 |
INTRODUCTION |
Insulin regulates the cellular uptake of glucose in adipose tissue
and muscle by acutely controlling the number of glucose transport
proteins present in the cell surface membrane (1-4). The major
insulin-responsive sugar transporter, GLUT4, recycles in endosomal and
exocytic membranes of these cells (5-7) and is mostly sequestered
within intracellular membranes in the unstimulated state (8-10).
Insulin acts primarily by enhancing the exocytosis of GLUT4, but the
hormone also appears to inhibit endocytosis as well (5-7). Recent
studies have revealed that the insulin-regulated intracellular
membranes containing GLUT4 are specialized and appear to exclude some
other cycling proteins such as the transferrin receptor and the GLUT1
glucose transporter (11, 12). These latter proteins are present at the
cell surface in higher abundance than GLUT4 in unstimulated cells and
move through a rapid constitutive endosomal recycling pathway (13, 14).
Complicating our understanding of the interrelationships between these
trafficking systems are findings suggesting that GLUT4 also partially
co-localizes with endosomal membranes containing the transferrin
receptor and GLUT1 (12, 15, 16). Because normal glucose homeostasis in
humans is dependent upon the dynamics of GLUT4 membrane trafficking, intensive efforts have been directed at understanding the basis for the
unique recycling characteristics and regulation of GLUT4.
Isolation of the intracellular membranes of adipocytes and muscle that
are enriched in GLUT4 has been achieved using immunoadsorption procedures with immobilized antibodies raised against GLUT4 (17, 18).
These membranes are likely a mixture of endosomal membrane fractions as
well as the specialized insulin-responsive membranes. Many proteins
that reside in the intracellular membranes containing GLUT4 have been
identified (for a review, see Ref. 19). Such membrane preparations have
been analyzed by both biochemical and immunological methods and contain
receptors that transport various ligands within cells such as the
insulin-like growth factor-II/mannose 6-phosphate receptor (17) as well
as several other transmembrane proteins, including
IRAP1 (20), amine oxidase
(21), and sortilin (22). Proteins that associate with GLUT4-containing
membranes also include several that are thought to function in
mechanisms of membrane trafficking such as VAMP2 (23, 24), secretory
carrier-associated membrane proteins (25, 26), syntaxin 4 (27), and
SNAP23 (28-30). Syntaxin 4-interacting protein-1 in turn has
been identified as a syntaxin 4-interacting protein, and its
interaction with syntaxin 4 is reported to be regulated by insulin
(31). A unifying hypothesis has been derived from the work of many
laboratories on these proteins, which suggests that fusion of
GLUT4-containing membranes with the plasma membrane involves
VAMP2/syntaxin 4 interaction regulated by SNAP23, MUNC 18, syntaxin
4-interacting protein-1, and other proteins (4). The molecular
mechanisms by which this process may be regulated by insulin are
not yet understood.
It seems likely that insulin also regulates events in the GLUT4
translocation process that occur prior to plasma membrane docking or
fusion. For example, in unstimulated 3T3-L1 adipocytes, an excellent
model system for insulin action on GLUT4, most of the GLUT4 is present
in perinuclear membranes and in other intracellular membranes that are
not apparently connected to the plasma membrane (32, 33). Also, the
time course of insulin action on GLUT4 translocation is rather slow
(several minutes) compared with the stimulation of neurotransmitter
release, which involves rapid fusion of synaptic vesicles near the cell
surface. These characteristics of GLUT4 regulation are consistent with
the hypothesis that insulin causes translocation of GLUT4-containing
membranes over some distance prior to docking and fusion with the
plasma membrane. However, no data are available on the mechanisms
whereby GLUT4 membranes are sequestered and restrained in a perinuclear
localization in 3T3-L1 adipocytes under basal conditions. Nor is there
an understanding of the processes whereby GLUT4-containing membranes
actually move to the plasma membrane.
In order to gain insight into these issues, we developed a method to
purify GLUT4-containing membranes that does not involve anti-GLUT4
antibodies that can contaminate membrane preparations. Mass
spectrometry analysis of proteins in the purified GLUT4-containing membranes identified the cytoskeletal proteins vimentin and
-tubulin, suggesting a role for intermediate filaments and
microtubules in their localization. We provide strong evidence in favor
of this concept. Importantly, our data support the hypothesis that dynein motor activity on microtubules is required to localize GLUT4
into perinuclear membranes that constitute the insulin-sensitive compartment. These results provide a new framework to test hypotheses on the movements of GLUT4 based on the regulation and coordination of
cellular motors and cytoskeletal structures.
| |
EXPERIMENTAL PROCEDURES |
Materials--
Rabbit polyclonal anti-GLUT4 antibody was raised
against the C-terminal 12-amino acid sequence of GLUT4. Mouse
anti-transferrin receptor was from Zymed Laboratories
Inc. Rabbit polyclonal anti-VAMP2 antibody was from StressGen
Biotechnologies Corp. Mouse monoclonal anti-vimentin antibody used in
immunoblots and immunoelectron microscopy analysis was from Santa Cruz
Biotechnology, Inc. (Santa Cruz, CA). Mouse monoclonal anti--tubulin
antibody, used in immunoblot and immunoelectron microscopy analysis and
the secondary antibodies conjugated to gold particles for
immunoelectron microscopy were from Amersham Pharmacia Biotech.
Cell Culture--
3T3-L1 fibroblasts were grown in Dulbecco's
modified Eagle's medium (DMEM) supplemented with 50 µg/ml
streptomycin, 50 units/ml penicillin and 10% fetal bovine serum.
3T3-L1 fibroblasts (2-3 days postconfluent) were differentiated into
adipocytes by incubating with DMEM supplemented with the same
antibiotics, 10% fetal bovine serum, 0.5 mM
isobutylmethylxantine, 0.25 µM dexamethasone, and 2.5 µg/ml insulin for 3 days, grown in DMEM with 10% fetal bovine serum
and 2.5 µg/ml insulin for 3 days, and then grown in DMEM with 10%
fetal bovine serum for an additional 5-8 days.
Isolation and Fractionation of GLUT4-containing
Vesicles--
Adipocytes were isolated from epididymal fat pads of
male Harlan Sprague-Dawley Rats (125-150 g) by collagenase digestion using Krebs-Ringer/HEPES (KRH; 80 mM NaCl, 10 mM HEPES, 10 mM glucose, 5 mM KCl,
2 mM CaCl2, 2 mM NaPO4,
1 mM MgCl2), pH 7.4, supplemented with 2%
bovine serum albumin and 2 mM pyruvate. Cells were washed
and incubated for a 30-min recovery period prior to initiation of
experiments. Cells were then incubated at 37 °C with or without 100 nM insulin for 20 min. The cells were then washed with PBS
and immediately homogenized in buffer A (50 mM HEPES, pH
7.4, 10 mM NaF, 1 mM NaPPi, 0.1 mM Na3VO4, 1 mM
phenylmethylsulfonyl fluoride, 10 µg/ml aprotinin, and 10 µg/ml
leupeptin) and then subjected to differential centrifugation as
described previously (6) to isolate the low density microsomal (LDM)
subcellular fractions. Protein was quantified using the bicinchoninic
acid protein determination kit (Pierce) with bovine serum albumin as standard. The GLUT4-enriched fractions were then isolated from LDM
fractions utilizing the sedimentation sucrose velocity gradient centrifugation method exactly as described previously (34, 35). Briefly, 1.5-2 mg of LDM fractions were loaded onto a 10-35% sucrose velocity gradient (buffer B: 20 mM HEPES, pH 7.4, 100 mM NaCl, 1 mM EDTA, 2 mM
dithiothreitol, 1 mM, 10 mM NaF, 1 mM NaPPi, 0.1 mM
Na3VO4, 1 mM phenylmethylsulfonyl
fluoride, 10 µg/ml aprotinin, and 10 µg/ml leupeptin) and
centrifuged for 3.5 h at 24,000 rpm in a SW28 rotor (Beckman).
Fractions containing GLUT4-membranes (fractions 8-18) were pooled,
pelleted by ultracentrifugation at 48,000 rpm for 1.5 h,
resuspended in buffer B, and then loaded onto an equilibrium density
sucrose gradient (10-65% (w/v) in buffer B and centrifuged at 35,000 rpm for 18 h in a SW 50.1 rotor (Beckman). After centrifugation,
fractions were collected starting from the top of the gradient and
analyzed for the total protein content determined by the Bradford assay
(Bio-Rad).
Immunoblotting--
Fractions from velocity gradients and
equilibrium density gradient were prepared as described above, and
aliquots from these fractions were subjected to SDS-PAGE on resolving
gels according to Laemmli (36). Separated proteins were
electrophoretically transferred to nitrocellulose membrane, blocked
with 3% nonfat milk and 1% BSA in TTBS (0.05% Tween 20 in
Tris-buffered saline), and then incubated with primary antibody in TTBS
containing 1% BSA. After incubation, membranes were washed with TTBS
and incubated with horseradish peroxidase-labeled anti-mouse IgG for
the detection of monoclonal antibodies or with horseradish
peroxidase-labeled anti-rabbit IgG for detection of polyclonal
antibodies. Proteins were visualized using an enhanced chemiluminescent
substrate kit (Amersham Pharmacia Biotech), and immunoblot intensities
were quantified by a scanning densitometer.
Electron Microscopy--
GLUT4-containing membranes of the
insulin-sensitive fractions from the equilibrium density gradient were
isolated as described above. Fractions were pooled, pelleted by
centrifugation at 48,000 rpm for 2 h, resuspended in PBS, and
fixed in a final concentration of 2% paraformaldehyde in PBS. GLUT4
vesicles were then adsorbed to Formvar-coated gold grids and processed
for double labeling as outlined by Martin et al. (24) and
Sleeman et al. (37). Grids were incubated with 50 µl of
primary antibody diluted in 1% BSA and PBS as follows: anti-GLUT4,
anti-IRAP, anti-vimentin, anti--tubulin, or nonimmune IgG, as a
negative control. After incubation with each IgG fraction, grids were
labeled with either 5- or 15-nm gold particles conjugated to the
secondary antibody (goat anti-rabbit or goat anti-mouse). Grids were
stained with 1% uranyl acetate, dried, and viewed using a transmission
electron microscope, Phillips CM.10.
MALDI-TOF Mass Spectrometry Analysis--
Proteins resolved by
SDS-PAGE were visualized by silver staining (Bio-Rad), and the bands
were excised from one single dimensional 5-15% gel. The
silver-stained proteins bands were destained and tryptically digested
(trypsin) in gel according to Gharahdaghi et al. (38) with
some slight modifications. The digested samples were further
concentrated and desalted with Millipore Zip Tip C18 microtips prior to
MALDI-TOF analysis. MALDI-TOF analyses were performed on a Kratos
Analytical Kompact SEQ Instrument, equipped with a curved field
reflectron. Peptide masses were searched against the nonredundant
protein data base using MS-Fit of the Protein Prospector program
developed by Clauser et al. (39) at the University of
California, San Francisco. Fragmentation information obtained from
individual peptides via postsource decay analysis was searched against
the nonredundant protein data base using the protein prospector program
MS-Tag.
Microinjection of 3T3-L1 Adipocytes--
3T3-L1 cells at 7-9
days postdifferentiation were released from the cell culture dishes
using 0.5 mg/ml collagenase D and 0.025% trypsin in PBS at 37 °C.
Cells were resuspended in DMEM containing 10% fetal bovine serum
(v/v), 50 units/ml penicillin and 50 µg/ml streptomycin,
centrifuged at 1000 rpm in a Beckman GPK centrifuge, and resuspended in
media once again. The cells were plated on grid coverslips
(Eppendorf) in 12-well dishes at a density of approximately 1.5 × 105 cells per well. The next day, approximately 300 cells
were impaled using the Eppendorf model 5171 micromanipulator and
injected using the Eppendorf model 5246 microinjector with
approximately 0.1 pl of a 2 mg/ml solution of vimentin peptide Vm-1A
(peptide sequence) or control peptide (Vimentin amino acids 1-18
mutated R10H) containing 20 µg/ml FITC-conjugated dextran. The cells
were allowed to recover for 60 min and then fixed with 3.7%
formaldehyde in PBS for 10 min. The coverslips were processed for
immunofluorescence as described below.
Immunofluorescence--
Formaldehyde-fixed cells were
permeabilized with methanol for 2 min and blocked in PBS, 1% fetal
bovine serum, and 0.5% Triton X-100 for 15 min. GLUT4 antibody diluted
1:1000 was applied to the coverslips for 2 h, washed, and detected
with rhodamine-conjugated anti-rabbit secondary antibody. Microinjected
cells, identified by FITC fluorescence were evaluated for GLUT4 dispersion.
Acetate Treatment and GLUT4 Translocation Assays--
Coverslips
containing 3T3-L1 adipocytes were incubated in KRH containing 70 mM sodium acetate, pH 6.4, for 15 min in a 37 °C
incubator. These conditions have previously been shown to selectively inhibit dynein activity (40). Acetate recovery was achieved by washing
the coverslips three times with KRH without acetate and incubating at
37 °C for 15 or 60 min, as indicated in the figure legends.
Cells were then fixed in 3.7% formaldehyde as described above, and
GLUT4 was visualized by immunofluorescence. For the effect of acetate
treatment on insulin-stimulated GLUT4 translocation, cells were treated
with KRH with or without acetate, allowed to recover for the indicated
times, and stimulated or not with insulin for 15 min.
GLUT4 translocation was assayed in plasma membrane lawns, generated by
using a technique similar to the one previously described (41, 42).
Briefly, membrane lawns were fixed on the coverslips with 3.7%
formaldehyde for 10 min, washed twice with PBS, and incubated for 45 min with 2% BSA in PBS. To quantify GLUT4 on lawns, the coverslips
were incubated with a 1:1000 dilution of rabbit anti-GLUT4 polyclonal
antibody in PBS and 0.05% Tween 20. Coverslips were washed five times
for 3 min each, and they were incubated with a 1:1000 dilution of
FITC-conjugated goat anti-rabbit mixed with 10 µg/ml
rhodamine-conjugated wheat germ agglutinin (to quantitate plasma
membrane). The coverslips were washed as before and postfixed for 10 min with 3.7% formaldehyde followed by a final wash with PBS. They
were then mounted on slides with 90% glycerol in PBS (DABCO) and
viewed with × 60 objective on a Nikon Diaphot 200 inverted
microscope coupled to a Bio-Rad MRC1024 processing unit. Images were
analyzed by the Lasersharp processing software.
| |
RESULTS |
Purification of Insulin-responsive GLUT4-containing
Membranes--
In order to purify GLUT4-containing membranes, low
density microsomes were first prepared from primary, unstimulated rat
adipocytes by standard techniques. This crude membrane fraction
contains most of the GLUT4 present in unstimulated adipocytes and is
composed primarily of intracellular membranes (6). These membranes were then subjected to sucrose velocity gradient centrifugation (Fig. 1), which reportedly yields a more
purified GLUT4-enriched membrane preparation (34, 35). Consistent with
previous results, this technique separates about 90% of the total
membrane protein (fractions 1-7) away from the GLUT4-enriched
membranes (fractions 8-18). Insulin treatment of rat adipocytes prior
to disruption of the cells and preparation of these membranes causes a
marked decrease in the yield of GLUT4 present in the latter fractions
(Fig. 1, A and upper panel of
B). However, no such insulin effect is observed when total
membrane protein is measured (Fig. 1B, lower
panel) because these membranes are still highly contaminated
with membranes that do not contain GLUT4 and are not
insulin-responsive.
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 1.
Isolation of GLUT4-containing membranes from
rat adipocytes by sucrose velocity gradient centrifugation. Rat
adipocytes were treated or not with insulin for 20 min, and low density
microsomes were prepared as described under "Experimental
Procedures." LDM subcellular fractions were centrifuged in a 28-ml
10-35% sucrose gradient, and 1-ml fractions were collected from the
top to the bottom. A, immunoblot analysis of fractions using
anti-GLUT4; B, the data shown in A for GLUT4
protein from control (open circles) or insulin-stimulated
(closed circles) cells were quantified by scanning
densitometry. C, total protein from control (open
square) or insulin-stimulated (closed square) fractions
was measured as described under "Experimental Procedures."
|
|
Fractions 8-18 containing most of the GLUT4 from the sucrose velocity
gradient were therefore subjected to equilibrium gradient centrifugation to further resolve the membrane species present (Fig.
2). Most of the membrane protein was
distributed over fractions 5-20 after this procedure, whereas most of
the GLUT4 was distributed within fractions 7-14. Importantly, this
GLUT4 was separated into two types of membranes that could be
distinguished based on their sensitivity to insulin. The amount of
GLUT4 in fractions 7-9 was decreased when the cells were treated with
insulin before homogenization and preparation of membranes, whereas the
GLUT4 in fractions 10-20 was not affected by insulin treatment of the
adipocytes (Fig. 1, A and upper panel
of B). Strikingly, measurement of total membrane protein in
the fractions of this gradient revealed a similar profile: about a 50%
reduction in fractions 7-9 due to insulin action, with no insulin
effect observed in fractions 10-20 (Fig. 1B,
lower panel). This observed insulin-mediated
decrease in total membranes recovered in fractions 7-9 indicates the
successful partial purification of membranes of the insulin-responsive
compartment or compartments in primary adipocytes. Similar data have
been obtained using 3T3-L1 adipocytes (not shown).
View larger version (23K):
[in this window]
[in a new window]
|
Fig. 2.
Fractionation of GLUT4-containing membranes
by equilibrium density gradient centrifugation. Fractions
containing GLUT4 membranes from Fig. 1 (fractions 8-18) were pooled,
pelleted, and then loaded on the top of a 5-ml equilibrium density
sucrose gradient (10-65%) as described under "Experimental
Procedures." A, immunoblot analysis of fractions using
anti-GLUT4. B, the data shown in A for GLUT4
protein from control (closed squares) or insulin stimulated
(open squares) cells were quantified using scanning
densitometry. C, total protein from control (closed
circles) or insulin-stimulated (open circles) fractions
were measured as described under "Experimental Procedures."
|
|
Two additional approaches were used to characterize the membranes
resolved by equilibrium gradient centrifugation. First, each of the
fractions from the gradient were analyzed by SDS-PAGE and silver
staining of the constituent proteins (Fig.
3). This analysis revealed that most of
the membrane proteins in fractions 7 and 8 were dramatically reduced
when membranes were derived from insulin-treated adipocytes. Certain
proteins in fractions 6 and 9 showed the same effect, whereas many did
not (Fig. 3). These results suggest that membranes resolved in
fractions 7 and 8 are highly purified insulin-responsive membranes,
while those in fractions 6 an 9 are only partially purified. Membranes
in higher density fractions show no detectable insulin sensitivity despite the presence of significant GLUT4 protein. Of note is the
finding that many of the protein bands in the insulin-sensitive membranes are also present in the membranes that are not responsive to
the hormone (compare lanes 7-9 with
lanes 10-13 in Fig. 3). These data are
consistent with the hypothesis that the insulin-sensitive membranes
containing GLUT4 contain many of the same constituent proteins as other
cell membranes that function in a hormone-insensitive mode.
View larger version (79K):
[in this window]
[in a new window]
|
Fig. 3.
Silver staining of the GLUT4-containing
membranes purified by equilibrium density gradient centrifugation.
Aliquots from fractions of the equilibrium gradients from Fig. 2 were
subjected to gradient SDS-PAGE (5-15% acrylamide), and proteins were
visualized by silver stain. Shown are fractions close to the
top of gradient that contain proteins sensitive (fractions
6-9) or not (fractions 10-13) to insulin treatment. No bands were
detected by silver stain in fractions 1-4. Depicted is a gel
representative of four different experiments.
|
|
A second set of experiments was conducted on the membranes fractionated
by equilibrium gradient centrifugation to determine the distribution of
transferrin receptors, thought to be present in endosomal membranes,
and VAMP2, thought to be associated with insulin-sensitive
GLUT4-containing membranes (18, 19). Surprisingly, both these proteins
were present in the fractions that were responsive to insulin and their
distributions were more restricted to these fractions than was GLUT4
itself (Fig. 4). These data suggest that the insulin-sensitive membranes in these fractions are contaminated by
recycling endosomes or that transferrin receptor is present in the
insulin-sensitive membranes or both. The restriction of VAMP2 to the
insulin-sensitive fractions is consistent with data showing that VAMP2
function is necessary for GLUT4 translocation to the plasma
membrane in response to insulin (23, 24).
View larger version (19K):
[in this window]
[in a new window]
|
Fig. 4.
Distribution of VAMP2, transferrin receptor,
and GLUT4 in equilibrium density gradient fractions. Aliquots from
fractions of the equilibrium gradients from Fig. 2 were subjected to
SDS-PAGE and Western blot analysis as described under "Experimental
Procedures," using anti-GLUT4, anti-transferrin receptor
(TfR), and anti-VAMP2 antibodies as indicated. Data shown
are representative of three different experiments.
|
|
Identification of Cytoskeletal Proteins in GLUT4-containing
Membranes--
In order to identify proteins present in the
insulin-sensitive membranes containing GLUT4, in separate experiments
the equivalent of fractions 7 and 8 shown in the experiment of Fig. 3
were pooled and analyzed by SDS-PAGE, and the gels were silver-stained
(Fig. 5). These results confirmed that
many of the resident proteins in the membranes derived from
insulin-treated cells were present at lower abundance compared with
controls (Fig. 5, arrowheads). Many of the protein bands,
combined from both lanes, were subjected to tryptic hydrolysis, and the
peptides were analyzed by mass spectrometry. Of the proteins identified
by this procedure, peptides derived from GLUT4 itself appeared in two
closely spaced bands. Remarkably, the lower of these bands also
contained a peptide corresponding to the phosphorylated form of the
COOH terminus of GLUT4 (Fig. 5), indicating that significant amounts of
phosphorylated GLUT4 are present in insulin-sensitive membranes. In
addition, peptides corresponding to several proteins previously
reported to be present in these membranes were identified, including
the insulin-like growth factor-II/mannose 6-phosphate receptor, IRAP, amine oxidase, long chain acyl-CoA synthetase, and secretory
carrier-associated membrane proteins (Fig. 5). Two proteins not
previously known to be present in insulin-sensitive GLUT4-containing
membranes were also identified: vimentin, an intermediate filament
subunit, and -tubulin, the microtubule protein.
View larger version (48K):
[in this window]
[in a new window]
|
Fig. 5.
Sequences of protein peptides obtained from
insulin-sensitive GLUT4-containing membrane. Fractionation of
GLUT4-containing membranes by equilibrium density gradient as described
in the legend to Fig. 2 was performed, and the insulin-sensitive
fractions, from control (CON) or insulin-stimulated cells
(INS), were pooled, pelleted, resolved by a 5-15%
gradient SDS-PAGE, and visualized with silver stain. Bands were excised
and subjected to proteolytic digestion, and peptides were identified by
mass spectrometry. The sequences of peptides obtained from some
insulin-sensitive bands are indicated. The arrows indicate
the insulin-sensitive bands. SCAMPs, secretory
carrier-associated membrane proteins.
|
|
Two approaches were taken to determine if vimentin and -tubulin are
actually directly associated with membrane vesicles that also contain
GLUT4 and are insulin-sensitive. First, the membrane preparations
obtained from the equilibrium gradient centrifugation were analyzed by
immunoelectron microscopy using anti-GLUT4, anti-vimentin, and
anti-tubulin antibodies. As shown in Fig.
6, most of the vesicles in our
preparations are reactive with anti-GLUT4 (large colloidal gold
particles), indicating relatively low contamination with membranes that
do not contain the transporter. A fraction of these GLUT4-positive
membrane vesicles also directly react with anti-vimentin (Fig. 6,
C and D, small colloidal gold particles) and
anti-tubulin (Fig. 6, E and F, small colloidal
gold particles). Nonimmune antibodies showed no detectable staining of
these membranes under the conditions of these experiments, while the
anti-GLUT4 was readily detected (Fig. 6, G and
H). These results indicate that some GLUT4-containing membrane vesicles are associated with the cytoskeletal protein vimentin, -tubulin, or both.
View larger version (71K):
[in this window]
[in a new window]
|
Fig. 6.
GLUT4, vimentin, and
-tubulin colocalization in GLUT4-containing
membranes. Immunoelectron microscopy of GLUT4 vesicles from
insulin-sensitive fractions, isolated by an equilibrium density
gradient as described in the legend to Fig. 2 was performed.
GLUT4-containing membranes were fixed in 2% paraformaldehyde and
adsorbed to Formvar-coated gold grids. The grids were labeled with
specific primary antibodies and were detected using secondary
antibodies conjugated to different sized gold particles. A
and B, colocalization of GLUT4 (15 nm, large
arrowheads) and IRAP (5 nm, small arrows). C
and D, colocalization of GLUT4 (15 nm, large
arrowheads) and vimentin (5 nm, small arrows).
E and F, colocalization of GLUT4 (15 nm,
large arrowheads) and -tubulin (5 nm, small
arrows). G and H, grids were labeled with
anti-GLUT4 antibody (15 nm, large arrowheads) and nonimmune
mouse IgG antibody.
|
|
To further assess association of vimentin and -tubulin with
insulin-sensitive membranes, the abundance of these cytoskeletal proteins was estimated using Western analysis in each of the membrane fractions obtained by equilibrium gradient centrifugation. Fig. 7 shows a comparison of the relative
abundance of GLUT4 protein versus vimentin and -tubulin
throughout these fractions. Both vimentin and -tubulin are present
in all of the membrane fractions of the gradient except for the top few
fractions. Strikingly, both these proteins are greatly reduced in
abundance in the same gradient fractions in which GLUT4 is also reduced
in response to the action of insulin. In membrane fractions of higher
density, the concentrations of GLUT4, vimentin, and -tubulin are all
unaffected by prior treatment of cells with insulin. Taken together,
the experiments depicted in Figs. 5-7 demonstrate that two
cytoskeletal proteins, vimentin, and -tubulin, are bound to
subpopulations of the GLUT4-containing membranes that are
insulin-responsive in rat adipocytes.
View larger version (30K):
[in this window]
[in a new window]
|
Fig. 7.
Vimentin and
-tubulin associate with insulin-responsive
GLUT4-containing membranes isolated by equilibrium density
gradient. Aliquots of fractions from equilibrium gradient
centrifugations performed as shown in Fig. 2 were subjected to SDS-PAGE
and Western blot analysis as described under "Experimental
Procedures." A, anti-vimentin and anti-GLUT4 immunoblots
of fractions from an experiment using primary rat adipocytes treated
(+) or not ( ) with insulin. B, anti--tubulin and
anti-GLUT4 immunoblots from another representative experiment using
primary rat adipocytes treated (+) or not () with insulin.
|
|
Perinuclear Localization of GLUT4 Depends on Cytoskeletal
Integrity--
The intermediate filament protein vimentin has been
previously implicated in playing a role in intracellular trafficking of proteins and membranes (43-45) and appears also to function in localization of intracellular structures (44, 45). It is also well
established that microtubules and their associated motor proteins
dynein and kinesins direct the cellular localization of cellular
organelles such as mitochondria, the endoplasmic reticulum, and Golgi
membranes (for a review, see Ref. 46). Based on the findings presented
in Figs. 5-7 indicating that some GLUT4-containing membranes prepared
from 3T3-L1 adipocytes are associated with tubulin as well as vimentin,
we tested whether the intact cytoskeleton is required for perinuclear
GLUT4 localization. To address this question, experiments were
conducted to disrupt intermediate filaments in 3T3-L1 adipocytes by
microinjection of peptide Vm-1A (RVTMQNLNDRLASYLDKV), derived from the
helix initiation 1A domain of vimentin, which has been previously shown
to cause disassembly of intermediate filaments in other cell types
(47). Fig. 8B confirms
previous findings that cultured adipocytes do contain an extensive
network of intermediate filaments (48, 49) and documents that
introduction of this peptide into 3T3-L1 adipocytes disrupts this
filamentous network (47).
View larger version (72K):
[in this window]
[in a new window]
|
Fig. 8.
Effect of microinjection of Vm-1A peptide on
the integrity of 3T3-L1 adipocyte microtubules and intermediate
filaments. 3T3-L1 adipocytes were microinjected with a solution
containing 20 µg/ml FITC-coupled dextran and 2 mg/ml Vm-1A peptide.
Following microinjection, the cells were allowed to recover for 60 min,
fixed, and stained with monoclonal anti--tubulin (A,
top panel) or anti-vimentin (B, lower
panel) antibodies. Primary antibodies were detected using
Cy5-conjugated anti-mouse and Cy3-conjugated anti-goat antibodies.
Microinjected cells were identified via FITC-coupled dextran.
|
|
Consistent with the known integration of intermediate filaments with
microtubules via interconnecting proteins in mammalian cells (50),
microinjection of peptide Vm-1A into cultured adipocytes was also found
to disassemble microtubules (Fig. 8A). This observation is
similar to published data in other cell systems (47). Microinjection into 3T3-L1 adipocytes of a peptide identical to Vm-1A except for a
single residue change (His substituted for Arg10), which
has been shown to be ineffective in disrupting vimentin dimerization
(47), failed to alter cytoskeletal morphology (not shown).
Fig. 9 depicts the results of
microinjection of Vm-1A peptide or its biologically inactive R10H
homolog (control peptide) on the intracellular GLUT4 distribution in
3T3-L1 adipocytes after 60 min. Uninjected adipocytes or those
microinjected with control peptide mostly display a normal, perinuclear
localization of GLUT4, with some of the glucose transporter also
distributed in vesicular structures throughout the cytoplasm. In
contrast, GLUT4 in many of the cultured adipocytes microinjected with
the Vm-1A peptide appears dispersed away from the juxtanuclear regions
of the cells (Fig. 9). In some cases, the GLUT4 in these latter cells
apparently remains associated with several large membrane clusters,
whereas in other cells the dispersion appears more complete. The Vm-1A peptide-induced movement of GLUT4 away from the juxtanuclear regions correlates with the disassembly of both intermediate filaments and
microtubules (Figs. 8 and 9).
View larger version (116K):
[in this window]
[in a new window]
|
Fig. 9.
Microinjection of Vm-1A peptide into 3T3-L1
adipocytes disperses perinuclear GLUT4 compartments. 3T3-L1
adipocytes were microinjected with a solution containing 20 µg/ml
FITC-coupled dextran and 2 mg/ml control peptide (middle
panel) or 2 mg/ml of Vm-1A peptide (lower panel).
Following microinjection, the cells were allowed to recover for 60 min,
and then they were fixed and stained with rabbit polyclonal anti-GLUT4
antibody. Primary antibody was detected using Cy3-conjugated
anti-rabbit antibody. Microinjected cells were identified by
FITC-coupled dextran.
|
|
Images similar to those obtained with the Vm-1A peptide above were also
observed when 3T3-L1 adipocytes are treated with the microtubule-disrupting agent nocodazole (not shown). The microtubule motor dynein is known to be required for the perinuclear localization of endoplasmic reticulum and Golgi membranes, as well as the recycling endosome that is transited by the transferrin receptor (51, 52). Heuser
(40) showed that dynein motor activity can be reversibly disrupted
without inhibition of kinesin function by slight acidification of the
cytoplasm, causing perinuclear lysosomes to disperse to the cell
periphery. Using this method, we monitored the disposition of GLUT4 in
3T3-L1 adipocytes immediately after a 15-min treatment with acetate
buffer, pH 6.4, or following a recovery period of 3 h in
physiological buffer. Fig. 10 shows
that cytoplasmic acidification of 3T3-L1 adipocytes causes dispersion of much of the perinuclear GLUT4 to vesicular structures throughout the
cytoplasm. Subsequent incubation of the acetate-treated adipocytes for
3 h at physiological pH restored normal perinuclear GLUT4 localization. Thus, the data in Fig. 10 show in cultured adipocytes a
tight correlation between a treatment known to disrupt the
microtubule-based motor dynein and the perinuclear disposition of
GLUT4.
View larger version (42K):
[in this window]
[in a new window]
|
Fig. 10.
Inhibition of dynein motor activity by
cytoplasmic acidification disperses perinuclear GLUT4. 3T3-L1
adipocytes were treated or not with 70 mM sodium acetate,
pH 6.4, for 15 min and washed with KHR, and then cells were fixed
(middle panel) or allowed to recover for 3 h and then
fixed (right panel). Cells were then stained with rabbit
anti-GLUT4 antibody followed by a FITC-coupled secondary anti-rabbit
antibody. Stained cells were observed as described under
"Experimental Procedures." The data shown are representative from
three different experiments.
|
|
The reversible dispersion of GLUT4 from the perinuclear region of
3T3-L1 adipocytes in response to cytoplasmic acidification offers the
opportunity to test whether perinuclear localization of GLUT4 is
necessary for the responsiveness of GLUT4 to insulin-mediated translocation to the cell surface. Fig.
11 shows that a 15-min acetate
treatment of 3T3-L1 adipocytes, which causes dispersion of GLUT4, prior
to washing at physiological pH and incubation with insulin for 10 min
blocks the hormone's effect on recruitment of GLUT4 to plasma
membranes. In these experiments, cell surface GLUT4 is estimated by the
binding of anti-GLUT4 antibody to plasma membrane sheets that are
adsorbed to glass coverslips following light sonication of attached
cells. When acetate-treated cells are allowed to recover for 60 min at
physiological pH, GLUT4 returns to a perinuclear localization (not
shown for these experiments; see Fig. 10), and insulin responsiveness
of GLUT4 measured by its translocation to the cell surface membrane is
also restored (Fig. 11). These data document a correlation between
normal localization of GLUT4 in the juxtanuclear region of cultured
adipocytes and sensitivity of intracellular GLUT4 to insulin
action.
View larger version (42K):
[in this window]
[in a new window]
|
Fig. 11.
Effect of inhibition of dynein activity by
cytoplasmic acidification on insulin-stimulated GLUT4
translocation. A, 3T3-L1 adipocytes were treated or not
with 70 mM sodium acetate for 15 min, washed with KHR, and
treated or not with 100 nM insulin for 10 min. Another
group of cells were allowed to recover for 60 min and then were treated
with 100 nM insulin for 10 min. Lawns of plasma membrane
were then generated as described under "Experimental Procedures."
The lawns were incubated with rabbit anti-GLUT4 antibody followed by
FITC-coupled secondary anti-rabbit antibody and GLUT4 stain visualized
as described under "Experimental Procedures." B, GLUT4
protein in the lawns shown in A were quantified by measuring
the fluorescence intensity using Photoshop analysis software. Data
correspond to the average of four experiments ± S.E.
|
|
| |
DISCUSSION |
The major conclusion of the studies presented here is that the
characteristic perinuclear localization of GLUT4 in 3T3-L1 adipocytes
requires an intact cytoskeleton and that this requirement may reflect
the action of the microtubule-based motor dynein. Disruption of the
intermediate filament and microtubule systems in these cells by
microinjection of the vimentin-derived peptide Vm-1A (Fig. 8) or by the
microtubule-disrupting agent nocodazole (not shown) leads to dispersion
of the perinuclear GLUT4 into the cell periphery (Fig. 9). It is well
established that intermediate filaments and microtubules are highly
integrated in mammalian cells, and disruption of either of these
cytoskeletal elements has deleterious effects on the other (47, 50).
Our data are consistent with this concept in that introduction of the
vimentin-derived peptide not only disperses intermediate filaments in
the cultured adipocytes but also disrupts microtubules (Fig. 8).
Interestingly, vimentin-containing intermediate filaments have been
reported to surround small lipid droplets in cultured adipocytes during differentiation and may have a function in lipogenesis and lipid storage (48). Also, reorganization of intermediate filaments is
apparently mediated by Cdc2-directed vimentin phosphorylation during mitosis and may play a role in vesicle orientation during cell
division (43). However, ablation of the vimentin gene does not cause
major phenotypic abnormalities in mice (53). Thus, while it is not
possible to determine whether intermediate filament disruption or
microtubule disruption or both are required to disperse perinuclear
GLUT4 to the cell periphery, it seems likely that microtubules play the
major role.
Our findings that vimentin-derived peptide or nocodazole causes
dispersion of perinuclear-localized GLUT4 are consistent with a large
body of literature strongly supporting a direct role of the
microtubule-based cytoskeleton in localizing cellular organelles (54,
55). Membranes of the endoplasmic reticulum and Golgi complex that are
retained in juxtanuclear regions of cells are dispersed to peripheral
regions upon disruption of microtubules (56, 57), similar to our
findings with GLUT4. Thus, the perinuclear membrane system that
sequesters GLUT4 in unstimulated 3T3-L1 adipocytes appears to be
localized within cells by the same mechanisms that regulate the
localization of other major cellular organelles.
A principal role for microtubules in membrane trafficking and
organization reflects the balance of molecular motor activity directed
toward the plus ends (cell periphery) versus minus ends (perinuclear region) of microtubules (for reviews see Refs. 46 and 54).
The microtubule-based motor dynein has been shown to direct many
membranes to the juxtanuclear regions of cells, including endosomes
(52), melanosomes (58), Golgi (51, 59), and lysosomes (60). Upon
inhibition of dynein activity, these membranes disperse out to the cell
periphery. Kinesins are microtubule-based molecular motors that in
general direct protein and organelle movement toward the plus ends of
microtubules (for a review, see Ref. 46). For example, the dispersion
of mitochondria throughout the cytoplasm is dependent upon kinesin
function (61). Heuser (40) showed that dynein activity as opposed to
kinesin activity could be selectively inhibited by a cytoplasmic
acidification procedure, which was subsequently confirmed by the
demonstration that the dynein motor complex is disrupted under these
conditions (62). As has been shown for other perinuclear organelles in response to brief treatment of cells at pH 6.4, perinuclear GLUT4 in
cultured adipocytes was found in the present study to disperse into the
cell periphery upon acidification of the cytoplasm (Fig. 10). This
response is apparently completely reversible upon incubation of the
cells at physiological pH for 3 h. These data are consistent with
the hypothesis that the microtubule-based motor dynein functions to
localize GLUT4-containing membranes into the perinuclear regions of
cultured adipocytes.
The endosomal membrane system in which the transferrin receptor and
other proteins continuously recycle between the cell surface membrane
and intracellular membranes is also partially localized in a
perinuclear disposition, which is disrupted by nocodazole (63, 64).
However, exocytosis of transferrin receptor from this compartment is
not impaired by this treatment, apparently dissociating the cellular
position of this organelle from its function (65). In the case of
regulated GLUT4 exocytosis, there may be such a relationship.
Disruption of dynein by incubation of 3T3-L1 adipocytes at low pH for
15 min followed by return to pH 7.4 did significantly inhibit the
ability of insulin to cause translocation of GLUT4 to the plasma
membrane (Fig. 11). Restoration of GLUT4 to its perinuclear disposition
following a 1-h recovery period at physiological pH occurred in
conjunction with a return of normal insulin responsiveness. While these
data will require confirmation with alternative techniques to disrupt
dynein function, they suggest the possibility that dynein motor
activity is necessary to confer insulin sensitivity to GLUT4-containing
membranes in the juxtanuclear region of adipocytes. Further, more
detailed experiments are under way to test this hypothesis.
One of the fundamental questions about the mechanism by which insulin
regulates GLUT4 is whether insulin causes intracellular membranes that
contain GLUT4 to actually move and fuse with the plasma membrane or
whether insulin causes only GLUT4 itself to move more rapidly through
constitutively recycling membrane compartments (3, 4, 66). The former
possibility is currently favored, because proteins such as VAMP2 that
are involved in membrane docking and fusion and are present in
GLUT4-containing membranes are also translocated to the plasma membrane
in response to insulin (23, 24). However, this data is indirect. Here
we provide strong evidence in favor of this hypothesis by directly
monitoring the mass of intracellular membranes enriched in GLUT4. A new
method to prepare GLUT4-containing membranes without the use of
anti-GLUT4 antibodies was devised for this purpose (Fig. 2). This
method is simple in that it employs a single sucrose equilibrium
gradient centrifugation step to further purify GLUT4-containing
membranes obtained by a previously published velocity gradient
centrifugation procedure (34, 35). Following both centrifugation steps,
the method yields about 10 µg of purified GLUT4-containing membranes from a starting preparation of about 1 mg of primary rat adipocyte low
density microsomes. Importantly, insulin treatment of adipocytes prior
to homogenization causes a lower yield of these intracellular membranes
(Figs. 2-4), consistent with the concept that insulin action directs
membranes contained in this fraction to the plasma membrane, where they fuse.
It is likely that the purified membrane preparations we obtain that
exhibit maximum insulin sensitivity (e.g. fractions 7 and 8 of Fig. 3) contain various other cell membranes in addition to the
GLUT4-containing membranes that represent the insulin-sensitive compartment. Thus, much of the transferrin receptor is present in these
fractions (Fig. 4), and this receptor is thought to be present mostly
in recycling endosomes that apparently show only modest sensitivity to
insulin (12, 16). However, this problem also applies to
GLUT4-containing membrane preparations obtained by immunoadsorption
with anti-GLUT4 antibodies because GLUT4 is also present in the
recycling endosome compartment (12, 15-17). Furthermore,
immunoelectron microscopy shows that virtually all of the membrane
vesicles obtained by the method described here do actually contain
GLUT4 (Fig. 6). This indicates that there is little contamination of
the insulin-sensitive membranes by membranes in which GLUT4 does not
transit. The great advantage of the present technique is that it allows
for the first time the purification of membranes that can be observed
to be insulin-sensitive without the use of reagents that can
contaminate the preparations such as immunoglobulin polypeptides. This
contrasts with other methods where membranes enriched with GLUT4 are
obtained but do not seem to be reduced in abundance when cells are
treated with insulin (16, 67). The GLUT4-containing membranes purified by the method described here thus appear to be particularly excellent for the application of microsequencing techniques to identify resident proteins.
It is interesting to note that significant amounts of the GLUT4 present
in the membrane fractions obtained by velocity gradient centrifugation
of low density microsomes is not apparently regulated by insulin, as
revealed by further separation of these membranes by sucrose
equilibrium gradient centrifugation (compare Figs. 1 and 2). Membranes
in fractions 10 and greater in the latter gradient show no detectable
decrease when obtained from insulin-treated cells. These membranes are
unlikely to be derived from the plasma membrane because there is little
contamination of low density microsomes by plasma membranes (6), and
there is little GLUT4 present at the cell surface in unstimulated
adipocytes. Thus, there appears to be GLUT4 in intracellular membrane
compartments that are exclusive of the recycling endosomal system and
the insulin-sensitive compartment. Further work will be required to
determine the cellular origin of these membranes that contain GLUT4.
The sucrose equilibrium density gradient step of the procedure
described here reveals interesting protein profiles of the various
membrane fractions that are separated. Surprisingly, many of the major
protein bands that are visualized in the gel lanes after SDS-PAGE and
silver staining are common throughout the gradient (Fig. 3). This
indicates that the membranes containing GLUT4 that are translocated to
the cell surface in response to insulin contain many proteins that are
also present in membranes not responsive to insulin. These results
imply that relatively few major proteins may be unique to the
insulin-sensitive membrane compartment or compartments. Perhaps a
relatively small number of proteins is required to confer insulin
sensitivity to this compartment. Alternatively, many additional
proteins may be present in the GLUT4-containing membranes that are
insulin-responsive but at much lower abundance than these major
proteins. Further work will be required to sort out this question. The
methodology developed here should greatly facilitate the identification
of low abundance proteins present in the insulin-sensitive
GLUT4-containing membranes. Current experiments are directed to this end.
,
TOP
ABSTRACT
INTRODUCTION
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
