ABSTRACT
The broad tissue distribution and evolutionary conservation of the glycosylphosphatidylinositol (GPI)-anchored prion protein (PrP, also known as PRNP) suggests that it plays a role in cellular homeostasis. Given that integrin adhesion determines cell behavior, the proposed role of PrP in cell adhesion might underlie the various in vitro and in vivo effects associated with PrP loss-of-function, including the immune phenotypes described in PrP−/− mice. Here, we investigated the role of PrP in the adhesion and (transendothelial) migration of human (pro)monocytes. We found that PrP regulates β1-integrin-mediated adhesion of monocytes. Additionally, PrP controls the cell morphology and migratory behavior of monocytes: PrP-silenced cells show deficient uropod formation on immobilized VCAM and display bleb-like protrusions on the endothelium. Our data further show that PrP regulates ligand-induced integrin activation. Finally, we found that PrP controls the activation of several proteins involved in cell adhesion and migration, including RhoA and its effector cofilin, as well as proteins of the ERM family. We propose that PrP modulates β1 integrin adhesion and migration of monocytes through RhoA-induced actin remodeling mediated by cofilin, and through the regulation of ERM-mediated membrane–cytoskeleton linkage.
INTRODUCTION
The cellular prion protein (PrP, also known as PRNP) is a glycosylphosphatidylinositol (GPI)-anchored glycoprotein that resides in detergent-resistant membrane domains. The conversion of PrP to a misfolded insoluble isoform (PrP Scrapie) occurs at the onset of neurodegenerative disorders known as transmissible spongiform encephalopathies (Colby and Prusiner, 2011). However, its physiological function is largely unclear (Biasini et al., 2012), although it has been suggested that it is involved in multiple cellular processes such as survival, apoptosis, differentiation and growth (Linden et al., 2008). This variety of functions together with the high evolutionary conservation and wide tissue expression of PrP suggest that this protein has a role in cellular homeostasis.
PrP is highly expressed in the hematopoietic system, particularly in hematopoietic stem cells and in mature mononuclear leukocytes (Isaacs et al., 2006). Accordingly, PrP has been reported to have a role in several hematopoietic and immune cell functions in vivo, such as the self-renewal of long-term repopulating hematopoietic stem cells (Kent et al., 2009; Zhang et al., 2006), macrophage phagocytosis (de Almeida et al., 2005) and T cell activation (Ballerini et al., 2006). However, the role of PrP in the innate immune response has hardly been explored with the exception of one study in which peritonitis was induced in PrP−/− mice with zymosan (de Almeida et al., 2005). PrP−/− mice showed decreased numbers of neutrophils and larger numbers of monocytes in peritoneal infiltrates. Although the basis for this phenotype was not investigated, one possibility is that PrP has an intrinsic role in leukocyte migration out of the circulation and/or into the peritoneum. In addition, several in vivo and in vitro studies have shown that PrP participates in the regulation of cell–cell (Málaga-Trillo et al., 2009; Mangé et al., 2002; Morel et al., 2008) and cell–matrix adhesion (Loubet et al., 2012; Schrock et al., 2009).
Leukocyte extravasation takes place through the binding of cytokine-activated leukocytes to the endothelium, followed by their migration across the endothelium and subsequently the vascular basement membrane (Nourshargh et al., 2010). Leukocyte integrin binding to endothelial ICAM-1 and VCAM-1 receptors is crucial for leukocyte arrest and firm adhesion to the endothelium, allowing leukocytes to resist flow-induced detachment. This requires integrin clustering upon ligand binding in order to increase avidity and strengthen adhesiveness, which is mediated by integrin association with the actin cytoskeleton through adaptor proteins (Alon et al., 2005). Migrating leukocytes polarize, with one side of the cell forming a leading edge and the other forming a uropod. The uropod is an adhesive structure containing cholesterol-rich membrane domains, β1 integrins and other adhesion receptors (i.e. ICAM-1 and ICAM-3) as well as actin-membrane linker proteins of the ezrin, radixin and moesin (ERM) family (Friedl and Weigelin, 2008). To migrate, leukocytes need to detach and retract their uropod, a process regulated by the small GTPase RhoA (Liu et al., 2002; Worthylake et al., 2001).
Here, we addressed the role of PrP in monocyte adhesion and migration and show that PrP silencing impairs monocyte adhesion to β1 integrin ligands and to endothelial cells while increasing their motility. We show that PrP regulates ligand-induced integrin activation and provide evidence that ERM proteins and the RhoA–cofilin pathway are involved.
RESULTS
PrP colocalizes with integrins at cholesterol-rich domains in monocytic cells
In vivo experiments in mice have shown that PrP regulates the recruitment of monocytes to inflamed peritoneum and the phagocytosis of dead cells by macrophages, suggesting a role for PrP in innate immunity (de Almeida et al., 2005). However, the mechanism underlying these effects has not been explored. Here, we first investigated the role of PrP in the adhesion and migration of the human pro-monocytic cell line U937. This model has been used extensively to study monocyte adhesion and CXCL12-induced migration (Chan et al., 2001; Hyduk and Cybulsky, 2009) and allows for gene expression manipulation by transfection and transduction techniques.
U937 cells express PrP on their cell surface as assayed by flow cytometry (Fig. 1A). To determine whether PrP accumulates in cholesterol-rich membrane domains (‘lipid rafts’), we performed a membrane flotation assay. PrP was primarily found in the low-density fractions together with the lipid raft marker flotillin-1 (Fig. 1B), indicating that PrP resides in lipid rafts in U937 cells. Further analysis of PrP distribution by confocal microscopy showed that a fraction of U937 cells displayed polarized PrP localization on the membrane (Fig. 1C). This fraction increased from ∼20% to ∼50% upon CXCL12 stimulation. In polarized cells, both PrP and β1 integrins clustered together in membrane caps where they colocalized (Pearson's coefficient larger than 0.5) (Fig. 1C, polarized cell, region of interest A). In contrast, PrP and β1 integrins failed to colocalize in the membrane area devoid of clustered integrins (Fig. 1C, polarized cell, region of interest B) and in non-polarized cells (Fig. 1C, non-polarized cell, regions of interest A and B) (Pearson's coefficient smaller than 0.5). PrP also colocalized with the lipid-raft-resident protein flotillin-1 in polarized cells (supplementary material Fig. S1). These data indicate that integrins are clustered in a fraction of U937 cells in steady-state conditions and that, upon cell polarization, PrP and β1 integrins are recruited to flotillin-1 lipid raft domains.
Both the localization of PrP in lipid rafts and its colocalization with integrins suggest a role for PrP in the regulation of monocyte adhesion and migration. To investigate this, we generated stable U937 cell lines with reduced PrP expression by transduction with lentiviruses expressing three different short hairpin RNAs (shRNAs) against human PrP. In parallel, stable cell lines were generated by transduction with lentivirus expressing a non-targeting shRNA (shCtrl-U937) as a negative control. Flow cytometry detection of surface PrP showed that shPrP-1 (complementary to the 3′UTR of PrP mRNA) was able to reduce PrP expression to ∼35%, whereas shPrP-2 and shPrP-3 (complementary to the PrP coding sequence) reduced PrP expression to ∼70% (Fig. 1D). Considering the moderate efficiency of shPrP-2 and shPrP-3 in silencing PrP expression, we continued our work using stable shPrP-1-transduced cell lines (shPrP-U937 cells).
PrP silencing did not affect the proliferation rate of U937 cells as evidenced by routine cell counting of cells transduced with control shRNA (shCtrl-U937) and shPrP-U937 cell cultures (data not shown). Accordingly, staining of apoptotic cells with annexinV–FITC showed no significant differences between shCtrl- and shPrP-U937 cells (0.3% and 0.2% positive cells, respectively). In addition, there was no difference in CXCL12-induced activation of the survival kinase Akt between control and PrP-silenced cells (data not shown).
PrP regulates β1 integrin adhesion of monocytes
We used stable shCtrl- and shPrP-U937 cell lines to explore the effect of PrP depletion on CXCL12-induced cell chemotaxis in a Transwell assay using fibronectin-coated filters. shPrP-U937 cells showed a significant increase in chemotaxis when compared with shCtrl-U937 cells (Fig. 2A), suggesting that PrP is a negative regulator of chemotaxis. To confirm this, we induced PrP-dependent signaling by cross-linking of surface PrP with the SAF61 antibody, previously shown to trigger signal transduction in T lymphocytes and in a neuronal cell line (Stuermer et al., 2004; Mouillet-Richard et al., 2000). U937 cells incubated with SAF61 displayed impaired migration (Fig. 2B), confirming the negative regulatory effect of PrP on monocytic cell migration and supporting the specificity of the shRNA-mediated PrP silencing. These effects of PrP on cell chemotaxis were not due to changes in CXCR4 surface expression or CXCL12-induced receptor internalization as tested by flow cytometry (data not shown).
We then investigated whether changes in U937 cell migration were paralleled by changes in cell adhesion by recording the electrical resistance at 32 kHz over time of cells adhering to electric cell–substrate impedance sensing (ECIS) electrodes coated with integrin ligands (fibronectin, Fc-ICAM-1 or Fc-VCAM-1) or with bovine serum albumin (BSA) (Fig. 2C–E). This assay is based on the increase in electrical resistance when cells spread on the electrodes following cell adhesion (Wegener et al., 2000). Addition of cells to the electrode arrays resulted in a slight increase in electrical resistance, which was consistently higher for shCtrl-U937 cells compared to shPrP-U937 cells (Fig. 2C), although differences were not significant (data not shown). Comparatively, a large increase in resistance was observed upon the stimulation of cell adhesion with PMA (Fig. 2C). shCtrl-U937 cells seeded on fibronectin-coated electrodes showed a larger increase in PMA-stimulated adhesion than shPrP-U937 cells (Fig. 2C). U937 cells express β1 [α4β1 (VLA-4) and α5β1 (VLA-5)] and β2 integrins [αLβ2 (LFA-1) and αMβ2 (Mac-1)] (Prieto et al., 1994), and under our experimental conditions, preferentially adhered to β1 integrin ligands (fibronectin and VCAM-1) (Fig. 2D,E). Consistent with a role for PrP in the regulation of β1-integrin-mediated adhesion, shPrP-U937 cells showed reduced adhesion to fibronectin (Fig. 2D) and Fc-VCAM-1 (Fig. 2E). Thus, PrP silencing impairs β1-integrin-mediated adhesion of monocytic cells. This reduced integrin-mediated adhesion might facilitate leukocyte motility and migration, which could explain our findings.
To confirm that our findings also apply to primary monocytes, we transduced freshly isolated human monocytes with shPrP-1. ECIS adhesion assays showed similar differences in adhesion to Fc-VCAM between shCtrl- and shPrP-transduced monocytes to that in U937 cells (Fig. 2F,G). Similar results were obtained using the human T lymphocyte cell line Jurkat lymphocytic cells (data not shown). To establish that the differences using the ECIS adhesion assay were due to differences in adhesion strength and not in cell shape (i.e. spreading), we performed conventional adhesion assays by seeding cells on Fc-VCAM-coated plates and counting adherent cells after thorough washing (supplementary material Fig. S2). Taken together, these data indicate that PrP regulates β1-integrin-mediated adhesion of monocytes and suggest that it has a similar role in other leukocyte types.
PrP regulates monocyte adhesion to the endothelium under flow
To assess the effects of PrP silencing in monocyte–endothelium interactions under near-physiological flow conditions we used parallel flow chamber assays. Mixed suspensions of shCtrl- and shPrP-U937 cells (1:1), differently labeled with either Calcein Red-Orange AM or Calcein Green were stimulated with either CXCL12 or PMA. After stimulation, labeled cells were perfused over TNFα-stimulated human umbilical vein endothelial cells (HUVECs), while recording fluorescent and bright-field time-lapse movies (a movie snapshot is shown in Fig. 3A). Adherent cells were quantified in each video frame as described in the Materials and Methods. shPrP-U937 cells adhered in lower numbers when compared with shCtrl-U937 cells, both after cell stimulation with CXCL12 (Fig. 3B) and with PMA (Fig. 3C). Our data show a significant reduction in firm adhesion of shPrP-U937 cells to the endothelium under flow in both conditions (Fig. 3D, CXCL12; Fig. 3E, PMA). Next, we transduced freshly isolated human monocytes with shPrP-1 or with shCtrl and quantified the percentage of monocytes that adhered to CXCL12-coated endothelium in the same assay as described for U937 cells. Our data show that PrP silencing also impaired firm adhesion of primary monocytes to the endothelium (Fig. 3F). We performed similar experiments with PMA-activated monocytes; however, we obtained no significant differences between the adhesion of shCtrl- and shPrP-transduced monocytes. We think this can be explained by the low numbers of adherent cells owing to monocyte activation and adhesion to the tubing of the flow system. Finally, similar results were obtained using PrP-silenced Jurkat cells (supplementary material Fig. S3A). These findings suggest that PrP positively regulates the firm adhesion of monocytes, and possibly other leukocyte types, to inflamed endothelium under physiological conditions of flow-induced shear stress.
PrP regulates cell shape and migratory behavior of U937 cells on immobilized VCAM
Our data above show that PrP regulates monocyte chemotaxis (Fig. 2). To analyze further the effects of PrP silencing on cell migration, we examined the migratory behavior of shCtrl- and shPrP-U937 cells on immobilized Fc-VCAM and CXCL12 by time-lapse video recording. Migrating shCtrl-U937 cells formed long and persistent uropods and moved in a polarized fashion, extending lamellipodia and filopodia at the cell front (Fig. 4A; supplementary material Movie 1). In contrast, shPrP-U937 cells failed to form long uropods (Fig. 4A; supplementary material Movie 1). Manual tracking of migrating cells and subsequent calculation of cell velocity and the accumulated distance of cell trajectories showed that shPrP-U937 cells were faster and covered longer distances than shCtrl-U937 cells (Fig. 4A). Lack of significant differences in directionality suggests that PrP silencing does not affect cell polarization (Fig. 4A). These data corroborate our findings using Transwell chemotaxis assays and suggest that PrP silencing enhances leukocyte motility through the downregulation of integrin-mediated adhesion.
Both β1 integrins and phosphorylated ERM (phospho-ERM) proteins localize to the uropod of migrating leukocytes (Lee et al., 2004; Martinelli et al., 2013). Similarly, phospho-ERM proteins accumulated in the uropod and partially colocalized with β1 integrins in shCtrl-U937 cells migrating on immobilized VCAM-1 (Fig. 4B). In contrast, shPrP-U937 cells mostly lacked uropods and contained reduced levels of phospho-ERM proteins, which localized in discrete spots and in thin membrane protrusions (Fig. 4B). Measurement of uropod length in cells migrating on Fc-VCAM-1 showed that the average length of the uropod was reduced from 26.5 µm to 14.5 µm upon PrP silencing (Fig. 4B). To test whether uropod loss could be rescued by overexpression of PrP in shPrP-U937 cells, we took advantage of the fact that shPrP-1 targets the 3′-UTR of PrP mRNA allowing for rescue with human PrP cDNA. We transfected cells with GFP–PrP or GFP cDNAs by electroporation and subsequently selected GFP-positive cells by flow cytometry. Given that GFP–PrP mainly localizes to the cell middle plane (in the Golgi, not shown), and the signal was therefore low at the bottom plane, we co-detected PrP with SAF32 in non-permeabilized cells to clearly show distribution of PrP on the membrane. Uropod formation in PrP-silenced cells was rescued by overexpression of GFP–PrP, but not GFP (Fig. 4C). In addition, PrP on the cell surface (detected with anti-PrP antibodies in non-permeabilized cells) partially co-localized with endogenous β1 integrin (Pearson's correlation coefficient=0.65±0.01, mean±s.e.m.), which corroborates our findings for cells in suspension.
ERM proteins link membrane proteins to the actin cytoskeletal cortex (Fehon et al., 2010) and active phospho-ERM proteins localize to and regulate formation of the uropod in migrating lymphoblasts (Lee et al., 2004; Martinelli et al., 2013). We observed that the intensity of phospho-ERM labeling appeared to be reduced in shPrP-U937 cells (Fig. 4B). To substantiate this observation, we assessed ERM protein phosphorylation by immunoblotting in cells stimulated with CXCL12 or PMA in a timecourse experiment. As moesin is the predominant ERM isoform expressed in leukocytes (Ivetic and Ridley, 2004), we detected this protein as a control for equal protein loading. CXCL12 induced ERM phosphorylation whereas PMA reduced phospho-ERM levels in a time-dependent manner (Fig. 5A). Consistent with the role of ERM proteins in uropod formation, levels of phospho-ERM proteins were significantly reduced in shPrP-U937 cells stimulated with CXCL12 and slightly decreased after stimulation with PMA (Fig. 5A). In addition, basal phospho-ERM levels were 1.5-fold lower in PrP-silenced cells when compared to control. These data suggest that PrP regulates uropod formation in monocytic cells through the regulation of ERM protein phosphorylation.
ERM proteins are known to link the uropod marker CD44 to the actin cytoskeleton (Friedl and Weigelin, 2008; Fehon et al., 2010). To determine whether the reduction in ERM protein activation observed in shPrP-U937 cells had any effects on CD44 distribution, we immunodetected CD44 and flotillin-1 in cells migrating on immobilized Fc-VCAM. In shCtrl-U937 cells, CD44 was enriched in uropods together with flotillin-1 (Fig. 5B). In contrast, CD44 was more uniformly distributed on the membrane of shPrP-U937 cells, although flotillin-1 still showed a polarized distribution (Fig. 5B). These data indicate that PrP silencing alters CD44 membrane distribution.
Next, we examined in detail the migratory behavior of shCtrl- and shPrP-U937 cells arrested on the endothelium under near-physiological flow conditions as described above. Firm adhesion of leukocytes to the endothelium is enhanced by PMA treatment, which stimulates integrin adhesion and consequently inhibits leukocyte motility (Cinamon et al., 2001). As expected, PMA-stimulated shCtrl-U937 cells remained mostly round and immotile after arrest and adhesion to the endothelium. In contrast, PMA-stimulated shPrP-U937 cells crawled over the endothelium while forming dynamic bleb-like membrane protrusions (Fig. 6A; supplementary material Movies 2–4). On average, 30% of shPrP-U937 cells arrested on the endothelium displayed dynamic protrusions in contrast to only 10% of shCtrl-U937 cells (Fig. 6B). PMA-activated shPrP-monocytes also showed increased membrane activity compared to shCtrl-transduced monocytes (supplementary material Movie 5), although differences were smaller (the percentages of adherent cells with membrane protrusions were 38.5% for shCtrl- transduced monocytes and 53.3% for shPrP-transduced monocytes). However, these data should be taken with caution owing to the low number of PMA-activated monocytes that were flown over the endothelium. Finally, this phenotype appears not to be unique to monocytic cells because PrP-silenced Jurkat cells displayed similar bleb-like motility on activated endothelium to that of U937 cells (supplementary material Fig. S3B).
Finally, we assessed the role of PrP in leukocyte transendothelial migration by quantification of the number of shCtrl-and shPrP-U937 cells undergoing diapedesis across a TNFα-activated CXCL12-coated endothelium under flow conditions (Grabovsky et al., 2000). Predictably, a significantly higher percentage of shPrP-U937 cells transmigrated across the endothelium when compared to control cells (Fig. 6C). Thus, following arrest on the endothelium, PrP-silenced cells appear to have a diminished capacity to establish firm adhesion, displaying increased motility together with high membrane protrusive activity, as well as increased diapedesis.
PrP regulates integrin activity
Integrin-mediated adhesion is regulated by conformational changes (affinity) as well as by the number of integrin molecules engaged in ligand binding (avidity) (Carman and Springer, 2003). Our data show that PrP silencing leads to reduced monocyte adhesion to β1 integrin ligands. To gain insight into the molecular mechanism underlying these effects, we first investigated whether PrP silencing impaired integrin affinity upregulation in in cells stimulated or not with CXCL12, PMA or Mn2+ for 10 min. PMA and CXCL12 induce intracellular signaling that leads to integrin activation, whereas Mn2+ binds to surface integrins triggering a signaling-independent conformational change (Carman and Springer, 2003). Total and activated β1 integrin on the surface of shCtrl- and shPrP-U937 cells were assessed by flow cytometry. No differences in surface expression of total or activated integrins were detected between shCtrl- and shPrP-U937 cells (Fig. 7A).To further examine the effect of PrP silencing on stimulus-induced upregulation of high-affinity β1 integrins, we stimulated cells with CXCL12, PMA or Mn2+ in the presence of soluble Fc-VCAM. Cell-bound Fc-VCAM was subsequently detected using APC-labeled F(ab′)2 goat anti-human-IgG antibody and flow cytometry analysis (Chan et al., 2003). shPrP-U937 cells consistently bound less Fc-VCAM (Fig. 7B). Direct integrin activation with Mn2+ induced higher Fc-VCAM binding and reduced the differences between shCtrl- and shPrP-U937 cells (Fig. 7B). These data suggest that PrP silencing does not affect the number of integrins on the cell surface but decreases ligand-induced integrin activation. We assessed this further by using multimeric Fc-VCAM complexes. For this, preformed soluble multimeric ligand complexes were generated by incubation of Fc-VCAM with IgG F(ab′)2 goat anti-human-IgG antibody (Konstandin et al., 2006). This assay was previously shown to detect both integrin affinity and avidity in leukocytes based on the fact that polyclonal F(ab′)2 fragments recognize different epitopes of the human Fc moiety of Fc-VCAM (Konstandin et al., 2006). The differences in ligand binding between control and PrP-deficient cells were more pronounced when multimeric Fc-VCAM complexes were used (Fig. 7C). The exacerbation of the differences between control and PrP-silenced cells when using a multivalent ligand suggests that PrP silencing primarily impairs ligand-induced integrin avidity.
PrP regulates RhoA activation and actin polymerization
Integrin avidity is enhanced upon increased lateral mobility of integrins on the plasma membrane, which allows them to form microclusters (Bakker et al., 2012; Buensuceso et al., 2003). Integrin binding to the actin cytoskeleton might impose constraints to the lateral mobility of integrins and reduce avidity. In addition, PrP-silencing has previously been shown to stabilize actin filaments in neuronal cells through a RhoA-dependent pathway (Loubet et al., 2012). Therefore, we explored RhoA activation in shCtrl- and shPrP-U937 cells stimulated with CXCL12 or PMA. shPrP-U937 cells displayed significantly elevated GTP-RhoA levels after 2 and 5 min of CXCL12 stimulation (Fig. 8A). In contrast to CXCL12, PMA stimulation decreased RhoA activation. However, the levels of GTP-RhoA were also slightly elevated in PMA-stimulated PrP-silenced cells when compared to control cells (Fig. 8A).
RhoA regulates the actin cytoskeleton by inducing both actin polymerization and actin microfilament crosslinking and remodeling. Therefore, we investigated actin polymerization dynamics in CXCL12-stimulated suspension cells using a more sensitive flow-cytometry-based assay (Voermans et al., 2001). In agreement with previous published data (Voermans et al., 2001), we observed a burst of F-actin after 0.5 min stimulation, followed by a decrease to unstimulated levels at 10 min (Fig. 8B). PrP-silenced cells showed higher F-actin content at every time point (Fig. 8B). Low concentrations of cytochalasin B are reported to enhance actin remodeling, whereas at high concentrations, cytochalasin B causes actin cytoskeleton depolymerization. To test whether altered actin dynamics underlies the adhesion defects of PrP-silenced cells, we performed an ECIS adhesion assay in the presence of low concentrations of cytochalasin B. At 1 µM, cytochalasin B increased adhesion of shPrP-U937 cells to control levels (Fig. 8C). At 5 µM, cytochalasin B, reduced PMA-induced adhesion of both shCtrl and shPrP-U937 cells (Fig. 8C). These data suggest that reduced actin dynamics mediates the effects of PrP silencing in cell adhesion.
It has been shown previously that the RhoA-regulated actin-severing protein cofilin 1 is involved in the regulation of actin microfilament stability by PrP in neurons (Loubet et al., 2012). To determine whether this pathway is operational in monocytes, we determined the levels of phosphorylated (inactive) cofilin 1 in cells adhering to immobilized Fc-VCAM. We found that PrP-silencing increases phosphorylated cofilin 1 levels (Fig. 8D) and thus inactivates cofilin1. Taken together, these findings provide evidence suggesting that PrP regulates monocyte adhesion and migration through the regulation of actin remodeling by the RhoA–cofilin pathway.
DISCUSSION
Leukocyte migration across the vascular wall is of crucial importance for the immune response. It has been previously shown that PrP is involved in the influx of immune cells into the peritoneal cavity during zymosan-induced peritonitis, where monocytes are recruited in larger numbers in PrP−/− mice compared to wild-type controls (de Almeida et al., 2005). Our study now provides evidence for the molecular mechanisms underlying the effects of PrP silencing on monocyte adhesion and migration, and corroborates the immunomodulatory role of PrP.
Our findings show that PrP silencing increases the chemotaxis and motility of monocytic U937 cells on β1 integrin ligands in static conditions as well as on endothelium in near-physiological flow conditions. In addition, we show that PrP-deficient U937 cells are able to transmigrate more efficiently than control cells. In contrast, activation of endogenous PrP signaling by antibody-mediated crosslinking of surface PrP impairs U937 cell chemotaxis. Collectively, these data suggest that PrP is a negative regulator of monocyte (transendothelial) migration.
Leukocyte migration is dictated by the strength of integrin-mediated cell adhesion. Although leukocyte attachment to the endothelial surface is required to resist flow-induced shear stress forces, excessive adhesion abrogates leukocyte motility on the endothelium as well as diapedesis across the endothelial monolayer (Cinamon et al., 2001). We show that PrP silencing impairs pro-monocyte adhesion to β1 integrin ligands (fibronectin and VCAM-1) in static conditions, as well as to TNFα-activated endothelial cells under flow. In addition, we show that primary monocytes and Jurkat cells also display defective adhesion and enhanced motility upon PrP silencing, suggesting that PrP regulates β1-integrin-mediated interactions of monocytes (and probably T cells) with the endothelium in vivo.
The regulation of integrin-mediated adhesion takes place through changes in affinity, avidity and/or trafficking (Margadant et al., 2011). We observed no changes in integrin affinity in PrP-silenced cells by using either an antibody specific for the high-affinity conformation of β1 integrin or a soluble β1 integrin ligand (Fc-VCAM1). Likewise, we detected no differences between control and PrP-silenced cells in β1 integrin internalization by using an antibody-feeding assay (data not shown). However, PrP silencing impaired the binding of multimeric Fc-VCAM complexes to cells. These data suggest that PrP controls β1 integrin adhesiveness through the regulation of ligand-induced changes in integrin activation, likely through the regulation of integrin avidity, which is determined by the number of integrins engaged in ligand binding (Carman and Springer, 2003). VLA-4 (α4β1) integrin microclustering can induce adhesiveness by increasing valency independently of changes in integrin affinity (Grabovsky et al., 2000). The attachment of the integrin cytosolic tail to the actin cytoskeleton restrains the lateral mobility of integrins on the membrane and consequently impairs microclustering (Bakker et al., 2012; Buensuceso et al., 2003). Thus, local microfilament remodeling might be required to allow disengagement of integrins from cortical actin and allow them to cluster, therefore increasing the strength of ligand binding (Bakker et al., 2012; Worthylake and Burridge, 2003). We show here that PrP silencing increases CXCL12-induced RhoA activation as well as F-actin content. Interestingly, although CXCL12-induced actin polymerization follows similar kinetics in shCtrl- and shPrP-U937 cells, actin depolymerization is slower in PrP-silenced cells (Fig. 8). Notably, increasing actin depolymerization rates with low concentrations of cytochalasin B abrogates the effects of PrP silencing on cell adhesion. Consistent with this, we found that PrP silencing induces the inactivation of the actin-severing protein cofilin 1. Inactivation of cofilin by phosphorylation on Ser3 is mediated by LIM kinase, which in turn is activated by RhoA (Bravo-Cordero et al., 2013). In view of the above findings, we propose that the stabilization of actin filaments by RhoA-induced cofilin inactivation prevents the release of activated integrins from the cytoskeleton in PrP-deficient cells, therefore impairing integrin microclustering and adhesiveness. Our findings are in agreement with previously reported data showing increased cytoskeletal stability, RhoA activation and cofilin inactivation in PrP-deficient neuronal cells (Loubet et al., 2012). Interestingly, overexpression of PrP in primary rat hippocampal neurons induces cofilin activation and the formation of actin–cofilin rods (Walsh et al., 2014). These structures are found in Alzheimer disease brain and have been shown to impair synaptic function (Minamide et al., 2000). It is tempting to speculate that cofilin regulation is crucial both for the physiological function of PrP and for prion-induced neurodegeneration.
We found that PrP colocalizes with integrins in pre-formed flotillin-1 domains (Langhorst et al., 2006). In contrast to in T cells, where PrP recruitment to flotilin-1 membrane caps can be induced by antibody-mediated PrP crosslinking (Stuermer et al., 2004), we observed that PrP co-clustered with integrins in a fraction of U937 cells in the absence of stimulation or antibody-mediated crosslinking. This suggests that a percentage of U937 cells are ‘primed’ in resting conditions, which could explain the differences observed in F-actin content and ligand binding in unstimulated cells. Despite the colocalization of PrP and β1 integrin in flotillin-1 domains, we were unable to show association of PrP and β1 integrin in co-immunoprecipitation assays (data not shown), although β1 integrin was previously identified as a PrP-interacting partner in a proteomic study (Watts et al., 2009). We found that PrP silencing dramatically altered the morphology and migratory behavior of cells moving on 2D surfaces. shCtrl-U937 cells migrating on immobilized VCAM accumulated β1 integrins, phospho-ERM, flotillin-1 and CD44 in the uropod, all bona fide uropod markers (Shulman et al., 2009; Gómez-Moutón and Mañes, 2007). In contrast, PrP-depleted cells generally failed to form uropods and displayed a round shape, although β1 integrin and flotillin-1 still showed a polarized distribution. Flotillins are essential for uropod formation in neutrophils (Ludwig et al., 2010). Although PrP colocalizes with flotillin in U937 cells, flotillin distribution is not affected upon PrP silencing, which suggests that PrP does not regulate uropod formation through the disruption of flotillin domains.
The uropod is an adhesive structure that needs to be retracted during migration (Liu et al., 2002; Smith et al., 2003; Worthylake et al., 2001). In agreement with this, PrP-deficient cells migrated faster than control cells. We found that PrP silencing alters the activation of ERM proteins and RhoA, which are regulators of uropod dynamics. First, PrP silencing reduces ERM protein phosphorylation. ERM proteins are activated by phosphorylation of a conserved C-terminal threonine residue, which enables them to crosslink membrane receptors with the underlying actin cytoskeleton (Fehon et al., 2010). ERM protein activation is crucial for uropod formation (Ivetic and Ridley, 2004; Lee et al., 2004; Martinelli et al., 2013) and for β1-integrin-dependent T cell adhesion and polarization (Chen et al., 2013; Liu et al., 2002). Therefore, the inhibition of ERM phosphorylation in the absence of PrP likely contributes to both the lack of uropods and the reduced β1 integrin adhesion observed for PrP-deficient monocytes.
RhoA is crucial for uropod de-adhesion and retraction during migration (Liu et al., 2002; Smith et al., 2003; Worthylake et al., 2001). In neutrophils, activation of myosin by RhoA has been implicated in actomyosin-driven uropod retraction (Eddy et al., 2000). However, we found that phospho-myosin levels were equal in shCtrl- and shPrP-U937 cells, and that they did not increase upon stimulating migration (data not shown). This suggests that actomyosin contraction does not play a role in uropod retraction in monocytic cells. In line with this, myosin inhibition did not affect uropod retraction in migrating monocytes (Worthylake et al., 2001).
PMA-dependent integrin activation induces firm leukocyte adhesion to the endothelium and blocks motility and migration (Cinamon et al., 2001). Thus ‘excessive’ integrin activation prevents migration. In line with this, PMA-stimulated control cells firmly adhered to the endothelium, whereas PMA-stimulated PrP-deficient cells crawled over the endothelium while protruding dynamic bleb-like extensions. This motility could be prompted by the weaker adherence of PMA-simulated PrP-silenced cells. In addition, decreased membrane tension due to lower levels of active ERM proteins might also contribute to this phenotype. In support of this, constitutively active ezrin has been shown to increase lymphocyte membrane tension, reduce chemokine-induced shape changes and block transendothelial migration (Liu et al., 2002).
In summary, we propose that PrP has a role in integrin-mediated adhesion and motility through the regulation of ERM-membrane–cortex adaptors and the RhoA–cofilin pathway, which controls cytoskeleton remodeling. This might be a general mechanism for the reported regulation of cell–cell and cell–matrix adhesion by PrP in different cell types (Petit et al., 2013). In addition, our data contributes to the understanding of additional roles of PrP in neurodegeneration.
MATERIALS AND METHODS
Cell culture and lentiviral transduction
Human pro-monocytic U937 cells were transduced with lentivirus containing the validated Sigma MISSION shRNA plasmids targeting human PrP (shPrP-1, TRCN0000083488; shPrP-2, TRCN0000083490; or shPrP-3, TRCN0000083491) or non-targeting shRNA (shCtrl, SHC002) as a negative control. Subsequently shRNA-expressing cells were selected by culturing in RPMI medium containing 10% fetal calf serum and 1.5 µg/ml puromycin. Before each experiment, cells were cultured in puromycin-free medium for at least 24 h. PrP expression in control and knockdown cell lines was tested before each experiment by flow cytometry with the SAF32 anti-PrP antibody. HUVEC (Primary Human Umbilical Vein Endothelial Cells) were cultured in endothelial growth medium-2, both from Lonza.
Human monocytes were isolated from PBMCs by positive selection with magnetic CD14 beads (Milteny Biotech). For transduction, shRNA lentiviruses were incubated with 4 µg/ml polybrene for 10 min at room temperature prior to their addition to monocyte suspensions (2×106 cells/ml) in complete RPMI. M-CSF was added at a concentration of 10 ng/ml to keep monocytes steady without inducing differentiation (Troegeler et al., 2014). After 48 h, PrP expression was analyzed by flow cytometry. The maximum efficiency of PrP knockdown was ∼30–60%. Only those monocytes with at least 50% knockdown were used for experimental assays.
Antibodies and reagents
Mouse monoclonal antibodies against PrP (SAF32 and SAF61) were from Spi-Bio. Anti-β1 integrin antibody (clone P5D4) was from Abcam. Activated β1 integrins were detected with HUTS-4 (Millipore). Flotillin-1 rabbit polyclonal antibody was from Abcam and transferrin receptor mouse monoclonal antibody from Life Technologies. RhoA monoclonal mouse antibody was from Santa Cruz Biotechnology. Rabbit antibodies against phospho-ERM proteins [ezrin (Thr567), radixin (Thr564), moesin (Thr558)], phospho-cofilin (Ser3), cofilin and myosin light chain (MLC) as well as mouse monoclonal antibody against phosphorylated myosin light chain (Ser19) were from Cell Signaling. Mouse monoclonal anti-moesin was from BD Biosciences. FITC-labeled anti-CXCR4 mouse monoclonal was from R&D Systems. Mouse monoclonal anti-actin was from Sigma. Horseradish peroxidase (HRP)- and Alexa-Fluor-dye-conjugated secondary antibodies were from Dako and Life Technologies, respectively. Human fibronectin was from Biopur. Recombinant human Fc-ICAM and Fc-VCAM were from R&D Systems. The APC-labeled polyclonal goat antibody human IgG F(ab′)2 was from Jackson ImmunoResearch Laboratories. AnnexinV–FITC was from Pharmingen. Human recombinant TNFα and CXCL12 were from Peprotech. Phorbol-12-myristate-13-acetate (PMA) and cytochalasin B were from Sigma. Calcein Green AM and Calcein Red-Orange AM were from Life Technologies.
Membrane flotation assay
20×106 cells were lysed on ice for 15 min in TXNE buffer containing 50 mM Tris-HCl pH 7.4, 150 mM NaCl, 5 mM EDTA, 1 mM DTT, 1% Triton X-100 and cOmplete Protease Inhibitor Cocktail Tablets (Roche). Lysates were mixed with 1.8 ml of 60% OptiPrep (Sigma) in TXNE buffer. Subsequently, 7.5 ml of 28% OptiPrep and 1.8 ml of TXNE buffer were layered on top. Gradients were centrifuged at 200,000 g for 3 h at 4°C. 1-ml fractions were taken from the top of the gradient and proteins were precipitated for 30 min on ice by addition of an equal volume of 20% trichloroacetic acid and pelleted at 10,000 g for 15 min at 4°C. Pellets were washed with cold acetone and solubilized in SDS sample buffer.
Chemotaxis
Transwell chemotaxis assays were performed and analyzed as previously described (Lorenowicz et al., 2006). Crosslinking of surface PrP was performed by incubating cells for 30 min at 37°C with 0.5 µg/ml of the SAF61 anti-PrP antibody.
ECIS adhesion assays
Cell adhesion to fibronectin, BSA, Fc-VCAM or Fc-ICAM was measured by using the electrical cell–substrate impedance sensing system (ECIS, Applied Biophysics). ECIS 8W10E electrode arrays were coated with 20 µg/ml fibronectin or BSA for 16 h at 4°C. For Fc-ICAM and VCAM coating, electrode arrays were first coated with 4 µg/ml of AffiniPure F(ab′)2 goat anti-human-IgG antibody (Jackson ImmunoResearch) for 1 h at 37°C in 20 mM Tris-HCl pH 8.0, 150 mM NaCl, 1 mM CaCl2 and 2 mM MgCl2. BSA (1%) was added for 30 min at 37°C to block non-specific binding. After blocking, human recombinant Fc-ICAM-1 or Fc-VCAM-1 were added at a concentration of 1 µg/ml and incubated for 16 h at 4°C. The electrical resistance of cells (500,000 cells per well) seeded on protein-coated arrays was measured in real-time at a frequency of 32 kHz.
Plate adhesion assays
Cells (100,000 cells per well) were added to F(ab′)2 and Fc-VCAM-coated wells of flat-bottomed 96-well Maxisorp plates (coating was performed as described for the ECIS arrays). Immediately after cell seeding, stimuli were added to the wells and cells were allowed to adhere for 30 min at 37°C. Subsequently, non-adherent cells were removed by washing three times with PBS containing 0.5% BSA. Adherent cells were fixed with 3.7% formaldehyde, incubated with Hoechst 33342 to stain nuclei and counted on an Axiovert wide-field microscope (Zeiss).
Adhesion and transendothelial migration under flow
Confluent HUVEC monolayers on fibronectin-coated parallel-plate flow chamber slides (µ-slide VI0.4, Ibidi) were treated with 10 ng/ml TNFα for 16 h. Flow chambers were mounted on an inverted wide-field fluorescence microscope (Axiovert 200, Zeiss) fitted in a 37°C and 5% CO2 chamber. shCtrl- and shPrP-U937 cells were differently labeled with either Calcein Green AM or Calcein Red-Orange AM (2.5 µM, Life Technologies). After washing, cells were resuspended in assay buffer [20 mM Hepes pH 7.4, 132 mM NaCl, 6 mM KCl, 1.2 mM K2HPO4, 1 mM MgSO4, 1 mM CaCl2, 0.1% D-(+)-glucose and 2.5% human albumin]. Differently calcein-labeled shCtrl- and shPrP-U937 cells were mixed at 1:1 ratio to a final total cell concentration of 1×106 cells/ml and subsequently stimulated with 100 ng/ml PMA or with 100 ng/ml CXCL12 for 10 min. Stimulated cell mixes were flown over TNFα-activated HUVEC monolayers at a flow rate of 0.5 ml/min, corresponding to a shear stress of 0.9 dynes/cm2. Time-lapse bright-field and fluorescence images were acquired from four random microscope fields for 30–60 min using an Orca R2 digital CCD camera (Hamamatsu). Quantification of the number of adherent cells per frame was performed using the ‘analyze particles’ application of ImageJ after thresholding of the fluorescence video images. Transendothelial migration of U937 cells was assessed in similar experiments as above except for the immobilization of CXCL12 on the endothelial surface (100 ng/ml, 15 min at 37°C) prior to U937 cell perfusion. Time-lapse microscopy was performed with a LSM510 META confocal microscope (Zeiss). Scoring of transmigrating cells was based on shape changes (from round to spread) and loss of brightness in the DIC channel.
Cell motility on immobilized Fc-VCAM-1
IBIDI µ-slides were coated with F(ab′)2 goat anti-human-IgG and Fc-VCAM-1 as described for the ECIS arrays. Subsequently, CXCL12 was absorbed on the dishes by incubation for 1 h at 37°C with 100 ng/ml CXCL12 in assay buffer. Motility of cells on Fc-VCAM-1 was recorded from four microscopic fields with a wide-field microscope as described above. Cell tracking on videos and calculation of accumulated distance and velocity was performed using Gradientech Tracking Tool software.
Flow cytometry
Cells were stimulated with CXCL-12 (100 ng/ml) for 15 min at 37°C, cooled on ice and labeled with SAF32 anti-PrP antibody and anti-β1- or anti-β2-integrin antibodies for 1 h at 4°C. Cells were then fixed in 4% paraformaldehyde, blocked in 50 mM NH4Cl and incubated with Alexa-Fluor-dye-labeled secondary antibodies. Labeled cells were analyzed by flow cytometry using a FACSCanto flow cytometer (BD Biosciences). Integrin activation was induced with 100 ng/ml PMA or with 1 mM MnCl2 for 10 min at 37°C in Hank's balanced salt solution containing 1 mM Ca2+ and 1 mM Mg2+.
Timecourse experiments for Fc-VCAM binding to cells were performed as described previously (Chan et al., 2003). For quantification of the binding to cells of Fc-VCAM and APC-labeled (Fab')2 complexes, we followed the protocol described by Konstandin et al. (Konstandin et al., 2006). Fluorescence was measured by flow cytometry using a FACSCanto flow cytometer (BD Biosciences).
Immunofluorescence
For confocal microscopy analysis, antibody-labeled cells as above were resuspended in mounting medium (Vectashield) containing DAPI, and mounted on a glass slide. Alternatively, cells were seeded on Fc-VCAM and CXCL12-coated IBIDI µ-slides, allowed to migrate and/or polarize for 20 min before being fixed with 4% paraformaldehyde in assay buffer, blocked using 50 mM NH4Cl in PBS and subsequently permeabilized with PBS-0.1% Triton X-100 for 2 min. After blocking in PBS containing 1% BSA, cells were subsequently incubated with primary antibodies for 30 min at room temperature and Alexa-Fluor-488-labeled goat anti-mouse-IgG antibodies. To stain nuclei, Hoechst 33342 was added in the washing buffer. Fluorescence images were obtained with the LSM510 META (Zeiss) or with a SP8 (Leica) confocal microscopes.
For colocalization analysis between PrP and β1 integrin, regions of interest (ROIs) were selected on digital pictures and Pearson's coefficients were determined using the Colocalisation Test plugin from ImageJ (Costes randomization method). In polarized cells, ROIs were selected in the capped and non-capped membrane domains. In non-polarized cells, ROIs were made around the membrane of each cell halves. We considered that colocalization existed at Pearson's coefficient values from 0.5 to 1.0.
DNA constructs
Monomeric EGFP (mEGFP) was inserted between the signal peptide (amino acids 1–22) and mature PrP (amino acids 23–253) by ligation independent cloning. A 121-bp fragment containing the PrP signal peptide was PCR amplified from human PrP cDNA (TrueClone from Origene) using primers MFB001 (5′-GCAGGGGCGCAACAGACCCCGGTGCCACCATGGCGAACCTTGGCTGCTGGATGC-3′) and MFB002 (5′-CCACCAGGCCGGCCAGCACCCGGTCCGCAGAGGCCCAGGTCACTCCATGTG-3′) and inserted in frame in front of mEGFP in XmaI-digested mEGFP-LIC as described previously (Bierings et al., 2012), resulting in sp-mEGFP-LIC. A 754-bp fragment containing mature PrP was PCR amplified using primers MFB003 (5′-GGGCGCGCCTGGTGGGGCCAAGAAGCGCCCGAAGCCTGGAGGATG-3′) and MFB004 (5′-GCGGCCGCCTGCTCGTCCATCATCCCACTATCAGGAAGATGAGG-3′) and inserted in frame behind mEGFP in SacII-digested sp-mEGFP-LIC using ligation independent cloning. All plasmids were verified by sequence analysis.
Rescue of PrP expression
U937 cells stably expressing shRNA targeting human PrP were cultured for 48 h in puromycin-free medium. After 48 h cells were transfected by electroporation (Neon Transfection System, Life Technologies) with a GFP–PrP or GFP construct according to the manufacturer's recommendations. GFP-positive cells were sorted using a FACSAria II cell sorter (BD Biosciences). Cells were allowed to recover for 2 h in complete medium before seeding on Fc-VCAM-coated IBIDI µ-slides.
GTPase activation assay
Cells were lysed in 25 mM Tris-HCl pH 7.2, 150 mM NaCl, 10 mM MgCl2, 1% NP-40 and 5% glycerol and cOmplete Protease Inhibitor Cocktail Tablets, Roche. RhoA-GTP was isolated from cell lysates using a pulldown assay as previously described (Alblas et al., 2001).
Immunoblotting
Cells were stimulated in suspension and lysed in cold NP-40 buffer (25 mM Tris-HCl, pH 7.2, 150 mM NaCl, 10 mM MgCl2, 10 mM CaCl2 1% NP-40 and 5% glycerol) containing Halt™ protease and phosphatase inhibitor cocktail (Thermo Scientific). Phospho-ERM and total moesin levels were determined by western blotting. Pixel density was determined using ImageJ. Data represent phospho-ERM-to-moesin ratios and were normalized to the value at t=0 in shCtrl cells.
To determine phospho-cofilin and total cofilin levels, U937 cells were lysed in sample buffer after 30 min of adhesion to VCAM-coated plates. Equal loading was controlled by detection of total cofilin. Pixel density values was determined using ImageJ. Data represent phospho-cofilin-to-cofilin ratios. Values were normalized to shCtrl.
Actin polymerization assay
CXCL12-induced changes in intracellular content of F-actin were measured by Alexa-Fluor-488–phalloidin staining as described previously (Voermans et al., 2001). Labeled cells were analyzed by flow cytometry using a FACSCanto flow cytometer (BD Biosciences)
Statistical analysis
Data are represented as mean±s.d. or mean±s.e.m. Statistical comparisons of means were performed using a two-tailed Student's t-test, one-way ANOVA with a Bonferroni post hoc test or two-way ANOVA with a Bonferroni's multiple comparisons post hoc test. Statistical significance is represented by asterisks (*P<0.05, **P<0.01, ***P<0.001, ****P<0.0001).
Acknowledgements
We are grateful to Dr Peter Hordijk for critically reading our manuscript. We also thank Erik Mul for his assistance with flow cytometry assays and Nicoletta Sorvillo for helping with monocyte isolation. We are grateful to Maaike Scheenstra for sharing with us her protocol for monocyte transduction and to Dr Keith Burridge and Dr Metello Innocenti for their helpful comments.
Author contributions
D.D.R., S.T., E.V.-E., C.P. and R.B. performed the research and analyzed the data; M.F.-B. designed the research, analyzed the data and wrote the paper. D.G. contributed reagents. D.D.R. and S.T. contributed equally to this study.
Funding
This work was financed with a grant from Sanquin Blood Supply [grant number PPOC 11-006]. R.B. was supported by a European Hematology Association Research Fellowship.
References
Competing interests
The authors declare no competing or financial interests.