Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Lysophosphatidic acid (LPA) is a member of the phospholipid growth factor (PLGF) family. PLGFs exert pleiotropic biological effects, such as activating platelet aggregation and affecting cell proliferation, apoptosis, migration, and cell shape (for reviews, see Goetzl et al., 2000; Tigyi, 2001). LPA elicits its biological effects through the activation of G protein-coupled receptors. In mammalian cells, three LPA-specific receptors have been identified, including LPA₁ (EDG-2), LPA₂ (EDG-4) and LPA₃ (EDG-7), all members of the endothelial differentiation gene (EDG) family (for review, see Contos et al., 2000). In addition to LPA₁, the PSP24 receptor was shown to elicit LPA-induced Ca²⁺-dependent Cl-currents in Xenopus laevis oocytes (Guo et al., 1996; Fischer et al., 1998; Kimura et al., 2001). Receptors LPA_1-3 share 50 to 54% sequence identity (Chun et al., 2002). The EDG receptor family also contains five other receptors, S1P_1-5 (EDG-1, -5, -3, -6, -8), specific for the sphingolipid PLGF sphingosine 1-phosphate (S1P) (Chun et al., 2002). The S1P receptors S1P_1-5 share 50% amino acid sequence identity and 35% identity with the LPA receptors (Chun et al., 2002), suggesting possible similarities in the characteristics of ligand recognition in these receptors. Most cells express a combination of these receptors, making it difficult to dissect the biological effects mediated by an individual receptor subtype. The need to understand the biological function of PLGF receptors and the desire to pharmacologically exploit the differences in their ligand recognition requires the development of receptor subtype-specific agonists and antagonists. Until recently, no such compounds were available. Local and general anesthetics have been reported to inhibit PLGF receptors in X. laevis oocytes (Chan and Durieux, 1997; Tigyi et al., 1997). Likewise, N-acyl serine phosphoric acid and N-acyl tyrosine phosphoric acid were shown to inhibit LPA-induced platelet aggregation, Cl-currents in X. laevis oocytes, and neutrophil adhesion to vascular endothelial cells (Sugiura et al., 1994; Liliom et al., 1996a; Hooks et al., 1998; Siess et al., 1999). However, in MDA MB231 cells, N-acyl serine phosphoric acid was shown to be a potent activator of LPA-like responses (Hooks et al., 1998), although the receptor activated by this compound was not identified. This same compound was a weak agonist of LPA₁ and LPA₃ receptors heterologously expressed in Jurkat T cells (An et al., 1998). Cyclic phosphatidic acid was also shown to inhibit LPA-induced platelet aggregation (Gueguen et al., 1999); however, it is an LPA receptor agonist in X. laevis oocytes (Liliom et al., 1996b; Fischer et al., 1998) and in cells that heterologously express EDG family LPA receptors (Bandoh et al., 2000).

Computational models for S1P₁ (Parrill et al., 2000), LPA₁, and LPA₂ (Wang et al., 2001; Sardar et al., 2002) confirmed the importance of two specific interactions between each receptor and its phospholipid ligand. One of these interactions involves ion pairing between the phosphate group of LPA and two positively charged conserved residues in the third and seventh putative transmembrane segments of the LPA receptor subfamily. A second interaction involves hydrophobic residues within the transmembrane segment of the receptor and the hydrophobic tail of LPA (Fischer et al., 2001). Based on these two points of interaction with the LPA pharmacophore, we identified dioctylglycerol pyrophosphate and dioctyl-phosphatidic acid as selective antagonists of LPA₁ and LPA₃ receptors, with an order of magnitude higher potency for LPA₃ (Fischer et al., 2001). A stereoisomer of the 2-substituted N-acyl ethanolamide phosphate LPA analog containing a bulky benzyl-4-oxybenzyl group was shown to be a dual LPA₁/LPA₃ competitive antagonist with higher inhibitory potency for LPA₁ (Heise et al., 2001).

Herein we report the synthesis and characterization of fatty alcohol phosphates (FAPs) that lack a glycerol backbone and therefore consist of only a polar phosphate head group and a hydrophobic tail and represent a minimal structure that satisfies the two-point pharmacophore introduced earlier for LPA-like structures (Fischer et al., 2001, Sardar et al., 2002). Pharmacological characterization of the ligand properties of FAPs showed an exquisite dependence on the length of the hydrocarbon chain. FAPs with 10, 12, and 14 carbons are agonists for LPA₂ and antagonists for the LPA₃ receptor. In contrast, other FAPs, with carbon chain lengths of 4, 8, 16, 18, and 22, showed no agonist effect on any mammalian LPA receptor. FAP-14 and FAP-18 were weak antagonists of the LPA₃ receptor, along with FAP-10.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Materials

Lipids were purchased from Avanti Polar Lipids (Alabaster, AL); other chemicals and fetal bovine serum were obtained from Sigma-Aldrich Chemical Co. (St. Louis, MO). Cell culture medium (Dulbecco's modified Eagle's medium) and G418 were purchased from Cellgro (Herndon, VA). Fura-2 acetoxymethyl ester was from Molecular Probes (Eugene, OR). Oocyte positive X. laevis frogs were from Xenopus I (Dexter, MI). Collagenase A was purchased from Roche Molecular Biochemicals (Mannheim, Germany). [³⁵S]GTP--S was purchased from Amersham Biosciences (Uppsala, Sweden). FLAG epitope-tagged cDNAs encoding human LPA₂ and LPA₃ receptors in pCDNA3 plasmids (Invitrogen, Carlsbad, CA) were a gift from Dr. Junken Aoki (University of Tokyo, Tokyo, Japan). S1P₁ cDNA was a gift from Dr. Timothy Hla (University of Connecticut, Storres, CT). PCDNA3.1 expression vector containing S1P₅ cDNA was a gift from Dr. Kevin Lynch (University of Virginia, Charlottesville, VA).

Chemical Synthesis of the FAP Compounds

All reagents and chromatography media were purchased from Sigma-Aldrich Chemical Co. or Fisher Scientific (Pittsburgh, PA) and were used without further purification. Thin-layer chromatography was performed on 200-µm alumina plates (Silica gel 60 Å; E.M. Science, Hawthorne, NY). Flash chromatography was performed on silica gel (60 Å, 200-425 mesh). Melting points were determined on a Thomas-Hoover capillary melting point apparatus and are uncorrected. ¹H, ¹³C, and ³¹P nuclear magnetic resonance spectra were obtained on an AX 300 spectrometer (Bruker, Billerica, MA). Chemical shifts for ¹H and ¹³C are reported as parts per million relative to tetramethylsilane. Spectra for ³¹P are reported as parts per million relative to 0.0485 M triphenylphosphate in CDCl₃. Electrospray ionization liquid chromatography/mass spectrometry was performed on a Bruker Esquire LC/MS system.

Synthesis of Phosphoric Acid Dibenzyl Ester Alkyl Esters (1-8a)

The synthesis of protected alkyl monophosphates (Fig. 1) was performed according to the method of Bittman et al. (1996) except that peracetic acid was used for the oxidation step instead of m-chloroperoxybenzoic acid. Each anhydrous alcohol (1.0 mmol) and 365 mg (5.17 mmol) of 1H-tetrazole were dissolved in 34 ml of anhydrous methylene chloride. A solution of 0.895 g (2.58 mmol) of dibenzyl-N,N-diisopropyl phosphoramidite in 5 ml of anhydrous methylene chloride was added under an argon atmosphere. The reaction mixture was stirred at room temperature for 2 h and was then cooled to 38°C in an isopropyl alcohol/dry ice bath. Peracetic acid [0.815 g (3.43 mmol) of 32% acid] in 28 ml of anhydrous methylene chloride was added drop-wise, and the temperature of the reaction mixture was raised to 0°C and stirred for 1 h. To the reaction mixture, 200 ml of methylene chloride was added, and the organic layer was washed with 10% sodium metabisulfite (2 × 40 ml), saturated sodium bicarbonate (2 × 40 ml), water (30 ml), and brine (40 ml). The organic layer was dried with anhydrous sodium sulfate, filtered, and concentrated under vacuum to dryness. The resulting crude products were purified by silica gel chromatography using hexane/ethyl acetate (1:1 for 1a and 7:3 for 2-8a) to elute the desired product.

View larger version (13K): [in this window] [in a new window]   Fig. 1.   The synthesis of FAPs. 1, FAP-4, n = 3; 2, FAP-8, n = 7; 3, FAP-10, n = 9; 4, FAP-12, n = 11; 5, FAP-14, n = 13; 6, FAP-16, n = 15; 7, FAP-18, n = 17; 8, FAP-22, n = 21.

Spectral Characterization of Phosphoric Acid Dibenzyl Ester Alkyl Esters (1-8a)

1a Isolated As a Clear Oil (309 mg) That Was Contaminated with Excess Phosphorylating Reagent). ¹H NMR (CDCl₃)  0.88 (t, J = 7.2 Hz, 3H, CH₃), 1.34 (sextet, J = 7.2 Hz, 2H, OCH₂CH₂CH₂CH₃), 1.59 (quintet, J = 6.6 Hz, 2H, OCH₂CH₂CH₂CH₃), 3.99 (q, 6.6 Hz, 2H, OCH₂CH₂CH₂CH₃), 5.02 (d, J = 1.8 Hz, 2H, OCH₂Ar), 5.05 (d, J = 2.1 Hz, 2H, OCH₂Ar), 7.35 (br s, 10H, 2 × ArH); ¹³C NMR (CDCl₃)  13.55, 18.60, 32.16 (d, J_C,P = 6.8 Hz), 67.72 (d, J_C,P = 6.1 Hz), 69.13(d, J_C,P = 5.5 Hz), 127.90, 128.47, 128.55, 136.00 (d, J_C,P = 6.8 Hz); ³¹P NMR (CDCl₃)  16.84; MS: [M + ²³Na] at m/z 357.3.

2a Isolated As a Clear Oil (351 mg, 90% Yield). ¹H NMR (CDCl₃)  0.88 (t, J = 6.9 Hz, 3H,CH₃), 1.24 (br s, 10H, OCH₂CH₂(CH₂)₅CH₃), 1.60 (quintet, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₅CH₃), 3.98 (q, J = 6.6 Hz, 6.9 Hz, 2H, OCH₂CH₂(CH₂)₅CH₃), 5.02 (d, J = 2.1 Hz, 2H, OCH₂Ar), 5.05 (d, J = 2.4 Hz, 2H OCH₂Ar), 7.34 (br s, 10H, 2 × ArH); ¹³C NMR (CDCl₃)  14.09, 22.62, 25.38, 29.06, 29.14, 30.17 (d, J_C,P = 6.9 Hz), 31.75, 68.05 (d, J_C,P = 6.2 Hz), 69.12 (d, J_C,P = 5.5 Hz), 127.90, 128.47, 128.56, 135.97 (d, J_C,P = 6.9 Hz); ³¹P NMR (CDCl₃)  16.83; MS: [M + ²³Na]⁺ at m/z 413.4.

3a Isolated As a Clear Oil (334 mg, 80% Yield). ¹H NMR (CDCl₃)  0.88 (t, J = 6.9 Hz, 3H,CH₃), 1.24 (br s, 14H, OCH₂CH₂(CH₂)₇CH₃), 1.58 (quintet, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₇CH₃), 3.98 (q, J = 6.7 Hz, 2H, OCH₂CH₂(CH₂)₇CH₃), 5.02 (d, J = 2.2 Hz, 2H, OCH₂Ar), 5.04 (d, J = 2.3 Hz, 2H OCH₂Ar), 7.34 (br s, 10H, 2 × ArH); ¹³C NMR (CDCl₃)  13.56, 22.13, 24.85, 28.57, 28.75, 28.95 (d, J_C,P = 1.6 Hz), 29.65 (d, J_C,P = 6.9 Hz), 31.34, 67.52 (d, J_C,P = 6.1 Hz), 68.59 (d, J_C,P = 5.6 Hz), 126.40, 126.97, 127.35, 127.96 (d, J_C,P = 6.6 Hz), 135.47 (d, J_C,P = 6.8 Hz); ³¹P NMR (CDCl₃)  16.82; MS: [M + ²³Na]⁺ at m/z 441.4.

4a Isolated As a Clear Oil (361 mg, 81% Yield). ¹H NMR (CDCl₃)  0.88 (t, J = 7.2 Hz, 3H, CH₃), 1.24 (br s, 18 H, OCH₂CH₂(CH₂)₉CH₃), 1.60 (quintet, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₉CH₃), 3.98 (q, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₉CH₃), 5.02 (d, J = 2.1 Hz, 2H, OCH₂Ar), 5.05 (d, J = 2.1 Hz, 2H, OCH₂Ar), 7.34 (br s, 10H, 2 × ArH); ¹³C NMR (CDCl₃)  14.13, 22.69, 25.38, 29.12, 29.35, 29.49, 29.56, 29.63, 30.18 (d, J_C,P = 7.0 Hz), 31.92, 68.05 (d, J_C,P = 6.1 Hz), 69.12 (d, J_C,P = 5.4 Hz), 127.89, 128.46, 128.55, 135.97 (d, J_C,P = 6.8 Hz); ³¹P NMR (CDCl₃)  16.84; MS: [M + ²³Na]⁺ at m/z 469.1.

5a Isolated As a Clear Oil (384 mg, 81% Yield). ¹H NMR (CDCl₃)  0.88 (t, J = 6.9 Hz, 3H, CH₃), 1.27 (br s, 22 H, OCH₂CH₂(CH₂)₁₁CH₃), 1.64 (quintet, J = 6.8 Hz, 2H, OCH₂CH₂(CH₂)₁₁CH₃), 3.98 (q, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₁₁CH₃), 5.04 (d, J = 2.1 Hz, 2H, OCH₂Ar), 5.06 (d, J = 2.1 Hz, 2H, OCH₂Ar), 7.34 (br s, 10H, 2 × ArH); ¹³C NMR (CDCl₃)  13.55, 22.14, 24.85, 25.26, 28.57, 28.80, 28.98 (d, J_C,P = 5.2 Hz), 29.12 (m), 29.65 (d, J_C,P = 6.9 Hz), 31.38, 32.31, 62.41, 67.50 (d, J_C,P = 6.1 Hz), 68.59 (d, J_C,P = 5.6 Hz), 127.34, 127.95 (d, J_C,P = 6.8 Hz), 135.48 (d, J_C,P = 6.8 Hz); ³¹P NMR (CDCl₃)  16.85; MS: [M + ²³Na]⁺ at m/z 497.2.

6a Isolated As a Clear Oil (427 mg, 85% Yield). ¹H NMR (CDCl₃)  0.88 (t, J = 6.9 Hz, 3H, CH₃), 1.28 (br s, 26H, OCH₂CH₂(CH₂)₁₃CH₃), 1.62 (quintet, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₁₃CH₃), 3.99 (q, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₁₃CH₃), 5.04 (d, J = 2.1 Hz, 2H, OCH₂Ar), 5.07 (d, J = 2.1 Hz, 2H, OCH₂Ar), 7.35 (br s, 10H, 2 × ArH); ¹³C NMR (CDCl₃)  13.57, 22.15, 24.85, 28.58, 28.83, 28.99 (d, J_C,P = 6.8 Hz), 29.17, 29.66 (d, J_C,P = 6.9 Hz), 31.39, 67.49 (d, J_C,P = 6.1 Hz), 68.59 (d, J_C,P = 5.6 Hz), 127.35, 127.95 (d, J_C,P = 6.6 Hz), 135.48 (d, J_C,P = 6.8 Hz); ³¹P NMR (CDCl₃)  16.88; MS: [M + ²³Na]⁺at m/z 525.3.

7a Isolated As a Hygroscopic White Solid (474 mg, 89% Yield), mp 32-33°C. ¹H NMR (CDCl₃)  0.88 (t, J = 6.9 Hz, 3H, CH₃), 1.25 (br s, 30H, OCH₂CH₂(CH₂)₁₅CH₃), 1.60 (quintet, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₁₅CH₃), 3.98 (q, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₁₅CH₃), 5.02 (d, J = 2.1 Hz, 2H, OCH₂Ar), 5.05 (d, J = 2.1 Hz, 2H, OCH₂Ar), 7.34 (br s, 10H, 2 × ArH); ¹³C NMR (CDCl₃)  14.12, 22.70, 25.40, 29.13, 29.38, 29.51, 29.58, 29.68, 29.72, 30.20 (d, J_C,P = 6.9 Hz), 31.94, 68.06 (d, J_C,P = 6.1 Hz), 69.14 (d, J_C,P = 5.4 Hz), 127.90, 128.47, 128.55, 136.00 (d, J_C,P = 6.8 Hz).; ³¹P NMR (CDCl₃)  16.83; MS: [M + ²³Na]⁺at m/z 553.3.

8a Isolated As a Hygroscopic White Solid (516 mg, 88% Yield), mp 43.5-44.5°C. ¹H NMR (CDCl₃)  0.88 (t, J = 6.9 Hz, 3H, CH₃), 1.25 (br s, 38H, OCH₂CH₂(CH₂)₁₉CH₃), 1.60 (quintet, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₁₉CH₃), 3.98 (q, J = 6.6 Hz, 2H, OCH₂CH₂(CH₂)₁₉CH₃), 5.02 (d, J = 2.4 Hz, 2H, OCH₂Ar), 5.05 (d, J = 2.4 Hz, 2H, , OCH₂Ar), 7.35 (br s, 10H, 2 × ArH); ¹³C NMR (CDCl₃)  14.13, 22.70, 25.39, 29.12, 29.37, 29.50, 29.57, 29.66, 29.71, 30.18(d, J_C,P = 6.9 Hz), 31.93, 68.06 (d, J_C,P = 6.0 Hz), 69.13 (d, J_C,P = 5.6 Hz), 127.89, 128.47, 128.55, 135.98 (d, J_C,P = 6.9 Hz); ³¹P NMR (CDCl₃)  16.83; MS: [M + ²³Na]⁺ at m/z 609.3.

Synthesis of Phosphoric Acid Mono Alkyl Esters (1-8b)

200 mg of 1-8a was dissolved in 30 ml of anhydrous methanol in a pressure vessel (Fig. 1). The vessel was purged with argon and ~200 mg of 10% Pd/C catalyst was added. The vessel was connected to a hydrogenation apparatus and a hydrogen atmosphere of ~ 50 psi was maintained inside the reaction vessel at room temperature for 8 h. The reaction mixture was then filtered by vacuum through a pad of methanol-washed celite. Solvent was evaporated under vacuum, yielding the desired product.

Spectral Characterization of Phosphoric Acid Mono alkyl Esters (1-8b)

1b Isolated As a Yellow Oil (70 mg, 86% Yield). ¹H NMR (CDCl₃/MeOH-d₄)  0.95 (t, J = 7.2 Hz, 3H, CH₃), 1.43 (sextet, J = 7.5 Hz, 2H, OCH₂CH₂CH₂CH₃), 1.66 (quintet, J = 6.9, 2H, OCH₂CH₂CH₂CH₃), 3.99 (q, J = 6.6 Hz, 2H, OCH₂CH₂CH₂CH₃); ¹³C NMR (CDCl₃/MeOH-d₄)  13.71, 19.02, 32.72 (d, J_C,P = 7.2 Hz), 66.86 (d, J_C,P = 5.5 Hz); ³¹P NMR (CDCl₃/MeOH-d₄)  18.84; MS: [M  H] at m/z 153.0.

2b Isolated As a White/Yellow Tacky Solid (100 mg, 93% Yield). ¹H NMR (CDCl₃/MeOH-d₄)  0.89 (t, J = 6.9 Hz, 3H, CH₃), 1.29 (br s, 10H, OCH₂CH₂(CH₂)₅CH₃), 1.67 (quintet, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₅CH₃), 3.97 (q, J = 6.6 Hz, 2H, OCH₂CH₂(CH₂)₅CH₃); ¹³C NMR (CDCl₃/MeOH-d₄)  14.18, 22.98, 25.89, 29.57, 29.58, 30.76 (d, J_C,P = 7.3 Hz), 32.18, 67.16 (d, J_C,P = 5.2 Hz); ³¹P NMR (CDCl₃/MeOH-d₄)  20.55; MS: [M  H] at m/z 209.1.

3b Isolated As a White/Yellow Tacky Solid (102 mg, 90% Yield). ¹H NMR (CDCl₃/MeOH-d₄)  0.89 (t, J = 6.9 Hz, 3H, CH₃), 1.28 (br s, 14 H, OCH₂CH₂(CH₂)₇CH₃), 1.67 (quintet, J = 6.8 Hz, 2H, OCH₂CH₂(CH₂)₇CH₃), 3.97 (q, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₇CH₃); ¹³C NMR (CDCl₃/MeOH-d₄)  12.83, 21.86, 24.79, 28.51 (d, J_C,P = 5.9 Hz), 28.79 (d, J_C,P = 1.3 Hz), 29.66 (d, J_C,P = 7.2 Hz), 31.12, 65.97 (d, J_C,P = 5.6 Hz); ³¹P NMR (DMSO-d₆)  16.55; MS: [M  H] at m/z 236.9.

4b Isolated As a White Solid (112 mg, 94% Yield). ¹H NMR (CDCl₃/MeOH-d₄)  0.88 (t, J = 6.6 Hz, 3H, CH₃), 1.27 (br s, 18 H, OCH₂CH₂(CH₂)₉CH₃), 1.67 (quintet, J = 6.6 Hz, 2H, OCH₂CH₂(CH₂)₉CH₃), 3.97 (q, J = 6.6 Hz, 2H, OCH₂CH₂(CH₂)₉CH₃); ¹³C NMR (CDCl₃/MeOH-d₄)  14.21, 22.98, 25.84, 29.57, 29.67, 29.89, 29.92, 29.96, 29.98, 30.69 (d, J_C,P = 7.4 Hz), 32.25, 67.22 (d, J_C,P = 5.7 Hz); ³¹P NMR (CDCl₃/MeOH-d₄)  21.22; MS: [M  H] at m/z 265.0.

5b Isolated As a White Solid (105 mg, 85% Yield), mp 58-60°C. ¹H NMR (CDCl₃/MeOH-d₄)  0.89 (t, J = 6.9 Hz, 3H, CH₃), 1.27 (br s, 22 H, OCH₂CH₂(CH₂)₁₁CH₃), 1.64 (quintet, J = 6.8 Hz, 2H, OCH₂CH₂(CH₂)₁₁CH₃), 3.96 (q, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₁₁CH₃); ¹³C NMR (CDCl₃/MeOH-d₄)  12.66, 21.82, 24.75, 28.48 (d, J_C,P = 8.5 Hz), 28.78 (d, J_C,P = 1.3 Hz), 28.85, 29.62 (d, J_C,P = 7.2 Hz), 31.14, 65.88 (d, J_C,P = 5.7 Hz); ³¹P NMR (DMSO-d₆)  16.51; MS: [M  H] at m/z 293.0.

6b Isolated As a White Solid (118 mg, 92% Yield), mp 71-72°C. ¹H NMR (CDCl₃/MeOH-d₄)  0.89 (t, J = 6.9 Hz, 3H, CH₃), 1.28 (br s, 26 H, OCH₂CH₂(CH₂)₁₃CH₃), 1.64 (quintet, J = 6.8 Hz, 2H, OCH₂CH₂(CH₂)₁₃CH₃), 3.96 (q, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₁₃CH₃); ¹³C NMR (CDCl₃/MeOH-d₄)  12.77, 21.85, 24.77, 28.48 (d, J_C,P = 8.5 Hz), 28.80 (d, J_C,P = 1.3 Hz), 28.88, 29.64 (d, J_C,P = 7.3 Hz), 31.16, 65.94 (d, J_C,P = 5.7 Hz); ³¹P NMR (DMSO-d₆)  16.51; MS: [M  H] at m/z 321.0.

7b Isolated As a White Solid (104 mg, 79% Yield). ¹H NMR (CDCl₃/MeOH-d₄)  0.89 (t, J = 6.9 Hz, 3H, CH₃), 1.27 (br s, 30H, OCH₂CH₂(CH₂)₁₅CH₃), 1.68 (quintet, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₁₅CH₃); 3.98 (q, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₁₅CH₃); ¹³C NMR (CDCl₃/MeOH-d₄)  14.26, 23.14, 26.01, 29.74, 29.84, 30.06, 30.09, 30.16, 30.87 (d, J_C,P = 7.2 Hz), 32.42, 67.32 (d, J_C,P = 5.8 Hz); ³¹P NMR (CDCl₃/MeOH-d₄)  21.69; MS: [M  H] at m/z 349.1.

8b Isolated As a White Solid (98 mg, 71% Yield). ¹H NMR (CDCl₃/MeOH-d₄)  0.88(t, J = 6.9 Hz, 3H), 1.26 (br s, 38H, OCH₂CH₂(CH₂)₁₉CH₃), 1.66 (quintet, J = 6.9 Hz, 2H, OCH₂CH₂(CH₂)₁₉CH₃), 3.97 (q, J = 6.6 Hz, 2H, OCH₂CH₂(CH₂)₁₉CH₃); ¹³C NMR (CDCl₃/MeOH-d₄)  14.22, 23.01, 25.87, 29.61, 29.71, 29.93, 29.97, 30.04, 30.73 (d, J _C,P = 7.4 Hz), 32.29, 67.27 (d, J_C,P = 5.6 Hz); ³¹P NMR (CDCl₃/MeOH-d₄)  20.66; MS: [M  H] at m/z 405.1.

Molecular Modeling of the LPA₂ Receptor

The inactive and active models of LPA₂ were previously developed in our research group (Sardar et al., 2002). Briefly, the inactive model of LPA₂ was developed by homology modeling using the MOE (molecular operating environment) program (version 2002:01; Chemical Computing Group, Montreal, ON, Canada) and was based on the bovine rhodopsin crystal structure (Palczewski et al., 2000). The active model of the LPA₂ receptor was developed via homology modeling using MOE and was based on the validated model of S1P₁ (Parrill et al., 2000). Autodock 3.0 (Morris et al., 1996, 1998) was used to calculate the docked energies for FAP-12 with both the inactive and active model of LPA₂. Automated docking using the Lamarckian genetic algorithm (Morris et al., 1998) was applied to generate 100 complexes of FAP with the LPA₂ receptor (inactive and active forms) and LPA with the LPA₂ receptor (active form), to evaluate the binding region of the ligand. The best complex for each receptor, in terms of energy and binding position of the ligand, was energy minimized with the MMFF94 force field (Halgren, 1996) to a root-mean-square gradient of 0.01 kcal/mol/Å to allow the receptor to adapt to the presence of the ligand. The ligands were then removed from the minimized receptor-ligand complexes and 100 additional complexes were generated with Autodock to re-examine the binding energy after allowing the receptor to acclimatize to the ligand. Autodock 3.0 allows full flexibility of ligand torsion angles but does not provide opportunity for protein side chains to adapt to the presence of a ligand. Thus, the best protein-ligand complex found in an initial docking run was geometry optimized to allow protein side chains to optimize in the presence of the ligand. The ligand was then removed and docked back into the protein to obtain a final docked energy that better reflects the induced fit that occurs when protein-ligand binding occurs. If a conformation similar to the one before minimization was not obtained, the docking run was repeated. When the preminimized conformation was located, the complex with the lowest docked energy was chosen as the best complex.

Cells and Cell Culture

Oocytes were harvested and treated as described earlier (Tigyi et al., 1999). RH7777 cells, stably expressing human LPA₂ receptors, were provided by Dr. Kevin Lynch (University of Virginia, Charlottesville, VA). All cell lines were maintained in Dulbecco's modified Eagle's medium, containing 10% fetal bovine serum and 2 mM glutamine, containing 250 µg/ml G418 for the stable transfectants. RH7777 cells stably expressing LPA₁ and LPA₃ receptors have been generated by our group and characterized elsewhere (Fischer et al., 2001). N1E-115 and IEC-6 cells were purchased from American Type Culture Collection (Manassas, VA).

Cellular Assays

Electrophysiological recording from X. laevis oocytes was done using the standard two-electrode voltage clamp technique (Tigyi et al., 1999). Monitoring of intracellular Ca²⁺ changes using Fura-2 acetoxymethyl ester fluorescent indicator (Tigyi et al., 1999) and ligand-induced [³⁵S]GTP--S binding assays (Parrill et al., 2000) was performed as described in our previous publications. FAP samples were prepared fresh from methanolic stock solutions by diluting them in perfusion buffer directly before application. LPA samples were prepared the same way from a bovine serum albumin-complexed stock solution.

Lipid Phosphate Phosphatase Assay

Lipid phosphate phosphatase (LPP) activity was measured as described previously (Yokoyama et al., 2002). Competitive inhibition of FAP-12 on LPP activity was analyzed according to the surface dilution kinetics model (Carman et al., 1995; Dillon et al., 1997). A 100,000g pellet from mouse fibroblasts overexpressing mouse LPP1 was used as the enzyme source, and the reaction was performed in the presence of 2 mol% ([lipid]/([lipid] + [Triton X-100]) [³²P]LPA (50 µM, 2 × 10⁴ cpm/assay), 2.5 mM Triton X-100, and indicated amount of competitors, including FAP-12. The released [³²P]phosphate was extracted and measured.

DNA Fragmentation Assay

The topoisomerase inhibitor camptothecin induces DNA fragmentation and apoptosis in rat IEC-6 intestinal epithelial cells that can be prevented by LPA (Deng et al., 2002). To assess the agonist properties of FAP-12, IEC-6 cells were exposed to 20 µM camptothecin for 6 h with or without FAP-12 (10 µM) or LPA (10 µM) pretreatment (15 min before camptothecin) and DNA fragmentation was measured using an enzyme-linked immunosorbent assay method with the cell death detection enzyme-linked immunosorbent assay kit from Roche (Indianapolis, IN). Briefly, cells were harvested and lysed in DNA lysis buffer for 30 min and centrifuged at 200 rpm for 10 min. An aliquot of the supernatant was incubated with anti-histone-biotin plus anti-DNA peroxidase conjugated antibody in 96-well streptavidin-coated plates on a shaker for 2 h. After washing with the incubation buffer, 100 µl of substrate buffer was added to each well and incubated for an additional 5 to 10 min. DNA absorbance was read at 405 nm in a microplate reader. Duplicates of the samples were used to quantify protein using the bicinchoninic acid assay kit from Pierce (Rockford, IL). DNA fragmentation was expressed as absorbance units per microgram of protein per minute.

RT-PCR Analysis of LPA Receptor Expression

To assess the expression of LPA receptor subtypes in N1E-115 cells, RT-PCR analysis was performed using a primer set and PCR protocol described in previous publications (Tigyi et al., 1999; Fischer et al., 2001).

Data Analysis

The significance of differences between the groups was determined using the Student's t test. Values were considered significantly different at p < 0.05. The antagonists' binding constant, K_i, was estimated by the method of Cheng (2002), as follows: K_i = IC₅₀/(1 + ([LPA]/EC₅₀)^n_H), where IC₅₀ is the half-effective concentration of the inhibitor, EC₅₀ is the half-effective concentration of the agonist (LPA), n_H is the slope function (Hill- or cooperativity-exponent) in the logistic function that describes the agonist's activation curve, and [LPA] is the concentration of LPA against which the antagonist was being tested.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

The synthetic pathway used to generate FAPs is shown in Fig. 1. Molecular structures were verified by NMR and mass spectrometry, and all spectral data were consistent with the assigned structures (see Materials and Methods).

FAPs with hydrocarbon chain lengths of 4, 8, 12, 18, and 22 were tested for their ability to enhance or inhibit LPA-induced Cl currents in X. laevis oocytes. None of the FAP compounds activated Cl currents in the oocytes when applied up to 10 µM (data not shown). In contrast, all FAPs inhibited LPA-induced Cl currents with varying potency. Dose-inhibition measurements revealed a strong correlation between chain length and inhibitory action (Fig. 2A). The lowest IC₅₀ was observed using FAP-12 and FAP-8 (IC₅₀ = 2.4 ± 1 and 6 ± 1 nM, respectively, against 5 nM oleoyl-LPA). FAPs with shorter or longer hydrocarbon chains than 12 were less effective in inhibiting the LPA responses.

View larger version (22K): [in this window] [in a new window]   Fig. 2.   Pharmacological characterization of the FAPs on LPA-elicited Cl currents in X. laevis oocytes. A, 5 nM LPA 18:1 mixed with increasing concentrations of FAP-4 (), -8 (), -12 (), -18 (), or -22 () was superfused over oocytes and peak Cl current amplitudes were measured. 100% represents the peak amplitude of the Cl current activated by 5 nM LPA, and 0% corresponds to baseline holding current. The bar graph represents the IC50 values for each FAP compound. IC50 values were 423 ± 150, 6 ± 1, 2.4 ± 1, 26 ± 12, and 32 ± 16 nM for FAP-4, -8, -12, -18, and -22, respectively. Data points represent the average of at least three measurements ± S.D. B, effect of FAP-12 on the LPA dose-response curve in X. laevis oocytes. Oocytes were treated with increasing concentrations of LPA (18:1) () alone, or mixed with 30 nM FAP-12 (). 100% represents the peak amplitude for the maximal LPA-activated Cl current, and 0% corresponds to baseline holding current. EC50 values were 30 ± 8 and 300 ± 40 nM for LPA and LPA + 30 nM FAP-12, respectively. Data points represent the average values of at least three experiments ± S.D.

When coapplied with LPA, FAP-12 shifted the LPA dose-response curve to the right, suggesting a competitive mechanism for the antagonist activity (Fig. 2B). In X. laevis oocytes the LPA dose-response curve has been shown to be biphasic and has been attributed to the presence of both high- and low-affinity binding sites (Guo et al., 1996; Liliom et al., 1996b). When the LPA activation curve was measured in the presence of FAP-12, it could be fitted by a simple Langmuir isotherm, suggestive of only one binding site (Fig. 2B). The activation curves measured in the absence and presence of the inhibitor run parallel at high LPA concentrations. Consequently, FAP-12 seems to inhibit only the high-affinity LPA binding site.

To characterize the effects of the FAP compounds in a mammalian system, we chose the RH7777 rat hepatoma cell line, which does not respond to LPA or S1P in a variety of cellular assays, including Ca²⁺ mobilization, and does not express any of the known LPA and S1P receptors as monitored by RT-PCR (Zhang et al., 1999). In cells stably expressing LPA₁, FAP-12 weakly inhibited LPA-induced Ca²⁺-mobilization (Fig. 3). In contrast, FAP-12 alone did not elicit intracellular Ca²⁺ transients when applied up to 10 µM (data not shown). Surprisingly, LPA₂-expressing cells showed a dose-dependent increase in intracellular Ca²⁺ mobilization in response to FAP-10 and FAP-12, with EC₅₀ values of 3.7 ± 0.2 µM and 700 ± 22 nM, respectively (Fig. 4A). FAP-14 was found to be a weak agonist, when applied at 10 µM. In contrast, no other FAP analog elicited Ca²⁺ mobilization, or inhibited the LPA response when applied in concentrations up to 10 µM. The maximal response to FAP-12 was 50% of that elicited by LPA, suggesting that FAP-12 was a partial agonist of the LPA₂ receptor. When a concentration of 10 µM was applied only FAP-10, FAP-12 and FAP-14 showed significant agonist activity (Fig. 4B) with FAP-10 being the most efficacious.

View larger version (21K): [in this window] [in a new window]   Fig. 3.   Inhibition of LPA-induced Ca2+ mobilization by FAP-12 in RH7777 cells expressing LPA1 receptors. Cells were exposed to 250 nM LPA (18:1) mixed with increasing concentrations of FAP-12. Peak areas of Ca2+ responses were measured. 100% represents the peak area of the Ca2+ transient elicited by 250 nM LPA, and 0% corresponds to baseline fluorescence. Ca2+ mobilization was measured as the ratio of fluorescence values at 340 and 380 nm. Data points represent the average from at least three experiments ± S.D. Inset, representative time-courses for Ca2+ transients induced by 250 nM LPA alone or by coapplication of 250 nM LPA with 10 µM FAP-12. R is the ratio of fluorescence measured at 340 and 380 nm.

View larger version (20K): [in this window] [in a new window]   Fig. 4.   A, dose-dependent activation of LPA2 by FAPs. Increasing concentrations of LPA (), FAP-10 (), or FAP-12 () were applied, and the peak areas of Ca2+ responses were measured in RH7777 cells stably expressing LPA2. The EC50 values for FAP-10 and FAP-12 were 700 ± 22 nM and 3.7 ± 0.2 µM, respectively. 100% represents the peak area for maximal LPA-activated Ca2+ mobilization, and 0% corresponds to baseline. Data points represent the average of at least three measurements ± S.D. Inset, comparison of time courses of maximal Ca2+ responses to LPA (18:1), FAP-10, and FAP-12 in LPA2-expressing RH7777 cells. LPA, FAP-10, or FAP-12 (30 µM) was applied. R is the ratio of fluorescence measured at 340 and 380 nm. B, chain length dependence of FAP agonist activity in RH7777 cells expressing LPA2. FAP compounds were applied at 10 µM, and peak areas of Ca2+ responses were measured. Data points represent the average of at least three measurements ± S.D. Inset, comparison of time courses of Ca2+ transients elicited by 10 µM FAP-8, -10, -12, and -14. R is the ratio of fluorescence measured at 340 and 380 nm.

Computational docking studies were used to evaluate the interactions of LPA and FAP with LPA₂. When LPA (18:1) was docked against the active LPA₂ model, the lowest docked free energy was 15.08 kcal/mol. The binding pocket of LPA (18:1) was in the transmembrane domain with the phosphate group of LPA forming ion pairs with arginine 107 and lysine 278 (Fig. 5A) and the 2-hydroxyl group of LPA hydrogen bonding with glutamine 108. 

View larger version (45K): [in this window] [in a new window]   Fig. 5.   Model of the LPA2 receptor docked with LPA (18:1) (A), FAP-12 (B), and FAP-10 (C). LPA2 is shown as a ribbon model (side view, top) with the ligands LPA (18:1) (blue), FAP-12 (green), and FAP-10 (yellow) as space-filling models. Residues R 107 (magenta), Q 108 (teal), and K 278 (orange) are also shown as space-filling models. Close-up view (top view, bottom) of critical interactions between LPA2 residues and the phosphate group of the three ligands is shown.

FAP-12 was docked against both the active and inactive models of LPA₂. The lowest docked energy obtained upon evaluation of FAP-12 with the inactive form of LPA₂ was 9.98 kcal/mol. A more favorable docked energy, 12.44 kcal/mol, was obtained when FAP-12 was docked against the active form of LPA₂, consistent with the agonist effect observed experimentally. The binding pocket of FAP-12 was located in the transmembrane domain and ion pairs were observed between the phosphate and two cationic residues, arginine 107 and lysine 278 of LPA₂ (Fig. 5B). These interactions are consistent with our previous studies on the binding of LPA to the LPA receptors (Wang et al., 2001; Sardar et al., 2002).

Docking studies between FAP-8, FAP-10 (Fig. 5C), FAP-14, and FAP-18 were performed using the active LPA₂ model, with resulting docked energies of 9.17, 10.46, 12.06, and 10.84 kcal/mol, respectively. The binding mode observed for these structures was similar to that observed for FAP-12. All four of these compounds had lower (less favorable) docked energies than FAP-12, as expected based on the experimental observation that FAP-12 had the lowest EC₅₀. However, the relative energies did not correlate further with the experimentally observed receptor activation. Several factors probably contribute to this discrepancy. First, energies from docking studies are most reflective of binding affinity rather than receptor activation. Second, binding is governed by the free energy change that occurs when a ligand moves from one environment (often aqueous solution) to another (a protein binding site). The docked energies represent only the enthalpic part of the free energy. The entropic part of the free energy, which consists of changes in conformational freedom and solvation effects, may be an important contributor to the differences in FAP compound activity.

RH7777 cells stably expressing the LPA₃ receptor did not respond to any of the FAPs by Ca²⁺ mobilization when applied up to 10 µM (data not shown). However, FAP-12 dose dependently inhibited LPA-induced Ca²⁺ mobilization with a K_i of 90 nM (Fig. 6A). Coapplication of 300 nM FAP-12 with LPA shifted the dose-response curve of LPA to the right, increasing the EC₅₀ value from 300 ± 14 to 1000 ± 30 nM and suggesting a competitive type of inhibition (Fig. 6B).

View larger version (16K): [in this window] [in a new window]   Fig. 6.   A, FAP-12 inhibits LPA (18:1)-induced Ca2+ mobilization in RH7777 cells stably expressing LPA3 in a dose-dependent manner. LPA (250 nM; 18:1) was mixed with increasing concentrations of FAP-12. The Ki value was 90 nM. Data points represent the average of at least three measurements ± S.D. Inset, time courses of Ca2+ mobilization after application of 250 nM LPA alone or together with 1 µM FAP-12. R is the ratio of fluorescence measured at 340 and 380 nm. B, FAP-12 shifts the LPA dose-response curve to the right. Cells were exposed to increasing concentrations of LPA () or mixed with 300 nM FAP-12 (). EC50 values were 300 ± 14 nM for LPA and 1000 ± 30 nM for LPA + 300 nM FAP-12. Data points represent the average of peak areas from at least three measurements ± S.D. Inset, time courses of Ca2+ mobilization after application of 1 µM LPA alone or together with 300 nM FAP-12. R is the ratio of fluorescence measured at 340 and 380 nm. C, the relationship between FAP chain length and inhibition of LPA-elicited Ca2+ mobilization through LPA3 in RH7777 cells. Each FAP compound (200 nM) was mixed with 250 nM LPA (18:1), and the elicited Ca2+ peak areas were measured. Data points represent the average of at least three measurements ± S.D. Inset, time courses of Ca2+ mobilization after coapplication of 250 nM LPA with 200 nM FAP-4, -10, and -12. R is the ratio of fluorescence measured at 340 and 380 nm.

All FAPs were tested for their ability to inhibit LPA-elicited Ca²⁺-mobilization in cells expressing LPA₃. Applied at a concentration of 200 nM, FAP-12 was the most effective (60% inhibition), followed by FAP-10, -14, and -18 (~25% inhibition) (Fig. 6C). FAP-4, -8, -16, and -22 did not inhibit the LPA response significantly at this concentration.

To determine whether the effects of FAPs were selective to LPA receptors, we examined the most potent analog, FAP-12, on responses elicited by various mammalian G-protein coupled receptors. In oocytes expressing poly(A)⁺ mRNA from rat brain, responses elicited by serotonin (10 µM), kainate (100 µM), and glutamate (10 µM), which are mediated through the inositol trisphosphate-Ca²⁺ pathway, were tested for interference by FAPs. In the presence of 10 µM FAP-12, these responses relative to control were 103 ± 8, 95 ± 12, and 106 ± 7% (n = 3), respectively. The lack of inhibition indicates that FAP-12 did not inhibit these neurotransmitter receptors or the second-messenger systems mediating Ca²⁺-activated Cl currents. Additionally, in RH7777 cells, FAP-12 (10 µM) did not affect ATP-induced (1 µM) Ca²⁺ mobilization (101 ± 10% of ATP alone control, n = 3) indicating that in this mammalian cell line, just as in X. laevis oocytes, the compound did not interfere with ligand-induced Ca²⁺ transients.

The effects of FAP-12 were also tested on S1P receptors. GTP--S loading assays were performed on RH7777 cells stably expressing S1P₁. FAP-12, when coapplied with S1P (300 nM) up to 10 µM, neither enhanced nor inhibited the substantial S1P-induced GTP loading. FAP-12 alone did not induce GTP loading in these cells (data not shown). Similar results were obtained with RH7777 cells transiently transfected with S1P5 (Table 1). S1P elicits dose-dependent Ca²⁺ transients in RH7777 cells expressing S1P₂ or S1P₃. FAP-12 did not induce Ca²⁺ mobilization in RH7777 cells transiently transfected with S1P₂ or S1P₃, when applied in concentrations as high as 10 µM, nor did it affect S1P-elicited Ca²⁺ transients (Table 1). We could not test FAP-12 on S1P₄ receptors, because, in our hands, this receptor did not respond to S1P in the GTP--S loading or Ca²⁺-mobilization assays.

To test whether FAP-12 interacts with LPP, a key enzyme of LPA degradation, enzymatic activity was measured in the absence or presence of FAP-12. LPP activity decreased to 73, 58, and 39% of control, when FAP-12 was added in 2-, 4-, and 8-fold excess, respectively (data not shown). This result suggests that FAP12 also interferes with the LPP-mediated degradative pathway of LPA.

To confirm the agonist properties of FAP-12 and FAP-10 on LPA₂, these compounds were tested in N1E-115 neuroblastoma cells for Ca2+ mobilization effect (Fig. 7A). N1E-115 cells express only LPA₂ receptor transcripts (Fig. 7B) and respond to LPA with a transient elevation in [Ca²⁺]_i. FAP-12 and FAP-10 both elicited Ca²⁺ transients in these neuroblastoma cells, confirming their agonist properties in a cell line that endogenously expresses only the LPA₂ receptor subtype. To further characterize the agonist properties of FAP-12 (Fig. 7C), we examined its effect in an apoptosis protection assay using IEC-6 cells. These cells express predominantly LPA₂ along with lesser amounts of LPA₁ (Deng et al., 2002). In these cells, FAP-12 applied at 10 µM elicited a significant decrease in camptothecin-induced DNA fragmentation that was comparable with the effect of LPA (10 µM). The results obtained from N1E-115 and IEC-6 cells lend strong support to the LPA-like agonist properties of FAPs in cell lines that express the LPA₂ receptor subtype.

View larger version (25K): [in this window] [in a new window]   Fig. 7.   Agonist properties of FAPs in cells that endogenously express LPA2. A, time course of a representative experiment monitoring intracellular Ca2+ concentration in N1E-115 cells. FAP-10 (30 µM), -12 (10 µM), and LPA 18:1 (5 µM) were applied consecutively to N1E-115 cells and the resulting changes in [Ca2+]i were recorded. B, RT-PCR study of LPA receptor transcripts in N1E-115 cells. M, molecular weight marker; 1, LPA1; 2, LPA2; 3, LPA3; 4, negative control (no template); 5, beta actin. C, FAP-12 mimics the antiapoptotic effect of LPA in IEC-6 cells. Camptothecin (20 µM) induces apoptosis accompanied by DNA fragmentation within 6 h. LPA (10 µM) or FAP-12 (10 µM) both significantly reduced camptothecin-induced DNA fragmentation (mean ± S.E.M., n = 3).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The phospholipid growth factors LPA and S1P are involved in numerous physiological and pathological processes, including regulation of cell proliferation and differentiation, apoptosis, Ca²⁺-homeostasis (Goetzl et al., 2000; Tigyi, 2001) and tumor cell invasion (Umezu-Goto et al., 2002), and atherosclerosis (Siess et al., 1999). Given that most cells express multiple PLGF receptor subtypes, the lack of subtype-specific agonists and antagonists for LPA receptors remains a limiting factor for the PLGF field. In an effort toward the rational design of such ligands, structural models of the PLGF receptors have been developed recently (Parrill et al., 2000; Wang et al., 2001; Sardar et al., 2002). Based on computational modeling of the ligand-receptor interactions, we deduced and partially validated a model for receptor activation. The model assigned distinct functions to the polar headgroup and the hydrophobic tail within the LPA pharmacophore. In this model, the phosphate headgroup interacts with two positively charged conserved residues in the third and seventh transmembrane helices (Wang et al., 2001), whereas the hydrocarbon tail interacts with hydrophobic side-chains of amino acid residues lining the interhelical pocket (Sardar et al., 2002). These hydrophobic interactions are predicted to be necessary for activation of the receptor (Fischer et al., 2001; Wang et al., 2001). Specific recognition of S1P versus LPA is achieved through hydrogen bonding between the hydroxyl group of LPA and a glutamine or by ion pairing between the amino group of S1P and a glutamate in the third transmembrane helix conserved in the corresponding receptor subfamilies (Wang et al., 2001). The two-point pharmacophore is consistent with the experimentally established structure-activity relationships of LPA (Lynch and Macdonald, 2002; Sardar et al., 2002).

Our group (Fischer et al., 2001) and Dr. Lynch's group (Heise et al., 2001) have reported on LPA antagonists that show receptor subtype selectivity. A systematic screening of 2-OH-substituted N-acyl ethanolamide phosphates led to the discovery of a benzyl-4-oxybenzyl derivative of ethanolamine phosphate, of which the S-enantiomer exhibited selective antagonism of LPA₁ over LPA₃, whereas it did not affect the LPA₂ receptor (Heise et al., 2001). We found that dioctyl phosphatidic acid and dioctylglycerol pyrophosphate were weak antagonists of LPA₁ and strong antagonists for LPA₃, whereas long-chain analogs (18:1) were not inhibitors of LPA receptors (Fischer et al., 2001). It is important to note that short-chain LPA (8:0) was neither an agonist nor an antagonist in this system (Fischer et al., 2001). Both dioctylglycerol pyrophosphate and (S)-2-benzyl-4-oxybenzyl-N-acyl ethanolamide phosphate have a negatively charged phosphate head group. They also have two hydrocarbon side chains, unlike the physiological ligand, LPA. Therefore, both the length and size of the hydrophobic tail seems to affect the ability to activate or inhibit LPA receptors. Based on the two-point pharmacophore, we hypothesized that modifications within the hydrophobic tail of the ligand might have profound effects in determining agonist or antagonist behavior and therefore might allow identification of molecules with selective antagonistic properties.

Our earlier work (Bittman et al., 1996; Liliom et al., 1996), as well as of that of others (Hooks et al., 1998; Heise et al., 2001), has shown that the glycerol backbone can be replaced by serine, tyrosine, or ethanolamine but maintain the LPA mimetic effect of these lipid phosphate analogs. Fatty alcohol phosphates provide the simplest set of easy to synthesize analogs required to delineate the minimal structural requirement for a ligand using this hypothesis. For this reason we synthesized a series of FAP molecules with carbon chain lengths from 4 to 22 and characterized their effects on individual LPA and S1P receptor subtypes.

Pharmacological analysis of the FAP series supports our hypothesis that the ionic headgroup attached to a hydrophobic tail are sufficient to make them ligands for all three LPA receptors, although with markedly different efficacies, potencies, and selectivities. FAP-10 and FAP-12 were specific agonists for LPA₂ with EC₅₀ values of 3.7 ± 0.2 µM and 700 ± 22 nM, respectively, and specific antagonists for LPA₃. FAP-12 also weakly inhibited LPA₁. The decyl and dodecyl chain seem to be unique, which suggests that LPA₂ differs from the other two receptors in that it is activated by these relatively short-chain analogs. FAP molecules with carbon chain lengths shorter than 10 or longer than 14 did not affect LPA₂ and were weaker inhibitors of LPA₃ than FAP-12. Thus, the present results emphasize that acyl chain length plays a significant role in the ligand properties of LPA-like pharmacophores and establishes a distinguishing role in ligand recognition between the different LPA receptor subtypes. Furthermore, FAPs lack the glycerol backbone; thus, the present findings provide compelling evidence that it is not required for ligand activity. These observations are in complete agreement with earlier structure-activity studies in which LPA receptors showed a high degree of tolerance to analogs with serine, tyrosine, or ethanolamine in place of the glycerol backbone (Sugiura et al., 1994; Jalink et al., 1995; Lynch et al., 1997; Hooks et al., 1998).

X. laevis oocytes express three pharmacologically distinguishable receptors for LPA (Liliom et al., 1996b; Fischer et al., 1998). This heterogeneity is reflected in the dose-response curve, which is best described by assuming multiple sites with different high and low affinity for LPA (Guo et al., 1996; Liliom et al., 1996b). Our experiments indicate that FAPs block only the high-affinity site (Fig. 2B). Thus far, two high-affinity LPA sites have been identified in oocytes: the PSP24 (Guo et al., 1996) and LPA₁ receptors (Kimura et al., 2001). Which of these is inhibited by FAPs remains unclear. The human ortholog of the LPA₁ receptor was only weakly inhibited by FAP-12 in concentrations in the high micromolar range (Fig. 3). The highly conserved nature of mammalian and X. laevis LPA₁ tends to argue against the hypothesis that the inhibitor would target LPA1. On the other hand, overexpression of X. laevis LPA₁ in X. laevis oocytes augmented the endogenous LPA response (Kimura et al., 2001), whereas the human ortholog did not (Yokoyama et al., 2002). Interestingly, the chain length-dependence of the inhibitory potencies of FAPs in X. laevis oocytes (Fig. 2A) and in LPA₃-expressing RH7777 cells (Fig. 6C) is very similar despite the consensus that oocytes do not express an LPA₃ receptor subtype (Kimura et al., 2001).

FAPs did not activate or inhibit S1P receptors, suggesting that this cluster within the EDG receptor family is fundamentally different from the LPA receptor cluster despite several similarities in ligand recognition. Based on earlier structure-activity measurements, a free amino group on the sphingoid backbone seems to be necessary for ligand recognition of the S1P receptors (Van Brocklyn et al., 1998). This group is necessary for ligand binding (Parrill et al., 2000) and provides selectivity for these receptors to distinguish S1P from LPA (Wang et al., 2001). The absence of this free amino group in the FAP structure might explain the lack of effect of these molecules on S1P receptors. The agonist and antagonist effects of FAP-12 seem to be specific for LPA receptors because it did not modify the function of other heterologously expressed GPCRs in the X. laevis bioassay or RH7777 cells. The agonist properties of FAP-12 were confirmed in N1E-115 and IEC-6 cells that endogenously express LPA₂ receptor subtype (Fig. 7). Thus, the present data identify FAP-12 as the first LPA₂-selective agonist. Computational docking studies of FAP-8, -10, -12, -14, -18, and LPA against the active form of LPA₂ suggest that the greatest affinity for all structures occurs in an overlapping region in the receptor. The energies observed for FAP-12 binding to the active and inactive models of the LPA₂ receptor, 12.44 kcal/mol versus 9.98 kcal/mol, are consistent with the experimental observation that FAP-12 is a partial agonist and thus interacts more strongly with the active conformation of the LPA₂ receptor. The lowest docked energy was observed for FAP-12 binding to the active model, a finding consistent with the observed lowest EC₅₀ value (700 ± 22 nM) for this structure. Thus, the present results provide new refinements to our computational models.

FAP-12 inhibits LPA degradation by LPP, indicating that the simplified structure of the FAP molecule is also recognized by a key enzyme in the established degradation pathway of LPA. This observation suggests that FAP-12 will have at least three molecular targets: the LPA₃ and LPA₂ receptors and LPP enzymes.

The two-point pharmacophore for LPA receptor activation, which is based on our atomic resolution structural models of the PLGF receptors, makes the rational design of receptor subtype-selective agonists and antagonists possible, opening an important new avenue of research in the PLGF field. This approach identified FAPs as selective inhibitors of LPA₃ and the first subtype-specific agonists of LPA₂ receptors. Although the compounds identified here are not themselves likely to be useful clinically, they could serve as lead compounds for further development. Moreover, the differential effects of the acyl chain length derivatives of FAPs point to an important concept in designing subtype-selective reagents.