p21-activated kinase (Pak) regulates NADPH oxidase activation in human neutrophils

Phagocytes represent the first line of cellular host defense against microorganisms. This function relies in part on the ability of these cells to generate large amounts of reactive oxygen species, including superoxide anion (O₂^·-), hydroxyl radical (OH^-), and hydrogen peroxide (H₂O₂). This phenomenon, known as the respiratory burst, results from activation of a multicomponent, O₂^·--producing enzyme known as the nicotinamide adenine dinucleotide phosphate (NADPH) oxidase. In resting phagocytes, the NADPH oxidase complex is dormant, separated into individual cytosolic and membrane-bound components.^1,2  Upon activation by a variety of stimuli, such as chemotactic peptides, phorbol esters, or opsonized particles, the cytosolic factors translocate to the membrane and associate with flavocytochrome b₅₅₈, a heterodimer composed of gp91^phox and p22^phox. The 4 minimally required cytosolic factors of the NADPH oxidase include p67^phox, p47^phox, p40^phox, and the Rho guanosine triphosphatase (GTPase) Rac.¹ 

Cell-free assay systems, in vivo experiments in Rac2 knock-out mice, as well as the clinical phenotype caused by the presence of a Rac2 mutation in an immunocompromised patient have confirmed the crucial role of Rac2 in O₂^·- generation by phagocytes.^3-6  Stimulation of Rac activity, followed by association of the active GTPase with the NADPH oxidase complex, coordinates essential steps of the electron flow and facilitates activation of downstream targets. NADPH oxidase activation is highly regulated, involving a number of phosphorylation events, conformational changes in oxidase components, and protein-protein interactions.¹  While all of the oxidase components have been reported to be targets of phosphorylation and several protein kinases have been implicated in various assay systems, the actual role of all these events in regulating the oxidase in intact neutrophils remains unclear.

We have previously shown that several isoforms of p21-activated kinases (Paks) are rapidly and transiently activated in chemoattractant-stimulated neutrophils.^7,8  Furthermore, we found Pak activation was dependent on heterotrimeric G-protein coupling to the N-formyl-methionyl-leucyl-phenylalanine (fMLF) receptor and occurred in concert with Rac activation. Subsequently, Dharmawardhane et al reported that Pak localizes to membrane ruffles and lamellipodia at the leading edge of polarized cells in fMLF-stimulated leukocytes.⁹  This localization closely resembles the pattern of Rac redistribution during NADPH oxidase activation. Pak activity and protein-protein interactions facilitated by proline-rich regions in the N-terminal domain of Pak are considered regulatory elements for actin reorganization and are involved in important neutrophil functions, such as chemotaxis and phagocytosis.¹⁰  For example, Pak regulates a number of cellular functions through phosphorylation of multiple targets, predominantly on serine residues surrounded by lysine or arginine residues. These targets include mitogen-activated protein kinase kinase-1 (MEK-1), Raf-1, LIM kinase, myosin light chain kinase, stathmin, merlin, Bad, RhoGDI (Rho guanosine diphosphate dissociation inhibitor), phosphoglycerate mutase B, and phosphoglucomutase.¹⁰ 

While several isoforms of Pak are highly expressed and rapidly activated by chemoattractants in neutrophils, participation of Pak in regulation of O₂^·- production by the NADPH oxidase has not been firmly established. Previously, we reported that the NADPH oxidase component p47^phox might represent a potential target for regulatory phosphorylation by activated Pak and that Pak-induced phosphorylation of the glycolytic enzyme phosphoglycerate mutase B (PGAM-B) increases O₂^·- generation.¹¹  However, further understanding of the role of Pak in NADPH oxidase function has been limited due to the absence of Pak-specific, small molecule inhibitors and the technical difficulties associated with efficient gene transfer into phagocytes. Thus, to address this issue, we used an HIV-Tat–mediated protein transduction approach to investigate the role of Pak in regulating the neutrophil NADPH oxidase in intact cells and provide direct evidence to substantiate a regulatory role for Pak in oxidase function. Furthermore, our data suggest that the basis of Pak-mediated regulation of the NADPH oxidase involves regulatory phosphorylation of the oxidase component p47^phox and direct association with flavocytochrome b₅₅₈.

The chemoattractant peptide fMLF was purchased from Sigma Chemical (St Louis, MO) and Molecular Probes (Eugene, OR). [γ-³²P] adenosine triphosphate (ATP) and [³²P] H₃PO₄ were from ICN (Irvine, CA) and Amersham-Pharmacia (Piscataway, NJ), respectively. Anti-p47^phox antibody was from BD Biosciences (San Diego, CA), anti–phospho Pak antibody was from Cell Signaling Technology (Beverly, MA), and anti-Pak antibodies,^7,8  anti-p22^phox, and anti-gp91^phox antibodies¹²  have been described previously. Anti–epitope tag antibodies or secondary antibodies were purchased from Covance (Berkeley CA), Promega (Madison, WI), and Molecular Probes. Peptides encompassing C-terminal sequences of p47^phox were synthesized and purified by the Scripps Research Institute Peptide facility.

The pTAT-HA bacterial expression vector was provided by S. Dowdy (Howard Hughes Medical Institute and University of California at San Diego) (Vocero-Akbani et al¹³ ) and used to insert sequences for green fluorescent protein (GFP), Pak–Pak-inhibitory domain (PID) (amino acids [aa] 83-149), PID L107F, and Rac2 T17N by polymerase chain reaction (PCR). The plasmids were validated by sequencing and used for transformation of Escherichia coli BL21(DE3) LysS competent cells (Invitrogen, Carlsbad, CA). Expression and Tat protein induction were essentially as described.¹³  Recombinant proteins were isolated from the bacterial pellet by sonication and denaturation in 8 M urea. Proteins were purified over a nickel-nitrilotriacetic acid (Ni²⁺-NTA)–agarose affinity column, followed by a fast performance liquid chromatography (FPLC) purification and/or desalting gel filtration. To assist in protein refolding, the urea concentration of Tat-PID and Tat-PID L107F was slowly lowered during the last 2 purification steps. Identity and purity of the proteins were assessed by Coomassie blue staining and immunoblotting. For some experiments, Tat proteins were labeled with fluorescein isothiocyanate (FITC) for 2 hours at room temperature (RT), and unbound FITC was removed with a Sephadex G-25 column. Expression and purification of glutathione-S–transferase (GST) fusion proteins, including GST-Pak1 and GST-p47^phox, have been described previously.⁷  Cytochrome b₅₅₈ was purified from human neutrophil membranes as described previously.¹⁴  Myc-tagged Pak1 plasmids for mammalian expression were prepared as reported in Shalom-Barak and Knaus.¹¹  Single and multiple mutations of p47^phox were introduced with the Quikchange Site-directed mutagenesis kit (Stratagene, La Jolla, CA) and verified by sequencing. Wild-type and mutant p47^phox were also subcloned into pJ₃H for mammalian expression. Endotoxin levels in the purified protein preparations were less than 0.01 EU/mL when measured using a commercially available limulus amebocyte lysate assay (Bio-Whittaker, Walkersville, MD).

Neutrophils were isolated under endotoxin-free conditions from heparinized venous blood of healthy volunteers by 2-step sedimentation on dextran and Ficoll-Hypaque.³  The purity was consistently higher than 95%. Monocyte isolation from human blood was performed as described.¹⁵  Monocytes were allowed to attach to coverslips for 2 hours, and nonadherent cells were removed by gentle washing. Adherent monocytes were cultured for 10 days in RPMI 1640 containing 10% heat-inactivated fetal bovine serum (FBS; Invitrogen) and recombinant macrophage colony-stimulating factor (rM-CSF, 10 g/mL), and medium was changed every 3 days. HL60 cells were obtained from American Type Culture Collection (ATCC, Rockville, MD) and differentiated with 15 ng/mL phorbol myristate acetate in RPMI 1640 containing 20% FBS for 4 days.

Neutrophils (1-5 × 10⁵) were suspended in RPMI and preincubated with Tat proteins or bovine serum albumin (BSA) for 10 minutes and washed with phosphate-buffered saline (PBS) before stimulation with fMLF (1 μM) or phorbol myristate acetate (PMA, 100 ng/mL). O₂·^- generation was measured using luminol-based chemiluminescence. Neutrophils were preincubated in 25 mM Hepes (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid, pH 7.5), 118 mM NaCl, 4.7 mM KCl, 1.2 mM KH₂PO₄, 1.2 mM MgSO₄, 5.5 mM glucose, and 1.0 mM CaCl₂ (KRHG) or in RPMI at 37°C for 5 minutes before adding 71 μM luminol and 50 μg/mL horseradish peroxidase. Stimulation with fMLF or PMA was initiated after 1 to 2 minutes of baseline recording by injecting the stimulus. Chemiluminescence was recorded every 20 seconds for up to 15 minutes, and results are expressed as relative light units (RLU). Experiments were performed 3 times in triplicates using a Wallac Berthhold plate luminometer (Gaithersburg, MD).

Cos7gp91^phox/p22^phox cells or Cos7p22^phox cells were maintained in Dulbecco modified Eagle medium (DMEM) low glucose supplemented with 10% FCS, 2 mM glutamine including 0.8 mg/mL neomycin (p22^phox), and 2 μg/mL puromycin (gp91^phox). Cells were transiently transfected with pCVM6, pCVM6-Pak1 wt, pCVM6-Pak1 L107F T423E, pCVM6-Pak1 (aa 1-201), or pCVM6-Pak1 (aa 206-545), alone or in combination with pJ₃H-p47^phox or pJ₃H-p47^phox S303, 304, 320, 328A using Lipofectamine Plus according to the manufacturer's instructions (Invitrogen). After 24 to 40 hours, cell lysates were prepared in 50 mM Tris (tris(hydroxymethyl)aminomethane, pH 7.5), 100 mM NaCl, 5 mM MgCl₂, 1 mM EDTA (ethylenediaminetetraacetic acid), 1% nonidet P-40 (NP-40), and 10% glycerol, and subjected to immunoprecipitation with either anti-Myc (for Pak), anti-p22^phox, or anti-HA (for p47^phox) antibodies. Immunoblots were probed with the corresponding antibodies to detect coimmunoprecipitation. For some experiments, cells were labeled with [³²P] H₃PO₄ (50 μCi/mL [1.85 MBq]) overnight in phosphate-free DMEM containing 0.5% dialyzed FBS, and phosphate incorporation into p47^phox was evaluated after immunoprecipitation of the protein from cell lysates, gel electrophoresis, and subsequent autoradiography of the gel.

Phosphorylation of peptide sequences derived from the C-terminus of p47^phox were carried out using recombinant, constitutively active GST-Pak1, as described,^7,11  followed by scintillation counting to determine radioactivity bound to phosphocellulose filter units (Pierce, St Louis, MO). For 2-dimensional (2-D) phosphopeptide mapping, in vitro kinase reaction-swith purified, recombinant GST-Pak1 using GST, GST-p47^phox, or GST-p47^phox mutant proteins as substrates were performed. The purity of proteins was assessed by Coomassie blue staining, and protein was quantified with the bicinchoninic acid (BCA) assay system. The phosphorylation reactions were performed in kinase buffer (50 mM Hepes [pH 7.5], 10 mM MgCl₂, 2 mM MnCl₂, 0.2 mM dithiothreitol [DTT], 50 μM ATP) with 10 μCi (0.37 MBq) [³²P]ATP/reaction and were analyzed using gel electrophoresis, blotting onto nitrocellulose, and detection by autoradiography. Band intensities were quantified by liquid scintillation counting. Tryptic digestion and phosphopeptide mapping were performed as described.^16,17 

HL60 cells were differentiated with PMA on coverslips for 4 days. The cells were carefully washed in warm medium without phenol red and incubated with buffer or 1 μM fMLF at 37°C for 1 minute. The reaction mixture was removed and cells were immediately fixed for 20 minutes in 2% paraformaldehyde. Following fixation, the cells were washed twice for 10 minutes in PBS, permeabilized for 5 minutes in 0.5% Triton X-100/PBS, washed again with PBS, and blocked with an avidin/biotin blocking kit (Vector Laboratories, Youngstown, OH) for 30 minutes. Cells were then immediately incubated for 1 hour in combinations of affinity-purified anti-Pak antibody and anti-p47^phox monoclonal antibody diluted in 2% FBS/PBS. Following this incubation, the cells were washed twice for 10 minutes in PBS, incubated for 1 hour with biotinylated donkey anti–rabbit immunoglobulin G (IgG; Jackson ImmunoResearch, West Grove, PA), washed twice for 10 minutes in PBS, incubated in the dark in fluorescent-conjugated goat anti–mouse IgG (Alexa Fluor 488; Molecular Probes) and streptavidinrhodamine red-X (Jackson ImmunoResearch), washed again with PBS, and fixed to glass slides using the Prolong Antifade Kit (Molecular Probes). Stained cells were examined with a × 63/1.4 NA objective under a Zeiss Axiovert S100TV confocal microscope (Carl Zeiss, Heidelberg, Germany) equipped with a krypton/argon mixed gas laser, visualized with Bio-Rad LaserSharp (v3.2) software (Bio-Rad, Hercules, CA), and processed using Adobe Photoshop (Adobe Systems, San Jose, CA). For human macrophages, stimulation with PMA was 3 minutes, followed by the procedure as described above. These cells were incubated with affinity-purified anti-Pak1 and anti-p22^phox monoclonal antibody. For transfected Cos7p22^phox cells, anti-Myc monoclonal antibody and anti-p22^phox polyclonal antibody were used.

The elucidation of intracellular signaling mechanisms in primary human neutrophils is hindered by the short life span and low transfection efficiency of neutrophils, and functional studies are primarily performed using pharmacologic inhibitors or inferring data from myeloid precursor cell lines, such as K562 cells. Since neither of these approaches is suitable for studies involving Pak function, we used a protein transduction method to study the involvement of Pak in O₂^·- generation. This approach relies on a short polybasic sequence derived from the HIV Tat protein to introduce proteins into intact cells.^18-20 

Currently, the best method to suppress receptor-stimulated Pak kinase activity without interfering with other downstream effectors of active Rac or Cdc42 is the expression of PID, which inhibits activation of 3 Pak isoforms (Paks 1-3).^21-23  FITC-labeled PID-Tat fusion protein was incubated with human neutrophils, and cellular uptake was evaluated by confocal microscopy after removal of externally bound protein. Nearly 100% of the neutrophils showed high levels of incorporated FITC-PID-Tat, which was predominantly cytosolic (Figure 1A). Furthermore, immunoblotting for supernatants and cell lysates from PID-Tat–treated neutrophils confirmed uptake of the fusion protein (Figure 1B).

Since Tat fusion proteins are prepared by expression and purification from E coli, the potential for a formyl-methionine addition to the amino terminal sequence exists.²⁴  To investigate the possibility of N-terminal modifications on purified Tat proteins, we analyzed their amino terminal start sequence MRGS (methionine-arginine-glycine-serine). According to studies by Hu et al and others, the formyl group will be removed from methionine in this sequence, while the methionine itself could be partially retained.^25-30  To evaluate this possibility, we performed functional competition binding studies on neutrophils using FITC-labeled fMLF and Tat protein. Cells were incubated with either fluorescently labeled peptide or Tat protein on ice and analyzed by flow cytometry. This procedure will prevent immediate uptake of Tat protein or internalization of the occupied formyl peptide receptor (FPR). The differences in percentage of fluorescent cells incubated with FITC-fMLF peptide alone or with FITC-fMLF peptide after Tat protein treatment was between 4% and 5%, which is in the range of background binding. Thus, these results indicate that Tat fusion proteins expressed from the described plasmid sequence and purified according to the published protocols will not bind to the FPR and alter FPR-mediated signaling by virtue of E coli–generated modifications.

To determine whether PID-Tat was properly refolded and retained its expected biologic activity, neutrophils were treated with PID-Tat or control GFP-Tat and stimulated with fMLF. Normal activation of Pak was detected in GFP-Tat–transduced neutrophils, whereas Pak activation was completely inhibited when PID-Tat was incorporated into neutrophils (Figure 1C). The peak of Pak activation occurred one minute after receptor stimulation and was followed by rapid down-regulation. This time course closely resembles Pak activation in nontransduced cells⁷  and shows, together with the PID-mediated inhibition of this response, that introduction of PID-Tat fusion protein is a valid approach to elucidate Pak function in neutrophils.

To investigate the role of Pak in regulating O₂^·- generation by the NADPH oxidase, we transduced cells with PID-Tat, as well as dominant-negative PID L107F-Tat and a dominant-negative Rac2 T17N-Tat mutant. It should be noted that we encountered substantial variation in respect to yield and aggregation of the PID and PID mutant constructs. These properties of PID-Tat proteins did not allow protein concentrations above 1 to 2 μM or freezer storage and necessitated continuous purification and immediate use of the proteins. Since preincubation of GFP-Tat with neutrophils for 10 minutes at RT before addition of fMLF resulted in slightly increased O₂^·- production (10%-20% depending on the protein and neutrophil preparation) (Figure 2A), the oxidative burst observed after incubation with GFP-Tat was used as the experimental control. Incubation of neutrophils with Rac2 T17N-Tat, a dominant-negative Rac mutant, decreased superoxide dismutase (SOD)–inhibitable O₂^·- production by approximately 50%, while transduction of cells with PID-Tat protein at the same concentration showed a 30% reduction (Figure 2B). In contrast, no inhibition was observed in neutrophils pretreated with inactive PID mutant protein (PID L107F-Tat) before fMLF stimulation (Figure 2B). The inhibition of fMLF-induced O₂^·- generation by PID-Tat protein was strictly dose dependent (Figure 2C), reaching close to 50% inhibition at a concentration of 2 μM PID-Tat. Under the same conditions, the oxidative burst remained unchanged when PMA was used as stimulus (data not shown). These results are consistent with our previous data indicating that PMA-mediated NADPH oxidase activation is Pak-independent.^7,8 

A correlation between the serine-threonine kinase activity of Pak and serine phosphorylation of p47^phox was suggested by our previous data.⁷  Since stimulus-induced phosphorylation of multiple serine residues at the C-terminus of p47^phox is essential for oxidase activation,^31-33  we tested multiple serine-containing peptides located in the C-terminal region of p47^phox to investigate potential Pak phosphorylation sites. Recombinant, constitutively active Pak phosphorylated 3 different peptides containing 6 serine residues located from amino acids 300 to 331 in p47^phox (Figure 3A). Furthermore, coexpression of active Pak and p47^phox in Cos cells resulted in incorporation of radioactive phosphorus into p47^phox (data not shown), and full-length recombinant p47^phox protein was phosphorylated by active GST-Pak in vitro (Figure 3B), indicating substantial Pak-mediated phosphorylation of intact p47^phox. Phospho–amino acid analysis of the radioactive p47^phox band indicated phosphorylation of serine residues, and tryptic digests of phosphorylated p47^phox, followed by 2-D phosphopeptide mapping, revealed multiple phosphorylation sites (Figure 3C). We used a combination of 2-D mapping followed by high-performance liquid chromatography (HPLC), matrix-assisted laser desorption/ionization–mass spectrometry (MALDI-MS), and mutational analysis to identify the location of these phosphoserines in p47^phox and found that serines 303, 304, 320, and 328 were targets of Pak-induced phosphorylation. Consistent with our peptide-based experiments (Figure 3A), the highest incorporation of ³²P was detected at serine 328. Note that we also found several phosphopeptide spots that originated from incomplete cleavage products of the same peptide. To confirm the phosphorylation sites identified, we generated a quadruple p47^phox S→A mutant protein and showed that this protein was not phosphorylated by active GST-Pak in vitro nor was it phosphorylated in vivo in Cos cells coexpressing constitutively active Pak (Figure 3B and data not shown).

It is conceivable that Pak activity is required for the phosphorylation of NADPH oxidase components other than p47^phox, and Ahmed et al reported that Pak phosphorylated p67^phox protein fragments.³⁴  However, we did not detect substantial phosphorylation of p67^phox by activated GST-Pak in vitro using recombinant p67^phox or in Cos cells coexpressing p67^phox and active Pak (data not shown).

Protein kinases can associate with their phosphorylation targets, as we have previously shown for Pak and PGAM.¹¹  Since recombinant p47^phox protein bound nonspecifically to various affinity beads (U.G.K., unpublished observations, 1999), we evaluated the potential interaction of Pak with p47^phox in intact cells. Wild-type Pak, as well as dominant-negative and constitutively active Pak mutants were coexpressed with wild-type p47^phox in Cos cells, followed by immunoprecipitation of Pak from cell lysates. Independently of the Pak construct used, we were not able to detect stable complex formation between Pak and p47^phox. Therefore, we analyzed possible colocalization of Pak with p47^phox in HL-60 cells, which had been differentiated with phorbol ester to a macrophage-like cell. Confocal microscopy analysis of differentiated HL-60 cells, treated with fMLF to stimulate Pak and the NADPH oxidase, showed stimulus-dependent colocalization of Pak and p47^phox in plasma membrane regions (Figure 4). Thus, while Pak and p47^phox do not appear to form a stable complex, their colocalization is consistent with the conclusion that p47^phox is a physiologic substrate of Pak.

Since stimulation with chemoattractant led to a significant translocation of Pak to the plasma membrane, but apparently not to a stable interaction with p47^phox, we hypothesized that Pak might interact directly with flavocytochrome b₅₅₈. To evaluate this possibility, we transfected Cos7 cell lines stably expressing gp91^phox and p22^phox or p22^phox alone with Pak wild-type and mutant constructs. We found that constitutively active Pak1, either as Pak1 mutant or Pak1 C-terminal domain, associated specifically with p22^phox but not with gp91^phox (Figure 5A). Furthermore, experiments using Cos7 cells stably expressing p22^phox alone confirmed that active Pak interacts with p22^phox independently of the presence of gp91^phox (data not shown). A direct association of active Pak with flavocytochrome b₅₅₈ was further established by the specific binding of the purified flavocytochrome b to purified, kinase-active GST-Pak1 but not to GST protein (Figure 5B).

Biochemical and structural studies have demonstrated that partial phosphorylation of p47^phox relieves an autoinhibitory intramolecular interaction, which then permits phospho-p47^phox binding to the cytoplasmic tail of p22^phox.³³  Thus, binding of p47^phox and active Pak to p22^phox must occur simultaneously without competition for the same binding site. To address this issue, we cotransfected active Pak1 or the constitutively active C-terminus of Pak1 with p47^phox into flavocytochrome b–expressing Cos7 cells and treated the cells with either vehicle control or PMA to induce p47^phox translocation to the plasma membrane. Immunoprecipitation of p22^phox revealed that the interaction of Pak with p22^phox was not affected by the presence of p47^phox or PMA stimulation of the cells (Figure 6A). To ensure p47^phox binding to p22^phox under our conditions, the association of p47^phox with p22^phox was determined. Our data clearly demonstrate a substantial increase in p47^phox/p22^phox complex formation when the cells were stimulated with PMA (Figure 6B).

The active NADPH oxidase complex localizes in immune cells to the plasma membrane or phagolysosome. To establish the cellular compartment of the Pak/p22^phox association, we initially determined the localization of wild-type Pak or constitutively active Pak in flavocytochrome b–expressing Cos cells. Wild-type Pak was widely dispersed over the cytoplasm, whereas p22^phox was localized to endomembranes, vesicles, and some plasma membrane sites. Introduction of kinase-active Pak1 initiated the translocation of a pool of Pak1 to the plasma membrane, where active Pak colocalized with p22^phox (Figure 7, top panel). Control experiments using p22^phox-deficient NCI H292 cells clearly demonstrated that both wild-type and kinase-active Pak localize in unstimulated cells to the cytosol (data not shown). To assess the localization of Pak in primary, chemoattractant-stimulated cells, human monocytes were isolated and differentiated to macrophages in vitro. In fMLF-stimulated macrophages, a substantial increase of plasma membrane–bound p22^phox was observed in areas resembling membrane ruffles (Figure 7, bottom panel). Concurrently, Pak translocated from the cytosol to the periphery, overlapping to a large extent with p22^phox staining. The fMLF-stimulated colocalization of Pak with p22^phox was most pronounced in well-spread, highly ruffling macrophages and resembled the colocalization of Pak with p47^phox in fMLF-stimulated macrophage-like cells (Figure 4).

Primary, terminally differentiated hematopoietic cells are exceedingly refractory to gene transfer. Neutrophils in particular have a very short life span and display negligible protein synthesis under quiescent conditions. To address this problem, methods for protein transduction into intact cells have recently been developed. More specifically, Dowdy and others demonstrated that recombinant proteins fused to a Tat leader sequence will enter the majority of mammalian cell types and retain their biologic functions upon intracellular refolding (Vocero-Akbani et al¹³ ; Ho et al¹⁹ ; and Schwarze et al²⁰ ). Using the protein transduction domain of HIV-1 Tat as a carrier for recombinant proteins, we observed an uptake efficiency of almost 100% in myeloid cells. Therefore, this approach allows altering signaling pathways in neutrophils when specific pharmacologic inhibitors are not available. In this study, we used Tat-mediated entry of an inhibitory domain of Pak to investigate the role of Pak in regulating O₂^·- generation by the phagocyte NADPH oxidase.

We addressed the effects of Pak in the context of chemotactic receptor signaling stimulated by bacterially derived N-formyl peptide agonist (fMLF), an event where rapid and transient Pak activation coincides with a burst of O₂^·-. Inhibition of Pak kinase activity in human neutrophils reduced fMLF-stimulated O₂^·- generation substantially and in a dose-dependent fashion, while a Tat fusion protein with the inactive Pak inhibitory domain did not affect O₂^·- generation. Thus, these data demonstrate that Pak activation not only correlates with oxidase activation but is directly involved in regulation of this enzyme complex.

The critical role of Rac2 in regulation of the NADPH oxidase has been established in cell-free studies, using neutrophils from Rac2-null mice, and through a dominant-negative Rac2 mutation identified in a pediatric patient with recurrent bacterial infections.⁵  Interestingly, Rac2 T17N-Tat protein transduced into neutrophils did not abolish O₂^·- production as predicted. The 50% to 60% inhibition of the fMLF-induced oxidative burst by dominant-negative Rac2 N17-Tat may be explained by incomplete refolding of a fraction of the Tat proteins to their active conformation inside the cell, thereby altering the bioavailability of the protein. Immunofluorescence microscopy of transduced PID-Tat protein showed some degree of intracellular protein aggregation, a phenomenon that could decrease the amount of bioactive protein. Additionally, the propensity of some Tat fusion proteins to form aggregates in solution did not allow us to use PID-Tat protein above a concentration of 1 to 2 μM.

A role for Pak in O₂^·- generation was proposed by our initial studies in neutrophils, by studies using Rac effector domain mutants in a Cos cell model, and by partial purification of a protein kinase with similar biochemical features.^7,35,36  One of several potential mechanisms to explain how Pak regulates O₂^·- production in neutrophils involves the phosphorylation of NADPH oxidase components. While we did not observe Pak-mediated phosphorylation of p67^phox or p22^phox, we detected a high level of p47^phox phosphorylation. The significance of p47^phox phosphorylation in conformational changes of the protein, association with other oxidase components, and the overall output of O₂^·- is well documented, and multiple serines have been identified as phosphorylation sites in stimulated, intact neutrophils.³⁷  Ago et al demonstrated in cell-free assay systems and in a model cell line that phosphorylation of serines 303, 304, and 328 is required for O₂^·- production.³⁸  In the present study, we used in vitro phosphorylation and coexpression experiments in radiolabeled cells to identify serines 303, 304, 320, and 328 of p47^phox as amino acids targeted for phosphorylation by Pak.

Several protein kinases have been implicated previously in the phosphorylation of p47^phox, including protein kinase C isoforms, mitogen-activated protein (MAP) kinases, phosphatidic acid–activated kinase, and the phosphatidylinositol-3 kinase target Akt.^39-43  In the present study, we conclusively show that Pak-mediated phosphorylation of p47^phox occurs at multiple sites relevant for O₂^·- generation and contributes to the regulation of oxidase activity in intact cells. However, because of the multiple kinases involved in p47^phox phosphorylation in neutrophils and the potential for some redundancy in this process, it was not possible to provide absolute confirmation of the essentiality of Pak-mediated phosphorylation of p47^phox phosphorylation in oxidase activation in intact, fMLF-stimulated neutrophils. Indeed, it is clear that p47^phox activation can be initiated by phosphorylation of only a subset of sites that is known to be phosphorylated in vivo, and removal of individual kinases or kinase targets does not necessarily eliminate p47^phox function. Likewise, treatment alone with individual kinases in vitro can fully activate p47^phox.^41-43  Thus, it is plausible that the neutrophil uses multiple kinases, including Pak, to ensure rapid and efficient phosphorylation of p47^phox, leading to rapid assembly of the oxidase. Furthermore, the essential role of Rac in oxidase activation and its demonstrated regulation of Pak strongly implicated Pak as a physiologically relevant mediator in the phosphorylation events.^7,35  Combined with these previous studies, the data presented here provide further direct evidence for a key role of Pak in NADPH oxidase regulation in intact cells via its involvement in p47^phox phosphorylation. While Tat transduction methodology resulted in a significant 40% to 50% reduction in oxidase activity, the low efficiency of neutrophil labeling with ³²P and concomitant preactivation during this procedure did not allow us to detect subtle differences in p47^phox phosphopeptides in stimulated neutrophils. Additionally, we were also not able to generate HL60 cell lines stably expressing PID, presumably due to the requirement of Pak activity for cell proliferation. Thus, future studies in neutrophils obtained from Pak isoform knock-out animals will be required to absolutely delineate the regulatory role of Pak in NADPH oxidase activation and its relation to the multiple kinases known to phosphorylate p47^phox in vitro and in vivo.

Our studies suggest that Pak may actually play multiple roles in the regulation of the phagocyte oxidase. It is likely that Rac activation and movement of active Rac to the plasma membrane stimulate Pak activity and possibly also affect its localization. Notably, a direct association of catalytically active Pak with p22^phox at the plasma membrane was observed in this study. Stimulus-dependent translocation of Pak and association with flavocytochrome b₅₅₈ may either precede p47^phox translocation leading to additional phosphorylation at the membrane or may occur simultaneously. In light of structural data³³  indicating that certain phosphorylations are required for p47^phox association with p22^phox, Pak-mediated p47^phox phosphorylations and translocation of both proteins are likely part of a rapid, highly synchronized process. One can envision the presence of a large multimeric signaling complex on membranes of activated neutrophils, where flavocytochrome b₅₅₈ and potentially Rac2 serve as the anchor, and proteins are held together by multiple protein-protein interactions. Additionally, Paks may regulate the NADPH oxidase at several levels, including phosphorylation of oxidase components, glycolytic enzymes, and structural or regulatory elements of the cytoskeleton, as well as by association with protein binding partners in specific membrane compartments, where formation of signaling networks occurs.

We thank Steve Dowdy for reagents and helpful discussions. We are grateful to Wolfgang Fischer for HPLC peptide sequence analysis and Charles King for MALDI-MS analysis. We also thank Abina Mulcahy and Monica Valo for technical assistance, Becky Diebold and Jonathan Chernoff for technical advice, and Katrina Schreiber for administrative assistance.