(Received for publication, January 24, 1997)
From the Department of Cell Biology, University of Alabama at Birmingham, Birmingham, Alabama 35294
During cell-matrix adhesion, both tyrosine and
serine/threonine kinases are activated. Integrin ligation correlates
with tyrosine phosphorylation, whereas the later stages of spreading
and focal adhesion and stress fiber formation in primary fibroblasts
requires interactions of cell surface proteoglycan with heparin-binding moieties. This correlates with protein kinase C (PKC) activation, and
PKC can become localized to focal adhesions in normal, but not
transformed, cells. PKC activation has been thought to be downstream of
initial receptor-ligand interactions. We now show, however, that
syndecan-4 transmembrane heparan sulfate proteoglycan and PKC
co-immunoprecipitate and co-patch in vivo. The core protein of syndecan-4 can directly bind the catalytic domain of PKC
and potentiate its activation by phospholipid mediators. It can also directly activate PKC
in the absence of other mediators. This activity resides in the sequence LGKKPIYKK in the center of the short
cytoplasmic domain, and other syndecans lack this sequence and PKC
regulatory properties. Syndecan-4 is a focal adhesion component, and
this interaction may both localize PKC and amplify its activity at
sites of forming adhesions. This represents the first report of direct
transmembrane signaling through cell surface proteoglycans.
In cell-matrix interactions, as with many cellular responses to
ligands, both tyrosine and serine/threonine kinases participate. During
adhesion to fibronectin, integrin clustering activates tyrosine kinases
(1-3), but we, and others, have noted that full spreading (4) and the
formation of stress fibers and focal adhesions requires additional
signals (3, 5-12). Adhesion to the "cell-binding" 105-kDa fragment
of fibronectin through ligation of integrin
5
1 is sufficient only for attachment and
spreading in anchorage-dependent primary fibroblasts in the
absence of serum and protein synthesis (8-11). To drive the
cytoskeletal and receptor clustering that accompany the formation of
stress fibers and focal adhesions, an additional activation of protein
kinase C (PKC)1 is needed (2, 8, 9). This
can be provided by addition of heparin-binding fibronectin moieties
(8-11), either of the whole heparin-binding domain of fibronectin or
of a peptide PRARI from this domain. These agents appear to signal
through a cell surface heparan sulfate proteoglycan, since treatment
with heparinase prevents the response (11), and mutant cells lacking
(13), or having undersulfated cell surface heparan sulfate
proteoglycans (14), have reduced capacity to form focal adhesions or
stress fibers. Similarly, addition of PRARI peptides to synovial
fibroblasts prespread on substrates of 105-kDa fragment markedly
increased the size of vinculin-positive adhesion plaques (12). The need for addition of heparin-binding agents can be circumvented by treatment
of cells with active, but not inactive, phorbol esters (8). In
addition, there is downstream activation of the RhoA G-protein (15),
which may initiate a contractile response (3).
Syndecan-4, one of the four members of transmembrane matrix-binding
proteoglycans (16-18), becomes inserted into the focal adhesions of a
range of cell types (19) when PKC is activated (20). PKC is also
localized in focal adhesions of normal cells (21) and is presumably
anchored there, but the mechanism of anchorage has not been elucidated.
Although activation of PKC is usually considered a downstream effect
following receptor-ligand interactions, these previous studies led us
to investigate whether syndecan-4 core protein could directly bind PKC
and effect its localization to forming adhesions. We now report the
first example of a mechanism for signaling through a transmembrane
proteoglycan. The cytoplasmic domain of syndecan-4 core protein can
directly regulate both the localization and activity of PKC through
interactions of a sequence unique to syndecan-4 with the catalytic
domain of PKC.
For co-immunoprecipitation experiments,
subconfluent rat embryo fibroblasts (REF) were scraped
(107 cells/immunoprecipitate) into 10 ml of lysis buffer
(1% Triton X-100 in 50 mM Tris-HCl, pH 7.5, containing 50 mM NaCl, 0.2 mM phenylmethylsulfonyl fluoride,
1 mM benzamidine HCl, 1 mM leupeptin, 750 µM CaCl2), precleared by centrifugation, and
incubated for 30 min at 4 °C with protein A-Sepharose beads
(Pharmacia Biotech Inc.) pre-blocked (30 min) with 10% fetal calf
serum in phosphate-buffered saline. After centrifugation, supernatants
were sequentially incubated (30 min each) with 20 µg/ml monoclonal
anti-syndecan-4 150.9 IgG raised against its NH2 terminus,
20 µg/ml rabbit anti-mouse IgG (Dako Corp. Carpinteria, CA) and fresh
preblocked protein A-Sepharose beads. Beads were centrifuged, washed
eight times with lysis buffer, boiled for 3 min, and released material
immunoblotted with pan-specific anti-PKC recognizing the COOH terminus
of the catalytic domains of PKC
, PKC
I, PKC
II, PKC
, and
PKC
(Upstate Biotechnology, Inc., Lake Placid, NY) following 3-15%
SDS-PAGE. Bound antibody was detected by goat anti-rabbit alkaline
phosphatase (Bio-Rad). Some cultures were pretreated before lysis with
phorbol 12-myristate 13-acetate (PMA: 200 nM, 15 min); in
others 4V or scrambled peptide (50 µg/ml) was added at lysis and was
present throughout. To demonstrate co-patching in vivo, REF
were plated for 20 min onto coverslips coated with fibronectin (19),
treated sequentially with 150.9 IgG and goat anti-mouse fluorescein
isothiocyanate (Organon Teknika Corp., Durham, NC), warmed for 20 min
in the presence or absence of 200 nM PMA (22), and stained
with anti-PKC followed by goat anti-rabbit conjugated to Texas Red
(Organon Teknika) after fixation and permeabilization (19). No
breakthrough or inappropriate cross-reactivity of antibodies was
detected in controls. For affinity chromatography on Affi-Gel-102 to
which peptides were coupled through NH2-terminal cysteine
by sulfo-LC-SPDP (Pierce), REF (
5 × 106
cells/sample) were scraped into 6 ml of lysis buffer lacking calcium,
centrifuged (100,000 × g, 1 h), and
CaCl2 (750 µM) and PMA (200 nM)
were added to the supernatant as indicated. Columns were washed with 50 mM NaCl and eluted with 1 mM EDTA, or 1 M NaCl, released material being concentrated with 10%
trichloroacetic acid, and immunoblotted with anti-PKC following 7.5%
SDS-PAGE. Bound antibody was detected by chemiluminescence (ECL;
Amersham Corp.). In solid phase binding assays (Pierce), enzyme-linked immunosorbent assay plate wells were coated with peptides (1 µg/well in NaHCO3, pH 9.6, 4 °C, overnight), washed, and
unoccupied sites blocked with 5% milk (1 h). Second, PKC
(10 ng)
was allowed to bind (37 °C, 30 min) in the presence or absence (for
binding conditions, see legend to Fig. 1D) of phospholipid
(50 µg/ml PS, 5 µg/ml DL) and/or 1 mM
CaCl2. Third, bound enzyme was activated with 0.2 mg/ml PS,
0.02 mg/ml DL, 750 µM CaCl2 to phosphorylate
added dye-labeled serine-substituted PKC pseudosubstrate (16 ng/well,
37 °C for 30 min). Phosphorylation was monitored by absorbance at
570 nm following affinity separation. Absorbance is normally
proportional to the amount of PKC
bound.
Fusion Protein Production
cDNA for full-length or truncated core proteins of syndecan-4 in glutathione S-transferase (GST) expression vector pGEX-5X-1 (Pharmacia) were expressed as fusion proteins in Escherichia coli, solubilized with 1% N-laurylsarcosine in 10 mM Tris, pH 7.5, 150 mM NaCl, 2 mM EDTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine HCl, purified on glutathione-agarose (Sigma), washed to exchange bound detergent to 0.1% Triton X-100, eluted with 50 mM Tris-HCl, pH 8.0, 5 mM reduced glutathione (Janssen Chimica, New Brunswick, NJ), subjected to 6% SDS-PAGE, and labeled by Coomassie Blue or immunoblotted with anti-GST (Molecular Probes, Eugene, OR).
Phosphorylation AssaysSimilar results were seen with
PKC (1.2 ng) from rat brain (Upstate Biotechnology, Inc.) or 1 ng of recombinant PKC
(Molecular Probes and Upstate Biotechnology,
Inc.). These were incubated (10 min, room temperature) in 20 µl of 50 mM HEPES, pH 7.3, containing 3 mM magnesium
acetate, 750 µM CaCl2, 200 µM
ATP (0.5 µCi of [
-32P]ATP), 3 µg of histone III-S
(Sigma) with preformed phospholipid (PL: 4 µg of PS
and 0.4 µg of DL, briefly sonicated) micelles as indicated.
Dose-response assays (not shown) determined that these conditions gave
maximum activation of PKC
. Fusion proteins (1 µg) or synthetic
peptides were added to preformed micelles, except for 4L, which was
only effective when added during their formation. In phosphorylation
assays with protein kinase M (PKM: 1.3 ng; PKC catalytic subunit from
rat brain, Calbiochem), calcium was omitted, but PL was present to
control for any contamination with whole PKC, which would be activated
by PL. The commercial PKM preparation did not contain PKC, since no
increased activity was seen on addition of PL (Fig. 3B,
compare lanes 1 and 2). Other PKM assays also,
therefore, contained PL for comparative purposes. Reactions were
stopped by addition of trichloroacetic acid (to 20%) and 0.5% SDS
(fusion protein assays) and the precipitates analyzed or by direct
analysis following addition of 5 × SDS-PAGE mixture (peptide
assays). Products were subjected to 20% SDS-PAGE and autoradiography
or PhosphorImager analysis (n = 2-10, values shown are
mean ± S.E., means compared by Student's t test). The presence of NH2-terminal cysteine to allow conjugation to
affinity matrices had no effect except to enhance direct activation.
All assays terminated when activity was maximal and linear.
The cytoplasmic domains of the syndecan family members are
highly homologous except for one central region (16-18). Since
syndecan-4 is the only syndecan present in focal adhesions of
fibroblasts, we tested in several ways whether this unique region (here
termed 4V) mediated interactions with PKC. First, PKC was
co-immunoprecipitated from REF cell lysate with syndecan-4 proteoglycan
in the presence, but not the absence, of phorbol ester (PMA) to
activate PKC (23). This was prevented by inclusion of the cytoplasmic
domain peptide 4V as a competitor, but not by a peptide of scrambled
sequence (Fig. 1A). Second, when syndecan-4
was patched on the surface of spreading cells, PKC was co-patched (Fig.
1B). This co-patching of initially diffuse syndecan-4 and
PKC was accelerated by PMA (not shown). Third, when cell lysates were
passed over an affinity matrix of syndecan 4V peptide (but not the
equivalent 2V peptide), two components corresponding in molecular mass
to PKC and PKM (the catalytic domain of PKC) were specifically
eluted and detected by immunoblotting with antibodies against PKC
catalytic domain (Fig. 1C). Again, binding of PKC or PKM was
not detected in the absence of calcium and PMA to activate PKC (23)
(compare lanes 3 and 4). PMA was the required
agent, however, since elution with EDTA (lane 2) had no
effect. Fourth, interaction was also seen in solid phase assays (Fig.
1D) where substrates of 4V specifically and directly bound
PKC
. This assay requires coating substrates with the peptide of
interest, adding PKC
under differing binding conditions, washing,
and then activating bound PKC
by addition of PL. Minimal binding
occurred to wells coated only with milk (M: negative
control; nonspecific binding). Maximal binding occurred to wells
blocked with milk after addition of PKC
(P: positive control), but phosphorylation was even more evident in wells coated with 4V peptide, since this reflects not just binding, but also potentiation of activity. Binding of PKC
was again optimal where the
enzyme had been activated with calcium and PL during binding to
4V-coated substrates (compare last three lanes in Fig.
1D).
Since syndecan-4 appeared to
bind PKC, both in vivo and in vitro, we monitored
whether this could affect enzyme activity. Fig.
2A shows a diagram of three GST-fusion
proteins containing either full-length syndecan-4 core protein (4W) or
with truncated cytoplasmic domains (4I, 4R). Fig. 2B shows
these purified fusion proteins by Coomassie Blue staining (lanes
1-3) and immunoblotting with antibodies against GST (lanes
4-6). All the fusion proteins migrated with
Mr much greater than the predicted molecular
mass (44-47 kDa), indicative of the formation of SDS-resistant
oligomers as seen in all syndecans and confirming that the cytoplasmic
domain is not essential for self-association (24). These fusion
proteins were used in in vitro assays to test for regulation
of activity of a mixture of PKC. When phosphorylation by PKC
of histone III-S was already maximal due to the presence of PL and
calcium, full-length syndecan-4 fusion protein potentiated the activity of a mixture of PKC
(lane 1, Fig. 2C) a
further 3-fold. This was not seen in the presence of 4I or 4R (Fig.
2C). Similar potentiation was seen with recombinant PKC
(not shown). None of the fusion proteins served as a PKC substrate.
Syndecan-4 appeared to act through the catalytic domain of PKC,
since 4W fusion protein, but not truncated fusion protein, could also
potentiate the activity of PKM (Fig. 3A),
which lacks the regulatory PL- and calcium-binding domains, and is,
therefore, constitutively active. It also bound both PKC and PKM (Fig.
1C). From binding studies and the different effects of the
truncated fusion proteins, it appeared that the active region in
syndecan-4 cytoplasmic domain was the V region. Similar assays were
performed, therefore, with synthetic peptides corresponding to the
whole cytoplasmic domain (4L) and V (4V, 2V) regions of syndecan-4 or its closest homolog, syndecan-2 (Fig. 3B). PKM is
constitutively active in the absence of PL, since it lacks the
regulatory domain. An absence of intact PKC in the PKM preparation was
confirmed by a comparison of histone phosphorylation in the absence or
presence of PL (compare lanes 1 and 2). Peptide
4L (lane 3) and 4V (lane 4) potentiated
phosphorylation of histones by PKM with an average increase of
3.5-4-fold. Neither a peptide of scrambled 4V sequence (Scr;
lane 5) nor the corresponding sequence from syndecan-2 (2V; lane 6) showed any enhancement of histone phosphorylation. The equivalent sequence from syndecan-1 was also ineffective (not shown).
Whole PKC
was inactive in the absence of PL (Fig. 3C, lane
1), but activated by PL (lane 2). This PL-mediated
PKC
activity was potentiated by the addition of peptide 4V (Fig.
3C, lane 3). Quantification indicated an average increase to
302 ± 41% in the presence of 4V over normally maximal
PL-mediated activity (Fig. 3C, lane 2), and 40-fold increase
over base-line activity (i.e. in the absence of PL; Fig.
3C, lane 1). This was dose-dependent, and
clearly detectable at 0.45 µM peptide (not shown). As a
further indication of the independent regulatory properties of
syndecan-4 V region, histone phosphorylation by PKC
was detected,
even in the absence of PL (Fig. 3C, lane 6) when 4V was
added. Direct activation was also dose-dependent with
maximal activation at 0.1 mM (not shown), but never reached
the levels seen in the presence of PL; the average increase was 3-fold
over base line. No activity was again seen with the scrambled peptide
or 2V (Fig. 3C, lanes 7 and 8). Similar
potentiation of PL-mediated activation, and direct activation of
PKC
, was seen in assays using a physiological substrate, an
epidermal growth factor receptor peptide (25) (not shown). Therefore,
the effects of 4V on potentiation of PL-mediated PKC activation were
not merely due to facilitation of histone interaction with PL vesicles.
Furthermore, direct activation of PKC by 4V peptide can occur in the
absence of PL (Fig. 3C, compare lanes 1 and
6).
The 4V region is highly cationic, and protamine has been shown
previously to activate PKC in a PL-independent manner (26). However,
further experiments with 4V peptides of altered or scrambled sequence
confirmed the specificity for PKC regulation by syndecan-4 (Fig.
4, A and B). Even conservative
substitutions markedly diminished the ability to potentiate PL-mediated
activation or to directly activate PKC
in the absence of PL.
Protein kinase C binding to syndecan-4 was seen using intact core
protein and cytoplasmic peptides in vitro and intact
glycosylated syndecan-4 in vivo. A specific region of the
cytoplasmic domain of syndecan-4 was mapped as a site that interacts
with and regulates the activity of PKC and PKM. This region of
syndecan-4 is the first example of a protein sequence from a cell
surface receptor that can directly regulate PKC localization and
activity. Syndecan-4 probably interacts through the catalytic domain of
PKC, since it elevated the activity of this isolated domain (PKM), and
PKC
activity was additive in the presence of both 4V and PL. Binding of PKC
to the V region of syndecan-4 in vitro was only
clearly detectable after enzyme activation by PMA, whereas co-patching occurred in vivo without additional treatment. Treatment of
cells with PMA did, however, accelerate the patching process,
indicating that binding of mediators to the regulatory domain may
enhance the availability of the catalytic domain of PKC for binding to syndecan-4. PKC is active early during cell spreading (4), and its
activation results in syndecan-4 translocation to stabilizing sites of
cell-matrix adhesion (20). Therefore, it is possible that, during cell
adhesion, conventional activation of PKC
facilitates its interaction
with ligated and clustered syndecan-4 and the colocalization of this
complex to focal adhesions. Syndecan-4 may then, at these sites,
compartmentalize PKC
and regulate its phosphorylation of other
cytoplasmic components.
Consistent with a role for syndecan-4 core protein regulating PKC activity during focal adhesion and stress fiber formation (8-11, 19), we have found2 that the ability of Chinese hamster ovary cells to form these structures is increased after stable transfection with cDNA for syndecan-4 core protein. In contrast, cells transfected with a cDNA encoding a truncated syndecan-4 core protein that terminates in the V region have greatly reduced focal adhesions and stress fibers. Further studies to investigate which cytoskeletal or other cytoplasmic proteins are phosphorylated at adhesions are underway.
We thank the Comprehensive Cancer Center (University of Alabama at Birmingham) for peptide synthesis and analysis, Drs. G. J. Cowling and J. T. Gallagher and A. Fleetwood (University of Manchester, Manchester, United Kingdom) for the syndecan-4 cDNA constructs.