Activation of the phosphatidylinositol 3-kinase-Akt/protein kinase B signaling pathway in arachidonic acid-stimulated human myeloid and endothelial cells: involvement of the ErbB receptor family.

Although arachidonic acid has been demonstrated to stimulate a wide variety of cellular functions, the responsible mechanisms remain poorly defined. We now report that arachidonic acid stimulated the activity of class Ia phosphatidylinositol 3-kinase (PI3K) in human umbilical vein endothelial cells, HL60 cells, and human neutrophils. Pretreatment of endothelial cells with AG-1478, an inhibitor of the ErbB receptor family, resulted in the suppression of PI3K activation by arachidonic acid. The fatty acid enhanced the tyrosine phosphorylation of ErbB4 but not of ErbB2 or ErbB3. The ability of arachidonic acid to stimulate PI3K activity in neutrophils was suppressed by indomethacin and nordihydroguaiaretic acid, inhibitors of the cyclooxygenases and lipoxygenases, respectively, but not by 17-octadecynoic acid, an inhibitor of omega-hydroxylation of arachidonic acid by cytochrome P450 monooxygenases. Consistent with this, the activity of PI3K in neutrophils was stimulated by 5-hydroxyeicosatetraenoic acid. Arachidonic acid also transiently stimulated the phosphorylation of Akt on Thr-308 and Ser-473. Although PI3K was not required for the activation of the mitogen-activated protein kinases, ERK1, ERK2, and p38, in arachidonic acid-stimulated neutrophils, the fatty acid acted via PI3K to stimulate the respiratory burst. These results not only define a novel mechanism through which some of the actions of arachidonic acid are mediated but also demonstrate that, in addition to ErbB1 (epidermal growth factor receptor), ErbB4 can also be transactivated by a non-epidermal growth factor-like ligand.

Arachidonic acid (20:46) 1 is a second messenger that is released by the action of phospholipase A 2 in activated cells (1).
The liberated 20:46 is either metabolized within the cells or released into the extracellular space where it can exert stimulatory or inhibitory effects on cellular responses. The actions of 20:46 include the stimulation of neutrophil superoxide production and degranulation (2,3), insulin secretion by isolated rat islets of Langerhans (4), induction of the MAP kinase phosphatase-1 gene in vascular smooth muscle cells (5), differentiation of HL60 cells (6) and zone chondrocytes (7), adherence of neutrophils and human breast cancer cells to various substrata (3,8), and proliferation of certain tumor and normal cell types (9,10). On the other hand, 20:46 has been reported to inhibit cell-cell communication via gap junctions (11,12), cytokineinduced adhesion molecule expression in endothelial cells (13), and proliferation of HL60 cells in conjunction with the induction of differentiation (6). In cell-free assays, 20:46 and other polyunsaturated fatty acids have been widely demonstrated to stimulate the activity of protein kinase C (PKC) (14,15). The ability of 20:46 to modulate the permeability of the Na ϩ , K ϩ , and H ϩ channels (16 -18) and stimulate Ca 2ϩ mobilization (3) has also been demonstrated. In intact cells, 20:46 has been found to stimulate the activity of the neutral sphingomyelinase (19), the tyrosine phosphorylation of a number of cytosolic proteins (20), and the activities and/or phosphorylation of the extracellular signal-regulated protein kinase (ERK)-1 and ERK2, p38, c-Jun N-terminal kinase (20 -23), the cytosolic and secretory phospholipase A 2 (24)-and MAP kinase-activated protein kinase 2 (8). On the other hand, 20:46 has been demonstrated to inhibit tumor necrosis factor-stimulated activation of NF-B (13). Although 20:46-derived metabolites are biologically active, the above actions are mostly unaffected by the addition of lipoxygenase and cyclooxygenase inhibitors.
In an endeavor to understand further the mechanisms through which 20:46 acts, we investigated whether 20:46 affected the activity of phosphatidylinositol 3-kinase (PI3K). PI3K is a family of lipid kinases that phosphorylate inositolcontaining phospholipids on the D-3 position of the inositol ring, resulting in the formation of phosphatidylinositol (PtdIns) 3 P, PtdIns 3,4 P 2 , and PtdIns 3,4,5 P 3 (25). PI3K is divided into three main classes based on structural similarity and in vitro substrate specificity. Class I kinases are heterodimers of a catalytic and regulatory subunit and are coupled to either tyrosine kinases (class Ia) or to heterotrimeric G protein-coupled receptors (class Ib) (25). Within cells, the preferred sub-strate for class I PI3K is PtdIns 4,5 P 2 , and only class I PI3Ks have been shown to activate Akt (also known as PKB), which is encoded by three closely related genes, PKB␣, PKB␤ and PKB␥ (25). Class Ia enzymes include the p110␣, p110␤, and p110␦ catalytic subunits that are complexed to one of seven adaptor (regulatory) proteins derived from three separate genes (p85␣, p85␤, and p55␥) (25). The adaptor proteins, via their two Src homology 2 domains, mediate the interaction of class Ia PI3K with tyrosine-phosphorylated residues. To date, only one class Ib PI3K, PI3K␥, has been identified, and this is composed of a p110␥ catalytic subunit and a 101-kDa regulatory subunit (25). Class I PI3Ks are sensitive to inhibition by wortmannin and LY294002 (25). Mammalian class II PI3K is exemplified by the 190-kDa PI3K-C2␣ (26). This kinase is refractory to inhibition by wortmannin and LY294002, and it predominantly uses Pt-dIns and PtdIns 4P as substrates in vitro and can phosphorylate PtdIns 4,5 P 2 in the presence of phosphatidylserine (26). The third class of PI3K is structurally related to Saccharomyces cerevisiae Vps34. This is a PtdIns-specific 3-kinase, and it associates with p150 for enzymatic activity (27). PI3K has a diverse range of roles in cell biology including the promotion of cell survival, mediation of the actions of insulin, and regulation of neutrophil chemotaxis in response to bacterial peptides (25,28). We now report that 20:46 stimulated the activity of PI3K in human endothelial and myeloid cells. A role for ErbB4 in the activation of PI3K by 20:46 in the endothelial cells was also established. Although -hydroxylation of the fatty acid by cytochrome P450 monooxygenases was not required for the stimulation of PI3K activity in neutrophils, metabolism of 20:46 by cyclooxygenases and lipoxygenases was required. These data suggest that some of the actions of 20:46 may be mediated by PI3K.

EXPERIMENTAL PROCEDURES
Materials-Polyunsaturated fatty acids were purchased from Sigma-Aldrich or from Cayman (Ann Arbor, MI). 17-octadecynoic acid was purchased from Cayman. Formyl methionyl leucyl phenylalanine (fMLP), myelin basic protein, protein A-Sepharose, phosphatidylinositol, and general reagents for kinase assays were from Sigma-Aldrich. Lavendustin A and B and AG-1478 were purchased from Biomol (Plymouth Meeting, PA), and AG-1296 was purchased from Calbiochem-Novabiochem (Victoria, Australia). [␥-32 P]ATP (specific activity 4000 Ci/mmol) was obtained from Geneworks (Adelaide, Australia). Polyclonal anti-p85␣, ERK2, ErbB2, ErbB3, ErbB4, and Akt antibodies and the anti-phosphotyrosine antibody, PY99, were purchased from Santa Cruz Biotechnology. The anti-phosphotyrosine antibody, PY100, was obtained from New England Biolabs. Anti-ACTIVE ERK, antiphospho-Akt (Thr-308) and anti-phospho-Akt (Ser-473) antibodies were obtained from Promega (Madison, WI), New England Biolabs, and Upstate Biotechnology (Lake Placid, NY), respectively. The Western blot recycling kit was obtained from Alpha Diagnostics (San Antonio, TX). Enhanced chemiluminescence solutions and reinforced nitrocellulose were from NEN Life Sciences Products (Boston, MA) and Schleicher and Schuell (Dassel, Germany), respectively. Arachidonic acid was dissolved in ethanol, and fMLP, 17-octadecynoic acid, and the tyrphostins, AG-1478 and AG-1296, were dissolved in dimethyl sulfoxide (Me 2 SO). The final concentrations of the vehicles were: 0.1% ethanol (v/v) and 0.1% Me 2 SO (v/v). Control cells received an equivalent amount of a vehicle that did not affect cellular responses.
Cell Culture-HL60 cells were maintained in RPMI 1640 medium supplemented with 10% fetal calf serum and antibiotics at up to 1 ϫ 10 6 cells/ml. Human neutrophils were isolated from the peripheral blood of healthy volunteers by the rapid single-step method of Ferrante and Thong (29). The preparations of neutrophils were at least 98% pure and 99% viable as judged by morphological examination of cytospin preparations and the ability of viable cells to exclude trypan blue. Human umbilical vein endothelial cells (HUVECs) were isolated by collagenase treatment and cultured as described previously (30). Second passage cells (0.25 ϫ 10 6 ) were plated in 10-cm culture dishes and were used after 4 days. Chinese hamster ovary cells stably expressing ErbB2 or ErbB3 were maintained as described previously (31). All cells were washed twice with Hanks' balanced salts solution (HBSS) 30 min before being incubated with 20:46 or vehicle.
Preparation of Cellular Extracts-Incubations were terminated by either removing the incubation medium and washing the cells once with HBSS (4°C) (HUVEC) or adding an equal volume of cold (4°C) HBSS into the incubation tubes followed immediately by centrifugation (3000 ϫ g, 3 min, 4°C). The cells were lysed by incubation in buffer A (20 mM Hepes, pH 7.4, 0.5% (v/v) Nonidet P-40, 100 mM NaCl, 1 mM EDTA, 2 mM Na 3 VO 4 , 2 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 10 g/ml each of leupeptin, aprotinin, pepstatin A, and benzamidine) for 2 h with constant mixing (4°C) as described previously (20). After lysis, cell debris was sedimented (12,000 ϫ g, 30 s), and the protein content of the soluble fractions was determined by the Lowry method. For detection of Akt phosphorylation and the tyrosine phosphorylation of cellular proteins by Western blot analysis, the incubations were terminated by the removal of the incubation medium, and Laemmli buffer was added. The samples were boiled (5 min) and stored (Ϫ20°C).
Immunoprecipitation-Lysates containing equal amounts of protein (0.5-1 mg) were precleared with protein A-Sepharose (15 l, 20 min, 4°C) before being incubated with the anti-p85␣ subunit antibody or PY99 (3 g/sample). After mixing for 2 h (4°C), the immune complexes were precipitated by the addition of protein A-Sepharose (15 l). The immunoprecipitates were collected by centrifugation (16,000 ϫ g, 15 s) and washed once with buffer A (4°C). For PI3K activity assays, samples that were immunoprecipitated with the anti-p85␣ subunit antibody were washed once with buffer A, followed by another wash with buffer B (10 mM Tris/HCl, pH 7.6, 100 mM NaCl, 1 mM EDTA, and 100 M Na 2 VO 4 ), and resuspended in 50 l of buffer B. For Western blot analyses, the samples that were immunoprecipitated with PY99 were mixed with Laemmli buffer and boiled (5 min).
PI3K Assay-The activity of PI3K was assayed essentially as described previously (32). Briefly, MgCl 2 (10 l, stock concentration of 100 mM) and PtdIns (10 l, stock concentration of 4 g/ml sonicated in 10 mM Tris/HCl containing 1 mM EDTA, pH 7.6) were added to the above prepared samples, and the reaction was started by the addition of 15 Ci of [ 32 P]ATP. The tubes were incubated at room temperature for 10 min with constant mixing. The reaction was terminated by the addition of 40 l of 8 M HCl and 320 l of CHCl 3 /CH 3 OH (1:1). After vigorous mixing and centrifugation the aqueous phase was removed, and the chloroform phase was washed (three times) with 500 l of CH 3 OH/H 2 O (1:1). The lipid samples were applied to Silica gel 60 thin layer chromatography plates that were developed with CHCl 3 /CH 3 OH/H 2 O/30% NH 4 OH (90:70:14.6:5.4 v/v) (32). The PtdIns 3 32 P spots were located, and the radioactivity was determined using an Instant Imager (Packard Instruments).
Western Blotting-Equal amounts of denatured proteins per lane were separated on 12% (8% for the analysis of ErbB receptors) polyacrylamide gels, transferred to nitrocellulose (100 V, 1.5 h), and immunodetected as described previously (20). Immediately after transfer, the blots were stained with Ponceau S (0.1% in 5% acetic acid) to confirm the even transfer of proteins within all the lanes of the gels. The blots that showed dissimilar protein levels between the lanes were discarded. Anti-ERK2, ACTIVE ERK, p85␣ subunit, Akt, and phospho-Akt antibodies were used to detect ERK2, dual-phosphorylated ERK1 and ERK2, the p85␣ subunit, Akt, and phosphorylated Akt, respectively. To detect ErbB receptor phosphorylation, samples that were immunoprecipitated with PY99 were Western blotted first with an anti-ErbB2 antibody. The blots were stripped using the Western blot recycling kit and reblotted with the anti-ErbB4 antibody, and the stripping and reprobing steps were repeated for the detection of ErbB3. To confirm that tyrosine-phosphorylated proteins remained bound to the nitrocellulose, the blots were finally probed with the anti-p85␣ subunit antibody. Immunocomplexes were detected by enhanced chemiluminescence (20).
Statistical Analysis-Where appropriate, differences were analyzed by analysis of variance or unpaired Student's t test and were considered significant at p Ͻ 0.05.

RESULTS
The effect of 20:46 on the activity of PI3K was investigated in cell types that had been shown previously to respond to this fatty acid (2,6,13). Incubation of neutrophils (Fig. 1a), HL60 cells (Fig. 1b), and HUVECs ( Fig. 1c) with 20:46 increased the activity of PI3K. Because the anti-p85␣ antibody that was used to immunoprecipitate the kinase cross-reacts with all the currently known forms of the regulatory subunit of class Ia PI3Ks, the data represent activation of class Ia PI3K. Although the effect of 20:46 on kinase activity was transient in neutrophils and HL60 cells, peaking at 2 and 5 min, respectively, PI3K activity in HUVECs increased with time over the 10-min incubation period (Fig. 1). The stimulatory effect of 20:46 on PI3K activity was detectable at 5 M fatty acid (Fig. 2). Although PI3K activity in neutrophils increased in a dose-dependent manner, increasing the fatty acid concentration above 5 M did not produce a greater effect on kinase activity in HL60 cells.
Consistent with the above data, incubation of neutrophils with 20:46 increased the level of tyrosine phosphorylation of the p85␣ subunit of PI3K (Fig. 3). The kinetics for this change in phosphorylation were similar to those found for the increase in kinase activity (Fig. 1a). To confirm that tyrosine phosphorylation was required for the activation of PI3K by 20:46, neutrophils and HUVECs were preincubated with lavendustin A, a tyrosine kinase inhibitor. The data in Fig. 3, b and c, show that pretreatment of HUVECs and neutrophils, respectively, with lavendustin A for 10 min caused a dose-dependent inhibition of the ability of 20:46 to stimulate PI3K. Lavendustin B, a negative control for lavendustin A, did not affect the ability of 20:46 to stimulate PI3K activity (data not shown). These data demonstrate that 20:46 stimulated the activity of PI3K in a tyrosine kinase-dependent manner.
Lavendustin A was discovered as an inhibitor of the epidermal growth factor (EGF) receptor (34). Our previous studies in WB rat liver epithelial cells have demonstrated that 20:46 caused the appearance of a major tyrosine-phosphorylated band that migrated on SDS-polyacrylamide gel electrophoresis (12% gels) with a molecular mass that was in excess of 138 kDa (20). This protein was also strongly tyrosine-phosphorylated in cells that were stimulated with either lysophosphatidic acid or EGF (20). Analyzed on 8% SDS gels, lysates from 20:46stimulated HUVECs were found to contain two major proteins (molecular masses of 180 -190 and 85 kDa) that exhibited  (20 M) for the indicated times at 37°C and lysed. Fractions were assayed for PI3K activity as described under "Experimental Procedures." To visualize the PtdIns 3P spots and quantify the radioactivity associated with the phosphorylated lipid, the thin layer chromatography plates were loaded onto an Instant Imager. A representative image is shown for each set of experiments. The histograms (-fold stimulation) represent means Ϯ S.E. of 3-4 separate experiments. Each experiment with neutrophils and HUVECs was conducted using cells from a different donor or cord, respectively. Significance of difference: p Ͻ 0.05 by analysis of variance. enhanced levels of tyrosine phosphorylation compared with those from control cells (Fig. 4a). Interestingly, in lysophosphatidic acid-stimulated COS and Vero cells (35), a tyrosine-phosphorylated protein of 170 -180 kDa has been identified recently as the EGF receptor. It was found that transactivation of the EGF receptor was necessary for the stimulation of PI3K by lysophosphatidic acid (35). Although the 85-kDa protein (Fig.  4a) was likely to be the p85␣ subunit of PI3K, the 180 -190-kDa protein could be a member of the ErbB receptor family. This receptor family is composed of ErbB1 (the EGF receptor), ErbB2, ErbB3, and ErbB4 (36), and HUVECs have been reported to express ErbB2, ErbB3, and ErbB4 but not ErbB1 (37). Lysates from HUVECs therefore were subjected to immunoprecipitation with PY99 and Western blotting with an antibody against ErbB2, ErbB3, or ErbB4. The data in Fig. 4c demonstrate that 20:46 selectively enhanced the tyrosine phosphorylation of ErbB4. This effect was partially reduced by AG-1478, an EGF receptor-specific inhibitor that also inhibits ErbB2, ErbB3, and ErbB4 (38). 20:46 did not stimulate the tyrosine phosphorylation of ErbB2 (Fig. 4b) or ErbB3 (Fig. 4d). The ability of the anti-ErbB2 and anti-ErbB3 antibodies to detect ErbB2 and ErbB3, respectively, was confirmed in Chi-nese hamster ovary cells transfected with ErbB2 or ErbB3 (31) (Fig. 4).
To determine whether transactivation of ErbB4 was required for the stimulation of PI3K activity by 20:46, HUVECs were pretreated with AG-1478. The data in Fig. 5 demonstrate that AG-1478 suppressed PI3K activation by 20:46. In contrast, AG-1296, a specific platelet-derived growth factor receptor inhibitor, did not affect the ability of 20:46 to stimulate PI3K activity in HUVECs (data not shown). These data are consistent with transactivation of ErbB4 being required for the activation of PI3K in HUVECs by 20:46. The data with lavendustin A (Fig. 3c) suggest that a member of the ErbB receptor family could also be involved in the activation of PI3K by the fatty acid in neutrophils.
Our previous studies have demonstrated that the ability of 20:46 to stimulate the activity of p38 was not suppressed by inhibitors of the cyclooxygenases and lipoxygenases (22). To investigate whether inhibitors of the cyclooxygenases and li- HUVECs were incubated with 20:46 (20 M) for 8 min and lysed, and the lysates were subjected to Western blot analysis using PY100 (a). In other experiments, the cells were also pretreated with AG-1478 (100 or 300 nM) for 10 min before being incubated with 20:46 (b-d). After lysis, tyrosine-phosphorylated proteins were immunoprecipitated (ip) with PY99, and phosphorylated ErbB receptors were detected by Western blot analyses (wb) (b-d). The nitrocellulose was probed sequentially with anti-ErbB2, anti-ErbB4, and anti-ErbB3 antibodies using a stripping procedure between each antibody as described under "Experimental Procedures." As a control for protein loading in addition to Ponceau S staining, an unidentified tyrosine-phosphorylated protein (arrow) that reacted with the anti-ErbB2 antibody within the same blot is shown. The level of phosphorylation of this protein was not affected by the treatments. The antigenicity of the anti-ErbB2 and anti-ErbB3 antibodies was confirmed in Chinese hamster ovary cells. Numbers on the side of the blots represent molecular weights. Numbers in parentheses represent band density (arbitrary units). Results are representative of three experiments.
poxygenases affected the ability of 20:46 to stimulate PI3K activity, neutrophils were preincubated with indomethacin and nordihydroguaiaretic acid, respectively. The data in Fig. 6a demonstrate that indomethacin and nordihydroguaiaretic acid each partially inhibited 20:46-stimulated PI3K activity. In contrast, the cytochrome P450 monooxygenase inhibitor, 17octadecynoic acid, did not affect kinase activity (data not shown). These data suggest that metabolism of 20:46 by cyclooxygenases and lipoxygenases but not by cytochrome P450 monooxygenases was required for stimulation of PI3K activity. The mechanisms by which 20:46 stimulated the activity of PI3K and p38 were therefore distinct. The 5-lipoxygenase is the most prominent lipoxygenase in neutrophils (3), and activated neutrophils liberate 5-hydroxyeicosatetraenoic acid (5-HETE) into the incubation medium. In support for a role for the lipoxygenases in the action of 20:46 on PI3K activity, incubation of neutrophils with 5-HETE resulted in the activation of class Ia PI3K (Fig. 6b). These data suggest that 5-HETE may mediate, at least in part, the effect of 20:46 on PI3K activity in neutrophils.
A hallmark of receptor-mediated activation of class Ia PI3K is the activation of Akt (25). This is mediated through the phosphorylation of Akt on Thr-308 and Ser-473 by PDK1 and PDK2 (possibly a modified PDK1) in the presence of PtdIns 3,4 P 2 and/or PtdIns 3,4,5 P 3 (25). Activation of PI3K is also accompanied by the activation of p70S6 kinase. Both PDK1 and Akt are implicated in mediating this response (39). Interestingly, it has been demonstrated that stimulation of p70S6 kinase activity by the 20:46 metabolites, 8,12-iso-isopro-staneF 2␣ -III and prostaglandin F 2␣ , in ventricular myocytes was not accompanied by Akt phosphorylation. This is despite the demonstration that the effects of 8,12-iso-isoprostaneF 2␣ -III and prostaglandin F 2␣ on p70S6 kinase were suppressed by the PI3K inhibitor, wortmannin (40). Because prostaglandin F 2␣ has been reported to stimulate PI3K activity in vascular smooth muscle cells (41), the data from the above study (40) imply that under certain circumstances activation of PI3K could be uncoupled from Akt activation. To investigate whether 20:46 was able to activate Akt, lysates from 20:46-stimulated neutrophils were examined for Akt phosphorylation. The data in Fig. 7 demonstrate that 20:46 caused the phosphorylation of Akt on Thr-308 (Fig. 7a) and Ser-473 (Fig. 7b). The bacterial tripeptide fMLP also stimulated the phosphorylation of Akt (Fig. 7a), consistent with its ability to activate class Ib PI3K␥ (28) and class Ia PI3K (Fig. 7c). Similar to the kinetics of PI3K activation by 20:46 (Fig. 1), the increase in Thr-308 phosphorylation was transient although Ser-473 phosphorylation was maintained for at least 10 min. The presence of approximately equal amounts of Akt within each lane was confirmed in duplicate blots stained with an anti-Akt antibody (data not shown). The data suggest a distinctive effect of 20:46 on the PI3K-Akt signaling pathway, one that is not observed with cyclooxygenase products of the fatty acid. The data also argue against of 8,12-iso-isoprostaneF 2␣ -III and prostaglandin F 2␣ being responsible for the stimulatory effect of 20:46 on PI3K and Akt activities in neutrophils.
PI3K has been reported to be an upstream regulator of the ERK1, ERK2, and p38 MAP kinase cascades when cells are stimulated by certain agonists (42,43). Given that 20:46 stimulates the activities of MAP kinases (20 -23), a possible role for PI3K in the regulation of ERK1 and ERK2 activation by 20:46 was investigated. The data in Fig. 8a demonstrate that the ability of 20:46 to stimulate ERK dual phosphorylation in neutrophils was not inhibited by wortmannin, even at a concentration of 100 nM. A duplicate blot stained with the anti-ERK2 antibody confirmed the presence of approximately equal amounts of ERK2 within each lane (data not shown). Although the data show strong ERK2 (p42-phospho-ERK band) phosphorylation in samples prepared from 20:46-stimulated cells, phospho-ERK1 (the p43-phospho-ERK band) was also detectable, albeit less consistently than phospho-ERK2. However, upon longer exposure of the nitrocellulose sheet to film, the phospho-ERK1 bands were readily visible (data not shown). Wortmannin (1-100 nM) also failed to suppress p38 activation by 20:46 in neutrophils (data not shown). These results reinforce the notion that different mechanisms were involved in the stimulation of PI3K and MAP kinase activities by 20:46. To determine whether PI3K plays a functional role in 20:46stimulated cells, the effect of wortmannin on 20:46-stimulated neutrophil respiratory burst was investigated. The data in Fig. 8b demonstrate that wortmannin dose-dependently inhibited 20:46-stimulated superoxide production with an IC 50 of ϳ1 nM (Fig. 8b). Neutrophil viability was not affected as determined by the trypan blue exclusion test.
In addition to 20:46, fatty acids in the 3 and 9 series are also biologically active. Both docosahexaenoic acid (22:63) and oleic acid (18:19) are released from membrane phospholipids by the group II and group V secretory phospholipase A 2 , respectively, in activated cells (44,45). Although some fatty acids may differ in their actions and their ability to affect cellular functions, different fatty acids have been demonstrated to mimic each other with respect to certain actions such as stim- ulation of the neutrophil respiratory burst (3,6,11). The data in Fig. 9 demonstrate that 22:63 and 18:19, at equimolar concentrations to 20:46, also stimulated the activity of PI3K, albeit to a slightly lower degree than 20:46. In contrast, the saturated fatty acid, stearic acid, which does not affect the neutrophil respiratory burst (3), did not increase PI3K activity. These data suggest that the ability of fatty acids to stimulate PI3K activity is restricted to mono-and polyunsaturated fatty acids.

DISCUSSION
Exogenously applied 20:46 alters cellular functions at concentrations reported to be present in stimulated cells (46). This is likely to be because of the ability of 20:46 or metabolites to stimulate the activities of intracellular signaling molecules such as PKC, the MAP kinases, and phospholipase A 2 or to modulate ion channel activity and cause an increase in intracellular Ca 2ϩ concentration (14 -24). The data in the present study demonstrate that the addition of exogenous 20:46 stimulated the activity of class Ia PI3K.
An interesting finding of the present study is that 20:46 activates ErbB4 in HUVECs and that inhibition of this recep-tor by AG-1478 suppressed the ability of the fatty acid to stimulate PI3K. Although neutrophils have not been demonstrated to express ErbB receptors, these leukocytes produce heparin-binding EGF-like growth factor (47) that binds ErbB1 or ErbB4 (36). The data with lavendustin A suggest that a member of the ErbB receptor family could also be involved in neutrophils. Although our previous studies in WB epithelial cells have demonstrated that 20:46, lysophosphatidic acid, and EGF each stimulated the tyrosine phosphorylation of a protein with a molecular mass in excess of 138 kDa (20), an accurate assessment of the molecular mass of the protein in the 12% gel was not possible. Nevertheless, our data (20) hinted at the possibility that the tyrosine-phosphorylated protein was the EGF receptor. Interestingly, 20:46 has been reported recently to stimulate the tyrosine phosphorylation of the EGF receptor in renal proximal tubule epithelial cells (48). Our demonstration that ErbB4 (HUVECs) or another member of the ErbB receptor family (neutrophils) is involved in the activation of PI3K not only provides a novel mechanism by which 20:46 stimulated the activity of this lipid kinase but also significantly enhances our understanding of how the actions of 20:46 are mediated. Although transactivation of the EGF receptor has been reported for an increasingly large number of non-EGF-like ligands, our novel results demonstrate that ErbB4 can also be transactivated. How 20:46 stimulated the tyrosine phosphorylation of ErbB4 in HUVECs remains to be investigated.
Although PI3K has been reported to be an upstream regulator of MAP kinases under certain circumstances (42,43), our data in neutrophils demonstrate that stimulation of ERK1, ERK2, and p38 activities by 20:46 did not require PI3K. This is in direct contrast to reports that fMLP-and interleukin 8-stimulated ERK1/ERK2 activation in neutrophils depends on PI3K (43,49). Similarly, insulin-stimulated ERK1/ERK2 acti- vation in rat adipocytes requires PI3K (42). However, this cross-talk between PI3K and the MAP kinase signaling pathways does not operate ubiquitously (25) and may depend on a number of factors. Thus, although antigen receptor activation of the MAP kinase kinase kinase was inhibited by wortmannin, suggesting a PI3K involvement, EGF stimulation of MAP kinase kinase kinase was wortmannin-insensitive (50). Even within the same cell type, the culture condition may affect the dependence of the MAP kinase pathways on PI3K. Thus, insulin-like growth factor 1-stimulated activation of ERK1 and ERK2 in adherent MCF-7 cells was inhibited by the PI3K inhibitor, LY294002, whereas stimulation of ERK1 and ERK2 activity by this growth factor in MCF-7 cells cultured in suspension was not (51). The ability of wortmannin to inhibit MAP kinase activation may depend on signal strength (52). Furthermore, because PI3K has been demonstrated to translocate to different cellular compartments in an agonist-dependent manner (53), it is conceivable that the coupling of PI3K to individual MAP kinase modules may also depend on the compartment in which members of the PI3K-Akt signaling pathway are localized. Based on our previous demonstrations that 20:46 stimulated the translocation of PKC to a particulate fraction and that depletion of cellular PKC prevented the activation of ERK1 and ERK2 by 20:46 (20), the evidence supports PKC but not PI3K being an upstream regulator of ERK1 and ERK2 in 20:46-stimulated cells. This is despite PDK1 being implicated in phosphorylating the activation or T-loop of AGC Ser/ Thr protein kinases (PKA, PKG, and PKC) including a number of PKC isozymes and MAP kinases (25).
Although cyclooxygenase and lipoxygenase inhibitors did not suppress 20:46-stimulated activation of p38 in neutrophils (22), stimulation of PI3K activity in this cell type depended, at least in part, on products of the cyclooxygenases and lipoxygenases. Metabolism of the fatty acid by cytochrome P450 CYP4A gene family members was not required. Despite the demonstration that prostaglandin F 2␣ (41) up-regulated the activity of PI3K and the implied activation of PI3K by 8,12-iso-isopro-staneF 2␣ -III (40), an involvement of these metabolites in the activation of the PI3K-Akt pathway by 20:46 was excluded. Neither 8,12-iso-isoprostaneF 2␣ -III nor prostaglandin F 2␣ stimulated Akt phosphorylation (40). Recently, the peptido-leukotriene, leukotriene D 4 , has also been reported to stimulate the activity of PI3K in renal mesangial cells (54). However, it is unlikely that leukotriene D 4 was involved in the stimulation of FIG. 8. Suppression by wortmannin of 20:46-stimulated superoxide production but not ERK dual phosphorylation. Neutrophils (1 ϫ 10 6 cells in 1 ml of HBSS for superoxide production or 3 ϫ 10 7 in 30 ml of HBSS for ERK phosphorylation) were preincubated with wortmannin or Me 2 SO (0.01%) for 10 min before being incubated with 20:46 (20 M) or ethanol. ERK dual phosphorylation (a) and the production of superoxide (b) were determined as described under "Experimental Procedures." Numbers in parentheses represent the density of phospho-ERK2 bands. The data on superoxide production (means Ϯ S.E., n ϭ 3) are expressed as chemiluminescence (mV). Significance of difference between wortmannin-pretreated and control cells: *, p Ͻ 0.001.  (55). In contrast to our data with 20:46, leukotriene D 4 -stimulated PI3K was not affected by AG-1478 but was suppressed by AG-1296 (54). Further studies will be needed to determine whether the abovedescribed differences are caused by tangible differences in action between leukotriene D 4 and prostaglandin F 2␣ on the one hand and 20:46 or a metabolite(s) of the fatty acid other than leukotriene D 4 , isoprostanes, and prostaglandins on the other. Our data obtained from neutrophils suggest that a 5-lipoxygenase product such as 5-HETE may be involved in mediating the effect of 20:46 on PI3K activity in this cell type, and this is consistent with the 5-lipoxygenase being the most prominent lipoxygenase in neutrophils (3).
It is therefore clear that a number of 20:46 metabolites can stimulate the activity of PI3K, albeit with differences in the ability of these metabolites to activate Akt, to couple PI3K to MAP kinase modules, and the differential utilization of the EGF and platelet-derived growth factor receptor (40,41,54). Interestingly, 20:46 metabolites that are capable of inhibiting PI3K have also been identified. Most notable is lipoxin A 4 . This metabolite, formed by a transcellular biosynthetic route (56), has been found to inhibit the ability of leukotriene D 4 to stimulate PI3K (54). One of the best characterized actions of lipoxin A 4 is its ability to negatively regulate neutrophil and eosinophil function (56). It is possible that in these cell types, the response to ligands that simultaneously mobilize 20:46 and activate PI3K will by modulated by the balance between the levels of lipoxins and those of the other metabolites in the milieu of the cells. 3 and 9 fatty acids, but not the saturated stearic acid, are also capable of activating PI3K despite being the sources of vastly different metabolites. This implies that 20:46-derived metabolites are not strictly required for the stimulation of PI3K by an unsaturated fatty acid. In in vitro assays, 20:46 and 18:19 directly activate purified PKC (14), whereas in intact cells, the 3 fatty acids are able to stimulate ERK1 and ERK2. 2 Thus, the ability of 22:63 and 18:19 to stimulate the activity of PI3K is consistent with their ability to evoke some of the cellular responses observed with 20:46. These responses include the induction of HL60 cell differentiation and the stimulation of neutrophil and macrophage superoxide production, adhesion to plastic, and up-regulation of CD11b/CD18 (3,6). In contrast, saturated fatty acids are devoid of these activities.
Although not generally considered to be second messenger molecules, there is evidence that 22:63 and 18:19 are released from membrane phospholipids in activated cells by the group II and group V secretory phospholipase A 2 , respectively, (44,45). The levels of the secretory phospholipase A 2 are known to be elevated in tissue fluids under certain pathophysiological conditions such as in the synovial fluid of rheumatoid arthritis patients (57). Thus, the presence of these unsaturated fatty acids in these fluids is a possibility. Given the current interest in diet manipulation in favor of monounsaturated and 3 fatty acid, our finding that 3 and 9 fatty acids stimulated PI3K activity may be physiologically relevant. We speculate that the release of these fatty acids by the secretory phospholipase A 2 in activated cells will continue to provide the modulation of PI3K by fatty acids, despite the shift in the membrane phospholipid fatty acid content away from 6 to 3 and 9 species. Our demonstration that PI3K plays an important role in neutrophil-mediated articular cartilage destruction in rheumatoid arthritis (32) highlights the dilemma in separating the antiinflammatory properties of 3 fatty acids from their stimulatory actions on neutrophil respiratory burst (3), a process that depends on PI3K.
A role for PI3K in 20:46-stimulated superoxide production was established in this study. Activation of the PI3K-Akt signaling pathway is known to exert an antiapoptotic effect on certain cell types, possibly as a consequence of the actions of Akt on components of the apoptotic machinery such as phosphorylation of BAD (25). The PI3K-Akt pathway has also been implicated in the promotion of cellular proliferation (25). Interestingly, 20:46 has been reported to promote the growth of certain tumor (9) and nontumor cell types (10). The enhancement of carcinosarcoma cell growth by 20:46 has also been shown to be accompanied by the suppression of apoptosis (9). Thus, it is possible that the growth-promoting and apoptosissuppressing effects of 20:46 on these cell types could be mediated, at least in part, via the activation of PI3K and Akt. In HL60 cells, 20:46 and 3 fatty acids have been reported to induce granulocytic differentiation, which is accompanied by the suppression of proliferation and the induction of cell death by apoptosis and necrosis (6). This action is similar to that induced by all-trans-retinoic acid, a known inducer of granulocytic differentiation (6). Because PI3K is required for the induction of granulocytic differentiation by retinoic acid (58) tion program may over-ride the survival signals provided by the PI3K-Akt axis, thereby giving rise to the cell death that accompanies differentiation.
We have suggested previously that 20:46 liberated through the activation of the cytosolic phospholipase A 2 could amplify or prolong intracellular signaling initiated by a receptor that is coupled to the enzyme (20,22). In this scenario, the binding of an agonist to its receptor results in the activation, on the one hand, of signaling molecules such as p21 ras , PKC, ERK1, ERK2, and/or p38, and on the other hand, the activities of other signaling molecules such as PI3K and Akt. Activated ERK1, ERK2, and/or p38 phosphorylate the cytosolic phospholipase A 2 (59,60), and if the intracellular Ca 2ϩ concentration is also elevated after receptor engagement, phospholipid hydrolysis by the cytosolic phospholipase A 2 occurs. This sets up a signaling loop. The demonstration that exogenous 20:46 stimulates the activities of PKC, ERK1, ERK2, p38, the cytosolic and secretory phospholipase A 2 , PI3K, and the phosphorylation of Akt (20,22,24) is consistent with such a loop. Current evidence, although incomplete, suggests that such a loop may exist, at least in human neutrophils. Thus, ligation of the fMLP or Fc␥ (Fc␥RIIa or Fc␥RIIIb) receptors results in the activation of all the above-listed signaling molecules and liberation of 20:46 (61)(62)(63). To further prove the existence of the loop, it is also essential to demonstrate that inhibition of phospholipase A 2 decreases the magnitude of receptor-mediated activation of signaling molecules that are upstream of phospholipase A 2 such as the MAP kinases. Although evidence for this in neutrophils is still lacking, studies in human monocytic cells have demonstrated that inhibition of either the activity or the expression of phospholipase A 2 resulted in the attenuation of ligand-stimulated activation of MAP kinases (64). The ability of inhibitors of phospholipase A 2 , ERK, or p38 modules to partially inhibit not only fMLP-(32, 63) but also 20:46-(24, 65) stimulated superoxide production suggests that this positive feedback loop in neutrophils is functional. The idea of positive feedback loops in intracellular signaling involving the PKC-ERK1 and ERK2-phospholipase A 2 loop has recently been the subject of a modeling study (66). Depending on the strength and duration of the initial signal at the cell surface, the model predicts that signaling via these loops can become self-sustaining, and this is supported by our observation that exogenous 20:46 stimulated the activation of the cytosolic and secretory phospholipase A 2 (24). The model proposes that the level of MAP kinase phosphatase-1 expression is one mode by which signaling via the PKC-ERK-phospholipase A 2 positive feedback loop can be terminated (66). Interestingly, 20:46 has been reported to induce the expression of MAP kinase phosphatase-1 (5). Because our data in neutrophils show that PI3K is not required for the activation of MAP kinases by 20:46, it is unlikely that PI3K is a participant in the loop, at least in neutrophils.
In summary, the present study establishes that unsaturated fatty acids such as 20:46 are capable of stimulating the class Ia PI3K-Akt signaling pathway. Although eicosanoid products of 20:46 are required for the activation of PI3K by 20:46 in myeloid cells, our data strongly suggest that a member of the ErbB receptor family is also required. In HUVECs, this was ErbB4. Fig. 10 summarizes the major signaling pathways that 20:46 has been found to activate in intact cells.