Tumor cell-derived 12(S)-hydroxyeicosatetraenoic acid induces microvascular endothelial cell retraction

[CANCER RESEARC[154. 565-574, January 15, 1994]

Tumor Cell-derived 12(S)-Hydroxyeicosatetraenoic Acid Induces Microvascular Endothelial Cell Retraction Kenneth V. Honn, 2 Dean G. Tang, Irma Grossi, 3 Zofia M. Duniec, Jozsef Timar, Colette Renaud, Marie Leithauser, Ian Blair, Carl R. Johnson, Clement A. Diglio, Victoria A. Kimler, John D. Taylor, and Lawrence J. Marnett Departments of Radiation Oncology [K. V. 1t., D. G. T., L G., Z. M. D., J. T., C. R.], Chemistry [K. V. H., C. R. J.], Pathology [C. A. D., K. V. H.], and Biological Science [V. A. K., J. D. T.], Wayne State University, Detroit, Michigan 48202; Departments of Biochemistry and Chemistry [M. L., L. J. M.], and Pharmacology [I. B.], Center for Molecular Toxicology, Vanderbilt University School of Medicine, Nashville, Tennessee 37232; and First Institute of Pathology and Experimental Cancer Research, Semmelweis Medical University, Budapest, Hungary [J. T.]

ABSTRACT

analyzed in the context of the "docking and locking" model (2). Following "tight" tumor cell adhesion (i.e., locking) tumor cells spread on the subendothelial matrix. The final outcome of tumor cell-endothelial cell interactions is the exodus of tumor cells from the circulation into the target organ parenchyma, a process frequently termed "extravasation." Thereafter, tumor cells interact with various components of the extracellular matrix and invade the interstitium by elaborating degradative proteases via the signals transduced through cell surface adhesion receptors (4, 5). The physical and functional integrity of endothelial cell monolayers is of paramount importance in maintaining hemostasis and retarding the dissemination of metastasizing tumor cells. A large number of experiments have documented that vascular integrity is compromised or disrupted under a variety of conditions by various bioactive agents. During hematogenous metastasis, some types of tumor cells may extravasate by intravascular proliferation followed by rupture of the vessel wall. However, the majority of experimental (morphological as well as functional) studies both in vivo (6-8) and in vitro (9-11) have demonstrated that vascular endothelial cells retract prior to the extravasation of most tumor cell types. Therefore, tumor cell-induced endothelial cell retraction has been proposed to facilitate tumor cell metastasis. Unfortunately, we are still far from understanding the molecular mechanisms underlying tumor cell-induced endothelial cell retraction. In addition, the factors responsible for endothelial cell retraction, which could derive from interacting host cells as well as tumor cells, remain elusive. One plausible candidate is arachidonic acid metabolites, which have been shown to significantly modulate tumor cell interactions with host platelets and endothelial cells (12, 13). Our previous experiments have identified a lipoxygenase metabolite of arachidonic acid, 12(S)-HETE, 4 which induces a dose- and time-dependent retraction of both large vessel and microvessel endothelial cells (13-16). Other experiments have established that solid tumor cells are capable of metabolizing arachidonic acid to 12(S)HETE (17, 18), therefore raising the possibility that tumor cells elaborate 12(S)-HETE to induce endothelial cell retraction. Here we present experimental evidence demonstrating that production of 12(S)-HETE by tumor cells during tumor cell-endothelial cell adhesion is indeed correlated with tumor cell-induced endothelial cell retraction, suggesting that this eicosanoid may play an essential role in facilitating cancer metastasis.

Our previous work demonstrated that the 12-1ipoxygenase metabolite of arachidonic acid, 12(S)-hydroxyeicosatetraenoic acid [12(S)-HETE] induced a nondestructive and reversible retraction of cultured endothelial cells. In the current study we tested the hypothesis that tumor cells produce 12(S)-HETE during their interactions with endothelial cells which in turn induces endothelial cell retraction. Coincubation of Lewis lung carcinoma cells or elutriated B16 amelanotic melanoma (B16a) cells but not 3T3 fibroblasts with microvascular endothelial cells (CD3) resulted in a time- and concentration-dependent retraction of the CD3 monolayers as revealed by quantitative binding assays and phase contrast microscopy. Lewis lung carcinoma cell-induced endothelial cell retraction was blocked by specific lipoxygenase inhibitors but not by cyclooxygenase inhibitors, suggesting the involvement of a lipoxygenase metabolite(s). Radioimmunoassay and high-performance liquid chromatography analysis of tumor cell extracts identified 12(S)-HETE as the major lipoxygenase metabolite of arachidonic acid and tumor cell generation of 12(S)-HETE was specifically blocked by a select 12-1ipoxygenase inhibitor N-benzyI-N-hydroxy5-phenyl-pentamide. The identity and stereochemistry of tumor cell-derived 12-HETE was substantiated by gas chromatography-mass spectrometry analysis and chiral phase high-performance liquid chromatography, respectively. Lewis lung carcinoma cell adhesion to CD3 monolayers was accompanied by an enhanced 12(S)-HETE biosynthesis by tumor cells, which paralleled the tumor cell-induced endothelial cell retraction in a cell number-dependent manner. Pretreatment of tumor cells with N-benzyI-N-hydroxy-5-phenylpentamide inhibited both increased 12(S)-HETE biosynthesis and tumor cell-induced endothelial cell retraction. Highly metastatic variants of elutriated B16a cells which had been shown to produce large quantities of 12(S)-HETE induced significant CD3 cell retraction, while low metastatic subpopulations of B16a cells which synthesized no or little 12(S)-HETE did not induce endothelial cell retraction. These results suggest that 12(S)-HETE synthesis during tumor cell-endothelial cell interactions may represent a key contributory factor in cancer metastasis.

INTRODUCTION Tumor cell interactions with the microvasculature constitute a rate regulator for hematogenous metastasis (1). Tumor cell adhesion to the target organ endothelial cells represents the first important event in tumor cell-vessel wall interactions and plays a decisive role in determining the organ preference of metastasis (2, 3). The adhesive interactions between blood-borne tumor cells and microvessel endothelial cells involve complicated yet well-concerted actions of a large array of cell surface adhesion molecules, whose hierarchical roles can be

MATERIALS

Received 2/18/93; accepted 11/5/93. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact. 1This work was supported by NIH Grants CA47115, CA 29997, and a grant from Harper Hospital, WayneState University,Detroit,MI (to K. V. H.), NIH Grants CA 47479 and ES00267 (to L. J. M.), Grant GM 15431 (to I. B.), and NATOLinkageGrant Program LG 921311 (to K. V. H. and J. T.). 2 To whom requests for reprints should be addressed, at Department of Radiation Oncology,Wayne State University,430 Chemistry,Detroit, MI 48202. 3 Present address: Upstate Biotechnotogy Inc., 89 Saranac Ave., Lake Placid, NY 12946. 565

AND M E T H O D S

Cell Lines and Culture. Murine pulmonary vascular endothelial cells (CD3; clone 1) were isolated, cloned, and characterized (phenotypically, enzymatically, and biochemically) as previously described (19). CD3 cells were grown in DMEM supplemented with 10% FCS in a 5% CO2-containing atmosphere. Cells were routinely passaged with a mixture of trypsin (0.1%) 4 The abbreviations used are: 12(S)-HETE, 12(S)-hydroxyeicosatetraenoicacid; 3LL, Lewis lung carcinoma; DMEM, Dulbecco's minimal essential medium; GC-MS, gas chromatography-mass spectrometry; HPLC, high performance liquid chromatography; BHPP, N-benzyl-N-hydroxy-5-phenylpentamide,PKC, protein kinase C; FCS, fetal calf serum; El, electron impact spectra.

12(S)-HETE AND ENDOTHELIAL CELL RETRACTION

and EDTA (2 mM). Only low-passage (i.e., passage 1 to 5) cells were utilized throughout the experiments. Murine 3LL cells and B16a melanoma cells were obtained from the Division of Cancer Treatment (NIH, Frederick, MD) and maintained in syngeneic C57BL/6J mice. 3LL cells were subcultured in DMEM containing 10% FCS as described previously (20) and passaged with 2 mM EDTA. Only low-passage (i.e., passage 1-3) tumor cells were used. In some experiments, B16a and 3LL cells were isolated by centrifugal elutriation from enzymatically dispersed solid tumor tissues as described previously (21, 22). 3T3 fibroblasts were obtained from the American Type Culture Collection and cultured in DMEM supplemented with 10% FCS and various antibiotics. Eicosanoids, Inhibitors, and Other Reagents. All eicosanoids used in the present studies were obtained from Cayman Chemical (Ann Arbor, MI) and stored in absolute ethanol. The purity of these eicosanoids was examined before use and was always greater than 99%. Lipoxygenase inhibitors BW755c and nordihydroguaiaretic acid were obtained from Burroughs Wellcome (Research Triangle Park, NC) and Sigma Chemicals (St. Louis, MO), respectively. A selective lipoxygenase inhibitor, BHPP was synthesized with the use of published procedures (23). Cyclooxygenase inhibitors, indomethacin and acetylsalicylic acid were purchased from the Upjohn Co. (Kalamazoo, MI) and Sigma Chemicals, respectively. Solid phase radioimmunoassay kit for 12(S)HETE was obtained from Oxford Biomedical Research, Inc. (Oxford, MI). Type IV collagen was isolated from Englebreath-Holm-Swarm murine sarcomas (Chemicon, Segundo, CA). Monoclonal antibody to collagen type IV was obtained from Collaborative Research (Waltham, MA). Arachidonic acid and [14C]arachidonic acid were purchased from NuChek-Prep (Elysian, MN) and NEN (Wilmington, DE), respectively. Antibody Radiolabeling. An appropriate amount (0.2 mg/200/xl) of anticollagen IV monoclonal antibody protein was added to a phosphate-buffered saline solution (800 p,1) containing IODO-BEADS (Pierce, Bockford, IL) and 0.5 mCi of 125I (specific activity, 17.4 Ci/mmol; NEN, Boston, MA). The iodination reaction was allowed to proceed for 20 min. The IODO-BEADS were then removed from the antibody solution to terminate the reaction and the antibody solution was dialyzed against phosphate-buffered saline (pH 7.3) containing 0.9% NaC1 solution to remove free label before use in assays. Specific activity of labeling was generally 4 • 104 cpm/~g antibody protein. Quantification of Endothelial Cell Retraction. Our established protocol (14) was used to quantitate endothelial cell retraction. CD3 endothelial cells (2.5 • 105/well) were added to 96-well fiat-bottomed plates coated with murine collagen type IV (20/~g/ml) and cultured to confluence. For experiments, CD3 monolayers were washed and then treated (10 min; 20~ with various eicosanoids (0.1 /.LMfinal concentration unless otherwise indicated) in DMEM containing 4% bovine serum albumin. Ethanol was used as the vehicle control. After treatment, endothelial cell monolayers were rinsed and incubated at 37~ for the time intervals indicated in "Results." Following incubation, cells were fixed, rinsed, and incubated with 125I-anti-collagen IV antibody (2 x 105 cpm/well) for 60 min at 20~ Then the wells were extensively washed, the contents were removed by trypsin (0.1%) and collagenase, and the radioactivity was measured as described (14). In some experiments, cultured or elutriated 3LL cells or different subpopulations of B16a cells (21, 22) which had or had not been pretreated with lipoxygenase or cyclooxygenase inhibitors (all used at a final concentration of 10 /~M) were coincubated (2.5 • 104 cells/well unless otherwise specified) with CD3 endothelial cell monolayers (60 min; 37~ Nonadherent cells were removed by aspiration and the monolayer was fixed and processed as described above. Results were expressed as total cpm/weU. The experiments were run in quadruplicate and each was repeated a minimum of three times with comparable results. In some experiments phase contrast micrographs of the monolayers were taken with a Nikon Optophot microscope, using Kodak T-MAX 400 panchromatic film. In some other experiments, whole mount transmission electron microscopy was used to confirm 12(S)-HETE as well as tumor cell-induced CD3 cell retraction, as previously described (24, 25). Briefly, CD3 cells were seed onto UV-sterilized Formavar-carbon-coated gold grids, cultured to confluence, and then treated with either solvent or 12(S)-HETE (0.1 /XM)for various time intervals. After treatment, cells were prefixed in 1.25% glutaraldehyde-l% paraformaldehyde mixture in 0.1 u sodium cacodylate. Following fixation, cells were sequentially treated (24, 25), finally dehydrated in Freon TF-113, and critical point dried by Freon 13 in a Bomar (Tacoma, WA) critical point drying apparatus. Finally, samples were examined and photographed on a Phillips 201 transmission electron microscope operating at accelerating voltages of either 80 or 100 kV.

Radioimmunoassay of 12(S)-HETE Production by Tumor Cells. A homogeneous population of 3LL cells was obtained by centrifugal elutriation (21, 22) and used to quantify 12(S)-HETE production following tumor cell adhesion to endothelium. Various numbers (for dose study) of tumor cells in a total volume of 500 txl of serum-free DMEM were added to confluent endothelial cell monolayers and incubated for the required amount of time (for time course study). A separate set of endothelial cell monolayers were treated identically but in the absence of tumor cells and used as background control. Following incubation, the contents (i.e., tumor cells plus CD3 cells in the experimental groups, or CD3 cells in the control groups) were harvested and extracted with a mixture of chloroform:methanol (1:2). The combined organic phase was used to quantitate the amount of 12(S)-HETE by immunoassay as described previously (26). Subcellular Fractionation, HPLC, and GC/MS. Freshly isolated 3LL tumor cells were homogenized in the buffer containing 100 mM potassium phosphate, 250 mM mannitol, 10 mM EDTA, and 300 /XM diethyldithiocarbamate (pH 7.5), using a Polytron homogenizer (Brinkmann Instruments, Westbury, NY). The resulting homogenate was filtered through at least two thicknesses of cheesecloth and centrifuged at 12,000 • g for 20 min. The supernatant was filtered through cheesecloth and was centrifuged at 100,000 • g for 90 min. The 100,000 • g supernatant (2 mg/ml) was incubated with [14C]arachidonic acid (17-52 mCi/mmol) at 37~ for 15 to 30 min. The mixture was extracted and biosynthetic 12-HETE was purified from the extract by reverse-phase HPLC (17, 18). In other experiments, elutriated 3LL cells (5 • 106) were preincubated for 10 min at 37~ with 1 ~M BHPP or ethanol (solvent control) and then 1 or 10 tXM concentrations of arachidonic acid plus 2/xCi [3H]arachidonic acid were added for an additional 15 min, followed by adding glacial acetic acid to terminate the incubation. Samples were centrifuged and cell pellets were resuspended in phosphate-buffered saline with or without 1 mM butylated hydroxytoluene. Samples were sonicated (15 s twice) on ice and extracted for 30 min at room temperature with 3.75 volumes of chloroform:methanol (1:2) containing or not containing 1 mM butylated hydroxytoluene. Then 1.25 volumes of chloroform and distilled H20 each were added and the samples were centrifuged to separate the layers. The organic phases were transferred to clean test tubes and aqueous layers were reextracted with 2.5 volumes of chloroform. Pooled extracts were dried under a stream of nitrogen and residues were redissolved in a solvent suitable for HPLC analysis. To confirm that the 12(S)-HETE generated by 3LL cells is a product of lipoxygenase pathway, straight-phase HPLC was performed by using Partisil silica 5 p,m column (250 x 4.6 mm; Alltech Associates, Inc., Deerfield, IL). Hexane:isopropanol:acetic acid (993:6:1) at a flow rate of 2 ml/min was used isocratically to separate 12-hydroxy- from 12-hydroperoxyeicosatetraenoic acid. For the BHPP inhibition studies, we used the reverse-phase HPLC with a C18 Ultrasphere ODS column (Beckman Instrument, Inc., Berkeley, CA). Samples were eluted with acetonitrile:acetic acid (100:0.05) solution A, and acetonitrile:water:acetic acid (20:100:0.05) solution B, at a flow rate of 1.5 ml/min. The gradient was as follows: 0 min, 42% solution B, 37 min, 42% solution B, 40 min, 100% solution B, and 50 min, 100% solution B. To determine the stereochemistry of tumor cell-derived 12-HETE, 10 p,g of authentic, racemic 12-HETE were added to the purified, biosynthetic, 14Clabeled 12-HETE and this mixture was derivatized by ethereal diazomethane. The sample was repurified by straight-phase HPLC on an Econosil silica column eluted with hexane:isopropyl alcohol (100:1) at a flow rate of 0.5 ml/min. The peak corresponding to the retention time of 12-HETE methyl ester (24 min) was collected. A portion of this sample was applied to a chiral HPLC column (Chiralcel-OC, 5 p~m, 0.46 x 25 cm; J. T. Baker Chemical Co., Phillipsburg, N J), eluted with hexane/isopropyl alcohol at a flow rate of 1 ml/min, and monitored at 235 nm. Under these conditions, 12(R)-HETE and 12(S)-HETE elute with approximate retention time of 24 and 26 min, respectively. Characterization of authentic 12(S)-HETE and the HPLC peak corresponding to biosynthetic 12-HETE by GC/MS was carried out on an INCOS 50 B mass spectrometer interfaced to a Hewlett Packard 5890 gas chromatograph, using positive ion chemical ionization and El mass spectrometry of their methyl ester, trimethylsilyl ether derivatives. Positive ion chemical ionization was performed by using conventional fused silica gas chromatography. The mass spectra of the compounds were recorded at 100 eV. Statistics. All data were analyzed for normal distribution as described previously (14). Data were analyzed for normality by using a Macintosh Plus

566

12(S)-HETE AND ENDOTHELIAL CELL RETRACTION

computer (Apple Computer, Cupertino, CA) and STATVIEW 512+ software (BrainPower, Calabasas, CA). When statistically significant differences (P < 0.05) between groups were noted, the results were further analyzed by the Kruskal-Wallis test. Groups with P < 0.05 differences were considered statistically significant. RESULTS 12(S)-HETE Induces a Reversible Retraction of Microvascular Endothelial Cells. We previously demonstrated that 12(S)-HETE induces large vessel endothelial cell retraction (14). In the present study we demonstrate that this eicosanoid similarly induces microvessel endothelial cell retraction. The 12(S)-HETE effect was stereospecific in that the enantiomer, 12(R)-HETE was ineffective (Fig. 1A). Eicosanoids derived from several different lipoxygenases were tested and only 12(S)-HETE was observed to induce endothelial cell retraction (Fig. 1A). Phase-contrast microscopy showed that 12(S)-HETE treatment (Fig. 2b) disrupted the normal "cobblestone" endothelial cell monolayer, which remained closely apposed with tight cell-cell borders even following 60-min treatment by the vehicle (ethanol) alone (Fig. 2a). 12(S)-HETE-treated endothelial cells demonstrated a retracted (arborized) cell phenotype with numerous extended cell filapodia, concomitant with the formation of easily visible cell-cell gaps exposing subendothelial matrix (Fig. 2b, asterisks). 12(S)-HETE-induced CD3 cell retraction was also confirmed by whole mount technique (Fig. 2, e and f). The 12(S)-HETE effect on endothelial cell retraction was both time (not shown) and dose dependent, approaching a maximal effect 1 h poststimulation with 0.1 /xM 12(S)-HETE (Fig. 1B). Retracted endothelial cell monolayers recovered their cobblestone morphology approximately 24 h after 12(S)-HETE treatment (data not shown). Tumor Cells Induce Time- and Concentration-dependent Endothelial Cell Retraction. When elutriated 3LL cells were coincubated with the confluent CD3 monolayers, endothelial cell retraction was observed, most notably at the sites of tumor cell-endothelial cell contact (Fig. 2c). Tumor cells were frequently observed to be in contact with the exposed subendothelial matrix in the retracted area (Fig. 2c). Quantitation of the tumor cell-induced endothelial cell retraction revealed a definitive time (Fig. 1C)- and concentration (Fig. 1B)-dependent effect. At a fixed concentration (i.e., 2.5 • 10 4) of tumor cells, endothelial cell retraction appeared at about 15 rain and peaked 1 h following the addition of tumor cells, and then gradually declined (Fig. 1C). At the longest time interval examined, i.e., 24 h postaddition of tumor cells, endothelial cell retraction was still apparent (Fig. 1C). In order to determine whether tumor cell-induced endothelial cell retraction is a cell line-restricted phenomenon, we chose another tumor cell system (i.e., B16a), well characterized in our laboratory, and performed similar retraction studies. B16a melanoma cells can be isolated from solid tumors grown in vivo by enzymatic dispersion and centrifugal elutriation. Four distinct subpopulations (designated 100, 180, 260, and 340 fractions, respectively) (21) which differ (340 > 260 > 180 > 100) in their ability to induce homologous platelet aggregation, express alIb/33 integrin receptors, adhere to endothelial cells and subendothelial matrix, produce 12(S)HETE, and form lung colonies following tail vein injection have been well characterized (21, 22, 27-29). Combined fractions of the 260 and 340 subpopulations were used to perform the retraction studies. Both the time study (Fig. 3A) and the dose study (Fig. 3B) demonstrated that B16a cells, similar to 3LL cells, induced a reversible endothelial cell retraction which was inhibited by pretreatment of B16a cells with BHPP (Fig. 3B). In contrast, when 3T3 fibroblasts were substituted for tumor cells, no retraction was observed (Fig. 3B).

Tumor Cell-induced Endothelial Cell Retraction Is a Lipoxygenase-dependent Process. Considering the similar morphology of endothelial cell retraction induced by 12(S)-HETE and tumor cells, we examined the effects of various lipoxygenase or cyclooxygenase inhibitors on tumor cell-induced endothelial cell retraction. Pretreatment of tumor cells with general lipoxygenase inhibitors BW755c or nordihydroguaiaretic acid, or with a selective 12-1ipoxygenase inhibitor (i.e., BHPP) (23), 5 completely blocked endothelial cell retraction induced by 3LL cells (Fig. 1D). Although 3LL cells pretreated with BHPP adhered to the endothelial cell monolayers at similar efficiency as the control tumor cells, the former were unable to induce endothelial cell retraction during the 60-min incubation interval (Fig. 2d). In contrast to the effect of lipoxygenase inhibitors, cyclooxygenase inhibitors indomethacin and acetylsalicylic acid showed little or no effect on tumor cell-induced endothelial cell retraction (Fig. 1D). The effects observed with these inhibitors were specific, since treatment of CD3 cell monolayer with the inhibitors alone did not have any impact on the morphology and integrity of the monolayer, as evidenced by both quantitative measurement (Fig. 1D) and morphological examination (data not shown). 12(S)-HETE Is the Major Lipoxygenase Metabolite of Arachidonic Acid in 3LL Cells. The above experiments suggest that a lipoxygenase product mediates the endothelial cell retraction induced by Lewis lung carcinoma cells. The most likely candidate for this product is 12(S)-HETE since BHPP, a selective 12-1ipoxygenase inhibitor, blocked the effects of 3LL and B16a cells. To explore the possibility that 3LL cells metabolize arachidonic acid to 12(S)-HETE through the action of 12-1ipoxygenase, we incubated elutriated 3LL cells with 2 /xCi of [3H]arachidonic acid (specific activity 100 Ci/ mmol) plus either 1 /XM or 10 /ZM concentrations of cold arachidonic acid followed by extraction with chloroform-methanol mixture and HPLC analysis. Both the straight-phase (Fig. 4b) and reverse-phase HPLC (Fig. 5a) analysis demonstrated that 12-HETE was the most predominant lipoxygenase metabolite in 3LL cells. In addition, 12(S)-hydroperoxyeicosatetraenoic acid, the metabolic precursor of 12HETE, was also detected in the extracts of 3LL cells (Fig. 4b), suggesting that 3LL cells possess a 12-1ipoxygenase metabolic pathway. Pretreatment of 3LL cells with BHPP (1 /XM) significantly reduced the amount of 12-HETE production (Fig. 5b), confirming that 3LL cells metabolize arachidonic acid to 12(S)-HETE. To determine the structural identity and stereochemical configuration of the 12-HETE produced by 3LL cells, we performed chiralphase HPLC and GC/MS analysis. Incubation of [14C]arachidonic acid with the 100,000 • g supernatant of 3LL cells yielded a single product that cochromatographed on HPLC and thin-layer chromatography with authentic 12-HETE (17). Other than a sizeable peak at the retention time of 12-HETE, the HPLC profile showed only minor UV-absorbing peaks in the HETE region and levels of radioactivity above background were only seen at the retention times of 12-HETE and arachidonic acid. In the HPLC system used for chromatography, 12-HETE (retention time, 23.6 min) was well separated from structurally similar arachidonic acid metabolites; retention times for 5-, 11-, and 15-HETE were 25.9, 21.6, and 19.4 min, respectively (data not shown). GC/MS revealed typical "saddle effect" peaks (30) in the reconstructed gas chromatograms at retention times of 9 min and 22 s as well as identical mass spectra for both compounds. EI spectra were obtained after chromatography on a short fused silica column to minimize the thermal decomposition, and peaks were observed in the reconstructed gas chromatograms at retention times of 6 rain and 30 s (Fig. 6). The EI mass spectra were identical with those reported previously (81) and were different from those derived from authentic 5 K. V. Honnand Z. M. Duniec,unpublishedobservations.

567

12(S)-HETE AND ENDOTHELIAL CELL RETRACTION

12000

10000 0 z

8oo0

z z 0 z E3 z m z uJ.-.