Assessment of a role for phospholipase A2 and arachidonic acid metabolism in human lymphocyte natural cytotoxicity

CELLULAR

IMMUNOLOGY

87,270-283 (1984)

Assessment of a Role for Phospholipase A2 and Arachidonic Metabolism in Human Lymphocyte Natural Cytotoxicity

Acid

KATHLEEN CARINE’ AND DOROTHY HUDIG* UniVerSitYof Cah~ornia. San Diego, Cancer Center, and the Department of Medicine, T-011, University of California, San Diego, La Jolla, California 92093 Received November 1, 1983; accepted March 27, 1984 The reagents quinacrine, hydrocortisone, and dexamethasone have been assumed to affect phospholipase A2 (PAZ) when they reduce natural killer (NK) activity. However, these reagents did not reduce lymphocyte incorporation of [‘4C]arachidonate, which implies that they are not acting as PA2 inhibitors in this lymphocyte system. However, pbromophenacyl bromide (BPB), which is an active site inhibitor of PA*, irreversibly abrogatedNK activity of pretreatedlymphocytes, disrupted target cell binding, and reduced [i4C]arachidonic acid incorporation by 70-80% as compared to controls. Other observations contrary to expectations for PA2 inhibitors were: (1) quinacrine inhibited NK lysis when lymphocytes were pretreated and (2) the glucocorticoids only inhibited NK activity when continuously present in the assay.Furthermore, NK inhibition by hydrocortisone did not require protein synthesis. The lipoxygenase inhibitors, nordihydroguaiaretic acid (NDGA), 5,&l 1,14eicosotetraynoic acid (ETYA), and hydroxyphenylretinamide, and not cycloxygenase inhibitors, reduced NK activity. These data suggest that arachidonate must be metabolized through the 5-lipoxygenase pathway in order to function in NK.

INTRODUCTION Human natural killer (NIQ3 activity is defined operationally as the ability of peripheral blood lymphocytes to lyse tumor cells in vitro. The mechanism of this lysis is not understood. One current theory suggeststhat fatty acid mobilization by phospholipase activity may be important to secretion and/or formation of a putative NK lytic substance (l-4). However, this hypothesis was supported by the assumption that quinacrine, glucocorticoids, and lipomodulin are truly inhibitors of lymphocyte phospholipase A2 (PA*) activity. Since the efficacy of these reagents on lymphocyte PA2 activity had never been verified, we evaluated the effects of these reagents (except for the regulatory protein lipomodulin) on lymphocyte arachidonate metabolism. In the course of this investigation, we found them not to affect the assay for PA2 activity. We chose the active site PA2 inhibitor pbromophenacyl bromide as an additional ’ To whom correspondence should be addressed,present address:Department of Biology, B-022 UCSD, La Jolla, California 92093. * Supported in part by NIH T32 CA 09290 and NIH CA 28196; present address: Department of Microbiology, School of Medicine, University of Nevada, Reno, Reno, Nevada 89557-0046. 3Abbreviations used NK, natural killer, PA*, phospholipaseA*; BPB, pbromophenacyl bromide; ETYA, 5,8,11,14-eicosatetraynoic acid; NDGA, nordihydroguaiaretic acid; BSA, bovine serum albumin; DMSO, dimethyl sulfoxide; TLC, thin-layer chromatography; PBMC, peripheral blood mononuclear cells; MW, molecular weight; TCGF, T-cell growth factor; HETE, hydroxyeicosatetraenoic acid. 270 0008-8749184$3.00 Copyright 0 1984 by Academic PI%%% Inc. All [email protected] of repdudon in any form reserved.

PHOSPHOLIPASE A2 AND ARACHIDONATE

METABOLISM

271

Phospholipids

phospholipase

1

T Arachidonic

glucocorticoids bromophenacyl quinacrine

bromide

acid

aspirin, indomethacin

LEUKOTRIENES

PROSTAGLANDINS THROMBOXANES

FIG. 1. Pathways for arachidonic acid metabolism. Double bars indicate points at which inhibitors act.

probe to evaluate this question. This reagent did affect both arachidonate metabolism and NK activity. In addition, we sought to establish whether specific metabolism of PA2 products was essential for NK activity. PA2 has specificity of cleavage for the fatty acid in the two position of the glycerol backbone of phospholipids. Since arachidonic acid is the principal fatty acid in this position, PA2 activity has been thought to be a crucial first step for generation of arachidonic acid products.4 Arachidonate is then metabolized by the cycloxygenase pathway to result in products such as prostaglandins, or by the lipoxygenase pathway to yield products such as leukotrienes. In this paper, several inhibitors of PAI, cycloxygenase, and lipoxygenase metabolism of arachidonate were used. The reported effects of these inhibitors on the metabolism of arachidonic acid are illustrated in Fig. 1. Bromophenacyl bromide (BPB), which binds at the active site pancreatic PA2 (7), irreversibly inhibits PA*. Quinacrine (mepacrine) can reduce the rate of hydrolysis of PA2 in vitro (8), and can inhibit 1-palmitoyl-2[ 1-‘4C]arachidonylphosphatidylcholine release from sonicated rat platelets (9). The glucocorticoids dexamethasone and hydrocortisone inhibit releaseof arachidonic acid from tissues and cells (10, 11). Also examined for effects on NK activity were: eicosatetraynoic acid (ETYA), which can inhibit production of both cycloxygenase and lipoxygenase products ( 12, 13); nordihydroguaiaretic acid (NDGA), which can reduce lipoxygenase activity (14, 15); and hydroxyphenylretinamide which preferentially inhibits 5-lipoxygenase at micromolar concentrations (16). Our contribution in this paper is to dispel previous misconceptions of the effects of quinacrine and glucocorticoids on lymphocyte PA2 activity, to investigate how these reagents might still affect NK activity, to demonstrate that lymphocyte NK and PA2 are affected by BPB, and to indicate in experiments with NDGA, ETYA, and hydroxyphenylretinamide that 5-lipoxygenase metabolites of arachidonate are likely to be required for NK activity. 4An alternate pathway has been proposed in which phospholipase C and diglyceride lipase act on phospholipids (particularly phosphatidyl inositol) to produce free arachidonic acid (5, 6).

272

CARINE AND HUDIG

MATERIALS

AND METHODS

Preparation of human peripheral blood NK efectors (PBMC). Peripheral blood from normal healthy donors was defibrinated, sedimented with 1% dextran (180,000 MW, Pharmachem Corp., Bethlehem, Pa.), and separatedby Ficoll-Hypaque density gradient centrifugation to yield mononuclear cells. These mononuclear cells (84-92% lymphocytes and 8- 16%esterase-positivemonocytes) were useddirectly asNK effecters or were depleted of monocytes by adherence to plastic for 1 hr at 37°C. Cells were resuspended in an assaymedium consisting of Dulbecco’s modified Eagle’s medium (DMEM, GIBCO, Grand Island, N.Y.) supplemented with 10 mM 4-(2-hydroxyethyl)I-piperazineethanesulfonic acid (Hepes, Calbiochem, La Jolla, Calif.) and 1% heatinactivated (56°C 30 min) fetal calf serum (GIBCO or Flow Laboratories, Rockville, Md.). Cell viability was determined by trypan blue exclusion. “Cr Cytoxicity Assay. NK activity was measured by monitoring 51Crrelease from K562 cells labeled with sodium chromate (200-900 Ci/g, New England Nuclear Corp., Boston, Mass.) ( 17). K562 is an erythroleukemic cell line derived from a patient with chronic myelogenous leukemia (18) and was used between passages270 and 375. The percentage specific chromium release was calculated as follows: [(experimental 51Crrelease - background “Cr release)/(maximum 51Crrelease - background “Cr release)] X 100. The rate of background 51Cr release was between 1.2 and 2.8%/hr, except in experiments using either ETYA or hydroxyphenylretinamide, where background 5’Cr releasevaried between 3 and 5%/hr. The results illustrated in the figures and tables are representative of at least three separateexperiments using lymphocytes from a different donor for each experiment. The ID5,, is the concentration of a reagent at which 50% inhibition of NK activity occurs, as indicated when twice as many treated cells are required to yield the same linear titration of cytotoxicity as expressed by untreated control cells. Statistical analyses.Statistical analyseswere done in order to evaluate the probability of two linear regression slopes being identical. A two-tailed Student t test was used to test the probability of the slopes being similar:

b = slope of the regression line s = standard deviation of b with rrl + n2 - 4 degreesof freedom. Minitab programs ( 19) were used to obtain the values needed for analyses. The criteria used in determining effective inhibition were P > 0.5 as an indication of failure to find a difference in the slopes, and P < 0.05 as an indication of a low probability of the lines being the same but appearing different. Chemicals. Cycloheximide, dexamethasone, pbromophenacyl bromide (BPB), fluores&n diacetate, hydrocortisone, nordihydroguaiaretic (NDGA), and quinacrine were obtained from Sigma Chemical Company (St. Louis, MO.). Eicosatetraynoic acid (ETYA) and hydroxyphenylretinamide were graciously supplied by Dr. W. E. Scott and Dr. Peter Sorter (Hoffmann-LaRoche Inc., Nutley, N.J.). Rosenthal’s inhibitor was obtained from Calbiochem. Fluorescein iodoacetamide was purchased from Molecular Probes (Junction City, Greg.). Chemical solutions were prepared fresh daily in either absolute ethanol or dimethyl sulfoxide. Dilutions were made in

PHOSPHOLIPASE

AZ AND

ARACHIDONATE

METABOLISM

273

assaymedium. Control dilutions contained corresponding concentrations of ethanol or DMSO. Toxic effects of chemicals on cells were routinely checked at the end of the NK assay by trypan blue exclusion. Pretreatment of effector cells (BPB for 1 hr at room temperature, quinacrine for 1 hr at room temperature or at 37°C) was followed by two washesprior to cytotoxicity assays.[ l-‘4C]Arachidonic acid (58 mCi/ mmol) was obtained from New England Nuclear Corporation or Amersham/Searle (Arlington Heights, Ill.). All other chemicals used were reagent grade. T-Cell growth factor (TCGF or Interleukin 2)-dependent cells. PBMC were cultured 48 hr in RPM1 1640 supplemented with 20% fetal calf serum, 10 mM 4-(2-hydroxyethyl)- 1-piperazineethanesulfonic acid (Hepes), 1% gentamycin (Irvine Scientific, Santa Ana, Calif.), and 0.5 &ml phytohemagglutinin (PHA, Burroughs Wellcome, Research Triangle Park, N.C.). After 48 hr, the ceils were centrifuged and the supematant was removed and stored asthe source of TCGF. Fresh medium supplemented with 20% fetal calf serum (without PHA) was added to some of the remaining cells. Following approximately 7 days in culture, half the volume was removed and replaced with medium containing 25% TCGF supematant. The medium was changed in this manner every 4-5 days until the day of use, which was normally 1O-l 5 days after initiation of culture. Conjugation assay. To facilitate visualization of conjugates, K562 cells were labeled with fluorescein diacetate (2 &ml per l-2 X IO6 cells) for 1 hr at 37°C washed three times, and added to effector cells (E:T = 2:l). Fifty microliters of effecters (8 X 106/ml) and 50 ~1 of K562 targets (4 X 106/ml) were added to 12 X 75-mm polypropylene tubes containing 0.1 ml of medium or inhibitor. The tubes were centrifuged at 40g for 3 min, and put on ice. Conjugates were determined by gently resuspending the cells and counting on a hemocytometer using a Zeiss epifluorescent microscope. Each lymphocyte attached to a K562 cell was scored as one conjugate. The tubes were coded using random numbers by a colleague and the tubes were identified after the experiment was complete. Thin-layer chromatography. Following [ 14C]arachidonic acid incorporation (0.1 &i/lo6 cells in 1.O ml assay medium at 37°C for 30 min), cells were washed twice in medium, then disrupted in distilled water. Lipids were extracted with one part chloroform-methanol-acetic acid (2/4/ 1, v/v) and 0.5 part chloroform. The chloroform layer was evaporated to dryness, reconstituted in 20-50 ~1 chloroform, and applied to Brinkmann Silica Gel 60 plates (0.25 mm, Brinkmann Instruments, Westbury, N.Y.). Chromatographic separation was accomplished in a saturated glass chamber using developing solvents of either chloroform-methanol-acetic acid-water (25/ 15/ 4/2 or 100/60/ 16/4, v/v) or chloroform-acetone-methanol-acetic acid-water (50/ 20/ 1O/10/5, v/v). Phospholipid standards were obtained from Supelco (Bellefonte, Pa.). After drying, plates were exposed to KODAX XAR-5 X-ray film at -70°C for 2-7 days. Following autoradiography, lipid standards were visualized under iodine vapor. Lanes were then subdivided into regions containing different lipids, scraped, and the radioactivity determined in a Beckman LS6800 scintillation counter. RESULTS Several Potential PA2 Inhibitors SuppressedNK Activity We began our studies by establishing that BPB inhibited NK, and by verifying that the reagents previously evaluated did affect NK. However, we found that aspects of

274

CARINE AND HUDIG

the NK inhibition by these reagentsindicated that they might not be alfecting lymphocyte PAI. Pretreatment of macrophage-depleted, nonadherent effector cells with the PA2 inhibitor BPB (5 X 10d6 M) for 1 hr at room temperature prior to cytolytic assay effectively abolished NK activity (Fig. 2). This effect remained after washing the lymphocytes and was therefore irreversible, as would be expected from the effect of BPB on isolated PA*. Quinacrine also inhibited NK, and was effective both in the NK assay (IDS0 = 5 N), and by pretreatment (Fig. 3) of the lymphocytes at 37’C (IDsc, = 5 N). This latter observation is anomalous to expectations for the reagent quinacrine as a PA2 inhibitor, since it should be fully reversible by pretreatment. In addition, when pretreatment was done at room temperature, quinacrine had no effect on NK activity (results not illustrated). Rosenthal’s inhibitor, a phosphatidylcholine analog and a competitive inhibitor for phospholipase A2 activity, was toxic to lymphocytes at inhibitory concentrations (1 X 10e4M) (results not shown). Thus only two of these three reagents which can inhibit isolated PA2 activity in micelles, BPB and quinacrine, were suitable reagents for cellular studies and only BPB conformed to expectations for a PA2 inhibitor. The glucocorticoid hormones, hydrocortisone and dexamethasone, inhibited NK activity at 10-5-10-4 M concentrations (Fig. 4). The continuous presence of the hormone was required for this inhibition. Pretreatment of effector cells for 1 hr at 37°C followed by washing prior to assay had no effect (not illustrated). Becausethe effects of these steroids on PA2 activity are inducible in several systems (20, 21), we first evaluated whether inhibition was as great in very short NK assaysas compared with those of longer duration. The hydrocortisone inhibition after 1 hr in assaywas the same as after 3 hr (results not shown). r

60c

BPB

P

FIG. 2. BPB-abolished NK activity. Fresh nonadherent PBMC were pretreated with BPB, washed, and tested for NK activity in a standard 4-hr Wr-release assay using K562 cells as tar@ cells. E:T, effector to target ratio. Five micromolar BPB caused complete abrogation of NK. The viabilities of the lymphocytes were not afktcd by these BPB concentrations.

PHOSPHOLIPASE A2 AND ARACHIDONATE

METABOLISM

275

E:T

FIG. 3. NK activity was irreversibly inhibited by quinacrine. Lymphocytes were pretreated at 37°C for 1 hr and washed. The subsequent NK activity of these cells was inhibited in a dose-dependent manner, P < 0.001 for 5 and 10 X low6 M. Pretreatment at room temperature at 10 X low6 M was without effect (not illustrated).

t 4 40d $ 230 ‘c 8 2 20s lo-

A

1.25

2.5

5 ET

10

20 E:T

FIG. 4. Inhibition of NK by the glucocorticoids hydrocortisone and dexamethasone. (A) Stock hydrocortisone was prepared in absolute ethanol, diluted in assaymedium, and incubated with PBMC and K562 cellsin a standard4-hr 5’Cr-releaseassay.Twofold inhibition of NK wasobservedat 5 X 10-5Mhydrocortisone, P < 0.002. (B) Dexamethasone similarly depressedNK activity when included in a standard assay.Twofold inhibition occurred at slightly lesserthan 10 X 10V5M, P i 0.001. These concentrations of hydrocortisone, dexamethasone, and ethanol were without detectable effects on PBMC or target cell viability.

276

CARINE AND HUDIG

The mechanism by which these steroids can affect phospholipase A2 activity is not clear, although in several cellular systems it appears that the synthesis of a protein PA2 inhibitor termed lipomodulin is induced (22,23). In order to determine whether the inhibition of NK activity required new protein synthesis, cycloheximide was included with hydrocortisone in the assaymedium. If new synthesis of a PA2 inhibitor were needed for NK inhibition, then cycloheximide should reverse the inhibition of NK cytolysis observed with hydrocortisone. As shown in Fig. 5, our results indicate that hydrocortisone effects on NK activity do not require de lzovo protein synthesis. Cycloheximide did have such effects on experiments using guinea pig lung, rat renal papilla, and rabbit peritoneal neutrophils in which hydrocortisone effectson PA2 and arachidonate release were blocked (22, 24, 25). The concentration of cycloheximide employed in these experiments effectively blocks the incorporation of radioactive leucine by human lymphocytes (17). The results of these initial experiments indicate that it is unlikely that the ability of quinacrine and glucocorticoids to inhibit NK was due to effects on PAZ. On the strength of the previous report of the effects of lipomodulin on NK lysis (3 1) and on our data with BPB, we continued to study the premise that PA2 is involved in NK activity. Also, we evaluated whether further metabolism of PA2 products was needed for natural killing. Lipoxygenase but Not Cycloxygenase Inhibitors Afect NK Activity It is possible that the product of phospholipase A2 activity, arachidonic acid, may play a key role in NK lysis, since arachidonic acid is the initial substrate for the enzymes that initiate formation of both prostaglandins and thromboxanes (which are

E:T

PIG. 5. Inhibition of NK by hydrocortisone was not dependent on new protein synthesis. Hydrocortisone (1 X 10e4M) and/or cycloheximide (2 &ml) were included in a 4-hr NK assay. As indicated by the diamond symbols, the combination of cycloheximide and hydrocortisone was slightly more inhibitory than hydrocortisone alone (indicated by squares).Cycloheximide was essentially without effect (triangles). PBMC and K562 cell viabihties were not atlkcted by these concentrations of hydrocortisone and cycloheximide. Treatment of PBMC with hydrocortisone and/or cycloheximide 1 hr prior to assayfolIowed by hvo washes, and assay in medium without hydrocortisone or cycloheximide also had no effect on NK activity (not illustrated~.

PHOSPHOLIPASE AZ AND ARACHIDONATE

METABOLISM

277

NDGA 50-

z!z 402 ;

30-

g fii a 20v) s0 IO-

Y.’ 1.25

2.5

5 E:T

IO

20

FIG. 6. Inhibition of NK by NDGA. Nonadherent PBMC were used as effecters. NDGA was initially dissolved in 100% ethanol and then added to assay medium. All control and NDGA dilutions contained a constant concentration of 0.1% ethanol. A concentration of 1 X lo-’ M NDGA reduced NK activity 50%. P < 0.001.

products of the cycloxygenasepathway) and leukotrienes (products of the lipoxygenase pathway). The cycloxygenase inhibitors, indomethacin and acetylsalicylate (aspirin), were investigated for their ability to inhibit NK. At concentrations between lO-‘j and 10m8A4 for indomethacin and up to 5.0 mM for aspirin, NK activity was not affected (data not shown). However, compounds which affect the lipoxygenase enzymes such as nordihydroguaiaretic acid (NDGA), 5,8,11,14-eicosatetraynoic acid (ETYA), and hydroxyphenylretinamide all reduced NK activity (Figs. 6-8). NDGA and especially hydroxyphenylretinamide can inhibit the 5-lipoxygenase enzyme preferentially. These

40

ETYA

E:T

FIG. 7. NK activity was inhibited by ETYA. Nonadherent PBMC were used as effecters. ETYA was initially dissolved in DMSO. All control and ETYA dilutions contained 0.25% DMSO. BSA (0.3 mg/ml, Cohn Fraction V, Sigma Chemical Co.) replaced fetal calf serum in the medium of these experiments. The IDSofor ETYA was 5 X lo-’ M; P < 0.05.

278

CARINE AND HUDIG

10 E:T

I

20

I

40

FIG. 8. Hydroxyphenylretinamide reduced NK activity. Nonadherent PBMC were used as effector cells. Twofold inhibition of NK activity occurred at 5 X 10e6M, the P < 0.005 in this representative experiment. Control dilutions contained 0.1% DMSO.

two compounds had similar IDso values (5- 10 PM) for inhibition of NK cytolysis. The reagent ETYA, which can all&t both the lipoxygenase and cycloxygenaseenzymes, was required in concentrations higher than those of NDGA and hydroxyphenylretinamide for reduction of NK activity (ID =,,,= 50 PA,&Fig. 7). It should be noted that NDGA can also have other cellular effects,such as scavengingof cellular oxidation products. Also, the reagents ETYA and hydroxyphenylretinamide are extremely susceptible to oxidation, resulting in toxicity to lymphocytes. Proper care was taken to reduce exposure to air by preparing dilutions immediately prior to use and by storing the chemicals under a nitrogen atmosphere.

E$ects of Inhibitors on Early Binding Events In each of two separateexperiments using different donors for PBMC, quinacrine, hydrocortisone, and NDGA had no effect on binding of lymphocytes to K562 target cells (Table 1). In striking contrast, BPB at a concentration of 5 X 10e6M, which abolished NK activity, also inhibited binding by 75% (Table 1). It must be emphasized, that lower concentrations of BPB (e.g., 0.2 @W),which did not affect binding, still inhibited NK activity. Previous studies from this laboratory have indicated that sulfhydryl groups are needed for binding of target cells (26). To address the possibility that 5 pit4 BPB might also be reacting with these sulfhydryl groups, we determined whether pretreatment with BPB blocked subsequent availability of thiol groups to alkyation with a fluorescent probe. Fresh PBMC were pretreated with 5 X lop6 M BPB and then allowed to react with fluorescein iodoacetamide (5 X lop4 M) for 20 min at 37°C. No reduction of the fluorescent label of pretreated cells was observed when compared to controls. Quantitation was made with an Ortho fluorescent-activated cell sorter, (Model 50H, Ortho Diagnostics Systems, Raritan, N.J.). Moreover, concomitant addition and incubation of PBMC with both BPB (5 H) and fluorescein iodoacetamide did not inhibit the cells from binding the fluorescent reagent.

PHOSPHOLIPASE A2 AND ARACHIDQNATE

279

METABOLISM

TABLE 1 Effect of Quinacrine, Hydrocortisone, NDGA, and BPB on NK-KS62 Conjugation

Experiment

Reagent

Percentage of lymphocytes in conjugates’b

Control Quinacrine ( 10 PM)

6 I

Control Hydrocortisone (100 &4)

9 8

Control NDGA (20 fl)

9 I

Control BPB (0.2 /&) BPB (1.0 pcM) BPB (5.0 ~uW)

12 12 8 3'

’ The percentage of lymphocytes in conjugates is (the number of lymphocytes in conjugates divided by the number of total lymphocytes) X 100. bControl conjugates were 103, 164, 169, and 212 bound lymphocytes of 1800 total lymphocytes for Experiments 1, 2, 3, and 4, respectively. These are representative experiments which were repeated using three different donors. 'P < 0.005.

Only the Reagent BPB Altered the Metabolism of Radiolabeled Arachidonic Acid by Lymphocytes The studies above indicated major reservations concerning the likelihood that several of the reagents were actually affecting lymphocyte PA2 activity. Therefore, we directly measured the effectson arachidonate metabolism for all reagents used in this study. Nonadherent PBMC were pretreated with either BPB (5 X lop6 M) or control solvent and then allowed to incorporate [‘4C]arachidonic acid for 20 min at room temperature. These BPB-treated lymphocytes did not metabolize the labeled fatty acid as effectively as did the controls. Most of the [‘4C]arachidonate remained unincorporated in the phospholipid fraction of the BPB-treated cells whereas in control cells the label was further metabolized (results not illustrated). The untreated cells incorporated the label into a neutral class of lipid which appeared at the solvent front observed after TLC and autoradiography. In order to eliminate a potential contribution of PA2activity from residual monocytes and polymorphonuclear cells of PBMC preparations, which can produce biologically active arachidonic acid metabolites (e.g., HETES, leukotrienes), TCGF-dependent lymphocyte lines were employed for further radioactive arachidonate incorporation studies. These lines contained no cells which stained positive for the nonspecific esterasesused to distinguish monocytes. The cells were pretreated with either BPB (5 X 10e6M), or quinacrine (1 X lop5 M), washed, and then allowed to incorporate [‘4C]arachidonic acid for 30 min at 37°C. If there were release of previously incorporated unlabeled arachidonic acid from the cells due to phospholipase activity, then there would be new sites for incorporation of labeled arachidonic acid. This approach is a more sensitive indication of PAZ activity than monitoring removal of previously

CARINE AND HUDIG

280

incorporated radioactive arachidonic acid. BPB caused a 70-8096 reduction in the amount of radioactivity incorporated into phosphotidylcholine and phosphotidylethanolamine (& = 0.3 and 0.53, respectively, in the chloroform-acetone-methanolacetic acid-water solvent solution) (Table 2). No differences were observed for the incorporation of arachidonic acid after pretreatment with quinacrine. Furthermore, quinacrine had no effect on arachidonate incorporation in an uptake study at 37°C in the continued presence of quinacrine. All of these effects were also reproducible in TCGF-dependent cells prepared from a different donor. DISCUSSION In this evaluation of the effects of potential PAP and lipoxygenase inhibitors on NK activity we found that (1) several of the putative PA2 inhibitors were not affecting lymphocyte arachidonate release although one (BPB) was, and (2) the inhibitors of the lipoxygenase pathway did inhibit NK activity as might be expected if PA2 activity were involved in NK. While quinacrine and glucocorticoids can inhibit NK, they do not appear to have any effect on the arachidonate assayfor lymphocyte PA2 activity, and thus cannot be considered PAZ inhibitors in this lymphocyte system. We also found that the toxicity associatedwith Rosenthal’s inhibitor made this an undesirable reagent for our studies. The first evidence we found that associatedNK cytolysis with a dependence on PA2 activity was that BPB, which reacts with histidine in the active site of pancreatic PAI and irreversibly inactivates it (7), also inhibited lymphocyte arachidonate metabolism at concentrations that abrogated NK activity. The second evidence for a requirement of PA2 activity for NK was that the lipoxygenase inhibitors ETYA, NDGA, and hydroxyphenylretinamide also inhibited NK, suggesting that further metabolism of arachidonate released by PAI might be needed for NK. It is possible to provide an explanation for the effects of quinacrine in the NK system. The concentration of quinacrine that affected human NK also inhibited mouse lymphocyte NK (I), and in both systems effector-target conjugation was not affected. However, we observed that pretreatment of effector cells with quinacrine at TABLE 2 [‘%]Arachidonic Acid Incorporation of Lymphocytes after Pretreatment with BPB or Quinacrine” Percentage of total radioactive incorporation into lymphocytes’ Phospholipid’

Control

BPB

Quinacrine

Phosphatidylcholine Phosphatidylethanolamine

33.5 Yk3.5 4.8 rf: 0.3

1.6 f 2.5c*d 1.5 f 0.1’

44.3 k 2.8 5.9 zk 0

BFollowing pretreatment as described in the text, 0.2 j&i of [ I-‘%J]arachidonate was added to 2 X 106 TCGFdependent cells in 0.5 ml assay medium; cells were incubated for 30 min at 37”C, washed twice, and extracted for phospholipids. Total radioactivity incorporated was 1680, 5730, and 5470 cpm for BPBtreated quinacrine-treated, and control cells, respectively. * Assigned relative to phospholipid standards. ’ Means of two experimental samples + standard deviations. dP < 0.01.

'P < 0.005.

PHOSPHOLIPASE A2 AND ARACHIDONATE

METABOLISM

281

37°C but not at room temperature, reduced subsequent NK activity. This was unexpected since quinacrine can reduce the rate of PA2 hydrolysis in liposomes (8) but does not bind irreversibly. Even more disturbing was the fact that quinacrine had no effect on [14C]arachidonate incorporation into TCGF-dependent cells, either by pretreatment at 37°C or when present during the am&donate uptake assay.Furthermore, NK was affected by concentrations of quinacrine (IDS0 = 5 pLM) which were well below the concentrations of quinacrine which slowed liposomal PA2 activity (approximately 1 rniV). One explanation is that the amine quinacrine was actually accumulating in the lysosomes and raising intralysosomal pH, which would interfere with normal lysosomal functions (27). In fact, a recent study demonstrated that several different lysosomotropic reagentsdepressedNK activity (28). Accumulation of amines within the lysosomes is a temperature-dependent function which is consistent with our temperature-dependent effect of quinacrine pretreatment of NK effecters. Quinacrine is also fluorescent. We observed that lymphocytes were bright with spots of fluorescence after treatment with quinacrine at 37°C but not after treatment at room temperature. Therefore, our conclusion is that while quinacrine does abrogate NK activity, this inhibition is probably due to interference with the functions of low pH granules and not due to a direct action on PAZ. It is more difficult to explain the results we observed for glucocorticoid inhibition of NK which are completely inconsistent with an effect of these hormones on lymphocyte PA2 activity. Glucocorticoid concentrations required to elicit biological responsesfrom cells and/or tissues usually occur in the lo-’ to lo-’ M range (23, 24). The relatively high concentrations needed to inhibit NK activity (i.e., 10m4M for both hydrocortisone and dexamethasone) suggesta non-receptor-induced mechanism (29). Moreover, both the absenceof a lag phase and the lack of a need for new protein synthesis for hydrocortisone suppression of NK are consistent with a non-receptormediated action of NK inhibition. In addition, [‘4C]arachidonate incorporation into TCGF-dependent cells was not assessable,due to the fact that these TCGF-dependent lymphocytes were refractory to NK inhibition by hydrocortisone, in contrast to the susceptibility of freshly isolated lymphocyte NK to low4 M hydrocortisone. At this time, we have no satisfactory explanation for glucocorticoid inhibition of NK activity. Thus, of all these reagents which might affect lymphocyte PA2 activity, only BPB did according to the criteria used. We also tested the effects of mastoparan, a peptide PA2 potentiator which enhances rat mast cell and human fibroblast PA2 activity (30), in the human NK system. While this reagent did enhance NK activity while present during the NK assay (data not shown), this peptide was also toxic to both effector and target cells at concentrations at or just above the 2-4 Kg/ml that enhanced NK activity. Together, the BPB and mastoparan data substantiate but do not prove a direct role for PA2 in NK activity. Along similar lines, the inhibition of NK by the regulatory protein lipomodulin (3 1) provides additional support, although purified lipomodulin could also affect other phospholipases (32). A strongly supportive casefor arachidonate metabolite involvement can be deduced from the results of the lipoxygenase inhibitors on NK activity. The relatively nonspecific lipoxygenase inhibitors, ETYA and NDGA, reduced NK activity in the range of their effective doses employed in other biological systems (12, 14). In independent experiments which were concurrent with ours, Seaman used the lipoxygenase inhibitors ETYA and BW755C (33, 34) to inhibit human NK and obtained inhibition of NK without any effectson conjugate formation. Our data with hydroxyphenylretinamide,

282

CARINE AND HUDIG

which preferentially affects the 5-lipoxygenase pathway at concentrations which also inhibited NK, provide evidence that it is a 5-lipoxygenase product which may be necessaryfor NK activity. Such products include 5-HETE and the leukotrienes LT&, LTB4, LTC4, LTD,, and LTE4. If the loss of NK activity by BPB treatment is due to a loss in supply of free arachidonate, then addition of exogenous arachidonate in the assaymight replenish NK ability. Preliminary experiments to explore this possibility were unsuccessful. One 5-lipoxygenase product, LTB4, is a strong candidate for involvement, since it can act as a calcium ionophore (35, 36) and can induce degranulation of rabbit polymorphonuclear enzymes (37). A cation-dependent release of specific granules in the extracellular spacehas been proposed as part of the mechanism of cytolysis from studies of electron micrographs of natural cytolysis (38). However, attempts to supply LTB4 to NK assays and overcome the effects of lipoxygenase inhibitors have not been successful (34). Exogenous LTB4, however, did augment human natural cytotoxicity to herpes simplex-infected target cells (39) in experiments in which the effecters were not compromised by PA2 inhibitors, thus suggestingthat suboptimal release in some systems may be augmented by LTB4. There is a potentially inconsistent observation in our data: BPB blocked effector,target conjugate formation, whereas the lipoxygenase inhibitors did not. Although we would like to propose that NK “receptor” mobilization might be necessary for binding and also dependent upon PAZ-induced lysolecithin formation, there is an alternate explanation which weakens our interpretation. This explanation is that BPB is reacting with cell surface thiol groups which are required for this binding (26), and it thus has two effectson the lymphocytes (one on PA2 and the second on sulfhydryls). Although BPB pretreatment did not reduce the subsequent ability of lymphocytes to react with the fluorescent thiol reagent fluorescein iodoacetamide, and thus did not support this reservation, selective reactivity of BPB remains a possibility. These reservations make reliance on composite evidence based on experiments with BPB, lipomodulin (3 1), and the lipoxygenase inhibitors, essential for support of the premise that PA2 activity is one of the events associated with NK. We propose that PA2 activity followed by the generation of a natural calcium ionophore may be the mechanism required for release of the NK “lytic substance.” This PA2 could also augment granule release by generation of localized lysolecithins as fusogens.Ultimately, isolation and characterization of both the lipoxygenase products and the final lytic substance from a homogeneous or cloned NK population will be necessaryto assign function to an NK-associated PA2 activity. At this time, these experiments are technically restricted becausein order to detect a “cytolytic-specific” change in arachidonate metabolism, at least lo6 cloned killer cells would be required to incorporate lessthan lo4 cpm of arachidonate. Unfortunately, the cost of producing tens of millions of cloned human NK cells neededfor such an investigation is presently prohibitive. ACKNOWLEDGMENTS The authors thank Dr. Edward Dennis and Mr. Ray Deems for technical advice and discussions.The excellent technical assistanceof Ms. Lory Minning is much appreciated. We also thank Dr. Carol MacLeod, Dr. John Mendelsohn, Dr. Doug Redelman, Dr. Ajit Varki, and Dr. Paul Patek for critical reviews of the manuscript, Mrs. Mary Donellan and Ms. Cenobia Abachiche for the manuscript preparation, and Ms. Phyllis Stookey for illustration.

PHOSPHOLIPASE Az AND ARACHIDONATE

METABOLISM

283

REFERENCES 1. 2. 3. 4.

Quan, P. C., Ishizaka, T., and Bloom, B. R., J. Irnmunol. 128, 1786, 1982. Carpen, O., Virtanen, I., and Saksela, E., J. Immunol. 128, 2691, 1982. Roder, J. C., Karre, K., and Kiessling, R., Prog. Allergy 28, 66, 1981. Hoffman, T., Hirata, F., Bougnoux, P., Fraser, B. A., Goldfarb, R. H., Herberman, R. B., and Axelrod, J., Proc. Natl. Acad. Sci. USA 78, 3839, 1981. 5. Rittenhouse-Simmons, S., J. Biol. Chem. 255, 2259, 1980. 6. Bell, R. L., and Majerus, P. W., J. Biol. Chem. 255, 1790, 1980. 7. Volwerk, J. J., Pieterson, W. A., and de Haas, G. H., Biochemistry 13, 1446, 1974. 8. Vigo, C., Lewis, G. P., and Piper, P. J., Biochem. Pharmacol. 29, 623, 1980. 9. Chan, A. C., Pritchard, E. T., Gerrard, J. M., Man, R. Y. K., and Choy, P. C. B&him. Biophys. Acta 713, 170, 1982. 10. Hong, S.-C., and Levine, L., Proc. Natl. Acad. Sci. USA 73, 1730, 1976. 11. Flower, R., Adv. Prostaglandin Thromboxane Res. 3, 105, 1978. 12. Sullivan, T. J., and Parker, C. W., J. Immunol. 122, 43 1, 1979. 13. Ahem, D. G., and Downing, D., B&him. Biophys. Acta 210,456, 1970. 14. Tappel, A. L., Lundberg, W. O., and Boyer, P. D., Arch. B&hem. Biophys. 42,293, 1953. 15. Sircar, J. C., Schwender, C. F., and Johnson, E. A., Prostaglandins 25, 393, 1983. 16. Fiedler-Nagy, C., Hamilton, J. G., Batula-Bemardo, C., and Coffey, J. W., Fed. Proc. 42, 9 19, 1983. 17. Hudig, D., Djobadze, M. D., Redelman, D., and Mendelsohn, J., Cancer Res. 41, 2803, 1981. 18. Lozzio, C. B., and Lozzio, B. B., Blood 45, 321, 1975. 19. Ryan, T. A., Jr., Jouiner, B. L., and Ryan, B. F., “Minitab Student Handbook.” Duxbury Press,North Scituate, Mass., 1976. 20. Peterkofsky, B., and Tomkins, G. M., Proc. Natl. Acad. Sci. USA 60, 222, 1968. 21. Thompson, E. B., and Lippman, M. E., Metab., Clin. Exp. 23, 159, 1974. 22. Flower, R. J., and Blackwell, G. J., Nature (London) 278, 456, 1979. 23. Daeron, M., Sterk, A. R., Hirata, F., and Ishizaka, T., J. Immunol. 129, 1212, 1982. 24. Hirata, F., Schiffman, E., Venkatagusramanian, K., Salomon, D., and Axelrod, J., Proc. Natl. Acad. Sci. USA 77, 2533, 1980.

25. Danon, A., and Assouline, G., Nature (London) 273, 553, 1978. 26. Hudig, D., Redelman, D., and Minning, L., In “NK Cells and Other Natural Effector Cells” (R. B. Herberman, Ed.), pp. 1021-1026. Academic Press, New York, 1982. 27. Seglen, P. O., and Gordon, P. B., Mol. Pharmacol. 18, 468, 1980. 28. Verhoef, J., and Shanna, S. D., J Immunol. 131, 125, 1983. 29. Patek, P. Q., Collins, J. L., and Cohn, M., Cell Immunol. 72, 113, 1982. 30. Argiolas, A., and P&no, J. J., Fed. Proc. 42, 423, 1983. 31. Hattori, T., Hirata, F., Hoffman, T., Hizuta, A., and Herberman, R. B., J Immunol. 131, 662, 1983. 32. Hirata, F., J. Biol. Chem. 256, 7730, 1981. 33. Seaman, W. E., Fed. Proc. 42, 1379, 1983. 34. Seaman, W. E., J. Immunol. 131, 2953, 1983. 35. Samuelsson, B., Science 220, 568, 1983. 36. Serhan, C. H., Fridovich, J., Goetzl, E. J., Dunham, P. B., and Weissmann, G., J, Biol. Chem. 257, 4746, 1982. 37. Showell, H. J., Ottemess, I. G., Marfat, A., and Corey, E. J., Biochim. Biophys. Acta 106, 741, 1982. 38. Frey, T., Petty, H. R., and McConnell, H. M., Proc. Natl. Acad. Sci. USA 79, 5317, 1982. 39. Rola-Pleszczynski, M., Gagnon, L., and Sirois, P., Biochem. Biophys. Res. Commun. 113, 531, 1983.