Endocrinology, doi:10.1210/en.2006-0446
Endocrinology Vol. 148, No. 1 374-385
Copyright © 2007 by The Endocrine Society
A Purine Analog Kinase Inhibitor, Calcium/Calmodulin-Dependent Protein Kinase II Inhibitor 59, Reveals a Role for Calcium/Calmodulin-Dependent Protein Kinase II in Insulin-Stimulated Glucose Transport
Nicky Konstantopoulos,
Seb Marcuccio,
Stella Kyi,
Violet Stoichevska,
Laura A. Castelli,
Colin W. Ward and
S. Lance Macaulay
CSIRO Molecular and Health Technologies (N.K., V.S., L.A.C., C.W.W., S.L.M.), Parkville, Australia 3052; and CSIRO Molecular and Health Technologies (S.M., S.K.), Clayton South, Australia 3169
Address all correspondence and requests for reprints to: Dr. S. Lance Macaulay, CSIRO Molecular and Health Technologies, 343 Royal Parade, Parkville 3052, Victoria, Australia. E-mail: lance.macaulay{at}csiro.au.
 |
Abstract
|
|---|
Olomoucine is known as a cyclin-dependent kinase inhibitor. We found that olomoucine blocked insulin’s ability to stimulate glucose transport. It did so without affecting the activity of known insulin signaling proteins. To identify the olomoucine-sensitive kinase(s), we prepared analogs that could be immobilized to an affinity resin to isolate binding proteins. One of the generated analogs inhibited insulin-stimulated glucose uptake with increased sensitivity compared with olomoucine. The IC50 for inhibition of insulin-stimulated glucose uptake occurred at analog concentrations as low as 0.1 µM. To identify proteins binding to the analog, [35S]-labeled cell lysates prepared from 3T3-L1 adipocytes were incubated with analog chemically cross-linked to a resin support and binding proteins analyzed by SDS-PAGE. The major binding species was a doublet at 50–60 kDa, which was identified as calcium/calmodulin-dependent protein kinase II (CaMKII) by N-terminal peptide analysis and confirmed by matrix-assisted laser desorption ionization-mass spectrometry as the 
- and ß-like isoforms. To investigate CaMKII involvement in insulin-stimulated glucose uptake, 3T3-L1 adipocytes were infected with retrovirus encoding green fluorescent protein (GFP)-hemagluttinin tag (HA)-tagged CaMKII wild-type or the ATP binding mutant, K42M. GFP-HA-CaMKII K42M cells had less kinase activity than cells expressing wild-type GFP-HA-CaMKII. Insulin-stimulated glucose transport was significantly decreased (
80%) in GFP-HA-CaMKII K42M cells, compared with nontransfected cells, and cells expressing either GFP-HA-CaMKII or GFP-HA. There was not a concomitant decrease in insulin-stimulated GLUT4 translocation in GFP-HA-CaMKII K42M cells when compared with GFP-HA alone. However, insulin-stimulated GLUT4 translocation in GFP-HA-CaMKII cells was significantly higher, compared with either GFP-HA or GFP-HA-CaMKII K42M cells. Our results implicate the involvement of CaMKII in glucose transport in a permissive role.
|
Introduction
|
|---|
INSULIN STIMULATION of glucose transport in fat and muscle is critical for the maintenance of glucose homeostasis. Its impairment is an early defect that contributes to insulin resistance, obesity and type 2 diabetes. The primary mechanism for insulin stimulation of glucose uptake in muscle and fat is the translocation of GLUT4 glucose transporters to the cell surface from intracellular storage vesicles within the cell. However, the later stages of the signaling pathway(s) that mediate this effect remain to be established. The initial trigger for insulin stimulation of glucose transport is the activation and autophosphorylation of the insulin receptor (IR) tyrosine protein kinase and subsequent phosphorylation of its downstream effectors, insulin receptor substrate (IRS)1 and IRS2 predominantly. Tyrosine phosphorylated IRS proteins dock several effectors including phosphoinositide 3-kinase (PI 3-kinase). A number of studies have clearly demonstrated that PI 3-kinase function is necessary, but not sufficient, for insulin-stimulated GLUT4 translocation (reviewed in Ref. 1). Activation of PI 3-kinase stimulates phosphoinositide-dependent kinase-1, leading to the activation of protein kinase B (Akt). Several studies, including our own, have established involvement of Akt as a downstream target of PI 3-kinase-mediating insulin stimulation of glucose transport (2). There are also substantial data to implicate the atypical protein kinase C (PKC) isoforms (
and 
) in this process (3). AS160 (a Rab GTPase-activating protein) has been reported as a potential but as yet uncharacterized target for Akt (4). Recent studies suggest that this protein interacts with insulin-responsive aminopeptidase on GLUT4-containing vesicles and may be involved in the translocation process (5). Other targets for these kinases in the signaling pathway have yet to be identified but are critical to understanding regulation of glucose transport.
p38-MAPK and calcium/calmodulin-dependent protein kinase II (CaMKII) have been implicated as potential regulators of glucose transport primarily through inhibitor studies. However, recent studies showed that the p38-MAPK inhibitor, SB203580, affected glucose transport via direct interaction with GLUT4 itself negating earlier reports that suggest involvement of p38-MAPK (6). The effects ascribed to CaMKII are less clear. The CaMKII inhibitor KN62 that binds to the calmodulin binding site of CaMKII, has been reported to inhibit both insulin- and hypoxia-stimulated glucose transport possibly at the level of vesicular trafficking (7). A variety of calmodulin antagonists including trifluoperazine, W13, W7, and Ca2+ chelating agents such as bis-(o-aminophenoxy)-ethane-N,N,N',N'-tetraacetic acid, tetra(acetoxymethyl)-ester have also been shown to inhibit both insulin- and hypoxia-stimulated glucose transport providing further support for a potential role of this family of kinases (7, 8). AMP-activated protein kinase also has a regulatory role but not in mediating insulin-stimulated glucose transport (9).
Leads to understanding the kinases that mediate the latter stages of insulin-stimulated GLUT4 translocation may be achieved through consideration of the machinery involved in the translocation, docking, and fusion process. This process involves a soluble N-ethylmaleimide sensitive fusion factor attachment protein receptor (SNARE) protein on the vesicle; vesicle-associated membrane protein 2 (VAMP2), which recognizes partner SNARE proteins on the plasma membrane; syntaxin (STX) 4; and synaptosomal-associated protein of 23 kDa (SNAP23) that together form a SNARE complex enabling fusion of the vesicle with the plasma membrane, placing GLUT4 on the cell surface, and enabling entry of glucose into the cell. A number of studies have examined the phosphorylation status of these proteins without consensus on mode of regulation (10). Phosphorylation of SNAP23 has been implicated in mast cell degranulation, possibly mediated via PKC (11), and PKC has been reported to phosphorylate VAMP2 (12).
Another point of regulation by phosphorylation could be mediated via SNARE protein accessory proteins. Two of these, namely synip and pantophysin, have been reported to be regulated by phosphorylation in response to insulin, but the role of their phosphorylation in regulation of GLUT4 translocation is still unclear (1, 12, 13). Munc18c (one of the mammalian homologues of the yeast Sec1 family of STX binding proteins) is required for GLUT4 trafficking and appears to regulate this process (14). Disruption of the function of any these SNARE proteins in adipocytes results in a specific block in the movement of GLUT4 to the cell surface, which emphasizes their importance in this process (15, 16).
Exactly how Munc18c regulates GLUT4 translocation is unclear. Recent studies showed that for an analogously regulated process, neurotransmitter release, the cell division kinase, cyclin-dependent kinase (CDK) 5, phosphorylates the neuronal munc homologue Munc18a and reduces its affinity for STX1A (17). This leads to its dissociation from STX1A and enables fusion of secretory vesicles and neurotransmitter release. A selective purine analog inhibitor of CDK, olomoucine, inhibits neurotransmitter release by competing for the ATP binding site of the kinase (17).
Here we report the inhibition of insulin-mediated glucose transport by olomoucine in 3T3-L1 adipocytes. Because known kinases inhibited by olomoucine (IC50 < 10 µM), namely p33CDK2/cyclin A, p34cdc2/cyclin B, p33CDK2/cyclin E, and CDK5/p35 kinase (18), are not responsive to insulin in 3T3-L1 adipocytes (19, 20), these data suggested that olomoucine inhibited another kinase. Olomoucine was found to have little effect on any of the known insulin-responsive kinases. This study therefore sought to identify the olomoucine-sensitive kinase activity that regulated glucose transport. An olomoucine analog, calcium/calmodulin-dependent protein kinase II inhibitor 59 (CK59), was generated for resin attachment to bind and isolate the activating kinase. CK59 markedly inhibited insulin stimulation of glucose transport with a lower IC50 than olomoucine itself. Affinity chromatography coupled with mass spectroscopy identified the major binding species in 3T3-L1 adipocyte extracts as CaMKII. Furthermore, the expression of a kinase-dead CaMKII in 3T3-L1 adipocytes inhibited insulin-stimulated glucose transport. These results thus support a role for CaMKII in insulin-stimulated GLUT4 translocation and glucose transport.
|
Materials and Methods
|
|---|
Materials
The following antibodies were used to detect phosphorylation, on tyrosine residues, PY20 or PY99 (mouse monoclonal; Santa Cruz Biotechnology, Santa Cruz, CA), of Akt on Ser 473 or Thr 308 (rabbit polyclonal; Cell Signaling, Beverly, MA); p42/p44-MAPK on Thr 202/Tyr 204 (E10 mouse monoclonal; Cell Signaling); ribosomal kinase 2 (RSK2) on Ser 227 or Ser 380 (goat polyclonal; Santa Cruz); CaMKII on Thr 286/287 (goat polyclonal; Santa Cruz); and p38-MAPK on Thr 180/Tyr 182 (rabbit polyclonal; Cell Signaling). The following antibodies were used to detect the presence of IRS1 (A-19 rabbit polyclonal; Santa Cruz), Akt1/2 (rabbit polyclonal; Cell Signaling), SHC (C-20 rabbit polyclonal; Transduction Laboratories, Lexington, KY), CDK5 (C-8 rabbit polyclonal, which also detected its proteolytic cleavage product p25; Santa Cruz), glycogen synthase kinase (GSK)3
/ß (mouse monoclonal; Santa Cruz), RSK2 (goat polyclonal antibody; Santa Cruz), CaMKII (rabbit polyclonal that reacts with all isoforms; Santa Cruz), p85 regulatory subunit of PI 3-kinase (rabbit polyclonal; Upstate Biotechnology, Lake Placid, NY), CDK4 (H-22 rabbit polyclonal; Santa Cruz), CDK2 (M2 rabbit polyclonal; Santa Cruz), and casein kinase (CK)1 (R-19 goat polyclonal and non-cross-reactive with CK1
; Santa Cruz). SR vectors that contain the hemagglutinin (HA) tag upstream of rat wild-type CaMKII or nontagged kinase-dead rat CaMKII (ATP binding site mutant, K42M) were kindly provided by Prof. H. Shulmann and Dr. A. Hudmon (Stanford University, Stanford, CA). The pBABE puro/HA-tagged GLUT4 retroviral plasmid (21) was a kind gift from Prof. D. James (Garvan Institute of Medical Research, Sydney, New South Wales, Australia).
Synthesis of CK59
A 2, 6, 9-tri-substituted purine analog of olomoucine, CK59, was synthesized (Fig. 1
). CK59 structure was based on comparison of olomoucine (Fig. 1) with its inactive isomer as well as the need to generate a compound that could be immobilized on a resin to enable binding of potential kinases after incubation with cell lysates. CK59 was synthesized with an alkyl spacer arm on its benzene ring for resin attachment to effectively clear the kinase pocket. The length of this required spacer arm was based on structural analysis of CDK2 in complex with olomoucine (22). The molecular weight of synthesized CK59 was 435.46 g/mol with an empirical formula C21H37N7O3. The scheme for synthesis of CK59 is supplied as an online supplement, published as supplemental data on The Endocrine Society’s Journals Online web site at http://endo.endojournals.org.
View larger version (12K):
[in this window]
[in a new window]
|
FIG. 1. Structure of CK59 for resin attachment to screen binding proteins. Structures of ATP and olomoucine taken from Furet et al. (43 ). *, IC50 for CDK1 (43 ).
|
|
Preparation of CK59 affinity resin
Mini-Leak Low (Kem-En-Tec; Helena Laboratories, Mt. Waverley, Victoria, Australia), a divinyl sulfone-activated matrix of spherical 6% (wt/vol) agarose beads, was used for immobilization of CK59 according to the manufacturer’s instructions. The typical capacity for protein on Mini-Leak Low is 10–20 mg protein per milliliter matrix with a 70–90% (wt/vol) coupling yield. Briefly, CK59 was cleaved with 95% (vol/vol) trifluoroacetic acid for 2 h at 22 C exposing its C6 NH group for coupling to Mini-Leak Low-activated matrix. Approximately 40 mg/ml (23 mM) CK59 were coupled to the matrix to obtain maximal exposure of the immobilized peptide. The coupling yield was determined by measurement of OD at 280 nm before and after coupling. The matrix immobilized to CK59 (termed CK59 resin in this paper) was stored at 4 C in 1x Tris-buffered saline (pH 7.5) containing 0.02% (wt/vol) sodium azide.
Cell culture
3T3-L1 fibroblasts obtained from the American Type Culture Collection (Manassas, VA) were passaged as preconfluent cultures in DMEM (Sigma, St. Louis, MO) containing 10% (vol/vol) fetal bovine serum (FBS; CSL, Parkville, Victoria, Australia), 50 IU/ml penicillin, and 50 µg/ml streptomycin (DMEM/FBS). Cells for differentiation were maintained at postconfluence for 2 d and then induced to differentiate by the addition of DMEM containing 10% FBS, 2 µg/ml insulin, 0.25 µM dexamethasone, and 0.5 mM 3-isobutyl-1-methylxanthine. After 3 d, the induction medium was replaced with fresh DMEM containing 10% FBS and 2 µg/ml insulin for another 3 d. Adipocytes were maintained in DMEM with 5% FBS for 2 d thereafter. Adipocytes were used 8–12 d after initiation of differentiation, after which time more than 80% of fibroblasts had differentiated into mature adipocytes. CHOT cells that stably overexpress approximately 1 x 106 human IRs per cell (23) were kindly provided by Dr. W. Rutter (University of California, San Francisco, San Francisco, CA) and maintained in -MEM containing 5% FBS, 50 IU/ml penicillin, and 50 µg/ml streptomycin. HEK293T cells were grown in DMEM that contained 5% FBS, penicillin, and streptomycin.
[35S]-metabolic labeling and affinity isolation of CK59 resin-binding proteins
Cells were metabolically labeled with 70% L-[35S]Met and 30% L-[35S]Cys (100 µCi per 10-cm dish) (PRO-mix; Amersham, Aylesbury, UK) in Cys and Met-free DMEM media (ICN Biomedicals, Sydney, Australia) supplemented with 2% (vol/vol) dialyzed fetal calf serum, 2 mM glutamine, and 0.069 g/liter L-proline (the latter required only for CHOT cells) overnight at 37 C. For studies investigating the ATP and CK59 dependence of protein interaction or binding specificity of the CK59 resin, cellular lysates were incubated with either free ATP (0.1, 1, or 5 mM) or CK59 (25, 100, or 250 µM). Cleared whole-cell lysates were incubated overnight on a rotating wheel at 4 C with 50 µl of CK59 resin per 10-cm dish of cells. Affinity isolated resin-binding proteins were then washed three times in 1 ml chilled PBS (pH 7.4) before analysis by SDS-PAGE [7.5 or 10% (wt/vol) acrylamide] under reducing conditions and autoradiography using Amplify fluorographic reagent (Amersham).
N-terminal peptide sequencing and mass spectroscopic analysis of CK59 resin-binding proteins
Affinity-isolated CK59 resin-binding protein bands separated by reducing SDS-PAGE and visualized with 0.03% (wt/vol) Coomassie R250 staining were excised and dried in a Speedyvac Plus (Savant, Hicksville, NY) and digested with modified sequencing grade trypsin (Roche, Stockholm, Sweden). In-gel tryptic digests were analyzed by Proteomics International (Perth, Western Australia, Australia) or in-house by Mr. D. Whelan. Protein identification by peptide mass fingerprinting was performed and mass spectra were analyzed against the NCBI protein database by Proteomics International.
Retroviral infection into 3T3-L1 adipocytes, cell sorting, and green fluorescent protein (GFP) selection
3T3-L1 fibroblasts were infected with retrovirus from BOSC23 packaging cells transfected with GFP- and HA-tagged CaMKII constructs in the replication-incompetent retroviral vector, pLXIN (CLONTECH, Palo Alto, CA). The pLXIN/GFP-HA-CaMKII constructs (6–8 µg of DNA per 10 cm dish) were transfected into BOSC23 cells using FuGENE transfection reagent (Roche). Medium was supplemented to 10 ml, and virus was harvested 48 h later and stored at –80 C. Virus supernatant was used to infect 3T3-L1 fibroblasts at 50–60% confluence. G418 (0.8 mg/ml; Life Technologies, Inc., Paisley, UK) selection was placed on cells 72 h after infection. Selection was further enhanced by flow cytometric selection of fluorescent cells on three successive passages, by which stage all cells were fluorescent. Cells were used as mixed populations. Parallel infections using virus from BOSC23 cells transfected with pLEIN (pLXIN vector encoding enhanced GFP; CLONTECH) were performed as a control. For all experiments described, 3T3-L1 cells were used in the differentiated state.
Glucose uptake assay
Glucose transport was measured as 2-deoxy-[U-14C] glucose uptake, as described previously (24). Adipocytes between d 8 and 12 postdifferentation in 24-well plates were washed twice in Krebs-Ringer bicarbonate buffer [25 mM HEPES (pH 7.4) containing 130 mM NaCl, 5 mM KCl, KH2PO4, 1.3 mM MgSO4.7H2O, 25 mM NaHCO3, and 1.15 mM CaCl2] supplemented with 1% (wt/vol) RIA-grade BSA and 2 mM sodium pyruvate. Adipocytes were equilibrated for 90 min at 37 C before insulin addition or for 30 min before CK59 or olomoucine addition. Insulin (Actrapid; Novo Nordisk, Baulkham Hills, New South Wales, Australia) was added over a concentration range of 0.7–70 nM for 30 min at 37 C. Olomoucine or CK59 (0–500 µM) was added to adipocytes for 90 min before the addition of insulin. Stocks of olomoucine and CK59 were prepared in either dimethylsulfoxide (DMSO) or methanol. Control incubations included these vehicle additions. Uptake of 50 µM 2-deoxyglucose and 0.5 µCi 2-deoxy-[U-14C] glucose (NEN Life Science Products, PerkinElmer Life Sciences, Boston, MA) per well was measured over the final 10 min of agonist stimulation and analyzed by scintillation counting.
GLUT4 translocation assays
Plasma membrane lawn assay.
Cell surface expression of GLUT4 was determined using the plasma membrane lawn assay as described previously (25, 26). 3T3-L1 fibroblasts stably overexpressing GFP-HA-CaMKII wild-type (WT), GFP-HA-CaMKII K42M, or GFP-HA and noninfected fibroblasts were grown on glass coverslips in 6-well plates and differentiated into adipocytes as detailed above. After 8–12 d after differentiation, adipocytes were serum starved for 18 h in DMEM containing 0.5% FBS. Cells were washed twice in Krebs-Ringer bicarbonate buffer (pH 7.4) and equilibrated for 90 min at 37 C before insulin (100 nM) addition or for 30 min before CK59 (100 µM) or olomoucine (100 µM) addition. After treatments, adipocytes were washed in 0.5 mg/ml poly-L-lysine in PBS, shocked hypotonically by three washes in 1:3 (vol/vol) membrane buffer [30 mM HEPES (pH 7.2), containing 70 mM KCl, 5 mM MgCl2, 3 mM EGTA, and freshly added 1 mM dithiothreitol (DTT) and 2 mM phenylmethylsulfonyl fluoride] on ice. The washed cells were then sonicated using a probe sonicator (Microson, Misonix, Farmingdale, NY) at setting 0 in 1:1 (vol/vol) membrane buffer on ice to generate a lawn of plasma membrane fragments that remained attached to the coverslip. The fragments were fixed in 2% (wt/vol) paraformaldehyde in membrane buffer for 20 min at 22 C and the fixative quenched by 100 mM glycine in PBS. The plasma membrane fragments were then blocked in 1% (wt/vol) skimmed milk powder (Blotto) in membrane buffer for 60 min at 22 C and immunolabeled with an in-house rabbit affinity purified anti-GLUT4 polyclonal antibody (clone R10, generated against a peptide encompassing the C-terminal 19 amino acids of GLUT4) and Alexa 488 goat antirabbit secondary antibody (1:200; Molecular Probes, Eugene, OR). Coverslips were mounted onto slides using FluoroSave reagent (Calbiochem, La Jolla, CA) and imaged using an OptiScan confocal laser-scanning immunofluoroscence microscope (OptiScan, Waverley, Victoria, Australia). Data were analyzed using ImageJ (National Institutes of Health, Bethesda, MD) imaging software. At least six fields were examined within each experiment for each condition, and the confocal microscope gain settings over the period of experiments were maintained to minimize between-experiment variability.
Intact cell GLUT4 translocation assay.
3T3-L1 fibroblasts that stably overexpress wild-type GFP-HA-CaMKII WT, mutant GFP-HA-CaMKII K42M, or GFP-HA were infected at 50% confluence with retrovirus from Plat-E packaging cells (27) transfected with HA tagged-GLUT4 in the replication-incompetent retroviral vector, pBABE puro, as detailed previously (21). Puromycin (2 µg/ml) selection was placed on cells 48 h after infection. Cells for differentiation into adipocytes were maintained at postconfluence for 2 d and then induced to differentiate as described above. Twenty-four hours before the translocation assay, adipocytes were seeded into gelatin-coated black, clear-bottom, 96-well plates. Cells were serum starved with DMEM (serum and bicarbonate free) containing 0.2% (wt/vol) BSA and 0.02 M HEPES (pH 7.4) for 2 h, followed by the addition of insulin (200 nM) at different time points at 37 C. Cells were then fixed with 3% (wt/vol) paraformaldehyde on ice for 15 min and for a further 30 min at 22 C and quenched with 50 mM glycine for 5 min at 22 C. Cells were washed with PBS and then blocked in the absence of 0.1% (wt/vol) saponin (to measure HA-GLUT4 moving to the cell surface) or presence of saponin (to measure total cellular amount of HA-GLUT4). The primary antibody, anti-HA.11-purified monoclonal antibody (2 µg/ml; Covance, Vienna, VA), or mouse IgG1
MOPC21 as a control nonrelevant isotype-specific antibody (2 µg/ml; Sigma) in PBS containing 2% normal swine serum was added for 45 min at 22 C. The cells were washed with PBS and blocked again in the presence of saponin (to permeabilize all cells so that the background labeling of secondary antibody is similar for all cells). Cells were incubated for 45 min at 22 C in the dark with the secondary antibody (20 µg/ml Alexa 488-conjugated goat antimouse fluorescein isothiocyanate antibody; Molecular Probes) to detect anti-HA.11 antibody. After extensive washing with PBS, the fluorescence intensity (excitation 485 nm, emission 520 nm) was measured using a Fusion analyzer (PerkinElmer). The percentage of HA-tagged GLUT4 at the cell surface was calculated as detailed previously (21).
Cell lysis and immunoblotting analysis
Differentiated 3T3-L1 adipocytes, in 10-cm cell culture dishes, were serum starved for 16 h at 37 C in DMEM containing 0.5% FBS. Adipocytes were pretreated with or without free CK59 (0–500 µM) diluted in serum-starvation medium for 2 h at 37 C and then stimulated with 100 nM insulin for 10 min at 37 C. After washing twice with ice-cold PBS (pH 7.4) containing 0.1% (wt/vol) BSA, cells were solubilized in a modified radioimmunoprecipitation assay lysis buffer [50 mM Tris-HCl (pH 7.4) containing 1% (vol/vol) Nonidet P-40, 0.25% (wt/vol) deoxycholate, 150 mM NaCl, 1 mM EGTA and freshly added phosphatase inhibitors; 2 mM Na3VO4, 100 mM NaF, 10 mM NaPPi and protease inhibitors; 2 mM phenylmethylsulfonyl fluoride, 10 µg/ml aprotinin, 10 µg/ml leupeptin, and 1 µg/ml pepstatin]. After 30 min at 4 C, the cell lysates were cleared by centrifugation at 8800 x g for 20 min at 4 C to remove cellular debris. Thirty microliters of protein A-Sepharose [50% (wt/vol)] with 1–5 µg anti-IR ß-subunit (CT-1 mouse monoclonal; gift from Dr. K. Siddle, University of Cambridge, Cambridge, UK) or anti-IRS1 (A-19 rabbit polyclonal; Santa Cruz) antibody were incubated with 400–500 µg cell lysate and the mixture incubated overnight on a rotating wheel at 4 C. After washing immunoprecipitates once with 1 ml ice-cold high-salt buffer [PBS (pH 7.4) supplemented with 0.2% (vol/vol) Triton X-100 and 0.5 M NaCl] and twice with 1 ml ice-cold low-salt buffer [PBS (pH 7.4) supplemented with 0.2% (vol/vol) Triton X-100], samples were separated by SDS-PAGE [7.5% wt/vol) acrylamide] under reducing conditions. Whole-cell lysates (20–40 µg) and affinity-isolated CK59 resin-binding proteins (as detailed above) were also separated by SDS-PAGE. Transfer of proteins [Towbin transfer buffer (pH 8.2)] onto nitrocellulose membranes was followed by immunoblotting with relevant antibodies.
To detect phosphorylation of either IR ß-subunit or IRS1 on tyrosine, membranes were blocked in 5% (wt/vol) BSA diluted in Tris-buffered saline (pH 7.4) containing 0.05% (vol/vol) Tween 20 (TBST), probed with 1 µg/ml PY20 or PY99 in BSA/TBST, followed by horseradish peroxidase (HRP)-conjugated antimouse antibody (1:10,000). To detect protein levels, membranes were washed with TBST for 2–3 h at 22 C and reprobed with either IR ß-subunit (CT-1 monoclonal at 1 µg/ml) or IRS1 C-terminal polyclonal antibody followed by HRP-conjugated antirabbit antibody (1:5000). Phosphorylation of Akt, p42/p44-MAPK, RSK2, CaMKII, or p38-MAPK was analyzed by immunoblotting proteins from whole-cell lysates resolved by SDS-PAGE and transferred to nitrocellulose membranes with the desired antibody overnight at 4 C in either 3% Blotto or 5% BSA in TBST. Membranes were probed with the relevant HRP-conjugated secondary antibodies for 90 min on a rocking platform at 22 C, followed by enhanced chemiluminescence for detection (Supersignal; Pierce, Rockford, IL).
CaMKII phosphorylation analysis
GFP-HA-CaMKII WT or GFP-HA-CaMKII K42M was immunoprecipitated with 5 µg/ml anti-HA antibody (clone 12CA5, in-house) and 50% (wt/vol) slurry of Protein A-Sepharose beads (Zymed, San Francisco, CA) by incubation overnight on a rotating wheel at 4 C with whole-cell lysates. The beads were pelleted by centrifugation at 12,500 x g for 1 min at 4 C and washed once in 1 ml of high-salt buffer and twice in 1 ml of low-salt buffer on ice. CaMKII activity was measured after incubating HA immunoprecipates in 20 µl CaMKII activation buffer [50 mM Tris-HCl (pH 7.5) containing 10 mM MgCL2, 2 mM DTT, 0.1 mM Na2EDTA, 2 mM CaCl2, 100 µM ATP, and 1.2 µM CaM] for 10 min at 30 C (to activate CaMKII by autophosphorylation). Assay was initiated by addition of CaMKII immunoprecipitates to a reaction mix of 100 µM Multitide 19S substrate peptide (KKRKAALRRWAPLAPRQMSFDC; synthesized in-house) (28), 100 µM ATP, and 100 µCi/µmol 
32P-ATP [in 50 mM Tris-HCl (pH 7.5), 10 mM MgCl2, 2 mM DTT, and 0.1 mM Na2EDTA] per sample for 5 min at 30 C. Triplicate 10-µl aliquots were spotted onto P81 filters (Whatman, Middlesex, UK), washed three times for 15 min in 0.75% (wt/vol) phosphoric acid, rinsed in 100% (vol/vol) ethanol, dried, and 32P incorporation into peptide measured in a liquid scintillation counter. Blank values, in the absence of 19S peptide, were subtracted from each result. In vitro autophosphorylation assays were carried out by incubating HA immunoprecipitates in 20 µl CaMKII activation buffer and the reaction started by the addition of 100 µCi 32P-ATP per sample for 10 min at 30 C. All reactions were stopped by adding 20 µl reducing sample buffer [0.19 M Tris (pH 6.8), 100 mM DTT, 6% (wt/vol) sodium dodecyl sulfate, 20% (vol/vol) glycerol, and 0.01% (vol/vol) bromophenol blue] and heating samples at 100 C for 5 min. Phosphorylated proteins were separated by SDS-PAGE and stained with Coomassie R250. Phosphorylated bands were visualized by autoradiography and the level of autophosphorylation quantitated by Cerenkov counting. To compare the relative amounts of GFP-HA-CaMKII in the HA immunoprecipitates during the in vitro kinase and autophosphorylation assays, anti-HA immunoprecipitates were blotted with GFP (CLONTECH), CaMKII (Santa Cruz), or HA Y-11 probe (Santa Cruz) antibodies and visualized by HRP/ECL detection.
Statistical analysis
Statistical analyses were performed using unpaired Student’s t test.
|
Results
|
|---|
Our initial studies (Fig. 2) show that the cyclin-dependent kinase inhibitor, olomoucine, markedly inhibited insulin stimulation of glucose transport by up to 80% over a range of insulin concentrations in 3T3-L1 adipocytes. Its inactive structural isomer was without effect (Fig. 2). Another structurally related cyclin-dependent kinase inhibitor, roscovitin, also inhibited insulin-stimulated glucose transport (data not shown). Olomoucine had no affect on basal glucose transport rate. It also had no affect on insulin stimulation of Akt phosphorylation or PKC/ activation (data not shown), the last known kinase steps in the signaling cascade for insulin stimulation of glucose transport. Because CDK2 and -5, the recognized targets of olomoucine inhibition, are either inactive or only weakly active in adipocytes (19, 20) and olomoucine was found not to affect insulin stimulation of Akt phosphorylation (data not shown), the data shown in Fig. 2 suggested the possibility that olomoucine inhibited an insulin-sensitive kinase acting downstream of Akt in the pathway to stimulation of glucose transport. To examine this possibility, an analog of olomoucine was synthesized, CK59, that could be coupled to a resin and used to affinity isolate binding proteins. The structure of CK59 is shown in Fig. 1
. The effect of CK59 on insulin stimulation of glucose transport in 3T3-L1 adipocytes was examined over a range of concentrations from 0 to 500 µM (Fig. 3). CK59 displayed a two-phase profile of inhibition of insulin-stimulated glucose transport, with an initial effect at concentrations less than 10 µM to inhibit insulin-stimulated glucose transport by up to 40% (n 
4). A secondary response at 20–100 µM CK59 resulted in almost complete inhibition of insulin-stimulated glucose transport (Fig. 3). There was no effect of either CK59 or olomoucine on basal glucose uptake (not shown). The CK59 effects were not batch specific because a second analog preparation also inhibited insulin stimulation of glucose transport in 3T3-L1 adipocytes (Fig. 3). The data in Fig. 3 demonstrated significant inhibition (by 37%, n = 6) of insulin-stimulated glucose transport at 0.1 µM CK59.
Because insulin mediates the majority of its effect on glucose transport in 3T3-L1 adipocytes via stimulation of GLUT4 translocation, the effects of olomoucine and CK59 on insulin-stimulated GLUT4 translocation were examined (Fig. 4). Consistent with the effects of olomoucine and CK59 on glucose transport (Fig. 3), both compounds at 100 µM markedly inhibited insulin-stimulated GLUT4 translocation (by 35 and 50%, respectively, when compared with vehicle at 100%, n = 3).
View larger version (14K):
[in this window]
[in a new window]
|
FIG. 4. Olomoucine and CK59 impair insulin-stimulated GLUT4 translocation in 3T3-L1 adipocytes. 3T3-L1 fibroblasts were grown on glass coverslips in 6-well plates and differentiated into adipocytes as described in Materials and Methods. Adipocytes were serum starved overnight and preincubated in Krebs-Ringer bicarbonate buffer for 2 h at 37 C before treatment with or without 100 µM olomoucine or CK59 for 90 min followed by incubation in the absence (clear bars) or presence (black bars) of insulin (100 nM) stimulation as indicated for 30 min at 37 C as detailed in the legend for Fig. 3. Plasma membrane fragments were prepared and analyzed for GLUT4 protein as explained in Materials and Methods. Each data point is expressed as the mean of six to eight field measurements ± SEM. This figure is representative of three independent experiments.
|
|
The effects of CK59 on insulin-stimulated protein phosphorylation were examined at 10 µM (first phase inhibition of glucose transport) and 100 µM (second phase inhibition) CK59 concentrations (Fig. 5). CK59 at 10 or 100 µM concentration had no effect on insulin-stimulated IRS1 and IR ß-subunit tyrosine phosphorylation, Akt phosphorylation on either T308/309 or S473/474 (the latter not shown here), RSK2 phosphorylation at S227 (phosphoinositide-dependent kinase-1 phosphorylation site), and p38 MAPK phosphorylation at T180/Y182 (RSK2 and p38 MAPK not shown). Insulin-stimulated p44/p42 MAPK phosphorylation at T202/Y204 was unaffected by 10 µM CK59 concentration but was markedly inhibited with 100 µM CK59. These studies suggest the first-phase inhibition of glucose transport by CK59 is not accounted for by effects on phosphorylation of these signaling proteins and therefore, by inference, may involve a kinase downstream of Akt.
View larger version (42K):
[in this window]
[in a new window]
|
FIG. 5. Effects of free CK59 pretreatment on insulin signaling proteins in 3T3-L1 adipocytes. 3T3-L1 adipocytes (shown here) or CHOT cells (data not shown), in 10-cm dishes, were serum starved overnight and pretreated with or without 10, 100, or 500 µM free CK59 or 10 µM CK59' (a separate preparation of CK59) in serum-starvation media for 90 min, followed by an acute insulin (100 nM) stimulation as indicated at 37 C. Adipocytes were solubilized in a modified radioimmunoprecipitation assay buffer, and whole cell lysates and immunoprecipitates of IR ß-subunit or IRS1 were prepared and analyzed by reducing SDS-PAGE and immunoblotting with the indicated antibodies to detect phosphorylation levels of tyrosine residues, Akt on Thr 308, p42/p44-MAPK on Thr 202/Tyr 204, and RSK2 on Ser 227 and p38-MAPK on Thr 180/Tyr 182 as detailed in Materials and Methods. In each case, experiments were performed at least four times and images shown are from one representative experiment.
|
|
To identify the kinase(s) inhibited by CK59 in the pathway to insulin stimulation of glucose transport, CK59 was coupled to Mini-Leak Low resin and used to affinity isolate binding proteins. CK59 resin was first incubated with supernatants prepared from [35S]-methionine/cysteine-labeled 3T3-L1 adipocytes (Fig. 6). Affinity-isolated binding proteins separated on SDS-PAGE were compared with those from cell lysates incubated with Mini-Leak Low resin alone as control (Fig. 6, left-hand side). The ability of free CK59 to compete with resin for protein binding was assessed as a measure of binding affinity. Thus, after lysate incubation and washing, the resins were incubated with free CK59 at concentrations up to 250 µM. Several proteins were found to bind to the CK59 resin (Fig. 6). The most abundant, as well as the highest affinity, binding protein had an apparent molecular mass of approximately 55 kDa (Fig. 6, band 4). It was not competed off with either 25 or 100 µM free CK59. Competition was observed only at 250 µM free CK59. This protein was also not competed off the CK59 resin in the presence of 0.1, 1, or 5 mM free ATP (data not shown). The 55-kDa protein band was also the major binding species in our parallel analyses of [35S]-labeled cell lysates from CHOT cells (data not shown). Of the other proteins that bound the CK59 resin, two proteins, 120–130 and 90–100 kDa (Fig. 6, bands 2 and 3, respectively), bound the CK59 resin in higher abundance after insulin treatment of cells (not shown); suggesting a modification such as phosphorylation state might regulate their association with the CK59 resin. The binding of three other proteins, the high-affinity 55-kDa binding protein, a 40- to 45-kDa protein, and a 32- to 36-kDa protein (Fig. 6, band 4, 5, and 6, respectively) were also unaffected by treatment of cells with insulin (not shown).
View larger version (60K):
[in this window]
[in a new window]
|
FIG. 6. . [35S]-metabolically labeled proteins from 3T3-L1 adipocytes bind specifically to CK59 resin. Whole-cell lysates from [35S]-metabolically labeled adipocytes were incubated with Mini-Leak Low resin alone (control; first four lanes), or Mini-Leak Low resin coupled to CK59 (last four lanes) in the presence of increasing concentration of free CK59 as indicated. Samples were resolved by reducing SDS-PAGE and visualized via autoradiography using Amplify fluorographic reagent. This experiment was performed with both CHOT cells (n = 4, data not shown) and 3T3-L1 adipocytes (n = 5) times. Similar results were obtained in five separate experiments with two different preparations of CK59 resin. Each arrow with its representative number indicates a CK59 resin-binding protein band, which was then excised and sent to Proteomics International for identification by mass spectrometry.
|
|
Immunoblotting analysis of published and potential olomoucine-sensitive kinases was performed as a preliminary attempt to identify some of the CK59 resin-binding bands. 3T3-L1 adipocyte lysate proteins binding to the CK59 resin included: CaMKII (50–60 kDa), GSK3/ß (47–51 kDa), phospho-p42/p44 MAPK (T202/Y204; 42–44 kDa), and CDK5 (33 kDa) (Fig. 7). All except CDK5 and CaMKII were competed off the CK59 resin with 25 µM free CK59. CDK5 was competed off the CK59 resin with 250 µM CK59, whereas CaMKII remained bound with high affinity. No binding was detected for the IR ß-subunit (95 kDa), SHC (66, 52, and 46 kDa), p85 subunit of PI 3-kinase, phospho-Akt (S473 or T308; 60 kDa), CDK4 (34 kDa), and CDK2 (33 kDa) (data not shown).
View larger version (49K):
[in this window]
[in a new window]
|
FIG. 7. Immunoblot assessment of CK59 resin-binding proteins and competition with increasing concentration of free CK59. Affinity-isolated proteins extracted from 3T3-L1 adipocyte supernatants were incubated with the CK59 resin in the presence or absence of increasing concentrations of free CK59 as indicated and separated by reducing SDS-PAGE (7.5 or 10%) followed by Western transfer and immunoblotting with the indicated antibodies to detect the protein levels of CaMKII, GSK3/ß, and CDK5 and phosphorylation level of p42/p44-MAPK as detailed in Materials and Methods. Similar results were obtained in at least four independent experiments.
|
|
Large-scale proteomic analysis of the CK59 resin-binding proteins extracted from both 3T3-L1 adipocytes and CHOT cell lysates (40 x 10 cm dishes) was performed to establish the identity of some of the bound proteins. Coomassie blue staining of CK59 resin-binding proteins were resolved by SDS-PAGE and transferred to nitrocellulose. Major protein bands at 50–60 kDa (Fig. 8), similar to the major resin-binding species isolated from [35S]-metabolic-labeled cell extracts, were isolated for identification by matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry. Peptide mass fingerprinting identified CaMKIIß-like and CaMKII as matches for the major doublet binding species with 23% (z score of 2.12) and 13% (z score of 1.17) sequence coverage, respectively. N-terminal peptide sequencing also matched the peptide (LHDSISEEGF) to CaMKII or CaMKIIß. This was further confirmed by the immunoblot analysis of CaMKII shown in Fig. 7. A number of low-abundance CK59 resin-binding proteins were also identified by MALDI-TOF mass spectrometry analysis, including phosphatidylinositol-4-phosphate 5-kinase type III; phosphoinositide kinase, fyve-containing (PIKfyve); never in mitosis gene A (NIMA)-related kinase 9 with 15 of 113 peptides matched yielding 13% sequence coverage of the protein; RSK2 with nine of 150 peptides matched covering 11% of the protein sequence; and CKI/ with nine of 53 peptides matched covering 14 and 15% of the protein sequencing, respectively. The data from [35S] labeling studies (Fig. 6) suggested that these were low-affinity binding proteins and clearly less of these proteins bound to the CK59 resin. Immunoblotting analysis of CK1 and RSK2 confirmed the binding was of low affinity, being effectively competed off the CK59 resin with 25 and 100 µM free CK59 (data not shown).
View larger version (61K):
[in this window]
[in a new window]
|
FIG. 8. Identification of CaMKII isoforms from CK59 resin-binding proteins by MALDI-mass spectrometry. CHOT (shown here) or differentiated 3T3-L1 cellular supernatants were batch-wise affinity purified on resin alone (control) or resin linked to CK59 (CK59 resin). Proteins were separated by reducing SDS-PAGE (10% Tris-glycine, Invitrogen, Carlsbad, CA) and visualized by Coomassie Blue (R250). In-gel tryptic digests of excised protein bands (from both CHOT and 3T3-L1 adipocyte supernatants) were sent to Proteomics International for protein identification by peptide mass fingerprinting and mass spectra analyzed against the NCBI protein database. NEK9, NIMA-related kinase 9. MWM, Molecular mass marker.
|
|
The MALDI-TOF analysis (Fig. 8) together with the [35S]-metabolic labeling (Fig. 6) and immunoblotting studies (Fig. 7) indicated that the highest affinity and most abundant CK59 binding protein was CaMKII. This protein is thus an attractive candidate kinase through which CK59 mediated its effects on glucose transport. Consistent with this, glucose transport was inhibited at CK59 concentrations below that which known insulin signaling kinase phosphorylation was inhibited (such as p44/p42 MAPK) or unaffected (Akt) (Fig. 5). The effect of free CK59 on CaMKII activity was assessed in vitro (Fig. 9). CK59 inhibited CaMKII activity by 32% at 0.1 µM and 55% at 10 µM (n = 5) with a similar concentration response curve to that observed for glucose transport. Olomoucine, at 10 and 100 µM, also inhibited CaMKII activity (by 42 and 60%, respectively, when compared with nontreated cells at 100%, n = 3, data not shown).
View larger version (10K):
[in this window]
[in a new window]
|
FIG. 9. CK59 inhibits CaMKII activity in 3T3-L1 adipocytes that overexpress wild-type GFP-HA-CaMKII WT. 3T3-L1 adipocytes that overexpress wild-type GFP-HA-CaMKII WT were solubilized and immunoprecipitated with anti-HA antibody (clone 12CA5) overnight at 4 C. Increasing concentrations of free CK59 (0–200 µM) were added to the anti-HA immunoprecipitates during the autophosphorylation/activation reaction mix for 10 min at 30 C. Immunoprecipitates were then assessed for in vitro CaMKII kinase activity toward exogenous Multitide 19S peptide (100 µM) in the presence of 100 µCi 32P-ATP as described in Materials and Methods. The graph shown is representative of five independent experiments and each data point is the mean ± SEM of triplicate aliquots. Negative controls also included anti-HA immunoprecipitates from 3T3-L1 adipocytes that overexpress GFP-HA tag only (data not shown).
|
|
To address the possible involvement of CaMKII in insulin stimulation of glucose transport, a knockdown approach was used. Small interfering RNA approaches were considered; however, because multiple isoforms of CaMKII were expressed in 3T3-L1 cells (see, for example, Fig. 8), a knockdown mutagenic strategy was preferred. 3T3-L1 fibroblasts were infected with retrovirus encoding GFP and HA-tagged wild-type CaMKII (GFP-HA-CaMKII WT) or a kinase-dead ATP binding site mutant K42M CaMKII (GFP-HA-CaMKII K42M) to test the level of CaMKII activity. Fibroblasts were selected for GFP fluorescence by cell sorting. Given caveats that in vitro kinase activity measurement does not always reflect in vivo activity, we surprisingly found that HA-tagged immunoprecipitates from adipocytes stably overexpressing the mutant CaMKII K42M displayed significant activity toward an exogenous peptide substrate (Fig. 10A); although it was 30% lower (P < 0.01, n = 4) than that seen with immunoprecipitates from wild-type CaMKII-infected adipocytes. Analysis of CaMKII autophosphosphorylation status of these HA-tagged immunoprecipitates revealed the explanation for this kinase activity because not only was the GFP-HA-CaMKII immunoprecipitated but also the endogenous CaMKII (Fig. 10B). This is not unexpected because CaMKII multimerizes in cells (29, 30). Nevertheless, autophosphorylation of endogenous CaMKII complexed with GFP-HA-CaMKII K42M was markedly reduced, despite the expression level of the mutant GFP-HA-CaMKII K42M, as assessed by immunoblotting, being equivalent or higher than the wild-type GFP-HA-CaMKII WT (Fig. 10C).
View larger version (19K):
[in this window]
[in a new window]
|
FIG. 10. In vitro CaMKII autophosphorylation and kinase activity of 3T3-L1 adipocytes that overexpress mutant GFP-HA-CaMKII K42M are reduced compared with wild-type GFP-HA-CaMKII WT. 3T3-L1 adipocytes that overexpress wild-type GFP-HA-CaMKII WT, mutant GFP-HA-CaMKII K42M, or GFP-HA vector alone were solubilized and immunoprecipitated with anti-HA antibody (clone 12CA5) overnight at 4 C. In vitro Multitide 19S peptide phosphorylation (A) results were quantitated as counts per minute (cpm), which have been background corrected, expressed as a percentage of the basal activity of wild-type GFP-HA-CaMKII WT (clear bars), and are the mean ± SEM of four separate experiments. Immunoprecipitates of cells overexpressing GFP-HA alone had no multitude phosphorylation activity and are not shown. Basal autophosphorylation activity is shown as a representative 32P autoradiograph of six independent experiments (B). Protein levels of immunoprecipitated GFP-HA-CaMKII WT and GFP-HA-CaMKII K42M were assessed by immunoblotting anti-HA immunoprecipitates with CaMKII polyclonal antibody (C). As a negative control, anti-HA immunoprecipitates were prepared from either noninfected 3T3-L1 adipocytes (L1) or 3T3-L1 adipocytes that overexpress GFP-HA tag only. **, P < 0.01 when compared with basal GFP-HA-CaMKII WT kinase activity.
|
|
The effects of expression of GFP-HA-CaMKII WT and mutant GFP-HA-CaMKII K42M on glucose transport and GLUT4 translocation were examined (Fig. 11). Insulin-stimulated glucose transport was significantly decreased in 3T3-L1 adipocytes expressing GFP-HA-CaMKII K42M (1.7 ± 0.5-fold at 7 nM insulin over basal, n = 5), compared with noninfected adipocytes (4.7 ± 0.6-fold at 7 nM insulin over basal, n = 5), adipocytes expressing GFP-HA-CaMKII WT (4.3 ± 0.2-fold at 7 nM insulin over basal, n = 5), or cells infected with the GFP-HA-tag vector alone (4.5 ± 0.6-fold at 7 nM insulin over basal, n = 5) (Fig. 11A).
View larger version (18K):
[in this window]
[in a new window]
|
FIG. 11. Insulin-stimulated glucose transport and GLUT4 translocation in 3T3-L1 adipocytes that overexpress wild-type GFP-HA-CaMKII WT or mutant GFP-HA-CaMKII K42M. Noninfected 3T3-L1 adipocytes ( ) or 3T3-L1 adipocytes that stably overexpress GFP-HA-CaMKII WT ( ) or mutant GFP-HA-CaMKII K42M ( ) or GFP-HA tag only ( ) were treated with increasing concentration of insulin (0, 0.7, 7, or 70 nM) for 30 min at 37 C in Krebs-Ringer bicarbonate buffer followed by measurement of glucose transport as detailed in Materials and Methods (A). Basal glucose transport was comparable between noninfected 3T3-L1 adipocytes and 3T3-L1 adipocytes that overexpress GFP-HA-CaMKII WT, mutant GFP-HA-CaMKII K42M, or GFP-HA tag only (data not shown). Results are expressed as fold change relative to basal values and represent mean ± SEM of five independent experiments, within which each point was assayed in triplicate. B, 3T3-L1 adipocytes that overexpress GFP-HA tag only, GFP-HA-CaMKII WT, or GFP-HA-CaMKII K42M were treated with or without 100 nM insulin for 30 min in Krebs-Ringer bicarbonate buffer. Translocation of GLUT4 was assessed by plasma membrane lawn formation and immunofluorescence as described in Materials and Methods. The graph shown is representative of three independent experiments, and each data point is expressed as the mean of six to eight field image measurements ± SEM. C, An exofacial tagged GLUT4 was coexpressed in GFP-HA tag-only-, GFP-HA-CaMKII WT-, or GFP-HA-CaMKII K42M-expressing cells as described in Materials and Methods and GLUT4 translocation measured in an intact cell assay that enables GLUT4-containing vesicle fusion to be distinguished from plasma membrane docking. Anti-HA-GLUT4-specific plasma membrane labeling is expressed as percent of total anti-HA-GLUT4-specific labeling (measured after cell permeabilization). The data for each point represent the mean ± SEM of six cell incubations within the one experiment. The graph shown is representative of two independent experiments. Total anti-HA-specific labeling for the three cell lines was similar [GFP-HA-CaMKII WT, 123027 ± 1980 relative fluorescence units (RFU); mutant GFP-HA-CaMKII K42M, 109453 ± 3078 RFU; GFP-HA, 101617 ± 2986 RFU].
|
|
Two separate approaches were used to assess the effects of expression of the CaMKII constructs on GLUT4 translocation (Fig. 11, B and C). In the first approach, plasma membrane lawns produced after stimulation of cells with insulin were probed for endogenous GLUT4 (Fig. 11B). Despite the significant and consistent inhibition of insulin-stimulated glucose transport in 3T3-L1 adipocytes that overexpress mutant GFP-HA-CaMKII K42M, insulin-stimulated GLUT4 translocation in these cells was not significantly inhibited when compared with adipocytes expressing GFP-HA only. Interestingly, insulin-stimulated GLUT4 translocation in adipocytes expressing GFP-HA-CaMKII WT was increased by 1.9 ± 0.2- or 2.2 ± 0.6-fold (n = 3) in comparison with either adipocytes expressing GFP-HA only or GFP-HA-CaMKII K42M, respectively. A second intact cell approach was used to measure GLUT4 translocation because it is likely that the GLUT4 lawn approach did not discriminate between docked and fused GLUT4-containing vesicles (Fig. 11C). Exofacial HA-tagged GLUT4 was expressed in the CaMKII-expressing cell lines by retroviral infection as previously described for 3T3-L1 cells (21). Cell surface HA-GLUT4 was then measured as described in Materials and Methods after insulin stimulation over 0–15 min time course and expressed as a percentage of the total cellular HA-GLUT4 (Fig. 11C). Cell surface HA-GLUT4 after a 15 min insulin stimulation was similar for each cell line. Therefore, impaired GLUT4 translocation to the cell surface cannot explain the impairment in glucose transport seen in the adipocytes expressing the GFP-HA-CaMKII K42M mutation.
|
Discussion
|
|---|
The present study showed that olomoucine inhibited insulin-stimulated glucose transport in 3T3-L1 adipocytes. This finding was surprising because CDKs such as CDK2 and -5, which are the known targets of olomoucine inhibition, are either not active or have very low activity in fully differentiated cells (19, 20). Olomoucine also had negligible effect on insulin stimulation of either Akt phosphorylation on Thr308 or Ser473 (data not shown), the last defined kinase involved in the signaling pathway to GLUT4 translocation. Therefore, these findings led us to propose that olomoucine inhibited a novel kinase acting downstream of Akt. In an effort to identify this kinase, an analog of olomoucine, CK59, was generated that could be coupled to a resin for affinity isolation of binding proteins. The concentration curve for CK59 inhibition of insulin-stimulated glucose transport showed a two-phase response. The first phase, approximately 40–50% inhibition of glucose transport, occurred at the 0.1- to 1-µM CK59 concentration range. The second phase occurred at concentrations above 20 µM. CaMKII was the only kinase found to bind the CK59 resin with appropriately high affinity, apart from CDK5. Phospho-p42/p44 MAPK, GSK3/ß, CK1, and RSK2 were identified as lower-affinity binding proteins (by immunoblotting). In the present study, CaMKII was established as a prime candidate for a regulatory role in insulin-stimulated glucose transport by the demonstration that CK59 inhibited glucose transport over a similar concentration range as its effects on CaMKII. This was further supported by the finding that cells expressing the ATP binding mutant CaMKII K42M had an impaired insulin-stimulated glucose transport response.
CaMKII is a serine/threonine kinase, which is activated by an increase in cytoplasmic calcium via its association with the ubiquitous calcium receptor, calmodulin. It is a heteromultimer composed of four different chains: , ß, , and that are encoded by distinct genes. The - and ß-subunits are expressed only in neural tissue at approximately 1–2% of total protein in the brain. The - and -subunits are expressed at relatively low levels in a wide variety of tissues at approximately 0.02% of the level of the brain-specific forms (29). The holoenzyme is composed of eight to 12 subunits with molecular masses of 50–65 kDa each mixed randomly (29, 30). All isoforms share approximately 89–93% sequence similarity in their N-terminal kinase and autoregulatory domains. The molecular weight differences between subunits results from a series of inserts C-terminal to the calmodulin-binding domain and within the C-terminal association domain/variable region (29). CaMKII autophosphorylation plays an important role in the regulation of its kinase activity, mediating effects on both catalytic activity and calmodulin binding (29, 30). The neuronal-specific -isoform was used for expression in the studies reported here because we were unable to easily source -isoform cDNA. This did not compromise the data obtained for CaMKII expression because the immunoprecipitation/Western blotting data shown in Fig. 10B demonstrate that the GFP-HA-CaMKII -isoform was incorporated effectively into the heteromultimer with the endogenous enzyme. Furthermore, the kinase-dead mutant integrated into complexes with the endogenous enzyme, which resulted in decreased autophosphorylation of the endogenous enzyme and decreased activity. However, it is acknowledged that an alternative, although less likely, explanation for the apparent decrease in phosphorylation of the endogenous enzyme is that WT expression of the GFP-HA-CaMKII isoform stimulated phosphorylation and activation of the endogenous enzyme.
Our findings suggest a permissive role for CaMKII in glucose uptake rather than a regulatory role because CaMKII activity was not stimulated by insulin (Konstantopoulos, N., and S. L. Macaulay, unpublished data). Previous studies showed that another CaMKII inhibitor, KN-62, inhibited both insulin- and hypoxia-stimulated glucose transport and cell surface GLUT4 expression by 49–54% in isolated rat skeletal muscle (7). The effect of KN-62, like that of CK59 in the present study, was relatively specific because there was no effect on insulin-stimulated IR, IRS1, PI 3-kinase, or Akt activity. Specific calmodulin antagonists (trifluoperazine, orphiobolin A, W13, and W7) have also been demonstrated to inhibit insulin-stimulated GLUT4 translocation in 3T3-L1 adipocytes (31). These inhibitors were reported to prevent formation of phosphatidylinositol 3, 4, 5-trisphosphate. In the present study and that of Brozinick et al. (7), no effect of CK59 or KN-62 on insulin-stimulated Akt phosphorylation was detected and therefore unlikely to be mediated through effects on phosphatidylinositol 3, 4, 5-trisphosphate formation. Thus, the various inhibitor studies together with our CK59 and CaMKII expression studies support a role for CaMKII signaling in insulin-stimulated glucose transport.
How could CaMKII mediate effects on glucose transport? A recent study demonstrated the calcium/ATP-dependent binding of autophosphorylated CaMKII to GST-STX1A in rat brain (32). CaMKII preferentially bound to the open form of STX1A (that also binds VAMP2 and SNAP25), whereas Munc18a bound the closed form. CaMKII displaced Munc18a binding to STX1A and vice versa. This led the authors to propose a priming role for CaMKII in fusion. Isoforms of this same fusion machinery are involved in the regulated fusion of GLUT4 to the plasma membrane, and thus, this represents an attractive site for CaMKII action in our studies. However, we were unable to demonstrate association of CaMKII with STX4 (the SNARE homolog involved in GLUT4 translocation) (Konstantopoulos, N., and S. L. Macaulay, unpublished data) consistent with a previous study (33). Interestingly, we found that recombinant SNAP23 (a SNARE homolog, which together with STX4 makes up a plasma membrane receptor for VAMP2 in the fusion of GLUT4 vesicles) can be phosphorylated by CaMKII (data not shown). It is possible that this phosphorylation could have a regulatory role.
A low abundance CK59 resin-binding protein kinase(s) could contribute to the effects on glucose transport measured in these studies. Alternatively, a combination of CK59 resin-binding proteins may be important in the regulation of glucose transport, suggested by the two phase effect of CK59 concentration on inhibition of insulin-stimulated glucose transport (Fig. 3). However, the [35S]-labeling and antibody binding studies shown in Figs. 6 and 7, respectively, indicate that the binding affinity of lower abundance binding proteins that were detected in the study is insufficient to mediate the initial phase inhibition of glucose transport seen at 0.1 mM CK59. Figure 5 shows phosphorylation of one of the low-affinity binding proteins, p42/p44 MAPK, was affected only by the addition of 100 µM CK59 to cells and not by the lower concentrations. Whereas acknowledging the limitation of assessing enzyme activity of immunoprecipitates, CaMKII activity was significantly inhibited by 0.1 µM CK59 (Fig. 9). Our finding that overexpression of a CaMKII mutant impairs insulin-stimulated glucose transport supports involvement of CaMKII in this process. However, this does not preclude significant involvement of other lower-abundance kinases mediating the CK59-sensitive step(s) in the insulin signaling pathway to glucose transport. PIKfyve was identified as one low-abundance CK59 resin-binding protein. PIKfyve has been shown to translocate from the cytosol to low-density microsomal fraction in response to insulin in 3T3-L1 adipocytes (34, 35). However, if the [35S]-labeled band 1 of appropriate molecular weight in Fig. 6 is PIKfyve, this would suggest the binding affinity is insufficient to mediate these effects. A number of CK59 resin-binding kinases are attractive targets for the second-phase inhibition, including CK1/, NIMA-related kinase 9, or p90 RSK2; however, their apparent binding affinity to CK59 resin in the [35S]-labeling studies (Fig. 6) appears too low to mediate first-phase effects on glucose transport.
One mechanism by which CK59 might inhibit insulin-stimulated glucose transport is via direct binding to GLUT4 itself. The recent findings of Ribé et al. (6) and Antonescu et al. (36) suggest the possibility that ATP analog inhibitors such as SB203580 (p38-MAPK inhibitor) bind to GLUT4 to mediate inhibition of transport. Moreover, Hellwig and Joost (37) demonstrated that a variety of ligands such as cytochalasin B, forskolin, dipyridamole, and 3-isobutyl-1-methylxanthine bind to GLUT4 with affinities in the 0.1- to 100-µM range similar to the inhibitory range of CK59 described here. It is possible that the effects of CK59 on insulin-stimulated glucose transport might also in part be mediated via binding to GLUT4. However, our preliminary studies did not detect binding of CK59 analog resin to GLUT4 in solubilized low-density microsomal fractions from 3T3-L1 adipocytes (data not shown). Furthermore, we were unable to demonstrate binding of CK59 resin to GFP-tagged GLUT4 in lysates from cells overexpressing GFP-GLUT4 (data not shown). It is therefore difficult to ascribe a major role for CK59 binding to GLUT4 in the initial phase of glucose transport inhibition. In contrast, the data from the present study showing CK59 effects on CaMKII kinase activity at the same concentration as the initial phase of glucose transport inhibition, as well as the inhibitory effects of the ATP binding mutant CaMKII K42M on insulin-stimulated glucose transport, support a role for this enzyme in glucose transport. Nevertheless, one cannot rule out the potential binding of CK59 to GLUT4, and this might possibly account for the second phase of inhibition.
Intriguingly, overexpression of wild-type CaMKII increased insulin-stimulated GLUT4 translocation as assessed by plasma membrane lawn assay without increasing insulin-stimulated glucose transport, compared with both mutant CaMKII K42M- or GFP-HA-only-expressing cells, although this effect was not seen in the intact cell assay approach. These data are consistent with either a saturable or permissive effect of CaMKII kinase activity on glucose transport. Overexpression of mutant CaMKII K42M did not affect insulin-stimulated redistribution of glucose transporters to the cell surface, compared with GFP-HA-only-infected cells. Because insulin-stimulated glucose transport was inhibited, these data raise the possibility that the intrinsic activity of glucose transporters that are already present at the cell surface is regulated by CaMKII. The results dissociate decreased glucose transporters at the cell surface from decreased glucose transport activity. One drawback with the plasma membrane lawn assay to determine GLUT4 translocation is that it does not only measure glucose transporters inserted into the plasma membrane bilayers because glucose transporter-containing vesicles docked to the membrane will also be detected.
To assess whether the effect of CaMKII was at the fusion step as opposed to the translocation of GLUT4, 3T3-L1 adipocytes that overexpress the CaMKII constructs were retrovirally infected with HA-tagged-GLUT4 (HA tag on the exofacial loop of human GLUT4), and optical detection of cell surface HA-GLUT4 was then measured in the absence and presence of insulin (21). Consistent with the results obtained from the plasma membrane lawn assay (Fig. 11), insulin-stimulated translocation of HA-GLUT4 to the plasma membrane in mutant CaMKII K42M-expressing cells was comparable at steady state to wild-type CaMKII- and GFP-HA-only-expressing cells (see Fig. 11C). Similar dissociation between translocation and transport has been noted in other studies. For example, SB203580 (p38-MAPK inhibitor) inhibited insulin-stimulated uptake of 2-deoxyglucose and 3-O-methylglucose in L6 myotubes and 3T3-L1 adipocytes without interfering translocation of GLUT1 and -4 to the plasma membrane (38, 39, 40). However, and as indicated above, recent findings (38, 41) suggest a more direct effect of this inhibitor on GLUT4. Clancy et al. (42) reported a marked increase in glucose transport due to protein synthesis inhibitors anisomycin or cycloheximide in 3T3-L1 adipocytes but unaltered amounts of glucose transporter proteins in the plasma membrane. A further possibility is that CaMKII causes association/dissociation of a protein with GLUT4 inserted in the plasma membrane after docking that mediates glucose transport. Despite several efforts to search for GLUT4-interacting proteins, none have been confirmed. A schematic depicting potential sites of CaMKII interaction is supplied within the online supplement.
These studies build on our finding that olomoucine, a selective nucleotide analog inhibitor of CDKs, inhibits insulin-stimulated glucose transport without effect on any of the known insulin-sensitive kinases, including the IR, PI 3-kinase, Akt, and atypical PKC isoforms. The use of analogs based on this inhibitor offered the potential to explore kinases acting downstream of Akt and PKC in the stimulation of glucose uptake. The studies present CaMKII as a major candidate for a permissive role in insulin-stimulated glucose transport. Identification of such kinases can offer alternative and new targets for the development of targeted therapies/intervention on glucose homeostasis and, ultimately, for the treatment of type 2 diabetes.
|
Acknowledgments
|
|---|
We thank Mr. John Bentley and Dr. Neil McKern for advice, Dr. Lindsay Sparrow for N-terminal peptide sequencing and initial identification of CaMKII, Mr. Dean Whelan and Dr. Jeff Gorman for preliminary MALDI-TOF analysis, Dr. Dean Hewish and Ms. Deborah Shapira for fluorescence-activated cell sorting, Professor Howard Shulmann and Dr. Andy Hudmon for providing CaMKII plasmids, Ms. Subdhadhcha for assistance with GLUT4 assays, and Professor David James for providing HA-GLUT4 plasmid. We also thank Dr. Victoria Foletta, Dr. David Segal, and Professor David James for helpful discussions of this manuscript.
|
Footnotes
|
|---|
This work was supported in part by grant from The Diabetes Australian Research Trust.
Current address for N.K.: Metabolic Research Unit, Deakin University, Geelong, Australia 3217.
Disclosure Summary: The authors have nothing to disclose.
First Published Online September 28, 2006
Abbreviations: Akt, Protein kinase B; CaMKII, calcium/calmodulin-dependent protein kinase II; CDK, cyclin-dependent kinase; CK, casein kinase; CK59, calcium/calmodulin-dependent protein kinase II inhibitor 59; DMSO, dimethylsulfoxide; DTT, dithiothreitol; FBS, fetal bovine serum; GFP, green fluorescent protein; GSK, glycogen synthase kinase; HA, hemagluttinin tag; HRP, horseradish peroxidase; IR, insulin receptor; IRS, insulin receptor substrate; MALDI-TOF, matrix-assisted laser desorption ionization-time of flight; PIKfyve, phosphoinositide kinase, fyve-containing; PI 3-kinase, phosphoinositide 3-kinase; PKC, protein kinase C; RSK2, ribosomal kinase 2; SNAP23, synaptosomal-associated protein of 23 kDa; SNARE, soluble N-ethylmaleimide sensitive fusion factor attachment protein receptor; STX, syntaxin; TBST, Tris-buffered saline with 0.05% Tween 20; VAMP2, vesicle-associated membrane protein 2; WT, wild type.
Received April 7, 2006.
Accepted for publication September 20, 2006.
|
References
|
|---|
Top
Abstract
Introduction
) or its inactive structural isomer, isoolomoucine (