Rac regulates PtdInsP₃ signaling and the chemotactic compass through a redox-mediated feedback loop.

Directional cell migration is an essential requirement for efficient neutrophil translocation to sites of infection and requires the establishment of a polarized cell characterized by an actin-rich leading edge facing the chemoattractant gradient. The asymmetrical accumulation of phosphatidylinositol(3,4,5)-trisphosphate [PtdIns(3,4,5)P(3)] in the up-gradient leading edge is a hallmark of polarization and regulates the recruitment and localization of various effector proteins at the leading-edge plasma membrane. How shallow gradients of chemoattractants trigger and maintain a much steeper intracellular gradient of PtdIns(3,4,5)P(3) is a critical question in the study of leukocyte chemotaxis. Our data demonstrate that the migration of neutrophils toward the chemoattractant N-formyl-L-methionyl-L-leucyl-L-phenylalanine depends on the generation of reactive oxygen species by the phagocytic NADPH oxidase (NOX2) and subsequent oxidation and inhibition of phosphatase and tensin homolog. Moreover, we show that events downstream of PtdIns(3,4,5)P(3), including phosphorylation of AKT, Rac activation, uncapping of actin filaments, and directional migration, can be attenuated by ROS scavengers or genetic ablation of NOX2. Using Rac mutants that are defective in their ability to activate NOX2, we show that Rac regulates a redox-mediated feedback loop that mediates directional migration of neutrophils.


Introduction
Neutrophils are important mediators of the innate immune system and are rapidly recruited from the circulation to sites of infection to eliminate pathogenic threats. Key to this recruitment is the release of chemoattractants by infected host tissue or pathogens, which subsequently form a chemical gradient that attracts neutrophils to the appropriate sites. 1 In response to a chemotactic gradient, neutrophils polarize and form a leading edge pointing toward the chemoattractant source and a posterior structure called the uropod. Actin polymerization within the leading edge drives up-gradient protrusion, whereas myosin activity in the uropod detaches the rear of the cell. The proper coordination of these polarized events is essential for directional migration and depends on the asymmetrical distribution of specific molecular determinants. A hallmark of neutrophil polarization is the asymmetrical accumulation of the phospholipid phosphatidylinositol (3,4,5)-trisphosphate [PtdIns (3,4,5)P 3 ] in the leading edge. 2,3 The formation of PtdIns(3,4,5)P 3 is catalyzed by PI3Ks and is counteracted by the activity of the lipid phosphatases phosphatase and tensin homolog (PTEN) and SH2-containing inositol phosphatase (SHIP), which generate the lipid products PtdIns(4,5)P 2 and PtdIns(3,4)P 2 , respectively. 4,5 Several studies have suggested that the establishment of a robust PtdIns(3,4,5)P 3 levels requires a positive feedback loop, and this is corroborated by the observation that the introduction of exogenous PtdIns(3,4,5)P 3 induces further accumulation of endogenous PtdIns(3,4,5)P 3 in a Rho GTPase-dependent fashion. 6,7 Rac and actin polymerization were both shown to be required for the establishment of PtdIns(3,4,5)P 3 polarization via a feedback loop mechanism. 8,9 However, the exact mechanism remains elusive. Because neutrophils produce reactive oxygen species (ROS) within the leading edge during chemotaxis, 10,11 and the activity of the lipid phosphatase PTEN is markedly reduced by ROS, 12,13 we hypothesized that local ROS production in neutrophils could regulate PtdIns(3,4,5)P 3 levels through redox regulation of PTEN. We indeed identified PTEN to be a major target of chemoattractantinduced ROS formation and demonstrate that its enzymatic activity decreases accordingly. Decreased PTEN activity led to an increase of PtdIns(3,4,5)P 3 , which could be reversed by ROS scavengers or genetic ablation of NADPH-oxidase 2 (NOX2). Consistent with this finding, reduction of ROS inhibited directional migration of neutrophils toward N-formyl-L-methionyl-L-leucyl-L-phenylalanine (fMLP), uncapping of actin filaments, and Rac activity. Although Rac GTPases are important downstream effectors of PtdIns(3,4,5)P 3 , they are also crucial for ROS production. Rescue experiments with small GTPase Rac effector mutants revealed that Rac facilitates PtdIns(3,4,5)P 3 accumulation and that this effect depends on its ability to activate NADPH-oxidase.
The publication costs of this article were defrayed in part by page charge payment. Therefore, and solely to indicate this fact, this article is hereby marked ''advertisement'' in accordance with 18 USC section 1734.

PTEN-activity assay
Primary neutrophils were partially permeabilized with 0.2% octyl glucoside for 15 seconds at room temperature using the procedure from Glogauer et al, 18 resuspended in HBSS, and treated with 100M Trolox (or mock-treated) for 15 minutes at 37°C. Cells were then stimulated with 1M fMLP for 1 minute and incubated with 1M of the PTEN substrate 3-methylenephosphonate, diC8 (Echelon Biosciences). Control reactions were set up using identical treatments omitting substrate. Reactions were incubated for 30 minutes at room temperature before measuring the released phosphate using the Malachite Green phosphate detection assay according to the manufacturer's protocol (Echelon Biosciences). A sodium phosphate standard curve was used to calculate the release of phosphate. Relative PTEN activity was calculated by subtracting the values of the samples with substrate minus background the background release (no substrate).

TAT-protein transduction
Murine wild-type Rac1 and Rac2 open reading frames were cloned into pTAT-HA containing a HIS tag (a kind gift from Dr G. Bokoch) using standard molecular techniques. Rac1-A27K (a kind gift from Dr E. Pick) was subcloned in pTAT-HA. Point mutations were introduced in the wild-type constructs using site-directed mutagenesis to obtain Rac1-T17N and Rac2-A27K. The resulting constructs were transformed into BL21(DE3)-pLysS (Stratagene). Protein expression and purification was performed as described previously. 19 In short, 500-mL cultures were inoculated 1:100 and grown for 3-4 hours at 37°C. Protein expression was induced overnight at 28°C after induction with isopropyl ␤-D-1-thiogalactopyranoside (1mM). Cell pellets were resuspended in lysis buffer (100mM NaH 2 PO 4 , and 10mM Tris, and 8M urea) and sonicated briefly. Cell debris was removed by centrifugation and HIS-tagged proteins were isolated from cleared lysates using Ni-NTA Superflow columns (QIAGEN) according to the manufacturer's protocol, and subsequently dialyzed for 24 hours 2ϫ at 4°C against PBS using 15-mL Slide-a-Lyzer cassettes (Pierce Protein Research Prod-ucts; ThermoFisher). Finally, proteins were concentrated in Pierce protein concentrators, and protein concentration and purity were assessed by SDS-PAGE followed by Coomassie staining. Control experiments and data from previous studies 20, 21 have verified that addition of TAT-Rac-WT rescues ROS generation and chemotaxis in leukocytes (data not shown).

SDS-PAGE and immunoblotting
Lysates were prepared in 4ϫ Laemmli loading buffer, boiled for 10 minutes at 99°C, subjected to reducing SDS-PAGE, and transferred to nitrocellulose membranes. Each membrane was blocked and incubated with the indicated primary Ab according to the manufacturer's specifications (1:2000 for Rac1, 1:1000 for p-AKT and AKT, and 1:1000 for PTEN) at 4°C overnight. Membranes were washed 3 times for 10 minutes with TBS-T, followed by incubation with donkey anti-rabbit or sheep anti-mouse IgG peroxidase conjugates (Amersham Pharmacia Biotech). Ag-Ab complexes were visualized on X-ray film by enhanced chemiluminescence (ECL Plus; Amersham, GE Healthcare).

Oxidative modification PTEN
Oxidated and reduced forms of PTEN were detected by a modified method adapted from Silva et al. 22 Cell pellets were resuspended in deoxygenated lysis buffer containing SDS (2%), Tris (50mM, pH 7.5), EDTA (0.5mM), iodoacetamide (50mM), and catalase (5000 u/mL), and incubated for 10 minutes in the dark. The insoluble fraction was sedimented by ultracentrifugation for 20 minutes at 175 000g at 8°C, and the soluble fraction was transferred to a new tube, mixed with nonreducing loading buffer, and separated by nonreducing SDS-PAGE. PTEN was detected by immunoblotting.

Rac activation assay
Active Rac1 was quantified using the PAK-binding domain pull-down assay as described previously. 14

CapZ quantification
CapZ were measured as described in Sun et al. 15 In short, 1 million murine neutrophils were permeabilized with 0.2% OG buffer (PHEM buffer containing 10M phallacidin, 42nM leupeptin, 10mM benzamidine, and 0.123mM aprotinin) and then stimulated with 1M fMLP for 60 seconds. The supernatants and lysates of the remaining cells were analyzed by SDS-PAGE and subsequent immunoblotting for CapZ. The release of capping proteins was evaluated by the calculating the ratio of CapZ in the supernatant divided by CapZ in the total lysate, as determined by densitometry using ImageJ 1.32j software in 3 independent experiments.

Results
Neutrophils migrating toward a chemoattractant source display polarized ROS production at the front of the cell 10 (supplemental Figure 1, available on the Blood Web site; see the Supplemental Materials link at the top of the online article). Because PTEN is a known target of oxidation, 12,13 we analyzed the redox state of PTEN by nonreducing SDS-PAGE. Stimulation of neutrophils with fMLP increased the fraction of oxidized PTEN by 32% Ϯ 12%, and this effect could be blocked by pretreating cells with Trolox, a potent ROS scavenger 23 ( Figure 1A). Cells treated with H 2 O 2 or lysates incubated with DTT were used as positive and negative controls for oxidative modification of PTEN, respectively. Interestingly, cells that lack the catalytic subunit (gp91) of the NADPH oxidase NOX2 24 also displayed reduced PTEN oxidation in response to fMLP (supplemental Figure 2). It has been reported that oxidation of PTEN inhibits enzymatic activity. 12,13,22 To assess the impact of fMLP-induced oxidation on enzymatic activity, we measured dephosphorylation of the PTEN substrate 3-methylenephosphonate (diC8) in partially permeabilized neutrophils. 18 diC8 BLOOD, 1 DECEMBER 2011 ⅐ VOLUME 118, NUMBER 23 For personal use only. on July 24, 2018. by guest www.bloodjournal.org From was dephosphorylated at a significant lower rate in cells treated with fMLP (53% Ϯ 12%) compared with control untreated cells ( Figure 1B). This decrease in activity was prevented by pretreating cells with Trolox, suggesting that the effect was due to oxidationdependent events. We noted an inverse correlation between the fraction of oxidized PTEN and the dephosphorylation of diC8 ( Figure 1A-B). Treatment with the selective PTEN inhibitor VO-OHpic 25 demonstrated the specificity of the assay for PTEN activity (supplemental Figure 3).
Oxidative modification of PTEN and subsequent decreased activity could result in increased PtdIns(3,4,5)P 3 levels. We quantified phosphorylation of AKT, a downstream target of PtdIns(3,4,5)P 3 that is widely used as an indicator of PtdIns(3,4,5)P 3 generation. 26 Stimulation of neutrophils with fMLP resulted in a robust increase in AKT phosphorylation (Figure 2A-B). Because a decrease in PTEN activity by oxidation would enhance PtdIns(3,4,5)P 3 accumulation, preventing oxidation by ROS scavengers would have the opposite effect: decreasing phosphorylation of AKT. Indeed, pretreatment of cells with the ROS scavengers NAC and Trolox decreased phosphorylation of AKT significantly (Figure 2A-B). To confirm that decreased AKT phosphorylation in response to ROS scavengers was because of decreased PtdIns(3,4,5)P 3 , we applied live-cell imaging using the PH domain of AKT fused to GFP (PH-AKT-GFP) as a reporter for subcellular PtdIns(3,4,5)P 3 accumulation and, to a lesser extent, PtdIns(3,4)P 2 . 27,28 In response to fMLP, PH-AKT-GFP was recruited to the leading edge in neutrophils ( Figure 2C). However, the distinct asymmetrical accumulation rapidly dissipated upon treatment with NAC, suggesting that ROS are required for robust PtdIns(3,4,5)P 3 levels at the front of the cell. fMLP stimulates ROS production by activating NOX2 in polymorphonuclear cells, and loss-of-function mutations in NOX2 lead to defective respiratory burst in neutrophils and macrophages, causing chronic granulomatous disease in humans. 29 To determine whether phosphorylation of AKT depends on fMLP-mediated activation of NOX2, we used mice that are deficient in gp91, the catalytic subunit of NOX2. 24 In neutrophils derived from C57/BL6 control mice, phosphorylation of AKT was strongly induced by fMLP ( Figure 2D and supplemental Figure 4), whereas in Gp91 Ϫ/Ϫ cells, phosphorylation remained at basal levels ( Figure 2E and supplemental Figure 4). Intriguingly, activation of AKT in gp91 Ϫ/Ϫ neutrophils could be (partially) rescued by the addition of exogenous H 2 O 2 , whereas in the presence of ROS scavengers (NAC and Trolox) phosphorylation of AKT was suppressed ( Figure 2E). We conclude from these experiments that ROS produced by NOX2 play an important role in modulating PtdIns(3,4,5)P 3 and its downstream target, AKT, upon fMLP stimulation.
Accumulation of PtdIns(3,4,5)P 3 at the leading edge of migrating neutrophils has been implicated in regulating directional cell migration by mediating the polarized recruitment of factors required for local actin polymerization and cell advancement up the chemotactic gradient. 3 Therefore, decreased PtdIns(3,4,5)P 3 levels induced by ROS scavengers could affect directionality. To test this, neutrophil chemotaxis toward fMLP was analyzed using Zigmond chambers in the presence of the ROS scavenger NAC ( Figure 3A,B,D). The majority of untreated control cells (74% Ϯ 6%) migrated up the fMLP gradient. However, pretreatment of cells with NAC inhibited directional sensing significantly, with only 56% Ϯ 4% of the cells migrating toward fMLP. We then tested directionality in neutrophils derived from gp91-deficient mice ( Figure 3C-D). Only 49% Ϯ 6% of the gp91 Ϫ/Ϫ cells migrated toward fMLP, whereas 75% Ϯ 6% of their wild-type counterparts (C57/BL6) displayed directional migration toward fMLP. Migration velocity was not affected in these experiments (data not shown). These data demonstrate that directional sensing in neutrophils requires ROS generated by NOX2. To demonstrate that ROS are important in directionality by inhibiting PTEN, we treated neutrophils with the ROS scavenger Trolox and the PTEN inhibitor VO-OHpic. Trolox inhibited directional migration of neutrophils to a similar extent as NAC ( Figure 3B,E), whereas the PTEN inhibitor VO-OHpic did not alter directionality compared with untreated cells ( Figure 3F). Cells treated with both Trolox and VO-OHpic did migrate normally toward fMLP. This demonstrates that inhibition of directionality by scavenging ROS can be overruled by inhibition of PTEN activity, supporting a role for redox regulation of PTEN in directional migration.
Because Rac is a principal regulator of NOX2 in neutrophils, 30 it is possible that Rac regulates PtdIns(3,4,5)P 3 levels through redox-mediated inhibition of PTEN. In earlier studies, we found that phospho-AKT levels were decreased in Rac1-null and, to a lesser extent, in Rac2-null neutrophils. 14 Moreover, directional migration was completely lost in Rac1-null cells. This led us to investigate the possibility that Rac1 is specifically required for activation of NOX2 during ROS-dependent directional migration. By making use of the Rac1 mutant (A27K), which is defective in p67phox binding, 31,32 we were able to selectively block Rac1mediated NOX2 activation while maintaining other Rac-effector functions. Rac1-null neutrophils were transduced with 200nM recombinant TAT-tagged wild-type Rac1, inactive Rac1-T17N, or Rac1-A27K before chemotaxis was analyzed using a Zigmond chamber (Figure 4 and supplemental Figure 5). Transduction of wild-type Rac1 completely rescued directionality toward fMLP in Rac1-null cells, whereas inactive Rac-T17N did not. However, Rac1-A27K also rescued directional migration, suggesting that ROS generation required for directionality does not depend on Rac1 and that Rac1 has alternative essential function(s) in directionality that are not related to ROS signaling. 15,33 To determine whether there is redundancy between Rac1 and Rac2 in ROS generation, we transduced neutrophils that are deficient for both Rac1 and Rac2 (double knockout) with combinations of Rac1-WT, Rac1-A27K, Rac2-WT, and Rac2-A27K ( Figure 4B,D) Transduction efficiencies were verified (supplemental Figure 6). Transduction of the wild-type versions of both Rac isoforms or a single wild-type isoform in combination with the A27K version of the other isoform was able to rescue directionality in Rac1/2-null cells.
However, when both isoforms contained the A27K mutation, cells displayed random migration, confirming that Rac-mediated generation of ROS through at least one Rac isoform is required for directional sensing during chemotaxis. Because H 2 O 2 has been reported to act as a chemoattractant for neutrophils in zebrafish 34 and murine neutrophils, 35 we investigated whether H 2 O 2 could serve as an exogenous source of ROS to rescue directionality in Rac1/2-null cells. Indeed, Rac1/2-null cells transduced with Rac1-A27K and Rac2-A27K regained some ability to migrate toward a combined gradient of fMLP and H 2 O 2 , whereas fMLP or H 2 O 2 alone was not sufficient for directional migration ( Figure 4C-D).
To determine whether Rac-mediated ROS generation is important for inhibiting PTEN activity, we performed enzymatic activity assays using neutrophils from Rac1-null and Rac2-null mice. In agreement with the chemotaxis results, PTEN activity was increased in both Rac1-null and Rac2-null cells after fMLP stimulation, whereas activity was decreased in wild-type cells (supplemental Figure 7). This is in accordance with earlier data showing that phospho-AKT was decreased in Rac1-null and Rac2-null cells, 14 which can be explained by increased PTEN activity. Indeed, cells deficient for Rac1/2 or gp91 display decreased oxidation of PTEN upon fMLP stimulation (supplemental Figure 2). We conclude from (A) Wild-type neutrophils were pretreated with NAC (10mM) or Trolox (100M) for 10 minutes after stimulation with 1M fMLP for 30, 60, and 120 seconds. Phosphorylation of AKT (Thr-308) and total AKT were detected by immunoblot assay. (B) The ratio of phospho-AKT/total AKT was determined using densitometry, and the average of 3 independent experiments Ϯ SD is shown. (C) Subcellular localization of a GFP-tagged PH domain of AKT was monitored in live neutrophils. Upon stimulation with fMLP (asterisk depicts the position of the microcapillary pipette tip), PH-AKT-GFP is recruited to the leading edge (panel 2). At t ϭ 82 seconds (panel 3), NAC (10mM) was added to the incubation medium, inducing redistribution of PH-AKT-GFP. Using image analysis software, density profiles were generated and the relative accumulation at the front and back of the cells was quantified (last panel, n Ͼ 10 cells). (D-E) Phosphorylation of AKT (Thr-308) was quantified by immunoblot assay in gp91 Ϫ/Ϫ neutrophils (E) and control cells (C57/ BL6; D). Cells were treated with 50M H2O2, 10mM NAC, and 100M Trolox, as indicated, and stimulated with 1M fMLP for 1 minute. Bar diagrams represent the average of 3 independent experiments Ϯ SD. *P Ͻ .05. BLOOD, 1 DECEMBER 2011 ⅐ VOLUME 118, NUMBER 23 For personal use only. on July 24, 2018. by guest www.bloodjournal.org From  these experiments that Rac-mediated activation of NOX2 is required for redox regulation of PTEN.
The results of several studies have supported the importance of small GTPases in the regulation of directional migration in neutrophils. 14,36 Although Rho GTPases have been shown to be activated downstream of PtdIns(3,4,5)P 3 via their respective guanine nucleotide exchange factors (GEFs) in neutrophils, 37,38 it is not clear how they act as upstream regulators of PtdIns(3,4,5)P 3 levels. 6 Our data demonstrate that Rac is able to inhibit PTEN via NOX2-mediated ROS production and thereby increase PtdIns(3,4,5)P 3 levels. This model also suggests that Rac activation itself depends on ROS, because they would be part of the same feedback loop. To test this, Rac1 activity was assessed in fMLPstimulated cells that were pretreated with NAC ( Figure 5A-B). Stimulation of neutrophils with fMLP induced a 2-fold increase in GTP-Rac1, as determined by PBD-PAK pull-down assays. In NAC-treated cells, this increase was significantly decreased, indicating that activation of Rac1 indeed depends on ROS signaling. Similar results were obtained in Rac2 activity assays (data not shown). In a similar fashion, we found that Rac1 activity was decreased in gp91 Ϫ/Ϫ cells upon fMLP stimulation, whereas control cells (C57/BL6) displayed robust activation of Rac1 ( Figure 5C). We demonstrated recently that Rac1 induces actin polymerization downstream of the fMLP receptor, which is important for directional migration in neutrophils. 14,15 Because Rac1-mediated actin polymerization depends on uncapping of existing barbed ends, 15 we investigated whether redox regulation of Rac1 activity also regulates actin filament-uncapping activity. Uncapping of actin filaments was assessed in neutrophils pretreated with Trolox or mock treated and stimulated. Upon stimulation with fMLP, cells were partially permeabilized and the capping protein CapZ was quantified in both supernatant and cell pellet. 15 Upon stimulation with fMLP, CapZ was released from actin filaments ( Figure 6A-B). However, when cells were treated with Trolox, fMLP-induced uncapping decreased significantly, demonstrating a requirement for ROS in uncapping. To confirm this, we quantified uncapping activity in gp91 Ϫ/Ϫ cells and wild-type control cells (C57/BL6). fMLP induced uncapping of filaments in polymorphonuclear cells derived from C57/BL6, which could be inhibited by the ROS scavenger Trolox ( Figure 6C). Moreover, exogenous addition of H 2 O 2 increased uncapping activity. In gp91 Ϫ/Ϫ cells, the release of CapZ in response to fMLP was decreased compared with C57/BL6 wild-type cells ( Figure 6C). Treatment with H 2 O 2 was able to partially restore uncapping activity. These experiments demonstrate that ROS generation and uncapping activity are functionally linked in fMLP-stimulated neutrophils. We therefore postulate a redox-regulated feedback mechanism in which PTEN activity is inhibited by ROS, which promotes the accumulation of PtdIns(3,4,5)P 3 and subsequent Rac activation and uncapping ( Figure 6C).

Discussion
Our data support a model in which Rac-mediated NOX2 activation is required to inhibit PTEN, resulting in the accumulation of PtdIns(3,4,5)P 3 , and this positive feedback loop stimulates Rac activation itself and promotes directional migration in neutrophils ( Figure 6C). The role of PTEN in chemotaxis has been somewhat controversial, with some studies suggesting a role for PTEN in directionality or migration and others finding PTEN to be dispensable for chemotaxis. [39][40][41] These discrepancies likely reflect differences in experimental setup, such as in vitro versus in vivo studies, differences in chemoattractant gradients, and the activation state of cells. In addition, the time frame of observation is important. We analyzed directionality during the first 15 minutes after fMLP exposure. Indeed, PI3K activity was shown to be specifically important during the initial polarization of neutrophils, but dispensable during later phases of migration. 42 Furthermore, parallel PtdIns(3,4,5)P 3 -independent mechanisms, such as the p38 MAPK pathway, add additional complexity to the signaling events regulating directional migration in neutrophils. 43 We propose a model in which active PTEN would normally negatively contribute to regulation of directional sensing by blocking the development/ maintenance of the PtdIns(3,4,5)P 3 gradient at the leading edge. PTEN therefore acts as the "brake" to PI3K's "accelerator" role. Our data support the conclusion that ROS are required at the earliest phase of chemotactic compass activation to deactivate PTEN, which is initially distributed throughout the cytoplasm before cell polarization. We propose that the initial activation of the chemotactic compass-the PtdIns(3,4,5)P 3 gradient-requires ROSmediated inhibition of PTEN activity to allow for the initial early development of the PtdIns(3,4,5)P 3 gradient and the subsequent directional sensing. Our data using the PTEN inhibitor to rescue activation of the chemotactic compass when ROS are sequestered or their generation is blocked support our conclusion that ROSmediated deactivation of PTEN is required for initial chemotaxis directional sensing.
Feedback loops are very common in biologic systems and can amplify internal or external signals. 44 It has been proposed that BLOOD, 1 DECEMBER 2011 ⅐ VOLUME 118, NUMBER 23 For personal use only. on July 24, 2018. by guest www.bloodjournal.org From robust PtdIns(3,4,5)P 3 accumulation requires a positive feedback mechanism. 6,9 Our data support such a mechanism and also suggest that ROS could fulfill a central role therein. Interestingly, ROS were recently identified as being important for directionality in a study of small-molecule inhibitors of directional migration. 45 That study also showed that neutrophils from chronic granulomatous disease patients, which lack significant ROS production, also exhibit defects in chemotaxis. In addition, an intriguing finding that neutrophils are able to migrate along H 2 O 2 gradients could be explained by this mechanism. 34 Because H 2 O 2 is membrane permeable, it is feasible that it could inhibit intracellular PTEN and stimulate self-amplification of the PtdIns(3,4,5)P 3 gradient and thereby facilitate directional sensing. In this case, H 2 O 2 could act as a chemoattractant and intracellular second messenger. In our experiments, exogenous H 2 O 2 was indeed able to partially rescue the NOX2 phenotype. However, it is important to note that although we used relatively low concentrations of exogenous H 2 O 2 , actual subcellular concentrations and localization might differ from endogenously produced ROS and trigger additional signaling pathways. 46 To elucidate the role of Rac GTPases in ROS-mediated PtdIns(3,4,5)P 3 accumulation, we used Rac-effector mutants that are unable to activate NOX2. Although Rac1 is required for the chemotactic compass in neutrophils and Rac2 is mainly responsible for actin polymerization and ROS generation, 14,47 we did not identify a unique role for Rac1 in activating NOX2 in directional migration ( Figure 4). Instead, Rac1 and Rac2 were both able to activate the ROS generation required for directional sensing and are therefore upstream regulators of PtdIns(3,4,5)P 3 . This might not be entirely surprising, because it has been shown that both Rac1 and Rac2 can activate NOX2, and that in Rac2-null neutrophils, Rac1 is still able to generate ROS albeit at lower levels. 48,49 The fact that Rac1-null cells are defective in directional migration is therefore due to isoform-specific functions that are independent of the ability of Rac1 to activate NOX2. 15,33 Because most guanine nucleotide exchange factors for Rac require PtdIns(3,4,5)P 3 for their activation, 37,38 Rac GTPases are also downstream effectors of PtdIns(3,4,5)P 3 . The implication that downstream Rac activation consequently depends on ROS was indeed supported by our data ( Figure 5). Moreover, fMLP-induced uncapping of actin filaments is a downstream event of Rac1 and is also inhibited by scavenging of ROS. These observations support a dual role of Rac in chemotaxis, both upstream (activation of NOX2) and downstream of PtdIns(3,4,5)P 3 (eg, uncapping of actin). We therefore propose a role for ROS in directional sensing through oxidation of PTEN and subsequent regulation of PtdIns(3,4,5)P 3 . Moreover, we identified a redox-regulated positive feedback loop that regulates Rac activity and its downstream effectors that facilitate directional migration.