| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
* The Beatson Institute for Cancer Research, Garscube Estate, Glasgow G61 1BD, U.K.;
GSF-Forschungszentrum für Umwelt und Gesundheit, Institut für Klinische Molekularbiologie und Tumorgenetik, D-81377 München, Germany;
Institute De Biochimie, Universite Lausanne, CH 1066 Epalinges, Switzerland;
§ Dana-Farber Cancer Institute, Boston, Massachusetts 02115, USA;
¶ The Salk Institute, La Jolla, California 92037, USA;
Institute of Microbiology and Genetics, Vienna Biocenter, A-1030 Vienna, Austria; and
Department of Nephrology, Medizinische Hochschule Hannover, 30625 Hannover, Germany
1Correspondence: The Beatson Institute for Cancer Research, Garscube Estate, Switchback Road, Bearsden, Glasgow G61 1BD, U.K. E-mail: pjanosch{at}beatson.gla.ac.uk
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: Raf • phosphorylation • cytoskeleton • casein kinase 2
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2
3
). The molecular
mechanism of Raf-1 activation is complex and still incompletely
understood, but in many situations involves the small G-protein Ras. In
its GTP-loaded state, Ras binds to Raf-1 with high affinity (reviewed
in refs 4
, 5
) resulting in the translocation of Raf-1 to
the plasma membrane (6
, 7)
, where Raf-1 becomes activated
by a still unknown mechanism, which involves phosphorylation on
tyrosine and/or serine residues (reviewed in ref 8
).
Activation may be further enhanced by a lipid cofactor
(9)
.
Activated Raf-1 phosphorylates and activates MEK (10
11
12)
,
a dual specificity kinase that in turn phosphorylates and activates
MAPK/ERK, which propagates the signal to nuclear transcription factors,
most notably the ternary complex factor, which is required for the
transcription of the c-fos gene (13
, 14)
. As
constitutively activated MEK mutants can transform NIH 3T3 cells and
induce differentiation of PC12 cells, thereby mimicking the effects of
v-raf, it has been suggested that activation of the MEK/ERK cascade is
the main, if not sole, biological function of Raf-1 (15)
.
Several lines of evidence indicate, however, that Raf-1 may signal
independently of the MEK/ERK pathway. First, v-raf can transform Rat
fibroblasts without activation of ERKs (16
, 17)
. Second,
Ras mutants have been isolated that still bind Raf-1 and consequently
lead to ERK activation, but fail to induce DNA synthesis or
transformation (18
, 19)
. Third, oncogenic Raf-1 can induce
the differentiation of 3T3L1 cells into adipocytes without activating
ERKs (20
, 21)
. Collectively, these findings suggest the
existence of other effectors and substrates of Raf-1.
Therefore, we set out to identify new Raf-1 signaling pathways and
discovered that Raf-1 associates with vimentin and vimentin kinases
that regulate the architecture of vimentin filaments. Vimentin is the
main constituent of intermediate filaments in mesenchymal cells.
Vimentin filaments are dynamic structures that are constantly being
rebuilt by an ongoing exchange of vimentin protomers between the
soluble and polymeric phases (reviewed in refs 22
, 23
).
Although the exact function of the vimentin skeleton is still unknown,
it is clear that its architecture is regulated in response to many
extracellular stimuli and during the cell cycle. In part this
regulation is exerted through phosphorylation, and vimentin has been
described as a target for several kinases (reviewed in ref
24
). Here we show that Raf-1 can signal vimentin
rearrangements via the regulation of associated vimentin kinases that
are independent of the MEK/ERK pathway.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. The MEK inhibitors PD098059 and U0126
were purchased from New England Biolabs (Beverly, Mass.) and Promega
(Madison, Wis.), respectively.
Isolation and identification of vimentin
3 x 109 logarithmically growing NIH
3T3 cells were lysed in TBS-1% Triton (TBST: 20 mM TrisHCl, pH 7.4,
150 mM NaCl, 2 mM EDTA, 1% Triton-X100, 1 mM PMSF, 10 µM leupeptin).
The lysate was cleared by centrifugation at 21,000 g for 10
min and incubated with glutathione Sepharose beads (Pharmacia,
Piscataway, N.J.) for 1 h at 4°C. The beads were loaded with
5 µg of GST or GST-Raf-1 proteins that were expressed in the
baculovirus/Sf-9 cell system and purified as described
(26)
. Beads were washed with TBST five times before
GST-Raf-1 and associated proteins were released with sodium dodecyl
sulfate (SDS) gel sample buffer (3% SDS, 10 mM DTT, 20 mM TrisHCl, pH
6.8), separated on a 10% SDS gel, and blotted onto nitrocellulose. The
blot was stained with Ponceau S. Proteins were cut out from the blot
and digested with sequencing grade trypsin (Boehringer Mannheim,
Mannheim, Germany) (27)
. Tryptic peptides were dissolved
in 60 µl of 0.1% trifluoroacetic acid and separated on an Aquapore
300 Angström C8 column (1x30 mm) using an Applied Biosystems
Model 172 Microbore high-performance liquid chromatography (HPLC).
Peptide peaks were microsequenced in a Model 477A gas phase sequenator
(Applied Biosystems, Foster City, Calif.).
Vimentin phosphorylation and polymerization assays
Vimentin was expressed in Escherichia coli and
purified as described (28)
. Vimentin preparations in 8M
urea were dialyzed against 10 mM TrisHCl, pH 7.4, 1 mM DTT, and 1 mM
EDTA. Insoluble material was removed by centrifugation at 21,000
g for 20 min. The supernatant containing soluble vimentin
protomers was incubated with purified GST-Raf-1 proteins bound to
glutathione Sepharose in a low-salt Raf-1 kinase buffer (20 mM TrisHCl,
pH 7.4, 20 mM NaCl, 10 mM MgCl2) supplemented
with 2 µM ATP and 2.5 µCi of [32P]-
-ATP.
In the course of experimentation, we discovered that low urea
concentrations do not affect Raf-1 kinase activity and yielded
identical results. Therefore, urea-containing vimentin preparations
were used in the kinase assays shown in Fig. 3
. GST-Raf proteins were
produced in Sf-9 cells, activated, and purified as described
(26)
. The protein kinase C
(PKC) -activated Raf
preparations were devoid of detectable contamination by PKC.
Inclusion of a specific PKC inhibitor (1 µM GF203109X, Biomol,
Plymouth Meeting, Pa.) did not alter Raf-1 mediated vimentin
phosphorylation. Kinase reactions were carried out as described
previously (26)
. After the phosphorylation reaction,
vimentin polymerization was induced by addition of NaCl to a final
concentration of 150 mM and allowed to proceed for 30 min at room
temperature. Vimentin polymers were pelleted by centrifugation at
21,000 g for 20 min (28)
. The pellet was washed
twice by vigorous resuspension in TBST before boiling in SDS gel sample
buffer. Samples were separated by SDS-gel electrophoresis and blotted
onto nitrocellulose membranes. The blots were autoradiographed and
subsequently stained with vimentin or Raf antisera as indicated.
Soluble activated Raf-1 was produced as follows. GST-Raf-1 was
activated by coexpression of Ras plus Lck in Sf-9 cells. After
purification the Raf-1 portion was released from the glutathione
Sepharose beads by digestion with 0.5 µg thrombin. Thrombin was
inactivated by addition of 10 µg/ml leupeptin.
|
Phosphopeptide mapping
To assay vimentin phosphorylation in vivo, COS-1
cells were transfected with a v-raf expression plasmid
(29)
. For transfection, 20–40% confluent COS-1 cells
were incubated in 6-well plates (Nunc, Roskilde, Denmark) containing 2
ml DMEM with glutamine and 10% NU-serum (Collaborative Research,
#55000) per well; 5 µg of 3611-v-raf plasmid DNA and 20 µl DC-mix
(10.3 mg/ml chloroquin phosphate, Sigma, and 80 mg/ml DEAE-dextrane,
Sigma D-9885) were added to each well. After 4 h the medium was
replaced by phosphate-buffered saline (PBS) containing 10% DMSO for 2
min. Then cells were fed with DMEM plus 10% FCS. Two days after
transfection, cells were incubated in serum-free medium overnight
before harvest. Two days after transfection, COS-1 cells were serum
starved overnight, followed by additional incubation in phosphate-free
medium (Sigma, St. Louis, Mo.) for 3 h. Then cells were
metabolically labeled by addition of 0.5 mCi/ml
[32P]-orthophosphoric acid for 3 h. Cells
were lysed and vimentin was extracted as described (30)
.
In parallel, purified vimentin was phosphorylated by Raf-1 in
vitro. Samples were resolved by SDS-gel electrophoresis. Vimentin
bands were cut out and digested with sequencing grade trypsin (Promega)
according to the instructions provided by the manufacturer. Tryptic
phosphopeptides were processed for 2-dimensional peptide mapping as
described (26)
and spotted onto 20 x 20 cm
phosphocellulose plates (Merck, Rahway, N.J.). In the first dimension,
peptides were resolved by electrophoresis in pH 8.9 buffer at 1250
volts for 15 min; the second dimension was chromatography in
n-butanol/pyridine/acetic acid/water (15/10/3/12) for 20 h.
Immunoprecipitation and Western blotting
Approximately 1 x 107 growing NIH
3T3 cells were lysed in TBST buffer. To increase the yield of soluble
vimentin, NaCl was omitted from the lysis buffer in some experiments.
Lysates were cleared by centrifugation at 21,000 g for 10
min and immunoprecipitated with crafVI antiserum (1.5 µl per ml of
lysate) or the corresponding preimmune serum. The crafVI serum was
raised by immunizing rabbits with a peptide corresponding to the unique
carboxy-terminal 12 amino acids of Raf-1 (26)
.
Immunoprecipitates were washed four times with TBST, then resolved by
electrophoresis on 7.5% SDS gels and transferred to nitrocellulose
membranes. Blots were stained with anti-vimentin antibody (Vim3B4,
Boehringer) using the enhanced chemiluminescence kit (Amersham,
Arlington Heights, Ill.) as described previously (26)
. The
antibodies were removed by incubating the membrane in 3% SDS and 10 mM
DTT for 15 min. Subsequently, Raf-1 was detected by staining with the
PBB1 monoclonal Raf antibody (31)
. Soluble vimentin was
precipitated with the V9 (DAKO, Carpinteria, Calif.) monoclonal
antibody and Raf-1 was detected by staining with PBB1.
Cell fractionation
For the experiment shown in Fig. 8d
, 1 x
106 COS-1 cells were transfected with GST-BXB and
MEK-DD (32)
expression plasmids using Superfect (Qiagen,
Chatsworth, Calif.) according to the manufacturer’s instructions. Two
days after transfection, the cells were treated with 50 µM U0126
(Promega) for 8 h and then lysed in 400 µl TBST. The lysates
were separated into soluble and insoluble fractions by centrifugation
at 20,000 g for 30 min. The insoluble pellet was sonicated
and boiled in 400 µl SDS-gel sample buffer. Blots were sequentially
stained with antibodies against vimentin (V9, DAKO), phospho-ERK
(Sigma) and pan-ERK (Sigma). For the experiment shown in Fig. 8c
, 1 x 107 BXB-ER cells were
labeled with 0.5 mCi/ml [32P]-orthophosphoric
acid for 3 h before cells were treated with 1 µM estradiol
(Sigma) as indicated. Lysates were prepared as above and vimentin was
immunoprecipitated from the soluble fraction with the V9 (DAKO)
antibody.
|
In vitro vimentin binding experiments
For the experiments shown in Figs. 2d
, e
, 1 x 108 logarithmically growing NIH 3T3 cells were
lysed in 12 ml of 20 mM TrisHCl, pH 7.5, 0.1 mM EDTA, 1% Triton X-100
plus phosphatase, and protease inhibitors. Lysates were cleared by
centrifugation at 20,000 g for 5 min and the supernatants
were incubated with the indicated GST-Raf proteins immobilized on
glutathione agarose beads (Pharmacia) for 12 h at 4°C. Beads
were washed in lysis buffer four times, separated on 12.5% SDS-gels,
and sequentially immunoblotted with antibodies to vimentin (V9, DAKO;
and goat anti-vimentin polyclonal antibody, Sigma) and GST (Santa Cruz,
Santa Cruz, Calif.). GST, GRS, GNX, and GST-Raf proteins were expressed
in Sf-9 insect cells and purified as described (26)
. For
the experiments shown in Fig. 2e
, the GST-tagged Raf-1
kinase domain GNX was expressed in E. coli. The GNX mutants
were generated by progressive deletions from the carboxyl terminus and
are described in (33)
. For the experiments shown in Fig. 2g
, NIH cells were serum starved overnight and stimulated
with TPA (100 ng/ml) as indicated. Synthetic peptides were coupled to a
chromatographic matrix using the Ultralink kit (Pierce, Rockford,
Ill.). The sequence of the 259 peptide is
QRQRSTS259TPNVHC, the sequence of the 621 peptide
is KINRSAS621EPSLHRC. Corresponding
phosphopeptides were synthesized with phosphoserine at positions 259
and 621, respectively. Incubation with cell lysates was done as
described above, but only for 2 h. Associated proteins were cut
out from Coomassie-stained gels, digested with trypsin, and identified
via mass determination of tryptic peptides by mass spectrometry on a
Bruker MALDI-TOF. MASCOT (www.matrixscience.com) was used for databases
queries.
|
Immunofluorescence
For immunofluorescence experiments, human HF-5 or mouse NIH 3T3
fibroblasts were grown on chamber slides (Nunc), washed with PBS, and
fixed in methanol for 10 min at -20°C. Samples were blocked with
10% serum in PBS for 30 min and incubated with primary antibodies
diluted in PBS plus 10% serum for 2 h at room temperature or for
16 h at 4°C. FITC-coupled anti-mouse immunoglobulin G (IgG)
(Jackson Laboratories, West Grove, Pa.) was used to detect
anti-vimentin monoclonal antibodies and TRITC-coupled anti-rabbit
IgG was used to visualize Raf-1. Between and after incubations the
slides were washed four times with PBS plus 0.25% Triton X-100. The
crafVI antiserum was used at a 1:100 dilution or 1:1000 after affinity
purification with the corresponding peptide, respectively. To ascertain
antibody specificity, the following control experiments were performed.
When slides were incubated with mismatched combinations of primary and
secondary antibodies, i.e., crafVI (rabbit) and FITC-coupled anti-mouse
IgG, or anti-vimentin (mouse) with TRITC-coupled anti-rabbit IgG, no
staining was observed. Preincubation of the crafVI serum with the same
volume of the peptide antigen (10 mg/ml) used for immunization
abolished the Raf stain. To assure the specificity of the vimentin
stain, three different monoclonal vimentin antibodies were used
including 7A3 (34)
, clone Vim3B4 (Boehringer) and clone
65E (Affinity, Neshanic Station, N.J.). All antibodies used
individually or as mixture gave indistinguishable results. Photographs
were taken at 1000x magnification through a Zeiss Axiovert
epifluorescence microscope. To exclude ‘bleed through’ artifacts
during photography, slides stained with crafVI or vimentin antibodies
alone were included as controls.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 36)
. Therefore, we tried to exploit this property to identify
new Raf-1 substrates by isolating Raf-1-associated proteins. For this
purpose, glutathione-S-transferase (GST) -tagged Raf-1 was expressed in
the baculovirus/Sf9 cell system and purified by adsorption to
glutathione-Sepharose beads. Subsequently, immobilized GST-Raf-1 and a
control protein (GST) were incubated with NIH 3T3 lysates. After washes
with TBS-1% Triton, the GST-Raf-1 beads were eluted with SDS,
separated by SDS-polyacrylamide gel electrophoresis (PAGE), and
blotted. Several bands were found to specifically associate with Raf-1,
but not with GST (Fig. 1a
). The most abundant band migrated at 58 kDa. It was cut out
from the blot, digested with trypsin, and the tryptic peptides were
separated by reversed phase HPLC. Two of the collected fractions that
appeared to contain only a single peptide were analyzed by
microsequencing. The sequence of both tryptic peptides exhibited 100%
homology with mouse vimentin (Fig. 1b
).
|
Interaction between Raf-1 and vimentin
To test whether Raf-1 and vimentin are associated in cells, NIH
3T3 lysates were immunoprecipitated with anti-Raf-1 (crafVI) or the
corresponding preimmune serum. The Immunoprecipitates were resolved by
SDS-PAGE, blotted, and stained with a monoclonal anti-vimentin
antibody. Subsequently, the blot was stripped and stained with Raf-1
specific antiserum (Fig. 2a
). As evident from Fig. 2a
, vimentin was found in
the Raf-1 immunoprecipitate. Due to its similar size, vimentin
comigrated with the immunoglobulin heavy chains, which somewhat
distorted the vimentin band. Therefore, the specificity of the vimentin
stain was confirmed by using several different anti-vimentin antibodies
(data not shown). In the reciprocal experiment, Raf-1 was
coimmunoprecipitated by anti-vimentin antibodies but not control
immunoglobulins.
The Raf-1 association with vimentin in intact cells was further
corroborated by immunofluorescence experiments in human and mouse
fibroblasts, where cells were simultaneously stained with anti-raf-1
and anti-vimentin antibodies (Fig. 2b
, c
). The specificity of the antibodies was assured as described in Materials
and Methods. Vimentin forms a filamentous reticular network throughout
the cytoplasm. Raf-1 was diffusely distributed in the cytosol, but also
detected in fibrillar structures that showed extensive overlap with
vimentin filaments. The spread-out morphology and larger size of HF-5
facilitated visualization of the colocalization (Fig. 2b
),
but consistent findings were obtained in NIH 3T3 cells (Fig. 2c
) and after nocodazole treatment of NIH 3T3 and HF-5.
Nocodazole disrupts the ordered architecture of intermediate filaments,
resulting in a cotton ball-like appearance of vimentin filaments, which
also colocalized with Raf-1 (data not shown). These data indicate that
Raf-1 can bind to vimentin in intact cells and colocalizes with the
vimentin scaffold.
However, in vitro experiments using highly purified vimentin
and Raf-1 proteins produced in E. coli or insect cells
failed to reveal interactions (data not shown), suggesting that the
interaction between Raf-1 and vimentin was indirect and mediated by a
yet unknown component. Since vimentin binding to Raf-1 was readily
observed in crude cell lysates, we used NIH 3T3 cell lysates to map the
vimentin binding domains in Raf-1. Although vimentin bound to both the
regulatory and catalytic portions of Raf-1, the main binding domain
appeared to be located in the kinase domain (Fig. 2d
). To
refine the binding site in the Raf kinase domain, we used deletion
mutants (Fig. 2e
, f
). Vimentin binding was severely
reduced by the deletion of 28 carboxyl-terminal Raf-1 amino acids and
virtually eliminated by further deletion, indicating that the
predominant vimentin binding domain resides in the extreme carboxyl
terminus of Raf-1. The existence of at least two vimentin binding sites
in Raf-1 was serendipitously and independently confirmed in the course
of experiments where we used small peptide domains as affinity baits to
isolate the protein components of Raf-1 signaling complexes. In these
experiments vimentin was identified as a protein that binding to
peptides encompassing two major Raf-1 phosphorylation sites, serine 259
and 621, respectively (Fig. 2g
).
Althoughthe phosphorylation of either residue did not affect vimentin binding,
binding was detected only when lysates from stimulated cells were used.
The specificity of these associations was further evidenced by the
failure of vimentin to bind to any other of 5 small Raf domains used in
the same experiment. The binding to the serine 621 peptide is fully
consistent with the mapping data derived from the Raf kinase deletion
mutants (Fig. 2e
, f
), which show that the removal of
amino acids 621–649 (GNX
28) compromises vimentin binding.
Phosphorylation of vimentin
Since vimentin is a phosphoprotein (22)
, we examined
whether it is a substrate for the Raf-1 kinase. Purified activated
GST-Raf-1 produced in Sf-9 cells was washed either with TBST or RIPA
buffer. In contrast to TBST, washes with RIPA remove 14–3-3, casein
kinase 2 (CK2), and other Raf-1-associated proteins (37
, 38)
. These Raf-1 proteins were tested for phosphorylation of
purified, solubilized vimentin and recombinant kinase negative MEK,
both produced in E. coli (Fig. 3a
). Since the vimentin preparations used in this experiment
contained urea, the effect of an equivalent amount of urea on the
kinase activity of Raf-1 was examined as control. Addition of vimentin
clearly results in the appearance of a new phosphorylated band.
However, the phosphorylation of vimentin by Raf-1 can efficiently be
prevented by washing with RIPA buffer, whereas neither the RIPA washes
nor the presence of urea significantly diminished MEK phosphorylation
by Raf-1.
These results suggested that vimentin phosphorylation did not require
Raf-1 kinase activity. To confirm this notion, we repeated this
experiment using a kinase negative GST-Raf-1 mutant, Raf-301 (Fig. 3b
, c
). In Raf-301, catalytic activity has been eliminated
by replacing lysine 375 with tryptophan (39)
. As observed
with wild-type Raf-1, the phosphorylation of vimentin by TBST-washed
Raf-301 could be completely abolished by washing with RIPA buffer. In
contrast, Raf-301 failed to autophosphorylate or phosphorylate MEK
under either condition. These results clearly demonstrate that vimentin
is phosphorylated by an associated kinase rather than by Raf-1
directly. To exclude that the vimentin kinase(s) associated with the
GST portion rather than with Raf-1 itself, we tested whether GST
preparations from Sf-9 insect cells could phosphorylate vimentin (Fig. 3b
, right panel). The result clearly demonstrated that
vimentin phosphorylation is dependent on Raf-1. To test whether the
Raf-1-associated vimentin kinases also are present in mammalian cells,
Raf-1 immunoprecipitates prepared from mammalian cell extracts were
examined for vimentin phosphorylation (Fig. 3c
, right
panel).
We have previously reported that the catalytic subunit of CK2
associates with Raf-1 (37)
. To test whether CK2 was
responsible for the vimentin kinase activity associated with Raf-1, we
used heparin, a CK2 inhibitor (40)
. Heparin did not affect
MEK phosphorylation by Raf-1 (37)
. However, the addition
of heparin to TBST washed GST-Raf-1 and GST-Raf301 led to a
significant, but not complete reduction of vimentin phosphorylation by
both Raf preparations as well as by Raf-1 immunoprecipitates (Fig. 3c
). On the other hand, purified CK2 phosphorylated vimentin
(Fig. 4
) and was completely inhibited by the same dose of heparin that
partially interfered with vimentin phosphorylation by Raf-1. In
summary, these experiments suggested that Raf-1 is associated with at
least two different vimentin kinases, one of which is likely to be CK2.
|
We found that the activation status of Raf-1 showed a positive
correlation with the phosphorylation rate of vimentin (Fig. 5
). Activated and nonactivated GST-Raf and GST-BXB were prepared in Sf-9
cells. GST-Raf-1 was activated by coexpression with Ras plus Lck and
GST-BXB by coexpression with PKC (26)
. These Raf
proteins were immobilized on glutathione Sepharose beads, washed with
TBST, and incubated with purified vimentin under Raf kinase conditions.
The moderate vimentin phosphorylation that was observed when using
unactivated preparations of Raf-1 proteins increased significantly when
activated Raf-1 or BXB preparations were used. To exclude that
undetected traces of PKC in the Raf preparations were responsible
for vimentin phosphorylation, the assays were also performed in the
presence of the specific PKC inhibitor GF 109203X, yielding identical
results (data not shown). Likewise, Lck did not phosphorylate vimentin
in vitro (data not shown). These observations suggest that
Raf-1 induces vimentin phosphorylation indirectly via activation of its
associated vimentin kinases.
|
To investigate whether Raf-1 can also stimulate the phosphorylation of
vimentin in vivo, COS-1 cells were transiently transfected
with v-raf, the constitutively active homologue of Raf-1. v-raf was
used in order to selectively display Raf-dependent phosphorylation
events and to circumvent the need to activate Raf-1 by mitogen
stimulation, which induces many other kinases independent of Raf-1.
Transfected cells were serum starved and labeled with
[32P]-orthophosphoric acid (Fig. 6
). Vimentin was isolated, digested with trypsin, and the tryptic
peptides were resolved on 2-dimensional phosphopeptide maps. In
vector-transfected control cells, vimentin exhibited basal
phosphorylation of three peptides (labeled 1, 2, and 5). v-raf
expression enhanced phosphorylation of the preexisting sites and
induced four novel sites. A subset of four phosphopeptides (peptides
1–4) was common to the tryptic phosphopeptide maps of the in
vitro and in vivo samples. These results show that
vimentin is a downstream target of activated Raf-1 in vivo
and in vitro. To confirm that CK2 is one of the
Raf-1-associated vimentin kinases and to investigate which
phosphorylation sites can be attributed to CK2, we compared
phosphopeptide maps of vimentin phosphorylated by GST-Raf-1 (in the
absence and presence of heparin) and CK2 in vitro
(Fig. 7
). CK2 phosphorylated only one peptide. This corresponded to a prominent
phosphopeptide that is phosphorylated by TBST-Raf-1 and eliminated by
heparin. Due to the complexity of the phosphopeptide maps, we could not
unequivocally identify this peptide in the in vivo map.
However, multiple comparisons of several experiments strongly suggest
that it corresponds to peptide 1, which is also a major in
vivo phosphorylation site.
|
|
Raf-1 activation induces the rearrangement of vimentin filaments
independent of MEK and ERK
Since Raf-1 induced vimentin phosphorylation in vitro
and in vivo, we tested whether Raf activation could impinge
on the structure of the vimentin network in cells. We therefore used
NIH 3T3 cells, which express BXB-ER, a Raf-1 mutant whose activity can
be conditionally regulated by estrogen (25)
. In BXB-ER,
the Raf-1 kinase domain is fused to the hormone binding domain of the
estrogen receptor. The kinase activity of this fusion protein can
readily be induced by administration of estrogen, resulting in
activation of the MEK/ERK pathway and transformation of the cells
(41)
. Thus, BXB-ER allows us to study the effects of
selective Raf activation independent of the pleiotropic effects caused
by growth factors. Serum-starved BXB-ER or NIH 3T3 control cells were
treated with estrogen and the vimentin filament network was examined by
immunofluorescence (Fig. 8a
). BXB-ER activation induced a rearrangement of vimentin filaments into
cable-like bundles and the concomitant dissolution of the reticular
scaffold. Bundle formation can be triggered by hyperphosphorylation and
is typically connected with the disassembly of the vimentin scaffold,
which collapses into tight bundles in the course of being dismantled
(30
, 42)
. These changes, which ensued 2 h after the
activation of BXB-ER, were most pronounced between 4 and 8 h and
were almost completely reversed after 24 h. This time frame
clearly precedes the induction of morphological transformation, which
takes at least 24 h to manifest (25)
. Thus, the
reorganization of the vimentin scaffold was not simply a consequence of
cellular transformation. Similar changes in the vimentin skeleton were
observed after serum treatment of serum-starved BXB-ER. A quantitative
evaluation showed that estrogen and serum induced vimentin
rearrangements with similar efficiencies and kinetics (Fig. 8b
). The MEK inhibitor PD98059 did not interfere with
BXB-ER-induced vimentin rearrangement (Fig. 8a
), indicating
that it was not due to the MEK/ERK pathway but occurred independently.
In addition, neither MEK nor MAPK could phosphorylate vimentin in
vitro, further confirming that the rearrangement of vimentin in
the cell was not mediated by MEK or ERK (data not shown).
To verify that the collapse of vimentin filaments in response to
activation of BXB-ER was accompanied by phosphorylation and
solubilization, we labeled BXB-ER cells with
[32P]-orthophosphoric acid and separated the
lysates into detergent-soluble and insoluble fractions (Fig. 8c
). The amount of vimentin that could be immunoprecipitated
from the soluble fraction increased dramatically upon BXB-ER
activation, whereas insoluble vimentin filaments decreased.
Solubilization was connected with phosphorylation as indicated by the
preferential incorporation of 32P label
into soluble vimentin. These results suggest that BXB-ER causes
vimentin rearrangements by induction of vimentin phosphorylation, which
leads to the solubilization of filaments. To further corroborate this
hypothesis we transiently transfected COS-1 cells with BXB or an
activated MEK-1 mutant, MEK-DD (32)
, and fractionated the
cells as above (Fig. 8d
). BXB led to a pronounced
redistribution of vimentin from the insoluble pellet to the soluble
phase, whereas MEK-DD was ineffective. The BXB-mediated vimentin
solubilization was completely resistant to the MEK inhibitor U0126,
although this inhibitor markedly suppressed ERK activation by BXB. In
summary, these experiments demonstrate a strong correlation
in vivo between Raf activity, vimentin phosphorylation, and
vimentin depolymerization, which, however, is independent of MEK and
ERK.
The above experiments pointed to a new effector that links Raf-1 to
vimentin, but of course could not reveal how many steps downstream of
Raf-1 this effector was. The simplest and immediately testable
possibility was that the effector(s) was one or more of the
Raf-1-associated vimentin kinases. Therefore, we assayed whether
phosphorylation by TBST-washed Raf-1 preparations could affect vimentin
filaments in vitro. Polymerized vimentin filaments were
incubated with activated GST-Raf-1 or purified CK2 under Raf kinase
conditions (Fig. 9a
). At different time points, the samples were separated into
soluble and insoluble fractions by centrifugation and the distribution
of vimentin was examined by Western blotting. Quantitative evaluation
of the Western blots by laser scanning densitometry showed that
incubation with activated Raf-1 resulted in the progressive
redistribution of vimentin into the soluble phase. In contrast, CK2
caused no significant redistribution of vimentin, and thus is unlikely
to mediate the effects of Raf-1 to the vimentin scaffold.
|
Vimentin filaments are dynamic structures that are constantly rebuilt
due to the exchange of soluble tetramers with polymerized filaments
(22
, 24)
. The slow kinetics of the reaction further
suggested that the Raf-1 induced phosphorylation of vimentin did not
disrupt vimentin polymers directly, but rather prohibited the
reintegration of phosphorylated soluble vimentin into the filaments. To
test this hypothesis, soluble vimentin oligomers were phosphorylated
with Raf-1 and activated Raf-1 (Fig. 9b
). Immediately after
the kinase reaction, vimentin polymerization was induced by raising the
NaCl concentration from 20 to 150 mM. Polymerized vimentin was
separated from soluble vimentin by centrifugation. The major fraction
of phosphorylated vimentin failed to polymerize and was recovered in
the soluble fraction. In contrast, unactivated Raf-1 induced little
vimentin phosphorylation and did not appreciably inhibit
polymerization. These in vitro results are consistent with
the in vivo effects of Raf on vimentin filaments, since
increasing the rate of disassembly causes the loss of the reticular
structure of the vimentin network and its collapse into tight bundles.
Furthermore, these data suggested that the effects of Raf-1 on vimentin
are mediated by a Raf-1-activated and -associated vimentin kinase other
than CK2. Several kinases have been described that regulate vimentin by
direct phosphorylation. These include PKC (43)
, cdc2
(44)
, and PKA (30
, 43
, 45)
. To examine
whether any of these could be the associated vimentin kinases of Raf-1,
the pattern of phosphorylation sites in vimentin induced by these
kinases were compared with the pattern induced by Raf-1 (data not
shown). The phosphopeptide maps showed clearly that the phosphorylation
sites in vimentin induced by Raf-1 do not match with the
phosphorylation sites in vimentin induced by PKC, PKA, and cdc2.
Thus, these kinases are unlikely candidates for the Raf-1-associated
vimentin kinase(s).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, whereas Raf-1
targeted to the mitochondrial membrane failed to stimulate the MEK/ERK
cascade but prevented apoptosis by inducing the phosphorylation and
inactivation of BAD, a proapoptotic molecule (47)
. Thus,
it appears that Raf-1 is distributed into different pools within the
cell and that the coupling of Raf-1 to substrates may be regulated by
subcellular colocalization.
We found that Raf-1 can induce vimentin phosphorylation in
vitro and in vivo. This activity is due to at least two
different Raf-1-associated kinases rather than to Raf-1 catalytic
activity. We could tentatively identify one of these kinases as CK2.
The removal of CK2 activity from Raf-1 kinase preparations by RIPA
buffer (37)
or inhibition with heparin resulted in a
significant reduction in the phosphorylation of vimentin. As shown by
phosphopeptide mapping, CK2 phosphorylated vimentin only on one peptide
in vitro, which is also phosphorylated in vivo.
Thus, we have identified a novel vimentin kinase that on the basis of
its biochemical behavior most likely corresponds to CK2. Inhibition of
CK2 diminished, but it did not abolish vimentin phosphorylation by
TBST-Raf-1 in vitro. As revealed by phosphopeptide mapping,
TBST-Raf-1 induced vimentin phosphorylation on multiple sites, some of
which were also induced in v-raf-transfected cells. Therefore, we
conclude that Raf-1 associates with at least one other vimentin kinase
besides CK2. We tried to identify the other kinase(s) by comparing the
phosphorylation site pattern induced by TBST-Raf-1 with the sites
phosphorylated by known vimentin kinases. Several kinases have been
reported to phosphorylate vimentin in vitro including cdc2
(44)
, mos (48)
, PKC (43)
, cAMP
dependent kinase, PKA (30
, 43
, 45)
, cGMP dependent kinase
(49)
, and calmodulin-regulated kinase II
(50)
. Since the roles of PKA, PKC, and cdc2 in vimentin
phosphorylation are much better characterized than those of other
kinases, we focused on these three. However, a comparison of the
phosphopeptide maps suggests that we can exclude PKC, PKA, and cdc2
as Raf-1-associated vimentin kinases, especially since we also were
unable to detect coprecipitation or in vitro association
(data not shown).
Vimentin phosphorylation by TBST-Raf-1 was dependent on its activation,
suggesting that Raf-1 can activate one or more of its associated
vimentin kinases. Indeed, we observed that Raf-1 was able to enhance
the activity of recombinant CK2 in vitro. This activation
increased with the amount of Raf-1 added, but was independent of
phosphorylation by Raf-1 (data not shown). An activation of CK2 by
Raf-1 is also suggested by the observation that v-Raf expression
enhanced the phosphorylation of vimentin on a peptide that is
phosphorylated by CK2 in vitro. Since v-raf expression also
enhanced vimentin phosphorylation on other sites, we believe that Raf-1
can also activate vimentin kinases in cells.
These kinases seem to regulate the architecture of vimentin filaments.
The activation of a conditional Raf-1 kinase, BXB-ER, in fibroblasts
induces the rearrangement of the vimentin scaffold. This was not a
consequence of morphological transformation, which ensues hours later
and is not caused by changes in the actin cytoskeleton to which
vimentin is linked, because BXB-ER does not induce actin rearrangement
(25)
. In this study, a bundling of microtubules was
observed in response to BXB-ER activation that slightly resembles the
bundling of vimentin filaments seen here. The effect on the
microtubules, however, was mediated via the MEK/ERK pathway and could
be blocked by MEK inhibition (25)
. In contrast, vimentin
rearrangement in response to BXB-ER activation is not affected by MEK
inhibition and therefore occurs independently of microtubule
rearrangement. Our in vitro experiments indicate that this
effect is due to Raf-1-associated kinases that phosphorylate vimentin
and thereby interfere with polymerization. Due to the complex
phosphorylation pattern, we have not been able to determine which
phosphorylation sites are responsible. CK2 phosphorylation does not
seem to play a role, since it fails to affect vimentin filaments
in vitro.
In the cell, vimentin filaments are not static structures but are
continuously rebuilt due to the incorporation of soluble vimentin
tetramers into the polymers. This exchange seems to be important for
maintaining the structure of the vimentin cytoskeleton. The assembly
and higher order arrangement are altered in response to cell cycle or
differentiation specific cues (22)
. These processes are
regulated at least partly by vimentin phosphorylation (22
, 51
, 52)
. Only the role of the cell cycle-regulated cdc-2 kinase in
the disassembly of vimentin filaments during mitosis is well
established (44
, 53
54
55)
. Mitosis is accompanied by an
increase in vimentin phosphorylation. Depending on the cell type, the
vimentin scaffold undergoes a major rearrangement or complete
disintegration during mitosis (44)
. These differences may
be due to variations in the stoichiometry of vimentin phosphorylation
by cdc-2 or could reflect the cell type-specific participation of other
vimentin kinases (44)
. Consistent with the latter
hypothesis, PKC (56)
and v-mos (57)
have been
reported as mitotic vimentin kinases.
The physiological meaning of vimentin phosphorylation during interphase
is less clear and may also be cell type dependent (22
, 58)
. This uncertainty stems primarily from the incomplete
knowledge about the precise functions of intermediate filaments.
Evidence suggests that intermediate filaments do not just serve "as
mechanical integrators of cellular space" (59)
, but may
also be involved in regulatory processes such as transformation
(60)
, differentiation (61
62
63
64)
, cellular
senescence (65)
, secretion (66)
, and even
control of gene expression (67)
. Given the diverse
functions of intermediate filaments, however, it appears plausible that
their structural organization is adapted to specific cellular
requirements by regulatory mechanisms, which conceivably could involve
different kinases. Under in vitro conditions, vimentin
filaments can be disrupted by PKC or cAMP-dependent kinase (PKA)
-mediated phosphorylation (68
, 69)
of vimentin on
sites other than cdc-2 (44)
. Microinjection of the
catalytic subunit of PKA into rat embryo fibroblasts caused the
disassembly and collapse of vimentin filaments into tight bundles
(30)
, and stimulation of Swiss 3T3 cells with cAMP
agonists resulted in a similar redistribution of vimentin filaments
(70)
. Unlike many other cell lines, including NIH 3T3,
cAMP acts as a growth-promoting factor in Swiss 3T3 cells, suggesting
that vimentin rearrangement may be part of the mitogenic response. This
idea is supported by our observation that in NIH 3T3 cells stimulation
with serum growth factors as well as the selective activation of BXB-ER
by estrogen both trigger vimentin bundling with similar kinetics.
Taken together, our data demonstrate that activation of the Raf-1 kinase induces extensive vimentin reorganization similar to that seen during growth factor stimulation. Whether the changes of the vimentin scaffold are required for the execution of the mitogenic response has to await future studies. The important finding of our study, however, is the identification of a novel branch point in Raf-1 signaling that links Raf-1 with the cytoskeleton via associated vimentin kinases.
|
ACKNOWLEDGMENTS |
|---|
Received for publication October 25, 1999.
Revision received March 23, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
