5-hydroxyicosatetraenoate stimulates neutrophils by a stereospecific, G protein-linked mechanism.

We examined how 5-hydroxyicosatetraenoate (5-HETE) activates human neutrophils (PMN). 5-HETE stimulates PMN to mobilize Ca2+ but has little effect on degranulation or superoxide anion production. It nonetheless stereospecifically induced these responses in cells primed with tumor necrosis factor-alpha and likewise induced PMN plasma membranes to bind 35S-labeled guanosine 5'-O-(thiotriphosphate) (GTP gamma S) and phosphohydrolyze [gamma-32P]GTP. Pertussis toxin blocked GTP gamma S binding responses. Scatchard analyses of GTP gamma S binding data indicated that 5-HETE raised the Ka of high affinity GTP gamma S binding sites without altering these sites' numbers or the parameters of low affinity GTP gamma S binding. Since N-formyl-Met-Leu-Phe, platelet-activating factor, and leukotriene (LT) B4 have these same bioactions, receptors for the latter agents might mediate responses to 5-HETE. However, 5-HETE desensitized degranulation responses to itself but not to the receptor agonists, the receptor agonists desensitized to themselves but not 5-HETE, and a LTB4 antagonist inhibited LTB4 but not 5-HETE in all assays. Finally, PMN and their membranes took up [3H] 5-HETE at 4 or 37 degrees C but, at both temperatures, also acylated the radiolabel into glycerolipids. Acylation nullified assessment of 5-HETE binding and questions reports that measure the cell binding, but not metabolism, of various HETEs. Our studies thus indicate 5-HETE acts by a down-regulatable, G protein-linked mechanism and represent the best available evidence that 5-HETE does not operate through, for example, LTB4 receptors.

The [Ca2+]i-raising actions of 5-HETE and LTB4 are equally sensitive to pertussis toxin (59,64). Here as elsewhere, evidence that a HETE operates v i a unique receptors, rather than by, e.g., acting as a weak LTB, receptor agonist or stimulator of effector enzymes, is inconclusive. This report expands the scope of 5-HETEs actions on PMN and provides evidence that the HETE does not operate through receptors involved in transmitting PMN responses to LTB,, PAF, or N-formyl-Met-Leu-Phe.
Preparation of PMN and Membranes-Leukocytes (>95% PMN, t 5 platelets/100 PMN, no erythrocytes) were isolated form normal human donor blood (59). Plasma membranes were prepared at 4 "C by suspending PMN in relaxation buffer, nitrogen-cavitating cells, removing unbroken cells and nuclear debris, and fractionating cavitates on Percoll discontinuous density gradients (65). Material sedimenting with plasmalemma markers was isolated, washed in binding, hydrolysis, or Hanks' buffer, and resuspended in the same buffer.
PMN Bioassays-For degranulation, 2.6 X lo6 PMN/ml of Hanks' buffer were incubated at 37 C for 15 min, exposed to TNF-a or equal amount of BSA for 0.75-57 min, treated with cytochalasin B (5 pg/ ml) for 3 min, challenged for 5 min, placed on ice, and centrifuged (11,000 X g, 15 s, 4 "C) to obtain supernatants, which were assayed for lysozyme and P-glucuronidase (66). Results are reported as the percentage of total cellular enzyme released. For superoxide anion (O;), lo7 PMN/ml of Hanks' buffer were incubated with 50 nM cytochrome c S_ 50 pg of superoxide dismutase at 37 "C for 15 min, treated with TNF-a or BSA for 12 min, exposed to 5 pg/ml cytochalasin B for 3 min, and challenged (66). Results are reported as the maximal rate of superoxide dismutase-inhibitable 0; formed/min/ lo' PMN based on a molar extinction coefficient of 21,000 for the difference between oxidized and reduced cytochrome c. For [Caz+],, PMN were loaded with Fura-2, suspended (SO7 PMN/ml) in Hanks' buffer, incubated at 37 "C for 20 min, and challenged for 5 min while being exited alternately at 340 and 380 nm and monitored at 510 nm with a spectrophotofluorometer. Results are reported as the nano-EDTA standards (59). The latter values were <0.1% of added radioactivity. As found else-where (67), reactions omitting membranes or M%+ did not bind GTPyS, and GTP+ binding was not stimulated when reactions lacked GDP. For Scatchard analyses, membranes were incubated with 3 PM to 3 p~ (0.5 log increasing increments) of labeled and unlabeled GTPyS for 120 min. Data were analyzed with LIGAND SCAPRE and SCAFIT programs. For GTP hydrolysis, membranes (200 ng of protein) were incubated in 80 pl of hydrolysis buffer at 30 'C for 20 min, treated with 20 p1 of hydrolysis buffer (containing 3.7 fmol of [y-3zP]GTP, 10 or 5000 pmol of GTP, 12.5 pg of BSA, and test stimuli), incubated for 1.25-10 min, diluted with 700 p1 of 4 "C buffer (pH 2) containing 10 mM NaH2P04 and 35 mg charcoal, and centrifuged (11,000 X g, 30 min, 4 "C) to obtain 400 p1 of supernatant, which was mixed with scintillation fluid and counted for radioactivity. Results are reported as low K,,, hydrolysis, i.e. the radioactivity liberated from reactions incubated with 10 pmol (100 nM) GTP minus that of reactions incubated with 5000 pmol (50 pM) GTP. The latter, high K,,, values were <8% of low K,,, and did not change during membrane stimulation.

5-HETE
Uptake and Metabolism-lo7 PMN or 5 pg of membrane protein were incubated in 1 ml of Hanks' buffer at 37 or 4 "C for 20 min, exposed to 100 p~ [3H]5-HETE k 5 p~ 5-HETE, and incubated for 5-80 min at 4 "C. For uptake, reactions were passed through GF/ C filters. Filters were washed with five volumes of 4 "C Hanks' buffer, air-dried, incubated in 0.5 ml of methanol for 10 min, mixed with scintillation fluid, and counted for radioactivity (65). Results are reported as the fraction of added label trapped by filters loaded with test reactions minus that trapped by filters loaded with reactions containing all reagents except PMN or membranes. For metabolic studies, reactions were placed on ice, centrifuged (11,000 X g, 5 s, 4 "C), and quickly washed four times with 1 ml of 4 "C Hanks' buffer.
Cells and pooled supernatants were extracted separately by an acidic Bligh and Dyer method and analyzed on TLC (pre-activated (180 ' C for 3 h) silica gel plates developed to 15 cm with ethyl ether/hexane/ glacial acetic acid (6040:1, v/v/v). Zones (0.5 cm) were scraped from the plates, incubated with 0.5 ml of methanol for 10 min, mixed with scintillation fluid, and counted for radioactivity. In selected experiments, identity of isolated material was verified by its elution with 5-HETE or 5,20-diHETE on reverse-phase HPLC or, alternatively, sensitivity to phospholipase Az, lipase, and saponification (51). The latter three reactions converted 90% of label migrating with glycerolipids to label migrating with 5-HETE on TLC. Membrane metabolic studies stopped reactions with acidic Bligh and Dyer extraction solvents and analyzed radiolabel as in cellular studies.
Pertussis Torin Treatment-Pertussis toxin (8 pg) was activated at 30 "C by incubation with 160 p1 of buffer (pH 8) containing 40 mM dithiothrietol, 100 mM triethanolamine, and 160 pg of BSA. After 30 min, reactions were diluted with 160 p1 of the same buffer containing 4 pg of membrane protein and 15 mM NAD'. After 30 min, 20-pl membrane samples were added to 80 p1 of binding buffer containing (final amounts) 50 pM [36S]GTPyS -+ 100 p~ GTPyS, 320 nM GTP, 12.5 pg of BSA, and test stimuli to assay for GTPyS-specific binding.
Results are compared to membranes that were identically treated except for being incubated with 8 pg of BSA in place of pertussis toxin. be released (first data point for each curve of Fig. 2). In combination, in contrast, the two agents proved capable of releasing remarkably large amounts of lysozyme. Responses increased with TNF-a (3-3200 units/ml) and 5(S)-HETE (5-5000 nM) concentrations. For all stimuli combinations, however, PMN incubated with TNF-a for 7.5-15 min were more responsive to 5(S)-HETE than cells treated with TNF-a for 22.5 or 30 min (Fig. 2), whereas PMN exposed to TNF-a for 3.75 or 60 min had minimal responses to 5-HETE (not shown). Companion experiments helped interpret these results. First, TNF-a and 5(S)-HETE also cooperated to release p-glucuronidase (Fig. 3, rightpanel) and trigger 0; production ( Table I). The TNF-a/5(S)-HETE interaction therefore is not limited to secondary granule enzyme (lysozyme) release but also includes release of a primary granule enzyme (8glucuronidase) as well as stimulation of the oxidative burst. Second, 5(R)-HETE was much weaker that 5(S)-HETE in triggering TNF-a-pretreated PMN to degranulate and produce 0; (Fig. 3 and Table I). 5(S)-HETE thus operated stereospecifically to activate TNF-a-primed PMN. Third, cytokine-primed PMN, when pretreated with 5(S)-HETE, degranulated minimally when re-challenged with 5(S)-HETE but responded fully to LTB4, fMLP, and PAF. Conversely, the latter agents desensitized PMN to themselves but not to each other or to S(S)-HETE (Table 11) TABLE I 01 production induced by 5-HETEs in TNF-a-primed PMN PMN (lO'/ml) were incubated with BSA f 1000 units/ml TNF-a for 12 min, exposed to 5 pg/ml cytochalasin @ for 3 min, and challenged with 500 nM of the indicated HETE.

TABLE I1
Desensitization of degranulation responses in TNF-a-primed PMN PMN were pre-incubated with 320 units/ml TNF-a plus 1.58 pM 5(S)-HETE, 1 nM LTB,, 1 nM PAF, or 1 DM fMLP for 4.5 min, exposed to 5 pg/ml cytochalasin B for 3 min, and challenged with these same concentrations of stimuli for 5 min. ' Values were significantly lower ( p < 0.01, Student's paired t test) than those for PMN not exposed to a desensitizing stimulus.
activate cytokine-primed cells by mutually independent mechanisms. Fourth, LY 255283 blocked LTB4, but not 5-HETE, in stimulating TNF-a-primed PMN to release lysozyme and @-glucuronidase ( Table 111). The antagonist by itself failed to elicit degranulation and, in contrast to results in [Ca"']; assays, did not suppress 5(S)-HETE-induced degranulation. This finding further differentiates the PMN-stimulating actions of 5(S)-HETE from LTB4. Finally, PMN did not release the cytosolic enzyme, lactic acid dehydrogenase, in any of the assays. Degranulation, therefore, was not due to PMN cytolysis. We conclude that 5(S)-HETE acts by a stereospecific, selectively desensitizable mechanism to stim- for 10 min, exposed to 0 or 1 PM LY 255283 for 2.5 min, treated with cytochalsin B for 2.5 min, and challenged for 5 min with 500 nM 5(S)-HETE or 3 nM LTB,. ulate TNF-a-primed PMN. Pathways transducing functional responses to 5(S)-HETE must diverge from those for LTB4, fMLP, and PAF at one or more points. G Protein Activities-PMN plasma membranes incubated with [36S]GTPyS bound radiolabel increasingly over 120 min after which apparent equilibrium occurred. 5(S)-HETE stimulated GTPyS binding at late ( t 2 30 min; Fig. 4, upper left panel) as well as early time intervals. Relative to the last point, control membranes (studies done as in Fig. 4)  min was significantly greater (p < 0.05) than control membranes at all time points.) In this assay, 5(R)-HETE was -100-fold weaker than 5(S)-HETE (Fig. 4, upper rightpanel); LY 255283 blocked LTB, but not 5(S)-HETE (Fig. 51, and LY 255283 lacked intrinsic effects (not shown). 5(S)-HETE also stimulated membranes to phosphohydrolyze [y-32P]GTP. Again, 5(R)-HETE was -100-fold weaker than 5(S)-HETE (Fig. 4, lower panels). 5(S)-HETE thus activated G proteins by a stereospecific mechanism differing from that used by LTB,. To define those G protein changes responsible for increased GTP binding and hydrolysis, we analyzed GTPyS binding data by the method of Scatchard. As with HL-60 membranes (67), resting PMN plasmalemma membranes possessed high and low affinity GTPyS binding sites. 5(S)-HETE increased the K, for high affinity binding sites but did not change these sites' numbers or the KO and Bmax for low affinity binding (Table IV). LTB4, fMLP, and PAF likewise promote GTPyS binding (Fig. 6), increase GTP hydrolysis rates (data not shown), and selectively raise high affinity GTPyS binding site K, without influencing other parameters of GTPyS binding (Table IV). In each of these assays, however, the latter agonists have not only much greater potency but also induce larger optimal responses than 5(S)-HETE. Nevertheless, pertussis toxin reduced GTPrS binding responses to all four agents (Table v). Evidently, then, 5(S)-HETE is less potent and powerful than LTB4, fMLP, or PAF in activating G proteins. However, it resembles the latter agonists in targeting pertussis toxin-sensitive G proteins and selectively increasing the affinity of high affinity guanine nucleotide binding. . , but, as indicated by asterisks, significantly reduced ( p < 0.05, Stu- PMN plasma membranes PMN plasma membranes were incubated (2 h, 30 "C) with 3 p~ to 3 PM [36S]GTP$3 plus GTPyS, 320 p~ GDP, 4 mM free M e , and 125 pg/ml BSA f the indicated stimulus. Reactions were filtered, counted for radiolabel binding, and evaluated for binding sites using LIGAND. In all cases, LIGAND indicated the presence of two binding sites ( p < 0.02, F-distribution). than membranes incubated with no pertussis toxin. uptake peaked at -0.12 (Fig. 7 ) . While such results might reflect receptor binding, PMN are known to process 5-HETE at 37 "C. For example, PMN suspensions exposed to 100 PM [3H]5(S)-HETE +-5 PM 5(S)-HETE at 37 "C for 5-80 min contain labels that migrate with triglyceride, phospholipid, 5(S),20-diHETE, and 5(S)-HETE on TLC. Labels migrating with 5,20-diHETE and 5(S)-HETE occur only in extracellular fluid, whereas cell-associated labels migrate with triglyceride and phospholipid (51). PMN incubated with 100 pM . The supernatants and cells were extracted, extracts were chromatographed on TLC, and 5-mm zones were scraped from the TLC plates and counted for radioactivity. Data are the mass of material recovered per fraction based upon the specific activity of added label (>94% recovery of starting radioactivity). Results are representative of 4 experiments. Note that the y ares for supernatants (left panels) are magnified 5-fold compared to cells (right paneki). Migration of 5-HETE, 5,20-diHETE, phospholipid, and triglyceride standards are indicated.

PHl5(S)-HETE
[3H]5-HETE at 4 "C for 80 min contained extracellular radiolabel that migrated with B(S)-HETE but not 5,20-diHETE (Fig. 8, upper left panel). The lower temperature thus blocks o-oxidation not only of LTB, (65) but also of 5(S)-HETE. Unexpectedly, however, cellular radioactivity in these experiments migrated exclusively with triglyceride and phospholipid (Fig. 8, upper rightpanel). Addition of 5 PM 5(S)-HETE to the PMN reduced label uptake without changing the general pattern of results (Fig. 8, lower panels). Similar findings occurred with PMN incubated for 40 min at 4 "C (data not shown). Indeed, plasmalemma exposed to 100 pM [3H]5-HETE at 37 or 4 "C for 40 min formed triglyceride-and phospholipid-co-migrating radioactivity. In the presence of 5 PM 5(S)-HETE, the membranes converted lower although still measurable percentages of the label to the latter species (data not shown). Further studies found that: ( a ) radioactivity in the supernatants of whole cell (4 "C) incubations migrated (>go%) with 5(S)-HETE but not 5,20-diHETE on HPLC; ( b ) lipase treatment or saponification of cell or membrane labels that migrated with triglyceride on TLC (from 37 or 4 "C incubates) yielded label (>98%) moving with 5(S)-HETE on TLC; and (c) phospholipase A2 treatment or saponification of labels that migrated with phospholipid on TLC (from 37 or 4 "C incubates) also yielded label moving (>98%) with 5(S)-HETE on TLC (not shown). Thus, PMN take up and acylate 5(S)-HETE into triglycerides and phospholipids at 4 as well as 37 "C. Acylation reactions keep pace with ligand uptake to effect virtually complete removal of intact 5(S)-HETE from PMN. Membranes similarly acylate 5(S)-HETE at both temperatures. In all cases, then, 5(S)-HETE uptake and metabolism occur more or less concomitantly. DISCUSSION 5(S)-HETE has limited actions on PMN. It induces Caz+ transients but has little or no ability to stimulate degranulation or oxidative metabolism. We broadened the biological spectrum of 5(S)-HETE using TNF-a, a cytokine that primes PMN to LTB,, PAF, and fMLP (66). TNF-a, in effect, converted the HETE from an incomplete agonist to one fully capable of eliciting exocytotic and 0 2 production responses (Fig. 2, Table I). 5(S)-HETE furthermore desensitized TNFa-primed PMN to itself (Table IT), stimulated PMN plasmalemma isolates to hydrolyze [-p3'P]GTP and bind
Our functional studies indicate that 5(S)-HETE acts by a down-regulatable, G protein-linked mechanism that does not * J. T. O'Flaherty and A. G. Rossi, unpublished observations. involve receptors for three typical agonists. Obviously, PMN receptors for other agonists, e.g. lipoxin A, (69), prostaglandin E', prostaglandin D2 (64), or ATP (70), could be S(S)-HETE targets. However, prostaglandins E2 and Dz block, rather than enhance, PMN function, and excess 5(S)-HETE does not interfere with PMN binding of these prostanoids (64). Likewise, lipoxin A4 and B4, at 0.02-2 MM, do not mimic 5(s)-HETE in promoting PMN responses to PAF, and various nucleotides including ATP, while enhancing PAF-induced responses, do not cross-desensitize with 5(S)-HETE in assays of degranulation.' Indeed, 5(S)-HETE is peculiar among Ca2+-mobilizing, pertussis toxin-sensitive stimuli in that it elicits functions like 0 ; production only when PMN are COchallenged with protein kinase C activators (601, P@ (611, or TNF-a (Figs. 2 and 3; Table I). We are not aware of any other PMN receptor agonist with these particular bioactions and therefore suspect that 5(S)-HETE bypasses all known receptors to stimulate PMN.
Although the data presented here are fully compatible with the notion that 5(S)-HETE acts through receptors, they do not exclude alternate PMN-stimulating mechanisms, e.g. direct G-protein activation. It is nevertheless interesting to speculate on putative 5(S)-HETE receptors. Such receptors would be directed toward cooperating with certain stimuli to elicit PMN function. We note that other PMN eicosanoid receptors localize to plasma membrane and associate with either G, (prostaglandins E' and D2 receptors) or Gi (LTB, and lipoxin A4 receptors) proteins (64,65,69). Our GTP binding and hydrolysis studies suggest that 5(S)-HETE receptors, if existing in PMN, are also plasmalemmal-bound and Gi protein-linked. As a first impression, then, the known PMN eicosanoid receptors may belong to the rhodopsin superfamily of cell surface recognition molecules and relate to each other in a manner formally analogous to, e.g., adrenergic rhodopsin superfamily receptors (71). Adrenergic receptors fall into different classes based in part on their affinities for a series of analogs. Ekosanoid receptors might also be profitable classified into, e.g., stimulatory El (LTB,, lipoxin &, putative 5(S)-HETE) and inhibitory E2 (prostaglandins EZ and Dz) receptors. Here, as in the adrenergic system, receptors within a single class may differ. LTB, receptors elicit a large array of functions; lipoxin A4 receptors cause a different and often inhibitory set of responses; and putative 5(S)-HETE receptors stimulate function only in PMN co-challenged with certain other stimuli. Relevant to the last point, it seems at least possible that the latter stimuli up-regulate 5(S)-HETE receptors. TNF-a induces PMN to increase their expression of fMLP and PAF receptors (66) and conceivably might have the same effects on putative 5(S)-HETE receptors. In any event, recent studies indicate that PMN synthesize various oxo analogs of arachidonic acid metabolites (55,56). Our studies recommend these and other novel analogs be tested for interaction with the full range of El receptors. Possibly, an eicosanoid analog will be found to interact with multiple E, receptor types and thereby indicate a close relationship between these fascinating receptor systems.