Secretion of macrophage urokinase plasminogen activator
is dependent on proteoglycans
Gunnar Pejler
1
, Jan-Olof Winberg
2
, Tram T. Vuong
3
, Frida Henningsson
1
, Lars Uhlin-Hansen
2
,
Koji Kimata
4
and Svein O. Kolset
5
1
Department of Veterinary Medical Chemistry, Swedish University of Agricultural Sciences, Uppsala, Sweden;
2
Department of
Biochemistry, Institute of Medical Biology, University of Tromsø, Norway;
3
Department of Biochemistry, University of Oslo,
Norway;
4
Institute for Molecular Science of Medicine, Aichi Medical University, Japan;
5
Institute for Nutrition Research,
University of Oslo, Norway
The importance of proteoglycans for secretion of proteolytic

tumor necrosis factor-a, lipoprotein lipase, proteoglycans,
leukotrienes, and various proteases [2]. The proteoglycans
expressed by monocytes and macrophages have been
characterized to some extent. The major product seems to
be serglycin, as shown by N-terminal sequencing of
proteoglycans released from the cultured monocytic cell
lines U937 and THP-1 [2,3]. Moreover, it has been shown
that activated murine and human macrophages express
syndecan-4 [4] and syndecan-2 [5], respectively, on the cell
surface.
The release of serglycin from monocytes and macro-
phages is the subject of regulation by inflammatory
signaling molecules such as interferon-c, transforming
growth factor-b, and platelet derived growth factor [2,6].
It is therefore likely that the secretion of proteoglycans in
these cells is linked to inflammatory reactions and that its
function(s) may be linked to the binding, transport and
regulation of other secretory products. Indeed, recent data
indicate that mice lacking functional heparin chains
attached to their serglycin proteoglycans show severe defects
in their capacities to store mast cell proteases in the secretory
granules [7,8], clearly demonstrating the importance of
intact proteoglycans for normal storage of proteases in these
cells. Serglycin proteoglycans have also been implicated in
the regulation of mast cell protease activities [9–11].
The biological functions of proteoglycans from activated
monocytes and macrophages have not been outlined in any
detail. It has however, been shown that serglycin may be
associated with chemokines and enzymes after release from
the cells [12]. It has furthermore been demonstrated that

Enzymes: chondroitinase ABC (EC 4.2.2.4)
(Received 17 June 2003, accepted 7 August 2003)
Eur. J. Biochem. 270, 3971–3980 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03785.x
of free glycosaminoglycan chains attached to the b-
D
-
xyloside rather than intact proteoglycans. Depending on the
concentration of xylosides used, endogenous proteoglycan
expression may be completely abrogated. b-
D
-Xylosides
seem to be more efficient in abrogating the expression of
chondroitin sulfate proteoglycans than heparan sulfate
proteoglycans. Results presented here show that the treat-
ment of macrophages with b-
D
-xylosides results in impaired
secretion of urokinase plasminogen activatior (uPA), indi-
cating that uPA is dependent on proteoglycans. The
secretion of matrix metalloproteinase 9 (MMP-9) was also
decreased by the xyloside treatment.
Materials and methods
Materials
Sephadex G50 Fine and Superose 6 were from Amer-
sham Pharmacia, Uppsala, Sweden. [
35
S]Sodium sulfate
was obtained from Amersham. The chromogenic peptide
substrates S-2288 (H-D-Ile-Pro-Arg-p-nitroanilide), S-2444
(pyroGlu-Gly-Arg-p-nitroanilide), S-2390 (H-

J774), was from the American Type Culture Collection,
Rockville, MD, USA. The cells were routinely kept in
Dulbecco’s modified Eagles medium (DMEM) with 2 m
M
L
-glutamine and gentamycin (0.1 mgÆmL
)1
), all from Bio
Whittaker, Verviers, Belgium. The medium was fortified
with 10% fetal bovine serum from Sigma Chemical Co.
The human histiocytic lymphoma cell line U937 clone 1
(U937-1) was cultured in RPMI medium with 10% fetal
bovine serum, 2 m
ML
-glutamine and gentamycin
(0.1 mgÆmL
)1
), all from Bio Whittaker.
Enzyme assays
J774 cells were established in medium with serum in
16 mm wells at cell densities between 0.5 and 1.0 · 10
6
cells
per well, or in 96-well plates at densities of approximately
1.5 · 10
5
cells per well. After reaching confluency, J774
cells were washed three times in medium without supple-
ments to remove serum proteins. The cells were thereafter
cultured in the serum-free medium QBSF 51 (Sigma). Cells

(200 lL final volume)
and 20 lL of either substrate S-2288 or S-2444, dissolved
in distilled water at stock concentrations of 20 m
M
.The
enzyme activities were recorded by reading the absorbance
at 405 nm at different time points using a Titertek
Multiscan spectrophotometer (Flow Laboratories, Irvine,
Scotland). The increase in absorbance showed linear
kinetics over a time period of 5 h, indicating that the
enzyme was stable for at least this period of time in
solution.
For inhibition studies, 50 lL of conditioned medium was
mixedwith150 lLofNaCl/P
i
in 96-well plates. Next, either
of the following protease inhibitors was added at a final
concentration of 0.2 l
M
:PAI-1,a
1
-anti-chymotrypsin,
a
1
-protease inhibitor or soybean trypsin inhibitor. The
effect of phenylmethanesulfonyl fluoride at a final con-
centration of 1 m
M
was also tested. After 30 min of
incubation, 20 lL of S-2288 (20 m

standard because it contains proMMP-9 monomer, giving
risetoamainbandat92kDaandtheproMMP-9
homodimer (a minor band at 225 kDa) [16]. In addition,
serum-free conditioned medium from normal human skin
fibroblasts [18] was used as a source for pro-MMP-2
standard (72 kDa). Ten microlitres of conditioned medium
3972 G. Pejler et al.(Eur. J. Biochem. 270) Ó FEBS 2003
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). Six microlitres of this nonheated mixture was
applied to the gel, which was run at 20 mA/gel at 4 °C.
Thereafter, the gel was washed twice in 50 mL 2.5% (v/v)
Triton X-100, and then incubated in 50 mL of assay
buffer (50 m
M
Tris/HCl, pH 7.5, 5 m
M
CaCl
2
,0.2
M
NaCl and 0.02% Brij-35) for approximately 20 h at
37 °C. In some cases 10 m
M
of EDTA was added to both
the washing and assay buffers to block potential metallo-
proteinase activity, but not serine proteinase activity. In
other cases samples were incubated with 10 m

0.2% of SBTI was also incorporated in both the stacking
and separating gels to prevent degradation of the incor-
porated gelatin substrate by trace amounts of trypsin that
may escape from the inhibitor complex during electro-
phoresis.
Western blotting
Media (5 mL) from nontreated cells (control) and cells
treated with PMA and xyloside, respectively, were concen-
trated 10 times on Millipore ultrafree-15, NMWL 10 000
(Biomax-10) centrifugal filter device. The concentrated
samples were mixed with SDS/PAGE sample buffer,
without 2-mercaptoethanol. Cells (1 · 10
6
) were solubilized
by adding 100 lL of SDS/PAGE sample buffer followed by
boiling for 3 min. Samples (40 lL) from medium- or cell
fractions were subjected to SDS/PAGE on 12% polyacryl-
amide gels under reducing conditions. Proteins were subse-
quently blotted onto nitrocellulose membranes, followed by
blocking with 5% milk powder in NaCl/P
i
for 1 h at 20 °C.
Next, the membranes were incubated with antiserum
(1 : 200) in 5% milk powder/Tris/NaCl/P
i
/0.1% Tween 20,
at 4 °C for 20 h. The rabbit anti-(mouse urokinase) Ig was
a kind gift from K. Danø, Rigshospitalet, Copenhagen,
University Hospital, Denmark. After extensive washing
with Tris/NaCl/P

magnetic beads (Dynal, Oslo Norway), and
separated on 1% agarose gels containing formaldehyde
and blotted to Hybond N nylon membranes (Amersham
Pharmacia Biotech). After prehybridization the blots were
hybridizedin0.5
M
sodium phosphate buffer with 7% SDS
and 1 m
M
EDTA and
32
P-labelled probes at 65 °C for 16 h.
The blots were washed three times at 65 °Cwith40m
M
sodium phosphate containing 1% SDS, sealed and exposed
to phosphorimage screen over night. The obtained screens
were analyzed in a phosphorimager (Molecular Dynamics,
Amersham Pharmacia Biotech). Probe for murine urokin-
ase was a kind gift from L. Hellman, Uppsala University. A
probe for the housekeeping gene, 36B4, obtained from
H. Nebb, University of Oslo, was used to compare mRNA
levels in different samples.
Proteoglycan expression
To analyze the effects of PMA and HX-xyl treatment on the
expression of proteoglycans, J774 cells were labelled with
[
35
S]sodium sulfate for 24 h. PMA and HX-xyl were present
only during the labeling period. The media were harvested
and loose cells pelleted by centrifugation. The cell fractions

To analyze the possible importance of proteoglycan
expression for the secretion of proteolytic enzyme activities
in activated macrophages, J774 cells were treated with HX-
xylorPMAaloneorwithPMAandHX-xylincombina-
tion. As can be seen in Table 1, PMA treatment resulted in a
50–80% increase in total proteoglycan synthesis. Further,
treatment of the cells with HX-xyl, both in the presence or
Ó FEBS 2003 Proteoglycans and urokinase (Eur. J. Biochem. 270) 3973
absence of PMA, resulted in a marked ( threefold)
increase in the synthesis of
35
S-labelled macromolecules
(Table 1). After HX-xyl treatment, the major part of the
35
S-labelled macromolecules expressed was recovered in the
culture medium, regardless if PMA was present or not
(Table 1). In contrast, control cells and cells treated with
PMA retained a major portion of the
35
S-labelled macro-
molecules in the cell fraction (Table 1).
35
S-labelled macro-
molecules recovered from the medium fractions were
analyzed by gel chromatography to discriminate between
intact proteoglycans and free glycosaminoglycan chains.
Further, samples were analyzed both before and after
treatment with alkali (NaOH), a treatment that is known to
release glycosaminoglycans from their respective protein
cores. In agreement with a previous study [14], treatment

macromolecules after treatment with chondroitinase ABC
(Fig. 1; panels two and four). However, small amounts of
HSPGs can also found in the medium of these cultures.
Both heparan and chondroitin sulfate proteoglycans
could be detected in the cell fractions of control- and PMA-
treated cells, as well as in cells treated with HX-xyl or PMA/
HX-xyl. When these fractions were analyzed by gel
chromatography, they displayed almost identical elution
profiles (results not shown), irrespective of treatment. The
ratio between heparan sulfate and chondroitin sulfate in the
cell fractions was therefore not affected by the xyloside
treatment. The shift from chondroitin sulfate/heparan
sulfate proteoglycans to mostly free chondroitin sulfate
chains is, accordingly, only seen in the medium fractions
after HX-xyl or PMA/HX-xyl treatment.
Xyloside and serine proteinases
Conditioned medium collected after 20 h incubation
under serum-free conditions did not contain any chymo-
trypsin-like activity, as no cleavage of the chromogenic
Table 1. [
35
S]-labelled macromolecules recovered from medium and cell
fractions of J774 cells. Cells were labelled with [
35
S]sodium sulfate for
20hwiththeindicatedtreatments.[
35
S]-Labelled macromolecules
were recovered from cell and medium fractions and the amount
determined by scintillation counting. The results presented are the

subjected to deaminative cleavage (HNO
2
) to degrade heparan sulfate,
chondroitinase ABC treatment to depolymerize chondroitin/dermatan
sulfate or alkali treatment to release free GAG chains and also ana-
lyzed by gel chromatography. Equal amounts of radioactivity were
taken from the different fractions for analyses by gel chromatography.
3974 G. Pejler et al.(Eur. J. Biochem. 270) Ó FEBS 2003
chymotrypsin substrate S-2586 was observed (result not
shown). Considerable activity, however, could be detected
when the chromogenic substrate S-2288 was used. This
substrate is cleaved by enzymes with trypsin-like substrate
specificities. From Fig. 2 it is evident that the secretion of
trypsin-like activity was increased approximately twofold
when the cells were treated with PMA.
When proteoglycan expression was compromised by
treatment with HX-xyl, the levels of trypsin-like activities
recovered in the conditioned media were reduced both in
untreated and in PMA-stimulated cells by  50%. The
effects of xyloside varied somewhat between different
experiments using different cell batches. In some experi-
ments the HX-xyl treatment reduced the secretion of
trypsin-like activities to an even larger extent, both in
control and PMA-stimulated cells (not shown). The reduc-
tion in trypsin-like activity in the medium upon HX-xyl
treatment was most pronounced after extended periods of
incubation. However, time course studies revealed a clearly
noticeable effect already 1 h after the addition of HX-Xyl,
with a gradually increased effect up to 20 h of incubation
(not shown). Furthermore, in experiments with the human

a serine proteinase. Further, the activity was completely
inhibited by plasminogen activator inhibitor 1 (PAI-1), but
not to any significant extent by neither a
1
-protease inhibitor,
a
1
-anti-chymotrypsin nor soybean trypsin inhibitor (Fig. 3).
This pattern of inhibition was seen in conditioned media
both from control- and PMA-stimulated cells.
To verify that the murine macrophage cell line J774
produced plasminogen activators, cell conditioned serum-
free medium was subjected to substrate zymography [17]. As
shown in Fig. 4 (left panel), a band at approximately
24 kDa was detected in the gel that contained both
plasminogen and gelatin, but not in the control gel that
contained only gelatin. This indicates that this band is a
plasminogen activator.
The figure also shows that the presence of PMA resulted
in a slight increase in the intensity of this plasminogen
activator band, which was verified in other experiments with
diluted conditioned medium (data not shown). Figure 4
(left panel) also shows that HX-xyl treatment of the cells
resulted in a reduction in the intensity of the plasminogen
activator band. This band was also drastically reduced
in conditioned medium (control as well as PMA- and
Fig. 2. Trypsin-like activities in conditioned media from J774 macro-
phages. Equal number of J774 macrophages were incubated with
PMA, HX-xyl or both. Conditioned media were harvested and the
levels of trypsin-like activities were assayed using the chromogenic

antigen was readily detected in conditioned medium both
from control- and PMA-treated cells. Strikingly, in medium
from cells incubated with either HX-xyl alone, or with
PMA together with HX-xyl, the uPA band was nearly
undetectable.
The M
r
of the uPA detected by Western blotting is
approximately twice as large as that detected by substrate
zymography. The 24 kDa form seen in zymography is most
likely the low M
r
form of uPA consisting of only the active
site serine proteinase (SP)-module as described previously
[21], while the antibody used in the Western blots only
recognized the N-terminal part of uPA. The lack of a band
at around 48 kDa in the substrate zymography gel (Fig. 4)
indicates that the 48 kDa band seen in the Western blot is
the inactive proform of uPA.
It is possible that the effect of HX-xyl could be mediated
through increased secretion of PAI-1. A decreased activity
of uPA due to complex formation with PAI-1 should be
evident through formation of a covalent complex with high
Fig. 4. Zymographic detection of plasminogen activators and matrix
metalloproteinases in supernatants from J774 cells. Supernatants from
J774 cells were subjected to SDS/PAGE using gels containing both
gelatin and plasminogen (left panel) or only gelatin (right panel). The
cells had been treated as described in the legend to Fig. 2 prior to
harvesting of the medium. After electrophoresis, the gels were treated
as described in Materials and methods. Standard 1 is conditioned

r
of approximately 250–300 kDa and
112 kDa and were not plasminogen activators, as they were
found in both the control gel containing only gelatin as well
as in the gel with plasminogen and gelatin. These bands did
not appear in gels that were washed and incubated in the
presence of EDTA, while the intensity of the bands in
harvested media treated with the serine proteinase inhibitor
pefabloc prior to electrophoresis was similar to the bands in
the untreated control media (data not shown). This indicates
that these bands are metalloproteinases, and most likely the
dimeric and monomeric forms of metalloproteinase 9
(MMP-9), as macrophages have previously been shown to
produce this enzyme [16,22]. Treatment of the conditioned
medium with trypsin prior to electrophoresis gave a new
bandwithanapproximateM
r
of 106 kDa (data not
shown). This suggests that the metalloproteinase in the J774
medium is most likely the proform of the gelatinase.
In the medium from PMA-treated cells, the two MMP
bands appeared somewhat stronger compared to the MMP
bands in the medium from the control cells (Fig. 4).
However, in the media from the HX-xyl-treated cells these
two bands were drastically reduced compared to the
controls (Fig. 4). Thus, the secretion of metalloproteinases
is also affected by HX-xyl treatment.
Transmission electron microscopy
To investigate if HX-xyl treatment of J774 cells would affect
the formation and organization of intracellular granules,

activated [6], as was also apparent in this study. Accord-
ingly, secretion of both uPA and proteoglycans increase in
activated monocytes and macrophages. Plasmin, generated
from the precursor plasminogen through the action of uPA,
can cleave matrix proteins such as fibronectin, laminin and
aggrecan, and also activate matrix- and membrane associ-
ated MMPs, fibroblast growth factor and transforming
growth factor b [26]. In atherosclerosis, lipid-rich macro-
phages increase uPA and plasmin expression and the release
of growth factors from the extracellular matrix [27]. Clearly,
the regulation of plasmin formation is important for
macrophages and metastasizing tumor cells, and cells
involved in tissue repair. Likewise, secretion of MMP-9
from macrophages is important in immune reactions and
atherosclerosis [28]. The results presented here thus indicate
that proteoglycans secreted from macrophages, e.g. sergly-
cin, may regulate the activity or availability of uPA and
MMP-9. However, HX-xyl treatment does not lead to a
complete inhibition of uPA release from the cells, despite an
essentially total abrogation of the synthesis of intact
proteoglycans. The reason for this is not known. However,
it is possible that preformed uPA and intact proteoglycans
are present in the cells and are being released during the
course of the experiments. Alternatively, uPA secretion may
be only partly dependent on the intact proteoglycans.
Control and PMA-stimulated J774 macrophages release
proteoglycans of both chondroitin sulfate and heparan
sulfate type. In the present study we show that xyloside
treatment of both control and PMA-stimulated J774 cells
completely blocks the assembly of intact heparan sulfate

possibilities implies that the protein core of the proteogly-
can, or the intact proteoglycan molecule, is an important
component of the secretory process, as the xyloside
treatment did not reduce the amount of secreted glycos-
aminoglycan chains available. The mechanism by which the
protein core could influence the secretion of proteolytic
enzymes is uncertain. It is possible, for example, that the
protein core in some way is involved in intracellular sorting
of uPA and MMP-9. Another possibility could be that the
protein core is attached to the vesicle membrane, and that
such a linkage may be important for formation or structural
integrity of the secretory vesicles. In this context it is noted
that proteoglycans, possibly GPI-linked to the granule
membrane, are important for the formation of zymogen
granules in pancreatic acinar cells [32]. Further, proteogly-
cans have been suggested to be important for the intracel-
lular transport of enzymes to the lysosomes in monocytes
[33]. A third possibility would be that the cell-surface
proteoglycans participate in the regulation of uPA. HX-xyl-
treated cells have reduced levels of cell surface-associated
proteoglycans compared to control macrophages. Possibly,
this may affect the cell association of uPA after release and/
or the level of activity. In fact, it has been shown previously
that cell association of uPA-generating activity enhances the
rate of formation of active uPA [34].
An alternative explanation for the effect of the xyloside
on uPA and MMP-9 secretion could be that the xyloside
treatment reduces the amount of heparan sulfate chains
synthesized in favor of chondroitin sulfate, and that uPA
and MMP-9 may be specifically dependent on glycosami-

macrophages, through the dependence of cellular proteo-
glycan expression and secretion.
Acknowledgements
The expert technical assistance of Eli Berg and Annicke Stranda is
acknowledged.
This work was supported by grants from The Norwegian Cancer
Society, The Throne-Holst Fund, The Swedish Medical Research
Council (grant no. 9913) and from King Gustaf V’s 80th anniversary
Fund.
References
1. Nathan, C.F. (1987) Secretory products of macrophages. J. Clin.
Invest. 79, 319–326.
2. 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.
3. Oynebraten, I., Hansen, B., Smedsrod, B. & Uhlin-Hansen, L.
(2000) Serglycin secreted by leukocytes is efficiently eliminated
from the circulation by sinusoidal scavenger endothelial cells in the
liver. J. Leukoc. Biol. 67, 183–188.
4. Yeaman, C. & Rapraeger, A.C. (1993) Membrane-anchored
proteoglycans of mouse macrophages: P388D1 cells express a
syndecan-4-like heparan sulfate proteoglycan and a distinct
chondroitin sulfate form. J. Cell Physiol. 157, 413–425.
5. Clasper,S.,Vekemans,S.,Fiore,M.,Plebanski,M.,Wordsworth,
P., David, G. & Jackson, D.G. (1999) Inducible expression of the
cell surface heparan sulfate proteoglycan syndecan-2 (fibroglycan)
on human activated macrophages can regulate fibroblast growth
factor action. J. Biol. Chem. 274, 24113–24123.
6. Uhlin-Hansen,L.,Eskeland,T.&Kolset,S.O.(1989)Modulation
of the expression of chondroitin sulfate proteoglycan in stimulated

-xylosides on the proliferation and
proteoglycan biosynthesis of monoblastic U-937 cells. Biochem.
J. 265, 637–645.
15. Halvorsen, B., Aas, U.K., Kulseth, M.A., Drevon, C.A.,
Christiansen, E.N. & Kolset, S.O. (1998) Proteoglycans in mac-
rophages: characterization and possible role in the cellular uptake
of lipoproteins. Biochem. J. 331, 743–752.
16. 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 sulfate proteoglycans. J. Mol.
Biol. 304, 669–680.
17. Heussen, C. & Dowdle, E.B. (1980) Electrophoretic analysis of
plasminogen activators in polyacrylamide gels containing sodium
dodecyl sulfate and copolymerized substrates. Anal. Biochem. 102,
196–202.
18. Svendsrud, D.H., Loennechen, T. & Winberg, J.O. (1997) Effect
of adenosine analogues on the expression of matrix metallopro-
teinases and their inhibitors from human dermal fibroblasts.
Biochem. Pharmacol. 53, 1511–1520.
19. Shively, J.E. & Conrad, H.E. (1976) Formation of anhydrosugars
in the chemical depolymerization of heparin. Biochemistry 15,
3932–3942.
20. Vassalli, J.D. & Belin, D. (1987) Amiloride selectively inhibits the
urokinase-type plasminogen activator. FEBS Lett. 214, 187–191.
21. Novokhatny, V., Medved, L., Mazar, A., Marcotte, P., Henkin, J.
& Ingham, K. (1992) Domain structure and interactions of
recombinant urokinase-type plasminogen activator. J. Biol. Chem.
267, 3878–3885.
22. Shapiro, S.D., Campbell, E.J., Senior, R.M. & Welgus, H.G.
(1991) Proteinases secreted by human mononuclear phagocytes.

(2000) Extracellular granzyme A, complexed to proteoglycans, is
Ó FEBS 2003 Proteoglycans and urokinase (Eur. J. Biochem. 270) 3979
protected against inactivation by protease inhibitors. Blood 95,
1465–1472.
32. Schmidt, K., Dartsch, H., Linder, D., Kern, H.F. & Kleene, R.
(2000) A submembranous matrix of proteoglycans on zymogen
granule membranes is involved in granule formation in rat pan-
creatic acinar cells. J. Cell Sci. 113, 2233–2242.
33. Lemansky, P. & Hasilik, A. (2001) Chondroitin sulfate is involved
in lysosomal transport of lysozyme in U937 cells. J. Cell Sci. 114,
345–352.
34. Duval-Jobe, C. & Parmely, M.J. (1994) Regulation of plasmino-
gen activation by human U937 promonocytic cells. J. Biol. Chem.
269, 21353–21357.
35. Henningsson, F., Ledin, J., Lunderius, C., Wilen, M., Hellman, L.
& Pejler, G. (2002) Altered storage of proteases in mast cells from
mice lacking heparin: a possible role for heparin in carboxy-
peptidase A processing. Biol. Chem. 383, 793–801.
36. Schick, B.P., Gradowski, J.F. & San Antonio, J.D. (2001)
Synthesis, secretion, and subcellular localization of serglycin pro-
teoglycan in human endothelial cells. Blood 97, 449–458.
37. Hildenbrand,R.,Jansen,C.,Wolf,G.,Bohme,B.,Berger,S.,von
Minckwitz, G., Horlin, A., Kaufmann, M. & Stutte, H.J. (1998)
Transforming growth factor-beta stimulates urokinase expression
in tumor-associated macrophages of the breast. Laboratory Invest.
78, 59–71.
3980 G. Pejler et al.(Eur. J. Biochem. 270) Ó FEBS 2003