Arachidonic acid activates mitogen-activated protein (MAP) kinase-activated protein kinase 2 and mediates adhesion of a human breast carcinoma cell line to collagen type IV through a p38 MAP kinase-dependent pathway.

Adhesion of metastatic human mammary carcinoma MDA-MB-435 cells to the basement membrane protein collagen type IV can be activated by treatment with arachidonic acid. We initially observed that this arachidonic acid-mediated adhesion was inhibited by the tyrosine kinase inhibitor genistein. Therefore, we examined the role of the mitogen-activated protein (MAP) kinase family tyrosine phosphorylation-regulated pathways in arachidonic acid-stimulated cell adhesion. Arachidonic acid stimulated the phosphorylation of p38, the activation of MAP kinase-activated protein kinase 2 (MAPKAPK2, a downstream substrate of p38), and the phosphorylation of heat shock protein 27 (a downstream substrate of MAP kinase-activated protein kinase 2). Treatment with the p38 inhibitor PD169316 completely and specifically inhibited arachidonic acid-mediated cell adhesion to collagen type IV. p38 activity was specifically associated with arachidonic acid-stimulated adhesion; this was demonstrated by the observation that 12-O-tetradecanoylphorbol 13-acetate-activated cell adhesion was not blocked by inhibiting p38 activity. Extracellular signal-regulated protein kinases (ERKs) 1 and 2 were also activated by arachidonic acid; however, cell adhesion to collagen type IV was not highly sensitive to PD98059, an inhibitor of MAP kinase kinase/ERK kinase 1 (MEK1) that blocks activation of the ERKs. c-Jun NH(2)-terminal kinase was not activated by arachidonic acid treatment of these cells. Together, these data suggest a novel role for p38 MAP kinase in regulating adhesion of breast cancer cells to collagen type IV.

and that in doing so alter the metastatic potential of these cells.
There are several families of adhesion molecules, the function of which can affect metastasis, including cadherins, selectins, the CD44 group, the immunoglobulin superfamily, and integrins (16,17). Integrins are heterodimeric, transmembrane cell surface glycoproteins that mediate adhesion to the extracellular matrix in a highly regulated manner and are linked to signal transduction pathways within the cell (18 -20). Because of the profound effect that integrins and their interactions with the extracellular matrix have on the biology of the cell, the proteins and second messengers that regulate these signal transduction pathways play critical roles in the behavior of both normal and transformed cells (6,8,(21)(22)(23).
The MAP 1 kinase signaling pathways are ubiquitous cascades that regulate a number of cellular responses. The MAP kinase family consists of three different subfamilies: ERK1 and ERK2, the JNK kinases, and the p38 MAP kinases. The MAP kinase family proteins are activated by dual phosphorylation of tyrosine and threonine residues within recognized motifs. These motifs are different for each of the different subfamilies (24,25). Proteins in the p38 subfamily contain a threonineglycine-tyrosine motif (26) that is phosphorylated in response to several types of stimuli, including different forms of cellular stress (26 -28), lipopolysaccaride (28 -30), several cytokines (26,(31)(32)(33), and ceramide (34 -36).
Arachidonic acid is a polyunsaturated fatty acid that is present in an esterified form in cell membranes and is released from the membrane by phospholipases, mainly phospholipase A2 (37)(38)(39)(40)(41). Direct addition of arachidonic acid or some of its metabolites can modulate cell-matrix interactions. Studies by Grossi et al. (42) and Timar et al. (43) have shown that 12(S)hydroxyeicosatetraenoic acid (12(S)-HETE), a lipoxygenase metabolite of arachidonic acid, increases the ability of tumor cells to adhere to fibronectin and subendothelial matrix (42) and to spread on fibronectin (43). Other researchers have demonstrated a role for arachidonic acid in HeLa cell spreading on collagen type IV (44). We, and others, have demonstrated that arachidonic acid (45) and its metabolic precursor, linoleic acid (46), activate the adhesion of metastatic human breast carcinoma cells to type IV collagen. Our previous studies using the PKC inhibitor calphostin C demonstrated PKC signaling involvement in arachidonic acid-mediated cell adhesion to colla-gen type IV (45). However, the mechanism by which arachidonic acid stimulates cell adhesion to collagen type IV has not been completely described.
Thus, the goal of this study was to define additional signaling pathways that are involved in arachidonic acid-stimulated cell adhesion of MDA-MB-435 human breast carcinoma cells to collagen type IV. Specifically, we were interested in the role that MAP kinase proteins may play in arachidonic acid-stimulated adhesion. We demonstrated that p38, its downstream substrate MAPKAPK2, and HSP27 (a downstream substrate of MAPKAPK2), were all activated by arachidonic acid and that activation of p38 was necessary for arachidonic acid-mediated cell adhesion. ERK1 and ERK2 were also activated by arachidonic acid, but these proteins do not appear to play a role in mediating adhesion. Finally, JNK was not activated by arachidonic acid treatment of these cells; thus, p38 is the only MAP kinase of which the activation appears to be necessary for arachidonic acid-mediated cell adhesion.

EXPERIMENTAL PROCEDURES
Cell Culture and Reagents-The human breast carcinoma cell line MDA-MB-435 was obtained from Dr. Janet Price (M. D. Anderson Cancer Center, Houston, TX) and was cultured as described (47). Cells were routinely tested for mycoplasma. Arachidonic acid (Cayman Chemical Co.) was purchased as a 328 mM stock in ethanol. The p38 inhibitor (PD169316) and a nonfunctional analogue of PD169316 (SB202474) were from Calbiochem. TPA and CHAPS (electrophoresis grade) were from Sigma. The MEK1 inhibitor (PD98059) was from New England Biolabs. Protease inhibitors were from Sigma or Roche Molecular Biochemicals. Collagen type IV and poly-D-lysine were from Becton Dickinson. BSA (fraction V) was from Life Technologies, Inc. (for adhesion studies) or ICN (for immunoblots). Primary antibodies were from Upstate Biotechnology (R2 and anti-p38), New England Biolabs (9101S, 9211S, and 9251S), Santa Cruz Biotechnology (sc-474), and Promega (anti-ACTIVE TM p38). MAPKAPK2 polyclonal antibody and recombinant HSP27 protein were from Stressgen (Victoria, British Columbia, Canada). Control 293 cell lysates for the JNK blots were from New England Biolabs. Ampholines were from Amersham Pharmacia Biotech. [␥-32 P]ATP was from Amersham Pharmacia Biotech. 32 P-Labeled orthophosphate was from ICN.
Cell Harvesting and Treatment-Subconfluent cells were harvested with 5 mM EDTA, washed twice with serum-free medium preequilibrated at 37°C in 7.5% CO 2 , resuspended at 5 ϫ 10 5 cells/ml, and allowed to recover at 37°C in 7.5% CO 2 for 20 min before being treated with arachidonic acid. Inhibitors and their analogues were dissolved in Me 2 SO and added to the cells 30 min (PD169316 and SB202474) or 2 h (PD98059) prior to treatment with arachidonic acid. Arachidonic acid was prepared immediately prior to its addition to the cells by adding an equal volume of 328 mM KOH to the arachidonic acid stock and diluting the mixture to 6 mM arachidonic acid in 0.9% NaCl. An appropriate volume was added to the cells to yield a final concentration of 30 M arachidonic acid. The maximum final concentration of ethanol was 0.009%.
Kinase Assays-Cells were treated with arachidonic acid and inhibitors as described above, and MAPKAPK2 assays were performed as described previously (50) with the following modifications. Protein Aagarose was used in place of protein G-Sepharose. The protein lysate used for each assay (0.5 mg of total protein per sample) was precleared for 1 h with protein A-agarose prior to immunoprecipitation. Leupeptin was added to the lysis buffer. E-64 and Brij were omitted from the procedure, and the proteins were resolved by electrophoresis on a 14% polyacrylamide gel. The specific activity of the [␥-32 P]ATP used was 6000 Ci/mmol. Two-dimensional Gels-Cells were harvested as described above and were labeled with 32 P-phosphate (0.5 mCi/ml of cells) during a 30-min recovery period prior to arachidonic acid treatment and for an additional 30 min of arachidonic acid treatment. Cells were collected as described above and lysed in 9.5 M urea, 2% CHAPS, 2% ampholines (75% pH 5-7, 25% pH 3.5-10), 0.7 M ␤-mercaptoethanol, 10 mM NaFl, 1 mM PMSF, and 1 mM EDTA (51). Lysates were passed through a 23 gauge needle 10 times, incubated for 40 min at 4°C with 10 l of a 50% slurry of protein G-agarose, and centrifuged at 16,000 ϫ g for 40 min at 4°. Supernatants were then analyzed by two-dimensional polyacrylamide gel electrophoresis as described previously (52) except that CHAPS was used in place of Nonidet P-40. The first dimension was run using 75% pH 5-7 and 25% pH 3.5-10 ampholines. Proteins on the second dimension 12% Laemmli gel were transferred to PVDF membrane prior to exposure to audoradiographic film. In order to determine the pH gradient of first-dimension tube gels, two extra gels were run that were cut into 1-cm sections. The ampholytes in these sections were extracted in 2 ml of degassed H 2 O for at least 2 h, and the pH of the solutions measured with a standard pH electrode.
Adhesion Assays-Cell attachment was assayed using a modification of the procedure described by Mould et al. (53). The wells of 96-well culture dishes (Costar) were coated with 100 l/well of either 6.4 g/ml collagen type IV, 32 g/ml poly-D-lysine, or 20 mg/ml BSA for either 2 h at room temperature or overnight at 4°C. Wells were then washed with calcium-and magnesium-free phosphate-buffered saline and blocked at room temperature for 2 h with 20 mg/ml heat-denatured BSA (Life Technologies, Inc., fraction V) (54). Cells were prepared and treated with TPA, arachidonic acid, or inhibitors as described above. Immediately after addition of arachidonic acid or TPA, 5 ϫ 10 4 cells in 100 l were added to each well. After a 45-min incubation at 37°C, nonadherent cells were washed from the wells using serum-free medium and aspiration. Poly-D-lysine coated wells were not washed. The remaining cells were fixed with gluteraldehyde and stained with 0.5% crystal violet in 20% methanol. The stain was then resolubilized with 1% SDS, and the absorbance was measured at 595 nm using a SpectraMax 250 microplate reader (Molecular Devices). Values are presented as the percentage of total cells assuming that the cells fixed to poly-D-lysine without washing represent 100%. In these adhesion assays, we typically see increases in adhesion in response to arachidonic acid of 1.5-3-fold; however, increases as high as 10-fold have been observed.
Statistics-Pairwise comparisons were made by Fisher's least significant difference test carried out at the p Ͻ 0.01 level of significance (55). Standard error for adhesion assays was calculated as the square root of the variance of the ratio of two means (56).

p38, MAPKAPK2, and HSP27 Are Targets of Arachidonic
Acid-In order to determine whether a tyrosine kinase-regulated signaling pathway was involved in arachidonic acid-activated adhesion of human breast carcinoma cells to collagen type IV, adhesion assays were performed in the presence of the phosphotyrosine inhibitor genistein or its nonfunctional analogue, genistin. Genistein was an effective inhibitor of arachidonic acid-activated cell adhesion, with half-maximal and maximal inhibition at concentrations of approximately 15 and 30 M genistein, respectively. In contrast, 30 M of the nonfunc-tional analogue, genistin, had no apparent inhibitory effect (data not shown).
Inasmuch as tyrosine phosphorylation is involved in the regulation of all three MAP kinase pathways, we examined whole cell extracts from arachidonic acid-treated human breast carcinoma cells for specific MAP kinase activity. Treatment of MDA-MB-435 cells with 30 M arachidonic acid-activated p38 MAP kinase as detected by an antibody that specifically recognizes the active, dually phosphorylated form of the protein (Fig.  1A, lanes 1 and 2). The increase in activated, phosphorylated p38 was not a result of an increase in total p38 protein (Fig. 1B,  lanes 1 and 2). That the protein being recognized is the active, dually phosphorylated form of p38 was further established by the observation that when cells were stimulated with arachidonic acid in the presence of the p38 inhibitor PD169316, active p38 was no longer detected (Fig. 1A, lane 2 versus lane 8). The total amount of p38 protein in these cells was unchanged as a result of treatment with PD169316 (Fig. 1B, lane 2 versus lane  8).
Kinase assays to measure MAPKAPK2 activity were performed to verify p38 activation and to determine whether downstream substrates of p38 were activated by arachidonic acid (50). MAPKAPK2 is a downstream substrate of p38 and phosphorylates HSP27 (27,57); thus, MAPKAPK2 activation, as detected by in vitro phosphorylation of HSP27, is a frequent outcome of p38 activation. As demonstrated in Fig. 2, arachidonic acid treatment induced a significant increase in MAP-KAPK2 activity (Fig. 2, compare lanes 1 and 4), which was inhibited by the p38 inhibitor PD169316 (Fig. 2, lane 5) but not by the nonfunctional analogue of this inhibitor, SB202474 (Fig.  2, lane 7). The specificity of this activity was demonstrated by the fact that when HSP27 was omitted from the reaction (Fig.  2, lane 2) or when the immunoprecipitation was performed with normal rabbit serum (Fig. 2, lane 3), phosphorylation of HSP27 was not detected. The ERK1 and ERK2 activator, MEK1, has also been observed to activate MAPKAPK2 (58). However, as indicated by the inability of the MEK1 inhibitor PD98059 to inhibit arachidonic acid activation of MAPKAPK2 (Fig. 2, lane 9), arachidonic acid activation of MAPKAPK2 occurred through p38 and not through a pathway involving MEK1. The observation that arachidonic acid activation of ERK1 and ERK2 does not occur until 15 min after arachidonic acid treatment of these cells (data not shown) also supports the conclusion that MAPKAPK2 activation, which occurs within 5 min of arachidonic acid treatment, does not occur through MEK1. These results demonstrate that the both p38 and its downstream substrate MAPKAPK2 are activated by arachidonic acid.
We next investigated whether arachidonic acid treatment induces the in vivo phosphorylation of HSP27. Human HSP27 can be phosphorylated at three different sites: serine 15, serine 78, and serine 82 (51,59,60). This phosphorylation results in the appearance of several different isoforms (61,62). Treatment with arachidonic acid led to an increase in the amount of the dually phosphorylated c form of HSP27 (Fig. 3), demonstrating that HSP27 is activated in these cells in response to arachidonic acid. Immunoblotting with a monoclonal antibody to HSP27 verified the identity of the indicated spots as HSP27 (data not shown).
Time and Dose Response of p38 Activation to Arachidonic Acid Treatment-Arachidonic acid activated p38 at concentrations as low as 5 M (Fig. 4, lane 3). Our previously published results demonstrated that the increase in arachidonic acidmediated adhesion becomes apparent with 10 M arachidonic acid treatment (45).
We performed kinetic studies of p38 activation to determine whether the time of arachidonic acid activation of p38 is consistent with its playing a role in arachidonic acid-stimulated cell adhesion. Arachidonic acid activation of p38 occurred rapidly, within 5 min of treatment (Fig. 5, lane 2), and was sustained for at least 60 min (Fig. 5, lane 5). These kinetics are consistent with the kinetics of activation of adhesion by arachidonic acid (data not shown).
Activation of p38 Is Required for Arachidonic Acid-stimulated Cell Adhesion-To determine whether activation of p38 is necessary for arachidonic acid-activated cell adhesion to collagen type IV, adhesion assays were performed using cells treated with the p38 inhibitor PD169316. As shown in Fig. 6A, 5 M PD169316 was sufficient to completely block the activation of p38, as judged by immunoblotting with antibody to the active, phosphorylated form of p38 (Fig. 6A, lane 5). An equal molar concentration of the nonfunctional analogue SB202474 had no effect on the detection of active p38 (Fig. 6A, lane 6). Both 5 and 10 M p38 inhibitor were tested for their ability to inhibit arachidonic acid-activated cell adhesion to collagen type IV ( Fig. 6B and data not shown). As shown in Fig. 6B, 10 M PD169316 inhibited the arachidonic acid-stimulated adhesion of MDA-MB-435 cells. In contrast, the same concentrations of the nonfunctional analogue had no effect on adhesion. Arachidonic acid-stimulated adhesion was also inhibited completely by 5 M PD169316 (data not shown). As judged by propidium iodine staining and flow cytometry, neither PD169316 nor SB202474 had a significant effect on cell viability under the conditions used in the cell adhesion assay (data not shown). A second p38 inhibitor, SB203580, with a higher IC 50, was also able to completely inhibit arachidonic acid-mediated adhesion when used at 20 M under the same conditions (data not shown).  lanes 1-4), the p38 inhibitor PD169316 (lanes 5 and 6), its nonfunctional analogue SB202474 ( lanes  7 and 8), or the MEK1 inhibitor PD98059 (lanes 9 and 10) and subsequently treated with 30 M arachidonic acid (lanes 1-3, 5, 7, and 9 (AA)) or its solvent (lanes 4, 6, 8, and 10 (C)) for 5 min. Cells were then lysed and assayed for MAPKAPK2 activity. In lane 2, the MAPKAPK2 substrate, HSP27, was omitted from the assay to demonstrate specificity. In lane 3, immunoprecipitation was performed using normal rabbit antisera. Similar results with arachidonic acid have been obtained from three separate experiments. Similar results using the inhibitors have been obtained from two separate experiments.
The specificity of the role of p38 in arachidonic acid-activated cell adhesion was further demonstrated by examining the role of p38 activation in TPA-induced cell adhesion to collagen type IV. As shown Fig. 6B, 10 M PD169316 had no significant effect on TPA-activated cell adhesion. In this same experiment, 10 M PD169316 completely inhibited arachidonic acid-mediated adhesion.
Arachidonic acid-mediated cell adhesion to collagen type IV was not highly sensitive to the MEK1 inhibitor PD98059. When cells were preincubated with 10 M PD98059 for 2 h, adhesion was not significantly inhibited (Fig. 7A), and activated ERK1 and ERK2 were not readily detected (Fig. 7B, lane 1), suggesting that high levels of activated ERK1 and ERK2 are not necessary for arachidonic acid-activated cell adhesion.
We also examined arachidonic acid-treated cells for activation of JNK. There are three separate JNK genes; expression from any of the three genes can yield 45-or 54-kDa isoforms (63). Immunoblotting with an antibody that recognizes the phosphorylated 46-and 54-kDa isoforms of all three JNK gene products did not detect phosphorylated JNK in MDA-MB-435 cells after arachidonic acid treatment for various times up to an hour (Fig. 8A). As shown in Fig. 8B, these cells do express at least some isoforms of JNK1 and JNK3. These results demonstrate that activation of a JNK MAP kinase-mediated pathway is not necessary for arachidonic acid-induced cell adhesion. DISCUSSION These studies established that arachidonic acid activates the MAP kinase family protein p38, its downstream substrate MAPKAPK2, and the phosphorylation of the MAPKAPK2 substrate HSP27 in MDA-MB-435 human breast carcinoma cells. In addition, we showed that p38 activation is necessary for arachidonic acid-mediated adhesion of these cells to collagen type IV. Thus, we have defined an arachidonic acid-stimulated signal transduction pathway that includes at least p38, MAP-KAPK2, and HSP27.
There are several possible mechanisms by which p38 may mediate arachidonic acid-stimulated adhesion to collagen type IV. First, p38 could alter the number of integrin receptors for collagen type IV present on the cell surface. This change could result from increased de novo expression of these integrins or from transport of such receptors from preexisting intracellular pools to the cell surface. The first mentioned mechanism is possible because multiple transcription factors, including cAMP-response element-binding protein (CREB) (64,65), activating transcription factor 1 (ATF-1) (64, 65), activating transcription factor 2 (ATF-2) (66, 67), C/EBP homologous protein (CHOP) (36,68), and myocyte-enhancer factor 2C (MEF2C) (69), can be activated by p38. In addition, p38 activates MAP kinase-interacting kinases 1 and 2 (MNK1 and -2) (70 -72). MAP kinase-interacting kinase 1 is a kinase that can phosphorylate eukaryotic initiation factor 4E (eIF4E), a translation factor the phosphorylation of which is necessary for the initiation of mRNA translation (72)(73)(74). Thus, p38 could increase the expression of cell surface molecules through transcriptional or translational mechanisms. However, we have observed that the levels of collagen type IV receptor ␤ 1 , ␣ 1 , and ␣ 2 integrin subunits are not altered when examined after a 45-min incubation with arachidonic acid. 2 Thus, it is more likely that p38 activates receptors already present at the cell surface. Activation of type IV collagen receptors could result from direct interaction of the adhesion receptors with signal transduction proteins or other cell surface adhesion molecules, from posttranslational modification of receptors, or from alterations in the cytoskeleton or cell membrane that could directly or indirectly affect adhesion receptor conformation. This activation could be achieved through downstream targets of p38, such as MAPKAPK2. Interestingly, activated MAPKAPK2 phosphorylates and activates heat shock protein 27 (57). The activation of HSP27 has been shown to stabilize and increase actin filament formation, a process that could result in cytoskeletal changes (75,76). We have verified that arachidonic acid stimulation of MAPKAPK2 leads to the in vivo phosphorylation of HSP27. Therefore, we hypothesize that p38 may mediate cell adhesion through a pathway involving MAPKAPK2, HSP27, and subsequent alteration of the cytoskeleton.
The upstream mediators of arachidonic acid-induced p38 activation have not yet been identified. Previous studies have shown that activation of p38 can occur through a protein kinase cascade involving either MAP kinase kinase 3, 4, or 6 (66,67,(77)(78)(79)(80). All of these proteins are potential mediators of p38 activation by arachidonic acid.
The MAP kinase proteins ERK1 and ERK2 were also activated by arachidonic acid treatment of these cells; this observation is consistent with results in other cell types (81)(82)(83). However, arachidonic acid-mediated cell adhesion does not appear to be highly sensitive to the MEK1 inhibitor PD98059, suggesting that high levels of activated ERK1 and ERK2 have little effect on adhesion.
Arachidonic acid does not appear to activate JNK in this cell type, demonstrating that the activation of JNK is not necessary for arachidonic acid-mediated adhesion. Other studies have shown that arachidonic acid activates JNK in rabbit epithelial (84) and human leukemia (83) cells; therefore, it is likely that there is cell type specificity with respect to arachidonic acid activation of JNK pathways.
Recent work by others has shown that arachidonic acid activates p38 in HL60 cells, HeLa cells, and neutrophils (83) and that p38 plays a role in the adhesion of neutrophils to fibrinogen (85). However, to our knowledge, this is the first study demonstrating the involvement of p38 in arachidonic acidmediated cell adhesion to collagen type IV and is the first to demonstrate that arachidonic acid activates MAPKAPK2 and causes the increased phosphorylation of HSP27. These observations are especially interesting because overexpression of HSP27 can alter the invasiveness of breast cancer cells in both tissue culture and animal models (86,87). Thus, it is possible that p38 activation of MAPKAPK2 and the ensuing posttranslational modification of HSP27 could play a role in regulating metastasis of these cells through the modulation of cell adhesion.
It has also been shown that mice fed a diet high in the arachidonic acid precursor, linoleic acid, exhibit increased susceptibility to MDA-MB-435 (88) or 4526 (89) tumor metastasis in vivo. Collagen type IV is a major component of most basement membranes and a critical barrier through which metastasizing cells must pass (16,90). Increased adhesion of cells to collagen type IV may facilitate their ability to invade through the basement membrane. Thus, our study suggests a possible mechanism by which increased linoleic acid in the diet of these mice could increase metastasis and suggests that p38 may be a useful target for antimetastatic drugs.