* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download 5 x buffer (50TB 25 7 - American Journal of Physiology
Oxidative phosphorylation wikipedia , lookup
Gene expression wikipedia , lookup
Paracrine signalling wikipedia , lookup
Lipid signaling wikipedia , lookup
Expression vector wikipedia , lookup
Ancestral sequence reconstruction wikipedia , lookup
Metalloprotein wikipedia , lookup
Size-exclusion chromatography wikipedia , lookup
G protein–coupled receptor wikipedia , lookup
Signal transduction wikipedia , lookup
SNARE (protein) wikipedia , lookup
Bimolecular fluorescence complementation wikipedia , lookup
Interactome wikipedia , lookup
Protein structure prediction wikipedia , lookup
Magnesium transporter wikipedia , lookup
Two-hybrid screening wikipedia , lookup
Protein–protein interaction wikipedia , lookup
AJP-Endo Articles in PresS. Published on April 9, 2002 as DOI 10.1152/ajpendo.00441.2001 E00441-R1 GLUT-4 containing vesicles are released from membranes by phospholipase D cleavage of a GPI anchor. Søren Kristiansen and Erik A. Richter Copenhagen Muscle Research Centre Department of Human Physiology Institute of Exercise and Sports Sciences University of Copenhagen 13 Universitetsparken 2100 Copenhagen Denmark Correspondence: Erik A. Richter At the above address or by phone + 45 35 32 16 26 or fax + 45 35 32 16 00 or E mail : [email protected] Running title: Glucose transport in skeletal muscle. Keywords: Insulin signaling, glucose metabolism, rat skeletal muscle. Copyright 2002 by the American Physiological Society. 2 ABSTRACT. We have previously developed a cell-free assay from rat skeletal muscle which displayed in vitro glucose transporter 4 (GLUT-4) transfer from large to small membrane structures by the addition of a cytosolic protein fraction. By combining protein fractionation and the in vitro GLUT-4 transfer assay, we have purified a glycosylphosphatidylinositol phospholipase D (GPI-PLD) which induces transfer of GLUT-4 from small to large membranes. The in vitro GLUT-4 transfer was activated and inhibited by suramin and 1,10 phenanthroline (an activator and an inhibitor of GPI PLD activity, respectively). Furthermore, upon purification of the GLUT-4 transporter protein, the protein displayed an elution profile where the molecular weight was related to the charge, suggesting the presence or absence of phosphate. Secondly, by photoaffinity labeling of the purified GLUT-4 with [125I] 3-(trifluromethyl)-3-(m-[125I]iodophenyl) diazirine both labeled phosphatidylethanolamine and fatty acids (constituents of a GPI link) were recovered. Third, by using phase transition of triton X –114, the purified GLUT-4 was found to be partly detergent resistent, which is a known characteristic of GPI-linked proteins. Fourth, the purified GLUT-4 protein was recognized by an antibody raised specifically against GPI links. In conclusion, GLUT-4 containing vesicles may be released from a membrane compartment by action of a GPI PLD. 3 INTRODUCTION. The insulin and muscle contraction-responsive glucose transporter 4 (GLUT-4) containing-vesicles are localized to tubulovesicular elements in the basal state (28). Upon stimulation, the GLUT-4 containing vesicles undergo translocation to the surface membrane. Despite extensive scrutiny the upstream insulin and contraction mediated signaling proteins leading to GLUT-4 translocation are not described in great detail. We have previously focused on immunoprecipitation of low spin membranes containing GLUT-4 protein and looked for other protein(s) associated within the membrane compartment (18). Using this approach, we found that the GLUT-4 protein was associated with an insulin-insensitive phosphatidylinositol (PI) 4- kinase. Furthermore, in a recent published paper (17) we also reported that in vivo insulin stimulation resulted in an increased production of phosphatidic acid in the GLUT-4 immunoprecipitate. Since phosphatidic acid is a known product of phospholipase D (PLD) action, this suggested to us that a PLD could be part of the mechanism leading to GLUT-4 mobilization. For a number of reasons we initially hypothesized that it could be the ADP ribosylation factor (ARF) sensitive PLD1 (14). Due to the restricted experimental models revealing ongoing enzymatic reactions in skeletal muscle, we were prompted to develop an in vitro assay in order to investigate whether ARF1 or 6 and subsequent PLD1 activation could lead to GLUT-4 translocation (17) which was also suggested by Emoto et al. (9). In brief, we reported that adding a cytosolic protein fraction (which may contain upstream signaling molecules such as ARFs) to membranes containing GLUT-4 resulted in a transfer of GLUT-4 proteins from large to small membrane structures. This transfer could be 4 inhibited by neomycin (a PLD inhibitor). Nevertheless, we were not able to replace the unknown cytosolic molecule(s) by adding recombinant myristoylated ARF1 or ARF6. Neither did an ARF1 inhibitory peptide block the GLUT-4 transfer. Taken together, the previous data (17) suggested that PLD could indeed be involved in the GLUT-4 mobilization process, but it was probably not an ARF sensitive PLD1. We have now taken advantage of the in vitro assay and by protein fractionation we have purified a protein fraction which induces in vitro GLUT-4 transfer. This fractionation procedure resulted in co-purification of a PLD which is known to cleave glycosylphosphatidylinositol anchors (7;15). 5 METHODS. We took advantage of the partial in vitro reconstitution of the GLUT-4 translocation mechanism (17). In brief, an enriched GLUT-4 membrane fraction (approximately 120 µg of protein) was isolated by density gradient centrifugation of a post nuclear supernatant. This enriched GLUT-4 fraction was added control buffer or a cytosolic protein fraction (approximately 225 µg of protein) (see below). Small and large membrane structures were then separated by differential centrifugation. The GLUT-4 protein content was determined in the supernatant containing the small membrane structures and in the pellet containing larger membrane structures. Method for isolation of the unknown cytosolic protein. It was found that the unknown cytosolic factor was present in large amounts in human blood. Thus, forty ml of plasma proteins were precipitated with polyethylene glycol (9 % wt/vol) and to the supernatant (10 min, 4000 g, 40C) an equal volume of 250 mM bis tris, pH 6.35 was added. All chromotographic steps were carried out with a Biologic HR workstation (Bio-Rad, Hercules, CA) operating at 50C. The purification process was monitored and controlled with a Biologic HR software system. First, the solution was run (flow 3 ml/min) through a column (2.5 cm x 20 cm) containing Q SepharoseTM fast flow (Amersham Pharmacia Biotech A/B, Buckinhamshire, UK). Non-captured proteins were washed out of the column with 150 ml of 50 mM bis tris, 10 mM NaCl, pH 6.35. Bound protein was eluted with a linear increase in NaCl to a final concentration of 500 mM. Active fractions eluted in a broad peak around 30 mS/cm. Solid NaCl was added to a final concentration of 500 mM to the active fractions 6 which were then subjected to immobilized metal affinity chromatography (IMAC). A column (1.5 cm x 5 cm) prepacked with iminodiacetic acid (Pierce, Rockford, Illinois) was activated with 500 mM zinc acetate and equilibrated with 50 mM tris, 500 mM NaCl, pH 7.5 (flow 1 ml/min). The sample was run through the column, and washed with equilibration buffer. Captured proteins were eluted with equilibration buffer containing 10 mM histidine. Active fractions were then separated by size exclusion chromatography (Sephacryl S-200 HR) on a 2.5 cm x 100 cm column (flow 1 ml/min) using equlibration buffer. The GLUT-4 mobilizing activity eluted in 1-2 fractions. Proteins in the active and neighboring fractions were then separated by SDS polyacrylamide gel (7.5%) electrophoresis. The gel was silver stained using a kit (Amersham Pharmacia Biotech A/B, Buckinghamshire, UK) in order to visualize all proteins. Isolation of the GLUT-4 protein. Approximately 20 g of fresh unfrozen rat skeletal muscle was transferred to 100 ml of homogenization buffer (50 mM tris base 25 mM, NaCl, 0.5 mM phenylmethylsulfonyl flouride, pH 7.4) and blended for 1 min. The homogenate was centrifuged (10 min, 1.500 g, 40C) and the resulting supernatant was centrifuged (30 min, 230.000 g, 40C). The membrane pellet was resuspended in 30 ml of ice cold strip buffer (100 mM NaCO3 pH 10.0) and repelleted. This pellet was resuspended in solubilization buffer (50 mM Mes, 50 mM bis tris, 25 mM NaCl, 0.20 % (vol/vol) triton X-114, pH 5.5) to a final protein concentration of 0.20 mg/ml. The samples were left on ice for 30 min with agitation. The samples were centrifuged (30 min, 230.000 g, 40C) followed by filtration of the supernatant through a 0.25 µm syringe driven filter (Millipore). The supernatant was run (flow 3.0 ml/min) through a 7 chromatography column (2.5 cm x 20 cm) containing Q SepharoseTM fast flow (Amersham Pharmacia Biotech AB, Uppsala, Sweden) equilibrated in solubilization buffer. The column was washed with 100 ml of solubilization buffer. The GLUT-4 protein was eluted with a linear increase in the NaCl concentration (25 mM – 1 M) in solubilization buffer. Fractions containing GLUT-4 were dissolved in a equal volume of 5 mM sodium phosphate buffer pH 6.8 containing 0.2 % (vol/vol) triton X-114 and subjected to a column containing ceramic hydroxyapatite (Bio-Rad, Hercules, CA). Bound GLUT-4 was eluted with a linear increase in sodium phosphate (5-200 mM) containing 0.2% triton X-114. Fractions containing GLUT-4 were captured on a Mono Q (5/5) column (Amersham Pharmacia Biotech) equilibrated with 50 mM tris, 10 mM NaCl, 0.2 % (vol/vol) triton X-114, pH 7.4. The GLUT-4 protein was eluted by a linear increase in the NaCl concentration (10 to 500 mM). Based on silver staining, the fractions contained exclusively the GLUT-4 protein. If the GLUT-4 protein was prepared for [125I] 3-(trifluromethyl)-3-(m-[125I]iodophenyl) diazirine (TID) photo affinity labeling, detergent was removed from the Mono Q column by extensively washing with sodium phosphate buffer, pH 8.5 before the GLUT-4 protein was eluted with an linear increase (0-1 M) in NaCl dissolved in sodium phosphate buffer. Photo affinity labeling of GLUT-4. The GLUT-4 was upconcentrated by centrifugation through an amicon 10 filter (Amicon). Ten µCi 3-(trifluromethyl)-3-(m-[125I]iodophenyl) diazirine (TID) (Amersham Pharmacia Biotech, Buckingham-shire, UK) were added to the sample which was photo affinity labeled for 30 min at 350 nm. (26;29). For the detection of phospholipids, the GLUT-4 sample was loaded on a chomotography column 8 (1.5 cm x 5.0 cm) containing Q SepharoseTM fast flow equilibrated in 5 mM sodium phosphate pH 8.5. The labeled GLUT-4 was loaded on the column, and most of the unlabelled 125I TID was washed out of the column by 50 ml of buffer and at a flow of 1 ml/min operated with a P1 pump device (Amersham Pharmacia Biotech, Buckinghamshire, UK). Fractions of 1 ml were collected by elution with 5 mM sodium phosphate pH 8.5 containing 1 M NaCl. Five µl 6 M HCl and 200 mM sodium nitrite were added to all fractions which were subsequently incubated at 600C for 4 hours. Positive GLUT-4 as well as negative control fractions were added 500 µl of chloroform: methanol (2:1), followed by centrifugation. The chloroform phase was collected and thin-layer chromatography was performed on 20 x 20 cm silica gel plates (Sigma) developed in a solvent of chloroform: methanol: water (65: 25: 4). For the detection of neutral lipids, the photoaffinity labeled GLUT-4 was dried under a stream of nitrogen to remove the volative unlabeled 125[I] TIDE. Two hundred µl 0.1 M NaOH were added to the sample which was then incubated overnight at 40C, and then added 500 µl chloroform: methanol (2:1) and acetic acid to neutralize the NaOH. The chloroform phase was loaded on a TLC plate and developed in solvent of hexane:diethylether:acetic acid (60:30:1). The developed plate was dried and radioactive components were detected with a STORM phosphoimager (Molecular Dynamics). The positions of phospholipid standards were detected by submerging the TLC plate in 10% CuSO4 8% phosphoric acid, and baking the plate for 1h at 1000C. The palmitic acid standard was detected by exposure of dried plates to iodine vapor. 9 SDS polyacrylamide gel electrophoresis and Western blotting. Proteins were separated on SDS polyacrylamide gels. Some gels were silver stained using a commercial kit (Amersham Pharmacia Biotech A/B) or proteins were electrotransferred to PVDF membranes. The PVDF membranes were Ponceaus S stained, destained and blocked with 3% (wt/vol) bovine serum albumin in TS buffer (50 mM tris, 150 mM NaCl, 0.05% (vol/vol) NP40, 0.05 % (vol/vol) Tween-20, pH 7.4) for 1.5 hour at room temperature. The PVDF membranes were then incubated with a primary antibody dissolved in TS buffer containing 1 % (wt/vol) bovine serum albumin and 0.02% NaN3 (the antibodies were a goat anti GLUT-4 IgG, a rabbit anti-cross reacting determinant (CRD) IgG, or a mouse monoclonal anti GPI-PLD IgG purchased from Santa Cruz Biotechnology, Santa Cruz, CA., Glyko, Inc., Novato, CA., or Transduction Laboratories, Lexington, KY., respectively). Antigen and primary antibody complexes were visualized upon binding with a secondary antibody conjugated with alkaline phosphatase (DAKO, Glostrup, Copenhagen, Denmark) and an enhanced chemofluorescense kit (Amersham Pharmacia Biotech A/B). Fluoroscense were detected with a STORMTM phosphoimager (Molecular Dynamics). Statistics. All results are shown as mean values ± standard error (SE) with a significance level at p < 0.05. Paired and un-paired t-tests were used to calculate the p level. 10 RESULTS In this paper we have used a previously developed assay which displays in vitro GLUT-4 transfer from large to small membrane structures (17). In brief, the assay is performed by incubation of a mixture of an enriched GLUT-4 membrane fraction and cytosolic proteins. The GLUT-4 protein content is then determined in the fast pelleting large membrane fraction and in the slow pelleting small membrane fraction, as shown in figure 1, upper panel. In particular, the increase in the GLUT-4 protein content in the small membrane fraction is not caused by adding GLUT-4 protein contained in the cytosol, since no GLUT-4 signal could be detected in the supernatant and pellet in experiments in which cytosol was added alone. The in vitro GLUT-4 assay is dependent upon the prior in vivo physiological condition of the rat. In brief, by adding unfrozen insulin-stimulated (1 min) cytosol, basal or no cytosol, respectively, there is an increase in GLUT-4 transfer from large to small membrane structures in an enriched GLUT-4 fraction isolated from a basal skeletal muscle (figure 1, middle panel). However, unfrozen cytosol from insulin-stimulated unfrozen skeletal muscle was only approximately 10-20% more potent when compared to unfrozen basal cytosol. This order of potency between basal and insulin stimulated cytosol is diminished with frozen cytosol (data not shown). In particular, adding basal or insulin-stimulated (1 min) cytosol to an enriched GLUT-4 membrane fraction from prior insulin-stimulated (10 min) skeletal muscle was not as potent as using basal skeletal 11 muscle as a source of the enriched GLUT-4 donor membrane fraction. We tested whether cytosolic protein fractions prepared from other tissue sources (liver, brain, blood plasma) than rat skeletal muscle could induce in vitro GLUT-4 transfer. We found that human blood plasma and to a lesser extent brain very potently induced in vitro GLUT-4 transfer, as shown in figure 1, lower panel. Thus, human blood plasma seemed to be a good source of the unknown stimulating factor and we therefore set out to identify the unknown stimulating factor found in the plasma. By protein chromatography of plasma we identified elute fractions, which were able to induce in vitro GLUT-4 transfer. First, to the resulting supernatant from polyethylen glykol precipitation, 50 mM of mes, bis-tris, or tris buffer was added to give different pH values (5.8, 6.3 and 7.8). The solutions were subjected to anion as well as cation exchange. In brief, it was found that the unknown factor did bind to a cation resin at pH 5.8 but not at 6.3 and 7.8. The factor did also bind to anion resin at pH 7.8 and at 6.3 with a lower affinity, but not at 5.8 (data not shown). This suggested that the isoelectric point for the protein(s) was between 5.8-6.3. We then performed anion exchange at pH 6.35. The procedure for identification of the unknown protein is outlined in figure 2. Elute fractions able to induce GLUT-4 transfer in the in vitro assay where then collected. Since we have previously observed that the in vitro GLUT-4 transfer is sensitive to the concentration of divalent cations, we performed metal affinity chromatography. Indeed, the fractions which eluted with the chelating amino acid histidine were able to induce GLUT-4 transfer. These fractions were then protein fractionated using a third parameter namely size exclusion chromatography. Interestingly, only 1-2 elute fraction(s) were able to induce GLUT-4 transfer. The elution profile suggested that the native protein had a molecular size higher than the range of 200-250 12 kDa. The active 1-2 fraction(s) and the neighboring non-active fractions were electrophoresed on 7.5 % gel, as shown in figure 2. All proteins were detected by silver staining in order to detect the presence of an extra band in the active fractions when compared to the non-active fractions. Only one extra band (around 110 kDa) appeared in the active fractions and these extra bands were identified by immunoblotting to be GPI PLD (figure 2c). Based on our observations regarding the source for purification (plasma), iso-electric point (5.8-6.3), presence of histidine in the protein, and the molecular weight of the native and denatured protein (200-250 kDa and 110 kDa), and our previous observation of production of phosphatidic acid in GLUT-4 immunoprecipitates, we hypothesized that the unknown protein could be GPI-PLD (20;30). This was tested with a GPI-PLD specific antibody. Indeed, GPI-PLD eluted exclusively in the active fraction(s) able to induce GLUT-4 transfer and the immunblot signal appeared at the same molecular weight as the extra band in the silver stain (data not shown). This suggested that the GLUT-4 containing vesicles could be linked to a large membrane compartment via a GPI link (see figure 9). We wanted to verify this hypothesis. First, we added 10 mM of suramin and 1 mM of 1,10 phenanthroline (an activator (21) and inhibitor (24) of membrane bound GPI PLD activity, respectively) to the GLUT-4 in vitro assay. As shown in figure 3, upper panel, suramin was able to potentiate the effect of plasma. Interestingly, suramin did also pontentiate the GLUT-4 transfer in the absence of plasma proteins, suggesting that some of the GPI-PLD could be associated with the large 13 membrane compartment. Finally, 1 mM of 1,10 phenanthroline did inhibit the in vitro GLUT-4 transfer, figure 3, lower panel. Furthermore, we wanted to purify the GLUT-4 protein in order to chemically verify the co-purification of a GPI link. The procedure for purifying the GLUT-4 protein is outlined in figure 4. In brief, the GLUT-4 protein was isolated from a rat skeletal muscle homogenate. Next, a membrane preparation was obtained by centrifugation and peripheral membrane proteins were stripped off. The integral membrane proteins were solubilized and purified by anion-exchange and hydroxyapatite chromatography. The GLUT-4 transporter was more than 99% pure based on silver staining of proteins in a 10% SDS gel (figure 4, upper panel). Interestingly, the GLUT-4 protein did bind to the positive anion resin at pH 5.5 which is a much lower pH value than the calculated isoelectric point based on the polypeptide sequence. Thus, an extra negative charge must be bound directly to the GLUT-4 protein. In fact, when the elute fraction from the final chromatography Mono Q step was separated on a gel and silver stained the GLUT-4 protein eluted in 3 bands around 50 kDa. All 3 bands were recognized by the anti-GLUT4 IgG. The triplets were not uniformly distributed in the elute fractions since the high molecular GLUT-4 eluted before the low molecular GLUT-4 isotype. The high molecular isotype may then have a higher isoelectric point (carrying relative more positive charges at the same pH) which causes elution prior to the relative more negative low molecular isoform. This could be explained by the presence (non-cleaved) or absence (cleaved) of a negatively charged phosphate molecule as part of a GPI link (see figure 9 in discussion). Furthermore, it may also explain why the GLUT-4 protein could bind to the positive 14 anion resin at a lower pH value than predicted from the calculated isoelectric point. The calculated isoelectric point is exclusively based on the polypeptide sequence and does not include conjugated site-groups. Next, we wanted to see whether we could recover constituents from a GPI link by labeling of all reactive site groups found on the isolated GLUT-4 protein. In brief, the GLUT-4 protein was captured on a chromatography column that was extensively washed with detergent in order to remove all lipids. The GLUT-4 protein was then eluted without detergent and the purified GLUT-4 transporter and its reactive chemical site groups were photo affinity labeled with [125I] TID. In experiments designed for detection of phospholipids, the labeled GLUT-4 protein was again captured on an anion-exchange column, and the excess of free unlabelled [125I] TID was washed away. The GLUT-4 protein was then eluted from the column and all elute fractions were nitrous deaminated. Fractions containing GLUT-4 were then identified. Lipids were then extracted from fractions with GLUT-4 and from fractions devoid of GLUT-4 and separated on a thin layer chromatography plate, as shown in figure 6. A clear radioactive signal appeared in the positive GLUT-4 fractions. No signal was detected in labeled control fractions devoid of GLUT-4 suggesting that the signal originated from the GLUT-4 protein. The signal had a migration pattern similar to phosphatidylethanolamine, an known constituent in a GPI link. Furthermore, for the detection of fatty acids the photo affinity labeled GLUT-4 was hydrolysed before any possible lipids were extracted and separated on a thin layer chromatography plate. Interestingly, now 2 radioactive bands appeared with a migration pattern similar to the palmitic acid standard. 15 Furthermore, we took advantage of the isolation of GLUT-4 protein in the presence of the detergent triton X-114. Logically, the GLUT-4 protein was detected in a Western blot in the water-detergent solution, as shown in figure 7, upper panel. By phase transition and separation of water from pure detergent, proteins could be separated into the detergent phase and water phase. The GLUT-4 could be immunodetected in the detergent phase, but hardly any signal was detected from the water phase. It should be noted that separation of detergent from the water phase results in a large volume of water and a minor volume of detergent. However, only a fraction of the water volume is run on the gel due to the size of the well in the gel. Thus, the GLUT-4 content in the water phase is higher than indicated in the figure. When the protein was upconcentrated by ammonium sulfate precipitation, the protein in the water phase did in fact react with a GLUT-4 specific antibody as shown in figure 7, lower panel. This strongly suggest the presence of GLUT-4 in the water phase. This is a known behaviour for GPI linked proteins. Finally, we separated the pure GLUT-4 transporter on a SDS gel and immunoblotted with an CRD antibody raised against an epitope which is uniquely formed on GPI proteins treated with phosphatidylinositol PLC (4;16). The CRD antibody reacted against a positive standard prepared from the variant surface glycoprotein prepared from T. Brucei. Interestingly, this antibody did also cross-react with the GLUT-4 transporter, as shown in figure 8. 16 DISCUSSION. We here report on the isolation of a GPI PLD, which may be able to release GLUT-4 containing vesicles from membranes by cleavage of a GPI link on the GLUT-4 transporter. Furthermore, we purified the GLUT-4 protein and provide immunological as well as biochemical evidence for the novel concept that GLUT-4 may indeed contain such a GPI link (figure 9). We have previously published evidence that a PLD may be involved in the mobilization of GLUT-4 containing vesicles in skeletal muscle (17). This was primarily based on the finding that in vivo insulin stimulation resulted in an increased production of phosphatidic acid in the GLUT-4 immunoprecipitate isolated from a low spin membrane fraction, but not in GLUT-4 immunoprecipitate isolated from a high spin membrane fraction. In the present study we have taken advantage of this in vitro assay and we have purified a protein fraction which induces GLUT-4 transfer in our assay. This fraction contained a PLD which is known to cleave glycosylphosphatidylinositol (GPI) anchors (7;15). The cleavage of GPI-linked proteins by GPI-PLD results in production of phosphatidic acid in the large membrane compartment harboring part of the GPI link (see figure 9). This may explain our previous finding of an insulin-induced production of phosphatidic acid in the GLUT-4 membrane compartment immunoprecipitated from a low spin membrane fraction (17). By adding PLD inhibitor/activators to the in vitro assay and by isolation of the pure GLUT-4 transporter we provide evidence that the GLUT-4 17 containing vesicles may be GPI linked to a large donor membrane compartment. Upon stimulation this GPI link may be cleaved which may release the GLUT-4 containing vesicles from binding to membranes. It should be noted that we used plasma as the source of purifying the GPI PLD. Unfortunately, we could not immunologically test whether GPI PLD is found in skeletal muscle, since there are no antibodies against rodent GPI PLD. Furthermore, due to the contamination of human muscle biopsies with plasma it is not possible to verify whether the GPI PLD originates from plasma or the muscle cell itself. However, GPI PLD mRNA expression is observed in skeletal muscle (20). Furthermore, the production of phosphatidic acid in GLUT4 immunoprecipitates (17), the co-isolation of the GPI constituents phosphatidylethanolamine and fatty acids with the pure GLUT-4 transporter (figure 6), and the effect of suramin and 1,10 phenanthroline on the in vitro GLUT-4 transfer carried out with cytosol isolated from rat skeletal muscle (figure 3), very strongly indicate the enzymatic involvement of a GPI PLD in skeletal muscle. Interestingly, there was no major difference between the potency of using cytosol from basal skeletal muscle when compared to insulin-stimulated (1 min) cytosol. This may suggest a constant active GPI PLD in the prepared cytosol. However, this is probably an artifact caused by homogenization and may not necessarily be a reflection of the in vivo GPI PLD activity in the intact muscle cell. For instance, the in vivo GPI PLD activity is related to the concentration of divalent cations (7;8;15;22;24). In fact, chelating the divalent cations by adding 1,10 phenanthroline in the in vitro GLUT-4 assay very 18 potently inhibited the GLUT-4 transfer (figure 3). Furthermore, GLUT-4 transfer was sensitive to the magnesium concentration (17). Although the concentrations of cations in the buffer are made comparable to the intracellular cytosolic concentrations, the true intracellular concentrations may differ due to membrane compartmentalization. Another alternative is that the GPI PLD in fact is constantly active in vivo in intact cells. If so the release of GLUT-4 containing vesicles may be clamped by another mechanism. We and others have observed docking to and phosphorylation of GLUT-4 protein by PKBβ in the process of GLUT-4 mobilization from membranes (3;17;19). This suggest to us that PKBβ (or possibly another protein kinase) may phosphorylate the GLUT-4 protein preparing it for GPI PLD cleavage. If GLUT-4 phosphorylation is a prerequisite for PLD cleavage the docking of PKBβ could be the regulated step in the GLUT-4 mobilization process. The GPI PLD also shows sequence similarity with the integrin family (15). This family is known for mechanochemical transduction properties via PI-3 kinase, which may suggest that GPI PLD could be activated by both insulin and muscle contractions which are the two physiological stimuli of GLUT-4 translocation to the surface membrane of muscle cells. In this regard, it should be noted that the production of phosphatidic acid in the GLUT-4 immunoprecipitate was insulin-sensitive (17) and furthermore that muscle contractions cause production of phoshatidic acid in rat skeletal muscle in vivo suggesting activation of a PLD (6). Interestingly, when GPI is added to cells it mimics many metabolic actions of insulin (25) suggesting that the GPI link has signaling properties of its own (2;13). 19 It was observed that GLUT-4 had an elution profile where charge was related to a change in molecular weight. It was speculated that this finding could be due to the presence (uncleaved) and absence (cleaved) of a phosphate as part of the GPI link (see figure 5). Given this, we were also curious to examine the elution profile for GLUT-4 isolated from basal and insulin stimulated skeletal muscle in order to try to detect a change in molecular weight and elution profile. Unfortunately, the poor linearity with silver staining and protein content and the strong signal in the Western blot made it difficult to make any measurements regarding small changes in molecular weight/charge and its possible relation to stimulation with insulin. It was also observed that the GLUT-4 protein could be separated into 2 different phases, namely a water phase and a detergent phase. This is at first glance very unexpected for an integral membrane protein. However, it is a very well known observation for GPI linked proteins (4;12;26), since the link is detergent resistant (11) and contains large and bulky side chains consisting of very water soluble glucosamine, carbohydrates and phosphate. Thus, the present observations strengthen the concept that GLUT-4 protein contains a GPI link. Finally, recent reports suggest that insulin-stimulated GLUT-4 translocation requires the translocation of cbl to lipid rafts (5) which are detergent resistant membrane areas enriched in GPI linked proteins (27). One question is whether all GLUT-4 transporters inside the muscle cell are released via this suggested mechanism. We have used an in vitro assay which may - or may not represent the whole population of GLUT-4. We have previously done marker enzyme 20 analysis on this membrane fraction (17). However, prior membrane homogenization causes cross contamination of membranes and enzyme marker analysis on the recovered membrane fraction does not discriminate between the GLUT-4 positive membrane and membranes devoid of GLUT-4. Thus, we are not in a position to answer this question with the techniques used in the present study. Future studies using microscopy of immunolabelled membrane organelles are necessary to answer this question. A number of consensus polypeptide sequences are known to signal for the attachment of a GPI (31). This attachment is mediated by a transamidase reaction and takes place during synthesis of the polypeptide in the endoplasmatic reticulum. By using computer based prediction analysis, it was suggested that the C-terminal of GLUT-4 (the amino acid # 480, SAT sequence) potentially could have a GPI anchor sequence. However, this prediction is based on calculation following the ω,ω+2 rule and may produce false positive results (31). The family of GPI linked proteins is broad (10) and many GPI linked proteins are localized on the exterior of the cells, where they can be released into the surrounding milieu. Given this, it might seem surprising that the intracellular located GLUT-4 protein should contain a GPI link. However, also ornithine decarboxylase, GPI proteins in granules and trafficking of GPI are related to the interior of the cell (1;10;12). Furthermore, some aminopeptidases are also known to be GPI linked and released by this mechanism (23). Interestingly, one of the constituents in the GLUT4 containing vesicles is IRAP/-aminopeptidase N. This is one of the few proteins found in the GLUT-4 immuno-precipitate which fully translocates together with GLUT-4 to the surface membrane. It is tempting to speculate that GLUT-4 and IRAP/aminopeptidase are 21 sequestered in the same vesicle because they are both released by cleavage of a GPI link. If IRAP and GLUT-4 indeed are sequestered in the same translocating vesicle it might be argued that in vitro GLUT-4 transfer (as observed when adding GPI PLD to the GLUT-4 donor membranes in the present study) may be due to cleavage of a GPI link to the IRAP rather than to GLUT-4. However, our finding of a GPI link on the pure GLUT-4 protein indicates that the GPI link indeed is on GLUT-4. Taken together, the present paper offers several novel findings. First, the further characterization and description of an assay strategy for studying ongoing enzymatic reactions in a demanding tissue. Second, a method was developed for purifying large and pure amounts of GLUT-4 proteins enabling GLUT-4 immobilization and studies of GLUT-4 phosphorylation and binding by other proteins. Third, we have shown that GPI PLD may cause the release of GLUT-4 containing vesicles from an enriched GLUT-4 fraction. This GLUT-4 release could be activated and inhibited with suramin and 1,10 phenanthroline, compounds known to activate and inhibit GPI PLD activity, respectively. By purifying the GLUT-4 transporter we also provide immunological and biochemical evidence for a GPI link. In conclusion, we suggest that a GPI PLD may be a novel important mediator for GLUT-4 mobilization in skeletal muscle. 22 Acknowledgements: The present study was supported by grants from the Danish National Research Foundation (504-14) and from Novo-Nordisk Research Foundation and Danish Diabetes Association. 23 Reference List 1. Bauman, N. A., J. Vidugiriene, C. E. Machamer, and A. Menon. Cell surface display and intracellular trafficking of free glycophosphatidylinositols in mammalian cells. J. Biol. Chem. 275: 7378-7389, 2001. 2. Bruni, P., E. Meacci, M. Avila, V. Vasta, M. Farnararo, J. M. Mato, and I. Varela. A phospho-oligosaccharide can reproduce the stimulatory effect of insulin on glycolytic flux in human fibroblasts. Biochem. Biophys. Res.Comm. 166: 765771, 1990. 3. Calera, M. R., C. Martinez, H. Liu, A. E. J. Jack, M. Birnbaum, and P. F. Pilch. Insulin increases the association of Akt-2 with GLUT4-containing vesicles. J. Biol. Chem. 273: 7201-7204, 1998. 4. Cardoso de Almeida, M. L. and M. J. Turner. The membrane form of variant surface glycoproteins of Trypanosoma brucei. Nature 302: 349-352, 1983. 5. Chiang, S., C. Bauman, M. Kanzaki, D. Thurmond, R. Watson, C. Neudauer, I. Macara, J. Pessin, and A. Saltiel. Insulin-stimulated GLUT4 translocation requires the CAP-dependent activation of TC10. Nature 410: 944-948, 2001. 24 6. Cleland, P. J., G. Appleby, S. Rattigan, and M. Clark. Exercise-induced translocation of protein kinase C and production of diacylglycerol and phosphatidic acid in rat skeletal muscle in vivo. J. Biol. Chem. 264: 17704-17711, 1989. 7. Davitz, M. A., J. Hom, and S. Schenkman. Purification of a glycosylphosphatridylinositol-specific phospholipase D from human plasma. J. Biol. Chem. 264: 13760-13764, 1989. 8. de Almeida, M. L., M. J. Turner, B. B. Stambuk, and S. Schenkman. Identification of an acid-lipase in human serum which is capable of solubilizing glycophosphatidylinositol-anchored proteins. Biochem. Biophys. Res. Comm. 150: 476-482, 1988. 9. Emoto, M., J. K. Klarlund, S. B. Waters, V. Hu, J. M. Buxton, A. Chawla, and M. P. Czech. A role for phospholipase D in GLUT4 glucose transporter translocation . J. Biol. Chem. 275: 7144-7151, 2000. 10. Ferguson, M. A. and A. F. Williams. Cell-surface anchoring of proteins via glycosyl-phosphatidylinositol structures. Ann. Rev. Biochem. 57: 285-320, 1988. 11. Fiedler, K., T. Kobayashi, T. V. Kurzchalia, and K. Simons. Glycosphingolipidenriched, detergent-insoluble complexes in protein sorting in epithelial cells. Biochemistry 32: 6365-6373, 1993. 25 12. Fouchier, F., P. Bastiani, T. Baltz, D. Aunis, and G. Rougon. Glycosylphosphatidylinositol is involved in the membrane attachment of proteins in granules of chromaffin cells. Biochem. J. 256: 103-108, 1988. 13. Frick, W., A. Bauer, J. Bauer, S. Wied, and G. Muller. Insulin-mimetic signalling of synthetic phosphoinositolglycans in isolated rat adipocytes. Biochem J. 336 ( Pt 1): 163-181, 1998. 14. Hammond, S. M., J. M. Jenco, S. Nakashima, K. Cadwallader, Q.-M. Gu, S. Cook, Y. Nozawa, G. D. Prestwich, M. A. Frohman, and A. J. Morris. Characterization of two alternatively spliced forms of phospholipase D1. J. Biol. Chem. 272: 3860-3868, 1997. 15. Huang, K. S., S. Li, W. J. Fung, J. D. Hulmes, L. Reik, Y. C. Pan, and M. G. Low. Purification and characterization of glycosyl-phosphatidylinositol- specific phospholipase D. J. Biol. Chem. 265: 17738-17745, 1990. 16. Jones, D. R., M. A. Avila, C. Sanz, and I. Varela-Nieto. Glycosylphosphatidylinositol-phospholipase type D: a possible candidate for the generation of second messengers. Biochem. Biophys. Res. Comm. 233: 432-437, 1997. 17. Kristiansen, S., Nielsen, JN., Bourgoin, S., Klip, A., Franco, M., and Richter, E. A. GLUT-4 translocation in skeletal muscle studied with a cell-free assay: involvement of phospholipase D. Am. J. Physiol. 281: E608-E618, 2001. 26 18. Kristiansen, S., T. Ramlal, and A. Klip. Phosphatidylinositol 4-kinase, but not phosphatidylinositol 3-kinase, is present in GLUT4-containing vesicles isolated from rat skeletal muscle. Biochem. J. 335 ( Pt 2): 351-356, 1998. 19. Kupriyanova, T. A. and K. V. Kandror. Akt-2 binds to Glut4-containing vesicles and phosphorylates their component proteins in response to insulin. J. Biol. Chem. 274: 1458-1464, 1999. 20. LeBoeuf, R. C., M. Caldwell, Y. Guo, C. Metz, M. A. Davitz, L. K. Olson, and M. A. Deeg. Mouse glycosylphosphatidylinositol-specific phospholipase D (Gpld1) characterization. Mamm.Genome 9: 710-714, 1998. 21. Lee, J. Y., M. R. Kim, and D. E. Sok. Enzymatic release of Zn2+glycerophosphocholine cholinephosphodiesterase from brain membranes by glycosylphosphatidylinositol-specific phospholipases and its regulation. Neurochem. Res. 23: 899-905, 1998. 22. Li, J. Y., K. Hollfelder, K. S. Huang, and M. G. Low. Structural features of GPIspecific phospholipase D revealed by proteolytic fragmentation and Ca2+ binding studies. J. Biol. Chem. 269: 28963-28971, 1994. 23. Low, M. G. The glycosyl-phosphatidylinositol anchor of membrane proteins. Biochim. Biophys. Acta 988: 427-454, 1989. 27 24. Metz, C. N., G. Brunner, N. H. Choi-Muira, H. Nguyen, J. Gabrilove, I. W. Caras, N. Altszuler, D. B. Rifkin, E. L. Wilson, and M. A. Davitz. Release of GPI-anchored membrane proteins by a cell-associated GPI- specific phospholipase D. EMBO J. 13: 1741-1751, 1994. 25. Misek, D. E. and A. R. Saltiel. An inositol phosphate glycan derived from a Trypanosoma brucei glycosyl- phosphatidylinositol mimics some of the metabolic actions of insulin. J. Biol. Chem. 267: 16266-16273, 1992. 26. N.M.Hooper. Identification of a glycosylphosphatidylinositol anchor on membrane proteins. In N.M.Hooper and A.J.Turner. Lipid modifications of proteins: A practical approach. Leeds, IRL press. 1992, 89-113. 27. Pfeffer, S. Caveolae on the move. Nature Cell Biology 3: E108-E110, 2001. 28. Ploug, T., B. Deurs, H. Ai, S. W. Cushman, and E. Ralston. Analysis of GLUT4 distribution in whole skeletal muscle fibers: Identification of distinct compartments that are recruited by insulin and muscle contractions. J. Cell Biol. 135: 415-430, 1998. 29. Roberts, W. L., B. H. Kim, and T. L. Rosenberry. Differences in the glycolipid membrane anchors of bovine and human erythrocyte acetylcholinesterases. Proc. Natl. Acad. Sci.U.S.A 84: 7817-7821, 1987. 28 30. Stadelmann, B., A. Zurbriggen, and U. Brodbeck. Distribution of glycosylphosphatidylinositol-specific phospholipase D mRNA in bovine tissue sections. Cell Tissue Res. 274: 547-552, 1993. 31. Udenfriend, U. and K. Kadukula. Prediction of site in nascent precusors of glycosylphosphatidylinositol protein. Casey, P. J. and J. E. Buss. Methods in Enzymology. San Diego, Academic Press. 1995, 571-581. 29 Legend 1. Upper panel. An enriched GLUT-4 membrane fraction was isolated from basal (B) rat skeletal muscle and incubated for 45 min at 370C with no cytosol (none), or a cytosolic protein fraction isolated from basal (B) skeletal muscle. The large and small membrane structures were subsequently separated by a differential centrifugation step. Proteins from the 2 fractions were subjected to GLUT-4 Western blotting. The upper panel shows one representative immunoblot (double determination) out of approximately 50. Middle panel: An enriched GLUT-4 membrane fraction was isolated from basal (B) rat skeletal muscle or insulin (Ins)-stimulated (10 min) rat skeletal muscle and incubated with no cytosol (none), or a cytosolic protein fraction isolated from basal (B) or insulin (Ins)-stimulated (1 min) rat skeletal muscle. The middle panel shows a representative GLUT-4 Western blot of 4 experiments, each done with double determination of each experimental condition. Bottom panel. An enriched GLUT-4 membrane fraction was isolated from basal skeletal muscle and incubated with no cytosol (none), cytosol prepared from rat skeletal muscle (Sm), rat liver (Li), rat brain (Br) and rat blood plasma (Bp). The immunoblot is representative of approximately 50, 2, 2, 28 experiments for skeletal muscle, rat liver, rat brain, and blood plasma, respectively. 30 Legend 2. The unknown protein factor was identified by testing different elute fractions for its capability to induce in vitro GLUT-4 transfer. The upper panel shows the increase in GLUT-4 content in the small vesicle fraction as a result of adding a control non-active fraction (lane 1), the active protein fraction from Q sepharose (lane 2), the active elute fraction from Zn++ immobilized metal chromatography (lane 3) and adding fraction # 39 and 40 from the size exclusion chromatography experiment (lane 4). The fractions ranging from 36 to 42 from the size exclusion chromatograhy step were examined closer by gel electrophoresis and silver staining as indicated in the middle panel. The active fractions # 39 and # 40 contained a distinct protein band at 110 kDa (marked with an arrow). The protein fractions were also electro transferred to a membrane and probed with an anti GPI PLD IgG as indicated in the lower panel. The elute fractions and stained gel is representative of 4 different experiments. 31 Legend 3. Upper panel: The upper panel shows the effect of 10 mM of suramin (a GPI PLD activator) on the in vitro GLUT-4 transfer . An enriched GLUT-4 membrane fraction was isolated from basal rat skeletal muscle and added nothing (-) or cytosol (+) prepared from basal skeletal muscle in the presence (+) or absence (-) of suramin. The mixtures were subsequently differential centrifugated. The graph shows the GLUT-4 protein content in the supernatant containing the small membrane structures. The calculated mean + SE are from 4 different experiments, each determination done in duplicate. The bars are no cytosol (empty bar), cytosol (filled bar), cytosol + suramin (cross hatched bar) and no cytosol + suramin (hatched bar). * p<0.05 when compared to no cytosol (empty bar). $ p < 0.05 compared to cytosol (filled bars). Lower panel. The lower panel shows the effect of 1 mM of 1,10 phenanthroline (an inhibitor of the GPI PLD activity). As explained in the text to the upper panel, the enriched GLUT-4 membrane fractions were incubated with or without cytosol and 1,10 phenanthroline, as indicated. The graph shows the calculated mean + SE from 4 different experiments, each in duplicate. The bars are no cytosol (empty bar), cytosol (filled bar), cytosol + 1,10 phenanthroline (cross-hatched bar). * p<0.05 when compared to no cytosol (empty bar). 32 Legend 4. The figure outlines the purification protocol for isolating the GLUT-4 transporter from rat skeletal muscle. The upper panel shows a silver stain of all proteins in a 10% gel loaded with a muscle homogenate (lane 1), a fraction containing peripheral and integral membrane proteins (lane 2), integral proteins (lane 3), positive GLUT-4 elute fractions from Q Sepharose (lane 4), hydroxyapatite (lane 5) and Mono Q chromatography (lane 6) and a molecular weight marker (lane 7). No other proteins than the immunoreactive GLUT-4 (lower panel) could be observed in lane 6. The silver stain and GLUT-4 immunoblot are representative of 10-15 determinations. Legend 5. The figure shows the change in GLUT-4 migration pattern on the gel with increasing charge loaded on the Mono Q (5/5) column. The silver stain is representative for 4 different elution experiments. 33 Legend 6. Pure GLUT-4 protein was photo affinity labelled with [125I] TID. In some experiments the labelled GLUT-4 protein was captured on a Q Sepharose column, washed and eluted with an increase in sodium chloride. Positive (+) and negative (-) GLUT-4 elute fractions were subjected to nitrous deamination. Alternatively, unlabelled [125I] TID were removed from GLUT-4 by drying the sample under a stream of N2. The residual sample was base hydrolysed. Lipids were extracted from the mixture and run on a thin layer chromatography plate with 2 different solvents for the separation of phospholipids (top panel : nitrous deamination) or fatty acids (bottom panel : base hydrolysis). Positive standards for phosphatidylserine (PS), phosphatidylcholine (PC), phosphatidylinositol (PI), and phosphatidylethanolamine (Peth.) and palmitic acid (PA) were separated and stained in the left top panel. An arrow is used to indicate the migration of palmitic acid standard in the solvent used for base hydrolysis (lower panel). The radioactive signals shown are representative of 3 different experiments. 34 Legend 7. The pure GLUT-4 transporter was eluted from a Mono Q (5/5) column in the presence of the detergent triton X-114. Fractions consisting of the water-detergent (WD) - , the water (w) or detergent (D) phases were separated on a gel and silver stained or Western blotted with an anti GLUT-4 IgG. The upper panel is representative of 3 different experiments. In some experiments, the GLUT-4 protein in the water phase was precipitated and subjected to GLUT-4 Western blotting (lower panel). Legend 8. The pure GLUT-4 transporter was separated on a gel and Western blotted with a cross reacting determinant (CRD) antibody raised against the GPI link on the variant surface glycoprotein (VSV) from T Brucei or blotted with an anti-GLUT-4 IgG. The immunoblot is representative of 4 different experiments. Legend 9. The figure shows a hypothetical model of a putative GPI link from the large membrane structure to the GLUT-4 protein in the vesicle. The link may consist of fatty acids, phosphate, inositol, glucosamine, several mannitol residues, and phosphatidylethanolamine. The PLD cleavage site is shown. 35 Figure 1 Donor membranes: B None B Large membrane structures Small membrane structures Cytosol : None Donor membranes: B B B Ins Small membrane structures Cytosol : None B Ins B Ins Small membrane structures Cytosol: None Sm Li. Br. Bp Figure 2. 36 1 2 3 4 GLUT-4 protein content in small membrane structures <50 kDa Silver stain GPI PLD Western blot 36………..………….42 SEC elute fractions 37 Figure 3 GLUT-4 content in small membra ne structures. $ 1.60 $ 1.40 1.20 1.00 * 0.80 0.60 0.40 0.20 0.00 Cyt: Suramin: - GLUT-4 content in small membra ne structures 0.45 + - + + + * 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 Cyt: 1,10 ph.thr.: - + - + + 38 Figure 4 1 2 3 4 5 6 Silver stain GLUT-4 Western blot 7 <50 kDa < 50 kDa 39 Figure 5 GLUT-4 silver stain 50 kDa > 40 Figure 6 Nitrous deamination - + - + PS PC PI PEth PA Base hydrolysis PA + - 41 Figure 7 WD W D Silver stain GLUT-4 Western blot <50 kDa GLUT-4 Western blot <50 kDa D D W W 42 Figure 8 VSG GLUT-4 < 50 kda CRD IgG GLUT-4 IgG 43 Figure 9 GLUT-4 containing vesicle GLUT-4 polypeptide Ethanolamine (Man)3 Glucosamine C-C-CO O O=C C=O P Inositol GPI PLD cleavage site Large membrane structure