Activation of NADPH oxidase by docosahexaenoic acid

Activation of NADPH Oxidase by Docosahexaenoic AcidHydroperoxide and Its Inhibition by a Novel RetinalPigment Epithelial Protein

Guey-Shuang Wu and Narsing A. Rao

PURPOSE. In an earlier study, a novel retinal pigment epithelial protective protein (RPP) wasdescribed, which suppresses the superoxide generation of activated polymorphonuclear leuko-cytes (PMNs). In experimental autoimmune uveitis, docosahexaenoic acid hydroperoxide (22:6OOH) has been shown to be the major lipid peroxidation product in photoreceptors. Thishydroperoxide was also found to be chemotactic to PMNs. This study was undertaken to evaluatethe activation capability of 22:6OOH in resting PMNs and the possible inhibition of this activationby RPP.

METHODS. The 22:6OOH was obtained from pure 22:6 and was purified by thin-layer and high-performance liquid chromatography. Intact rabbit peritoneal PMNs (7-8 X 105) were coincubatedwith 0.5 JWM formyl-methionyl-leucyl-phenylalanine (fMLP), 1.3 /xM 22:6OOH, or 5.0 ixM 22:6.These systems were coincubated with and without 0.25 /xg/ml RPP. From PMN cell-free prepara-tions, the reconstitutes each containing 21 /ag plasma membranes and 276 jug cytosolic factorswere coincubated with arachidonate, 22:6OOH, or 22:6, each at 100 /xM. The inhibition ofsuperoxide production was estimated by adding 0.20 jLtg/ml RPP. Superoxide generation wasmeasured by superoxide dismutase-inhibitable cytochrome C reduction.

RESULTS. In 30 minutes, 22:6OOH-activated PMNs produced 11.10 ± 0.68 nanomoles superoxide,and production was suppressed 72% by RPP. Under the same conditions, fMLP induced productionof 34.6 ± 2.77 nanomoles superoxide, and RPP inhibited 60% of production. In the PMN cell-freesystems, 100 fxM 22:6OOH induced 74.7 nanomoles superoxide per milligram plasma membraneproteins per 5 minutes, and RPP suppressed 50% of production. These results were comparablewith those generated by arachidonate, a known stimulator for this system. RPP was effective onlywhen it was added before assembly of reduced nicotinamide adenine dinucleotide phosphate(NADPH) oxidase.

CONCLUSIONS. The inflammation-mediated retinal peroxidation product 22:6OOH significantly acti-vates resting PMNs, either in intact cells or in cell-free preparations, to increase further the releaseof superoxide from PMNs, thus accelerating inflammation-mediated tissue damage. This profoundamplification process seems to be effectively downregulated by an RPE-generated protein RPP.(Invest Ophthalmol Vis Sci. 1999;40:831-839)

Photoreceptor membranes require certain polyunsatu-rated fatty acids (PUFAs) to provide the precise micro-environment necessary for the proper function of the

visual pigment rhodopsin. These functions depend not only onthe fluidity of the membrane lipids, but on the length of fattyacid chains in phospholipids as well.1 The presence of nearly50% docosahexaenoic acid (22:6, omega 3) in photorecep-tors,2'3 which may be a reflection of these internal regulations,is a unique feature found in no other organ, except cerebralgray matter.4

From the Doheny Eye Institute and the Department of Ophthal-mology, University of Southern California, School of Medicine, LosAngeles.

Supported in part by Grants EY10212 and EYO3O4O from theNational Institutes of Health, Bethesda, Maryland.

Submitted for publication October 2, 1998; accepted December 7,1998.

Proprietary interest category: N.Reprint requests: Narsing A. Rao, Doheny Eye Institute, 1450 San

Pablo Street, DVRC Rm 211, Los Angeles, CA 90033-1088.

The unusual susceptibility of photoreceptor membranesto lipid peroxidation has been attributed to the presence of ahigh concentration of PUFAs, especially of 22:6.5r> The ar-rangement of photoreceptor membrane disks in unusual prox-imity also seems to favor peroxidation chain propagation steps,such as abstraction of hydrogen atoms from the neighboringmolecules. For example, the rat rod outer segment, with di-mensions of 1.5 jam to 2.0 /xm diameter and 20 /xm to 40 /xmlength, contains approximately 600 to 1000 bilayer discs.7

We have used an experimental autoimmune uveitis modelto demonstrate the occurrence of membrane lipid peroxida-tion and the accumulation of 22:6 hydroperoxide (22:6OOH)in the retina.8"10 The lipid peroxidation and hydroperoxida-tion of PUFAs, especially of 22:6, are presumably initiated bythe reactive oxygen and nitrogen metabolites released by theinfiltrating phagocytes, including polymorphonuclear leuko-cytes (PMNs) and macrophages. The peroxidation of retinalmembranes was demonstrated by measuring lipid peroxidationparameters8'9 and by isolating 22:6OOH from the inflamedretina. The identity of 22:6OOH was confirmed by gas chro-

Investigative Ophthalmology & Visual Science, April 1999, Vol. 40, No. 5Copyright © Association for Research in Vision and Ophthalmology 831

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933211/ on 04/07/2018

832 Wu and Rao IOVS, April 1999, Vol. 40, No. 5

matography-mass spectrometry.10 These oxidized retinal lip-ids, especially 22:6OOH, were subsequently found to be che-motactic to neutrophils and may therefore be a factor inamplification and perpetuation of inflammation.''

Regarding superoxide production by fatty acid-activatedneutrophils, only arachidonic acid (20:4), one of the lipidmediators generated early in the signal transduction cas-cade,1213 has been studied extensively. In intact PMNs, methylesters or sodium salts of 20:4 and hydroxy 20:4 were found tobe effective stimulators; and, in PMN cell-free preparations,the sodium salt of 20:4 was found to be effective. More re-cently, it was shown that the czs-polyunsaturated fatty acids(a's-PUFAs) display varying degrees of stimulatory effect, butneither frarcs-unsaturated nor saturated fatty acids is effectivefor this purpose. The stimulatory capability of czs-PUFAs seemsto correlate with their chain lengths, with C20 and C22 fattyacids being the most effective.121415 Regarding the stimula-tory effect of 22:6, investigators in one study demonstrated thestimulatory capability of 22:6 alone and in synergism withyV-formyl-methionyl-leucyl-phenylalanine (fMLP) or phorbolmyristate acetate.15 Because this 22:6-induced activation wassensitive to the calmodulin antagonist W-7, it was concludedthat the calmodulin-dependent process was involved in theresponse to 22:6.15 Reports of neutrophil activation by oxi-dized fatty acid derivatives such as hydroperoxides are scarce.There is only one report in which linoleic hydroperoxide (0.5to 1.0 jLtM) are shown to enhance superoxide production ineither phorbol myristate acetate- or fMLP-stimulated neutro-phils. Linoleic acid hydroperoxide, however, does not affectmembrane potential, lysozyme release, membrane fluidity, orF-actin polymerization.l6 The stimulator)' characteristics of 22:6OOH in particular have yet to be described.

Using cultured retinal pigment epithelial (RPE) cells, wehave isolated an RPE-protective protein (RPP), which inhibits70% to 80% of the superoxide generated by the stimulatedPMNs.17 This protein is secreted specifically by cultured RPEcells and noncultivated intact RPE, but not by fibroblasts,cornea] epithelial cells, or intact choroidal tissues. It has alsobeen shown that this factor does not scavenge superoxide;rather, it acts directly on the PMNs, intervening in their acti-vation responses. This inhibitory activity is concentration de-pendent and requires endogenous protein synthesis withinRPE cells. The RPP was subsequently purified and sequenced.The primary amino acid sequences indicated the RPP to benovel: No identical sequence homology was found in thedatabase.18

In this study, we evaluated the role of 22:6OOH as a lipidmediator in inflammation by assessing the possible stimulatorycapability of this hydroperoxide in superoxide generation frominflammatory infiltrates. After the establishment of 22:6OOHactivation, the suppressive effect of RPP on superoxide gener-ation in this system was also evaluated. The incorporation ofRPP in this system serves to delineate the inhibitory activity ofthis novel protein and to aid in understanding the mechanismof reduced nicotinamide adenine dinucleotide phosphate(NADPH) oxidase activation by 22:6OOH. The stimulatory ca-pacity of 22:6OOH and the inhibitory effect of RPP wereassessed using intact PMNs and PMN cell-free preparations.NADPH oxidase, which catalyzes superO/Xide generation, canalso be activated in a PMN cell-free system consisting of frac-tionated plasma membranes and cytosolic factors in the pres-ence of 20:4 or sodium dodecyl sulfate (SDS).13 Activation

using the whole-cell system does not always permit a distinc-tion between an effect on the membrane receptor and signaltransduction and a direct effect on the components of theNADPH oxidase complex. The cell-free system circumventsthis difficulty by allowing investigation of the direct effect onthe components and the assembly of NADPH oxidase.

Superoxide and nitric oxide are known to be produced ininflammation.19 Although chemical reactivity of both theseentities is low, their facile combination to yield the potentoxidizing/nitrating agent peroxynitrite is known to cause con-siderable damage to photoreceptors in experimental autoium-mune uveitis.20

MATERIALS AND METHODS

Preparation, Purification, and Quantitationof 22:6OOH

22:6OOH was prepared according to a previously describedmethod.1 x Briefly, an aliquot of pure methyl docosahexaenoate(>99% pure; Nu-Chek, Elysian, MN) as a thin film coating asmall flask, was air oxidized at room temperature until a con-version of 25% to 30% was achieved. This extent of conversionwas found to yield a maximum amount of hydroperoxidewithout introducing side products, thus simplifying the subse-quent purification of hydroperoxide. The quantity of 22:6OOHin the product mixture was determined by UV detection ofconjugated dienes at 233 nm, using a molar extinction coeffi-cient of 25,200 for calculation.9 Pure 22:6OOH was isolated bythin-layer chromatography (TLC).10 In a 5 X 20-cm, 0.25-mmthick, precoated silica gel 60 plate (Analtech, Newark, DE), theoxidized 22:6 dissolved in acetone was spotted across theentire plate (20 spots; approximately 1 mg/spot) and devel-oped by a solvent system consisting of petroleum ether/diethylether/acetic acid (70:30:1). The analytical TLC used to indicatethe position of spots was visualized by dipping the plate in asolution of 3% cupric acetate in 8.5% phosphoric acid andcharring at l40°C (Fig. 1). The area containing the 22:6OOHspots (Fig. IB; relative mobility = 0.28 and 0.33) in the pre-parative plate (without visualization treatment) was scrapedoff, and 22:6OOH was extracted from the silica gel, initiallywith methanol and then with diethyl ether to remove theresidual silica gel. The chromatography was repeated severaltimes until the desired quantity of 22:6OOH was obtained. Thepurified 22:6OOH was rechecked for its purity by UV absorp-tion and TLC and used immediately in the assays.

To prepare 22:6OOH sodium salt, 22:6 (free acid; Nu-Chek) was air oxidized, and the hydroperoxide was quanti-tated by UV absorption, as described for methyl esters. Analiquot of the oxidation product was methylated and analyzedto check for any byproduct formed during oxidation.10 The22:6OOH (free acid) was purified by high-performance liquidchromatography (HPLC) using a C18 re versed-phase column,and hydroperoxides were eluted with methyl formate.21 Themethyl formate fraction was evaporated under nitrogen, andthe sodium salt of 22:6OOH was prepared by dissolving theresidue in an equimolar quantity of sodium hydroxide. The22:6OOH sodium salt was also prepared directly from pure22:6 sodium salt (Sigma, St Louis, MO). With hydroperoxideyield of approximately 28%, the unpurified hydroperoxide pro-duced results similar to those of purified hydroperoxide in the


IOVS, April 1999, Vol. 40, No. 5 Activation of NADPH Oxidase by Hydroperoxide 833

1 2FIGURE 1. Analysis by TLC of oxidized 22:6. Pure 22:6 was air oxi-dized to yield 28% hydroperoxide. The total oxidized mixture wasspotted as methyl esters, and the plate was eluted with petroleumether/diethyl ether/acetic acid (70:30:1). Lane 1: pure 22:6; lane 2:total oxidized product from 22:6. (A) Unchanged 22:6; (B) 22:6 hy-droperoxides. Note that the positional isomers of 22:6 hydroperoxidesare separated as two spots, and no other by-product is seen in themixture. The analyses were repeated four times; a representativechromatogram is shown.

activation assays, provided that the hydroperoxide content inthe product mixture was corrected.

Isolation and Purification of RPPfor the Activation AssaysAll animals used in the study were maintained and treated inaccordance with the ARVO Resolution for the Use of Animalsin Ophthalmic and Vision Research. The establishment andcharacterization of rabbit RPE culture has been published else-where.17 The collection and purification of RPP have also beenpublished.1718 Briefly, the confluent RPE cells in 25-cm2 cul-ture flasks were washed several times with Hanks' balancedsalt solution (HBSS) to assure the absence of tissue cultureserum before incubating with 2 ml HBSS for 4 hours to collectsecreted proteins from RPE cells. The total secreted proteinswere quantitated and used as is or were concentrated byultrafiltration through a YM-10 membrane (Amicon, Lexington,MA) before use. In some experiments, RPP was also purified byanion-exchange chromatography (diethylaminoethyl Sepha-

rose CL-6B)18 before usage. The center-cut fractions from thispurification provided at least 90% pure RPP (as indicated by thereversed-phase HPLC) without losing a significant amount ofactivity.18 These RPP preparations were either used immedi-ately or stored at — 70°C and used within 1 week. Either nativeor purified RPP provided comparative results for superoxideinhibition per microgram RPP.

Neutrophil Superoxide Production Stimulatedby Docosahexaenoic Acid Hydroperoxide

Rabbit peritoneal PMNs were collected according to the pub-lished procedure.22 The purity of PMNs, determined byWright-Giemsa staining, was more than 95%, and there was nored blood cell contamination. The viability of PMNs, deter-mined by trypan blue exclusion, was generally greater than98% of total counts. The cells were used immediately aftercollection.

Generation of superoxide was determined by the super-oxide dismutase (SOD)-inhibitable reduction of cytochromeC.23 A typical assay procedure was as follows: Two identicaltubes of PMN suspensions (105-106 cells) in HBSS, one con-taining 10 jid SOD (3 mg/ml; Sigma) and the other 10 /ml water,were incubated for 2 minutes at 37°C before die addition of 50jal cytochrome C (30 mg/ml; Sigma) and fMLP, 22:6HP, or 22:6.The mixture was then incubated for the desired length of time.The final concentrations of cytochrome C, fMLP, 22:6OOH,and 22:6 in a total incubation mixture of 1.5 ml were 80 JU-M,0.5 fxM, 1.3 fxM, and 5.0 ju,M, respectively. The reduced cyto-chrome C was measured with a double-beam spectrophotom-eter (model UV-160; Shimadzu, Kyoto, Japan), using the SOD-containing sample as the reference. The amount of superoxideproduced was calculated using the molar extinction coefficientof 21,000/M/cm.23 In the experiments with 10 /xM trifluoper-azine (TFP),24 20 JUM chlorpromazine (CP)24 and 0.25 /u-g/mlRPP,17 the inhibitors were added simultaneously with the cho-sen stimulus. Methyl esters and sodium salts were both used forthe stimulation of intact cells. Sodium salts were added to HBSSdirectly. The hydroperoxy methyl ester (6.1 mg) was dissolvedin 1.75 ml dimethyl sulfoxide, and 75 JU,1 of this solution wasadded to the incubating mixture until a final concentration of1.3 ju,M was achieved.

Generation of Superoxide in a Cell-Free SystemStimulated by 22:6OOH

To prepare the cell-free system from freshly harvested rabbitperitoneal PMNs, 1 X 108 cells/ml suspended in buffer A (138mM NaCl, 2.7 mM KC1, 16.2 raM Na2HPO4) 1.47 mM KH2PO4,2.0 mM MgCl2, 1.25 mM EGTA, and 1.0 mM phenylmethylsul-fonyl fluoride) were disrupted by sonication using a sonicdismembrator (model 250; Branson, Danbury, CT) in a melting-ice bath with a discontinuous 10-second burst and 60% power.Unbroken cells and debris were removed by centrifugation at800,g for 15 minutes at 4°C. Aliquots of supernatant (0.9 ml)were then layered on a discontinuous sucrose gradient consist-ing of 1.5 ml 15% (wt/vol) sucrose and 1.5 ml 40% (wt/vol)sucrose. After ultracentrifugation in a swinging-bucket rotor(model SW 60Ti; Beckman, Fullerton, CA) at lOO.OOOg for 60minutes, the cytosolic cofactors were in the top layer, and thelight membranes were concentrated at the interface of the twosucrose layers. In some preparations, the light membraneswere collected slightly lower than the 15%/40% sucrose inter-



c"EoCO

"oEc

3T3O

0)

'I(1)Q.3

C/D

401

35"

30-

25"

20-

15-

ID-

None fMLP fMLP + RPP 22:6 OOH 22:6OOH + RPP 22:6

FIGURE 2. Activation of intact rabbit peritoneal PMNs by 22:6OOH and its inhibition by RPP. Polymor-phonuclear leukocytes (7.0 X 105 cells) were activated for 30 minutes with 0.5 jaM fMLP, 1.3 /xM22:6OOH, or 5.0 \iM 22:6, and the fMLP- and 22:6OOH-induced activations were inhibited by 0.25 jig/mleach of RPP. The superoxide production was measured by the SOD-inhibitable reduction of cytochromeC. The data are expressed as the mean ± SD of three experiments.

face, and these membranes, as reported by others,25 providedsimilar results in our system. There was a pellet visible at thebottom of the 40% sucrose layer; presumably consisting ofheavier membranes and other cellular components; these werenot used for reconstitution. Both preparations yielded similarresults. Cytosolic fractions were recentrifuged at 150,000^ for30 minutes to remove residual membranes. Protein aggregatesformed during storage at — 70°C were also removed by centrif-ugation before use.

The superoxide assays in the cell-free system were per-formed in a two-step procedure, permitting at least partialseparation of the activation and catalytic stage of NADPHoxidase. In a typical assay, 295 jug cytosolic factors (cell equiv-alent of 6 X 106 cells) and 21 \x% membrane fraction (cellequivalent of 6 X 106 cells) in 200 y\ buffer A were incubatedat 27°C for 2 minutes and then activated at 27°C for 4 minuteswith 100 JUM 20:4, 22:6OOH, or 22:6 in the presence of 2 mMMgCl2 and 100 \xM cytochrome C. The assay mixture was thentransferred into the cuvettes, and superoxide production wasinitiated by adding 200 pM. NADPH (Sigma). The superoxideproduced was measured by SOD-inhibitable cytochrome Creduction, the same method used in the whole-cell system. Thecuvette containing 45 ju,g/ml SOD was used as the referencecell. Sodium salt of fatty acids was used for all determinations.SDS and 20:4 were used as positive controls. The incubationtemperature of 27°C was chosen because the activity ofNADPH oxidase has been found to decline at 37°C.26

RESULTS

22:6OOH Preparation

With conversion of 25% to 28%, the hydroperoxide productwas found to contain no detectable quantity of other oxidation

product in the preparation (Fig. 1). The positional isomers of22:6OOH were separated as two spots (Rf = 0.28 and 0.33; Fig.IB).10 The subsequent redetermination by UV absorption andTLC indicated the 22:6OOH to be stable through these purifi-cation procedures and storage at — 70°C under nitrogen atmo-sphere. Commercial 22:6 was found to be pure by TLC; only22:6 methyl ester was detected (Fig. 1A).

Stimulation of Superoxide Release from IntactPolymorphonuclear Leukocytes

22:6OOH (1.3 \xM) was found to stimulate rabbit peritonealPMNs (7.0 X 105) to produce a significant quantity of super-oxide (11.1 ± 0.68 nanomoles/30 minutes; Fig. 2). Under theseconditions, 0.5 /xM fMLP generated 34.6 ± 2.77 nanomolessuperoxide/30 minutes. Pure 22:6 stimulated superoxide pro-duction to a much smaller extent; the superoxide generated,0.95 ± 0.08 nanomoles/30 minutes, amounted to only 9% ofthat generated by 22:6OOH (Fig. 2). Coincubation with 0.25jag/ml RPP suppressed 72% of 22:6OOH-induced superoxidegeneration and 60% of that induced by fMLP (Fig. 2). Withoutadded stimulus, superoxide generation was 0.6 nanomoles/30minutes. In the 22:6OOH-stimulation, a higher concentrationof 55 /xM produced only a minor increase (28%) in superoxideproduction at 30 minutes, and 7.3 JLIM produced no increase in10 minutes and a decrease after a 30-minute incubation (Fig. 3).

In the fMLP-stimulation, the calmodulin antagonists, either10 JULM TFP or 20 /xM CP totally inhibited PMN superoxidegeneration. In 22:600H-stimulation, a similar total inhibitionwas seen with 20 jaM CP treatment. The effect of TFP, how-ever, was less, with inhibition of only 43% (Fig. 4). Increasedconcentrations of TFP (up to 50 /xM) in 22:6OOH stimulationresulted in no apparent increase in inhibition (data notshown).27



Concentration

FIGURE 3. Concentration-dependent activation of intact rabbit peritoneal PMNs by 22:6OOH. Polymor-phonuclear leukocytes (7.5 X 105 cells) were activated by 0, 1.3, 5.5, and 7.3 /xM of 22:6OOH, and thesuperoxide production was measured at 10- and 30-minute intervals by SOD-inhibitable cytochrome Creduction. The results are expressed as the mean ± SD of three determinations.

Stimulation of Superoxide Release in Cell-FreePreparations

To confirm the effect of 22:6OOH in the assembly of NADPH

cell-free preparations. The cell-free system was obtained fromresting rabbit peritoneal PMNs and fractionated into lightplasma membranes and cytosols. In a typical run, beginningwith 1 X 108 cells, total protein levels of 300 ixg for light

oxidase directly, the assays were also performed using PMN membranes and 2750 /xg for cytosolic fraction were obtained.

T3oa.a)•o

1<DQ.

CO

40-

35-

30-

25"

20

15

10

fMLP fMLP + TFP fMLP+CP 22:6 OOH 22:6 OOH+ TFP

22:6 OOH+ CP

FIGURE 4. Effect of calmodulin antagonists in the activation of intact rabbit peritoneal PMNs by fMLP and by22:6OOH. Polymorphonuclear leukocytes (7.2 X 105 cells) were activated by 0.5 yM fMLP or 1.3 yM22:6OOH, and these activations were inhibited by 10 yM TFP or 20 yM chlorpromazine. The superoxideproduction was measured by the SOD-inhibitable reduction of cytochrome C. The results are shown asnanomoles of superoxide produced in 30 minutes and are expressed as the mean ± SD of three experiments.



CO

8 10 2 4 6 8

Time (min)

10 0 8 10

FIGURE 5- Activation of PMN cell-free preparations by 20:4, 22:6OOH, or 22:6 and their inhibition byRPP. Plasma membranes (21 /xg protein) and cytosolic factors (276 /ig protein) obtained from the restingrabbit peritoneal PMNs were reconstituted and activated by 100 ixM (A) 20:4, (B) 22:6OOH, or (C) 22:6.Superoxide generation was started by the addition of 200 JUM NADPH. The inhibition was performed by0.20 p,g/ml RPP. Superoxide generation was estimated by the SOD-inhibitable cytochrome C reduction.The entire measurement was repeated three times, and the representative curves are shown.

These quantities were similar to those reported by others.28

These two fractions were reconstituted in the presence ofanionic lipids (anion derived from sodium arachidonate orsodium salt of 22:6OOH) and NADPH for superoxide genera-tion. The amount of superoxide released in the absence ofNADPH was essentially zero. Arachidonate was used as a pos-itive control in this system.29 In our preparations, 100 /xM 20:4caused almost a linear increase in production, reaching 97.4nanomoles superoxide per milligram plasma membrane pro-teins in 10 minutes (Fig. 5). After the initial 10-minute period,the increase in the rate curve was minimal (data not shown).These observed rates and the initial specific activity of NADPHper minute were comparable with those reported by others for2Q-4 25,26,29 stimulation using 100 JLLM SDS27 produced resultssimilar to those obtained with 20:4 (data not shown). Underthe same conditions, 100 juM 22:6OOH produced 86.1 nano-moles superoxide per milligram plasma membrane proteinswith 10 minutes of incubation. The superoxide release wasmuch less with 100 ixM 22:6; at 10 minutes: The release wasonly 25 nanomoles per milligram plasma membrane proteins(Fig. 5). In the 22:6OOH-stimulation, the rate of superoxideproduction was comparable with that of 20:4 in the first 10-minute incubation (Fig. 5). The duration of superoxide pro-duction by 22:6OOH, however, was longer than 20:4, with thenear linear production continuing up to 25 minutes (data notshown). In 22:6OOH stimulation, the addition of 10 /xMguanosine 5'-O-(thiotriphosphate) (GTPyS) before assembly in-creased the superoxide generation from 72.7 ± 7.0 to 145.0 ±17.4 nanomoles per milligram plasma membrane protein.

In the cell-free system, coincubation with 0.25 to 0.30/jig/ml RPP also suppressed 50% of 22:6OOH-stimulated super-oxide generation in a 5-minute incubation. The same concen-

tration of RPP suppressed superoxide production 68% in 20:4and 45% in 22:6 (Fig. 5). This suppression of 22:6OOH stimu-lation by RPP could be reversed, mostly by adding more cyto-solic factors.

In the cell-free preparations, the reconstitution of plasmamembranes with cytosol was performed with two incubationsteps, thus separating the assembly/activation stage from thecatalytic stage of NADPH oxidase. We found that the RPPadded at the beginning of the assembly/activation resulted insignificant suppression, whereas RPP added after the assemblycaused no detectable suppression (Fig. 6). In these two-stageexperiments, TFP, a calmodulin antagonist, was found to sup-press the assembly/activation totally (Fig. 6).

DISCUSSION

In this study, we found that 22:6OOH in concentrations as lowas 1.3 juM is capable of stimulating resting intact PMNs torelease superoxide. The amounts released were lower thanthose produced by the most potent stimulus, fMLP (0.5 i^M),but were several times higher than those previously reportedfor 82 jaM 20:4 stimulation. Retinal pigment epithelial protec-tive protein, which intervenes in the activation processes ofPMNs,17 abolished a major part (70%) of this activation. Usinga PMN cell-free system, we also showed that 22:6OOH acti-vated the reconstitutes of light plasma membranes and cytoso-lic factors to produce superoxide; the extent of this activationwas comparable with that obtained with the well-documentedactivator 20:4 in our system.24 RPP also inhibited this activa-tion in the cell-free system, although to a lesser extent than inintact cells. The calmodulin antagonists TFP and CP24 inhibited22:6OOH activation in intact cells and in cell-free systems.


IOVS, April 1999, Vol. 40, No. 5

100n

Activation of NADPH Oxidase by Hydroperoxide 837

c"E

1CD

8O

0)Q.

CO

80"

Q.

| 601

"5E

40"

20"

None +RPP pre-assembly

+RPP post-assembly

+TFP

FIGURE 6. Retinal pigment epithelial protective protein (RPP) inhibits the assembly but not the activityof NADPH oxidase activated by 22:6OOH. The rabbit PMN cell-free reconstitutes were prepared by theprocedure described in the Methods section. RPP was added before or after the assembly/activation ofNADPH oxidase in the reconstitutes. Trifluoperazine (50 JLIM) was added before assembly. Superoxideproduction was determined by the SOD-inhibitable cytochrome C reduction. Data are expressed asmean ± SD of three separate experiments.

Although it has been known for some time that intactneutrophils can be activated by unoxidized czs-PUFAs, the fattyacid- dependent mechanism of activation has not yet been fullyelucidated. When this phenomenon was first discovered, it wasthought that the structural asymmetry of czs-PUFAs could per-turb the bilayer order and change the polarization, thus causingactivation of components necessary for the respiratory burst.29

Therefore, only the czs-PUFAs are effective, whereas trans-fattyacids and saturated fatty acids are not.29 No further viewpointon the mechanism has been advanced regarding the molecularevents after this initial structural perturbation.

In the receptor-mediated activation of intact neutrophilsby a stimulus, such as fMLP, the important signal transductionevents implicated in recent years include the following: Afterreceptor occupancy, phospholipase C is activated in the mem-branes to release diacylglycerol (DAG) and inositol 1,4,5-Tris-phosphate (IPs)- The IP3 then moves into the cytosol to releasethe calcium ion from its intracellular storage. The increasedcalcium levels, in turn, may activate calmodulin and othercalcium-dependent processes. The protein kinase C (PKQ-mediated phosphorylation of cytosolic factors follows, andphosphorylated cytosolic factors (including p47-phox, p67-phox, Racl p21, and Rac2 p21) translocate to the plasmamembranes and integrate with NADPH oxidase components toform an active superoxide-generating complex.30

Considering the several new findings in recent years, itseems that activation of intact neutrophils by czs-PUFAs, andtherefore by 22:6OOH, can best be rationalized by the coacti-vation of phospholipase D (PLD) and the signal transductionevents that follow. The activation of phospholipase C alone bymost of the physiological stimuli was invariably transient and

temporary because of the rapid metabolism and disappearanceof IP3 and DAG from the signal transduction cascade. Becausethe activation of 22:6OOH was sustained for a relatively longperiod, the involvement of PLD activation would support thisnotion. Coactivation of PLD results in more sustained produc-tion of DAG, occasionally lasting for more than a hour.31 Theprimary product from PLD activation, phosphatidic acid (PA),has been shown to activate several protein kinases32 and trig-ger a respiratory burst in human neutrophils.33 From PA, DAGcan also be generated by the action of phosphatidic acidphosphomono esterase.34 To substantiate further the capabil-ity of 22:6OOH in activating PLD, it was recently shown thatthe fatty acid hydroperoxides activate PLD effectively in a dose-and time-dependent manner to produce phosphatidylethanoland PA.35 These molecular events, therefore, may serve asinitial activation steps for 22:6OOH. Also, the occurrence ofthese events supports our observation that the hydroperoxidesare more effective than cz's-PUFAs in this process.

In this study, CP totally inhibited fMLP- and 22:6OOH-induced activations, whereas TFP totally inhibited fMLP butonly partially inhibited 22:6OOH. Although both TFP and CPare generally classified as anti-calmodulin drugs, their mecha-nisms of action are not entirely the same, largely because of theconsiderable difference in their structures. Aside from inhibit-ing calcium-binding proteins, CP seems to show other lesswell-known effects, such as inhibition of phospholipase A2,thus possibly reducing subsequent release of 20:4 in the earlysignal transduction cascade.36 Chlorpromazine also causednonspecific alteration of the receptor affinity for fMLP bindingand therefore depressed superoxide release.37 Thus, it is con-ceivable that the inhibition of calmodulin alone is sufficient to



cause total suppression in the fMLP + TFP system, but notsufficient to suppress the 22:6OOH + TFP system, suggestingthat there is a second activation pathway in 22:6OOH activa-tion that is calmodulin-independent.

In the PMN cell-free system, 20:4 is known to activate thereconstitutes of plasma membranes and cytosolic fractions; themechanistic pathways for this activation, however, are not thesame as those in intact cells. In this system, the phosphoryla-tion of cytosolic factors in the activation sequence was origi-nally discounted, and the neutralization of these factors by theanionic amphiphiles was thought to be enough to preparethese factors for assembly with cytochrome b558.25 However,more recently with this system, PA was found to activateprotein kinases and induce phosphorylation of a wide array ofendogenous proteins, including p47-phox.38 These findings,therefore, also support the role of hydroperoxide-initiated PLDactivation in this system. In the PMN cell-free system, we(using 22:6OOH) and others (using as-PUFAs)25 have per-formed the activation without exogenous GTPyS. It has beenshown that czs-PUFAs (20:4) alone cause significant transloca-tion and binding of p47-phox and p67-phox and slight trans-location and binding of Rac p21s to the plasma membranes.Additional GTPyS further increases the translocation of Racp21s.39 In our initial 22:6OOH activation without GTPyS, su-peroxide generation was higher than that reported with 20:4plus GTPyS.40 However, addition of GTPyS induced a twofoldincrease in superoxide production. Therefore, it is clear that22:6OOH is an effective stimulator, and this effect is substan-tially enhanced in the presence of GTPyS to facilitate the trans-location and binding of Rac p21 in NADPH oxidase assembly.

In the activation of NADPH oxidase by czs-PUFAs, in eitherintact cells or a cell-free system, the unoxidized fatty acidswere the most frequently evaluated in the past. As we found inthis study, the effect of hydroperoxy fatty acid in the signaltransduction pathways surpass that of the unoxidized PUFA.The fatty acid hydroperoxides have been shown to activate ratbrain PKC more efficiently than the parent unoxidized fattyacids or other secondary oxidation products, such as hydroxyfatty acids and keto fatty acids. Moreover, unsaturated fattyacids have been shown to activate PKC with effective concen-trations between 100 JUM and 300 JLLM, and hydroperoxy fattyacids with only half of these concentrations.41 Because PKC isknown to participate in the signal transduction cascade forsuperoxide generation, these results for PKC39 support ourfindings that the hydroperoxides are effective mediators inthese processes.

In the present study, we vised the 22:6OOH derived fromfree fatty acid 22:6; therefore, it is not a phospholipid hy-droperoxide. In intraocular inflammation, the hydroperoxidesgenerated in the retinas are mostly phospholipid hydroperox-ides; therefore, most of the PUFAs are esterified in the mem-brane phospholipids, and the unesterified pool is extremelysmall.910 Extrapolation of the present data to an in vivo inflam-matory system is possible, however, considering the recentfindings regarding the peroxidation of phospholipids invivo.42'43 It seems that in the event of membrane pertubationby exogenous factors, such as in oxidative modifications, mem-brane phospholipase A2 and C act rapidly to hydrolyze prefer-entially the oxidized phospholipids to release hydroperoxyfatty acids to intracellular space.42'43 Because retinal mem-brane lipids contain 50% 22:6,2'4 and because 22:6 is ex-tremely susceptible to peroxidation,8"10 the hydroperoxides

released from retinal cells in experimental uveitis are likely tocontain high concentrations of 22:6 hydroperoxides.

Previously, we demonstrated that RPP suppresses 60% to70% of superoxide generation in intact neutrophils stimulatedby fMLP or phorbol ester.17 The present study further con-firmed and defined the inhibitory characteristics of this novelprotein by intervening with the 22:6OOH-induced activation inintact cells and the cell-free system. Although the exact natureof this inhibition in the signal transduction pathways has notbeen fully elucidated, RPP has recently been found to inhibitPKC-mediated p47-phox phosphorylation effectively in intactcells.44 In the cell-free system, RPP was found to block super-oxide release effectively during the initial 5-minute activationby 22:6OOH but less effectively during the subsequent stage,suggesting that the propagation step of this activation mayinvolve a different mechanism in this system (Fig. 5).45 It hasalso been found that in the cell-free system, RPP interferes withthe assembly of NADPH oxidase rather than modifying theNADPH oxidase activity after assembly. Aside from affectingthe phosphorylation of p47-phox, RPP may also bind to p47-phox to induce a conformational change in this protein, thuspreventing it from interacting with other molecules. Initialevidence to support this assumption was found in this study inwhich the effect of RPP was alleviated by increased concen-trations of cytosolic factors.

In the past, it was thought that the cytotoxicity of thesuperoxide was exerted through the generation of hydroxylradicals because of the low chemical reactivity of the superox-ide anion.46 However, in recent years, it has been shown thatthe experimental results are more consistently explained bythe cytotoxicity of peroxynitrite, a combination product ofsuperoxide and nitric oxide.47 The production of nitric oxideby neutrophils stimulated by 22:6OOH has recently been dem-onstrated in this laboratory.48 In the present study, we showedthat the 22:6OOH produced by the reactive oxygen metabo-lites of inflammatory cells,10 fed back to the inflammatory sitesto elicit PMNs to produce more superoxide, resulting in in-creased local concentrations of peroxynitrite. The presence ofperoxynitrite in the retina and its consequential peroxidativedamage in experimental autoimmune uveitis has recently beenshown in this laboratory.20 The 22:6OOH-derived pathwaysmay therefore constitute the potential routes leading to theamplification of pathologic retinal degeneration in intraocularinflammation.

References

1. Davoust J, Bienvenue A, Fellmann P, Devaux PF. Boundary lipidsand protein mobility in rhodopsin-phosphatidylcholine vesicles.Effect of lipid phase transitions. Biochim Biophys Acta. 1980;596:28-42.

2. Fliesler SJ, Anderson RE. Chemistry and metabolism of lipids in thevertebrate retina. Prog Lipid Res. 1983;22:79-131.

3. Stone WL, Farnsworth CC, Dratz EA. A reinvestigation of the fattyacid content of bovine, rat and frog retinal rod outer segment. ExpEye Res. 1979;28:387-397.

4. Neuringer M, Connor WE, Lin DS, Barstad L, Luck S. Biochemicaland functional effects of prenatal and postnatal omega 3 fatty aciddeficiency on retina and brain in rhesus monkeys. Proc NatlAcadSd USA. 1986;83:4021-4025.

5. Slater TF. Lipid peroxidation. Biochem Soc Trans. 1982;10:70-71.6. Witting L. lipid peroxidation in vivo. J Am Oil Cbem Soc. 1965;

42:908-913.7. Wiegand RD, Jose JG, Rapp LM, Anderson RE. Free radicals and

damage to ocular tissues. In: Armstrong D, ed. Free Radicals in



Molecular Biology, Aging and Disease. New York: Raven Press;1984:317-353.

8. Rao NA. Role of oxygen free radicals in retinal damage associatedwith experimental uveitis. Trans Am Ophthalmol Soc. 1990;88:797-845.

9. Goto H, Wu GS, Chen F, Kristeva M, Sevanian A, Rao NA. Lipidperoxidation in experimental uveitis: sequential studies. Curr EyeRes. 1992; 11:489-499.

10. Wu GS, Sevanian A, Rao NA. Detection of retinal lipid hydroper-oxides in experimental uveitis. Free RadicBiolMed. 1992;12:19-27.

11. Goto H, Wu GS, Gritz DC, Attalla LR, Rao NA. Chemotactic activityof the peroxidized retinal membrane lipids in experimental auto-immune uveitis. Curr Eye Res. 1991;10:1009-10l4.

12. Badwey, JA, Curnutte JT, Robinson JM, Berde CB, Karnovsky MJ,Karnovsky ML. Effect of free fatty acids on release of superoxideand on change of shape by human neutrophils: reversibility byalbumin. / Biol Chem. 1984;259:7870-7877.

13. Curnutte JT. Activation of human neutrophil nicotinamide-adeninedinucleotide phosphate, reduced (triphosphopyridine nucleotide,reduced) oxidase by arachidonic acid in a cell-free system. / ClinInvest. 1985;75:1740-1743.

14. Hardy SJ, Robinson BS, Ferrante A, et al. Polyenoic very-long-chainfatty acids mobilize intracellular calcium from a thapsigargin-insen-sitive pool in human neutrophils. The relationship between Ca2+

mobilization and superoxide production induced by long- andvery-long-chain fatty acids. BiochemJ. 1995;311:689 -697.

15. Poulos A, Robinson BS, Ferrante A, Harvey DP, Hardy SJ, MurrayAW. Effect of 22-32 carbon n-3 polyunsaturated fatty acids onsuperoxide production in human neutrophils: synergism of doco-sahexaenoic acid with f-met-leu-phe and phorbol ester. Immunol-ogy. 1991;73:102-108.

16. Torkildson, JC, Torres M, Forman HJ, Coates TD. linoleic acidhydroperoxide modulates the production of reactive oxygen spe-cies by human neutrophils. Pediatr Res. 1990;27:151a.

17. Wu GS, Rao NA. A novel retinal pigment epithelial protein sup-presses neutrophil superoxide generation, I: characterization ofthe suppressive factor. Exp Eye Res. 1996;63:713-725.

18. Wu GS, Swiderek KM, Rao NA. A novel retinal pigment epithelialprotein suppresses neutrophil superoxide generation, II: purifica-tion and microsequencing analysis. Exp Eye Res. 1996;63:727-737.

19. Rodenas J, Mitjavila MT, Carbonell T. Simultaneous generation ofnitric oxide and superoxide by inflammatory cells in rats. FreeRadic Biol Med. 1995;18:869-875.

20. Wu GS, Zhang J, Rao NA. Peroxynitrite and oxidative damage inexperimental autoimmune uveitis. Invest Ophthalmol Vis Sci.1997;38:1333-1339.

21. Graff G, Anderson LA, Jaques LW. Preparation and purification ofsoybean lipoxygenase-derived unsaturated hydroperoxy and hy-droxy fatty acids and determination of molar absorptivities ofhydroxy fatty acids. Anal Biochem. 1990;188:38-47.

22. McCarron RM, Goroff DK, Luhr JE, Murphy MA, Herscowitz HB.Methods for the collection of peritoneal and alveolar macro-phages. Methods Enzymol. 1984;108:274-284.

23. Markert M, Andrews PC, Babior BM. Measurement of O2" produc-tion by human neutrophils. The preparation and assay of NADPHoxidase-containing particles from human neutrophils. MethodsEnzymol. 1984;105:358-365.

24. Curnutte JT, Badwey JA, Robinson JM, Karnovsky MJ, KarnovskyML. Studies of the mechanism of superoxide release from humanneutrophils stimulated with arachidonate./Zfro/ Chem. 1984;259:11851-11857.

25. Gabig TG, English D, Akard LP, Schell MJ. Regulation of neutrophilNADPH oxidase activation in a cell-free system by guanine nucle-otides and fluoride. Evidence for participation of a pertussis andcholera toxin-insensitive G protein. J Biol Chem. 1987;262:l685-1690.

26. Tamura M, Takeshita M, Curnutte JT, Uhlinger DJ, Lambeth JD.Stabilization of human neutrophil NADPH oxidase activated in acell-free system by cytosolic proteins and by l-ethyl-3-(3-dimethyl-aminopropyl) carbodiimide. / Biol Chem. 1992;267:7529-7538.

27. Bromberg Y, Pick E. Activation of NADPH-dependent superoxideproduction in a cell-free system by sodium dodecyl sulfate. / BiolChem. 1985;260:13539-13545.

28. Ogata K, Nishimoto N, Uhlinger DJ, Igarashi K, Takeshita M,Tamura M. Spermine suppresses the activation of human neutro-phil NADPH oxidase in cell-free and semi-recombinant systems.BiochemJ. 1996;313:549-554.

29- Steinbeck MJ, Hegg GG, Karnovsky MJ. Arachidonate activation ofthe neutrophil NADPH-oxidase. Synergistic effects of protein phos-phatase inhibitors compared with protein kinase activators. / BiolChem. 1991 ;266:16336-16342.

30. Edward SW. Regulation of neutrophil oxidant production. In: Dun-can CJ, ed. Calcium, Oxygen Radicals and Cellular Damage.New York: Cambridge University Press; 1991:35-76.

31. Nakamura S, Nishizuka Y. Lipid mediators and protein kinase Cactivation for the intracellular signaling network. J Biochem. 1994;115:1029-1034.

32. Limatola C, Schaap D, Moolenaar WH, van Blitterswijk WJ. Phos-phatidic acid activation of protein kinase C-zeta overexpressed inCOS cells. Comparison with other protein kinase C isotypes andother acidic lipids. BiochemJ. 1994;304:1001-1008.

33. McPhail LC, Qualliotine-Mann D, Agwii DE, McCall CE. Phospho-lipases and activation of the NADPH oxidase. Eur J Haematol.1993;51:294-3OO.

34. Nishizuka Y. Intracellular signaling by hydrolysis of phospholipidsand activation of protein kinase C. Science. 1992;258:607-6l4.

35. Natarajan V, Taher MM, Roehm B, et al. Activation of endothelialcell phospholipase D by hydrogen peroxide and fatty acid hy-droperoxide. / Biol Chem. 1993;268:930-937.

36. Bertini R, WangJM, Mengozzi M, et al. Effect of chlorpromazine onPMN-mediated activities in vivo and in vitro. Immunology. 1991;72:138-143.

37. Lohr KM, Feix JB, Kurth C. Chlorpromazine inhibits neutrophilchemotaxis beyond the chemotactic receptor-ligand interactions.J Infect Dis. 1984; 150:643-652.

38. McPhail LC, Qualliotine-Mann D, Waite KA. Cell-free activation ofneutrophil NADPH oxidase by a phosphatidic acid-regulated pro-tein kinase. Proc Natl Acad Sci USA. 1995;92:7931-7935.

39. Sawai T, Asada M, Nunoi H, et al. Combination of arachidonic acidand guanosine 5'-O-(3-thiotriphosphate) induce translocation ofrac p21s to membrane and activation of NADPH oxidase in acell-free system. Biochem Biophys Res Commun. 1993; 195:264-269.

40. Schmid E, Figala V, Ullrich V. Inhibition of NADPH-oxidase activityin human polymorphonuclear neutrophils by lipophilic ascorbicacid derivatives. Mol Pharmocol. 1994;45:815-825.

41. O'Brian CA, Ward NE, Weinstein IB, Bull AW, Marnett LJ. Activa-tion of rat brain protein kinase C by lipid oxidation products.Biochem Biophys Res Commun. 1988; 155:1374-1380.

42. Sevanian A, Kim E. Phospholipase A2 dependent release of fattyacids from peroxidized membranes. / Free Radic Biol Med. 1985;1:263-271.

43. Gamache DA, Fawzy AA, Franson RC. Preferential hydrolysis ofperoxidized phospholipid by lysosomal phospholipase C. BiochimBiophys Acta. 1988;958:116-124.

44. Rao NA, Sultana C, Gullapalli VK, Numaga J, Kalra VJ. RPE protec-tive protein (RPP) inhibits p47-phox phosphorylation in inflamma-tory cells [ARVO Abstract]. Invest Ophthalmol Vis Sci. 1997;38(4):SI86. Abstract nr 922.

45. Wright CD, Hoffman MD. Comparison of the roles of calmodulinand protein kinase C in activation of the human neutrophil respi-ratory burst. Biochem Biophys Res Commun. 1987; 142:53-62.

46. Weiss SJ, LoBuglio AF. Biology of disease. Phagocyte-generatedoxygen metabolites and cellular injury. Lab Invest. 1982;47:5-18.

47. Radi R, Beckman JS, Bush KM, Freeman BA. Peroxynitrite-inducedmembrane lipid peroxidation: the cytotoxic potential of superox-ide and nitric oxide. Arch Biochem Biophys. 1991;288:481-487.

48. Wu GS, Riono WP, Zhang J, Rao NA. An RPE generated protectiveprotein (RPP) suppresses the docosahexaenoic acid hydroperox-ide-induced activation of PMNs [ARVO Abstract]. Invest Ophthal-mol Vis Sci. 1997;38(4):S186. Abstract nr 921.


Documents

Activation of NADPH oxidase by docosahexaenoic acid