Translocation of Rac Correlates with NADPH Oxidase Activation EVIDENCE FOR EQUIMOLAR TRANSLOCATION OF OXIDASE COMPONENTS*

Activation of the superoxide-generating NADPH oxi- dase system of human neutrophils involves the assembly of several neutrophil components, some located on the plasma membrane and others in the cytosol. It has re-cently been established that one of the required components for NADPH oxidase activity is the GTP-binding protein Rac. To further investigate the role of Rac in the NADPH oxidase system, studies were carried out to de- termine its subcellular distribution in resting and activated human neutrophils. In resting cells, Rac and an associated guanine nucleotide regulatory factor, GDP dissociation inhibitor (GDI), were located only in the cytosol, along with other known oxidase factors, p47-phar and p67-phar. After activation of neutrophils with phorbol 12-myristate 13-acetate or formyl-methionyl- leucyl-phenylalanine, Rac was translocated from the cytosol to the plasma membrane, and this translocation corresponded temporally with the translocation of p47- phox and p67-phox and with the generation of superoxide, GDI to the cytosol, suggesting activation of the oxidase involved dissociation of the Rac-GDI complex prior to Rac translocation. Determi-nation of the quantities of cytosolic factors associated with the plasma membrane indicated that Rac, p47-phox, and p67-phoz are translocated to the plasma mem- brane simultaneously in equimolar amounts, but that the membrane-associated cytochrome b was present at 34-fold molar excess. These findings suggest that Rac may play a role in assembly of the active NADPH oxidase complex. Superoxide in cells celldml Dulbecco's 0.65 mg/ml cytochrome with 1 PMA or PM fMLP at 25 "C incubating cells stirred cuvettes and the reduction of Cytochrome c. Superoxide dismutase-inhibitable cytochrome C reduction calculated using E = 18.5 mM-' cm-l. Antibody Preparations-Previously characterized anti-peptide antibodies used for Western blotting included antibodies to peptides from gp9l-phox (residues 546-558) (281, p22-phox (residues 162-174) (30), RaplA(residues 131-140) (31), and GDI (residues 17-28) (23). Antibod- ies against the entire Rac2 and GDI proteins were prepared as described (32) using purified recombinant Rac2 (23) or human (Rho) GDI (33) as antigens. These antibodies reacted specifically with Rac2

* This work was supported in part by National Institutes of Health (to A. J. J.) and GM44428 and HL48008 (to G. M. B.). 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.
$ Recipient of an Arthritis Foundation investigator award and an American Lung Association research grant. To whom correspondence should be addressed. Tel.:  (1 Recipient of California Tobacco Related Disease Research Award  plicated in the flow of electrons from NADPH to oxygen a r e a flavoprotein (4,5) coupled to a low potential b-type cytochrome, which is believed to directly reduce molecular oxygen to 0 ; (6). Recent studies (7,8 ) suggest that the cytochrome b and flavoprotein may not be distinct components but may be integrated into a single flavocytochrome. Biochemical and genetic analyses have also demonstrated the participation of two cytosolic proteins, p47-phox and p67-phox (9-11). These proteins are localized to the cytosol in unstimulated cells, possibly as part of a larger macromolecular complex (12, 13), but become tightly associated with the membrane-bound components upon cell activation (14,15).
The participation of GTP-binding proteins in the NADPH oxidase system was indicated by the dependence of activation in cell-free systems upon . Ligeti et al. (17) and Gabig et al. (20) reported that activation in the cell-free system resulted from the GTP-dependent formation of a complex between a cytosolic protein and a membrane protein and suggested that there was a cytosolic GTP-binding protein involved in oxidase activation. This cytosolic GTP-binding protein has been identified as the Rho-related protein, Rac (21-23). Rac (Racl and/or Rac2) has been shown to be required for NADPH oxidase activation in cell-free systems consisting of recombinant p47-phox, p67-phox, and relipidated cytochrome b (7,24) or purified neutrophil plasma membranes (25, 26), indicating that Rac is responsible for at least part of the GTP sensitivity of the NADPH oxidase system and is a third required cytosolic cofactor.
Since changes in the subcellular distribution of NADPH oxidase components take place during assembly of the active oxidase, we have carried out studies to determine the subcellular distribution of Rac and the associated GDP-dissociation inhibitor, GDI' (21,23), in resting and activated human neutrophils. The kinetics and quantity of each of the other cytosolic factors that translocated to the plasma membrane were determined. These studies indicate that Rac does translocate to the plasma membrane from the cytosol upon neutrophil activation, and this translocation corresponds well both temporally and quantitatively with p47-phox and p67-phox translocation and NADPH oxidase activation.
MATERIALS AND METHODS Preparation, Activation, and Fractionation of Neutrophils-Purified human neutrophils (los cells/ml), isolated as previously described (27).
were resuspended in a modified Dulbecco's phosphate-buffered saline, warmed to 37 "C, and stimulated for the indicated times with 1 pg/ml PMA (phorbol 12-myristate 13-acetate) or 1 V M fMLP with gentle agitation. The reaction was stopped by the addition of ice-cold buffer, and ~ The abbreviations used are: GDI, GDP dissociation inhibitor; PMA, phorbol 12-myristate 13-acetate; fMLP, formyl-methionyl-leucyl-phenylalanine; MCID, Microcomputer Imaging Device (Imaging Research, Inc.). the cells were disrupted by Nz cavitation at 4 "C and fractionated either by isopycnic or discontinuous sucrose density gradient sedimentation as described previously (28).
Biochemical AssayeNeutrophil cytochrome b was quantitated by reduced-minus-oxidized difference spectroscopy on a Cary 3 dual beam spectrophotometer, using a reduced-minus-oxidized Soret band (427 nm) extinction coefficient of 161 m"l cm-'. Alkaline phosphatase, myeloperoxidase, total protein, and other markers were measured as described (27, 29).
Superoxide generation in sucrose gradient fractions was measured at 25 "C as described previously (28) in 1.6-ml microcuvettes containing 650 pl of detection buffer (0.65 mg/ml cytochrome c, 2 m~ MgCl,, 2 m~ NaNs, and 10 l l l~ Hepes, pH 7.4) and 50 pl of membrane. Superoxide generation in intact cells (2 x lo6 celldml in Dulbecco's phosphatebuffered saline + 0.65 mg/ml cytochrome c ) stimulated with either 1 pg/ml PMA or 1 PM fMLP was determined at 25 "C by incubating the cells in stirred cuvettes and monitoring the reduction of Cytochrome c. Superoxide dismutase-inhibitable cytochrome C reduction was calculated using E = 18.5 mM-' cm-l.
Danslocation I Western Blotting Experiments-To determine kinetics of translocation, cells were stimulated with 1 pg/ml PMA or 1 PM fMLP and the reaction was terminated at various time points (see Fig. 2 legend) by the addition of four volumes of ice-cold buffer. Plasma membranes were isolated from cell cavitates using sucrose step gradients as described previously (27), and samples were analyzed by SDS-polyacrylamide gel electrophoresis and Western blotting (27, 28). Western blotting was performed as previously described (27.28) using an alkaline phosphatase-conjugated goat anti-rabbit IgG secondary antibody (Bio-Rad). Western blots were quantitated by video densitometry using an image analysis system (Microcomputer Imaging Device (MCID) with Image Analysis software) (Imaging Research, Inc., Brock University, Ontario, Canada). The relative density of a sample (expressed in arbitrary units) represents the density of the sample relative t o an arbitrary gray scale defined by the image analysis system to cover the total gray shade range of the image.

Subcellular Distribution of Rac-
The subcellular distribution of Rac, GDI, and other oxidase-associated proteins in resting or PMA-stimulated human neutrophils was determined by isopycnic sucrose density gradient sedimentation (Fig. 1). Distinct subcellular organelles were localized using specific markers that have been characterized previously (i.e. plasma membrane (PM) markers (alkaline phosphatase, RaplA, and cytochrome b; peak = 30-32% sucrose, fractions 12-14), specific granule (SG) markers (cytochrome b and RaplA; peak = 41-43% sucrose, fractions 18-20), and azurophil granule (AG) marker (myeloperoxidase; peak = 48-50% sucrose, fractions 22-24)) (27, 29,31,37). Fig. 1 (panel C) also shows the subcellular distribution of superoxide-generating activity in PMA stimulated cells (peak activity = 3636% sucrose, fractions 13-15). As was observed previously (271, the peak 0;-generating activity occurred in the heavy plasma membrane (enriched in G. M. Bokoch, unpublished data. tions were determined by densitometric scans of Western blots as described. Superoxide dismutase-inhibitable superoxide-generating activity in fractions from PMA-stimulated cells (W) (panel C) was determined as described under "Materials and Methods." In all panels, percent of the maximum level of the specific oxidase proteins or oxidase activity (panel C) recovered from the gradients (including cytosolic fractions) is plotted as a function of fraction number (fraction 1 = 10% sucrose; fraction 29 = 55% sucrose). The specific activity of the peak superoxide-generating fraction was 28.3 nmol of O&nin/mg. Locations of the cytosolic fractions and the peak enzymdmarker activities for the plasma membrane (I"), specific granules (SG), and azurophil granules (AG) are indicated at the top ofpanel A. The results are representative of four separate experiments (variability in sedimentation profiles between experiments was 52%).
The fractions from these gradients were analyzed for the distribution of Rac using Western blotting techniques with antibodies prepared against recombinant Rac as described. As Fig. 1 (panel A) shows, Rac is localized completely to the cytosolic compartment of resting cells (fractions 1-10], but is translocated to the plasma membrane and 0;-generating fractions (fractions 12-15) after stimulation with PMA (panel B 1. Quantitation of the amount of translocated Rac using densitometric scans of the Western blots indicated that -10% of the total cytosolic Rac was translocated in these gradients. In addition, if the cells were stimulated with the chemotactic peptide MLP, a similar redistribution of Rac to the plasma membrane was observed (see Fig. 3). This indicates that Rac translocation is not stimulus-specific and that translocation of this GTP-binding protein is a general phenomenon associated with oxidase activation for both a receptor-mediated and a non-receptormediated stimulus.
Because Rac is associated with a guanine nucleotide regulatory protein (GDI) in human neutrophil cytosol (24, 381, the subcellular distribution of GDI was also analyzed to determine if GDI could be translocating in association with Rac, or by itself, during neutrophil activation. The localization of GDI was assessed using both the GDI-specific anti-peptide antibody and a more sensitive anti-protein antibody (see "Materials and Methods"). Our detection level was down to <lo ng, which would have readily detected any GDI translocated at levels even approximating those of Rac. As Fig. 1 (panel A ) shows, GDI is localized completely to the cytosolic compartment (fractions 1-10] in resting cells, and it remains in the cytosol after activation (fractions 1-10; panel B ) .
Since the oxidase cofactors p47-phox and p67-phox are known to be translocated during oxidase activation, we analyzed the subcellular distribution of p47-phox and p67-phox in resting and PMA-stimulated cells. Fig. 1 (panel B ) shows that p47-phox and p67-phox translocated from the cytosol (fractions 1-10] to the plasma membrane (fractions 12-15) after the cells were activated with PMA. Quantitation of this movement using densitometric scans of the Western blots indicated that approximately 10-18% of the total amounts of each of these proteins in the cytosol was translocated. This result is consistent with those previously reported (14, 15). Fig. 2A, 0; generation in PMA-stimulated neutrophils began -1 min after the addition of PMA and increased steadily over the next 20 min, although the overall rate of 0; production decreased gradually over the time studied. In cells stimulated with fMLP ( Fig. 3A), 0; production began within 15 seconds and plateaued approximately 3-4 min after addition of the stimulus. These activation kinetics are typically observed with such nonreceptor-and receptor-mediated stimuli, respectively. Analysis of the plasma membranes from these cells by Western blotting for Rac, p47phox, and p67-phox showed that all three proteins were stably translocated to the plasma membrane in samples prepared from PMA-stimulated cells (Fig. 2 B ) or in membranes prepared from fh4LP-stimulated cells (Fig. 3B). In addition, the time course of Rac, p47-phox, and p67-phox translocation closely paralleled the kinetics of 0; production in both PMA and fMLPactivated cells. Thus, in PMA-stimulated cells accumulation at the membrane occurred relatively slowly and increased over the 20 min with a gradual tapering off at the longer time points, while in fMLP-activated cells all three cytosolic proteins translocated to a near-maximal membrane-associated level within the first minute of stimulation and then increased only slightly over the remainder of the time studied. These data suggest that translocation of all three components is necessary to maintain a continuous production of superoxide.

Quantitation of the Danslocation of Cytosolic Oxidase
Factors-In order to determine the relative quantities of Rac, p67phox, and p47-phox that translocate to the membrane during activation, membrane samples prepared from the 6-min time points of fMLP-and PMA-stimulated cells (see legend to Fig. 2) were analyzed by quantitative Western blotting. Representative blots and standard curves obtained by quantitative densitometry of recombinant protein standards are shown in Fig. 4. Quantitative analysis revealed that Rac, p47-phox, and p67phox translocated to the plasma membrane in equimolar quantities, with approximately 200,000 molecules of each protein being translocated per cell in cells stimulated with fMLP for 6 min (actual measured values indicated by an F on the standard curves were 23 * 3.1,49 * 5, and 70 * 7.1 ng of proteidsample, which calculates to 7.0 * 0.9, '14.8 * 1.5, and 21. 20,358 moleculedcell for Rac, p47-phox, and p67-phox, respectively; mean f S.D.; n = 2) and approximately 250,000 molecules of each protein being translocated in cells stimulated with PMA for 6 min (actual measured values indicated by a P on the standard curves were 32 * 1.7,59 * 5, and 91 * 2.1 ng of proteidsample, which calculates to 9.4 + 0.5, 17.4 f 1.5, and 26.8 f 0.6 ng/106 cell eq or 265,241 f 14,000,230,000 + 20,000, and 247,673 + 5715 moleculedcell for Rac, p47-phox, and p67phox, respectively; mean + S.D.; n = 3). In comparison, approximately 800,000-1,000,000 molecules of cytochrome blcell were present in the same samples, as determined spectrophotometrically.

DISCUSSION
Racl and/or Rac2, which are Rho-related GTP-binding proteins, have been shown to be required for activation of the Kinetics of Rac translocation in f"-stimulated hum a n neutrophils. Purified human neutrophils were treated with 1 PM fMLP ( arrow ) and superoxide generation (panel A ) was determined as described under "Materials and Methods." In parallel incubations, cells were removed after the indicated time of stimulation and membranes were prepared and analyzed for Rac, p47-phox, and p67-phox using NADPH oxidase in cell-free reconstitution assays (21-26) and in intact cells (39). "he present studies address the mechanisms by which Rac may regulate oxidase activity by showing that Rac translocates to the plasma membrane in PMA-and fMLP-stimulated human neutrophils, and that the kinetics of Rac translocation parallel the kinetics of 0; production by these cells. It should be noted, however, that the kinetics of translocation do not correlate exactly with the kinetics of 0; production, especially a t early time points after activation where translocation seems to precede 0, production. One explanation for these results is that there may be a lag time required for the translocated proteins to interact with each other and with cytochrome b to assemble into active 0;-generating complexes. This is suggested in fMLP-stimulated cells where 6040% of the total translocation occurs during the first 30 s of stimulation, while 0; production rises more slowly and in PMA-stimulated cells where translocation precedes activity in the first minute. A second possibility is that there may also be a threshold amount of cytosolic components that need to translocate to initiate the oxidative burst, as is evident in PMA-stimulated cells where ?80,000 molecules of cytosolic proteins are translocated in the first minute before any noticeable amount of 0; is produced. Further experiments are in progress to analyze these possibilities. The Rac regulatory modulator GDI, which binds Rac and maintains it in a cytosolic complex (33, 38), is not translocated along with Rac. Since we find that all the cytosolic Rac in resting neutrophils exists in a GDI complex (381, these results indicate that cell activation causes dissociation of these proteins, allowing Rac to then associate with the plasma membrane. Indeed, we have observed that activation of the oxidase under cell-free assay conditions is accompanied by Rac-GDI dissociation (38). "hus, activated Rac, with its isoprenylated C terminus freed up by release of GDI, can bind to the plasma membrane in conjunction with interacting cytosolic components such as p47-phox or p67-phox. As such, Rac could serve as a "chaperone" that helps with the translocation of the latter proteins to the plasma membrane and/or membrane skeleton, where they are required for oxidase assembly. Alternatively, translocated Rac could itself serve as a membrane target for the translocation and binding of p47-phox and p67-phox. Although we cannot conclusively rule out the latter possibility, our data indicate that all three proteins translocate with similar kinetics and, therefore, fit better with the chaperone hypothesis.
This would also be consistent with data from Park and Babior (40) and Uhlinger et al. (411, who reported that non-hydrolyzable guanine nucleotides stimulated the association of p47phox and p67-phox with the membrane in cell-free systems. Previous studies showing that components of the oxidase are organized by and associated with the membrane skeletal ma-trix (27,42,43) raise the additional possibility that Rac may play some role at the level of cytoskeletal assembly/association (44).
In this report, we provide a direct measurement of the actual amounts of cytosolic oxidase proteins translocating to the plasma membrane in activated neutrophils. Our measurements indicate that the three cytosolic cofactors (Rac, p47phoz, and p67-phox) translocate in equimolar quantities, while there is a 34-fold molar excess of cytochrome b present on the membrane. These data confirm our previous results suggesting that there is excess cytochrome b in the plasma membrane that is either not involved in generating superoxide or that multiple cytochromes participate in one superoxide-generating complex (27,37). The overall cytosolic levels of total Rac (Racl and Rac2) and GDI have been determined by quantitative Western blotting as 80.0 * 10.3 and 337.0 * 150 ng/106 cell eq, which corresponds to -2.3 * 0.3 x lo6 and 8.2 * 3.6 x lo6 molecules/ cell for Rac (12) and GDI, respectively. Therefore, we can fairly accurately determine that 11.5% (265,241 molecules/2.3 x lo6 molecules) of Rac is translocated to the membrane in PMAstimulated neutrophils after 6 min of stimulation. This value is in good agreement with the estimates from the sucrose gradients (7-10%). Using the data obtained for the 6-min time points from PMA-and fMLP-stimulated cells (see Fig. 4 and "Results") and the kinetics data from Figs. 2 and 3, one can calculate the amount of translocation at each time point. For example, the amount of translocated proteins calculated from Fig.  2 for 20 min of PMA stimulation is -500,000 molecules, and this value agrees (*lo%) to that determined by quantitative blotting of this sample (data not shown).
It is of interest that the translocation of Radp47-phodp67phox closely follows the kinetics of activation of the NADPH oxidase and that termination of translocation is associated with termination of superoxide production. These results confirm recent reports showing that continuous translocation (45) of p47-and p67-phox and continuous association with the active 0;-generating complex (46) is necessary to maintain linearity of the respiratory burst. The translocated proteins are still stably associated with the membrane after the oxidase is turned off, indicating that inactivation of the oxidase is not solely due to the cytosolic factors falling back off of the membrane. Our findings raise the possibility that Rac may play an additional role in the assembled oxidase complex of regulating 0 ; production.