Platelet-Activating Factor-Mediated Endosome Formation Causes Membrane Translocation of p67^phox and p40^phox That Requires Recruitment and Activation of p38 MAPK, Rab5a, and Phosphatidylinositol 3-Kinase in Human Neutrophils¹

Clathrin-mediated endocytosis (CME)³ occurs constitutively in all mammalian cells and modulates signal transduction by controlling the levels of cellular receptors by the rapid clearance and down-regulation of activated receptors (1, 2). Ligand activation of G protein-coupled receptors (GPCRs) concentrates the ligand-receptor complexes into clathrin-coated pits in the plasma membrane followed by the formation of clathrin-coated vesicles or endosomes (1). Subsequent endosome formation does not occur randomly but appears to take place in endocytotic hot spots, which appear to be constrained by the actin cytoskeleton (3, 4). These endosomes contain assembly proteins that aid in clathrin cage assembly; however, only assembly protein-2 is important for endocytotic clathrin-coated vesicle (CCV) formation at the plasma membrane (1, 5, 6). The other adaptor proteins are important for vesicle formation involving intracellular organelles (1, 5, 6).

The GTPase dynamin regulates endosome formation through either a molecular garrote that causes pinching off of the membranes to form the endosome or via an inherent mechanochemical activity in which dynamin molecules form spirals in the presence of GTP, which function as a coiled spring and upon release propel the newly formed endosome into the cytosol (1, 7, 8, 9, 10). Irrespective of its exact function, dynamin activation is required for CME, and endosome formation requires the recruitment of a number of specific proteins including the small GTPases, especially Rab5a, which has defined roles in the development of the soluble NSF attachment protein receptor (SNARE) complex, a prerequisite for vesicle fusion, and is also known to be important in endosome motility (8, 11, 12, 13). Additionally, Rab5a also causes the recruitment of early endosome Ag-1 (EEA-1), a tethering factor known to be involved with intracellular trafficking (14).

Signaling through GPCRs is essential to all cells but especially to leukocytes, which must respond rapidly to infection (1, 2, 15, 16, 17). Neutrophils (polymorphonuclear leukocytes, PMNs) are a vital part of innate immunity and respond to a variety of chemoattractants via cell surface GPCRs, allowing PMNs to marginate to the tissues and migrate to the site of infection to eradicate the pathogens (1, 4, 15, 16). This migration to the tissues causes PMN priming resulting in physiologic changes in the PMN, including: changing the phenotype from nonadhesive to adhesive, alteration in the actin cytoskeleton manifested as changes in PMN morphology, and augmentation of the oxidative response to a given stimulus through translocation of specific cytosolic oxidase components to the plasma membrane (18, 19, 20, 21, 22, 23). Platelet activating factor (PAF) is a prototypical lipid chemoattractant that binds to a specific seven-transmembrane domain receptor (PAFR) (24). PAF occupation of this receptor results in release of the Gα_i and Gα_q heterotrimeric subunits, which mediate chemotactic, cytotoxic, and cross-regulatory signals as well as initiating inhibitory signals, β-arrestins and G protein receptor kinases, to desensitize these activated receptors to repeated stimulation (25, 26). Ligand occupancy of many GPCRs initiates signal transduction through recruitment of the β-arrestins and CME, with β-arrestins being responsible for both transduction and termination of signaling events from the activated receptor (1, 27). Previous work has demonstrated that activation of the PAFR induces CME, recruits a β-arrestin-1 scaffold for assembly of the p38 MAPK signalosome, and activation of p38 MAPK causes the formation of actin bundles at the PAFR, all before activation of dynamin-2 at the PMN membrane (4, 28). We hypothesize that PAF ligation of the PAFR induces dynamin-2 scission of the endosome from the plasma membrane and recruitment of PI3K, Rab5a, EEA-1, and the p38 MAPK signalosome to the nascent endosome, triggering phosphorylation and translocation of the cytosolic 67-kDa phagocyte oxidase protein (p67^phox) with the 40-kDa phagocyte oxidase protein (p40^phox) to the plasma membrane.

Unless otherwise indicated, all chemicals were purchased from Sigma-Aldrich. All reagents were endotoxin-free, and all solutions were made from sterile water (United States Pharmacopeia). Acrylamide, N,N′- methylene-bis-acrylamide, and N,N,N′,N′-tetramethylethylenediamine (TEMED) were obtained from Bio-Rad. ECL reagents were obtained from Amersham Biosciences. Nitrocellulose paper, x-ray film, and Nunc 96-well plates were purchased from Life Science Products. Abs to the PAFR, dynamin-2, ASK1, MAPK kinase-3 (MKK3), Rab5a, Rab GDP dissociation inhibitor (RabGDI), EEA-1, p40^phox, p47^phox, p67^phox, gp91^phox, Rac2, PI3K, both the 85-kDa regulatory and the 110-kDa catalytic domains, Akt1, secondary anti-goat IgG-HRP, anti-mouse IgG-HRP, and Protein A/G PLUS-Agarose beads were purchased from Santa Cruz Biotechnology. Abs to p38 MAPK, dual phosphorylated p38 MAPK, p42/44 MAPK, dual phosphorylated p42/44 MAPK, and secondary anti-rabbit IgG-HRP were obtained from Cell Signaling Technology, and phosphoserine (PS) and phosphotyrosine (PY) Abs were purchased from Invitrogen and Abcam, respectively. Anti-PAFR for immunohistochemistry was obtained from Cayman Chemical. Reagents for electron microscopy, including cacodylate-buffered glutaraldehyde, immunogold particles, cacodylate-buffered osmium tetroxide, 2% BSA-5% normal donkey serum, lead citrate, EMbed 812, and 8-well Permanex chambered slides (Nunc), were purchased from Electron Microscopy Sciences.

Heparinized whole blood was drawn from healthy human volunteers by venipuncture consent using a protocol approved by the Combined Internal Review Board and Human Subject Committee at the University of Colorado School of Medicine, and PMNs were isolated as previously described (29). Cells were resuspended to a concentration of 2.5 × 10⁷ cells/ml in Krebs-Ringer phosphate buffer (pH 7.35) with 2% dextrose and used immediately for all subsequent manipulations.

PMNs (1.25 × 10⁷ cells (whole-cell lysate and immunoprecipitation) or 1 × 10⁸ cells (subcellular fractionation)) were incubated with buffer or 2 μM PAF for 1 min. Reactions were stopped with the addition of ice-cold relaxation buffer (10 mM PIPES (pH 7.4), 3 mM NaCl, 100 mM KCl, 3.5 MgCl₂, 1.2 mM EGTA, 10 μg/ml leupeptin, 40 mM sodium orthovanadate, 1 M nitrophenylphosphate, and 50 μg/ml PMSF) and immediately sonicated (2 × 30 s). Lysates were cleared and used for Western blot analysis, immunoprecipitation, or subjected to subcellular fractionation as previously described (29, 30). Purity of fractions was determined using p47^phox as a cytosol marker and the PAFR as a membrane marker (data not shown). Proteins were separated with 7.5 or 10% acrylamide gel with SDS-PAGE and transferred to a nitrocellulose membrane; these membranes were cross-linked and probed with the specific Abs indicated in the text or figure legends.

Primary Abs of interest against signaling proteins or isotype controls were labeled by first solubilizing the dye of interest (Molecular Probes) in methanol and drying under N₂. The dye is resuspended in DMSO and conjugation occurs in a mixture of 10 μl dye + 20 μl NaHCO₃ solution (pH 8.3) + 180 μl Ab (2 mg/ml). The mixture is stirred for 60 min at room temperature in the dark, loaded onto a G-25 desalting column (Bio-Rad) with a 25-kDa molecular mass exclusion to separate free dye from labeled protein, and washed at 1100 × g for 5 min and stored at 4°C (31). The labeling efficiency is from 90 to 95% as determined by standard spectrofluorimetry at appropriate UV wavelengths for each fluorophore. The labeling of all primary Abs was performed under the described conditions twice separately for each Ab employed against the proteins of interest.

PMNs (5.0 × 10⁵ cells) were warmed to 37°C and then incubated with either buffer or 2 μM PAF for 1 min and prepared as previously described (4). Images were acquired with a Zeiss Axiovert fitted with a Cooke CCD SensiCam using a Chroma Sedat Multiple Bandpass filter wheel and Sutter filter control. Images were acquired using Intelligent Imaging Innovations SlideBook software. All images compared within a single figure were acquired as Z-stacks in 0.2-μm intervals and were deconvolved by applying constrained iterative deconvolution and Gaussian noise smoothing from system-specific point spread functions (32, 33, 34, 35, 36). Following deconvolution, images were cropped to represent the middle-most planes (center ± 10 planes), and the proteins in question were masked to represent zero fluorescence in IgG-negative controls. Cell polarization was assessed using Nomarski differential interference contrast microscopy (Nomarski images) (37).

FRET determinations were obtained using the 3-cube method to account for possible bleed-through of light (38). Images are acquired sequentially through three filter settings: donor filter, acceptor filter, and the FRET transfer filter, which is excitation through the donor filter and emitted light collected through the acceptor filter. To account for bleed-through from the donor to the acceptor or vice versa, images of samples with a single Ab (either donor or acceptor) are acquired under identical conditions as the experimental group, with correction coefficients being calculated using the “fit data” operation within SlideBook imaging software. In all microscopy experiments, the fluorophore-antibody combination was varied, and similar results were found. These data would suggest that even if the donor fluorophore vastly outnumbered the acceptor, giving a false-positive, or the acceptor outnumbered the donor, giving a false-negative, the proximity of the Abs against proteins of interest was sufficient and necessary to allow for real energy transfer. FRET indices may then be calculated using: FRET channel (FRET^C) = F_f − D_f(F_d/D_d) − A_f(F_a/A_a), where F_f, D_f, and A_f represent the transfer, donor, and acceptor channels in the presence of the all three fluorophores (Cy3, FITC, Cy5), and (F_d/D_d) and (F_a/A_a) represent the bleed-through corrections. The FRET^C is the energy transfer from donor to acceptor, which is then corrected for bleed-through as follows. Because the method of data processing is paramount to its interpretation, each image is acquired as a Z-stack of ≥25 planes. Bleed-through is corrected by a constraint deconvolution iterative in which the imaging properties of the optical system are used as a measured point spread function such that one mathematically “puts light back where it came from” (39). This point spread function can be used for a calculation of a likely model of the object from the recorded data set in an iterative process (39). Deconvolution eliminates effective blur caused by distortion, and by assuming a Poisson distribution of stray light it suppresses background light to very low levels (39). Not only is bleed-through eliminated but resolution is increased, which is of paramount importance in cases involving proteins of objects near resolution size (39). FRET efficiencies (E_i) were calculated using previously published techniques, and images are displayed in pseudocolor, using arbitrary linear units of fluorescent intensity (a.l.u.f.i.), where blue is “cold” (no FRET) and red is “hot” (most FRET). Such a visual representation was used rather than “white light” in which brightness is the representation of high FRET (4). Additionally, for all the FRET images there are a number of FRET-positive voxels even in the control (2, 3, 4, 5) if one looks closely. A positive result was only considered if the number of FRET-positive voxels increased by at least 2 logs over the control, and in each of the FRETs in the present paper the numbers of positive FRET voxels were between 2 and 3 logs greater than the paired controls. Furthermore, at least two Abs were used for each and every protein of interest and these Abs are suitable for immunoprecipitation, for most figures have such immunoprecipitations done in concert with the FRETs. Therefore, the immunoreactivity is for the native protein and not an epitope that may be “unmasked” by separation using reducing conditions and immunoblotting. Lastly, in a number of figures we overlay the FRET images, termed “FRET overlay”, or the cellular immunoreactivity of a single protein (p38 MAPK) with a FRET image, termed “overlay”, and examine whether the maximal FRET-positive physical interactions between proteins or the cellular immunoreactivity of a specific kinase and a possible substrate occurs in the identical cellular locales. These overlays visually demonstrate the regions that contain maximal overlay of FRET-positive voxels (e.g., proteins A and B and proteins A and C; all proteins must be on different fluorescent channels) to determine whether these interactions are occurring in the same cellular area. The maximal overlay of voxels is red with areas in high colocalization or blue in areas with little colocalization as visually expressed in a.l.u.f.i.

BioPORTER was reconstituted according to the manufacturer’s directions. Briefly, individual tubes were reconstituted to a total volume of 40 μl with Krebs-Ringer phosphate buffer with 2% dextrose ± 4 μg of the Ab specified in the text for 5 min at room temperature. PMNs (5 × 10⁶ cells) were incubated with buffer, vehicle only, or vehicle with the Ab for 2 h at 37°C. Following incubation, PMNs were centrifuged for 3 min at 400 × g at 4°C and resuspended to 5 × 10⁶ cells/ml. To control for IgG introduction, a FITC-IgG was introduced, and Z-stack images were acquired as described above. β-arrestin-1 and dynamin-2 were sufficient for native protein as demonstrated by use in immunoprecipitation. Furthermore, for intracellular neutralization two different Abs to different epitopes were used.

Isolated PMNs were primed with 2 μM PAF or buffer control for 1 and 3 min and immediately fixed with 4% phosphate-buffered paraformaldehyde (pH 7.35). PMNs (7.5 × 10⁵) were added to the 8-well Permanex chambered slides (Nunc) that were shaken to allow for equal distribution of PMNs. The slides were dried, washed with PBS (pH 7.0), and the PMNs permeabilized with 0.1% Triton-X. The permeablized PMNs were blocked with 200 μl 2% BSA-5% NDS (BSA-C) for 60 min at room temperature. The PMNs were incubated with the primary Abs (1/40 dilution in BSA-C) overnight and then washed with PBS. The secondary Abs, bound to immunogold (5- or 25-nm particles) by standard techniques, were then added at a dilution of 1/100 in BSA-C and incubated for 15 min at room temperature (40, 41). The cells were washed with PBS and fixed with 2.5% cacodylate-buffered glutaraldehyde for 60 min. The PMNs were silver enhanced for 60 min at room temperature and then washed with deionized water (Nanopure filtered) (42). Finally, the PMNs were again washed with 2% cacodylate-buffered osmium tetroxide for 60 min and then dehydrated in a graded series of ethanol, embedded in Embed 812, and sectioned at a thickness of 80 nm. Sections were stained with uranyl acetate and lead citrate and examined with a Zeiss EM-10CA transmission electron microscope at 60 kV accelerating voltage (43).

The data are expressed as the means ± the SEM. For FRETs the data are expressed as the mean fluorescent intensity of the FRET^C, and for better visualization the FRETs are represented in color (a.l.u.f.i.) as described previously. Statistical significance among groups was calculated by paired ANOVA with a Bonferroni-Dunn post hoc analysis using p ≤ 0.05 or p ≤ 0.01 as the level of statistical significance.

Dynamin-2 is responsible for pinching off the endosome from the plasma membrane and propelling it into the cytosol (1, 7, 8, 9, 10). PAF caused activation of dynamin-2 as demonstrated by tyrosine Y-phosphorylation of dynamin (Fig. 1) and colocalization with Grb2, a known signaling protein associated with activated dynamin-2, which was present at 1 and 3 min (Fig. 1) (44, 45). Importantly, intracellular neutralization of dynamin using two distinct neutralizing Abs inhibited internalization of the PAF receptor by 97 ± 4% and 99 ± 8%, respectively (data not shown).

PI3K has been implicated in the formation of the nascent endosome; therefore, it was important to discern whether it was required for colocalization of the Rab5a-GTPase with the sorting protein EEA-1 (14, 46, 47). PAF-induced dynamin-2 activation caused colocalization of PI3K with the Rab5a-GTPase in the cytosol at 3 min, as demonstrated by a positive FRET between Rab5a and PI3K, which is inhibited by the intracellular neutralization of dynamin-2 but unaffected by the isotypic control IgG (Fig. 2,A). PI3K also has a FRET-positive colocalization with EEA-1, which was corroborated by coprecipitation of PI3K with EEA-1 in PMNs primed by PAF as compared with buffer-treated controls of EEA-1 (Fig. 2, B and C). The FRET efficiencies for Fig. 2, A and C, were 34 ± 8% and 32 ± 4%, which correlate with a distance of <8 nm between these proteins. Furthermore, PAF caused a FRET-positive colocalization of PI3K with both EEA-1 and Rab5a in the identical PMNs and localized these protein interactions to the cytosol as demonstrated by light microscopy (Nomarski image, Fig. 3,A). The microscopy was reinforced by immunoprecipitating Rab5a and demonstrating colocalization of both EEA-1 and PI3K at 1 min that increased at 3 min (Fig. 3,B) and by immunoprecipitating EEA-1 and finding coprecipitation of Rab5a (Fig. 3 C).

Because Rab5a requires dissociation from RabGDI for activity, it was imperative to demonstrate that PAF caused Rab5a activation (48, 49). In buffer-treated control PMNs, Rab5a and RabGDI evidenced a FRET-positive interaction (Fig. 4,A). PAF caused dissociation of this interaction as demonstrated by the loss of the FRET-positive interaction (Fig. 4,A). Importantly, these experiments were also done with the introduction of an IgG isotypic control because in the third group of PMNs we neutralized dynamin-2 by intracellular introduction of an Ab directed against dynamin-2. Intracellular delivery of specific Abs to dynamin-2 abrogated the dissociation of RabGDI from Rab5a, as demonstrated by the FRET-positive interaction between Rab5a and RabGDI (Fig. 4,A). This microscopy was reinforced by immunoprecipitations of Rab5a and probing for EEA-1 and RabGDI, which demonstrated that PAF caused a decrease colocalization of Rab5a with RabGDI and an increased coprecipitation with EEA-1, as compared with buffer-treated controls (Fig. 4 B).

Previous studies have demonstrated that a potential role for MAPKs is the activation of Rab5a via phosphorylation of RabGDI, which facilitates the release of Rab5a (49, 50, 51, 52, 53). To determine whether this was true of PMNs in response to PAF, the relationship between Rab5a and the MKK3/p38 MAPK signaling cassette was first investigated. The role of the p38 MAPK signaling cassette in the activation of Rab5a was then investigated. In lysates immunoprecipitated for p38 MAPK from buffer-treated PMNs, there was coprecipitation of ASK1, a MAP kinase kinase kinase (MAP3K), which did not increase upon PAF stimulation (Fig. 5,A). Upon PAF stimulation (3 min), MKK3 coprecipitated with p38 MAPK and Rab5a (Fig. 5 B), implying that assembly of the p38 MAPK signalosome occurs on the endosome in the proximity of Rab5a and becomes activated by the recruitment of MKK3. Moreover, pretreatment of PMNs with 1–10 μM PD98059, a selective p42/44 MAPK inhibitor, did not affect PAF priming of the PMA respiratory burst (4.0 ± 0.7 to 8.2 ± 4.1% inhibition), whereas using a selective p38 MAPK inhibitor SB203580 (1 μM) significantly inhibited PAF priming of the PMA-activated respiratory burst by 74 ± 6% as compared with DMSO-treated control PMNs activated with 200 ng/ml PMA (p < 0.05, n = 7). Furthermore, p42/44 MAPK did not localize with RabGDI (data not shown).

PAF also elicited the following events: 1) a FRET-positive association of RabGDI with activated (diphosphorylated: Y138, T142) p38 MAPK; 2) a FRET-positive interaction of RabGDI and PS; and 3) whether one overlays the cellular immunoreactivity of activated p38 MAPK and the FRET of RabGDI and PS, there are obvious areas of strong (red pseudocolor) colocalization between the phosphorylated RabGDI and activated p38 MAPK (Fig. 6,A). These data directly implicate p38 MAPK in the S-phosphorylation of RabGDI by a FRET-positive interaction of active p38 MAPK and RabGDI leading to activation of Rab5a. Lastly, because a number of reports have documented that EEA-1 is phosphorylated by MAPKs, the phosphorylation of EEA-1 by p38 MAPK was also investigated. Fig. 6,B shows that PAF caused a FRET-positive interaction between active p38 MAPK and EEA-1 compared with buffer-treated controls. In these same PMNs there is an increase in PS, compared with control PMNs, which demonstrates a FRET-positive interaction with EEA-1 (Fig. 6,B). If one then overlays these two FRETs (activated p38 MAPK/EEA-1 and EEA-1/PS) and displays the cellular colocalization in pseudocolor, it appears that these events are occurring in the same cellular location (Fig. 6 B). However, one may not expect an interaction of p38 MAPK with PS because the kinase has already phosphorylated the substrate (EEA-1) and has dissociated from it. Thus, these data implicate p38 MAPK in the phosphorylation of EEA-1 based on a FRET-positive interaction of activated p38 MAPK and EEA-1, which demonstrates that these proteins were within 5–8 nm. This physical association of an active S/T kinase with its substrate provides provocative evidence that p38 MAPK is responsible for the observed S-phosphorylation of EEA-1.

Because p38 MAPK has a prominent role in the formation of actin bundles in the plasma membrane at sites for endosomal scission, it was necessary to ensure the spatial locality of activated p38 MAPK, which colocalizes with β-arrestin-1 at 1 min, and the activated p38 MAPK, which colocalizes with EEA-1 in the endosome at 3 min (4). Small immunogold particles (5 nm) were bound to Abs specific for active, diphosphorylated p38 MAPK, and large immunogold particles (25 nm) were linked to either β-arrestin-1 or Rab5a. In buffer-treated PMNs, minimally activated p38 MAPK was visualized with scattered immunoreactivity for either β-arrestin-1 or Rab5a (Fig. 7, A and C). In contrast, at 1 min, activated p38 MAPK (small immunogold particle, 5 μm) was present near the membrane and colocalized with β-arrestin-1 (large immunogold particle, 25 μm), and in this case two active p38 MAPKs are seen with two β-arrestin-1 molecules (Fig. 7,B). At 3 min, activated p38 MAPK (small particle) in the cytosol colocalized with Rab5a (large particle) in an endosome, and, as illustrated, there are two large immunogold particles with one small particle (Fig. 7 D). These data provide qualitative evidence that PAF induces sequential activation of p38 MAPK, which occurs on the β-arrestin-1 scaffold at or near the membrane at 1 min and in the cytosol localized with Rab5a at 3 min.

The cytosolic oxidase components p40^phox and p67^phox are present as a complex in the cytosol ± p47^phox (54, 55, 56, 57). As demonstrated in Fig. 8,A, the FRET-positive interaction between p40^phox and p67^phox remained unaltered with PAF priming; however, PAF priming does appear to modestly change the cellular location of this complex as compared with the buffer-treated controls. Importantly, p67^phox does not contain a lipid-binding domain, and because p40^phox does contain this lipid-binding domain, it has the capacity to serve as a chaperone for p67^phox to allow this p40^phox-p67^phox complex to bind to phosphatidylinositol 3,4,5-triphosphate (PIP3) domains in the membrane, or like structures, for translocation and binding to the plasma membrane (58, 59, 60, 61). In comparison to buffer-treated controls, PAF caused increased colocalization of p40^phox and p67^phox with EEA-1 (Fig. 8,B) as determined by immunoprecipitation of EEA-1 and probing for p40^phox and p67^phox (Fig. 8,B). Additionally, PAF (3 min) also elicited a FRET-positive colocalization of p67^phox with Rab5a, which was blocked with intracellular neutralization of dynamin-2 (Fig. 8,C); moreover, PAF induced a coprecipitation of Rab5a with p67^phox, which was not present in buffer-treated controls (Fig. 8,D), and this further reinforces the data that implicated PAF-induced recruitment of the p40^phox-p67^phox complex to both EEA-1 and Rab5a, which are constituents of the endosome. Thus, PAF appears to cause localization of the p40^phox-p67^phox complex to the endosome; however, one must consider how this complex attaches to the endosome and why. To gain insight into this question we used an Ab against p40^phox that spanned the PX (lipid binding) domain; therefore, if the p40^phox PX domain is already bound to PIP3 residues, the epitope would be “hidden” and there would be no immunoreactivity. As demonstrated previously, control PMNs demonstrated a FRET-positive interaction between the p40^phox and p67^phox complex, which does not colocalize with EEA-1 (Fig. 9,A). At 3 min of PAF stimulation, three important protein interactions occur: 1) the p40^phox-p67^phox FRET disappears, 2) the total cellular immunoreactivity of p40^phox decreases dramatically, and 3) p67^phox displays colocalization with EEA-1 (Fig. 9,B). Such findings may be explained by whether most p40^phox becomes lipid bound to the endosome via PIP3 residues, which could be available on the FYVE domain of EEA-1, or to other PIP3 sites (14, 46). If this supposition is true, then PI3K inhibition would 1) return the FRET-positive interaction of the p40^phox-p67^phox seen in control PMNs, 2) increase total cellular p40^phox immunoreactivity because the PX domain is now free to bind the Ab, and 3) decrease the colocalization of EEA-1 and p67^phox. As shown in Fig. 9 C, these described events all occurred with PI3K inhibition. Importantly, no other oxidase components were associated with either EEA-1 or Rab5a including p47^phox, Rac2, or gp91^phox (data not shown), confirming that PAF caused recruitment of substrates to the endosome, namely the p40^phox-p67^phox complex, and not oxidase assembly.

Because phosphorylation of oxidase components has been reported as a necessary prerequisite for translocation and also appears to be important in priming of the NADPH oxidase, PAF-induced phosphorylation of p40^phox and p67^phox was investigated (19, 23, 62, 63, 64). At 3 min, PAF markedly enhanced S-phosphorylation of p67^phox and colocalization with p40^phox (Fig. 10,A). To further delineate the role of PI3K, we investigated its relationship with both the phosphorylated and nonphosphorylated forms of p40^phox. As compared with controls, PAF elicited a FRET-positive interaction between PI3K and p40^phox and to a lesser extent between PI3K and phosphorylated p40^phox (Fig. 10,B). However, these data make little intuitive sense unless another kinase is involved, because PI3K phosphorylates lipids on tyrosine residues, and the p40^phox and p67^phox require S-phosphorylation for translocation (14, 54, 62, 63, 65, 66, 67). Immunoprecipitation of p40^phox from controls and PAF-primed PMNs was completed and demonstrated a PAF-induced colocalization of Akt1 with p40^phox, and identical immunoprecipitations were completed for p67^phox that also showed coprecipitation with Akt1 (Fig. 10, C and D). Additionally, immunoprecipitation of Akt1 also demonstrated coprecipitation of both p40^phox and p67^phox (data not shown). Thus, PI3K appears to recruit Akt1 for phosphorylation of the p40^phox and p67^phox and consequently the FRET-positive interaction of PI3K with the nonphosphorylated p40^phox and p67^phox and a decrease in the FRET-positive interaction with these phox proteins after S-phosphorylation (Fig. 10,B). The latter decrease in the FRET-positive interaction between the cytosolic oxidase components and PI3K is to be expected because if PI3K is necessary for the phosphorylation of p40^phox, then following phosphorylation the physical relationship between p40^phox and PI3K should diminish as demonstrated (Fig. 10 B).

Previous data have demonstrated that PAF causes translocation of p67^phox to the plasma membrane without concomitant translocation of p47^phox (14, 54, 65, 68). The amount of p67^phox is sufficient to run the NADPH oxidase in an SDS cell-free system using both membranes for PAF-primed PMNs with cytosol that was p67^phox depleted (14, 54, 65, 68). Such p67^phox translocation alone appears counterintuitive due to the inability of p67^phox to bind lipids and to a relative dearth of data with regard to p67^phox membrane translocation without p47^phox (14, 54, 65, 68). Therefore, PAF-mediated translocation of the p40^phox-p67^phox complex was investigated. At 3 min, PAF caused the translocation of this complex to the membrane, compared with buffer-treated controls, and this complex colocalized with the gp91^phox membrane oxidase component, which also exhibited a FRET-positive relationship with PI3K at the plasma membrane (data not shown) (68). Additionally, because the small GTPase Rac2 has been associated with the translocation of p47^phox to the membrane with priming and activation of the oxidase, PAF-mediated translocation of Rac2 was also investigated. PAF did not cause membrane translocation of Rac2 with the p40^phox-p67^phox complex (results not shown, n = 5) (32, 60, 69). Lastly, as stated above, p67^phox translocation is dependent on p47^phox translocation, as demonstrated in the PMNs from patients with chronic granulomatous disease who lack p47^phox (70). The roles of p40^phox and p47^phox were investigated in PAF-mediated translocation of p67^phox in intact PMNs using intracellular immunodepletion of p47^phox and p40^phox with two disparate neutralizing Abs for each phox protein and comparing these PMNs to controls loaded with isotypic IgG. PAF (3 min) caused the translocation of p67^phox, which was inhibited by intracellular neutralization of p40^phox but was not affected by identical intracellular neutralization of p47^phox, when the colocalization of p67^phox and the membrane counterstain wheat germ agglutinin were analyzed (Figs. 11). Additionally, intracellular neutralization of p40^phox inhibited PAF priming of the PMA-activated respiratory burst by 74 ± 9% as compared with PAF priming of the PMA-activated respiratory burst in PMNs loaded with isotypic IgG controls (n = 5, p < 0.05). Conversely, intracellular neutralization of the p47^phox only inhibited PAF priming by 21 ± 2% (n = 5), which was not different from the IgG-loaded control PMNs (data not shown).

The data presented demonstrated that PAF ligation of its receptor on PMNs resulted in dynamin-2 activation, as shown by tyrosine phosphorylation, colocalization with Grb2, and endosome formation (Fig. 12). Assembly of the endosome consisted of PI3K, Rab5a, and EEA-1, and activation of Rab5a occurred through likely p38 MAPK phosphorylation of the RabGDI, resulting in dissociation from Rab5a. The p38 MAPK signalosome assembled on the endosome, as documented by its colocalization with Rab5a, consisting of ASK1, MKK3, and p38 MAPK, identical with our previous work with PAF-mediated p38 MAPK activation on the β-arrestin-1 scaffold (4) (Fig. 12). The tethering protein EEA-1 served as a scaffold, which first required PI3K phosphorylation to change the phosphatidylinositol 4,5-diphosphate (PIP2) sites to PIP3, required for binding of the p40^phox-p67^phox complex through the PX domain on p40^phox (Fig. 12). Following this tethering to EEA-1, these cytosolic phox proteins were most likely phosphorylated by Akt1, as demonstrated by coprecipitation of the p40^phox and p67^phox with Akt1, which is a common finding among priming agents (18, 19, 23). The p40^phox-p67^phox complex subsequently translocated to the plasma membrane and localized to gp91^phox along with PI3K and did not require p47^phox or Rac2 translocation.

Dynamin activation as a function of Y-phosphorylation and its colocalization with Grb2 is congruent with previous reports (1, 44, 45, 71). Furthermore, inhibition of dynamin disrupted endosome formation as demonstrated by the loss of Rab5a activity, the FRET-positive colocalization of PI3K with Rab5a, and the FRET-positive interaction between Rab5a and p67^phox (1, 44, 45, 71). Additionally, the colocalization of Rab5a and EEA-1 in the endosome and PI3K-dependent activation of Akt1 also confirm prior reports in transformed cells or cells that have been manipulated to specifically study the interactions of these specific signaling proteins that are integral to CME (46, 47, 72, 73, 74, 75, 76). The Rab GTPases, especially Rab5a, are mediators of intracellular trafficking, control the fusion of vesicles and/or lysosomes with intracellular organelles, and play important roles in CME, and their mutation results in human disease (17, 47, 72, 75, 76, 77, 78, 79, 80). Moreover, Rab5a is required for CME of activated receptors and is regulated by binding of RabGDI in the cytosol (17, 47, 49, 53, 72, 75, 76, 78, 79, 80). Previous data have demonstrated that activation of Rab5a is controlled by the p42/44 MAPKs that phosphorylate the RabGDI and cause its dissociation (49, 50, 51, 52, 53). Contrary to these data, PAF caused a likely p38 MAPK-dependent activation of Rab5a by S-phosphorylation of RabGDI and its subsequent dissociation from Rab5a. This conclusion is drawn from an overlay of the cellular immunoreactivity of active p38 MAPK and the FRET-positive physical interaction of RabGDI with phosphoserine demonstrating a high degree of identical cellular locale. However PAF-induced S-phosphorylation of RabGDI is based on: 1) the physical interactions between proteins, RabGDI/active p38 MAPK, and RabGDI/PS; and 2) the locality of active p38 MAPK with the FRET between RabGDI and PS. Importantly, RabGDI has been reported to be a MAPK substrate, congruent with these observations (49, 50, 51, 52, 53). This p38 MAPK phosphorylation is novel in GPCR signaling in primary cells and confirms the in vivo data from peroxide or UV-stimulated PMNs that suggested a role for p38 MAPK in Rab5a activation (50). The assembly of the ASK1/MKK3/p38 MAPK signalosome on the endosome, associated with Rab5a, is also novel in CME, as is the PAF-induced p38 MAPK activation in two cellular locales at disparate times: 1) at the plasma membrane on the β-arrestin-1 scaffold at 1 min (4), and 2) in the cytosol on the endosome at 3 min as delineated by electron microscopy. Activation of p38 MAPK on both the β-arrestin-1 scaffold and on the endosome associated with Rab5a occurred through the recruitment of MKK3 into the signalosome, as in resting PMNs, both ASK1 and p38 MAPK colocalized with β-arrestin-1 and Rab5a, respectively (4). In our initial experiments, PAK1 was suspected to be the MAP3K in the p38 MAPK signalosome, but we were unable to colocalize PAK1 to either the β-arrestin-1 scaffold or to Rab5a (4). For PAF priming of the PMN NADPH oxidase, inhibition of p42/44 MAPK with PD98059 had little effect on the PAF-induced changes in PMN physiology, as others have demonstrated (81, 82). Conversely, p38 MAPK inhibitors did inhibit the PAF priming of the PMA-activated respiratory burst; however, one must interpret these data with caution, as p38 MAPK is required for actin bundle formation at the site of PAFR CME, and thus one may not be inhibiting the activation of Rab5a, which is downstream of these initial signaling events (4).

PAF elicited the tethering of the p40^phox-p67^phox complex to the endosome, as shown by its colocalization with both EEA-1 and Rab5a. This colocalization occurs through binding of the p40^phox PX domain, because it required PI3K activity and the lipid-binding PX domain of the p40^phox as based on Ab inhibition of this PX domain and PI3K and the lack of a lipid-binding PX domain on p67^phox. These data implicate an interaction between the PX domain of p40^phox and PIP3 residues in the endosome, which are rich in PIP2 domains (83, 84). Upon specific receptor activation, these PIP2 domains may be phosphorylated by PI3K and then may play a role in the regulation of endocytotic membrane trafficking, signal transduction, and cytoskeletal reorganization by interacting with the PX domains of the phox proteins and other effectors, including EEA-1, through the phox homology region in its FYVE domain, and the sorting nexins (84, 85, 86, 87, 88). Moreover, the recruitment of the p40^phox-p67^phox complex to EEA-1 allowed for their S-phosphorylation by Akt1 in a PI3K-dependent manner. These data also extend the relevance of Akt1 from phosphorylation of p47^phox to its ability to phosphorylate the p40^phox-p67^phox complex, which then resulted in the membrane translocation during PAF priming of the NADPH oxidase, both of which were abrogated by intracellular neutralization of the catalytic domain of PI3K (89, 90). Additionally, no other oxidase components were recruited to the endosome; therefore, PAF-mediated enlistment of the p40^phox-p67^phox complex to the EEA-1 sorter protein is indicative of substrate recruitment rather than assembly or partial assembly of the NADPH oxidase.

Several priming agents, including endotoxin and IL-18, cause the translocation of p47^phox to the plasma membrane (18, 23). In contrast, PAF elicited translocation of the p40^phox-p67^phox complex to the PMN membrane in the region of gp91^phox. Additionally, PI3K-dependent formation of PIP3 domains on EEA-1 was required for phosphorylation of the p40^phox-p67^phox complex, a prerequisite for translocation, with binding of the PX domain of p40^phox to the FYVE domain of EEA-1. One may expect a similar regulatory role for p40^phox with membrane translocation because p67^phox does not contain the PX domain; however, more work is required to test this hypothesis (56, 58, 91). Furthermore, the amount of the p40^phox-p67^phox complex translocated to the plasma membrane was not trivial based upon the ability of PAF-primed membranes to provoke oxidase activity when combined with p67^phox immunodepleted cytosol in a cell-free oxidase system (14, 54, 65, 68). Moreover, intracellular neutralization of p40^phox significantly inhibited PAF priming of the PMA-activated respiratory burst, as compared with PMNs that have had p47^phox neutralized or with control PMNs loaded with IgG isotype controls. Recently, the enzymatic importance of the p40^phox component has been described to regulate Fc receptor-mediated activation of the NADPH oxidase (60), and the presented data imply a regulatory role for p40^phox in PAF priming of the NADPH oxidase. Additionally, we were unable to detect translocation of Rac2 to the membrane; however, most data that have delineated a role for Rac2 in the membrane assembly of the NADPH oxidase are based pm activation and not priming (32, 33, 34, 36). PMN priming does not result in activation of the NADPH oxidase and may not require the small GTPases for phox protein translocation, especially Rac2.

In conclusion, these data have delineated the role of CME in signal transduction in human PMNs in response to a well-described PMN priming agent. Most of the data with regard to CME in leukocytes are derived from transformed cells or differentiated myelogenous leukemia cells, and many of the differences between these data and those of others may be due to their use of transformed cells or differentiated myelogenous leukemia cells. In syndromes of inappropriate PMN priming and activation, including acute lung injury or postinjury multiple organ failure, many of the implicated agents are known to signal through GPCRs, the majority of which transduce these signals through CME. There are a number of pharmacologic inhibitors of CME that may attenuate or inhibit PMN-mediated damage by abrogating CME. Further preclinical and clinical trails with these agents may result in better outcome for patients with PMN-mediated organ damage.

The authors have no financial conflicts of interest.