Nucleoside triphosphate requirements for superoxide generation and phosphorylation in a cell-free system from human neutrophils. Sodium dodecyl sulfate and diacylglycerol activate independently of protein kinase C.

The NADPH-oxidase of human neutrophils can be activated in a cell-free system comprised of plasma membrane, cytosol, and an anionic amphiphile such as arachidonate or sodium dodecyl sulfate (SDS). Recently, we showed that diacylglycerol acts synergistically with SDS in the cell-free system to stimulate superoxide generation, with concurrent phosphorylation of a 47-kDa cytosolic protein which is thought to be a component of the oxidase (Burnham, D. N., Uhlinger, D. J., and Lambeth, J. D. (1990) J. Biol. Chem. 265, 17550-17559). We report herein that when undialyzed cytosol is used along with either SDS alone or SDS plus diacylglycerol as activators, adenosine 5'-(gamma-thio)triphosphate (ATP gamma S) and guanosine 5'-(gamma-thio)triphosphate (GTP gamma S) both stimulated superoxide generation several fold, yielding about the same maximal velocity. ATP and GTP showed lower levels of stimulation. Stimulation by ATP gamma S and GTP gamma S was nonadditive, and showed a 5-7-fold greater specificity for GTP gamma S. ATP gamma S stimulation was inhibited by the nucleoside diphosphate (NDP) kinase inhibitor UDP. In contrast, when extensively dialyzed cytosol was used, most of the stimulation by ATP gamma S was lost, while most of that by GTP gamma S was retained. Addition of GDP restored the ability of ATP gamma S to stimulate, consistent with NDP kinase-catalyzed formation of GTP gamma S from ATP gamma S plus GDP. This activity was demonstrated directly in both cytosol and plasma membrane. Using undialyzed cytosol, phosphorylation of p47 showed a similar nonspecificity for nucleoside triphosphates, due to NDP kinase activity, but revealed the expected ATP specificity when dialyzed cytosol was used. Neither ATP gamma S nor GTP gamma S were good substrates for protein phosphorylation. Under a variety of conditions, phosphorylation of p47 or other neutrophil proteins failed to correlate with oxidase activation. The present studies indicate that SDS and diacylglycerol stimulation of superoxide generation in the cell-free system is independent of protein kinase C or other protein kinase activity, and suggest a novel role for diacylglycerol in cell regulation.

The NADPH-oxidase of human neutrophils can be activated in a cell-free system comprised of plasma membrane, cytosol, and an anionic amphiphile such as arachidonate or sodium dodecyl sulfate (SDS). Recently, we showed that diacylglycerol acts synergistically with SDS in the cell-free system to stimulate superoxide generation, with concurrent phosphorylation of a 47-kDa cytosolic protein which is thought to be a component of the oxidase (Burnham, D. N., Uhlinger, D. J., and Lambeth, J. D. (1990) J. Biol. Chem. 265,[17550][17551][17552][17553][17554][17555][17556][17557][17558][17559]. We report herein that when undialyzed cytosol is used along with either SDS alone or SDS plus diacylglycerol as activators, adenosine 5'-(y-thio)triphosphate (ATPyS) and guanosine 5'47thio)triphosphate (GTPrS) both stimulated superoxide generation several fold, yielding about the same maximal velocity. ATP and GTP showed lower levels of stimulation. Stimulation by ATPyS and GTPyS was nonadditive, and showed a 5-7-fold greater specificity for GTPyS. ATPyS stimulation was inhibited by the nucleoside diphosphate (NDP) kinase inhibitor UDP. In contrast, when extensively dialyzed cytosol was used, most of the stimulation by ATPyS was lost, while most of that by GTPyS was retained. Addition of GDP restored the ability of ATPyS to stimulate, consistent with NDP kinase-catalyzed formation of GTPyS from ATPyS plus GDP. This activity was demonstrated directly in both cytosol and plasma membrane. Using undialyzed cytosol, phosphorylation of p47 showed a similar nonspecificity for nucleoside triphosphates, due to NDP kinase activity, but revealed the expected ATP specificity when dialyzed cytosol was used. Neither ATPyS nor GTPyS were good substrates for protein phosphorylation. Under a variety of conditions, phosphorylation of p47 or other neutrophil proteins failed to correlate with oxidase activation. The present studies indicate that SDS and diacylglycerol stimulation of superoxide generation in the cell-free system is independent of protein kinase C or other protein kinase activity, and suggest a novel role for diacylglycerol in cell regulation.
The NADPH-dependent respiratory burst oxidase (NADPH-oxidase) of human polymorphonuclear leukocytes provides the critical cellular defense against invading micro-* This work was supported by Grants A122809 and CA46508 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
To whom correspondence should be addressed.
organisms. The oxidase exists in a dormant state in unstimulated neutrophils, but can be activated by a variety of stimuli, including the chemoattractant formyl-methionyl-leucylphenylalanine, protein kinase C activators such as phorbol 12-myristate 13-acetate, and particulates such as opsonized bacteria or zymosan (see Ref. 1 for review). The activated oxidase catalyzes the univalent reduction of molecular oxygen to superoxide which secondarily generates other reactive oxygen species (e.g. H2O2, HOC1, . OH) which participate in oxidative killing of ingested microbes (2). The importance of the NADPH-oxidase is illustrated by the inherited condition chronic granulomatous disease, a family of biochemical diseases wherein defects in one of several proteins of the oxidase result in an inability to activate the respiratory burst, with a consequent impairment of affected individuals to combat infections (3). Identified by their absence in variants of chronic granulomatous disease, protein components of the respiratory burst oxidase (or related to its activation) include a plasma membrane-bound heterodimeric hemoprotein, cytochrome b558 (23-and 92-kDa subunits) (4, 5 ) , and two cytosolic components, p47phox and p67,1,,, (6, 7). The latter translocate to the plasma membrane upon activation, and fail to do so in chronic granulocytomous disease variants lacking cytochrome bss8, suggesting that these components participate in a complex with the cytochrome (8-10). A role for phosphorylation in activation in intact cells has been proposed, based on the ability of protein kinase C agonists to activate the respiratory burst (for reviews, see Ref. 1). A potential mechanism for activation involves the phosphorylation of p47ph,,, which occurs in parallel with activation by both phorbol 12-myristate 13-acetate and formyl-methionyl-leucyl-phenylalanine (11)(12)(13)(14). Recent studies using intact cells have demonstrated multiple phosphorylations of this protein (13,14), some of which require association with the cytochrome (15). Nevertheless, other studies (16, 17) have noted a lack of correspondence between phosphorylation and activation under certain conditions, and a causal role for phosphorylation in activation remains unclear. Indeed, a role for phosphorylation in down-regulation is possible, and specific phosphorylations of p47pho. might either activate or inhibit, depending upon site-specific phosphorylation.
Because of its extreme lability (18), the activated oxidase has been difficult to characterize in broken cell preparations from activated neutrophils. However, a cell-free activation system has recently been described (19)(20)(21), and consists of plasma membrane and cytosol, both from unstimulated neutrophils, along with an anionic amphiphile such as SDS' or arachidonate. Although the relevance of this cell-free activation mechanism to in vivo activation is not yet clear, the system has been invaluable in demonstrating and partially characterizing several oxidase-related components, and in investigating some cofactor/coenzyme requirements and catalytic properties. Concomitant phosphorylation of p47phox has been observed in this system (22), but, as in the intact cell, a causal role in activation remains unclear.
The nucleoside triphosphate and metal ion requirements for oxidase activation have been examined, both in electropermeabilized neutrophils and in the cell-free activation system. Data from permeabilized cells indicate that both ATP and GTP participate in activation (23-25), and pharmacological approaches in this system implicate a protein kinase, perhaps protein kinase C (25). Nevertheless, in permeabilized cells, it is not clear whether the nucleoside triphosphate requirements relate to the oxidase itself, or to upstream signal transduction components. In the cell-free system, which may reflect only the down-stream oxidase components, there is clearly a magnesium requirement. In addition, there is general agreement that guanine nucleotide analogs such as GTPyS and Gpp(NH)p stimulate superoxide generation by 2-4-fold above the rate seen with anionic amphiphile alone (26)(27)(28)(29). This, together with the observation that fluoride augments activity, has lead to the proposal of an oxidase-linked guanine nucleotide regulatory protein (G protein). Data regarding an adenosine triphosphate requirement are less clear. Several groups (26,28,30,31) report no requirement for ATP or ATP analogs, while others (29,32) have noted up to a 2-fold stimulation by adenosine triphosphate or its thiophosphoryl analog ATP$% In some cases, interpretations may have been affected by the presence of endogenous nucleotides in isolated cytosol (33). Thus, while GTP analogs clearly augment activation, the participation of adenosine triphosphate seems less clear.
We (34)(35)(36)(37) and others (38)(39)(40) have shown that under a variety of activation conditions, diacylglycerol generation correlates with superoxide generation. Because of previous data implicating the involvement of both protein kinase C and p47phox phosphorylation in this process, we recently investigated the effects of diacylglycerol and other protein kinase C activators on superoxide generation in the cell-free system (17). We discovered that although they are ineffective alone, short-chain diradylglycerols synergize with the anionic amphiphile to augment oxidase activities from 2.5 to 7-fold. In these studies, diacylglycerol also augmented phosphorylation of p47,h0,. Although it was tempting to speculate that protein kinase C was involved in the synergistic activation, several features of activation (e.g. calcium requirements, diacylglycerol specificity, lack of effect of inhibitors) did not appear to be consistent with a role for classical protein kinase C isoforms.' In the present studies, we have undertaken the examination of the nucleoside triphosphate (NTP) requirements for activation and phosphorylation in the cell-free system, using SDS, both alone and in combination with diC8. We find that for both activation conditions, the actual nucleoside triphosphate requirement is for guanosine rather than adenosine, and the apparent nonspecificity for NTPs is exadenosine 5'-(y-thio)triphosphate; App(NH)p, adenyl-5"yl B,-yimidodiphosphate; GTP-yS, guanosine 5'-(y-thio)triphosphate; Gpp(NH)p, guanosine 5'-(P,y-imido)triphosphate; NDP kinase, nucleoside-diphosphate kinase.
* For example, diacylglycerol analogs (e.g., 1,3-diacylglycerol, 1-0alkyl-2-acylglycerol) which are not known to activate protein kinase C are partially effective in the cell-free system. A variety of other diradylglycerol analogs and detergents, however, were ineffective, indicating that the diradylglycerol effect is structurally specific. plained by the presence of a nucleoside diphosphate kinase activity in cytosol and plasma membrane. The data indicate that diacylglycerol/SDS costimulation of the respiratory burst oxidase is not protein kinase-mediated, and suggest a novel, protein kinase C-independent role for diacylglycerol in cell activation.
Isolation of Human Neutrophils-Human neutrophils were isolated from peripheral blood from healthy adult donors as described previously (41). Informed consent was obtained from all subjects. Isolated cells were suspended in phosphate-buffered saline (2.6 mM KC1, 1.5 mM KH2P04, 0.5 mM MgCl,, 136 mM NaCl, 8 mM Na2HP04, 0.6 mM CaC12) containing 5.5 mM glucose and were used immediately.
Preparation of Plasma Membrane and Cytosolic Fractiom-Neutrophils (1-1.5 X lo9 cells) were suspended in 10 ml of phosphatebuffered saline-glucose and incubated on ice with 4 mM diisopropylfluorophosphate for 25 min. The cells were pelleted a t 600 X g for 7 min a t 4 "C and were resuspended in incubation buffer A (100 mM KCI, 3 mM NaC1,4 mM MgCl,, and 10 mM PIPES, pH 7.0, containing 1 mM EGTA, 2 p~ leupeptin, 2 p M pepstatin, and 0.5 mM phenylmethylsulfonyl fluoride). The cells were adjusted to 1.0 X 10' cells/ ml with the same buffer and were disrupted by nitrogen cavitation after being pressurized at 450 p.s.i. for 25 min at 4 "C (17). The cavitate was centrifuged a t 600 X g X 7 min, and the supernatant (-6 ml) was layered onto sucrose step gradients (3 ml each of 30 and 50% sucrose) and then centrifuged a t 210,000 X g for 75 min in a Beckman SW-41 rotor. The plasma membrane was recovered from the interface between the sucrose layers, and was pelleted by centrifuging a t 400,000 X g for 90 min, using a Beckman TL-100 ultracentrifuge.
The pellet was resuspended in 50 mM KC1, 1.5 mM NaC1, 5 mM PIPES (pH 7.0), 2.0 mM MgCl,, 0.34 M sucrose and 1 mM EGTA to a concentration of 3-5 mg/ml. The cytosol fraction, recovered from the top of the gradient, was centrifuged a t 230,000 X g for 1 h a t 4 "C to remove any residual particulate material. The resulting cytosol was concentrated to 10-15 mg/ml using Centricon (Amicon) filtration devices equipped with 10-kDa molecular mass cut-off filters. Both the membrane and cytosol fractions were stored in aliquots (50-150 p1) a t -80 "C. Samples stored in this manner were stable for a t least 3 months. Extensively dialyzed cytosol was prepared by dialysis (12-14-kDa cut-off) a t 4 "C against a 2000-fold excess volume of buffer A, with three changes of buffer a t -6-h intervals.
Zncubations-Reaction mixtures contained 20 pg of plasma membrane, 100 pg of cytosol, and varying amounts of agonists (e.g. nucleotide triphosphates, SDS, and diC8) in a total volume of 50 pl. Three 10-pl aliquots of each reaction mixture was transferred to 96well assay plates (Corning) and preincubated for 5 min a t 25 "C. The preincubation step prior to assay permits activation in the presence of concentrated components, allowing conservation of materials and high activities, as described previously (17). At the end of the preincubation period, NADPH (200 pM), cytochrome c (80 f i~) , and 240 p l of buffer A were added to initiate superoxide generation.
Superoxide Generation-Superoxide was quantified at 25 "C by monitoring cytochrome c reduction using a thermostatted Molecular Devices Thermomax Kinetic microplate reader equipped with a 550nm wavelength filter with a l-nm bandpass, as described (17). In parallel samples, superoxide dismutase (10 pg) was added after approximately 30 s to inhibit superoxide-dependent cytochrome c reduction. An extinction coefficient at 550 nm of 2 1 mM"cm" was used to calculate the quantity of cytochrome c reduced (42). The rate of cytochrome c reduction in the presence of superoxide dismutase was subtracted from that in its absence to calculate superoxidemediated cytochrome c reduction. For each preparation of cytosol and membranes, the optimal concentration of SDS for activation was determined prior to carrying out experiments. This concentration varied from 150 to 35 p~, depending on the preparation.
Protein Phospho~~at~on-Reaction mixtures containing cytosol, plasma membrane, and various effectors were as described above. The 45-p1 mixtures were kept on ice until the reaction was initiated with 5 pl of 2.7 mM ~-~~P -l a b e l e d ATP or GTP (approximately 1 pCi of label added, 270 p~ nucleotide final concentration). For some experiments, [y3'SS]ATP or [T-~~SIGTP were used at the same concentration and specific activity. The mixture was incubated for 10 min at 25 "C with shaking, and 50 pl of 2 X SDS Laemmli sample buffer was then added (43). Samples were heated at 100 "C for 3 min and centrifuged briefly prior to loading onto 12.5% gels for SDSpolyacrylamide gel electrophoresis.
Polyacrylamide Gel Electrophoresis and Autoradiography-Samples, approximately 12 gg of protein per lane, were subjected to electrophoresis on minigels (8 cm x 10 cm x 1 mm) using a modified Laemmli system (43). The running gel and stacking gel consisted of 12.5 and 5% acrylamide, respectively, with bis-acry1amide:acrylamide ratios of 1:180 and 1:37.5, respectively. After staining with Coomassie Blue, the gels were destained and dried onto Whatman 3 filter paper. The dried gels were subjected to autoradiography on Kodak X-Omat AR film overnight at -80 "C, with an intensifying screen. For 35Sprotein (thio)phosphorylation experiments, the film was allowed to develop for 4 days. Densitometry was performed on a Bio-Rad Model 620 densitometer equipped with one-dimensional Analyst data analysis software, and confirmed visual impressions. The major phosphorylated band at 47 kDa was confirmed as ~4 7 ,~~~ in earlier studies using two-dimensional gel electrophoresis and immunochemical methods, as described (17).
~~~e o s~d e -d i p h o s p~t~e Kinase Activity-Reaction mixtures (45 p l ) contained 40 pg of plasma membrane and/or 200 pg of either native or dialyzed cytosol, 4 PM [8-3H]GDP (approximately 2 pCi per reaction), and buffer A. ATPrS (5 pl of 1 mM) was added to initiate the reaction, and the mixture was incubated for 10 min at 25 "C with shaking. The reaction was terminated by the addition of EDTA to a final concentration of 17 mM. Samples were centrifuged for 15 min at 4,000 X g, and a 10-gl aliquot of the supernatant was mixed with 0.5 pl of carrier unlabeled GTP, GDP, and GMP (each at 1 mM final concentration) and applied to polyethyleneimine-cellulose thin-layer chromatography plates. The plates were developed in 0.75 M KHzP04 at room temperature as described (29). The nucleotides were visualized by fluorescence at 254 nm and the areas corresponding to standards were scraped and eluted for liquid scintillation counting.
Data Analysis-The maximal rates of cytochrome c reduction from triplicate aliquots from each reaction mixture were averaged to obtain a single rate value for each incubation. A minimum of three separate incubations were then averaged to provide standard error information. The data presented graphically are representative of results obtained from three preparations of membranes and cytosols obtained from separate donors. ECS and Vmex values were determined using a nonlinear regression data analysis as described previously (44).

Effect of ~~l e o s i d e ~r i p h o s p~t e s on Superoxide Generation in the Cell-free
System-Results using plasma membrane plus cytosol, activated with SDS, are shown in the two left panels of Fig. 1. In the absence of added nucelotide, a basal activity of 250-300 nmol of cytochrome c reduced/min/mg plasma membrane protein was seen. ATP (100 p M ) augmented the activity slightly less than 2-fold (Fig 1, upper left panel,  filled circles), and there was no further stimulation by concentrations up to 400 p~ (not shown). A larger (2.5-fold) augmentation which occurred at lower concentrations (optimum at 20 p~) was seen with ATPyS ( Fig. 1, upper leftpanel, open  circles). Using SDS alone, GTP had little or no stimulatory effect at levels up to 100 pM (Fig. 1, lower left panel, filled  squares). GTPyS, however, augmented oxidase activity (open squares) to a similar degree as ATPyS.
The activating effects of nucleoside triphosphates were also seen when an optimal concentration of diC8 was included along with SDS in the preincubation (Fig. 1, right panels). diC8 alone augmented the basal rate by more than 2-fold to 544 nmol/min/mg plasma membrane protein. and cytochrome c reduction was monitored continuously as described under "Experimental Procedures." Results with adenosine triphosphates are shown in the upper panels, and those with guanosine triphosphates in the lower panels. Note the different x-axis scales in upper and lower panels. Data points represent the mean f S.E. (n = 3) from one preparation. Where not shown, the error bar did not exceed the size of the symbol. Similar results were obtained using three separate preparations of cytosol and plasma membrane. oxide generation in the cell-free system. Data from the experiment in Fig. 1, right panels, using ATPyS and GTP-yS were replotted in a Lineweaver-Burk format. Because GTP-yS exhibited inhibition at high concentrations, only the data a t 100 p~ and less were used. The basal rate in the absence of nucleoside triphosphates was first subtracted. and higher concentrations increased the rate 1.6-fold (upper right, filled circles). ATPyS resulted in a larger (2.6-fold) enhancement of oxidase activity to more than 1400 nmol/ min/mg plasma membrane protein (upper right, open circles).
The lower rightpunel shows the effects of guanine nucleotides, using diC8 along with SDS to activate. There was a modest increase with GTP alone (1.5-fold at 100 p~) but a larger (3.3-fold) effect was seen with GTPyS (10 pM), which resulted in a rate of more than 1500 nmoi/min/mg plasma membrane protein. Thus, the presence of diacylglycerol increases both basal and nucleoside triphospha~-augmented activities. Data from the right panels in Fig. 1 using the optimal activators ATPyS and GTPyS were replotted in Fig. 2 in a Lineweaver-Burk format after subtraction of basal activities.

TABLE I Effect of ATPyS and GTPyS on the kinetics of NADPH-oxidase activity in cell-free systems
Preincubation and assay conditions were as in Figs. 1 and 4, and "Experimental Procedures." To permit a good kinetic fit to the data, rates were initially corrected by subtracting the basal activity (Le. that seen with no added nucleotide) prior to determination of kinetic parameters by a nonlinear least-squares method, as under "Experimental Procedures." The Vmax value shown represents the sum of the basal activity and the extrapolated maximal stimulation by nucleotide. The basal rates in nmol cytochrome c reduced/min/mg plasma membrane (PM) protein were as follows: in the native (undialyzed system) the rate with SDS alone with 245, and 544 with SDS plus diC8, and with dialyzed cytosol the rate with SDS was 56, and that with SDS Dlus diC8 was 155. -

Cell-free sys-Nucleotide
Echo tem/agonist added   Table I, under "SDS + diC8." The for GTPyS was 1.1 pM, 7-fold lower than that for ATPyS. The V,,, values using both thiophosphoryl nucleotides were essentially identical. Also shown in Table I are ECso and VmSx values for ATPyS and GTPyS stimulation using SDS alone as activator. ECso values for both nucleoside triphosphates were the same within experimental error as those seen when diC8 was included, again demonstrating a preference for the guanosine triphosphate analog. VmaX values were about half of those seen when diC8 was included, and showed a slight preference for the guanine nucleotide. Thus, both nucleotides activated superoxide generation to an approximately equal extent, and stimulation was more specific for GTPyS than ATPyS.
The similarity in V,,, values suggested that ATPyS and GTPyS might be activating by the same mechanism. If this were the case, then simultaneous addition of both nucleotides should be nonadditive. Results using combinations of the two nucleotides are summarized in Fig. 3. The concentrations of nucleotides utilized were those that elicited the maximal responses in Fig. 1. As shown in Fig. 3, there was no additivity observed when SDS alone was used to activate. Although there was a slight additional activation seen by the combined nucleotides when diC8 was included, the rate was considerably less (-1750 nmol/min/mg plasma membrane protein) than that which would be predicted for additive rates (-2400 nmol/ min/mg plasma membrane protein). Thus, ATPyS and GTPyS do not stimulate in an additive manner, suggesting that they may act by the same mechanism.
Effect of Dialyzed Cytosol on the Nucleoside Triphosphate Dependence of Superoxide Generation in the Cell-free System-Because nucleotides are known to be present in neutrophil cytosol (33) and might complicate the inte~retation of nucleotide-dependent activation, we repeated the experiments shown in Fig. 1 using extensively dialyzed cytosol. Basal activities were 4-6-fold lower using dialyzed cytosol, both with SDS and with SDS plus diC8 as activators, consistent with the loss of low molecular weight stimulatory factor(s). With SDS alone (Fig. 4, upper leftpanel), neither ATP (filled circles) nor ATPyS (open circles) activated to an appreciable degree. Results with guanine nucleotides, however, were essentially the same as those obtained with undialyzed cytosol, although the maximal activity achieved was somewhat lower. GTPyS (10 p~) stimulated oxidase activity about 10-fold, from a basal rate of 56 nmol of cytochrome c reduced/min/ mg plasma membrane protein. The degree of stimulation was higher in the dialyzed system (9us. 2.5-fold), primarily due to a lower basal activity. Presumably, the presence of endogenous nucleoside t~phosphates in the undialyzed cytosol elevates the basal rate significantly.
When die8 was used along with SDS in the dialyzed system (upper rightpanel, Fig. 4), there was a very modest stimulation of activity by ATP and a slightly greater effect of ATPyS, but this response remained much less than that observed when native cytosol was used (compare with Fig. 1). As shown in the lower right panel of Fig. 4, GTP again had only a small (40%) stimulatory effect (filled squares). However, as in the  effect (open squares), and achieved a level of stimulation almost as great as that obtained when undialyzed cytosol was used (compare lower right panels of Figs. 1 and   4). As summarized in Table I, the EC5, and V,,, values for GTPyS were relatively unchanged by dialysis. However, the VmaX for ATPyS stimulation was about 6-fold less when dialyzed cytosol was used. Thus the removal of soluble low molecular weight components by dialysis drastically reduces the ability of adenine nucleotides to stimulate in the cell-free system, but activation by guanine nucleotides remains largely intact.
GDP Restores the Ability of ATPyS to Stimulate Activity Using Dialyzed Cytosol, and UDP Inhibits ATPyS Stimulation-A possible explanation for the apparent nonspecificity for nucleotides in the native cytosol is that NDP kinase has catalyzed the transfer of the thiophosphoryl groups among nucleotides, e.g. from ATPyS to endogenous GDP to form GTPyS. Thus when, dialyzed cytosol is used, there is no longer free GDP to act as substrate acceptor for the (thio)phosphoryl group. To test this hypothesis we used the cell-free system consisting of plasma membrane and dialyzed cytosol, as in Fig. 4. Results with SDS plus diC8 are presented in Fig. 5. GTPyS (open squares) again stimulated activity markedly (7.8-fold), and showed an optimum at 10 pM. As in Fig. 4, there was only a modest stimulatory effect of ATPyS alone (open circles, Fig. 5). However, inclusion of 10 NM GDP along with ATPyS restored the ability of the ATPyS to activate (filled triangles), and rates near those seen with GTPyS were observed. The concentration of GDP used was critical to observe restoration of activity by ATPyS, since 2fold higher or lower concentrations resulted in a much smaller degree of stimulation (data not shown). These data support the idea that the loss of ATPyS stimulation upon dialysis of cytosol is due at least in large part to removal of GDP, which is functioning as a thiophosphoryl acceptor. Thus, following dialysis, the actual nucleoside triphosphate dependence of the system is revealed. In support of this interpretation, UDP, which can function as an inhibitor of NDP kinase, inhibited ATP and ATPyS but not GTPyS stimulation of oxidase activity (IC50 = 0.75 mM using 200 phi ATP or ATPyS, 350 PM SDS and 200 PM diC8) when undialyzed cytosol was used (data not shown), confirming results by Seifert et al. (29) using arachidonate.   (CYT) and

plasma membrane (PM)
Incubation conditions and concentrations were as in Fig. 1 and under "Experimental Procedures." DCYT represents dialyzed cytosol. Results are expressed as picomoles of [3H]GTPrS produced from ATPrS and [3H]GDP per mg of protein in 10 min.

Nucleoside-diphosphate Kinase Activity in P l a s m Membrane and
Cytosol-To test directly whether NDP kinase activity in cytosol and plasma membrane preparations could account for the observed nucleotide specificity, we examined the ability of these fractions to catalyze thiophosphoryl transfer from ATPyS to [8-3H]GDP to form [8-3H]GTPyS. Results are shown in Table 11. NDP kinase activity was present in both cytosol and plasma membrane, although the activity in the combined fractions was somewhat lower than that expected from the individual activities. The activity was higher when dialyzed cytosol was used, presumably due to dilution of the tritium label in the undialyzed sample with unlabeled GDP and/or other nucleotides. Neither SDS nor SDS plus diC8 at concentrations used in the superoxide assays affected the nucleoside-diphosphate kinase activity (data not shown).

Nucleoside Triphosphate Dependence for Phosphorylation of p47,,b,, Using Undialyzed and Dialyzed Cytosol-Both [y-
32P]ATP and [T-~*P]GTP were used as phosphoryl donors under activation conditions given in Figs. 1 and 4, and phosphorylation was monitored by SDS-polyacrylamide gel electrophoresis followed by autoradiography, and under "Experimental Procedures." As shown in Fig. 6A, when native (undialyzed) cytosol was used, both SDS (S) and diacylglycerol ( D ) alone and in combination ( S / D ) enhanced the phosphorylation of both a 47-kDa protein and a 68-kDa p r~t e i n .~ [T-~*P]ATP and [y-"PIGTP functioned equally well as phosphoryl donors under these conditions. Comparison of Figs. 1 and 6 reveals that phosphorylation failed to correlate with activation. For example, with SDS alone or combined with diacylglycerol, ATP and GTP served equally as phosphoryl donors to protein, but only ATP activated to a significant extent. Diacylglycerol alone promoted heavy phosphorylation with either nucleotide, but in the absence of SDS fails to activate superoxide generation (17). With dialyzed cytosol (Fig. 6, lower panel), GTP lost much of its ability to serve as a phosphoryl donor to protein, while ATP-dependent p47 phosphorylation remained largely intact. Phosphorylation of the p47 again failed to correlate with activation (compare Fig.   6B with Fig. 4). Using all combinations of activators, ATP remained a good phosphoryl donor to ~47,~,,, but lost most of It should be noted that an apparently synergistic effect of diacylglycerol and SDS on p47 phosphorylation was seen previously by US (17), but conditions (tracer concentration of added [y-32P]ATP and shorter reaction time) differed considerably from those of the present studies. In the present studies which use more physiological concentrations of ATP and GTP, we found a moderate stimulation of p47 phosphorylation by SDS, a generally higher activation by diacylglycerol alone, and little or no additive or synergistic effect of the two together. Densitometry was carried out for all experiments and confirmed the visual impression. FIG. 6. Dependence of protein phosphorylation on nucleoside triphosphates in the cell-free system consisting of plasma membrane plus either native or dialyzed cytosol. Incubations, SDS-polyacrylamide gel electrophoresis, and autoradiography were as detailed in Fig. 1 and under "Experimental Procedures." A, autoradiogram obtained from SDS-polyacrylamide gel electrophoresisresolved phosphoproteins derived from incubations containing native (undialyzed) cytosol while B is that using extensively dialyzed cytosol, both in addition to plasma membrane. Parallel incubations were carried out using 270 PM of either ATP ( A ) or GTP ( G ) , and 10 PM ADP was included as indicated.

Requirements for NADPH-oxidase
Incubations its effectiveness in supporting superoxide generation. Thus, ATP rather than GTP is the primary phosphoryl donor for protein phosphorylation, but phosphorylation fails to correlate with activation.
In an experiment analogous to that shown in Fig. 5, ADP was included along with [y-32P]GTP in the system utilizing dialyzed cytosol (Fig. 6B). Under these conditions, phosphorylation of p47 by [y-"PIGTP was restored to a level near that seen using undialyzed cytosol. Thus, the nonspecificity for nucleotide triphosphate-dependent protein phosphorylation in the system using undialyzed cytosol was due to nucleoside-diphosphate kinase-dependent transfer of the [Y-~'P] phosphate from [y-32P]GTP to ADP to form [y-"PIATP, which was then utilized for protein phosphorylation. To address the possibility that the large activations seen with ATPyS and GTPyS were the result of (thio)phosphorylations catalyzed by an unknown kinase, experiments exactly analogous to those presented in Fig. 6 were performed except that [y-35S]ATPyS and [y3'SS]GTPyS were used as thiophosphoryl donors. Virtually no label was incorporated using either nucleotide with SDS or diC8, alone or in combination.

DISCUSSION
We recently described (17) the stimulatory effect of diacylglycerol on superoxide generation in a cell-free system consisting of plasma membranes, cytosol, and an anionic amphi-phile (SDS). Our assumption at the outset, based on the wellknown function of diacylglycerol as a protein kinase C activator, was that despite some anomalies with regard to specificity (17), the activation mechanism would likely turn out to involve protein kinase C, or perhaps one of the recently described related isoforms. The expectation was, therefore, that the stimulation by diacylglycerol would require ATP. Tauber and colleagues (45, 46) had previously described activation of superoxide generation by phorbol esters in a cellfree system in which the cytosolic requirement could be replaced by purified protein kinase C and showed an ATP requirement in this system. Nevertheless, the activity achieved was very low (less than 0.7% of that seen in the present studies), and we (17) and others (47) have repeatedly failed to observe significant activation by phorbol esters, either in the presence of cytosol or purified protein kinase C. In the face of conflicting reports with regard to an ATP requirement in the standard cell-free system (i.e. without diacylglycerol), we undertook the present studies to determine under our assay conditions the NTP requirements with SDS alone, and to define the NTP requirements for the diacylglycerol-linked stimulation.
As shown in Figs. 1 and 4, qualitatively similar NTP requirements were seen regardless of whether SDS alone or SDS plus diacylglycerol were used as activators. In the undialyzed cytosol (Fig. l), ATPyS and GTPyS were both effective in stimulating superoxide generation under both activation conditions. ATP was somewhat effective, while GTP was the least effective, and in the case of SDS activation, did not activate a t all. Diacylglycerol did not alter the NTP specificity; rather its effect was to increase the maximal velocity at optimal NTP concentrations. In the undialyzed system, the similarity of V,,, values using both ATPyS and GTPyS as well as their nonadditive effects were consistent with a common stimulatory mechanism. The lower ECso for GTPyS compared with ATPyS suggested a specificity for guanine over adenine nucleotides. When the dialyzed cytosol was used, the nucleoside triphosphate requirement was radically altered in favor of guanine nucleotides, and the effect was at the level of the V,,,. As with the dialyzed cytosol, the NTP specificity was independent of whether SDS alone or SDS plus diacylglycerol was used to activate.
The present studies indicate that the nucleoside triphosphate specificity is primarily for the guanine nucleotide, and imply that the effects of adenosine triphosphates in the undialyzed system were due to a NDP kinase activity which can transfer the y-phosphoryl or y-(thio)phosphoryl group from ATP or ATPyS to GDP to form the corresponding GTP derivative. In dialyzed cytosol, free GDP is removed, thus eliminating this pathway. In support of this explanation, we demonstrated directly a NDP kinase activity in both cytosol and plasma membrane which was capable of transferring the thiophosphoryl group from ATPyS to GDP. In addition, GDP restored the ability of ATPyS to activate maximally. Seifert et al. (29) previously demonstrated a nucleoside diphosphate kinase activity in differentiated HL-60 cells. This group observed ATPyS stimulation of superoxide generation in the arachidonate-activated cell-free system, and proposed that this was due to NDP kinase-catalyzed generation of GTPyS, based on a lack of activation by App(NH)p and inhibition by UDP. As in the Seifert studies, we also found UDP to be an effective inhibitor of ATP and ATPyS stimulation of oxidase activity. However, stimulation by GDP was not observed in (29), perhaps because of endogenous GDP in undialyzed cytosols. In our experiments, we found a narrow window of concentrations in which GDP affected ATPyS stimulation.

Requirements for NADPH-oxidase
The reason for the narrow activity window for GDP is not clear, but may relate to inhibition of activity by excess GDP which may not be completely converted to GDP. We suggest that the presence of NDP kinase and variable endogenous nucleotide levels, which may occur using different preparation methods and dilutions, accounts for contradictory reports regarding the effects of ATP and ATPyS on activation.
Two quantitatively small but apparently anomalous results require explanation. First, in the dialyzed system with SDS plus diacylglycerol, there is a small residual stimulation by ATPyS, and to a lesser extent by ATP (Fig. 4, upper right   panel). The stimulation retains the same EC5, for ATPyS, but shows a markedly lower Vmar compared with the undialyzed system. This observation could indicate a modest activating effect of a kinase. Alternatively, there may be residual GDP which can serve as phosphoryl acceptor. Although residual free GDP is not expected given the extensive dialysis proto~ol,~ a small quantity of tightly protein-bound GDP may be retained. In this regard, many G proteins are known to retain bound GDP. Although the classical mechanism for activation of G proteins involves dissociation of GDP with subsequent binding of GTP, recent evidence indicates that G, can be physically associated with NDP kinase (48, 49) and that the latter can catalyze directly the phosphorylation of GDP by ATP or ATPyS without nucleotide exchange (50, 51). Thus, if residual GDP is bound to a G protein, ATP (or ATPyS) might serve as direct phosphoryl donor to the bound nucleotide. This mechanism could also account for a second apparent anomaly: while there is a marked specificity for GTPyS over ATPyS, there is a preference for ATP over GTP when native nucleotides are used (e.g. see Fig. 1). This could be explained if a G protein-associated NDP kinase shows a nucleotide donor preference for ATP. We are not aware of published studies which relate to this point.
Phosphorylation experiments also support the idea that at least in the cell-free system, a protein kinase is not sufficient for activation. Using undialyzed cytosol, there was no preference for ATP over GTP as phosphoryl donor to p47,h,, under any activation condition, despite the preference for ATP for activation. Using dialyzed cytosol, the preference for protein phosphorylation was for ATP, as expected for a kinase, and addition of ADP restored the ability of GTP to function as a phosphoryl donor. Thus, the apparent nonspecificity for NTP seen in the undialyzed system was due to the NDP kinasecatalyzed phosphoryl transfer from added [y-32P]GTP to endogenous ADP to form ATP. In both dialyzed and undialyzed systems, a high degree of phosphorylation of p47,h0. was often seen under conditions where little or no activation was observed. Thus, phosphorylation of p47ph0, in the cell-free system is not a sufficient stimulus for activation. In addition, using inhibitors of protein kinases, Badwey and colleagues (16) reported activation in the absence of p47ph0, phosphorylation in a guinea pig (intact cell) system (16). Our data in the context of other reports do not support a role for a protein kinase. Rather, our data are consistent with reports from several laboratories (see Introduction) implicating a G protein.
The present data also imply a novel, protein kinase Cindependent regulatory role for diacylglycerol. Diacylglycerol is well-known as a physiological activator of protein kinase C, and is also reported to regulate the translocation of this kinase to the plasma membrane (52). Additionally diacylglycerol may participate in other regulatory functions. For example, diacylglycerol promotes the translocation of diacyl-