Regulation of PTEN activity by p38δ-PKD1 signaling in neutrophils confers inflammatory responses in the lung

We first explored the expression of p38δ in different immune compartments and cells. We confirmed high expression of p38δ in neutrophils at the mRNA and protein level (Fig. 1, a and b). Strikingly, however, p38δ was undetectable in peritoneal macrophages on both mRNA and protein levels (Fig. 1 b and not depicted). Furthermore, very low expression of p38δ was detected in spleen, thymus, and in CD4⁺ or CD8⁺ splenocytes (Fig. 1, a and b). This is in strong contrast to p38α, which is ubiquitously and highly expressed in all immune compartments and cells (Kumar et al., 2003). Therefore, this observation prompted us to assess the function of p38δ in neutrophils in vivo. We used sterile peritonitis in global p38δ knockout (p38δ^−/−) and wild-type (p38δ^+/+) control mice as a model of inflammation. We observed markedly reduced neutrophil recruitment to the peritoneal cavity in p38δ^−/− mice as compared with control p38δ^+/+ mice. Importantly however, recruitment of macrophages and lymphocytes remained unaffected (Fig. 1 c). Blood leukocyte and bone marrow composition were similar in p38δ^−/− and p38δ^+/+ mice (Table S1 and Table S2).

To study consequences of reduced neutrophil recruitment, we next used intratracheal instillation of endotoxin in mice as a widely accepted in vivo model of ALI (Matute-Bello et al., 2008). We compared lung damage in p38δ^−/− to p38δ^+/+ control mice and found that p38δ^−/− mice were largely protected against endotoxin-induced damage and lung dysfunction. i.v. administration of Evans blue dye indicated pronounced reduction in vascular injury in p38δ^−/− lungs upon ALI as compared with p38δ^+/+ lungs (Fig. 1 d). Lung histology showed considerable alveolar damage and inflammatory cell infiltration in p38δ^+/+ mice in response to LPS. In contrast, lungs of p38δ^−/− mice were almost indistinguishable from saline-treated control mice (Fig. 1 e). Albumin levels in BAL fluids of p38δ^−/− lungs were significantly lower, indicating largely maintained vascular integrity (Fig. 1 f). Pulmonary edema formation was markedly lower in p38δ^−/− mice (Fig. 1 g). Remarkably, lung function assessed by arterial blood oxygen pressure was significantly impaired in p38δ^+/+ mice but was unaltered in p38δ^−/− mice (Fig. 1 h). Quantification of neutrophils in BAL revealed that their alveolar accumulation was notably diminished in p38δ^−/− mice as compared with p38δ^+/+ mice (Fig. 1 i). The neutrophil chemoattractants CXCL1 and CCL3 were induced to a similar extent in p38δ^−/− and p38δ^+/+ lungs (unpublished data). Lower BAL neutrophil numbers were also observed upon Gram-negative bacterial lung infection in p38δ^−/− as compared with p38δ^+/+ lungs (Fig. 1 j). Importantly, bacterial burden was markedly higher in p38δ^−/− lungs and BAL fluid (Fig. 1, k and l). Substantial splenic bacterial dissemination was only observed in p38δ^−/− mice (Fig. 1 m).

To exclude contributions of p38δ in cells of nonmyeloid origin, we deleted p38δ specifically in the myeloid compartment by crossing floxed p38δ with LysM-Cre mice. Deletion of p38δ in neutrophils was efficient using the latter deletion approach (unpublished data). Myeloid-restricted deletion of p38δ (p38δ^Δmy) was sufficient to reduce neutrophil recruitment to the inflamed peritoneal cavity as compared with control mice without affecting recruitment of macrophages and lymphocytes (unpublished data). Moreover, p38δ^Δmy mice showed attenuated neutrophil accumulation during ALI resulting in decreased pulmonary tissue damage (Fig. 1, n and o).

These data indicate that p38δ regulates neutrophil recruitment during lung inflammation to impact on pulmonary damage and bacterial clearance.

Defective extravasation, occurring in a multistep process, including capturing, rolling, arrest, firm adherence, and transmigration (Nathan, 2006), might underlie reduced neutrophil recruitment in mice lacking p38δ. We therefore quantified extravasation steps by intravital microscopy (Rossaint et al., 2011). For this purpose, we used global p38δ knockout mice, myeloid-specific p38δ knockout, bone marrow transfer from p38δ^+/+ or p38δ^−/−, as well as mixed bone marrow transfer from WT mice expressing GFP under the control of the LysM promoter (LysM-GFP) and p38δ^−/− mice into C57BL/6J WT mice. Hemodynamic parameters and systemic leukocyte counts were similar between experimental groups in all intravital microscopy experiments (unpublished data).

Numbers of rolling leukocytes were similar upon stimulation with chemoattractant N-formylmethionyl-leucyl-phenylalanine (fMLP) in p38δ^−/−, p38δ^Δmy and chimeric mice as compared with control mice (data not shown). However, the number of adherent leukocytes upon stimulation with fMLP was around 50% lower in p38δ^−/−, in p38δ^Δmy and in chimeric mice as compared with control mice (Fig. 2 a and data not shown).

Emigration of neutrophils from the circulation requires efficient chemokine-directed transmigration across the endothelial cell layer (Nathan, 2006). The numbers of transmigrated leukocytes was significantly reduced by 70% in p38δ^−/− (Fig. 2 b), in p38δ^Δmy (Fig. 2 c), in mice with single bone marrow transfers (Fig. 2, d and e) as well as mice with mixed bone marrow transfers (Fig. 2 f) as compared with corresponding control mice upon chemokine stimulation.

The reduction in transmigration is considerably higher than the one observed for cell adherence, and thus suggests that the process of migration across the endothelium itself is affected by p38δ. We therefore quantified the capacity of p38δ knockout neutrophils to transmigrate through endothelial cells in vitro, thereby analyzing transmigration independent of blood circulation. Transmigration of p38δ^−/− neutrophils toward a chemokine across an endothelial cell layer in vitro was significantly reduced as compared with p38δ^+/+ cells (Fig. 2 g).

Considering the defects in transmigration of p38δ^−/− neutrophils, we reasoned that p38δ might affect neutrophil chemotaxis on a single-cell level. Therefore, we used time-lapse microscopy to trace neutrophils during chemotaxis in vitro. Although p38δ^+/+ neutrophils readily migrated toward a gradient of fMLP, p38δ^−/− cells showed a profound reduction in migration (Fig. 2 h and Videos 1 and 2). Distances covered by migrating p38δ^−/− neutrophils were reduced compared with p38δ^+/+ cells due to lower velocity (Fig. 2, i and j). Furthermore, p38δ^−/− neutrophils showed a significant aberration in directionality (Fig. 2 k). Thus, p38δ is critical for neutrophil extravasation, regulating adherence, cell motility, and directional interpretation of chemoattractant gradients.

We previously demonstrated that p38δ inhibits the activity of its target PKD1 through phosphorylation on serine 397/401 (Sumara et al., 2009). Strikingly, this regulatory mechanism seemed to also exist in neutrophils, as we found markedly enhanced activity of PKD1 in p38δ^−/− neutrophils as compared with p38δ^+/+ cells (Fig. 3 a). PKD1, among other protein kinase C family members, is activated through PLC (Wood et al., 2005). Inhibition of PLC by the compound U73122 was sufficient to restore accumulation of neutrophils in p38δ^−/− mice during sterile peritonitis (Fig. 3 b). To corroborate the importance of PKD1, we subsequently addressed the relationship between p38δ and PKD1 using conditional gene targeting in mice. We performed ALI with mice that lack p38δ, PKD1 or both genes in the myeloid cell lineage (p38^Δmy, PKD1^Δmy and p38δ PKD1^Δmy mice). Again, p38δ^Δmy mice showed mitigated lung inflammation as compared with control mice. In strong contrast, PKD1^Δmy mice displayed markedly enhanced recruitment of neutrophils and exacerbated lung injury (Fig. 3, c–e). Enhanced recruitment of neutrophils in PKD1^Δmy mice was also observed in the sterile peritonitis model (unpublished data). Importantly, p38δ PKD1^Δmy mice showed similarly enhanced levels of lung injury and cell recruitment as compared with PKD1^Δmy mice (Fig. 3, c–e) providing genetic evidence in vivo that PKD1 is downstream of p38δ. These results suggest that p38δ negatively regulates PKD1 signaling to modulate neutrophil recruitment in vivo.

We subsequently addressed effects of PKD1 deletion in myeloid cells on steps of extravasation using mice with single and mixed bone marrow transfers for intravital microscopy. Micro-hemodynamic parameters and systemic leukocyte counts were comparable across the experimental groups in these experiments (data not shown). Interestingly, leukocyte rolling flux was enhanced in chimeric as compared with control mice (Fig. 4 a and not depicted). In chimeric mice, adherent leukocyte numbers were significantly increased as compared with control mice in response to CXCL1/keratinocyte-derived chemokine (KC; Fig. 4 b and not depicted). Moreover, leukocyte transmigration was significantly elevated as compared with control mice (Fig. 4, c–e). Transmigration of PKD1^Δmy neutrophils toward a chemokine across an endothelial cell layer in vitro was significantly enhanced as compared with PKD1^f/f cells (Fig. 4 f).

We then aimed at corroborating the functions of PKD1 in neutrophil chemotaxis using time-lapse microscopy. Strikingly, neutrophils isolated from PKD1^Δmy mice showed markedly enhanced migration in a gradient of fMLP (Fig. 4 g and Videos 3 and 4). Primary PKD1^Δmy neutrophils migrated with enhanced velocity (Fig. 4 h), covering larger distances (Fig. 4 i). Moreover, these cells had defective directional sensing (Fig. 4 j). These results show that PKD1 elicits opposite effects on neutrophil extravasation, adherence, and chemotaxis as compared with p38δ.

Efficient directed migration of neutrophils requires polarization of signaling molecules, for example by generating a gradient of specific phosphoinositol phosphates through a localized action of PTEN and PI3K at the back and the front of the cell, respectively (Stephens et al., 2008). Interestingly, cells from myeloid-specific PTEN knockout mice showed similarly enhanced neutrophil recruitment as cells lacking PKD1 (Subramanian et al., 2007; Heit et al., 2008b; Sarraj et al., 2009). Therefore, we addressed whether PTEN abundance, subcellular distribution, and activity depended on PKD1. Immunofluorescence staining showed that localization of PTEN in PKD1^Δmy cells was not restricted to the uropod as seen in control PKD1^f/f cells (Fig. 5 a). Quantification corroborated impaired PTEN localization in PKD1^Δmy cells as compared with control PKD1^f/f cells (Fig. 5 b). However, polarization of F-actin to the front was similar in PKD1^Δmy cells and PKD1^f/f cells (Fig. 5, a and c). Specificity of PTEN staining was confirmed using PTEN-negative PC3 cells (unpublished data).

To assess PTEN activity, we measured in vitro PIP₃ hydrolysis by PTEN immunoprecipitated from primary neutrophils. Activity of PTEN was lower in primary PKD1^Δmy neutrophils upon stimulation with fMLP as compared with control PKD1^f/f cells (Fig. 5 d). Total PTEN protein levels were similar in PKD1^Δmy and PKD1^f/f primary neutrophils. Phosphorylation of Akt downstream of PIP₃ formation is a key signaling event, which is inhibited by conversion of PIP₃ to PIP₂ by PTEN (Cantley, 2002). We analyzed phosphorylation of serine 473 on Akt, which serves as an indicator of upstream PI3 kinase activity and PIP₃ levels in neutrophils (Strassheim et al., 2004; Li et al., 2009; Prasad et al., 2011). Primary PKD1^Δmy neutrophils showed markedly stronger induction of Akt phosphorylation during chemokine stimulation as compared with control PKD1^f/f cells (Fig. 5 e).

Neutrophils lacking p38δ show increased activity of PKD (Fig. 3 a). Thus, opposite results for PTEN and AKT activity are to be expected. Indeed, activity of immunoprecipitated PTEN was higher in p38δ^−/− neutrophils upon stimulation with fMLP as compared with p38δ^+/+ cells (Fig. 5 f). Moreover, p38δ^−/− neutrophils showed lower levels of Akt phosphorylation during chemokine stimulation as compared with p38δ^+/+ cells (Fig. 5 g). Collectively, these results imply that modulation of p38δ and PKD1 activity conversely regulates PTEN function in migrating neutrophils.

We next aimed at finding direct biochemical evidence for regulation of PTEN by PKD1. Therefore, we searched for novel candidate targets of PKD1 that potentially link PKD1 to PTEN. We used a specific PKD phosphorylation consensus site in an in silico screen (Table S3). This analysis revealed the product of the pik3r1 gene p85α, which is a regulatory subunit of class I PI3 kinases (Cantley, 2002). Recent studies showed that p85α is also a positive regulator of PTEN activity (Rabinovsky et al., 2009; Chagpar et al., 2010), indicating that p85α exerts both positive and negative regulation of PI3 kinase signaling. Indeed, we found lipid phosphatase activity of PTEN markedly reduced in adherent p85α-knockdown HL60 cells as compared with nonsilencing shRNA control cells (Fig. 6 a). p85α directly binds and enhances PTEN activity through the Bcr homology (BH) domain (Chagpar et al., 2010). The BH domain contains the PKD1 consensus site at serine 154 (S154) that we identified in our in silico screen. This site is conserved throughout mammalian pik3r1 genes (unpublished data). In an in vitro kinase assay with PKD1 and radioactively labeled ATP, recombinant p85α showed incorporation of radioactivity, which was suppressed by an inhibitor of PKD1 (Gö6976; Fig. 6 b). Phosphopeptide mapping by mass spectrometry identified S154 phosphorylation on p85α expressed in Sf9 cells, as well as other previously published phosphorylation sites (Oppermann et al., 2009). The additional S565 phosphorylation detected in the in vitro reaction may reflect an artifact, as it was not detected in Sf9 cells and does not correspond to the phosphorylation consensus site of PKD (Fig. 6 c and Table S4). We next produced recombinant human WT and mutant serine 154 to Alanine (S154A) p85α and subjected these proteins to in vitro kinase assay with PKD1. Phosphoincorporation into mutant S154A p85α was markedly reduced, confirming S154 as phosphorylation site on p85α in vitro (Fig. 6 d).

Next, we addressed whether phosphorylation of p85α affects its interaction with PTEN and the activity of PTEN. Addition of active PKD1 enhanced binding of purified HIS-p85α to recombinant GST-PTEN in vitro (Fig. 6 e). Consistently, co-transfection of active PKD1 into cells transiently expressing PTEN and p85α enhanced co-immunoprecipitation of p85α with PTEN (Fig. 6 f). Co-expression of WT p85α, S154A p85α, and phospho-mimicking p85α (S154D) with PTEN, followed by immunoprecipitation, revealed enhanced interaction of PTEN with the phospho-mimicking p85α, whereas the interaction was decreased with S154A p85α as compared with WT p85α (Fig. 6 g). Recombinant S154D p85α showed similarly enhanced binding with recombinant GST-PTEN in vitro as compared with WT p85α (unpublished data).

We then measured lipid phosphatase activity of PTEN in vitro with addition of different forms of p85α. Addition of WT p85α significantly enhanced PTEN activity as previously shown (Chagpar et al., 2010). Addition of S154D p85α significantly potentiated PTEN activity seen with addition of WT p85α, although it was significantly reduced when S154A p85α was added instead (Fig. 6 h). To confirm phosphorylation, we generated a monoclonal phospho-S154-p85α antibody. Expression of constitutive active PKD enhanced phospho-S154 signal of WT p85α, whereas kinase-dead PKD had no effect. Importantly, phospho-S154 signal was abolished when S154A p85α was expressed, indicating the specificity of the antibody (Fig. 7 a). We addressed whether phosphorylation occurs in primary neutrophils stimulated with fMLP and whether it is dependent on the presence of p38δ or PKD1. Indeed, phosphorylation of S154 p85α was induced in primary neutrophils upon chemokine-stimulation (Fig. 7, b and c). However, phosphorylation of S154 was not detected in primary neutrophils isolated from PKD1^Δmy mice (Fig. 7 b). In contrast, neutrophils from p38δ^−/− mice showed enhanced phosphorylation of p85α as compared with p38δ^+/+ cells, which is consistent with increased PKD activity in the absence of p38δ (Fig. 7 c). Overall, these experiments suggest that PKD1-dependent phosphorylation of p85α occurs in stimulated primary neutrophils. Phosphorylation of p85α may lead to its localization in the uropod of neutrophils, where it binds and activates PTEN. Polarized colocalization of p85α and PTEN was frequently found in the uropods of PKD1^f/f cells, which was much less evident in PKD1^Δmy cells. In contrast, PKD1^Δmy cells localized p85α preferentially at the front of the cell, whereas PTEN was dispersed all over the cell body (Fig. 7 d).

To address whether S154 phosphorylation of p85α is involved in chemotaxis of neutrophils, we expressed WT or mutant p85α in HL60 cells (Fig. 7 e). Live cell tracking of transfected chemotaxing HL60 cells revealed that expression of S154A p85α enhanced migration, whereas S154D p85α expression decreased migration (Fig. 7 f and Videos 5–8), with significantly higher mean velocity of S154A p85α-expressing cells and lower velocity of S154D p85α-expressing cells (Fig. 7 g). WT p85α-expressing cells showed similar migratory behavior compared with empty vector control cells (Fig. 7, f and g). Furthermore, WT p85α-expressing cells showed transwell migration similar to that of empty vector-transfected cells, whereas cells expressing S154A p85α mutant showed higher and S154D p85α-expressing cells showed lower migratory capacity compared with WT and control cells in the same assay (Fig. 7 h). These results indicate that phosphorylation of p85α on S154 is involved in regulation of neutrophil migration.

In summary, phosphorylation of p85α on serine 154 by PKD1 increases the ability of p85α to induce PTEN activity and regulate migration (Fig. 7 i). Thus, modulation of p38δ and PKD1 activities may change PTEN activity in migrating neutrophils, altering neutrophil recruitment during acute inflammatory insults.

The degree of inflammation and associated organ damage is a result of complex pro- and antiinflammatory responses. In acute lung inflammation, this involves tight regulation of neutrophil recruitment and migration in a cell-autonomous manner. Neutrophil recruitment is governed by signals of extracellular pro- and antiinflammatory cues. Appropriate signaling in neutrophils downstream of these stimuli is thus essential to resolve inflammatory insults without causing inappropriate organ damage. We suggest that neutrophils might dynamically adapt to pro- and antiinflammatory inputs by modulating activities of p38δ and PKD1. We also provide evidence that this signaling pathway regulates PTEN, a master regulator of neutrophil recruitment. Direct regulation of PKD1 by p38δ and corresponding effects on PTEN activity allow for efficient regulation of neutrophil migration (Fig. 7 i).

A previous study identified expression of p38δ in neutrophils (Nick et al., 1999). Yet, a nonredundant function of p38δ in this cell type was not evidenced thus far. We thus provide first in vivo evidence that p38δ regulates recruitment and chemotaxis of primary neutrophils, thereby controlling pulmonary tissue injury and bacterial clearance. Because our deletion approaches in mice could not entirely exclude a potential function of p38δ in macrophages that contributed to the observed phenotypes, we performed experiments in mixed bone marrow chimeric mice to support our results with global and conditional gene deletion. p38δ-deficient neutrophils showed reduced recruitment compared with WT control cells in these mice, supporting a cell-autonomous role of p38δ in neutrophils. Furthermore, isolated p38δ-deficient neutrophils show a clear defect in chemotaxis in vitro, which is in line with reduced recruitment of neutrophils in vivo. p38δ is highly expressed in neutrophils, whereas its expression in macrophages is comparatively low. Very low expression of p38δ in macrophages as opposed to p38α has been confirmed by a recent study (Risco et al., 2012). This study also demonstrated that macrophages lacking p38δ exhibited minor changes in cytokine expression in response to LPS and that additional inactivation of p38γ was necessary to obtain more significant effects. Hence, we conclude that p38δ-dependent reduction of neutrophil recruitment in response to all tested inflammatory stimuli and mitigation of lung injury in response to LPS is likely to be independent of effects by macrophages.

Our study also provides strong experimental evidence that effects seen in p38δ-deficient mice are caused by increased PKD1 activity in neutrophils. PKD1 is expressed in both, macrophages and neutrophils. In fact, PKD1 in murine peritoneal macrophages and isolated bone-marrow–derived dendritic cells was shown to promote LPS-stimulated release of proinflammatory mediators (Park et al., 2008). Importantly, these two studies lack any data on inflammation in vivo. Using myeloid-specific deletion of PKD1, we found strikingly increased inflammatory injury in the lung in response to LPS. This observation is perfectly in line with enhanced recruitment of neutrophils to the lung and enhanced extravasation indicated by intravital microscopy in different mouse models, including mixed bone marrow chimera. In addition, the results in mixed bone marrow chimeras strongly support a cell-autonomous role of PKD1 in neutrophils rather than in other myeloid lineages. Considering a potential proinflammatory role of PKD1 in macrophages as opposed to markedly enhanced lung inflammation in myeloid-specific PKD1-null mice, we conclude that the observed effects in our study are independent of macrophages and instead largely caused by an uncontrolled influx of neutrophils.

Our results implicate PKD1 in neutrophil chemotaxis, regulating both directionality and speed of migration. PKD1^Δmy cells show enhanced cell speed, which may compensate for a defect in directionality of movement, and thus result in efficient transmigration of PKD1-deficient cells in vivo and in vitro. In line with this assumption, we found higher numbers of extravasated neutrophils in in vivo inflammation models in PKD1^Δmy mice. Our study is in line with studies that investigated tumor cell motility in the absence and presence of PKD1. These studies provide a mechanism that involves PKD1-mediated negative regulation of slingshot phosphatase SSH1L, which dephosphorylates and activates actin-regulator cofilin. Reduction of PKD1 led to reduced phosphorylation of cofilin and enhanced migration in fibroblasts (Eiseler et al., 2009; Peterburs et al., 2009). We did not detect differences in cofilin phosphorylation in PKD1-deficient neutrophils (unpublished data). In contrast to neutrophils, adherent cells such as fibroblasts and epithelial cells tend to form stable focal adhesions and stress fibers during migration (Stephens et al., 2008). Thus, given these differences in modes of migration, the existence of specific signaling events in different cell types is likely.

In this study, we also discovered p85α as a novel target of PKD1, providing a stunning link between PKD and PI3 kinase–PTEN signaling. p85α serves as regulatory subunit for the PI3 kinase catalytic subunits p110α, β, and δ, whereas p101 is the regulatory subunit for p110γ (Okkenhaug and Vanhaesebroeck, 2001; Vanhaesebroeck et al., 2001). Neutrophils express p110α, δ, and γ (Sadhu et al., 2003a; Condliffe et al., 2005). p110γ is responsible for initial PIP₃ accumulation at the leading edge upon chemokine stimulation (Condliffe et al., 2005; Van Keymeulen et al., 2006). Amplification of PIP₃ levels, however, depends on p110δ (Sadhu et al., 2003b; Condliffe et al., 2005). p85 has a dual function on p110; it promotes activation of PI3 kinase, but it also inhibits its activity through p85–p110 complex sequestration (Fruman et al., 1998; Vanhaesebroeck et al., 2001). Recent studies showed that p85α can also directly bind and enhance the activity of PTEN (Rabinovsky et al., 2009; Chagpar et al., 2010). Direct binding of PTEN to the BH domain of p85α promotes the activity of PTEN (Chagpar et al., 2010). We found that PKD1 phosphorylates p85α at serine 154 within the BH domain and that phosphorylation of S154 enhances binding of p85α to PTEN, and thus increases the capacity of p85α to activate PTEN. S154 phosphorylation also occurs in vivo upon chemokine stimulation of neutrophils. Overexpression of a nonphosphorylatable S154A p85α mutant was sufficient to enhance chemotaxis and migration speed of neutrophil-like HL60 cells, indicating that phosphorylation of p85α is required to control neutrophil recruitment. Overexpressing the phospho-mimicking mutant S154D p85α reduced the migratory response of HL60 cells and particularly lowered cell speed, implicating that S154 phosphorylation is an important new mechanism in migrating cells. The molecular link between PKD1, p85α, and PTEN might also be essential in other processes in which both PKD1 and PI3 kinase signaling play an essential role, including invasion and neovascularization of tumors (Di Cristofano and Pandolfi, 2000).

Our study also revealed specific localization of endogenous p85α in primary neutrophils. Localization of p85α to the back of the cell was dependent on phosphorylation by PKD1. Nonphosphorylated p85α preferentially localized to the front of the cell. PTEN cannot bind to nonphosphorylated p85α at the leading edge, but it binds specifically to phosphorylated p85α at the back of the cell. Localization of the phosphorylation site in the BH domain implies that it will not directly interfere with binding to catalytic subunits of PI3K. High affinity of PTEN to phosphorylated p85α may thus create competition with p85α binding to catalytic subunits of PI3K, specifically preventing PIP3 formation at the rear of the cell. It is, however, currently unclear how phosphorylated and nonphosphorylated p85α distinguishes the front and back of the cell. Specifically localized adaptor proteins might be necessary.

Altogether, phosphorylation of p85α results in its specific localization at the rear of the cell, leading to enrichment of PTEN and enhancement of its activity at the uropod. This may occur during the initial steps of polarization, at which point p101, and not p85α, is needed in the leading edge. Non-phosphorylated p85α localizes to the front of the cell. This may become important at later steps of polarization, when amplification of PIP3 requires p110δ, with the latter binding to p85α.

The involvement of PI3 kinase signaling in adherence is a controversial topic among researchers, which is probably a result of different experimental settings used in previous studies. p110γ-deficient neutrophils showed strongly reduced chemokine-stimulated adherence (Ferguson et al., 2007; Heit et al., 2008a). Another study found that p110γ or p110δ are dispensable for neutrophil adherence under shear stress (Puri et al., 2005). A recent study reported no changes in the adherence of neutrophils upon inhibition of PI3 kinase or deletion of PTEN in vivo (Sarraj et al., 2009). We found reduced chemokine-stimulated adherence of neutrophils in p38δ^−/− mice and increased adherence of PKD1^Δmy neutrophils in vivo using intravital microscopy. Given the possibility that neither PI3 kinase nor PTEN are essential in adherence of neutrophils under physiological shear stress, a PTEN-independent mechanism might account for the role of p38δ and PKD1 in adherence. Nevertheless, PTEN was described as a critical regulator of cell migration (Leslie et al., 2008). Spatial and temporal regulation of PTEN is involved in regulation of chemotaxis (Funamoto et al., 2002). The requirement of PTEN in chemotaxis of neutrophils in response to fMLP is controversial. In fact, migration of murine PTEN-deficient neutrophils in response to fMLP was indistinguishable from control cells, whereas granulocytes lacking the lipid phosphatase SH2 domain-containing inositol SHIP-1 showed clear defects (Nishio et al., 2007). This study thus establishes SHIP-1 as a predominant regulator of fMLP-mediated neutrophil chemotaxis. Nonetheless, occurrence of polarization of PTEN in neutrophils in response to fMLP is widely accepted (Funamoto et al., 2002), and PTEN is indeed required for chemotaxis of Dictyostelium (Iijima and Devreotes, 2002), an organism sharing many mechanisms of migration with neutrophils (Stephens et al., 2008). Based on these observations, investigators continued searching for a function of PTEN in neutrophil chemotaxis. Indeed, an important subsequent study found that PTEN is not required for fMLP-induced neutrophil chemotaxis, but that it regulates migration in response to the chemokine CXCL2. By exposing neutrophils simultaneously to both chemotactic factors, thereby engaging cells to prioritize chemotactic cues, a recent study established a very precise role of PTEN in neutrophil chemotaxis (Heit et al., 2008b). The importance of these findings is highlighted by the fact, indicated by several independent research teams, that recruitment of neutrophils to sites of inflammation in vivo in mice is clearly dependent on the absence and presence of PTEN (Heit et al., 2008b; Li et al., 2009; Sarraj et al., 2009; Schabbauer et al., 2010). Hence, more recent studies established PTEN as a crucial regulator of neutrophil chemotaxis and of their recruitment to inflammatory sites in vivo (Phillipson and Kubes, 2011), confirming PTEN as a potential downstream effector of p38δ-PKD1 signaling. The question remains, however, why chemotaxis of neutrophils lacking p38δ or PKD1 is altered in response to fMLP, if their target PTEN seems not to be required for the latter process. In fact, defects in the polarization and activity of PTEN seen in our experiments are very different from the complete deletion of PTEN that was used in previous studies to investigate its requirement. Therefore, these previous experiments do not challenge our findings. In contrast, our data highlight the importance of proper PTEN localization and efficient PTEN activity in neutrophil chemotaxis to fMLP. However, our data do not exclude a complex role of p38δ and PKD1 in PTEN-dependent neutrophil chemotaxis in response to different chemotactic stimuli. This needs to be addressed further in future studies.

The leading edge of migrating cells contains high levels of PIP₃, whereas the back and sides localize PTEN and PIP₂ (Funamoto et al., 2002). PTEN was shown to bind PIP₂ within membranes, which leads to allosteric PTEN activation (Leslie et al., 2008). It was suggested that PLC acts upstream of PTEN in chemotaxing Dictyostelium (Kortholt et al., 2007). Mechanistically, it was proposed that PIP₂ hydrolysis by PLC regulates PTEN by creating low PIP₂ levels at the cell front (Kortholt et al., 2007). However, PIP₂ is found at much higher levels than other phosphoinositide species (Leslie et al., 2008). Because of the excess of PIP₂, specific activation of PTEN by PLC might require direct activation, in addition to binding site limitation in the membrane, in migrating cells. Indeed, our data support a direct mechanism in which PLC, through PKD1, regulates colocalization and complex formation of p85α and PTEN, and thereby PTEN activity. This mechanism is corroborated by the fact that deletion of PLCβ2/3 in neutrophils leads do similar effects in neutrophil recruitment and chemotaxis as seen in neutrophils deficient for PTEN and PKD1 (Li et al., 2000).

ALI and ARDS are characterized by strong recruitment of neutrophils to lungs in humans (Ware and Matthay, 2000). Clinically, ARDS has a poor prognosis and high mortality worldwide as there is no efficient pharmacologic therapy available that supports intensive care measures (Jain and DalNogare, 2006). Given the inverse effects on tissue injury and pathogen burden, efficient therapeutic interventions at the level of neutrophils might be difficult to achieve. Eventually, implementation of potent antimicrobials together with alteration of neutrophil recruitment through modulation of the p38δ–PKD1 signaling axis may constitute a valuable therapeutic avenue in conditions of acute inflammation including ARDS in humans.

Mice with targeted alleles for p38δ (p38δ^f/f) and p38δ^−/− mice on a C57BL/6J background were described previously (Sumara et al., 2009). To obtain p38δ^Δmy mice, p38δ^f/f mice were crossed with LysM-Cre deleter mice (Lyz2^tm1(cre)Ifo) on a C57BL/6J background (Clausen et al., 1999). Mice with targeted alleles for PKD1 (PKD1^f/f) were described previously (Fielitz et al., 2008). R. Bassel-Duby and E.N. Olson (University of Texas Southwestern, Dallas, TX) provided floxed PKD1 mice. We crossed PKD1^f/f mice on a C57BL/6J background with LysM-Cre mice to obtain myeloid-specific PKD1 knockout mice. To obtain mice with myeloid-specific deletion of both p38δ and PKD1, p38δ^f/f mice were crossed with PKD1^f/f and LysM-Cre mice. Experiments were performed using littermate mice as controls. Mice were housed under specific pathogen–free conditions. The Swiss Federal Veterinary Office approved all animal experiments.

Antibody against p38δ was raised in rabbit by immunization with a peptide specific for murine p38δ (Eurogentec). Antibodies for Tubulin (Sigma-Aldrich), GAPDH, phospho-serine 916 PKD (Cell Signaling Technology), PKD (Santa Cruz Biotechnology, Inc.), FITC-labeled anti–mouse Gr1 (Ly6G/C; Miltenyi Biotec), PE-labeled anti–mouse CD11b (BioLegend), PTEN (Cell Signaling Technology), PTEN-HRP conjugate (Santa Cruz Biotechnology, Inc.), p85 (Cell Signaling Technology), phospho-serine 473 Akt (Cell Signaling Technology), phospho-Threonine 308 Akt (Cell Signaling Technology), pan-Akt (Cell Signaling Technology) are commercially available. Mouse monoclonal anti-Phospho-serine 154 p85α antibody was raised against a peptide surrounding S154 in p85α (kKGLECSTLYRTQ-pSSSNPAELRQLL) and produced using hybridoma technique. fMLP, U73122, fibronectin, fatty acid–free bovine serum albumin, Evans blue, May-Grünwald Giemsa, and Wright-Giemsa solutions were purchased from Sigma-Aldrich.

Age- and sex-matched mice were instilled intratracheally (i.t.) with 50 µg of LPS (E. coli LPS serotype O111:B4; Sigma-Aldrich) in 50 µl saline under anesthesia with ketamine:xylazine (80:8 mg/kg body weight). After 6 h, mice were sacrificed and lungs were lavaged with cold phosphate-buffered saline. RBCs were lysed in 0.8% NH₄Cl 0.1 mM EDTA, pH 7.4. Total cells were counted in bronchoalveolar lavage (BAL) fluids using a hemocytometer. Percentage of neutrophils of total BAL cells was determined by flow cytometry after F_c-Receptor blocking for CD11b and Gr1 with neutrophils as Gr1^hi CD11b^pos cells. Albumin was measured by ELISA (Bethyl Laboratories, Inc). Wet-to-dry weight ratio was determined with lung tissue dried at 70°C. Aortic blood oxygenation was measured on a blood gas analyzer (Bayer Diagnostics). 40 µg/g Evans blue was administered i.v., and lungs were excised after 15 min. Lungs were fixed with paraformaldehyde (PFA; 4%, pH 7.4, in PBS) under 20 mm H₂O and paraffin processed. Histological sections (5 µm) were stained with hematoxylin and eosin.

Age- and sex-matched mice were injected i.p. with 1 ml of 4% thioglycolate broth (BD). After 4 h, mice were sacrificed, and infiltrated cells were recovered in 8 ml PBS. RBCs were lysed. Total cells were counted on a hemocytometer. Ratios of neutrophils, monocytes/macrophages, and lymphocytes were distinguished by morphology on Wright-Giemsa–stained cytospins of 5 × 10⁴ cells or, where indicated, percentages of neutrophils and monocytes/macrophages in peritoneal infiltrates were determined by flow cytometry. Neutrophils were determined as Gr1^hi CD11b⁺ and monocytes/macrophages as Gr1^lo CD11b⁺ cells in a leukocyte scatter gate. For treatment with U73122 (Sigma-Aldrich), mice were injected i.p. with U73122 (2.5 mg/kg body weight) or vehicle (DMSO) 1 h before thioglycolate.

Overnight culture (37°C) of E. coli (American Type Culture Collection 25922) was grown in Tryptic Soy medium (Merck), and then washed and resuspended in sterile saline (0.9%) at a concentration of 2 × 10⁷ CFU/ml. Anesthetized mice (ketamine:xylazine) were inoculated i.t. with 10⁶ CFU/mouse. Initial inocula were confirmed by plating serial 10-fold dilutions on tryptic soy agar (TSA). After 24 h, CFUs were determined in BAL fluid (1 ml saline), homogenized lungs, and spleens by plating serial 10-fold dilutions on TSA. Neutrophils in BAL fluid were determined by flow cytometry for Gr1 and CD11b.

Using near-infrared reflected light oblique transillumination in vivo microscopy (Mempel et al., 2003), leukocyte recruitment was analyzed in cremaster muscle of PBS-treated p38δ^+/+ control mice, as well as that of p38δ^+/+ and p38δ^−/− mice stimulated intrascrotally with 100 µM fMLP. Chimeric mice cremaster were stimulated by superfusion at 37°C with KC (5 nM; R&D Systems) and neutrophil recruitment was monitored 1 h after preparation of cremaster muscle (Hickey et al., 2000). Blood flow velocity was measured using intraarterial fluorescence-labeled microspheres. To assure intergroup comparability, systemic leukocyte counts, inner vessel diameter, blood flow velocity, and wall shear rate were determined. Leukocyte arrest was determined before and 1 min after i.v. injection of 600 ng CXCL1 (PeproTech), as described previously (Zarbock et al., 2007). Arrest was defined as leukocyte adhesion >30 s and expressed as cells per surface area. Surface area, S, was calculated for each vessel using S = π × d × l_ν, where d is the diameter and l_ν is the length of the vessel. Postadhesion strengthening was determined by tracking the adherent cells over time.

All cells were isolated from 8–12 wk old sex-matched mice. Mouse bone marrow neutrophils were isolated as previously described (Fumagalli et al., 2007). Purity of isolated cells was assessed by cytospin and by flow cytometry. Purity of isolated neutrophils was 90–95%. Cell viability (Trypan blue exclusion) was >95%. Peritoneal neutrophils were elicited by i.p. injection of thioglycolate medium and, 4 h later, peritoneal cells were recovered in complete RPMI medium (with 10% fetal calf serum and 1% penicillin/streptomycin). Peritoneal macrophages were harvested after 4 d of peritonitis in complete RPMI medium. Macrophages were enriched by differential plating. Splenocytes were enriched for CD4⁺ and CD8⁺ cells using a kit (Stem Cell Technologies).

Chemotaxis assays were performed using peritoneal neutrophils, which show good chemotactic response (Li et al., 2000). Cells were seeded on chemotaxis μ-slides (Ibidi) coated with fibronectin (50 µg/ml) applying a gradient of fMLP (10 µM) for primary neutrophils or CXCL-1 (1ng/ml) for HL-60 cells. Time-lapse microscopy was recorded on CellR (Olympus) at 2 frames per minute for 30 min. Cells tracks were done with Manual Tracking (ImageJ) and analyzed with Chemotaxis plug-in (Ibidi) for accumulated distance and average cell velocity. Chemotaxis index is defined as overall distance migrated by cells divided by migration toward the gradient.

HL-60 cells (CCL-240; American Type Culture Collection) and RAW264.7 (TIB-71; American Type Culture Collection) cells were grown in complete RPMI) medium (with 10% fetal calf serum and 1% penicillin/streptomycin). Differentiation of HL-60 cells (CCL-240; American Type Culture Collection) was performed with addition of 1.5% DMSO for 4 d. RAW cells spontaneously polarized after seeding for 2 h as described (Evans and Falke, 2007). Primary murine macrophages were cultured in complete RPMI. For maintenance of puromycin resistance and shRNA, cells were cultured in presence of 3 µg/ml puromycin (Sigma-Aldrich). Cells were kept at 37°C and 5% CO₂. For transient expression and lentivirus production, 293T cells were transfected by calcium precipitation method. Lentiviral vectors encoding shRNAs against p38δ, PKD1, p85α, and nonsilencing shRNA were obtained from the MISSION shRNA library (Sigma-Aldrich). HL-60 cells were infected for 24 h with lentivirus containing pLVX with p85α cDNA C-terminally fused to V5 tag. After infection, medium was replaced for differentiation medium for 3 d.

Chimeric mice and mixed chimeric mice were generated by bone marrow transplantation as described previously (Mueller et al.; Zarbock et al., 2008). In brief, bone marrow cells isolated from p38δ^+/+, p38δ^−/−, PKD1^f/f, PKD1^Δmy, and/or Lys-M-GFP⁺ mice were injected i.v. into lethally irradiated C57BL/6J (wt) mice. Experiments were performed 6–8 wk after bone marrow transplantation. In mixed chimeric mice, adhesion and transmigration of both Lys-M-GFP⁺ and GFP-negative cells was determined by intravital microscopy as described above. Cell numbers were normalized to the percentage of Lys-M-GFP⁺ neutrophils in the blood. Data were expressed as adhesion and transmigration of GFP-negative cells in relation to Lys-M-GFP⁺ cells.

Total bone marrow from femurs of mice was resuspended in Hanks’ balanced salt solution (HBSS, Invitrogen). Red blood cells were lysed using lysis buffer (0.8% NH₄Cl 0.1mM EDTA pH7.4). Overall morphology was assessed on cytospins stained with Wright-Giemsa (Sigma). Bone marrow cells were double-labeled with FITC-conjugated anti-CD11b (Miltenyi Biotec) and PE-conjugated anti-Gr1 (Miltenyi Biotec) and analyzed by flow cytometry. Blood was sampled in EDTA-coated collection tubes (Sarstedt) and analyzed on a hemocytometer. Differential blood cell counts were obtained from May-Grünwald-Giemsa-stained blood smears.

Adhesion flow chamber experiments were performed as described previously (Kuwano et al., 2010). In brief, protein-G precoated glass capillaries were coated with P-selectin (20 µg/ml), IL-8 (50 µg/ml), and IgG1 (15 µg/ml) or m24 (15 µg/ml) for one hour and blocked with casein (Thermo Fisher Scientific). HL60 cells were resuspended in PBS containing MgCl₂ and CaCl₂ (1 mM) with a density of 5 × 10⁶/ml living cells. The flow chamber was perfused with the cell suspension for 2 min and rinsed with PBS (1 mM MgCl₂/CaCl₂) for 1 min. In representative images the number of cells per field of view was determined.

RNA was extracted using TRIzol (Invitrogen) and reverse transcribed using first-strand cDNA synthesis (Fermentas). Primers for p38δ and 18S rRNA were described previously(Sumara et al., 2009). Primer sequences for CXCL-1 (NM_008176) were as follows: fwd 5′-CTGGGATTCACCTCAAGAACATC-3′, rev 5′-CAGGGTCAAGGCAAGCCTC-3′, and for CCL3 (NM_011337) fwd 5′-TTCTCTGTACCATGACACTCTGC-3′, rev 5′-CGTGGAATCTTCCGGCTGTAG-3′.

Cells were fixed with PFA for 10 min at room temperature. F-actin was stained using Phalloidin-Rhodamine (Molecular probes). Confocal microscopy was performed using LSM510-NLO (Carl Zeiss) using Zen2008 software. To quantify PTEN and F-actin polarization in neutrophils, cell front including the lamellipodium and cell back including the uropod were distinguished morphologically. The area anterior of a median between lamellipodium and uropod was defined as belonging to the cell front, whereas the area posterior of this median was defined as belonging to the uropod. ImageJ was used to determine mean fluorescence intensities within these areas and the relative intensities were calculated with respect to the cell front.

Phosphopeptides from in vitro kinase assays were identified using LC-MSⁿ (LTQ Orbitrap XL; Thermo Fisher Scientific), with or without enrichment by TiO₂ (Bodenmiller et al., 2007). MS/MS data were searched against UniProt database (uniprot_sprot_human, May 18th, 2010) using Mascot (Matrix Science; Perkins et al., 1999).

Activity of PTEN in cell lysates and of recombinant or immunoprecipitated PTEN was determined as previously described (Vemula et al., 2010).

In vitro kinase assays were performed as previously described (Hausser et al., 2005).

GST-tagged proteins were expressed in BL21DE3 cells growing in presence of ampicillin and chloramphenicol. At log phase (OD₆₀₀ 0.6), expression was induced by addition of isopropyl-β-D-thiogalactopyranoside (IPTG; 0.1 mM) at 25°C overnight. Cells were resuspended in 15 ml NETN buffer containing protease inhibitors (Complete, Roche) and subjected to lysis by French press. Insoluble fraction was pelleted by centrifugation (12000g, 15 min, 4°C). The supernatant was mixed with 500 µl bed volume glutathione (GSH)-bound beads and orbital mixed for 3 h at 4°C. The beads were collected and washed 4 times with NETN. The GST-tagged protein was eluted with GSH (15 mM in Tris pH9) for 30 min. Polyhistidine (HIS)-tagged proteins were purified using TALON Cobalt resin (Takara Bio Inc.) according to manufacturer’s description. Expression and purity of protein was determined by SDS-PAGE using Coomassie Brilliant Blue staining.

Recombinant GST-PTEN (3 µg) was incubated with recombinant HIS-p85α (6 µg) for 2 h at 4°C in GST binding buffer (20mM HEPES pH7.5, 150mM NaCl, 0.1% NP-40 Substitute, 5mM MgCL₂, 10% Glycerol, 2mM DTT, 1mM PMSF, Complete protease inhibitors (ROCHE). Samples were incubated on buffer-equilibrated GSH-beads, washed, and prepared for SDS-PAGE using SDS sample buffer. For binding assays, in presence of active PKD1, recombinant PKD1 (1 µg) was added to the samples together with ATP (200nM) and samples were first incubated at 25°C for 10 min.

Transient transfection of 293T cells was performed using phosphate precipitation method using 15 µg of plasmid DNA per 10 cm dish. Cells were lysed after 24 h in TNN Buffer (20 mM Tris, pH 7.4, 150 mM NaCl, 5 mM EDTA, 1 mM Vanadate, 10 mM NaF, 0.5 mM Pyrophosphate, Complete protease inhibitors [ROCHE]) for 20 min on ice. Lysates were centrifuged for 10 min at 11,000 g. Supernatants were incubated with antibodies (7 µg) overnight at 4°C. Antibody–protein complexes were precipitated with blocked Protein G beads for 2–4 h at 4°C, washed, and prepared for SDS-PAGE using SDS sample buffer.

In silico screening was performed using PROSITE tool (www.expasy.org) Our input consensus PKD phosphorylation site [L]-X-[R]-T-[QA]-pS with predicted phosphorylation on serine or Threonine corresponds to the sites in class II histone deacetylases, which are established PKD1 targets (Vega et al., 2004; Huynh and McKinsey, 2006; Jaggi et al., 2007). Using this consensus, our search identified a list of primary sequences, including HDACs 4, 5, 7 and phosphatidylinositol 4-kinase β, which are established targets of PKD1.

One-way ANOVA with post-tests was used for multiple groups. Student’s t test was used for two groups. Statistical significance was considered when P < 0.05.

Table S1 shows that bone marrow composition in p38δ knockout mice and WT controls is comparable. Table S2 shows that basal blood parameters and leukocyte counts in whole blood of p38δ knockout mice and WT controls are comparable. Table S3 shows that an in silico screen for PKD phosphorylation consensus sites revealed several hits. Table S4 shows several phosphopeptides identified in an in vitro kinase assay by mass spectrometry. Videos S1-S8 show neutrophil migration and differntiated HL60 cell expression.

We thank Susann Kumpf, Julius Müller and Ivan Formentini for helping with experiments. We thank Darya Zibrova for helping for the generation of the phospho-p85α antibody. We also thank Rhonda Bassel-Duby and Eric N. Olson at University of Texas Southwestern for PKD1 floxed mice.

Studies were supported by Swiss National Science Foundation SNSF (PP0033-114856), Roche Research Foundation, Bonizzi-Theler Foundation, an ETHIIRA grant by ETH (ETH-07 09-1) and by an ERC starting grant (ERC-2011-StG, 281271-STRESSMETABOL). Experiments by the laboratory of A.Z. were supported by German Research Foundation (AZ 428/3-1, AZ 428/6-1, AZ 428/8-1 and to A.Z.) and Interdisciplinary Clinical Research Center (IZKF; Münster, Germany, to A.Z.). Experiments by the laboratory of F.K. were supported by Deutsche Forschungsgemeinschaft (DFG; RE 2885/1-1 to C.A.R) and Sonderforschungsbereich 914 of DFG to C.A.R. and F.K. We thank Huiping Jiang at Boehringer Ingelheim, USA for p38δ floxed mice. The authors declare no competing financial interests.

Author contributions: A.I. conceived experiments and conducted the majority of experiments and helped write the manuscript. H.B., C.A.R., A.Z and F.K. carried out in vivo microscopy experiments. M.V. and M.G. performed proteomic analyses. H.G. and G.S. helped with mouse experiments. A.Z. conceived of important experiments. R.R. conceived of and supervised the project and wrote the manuscript.