15( S )-Lipoxygenase-2 Mediates Arachidonic Acid-stimulated Adhesion of Human Breast Carcinoma Cells through the Activation of TAK1, MKK6, and p38 MAPK*

The dietary cis -polyunsaturated fatty acid, arachidonic acid, stimulates adhesion of metastatic human breast carcinoma cells (MDA-MB-435) to the extracellular matrix, but the molecular mechanisms by which fatty acids modify the behavior of these cells are unclear. Exposure to arachidonic acid activates multiple signaling pathways. Activation of p38 mitogen-activated protein kinase (p38 MAPK) is required for increased cell adhesion to type IV collagen, and this activation is sensitive to inhibitors of lipoxygenases, suggesting a requirement for arachidonic acid metabolism. The goals of the current study were to identify the one or more key metabolites of arachidonic acid that are responsible for activation of p38 MAPK and to elucidate the upstream kinases that lead to p38 MAPK activation. High performance liquid chromatographic analysis revealed that MDA-MB-435 cells metabolize exogenous arachidonic acid predominantly to 15( S )-hydroxyeicosatetraenoic acid (15( S )-HETE). Immunoblot analysis with antibodies specific to 15( S )-lipoxy-genase-1 (LOX-1) and 15( S )-lipoxygenase-2 (LOX-2) demonstrated the expression protein type IV collagen through activation of p38 MAPK and protein kinase C pathways (21). In this study, we show that 15( S )-HETE is the primary metabolite of arachidonic acid exogenously added to these cells and that the addition of 15( S )-HETE recapitulated the effects of arachidonic acid. We report for the first time the identification of 15( S )-LOX-2 in MDA-MB-435 cells, a human metastatic breast carcinoma cell line, and, because these cells lack 15( S )-LOX-1, suggest that the LOX-2 isozyme is responsible for the generation of the active eicosanoid product that initiates a signal transduction cascade. of the sensitizes the cells to the addition of arachidonic acid, leading to higher levels of p38 MAPK activation. We have also defined the upstream kinases that lead to p38 MAPK activation after the generation of 15( S )-HETE and show that a growth factor receptor related kinase, is critical for the response of these cells to eicosanoid metabolites. These data show that a pathway usually associated with growth factor receptor signaling is activated by metabolites of dietary fatty acids.

The cis-polyunsaturated fatty acid arachidonic acid and its many metabolites are important mediators of cell signaling with roles in inflammation, platelet aggregation, tissue development, and carcinogenesis (1)(2)(3)(4)(5). The cis-polyunsaturated fatty acids have been implicated in a number of in vivo and in vitro rodent studies that link fat intake and cellular fatty acid levels with carcinogenesis, tumor development, and metastasis (6 -8). However, the specific pathways by which arachidonic acid and its metabolites alter the development and behavior of certain cancers are unclear (9,10).
The regulation of cell-cell and cell-matrix adhesion plays a key role in the migratory and invasive potential of tumor cells (17). The fate of these cell-cell and cell-matrix interactions depends largely on the cellular profile of adhesion molecules and matrix metalloproteases and on the signaling mechanisms that regulate adhesive behavior and cellular responses to environmental stimuli (18 -20). Thus, elucidating the effects of environmental components, such as fatty acids, on tumor cell adhesion is important for the understanding of metastasis.
Our laboratory has used the metastatic human breast carcinoma cell line, MDA-MB-435, to examine the cell-matrix adhesion-related effects of fatty acids on metastatic tumor cells. We first demonstrated that exogenous arachidonic acid rapidly stimulates an increase in the adhesion of MDA-MB-435 cells to the prevalent basement membrane protein, type IV collagen, in vitro (21). The inhibition of LOX activity blocks the ability of arachidonic acid to promote cell adhesion. These results suggested that arachidonic acid or its metabolites could affect metastasis through alteration of the adhesive properties of tumor cells and encouraged us to investigate the lipid-induced signaling pathways involved in this process. Our investigations showed that arachidonic acid activates both p38 MAPK and protein kinase C isozymes, leading to enhanced cell adhesion (22,23). This work demonstrated that both pathways are required for the fatty acid-induced adhesion. Our model has been previously described in detail (24). We have not yet determined the metabolites generated by the lipoxygenases, nor the isoforms of the LOX involved in this process. Nor is it clear which upstream kinases are responsible for the activation of the p38 MAPK in this system.
In this study, we examined how MDA-MB-435 human carcinoma cells metabolize arachidonic acid and how the resulting lipid signals lead to activation of p38 MAPK. We show for the first time the presence of 15(S)-LOX-2 in breast carcinoma cells and demonstrate the involvement of upstream kinases TAK1 and MKK6 in the activation of p38 MAPK by the arachidonic acid metabolite, 15(S)-HETE.  (Thr-180 and Tyr-182), ASK1, and phosphorylated ASK1 (Ser-83) were from Cell Signaling (Madison, WI). Antibodies against total p38␣ MAPK (anti-SAPK2a) and TAK1 were from Upstate Biotechnology (Lake Placid, NY). Cell culture media and supplements, electrophoresis buffers, LDS sample buffer (141 mM Tris, 2% (w/v) lithium dodecylsulfate, 10% (v/v) glycerol, 0.51 mM EDTA, 0.22 mM Serva® Blue G250, 0.175 mM phenol red, pH 8.5) and pre-cast gels were from Invitrogen (Carlsbad, CA). Fetal bovine serum was from HyClone (Logan, UT). BCA assay reagents and Super Signal chemiluminescent detection reagents were from Pierce. All other reagents were from Sigma. The pcDNA3.1 and pcDNA3.1-LOX2 vectors were the kind gift of Dr. Thomas Eling (NIEHS, NIH) and have been previously described (14). The TAK1 dominant negative construct (pFLAG-TAK1K63W) was the kind gift of Dr. Marcello Arsura (25) with permission from Dr. Takahisa Sugita. FuGENE 6 was from Roche Applied Science.

Reagents-Arachidonic
Cell Culture-MDA-MB-435 cells were obtained from Dr. Janet Price (M.D. Anderson Cancer Center, Houston, TX). Cells were cultured in Eagle's minimal essential medium (MEM) supplemented with MEM vitamin solution, sodium pyruvate, L-glutamine, and 5% (v/v) fetal bovine serum. For most experiments, cells were harvested by incubation with Versene and resuspended in serum-free MEM.
HPLC Analysis of Arachidonic Acid Metabolites-MDA-MB-435 cells suspended in serum-free MEM (3.5 ϫ 10 5 cells/ml) were equilibrated at 37°C and treated with [ 3 H]arachidonic acid (15 Ci) plus unlabeled arachidonic acid (30 M), A23187 (5 M), or ethanol vehicle. Some groups were pretreated with NDGA (10 M) for 15 min prior to addition of the radiolabel. At 2, 5, 10, and 15 min after arachidonic acid exposure, an equal volume of methanol was added to quench enzymatic reactions. Radiolabeled fatty acids were isolated from the incubation mixture by a C18-PrepSep solid-phase extraction column (Waters Associates, Milford, MA) after acidification to pH 3.5 with acetic acid. The column was washed with acidified water, and the metabolites were eluted with methanol, evaporated to dryness, and reconstituted with HPLC solvent. Reverse phase-HPLC analysis was conducted with an Ultrasphere ODS column (5 m, 4.6 ϫ 250 mm, Beckman) equipped with a Waters model 6000A pump and a Waters WISP 710B automatic injector. The mobile phase consisted of a methanol/water/acetic acid mixture (70/30/0.01, v/v/v) with a flow rate of 1.0 ml/min. Eluted radioactivity was monitored with a Flo-One Beta detector using Ecolume (ICN Biochemicals, Costa Mesa, CA) as the liquid scintillate.
Cell Adhesion Assay-MDA-MB-435 cells suspended in serum-free MEM (3.5 ϫ 10 5 cells/ml) were equilibrated at 37°C and exposed to fatty acids, inhibitors, or ethanol vehicle, transferred in triplicate to 48-well plates coated with either heat-denatured bovine serum albumin (negative control), type IV collagen, or poly-D-lysine (positive control) and incubated for 45 min at 37°C. Non-adherent cells were washed away; adherent cells were fixed in 6% (v/v) glutaraldehyde, washed, and quantified spectrophotometrically after staining with crystal violet. Cell adhesion to type IV collagen was calculated as a percentage of adhesion to poly-D-lysine after subtracting nonspecific cell adhesion in bovine serum albumin-coated wells.
Immunoblotting-MDA-MB-435 cells were solubilized in lysis buffer (50 mM Tris, pH 7.4, 1% (v/v) Nonidet P-40, 6 mM sodium deoxycholate, 150 mM NaCl, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM Na 3 VO 4 , 1 mM NaF, with 1 g/ml protease inhibitors: aprotinin, leupeptin, and pepstatin) for 30 min on ice. Extracts were cleared by centrifugation and assayed for protein content. Equal protein amounts were boiled in LDS sample buffer with 10% (v/v) ␤-mercaptoethanol, separated by SDS-PAGE, and transferred to nitrocellulose. Blocking and antibody incubations were carried out at recommended dilutions in 5% (w/v) bovine serum albumin or 5% (w/v) nonfat milk. Primary antibody complexes were detected using horseradish peroxidase-conjugated secondary antibodies, SuperSignal horseradish peroxidase substrate, and exposure to Hyperfilm. For detection of phosphorylated proteins, blots were probed and analyzed initially with antibodies specific to phosphorylated proteins, then incubated for 30 min at room temperature in stripping buffer (tris-buffered saline, 100 mM ␤-mercaptoethanol, 2% (w/v) SDS) to remove activation-specific antibodies, blocked in 5% nonfat milk, and reprobed with non-activation specific antibodies. Immunoblots were digitized and analyzed using Scion Image analysis software (Scion Corp., Frederick, MD).
Uptake of Radiolabeled Fatty Acids-MDA-MB-435 cells were grown as described and resuspended in serum-free MEM at 3.5 ϫ 10 5 cells/ml. [ 3 H]arachidonic acid (15 Ci) or [ 3 H]15(S)-HETE (15 Ci) was added to 5 ml of cells, and the mixture was allowed to incubate at 37°C for 2, 5, 10, 20, 30, or 45 min. Cells were centrifuged and washed with media twice. Aliquots of each fraction (media, wash, and cells) were transferred to liquid scintillation vials containing scintillation fluid, vortexed, and analyzed for radioactive content in a Beckman LS 6000LL liquid scintillation counter.
Radioimmunoprecipitation and Autoradiography-MDA-MB-435 cells were suspended in serum-free MEM (3.5 ϫ 10 5 cells/ml), equilibrated at 37°C, incubated for 30 min with [ 32 P]orthophosphate (2.0 mCi) and treated with 15(S)-HETE (0, 10, or 20 M), arachidonic acid (30 M), or transforming growth factor-␤ 1 (0.1 ng/ml) for 10 min. Cells were centrifuged, washed, and lysed in lysis buffer as stated previously. Lysates were pre-cleared with protein A-agarose beads and assayed for total protein. Equal amounts of protein were added to overnight reactions with 4 g of TAK1 polyclonal antibody. Immune complexes were precipitated with protein A-agarose beads, washed, boiled in LDS sample buffer with 10% ␤-mercaptoethanol, separated by SDS-PAGE, and transferred to nitrocellulose membranes for autoradiography. Bands corresponding to phosphorylated TAK1 were digitized and quantitated using Scion Image analysis software. The amount of 32 P incorporated into TAK1 was such that short exposures of TAK1-containing nitrocellulose membranes to Hyperfilm (up to 24 h) did not produce visible bands. Therefore, we performed immunoblot analysis to detect total TAK1 on the same membranes using rabbit anti-TAK1 and donkey anti-rabbit IgG and exposures of Ͻ1 h.
Transient Transfections-MDA-MB-435 cells were transfected with pFLAG-TAK1K63W (2 g/60-mm dish) using FuGENE per the manufacturer's instructions. After 12 h, cells were treated and lysed as stated previously. Transfections with the pcDNA3.1 and pcDNA3.1-LOX2 were also done with FuGENE, but the cells were allowed to express the LOX-2 for 48 h prior to treatment with arachidonic acid.
Statistical Analysis-One-way analysis of variance was used to analyze data from three to five separate experiments. Significant differences were determined with a p value set at Ͻ0.05.

HPLC Analysis of Lipoxygenase Metabolites of Arachidonic
Acid in MDA-MB-435 Cells-Arachidonic acid stimulates a number of changes in the behavior of MDA-MB-435 human breast tumor cells in a dose-dependent manner (22,23), but the metabolite responsible for these effects has not yet been identified. Thus, we identified lipoxygenase metabolites of arachidonic acid in cells treated with a stimulatory dose of arachidonic acid (30 M) plus [ 3 H]arachidonic acid (15 Ci). Reverse-phase HPLC revealed a single major peak of radiolabeled material in cell extracts prepared 5, 10, and 15 min after arachidonic acid exposure (Fig. 1A). The same peak was evident following exposure to the calcium ionophore A23187 (5 M), which also stimulates arachidonic acid release and metabolism in these cells. The radiolabeled material co-eluted with standard 15(S)-HETE, which is an arachidonic acid metabolite produced by 15(S)-LOX (data not shown). 15(S)-HETE production following arachidonic acid exposure was abolished by the LOX inhibitor NDGA in a dose-dependent manner, providing evidence that 15(S)-HETE production in MDA-MB-435 cells results from LOX-mediated metabolism of arachidonic acid ( Fig. 1B and Table I).
Under the conditions of our experiments, there are two possible pathways by which arachidonic acid may become accessible to cellular lipoxygenases: the exogenously added arachidonic acid may be metabolized directly as it is taken up, or it may first be incorporated into membrane phospholipids and consequently released by phospholipase activity. To determine whether phospholipase-dependent release of arachidonic acid is critical for the generation of this 15-(S)-HETE, we pretreated cells with the phospholipase A 2 inhibitor arachidonyltrifluoromethyl ketone and measured the production of LOX metab-olites following addition of exogenous [ 3 H]arachidonic acid. Arachidonyltrifluoromethyl ketone did not alter the production of [ 3 H]15(S)-HETE, nor did it block arachidonic acid-stimulated adhesion to type IV collagen (data not shown), suggesting that the arachidonic acid metabolized under these conditions was taken up from the exogenous pool and not released from membrane phospholipids by phospholipase A 2 .
Characterization of 15-Lipoxygenase Enzymes in MDA-MB-435 Cells-Two 15-LOX isoforms have been identified in human cells to date (13). To identify the 15-lipoxygenase isoform responsible for 15(S)-HETE production in MDA-MB-435 cells following arachidonic acid exposure, cell extracts from appropriate reference tissues and from MDA-MB-435 cells were separated by SDS-PAGE, transferred to nitrocellulose, and probed with specific antibodies directed against 15-LOX-1 or 15-LOX-2. Chemiluminescent detection of immune complexes showed that MDA-MB-435 cells did not express 15-LOX-1 at detectable levels as compared with an extract of normal human tracheobronchial epithelial cells, an abundant source of 15-LOX-1 ( Fig. 2A). In comparison, MDA-MB-435 cells express a 78-kDa protein corresponding to 15-LOX-2 (Fig. 2B). This conclusion was supported by immunodetection of a protein of similar size in an extract of normal human prostate epithelium, the tissue from which 15-LOX-2 was originally cloned (14). These results suggest that 15-LOX-2 is the isozyme responsible for the metabolism of arachidonic acid to 15 (Fig. 3). 15(R)-HETE did not increase cell adhesion to type IV collagen (data not shown), supporting the conclusion that the 15(S)-LOX metabolism of arachidonic acid is required to stimulate adhesion.
The stimulation of cell adhesion by exogenous 15(S)-HETE was slightly less than that observed with exogenous arachidonic acid (Fig. 3). One possibility for this difference is that the polarity of 15(S)-HETE hinders its uptake into MDA-MB-435 cells compared with the uptake of arachidonic acid. To address this issue, we measured the uptake of exogenous [ 3 H]15(S)-HETE and exogenous [ 3 H]arachidonic acid by MDA-MB-435 cells in suspension. Cells exposed to [ 3 H]arachidonic acid for 2 min incorporated Ͼ60% of the available lipid; whereas Ͻ5% of [ 3 H]15(S)-HETE became cell-associated (Fig. 4A). When the incubation time was extended to 45 min as in the adhesion assay (Fig. 3), Ͼ33% of the available 15(S)-HETE was incorpo-  The area under the 15(S)-HETE peaks from HPLC chromatographs (as in Fig. 1 15-LOX-2 Mediates Cell Adhesion through p38 MAPK Pathway rated into the cells (Fig. 4B). This result suggests that the diminished effect of exogenous 15(S)-HETE on cell adhesion to type IV collagen is due to decreased intracellular uptake of 15(S)-HETE compared with arachidonic acid.
p38 MAPK Activation following Lipoxygenase Inhibition-Our previous studies have shown that arachidonic acid treatment leads to activation of p38 MAPK and several proteins downstream of this kinase (22,23). This work showed that there was a dose response of p38 activation to arachidonic acid and that the inhibition of p38 activation completely blocked cell adhesion to collagen IV. To determine whether lipoxygenase metabolism is critical for the arachidonic acid-stimulated activation of p38 MAPK, immunoblot analysis was performed on extracts from MDA-MB-435 cell pretreated with NDGA then exposed to arachidonic acid. The amount of p38 MAPK activated after stimulation with arachidonic acid (Fig. 5, upper panels) was normalized to total p38 MAPK expression (Fig. 5, lower panels). At 1 and 5 min post treatment, arachidonic acid stimulated a 4-and 7-fold increase, respectively, in the activation of p38 MAPK in MDA-MB-435 cells compared with vehicle-treated cells. In the presence of NDGA, however, p38 MAPK activation by arachidonic acid at 1 and 5 min was inhibited by 65 and 55%, respectively, suggesting that lipoxygenase metabolism of arachidonic acid is required for the activation of p38 MAPK.
The lipoxygenase inhibitor, NDGA, is not specific for a single isoform of the enzyme. Thus, we used a vector that carries the 15(S)-LOX-2 gene to test whether increased expression of this isoform in the MDA-MB-435 cells would lead to increased activation of the p38 MAPK pathway following addition of arachidonic acid. Cells that were transfected with this vector expressed high levels of the 15(S)-LOX-2, as shown by immunoblots of cell extracts (Fig. 6). (Although not visible in this figure, the 15(S)-LOX-2 protein was detectable in untransfected cells upon longer exposures (data not shown).) Upon addition of arachidonic acid, these cells displayed a significantly higher level of p38 MAPK activation than in cells carrying the empty vector control. A quantitative analysis by densitometry indicated that the cells overexpressing the 15(S)-LOX-2 showed at least twice as much p38 MAPK activation when normalized to the amount of total p38 in the cells. We also observed increased activation of p38 at lower concentra- Cells were pretreated for 15 min with NDGA or vehicle, then exposed to arachidonic acid or vehicle for 1 or 5 min. Blots were probed with antibodies directed against phosphorylated p38 MAPK (active p38, top panels), stripped, and re-probed with antibodies directed against total p38 MAPK (total p38, bottom panels). Densitometric analysis was used to calculate the ratio of active:total p38 MAPK.
MKK6, and MKK3 revealed that 15(S)-HETE stimulated a dose-dependent activation of p38 MAPK that was similar to the stimulation of p38 MAPK by arachidonic acid (Fig. 7A). Neither 15(S)-HETE nor arachidonic acid activated the MAPK kinase MKK3 (Fig. 7B); furthermore, treatment with doses (30 M) of arachidonic acid that activated p38 did not activate MKK4 (data not shown). However, 15(S)-HETE stimulated a dose-dependent increase in the activation of MKK6 (Fig. 7B) similar to that produced by arachidonic acid, suggesting that MKK6 is activated in response to arachidonic acid exposure and metabolism to 15(S)-HETE.
Activation of TAK1 and ASK1 following Exogenous Arachidonic Acid or 15(S)-HETE Exposure-Given the activation of MKK6 in MDA-MB-435 cells following arachidonic acid or 15(S)-HETE exposure, we measured the activation of two MKKKs immediately upstream of MKK6: TAK1 and ASK1. Immunoprecipitation followed by autoradiography of 32 P-labeled protein showed 70-kDa proteins corresponding to TAK1 (Fig. 8A, top panels). This conclusion was verified by immunoblotting with TAK1 antibodies (Fig. 8A, bottom panels). Densitometric analysis revealed that a dose-dependent increase in the ratio of active:total TAK1 was produced by the addition of exogenous 15(S)-HETE. Exogenous arachidonic acid (30 M) and transforming growth factor-␤ 1 (1.0 ng/ml, positive control) exerted similar effects on TAK1 activation. Immunoblotting with phospho-ASK1-specific and total ASK1 antibodies showed a basal level of ASK1 phosphorylation in untreated cells; however, the ratio of phosphorylated:total ASK1 did not change with the addition of arachidonic acid or 15(S)-HETE (Fig. 8B).
Inhibition of p38 MAPK Phosphorylation by Dominant Negative TAK1-If activation of TAK1 were critical for the activation of the p38 MAPK pathway, inhibition of TAK1 by a dominant negative form of the kinase would be expected to block arachidonic acid-induced phosphorylation of p38 MAPK. To test this hypothesis, we transfected MDA-MB-435 cells with a pFLAG-TAK1K63W construct that expresses a dominant negative form of TAK1. Cells were then harvested and treated with arachidonic acid. The mock transfected cells showed a significant increase in the phosphorylation of p38 MAPK after treatment with arachidonic acid (Fig. 9). The expression of the TAK1 dominant negative protein blocked this activation, demonstrating that an active TAK1 is required for the fatty acidstimulated phosphorylation of p38 MAPK. DISCUSSION The adhesion of metastatic tumor cells to extracellular matrix proteins plays a key role in the metastatic process. Dietary fatty acid-mediated regulation of this process through specific signaling pathways is not well understood. We have shown previously that dietary fatty acids promote the adhesion of metastatic human carcinoma cells to the extracellular matrix  vector, lanes 2 and 3), pcDNA3.1-LOX2 (lanes 4 -6), or were mock transfected (lane 1) and incubated for 48 h. Cells were treated with vehicle (EtOH, Ϫ) or 30 M arachidonic acid (ϩ) for 5 min. Proteins from whole cell lysates (25 g) were analyzed by SDS-PAGE and immunoblotting. The blot was divided, and the high molecular weight portion was probed with antibodies to 15(S)-LOX-2 (upper panel), whereas the low molecular weight portion was probed with antibodies to phosphorylated p38 MAPK (middle panel). The phospho-p38 blot was stripped and re-probed with antibodies to total p38 (bottom panel). Nitrocellulose membranes containing immunoprecipitated, 32 P-labeled TAK1 were exposed to film (top panels) and probed with antibodies to total TAK1 (bottom panels). B, cells were exposed to 15(S)-HETE (0, 10, and 20 M) or arachidonic acid (30 M). Immunoblots were probed with antibodies to phosphorylated ASK 1 (top panel), stripped, and re-probed with antibodies to total ASK1 (bottom panel). Bands were analyzed densitometrically, and the ratio of phosphorylated:total kinase was calculated. For comparison of treatment groups, this ratio was adjusted to 1.00 in cells exposed to vehicle alone. Ratios for other treatment groups were then expressed relative to this ratio. protein type IV collagen through activation of p38 MAPK and protein kinase C pathways (21). In this study, we show that 15(S)-HETE is the primary metabolite of arachidonic acid exogenously added to these cells and that the addition of 15(S)-HETE recapitulated the effects of arachidonic acid. We report for the first time the identification of 15(S)-LOX-2 in MDA-MB-435 cells, a human metastatic breast carcinoma cell line, and, because these cells lack 15(S)-LOX-1, suggest that the LOX-2 isozyme is responsible for the generation of the active eicosanoid product that initiates a signal transduction cascade. Furthermore, overexpression of the 15(S)-LOX-2 sensitizes the cells to the addition of arachidonic acid, leading to higher levels of p38 MAPK activation. We have also defined the upstream kinases that lead to p38 MAPK activation after the generation of 15(S)-HETE and show that a growth factor receptor related kinase, TAK1 is critical for the response of these cells to eicosanoid metabolites. These data show that a pathway usually associated with growth factor receptor signaling is activated by metabolites of dietary fatty acids.
The stimulation of adhesion and invasion pathways by dietary fatty acids has important implications for tumor cell behavior, because environmental factors, such as local fatty acid concentrations, may alter the ability of metastatic cells to arrest and populate a secondary tumor site. Thus, understanding the metabolic and signaling pathways that control these effects is critical to developing strategies to combat metastasis. We set out first to establish the enzymes that metabolize arachidonic acid in these human carcinoma cells. LOX enzymes produce characteristic metabolites of arachidonic acid according to the position where the fatty acid chain is oxidized. HPLC analysis of LOX metabolites produced in MDA-MB-435 cells following exogenous arachidonic acid exposure showed that only one major LOX metabolite, 15(S)-HETE, is produced in these cells. Together, our HPLC and immunoblot data demonstrate for the first time to our knowledge the activity and presence of 15(S)-LOX-2 in these human carcinoma cells. Furthermore, the LOX inhibitor NDGA, previously shown to block arachidonic acid-stimulated cell adhesion to type IV collagen, inhibited the formation of 15(S)-HETE, and the adhesive response to exogenously added 15(S)-HETE mimicked the response to arachidonic acid. It was interesting that the level of 15(S)-HETE required to stimulate adhesion and p38 activation was similar to that of arachidonic acid, essentially micromolar levels. Our finding that the more polar HETE is not taken up by the carcinoma cells as readily as arachidonic acid probably explains this need, although it is also possible that 15(S)-HETE generated in vivo may be present at relatively high local concentrations near its immediate target. Therefore, the activity of 15(S)-LOX-2 in catalyzing the metabolism of exogenous arachidonic acid to 15(S)-HETE appears to play a significant role in arachidonic acid-stimulated cell adhesion of MDA-MB-435 cells to type IV collagen, a role not previously described for a 15(S)-LOX.
Recent studies demonstrated that inhibitors of LOX enzymes have impressive anti-tumorigenic properties, suggesting that some LOXs and their eicosanoid metabolites may promote tumor development and/or progression (28 -30). In rat carcinosarcoma cells and in prostate cell lines, 12(S)-HETE has been shown to regulate apoptosis and cancer progression, and inhibitors of 12(S)-HETE production caused apoptosis and growth arrest (31,32). These findings complement studies that demonstrate increased expression of certain LOXs in cancerous tissue. For example, leukocyte-type 12(S)-LOX was shown to be overexpressed in malignant breast tissue and breast cancer cell lines compared with normal breast tissue and cells (28,33), and expression of 15(S)-LOX-1 is increased in human colorectal tumors compared with adjacent normal tissue (34). In contrast, the expression of other LOXs is decreased or absent in certain tumors compared with surrounding non-cancerous tissues, suggesting that, in some settings, these enzymes also suppress tumorigenesis (30, 34 -37). In prostate cancer, different 15(S)-LOXs are thought to exert differing effects on tumor growth. 15(S)-LOX-1 expression is increased in prostate tumors over levels found in normal cells. This increase correlates with the degree of tumor progression according to Gleason grading (38). Conversely, there is low expression of 15(S)-LOX-2 and decreased 15(S)-HETE production in prostate carcinoma relative to normal prostate tissue (39). In addition, some LOX metabolites have directly opposing effects on signaling in prostate carcinoma cells. Specifically, the 15(S)-LOX-1 metabolite of linoleic acid, 13(S)-hydroxyoctadecadienoic acid, was shown by Hsi et al. (40) to down-regulate peroxisome proliferator-activated receptor ␥, a putative tumor suppressor, in prostate carcinoma cells; however, the 15(S)-LOX-2 metabolite 15-(S)-HETE up-regulated peroxisome proliferator-activated receptor ␥. This finding demonstrated that increased levels of 15(S)-LOX-1, and not just decreased expression of 15(S)-LOX-2, can shift the balance of specific LOX metabolites to favor those that promote carcinogenic changes in cells. Collectively, these studies suggest that a balance between the production of different eicosanoid metabolites is required to maintain normal cellular growth characteristics and that the alteration of this balance can have implications for neoplastic development (41). For example, 12(S)-HETE was shown to mediate the adhesion of B16 melanoma cells exposed to extracellular matrix following retraction of vascular endothelial cells by a mechanism that requires protein kinase C (42,43). Subsequently, 12(S)-HETE was shown to stimulate ␣ IIb ␤ 3 integrin activation and tumor cell spreading on fibronectin (44) as well as ␣ v ␤ 3 integrin activation and lung endothelial cell adhesion to vitronectin (45). 12(S)-LOX is also associated with serine phosphorylation events generated upon activation of ␣ IIb ␤ 3 integrins in B16 melanoma cells (46).
Previously, our laboratory demonstrated that arachidonic acid-stimulated cell adhesion requires the activation of p38 MAPK and that p38 MAPK activates its downstream substrates, MAPK-activated protein kinase 2 and heat-shock pro- tein 27 (23). Given the evidence that 15-LOX-2 activity mediates the stimulation of cell adhesion to type IV collagen, we hypothesized that 15-LOX-2 activity leads to the activation of p38 MAPK following exposure to arachidonic acid. To test this hypothesis, we measured the arachidonic acid-induced activation of p38 MAPK in the presence of the LOX inhibitor NDGA in MDA-MB-435 cells. We found that LOX inhibition caused a pronounced inhibition of p38 MAPK activation that was evident following exposure to exogenous arachidonic acid, supporting the hypothesis that 15(S)-LOX-2 metabolism of arachidonic acid leads to the activation of p38 MAPK in the process of stimulation of cell adhesion to type IV collagen.
To date, the mechanism by which fatty acids and their metabolites initiate the activation of kinase signaling pathways is unknown, and their ability to associate with numerous cellular components makes demonstrating critical interactions difficult. The MAPK kinases MKK3, MKK4, and MKK6 have been shown to lie immediately upstream of p38 MAPK and to activate p38 in response to a variety of stimuli (26,27). MKK3 and MKK4 were not activated following exposure to fatty acids, whereas phosphorylation of MKK6 occurred within minutes of the addition of either arachidonic acid or 15(S)-HETE. These results implicate MKK6 in the activation of p38 MAPK in this system. TAK1 and ASK1 are two MKKKs that have been demonstrated to activate the p38 MAPK pathway (47,48). We found that 15(S)-HETE induced a dose-dependent activation of TAK1 in MDA-MB-435 cells that was similar to TAK1 activation by arachidonic acid or transforming growth factor-␤ 1 . In addition, we found that ASK1 phosphorylation did not change after exposure to arachidonic acid or 15(S)-HETE. These findings suggest that arachidonic acid-mediated TAK1 activation is responsible for MKK6 activation and the ensuing downstream signaling. p38 MAPK and TAK1 have been shown to be autophosphorylated through a mechanism that requires binding to TAK1-binding protein-1 (49,50). These results help us to refine our published model (24) for how p38 is activated, leading to both direct and indirect effects on cytoskeletal components that coordinate adhesion, migration, and invasion of carcinoma cells. We do not yet know how 15(S)-HETE initiates activation of TAK1 and the p38 MAPK pathway or whether TAK1-binding protein-1 is involved in the activation of TAK1 in arachidonic acid-exposed MDA-MB-435 cells. We are currently looking for signaling molecules that may be the initial interacting partner for 15(S)-HETE. One possible target is the transforming growth factor-␤ receptor, which is expressed on the surface of MDA-MB-435 cells.
These findings demonstrate for the first time the activation of an MKKK or a growth factor-associated kinase, such as TAK1, in response to exposure to arachidonic acid or an eicosanoid metabolite and imply that the step at which lipid signals can initiate MAPK signal transduction is at least as high as the MKKKs.