(CANCER RESEARCH 52, 5046-5053, September 15, 1992]
Multiple Secretion of Matrix Serine Proteinases by Human Gastric Carcinoma
Cell Lines1
Naohiko Koshikawa, Hidetaro Yasumitsu, Makoto Umeda, and Kaoru Miyazaki2
Division of Cell Biology, Kihara Institute for Biological Research, Yokohama City University, 2-120-3 Nakamura-cho, Minami-ku,
ABSTRACT
Proli'¡misespecies secreted by 10 human gastric carcinoma cell lines
were analyzed by gelatin zymography and immunoblotting. These cell
lines were classified into the following three groups with respect to
proteinase secretion: cell lines secreting mainly gelatinases A and/or B;
those secreting multiple types of serine proteinases; and those scarcely
secreting these enzymes. Two cell lines of the second group, STKM-1
and MKN28, hardly secreted metalloproteinases but secreted the
following four types of serine proteinases: (a) two trypsin-like enzymes
(A/, 26,000 and 24,000 in proenzyme forms); (b) a tissue kallikreinlike enzyme (M, 150,000 in a complex form); (c) a plasmin-like enzyme
(\l, 70,000); and (it) a plasminogen activator (urokinase-type, A/,
57,000, from STKM-1 and tissue-type, M, 70,000, from MKN28). The
A/, 70,000 plasmin-like enzyme was also detected at lower levels in the
conditioned media of four other cell lines (MKN1, MKN45, NUGC-3,
and KATO III). The \l, 24,000 proenzyme of the trypsin-like enzyme
was purified from the serum-free conditioned medium of STKM-1. The
proenzyme was activated by enterokinase treatment or autolytically by
incubation at neutral pH, decreasing its apparent molecular weight from
24,000 to 23,000 on nonreducing sodium dodecyl sulfate-polyacrylamide
gel electrophoresis. The activated enzyme extensively degraded fibronectin, laminin, and gelatins and to lesser extents type I, III, IV, and
V collagens at 30°C.These results suggest that the matrix serine pro
teinases may play a major role in the matrix degradation by some kinds
of human cancer cells.
plasminogen activators are often secreted from tumorous and
malignantly transformed cells, and they can convert blood plas
minogen to plasmin by limited proteolysis (7-9). It has been
reported that plasmin not only directly degrades extracellular
matrix proteins but also activates some metalloproteinase pre
cursors (10-12). Recently, Koivunen et al. (13-15) purified two
trypsinogen-like proteins from cyst fluid of an ovarian tumor
patient and named them TAT-1 (A/r 25,000, pi 5) and TAT-2
(Mr 28,000, pi 4). TAT-1 and TAT-2 are very similar to human
trypsinogens 1 and 2, respectively, in structure and enzymatic
properties, but there are differences in their isoelectric points
and substrate specificities.
Although various types of tumor cells have been studied con
cerning proteinase production, there have been few reports of
studies on proteinase production by gastric cancer, which is the
most common cancer in Costa Rica, Japan, Chile, and some
other countries. In the present study, we examined proteinase
secretion from 10 kinds of human gastric carcinoma cell lines
using the zymographic analysis and found that two of them
secreted high activities of four types of serine proteinases (two
trypsin-like enzymes, a plasmin-like enzyme, a kallikrein-like
enzyme, and a plasminogen activator). Some properties of a
trypsin-like enzyme are also reported.
MATERIALS
INTRODUCTION
The process of metastasis and invasion of tumor cells is
thought to require the proteolytic degradation of extracellular
matrix components. Recently, many studies have suggested
that a family of metalloproteinases, often called "matrix met
alloproteinases," plays an essential role in the matrix degrada
tion (reviewed in Refs. 1 and 2). The matrix metalloproteinases
include interstitial collagenase (EC 3.4.24.7), gelatinase A
(Mr 72,000 or 64,000 gelatinase/type IV collagenase; EC 3.4.
24.24), gelatinase B (A/r 92,000 or Mr 90,000 gelatinase/type
IV collagenase; EC 3.4.24.35), stromelysin (EC 3.4.24.17), and
matrilysin (matrin/pump-1; EC 3.4.24.23). We previously pu
rified and characterized human gelatinases A and B (3), human
matrilysin (4), rat stromelysin (5), and their inhibitor [tissue
inhibitor of metalloproteinase 2 (TIMP-2)] (6) from a condi
tioned medium of tumor cell lines. These enzymes are secreted
in latent proenzyme forms and hence must be activated for the
expression of their activities. The activated enzymes potently
degrade various extracellular matrix proteins such as collagens,
fibronectin, laminin, and proteoglycans.
In contrast to the metalloproteinases, only a few studies have
been reported about the role of serine proteinases in tumor
invasion/metastasis, except t-PA3 and u-PA, respectively. Both
Received 3/30/92; accepted 7/8/92.
The costs of publication of this article were defrayed in part by the payment of
page charges. This article must therefore be hereby marked advertisement in accord
ance with 18 U.S.C. Section 1734 solely to indicate this fact.
1This work was supported by a Grant-in-Aid from the Special Coordination
Funds of the Science and Technology Agency of Japan.
2 To whom requests for reprints should be addressed.
'The abbreviations used are: t-PA. tissue-type plasminogen activator; HPLC,
Yokohama 232, Japan
AND METHODS
Cells and Culture Conditions. Human gastric cancer cell lines tested
for the secretion of proteinases were MKN1 (adenosquamous carcino
ma), MKN28 (6TG-r; adenocarcinoma), MKN45 (adenocarcinoma),
MKN74 (adenocarcinoma), NUGC-2 (adenocarcinoma), NUGC-3 (ad
enocarcinoma), AZ-521 (unspecified). KATO III (signet ring cell car
cinoma), SCH (chorionic carcinoma), and STKM-1 (adenocarcinoma).
STKM-1, which had been established from the pleural effusion of a
41-year-old female patient with poorly differentiated gastric carcinoma
(16), was a kind gift from Dr. S. Yanoma, Kanagawa Cancer Center,
Japan. The other 9 cell lines were provided by the Japanese Cancer
Research Resources Bank. These cells were cultured at 37°Cin a hu
midified atmosphere of 5% CO2 and 95% air. RPMI 1640 supple
mented with 15 HIM A'-2-hydroxyethylpiperazine-Ar-ethanesulfonic
acid, 1.2 mg/ml NaHCO-,, 100 units/ml penicillin G, and 0.1 mg/ml
streptomycin sulfate was used as the basal medium. Cultures were
maintained in the basal medium supplemented with 10% fetal calf
serum (Hyclone, Logan, UT) (RPMI 1640 plus 10% fetal calf serum).
Plastic culture dishes were generous gifts from Sumibe Medical (Tokyo,
Japan), and plastic roller bottles were purchased from Becton Dickin
son (Oxnard, CA).
Preparation of Serum-free Conditioned Medium of 10 Kinds of
Gastric Carcinoma Cells. The human gastric carcinoma cell lines
were grown to confluence in 90-mm dishes containing 10 ml of RPMI
1640 plus 10% fetal calf serum. The cultures were rinsed three times
with Ca2+- and Mg2+-free Hanks' balanced salt solution and then re
placed with 10 ml of serum-free RPMI 1640. After each culture was
high performance liquid chromatography; PAGE, polyacrylamide gel electro
phoresis; SDS, sodium dodecyl sulfate; TAT. tumor-associated trypsinogen;
u-PA, urokinase-type plasminogen activator; PBS, phosphate-buffered saline;
TAME, A/-a-p-tosyl-L-arginine methylester hydrochloride; Cls, first component
of complement.
5046
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1992 American Association for Cancer Research.
MATRIX SERINE PROTEINASES
incubated for 2 days, the serum-free conditioned medium was clari
fied by sequential centrifugaron at 1,500 rpm for 15 min and at 19,000
rpm for 30 min. The supernatant was added with ammonium sulfate to
a final saturation of 80% and allowed to stand overnight at 4°C.The
resultant protein precipitate was collected by centrifugation at 19,000
rpm for 30 min, dissolved in and dialyzed against 10 IHMTris-HCl (pH
7.5) containing 0.01% (w/v) Briji-35, and used as concentrated condi
tioned medium.
Column Chromatographies. All procedures for proteinase purifica
tion, except reverse-phase HPLC, were performed at 4°C.Conditioned
medium was prepared from confluent serum-free culture of STKM-1
cells in roller bottles and concentrated as described above. The concen
trated conditioned medium was dialyzed against 20 mm Tris-HCl (pH
7.5) containing 0.5 M NaCl and 0.01% (w/v) Briji-35 and subjected to
molecular sieve chromatography on a Cellulofine GCL-2000m column
(2.6 x 98 cm) (Chisso, Tokyo, Japan) as reported previously (4). Rabbit
polyclonal antibody against human trypsin was chemically conjugated
to protein A-Sepharose by an ImmunoPure IgG Orientation kit (Pierce,
Rockford, IL). The proteinase sample was applied to the antibodyconjugated column (0.8 x 4.0 cm), previously equilibrated with 20 IHM
Tris-HCl (pH 7.5) containing 0.01% Briji-35. After the column was
washed with the buffer, the adsorbed proteinases were eluted with 10 ml
of 0.1 Mglycine (pH 2.5) at a flow rate of 8 ml/h. The eluted proteinase
fraction was dialyzed against 0.05% (v/v) trifluoroacetic acid. The dia
lyzed material was subjected to reverse-phase HPLC on a SynCropack
RP-4 column (0.41 x 25 cm) (SynChrom), preequilibrated with 0.05%
trifluoroacetic acid, and eluted with a linear gradient of 0-80% acetonitrile in 30 ml of 0.05% trifluoroacetic acid at a flow rate of 0.5
ml/min.
SDS-Polyacrylamide Gel Electrophoresis and Zymographic Analy
ses of Proteinases. Unless otherwise noted, SDS-PAGE and gelatin
zymography of proteinases were carried out on 12.5% polyacrylamide
slab gels (90 mm long, 90 mm wide, 0.75 mm thick) under nonreducing
conditions as described previously (3-6). For the zymography, protein
ases separated on the gels containing 1 mg/ml gelatin were renatured
and then incubated in 50 mM Tris-HCl (pH 7.5) with or without 10
HIMCaCl2 at 37°Cfor 18 h. The gels were stained with Coomassie
Brilliant Blue R-250. The molecular weight markers used are rabbit
muscle phosphorylase b (Mr 97,400), bovine serum albumin (Mr
66,200), hen egg albumin (M, 42,700), bovine carbonic anhydrase (M,
29,000), soybean trypsin inhibitor (A/r 21,500), and hen egg lysozyme
(Mr 14,000).
Zymography of plasminogen activator was performed on 1.25%
(w/v) agarose plates containing 2 mg/ml fibrin and 5 Mg/ml human
plasminogen as described before (17). Sample proteins were separated
by nonreducing SDS-PAGE, and the resultant SDS-polyacrylamide gel
was washed with 2.5% (v/v) Triton X-100, layered on the fibrin-agarose
plate, and incubated overnight at 37"C. The activity of plasminogen
1 HIMCaCl2 at room temperature for 20 min. The activated enzyme was
incubated in a reaction mixture containing 10 miviTAME and 46 HIM
Tris-HCl (pH 8.1) at 30°C,and the increase of absorbance at 247 nm
was measured with a spectrophotometer.
Determination of Protein Concentration. Protein concentration was
determined by the dye method with a Bio-Rad protein assay kit, using
bovine serum albumin as the standard.
Reagents. Bovine plasma fibronectin, mouse laminin, and pepsintreated bovine lens type IV collagen were purchased from Nitta Gelatin
(Tokyo, Japan); pepsin-treated bovine skin type I collagen and pepsintreated bovine placenta type III and V collagens from Koken (Tokyo,
Japan); and human plasma plasminogen and human kidney cell-derived
u-PA from Sigma (St. Louis, MO). Rabbit polyclonal antibodies
against human trypsin and human neutrophil elastase were purchased
from Athens Research and Technology (Athens, GA); goat polyclonal
antibody against human tissue kallikrein from Protogen AG (Switzer
land); rabbit polyclonal antibody against human u-PA from Japan
Chemical Research (Kobe, Japan); and goat polyclonal antibodies
against human melanoma t-PA and human plasminogen from Biopool
AB (Umeá, Sweden). Two monoclonal antibodies against Cls were
generous gifts from Dr. M. Matsumoto (The Center for Adult Diseases,
Osaka, Japan). Recombinant human t-PA was kindly provided by Phar
maceuticals Research Center, Toyobo Co., Ltd. (Shiga, Japan).
RESULTS
Secretion of Proteinases from Human Gastric Carcinoma
Cell Lines. Serum-free culture medium conditioned by 10
kinds of human gastric cancer cell lines was assayed for gelatinolytic activity by gelatin zymography in the presence and
absence of Ca2+. In the absence of Ca2+, two cell lines, MKN28
and STKM-1, showed strong and heterogeneous gelatinolytic
bands (Fig. \A). Three bands with molecular weights of 70,000,
26,000, and 24,000 were common to both cell lines. The Mr
26,000 activity was separated into three distinct bands when a
smaller amount of the sample was analyzed. Additional activi
ties were observed at M, 33,000 and 28,000 in MKN28 and at
Mr 150,000 and 45,000 in STKM-1. Such gelatinolytic activi
ties, especially those of Mr 150,000,45,000,33,000,
and 28,000
bands, varied considerably among different preparations of the
respective conditioned media, suggesting that some of them
might be proteolytic fragments of other proteinases. The MT
\ 50,000 activity was detected in different preparations of the
MKN28 conditioned medium. The Mr 70,000 activity was de
tected at lower levels in the conditioned media of MKN1,
MKN45, NUGC-3, and KATO III.
When the reaction mixture was supplemented with 10 IHM
Ca2+, two gelatinolytic activities at bands with molecular
activator was visualized as a transparent fibrinolysis band.
Immunoblotting Analysis. Sample proteins were separated by SDSPAGE. After the electrophoresis, the proteins separated on the gels
weights of 90,000 and 64,000 appeared in the conditioned me
were transferred onto nitrocellulose filters using a Bio-Rad Mini TransBlot apparatus (Richmond, CA), according to the method of Towbin et dia of four cell lines: the Afr 64,000 activity in SCH, the A/r
90,000 activity in NUGC-3, and both activities in MKN1 and
al. (18). The blotted filters were blocked with PBS containing 5% (w/v)
skim milk at 37°Cfor 2 h, washed with a mixture of PBS and 0.05%
MKN74 (Fig. IB). When the effects of various proteinase in
hibitors on these activities were examined, 1 HIM1,10-phenanTween 20 (Tween-PBS), and then incubated overnight at room temper
ature with the first antibody, which had been diluted 1000-fold with
throline, but not inhibitors for serine, cysteine, and aspartic
PBS. After a washing with Tween-PBS, the filters were incubated with
proteinases, completely inhibited the activities of Mr 64,000
a 1000-fold diluted biotinylated anti-rabbit or anti-goat IgG antibody
and 90,000 bands, showing that both gelatinolytic enzymes
(Vector Laboratories), washed with Tween-PBS, and then incubated
were metalloproteinases (data not shown). The electrophoretic
with avidin-alkaline phosphatase (Vector) at room temperature for 1 h.
mobilities of the Mr 64,000 and M, 90,000 enzymes in zymog
The filters were washed with Tween-PBS and then incubated at room
raphy were identical to those of gelatinase A (M, 72,000 gelatitemperature for about 5 min in a reaction mixture containing 5-bromonase) and gelatinase B (Mr 92,000 gelatinase), respectively,
4-chloro-3-indolylphosphate and nitro blue tetra/oliimi to develop col
which we had previously isolated from the conditioned medium
ored product on the filters.
Activity Assay of Trypsin-like Enzyme. The purified proenzyme
of human schwanoma cell line (3) (data not shown). These
form of trypsin-like enzyme was activated by incubating with bovine
results indicated that the A/r 64,000 and MT90,000 gelatinolytic
enterokinase (1:1, w/w) in 28 mw succinate buffer (pH 5.6) containing
activities corresponded to gelatinases A and B, respectively.
5047
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1992 American Association for Cancer Research.
MATRIX SERINE PROTEINASES
positions as on the gelatin-containing gel (data not shown),
indicating that the serine proteinases with molecular weights of
150,000, 70,000, 26,000, and 24,000 could digest both gelatin
and casein.
In order to identify these serine proteinases, the concentrated
conditioned medium of STKM-1 was analyzed by immunoblotting analyses with antibodies against some serine proteinases.
Polyclonal antibody against human trypsin showed two immunochemically stained bands with Mr 26,000 and 24,000 under
nonreducing conditions, suggesting that the Mr 26,000 and
24,000 gelatinolytic enzymes were trypsin-like enzymes (see
Fig. 6). These enzymes were further characterized as described
later.
When human plasmin and the STKM-1-conditioned medium
were run on the same gelatin-containing gel under nonreducing
conditions, plasmin showed a major gelatinolytic band with the
same mobility as the Mr 70,000 activity of the conditioned
medium (Fig. 3). Plasmin showed two minor additional bands
with molecular weights of 45,000 and 33,000. These results
suggested that the A/r 70,000 proteinase secreted from STKM-1
and other carcinoma lines might be plasmin and that the minor
activities of A/r 45,000, 33,000, and 28,000 bands in the con
ditioned media of STKM-1 and/or MKN28 were due to plas
min fragments. Immunoblotting
analyses with antibodies
against human plasminogen and human tissue kallikrein
showed multiple immunostained bands and were unable to
identify specific bands (data not shown).
The activity of the A/r 70,000 proteinase secreted from
STKM-1 cells did not significantly change during at least 2
weeks in serum-free culture with 4 medium changes (data not
shown). This ruled out the possibility that the enzyme was a
contaminant from fetal calf serum which had been used in the
initial cell plating. The activity of plasminogen, which showed
B
-29
7C
Z
*
Z
*
Z
_L
M
*.
oo
cn
C
C
o
o
o
IV)
o
I
co
co
>
N
co
o
I
=
T
Fig. 1. Zymographic analysis of gelatinolytic enzymes secreted from 10 kinds
of gastric carcinoma cells. The serum-free conditioned media of the 10 carcinoma
cell lines were concentrated 30-fold by ammonium sulfate precipitation, and
aliquots (10 M!)of the concentrated conditioned media were subjected to zymography on gelatin-containing gels. After electrophoresis, proteins on the gels were
renatured and then incubated at 37°Cfor 18 h in the reaction mixture without
(A) or with (B) 10 mM CaCl2. Left to right, MKN1, MKN28, MKN45, MKN74,
NUGC-2, NUGC-3, AZ-52I, KATO III, STKM-1, and SCH. Ordinate, molec
ular weight in thousands. Arrows, positions of gelatinolytic enzymes with molec
ular weights of 150,000, 70,000, 26,000, and 24,000 in A, and gelatinases A
(M, 64,000) and B (M, 90,000) in B. Other experimental conditions are given in
the text.
Partial Characterization of Ca2+-independent Proteinases.
To classify the Ca2+-independent proteinases, the effects of
various proteinase inhibitors on their gelatinolytic activity were
tested by gelatin zymography of the STKM-1-conditioned me
dium. Diisopropylfluorophosphate
effectively inhibited all ge
latinolytic activities including those of Mr 150,000, 70,000,
26,000, and 24,000 (Fig. 2). Leupeptin, but not chymostatin,
inhibited these activities, indicating that they had trypsin-like
sensitivity to the proteinase inhibitors (data not shown). These
activities were not affected by 1,10-phenanthroline
(a metalloproteinase inhibitor), p-aminophenyl mercuric acetate (a
cysteine proteinase inhibitor), and pepstatin (an aspartic pro
teinase inhibitor) (data not shown). When the STKM-1-condi
tioned medium was subjected to zymography on a casein-con
taining gel, proteolytic bands were observed at the same
24
DFP
Fig. 2. Effect of diisopropylfluorophosphate (DFP) on gelatinolytic activities
secreted by STKM-1 cells. The concentrated conditioned medium of STKM-1
was electrophoresed on a gelatin-containing gel, and the resultant gel was rena
tured and cut into strips, followed by incubation in the Ca2+-free reaction mixture
without (left) or with (right) 5 mM diisopropylfluorophosphate. Arrows, gelati
nolytic activities at M, 150,000, 70,000, 45,000, 26,000, and 24,000. Other ex
perimental conditions are the same as given in Fig. 1.
5048
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1992 American Association for Cancer Research.
MATRIX SERINE PROTEINASES
^26
Fig. 3. Gelatin zymography of STKM-1-conditioned medium and u-PA-activated human plasmin. Lane I, plasmin (approximately 0.25 i¡%/Lane)
prepared by
incubating human plasminogen with human u-PA (10:1, w/w) at room tempera
ture for 60 min; Lane 2, 100-fold concentrated conditioned medium of STKM-1.
Arrows, gelatinolytic activities at M, 150,000, 70,000. 45,000, 33,000, and
26,000; arrowhead, plasmin.
'
57*-
hardly showed gelatinolytic activity on zymography, immunoblotting analyses with anti-u-PA and anti-t-PA polyclonal an
tibodies and zymography on a fibrin-agarose plate were carried
out with the STKM-1-conditioned medium. The anti-u-PA an
tibody showed a major immunostained A/r 57,000 band at the
same position as control two-chain u-PA (Fig. 4, Lane /). The
fibrin zymography in the presence of plasminogen showed a
major fibrinolytic band at Mr 57,000 (Fig. 4, Lane 5). The
activity was not detected when the same volume of the condi
tioned medium was analyzed on the fibrin-agarose plate with
out plasminogen. These results revealed that STKM-1 cells
secreted u-PA. Any specific band was not detected by immunoblotting with the anti-t-PA antibody (Fig. 4, Lane 3). In con
trast to the STKM-1-conditioned medium, the conditioned me
dium of MKN28 cells exhibited an immunostained band at A/r
70,000 on immunoblotting analysis with the anti-t-PA antibody
but no band resulting from the analysis with the anti-u-PA
antibody (data not shown).
Purification of Trypsin-like Proteinases. STKM-1 cells were
cultured in serum-free RPMI 1640 medium. The resultant con
ditioned medium (2.6 liters) was concentrated by ammonium
sulfate precipitation and subjected to molecular sieve chroma
tography on a Cellulofine GCL-2000m column (Fig. 5). The
fractions of A/r 26,000 and 24,000 proteinases (fractions 5870), which also contained a part of the MT 70,000 and 150,000
activities, were applied to an immunoaffinity column conju
gated with anti-trypsin polyclonal antibody. Most of proteins
passed through the column, whereas the majority of the serine
proteinase activities of Mr 24,000, 26,000, 70,000, and 150,000
bands were bound to the column and eluted with 0.1 M glycine
(pH 2.5). When analyzed by SDS-PAGE under nonreducing
conditions, the eluted fraction showed a major band at A/,
24,000 and a minor band at A/r 26,000 (Fig. 6, Lane /). Gelatin
zymography of the eluted fraction showed two strong bands
with molecular weights of 26,000 and 24,000 corresponding to
-29
(1)
(2)
(3)
(4)
(5)
Fig. 4. Immunoblotting analysis and fibrin zymography of plasminogen acti
vator secreted by STKM-1 cells. The conditioned medium of STKM-1 was con
centrated 100-fold by ammonium sulfate precipitation, and subjected to inumi
noblotting analyses with anti-u-PA antibody (Lane I) and with anti-t-PA antibody
(I.am13) and zymography on a fibrin-agarose plate containing plasminogen (Lane
5), as described in the text. Lane 2, immunoblot of human kidney cell-derived
u-PA (two-chain form) with anti-u-PA antibody; Lane 4. immunoblot of recom
binant human t-PA with anti-t-PA antibody. Ordinate, molecular weight in thou
sands; arrowhead, an immunostained band at M, 57.000. In Lane 2, the control
recombinant u-PA showed several minor bands in addition to the M, 57.000 band.
These proteins might be produced by differences in proteolytic processing or
glycosylation. The M, 33,000 and 31,000 bands are very likely the low molecular
weight u-PA.
a gelatinolytic band with a molecular weight of 78,000 on zy
mography (data not shown), was hardly detected with the
STKM-1-conditioned medium.
We also attempted to detect plasminogen activators in the
conditioned medium of STKM-1 cells. Since u-PA and t-PA
50
Fraction number
65
80
5.5 ml each)
Fig. 5. Molecular sieve chromatography of STKM-1 conditioned medium on
Cellulofine GCL-2000m column. The serum-free conditioned medium (2.6 liters)
was concentrated to 30 ml and divided into two equal portions. Each portion,
which contained about 150 mg of protein, was applied to the column, and the
eluted fractions were measured for absorbance at 280 nm (Aim). Arrows, elution
positions of ferritin (M, 450,000), albumin (M, 66,000), and cytochrome c (M,
12,500). Fractions 58-70 (bar) from the two runs of the chromatography were
pooled and used for further purification. Inset, profiles of SDS-PAGE (Lane I)
and gelatin zymography (Lane 2) of the pooled fraction. Other experimental
conditions are given in the text.
5049
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1992 American Association for Cancer Research.
MATRIX SERINE PROTEINASES
Substrate specificity of the autolytically activated Mr 24,000
enzyme was examined against two cell-adhesive glycoproteins
and four types of pepsin-treated collagens (Fig. 8). The enzyme
potently degraded fibronectin and laminin. while it partially
degraded the four collagens. Type IV and V collagens were
more susceptible to the enzyme than type I and III collagens.
Immunochemical Identification of Tissue Kallikrein-like En
zyme and Plasmin-like Enzyme. As shown in Fig. 6, Lane 2,
the proteinase fraction eluted from the anti-trypsin antibody
-97
-66
-43
-97
-66
-29
i
-43
l
-20
-29
(1)
(2)
(3)
-21
Fig. 6. SDS-PAGE (Lane /), gelatin zymography (Lane 2), and immunoblotting with anti-trypsin antibody (Lane 3) of proteinase fraction obtained from the
anti-trypsin antibody column. Ordinate, molecular weight in thousands. Other
experimental conditions are given in the text.
the protein bands (Fig. 6, Lane 2). Together with the SDSPAGE result, this analysis indicated that the A/r 26,000 enzyme
had a higher specific activity than the A/r 24,000 enzyme. The
relative increase of the A/r 24,000 activity compared to the A/r
26,000 activity seemed due to the partial leakage of the latter
activity from the affinity column. Both proteins were strongly
stained by immunoblotting with anti-trypsin antibody (Fig. 6,
Lane 3).
The gelatin zymography of the proteinase fraction showed
additional gelatinolytic bands such as two bands at A/, 66,000
and 70,000 and a broad band on the top of the gel (Mr >
150,000) (Fig. 6, Lane 2), although their corresponding pro
teins were not detected by protein staining (Fig. 6, Lane 1).
These gelatinolytic activities were further characterized by im
munoblotting analysis as described later (see Fig. 9).
The proteinase fraction from the immunoaffinity column was
further subjected to reverse-phase HPLC on a C4 column. This
chromatography separated the Mr 24,000 protein from the MT
26,000 protein and other proteinases. The purified protein
hardly hydrolyzed TAME, the synthetic substrate of trypsin.
When treated with enterokinase, the natural activator of trypsinogen, it effectively hydrolyzed TAME; the approximate spe
cific activities (units/mg protein) before and after the treatment
were 12 and 110, respectively. This indicated that the Mr
24,000 protein was a proenzyme (zymogen). The enterokinaseactivated enzyme showed the same mobility as trypsin on SDSPAGE; its estimated molecular weight was 23,000 under nonreducing conditions (A/r 27,000 under reducing conditions)
(Fig. 7). The Mr 24,000 proenzyme was autolytically activated
when incubated in a pH 7.5 solution at 37°Cfor more than 1 h.
These results showed that the Mr 24,000 protein was identical
or very similar to trypsinogen. Analysis by isoelectric focusing
on a thin layer polyacrylamide gel showed that the pi of the A/,24,000 protein was 6.6 (data not shown).
(1)
(2)
(3)
Fig. 7. Effect of enterokinase treatment on molecular size of purified M,
24,000 enzyme. The purified protein was incubated with an equal amount (w/w)
of enterokinase in a reaction mixture containing 28 HIMsodium succinate (pH 5.6)
and 1 mm CaCl2 at 37'C for 10 min and then subjected to SDS-PAGE on a 15%
gel under reducing conditions. Lane 1, the purified protein alone; Lane 2, the
purified protein treated with enterokinase; Lane 3, human pancreatic trypsin.
Ordinate, molecular weight in thousands. The protein band at A/r 130,000 in Lane
2 is enterokinase.
FN
Collagen
LN
•¿
m
IV
IM*
<«•*.
-200
i.,. i+&~ -116
-97
—¿â€”¿-66
-43
—¿
+
—¿
+
—¿
+
—¿
+
—¿
+
—¿
+ ENZYME
Fig. 8. Substrate specificity of purified M, 24,000 enzyme. Each substrate
protein was incubated with (+) or without (-) 1.0 ng/ml of the autoactivated
M, 24,000 enzyme in 50 p\ of a reaction mixture containing 20 min Tris-HCl
(pH 7.5), 0.01% Briji-35, and a substrate protein at 30°Cfor 18 h. After the
incubation, the reaction mixture was mixed with the SDS-sample buffer contain
ing 2-mercaptoethanol, heated in boiling water for 3 min, and then subjected to
SDS-PAGE on 7.5% polyacrylamide gels. FN, 0.2 mg/ml bovine plasma fibronec
tin; LN, 0.2 mg/ml mouse laminin; /, 0.5 mg/ml pepsin-treated bovine skin type
I collagen; ///, 0.5 mg/ml pepsin-treated bovine placenta type III collagen; IV, 0.5
mg/ml pepsin-treated bovine lens type IV collagen; V, 0.5 mg/ml pepsin-treated
bovine placenta type V collagen. Ordinate, molecular weight in thousands; arrow
heads, proteolytic fragments of collagens.
5050
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1992 American Association for Cancer Research.
MATRIX SERINE PROTEINASES
column contained gelatinolytic activities with Mr 66,000,
70,000, and > 150,000 in addition to the trypsinogen-like pro
teins with M, 24,000 and 26,000. It appeared that during the
affinity chromatography the A/r 150,000 enzyme present in the
pooled fraction from the molecular sieve chromatography was
aggregated to show a broad band on the top of the SDS-gel,
while the plasmin-like enzyme of M, 70,000 partly converted to
a lower molecular form of M, 66,000 by limited proteolysis. To
identify these gelatinolytic enzymes, their reactivity with polyclonal antibodies against tissue- and plasma-types kallikreins,
plasminogen, elastase, and Cls was examined by immunoblotting analysis.
The anti-plasminogen
antibody exhibited strong doublet
bands at Mr 60,000 and 56,000 and some minor bands under
reducing conditions (Fig. 9, Lane 1). This profile is very similar
to that of u-PA-activated human plasmin (Fig. 9, Lane 2).
These immunostained bands appeared to correspond to the
large subunits of Glu- and Lys-plasmins (Mr 60,000 and
56,000), their proteolytic fragments (Mr 45,000 and 43,000),
their small subunit (A/r 23,000, and intact plasminogen (Mr
83,000).
The antibody against tissue-type kallikrein showed a strongly
stained, broad band on the top of the immunoblot under nonreducing conditions (Fig. 9, Lane 3). When the sample was
reduced by 2-mercaptoethanol,
the immunoblot showed a
sharp, single band with A/r 50,000 (Fig. 9, Lane ¥),which was
consistent with the reported molecular weight value of tissue-
type kallikrein (19). These results indicated that the A/r 150,000
gelatinolytic activity found in the STKM-1 conditioned me
dium was identical or closely related to tissue-type kallikrein
and that the A/r 150,000 and higher molecular weight forms
might be produced by self-aggregation or binding with a carrier
protein.
The antibodies against human elastase, human plasma kal
likrein, and human Cls did not show immunostained band with
the serine proteinase fraction used in the above experiments
(data not shown). The two trypsinogen-like proteins, the plas
min-like protein and the tissue-kallikrein-like protein, were also
detected immunochemically in the proteinase fraction obtained
from the conditioned medium of MKN28 cells by the antitrypsin antibody column chromatography (data not shown).
DISCUSSION
In the present study we revealed proteinase species secreted
by 10 human gastric carcinoma cell lines using gelatin zymography and immunoblotting analysis. These cell lines were di
vided into the following three groups with respect to proteinase
secretion: (a) cell lines secreting gelatinases A and/or B as ma
jor components; (b) those secreting multiple serine proteinases;
and (c) those scarcely secreting these enzymes. The two cell
lines STKM-1 and MKN28, which belonged to the second
group, secreted at least four types of serine proteinases: (a) two
trypsin-like enzymes (Mr 26,000 and 24,000 in proenzyme
forms); (b) a tissue kallikrein-like enzyme; (c) a plasmin-like
enzyme; and (d) a plasminogen activator. Such cancer cell lines
Antihave not been reported before.
Trypsinogen is normally synthesized in the pancreas, and its
plasminogen AnthkalHkrein
activated form trypsin nonspecifically digests various proteins.
Koivunen et al. (15) have immunologically detected TAT-1
and/or TAT-2 in the conditioned media of three fibrosarcoma
cell lines, a colon carcinoma cell line, and a erythroleukemia
cell line. The trypsinogen-like proteins with molecular weights
of 24,000 and 26,000 secreted by STKM-1 cells are likely to
correspond to TAT-1 and TAT-2, respectively. Although
TAT-1 and TAT-2 have been believed to be distinct from trypsinogen 1 and trypsinogen 2, respectively, the possibility seems to
remain that they are the modified forms of trypsinogen 1 and 2.
Further structural analysis is required to prove this possibility.
The
present study, together with the studies of Koivunen et a/.,
-43
demonstrates that trypsin-like enzymes are ectopically synthe
sized and secreted by various kinds of tumor cells.
Kallikreins are classified into plasma kallikrein and tissue
kallikrein (or glandular kallikrein), the latter of which is rich in
-29
exocrine glands such as the salivary gland and pancreas. Tissue
kallikrein is a member of a closely related gene family. This
family consists of tissue kallikrein and at least three structurally
related proteins in humans. Besides the exocrine tissues, tissue
kallikrein or its related proteins or their mRNA have been
found in the kidney, arterial walls, prostate, brain, colon,
spleen, heart, plasma, and urine (reviewed in Ref. 20). However,
production and secretion of kallikrein by cancer cells have
(1)
(2)
(3)
(4)
rarely been reported except for pancreatic carcinoma cells
Fig. 9. Immunochemical identification of plasmin-like and tissue kallikrein(21). Kallikrein produces the vasodilator kinin from kininogen
like proteins secreted by STKM-I cells. The fraction of serine proteinases ob
tained from the anti-trypsin antibody column (see Fig. 6) was subjected to iininii
by limited proteolysis. Kinin is involved in various pathophysnoblotting analyses with anti-plasminogen antibody under reducing conditions
iological processes such as hypertension, inflammation, and
(Lane I) and with anti-tissue kallikrein antibody under nonreducing (Lane 3) or
reducing (Lane 4) conditions. Lane 2, immunoblot of u-PA-activated human
allergy. Although kallikrein is known to hydrolyze other bio
plasmin as control (see the legend to Fig. 3) under the same conditions as Lane I.
logically important peptides such as pro-insulin, atrial natriOrdinate, molecular weight in thousands. Arrowheads, immunostained bands in
uretic factor, and vasoactive intestinal peptides (20), its direct
Lane 1 (from top to bottom, M, 83.000, 60,000, 56,000, 45,000, 43,000, and
23,000). Other experimental conditions are given in the text.
effect on extracellular matrix proteins has not been reported.
5051
r
4
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1992 American Association for Cancer Research.
MATRIX SERINE PROTEINASES
We have obtained the data that porcine urinary tissue kallikrein
(Sigma) potently hydrolyzes laminin and fibronectin and to
lesser extents gelatin and type I, III, and IV collagens (data not
shown). In the present study, the gelatinolytic activities with A/r
150,000 or higher present in the STKM-1 conditioned medium
and its Chromatographie fractions were attributed to tissue kal
likrein or its related proteinase. Such high molecular weight
forms of tissue kallikrein have not been reported before (19,
20). However, human urinary kallikrein (A/r 42,000), which is
smaller than pancreatic tissue kallikrein (Mr 55,000) because of
the difference in the sugar moiety (19), is known to form a
SDS-stable complex at MT 92,000 with a specific binding pro
tein (serpin) at Mr 60,000 (22, 23). It seems likely that the
tumor-derived tissue kallikrein forms a larger complex with an
unidentified binding protein. In addition, there is another pos
sibility that the Mr 150,000 form and other high molecular
weight forms of tissue kallikrein are oligomeric or polymeric
forms.
STKM-1 and MKN28 cells also secreted a high gelatinolytic
activity with Mr 70,000 into culture media. Immunoblotting
analysis suggested that the enzyme might be plasmin, although
the possibility cannot be ruled out that it is a different trypsintype proteinase. We could not detect plasminogen in their con
ditioned media by gelatin zymography, although a plasminogen-like protein was slightly observed by immunoblotting
analysis. It seems likely that plasminogen had been mostly con
verted to plasmin by an endogenous plasminogen activator, the
latter of which was also secreted from these cells. Plasmin has
been reported to degrade extracellular glycoproteins and base
ment membrane type IV collagen (10). There are many studies
suggesting an important role of the PA/plasmin system in tu
mor invasion and metastasis (10, 12, 24). To our knowledge,
however, the synthesis and secretion of plasminogen by nonhepatic tumors have not been reported before. It should be noted
that a similar plasmili-1 ike activity was detected in 5 of the 10
gastric carcinoma cell lines examined. Sakiyama et al. (25, 26)
have isolated an extracellular matrix-degrading A/r 88,000
serine proteinase from the conditioned medium of malignant
hamster fibroblasts and identified it as C Is. The A/r 70,000
proteinase secreted by STKM-1 cells was clearly distinguished
from human Cls by electrophoretic mobility on zymography,
and the former hardly reacted with monoclonal antibodies
against human Cls on immunoblotting analysis (data not
shown).
The three enzymes trypsin, tissue-type kallikrein, and plas
min can proteolytically activate each other's proenzymes, and
the activated enzymes directly degrade extracellular matrix pro
teins. Moreover, they can activate the proenzymes of some
matrix metalloproteinases,
which also hydrolyze the matrix
proteins (5, 11, 12). The two carcinoma cell lines STKM-1 and
MKN28 secreting these enzymes or their related enzymes are
expected to exhibit a high matrix-degrading activity in vivo.
Indeed, we have preliminarily observed that STKM-1 cells ex
hibit invasive growth in the peritoneal cavity of SCID mice and
in vitro invasion through the reconstituted basement membrane
(Matrigel) in Boyden chambers. Under culture conditions, both
cell lines showed very poor cell-to-cell and cell-to-substratum
interactions compared to other cell lines, except KATO III.
KATO III cells grew in suspension in spite of the poor secretion
of proteinases. Further studies are needed to clarify the rela
tionship between the matrix-invasive potential of the gastric
carcinoma cell lines and the proteinase species secreted by
them.
In preliminary studies we have detected the A/r 24,000 and
26,000 trypsinogen-like activities and the M, 70,000 plasmin
like activity in the conditioned media of many human cancer
cell lines other than gastric carcinomas but to lesser extents
than STKM-1 and MKN28. We have not been able to show the
secretion of the tissue kallikrein-like enzyme in cancer cell lines
other than the two gastric carcinoma cell lines on gelatin zy
mography. Immunochemical analysis with ami kallikrein anti
bodies having a high titer may reveal its wider distribution.
As summary, the present study, together with the preliminary
results, suggests the possibilities that in some kinds of tumor
cells these serine proteinases, which may be called "matrix
serine proteinases," play a major role in the matrix degradation
and subsequent tumor cell invasion, and that in some other
tumor cells they act as the activator of the latent proenzymes of
the matrix metalloproteinases.
ACKNOWLEDGMENTS
We thank A. Nakamura and N. Tsuchiya for technical support. We
are grateful to Dr. H. Sakiyama, National Institute of Radiological
Sciences, Japan, for the generous gift of purified human Cls and help
ful discussions.
REFERENCES
1 Woessner, F. .1.. Jr. Matrix metalloproteinases and their inhibitors of con
nective tissue remodeling. FASEB J., 5: 2145-2154, 1991.
2. Matrisian, M. L. Metalloproteinases and their inhibitors in matrix remodel
ing. Trends Genet., 6: 121-125, 1990.
3. Yasumitsu, H., Miyazaki, K., Umenishi, F., Koshikawa, N., and Umeda, M.
Comparison of extracellular matrix-degrading activities between 64-kDa and
90-kDa gelatinases purified in inhibitor-free forms from human schwannoma
cells. J. Biochem. (Tokyo), ///: 74-80, 1992.
4. Miyazaki, K., Hattori, Y., Umenishi, F., Yasumitsu, H., and Umeda, M.
Purification and characterization of extracellular matrix-degrading metalloproteinase, matrin (pump-1), secreted from human rectal carcinoma cell line.
Cancer Res., 50: 7758-7764, 1990.
5. Umenishi, F., Yasumitsu, H., Ashida, Y., Yamashita, J., Umeda, M., and
Miyazaki, K. Purification and properties of extracellular matrix-degrading
metallo-proteinase overproduced by Rous sarcoma virus-transformed rat
liver cell line, and its identification as transin. J. Biochem. (Tokyo), 108:
537-543, 1990.
6. Umenishi, F., Umeda, M., and Miyazaki, K. Efficient purification of TIMP-2
from culture medium conditioned by human hepatoma cell line, and its
inhibitory effect on metalloproteinases and in vitro tumor invasion. J. Bio
chem. (Tokyo), 110: 189-195, 1991.
7. Quax, P. H. A., van Leeuwen, R. T. J., Verspaget, H. W., and Verheijen, J.
H. Protein and messenger RNA levels of plasminogen activators and inhib
itors analyzed in 22 human tumor cell lines. Cancer Res., 50: 1488-1494,
1990.
8. Quigley. J. P., Berkenpas, M. B., Aimes, R. T., and Chen, J. M. Serine
protease and metallo protease cascade systems involved in pericellular proteolysis. Cell Differ. Dev., 32: 263-276, 1990.
9. Saksela, O. Plasminogen activation and regulation of pericellular proteolysis.
Biochim. Biophys. Acta, 823: 35-65, 1985.
10. Mackay, A. R., Corbitt, R. H., Hartzler, J. L., and Thorgeirsson, U. P.
Basement membrane type IV collagen degradation: evidence for the involve
ment of a proteolytic cascade independent of metalloproteinases. Cancer
Res., 50:5997-6001, 1990.
11. Suzuki, K., Enghild, J. J., Morodomi, T., Salvesen, G., and Nagase, H.
Mechanisms of activation of tissue procollagenase by matrix metalloproteinase 3 (stromelysin). Biochemistry, 29: 10261-10270, 1990.
12. He, C, Wilhelm, S. M., Pentland, A. P., Marmer, B. L., Grant, G. A., Eisen,
A. Z., and Goldberg, G. I. Tissue cooperation in a proteolytic cascade acti
vating human interstitial collagenase. Proc. Nati. Acad. Sci. USA, 86:26322636, 1989.
13. Koivunen, E., Huhtala, M-I... Sten num. and U-H. Human ovarian tumorassociated trypsin. J. Biol. Chem., 264: 14095-14099, 1989.
14. Koivunen, E., Ristimäki,A., Osman, S., Vuento, M., and Stenman, U-H.
Tumor-associated trypsin participates in cancer cell-mediated degradation of
extracellular matrix. Cancer Res., 51: 2107-2112, 1991.
15. Koivunen, E., Saksela, O., Itkonen, O., Osman, S., Huhtala, M-L., and
Stenman, U-H. Human colon carcinoma, fibrosarcoma and leukemia cell
lines produce tumor-associated trypsinogen. Int. J. Cancer (Phila.), 47:592596, 1991.
16. Arim um. A., Nakamura, Y., Shimizu, A., Harada, M., and Yanoma, S.
Establishment and characterization of a CAI9-9 producing human gastric
5052
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1992 American Association for Cancer Research.
MATRIX SERINE PROTEINASES
cancer cell line, STKM-1. Hum. Cell, 4: 67-70, 1991.
17. Miyazaki, K., Ashida, Y., Kihira,Y., Mashima, K., Yamashita, J.,and Horio,
T. Transformation of rat liver cell line by Rous sarcoma virus causes loss of
cell surface fibronectin, accompanied with secretion of metallo-proteinase
that preferentially digests the fibronectin. J. Biochem. (Tokyo), 102: 569582, 1987.
18. Towbin, H., Staehelin, T., and Gordon, J. Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets. Proc. Nati. Acad. Sci.
USA, 76: 4350-4354, 1979
..,.,.,,
.,
,
..
19. Kageyama, H.. Okumura, Y., Adach,, T., Ito, Y., Hirano. K.. and Sug.ura, M.
Purification and properties of kallikreins from human urine and pancreas.
Chem. Pharm. Bull. (Tokyo), 31: 2386-2394, 1983.
20. McDonald, R. J., Margolius, H. S., and Erdos, E. G. Molecular biology of
tissue kallikrein. Biochem. J., 253: 313-321, 1988.
21. Berg, T., Johansen, L., Bergundhaugen, H., Hansen, L. J., Reddy, J. K., and
Poulsen, K. Demonstration of kallikrein in rat pancreatic acinar cell cardnoma. Cancer Res.,45: 226-234, 1985.
22. Chen, L-M., Mayfield, R. K., and Chao, J. Differential interaction of human
kallikrein-binding protein and a,-anlitrypsin with human tissue kallikrein.
Biochem. J., 267: 79-84, 1990.
23. Chao, J., Chai, K. X., Chen, L-M.. Xiong, W., Chao, S., Woodley-Miller, C.,
Wang, L., Lu, H. S., and Chao, L. Tissue kallikrein-binding protein is a
serpin. J. Biol. Chem., 265: 16394-16401, 1990.
24. Reich, R., Thompson, E. W., Iwamoto, Y., Martin, G. R., Deason, J. R.,
Fu|ler G c and Miskin- R EfTec,s of inhibitors of plasminogen activator,
ser¡neproteinases> and col|age„aseIV on the invasion of basement membranes by metastatic cells. Cancer Res., ¿«.-3307-3312,1988.
Sakiyama, H., Nishino, T., Tanaka, T., Tomosawa, T., Kinoshita, H., Na'
'
'
jcic V. T
,
-r
»"a, K" Chlba- K" and Saklvama' s- Purification and characterization of
novel calcium-dependent senne protemase secreted from malignant hamster
embryo fibroblast Nil2C2. Biochim. Biophys. Acta, 990: 156-161, 1989.
26. Kinoshita, H., Sakiyama, H., Tokunaga, K., Imojoh-Ohmi, S., Hamada, Y.,
Isono, K., and Sakiyama, S. Complete primary structure of a calcium-dependent serine proteinase capable of degrading extracellular matrix protein.
FEBS Lett., 250: 411-415, 1989.
5053
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1992 American Association for Cancer Research.
Multiple Secretion of Matrix Serine Proteinases by Human
Gastric Carcinoma Cell Lines
Naohiko Koshikawa, Hidetaro Yasumitsu, Makoto Umeda, et al.
Cancer Res 1992;52:5046-5053.
Updated version
E-mail alerts
Reprints and
Subscriptions
Permissions
Access the most recent version of this article at:
http://cancerres.aacrjournals.org/content/52/18/5046
Sign up to receive free email-alerts related to this article or journal.
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Department at [email protected].
To request permission to re-use all or part of this article, contact the AACR Publications
Department at [email protected].
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1992 American Association for Cancer Research.