligonucleotides, indicating a critical role of LPA signaling in arterial development.

In cell culture systems, LPA can promote endothelial cell migration and proliferation [44, 45]. LPA stimulated fetal bovine heart endothelial cells and β1GD25 cells migration is sensitive to both C3 toxin and pertussis toxin, suggesting that its effects are through G_12/13/Rho and G_i/Ras pathways [46]. LPA may also promote angiogenesis by suppressing endothelial cell CD36 surface expression via Ca²⁺/calmodulin-dependent Ser/Thr kinase PKD-1 [47]. In cultured allantois explants, LPA prevents the disassembly of blood vessels, supporting a role for LPA signaling in the maintenance of existing vasculature [48]. LPA can stimulate endothelial cell invasion in matrigel by inducing matrix metalloproteinase-2 expression and degradation of extracellular matrix [49]. LPA can also stimulate vascular endothelial growth factor (VEGF) expression through hypoxia-inducible factor-1-alpha dependent and independent pathways in cancer cells, suggesting a role for LPA signaling in pathologic angiogenesis [50].

S1P in the vasculature

S1P, another important substrate for LPP3, has been widely implicated in cellular differentiation, proliferation, migration and contributes to angiogenesis [51–53]. S1P is generated intracellularly from sphingosine by sphingosine kinase 1 and 2. Kinase activity is influenced by various stimuli, including VEGF, platelet-derived growth factor, tumor necrosis factor-alpha, transforming growth factor-beta, epidermal growth factor and cytokines [54]. S1P requires transport out of the cell to act on its specific receptors [55]. Plasma contains high nM concentrations of S1P [56]. Multiple cell types likely contribute to circulating S1P, including blood and endothelial cells. Both platelets and RBCs store and release S1P [57, 58]. After release from cells, plasma S1P circulates largely bound to apoliprotein M in HDL particles [59].

Analogous to LPA signaling, the extracellular effects of S1P are mediated by G-protein coupled receptors. Five S1P receptors have been identified: S1P₁/Edg1, S1P₂/Edg5, S1P₃/Edg3, S1P₄/Edg6, and S1P₅/Edg8 [60–62]. S1P_1–3 are expressed on vascular cells, including endothelial and smooth muscle cells, cardiomyocytes, and cardiac fibroblasts [63]. S1P₁ couples exclusively to G_i to activate Ras/ERK and PI 3-kinase/Akt signaling pathways, and Rho family small GTPase Rac, whereas S1P₂ couples to G_12/13 - RhoA to inhibit Rac and Akt activity [64–66]. S1P₃ can elicit signaling through multiple G proteins, such as G_i, G_q and, to a lesser extent, G_12/13 [64, 66–73]. Compared with S1P_{1 – 3}, the functions of S1P₄ and S1P₅ are less well-characterized.

Deletion of S1pr1 in mice results in embryonic lethality around E12.5 due to massive hemorrhage as a consequence of failure to recruit vascular smooth muscle cells to developing vessels [74]. Tissue-specific deletion of S1pr1 in the endothelium produces an identical phenotype, demonstrating an essential role for endothelial S1P1 [75]. In contrast, tumor angiogenesis is enhanced in S1pr2-null mice, which demonstrate increased mural cell recruitment and myeloid cell mobilization [67]. S1pr2-null mice also develop deafness that may be due to vascular abnormalities within the stria vascularis [76]. Lung endothelial cells isolated from S1pr2-null mice have enhanced Rac activity, Akt phosphorylation, cell migration, proliferation, and tube formation in vitro [77]. S1pr3-deficient mice have no obvious vascular defects [73]. However, loss of both S1P₂ and S1P₃ results in bleeding and edema in the subcutaneous regions and reduced viability at E13.5 [78]. Ultrastructural analysis of microvessels in the combined S1pr2^−/−S1pr3^−/− mice revealed abnormal endothelial cells with thin cell bodies that may result in fragile vessels prone to rupture.

S1P plays a critical role in maintaining endothelial barrier integrity. S1P₁ promotes barrier function by stabilizing tight and adherent junctions. Genetic manipulations that lower S1P levels [79] and pharmacologic antagonism of S1P₁ signaling elicits vascular leak in mice [80], whereas acute administration of S1P or S1P₁ agonists prevent vascular leak in acute lung injury models [81–83]. Interestingly, prolonged activation of S1P₁ signaling has the opposite effect and increases vascular permeability [79, 84].

Based on the results described above, it is evident that both LPA and S1P signaling pathways regulate vascular development and endothelial barrier function. However, given the redundancy and opposing effects within both the LPA and S1P receptor systems, it is not currently possible to determine if the role of LPP3 in vascular development is due to its ability to dephosphorylate LPA and/or S1P or other lipid phosphate mediators. Moreover, LPP3 may contribute to vascular cell function in a manner independent of its lipid phosphate phosphatase activity.

LPP3 and adhesive interactions

The phenotype of axis duplication that occurs in mice globally lacking Ppap2b resembles that observed in animals with altered Wnt signaling, suggesting cross-talk between the two pathways [12]. Wnt signaling is regulated by β-catenin and its interactions with p120-catenin and VE-cadherins. Tyrosine phosphorylation of β-catenin reduces its affinity for cadherins and redistributes the protein to the cytosol, where it can either be targeted for ubiquitin-mediated degradation or it can translocate to the nucleus and, via interactions with T cell factor (TCF)/ lymphoid enhancer-binding factor (LEF), alter the transcription of Wnt target genes including fibronectin, cyclin D, and LEF-1 [85, 86]. β-catenin-mediated TCF/LEF-transcription is upregulated in Ppap2b^−/− embryonic stem cells, as is expression of the Wnt target gene brachyury [87], implicating LPP3 as a negative regulator of the Wnt pathway. In HEK cells, catalytically inactive LPP3 inhibits β-catenin-mediated TCF/LEF transcription and also suppresses axis duplication, although not as effectively as active LPP3, indicating that phosphatase activity may not be required for LPP3’s effects on β-catenin [12]. However, in subconfluent endothelial cells, overexpression of LPP3 increases TCF/LEF activity and fibronectin expression, in association with a reduction in β-catenin phosphorylation. siRNA-mediated knock-down of LPP3 reduces levels of VE cadherin, p120 catenin and fibronectin and impairs branch point formation in collagen matrix. No effect of LPP3 on TCF/LEF transription was observed in confluent endothelial cells. [88].

LPP3 may also affect vessel development and function through integrin interactions. Both human and rodent LPP3 recognize α_vβ₃ and α₅β₁ integrins [89]. Human LPP3 contains an arginine-glycine-aspartate (RGD) cell adhesion sequence in its third extramembrane loop, which may mediate the interactions [90, 91]. In mouse and rat, the corresponding sequence is RGE [92]. Antibody inhibition of LPP3 blocks vascular endothelial cell growth factor (VEGF) and basic fibroblast growth factor (bFGF)-induced capillary morphogenesis in vitro [93]. These growth factors stimulate new vessel growth, in part via integrin-mediated adhesion to the extracellular matrix. Antibodies to LPP3 also inhibit endothelial cell aggregation mediated by α_vβ₃ and α₅β₁ [92]. Recognition of LPP3 by integrins does not require LPP3 catalytic activity.

LPP3 in human vascular disease

The most compelling evidence of a role for LPP3 in human disease comes from a meta-analysis of multiple, large genome-wide association studies [94]. The analysis identified heritable single nucleotide sequence variants that predict the development coronary artery disease (CAD) and used data from over 86,000 individuals. PPAP2B emerged as 1 of 13 new loci that associate with CAD. A single nucleotide polymorphism in the PPAP2B gene is associated with a 1.17-odds ratio for CAD (P= 3.81 X 10⁻¹⁹). The polymorphism lacked association with traditional risk factors such as hypertension, cholesterol, diabetes, obesity or smoking. Interestingly, CAD risk is associated with the minor allele of the polymorphism, termed rs17114036, which is located in the final intron of the gene. Intronic variants can be associated with alterations in gene expression through effects on transcription, RNA processing or stability [95–98]. Thus, it is possible that the polymorphism or a linked allele influences LPP3 expression levels. In support of this, we have interrogated publically available data sets for PPAP2B gene expression and observed lower mRNA levels in leukocytes from individuals with a polymorphism that is in linkage disequilibrium with rs17114036. Taken together, these findings suggest that LPP3 may play a role as a predictor in genetic screening for early prevention and treatment for CAD. At present, essentially no information is available about how LPP3 may contribute to the pathogenesis of CAD, however, it is possible that LPP3 expression in vascular smooth muscle and endothelial cells is necessary for normal adult vessel function analogous to its requirement in vascular development.

Conclusion

LPP3 is essential for vascular development by regulating cell proliferation, cell migration, invasion and morphology. Recent exciting findings suggest that an allelic variant of PPAP2b is associated with elevated risk of CAD and PPAP2b expression is also increased in atherosclerotic plaques. However, current understanding of LPP3 function in the vasculature is largely limited to a large body of evidence implicating its bioactive substrates, S1P and LPA in vascular development and function. Further studies to develop LPP isoform-specific inhibitors or tissue specific knockout models could facilitate the understanding and the development of effective pharmacologic approaches to treat CAD and other cardiovascular diseases.