|Home | About | Journals | Submit | Contact Us | Français|
Neutrophils are important in innate immunity and acute inflammatory responses. However, the regulation of their recruitment to sites of inflammation has not been well characterized. Here, we investigated the kinase PIP5K1C and showed that PIP5K1C-deficiency impaired neutrophil recruitment due to an adhesion defect. PIP5K1C regulated the adhesion through facilitating RhoA GTPase and integrin activation by chemoattractants. Integrins could induce an isoform of PIP5K1C, PIP5K1C-90, polarization in neutrophils through intracellular vesicle transport independently of exogenous chemoattractant. PIP5K1C-90 polarization was required for polarized RhoA activation at uropods and provided an initial directional cue for neutrophil polarization on the endothelium. Importantly, the polarization was also required for circumventing the inhibition of lamellipodium formation by RhoA so that neutrophils could form leading edges required for transendothelial migration. Because integrins are not known to regulate neutrophil polarization, our study revealed a previously underappreciated role of integrin signaling in neutrophil regulation.
Neutrophils are one of the key players in acute inflammatory responses. They play an important role in host defense and contribute to inflammation-related tissue injuries. During the inflammation, neutrophils extravasate across the endothelium that lines the blood vessel wall through a multi-step process (Luo et al., 2007; Rose et al., 2007), including the rolling on and subsequent firm adhesion to endothelial cells. In mouse neutrophils, the β2 integrins are important in mediating the adhesion of neutrophils to endothelial cells. After neutrophils transmigrate through the endothelium, they migrate to the sites of injury and infection in response to chemoattractant gradients. Thus, understanding the mechanisms for neutrophil recruitment is of great physiological and pathophysiological significance.
Neutrophils are the most motile cells in higher organisms and can efficiently interpret and chemotax under a shallow chemoattractant gradient. Although it remains unclear how, these cells can translate a small chemoattractant gradient into intracellular biochemical polarization and align the biochemical polarity with the chemoattractant gradients (Franca-Koh et al., 2007; Janetopoulos and Firtel, 2008; Rericha and Parent, 2008; Wu, 2005). In neutrophils, the biochemical polarization includes the localization of molecules such as phosphatidylinositol 3,4,5-trisphosphate and the small GTPases Rac and Cdc42 in the front and the GTPase RhoA, phosphorylated myosin light chain (pMLC) (Weiner, 2002; Wu, 2005; Xu et al., 2003), the RhoA guanine nucleotide exchange factor PDZRhoGEF (Wong et al., 2007), phosphorylated ezrin-radixin-moesin (pERM) (Lacalle et al., 2007; Lokuta et al., 2007), and phosphatase and tensin homolog (PTEN) (Heit et al., 2008; Li et al., 2005; Li et al., 2003; Wu et al., 2004) in the back. The biochemical polarization subsequently leads to cellular polarization into a leading edge (front) and uropod (back). The leading edge contains lamellipodia composed of unbundled F-actin, continuous formation of which provides a driving force for cell locomotion. The uropod contains actomyosin filaments. Contraction of these filaments can provide another locomotive force to push cells forward. Thus far, chemoattractant signaling is believed to be exclusively responsible for regulating the directionality of neutrophil migration.
Phosphatidylinositol 4,5-bisphosphate [PtdIns (4,5)P2] represents about 1% of plasma membrane phospholipids and is important in various cellular functions (De Matteis and Godi, 2004; Di Paolo and De Camilli, 2006; Ling et al., 2006). In mammalian cells, PtdIns (4,5)P2 is primarily synthesized by sequential phosphorylation of PtdIns by two PtdIns kinases, PI4K and PIP5K1. To date, three mammalian PIP5K1 isoforms (A, B, & C) have been identified. PIP5K1C has two major splicing variants: a short 87-kDa protein (PIP5K1C-87) and a longer one with 28 additional amino acids at its C-terminus (PIP5K1C-90). PIP5K1C-90 was shown to be localized in the uropods of chemotaxing neutrophils (Lokuta et al., 2007), whereas PIP5K1B was found in the uropods of polarized human neutrophil-like HL-60 cells (Lacalle et al., 2007). There are two reported Pip5k1c−/− mouse lines, both of which show early lethality (Di Paolo et al., 2004; Wang et al., 2007a).
In this report, we studied PIP5K1C-deficient mouse neutrophils and found that PIP5K1C-deficiency did not impair neutrophil chemotactic activities in vitro, but it compromised neutrophil infiltration in vivo. In our investigation, we discovered that integrins could induce PIP5K1C-90 polarization independently of chemoattractants. This integrin-induced PIP5K1C-90 polarization works together with chemoattractant signaling in regulating neutrophil polarization and directionality in vitro and infiltration in vivo.
To investigate the role of PIP5K1C in neutrophil regulation, we generated neutrophils lacking PIP5K1C by reconstituting lethally irradiated mice with fetal liver cells from a Pip5k1c−/− line (Di Paolo et al., 2004). Neither Pip5k1c mRNA nor protein could be detected in neutrophils prepared from the transplanted mice (Figure S1A–B).
The regulation of PtdIns(4,5)P2 by PIP5K1C in mouse neutrophils was examined by visualizing the localization of the phospholipase C δ-pleckstrin homology (PLC δ-PH) domain, which binds PtdIns(4,5)P2 in a specific manner (Lemmon et al., 1995). Most of wild-type neutrophils undergoing chemotaxis exhibited uropod polarized PLC δ-PH GFP distribution (Fig. S1C upper panels and Movie S1A). However, in Pip5k1c−/− neutrophils, there was a lack of increased localization of PLC δ-PH-GFP in the uropods (Fig. S1C lower panels and Movie S1B), suggesting that PIP5K1C may be responsible for the increased amount of PtdIns(4,5)P2 at the uropods.
Next, we investigated the effect of PIP5K1C-deficiency on neutrophil recruitment in vivo using a peritonitis model. Purified wild-type neutrophils and Pip5k1c−/− neutrophils were labeled with different cell tracing dyes or vice versa (Jia et al., 2007) and mixed at a 1:1 ratio. The mixed cells were injected into tail veins of wild-type mice, in which acute peritoneal inflammation was induced by intraperitoneal injection of thioglycolate. Significantly lower numbers of Pip5k1c −/− than wild type transplanted neutrophils were recruited into the peritonea (Figure 1 A–B). We next extended our observations to a gout model by examining the effects of PIP5K1C-deficiency in neutrophil recruitment into preformed air pouches after injection of monosodium urate (MSU) crystals (Chen et al., 2006). There were significantly lower numbers of neutrophils (CD11b+Ly6G+) in the lavages from Pip5k1c−/− cell transplanted mice than those from the wild-type cell transplanted mice (Fig. 1C). Thus, PIP5K1C-deficiency impairs the in vivo neutrophil infiltration in both of the models.
Previous studies using overexpression of PIP5K mutants suggest that PIP5K including PIP5K1C plays a positive role in regulating neutrophil chemotaxis (Lacalle et al., 2007; Lokuta et al., 2007). We found that Pip5k1c−/− cells showed no defects in their directionality and motility compared to wild-type cells under an fMLP gradient (Fig. 2A–B), though the mutant cells appeared to be more elongated than the wild-type cells (data not shown). On the contrary, P Pip5k1c−/− neutrophils appeared to follow the chemoattractant gradient more faithfully than the wildtype cells because they had significantly smaller average directional errors than the wildtype cells on fibrinogen-coated surface (Fig.2C). However, this directional error difference became less significant on the polylysine surface (Fig. 2C). These results in Fig. 2C can also be interpreted to suggest that there is a substantial difference in gradient errors or directionality between fibrinogen-coated and poly-lysine coated surfaces for wildtype neutrophils, whereas the difference became insignificant for Pip5k1c−/− cells. Because there may be more integrin activation in cells on fibrinogen than on polylysine (Kim et al., 2006), these results imply that there might be a connection between integrins and PIP5K1C in regulating neutrophil directionality.
PIP5K1C-90, which is expressed five times more than PIP5K1C-87 in mouse neutrophils based on quantitative RT-PCR analysis (data not shown), was shown to polarize in neutrophils upon fMLP stimulation (Lokuta et al., 2007). In reproducing these findings, we observe that GFP-PIP5K1C-90 could polarize in mouse neutrophils in the absence of any exogenous chemoattractant (Fig. 3A). Up to 80% of neutrophils expressing GFP-PIP5K1C-90 showed the polarized localization of PIP5K1C on the fibrinogen surface compared to less than 20% on the polylysine surface (Fig. 3B). Fig. S2A and Movie S2 show three-dimensional reconstruction of a neutrophil in which GFP-PIP5K1C-90 is polarized. Moreover, endogenous PIP5K1C-90 could polarize on fibrinogen (Fig. S2B). It is important to note that, in the absence of a chemoattractant, fibrinogen did not overtly polarize the distribution of F-actin, which was primarily localized in the cortex (Fig. 3A and Fig. S2A, B). The differential effects of fibrinogen and polylysine prompted us to hypothesize that integrins may regulate PIP5K1C-90 polarization. We thus tested ICAM-1, a ligand for the β2 integrins abundantly found on neutrophils. ICAM-1 could also induce GFP-PIP5K1C-90 polarization (Fig. S2C). In addition, ICAM-induced polarization was reduced in neutrophils isolated from a mouse line in which β2-intergrin is expressed at ~10% of the normal amount (Wilson et al., 1993). Moreover, a neutralizing β2 integrin antibody or expression of an integrin dominant negative mutant, in which the intracellular domain of the IL-2 receptor was replaced with the β2 integrin intracellular domain (Smilenov et al., 1994), inhibited PIP5K1C-90 polarization (Fig. 3B & S2C). Thus, integrins, particularly β2-integrin, are involved in polarized localization of PIP5K1C-90 in mouse neutrophils. This conclusion is further confirmed by the fact that neutralizing αL or αM integrin antibody inhibited ICAM1-induced PIP5K1C-90 polarization (Fig. S2D). Thus, both αLβ2 (LFA-1) and αMβ2 (MAC-1) integrins participate in PIP5K1C-90 polarization.
To examine the kinetics of GFP-PIP5K1C-90 polarization, we acquired time-lapsed images of GFP-PIP5K1C-90-expressing neutrophils that flowed over fibrinogen-coated surfaces. Polarization appeared to start only when cells stopped or were close to stopping moving presumably due to integrin engagement (Fig. 3C and Movies S3). The polarization process was completed in 1–2 minutes. Of note, during this polarization process, the shape of these cells remained largely round rather than overtly elongated found in chemoattractant-stimulated cells. Intriguingly, GFP-PIP5K1C-90 always polarized in the alignment of the cell movement direction in all of the cells we examined. We also tested neutrophils expressing GFP-PIP5K1C-90 using a flow chamber coated with mouse endothelial cells. In all of the cells we observed (n=35), GFP-PIP5K1C-90 was polarized in general alignment of the flow direction (Fig. 3D). These results together suggest that the cell movement direction on the substrate prior to their arrest may provide a directional cue for GFP-PIP5K1C-90 polarization.
We tested the short splicing variant of PIP5K1C, PIP5K1C-87 (Fig. S3A) and showed it could not polarize (Fig. 3B and S3B), suggesting that the 28 extra amino acids found only in PIP5K1C-90 are required for integrin-induced polarization. These 28 amino acids are required for the binding to talin and adaptin-2 (AP2) proteins, and various point mutations in this sequence can disrupt these interactions (Di Paolo et al., 2002; Lee et al., 2005; Ling et al., 2002; Thieman et al., 2009). We confirmed L652S, S650D, Y649F, and W647F mutations disrupted the interaction of PIP5K1C-90 with the AP2β subunit (Fig. S3C), whereas these mutations except Y649F also disrupted the interaction with talin (Fig. S3D). Because all of these mutations including Y649F impaired the polarization (Fig. 4A), we suspected that the interaction of PIP5K1C-90 with AP2 might have an important role in PIP5K1C-90 polarization upon integrin engagement.
AP2 is a key adaptor protein in the formation of intracellular transport vesicles, including those coated with clathrin. The localization of AP2 and clathrin relative to that of PIP5K1C-90 in neutrophils was examined. AP2 also showed polarized distribution in neutrophils placed on fibrinogen, but not poly-lysine, and there was colocalization of AP2 and PIP5K1C-90 (Fig. 4B). Similar observations were also made with endogenous clathrin (Fig. S3E) and clathrin-GFP (Fig. S3F). There is also a partial colocalization of β2-integrin and PIP5K1C90 (Fig. S3G). These results are consistent with the idea of an involvement of vesicle transport in PIP5K1C-90 polarization. Vesicle transport is regulated by various small GTPases including Arfs (ADP-ribosylation factor) and Rabs (D'Souza-Schorey and Chavrier, 2006; Myers and Casanova, 2008). We tested several dominant negative (dn) mutants of small GTPases and found that dnArf6, but not dnArf4 (Fig. 4C) or dnRab5 (data not shown), inhibited integrin-dependent PIP5K1C-90 polarization.
Actin filaments also play an important role in vesicle transport (Myers and Casanova, 2008). Treatment with Cytochalasin D, an actin polymerization inhibitor, resulted in reduction in GFP-PIP5K1C-90 polarization on fibrinogen (Fig. 4D). Along the same line, expression of dnRac1, a small GTPase known to stimulate actin polymerization in neutrophils (Gu et al., 2003), also inhibited PIP5K1C-90 polarization on fibrinogen (Fig. 4E). We also tested pertussis toxin, which inhibits the Gi class of heterotrimeric G proteins. It failed to inhibit PIP5K1C-90 polarization (Fig. 4D), further confirming the non-involvement of these G proteins in integrin-dependent PIP5K1C-90 polarization. As a control, the same pertussis toxin treatment abolished fMLP-induced increases in cytosolic Ca2+ concentrations in the neutrophils (data not shown). Thus, vesicle transport may be involved in PIP5K1C-90-polarization.
The importance of Tyr649, which can be phosphorylated by Src (Ling et al., 2003), in PIP5K1C-90 polarization suggests a possible involvement of Src in the polarization process. We hence tested a Src inhibitor PP2, which inhibited PIP5K1C-90 polarization on fibrinogen (Fig. 4F). Focal adhesion kinase (FAK) was shown to participate in integrin-mediated Src activation and PIP5K1C-90 phosphorylation at Tyr649 (Ling et al., 2003). Expression of two FAK inhibitors, FRNK and Y397F-FAK, inhibited PIP5K1C-90 polarization (Fig. 4F). Furthermore, RhoB was shown to be involved in Src activation by integrins (Sandilands et al., 2004). Expression of dnRhoB, but not dnRhoA, inhibited PIP5K1C-90 polarization (Fig. 4E). These results together support the conclusion that integrin-mediated Src activation has an important role in PIP5K1C-90 polarization.
Phosphorylation of PIP5K1C-90 at Tyr649 in neutrophils was further confirmed by using an antibody specific for phosphorylated Tyr649. The antibody only recognized PIP5K1C-90 coexpressed with an activated Src mutant, but not Y649F mutant or PIP5K1C-87 (Fig. S3H). Immunostaining shows that Tyr649-phosphorylated PIP5K1C was detected at the same location as polarized PIP5K1C-90 in neutrophils (Fig. 4G,H). Because the phospho-mimetic mutation of Tyr649 to Glu showed reduced interaction with AP2-β (Fig. S3C), it is possible that Tyr649-phosphoryated PIP5K1C-90 also has a reduced affinity for AP2-β. Thus, Tyr649 phosphorylation may lead to the dissociation of PIP5K1C-90 from AP2 so that AP2 can be recycled while PIP5K1C stays. This idea is consistent with the partial colocalization of AP2 or clathrin with PIP5K1C-90 (Fig. 4B, & Fig S3E–F)
As integrins induce PIP5K1C-90 polarization independently of chemoattractants, it would be important to know the relationship of this integrin-induced polarity with the one induced by chemoattractants. As shown in a previous study (Lokuta et al., 2007), GFP-PIP5K1C-90 is colocalized with uropod makers including pERM (Fig S4A) and pMLC (Fig. S4B) upon fMLP stimulation. Because fibrinogen alone could not induce polarized distribution of pMLC (Fig. S4C) or GFP-moesin (Fig. S4D) [the localization of GFP-moesin correlates with phosphorylated ERM localization in neutrophils (Lacalle et al., 2007; Lokuta et al., 2007)], integrin-regulated PIP5K1C-90 polarization does not depend on either pERM or pMLC polarization. These observations are consistent with the fact that neither dnRhoA (Fig. 4E) nor the Rho kinase inhibitor Y-27632 (Fig. 4D) inhibited PIP5K1C-90 polarization. RhoA, a small GTPase, is localized at the uropod of chemoattractant-polarized neutrophils and regulates actomyosin structure formation and functions including MLC phosphorylation via Rho kinase (Li et al., 2005; Li et al., 2003; Wheeler and Ridley, 2004; Xu et al., 2003). Thus, although integrin-induced PIP5K1C-90 polarization is colocalized with the uropod markers induced by chemoattractants, it does not depend on these chemoattractant-regulated uropod markers.
Knowing that the integrin-induced PIP5K1C-90 polarization is independent of chemoattractant signaling, we investigated the impact of PIP5K1C-90 polarization on chemoattractant-regulated neutrophil polarization and directional response. Mouse neutrophils expressing RFP-PIP5K1C-90 were placed on fibrinogen, followed by directional stimulation with fMLP. To assess real-time neutrophil responses, the cells were also cotransfected with YFP-actin (Fig. 5A). The stimulation was applied either proximally or distally to polarized RFP-PIP5K1C-90 as shown in Fig. 5A. Upon stimulation, the cells responded in about 30 seconds evidenced by polarized distribution of YFP-actin toward the micropipette (Movie S4A). This polarized distribution of YFP-actin is presumably the result of rapid formation of F-actin, which primarily occurs at the lamellipodia upon chemoattractant stimulation.
Regardless of the positions of the stimulation, which did not markedly affect the rate of YFP-actin polarization, YFP-actin always started to polarize at the sites opposite to polarized RFP-PIP5K1C-90 (Fig. 5A, S4E, Movie S4). These observations indicate that the initial polarization of YFP-actin or formation of leading edges is determined by the location of PIP5K1C-90 polarization rather than the fMLP gradient. It is also important to note that cells with their PIP5K1C-90-rich structures near the pipette often started with more than one pseudopod pointing to different directions, which were always against the gradient (lower panels of Fig. 5A & S4E, arrow heads). Although these pseudopods eventually formed consolidated singular lamellipodia, it took significantly longer for them to do so than for those with their PIP5K1C-90-rich structures distal to the pipette (Fig. 5B). In a majority of cells that were stimulated by a pipette proximal to the PIP5K1C-90-rich structures, their lamellipodia were able, at the end, to turn toward the stimulation. Thus, we conclude that integrin-induced polarization of PIP5K1C-90 has an important role in regulating neutrophil response to directional chemoattractant stimulation. It determines where the leading edge, whose formation is stimulated by chemoattractants, can be initially formed regardless of the directional cue of a chemoattractant gradient probably by specifying the uropod.
Knowing that PIP5K1C-deficiency does not impede neutrophil chemotaxis, we sought other possible causes for the in vivo infiltration defects of Pip5k1c−/− neutrophils observed in Fig. 1. We examined the interaction between neutrophils and endothelial cells, the first step in neutrophil infiltration in vivo, using a flow chamber assay. Pip5k1c-deficiency significantly reduced the number of neutrophils that can firmly adhere to endothelial cells (Fig. S5A). This defect may provide an explanation to the in vivo infiltration defects.
Intravital microscopic examination of the cremaster muscle venules was carried out to confirm the in vivo significance of the adhesion defect. Although PIP5K1C-deficiency increased rolling flux, it reduced the number of cells adherent to the endothelium upon the treatment of TNFα particularly in the smaller vessels (Figure 6A–B). PIP5K1C-deficiency had little effects on rolling velocity or rate of emigration (Fig. 6C & D). These observations are consistent with those observed in the flow chambers and demonstrate that PIP5K1C is important for neutrophil firm adherence to endothelial cells in vivo.
Because there were no differences in the cell surface expression of CD18 (β2-integrins) between wildtype and Pip5k1c−/− cells (Fig. S5B), the differences in their adherence to endothelial cells have to be attributed to other factors. RhoA, which is activated by endothelial cell-tethered chemokines, plays an important role in regulating monocyte and T cell adherence to endothelial cells under flow conditions (Giagulli et al., 2004; Honing et al., 2004). We found that PIP5K1C -deficiency significantly reduced fMLP-induced activation of RhoA (Fig. 7A). Consistent with this result, PIP5K1C -deficiency decreased RhoA-dependent phosphorylation of MLC (Fig. 7B, C). Moreover, PIP5K1C -deficiency decreased RBD-staining at the uropods of neutrophils stimulated with fMLP (Fig. S6A). RBD is a protein domain of rhotekin and preferentially binds to activated RhoA (Ren et al., 1999). All of these results indicate that PIP5K1C has an important role in RhoA activation by fMLP in mouse neutrophils.
RhoA is involved in chemokine-stimulated enhancement of integrin affinity in T lymphocytes (Giagulli et al., 2004). Consistent with the finding, PIP5K1C -deficiency attenuated the binding of sICAM-Fc complexed with an anti-Fc antibody (Fig. 7D, S6B) or sICAM directly conjugated with a fluorochrome (Fig. S6C) to fMLP-stimulated neutrophils, without effects on the expression of cell surface β2-integrins (Fig. S5B; data not shown). In addition, dnRhoA expression (Fig. 7E) or Y-27632 treatment (Fig. S6D) could reduce ICAM binding in response to fMLP. Moreover, PIP5K1C -deficiency or inhibition of Rock reduced the ability of neutrophils to retain their adhesion to the endothelial cells under high shear flow (Fig. S6E). Together with the lack of obvious effect of PIP5K1C -deficiency on integrin clustering or lateral mobility (Fig. S6F), we conclude that PIP5K1C may regulate neutrophil firm adhesion primarily through facilitating RhoA activation and integrin affinity increase.
The next key question is whether PIP5K1C polarization has biological significance. We addressed this question by comparing the two PIP5K1C isoforms; PIP5K1C-90 that can be polarized by integrins and PIP5K1C-87 that cannot. Expression of RFP-PIP5K1C-90 in Pip5k1c −/− neutrophils restored polarized localization of PLC δ-PH-GFP or PtdIns(4,5)P2 production at uropods (Fig. S6G), whereas PLC δ-PH-GFP showed even distribution in cells expressing RFP-PIP5K1C-87 (Fig. S6G). Upon fMLP stimulation, expression of either GFP-PIP5K1C-90 or −87 increased pMLC staining, a surrogate marker for RhoA activity (Fig. 7F), which was only weakly detected in Pip5k1c−/−neutrophils (Fig. 7C). Expression of a kinase dead form of GFP-PIP5KIC-87 led to little increases in pMLC staining in Pip5k1c−/− neutrophils (Fig. S6H), suggesting that the lipid kinase activity or PtdIns(4,5)P2 may be responsible for the increase in pMLC and RhoA activity. However, there were two important distinctions between cells expressing GFP-PIP5K1C-90 and those expressing GFP-PIP5K1C-87. Firstly, pMLC staining in GFP-PIP5K1C-90-expressing cells was highly polarized and concentrated at the vicinity of polarized GFP-PIP5K1C-90 (Fig. 7F), whereas in cells expressing GFP-PIP5K1C-87 pMLC staining was broadly distributed similarly to the distribution of GFP-PIP5K1C-87 (Fig. 7F). Secondly and more importantly, neutrophils expressing GFP-PIP5K1C-87 failed to form F-actin-rich lamellipodia, whereas those expressing GFP-PIP5K1C-90 did, in response to fMLP (Fig. 7F). F-actin detected in GFP-PIP5K1C-87-expressing neutrophils was colocalized with GFP-PIP5K1C-87 and pMLC and appeared to resemble cortical actin. Expression of the kinase-dead form of GFP-PIP5K1C-87 did not affect the formation of F-actin-rich lamellipodia (Fig. S6H), indicating that the effect of PIP5K1C-87 expression on lamellipodium formation depends on PtdIns(4,5)P2. These results collectively suggest that although expression of PIP5K1C was able to restore fMLP-induced phosphorylation of MLC and probably RhoA activation, polarized localization of PIP5K1C-90, which is expressed much more than PIP5K1C-87 in mouse neutrophils, is required for polarized activation of RhoA. Failure to do so as in the case of expression of the non-polarizable PIP5K1C-87 isoform would result in broad activation of RhoA and the inability to form lamellipodia. Consistent with the idea that lamellipodia are required for neutrophils to undergo transendothelial migration, neutrophils expressing PIP5K1C-87 showed markedly attenuated ability to migrate across a layer of endothelial cells in response to fMLP (Fig. 7G). Therefore, polarized localization of PIP5K1C, manifested by integrin-induced PIP5K1C-90 polarization (Fig. S6I), has important roles in regulating neutrophil polarization and infiltration.
Here we showed integrins can confer a polarity to neutrophils by inducing polarized localization of PIP5K1C-90 independently of chemoattractants. This integrin-induced polarity has important roles in neutrophil polarization and infiltration. Although integrins have been shown to regulate cell motility, they are not known to regulate neutrophil polarization or directionality. Thus, our findings have revealed previously underappreciated roles of integrin signaling in regulating neutrophil functions.
The evidence presented in this report implicated an important role of vesicle transport in integrin-induced PIP5K1C-90 polarization. The evidence includes an impediment of PIP5K1C-90 polarization by AP2 interaction mutations on PIP5k1C-90, polarized distributions of AP2 and clathrin by integrin engagement and their colocalization with PIP5K1C-90, and inhibition of the polarization by dnArf6. Because the vesicle transport may be a continuous process, clathrin and Ap2 may have to be recycled. Thus, there should be a mechanism for the dissociation of PIP5K1C-90 from Ap2. Tyr649 Phosphorylation of PIP5K1C-90, which was probably carried out by Src and Fak and detected in polarized PIP5K1C-90 structure, may provide such a mechanism because the phospho-mimetic mutation of Tyr649 reduces the interaction with AP2. Based on these data, we propose a model to suggest that integrin engagement may stimulate Arf6-dependent vesicle transport, which brings Ap2-associated PIP5K1C-90 to one side of a cell. The directionality of the transport may be determined by the cell movement direction prior to its arrest. Fak and Src-mediated phosphorylation may subsequently result in the dissociation of PIP5K1C-90 from Ap2. Ap2 might be recycled, whereas PIP5K1C-90 stayed. It remains unknown how cell movement direction determines the directionality of vesicle transport and the location of PIP5K1C-90 polarization and whether additional modifications of PIP5K1C-90 or its interaction with other molecules are required for its final, more consolidated localization. It is also not known how integrin engagement triggers the polarization process. Engaged Integrins may either accelerate basal endocytic or endosomal vesicle trafficking or are the cargos that initiate the trafficking.
Current concepts regard chemoattractants as the sole regulators of polarization and directionality in chemotaxis. It is also believed that chemoattractants polarize the cell by specifying the “front” and “back” through polarized localization and/or activation of signaling and structural molecules at both leading edges and uropods. The results of this study extend the concepts to suggest that signaling other than elicited by chemoattractants can also break the symmetry and polarize the neutrophils. In this case, integrin-induced polarization specifies the “back” without mobilizing the “front” signaling molecules. Importantly, this integrin-regulated “back” signaling is dominant enough to determine the initial polarity, along which chemoattractants have to polarize their front and back signals. Our results also confirm that chemoattractant-controlled fronts, once formed, become more dominant, which can lead a direction change if the cell polarity is not aligned with the chemoattractant gradient. Therefore, chemotactic directionality determination may the result of the summation of signaling inputs of multiple pathways in a context-dependent manner. On a different note, this effect of integrin-PIP5K1C-90-regulated polarization on initial neutrophil directionality provides an explanation to the poor initial directionality on fibrinogen as well as the heterogeneity in neutrophil chemotactic directionality often observed in many of the in vitro assays. In these assays, the neutrophils may have taken an initial polarity due to integrin signaling, which is random to the chemoattractant gradient applied afterward. The more integrin activation as in the case of fibrinogen coating may cause more cells to take up the initial polarity and thus higher directional errors.
The integrin-PIP5K1C-90-regulated polarization intersects chemoattractant-regulated polarity at RhoA regulation. Chemoattractants are long known to activate RhoA at the uropods, but the mechanisms for such polarized activation remain unclear. PDZ-RhoGEF, which contains a PH domain, was found to be localized at uropods and appeared to regulate uropod RhoA activation in neutrophil-like HL-60 cells (Wong et al., 2007; Xu et al., 2003). We are currently investigating whether PIP5K1C-90 and PtdIns(4,5)P2 can regulate PDZ-RhoGEF localization or activity.
Our data also showed an incomplete abrogation of RhoA activation by PIP5K1C deficiency. This may be attributed to the existence of other RhoA activation mechanisms and/or PIP5K1 isoforms. Human PIP5K1B is also polarized at the uropod upon chemoattractant stimulation in neutrophil-like HL-60 cells (Lacalle et al., 2007). However, our analysis of neutrophils isolated from mice lacking human PIP5K1B ortholog Pip5k1a (Sasaki et al., 2005) revealed no chemotactic, RhoA activation, or adhesion defects compared to the wildtype neutrophils (data not shown). It also remains unclear whether the kinase activity of Pip5k1c is regulated during the polarization process. Both Arf6 and RhoA have been shown to activate its activity (Bolomini-Vittori et al., 2009; Honda et al., 1999; Krauss et al., 2003). It is possible that RhoA and PIP5K1C constitute a positive feed-forward mechanism for the production of PtdIns(4,5)P2 at uropods.
The results in this study suggest that the biological significance of PIP5K1C and its polarization upon integrin engagement may not lie in its regulation of neutrophil chemotaxis rather its infiltration through the endothelium. On one hand, PIP5K1C facilitates RhoA activation, which leads to the increase in integrin affinity required for neutrophil firm adhesion to the endothelium. In contrast, PIP5K1C-90 polarization is required for polarized RhoA activation. RhoA is known to antagonize Rac, which is required for the formation of F-actin and hence lamellipodium. Without lamellipodia, cells cannot undergo migration. Thus, two competing activities (RhoA-mediated firm adhesion and Rac-mediated formation of lamellipodia for cell migration) have to occur in the same cell for successful neutrophil infiltration. Polarized activation of RhoA through integrin-induced Pip5k1c-90 polarization provides a solution for these two competing biological activities to occur concomitantly in the same cell. Thus, PIP5K1C has two critical roles in regulating neutrophil infiltration in vivo; while its role in facilitating RhoA activation by chemoattractants regulates endothelial cell adhesion, its role in polarizing RhoA activation helps evade the suppressive effect of RhoA activation on the formation of lamellipodia that is required for neutrophil extravasation and migration.
Wild-type GFP-PIP5K1C-90 and the antibody specific for Pip5k1c-90 (Di Paolo et al., 2002) were kindly provided by Dr. Pietro De Camilli. Plasmids encoding FRNK and Y397-FAK were gifts from Dr. Jun-lin Guan. The antibody specific for Tyr-649 of PIP5K1C-90 was generated by AbMax (Beijing, China). The PIP5K1C mutants were generated by PCR-based mutagenesis and verified by nucleotide sequencing. GFP-PH-PLCδ was a gift from Alan Smrcka. Anti-Talin antibody was purchased from Sigma. Phospho-MLC antibody was obtained from Cell Signaling. Anti-mouse β2 integrin blocking antibody (GAME-46) was purchased from BD Biosciences Pharmingen. Blocking antibodies to αM (Clone M1/70) and αL (Clone M17/4) integrins were purchased from eBiosciences. YFP-β-actin was described previously (Kress et al., 2009).
Pip5k1c-deficient mice have been previously described (Di Paolo et al., 2004). Liver cells (2 million) from neonatal wild type or Pip5k1c-null mice were transplanted into wildtype recipient mice that had been subjected to 1000cGy X-Ray irradiation. Eight weeks later, the transplanted mice were used for neutrophil preparation.
Bone marrow neutrophils were purified from mouse bone marrow, transfected, stained, and assayed for its migration in Dunn chambers and transwell plates as described previously (Zhang et al., 2010). The detailed protocols are also described in the Supplemental Experimental Procedures.
Coverslips were coated with 1mM polylysine or 100µg/ml Fibrinogen at 37°C for 1 hour. Neutrophils were plated onto the coverslips for 10 mins. In some experiment, neutrophils were pretreated with PTX (1µg/ml for 2 hrs at 37°C), PP2 (10µM for 15 mins), Y-27632 (10µM for 15 mins) or Cytochalasin D (10µM for 15 mins). The cells were then fixed with 4% paraformaldehyde and examined by using Leica SP5 confocal microscope.
Mouse endothelial cells (Wang et al., 2007b) were cultured to confluency on 10µg/ml fibronectin coated coverslips and treated with 50ng/ml TNFα for 4 hours. The coverslips containing the endothelial cell layer were washed with PBS and placed in a flow chamber apparatus (GlycoTech). Purified wildtype and Pip5k1c-null neutrophils were labeled with CFSE and FarRed SSAO SE Dye, respectively, at 37°C for 15 mins and then mixed at a 1:1 ratio. The mixed neutrophils were placed on top of the endothelial cells and subjected to sheer flow of 1 dyne/cm2 for 1min. The cells were then fixed, and the number of neutrophils adhering to the endothelial cells was counted using a fluorescence microscope. For examining directionality, neutrophils transiently transfected with GFP-PIP5K1C-90 were flowed through the chamber coated with a monolayer of mouse endothelial cells as described above.
To test the adherence under high sheer stress, wildtype and Pip5k1c-null or Y-27632 treated neutrophils labeled with CFSE and FarRed SSAO SE Dye, respectively, or vice versa were allowed to sediment to the monolayer endothelial cells for 10 min in the chamber. The sheer stress was gradually ramped up to 4 dyn/cm2 in 10 min. Images sequences were taken at 15-sec intervals. The numbers of wildtype and Pip5k1c-null cells or Y-27632 treated WT cells attached to the endothelium at first min were confirmed to be no more than 2.5% different. The numbers of untreated wildtype cells adhered to endothelial at the end of recording are taken as 1.
The levels of active GTP-bound RhoA were determined using a G-LISA RhoA Activation Assay kit (Cytoskeleton, Inc). One million of wildtype or Pip5k1c -null neutrophils were stimulated with mock or 4µM fMLP for 3 min before the assay. The ICAM binding assay was carried out as previously described (Konstandin et al., 2006). Detailed protocols for ICAM binding and integrin clustering are described in the Supplemental Experimental Procedures.
Neutrophil recruitment to the peritonitis and MSU-induced gout was carried out as described in (Di Lorenzo et al., 2009; Jia et al., 2007). Neutrophil infiltration in the mouse cremaster muscle venules was performed as previously described (Liu et al., 2005). The detailed protocols are also described in the Supplemental Experimental Procedures.
Total RNA was isolated from purified wild type or Pip5k1c -null neutrophils using TRIzol. The cDNA was synthesized using iScript cDNA synthesis kit (Bio-Rad), and qPCR was carried out using Pip51a, 1b and 1c specific oligos.
We thank Michelle Orsulak and other members of the Wu lab for various assistances, Pietro De Camilli for the pip5k1c mutant mice, plasmids, and reagents, and David Calderwood for plasmids. This work was supported in part by NIH grants (HL080706, HL070694, 1U54 RR022232-01, NS36251, DA018343), by German Academy of Sciences Leopoldina (BMBF-LPD 9901/8-162)(H.K.), and by AHA and NSFC30800587, 30971521, 2010CB529704, 09QA1401900, 06DZ22923 (P.W.).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.