Intramolecular and Intermolecular Interactions of Protein Kinase B Define Its Activation In Vivo

Intramolecular and Intermolecular Interactions of Protein Kinase B Define Its Activation In Vivo
of 12
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  Intramolecular and IntermolecularInteractions of Protein Kinase BDefine Its Activation In Vivo Ve´ronique Calleja 1 [ , Damien Alcor 1 [ , Michel Laguerre 2 , Jongsun Park  3¤ , Borivoj Vojnovic 4 , Brian A. Hemmings 3 ,Julian Downward 5 , Peter J. Parker 6* , Banafshe´ Larijani 1* 1  Cell Biophysics Laboratory, Lincoln’s Inn Fields Laboratories, London Research Institute, Cancer Research UK, London, United Kingdom,  2  Institut Europe´en de Chimie etBiologie, Pessac Cedex, France,  3  Friedrich Miescher Institute for Biomedical Research, Basel, Switzerland,  4  Advanced Technology Development Group, Gray Cancer Institute,Mount Vernon Hospital, Northwood, United Kingdom,  5  Signal Transduction Laboratory, Lincoln’s Inn Fields Laboratories, London Research Institute, Cancer Research UK,London, United Kingdom,  6  Protein Phosphorylation Laboratory, Lincoln’s Inn Fields Laboratories, London Research Institute, Cancer Research UK, London, United Kingdom Protein kinase B (PKB/Akt) is a pivotal regulator of diverse metabolic, phenotypic, and antiapoptotic cellular controlsand has been shown to be a key player in cancer progression. Here, using fluorescent reporters, we shown in cells that,contrary to in vitro analyses, 3-phosphoinositide–dependent protein kinase 1 (PDK1) is complexed to its substrate,PKB. The use of Fo ¨ rster resonance energy transfer detected by both frequency domain and two-photon time domainfluorescence lifetime imaging microscopy has lead to novel in vivo findings. The preactivation complex of PKB andPDK1 is maintained in an inactive state through a PKB intramolecular interaction between its pleckstrin homology (PH)and kinase domains, in a  ‘‘ PH-in ’’  conformer. This domain–domain interaction prevents the PKB activation loop frombeing phosphorylated by PDK1. The interactive regions for this intramolecular PKB interaction were predicted throughmolecular modeling and tested through mutagenesis, supporting the derived model. Physiologically, agonist-inducedphosphorylation of PKB by PDK1 occurs coincident to plasma membrane recruitment, and we further shown here thatthis process is associated with a conformational change in PKB at the membrane, producing a  ‘‘ PH-out ’’  conformer andenabling PDK1 access the activation loop. The active, phosphorylated,  ‘‘ PH-out ’’  conformer can dissociate from themembrane and retain this conformation to phosphorylate substrates distal to the membrane. These in vivo studiesprovide a new model for the mechanism of activation of PKB. This study takes a crucial widely studied regulator(physiology and pathology) and addresses the fundamental question of the dynamic in vivo behaviour of PKB with adetailed molecular mechanism. This has important implications not only in extending our understanding of thisoncogenic protein kinase but also in opening up distinct opportunities for therapeutic intervention. Citation: Calleja V, Alcor D, Laguerre M, Park J, Vojnovic B, et al. (2007) Intramolecular and intermolecular interactions of protein kinase B define its activation in vivo. PLoSBiol 5(4): e95. doi:10.1371/journal.pbio.0050095 Introduction A key downstream relay in various growth factors andhormones is the activation of the serine/threonine proteinkinase (PKB/Akt). PKB activates a plethora of proteins thatare involved in metabolism, proliferation, growth, andsurvival [1–3]. Several lines of evidence indicate that the PKB pathway is involved in human cancer, and in particular,its overexpression induces malignant transformation andchemoresistance [2,4,5]. Its activation is thought to proceedthrough the recruitment of the protein to membranes viainteraction of its PH domain with the phosphoinositidesproduced by phosphoinositide-3-kinase [specifically PtdIns(3,4,5)P 3  and PtdIns (3,4)P 2 ] [6,7]. The lipid-bound PKB is then phosphorylated by 3-phosphoinositide–dependent proteinkinase 1 (PDK1), which is also recruited through its PHdomain binding to PtdIns (3,4,5)P 3 . The PDK1 phosphoryla-tion, critical for activation, occurs at Thr308 in the activationT-loop of PKB a  [8]. A second phosphorylation within a C-terminal hydrophobic motif at Ser473 acts in synergy to fullyactivate the protein kinase. It is believed that this phosphor-ylation occurs via the mTor:rictor pathway [9]. Our under-standing of the mechanism of activation of PKB remainslimited as, unlike other AGC kinases that are substrates forPDK1 [10], direct in vitro or in vivo interaction of PKB withPDK1 has not been observed. The current models of PKBactivation only speculate, based on the regulation of otherAGC kinases, how PKB may change its conformation tointeract with PtdIns at the plasma membrane and thereafterbe activated by PDK1. The functionally independent behav-iour of the PH and kinase domains of PKB is well Academic Editor:  John Kuriyan, University of California, United States of America Received  December 11, 2006;  Accepted  January 6, 2007;  Published  April 3, 2007 Copyright:  2007 Calleja et al. This is an open-access article distributed under theterms of the Creative Commons Attribution License, which permits unrestricteduse, distribution, and reproduction in any medium, provided the srcinal authorand source are credited. Abbreviations:  Cy3, Cyan 3 dye; EGFP, enhanced green fluorescent protein; FLIM,fluorescence lifetime imaging microscopy; FRET, Fo¨rster resonance energy transfer;mRFP, monomeric red fluorescent protein; PDGF, platelet-derived growth factor;PDK1, 3-phosphoinositide–dependent protein kinase 1; PH, pleckstrin homology;PIF, PDK1 interacting fragment; PKB, protein kinase B; PtdIns, phosphoinositides;tau (or  s ), refers to the lifetimes in the figures or in the text* To whom correspondence should be addressed. E-mail: (PJP); (BL) [ These authors contributed equally to this work.¤ Current address: Cell Signaling Laboratory, Cancer Research Institute, College of Medicine, Chungnam National University, Taejon, South Korea PLoS Biology | April 2007 | Volume 5 | Issue 4 | e950780   o   BIOLOGY  characterised; the structure of both has been solved sepa-rately [6,7,11–13]. In isolation, the PH domain retains intrinsic lipid binding properties and, similarly, the kinasedomain retains function [14,15]. The kinase domain is well understood structurally with respect to its inactive and activeconformers [12,13]. However, the more complex interactions between the PH and kinase domains have not succumbed todirect structural analysis and remain to be elucidated. Therecent advent of an in vivo probe to monitor PKB conforma-tional changes in conjunction with molecular modeling haspermitted analysis of the mechanism by which these twodomains interact and how PKB changes conformation inrelation to its activation by PDK1. We have addressed thepossible molecular mechanisms involved in the conforma-tional dynamics of PKB and dissected its activation in intactcells. This has important implications not only in extendingour understanding of this critical regulator but also inopening distinct opportunities for therapeutic intervention. Results PDK1 Precomplexes with Its Substrate PKB in theCytoplasm It has been assumed that upon activation of various growthfactor pathways, PDK1 and PKB uniquely colocalise at theplasma membrane, permitting PDK1 to phosphorylate PKB.To assess in single cells the dynamic relationship betweenPKB a  and PDK1, Fo ¨ rster resonance energy transfer (FRET)was exploited. Two fusion proteins were generated bygenetically encoding enhanced green fluorescent protein(EGFP) (donor) at the N terminus of PDK1 and monomericRFP (acceptor) at the N terminus of PKB a . NIH3T3 cells weretransfected with the expression vectors for GFP-PDK1 andRFP-PKB, and the expressed proteins behaviours werefollowed upon stimulation of the platelet-derived growthfactor (PDGF) receptor pathway (note that the levels of expression were the minimum required for data acquisition).For optimum spatial resolution, FRET was monitored by two-photon fluorescence lifetime imaging microscopy (FLIM).Resonance energy transfer was detected by the decrease inthe donor lifetime in the presence of the acceptor. Thevariations in lifetime are presented by lifetime distributionscurves and mean FRET efficiency. Coexpression of RFP-PKBwith GFP-PDK1 in NIH3T3 cells (Figure 1A) induced adecrease in the lifetime of GFP-PDK1. The decrease in thedonor lifetime is quantitatively presented by the lifetimedistribution curves. The blue curve is the GFP-PDK1 whenexpressed alone, and the green curve is when it is expressedwith RFP-PKB (acceptor). From this reduction in lifetime, wededuced that, at steady state, GFP-PDK1 and RFP-PKBinteract in the cytoplasm. The cytoplasmic mean FRETefficiency presented as box and whiskers plots is 2.5 (Figure1C). Upon PDGF stimulation, both PKB and PDK1 trans-located to the plasma membrane, and a further decrease inthe average lifetime was detected (orange curve). The trans-location of RFP-PKB to the plasma membrane is clearlyshown in Figure S6. The separation of the lifetime distribu-tions at the plasma membrane versus the cytoplasm (redcurve: plasma membrane; light green curve: cytoplasm)showed that at the plasma membrane, the mean FRETefficiency is 7.1, and in the cytoplasm, it is 4.9 (Figure 1C).These results indicated that PDK1 and PKB interacted underbasal conditions and that the stimulation of the PDGFreceptor pathway promoted an enrichment of the PKB–PDK1 complex at the plasma membrane. This intermolecularinteraction was also demonstrated by frequency domainFLIM (Figure S1). This behaviour indicates that in thecytoplasm, the PKB–PDK1 complex is at equilibrium deter-mined by the bulk phase concentrations. Upon stimulation,the local concentration of both proteins is enhanced at theplasma membrane, whereby the equilibrium is shifted towardthe formation of the PKB–PDK1 complex.To verify that the PKB–PDK1 interaction was due todocking and not purely concentration effects in thecytoplasm due to transfection, GFP-PKB alone or togetherwith RFP-PKB was expressed in NIH3T3 cells (Figure 1B). Thelifetimes from cells with expression levels of donor (GFP-PKB) and acceptor (RFP-PKB), comparable to those of GFP-PDK1 and RFP-PKB, were identified. Despite the colocalisa-tion of GFP-PKB and RFP-PKB in the cytoplasm and thecorecruitment of both PKBs at the plasma membrane, uponactivation, the FRET efficiency did not change. This is clearlyillustrated by the complete lack of lifetime variation (Figure1B and 1C). These data indicated that PDK1 and PKB weredocking in the cytoplasm and at the plasma membrane understeady state and stimulated conditions.To further investigate whether the docking was PtdIns(3,4,5)P 3  dependent, cells were pretreated with the phospha-tidylinositol 3-kinase inhibitor LY294002 prior to PDGFstimulation. Figure 1D shows that with LY294002, thetranslocation of PKB and PDK1 is prevented. However, acytoplasmic FRET efficiency of 3.4, similar to basal con-ditions, was still observed (Figure 1E). The PDGF-inducedphosphorylation of Thr308 and Ser473 was reduced onpretreatment with LY294002 (Figure S2E). This indicatedthat the recruitment of the PKB–PDK1 complex to theplasma membrane was dependent on PtdIns (3,4,5)P 3  pro-duction, while the docking of PKB to PDK1 in the cytoplasmwas PtdIns (3,4,5)P 3  independent. To assess the role of PtdIns(3,4,5)P 3  in an inhibitor-independent manner, PKB and PDK1 PLoS Biology | April 2007 | Volume 5 | Issue 4 | e950781In Vivo PKB Molecular Mechanism Author Summary Regulation of intracellular signaling depends on the preciseoperation of molecular switches such as kinases and phosphatases.Disruption of their activities leads to inappropriate cellularproliferation, growth, and survival. Protein kinase B (PKB) is a criticalkinase that regulates events downstream of growth factor receptors.It is also involved in human cancer, where its overexpression inducesmalignant transformation. We studied the molecular mechanisms of PKB’s interaction with its upstream regulator, 3-phosphoinositide–dependent protein kinase 1 (PDK1). By using a fluorescent probe, wemonitored the conformational changes of PKB in cells using Fo¨rsterresonance energy transfer detected by fluorescence lifetimeimaging microscopy. Applying this approach, we show that PKBand PDK1 are found as complexes in the cytoplasm. Despite thisproximity to its regulator, we show that PKB remains inactivethrough an intramolecular interaction of its pleckstrin homology(PH) domain and kinase domain. We refer to this inactive state as the ‘‘ PH-in ’’  conformer. Following growth factor activation, PKB changesconformation to the  ‘‘ PH-out ’’  conformer and is phosphorylated byPDK1. The active  ‘‘ PH-out ’’  conformer dissociates from the plasmamembrane to phosphorylate downstream proteins. Our in vivostudies provide a new model for the mechanism of PKB.  PH domain mutants (mutants that do not efficiently bind tophosphoinositide lipids: PDK1 RRR and PKB R25C [16,17])were used to determine their influence on interaction withtheir wild-type binding partners. Cells were cotransfectedwith GFP-PDK1/RFP-PKB R25C or GFP-PDK1 RRR/RFP-PKB.Under basal conditions, both wild-type proteins interactedwith their mutant partner (Figure S3A and S3B). However,upon activation, the PH mutants did not translocate to theplasma membrane and a change in FRET efficiency was notdetected (Figure S3C). It is concluded that complex formationin the cytosol is independent of lipid binding.PDK1 docks with phosphorylated/acidic hydrophobic mo-tifs at the C termini of some AGC kinases. This is affectedthrough a PDK1 interacting fragment domain (PIF pocket)and additionally through a phosphate-binding site within theupper lobe of the PDK1 kinase domain. PDK1 mutantsdefective in these interactions, namely the L155E mutant inthe PIF pocket and the R131A mutant in the phosphate-binding pocket, still interacted with PKB in the cytoplasm oron PDGF stimulation at the plasma membrane (Figure S4Aand S4B). Moreover, the PKB mutant that is not phosphory-lated in response to agonists, T308A/S473A (Figure S4C), alsoretained binding capacity, confirming independently thatPKB Ser473 phosphorylation was not required for dockinginto PDK1 9 s phosphate-binding pocket. Therefore, theinteraction of PKB and PDK1 is independent of the PIF Figure 1.  PDK1 Interacts with PKB in a Phosphatidylinositol 3-Kinase–Dependent Manner at the Membrane and in a Phosphoinositide-IndependentManner in the Cytoplasm(A) NIH3T3 cells were transfected with GFP-PDK1 alone or with RFP-PKB. The lifetimes are presented in pseudo-colour scale ranging from 2.0 to 2.7 ns.The graph presents the lifetime distributions of the cells for the following conditions: GFP-PDK1 (blue), GFP-PDK1 with RFP-PKB prior (green) and uponPDGF stimulation (orange). The red and the light green lines indicate the lifetime distributions of the plasma membrane and cytoplasm pixels,respectively, after separation of the compartmental pixels (see decomposition).(B) NIH3T3 cells were transfected either with GFP-PKB alone or together with RFP-PKB. The FRET efficiency variation is statistically insignificant.(C) The statistical analysis of experiments in A and B are presented as box and whiskers plots (see Supporting Information). The FRET efficiency wascalculated for at least ten cells per condition for the cytoplasmic (Cyto) and plasma membrane (PM) components, as indicated. The first column showsthe variation of the donor lifetime (blue) centred at zero. The dispersion is less than  6 1%. The FRET efficiency of GFP-PDK1 þ RFP-PKB condition ispresented before (green boxes) and after (red boxes) PDGF stimulation. The FRET efficiency of the control GFP-PKB þ RFP-PKB upon PDGF is presentedby the two last boxes. The mean value of the FRET efficiency is indicated on top of each box. The highly significant FRET efficiency variations at (PM), forexperiment A, is indicated by a  p- value of 0.002 (see Supporting Information).(D) Similar experiment as in (A) was performed but cells were first pretreated for 20 min with LY294002. The graph shows the superposition of thelifetime distributions of coexpressed GFP-PDK1 and RFP-PKB pretreated with LY294002 alone (green line) or pretreated with LY294002 before PDGFstimulation (dashed green line). Control GFP-PDK1 lifetime distribution is shown in blue.(E) The FRET efficiency between LY294002- and LY294002 þ PDGF–treated cells do not vary significantly (  p ¼ 0.8).doi:10.1371/journal.pbio.0050095.g001PLoS Biology | April 2007 | Volume 5 | Issue 4 | e950782In Vivo PKB Molecular Mechanism  PLoS Biology | April 2007 | Volume 5 | Issue 4 | e950783In Vivo PKB Molecular Mechanism  and the phosphate-binding pockets and does not requirephosphorylation of Thr308 or Ser473 (Figure S4D). In linewith this observation, it has also been shown by using knock-in mutants of PDK1 that the PIF pocket and the phosphate-binding pocket of PDK1 were not implicated in the activationof PKB [18,19]. Therefore, the interaction in the cytoplasm occurs via mechanisms that are distinct from those of otherAGC kinases.Since these kinases interacted in the cytoplasm, it was of interest to determine why PKB was not constitutivelyphosphorylated by the associated PDK1 in the basal state.One of the mechanisms that regulate the basal levels of Ser473 is the dephosphorylation of this residue by a serine/ threonine protein phosphatase, PP2A, implicated in thedownregulation of PKB activation [20]. It was postulated thatdephosphorylation may dominate the cytoplasmic phosphor-ylation of PKB. To test this hypothesis, PP2A activity wasblocked by treating cells with okadaic acid. The time-dependent increase of RFP-PKB Ser473 phosphorylationupon treatment indicated that in the absence of PP2A,phosphorylation of Ser473 was enhanced, showing that inNIH3T3 cells PP2A regulates Ser473 phosphorylation in thecytoplasm (Figure S2A). The okadaic acid–insensitive phos-phatase PHLPP [21] cannot therefore regulate the basalphosphorylation process. Unlike Ser473, the okadaic acidstimulation of Thr308 was modest. It was possible that the ‘‘ basal state ’’  phosphorylation induced by okadaic acidoccurred via a transient PtdIns (3,4,5)P 3 –associated form.To investigate this, we exploited the nonbinding PH domainmutant of PKB (RFP-PKB R25C). The okadaic acid–inducedphosphorylation of Ser473 and Thr308 did not differsignificantly from that of wild-type PKB, whereas the PDGF-induced phosphorylation of RFP-PKB R25C was significantlyreduced (Figure S2B). These data indicated that the okadaicacid–induced phosphorylations occurred in the cytoplasm. InRFP-PKB–transfected cells, the okadaic acid–induced Ser473phosphorylation was at the same level as the Ser473phosphorylation induced by PDGF (Figure S2A, densitom-etry). However, in the case of phospho-Thr308, the okadaicacid signal was reduced compared to phosphorylationtriggered by PDGF stimulation (Figure S2A, densitometry).The reduced phosphorylation of Thr308 led to the hypothesisthat Thr308 was inaccessible in the cytoplasmic PDK1complex due to PH domain steric hindrance. If thishypothesis is correct, then the removal of the PH domainshould alter the kinetics of Thr308 phosphorylation inresponse to okadaic acid. Thr308 phosphorylation of thedeletion mutant RFP- D PH PKB was more extensive than thewild-type PKB (Figure S2C). This provided evidence that, inthe cytoplasm, Thr308 was not fully accessible to PDK1 andhence could not be phosphorylated.To verify that in the RFP- D PH-PKB mutant Thr308phosphorylation was due to PDK1, NIH3T3 cells werecotransfected with RFP- D PH PKB and GFP-PDK1 wild-typeor the PH domain mutant GFP-PDK1 RRR (Figure S2D). In allcases, a prominent phosphorylation of Thr308 independentof PDGF was detected. Therefore, it is established that whenthe PKB PH domain is absent, Thr308 could be phosphory-lated by PDK1 within the cytosolic complex. Thus, formaintenance of PKB in a Thr308-dephosphorylated stateunder basal conditions, its PH and kinase domains arepredicted to interact to obscure Thr308 accessibility. PKB PH and Catalytic Domains Interact in the PKBInactive State To test the above hypothesis and determine the possibleinteractive sites of the PH and kinase domains, molecularmodeling was exploited. The calculations of lipophilicitypotentials of the PH and kinase domains resulted in findingtwo isolated hydrophobic patches that were located on thekinase domain (Figure 2A, in blue). On the PH domain, foursmall hydrophobic patches were found (Figure 2A, in cyan)and were located on the same side of the protein. To test thevalidity of this molecular model, the electrostatic potentialmaps of each domain were calculated. The complementarityobserved between the positive (red) and negative (blue) lobesfor the two domains (Figure 2B) was almost perfect. Theupper positive lobe of the PH domain corresponded to theupper negative lobe of the kinase domain, and reciprocally,the negative lobe on the PH domain corresponded to apositive lobe on the kinase domain. Figure 2C shows thecomplementarity between the acidic and the basic residues inthe kinase and PH domains of PKB. The yellow arrowsindicate the pair of complementary residues on each domain.The fit of the hydrophobic interactions predicts that theTrp80 on the PH domain, at the extremity of the variableloop 3, inserts inside a deep cleft in the kinase domainaround residues Lys297, Glu298, and Glu314 (Figure 2D).Furthermore, the crystal structure of the isolated PKB PHdomain showed that Trp80 may have an important role in theinteraction of the two domains of PKB [7].From the complementarity of the basic and acidic residueslocated around Trp80 in the PH and kinase domains, it wasenvisaged that mutations of these regions would disrupt the Figure 2.  PKB PH and Catalytic Domains Interact in the PKB Inactive State(A) The solvent-accessible surface of the domains are shown in green (PH) and in yellow (kinase) and the hydrophobic patches are indicated. The IP 4 binding site is in magenta (located to the rear of PH domain) and the ATP active site is in red.(B) The electrostatic potential maps of PKB PH and kinase domains are calculated in the complexed form, based on crystal structures of PKB a  PHdomain. A strong complementarity between the positive and negative lobes of the PH and kinase domains is seen.(C) Acidic (red), basic (blue), and polar (green) residues are presented here. The Trp80 is shown in purple and Arg25 is shown in light blue. The yellowarrows show the complementarity between the acidic and basic residues in each domain. The orange arrow shows the docking site of Trp80 in thekinase domain cleft.(D) The model of complexed PKB a  PH domain (yellow) and the rebuilt catalytic domain model (green) is based on the hydrophobicity in an aqueousenvironment. To visualise the binding site of the lipid head group in the complex, IP 4  is represented as CPK (Corey-Pauling-Koltun) orange spheres. Notethat IP 4  cannot be positioned unless the PH and kinase domains are separated.(E) NIH3T3 cells were transfected either with GFP-PKB or GFP-PKB-RFP (WT) or mutant (EE). The lifetimes are in pseudo-colour scale ranging from 2.30 to2.67 ns. The graph shows the lifetime distributions: GFP-PKB (blue), GFP-PKB-RFP WT (green), and EE (red). The statistical analysis shows the highlysignificant variation of FRET efficiency between WT and EE (  p ¼ 0.004).(F) Expression of GFP-PKB-RFP and phosphorylation on Thr308 and Ser473 sites is determined by Western blot using total protein (WB Akt) orphosphospecific antibodies as indicated.doi:10.1371/journal.pbio.0050095.g002PLoS Biology | April 2007 | Volume 5 | Issue 4 | e950784In Vivo PKB Molecular Mechanism
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks