The p110δ isoform of PI 3-kinase negatively controls RhoA and PTEN
2007; Springer Nature; Volume: 26; Issue: 13 Linguagem: Inglês
10.1038/sj.emboj.7601763
ISSN1460-2075
AutoresEvangelia A. Papakonstanti, Anne J. Ridley, Bart Vanhaesebroeck,
Tópico(s)Chronic Myeloid Leukemia Treatments
ResumoArticle21 June 2007free access The p110δ isoform of PI 3-kinase negatively controls RhoA and PTEN Evangelia A Papakonstanti Evangelia A Papakonstanti Ludwig Institute for Cancer Research, London, UK Search for more papers by this author Anne J Ridley Anne J Ridley Ludwig Institute for Cancer Research, London, UK Department of Biochemistry and Molecular Biology, University College London, London, UK Search for more papers by this author Bart Vanhaesebroeck Corresponding Author Bart Vanhaesebroeck Ludwig Institute for Cancer Research, London, UK Department of Biochemistry and Molecular Biology, University College London, London, UK Search for more papers by this author Evangelia A Papakonstanti Evangelia A Papakonstanti Ludwig Institute for Cancer Research, London, UK Search for more papers by this author Anne J Ridley Anne J Ridley Ludwig Institute for Cancer Research, London, UK Department of Biochemistry and Molecular Biology, University College London, London, UK Search for more papers by this author Bart Vanhaesebroeck Corresponding Author Bart Vanhaesebroeck Ludwig Institute for Cancer Research, London, UK Department of Biochemistry and Molecular Biology, University College London, London, UK Search for more papers by this author Author Information Evangelia A Papakonstanti1, Anne J Ridley1,2 and Bart Vanhaesebroeck 1,2 1Ludwig Institute for Cancer Research, London, UK 2Department of Biochemistry and Molecular Biology, University College London, London, UK *Corresponding author. Ludwig Institute for Cancer Research, 91 Riding House Street, London W1W 7BS, UK. Tel.: +44 207 878 4066; Fax: +44 207 878 4040; E-mail: [email protected] The EMBO Journal (2007)26:3050-3061https://doi.org/10.1038/sj.emboj.7601763 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Inactivation of PI 3-kinase (PI3K) signalling is critical for tumour suppression by PTEN. This is thought to be a unidirectional relationship in which PTEN degrades the lipids produced by PI3K, thus controlling cell proliferation, survival and migration. We now show that this relationship is in fact bidirectional, whereby PI3K reciprocally controls PTEN. We report that the p110δ PI3K negatively regulates PTEN, through a pathway involving inhibition of RhoA. Inactivation of p110δ in macrophages led to reduced Akt and Rac1 activation, but paradoxically to increased RhoA and PTEN activity. Partial inactivation of p190RhoGAP and a reduced binding of cytoplasmic RhoA to the cyclin-dependent kinase inhibitor p27 both contributed to the increased RhoA-GTP levels upon p110δ inactivation. Pharmacological inhibition of ROCK, a downstream effector kinase of RhoA, restored all signalling and functional defects of p110δ inactivation, including Akt phosphorylation, chemotaxis and proliferation. This work identifies the RhoA/ROCK pathway as a major target of p110δ-mediated PI3K signalling, and establishes for the first time that PI3K controls itself, via a feedback loop involving PTEN. Introduction Ligation of a wide variety of cell surface receptors leads to the generation of the PIP3 lipid second messenger, which affects cell growth, cell cycle progression, survival, intracellular traffic, cytoskeletal changes and migration (Vanhaesebroeck et al, 2001; Hawkins et al, 2006). PIP3 targets diverse downstream effector proteins through interaction with pleckstrin homology (PH) domains present in Ser/Thr kinases (such as Akt/PKB), Tyr kinases (such as Btk), adaptor proteins (such as Gab2) and regulators of small GTPases (such as guanosine nucleotide exchange factors (GEFs) and GTPase-activating proteins (GAPs)) (Vanhaesebroeck et al, 2001; Hawkins et al, 2006). PIP3 action in cells is counteracted by lipid phosphatases, among which PTEN is the most extensively investigated. Indeed, PTEN is mutationally inactivated in a wide variety of cancers, contributing to the constitutive activation of the PI3K pathway in these cells (Leslie and Downes, 2004; Parsons, 2004; Sansal and Sellers, 2004; Cully et al, 2006). Despite its importance, however, a clear picture on the biochemical regulation of PTEN, its responsiveness to extracellular stimuli and its control by PI3K itself has yet to emerge (Leslie and Downes, 2004; Gericke et al, 2006). Processes that have been implicated in PTEN regulation include oxidation, protein:protein interaction and phosphorylation on Tyr by unknown kinases (Koul et al, 2002; Sanchez et al, 2005) or phosphorylation on Ser/Thr, for example by CK2 and GSK3 (Vazquez et al, 2000; Torres and Pulido, 2001; Miller et al, 2002). All of these events may impact on PTEN stability and subcellular localisation. Recently, a role for RhoA and its effector kinase ROCK in positive regulation of PTEN activity has been reported (Li et al, 2005; Sanchez et al, 2005). Indeed, ROCK can increase the phosphorylation and activity of PTEN, through an unknown mechanism but most likely involving a physical interaction with PTEN (Li et al, 2005; Meili et al, 2005). The upstream signals that control this RhoA/ROCK/PTEN pathway are unknown. It is also unclear whether this pathway is operational in cells under non-overexpression conditions. Downstream of tyrosine kinases and Ras, PIP3 is produced by the class IA subset of PI3Ks. These are heterodimers made up of a 110 kDa catalytic subunit (p110α, p110β and p110δ) in complex with one of five regulatory subunits (collectively called the ‘p85s’). Whereas p110α and p110β are found in most tissues, the expression of p110δ is most abundant in leukocytes. We previously presented evidence that the class IA PI3K isoforms have nonredundant functions downstream of the receptors for colony-stimulating factor-1 (CSF-1) in macrophages (Vanhaesebroeck et al, 1999b) and epidermal growth factor (EGF) in breast cancer cell lines (Sawyer et al, 2003). Using microinjection of p110δ-selective antibodies in the immortalised BAC1.2F5 macrophage cell line (Morgan et al, 1987), we showed that p110δ controls actin reorganisation and chemotaxis downstream of the CSF-1 receptor (CSF-1R) (Vanhaesebroeck et al, 1999b). The intracellular signalling mechanism by which p110δ achieves its biological functions in these cells has thus far not been delineated. Here we use genetic and pharmacological approaches to dissect the molecular mechanism of p110δ signalling, using primary mouse macrophages as well as other cell types that express p110δ, including breast cancer cell lines (Sawyer et al, 2003). We report the unexpected finding that p110δ inactivates PTEN through a pathway involving RhoA and ROCK. This inhibition of PTEN is essential for p110δ to exert its biological functions. This is the first documented pathway whereby PTEN activity is controlled by PI3K itself. Results Derivation of macrophages with inactive p110δ Bone-marrow-derived macrophages (BMMs) were derived from wild-type (WT) mice, or from mice in which both p110δ alleles are replaced by a kinase-dead version of p110δ (called p110δD910A), as a result of knock-in mouse gene targeting (Okkenhaug et al, 2002). Pharmacological intervention with p110δ was achieved by the use of IC87114, a small molecule inhibitor with selectivity for p110δ (Sadhu et al, 2003). We first tested the effect of p110δ inactivation on BMM biology. Culture of bone marrow precursor cells from WT and δD910A/D910A mice in the presence of CSF-1 gave rise to similar numbers of mature BMMs (data not shown), expressing similar levels of the CSF-1R and the macrophage-specific marker F4/80 (Supplementary Figure S1). Mutant BMMs did not show compensatory expression of nontargeted class I PI3K isoforms (Supplementary Figure S1b) or other genes (as assessed by gene array studies; Clotilde Billottet and Bart Vanhaesebroeck, unpublished data). Under proliferating conditions in the presence of CSF-1, no changes in cell morphology, size and actin cytoskeleton were observed (data not shown). These data indicate that inactivation of p110δ has no major effects on the differentiation of bone marrow precursor cells to mature macrophages. This conclusion is in line with previous reports that CSF-1-induced differentiation of myeloid precursors to macrophages does not depend upon PI3K activity, but rather on Erk activity (Gobert Gosse et al, 2005), which was unaltered in δD910A/D910A BMMs (see below). CSF-1-induced BMM chemotaxis and DNA synthesis depend on p110δ activity Compared to WT cells, CSF-1-starved δD910A/D910A BMMs had a more rounded appearance (reduced ratio of length/breadth) and a small but significantly reduced adhesive area (Supplementary Figure S2a). Upon stimulation with a uniform and saturating concentration of CSF-1 (30 ng/ml), δD910A/D910A cells also showed reduced actin polymerisation, membrane ruffling and cell spreading (Figure 1A and Supplementary Figure S2a, b). Figure 1.Impact of p110δ inactivation on CSF-1-driven biological responses in BMMs. (A) Acute morphological changes in BMMs upon random stimulation with CSF-1. BMMs were starved of CSF-1 for 16–20 h, stimulated with 30 ng/ml of CSF-1 for the indicated time points, fixed and stained for F-actin, followed by confocal imaging/photography (scale bar: 20 μm). (B) Cell polarisation and PIP3 gradient formation in BMMs upon gradient stimulation with CSF-1. BMMs were exposed to a gradient of CSF-1, delivered via a microtip, for the indicated time periods, followed by fixation and indirect immunofluorescence using an anti-PIP3 antibody (scale bar: 30 μm). Cells with the pan-PI3K inhibitor LY294002 were used as control. (C) Chemotaxis in BMMs exposed to a gradient of CSF-1 in Dunn chambers, monitored for 16 h by time-lapse microscopy. The cell tracks from three separate experiments were merged into a single file for analysis. Circular histograms (upper panels) show the proportion of cells migrating into each of 20 segments of the angular trajectory plot (measured when each cell migrated past a horizon of 30 μm from its starting point with the source of CSF-1 at the top). Arrows indicate significant mean directionality of the cell population. The shaded area marks the 95% confidence intervals of statistical significance. n=x (y) below the histograms indicates the number of cells migrated past the particular horizon (x), out of the total number of cells tracked (y). Similar chemotaxis plots were obtained for horizon limits of 50, 80 and 100 μm, although the cell numbers varied (not shown). Vector plots (lower panel) show the end point of the cells with the starting point of each cell at the intersection between X and Y axes and with the source of CSF-1 at the top of each plot. (D) Number of cells tested under (C) that migrated past the indicated horizons from their point of origin, relative to the total number of cells tracked. (E) Effect of genetic or pharmacological inactivation of p110δ on DNA synthesis of BMMs. Data show a representative experiment performed with quadruplicate wells. LY=LY294002, IC=IC87114. Download figure Download PowerPoint Upon gradient stimulation with CSF-1, administered by a micropipette (Figure 1B), WT cells rapidly polarised towards the source of CSF-1 and accumulated PIP3 at the leading edge. At very early time points, a number of cells in δD910A/D910A BMM cultures polarised but this was followed by quick retraction, so that at later time points polarised cells were no longer observed (Figure 1B). Similarly, IC87114 treatment of BMMs derived from WT mice, transgenic for an intracellular PIP3/PI(3,4)P2-binding GFP-Akt-PH probe (Nishio et al, 2007), blocked CSF-1-induced GFP recruitment to the leading edge (Supplementary Figure S2c). Migration induced by a uniform concentration of CSF-1 was mildly affected upon p110δ inactivation. Although WT and δD910A/D910A BMMs migrated with the same speed (Supplementary Figure S2d), the latter cells stayed closer to their starting point, as revealed by plots whereby the number of cells that migrated past a given distance from their point of origin is expressed, relative to the total number of cells tracked (Supplementary Figure S2d, bottom graph). Chemotaxis of BMMs towards a gradient of CSF-1 in Dunn chemotaxis chambers was blocked by genetic or pharmacological inactivation of p110δ (Figure 1C). Also under these conditions of directional migration, p110δ inactivation led to a reduction in the number of cells that migrated past a given distance from their point of origin (Figure 1D). In fully differentiated BMMs, inactivation of p110δ led to reduced CSF-1-induced DNA synthesis, especially at higher concentrations of CSF-1 (Figure 1E). Treatment with IC87114 also decreased DNA synthesis of WT BMMs to the levels seen in δD910A/D910A cells, but did not affect DNA synthesis of the latter (Figure 1E). The pan-PI3K inhibitor LY294002 further reduced DNA synthesis of δD910A/D910A and WT cells (Figure 1E), suggesting that the remaining DNA synthesis in the absence of p110δ is controlled by PI3K isoforms other than p110δ. However, it is also possible that inhibition of non-PI3K signalling proteins (such as mTOR) by LY294002 (Knight et al, 2006) contributes to the antiproliferative effect of this compound. Active p110δ suppresses PTEN lipid phosphatase activity As expected, CSF-1-induced Akt phosphorylation was reduced in δD910A/D910A cells (Figure 2A), indicative for reduced PIP3 production in these cells upon CSF-1 stimulation. We aimed to overcome this effect of p110δ inactivation by inhibiting PTEN, a major PIP3 phosphatase. Pretreatment with VO-OHpic (3-hydroxypicolinate vanadium(IV) complex), a selective small molecule inhibitor of PTEN (Rosivatz et al, 2007), partially restored CSF-1-induced Akt phosphorylation in δD910A/D910A cells (Figure 2A). Figure 2.p110δ inhibits PTEN activity. (A) Effect of PTEN inhibition on Akt phosphorylation. BMMs were pretreated for 30 min with vehicle or 500 nM of the PTEN inhibitor VO-OHpic (Rosivatz et al, 2007), stimulated with CSF-1 (30 ng/ml) for the indicated time points followed by WB for P-Akt (S473) or GAPDH. Two independent experiments were performed with similar results. (B) Effect of p110δ inactivation on Tyr phosphorylation of PTEN. BMMs were stimulated with CSF-1 (30 ng/ml) for the indicated time points, followed by IP of PTEN and WB for P-Tyr or total PTEN. Bottom panel: ratio of WB signal of Tyr-phosphorylated PTEN over that of total PTEN. Two independent experiments were performed with similar results. (C) Effect of p110δ inactivation on PTEN lipid phosphatase activity. PTEN was immunoprecipitated from BMM lysates and its PIP3 phosphatase activity determined using L-α-phosphatidylinositol(3,4,5)-triphosphate as a substrate and the Biomol Green reagent. One representative experiment out of two independent experiments carried out in triplicate is shown (**P<0.01). The blot above the graph shows WB of PTEN in the BMM lysates used for PTEN immunoprecipitation. (D) PTEN immunoprecipitates were incubated for 30 min with GST-SHP1 or with buffer control, followed by measurement of Tyr phosphorylation and lipid phosphatase activity of PTEN. (E) PTEN activity induced by IC87114 in breast cancer lines. MDA-MB-231 or T47D cancer cells were pretreated with 5 μM of IC87114 or vehicle for 60 min, followed by PTEN IP and determination of its PIP3 phosphatase activity by ELISA (Echelon). One representative experiment out of two independent experiments carried out in triplicate is shown (*P<0.05). The blot above the graph shows a WB of PTEN in the lysates used for PTEN immunoprecipitation. Download figure Download PowerPoint The regulation of PTEN is complex and involves phosphorylation on tyrosine (Tyr) by unknown kinases (Koul et al, 2002; Sanchez et al, 2005) and on Ser/Thr, possibly by CK2, GSK3 and other kinases (Vazquez et al, 2000; Miller et al, 2002). Previous work has also implicated a RhoA/ROCK pathway in the control of PTEN whereby the Ser/Thr kinase ROCK activates PTEN, through an unknown mechanism (Li et al, 2005; Meili et al, 2005; Sanchez et al, 2005). Upstream signalling pathways controlling this RhoA/PTEN axis are also unknown. Genetic or pharmacological inactivation of p110δ led to increased (yet still low) Tyr phosphorylation of PTEN, both under basal and CSF-1-stimulated conditions (Figure 2B; Supplementary Figure S3a). Ser/Thr phosphorylation of PTEN was not affected under those conditions (Supplementary Figure S3b). Inactivation of p110δ also led to increased PTEN activity under basal conditions (Figure 2C). CSF-1 stimulation did not affect PTEN activity in cells with inactive p110δ, in contrast to WT cells in which CSF-1 stimulation reduced PTEN activity (Figure 2C). Treatment of PTEN, immunoprecipitated from cells with inactive p110δ, with the tyrosine phosphatase SHP1 reduced PTEN activity to the levels seen in WT cells (Figure 2D). This indicates a correlation between increased Tyr phosphorylation and PTEN hyperactivity upon p110δ inactivation. PTEN activity was also constitutively increased in brain tissues from mice expressing inactive p110δ (data not shown). Moreover, short-term treatment of p110δ-positive breast cancer cell lines with IC87114 also led to increased PTEN activity (Figure 2E). These data indicate that p110δ action inhibits PTEN activity. p110δ negatively regulates RhoA To investigate how p110δ could affect PTEN activity, we first investigated the effect of p110δ inactivation on early signalling induced by CSF-1. Akt phosphorylation upon stimulation of δD910A/D910A BMMs with a saturating dose of CSF-1 (30 ng/ml) was reduced by approximately 50% (Figures 2A and 3A), whereas Erk phosphorylation was unaffected (Figure 3A and Supplementary Figure S4a). IC87114 mimicked this effect in WT cells (Figure 3A and Supplementary Figure S4a) but did not reduce Akt phosphorylation in δD910A/D910A BMMs (Figure 3A and Supplementary Figure S4a), in line with the p110δ selectivity of this compound. Figure 3.Effect of p110δ inactivation on CSF-1-induced activation of protein kinases and small GTPases. (A) Effect of genetic or pharmacological inactivation of p110δ on CSF-1-induced phosphorylation of Akt and Erk1/2 in BMMs. Top graphs: the indicated BMMs were stimulated with 30 ng/ml CSF-1 for different time points, followed by analysis of phosphorylation of Akt (on T308 and S473) or Erk1/2 (on T202/Y204) by WB of total cell lysates (80 μg/lane). Graphs represent the mean±s.e.m. of three separate experiments. Bottom graphs: WT BMMs were incubated with the indicated concentrations of IC87114 30 min before stimulation with CSF-1 (30 ng/ml, 5 min). Akt and Erk1/2 phosphorylation were assessed by WB of total cell lysates (80 μg/lane). Graphs represent the mean±s.e.m. of three separate experiments. (B) Effect of genetic or pharmacological inactivation of p110δ on basal and CSF-1-induced Rac1 activation in BMMs. Equal volumes of cell lysates of the indicated BMMs were subjected to pull down assay with GTP-PBD, followed by detection of precipitated Rac1 by WB. Total cell lysates were resolved on the same SDS–PAGE gel and immunoblotted for Rac1. Graphs represent the mean±s.e.m. of three separate experiments. (C) Effect of genetic or pharmacological inactivation of p110δ on basal and stimulus-induced RhoA activation in macrophages and breast cancer lines. BMMs or BAC1.2.F5 cells were stimulated with CSF-1 (30 ng/ml), whereas MDA-MB-231 or T47D breast cancer cells were stimulated with EGF (50 ng/ml). When indicated, cells were pretreated with IC87114 (5 μM, 1 h). Equal volumes of cell lysates of the indicated BMMs or breast cancer cell lines were subjected to pull down assay with GST-RBD, followed by WB detection of precipitated RhoA. Total cell lysates were resolved on the same SDS–PAGE gel and immunoblotted for RhoA. Graphs for BMMs represent the mean±s.e.m. of three separate experiments. For breast cancer cell lines, two independent experiments were performed with similar results. Download figure Download PowerPoint Compared to WT cells, CSF-1-induced GTP loading of Rac1 was decreased both in δD910A/D910A and IC87114-treated WT BMMs (Figure 3B and Supplementary Figure S4b). These observations are in line with the known positive role of PI3K in the regulation of Akt and Rac1 (Vanhaesebroeck et al, 2001; Welch et al, 2003). Surprisingly, both basal as well as CSF-1-stimulated levels of RhoA-GTP were significantly increased in δD910A/D910A BMMs, compared to WT cells (Figure 3C and Supplementary Figure S4c). Pretreatment of WT cells with IC87114 also increased RhoA-GTP (Figure 3C and Supplementary Figure S4c). A similar increase in RhoA-GTP was also observed in CSF-1-stimulated BAC1.2F.5 macrophages and EGF-treated MDA-MB-231 and T47D p110δ-expressing breast cancer cell lines (Sawyer et al, 2003) in response to IC87114 (Figure 3C and Supplementary Figure S4d). Similar observations were made in brain cell lysates from δD910A/D910A mice (data not shown). These data imply that, under normal conditions, p110δ activity suppresses RhoA function in cells. RhoA and Rac1 have been shown to oppose each other in several biological settings, and low cellular levels of Rac-GTP most often correlate with high levels of RhoA-GTP, and vice versa (Sander et al, 1999; Raftopoulou and Hall, 2004). This appears not to be the case in BMMs given that basal RhoA-GTP levels were increased in p110δ mutant BMMs, without an effect on basal Rac1-GTP levels (Figure 3B and C). This suggests that p110δ-driven negative regulation of RhoA in these cells under basal conditions is independent of Rac1. Inhibition of ROCK overcomes the biological effects of p110δ inactivation We next set out to investigate the importance of the increased RhoA activity in the biological responses controlled by p110δ and regulation of PTEN. An important effector of RhoA-GTP is the Ser/Thr kinase ROCK (Riento and Ridley, 2003), which is sensitive to Y27632 (Uehata et al, 1997; Nobes and Hall, 1999; Ishizaki et al, 2000). Treatment of δD910A/D910A BMMs with Y27632 reverted p110δ phenotypes to those of WT cells. This included the capacity of these cells to undergo CSF-1-induced cell polarisation and accumulation of PIP3 at the leading edge (Figure 4A), chemotaxis (Figure 4B) and migration in a wound-healing assay (Supplementary Figure S5a). Y27632 also partially recovered the defects in Akt phosphorylation (Figure 4C) and DNA synthesis (Figure 4D), and blocked the increase in Tyr phosphorylation (Figure 4E) and lipid phosphatase activity of PTEN (Figure 4F). Y27632 also inhibited the increased phosphorylation of MLC, a downstream target of ROCK, observed in δD910A/D910A cells (Supplementary Figure S5b). Figure 4.Effect of the ROCK inhibitor Y27632 on CSF-1-induced biological responses of WT and δD910A/D910A BMMs. (A) δD910A/910A BMMs were pretreated with 10 μM Y27632 or vehicle for 30 min, exposed to a gradient of CSF-1 delivered through a microtip for the indicated time points, followed by fixation and indirect immunofluorescence using an anti-PIP3 antibody (scale bar: 20 μm). (B) δD910A/D910A BMMs were treated with 10 μM of Y27632 and exposed to a gradient of CSF-1 in Dunn chambers. Cell migration was followed by time-lapse microscopy for 16 h. The circular histogram (left panel) shows the proportion of cells migrating into each of 20 segments of the angular trajectory plot measured when each cell migrated past a horizon of 70 μm from its starting point with the source of CSF-1 at the top. Arrows indicate significant mean directionality of the cell population and the shaded areas mark the 95% confidence intervals of statistical significance. n=x (y) below the histograms indicates the number of cells migrated past the particular horizon (x), out of the total number of cells tracked (y). Vector plot (right panel) shows the end point of the cells with the starting point of each cell at the intersection between X and Y axes and with the source of CSF-1 at the top of each plot. (C) BMMs were pretreated with 25 μM of Y27632 or its vehicle for 15 min, stimulated with CSF-1 (30 ng/ml) for the indicated time points followed by WB for P-Akt (S473) or total Akt. Graph represents the mean±s.e.m. of three separate experiments. (D) BMMs were incubated with the indicated concentrations of CSF-1 in the presence or absence of Y27632 (20 μM), followed by measurement of [3H]thymidine incorporation 48 h later. Data show a representative experiment carried out in quadruplicate. (E) BMMs were stimulated with CSF-1 (30 ng/ml) for the indicated time points, followed by IP of PTEN and WB for P-Tyr or total PTEN. Bottom panel: ratio of WB signal of Tyr-phosphorylated PTEN over that of total PTEN. Two independent experiments were performed with similar results. (F) BMMs were pretreated with 25 μM of Y27632 or vehicle for 15 min, followed by PTEN IP and determination of its PIP3 phosphatase activity using L-α-phosphatidylinositol (3,4,5) triphosphate as a substrate and the Biomol Green reagent. One representative experiment out of two independent experiments carried out in triplicate is shown. The blot below the graph shows a WB of PTEN in the BMM lysates used for PTEN immunoprecipitation. Download figure Download PowerPoint Alternative ways to interfere with Rho/ROCK pathways had similar effects as Y27632. For example, H1152, a ROCK inhibitor from a different chemical class to Y27632 (Sasaki et al, 2002), increased Akt phosphorylation in δD910A/D910A cells (Supplementary Figure S5c). Direct inhibition of RhoA by C3 transferase also annihilated the increased lipid phosphatase activity of PTEN in δD910A/D910A BMMs (Supplementary Figure S5d). Similar data were observed in p110δ-expressing breast cancer cells in which Y27632 restored IC87114-induced inhibition of EGF-stimulated Akt phosphorylation (Supplementary Figure S5e), demonstrating the generality of the link between p110δ activity and RhoA inhibition. These data are consistent with a model that inactivation of p110δ leads to enhanced ROCK activity, as a consequence of increased RhoA-GTP levels, contributing to the observed phenotypes. p110δ inactivation leads to reduced p190RhoGAP activity RhoA activity could be inhibited by p110δ either through inhibition of a GEF or activation of GAP. Given that PI3K-dependent Rho-GEFs have not been reported, we first focused on the activity of p190RhoGAP, one of the key regulators of RhoA (Ren et al, 1999, 2000; Arthur and Burridge, 2001; Wakatsuki et al, 2003; Continolo et al, 2005). The in vitro GAP activity of p190RhoGAP immunoprecipitated from δD910A/D910A BMMs under CSF-1-starved conditions towards RhoA was significantly reduced (Figure 5A). Given that the expression of p190RhoGAP was not altered in δD910A/D910A BMMs (Figure 5A, lower panel), this observation suggests an intrinsic change in p190RhoGAP in these cells. This was indeed the case, as p190RhoGAP showed reduced basal and CSF-1-stimulated Tyr phosphorylation in δD910A/D910A BMMs (Figure 5B). Figure 5.Effect of p110δ inactivation on p190RhoGAP activity. (A) RhoA GAP activity in p190RhoGAP immunoprecipitates from unstimulated WT or p110δD910A/D910A BMMs. The blot below the graph shows a WB of p190RhoGAP in the BMM lysates used for the GAP assay. Purified p50RhoGAP was used as positive control. (B) Top panel: p190RhoGAP immunoprecipitates were immunoblotted for pTyr or p190RhoGAP. Bottom panel: ratio of WB signal of Tyr-phosphorylated p190RhoGAP over that of total p190RhoGAP. (C) Ratio of the PYK2 WB signal in PTyr IP over that of total PYK2 in total lysates of the indicated cell types. Graph represents the mean±s.e.m. of three separate experiments. (D) Ratio of the WB signal of Tyr (418)-phosphorylated Src over that of tubulin in total lysates of the indicated cell types. Download figure Download PowerPoint The data above are indicative of defective Tyr kinase signalling downstream of the CSF1R upon p110δ inactivation. p190RhoGAP is known to be phosphorylated and activated in an Src-dependent manner (Zrihan-Licht et al, 2000), via a complex mechanism that also involves PYK2, a FAK Tyr kinase homologue expressed in leucocytes (Ganju et al, 1997). Indeed, PYK2 has been reported to activate Src (Dikic et al, 1996) and to bind constitutively to p190RhoGAP, facilitating the activation of the latter by Src (Zrihan-Licht et al, 2000). Complex formation between PYK2 and p190RhoGAP was unaltered in δD910A/D910A BMMs (Supplementary Figure S6a), both under basal and CSF-1-stimulated conditions. However, CSF-1-induced Tyr phosphorylation of PYK2 (Figure 5C and Supplementary Figure S6b) was reduced in δD910A/D910A cells and IC87114-treated WT BMMs. Likewise, also Tyr phosphorylation of Src was decreased under these conditions (Figure 5D and Supplementary Figure S6c). Taken together, these results are consistent with a model in which p110δ activity leads to increased PYK2/Src Tyr phosphorylation and p190RhoGAP activity, leading to decreased RhoA-GTP levels. Conversely, inactivation of p110δ leads to increased RhoA activity (schematically shown in Figure 8). Figure 6.Effect of p110δ inactivation on p27 subcellular localisation and binding to cytosolic RhoA in BMMs. (A) Top panel: BMMs were stimulated with CSF-1 (30 ng/ml) for the indicated time points, followed by isolation of cytosolic and nuclear fractions and WB for p27, tubulin (marker for cytosol) or retinoblastoma protein (Rb; marker for nucleus). Bottom panel: ratio of WB signal of cytosolic vs nuclear p27. (B) Top panel: RhoA was immunoprecipitated from the cytosolic fraction of BMMs using a monoclonal antibody, followed by WB for p27 or RhoA. Bottom panel: ratio of WB signal of cytosolic p27 over that of cytosolic
Referência(s)