Báo cáo khoa học: Calcium-induced activation and truncation of promatrix metalloproteinase-9 linked to the core protein of chondroitin sulfate proteoglycans pot

Calcium-induced activation and truncation of promatrix
metalloproteinase-9 linked to the core protein
of chondroitin sulfate proteoglycans
Jan-Olof Winberg
1
, Eli Berg
1
, Svein O. Kolset
2
and Lars Uhlin-Hansen
1
1
Department of Biochemistry, Institute of Medical Biology, University of Tromsø, Norway;
2
Institute of Nutrition Research,
University of Oslo, Norway
In the leukemic macrophage cell-line THP-1, a fraction of
the secreted matrix metalloproteinase 9 (MMP-9) is linked
to the core protein of chondroitin sulfate proteoglycans
(CSPG). Unlike the monomeric and homodimeric forms
of MMP-9, the addition of exogenous CaCl
2
to the
proMMP-9/CSPG complex resulted in an active gelatinase
due to the induction of an autocatalytic removal of the
N-terminal prodomain. In addition, the MMP-9 was
released from the CSPG through a process that appeared
to be a stepwise truncation of both the CSPG core protein
and a part of the C-terminal domain of the gelatinase. The
calcium-induced activation and truncation of the MMP-9/
CSPG complex was independent of the concentration of

the complex, inhibited by the MMP inhibitors EDTA,
1,10-phenanthroline and TIMP-1, but not by general
inhibitors of serine, thiol and acid proteinases. This indi-
cated that the activation and truncation process was not
due to a bimolecular reaction, but more likely an intra-
molecular reaction. The negatively charged chondroitin
sulfate chains in the proteoglycan were not involved in this
process. Other metal-containing compounds like amino-
phenylmercuric acetate (APMA), NaCl, ZnCl
2
and MgCl
2
were not able to induce activation and truncation of the
proMMP-9 in this heterodimer. On the contrary, APMA
inhibited the calcium-induced process, whereas high con-
centrations of either MgCl
2
or NaCl had no effect. Our
results indicate that the interaction between the MMP-9
and the core protein of the CSPG was the causal factor in
the calcium-induced activation and truncation of the gel-
atinase, and that this process was not due to a general
electrostatic effect.
Keywords: gelatinase B; MMP-9; proteoglycan; activation;
calcium.
The superfamily of matrixins or matrix metalloproteinases
consists of at least 18 different mammalian zinc- and calcium-
dependent metalloproteinases (MMPs) [1–4], of which
monocytes/macrophages can express several types [5–9].
Together, the MMPs are able to degrade most extracellular

matrix proteins [3,4,10], as well as regulating the activity of
serine proteinases by digesting various serpins (serine
proteinase inhibitors) [11], and the growth factor activity of
insulin-like growth factor (IGF) by the ability to degrade
IGF binding protein (IGHBP) [12]. Thus MMPs have broad
substrate specificity, and have been shown to be involved in
various regulatory processes in normal and pathological
conditions in different tissues and organs.
The activity of MMPs is regulated at the transcriptional,
translational and post-translational levels. Most of the
MMPs are synthesized in their latent pro-form, and must
be converted to their active forms in the extracellular
space. The cysteine in the conserved PRCG(V/N)PD
sequence in the pro-domain binds to the active site zinc
as a fourth ligand, and hence is involved in the mainten-
ance of the latency of the enzymes [3,13]. During the
activation, either parts of or the entire N-terminal pro-
domain are removed. This process can be performed by
various agents in vitro, including p-aminophenylmercuric
acetate (APMA), SDS, urea, chaotropic agents, heat
treatment and by proteinases [3,10,13–17]. A model for
the latency and activation of MMPs has been proposed,
called the Ôcysteine-switchÕ or ÔvelcroÕ model, which suggests
that there is an equilibrium between a Ôswitch-openÕ and a
Ôswitch-closedÕ form of the pro-enzyme [16]. The reaction
of the free thiol group in the switch-open form with for
example organomercurials has been suggested to drive the
equilibrium toward the open form, which then undergoes
an autolytic conversion to an active form. Once activated,
the activity of MMPs can be regulated by endogenous

inhibitors such as a2-macroglobulin and tissue inhibitors
of MMPs (TIMPS) [3,4,10,13,18].
MMP-9 has been found as a monomer as well as in
various dimeric forms [19–23]. In the homo- and hetero-
dimeric forms, the proteins are either covalently linked to
each other through disulfide bonds [20,22,23] or through
a strong noncovalent and predominantly hydrophobic
Correspondence to J O. Winberg, Department of Biochemistry,
Institute of Medical Biology, University of Tromsø, 9037 Tromsø,
Norway. Fax: + 47 77 646222, Tel.: + 47 77 645488,
E-mail:
Abbreviations: APMA, amino-phenylmercuric acetate; cABC, chon-
droitin ABC lyase; CS, chondroitin sulfate; GAG, glycosaminoglycan;
MMP, matrix metalloproteinase; PG, proteoglycan; TIMP, tissue
inhibitors of MMP; SG, serglycin; SBTI, soybean trypsin inhibitor.
(Received 1 July 2003, revised 4 August 2003, accepted 8 August 2003)
Eur. J. Biochem. 270, 3996–4007 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03788.x
dimerization contact [24]. Recently, we discovered that
the leukemic monocyte cell line THP-1 produced a new
type of reduction-sensitive heterodimer, in which MMP-9
is strongly linked to the core protein of a chondroitin
sulfate proteoglycan (CSPG) [23]. Proteoglycans (PGs)
constitute a distinct class of glycoconjugates, character-
ized by a core protein substituted with highly negatively
charged glycosaminoglycan (GAG) chains of which
chondroitin sulfate (CS) is a major type [25–27]. The
expression of various types of PG in THP-1 cells are not
extensively studied, but it has previously been reported
that serglycin is the major CSPG released from these
cells [27]. As most cells produce several types of PG, one

can expect that THP-1 cells also express various PG core
proteins such as versican and syndecans. At present, it is
not known which of these PG core proteins are bound
covalently to MMP-9. The discovery of complexes of
MMP-9 covalently linked to the core protein of CSPG
expanded our view of how MMP-9 can interact with
PGs, as it was known previously that both MMP-9 and
MMP-2 binds to the negatively charged GAG chains in
PGs through positively charged clusters in the C-terminal
hemopexin-like domains of the MMPs [28,29]. Because
calcium is known to maintain the structural integrity of
MMPs including MMP-9 [30,31] and to be of import-
ance for the activity in APMA activated MMP-9 [32,33],
we investigated if either APMA, APMA in combination
with calcium, or calcium alone could induce an in vitro
activation of the proMMP-9/CSPG heterodimer and
eventually influence on the enzymatic activity of the
complex. The results presented show that proMMP-9
bound covalently to the CSPG has different character-
istics compared to other MMP-9 forms.
Experimental procedures
Materials
Safranin O (number S-2255), cetylpyridinuim chloride,
phorbol 12-myristate 13-acetate, Triton X-100, chondroitin
sulfate C, trypsin, soybean trypsin inhibitor (SBTI),
EDTA, 1,10-phenanthroline, gelatin, calf skin collagen,
pepstatin, leupeptin, N-ethylmaleimide and alkaline phos-
phatase-conjugated antibody were purchased from Sigma.
Pefabloc was from Pentapharm Ltd. Human recombinant
TIMP-1, Q-Sepharose, Sephadex 200, Sephadex G-50

(fine), Gelatin-Sepharose, Heparin-Sepharose, Amplify,
14
C-labeled Rainbow
TM
protein molecular mass standards,
[
35
S]sulfate and mouse monoclonal antibodies against
human MMP-9 (#IM10L) were obtained from Amersham
Pharmacia Biotech. According to the manufacturer, the
MMP-9 antibody (#IM10L) detects only the latent
92 kDa form under both reducing and nonreducing
conditions. Polyclonal antibodies against TIMP-1,
MMP-7 and the C-terminal region and the hinge region
of MMP-9 were obtained from Chemicon International,
Inc. Chondroitin ABC lyase (proteinase free) was pur-
chased from Seikagaku Kogyo Co. CDP-Star
TM
chemi-
luminescent substrate was obtained from New England
Biolabs. Unlabeled molecular mass standards were from
Bio-Rad. Cronex 4 medical X-ray film was obtained from
Sterling Diagnostic Imaging.
Cells
The human leukemic macrophage cell-line THP-1 was a
kind gift from K. Nilsson, Department of Pathology,
Uppsala University, Sweden. The cells were cultured in
RPMI 1640 medium with 10% fetal bovine serum,
100 lgÆmL
)1

of streptomycin, and 100 UÆmL
)1
of penicillin.
To isolate cell-synthesized MMP-9/CSPG complex, the cells
were washed three times in serum-free medium and then
cultured for 72 h in serum-free RPMI 1640 medium with
0.1 l
M
phorbol 12-myristate 13-acetate, and the conditioned
medium was thereafter harvested and treated as described
earlier [23]. This medium was then used directly for analyses
or purification of CSPG (see below).
Purification of proMMP-9/CSPG complexes
The proMMP-9/CSPG complex was purified by Q-Seph-
arose anion-exchange chromatography as described previ-
ously [23]. Briefly, the column (bed volume 2 mL) was
washed with 50 mL of 0.05
M
sodium acetate, pH 6.0,
containing 6
M
urea and 0.35
M
NaCl. During these
conditions, both the 92 and 225 kDa forms of the
MMP-9 passed through the column. Bound material was
eluted with 1.5
M
NaCl in 0.05
M

sodium acetate, 6
M
urea,
pH 6.0. The CSPG-containing fractions, detected by the
Safranin O method (see below), were pooled and diluted
with 0.05
M
sodium acetate/6
M
urea to give a final NaCl
concentration of 0.35
M
. The material was then re-subjected
to another column of Q-Sepharose. After extensive wash
with the buffer containing 0.05
M
sodium acetate, 6
M
urea
and 0.35
M
NaCl, bound material was eluted with a
gradient of 0.35–1.5
M
NaCl in 0.05
M
sodium acetate,
6
M
urea, pH 6.0. The column was run at a flow rate of

0.5 mLÆmin
)1
and fractions of 1 mL were collected. The
fractions containing most CSPGs, as determined by Safr-
anin O, were pooled and desalted on Sephadex G-50 (fine)
columns run in H
2
O. The volume was reduced in a Speed
Vac (Savant).
In other experiments, the Q-Sepharose purified
proMMP-9/CSPG complex was further purified by first
incubating the complex in the presence of 10 m
M
of EDTA
for 30 min at 4 °C. The EDTA and EDTA extracted
material was then separated from the intact proMMP-9/
CSPG complex on gel permeation chromatography. Two
hundred microlitres of this material was added to an
800 · 0.4 cm Sephacryl 200 column, and fractions of 50 lL
were collected. The intact proMMP-9/CSPG complex was
eluted in the void-volume.
Purification of MMP-9 from the THP-1 cells
The proMMP-9 in conditioned medium from the THP-1
cells was partly purified by subjecting the culture medium to
a Gelatin-Sepharose column. Both the MMP-9 monomer
and dimer forms bound to the column, while the MMP-9/
CSPG complex was detected in the pass-through fractions.
Prior to elution of the bound proMMP-9 from the column
with 10 m
M

of dimethylsulfoxide, the column was thor-
oughly washed with 0.1
M
Hepes buffer, pH 7.5. The eluted
and pooled MMP-9 fractions were passed over a Sephadex
G-50 (fine) column, run in 0.1
M
Hepes, pH 7.5.
Ó FEBS 2003 Activation of the proMMP-9/CSPG complex (Eur. J. Biochem. 270) 3997
Degradation of PG-bound CS-chains by chondroitin
ABC lyase (cABC) treatment
The PG-bound CS-chains were degraded by digestion for
2hat37°C with 0.2–1.0 units of cABCÆmL
)1
of 0.05
M
Tris/HCl, pH 8.0, containing 0.05
M
sodium acetate. In
some experiments, the degraded CS-chains were removed
from the proMMP-9/PG core protein complex by gel
chromatography on a Sephadex G-50 column. In other
experiments, the resulting proMMP-9/core protein complex
was separated from remaining intact complex or other
impurities on either a new Q-Sepharose column pre-
equilibrated with 0.35
M
NaCl, or on a Gelatin-Sepharose
column, or alternatively a combination of these two
columns. During the conditions used, the proMMP-9/PG-

core protein complex did not bind to the Q-Sepharose
column. The proMMP-9/core protein complex was bound
to the Gelatin-Sepharose column, and was eluted from the
column using 10% dimethylsulfoxide, while the intact
proMMP-9/CSPG complex passed through this column.
Detection of PG-bound CS-chains
PG-bound CS-chains were quantitated spectrophotometri-
cally by the Safranin O method [34] as described previously
[23]. Briefly, 30 lL was mixed with 300 lLof50m
M
sodium acetate, pH 4.75, containing 0.02% Safranin O.
The mixture was subjected to a microsample filtration
manifold using the slot-blotting process. After filtration
through nitrocellulose filter (Millipore HA 0.45 lm), each
sample was washed twice with 100 lLH
2
O by filling the
wells and reapplying the vacuum. The nitrocellulose filter
was removed and the individual dots were cut out and
transferred to tubes containing 200 lL of 10% cetylpyridi-
nium chloride in H
2
O. The precipitates were solubilized by
incubation at 37 °C for 30 min. Vortexing was performed
every 10 min during the incubation. The absorbance of the
solubilized color was measured in a Pharmacia Ultrospec III
spectrophotometer at 536 nm. The amount of GAGs in
each sample was estimated from a standard curve of
4–40 lgÆmL
)1

of chondroitin sulfate C.
Gelatin zymography
SDS/substrate PAGE was carried out as described previ-
ously [23] with gels (7.5 · 8.5 cm · 0.75 mm) containing
0.1% (w/v) gelatin in both the stacking and the separating
gel, 4 and 7.5% (w/v) of polyacrylamide, respectively. The
gelatin zymograms were calibrated with both human
gelatinase standards from capillary whole blood as des-
cribed previously [35], protein standards and the condi-
tioned serum-free THP-1 medium. Ten microlitres of
conditioned medium or purified CSPG was mixed with
3 lL of loading buffer (333 m
M
Tris/HCl, pH 6.8, 11%
SDS, 0.03% bromophenol blue and 50% glycerol). Eight
microlitres of this nonheated mixture was applied to each
gel, which was then run at 20 mA at 4 °C. Thereafter, the gel
was washed twice in 100 mL of washing buffer (50 m
M
Tris/
HCl, pH 7.5, 5 m
M
CaCl
2
,1l
M
ZnCl
2
and 2.5% (v/v)
Triton X-100) and then incubated in 100 mL of assay buffer

(50 m
M
Tris/HCl, pH 7.5, 5 m
M
CaCl
2
,1l
M
ZnCl
2
and
1m
M
APMA) for approximately 20 h at 37 °C. Gels were
stained with 0.2% Coomassie Brilliant Blue R-250 (30%
methanol) and destained in a solution containing 30%
methanol and 10% acetic acid. Gelatinase activity was
evident as cleared (unstained) regions. The area of the
cleared zones was analyzed with the GelBase/GelBlot
TM
Pro
computer program from Ultra Violet Products.
Western immunoblotting analysis
Purified CSPG was electrophoresed on SDS/PAGE 4%
(w/v) in stacking gel and 7.5% (w/v) in separating (gel) and
electroblotted to a poly(vinylidene difluoride) membrane.
After blockage of nonspecific binding sites with non fat milk
(5% in Tris-buffered saline), blots were incubated for 1 h at
room temperature with the appropriate antibody against
human MMP-9. After washing, the blots were incubated for

1 h at room temperature with an alkaline phosphatase-
conjugated antibody, diluted 1 : 20 000 in blockage solu-
tion and developed with CDP-Star
TM
chemiluminescent
substrate. All procedures were performed according to the
manufacturer. The area and intensity of the stained bands
was also analyzed with the
GELBASE
/
GELBLOT
TM
PRO
computer program from Ultra Violet Products.
Activation of latent gelatinases
The gelatin-degrading enzymes are secreted from THP-1
cells into the culture medium in a latent form and require
proteolytic activation. The trypsin-titration of the latent
enzymes in the THP-1 conditioned medium was mainly
achieved as described previously [36], except that the
activation time in the present work was between 15 and
30 min at 37 °C.
The gelatinases associated with CSPG were activated by
incubating the proMMP-9/CSPG complex at 37 °Cwith
0.1–100 m
M
of CaCl
2
for 2 h. In some experiments, the
complex was incubated with either 1 m

M
APMA, 10 m
M
MgCl
2
, 0.001–10 m
M
ZnCl
2
or 10–200 m
M
NaCl for 2 h at
37 °C. To test for a possible prevention of the activation
process, either 10 m
M
EDTA, 1 m
M
1,10-phenanthroline,
1–20 n
M
human recombinant TIMP-1, 0.1–4.5
M
urea,
1m
M
pefabloc, 2 lgÆmL
)1
leupeptin, 1 lgÆmL
)1
pepstatin

or 1 m
M
N-ethylmaleimide was added to the proMMP-9/
CSPG complex prior to incubation with 10 m
M
of CaCl
2
for 2 h at 37 °C.
3
H-labeling of calf skin collagen
Acid-soluble calf skin collagen was labeled with tritium by
reductive methylation of the amino groups as described
previously [37]. Collagen denatured for 5–10 min at 90 °C
resulted in gelatin.
Gelatinolytic proteinase activity
Briefly, 50 lL of activated or non activated cell-conditioned
medium or, activated or non activated purified MMP-9/
CSPG (5 lg), was mixed with 50 lLof0.1
M
Hepes buffer,
pH 7.5 and 50 lLofthe
3
H-labeled gelatin solution
(2.3 mgÆmL
)1
or 10
7
c.p.m.Æmg
)1
). In inhibition experi-

ments, either 10 m
M
EDTA, 1 m
M
1,10-phenanthroline or
3–20 n
M
of human recombinant TIMP-1 was added to
3998 J O. Winberg et al. (Eur. J. Biochem. 270) Ó FEBS 2003
CaCl
2
activated MMP-9/CSPG complex. The gelatinase
assays were carried out at 37 °C for approximately 20 h.
Twenty microlitres of the supernatants were subjected to
SDS/PAGE. Thereafter, the gel was soaked in Amplify and
dried. Nondegraded and degraded [
3
H]gelatin were detected
with autoradiography.
Results
Calcium-induced activation of the MMP-9/CSPG
complex
Experiments were performed to determine whether purified
MMP-9/CSPG complex was active and able to degrade
gelatin.
3
H-labeled gelatin was incubated with the MMP-9/
CSPG complex at 37 °C in the absence or presence of
exogenously added CaCl
2

.After24hthesamplewas
appliedtoSDS/PAGEandthegelwasanalyzedby
autoradiography. No fragmentation of gelatin could be
detected in the absence of CaCl
2
, while in the presence of
10 m
M
CaCl
2
, the gelatin was degraded to smaller frag-
ments (Fig. 1A, lane 3 and Fig. 1B, lane 4). When APMA-
treated MMP-9/CSPG complex was incubated in the
absence or presence of 10 m
M
CaCl
2
, no degradation of
gelatin was obtained (Fig. 1B, lanes 3 and 5).
The CS chains were not involved in the activation of
the complex, as the same results were obtained when the
CS-chains were enzymatically degraded by cABC lyase and
removed from the complex by Sephadex G-50 gel chroma-
tography prior to incubation with gelatin.
ThegelatinaseactivityoftheCaCl
2
activated MMP-9/
CSPG complex was totally inhibited in the presence of
either 10 m
M

of EDTA (Fig. 2A) or 1 m
M
of 1,10-
phenanthroline (data not shown). Also human recombinant
TIMP-1 (3–20 n
M
) inhibited the gelatinase activity in a
concentration-dependent manner (Fig. 2B).
The addition of calcium released low molecular size
forms of MMP-9 from the MMP-9/CSPG complex
As shown earlier [23], gelatin zymography of the MMP-9/
CSPG complex reveal bands in the stacking gel and a band
around 300 kDa (Fig. 3A, lane 1). When the MMP-9/
CSPG complex was incubated with 10 m
M
CaCl
2
for 2 h
at 37 °C prior to electrophoresis, both these bands either
disappeared or were strongly reduced, and new bands with
lower M
r
appeared (Fig. 3A, lane 2). Two weak bands at 80
and 85 kDa were seen along with a strong doublet at
74/76 kDa. The same pattern occurred when the MMP-9/
CSPG complex was incubated for 2 h at 37 °C with varying
CaCl
2
concentrations from 0.1 to 100 m
M

(data not shown).
However, the process was concentration dependent. The
Fig. 1. Activation of proMMP-9/CSPG with CaCl
2
. Purified CSPG
was incubated for 2 h at 37 °C with (+) or without (–) CaCl
2
(10 m
M
),
cABC or APMA (1 m
M
) as indicated under each lane. Five
micrograms of the CSPG was then mixed with [
3
H]gelatin and incu-
bated for 24 h at 37 °C as described in the method section. These
samples (20 lL per lane) were then separated on a 7.5% SDS/PAGE
gels, and the radioactivity of the labeled gelatin and its degradation
products were detected by autoradiography. Lane 1 in (A) and (B)
shows a [
3
H]gelatin control, and both gels contained [
3
H]gelatin
incubated with either trypsin-activated THP-1 conditioned serum-free
medium or trypsin as a positive control (not shown). At the left in each
figure is shown the position of the rainbow standard markers and their
M
r

in kDa. The arrowheads indicate the bottom of the application
well.
Fig. 2. Inhibition of the CaCl
2
-activated MMP-9/CSPG complex.
(A) Five micrograms of purified MMP-9/CSPG was incubated for 2 h
at 37 °C in the presence of 10 m
M
of CaCl
2
and then mixed with
[
3
H]gelatin, either with (+) or without (–) EDTA (10 m
M
) or human
recombinant TIMP-1 as indicated under each figure. These mixtures
were thereafter incubated for 24 h at 37 °C as described in the
Experimental procedures section. (B) The amount of TIMP-1 used was
3.3 n
M
(lane 5), 6.7 n
M
(lane 6) and 20 n
M
(lane 7). In (A) and (B),
negative controls are shown in lane 1 ([
3
H]gelatin) and in lane 3
([

3
H]gelatin incubated with 5 lg of unactivated MMP-9/CSPG). Lane
2 shows a positive control of [
3
H]gelatin degraded by trypsin-activated
THP-1 conditioned serum-free medium. At the left in each figure is
shown the position of the rainbow standard markers and their M
r
in
kDa. The arrowheads indicate the bottom of the application well.
Ó FEBS 2003 Activation of the proMMP-9/CSPG complex (Eur. J. Biochem. 270) 3999
low M
r
bands were much weaker at 0.1 m
M
calcium
compared to the higher concentrations, and at 0.01 m
M
calcium, no low M
r
bands appeared. This calcium-induced
conversion of the MMP-9/CSPG complex to lower M
r
forms was not affected by the presence of 0.05% Brij-35, a
compound known to inhibit autoactivation and autolytic
degradation of MMP-2 and MMP-9 [38]. Thus, treatment
of the MMP-9/CSPG complex with calcium resulted in
proteolytic cleavage and the release of the gelatinase from
the complex. When the complex was incubated for 2 h at
37 °C in the presence of either 0.001–10 m

M
ZnCl
2
,10m
M
MgCl
2
or 10–200 m
M
of NaCl, no conversion to low
molecular size forms could be detected in gelatin zymogra-
phy (data not shown). The calcium-induced activation and
truncation of the MMP-9/CSPG complex was not affected
by the presence of either 10 m
M
MgCl
2
or 200 m
M
NaCl
(data not shown). Thus, the salt-induced processing of the
complex is not due to a general electrostatic or ionic strength
effect, but appears to be a unique effect of the chemical
properties of calcium.
As expected, the degradation and removal of CS-chains
from the complex resulted in the disappearance or a large
reduction of the bands in the stacking gel and the band at
300 kDa, and new bands around 120–150 kDa appeared
(Fig. 3A, lane 4). Treatment of this MMP-9/PG-core
protein complex with 10 m

M
of CaCl
2
for 2 h at 37 °C
prior to electrophoresis resulted in the appearance of new
bands with the same M
r
as the bands from the cABC
untreated material (Fig. 3A, lane 5). The same result was
obtained if the degraded CS-chains were removed or not
removed from the sample, demonstrating that the
CS-chains were not involved in the calcium-induced
processing and release of the gelatinase from the complex.
The CaCl
2
-induced conversion of the MMP-9/CSPG
complex to lower molecular size forms was inhibited by the
presence of either 10 m
M
of EDTA (Fig. 3A, lanes 3 and 6),
1–20 n
M
of human recombinant TIMP-1 (Fig. 3B) or 1 m
M
of 1,10-phenanthroline (data not shown). The addition of
these inhibitors to the complex after calcium activation, but
prior to electrophoresis gave the same pattern as without
inhibitors (data not shown). The CaCl
2
induced conversion

of the MMP-9/CSPG complex to lower molecular size
forms was not inhibited by the presence of either pefabloc,
leupeptin, N-ethylmaleimide or pepstatin (data not shown),
i.e. general inhibitors of serine, thiol and acid proteinases.
As EDTA inhibited the calcium-induced conversion of
the MMP-9/CSPG complex to lower M
r
forms, we used this
inhibitor to investigate the kinetics of the processing of the
complex. The CaCl
2
-induced conversion to lower molecular
size forms was stopped by 10 m
M
EDTA after 1, 5, 10, 15,
30, 60 and 120-min incubation, which showed that the
bands at 80, 85 and 100 kDa were intermediates, with
maximum intensity between 5 and 15 min (Fig. 4A). The
intensity of the 74/76 kDa doublet increased during the
entire incubation period (Fig. 4A). However, in a few other
preparations of the MMP-9/CSPG complex, the induced
conversion to lower M
r
forms by CaCl
2
was slower, and the
conversion seemed to involve at least one additional step, i.e.
the formation of a transient species at 180 kDa (Fig. 4B). As
with the other preparations, the intensity of the 74/76 kDa
bands increased during the entire incubation period.

The released forms of MMP-9 from the proMMP-9/CSPG
complex were N- and C-terminally truncated
To determine whether the different bands obtained after
CaCl
2
treatment lacked either the N- or the C-terminal
regions, Western blots were performed using various
antibodies against MMP-9. Under reducing conditions,
only the 92 kDa band was obtained with all antibodies used
(Fig. 5). When the complex had been incubated for 2 h in
the presence of CaCl
2
, this 92 kDa band was weaker than in
the untreated control (Fig. 5). As no bands with an M
r
lower than 92 kDa appeared in the blots treated with
antibodies against the proform of MMP-9 (Fig. 5A), these
results show that the calcium-induced truncated forms of
MMP-9 must lack the N-terminal pro-domain. A strong
band at approximately 70 kDa appeared in the CaCl
2
-
treated material when the polyclonal antibody that recog-
nizes the hinge region was used (Fig. 5B, lane 2). Intact
MMP-9 was only weakly stained by this antibody (Fig. 5B,
lane 1). As proMMP-9 in the serum-free conditioned
medium from THP-1 cells also was weakly stained by this
antibody (data not shown), it appears that the epitope in the
hinge region is partly hidden in the intact enzyme. The fact
Fig. 3. Activation of proMMP-9/CSPG with CaCl

2
results in the
release of low M
r
forms of the gelatinase. Gelatin zymography of 4 lg
per lane of MMP-9/CSPG, which has been incubated for 2 h at 37 °C
with (+) or without (–) cABC, CaCl
2
(10 m
M
), EDTA (10 m
M
)or
human recombinant TIMP-1 as indicated under the figure. (A)
Arrowheads show the 74/76 kDa doublet and 80 and 85 kDa forms of
gelatinase in the CaCl
2
treated material. (B) The amount of TIMP-1
used was 1 (lane 3), 10 (lane 4) and 20 n
M
(lane 5). In (A) and (B), at
the left side is shown the position of the 225 and 92 kDa forms of
proMMP-9 in serum-free culture medium of THP-1 cells and the
72 kDa form of proMMP-2 in serum-free culture medium of human
skin fibroblasts. Arrow shows the border between the stacking and the
separating gel. Due to the high glycosylation of the proMMP-9/CSPG
complex, the proteins migrate as if they initially are distributed to the
edges of the stacking gel well, and two spots in the separating gel
appears instead of a clear band. This is typical for highly glycosylated
proteins as described by Carlsson [50].

4000 J O. Winberg et al. (Eur. J. Biochem. 270) Ó FEBS 2003
that no band was detected at 80/84 kDa in the CaCl
2
-
treated sample, but only a strong band at approximately
70 kDa indicates that also a certain degree of C-terminal
processing is necessary to fully expose the epitope. The
70 kDa band was not detected by the polyclonal antibody
against the C-terminal region of MMP-9 (Fig. 5C, lane 2),
showing that the 74/76 kDa bands lack large parts of their
C-terminal region. This antibody detected a weak band
at around 80 kDa in the CaCl
2
treated material (Fig. 5C,
lane 2). Thus, calcium induced both an N- and a C-terminal
truncation of CSPG bound proMMP-9.
As a control, the proMMP-9/CSPG complex was treated
with either 10 m
M
EDTA, 1 m
M
1,10-phenanthroline or
various amounts of TIMP-1 (50, 100 and 200 n
M
)priorto
incubation of these mixtures with 10 m
M
of CaCl
2
in 2 h at

37 °C. These samples were thereafter treated with 0.1
M
dithiothreitol and subjected to SDS electrophoresis and
analyzed by Western blotting, using the two polyclonal
antibodies against the hinge and C-terminal region, respect-
ively. As seen in Fig. 5B,C (lanes 3 and 4), only the 92 kDa
species is seen in the samples treated with either EDTA or
1,10-phenanthroline, and the intensity corresponds to the
untreated sample. Increasing concentrations of TIMP-1
resulted in a successively reduced amount of the 70 kDa
form of the enzyme (using the antibody against the hinge
region, data not shown).
The released forms of MMP-9 from the heterodimer
are active
The proMMP-9/CSPG complex treated with 10 m
M
CaCl
2
for 2 h at 37 °C was applied to Q-Sepharose chromato-
graphy. The released truncated forms of MMP-9 were
collected in the flow-through fraction, whereas the MMP-9
bound to CSPG were eluted from the column by 1.5
M
NaCl. The released forms of MMP-9 degraded [
3
H]-labeled
gelatin, both in the presence and absence of 10 m
M
CaCl
2

(Fig. 6A, lanes 5 and 6), whereas the MMP-9 complexed to
the CSPG needed CaCl
2
for activity (Fig. 6A, lanes 7 and
8). Gelatin zymography revealed that this second addition
of CaCl
2
resulted in a further release of low M
r
forms of the
gelatinase from the complex (data not shown).
The flow-through fraction from the Q-Sepharose column
of the CaCl
2
-treated proMMP-9/CSPG complex contained
the 74/76 kDa, 80/85 kDa and 100 kDa forms of MMP-9.
To determine which of these forms were active in solution,
proMMP-9/CSPG complex was incubated with CaCl
2
at
various time intervals and then mixed with a2-macroglo-
bulin, an inhibitor known to bind and trap active MMPs but
not the proform of the enzymes [39]. Gelatin zymography
(Fig. 6B) showed that all the released forms of MMP-9
(74/76, 80/85 and 100 kDa) reacted with a2-macroglobulin,
which resulted in a partial or total disappearance of the
gelatinolytic zones. This indicated that all four forms were
active in solution. However, no change in intensity of the
MMP-9 complexed to CSPG was observed in the presence
of a2-macroglobulin.

CaCl
2
-induced activation of the MMP-9/CSPG complex
was not abolished by removal of potential contaminating
proteins bound to the CS-chains
Some MMP-9/CSPG preparations needed about 20–24 h
incubation with 10 m
M
of CaCl
2
at 37 °C to release low M
r
forms from the complex, while for most preparations release
of low M
r
forms was obtained after 2 h incubation. It is
known that CS chains can bind various proteins [40]. The
variation in time needed for activation of the proMMP-9 in
different preparations could therefore be due to different
amounts of contaminating proteins that inhibits the
CaCl
2
-induced activation of the proMMP-9 in the CSPG
Fig. 4. Time dependent release of low M
r
forms of the gelatinase during
the activation of the proMMP-9/CSPG complex with CaCl
2
. Gelatin
zymography of 4 lg per lane of CSPG (A and B represent two dif-

ferent CSPG preparations), which has been incubated with 10 m
M
of
CaCl
2
for various time points at 37 °C as indicated under the figure.
The activation was stopped at the indicated time points by the addition
of 10 m
M
EDTA to the reaction mixture. At the right is shown the
position of the 225 and 92 kDa forms of proMMP-9 in serum-free
culture medium of THP-1 cells and the 72 kDa form of proMMP-2 in
serum-free culture medium of human skin fibroblasts. Arrow shows
the border between the stacking and the separating gel.
Ó FEBS 2003 Activation of the proMMP-9/CSPG complex (Eur. J. Biochem. 270) 4001
complex. Alternatively, the activation could be due to the
presence of another metalloproteinase that cleaves and
activates the proMMP-9 in the complex in the presence of
CaCl
2
.
As TIMP-1 inhibits the CaCl
2
-induced activation of the
MMP-9/CSPG complex, we investigated whether the
THP-1 cells produced TIMP-1. Western blots showed that
the serum-free conditioned THP-1 medium contained
TIMP-1 (Fig. 7, lane 2), and from ELISA the amount of
TIMP-1 in the conditioned medium was estimated to be
approximately 8.4 lgÆmL

)1
. We also investigated whether
some of the TIMP-1 was bound to the proMMP-9/CSPG
complex in spite of the dissociating conditions used to avoid
unspecific binding during the isolation procedure. In some
of the purified MMP-9/CSPG preparations, a small amount
of TIMP-1 was detected (Fig. 7, lane 1), while no TIMP-1
was detected in other preparations (data not shown). The
various amounts of TIMP-1 in the different proMMP-9/
CSPG preparations may therefore explain the variations in
the time needed to activate the complex with CaCl
2
.
During purification, 6
M
urea was present to prevent
aggregation and to remove proteins reversibly bound to the
proMMP-9/CSPG complex. Urea and salts were finally
removed by gel filtration, followed by a concentration step
of the purified complex. If urea and salts were not properly
removed in all preparations, this might affect the CaCl
2
-
induced conversion of the complex to lower M
r
forms of
the gelatinase, as well as the cABC degradation of the
CS-chains of the PG. The presence of 4.5
M
urea in the

proMMP-9/CSPG preparation had no effect on the posi-
tion of the two main gelatinase bands (Fig. 8, lanes 1 and 2),
suggesting that neither the band seen in the stacking gel nor
the 300-kDa band at the top of the separating gel are due
to aggregation. However, 4.5
M
urea inhibited both the
CaCl
2
-induced conversion of the complex to lower M
r
forms (Fig. 8, lanes 5 and 6) and also cABC-degradation of
the CS-chains (Fig. 8, lanes 3 and 4). Up to 0.5
M
urea had
no effect on the CaCl
2
-induced conversion to lower M
r
forms, while a concentration-dependent effect was seen
from 1
M
andupto4.5
M
(data not shown). In contrast to
this, it was first at a concentration of 4.5
M
that urea had an
effect on the cABC degradation of the CS-chains. Thus,
remnants of urea in some preparations may explain the

delay in the kinetics of the CaCl
2
-induced conversion of the
proMMP-9/CSPG complex to lower M
r
forms.
Previously, we have shown that conditioned THP-1
medium did not contain MMP-1, MMP-2 and MMP-8 [23].
The two former enzymes are known activators of MMP-9
[13]. Another MMP that is known to activate MMP-9 is
matrilysin (MMP-7) [13] that is also known to bind strongly
to GAG-chains, especially to heparin and heparan sulphate
[41]. However, neither the THP-1 medium nor the purified
proMMP-9/CSPG complex revealed bands between 18 and
30 kDa in zymograms of gels containing either gelatin,
casein or carboxymethylated-transferin (data not shown);
nor was MMP-7 detected with Western blots (data not
shown).
The proMMP-9/CSPG complex was treated in ways that
were expected to result in the release and separation of
potential contaminating proteins bound to the CS-chains of
the complex. In those experiments both intact and cABC-
treated proMMP-9/CSPG complexes were subjected to the
same chromatography columns, but in different experi-
ments. Conditions were used so that only the intact or the
cABC-treated complexes bound to the column. Intact
proMMP-9/CSPG bound to the Q-Sepharose column,
while the proMMP-9/PG-core protein passed through this
column. The opposite was the case when these complexes
Fig. 5. Western blots showing N- and C-terminal truncation of the low M

r
forms from the CaCl
2
-activated MMP-9/CSPG complex. Purified CSPG
(30 lg) was treated for 2 h at 37 °Cwith(+)orwithout(–)10m
M
CaCl
2
,10m
M
EDTA, 1 m
M
1,10-phenanthroline (OP) as indicated under the
figure. Western blots were run under reducing conditions, i.e. samples were treated with 0.1
M
dithiothreitol prior to electrophoresis. In A, the
monoclonal MMP-9 antibody (IM10L) from Amersham was used. This antibody detects only the pro-form of the enzyme. In B, a polyclonal
antibody against the hinge region of MMP-9 was used. In C, a polyclonal antibody against the C-terminal region of MMP-9 was used. M
r
standard
markers are shown at the left, and the arrowhead indicates the position of the 92 kDa band in the serum-free culture medium from THP-1 cells.
4002 J O. Winberg et al. (Eur. J. Biochem. 270) Ó FEBS 2003
were applied to Gelatin-Sepharose and Heparin-Sepharose
columns. In other experiments, intact proMMP-9/CSPG
was subjected to gel chromatography (Sephacryl 200) in the
presence and absence of EDTA. In all these experiments,
the obtained proMMP-9/CSPG complex and proMMP-9/
PG-core protein complex could be activated by the addition
of CaCl
2

. However, none of the complexes could be
activated by APMA or APMA plus CaCl
2
as shown by
zymography and degradation of
3
H-labeled gelatin (data
not shown). This strongly indicates that the calcium-induced
activation of the proMMP-9 complexed to CSPG was not
due to any CS-chain bound contaminants.
Fig. 8. Urea inhibits both the CaCl
2
-induced release of low M
r
forms of
the gelatinase and cABC degradation of the CS-chains in the complex.
Four micrograms of purified CSPG was incubated for 2 h at 37 °C
with (+) or without (–) 4.5
M
urea, cABC and 10 m
M
CaCl
2
as indi-
cated under the figure. Arrow shows the border between the stacking
and the separating gel. At the left side is shown the position of the 225
and 92 kDa forms of proMMP-9 in serum-free culture medium of
THP-1 cells and the 72 kDa form of proMMP-2 in serum-free culture
medium of human skin fibroblasts.
Fig. 7. Western blots showing TIMP-1 in purified MMP-9/CSPG and

in serum-free THP-1 conditioned medium. Both the purified CSPG
(30 lg) and the conditioned medium was treated with 0.1
M
dithio-
threitol prior to electrophoresis. M
r
standard markers are shown at the
left. The position of commercial human recombinant TIMP-1 was
identical with the bands seen in the figure.
Fig. 6. Released forms of MMP-9 from the CaCl
2
-treated heterodimer
are active. (A) Purified CSPG was treated with CaCl
2
for 2 h at 37 °C,
and thereafter applied to a Q-Sepharose column as described in
Experimental procedures. MMP-9 forms that were released from the
CSPG were collected in the flow through (F) fractions, whereas the
MMP-9 that was still bound to CSPG (B) was attached to the column.
The bound material was eluted with 1.5
M
NaCl. The flow through
material (F), the bound material (B) and the starting material of intact
CSPG (U) was either treated for 2 h at 37 °Cwith(+)orwithout(–)
10 m
M
CaCl
2
as indicated under the figure, and then incubated with
[

3
H]gelatin for approximately 24 h at 37 °C as described in Experi-
mental procedures. At the left is shown the M
r
of the rainbow M
r
standard markers in lane 1, and the arrowhead indicates the bottom of
the application well. Lane 2 shows the negative control of nondegraded
[
3
H]gelatin, and lane 3 shows a positive control of trypsin digested
[
3
H]gelatin. In lane 5, the bands at 30 kDa and below appear as two
spots instead of a band due to a crack in the dried gel. (B) Gelatin
zymography of CSPG treated with CaCl
2
at different time intervals as
indicated under the figure. At the indicated time points, each sample was
either untreated (–) or treated (+) with 800 lgÆmL
)1
of a2-macro-
globulin (a2MG) for 10 min at 37 °C, after which 10 m
M
EDTA was
added to the reaction mixture to stop the activation. At the left is shown
the position of the 225 and 92 kDa forms of proMMP-9 in serum-free
culture medium of THP-1 cells and the 72 kDa form of proMMP-2 in
serum-free culture medium of human skin fibroblasts. The arrow shows
the border between the stacking and the separating gel.

Ó FEBS 2003 Activation of the proMMP-9/CSPG complex (Eur. J. Biochem. 270) 4003
The calcium-induced activation of proMMP-9 is restricted
to the proMMP-9 covalently bound to CSPG
If the calcium-induced activation of proMMP-9 com-
plexed to CSPG was due to any contaminations, it should
be expected that such contaminants also could activate the
monomeric/homodimeric proMMP-9. Monomeric and
homodimeric forms of proMMP-9 were therefore purified
from THP-1 conditioned medium. As shown in Fig. 9, the
same pattern occurred when proMMP-9 was incubated
with either CaCl
2
treated or untreated proMMP-9/CSPG
complex in the presence or absence of 1,10-phenanthro-
line. This was also the case when these mixtures were
treated with or without Brij-35 (0.05%) or SBTI (data not
shown). These experiments show that the MMP-9/CSPG
complex in the presence of calcium is not able to activate
and process externally added monomeric/homodimeric
proMMP-9, which indicates that the calcium-induced
activation and processing of the proMMP-9 is restricted
to proMMP-9 complexed to CSPG. Further, the results
strongly suggest that the calcium-induced activation does
not involve any contaminating calcium-dependent pro-
teinase, but is due to an autoproteolysis of the proMMP-
9/CSPG complex.
To investigate if the calcium-induced activation of the
proMMP-9/CSPG complex is a bimolecular reaction or
not, various concentrations of the complex were incubated
with and without 10 m

M
of exogenous CaCl
2
for 0–4 h at
37 °C. As shown in Fig. 10, the calcium-induced activation
was concentration independent, which strongly suggests
that the autoactivation and the truncation process is not a
bimolecular reaction, but rather an intramolecular reaction.
Discussion
Previously, we have shown that a significant amount of the
MMP-9 produced by THP-1 cells is linked to the core
protein of CSPG [23]. Western blots indicated that it was the
92 kDa proform that was bound to the CSPG. In the
present work we have investigated the activity and condi-
tions for inducing activation of the MMP-9 bound to the
CSPG core protein. The enzyme in the complex was inactive
in the soluble activity assay in the absence of exogenously
added calcium, supporting that the synthesized complex
contains only the 92 kDa pro-form of the gelatinase. The
addition of exogenous calcium resulted in the generation of
N- and C-terminally truncated forms of MMP-9 which were
enzymatically active. Inhibition studies showed that the
truncation and activation of the pro-MMP-9 in the complex
was due to a metalloproteinase, and most likely a matrix
metalloproteinase as TIMP-1 inhibited the activation. Our
results exclude calcium-induced truncation and activation as
a bimolecular process that involves either another metallo-
proteinase or an N-terminally truncated form of MMP-9.
The process is more likely to be due either to an
intramolecular autoactivation process (i.e. within one het-

erodimer) or to two MMP-9/PG molecules existing as
dimers where the MMP-9 in each subunit might proteo-
lytically cross-activate each other. Although the autoacti-
vation in the latter scenario involves two distinct MMP-9
molecules, the process will not follow the kinetics of a
bimolecular reaction as the two MMP-9 molecules that act
on each other occurs within the same dimer. This conclusion
Fig. 10. Calcium-induced truncation and release of MMP-9 from the
proMMP-9/CSPG complex was not dependent on the concentration of
the complex. Various concentrations of the proMMP-9/CSPG com-
plex were incubated for 2 h at 37 °C in the absence (–) or presence (+)
of CaCl
2
(10 m
M
) as indicated in the figure, after which 2.1 lgof
CSPG was withdrawn and loaded to the gel. Shown are the released
forms from 74 to 100 kDa in the calcium-treated material, while a very
faintbandisseenintheuntreatedmaterial.Totheleftisshownthe
position of the 92 kDa form of proMMP-9 in serum-free culture
medium of THP-1 cells and the 72 kDa form of proMMP-2 in serum-
free culture medium of human skin fibroblasts. At the bottom of the
figure is shown the relative amount (± SD) of the CaCl
2
-induced
74–100 kDa forms from four independent experiments. The amount
released from the CSPG with the lowest concentration (0.5 mgÆmL
)1
)
was set to 100%.

Fig. 9. Neither proMMP-9/CSPG nor calcium-activated proMMP-9/
CSPG activates endogenously added proMMP-9 monomer and
homodimer. The monomer (92 kDa) and dimer (225 kDa) forms of
proMMP-9 was isolated from conditioned THP-1 medium as des-
cribed in Experimental procedures. The partly purified proMMP-9
was incubated for 24 h at 37 °Cwith3lgofpurifiedCSPGinthe
presence (+) or absence (–) of 10 m
M
CaCl
2
and 1 m
M
1,10-phen-
anthroline (OP) as indicated under the figure. Aliquots corresponding
to 0.27 lg of CSPG were then added to the gel, in order to prevent the
appearance of the released forms of MMP-9 from the complex. The
arrowhead shows the border between the stacking and separating gels,
and the arrow shows the position of the 72 kDa form of proMMP-2 in
serum-free culture medium of human skin fibroblasts. Similar results
appeared when the CSPG was treated with CaCl
2
for 2 h at 37 °C
prior to mixing with the partly purified proMMP-9.
4004 J O. Winberg et al. (Eur. J. Biochem. 270) Ó FEBS 2003
is based on the following observations: (a) neither calcium-
treated nor -untreated proMMP-9/CSPG activated exo-
genously added monomeric or homodimeric proMMP-9;
(b) the calcium-induced activation of the proMMP-9 in the
complex was not dependent on the concentration of the
complex; (c) calcium-induced activation took place even in

preparations of the proMMP-9/CSPG and proMMP-9/
PG-core protein complexes that were treated in such ways
that possible CS-chain bound activators were released and
removed from the complex. Large amounts of endogen-
ously produced TIMP-1 in the conditioned medium prob-
ably explains why the proMMP-9 in the complex was not
activated and processed to low M
r
forms already during the
72 h cell synthesis period, and hence before the isolation
procedure started.
In the present work, it is shown that the interaction
between proMMP-9 and the CSPG core protein causes
changes in the proMMP-9 with respect to its ability to
autoactivate. The first example is the response to urea.
During the purification of the complex, large amounts of
urea was added to the preparation to dissolve and remove
reversibly bound contaminants from the complex. Although
urea is known to induce autoactivation and processing of
92 kDa proMMP-9 to lower M
r
forms [42], the treatment
of the proMMP-9/CSPG complex with urea during the
purification procedure did not induce activation and
processing of the proenzyme in the complex. Only the
92 kDa pro-form of MMP-9 was detected in the purified
material under reducing conditions.
The second example is the response to APMA alone or in
combination with CaCl
2

, which did not result in truncation
and activation of the enzyme in the proMMP-9/CSPG
complex. Previous studies have shown that treatment of
calcium-depleted proMMP-9 with APMA resulted in an
inactive enzyme that had lost approximately 8–9 kDa of its
N-terminal prodomain, with Met75 as the N-terminal
residue [32,33]. This form of the enzyme had retained the
78PRCGVPD sequence that blocks the active site
[32,43,44]. However, in the presence of Ca
2+
the APMA-
treated enzyme was active. This was due to a further
processing of the enzyme, such that its C-terminal end was
autocatalytically removed [32,43,44]. It has been suggested
that calcium induced a conformational change in the
N-terminally truncated enzyme and unblocked the active
site. Site-directed mutagenetic studies indicate that calcium
interacted with Asp432 and probably a residue in the
remaining prodomain or in the catalytic domain [33].
Several reports show that the APMA-induced autoactiva-
tion of various proMMPs is a complicated process. To
achieve an understanding of the mechanism behind the
APMA induced activation, Cys75 was chemically modified,
the prodomain was successively deleted and the amino acids
in the 73PRCGVPD were changed through site-directed
mutagenesis of proMMP-3 [45–47]. These studies indicated
that APMA was first bound to residues other than Cys75 in
the prodomain and induced a conformational change prior
to binding to the Cys75, followed by autoactivation. It was
also shown that if 63 or more amino acids in the prodomain

of MMP-3 were deleted, addition of APMA no longer
accelerated, but rather inhibited the autoactivation process.
The third example that shows the interaction between
proMMP-9 and the PG-core protein alters the ability of the
gelatinase to autoactivate is the effect of calcium, which is a
stabilizer of proMMP-9 and other MMPs, but induces
truncation and activation of the proMMP-9/CSPG com-
plex. The various and complex effects metals exert on
MMPs can be visualized by a recent study, where it was
shown that calcium and zinc, but neither of the metals
alone, could activate a truncated form of human proMMP-
3 that lacked the first 34 N-terminal amino acids and the
entire C-terminal hemopexin domain [48]. The difference in
response between proMMP-9 and the proMMP-9/CSPG
complex to the treatment of urea, APMA, APMA in
combination with calcium, and calcium alone is most likely
due to the interaction between the enzyme and the core
protein of the CSPG, and not through a general electrostatic
effect involving the CS-chains as 200 m
M
of NaCl alone
could not induce activation of the complex, nor did high salt
concentrations affect the calcium-induced activation. Our
observations that the ionic strength is not of importance for
the activation was also reflected by the fact that the
activation was equally as effective at 100 m
M
as at 10 m
M
of

calcium. The metal-induced activation of the MMP-9/
CSPG complex was specific for calcium as the activation
process could not be mimicked by either NaCl, MgCl
2
,
ZnCl
2
or mercury (APMA). Thus, the interaction between
proMMP-9 and the CSPG core protein probably generates
a binding site for exogenous calcium that causes destabili-
zation in the proenzyme and allows for autoactivation of the
enzyme. This interaction must also hide the epitopes that are
normally involved in the APMA-induced activation of
MMP-9, and expose epitopes that results in an APMA-
induced inhibition of the calcium-induced activation.
Recently it has been shown that the various forms of
proMMP-9 are not equally susceptible to activation. The
monomeric and homodimeric forms of MMP-9 respond
differently to MMP-3-induced activation of these enzymes
[22], while the proMMP-9/NGAL heterodimer was more
effectively activated by mercurial compounds in the pre-
sence of human neutrophil lipocalin (HNL) than both the
monomeric and homodimeric forms of proMMP-9 [49].
Thus, formation of various dimers of MMP-9 results in
enzyme variants with altered biochemical properties, which
expand the biological properties and function of the enzyme
that might be optimal under various conditions.
The calcium-induced activation of the proMMP-9/CSPG
complex resulted in truncated variants of MMP-9 that had
lost both the N-terminal domain and large parts of the

C-terminal domain. The truncated variants that were
released from the CSPG had lost their N-terminal part
and were active in solution, as they reacted with a2-macro-
globulin and degraded
3
H-labeled gelatin. In addition, the
smallest truncated variants contained the hinge region that
connects the catalytic site and the C-terminal hemopexin-
like domain, but at least a part of the C-terminal domain
was lacking. These results indicate that the interaction
between MMP-9 and the CSPG core protein must involve
the most C-terminal part of the MMP-9 hemopexin-like
domain. Some of the truncated variants that had lost their
N-terminal pro-domain, were active and reacted with
a2-macroglobulin, and had an M
r
that was larger than the
92 kDa proform of MMP-9. In those variants, a part of
the core protein from the CSPG must still be linked to
the N-terminal truncated MMP-9 molecule. Thus, the
Ó FEBS 2003 Activation of the proMMP-9/CSPG complex (Eur. J. Biochem. 270) 4005
autoactivation process must also involve cleavage of the
CSPG core protein in such a way that the part that contains
the CS-chains no longer binds to the gelatinase, but only a
small processed part of the core protein remains bound to
the enzyme.
The calcium-induced activation and processing of the
proMMP-9 bound to the CSPG core protein appeared to be
a stepwise process. First the N-terminal prodomain was
removed, followed by a stepwise shortening of the

C-terminal domain. It appeared that the core protein of
the CSPG was cleaved prior to the processing and
shortening of the C-terminal hemopexin-like domain, as
the 100 kDa form appeared to be an intermediate form that
preceded the final 74/76 kDa forms. This also implies that
both the C-terminal end of the MMP-9, as well as the
covalently bound CSPG core protein, must be able to
interact with the active site of the bound gelatinase in order
to be cleaved through an intramolecular process.
The fact that MMP-9 can exist in various dimeric
forms, in addition to the monomeric form, suggests that
the regulation of activity and targeting of the gelatinase is
complex. When MMP-9 is linked to the core protein of
CSPG one can anticipate that the CSPG can localize and
concentrate the bound gelatinase to target sites other than
those usually targeted by the free monomer or its other
dimeric forms. One such target site may be the CD44
receptor located on the surface of various cells. The
function and activity of the MMP-9/CSPG complex at a
target site is probably, at least partly, regulated by calcium
and TIMP-1. The calcium-induced activation and trunca-
tion is of potential physiological relevance as the concen-
tration needed is within the physiological level of calcium
in the extracellular matrix. Once activated, the MMP-9
will be released from the complex and can diffuse into the
surrounding tissue where it may act distantly from its
original attachment site. Such behavior may be beneficial
for migration of cells. Enzymes attached to the cell surface
can act on its immediate environment, while the released
enzymes can degrade barriers at a distance from the cell.

The fact that the various forms of MMP-9 respond
differently to known activators and stabilizing reagents
makes MMP-9 capable of acting optimally during various
conditions.
Acknowledgements
This work was supported in part by grants from The Norwegian
Research Council, The Norwegian Cancer Society and the Erna and
Olav Aakre Foundation for Cancer Research.
References
1. Nagase, H. & Woessner, J.F. Jr (1999) Matrix metalloproteinases.
J. Biol. Chem. 274, 21491–21494.
2. Johansson, N., Ahonen, M. & Kahari, V.M. (2000) Matrix
metalloproteinases in tumor invasion. Cell Mol. Life Sci. 57, 5–15.
3. Birkedal-Hansen, H., Moore, W.G., Bodden, M.K., Windsor,
L.J., Birkedal-Hansen, B., DeCarlo, A. & Engler, J.A. (1993)
Matrix metalloproteinases: a review. Crit.Rev.OralBiol.Medical
4, 197–250.
4. Coussens, L.M. & Werb, Z. (1996) Matrix metalloproteinases and
the development of cancer. Chem. Biol. 3, 895–904.
5. Machein, U. & Conca, W. (1997) Expression of several matrix
metalloproteinase genes in human monocytic cells. Adv. Exp Med.
Biol. 421, 247–251.
6. Shapiro, S.D., Campbell, E.J., Senior, R.M. & Welgus, H.G.
(1991) Proteinases secreted by human mononuclear phagocytes.
J. Rheumatol. Suppl. 27, 95–98.
7. Welgus, H.G., Campbell, E.J., Cury, J.D., Eisen, A.Z., Senior,
R.M., Wilhelm, S.M. & Goldberg, G.I. (1990) Neutral metallo-
proteinases produced by human mononuclear phagocytes: enzyme
profile, regulation, and expression during cellular development.
J. Clin. Invest. 86, 1496–1502.

8. Welgus, H.G., Senior, R.M., Parks, W.C., Kahn, A.J., Ley, T.J.,
Shapiro, S.D. & Campbell, E.J. (1992) Neutral proteinase
expression by human mononuclear phagocytes: a prominent role
of cellular differentiation. Matrix Suppl. 1, 363–367.
9. Gibbs, D.F., Warner, R.L., Weiss, S.J., Johnson, K.J. & Varani, J.
(1999) Characterization of matrix metalloproteinases produced
by rat alveolar macrophages. Am.J.Respir.CellMol.Biol.20,
1136–1144.
10. Murphy, G. & Reynolds, J.J. (1993) Extracellular matrix
degradation. In Connective Tissue and its Heritable Disorders.
Molecular, Genetic and Medical Aspects (Royce, P.M. & Steinman,
B., eds), pp. 287–316. Wiley-Liss Inc, New York.
11. Liu, Z., Zhou, X., Shapiro, S.D., Shipley, J.M., Twining, S.S.,
Diaz, L.A., Senior, R.M. & Werb, Z. (2000) The serpin alpha1-
proteinase inhibitor is a critical substrate for gelatinase B/MMP-9
in vivo. Cell 102, 647–655.
12. Manes, S., Llorente, M., Lacalle, R.A., Gomez-Mouton, C.,
Kremer, L., Mira, E. & Martinez, A.C. (1999) The matrix
metalloproteinase-9 regulates the insulin-like growth factor-
triggered autocrine response in DU-145 carcinoma cells. J. Biol.
Chem. 274, 6935–6945.
13. Nagase, H. (1997) Activation mechanisms of matrix metallopro-
teinases. Biol. Chem. 378, 151–160.
14. Matrisian, L.M. (1990) Metalloproteinases and their inhibitors in
matrix remodeling. Trends Genet. 6, 121–125.
15. Woessner, J.F. Jr (1991) Matrix metalloproteinases and their
inhibitors in connective tissue remodeling. FASEB J. 5, 2145–
2154.
16. Van Wart, H.E. & Birkedal-Hansen, H. (1990) The cysteine
switch: a principle of regulation of metalloproteinase activity with

potential applicability to the entire matrix metalloproteinase gene
family. Proc. Natl Acad. Sci. USA 87, 5578–5582.
17. Nagase, H., Suzuki, K., Morodomi, T., Enghild, J.J. & Salvesen,
G. (1992) Activation mechanisms of the precursors of matrix
metalloproteinases 1, 2 and 3. Matrix Suppl. 1, 237–244.
18. Liotta, L.A. & Stetler-Stevenson, W.G. (1991) Tumor invasion
and metastasis: an imbalance of positive and negative regulation.
Cancer Res. 51, 5054s–5059s.
19. Hibbs,M.S.,Hasty,K.A.,Seyer,J.M.,Kang,A.H.&Mainardi,
C.L. (1985) Biochemical and immunological characterization of
the secreted forms of human neutrophil gelatinase. J. Biol. Chem.
260, 2493–2500.
20. Triebel, S., Blaser, J., Reinke, H. & Tschesche, H. (1992) A 25 kDa
alpha 2-microglobulin-related protein is a component of the 125
kDa form of human gelatinase. FEBS Lett. 314, 386–388.
21. Kjeldsen, L., Johnsen, A.H., Sengelov, H. & Borregaard, N.
(1993) Isolation and primary structure of NGAL, a novel protein
associated with human neutrophil gelatinase. J. Biol. Chem. 268,
10425–10432.
22. Olson, M.W., Bernardo, M.M., Pietila, M., Gervasi, D.C., Toth,
M., Kotra, L.P., Massova, I., Mobashery, S. & Fridman, R.
(2000) Characterization of the monomeric and dimeric forms of
latent and active matrix metalloproteinase-9: differential rates for
activation by stromelysin 1. J. Biol. Chem. 275, 2661–2668.
4006 J O. Winberg et al. (Eur. J. Biochem. 270) Ó FEBS 2003
23. Winberg, J.O., Kolset, S.O., Berg, E. & Uhlin-Hansen, L. (2000)
Macrophages secrete matrix metalloproteinase 9 covalently linked
to the core protein of chondroitin sulphate proteoglycans. J. Mol.
Biol. 304, 669–680.
24. Cha, H., Kopetzki, E., Huber, R., Lanzendorfer, M. & Brand-

stetter, H. (2002) Structural basis of the adaptive molecular
recognition by MMP9. J. Mol. Biol. 320, 1065–1079.
25. Kjellen, L. & Lindahl, U. (1991) Proteoglycans: structures and
interactions. Annu. Rev. Biochem. 60, 443–475.
26. Uhlin-Hansen, L., Wik, T., Kjellen, L., Berg, E., Forsdahl, F. &
Kolset, S.O. (1993) Proteoglycan metabolism in normal and
inflammatory human macrophages. Blood 82, 2880–2889.
27. Øynebra
˚
ten, I., Hansen, B., Smedsrød, B. & Uhlin-Hansen, L.
(2000) Serglycin secreted by leucocytes is efficiently eliminated
from the circulation by sinusoidal scavanger endothelial cells in the
liver. J. Leukoc. Biol. 67, 183–188.
28. Butler, G.S., Butler, M.J., Atkinson, S.J., Will, H., Tamura, T.,
van Westrum, S.S., Crabbe, T., Clements, J., d’Ortho, M.P. &
Murphy, G. (1998) The TIMP2 membrane type 1 metallopro-
teinase ÔreceptorÕ regulates the concentration and efficient activa-
tion of progelatinase A: a kinetic study. J. Biol. Chem. 273,
871–880.
29. Wallon, U.M. & Overall, C.M. (1997) The hemopexin-like
domain (C domain) of human gelatinase A (matrix metallopro-
teinase-2) requires Ca
2+
for fibronectin and heparin binding:
binding properties of recombinant gelatinase AC domain to
extracellular matrix and basement membrane components. J. Biol.
Chem. 272, 7473–7481.
30. Housley, T.J., Baumann, A.P., Braun, I.D., Davis, G., Seperack,
P.K. & Wilhelm, S.M. (1993) Recombinant Chinese hamster
ovary cell matrix metalloprotease-3 (MMP-3, stromelysin-1): role

of calcium in promatrix metalloprotease-3 (pro-MMP-3, pros-
tromelysin-1) activation and thermostability of the low mass cata-
lytic domain of MMP-3. J. Biol. Chem. 268, 4481–4487.
31. Seltzer, J.L., Welgus, H.G., Jeffrey, J.J. & Eisen, A.Z. (1976) The
function of Ca
2+
in the action of mammalian collagenases. Arch.
Biochem. Biophys. 173, 355–361.
32. Bu, C.H. & Pourmotabbed, T. (1995) Mechanism of activation of
human neutrophil gelatinase B: discriminating between the role of
Ca
2+
in activation and catalysis. J. Biol. Chem. 270, 18563–18569.
33. Bu, C.H. & Pourmotabbed, T. (1996) Mechanism of Ca
2+
-
dependent activity of human neutrophil gelatinase B. J. Biol.
Chem. 271, 14308–14315.
34. Carrino, D.A., Arias, J.L. & Caplan, A.I. (1991) A spectro-
photometric modification of a sensitive densitometric Safranin O
assay for glycosaminoglycans. Biochem. Int. 24, 485–495.
35. Makowski, G.S. & Ramsby, M.L. (1996) Calibrating gelatin
zymograms with human gelatinase standards. Anal. Biochem. 236,
353–356.
36. Winberg, J.O. & Gedde-Dahl, T. Jr (1992) Epidermolysis bullosa
simplex: expression of gelatinase activity in cultured human skin
fibroblasts. Biochem. Genet. 30, 401–420.
37. Winberg, J.O. & Gedde-Dahl, T. Jr (1986) Gelatinase expression
in generalized epidermolysis bullosa simplex fibroblasts. J. Invest.
Dermatol. 87, 326–329.

38. Murphy, G. & Crabbe, T. (1995) Gelatinases A and B. Methods
Enzymol. 248, 470–484.
39. Nagase, H., Itoh, Y. & Binner, S. (1994) Interaction of alpha
2-macroglobulin with matrix metalloproteinases and its use for
identification of their active forms. Ann. NY Acad. Sci. 732,
294–302.
40. Kolset, S.O., Mann, D.M., Uhlin-Hansen, L., Winberg, J.O. &
Ruoslahti, E. (1996) Serglycin-binding proteins in activated mac-
rophages and platelets. J. Leukoc. Biol. 59, 545–554.
41. Yu, W.H. & Woessner, J.F. Jr (2000) Heparan sulfate pro-
teoglycans as extracellular docking molecules for matrilysin
(matrix metalloproteinase 7). J. Biol. Chem. 275, 4183–4191.
42. Sopata, I. & Maslinski, S. (1991) Activation of the latent human
neutrophil gelatinase by urea. Acta Biochim. Pol. 38, 67–70.
43. Triebel, S., Blaser, J., Reinke, H., Knauper, V. & Tschesche, H.
(1992) Mercurial activation of human PMN leucocyte type IV
procollagenase (gelatinase). FEBS Lett. 298, 280–284.
44. Okada, Y., Gonoji, Y., Naka, K., Tomita, K., Nakanishi, I.,
Iwata, K., Yamashita, K. & Hayakawa, T. (1992) Matrix
metalloproteinase 9 (92-kDa gelatinase/type IV collagenase) from
HT 1080 human fibrosarcoma cells: purification and activation of
the precursor and enzymic properties. J. Biol. Chem. 267, 21712–
21719.
45. Chen, L.C., Noelken, M.E. & Nagase, H. (1993) Disruption of the
cysteine-75 and zinc ion coordination is not sufficient to activate
the precursor of human matrix metalloproteinase 3 (stromelysin
1). Biochemistry 32, 10289–10295.
46. Galazka, G., Windsor, L.J., Birkedal-Hansen, H. & Engler,
J.A. (1996) APMA (4-aminophenylmercuric acetate) activation
of stromelysin-1 involves protein interactions in addition to

those with cysteine-75 in the propeptide. Biochemistry 35, 11221–
11227.
47. Galazka, G., Windsor, L.J., Birkedal-Hansen, H. & Engler, J.A.
(1999) Spontaneous propeptide processing of mini-stromelysin-1
mutants blocked by APMA (4-aminophenyl) mercuric acetate).
Biochemistry 38, 1316–1322.
48. Suzuki,K.,Kan,C.C.,Hung,W.,Gehring,M.R.,Brew,K.&
Nagase, H. (1998) Expression of human pro-matrix metallopro-
teinase 3 that lacks the N-terminal 34 residues in Escherichia coli:
autoactivation and interaction with tissue inhibitor of metallo-
proteinase 1 (TIMP-1). Biol. Chem. 379, 185–191.
49. Tschesche, H., Zolzer, V., Triebel, S. & Bartsch, S. (2001) The
human neutrophil lipocalin supports the allosteric activation of
matrix metalloproteinases. Eur. J. Biochem. 268, 1918–1928.
50. Carlsson, S.R. (1993) Isolation and characterization of glycopro-
teins. In Glycobiology a Practical Approach (Fukuda, M. &
Kobata, A., eds), pp. 1–26. IRL Press, Oxford.
Ó FEBS 2003 Activation of the proMMP-9/CSPG complex (Eur. J. Biochem. 270) 4007