2 and Akt in macrophages

Cellular Signalling 24 (2012) 1753–1761

Contents lists available at SciVerse ScienceDirect

Cellular Signalling journal homepage: www.elsevier.com/locate/cellsig

CSF-1 receptor signalling from endosomes mediates the sustained activation of Erk1/2 and Akt in macrophages Jennifer Huynh, Mei Qi Kwa, Andrew D. Cook, John A. Hamilton, Glen M. Scholz ⁎ Department of Medicine, The University of Melbourne, Royal Melbourne Hospital, Parkville, Victoria 3050, Australia

a r t i c l e

i n f o

Article history: Received 17 April 2012 Accepted 27 April 2012 Available online 7 May 2012 Keywords: CSF-1 receptor Endocytosis Erk1/2 Stat3 Cyclin D1 Spatiotemporal signalling

a b s t r a c t Colony stimulating factor-1 (CSF-1) mediates its pleiotropic effects on macrophages through the CSF-1 receptor (CSF-1R), a receptor tyrosine kinase. Current models of CSF-1 signalling imply that the CSF-1R activates signalling pathways exclusively at the plasma membrane and the subsequent internalisation of the CSF-1R simply facilitates its lysosomal degradation in order to prevent on-going signalling. Here, we sought to establish if the CSF-1R may in fact continue to signal following its internalisation. Erk1/2, Akt and Stat3 activation were abrogated when the internalisation of the CSF-1R was impaired, with the effects on Stat3 distinct from those for Erk1/2 and Akt. Pharmacologic inhibition of the CSF-1R following its internalisation resulted in less sustained Erk1/2 and Akt activity, whereas Stat3 activity was unaffected. Significantly, the suppressive effects of the CSF-1R inhibitor on the up-regulation of gene expression by CSF-1 (e.g. cyclin D1 and Bcl-xL gene expression) were comparable irrespective of whether the inhibitor was added prior to CSF-1 stimulation or following the internalisation of the CSF-1R. Similarly, pharmacologic inhibition of Erk1/2 (or Akt) activity either prior to CSF-1 stimulation or subsequent to CSF-1R internalisation had comparable effects on the regulation of gene expression by CSF-1. Together, our data argue that key signalling responses to CSF-1 depend on the ability of the CSF-1R to signal from endosomes following its internalisation, thus adding an important spatiotemporal aspect to CSF-1R signalling. © 2012 Elsevier Inc. All rights reserved.

1. Introduction Macrophages serve important roles in both host defence and tissue homeostasis [1–4]. These roles include the secretion of inflammatory cytokines (e.g. TNF, IL-1β and IL-10) and the phagocytosis of pathogens. Macrophages are also central to the clearance of dead host cells, which can arise as a result of infection, injury, or normal tissue turnover and/or remodelling. However, macrophages also directly contribute to a number of important diseases. For example, macrophage-derived inflammatory cytokines (e.g. TNF) drive the pathology of septic shock, rheumatoid arthritis, and inflammatory bowel disease [5–7]. Similarly, the secretion of angiogenic factors and growth factors (e.g. VEGF and EGF) by macrophages is considered to be important in breast cancer [8]. Colony stimulating factor-1 (CSF-1) is a critical regulator of macrophage survival, proliferation and differentiation [9]. It can also regulate important macrophage functions, including inflammatory cytokine production and phagocytic activity [9]. In vivo CSF-1 exists as three biologically active isoforms — a secreted glycoprotein, secreted proteoglycan, and membrane-bound glycoprotein [9–12]. All three isoforms mediate

Abbreviations: BMM, bone marrow-derived macrophages; ccl2, chemokine (C–C motif) ligand 2 (also known as MCP-1, monocyte chemotactic protein-1); CSF-1, colony stimulating factor-1; CSF-1R, CSF-1 receptor; HGF, hepatocyte growth factor. ⁎ Corresponding author. Tel.: + 61 3 8344 3298; fax: + 61 3 9347 1863. E-mail address: [email protected] (G.M. Scholz). 0898-6568/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.cellsig.2012.04.022

their effects through the CSF-1 receptor (CSF-1R), a receptor tyrosine kinase (RTK) that is structurally related to the PDGF receptor, Flt3 and c-Kit [13,14]. CSF-1 triggers the dimerisation of the CSF-1R and the subsequent autophosphorylation of specific tyrosine residues (e.g. Tyr723) in the intracellular domain of the CSF-1R that can serve as docking sites for SH2 domain containing proteins, such as Grb2, PI-3 kinase, and Src-family kinases. The recruitment of these proteins to the CSF-1R facilitates the activation of additional signalling proteins, including Erk1/2, Akt and Stat3, which ultimately leads to changes in the expression of CSF-1 target genes (e.g. cyclin D1 and Bcl-xL) and a cellular response to CSF-1 [13,14]. Notably, current models of CSF-1 signalling suggest that the CSF-1R activates signalling pathways exclusively at the plasma membrane and that the ensuing internalisation and endocytic trafficking of the CSF-1R to lysosomes merely facilitate its degradation in order to prevent on-going signalling [13,14]. However, such models of CSF-1R signalling largely fail to explain how CSF-1 mediates its pleiotropic effects on macrophages or, for that matter, how the secreted and membrane-bound isoforms of CSF-1 are able to induce overlapping as well as different biological responses [9–12]. The endocytic system has traditionally been viewed as a conduit through which RTKs are recycled or degraded [15]. Mounting evidence now suggests that endosomes can also fulfil important signalling functions [16–21]. For instance, they may facilitate the spatiotemporal propagation of signalling responses that are first activated by RTKs at the plasma membrane. Endosomes may also allow RTKs to activate signalling

1754

J. Huynh et al. / Cellular Signalling 24 (2012) 1753–1761

pathways that are different from those they activate at the plasma membrane via signalling proteins uniquely localised to endosomes. The cellular response/s to a growth factor may therefore not only depend on which specific signalling proteins are activated, but also from where in the cell the signalling proteins are activated (e.g. plasma membrane and/or endosomes) and for how long their activation is maintained [16–21]. Only limited information about endosomal signalling by a few RTKs (e.g. EGFR and c-Met) has emerged so far [22–27]. Moreover, the general applicability of endosomal signalling by RTKs is still unclear. Here, we present data which suggest the CSF-1R signals from endosomes following its CSF-1-induced internalisation. Specifically, our data suggest that the sustained activation of Erk1/2 and Akt by CSF-1, as well as the up-regulation of genes that govern macrophage proliferation and survival (e.g. cyclin D1 and Bcl-xL), are dependent on the ability of the CSF-1R to signal from endosomes. These findings add an important and previously unrecognised spatiotemporal aspect to CSF-1R signalling that may help to explain at a molecular level how CSF-1 exerts its pleiotropic effects on macrophages. 2. Material and methods 2.1. Material Cell culture medium and supplements, foetal calf serum (FCS), precast 10% Nu-PAGE gels, SuperScript III reverse transcriptase, random primers, an Alexa Fluor 488-conjugated goat anti-rat IgG antibody, and ProLong® Gold Antifade reagent containing DAPI were from Invitrogen. The rabbit anti-CSF-1R antibody was obtained from Santa Cruz Biotechnology, while the PE-conjugated and unconjugated rat antiCSF-1R (AFS98) monoclonal antibodies, as well as the PE-conjugated rat IgG2a isotype control monoclonal antibody, were from eBioscience. The anti-phospho-Tyr809 CSF-1R, anti-phospho-Tyr723 CSF-1R, antiphospho-Thr202/Tyr204 Erk1/2, anti-phospho-Ser473 Akt, and antiphospho-Tyr705 Stat3 antibodies were from Cell Signaling Technology. The mouse monoclonal anti-Hsp90 antibody was purchased from BD Biosciences. Complete™ protease inhibitors were supplied by Roche. Dynasore was from Sigma, while GW2580, UO126 and AktVIII were purchased from Merck. Recombinant CSF-1 was a generous gift from Chiron. 2.2. Bone marrow-derived macrophages The use of mice in this study was approved by the University of Melbourne's Animal Ethics Committee (AEC Projects 0707247 and 1011748). Mouse bone marrow-derived macrophages (BMM) were obtained as previously described [28–30]. Briefly, bone marrow cells from the femurs of 8 to 12-week old C57BL/6 mice were cultured (1 × 10 6 cells/mL) for 3 days in RPMI 1640 medium supplemented with 5000 U/mL CSF-1, 10% FCS, 100 U/mL Penicillin, 100 μg/mL Streptomycin, and 2 mM GlutaMax-1™ in a humidified atmosphere of 5% CO2 at 37 °C. Non-adherent macrophage precursor cells were then collected and seeded (2 × 10 6 cells/dish) into 10 cm non-tissue culture treated dishes (Iwaki) for an additional 3 days in the presence of CSF-1 by which time a homogenous population of adherent BMM were obtained. BMM were harvested by incubation in PBS for 15–20 min at 37 °C followed by gentle pipetting. The cells were washed with growth medium, seeded (1× 105 cells/cm2) into tissue culture plates/dishes (Falcon) as required, and cultured for 24 h in the presence of CSF-1. BMM were deprived of CSF-1 for 14–16 h and then stimulated with 10,000 U/mL CSF-1. 2.3. Cell lysis and Western blotting BMM were washed twice with ice-cold phosphate-buffered saline (PBS) and then lysed (20 mM Tris–HCl [pH 7.4], 150 mM NaCl, 1 mM EDTA, 1% IGEPAL CA-630, 10% glycerol, 1 mM sodium orthovanadate, 10 mM NaF, 10 mM β-glycerophosphate, and Complete™ protease

inhibitors) on ice for 30–60 min. The cell lysates were clarified by centrifugation at 13,000 × g for 10 min at 4 °C and protein concentrations measured using a protein assay kit (Bio-Rad). Cell lysates were subjected to electrophoresis on 10% Nu-PAGE gels (MOPS running buffer) followed by Western blotting. Immunoreactive bands were visualised using ECL reagents (Millipore) and exposure to X-ray film (Fuji). Developed films were scanned using a GS800 Calibrated Imaging Densitometer (Bio-Rad) and the images exported as TIFF files. If membranes needed to be probed with additional antibodies, they were first incubated in stripping buffer (50 mM Tris–HCl [pH 7.0], 2% SDS, and 100 mM β-mercaptoethanol) for 30 min at 55 °C. 2.4. Real-time PCR Total RNA was purified using RNAeasy Mini kits (QIAGEN) and then reverse-transcribed into single-stranded cDNA using random primers and SuperScript III reverse transcriptase. Real-Time PCR was performed (in triplicate) using an Applied Biosystems Prism 7900HT sequence detection system and TaqMan assays for the following mouse genes: cyclin D1 (Mm00432359_m1), Bcl-xL (Mm00437783_m1), cMyc (Mm00487803_m1), and ccl2 (Mm00441242_m1). Changes in mRNA levels, relative to those of the endogenous control gene, hprt (Mm446968_m1), were calculated using the ΔΔCt (cycle threshold) method. 2.5. Fluorescence-activated cell sorting (FACS) analysis BMM were stimulated with CSF-1, washed with ice-cold PBS, and then harvested by gentle pipetting. Non-specific antibody binding sites were blocked using 20% normal mouse serum in PBS. The BMM were then incubated with either a PE-conjugated rat anti-CSF-1R (AFS98) monoclonal antibody or a PE-conjugated rat IgG2a isotype control monoclonal antibody for 30 min on ice [31]. The cells were washed with ice-cold PBS and CSF-1R cell-surface levels measured using a CyAn™ ADP Analyzer and Summit software (Beckman Coulter). 2.6. Immunofluorescent staining BMM, which had been seeded onto glass coverslips, were stimulated with CSF-1, fixed with 4% paraformaldehyde for 30 min at room temperature, and then blocked in 5% goat serum for 60 min at room temperature. The BMM were subsequently stained with a rat anti-CSF-1R monoclonal antibody for 60 min at room temperature followed by three washes with PBS. The cells were then probed with an Alexa Fluor 488-conjugated goat anti-rat IgG antibody for 60 min at room temperature, washed three times with PBS, and finally mounted on glass microscope slides using ProLong® Gold Antifade reagent containing DAPI. Mounted coverslips were allowed to cure for 24 h in the dark prior to fluorescent images of the cells being acquired on an Olympus FV1000 scanning confocal microscope. 2.7. Statistical analysis Data are given as the mean ± SEM, and statistical significance (p-values) evaluated using the Student's t test (GraphPad Prism). 3. Results 3.1. Differential effects of temperature on CSF-1-induced signalling Emerging evidence suggests that RTK signalling from endosomes may be critical in dictating the effects of specific growth factors on cells [23–27]. Accordingly, an ability by the CSF-1R to continue to signal from endosomes following its CSF-1-induced internalisation could help to explain how CSF-1 exerts its pleiotropic effects on macrophages. In order to begin to address this important question we took advantage

J. Huynh et al. / Cellular Signalling 24 (2012) 1753–1761

of the fact that the stimulation of cells with growth factors at reduced temperatures (e.g. 4 °C) can result in RTK activation, including that of the CSF-1R [32], but with the internalisation of the RTK being prevented or greatly impaired [33–36]. Cell-surface levels of the CSF-1R were reduced by >70% within 5 min of CSF-1 stimulation at 37 °C; a >95% reduction in CSF-1R cell-surface levels occurred by 30 min (Fig. 1A). As shown in Fig. 1B, the rapid down-regulation of cell-surface CSF-1R levels at 37 °C was accompanied by a marked decline in total CSF-1R protein levels as a result of the lysosomal-mediated degradation of the CSF-1R [35]. Although CSF-1 triggered the down-regulation of cellsurface CSF-1R levels at 4 °C, the rate was slower than that at 37 °C (Fig. 1A). By 5 min post-CSF-1 stimulation at 4 °C, CSF-1R cell-surface levels had only declined by 35%, with around 80% of cell-surface CSF1R down-regulated after 30 min (Fig. 1A). In contrast to the situation at 37 °C, no decreases in total CSF-1R protein levels were observed for at least the first 2–4 h following CSF-1 stimulation at 4 °C (Fig. 1B). Together, these findings suggest that while CSF-1 induces the internalisation of the CSF-1R at 4 °C, albeit more slowly than at 37 °C, the internalised CSF-1R likely remains associated with the plasma membrane rather than being trafficked to lysosomes via the endocytic pathway. The stimulation of BMM with CSF-1 at 37 °C resulted in the robust activation of the CSF-1R as judged by the phosphorylation of Tyr809 and Tyr723 in the catalytic loop and kinase-insert domain, respectively, of the CSF-1R (Fig. 1B and data not shown). Maximal CSF-1-induced

1755

activation of the CSF-1R at 37 °C occurred within 1–5 min of CSF-1 stimulation and declined thereafter (Fig. 1B). In contrast, CSF-1induced activation of the CSF-1R occurred more slowly at 4 °C; it was also more sustained at this temperature (Fig. 1B). The observed effects of reduced temperature on CSF-1R activation were consistent with a previous report [32]. Notably though, neither Erk1/2 nor Akt activation was detected when BMM were stimulated with CSF-1 at 4 °C (Fig. 1B). This contrasted with the strong CSF-1-induced activation of Erk1/2 and Akt at 37 °C. In the case of Stat3, its CSF-1-induced activation occurred at both 37 °C and 4 °C (Fig. 1B). However, like that for the CSF-1R, CSF1-induced Stat3 activation occurred more slowly at 4 °C and was more sustained. As shown in Fig. 1C, Erk1/2 and Akt were rapidly activated when BMM that had initially been stimulated with CSF-1 for 2 h at 4 °C were subsequently shifted to 37 °C. Taken together, our findings are consistent with the entry of the activated CSF-1R into the endocytic pathway being required for the optimal activation of select signalling proteins (e.g. Erk1/2 and Akt). 3.2. Dynasore impairs CSF-1R signalling and the up-regulation of CSF-1 target genes In light of the above findings we also made use of the cell-permeable dynamin inhibitor, Dynasore, to establish the importance of the entry of the activated CSF-1R into the endocytic pathway for downstream signalling and gene expression responses to CSF-1. Dynasore inhibits

Fig. 1. Effects of temperature on CSF-1-induced signalling. (A–B) CSF-1-deprived BMM were incubated at 4 °C for 45 min or kept at 37 °C; they were then stimulated with CSF-1 at 4 °C or 37 °C. (A) CSF-1R cell-surface levels were measured by FACS analysis. (B) Cell lysates were subjected to Western blotting with anti-CSF-1R, anti-phospho-Y809 CSF-1R, antiphospho-Erk1/2, anti-phospho-Akt, and anti-phospho-Stat3 antibodies. The membranes were also probed with an anti-Hsp90 antibody to determine protein loading levels. (C) CSF-1-deprived BMM were incubated at 4 °C for 45 min and then stimulated with CSF-1 for 2 h at 4 °C or left unstimulated. The BMM were subsequently shifted to 37 °C, followed by Western blotting. Data are representative of three independent experiments.

1756

J. Huynh et al. / Cellular Signalling 24 (2012) 1753–1761

the dynamin-mediated release of plasma membrane-bound intracellular vesicles into the endocytic pathway [37]. Treatment of BMM with Dynasore impaired a number of CSF-1-induced signalling events (Fig. 2A). At 1 min post-CSF-1 stimulation, Erk1/2, Akt and Stat3 activity were lower in Dynasore-treated BMM. Although the levels of Erk1/2 and Akt activity in Dynasore- and vehicle-treated BMM were comparable at 5–10 min post-CSF-1 stimulation, Erk1/2 and Akt activity were less sustained in Dynasore-treated BMM (Fig. 2A). In the case of Stat3, both the magnitude and duration of its CSF-1-induced activation were impaired in BMM that had been treated with Dynasore (Fig. 2A). In order to examine the consequences of Dynasore on CSF-1 signalling further the effects of the dynamin inhibitor on the induction of a number of CSF-1-target genes, namely those that regulate macrophage proliferation, survival and chemotaxis (cyclin D1, Bcl-xL, c-Myc

and ccl2) [38–41], were examined (Fig. 2B–E). The CSF-1-induced up-regulation of cyclin D1 gene expression was strongly inhibited in Dynasore-treated BMM (Fig. 2B). CSF-1-induced up-regulation of Bcl-xL, c-Myc and ccl2 gene expression were significantly impaired at early time-points (e.g. 2 h) post-CSF-1 stimulation in Dynasore-treated BMM (Fig. 2C–E). At later time-points, however, Bcl-xL, c-Myc and ccl2 mRNA levels in Dynasore- and vehicle-treated BMM were largely comparable (Fig. 2C–E). These data again point to an important link between the internalisation of the CSF-1R and optimal signalling responses to CSF-1.

3.3. Inhibition of the CSF-1R following its internalisation abrogates CSF-1-induced signalling and gene expression The importance of endosomal signalling by the RTK, c-Met, for the sustained activation of Erk1/2 and Stat3 in response to hepatocyte growth factor (HGF) in HeLa cells was recently demonstrated by pharmacologically inhibiting c-Met following its internalisation and entry into the endocytic pathway [23]. Here, we used a comparable approach to further establish if post-internalisation signalling by the CSF-1R is required for the temporal propagation of Erk1/2, Akt and/or Stat3 activity. GW2580 is a potent and highly specific small molecule inhibitor of the CSF-1R [42], being inactive against more than 150 other kinases, including Erk1/2 and Akt [43]. As shown in Fig. 3A, GW2580 inhibited the CSF-1-induced activation of the CSF-1R in BMM; it also inhibited CSF-1-induced Erk1/2, Akt and Stat3 activation. The addition of GW2580 as little as 1 min prior to CSF-1 stimulation was sufficient to inhibit CSF-1R, Erk1/2, Akt and Stat3 activation (data not shown), confirming the potency of GW2580 towards the CSF-1R. The ability of GW2580 to inhibit the up-regulation of specific genes (i.e. cyclin D1, Bcl-xL, c-Myc and ccl2) by CSF-1 was also confirmed (Fig. 3B–E).

Fig. 2. Effects of Dynasore on CSF-1-induced signalling and gene expression (A) CSF-1deprived BMM were washed with serum-free medium and then incubated with 100 μM Dynasore or 0.1% DMSO (vehicle) for 30 min. The BMM were subsequently stimulated with CSF-1, followed by Western blotting with anti-CSF-1R, anti-phospho-Y809 CSF-1R, anti-phospho-Erk1/2, anti-phospho-Akt, and anti-phospho-Stat3 antibodies. The membranes were also probed with an anti-Hsp90 antibody to determine protein loading levels. Data are representative of three independent experiments. (B-E) CSF-1-deprived BMM were washed with serum-free medium, incubated with 100 μM Dynasore or 0.1% DMSO for 30 min, and then stimulated with CSF-1. Real-Time PCR was used to measure changes in: (B) cyclin D1, (C) Bcl-xL, (D) c-Myc, and (E) ccl2 mRNA levels relative to unstimulated BMM. Data from four independent experiments are presented as the mean± SEM (** = p b 0.01; *= p b 0.05; ns = not significant).

Fig. 3. Effects of GW2580 on CSF-1-induced signalling and gene expression. (A) CSF-1deprived BMM were incubated with 5 μM GW2580 or 0.1% DMSO (vehicle) for 15 min, stimulated with CSF-1 for 5 min, and then subjected to Western blotting with anti-CSF1R, anti-phospho-Y809 CSF-1R, anti-phospho-Erk1/2, anti-phospho-Akt, and antiphospho-Stat3 antibodies. The membranes were also probed with an anti-Hsp90 antibody to determine protein loading levels. Data are representative of two independent experiments. (B-E) CSF-1-deprived BMM were incubated with 5 μM GW2580 or 0.1% DMSO for 15 min and then stimulated with CSF-1 for 4 h. Real-Time PCR was used to measure changes in: (B) cyclin D1, (C) Bcl-xL, (D) c-Myc, and (E) ccl2 mRNA levels relative to unstimulated BMM. Data from three independent experiments are presented as the mean ± SEM (** = p b 0.01; * = p b 0.05).

J. Huynh et al. / Cellular Signalling 24 (2012) 1753–1761

Fluorescence-activated cell sorting (FACS) analysis was then used to determine the earliest time-point at which cell-surface CSF-1R levels are down-regulated post-CSF-1 stimulation. Treating CSF-1stimulated BMM with GW2580 at this time-point should prevent on-going signalling by the internalised CSF-1R without impairing the initial activation of the CSF-1R at the cell-surface. FACS analysis indicated that 90–95% of cell-surface CSF-1R had been internalised within 15 min of CSF-1 stimulation (Fig. 4A). The CSF-1-induced internalisation of the CSF-1R was also assessed by immunofluorescencebased microscopy (Fig. 4B). The majority of the CSF-1R was localised to the cell periphery in CSF-1-deprived BMM, although some diffuse cytoplasmic CSF-1R staining was also apparent (Fig. 4B). By 5 min postCSF-1 stimulation a substantial proportion of the CSF-1R exhibited a more vesicular form of staining with a corresponding decrease in CSF-1R staining at the cell periphery. The majority of the CSF-1R appeared to be localised to relatively large vesicular bodies, which primarily adopted a perinuclear subcellular localisation, by 10–15 min post-CSF-1 stimulation (Fig. 4B). The observed changes in CSF-1R subcellular localisation in response to CSF-1 stimulation were consistent with an earlier report using the CSF-1-dependent mouse macrophage cell line, BAC1.2F5 [36]. If the CSF-1R only signals from the cell-surface, then GW2580 addition 15 min post-CSF-1 stimulation would not be expected to perturb

Fig. 4. Kinetics of CSF-1-induced internalisation of the CSF-1R. CSF-1-deprived BMM were stimulated with CSF-1 for the indicated times. (A) CSF-1R cell-surface levels were measured by FACS analysis. Data are representative of four independent experiments. (B) The BMM were fixed, permeabilised and stained with a rat anti-CSF-1R monoclonal antibody (red staining); nuclei were stained with DAPI (blue staining). Immunofluorescent images of the cells were acquired using a confocal microscope. Data are representative of three independent experiments.

1757

downstream signalling events. However, if the activity of specific signalling proteins were less sustained following GW2580 addition then this would suggest that these proteins are to some extent dependent on the internalised CSF-1R signalling from endosomes for maintenance of their activity. GW2580 addition 15 min post-CSF-1 stimulation resulted in the rapid loss of CSF-1R tyrosine phosphorylation (Fig. 5A), a finding which was consistent with the inhibition of the activated CSF-1R by GW2580. Notably, inhibition of the CSF-1R following its internalisation resulted in less sustained Erk1/2 and Akt activity, with the effects on Erk1/2 activity greater than those for Akt (Fig. 5A). In contrast to the adverse effects on the duration of Erk1/2 and Akt activity, GW2580 addition following the internalisation of the CSF-1R did not appear to affect the magnitude and/or duration of Stat3 activity (Fig. 5A).

Fig. 5. Effects of inhibiting the CSF-1R following its internalisation on CSF-1-induced signalling and gene expression. (A) CSF-1-deprived BMM were stimulated with CSF1, followed by addition of 5 μM GW2580 or 0.1% DMSO (vehicle) 15 min post-CSF-1 stimulation. The BMM were subsequently subjected to Western blotting with antiCSF-1R, anti-phospho-Y809 CSF-1R, anti-phospho-Erk1/2, anti-phospho-Akt, and anti-phospho-Stat3 antibodies. The membranes were also probed with an anti-Hsp90 antibody to determine protein loading levels. Data are representative of four independent experiments. (B-E) CSF-1-deprived BMM were stimulated with CSF-1, followed by addition of 5 μM GW2580 or 0.1% DMSO 15 min post-CSF-1 stimulation. Real-Time PCR was used to measure changes in: (B) cyclin D1, (C) Bcl-xL, (D) c-Myc, and (E) ccl2 mRNA levels relative to unstimulated BMM. Data from four independent experiments are presented as the mean±SEM (**=pb 0.01; *=pb 0.05; ns = not significant).

1758

J. Huynh et al. / Cellular Signalling 24 (2012) 1753–1761

The consequences of inhibiting the CSF-1R following its internalisation were also examined at the level of gene expression (Fig. 5B–E). CSF-1induced up-regulation of cyclin D1 gene expression was inhibited in BMM that had been treated with GW2580 following the internalisation of the CSF-1R (Fig. 5B). Likewise, up-regulation of Bcl-xL, c-Myc and ccl2 gene expression was also significantly abrogated when the CSF-1R was inhibited with GW2580 following its CSF-1-induced internalisation (Fig. 5C–E). 3.4. Inhibition of Erk1/2 and Akt following the internalisation of the CSF-1R has different effects on CSF-1 target gene expression Given that the duration of CSF-1-induced Erk1/2 and Akt activity was dependent on the CSF-1R continuing to signal following its internalisation (Fig. 5A), the effects of inhibiting Erk1/2 or Akt activity following the internalisation of the CSF-1R on the up-regulation of target gene expression by CSF-1 were examined. However, the effects of inhibiting Erk1/2 and Akt prior to the activation of cell-surface CSF-1R on the induction of CSF-1 target genes were first assessed (Fig. 6). Because MEK mediates the activation of Erk1/2 in response to CSF-1 stimulation [14], Erk1/2 was indirectly inhibited with the MEK inhibitor, UO126 [44]. Akt was directly inhibited with the small chemical inhibitor, AktVIII [45]. Treatment of BMM with UO126 prior to CSF-1 stimulation inhibited Erk1/2 activation without affecting the activation of the CSF-1R, Akt or Stat3 (Fig. 6A). Similarly, AktVIII inhibited the CSF-1-induced activation of Akt but not CSF-1R, Erk1/2 or Stat3 activation (Fig. 6B). The effects of UO126 and AktVIII on the up-regulation of target genes by CSF-1 were then tested (Fig. 6C–F). Induction of cyclin D1 gene expression by CSF-1 was reduced by approximately 70% in BMM that had been treated with UO126 prior to CSF-1 stimulation when compared to vehicle-treated BMM (Fig. 6C). Treatment of BMM with AktVIII inhibited the CSF-1-induced up-regulation of cyclin D1 gene induction by around 50% (Fig. 6C). UO126 moderately inhibited the induction of Bcl-xL gene expression by CSF-1 (Fig. 6D). However, AktVIII enhanced the CSF-1-induced up-regulation of Bcl-xL gene expression (Fig. 6D). The effects of UO126 and AktVIII on CSF-1-induced c-Myc gene expression were largely comparable with those exerted by the inhibitors on cyclin D1 gene expression (Fig. 6C and E). Treatment of BMM with UO126 inhibited CSF-1-induced ccl2 gene expression by approximately 75% (Fig. 6F), whereas the induction of ccl2 gene expression by CSF-1 was not affected by the Akt inhibitor (Fig. 6F). The ability of UO126 and AktVIII to inhibit Erk1/2 and Akt following their CSF-1-induced activation was next assessed. As can be seen in Fig. 7A, the addition of UO126 following the internalisation of the CSF-1R (i.e. 15 min post-CSF-1 stimulation) resulted in a rapid loss in Erk1/2 activity. Importantly, UO126 had no significant effects on the magnitude and/or duration of CSF-1-induced CSF-1R, Akt or Stat3 activation (data not shown). Addition of AktVIII 15 min post-CSF-1 stimulation resulted in a rapid loss in Akt activity (Fig. 7B). However, addition of the Akt inhibitor at this time-point did not alter the magnitude and/or kinetics of CSF-1-induced CSF-1R, Erk1/2 or Stat3 activation (data not shown). As shown in Fig. 7, adding UO126 or AktVIII following the internalisation of the CSF-1R had largely the same effects on CSF-1-induced up-regulation of cyclin D1, Bcl-xL, c-Myc and ccl2 gene expression as when the inhibitors were added prior to the activation of cell-surfacelocalised CSF-1R (Fig. 6C–F). Specifically, inhibition of Erk1/2 activity post-CSF-1R internalisation abrogated the CSF-1-induced up-regulation of cyclin D1, Bcl-xL, c-Myc and ccl2 gene expression, although the effects on Bcl-xL and c-Myc gene expression were less than those for cyclin D1 and ccl2 (Fig. 7C–F). Notably, the inhibitory effects of UO126 on the upregulation of cyclin D1, Bcl-xL, c-Myc and ccl2 gene expression were not as great as those exerted by GW2580 (Fig. 7C–F). Inhibition of Akt activity following the internalisation of the activated CSF-1R partially blocked the CSF-1-induced up-regulation of cyclin D1 and c-Myc gene expression (Fig. 7C and E), whereas Bcl-xL gene expression was

Fig. 6. Effects of UO126 and AktVIII on CSF-1-induced signalling and gene expression. (A–B) CSF-1-deprived BMM were incubated with (A) 10 μM UO126 or (B) 10 μM AktVIII for 15 min, stimulated with CSF-1 for 5 min, and then subjected to Western blotting with anti-CSF-1R, anti-phospho-Y809 CSF-1R, anti-phospho-Erk1/2, antiphospho-Akt, and anti-phospho-Stat3 antibodies. The membranes were also probed with an anti-Hsp90 antibody to determine protein loading levels. Data are representative of two independent experiments. (C-F) CSF-1-deprived BMM were incubated with 0.1% DMSO, 10 μM UO126 or 10 μM AktVIII for 15 min, and then stimulated with CSF-1 for 4 h. Real-Time PCR was used to measure: (C) cyclin D1, (D) Bcl-xL, (E) c-Myc, and (F) ccl2 mRNA levels relative to those in BMM that had been treated with DMSO prior to CSF-1 stimulation. Data from three independent experiments are presented as the mean ± SEM (** = p b 0.01; * = p b 0.05; ns = not significant).

potentiated (Fig. 7D). The up-regulation of ccl2 gene expression by CSF-1 was unaffected when Akt activity was inhibited post-CSF-1R internalisation (Fig. 7F). 4. Discussion The CSF-1R plays a critical role in regulating the development of mature macrophages from myeloid progenitor cells [46]. It is generally assumed that the CSF-1R activates signalling pathways exclusively at

J. Huynh et al. / Cellular Signalling 24 (2012) 1753–1761

Fig. 7. Effects of inhibiting Erk1/2 and Akt activity following the internalisation of the CSF-1R on CSF-1-induced gene expression. (A–B) CSF-1-deprived BMM were stimulated with CSF-1, followed by addition of 10 μM UO126 (A) or 10 μM AktVIII (B) 15 min postCSF-1 stimulation. The BMM were subsequently subjected to Western blotting with anti-phospho-Erk1/2 and anti-phospho-Akt antibodies. The membranes were also probed with an anti-Hsp90 antibody to determine protein loading levels. Data are representative of three independent experiments. (C–F) CSF-1-deprived BMM were stimulated with CSF1 for 15 min, followed by addition of 0.1% DMSO, 10 μM UO126, 10 μM AktVIII, or 5 μM GW2580 15 min post-CSF-1 stimulation. Real-Time PCR was used to measure: (C) cyclin D1, (D) Bcl-xL, (E) c-Myc, and (F) ccl2 mRNA levels relative to those in BMM that had been treated with DMSO post-CSF-1 stimulation. Data from three independent experiments are presented as the mean ± SEM (** = p b 0.01; * = p b 0.05; ns = not significant).

the plasma membrane and the subsequent internalisation of the CSF-1R, together with its ligand, simply serves to prevent on-going signalling by facilitating the endocytic trafficking and lysosomal degradation of the

1759

CSF-1R [13,14]. Here, a combination of approaches was used to determine if the CSF-1R may in fact continue to signal following its internalisation, as has been suggested for several other RTKs [22–27]. These approaches included: (i) stimulating macrophages at 4 °C in order to prevent/perturb the internalisation and endocytic trafficking of the CSF-1R, (ii) blocking the entry of the CSF-1R into the endocytic pathway with the dynamin inhibitor, Dynasore, and (iii) pharmacologic inhibition of the CSF-1R following its internalisation. Our data suggest that the sustained activation of key signalling proteins and the ensuing up-regulation of important target genes by CSF-1 are dependent on the ability of the CSF-1R to signal from endosomes following its internalisation. CSF-1 failed to trigger the activation of Erk1/2 and Akt when BMM were stimulated at 4 °C. At the same time, CSF-1-induced tyrosine phosphorylation of the CSF-1R was robust and prolonged, albeit initially delayed, which was in line with a previous study [32]. Thus, the failure of CSF-1 to activate Erk1/2 and Akt at 4 °C does not appear to have been related to the activation of the CSF-1R per se. Indeed, Stat3 activation in response to CSF-1 stimulation was more sustained at 4 °C. These observations suggest that the activation of Erk1/2 and Akt by the CSF-1R may not occur at the cell-surface but instead occur following the CSF-1-induced internalisation of the CSF-1R. This conclusion is consistent with reports which have suggested that the internalisation of other RTKs (e.g. EGFR and RET) is required for optimal Erk1/2 activation [47,48]. However, the stimulation of BMM at 4 °C with CSF-1 appeared to delay rather than prevent the internalisation of the CSF-1R. Thus, we cannot exclude the possibility that the absence of Erk1/2 and Akt activation could also potentially be explained by the failure of proteins, which are required for the activation of Erk1/2 and Akt, to be recruited to the CSF-1R. In this regard, recent evidence suggests that some CSF-1-induced signalling events may be regulated by the CSF-1R from lipid rafts [49]. A role for lipid rafts in regulating EGFR-mediated Erk1/2 and Akt activation has been reported [50]. Lipid rafts are cholesterol- and sphingolipid-rich regions in cell membranes that can function as signalling platforms through their sequestration or exclusion of specific signalling proteins [51]. Because the coalescing of dynamic, nanoscale lipid rafts into larger more stable rafts is impaired at lower temperatures [52,53], the observed changes in CSF-1R signalling at 4 °C may be due in part to effects on lipid raft formation. Several recent studies have used Dynasore to demonstrate that select signalling responses to growth factors (e.g. HGF and VEGF) are dependent on endosomal signalling by their cognate RTK [23,24,26]. Despite the modest inhibitory effect of Dynasore on the CSF-1-induced degradation of the CSF-1R, suggesting that Dynasore was not overly effective at inhibiting the entry of the CSF-1R into the endocytic pathway, CSF-1-induced Erk1/2, Akt and Stat3 activation were nonetheless perturbed. That Erk1/2 and Akt activation were less sustained in Dynasore-treated BMM is consistent with CSF-1R signalling from endosomes being necessary for the optimal activation of Erk1/2 and Akt by CSF-1. The marked inhibition of CSF-1-induced Stat3 activation by Dynasore was somewhat surprising given that Stat3 was robustly activated, notwithstanding its delayed activation, when the internalisation of the CSF-1R was perturbed by stimulating BMM at 4 °C rather than by Dynasore treatment. However, our results with Dynasore are consistent with a recent report which concluded that endosomal signalling by c-Met was required for optimal Stat3, as well as Erk1/2, activation [23]. Similarly, Dynasore was shown to impair VEGF-induced Akt activation through its inhibitory effects on VEGFR2 endocytosis [24]. An important regulatory link between CSF-1-induced signalling and the endocytic trafficking of the CSF-1R was also apparent from the inhibitory effects of Dynasore on CSF-1 target gene expression. Dynasore primarily abrogated the ‘early’ up-regulation of Bcl-xL, cMyc and ccl2 gene expression by CSF-1. Such an outcome is consistent with Dynasore having exerted a partial inhibitory effect on the entry of the CSF-1R into the endocytic pathway. The more pronounced inhibitory effect of Dynasore on CSF-1-induced cyclin D1 gene

1760

J. Huynh et al. / Cellular Signalling 24 (2012) 1753–1761

expression is potentially explained by the marked impact of Dynasore on the activation of Stat3, a key transcriptional regulator of cyclin D1 expression [54]. Given the differences in the magnitudes of the inhibitory effects of Dynasore on CSF-1-induced CSF-1R endocytosis and Stat3 activation it remains possible that Dynasore also impairs CSF-1-stimulated Stat3 activation independently of its inhibitory effects on dynamin. Further evidence that select responses to CSF-1 are dependent on the CSF-1R signalling from endosomes came from our studies with the CSF-1R inhibitor, GW2580. Specifically, the duration of Erk1/2 and Akt activity was shown to be dependent on the CSF-1R continuing to signal following its internalisation. Indeed, that the inhibitory effects of GW2580 on the up-regulation of CSF-1 target genes (e.g. cyclin D1) were comparable irrespective of whether the CSF-1R inhibitor was added prior to CSF-1 stimulation or following the internalisation of the CSF-1R argues as to the importance of CSF-1R signalling from both the plasma membrane and endosomes. Such a model for CSF-1R signalling is also supported by the demonstration that pharmacologic inhibition of Erk1/2 activity had essentially the same inhibitory effects on CSF-1induced gene expression regardless of whether the inhibitor was added prior to CSF-1 stimulation or subsequent to CSF-1R internalisation. Likewise, inhibition of Akt either before or after the internalisation of the CSF-1R had comparable effects on CSF-1 target gene expression. In the case of Stat3, the duration of its CSF-1-induced activation does not appear to be dependent on post-internalisation signalling by the CSF-1R. This contrasts with the situation in HeLa cells where the sustained activation of Stat3 in response to HGF was shown to be dependent on c-Met signalling from endosomes [23]. Therefore, the dependence on RTKs signalling from endosomes for the propagation of particular signalling responses, such as Erk1/2 and Stat3 activity, may be cell context and/or RTK-specific. As has previously been shown [55], the intensity and duration of a particular signal (e.g. Erk1/2 activity) can be critical in dictating the cellular response elicited by a growth factor (e.g. proliferation versus differentiation). Accordingly, a spatiotemporal aspect to CSF-1R signalling, as suggested by our data, could help to explain how secreted and membrane-bound CSF-1 are able to exert different, yet overlapping, effects on macrophages [10–12,56]. For example, membrane-bound CSF-1 may be capable of inducing more sustained activation of signalling proteins that are exclusively activated at the plasma membrane if, as might be expected, the internalisation of the CSF-1R occurs more slowly when the receptor is activated by membrane-bound CSF-1. This appears to be the case for the activation of c-Kit by membranebound stem cell factor [57]. However, signalling responses that require CSF-1R signalling from endosomes for their spatiotemporal propagation (e.g. Erk1/2 activity) may be less sustained than following stimulation with secreted CSF-1 and hence result in a biological response/s different to that elicited by secreted CSF-1. It has been suggested that some receptors may also be capable of activating specific signalling proteins exclusively at endosomes, thereby providing a mechanism for greater diversification of downstream signalling [17,19–21]. Whether this is true for the CSF-1R has not been established. Nonetheless, the ability of the CSF-1R to signal from both the plasma membrane and endosomes could provide the level of ‘signalling dexterity’ necessary for CSF-1 to mediate its pleiotropic actions on macrophages (e.g. regulation of macrophage survival, proliferation, differentiation, migration, and immune/homeostatic functions). Understanding the spatiotemporal aspects of CSF-1R signalling could be clinically important as it may enable, through the development of ‘endosome-specific targeting’ strategies [20], selective targeting of particular CSF-1R-regulated signalling pathways while leaving other pathways intact. 5. Conclusions In summary, the effects of endocytosis on signalling by the CSF-1R are more complex than previously recognised. Our data suggest that the internalisation of the CSF-1R is necessary for the optimal activation

of select signalling pathways (e.g. Erk1/2 pathway) by CSF-1. Moreover, on-going signalling by the CSF-1R following its internalisation appears to be required for the sustained activation of the signalling pathways that control macrophage proliferation and survival through their regulation of such genes as cyclin D1 and Bcl-xL. Thus, key signalling responses to CSF-1 depend on the ability of the CSF-1R to signal from endosomes. Conflict of interest The authors declare no conflict of interest. Author contributions GS, JH and JAH conceived and designed the experiments. JH, GS, MK and AC performed the experiments. GS, together with JH and JAH, wrote the paper. All authors have approved the manuscript. Acknowledgements JH was supported by an Australian Postgraduate Award (Australian Government), MK was supported by a Melbourne International Research Scholarship (University of Melbourne), and JAH was supported by a Senior Principal Research Fellowship (NHMRC). References [1] D.A. Hume, Current Opinion in Immunology 18 (2006) 49–53. [2] C.A. Janeway Jr., R. Medzhitov, Annual Review of Immunology 20 (2002) 197–216. [3] P.J. Murray, T.A. Wynn, Nature Reviews Immunology 11 (2011) 723–737. [4] J.W. Pollard, Nature Reviews Immunology 9 (2009) 259–270. [5] J. Cohen, Nature 420 (2002) 885–891. [6] R.W. Kinne, R. Brauer, B. Stuhlmuller, E. Palombo-Kinne, G.R. Burmester, Arthritis Research 2 (2000) 189–202. [7] Y.R. Mahida, Inflammatory Bowel Diseases 6 (2000) 21–33. [8] J.W. Pollard, Nature Reviews Cancer 4 (2004) 71–78. [9] V. Chitu, E.R. Stanley, Current Opinion in Immunology 18 (2006) 39–48. [10] X.M. Dai, X.H. Zong, V. Sylvestre, E.R. Stanley, Blood 103 (2004) 1114–1123. [11] S. Nandi, M.P. Akhter, M.F. Seifert, X.M. Dai, E.R. Stanley, Blood 107 (2006) 786–795. [12] G.R. Ryan, X.M. Dai, M.G. Dominguez, W. Tong, F. Chuan, O. Chisholm, R.G. Russell, J.W. Pollard, E.R. Stanley, Blood 98 (2001) 74–84. [13] J.A. Hamilton, Journal of Leukocyte Biology 62 (1997) 145–155. [14] F.J. Pixley, E.R. Stanley, Trends in Cell Biology 14 (2004) 628–638. [15] A. Sorkin, C.M. Waters, Bioessays 15 (1993) 375–382. [16] B.P. Ceresa, S.L. Schmid, Current Opinion in Cell Biology 12 (2000) 204–210. [17] D. Hoeller, S. Volarevic, I. Dikic, Current Opinion in Cell Biology 17 (2005) 107–111. [18] B.N. Kholodenko, Trends in Cell Biology 12 (2002) 173–177. [19] M. Miaczynska, L. Pelkmans, M. Zerial, Current Opinion in Cell Biology 16 (2004) 400–406. [20] J.E. Murphy, B.E. Padilla, B. Hasdemir, G.S. Cottrell, N.W. Bunnett, Proceedings of the National Academy of Sciences of the United States of America 106 (2009) 17615–17622. [21] M. von Zastrow, A. Sorkin, Current Opinion in Cell Biology 19 (2007) 436–445. [22] C. Joffre, R. Barrow, L. Menard, V. Calleja, I.R. Hart, S. Kermorgant, Nature Cell Biology 13 (2011) 827–837. [23] S. Kermorgant, P.J. Parker, The Journal of Cell Biology 182 (2008) 855–863. [24] S. Sawamiphak, S. Seidel, C.L. Essmann, G.A. Wilkinson, M.E. Pitulescu, T. Acker, A. Acker-Palmer, Nature 465 (2010) 487–491. [25] N. Taub, D. Teis, H.L. Ebner, M.W. Hess, L.A. Huber, Molecular Biology of the Cell 18 (2007) 4698–4710. [26] Y. Wang, M. Nakayama, M.E. Pitulescu, T.S. Schmidt, M.L. Bochenek, A. Sakakibara, S. Adams, A. Davy, U. Deutsch, U. Luthi, A. Barberis, L.E. Benjamin, T. Makinen, C.D. Nobes, R.H. Adams, Nature 465 (2010) 483–486. [27] Y. Wang, S. Pennock, X. Chen, Z. Wang, Molecular and Cellular Biology 22 (2002) 7279–7290. [28] A. Achuthan, P. Masendycz, J.A. Lopez, T. Nguyen, D.E. James, M.J. Sweet, J.A. Hamilton, G.M. Scholz, Molecular and Cellular Biology 28 (2008) 6149–6159. [29] D. de Nardo, P. Masendycz, S. Ho, M. Cross, A.J. Fleetwood, E.C. Reynolds, J.A. Hamilton, G.M. Scholz, The Journal of Biological Chemistry 280 (2005) 9813–9822. [30] G.A. Manes, P. Masendycz, T. Nguyen, A. Achuthan, H. Dinh, J.A. Hamilton, G.M. Scholz, The FEBS Journal 273 (2006) 1791–1804. [31] D. de Nardo, C.M. de Nardo, T. Nguyen, J.A. Hamilton, G.M. Scholz, Journal of Immunology 183 (183) (2009) 8110–8118.

J. Huynh et al. / Cellular Signalling 24 (2012) 1753–1761 [32] A. Sengupta, W.K. Liu, Y.G. Yeung, D.C. Yeung, A.R. Frackelton Jr., E.R. Stanley, Proceedings of the National Academy of Sciences of the United States of America 85 (1988) 8062–8066. [33] J.C. Chow, G. Condorelli, R.J. Smith, The Journal of Biological Chemistry 273 (1998) 4672–4680. [34] A.R. Frackelton Jr., P.M. Tremble, L.T. Williams, The Journal of Biological Chemistry 259 (1984) 7909–7915. [35] L.J. Guilbert, E.R. Stanley, The Journal of Biological Chemistry 261 (1986) 4024–4032. [36] J. Murray, L. Wilson, S. Kellie, Journal of Cell Science 113 (2000) 337–348. [37] E. Macia, M. Ehrlich, R. Massol, E. Boucrot, C. Brunner, T. Kirchhausen, Developmental Cell 10 (2006) 839–850. [38] A. Orlofsky, E.R. Stanley, The EMBO Journal 6 (1987) 2947–2952. [39] L. Sevilla, C. Aperlo, V. Dulic, J.C. Chambard, C. Boutonnet, O. Pasquier, P. Pognonec, K.E. Boulukos, Molecular and Cellular Biology 19 (1999) 2624–2634. [40] C.J. Sherr, H. Matsushime, M.F. Roussel, Ciba Foundation Symposium 170 (1992) 209–219. [41] Y.J. Shyy, L.L. Wickham, J.P. Hagan, H.J. Hsieh, Y.L. Hu, S.H. Telian, A.J. Valente, K.L. Sung, S. Chien, The Journal of Clinical Investigation 92 (1993) 1745–1751. [42] J.G. Conway, B. McDonald, J. Parham, B. Keith, D.W. Rusnak, E. Shaw, M. Jansen, P. Lin, A. Payne, R.M. Crosby, J.H. Johnson, L. Frick, M.H. Lin, S. Depee, S. Tadepalli, B. Votta, I. James, K. Fuller, T.J. Chambers, F.C. Kull, S.D. Chamberlain, J.T. Hutchins, Proceedings of the National Academy of Sciences of the United States of America 102 (2005) 16078–16083. [43] J.G. Conway, H. Pink, M.L. Bergquist, B. Han, S. Depee, S. Tadepalli, P. Lin, R.C. Crumrine, J. Binz, R.L. Clark, J.L. Selph, S.A. Stimpson, J.T. Hutchins, S.D. Chamberlain, T.A. Brodie, The Journal of Pharmacology and Experimental Therapeutics 326 (2008) 41–50. [44] J.V. Duncia, J.B. Santella III, C.A. Higley, W.J. Pitts, J. Wityak, W.E. Frietze, F.W. Rankin, J.H. Sun, R.A. Earl, A.C. Tabaka, C.A. Teleha, K.F. Blom, M.F. Favata, E.J.

[45]

[46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57]

1761

Manos, A.J. Daulerio, D.A. Stradley, K. Horiuchi, R.A. Copeland, P.A. Scherle, J.M. Trzaskos, R.L. Magolda, G.L. Trainor, R.R. Wexler, F.W. Hobbs, R.E. Olson, Bioorganic & Medicinal Chemistry Letters 8 (1998) 2839–2844. C.W. Lindsley, Z. Zhao, W.H. Leister, R.G. Robinson, S.F. Barnett, D. Defeo-Jones, R.E. Jones, G.D. Hartman, J.R. Huff, H.E. Huber, M.E. Duggan, Bioorganic & Medicinal Chemistry Letters 15 (2005) 761–764. X.M. Dai, G.R. Ryan, A.J. Hapel, M.G. Dominguez, R.G. Russell, S. Kapp, V. Sylvestre, E.R. Stanley, Blood 99 (2002) 111–120. D.S. Richardson, A.Z. Lai, L.M. Mulligan, Oncogene 25 (2006) 3206–3211. A.V. Vieira, C. Lamaze, S.L. Schmid, Science 274 (1996) 2086–2089. H.J. Kim, W. Zou, Y. Ito, S.Y. Kim, J. Chappel, F.P. Ross, S.L. Teitelbaum, Journal of Cellular Biochemistry 110 (2010) 201–209. H.Y. Oh, E.J. Lee, S. Yoon, B.H. Chung, K.S. Cho, S.J. Hong, The Prostate 67 (2007) 1061–1069. K. Simons, D. Toomre, Nature Reviews Molecular Cell Biology 1 (2000) 31–39. A.J. Garcia-Saez, S. Chiantia, P. Schwille, The Journal of Biological Chemistry 282 (2007) 33537–33544. K. Weise, G. Triola, S. Janosch, H. Waldmann, R. Winter, Biochimica et Biophysica Acta 1798 (2010) 1409–1417. J.F. Bromberg, M.H. Wrzeszczynska, G. Devgan, Y. Zhao, R.G. Pestell, C. Albanese, J.E. Darnell Jr., Cell 98 (1999) 295–303. S. Traverse, K. Seedorf, H. Paterson, C.J. Marshall, P. Cohen, A. Ullrich, Current Biology 4 (1994) 694–701. J. Friel, C. Heberlein, M. Geldmacher, W. Ostertag, Journal of Cellular Physiology 204 (2005) 247–259. K. Miyazawa, D.A. Williams, A. Gotoh, J. Nishimaki, H.E. Broxmeyer, K. Toyama, Blood 85 (1995) 641–649.