EXPERIMENTAL
CELL
RESEARCH
192,
574-580
(1991)
Tumor Necrosis Factor and Epidermal Growth Factor Modulate Migration of Human Microvascular Endothelial Cells and Production of Tissue-Type Plasminogen Activator and Its Inhibitor MASASUMI
MAWATARI,*
KAZUKI OKAMURA,* TAKAO MATSUDA,* RYOJI HAMANAKA,* HIROMOTO KANJI HIGASHIO,? KIMITOSHI KOHNO,* AND MICHIHIKO KUWANO*,’
*Department
of Riochcmistry,
Oita Medical School, Oita X79-.56, Japan; and tfficwarch Institute, Snow Brand Milk Products Company. 7’ochigi 329-0.5. Japan
INTRODUCTION
The production of proteases and cell migration are essential during angiogenesis. Proteases, such as plas-
0014.4827191
Copyright All
rights
$3.00 (c 1991 by of reproduction
and reprint
requests
should
’ Ahhreviations used: PA, plasminogen activator; t-PA, tissue-type plasminogen activator; u-PA, urokinase-type plasminogen activitor; PAI-1, plasminogen activator inhibitor-l, EGF, epidermal growth factor; TNF, tumor necrosis f’actor; hFGF, basic fibroblast growth factor; HME cell, human microvascular endothelial cell.
be ad-
574
Academic Press, Inc. in any
form
reserved.
of Life Sciwzw,
minogen activator (PA)’ and collagenase, are produced from endothelial cells when angiogenic factors are present [l-3]. In addition, PA can convert the latent collagenase to active collagenase through converting plasminogen to plasmin. Plasmin itself is also active in de[4]. The grading extracellular matrix components production of PA may be regulated in concert with cell migration [S-7], since basic fibroblast growth factor (bFGF) stimulates cell migration of bovine aortic endothelial cells and causes a concomitant increase of tissuetype plasminogen activator (t-PA) induction [S]. An inhibitor of PA, plasminogen activator inhibitor-l (PAIl), is also synthesized and secreted from endothelial cells [9, 101 and hepatocytes [ll]. PAI- rapidly binds with PA, and thus can modulate PA activity. Tumor necrosis factor (TNF), which is secreted from activated macrophages and monocytes, causes hemorrhagic necrosis in mice with transplantable tumors [ 121. TNF shows cytotoxic or cytostatic effects and induces morphologic changes in endothelial cells [ 131. We have previously reported that epidermal growth factor (EGF) induces the vessel-like structures of human microvascular endothelial (HME) cells in a type I collagen gel, but TNF abrogates this induction of vessel-like structures [ 141. Coadministration of EGF and TNF synergistically increases the production of tissue inhibitor of metalloproteinases and interleukin-6 [ 141. The mechanisms for the effects of EGF and TNF on endothelial cells are not understood but could be related to cell migration and the production of PA. EGF has been shown to promote cell migration [ 151, but its activity as an autocrine motility factor has not been demonstrated. In this study, we have examined the effect of TNF on expression of PA and its inhibitor, PAI-1, in HME cells and also on migration in the presence or absence of EGF.
Epidermal growth factor (EGF) induces tubular formation of cultured human microvascular endothelial (HME) cells in the gel matrix containing collagen, and tumor necrosis factor (TNF) disrupts the tubular formation (Mawatari et al. (1989) J. Immunol. 143, 1619-1627). Here we studied the effects of EGF and TNF on endothelial cell migration and on the production of proteases. Confluent HME cells, when wounded with a razor blade, moved into the denuded space. This migration was stimulated by EGF and inhibited by TNF in this assay and in the Boyden chamber assay. Antibody against tissue-type plasminogen activator (t-PA) inhibited the EGF-stimulated cell migration in both assays by approximately 70%, but antibody against urokinase-type plasminogen activator (u-PA) could not inhibit its migration. Quantitative immunoreactive assays showed an approximately threeto fourfold increase of t-PA at 6 to 12 h after EGF addition, and TNF inhibited the production of t-PA by 50%. Northern blot analysis showed increased expression of t-PA mRNA by EGF alone in a time- and dose-dependent manner, whereas TNF alone inhibited its expression in a time- and dose-dependent manner. Northern blot analysis showed a significant increase of plasminogen activator inhibitor-l (PAI-1) mRNA when EGF or TNF was present. Stimulation by EGF of cell migration of HME cells and its inhibition by TNF appear to be closely correlated with the cellular modulation of t-PA and 6’ 1991 Academic Press, Inc. PAIactivities.
1 To whom correspondence dressed. Fax: 0975-49-1706.
MIZOGUCHI,*
EFFECT
MATERIALS
AND
OF TNF
ON ENDOTHELIAL
METHODS
Cells. HME cells were isolated from omental tissue and were cultured in Medium 199 (M-199) supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 pg/ml kanamycin as previously reported (14). The cells at passages (3-5) were used. Wound migration assay. The wound migration assay was performed according to Sato and Rifkin [8]. Confluent HME cells on a 35.mm plate were scraped with a razor blade, washed twice with phosphate-buffered saline (PBS) and incubated with M-199 containing 1% FBS and either 10 rig/ml of EGF or 100 U/ml of TNF. To assay the effect of t-PA antibody on HME cell migration, 100 fig/ml of rabbit anti-t-PA antibody or nonimmune rabbit IgG was incubated in the absence or presence of 10 rig/ml of EGF. After 24 h, cells were fixed with methanol and stained with Giemsa. We did not detect any cell division in the presence of a reduced amount of serum (1%). The numbers of cell nuclei crossing the starting line marked on the plate were counted in a 100 X lo-pm square. Several different parts of the plate were counted and the mean values were determined. Chemota.& of HME cells. Chemotactic assays were performed in blind well chambers (Ieda Boeki Co., Tokyo, Japan). M-199 containing 1% FBS and various doses of either EGF or TNF was placed in the lower chambers which were covered with nucleopore filters (5-wcm pore size) and then 200 ~1 of 2.5 X lo5 cells/ml HME cells was placed in the upper chambers. To assay the effect of t-PA, u-PA, and PAIL1 antibody on the chemotaxis of HME cells, 100 fig/ml of rabbit or goat anti-t-PA antibody or 100 pg/ml of goat anti-PAIantibody or nonimmune goat IgG, or 100 pg/ml of mouse anti-u-PA antibody or nonimmune mouse IgG was incubated with or without 10 rig/ml EGF in the lower chambers. After 24 h incubation at 37”C, media in the upper chambers were aspirated and cells on the upper surface of the filter were removed with a cotton swab. Cells on the lower surface were fixed with methanol and were stained with Giemsa. The number of stained nuclei was counted. Four high-power fields (X100) per each chamber were counted, and average values were determined from assays with four chambers. t-PA antigen assn~. Enzyme immunoassays with anti-t-PA antibody were carried out according to the procedure published by Mizoguchi et al. [15]. In brief, each well of a 96.well microplate was coated with 100 ~1 of anti-t-PA antibody F(ah), per well at 4°C for overnight. After samples of 100 ~1 were incubated at 37°C for 3 h, the plates were washed with BT-PBS (0.1% BSA and 0.05% Tween 20 in PBS) three times. The microplates were further incubated with 100 ~1 of peroxidase-labeled antibody diluted 1:lOO with BT-PBS at 37°C for 3 h, washed with BT-PBS three times, and then incubated with 100 ~1 of solution containing 0.4 mg/ml o-phenylenediamine and 0.006% H,O, in EIA buffer (0.1 M citrate-O.2 M sodium phosphate, pH 4.5) for 30 min at 37°C. The reaction was terminated by the addition of50 ~1 of 6 N H,SO, and absorbance was measured at 492 nm. Northern blot analysis. Northern blot analysis was carried out as described previously [15]. Total RNAs were isolated from HME cells incubated with or without EGF and/or TNF as described by Chomczynski and Sacchi [16]. Briefly, harvested cells were suspended in 4 ml of 4 M guanidinium thiocyanate, 25 mM sodium citrate, pH 7.0, 0.5% sarkosyl, and 0.1 M mercaptoethanol. Four hundred microliters of 2 M sodium acetate, pH 4.0, 4 ml of water-saturated phenol, and 800 ~1 of chloroform were added successively to the sample. After mixing vigorously, the mixture was left on ice for 20 min, and then samples were centrifuged at 10,OOOg for 30 min. The pelleted RNA was dissolved in sterile water. Fifteen micrograms of total RNA was fractionated on 1% agarose gels containing 2.2 M formaldehyde and then transferred to a Nytran filter (Schleicher & Schuell, Inc., Keene, NH). The filter-bound RNA was hybridized to 32P-labeled probes. Hybridization was carried out in Hybrisol 1 (Oncoruw.) for 24 h at 42°C. The filter was washed at room temperature in 2~ SSC andO.l% SDS and then washed in 0.2~ SSC and 0.1% SDS. Autoradiography was carried out using Kodak XAR film [14].
CELL
575
MIGRATION
DNA probes, antibodies, and other materials. Human t-PA cDNA was obtained from Dr. W-D. Schleuning (Schering Aktiengeselshaft Pharma Forshung, Berlin, West Germany) and a PAI- cDNA probe consisting of 3 kb PAI- human cDNA inserted in the EcoRl site of plasmid pGEM, was obtained from Dr. D. J. Loskutoff (Research Institute of Scripps Clinic, La Jolla, CA). Anti-human t-PA antibody and immunoassay (ELISA) kits of t-PA were used for t-PA assays. Poiyclonal rabbit IgG to human t-PA, which recognizes t-PA and t-PA-PAIcomplex, was prepared in Snow Brand Milk Products Co. Polyclonal goat IgG to t-PA or PAI- was purchased from Biopool AB, Sweden. Monoclonal mouse IgG to u-PA was purchased from Immunoteck Lab., France. An ELISA kit for PAI- was purchased from American Diagnostica Inc., New York. EGF (Toyobo Co., Osaka, Japan) and human recombinant TNF-n (TNF) 114,171 (Asahi Kasei Chemical lnd. Co., Tokyo, Japan) were used.
RESULTS
Effect of TNF andlor EGF on Migration Cells
of Endothelial
The migration of endothelial cells is a critical process in angiogenesis. Our previous study showed that EGF increased and TNF inhibited the formation of vessellike structures by HME cells in collagen gels [14]. Here we examined the effect of EGF and/or TNF on cell migration in a migration assay [8]. ConAuent monolayers of HME cells were wounded with a razor blade and further incubated in M-199 containing 1% FBS and either EGF (10 rig/ml) or TNF (100 U/ml) or both for 24 h at 37°C. HME cells that migrated from the edge of the wound were fixed and stained (Fig. 1). HME cell migration was stimulated over control 2.2-fold in the presence of EGF and was inhibited in the presence of TNF (Fig. 2). HME cell migration was reduced by 43% in the presence of both EGF and TNF over that observed with EGF. Consistent with a previous report [ 141, coadministration of EGF and TNF caused morphological changes of HME cells. Suramin, a polyanionic trypanocidal drug, which inhibits binding of growth factors to their receptor [ 181, almost completely inhibited the migration of HME cells in the presence of EGF (data not shown). Almost identical data for HME cell migration as seen in Fig. 2 were observed when DNA synthesis was inhibited by mitomycin (0.1 pg/ml) by more than 90% (data not shown). We concluded that EGF stimulates the migration of HME cells in the absence of active cell division. We next examined the chemotactic activity of HME cells using Boyden chambers. As seen in Fig. 3,10 rig/ml EGF stimulated the chemotactic activity of HME cells 4.3-fold higher than control in the presence of serum alone while 5 rig/ml EGF showed only a slight effect. In our assay system, EGF at lo-20 rig/ml showed a maximal effect on chemotactic activity as well as on wound migration activity (unpublished data). Addition of various doses of TNF alone did not stimulate the chemotactic activity (Fig. 3). The chemotactic activity of HME cells was enhanced by 10 rig/ml EGF, whereas the activity was inhibited by coadministration of 100 U/ml
576
MAWATARI
ET AL.
FIG. 1. Wound migration assay. Confluent HME cells were scraped with a razor blade and then washed with PBS. They were incubated for 24 h in the absence of added factor (A) or in the presence of 10 rig/ml EGF (B), 100 U/ml TNF (C), or EGF (10 rig/ml) plus TNF (100 U/ml) (D). They were fixed with methanol, stained with Giemsa, and photographed. The arrow points to the wound edge.
TNF by 60%. TNF alone had little effect on the HME cell migration and chemotactic activity, but TNF inhibited EGF-induced stimulation of HME cell migration. of Cell Migration
Inhibition
by Anti-t-PA
Antibody
Cell invasion [2] and endothelial cell migration [S] are often correlated with cellular levels of PA activities. To examine whether HME cell migration is correlated with PAS, we assayed the effect of anti-PA antibodies or anti-PAIantibody on the cell migration. In the Boyden chamber assay system, EGF at 10 rig/ml stimulated about four-fold the HME cell migration over the control, and the HME cell migration with EGF was significantly reduced in the presence of 100 pug/ml goat anti-tPA antibody, but not in the presence of 10 pg/ml t-PA antibody (Figs. 4A and 4B). By contrast, the HME cell migration in the presence of EGF was not inhibited
when 100 wg/ml mouse anti-u-PA antibody was present (Fig. 4C). In the presence of goat antibody against PAI1, the HME cell migration was stimulated by about 140% over that in the presence of EGF alone (Fig. 4B), but the difference was found to be statistically insignificant. These results suggest that the HME cell migration may be correlated with the expression of t-PA, rather than that of u-PA. Moreover, we further examined the possible involvement of t-PA in cell migration by using antibody against t-PA in both wound migration and Boyden chamber assay systems (Fig. 5). In these assay systems, we used anti-t-PA antibody developed in rabbit. In the wound migration assay system, incubation of HME cells with EGF and rabbit antibody to t-PA significantly reduced the HME cell migration to less than 40% of that in the presence of EGF alone (Fig. 5A). The che-
t ( I I .
50
= 2
40 30 20 10 0
C
A
TNF(Ulml 0
20
40
60
en DlSTANCE
100
0
20
40
60
80
(sum)
FIG. 2. Quantitation of EGF and/or TNF on wound migration activity by HME cells. Confluent HME cells were scraped with a razor blade and then washed with PBS. They were incubated for 24 h in the absence of added factor (A) or in the presence of 10 rig/ml EGF (B), 100 U/ml TNF (C), or EGF (10 rig/ml) plus TNF (100 U/ml) (D). Migration was quantitated by counting the number of cells within a 100 X lo-+rn area in six fields. The results are presented as mean number of cells per field and each result varied less than 10% of the mean values.
1
0
0
0
0
0
50
100 500
0
L
0
100
100
FIG. 3. Effect of EGF and/or TNF on chemotactic activity of HME cells in a Boyden chamber. EGF (0,5, 10, 50 rig/ml) (A), TNF (0, 50, 100, 500 U/ml) (B), and EGF (10 rig/ml) and/or TNF (100 U/ml) (C) was incubated in the lower chamber, then 2.5 X lo5 cells/ml of HME cells were placed in the upper chamber for 24 h. The chemotactic activity was quantitated by counting the number of cells within a high-power field (X100 magnification). These data are presented as mean number of cells from determination of cell number of 16 different fields in four chambers. Columns, mean; bars, SD. In (A), *, significantly (P < 0.01) different from the value in the absence of EGF. In (P < 0.01) different from the value in the presence ((3, *> significantly of 10 rig/ml EGF and 100 U/ml TNF by Student’s f test.
EFFECT
OF TNF
ON ENDOTHELIAL
CELL
E
577
MIGRATION
A
300
i
t
I
1
ii ”
x 100
T u
Ab (*lglmll
O
I-PA (IO)
t-PA (100)
PAl-1 (100)
O
EGF
0
0
0
0
++++
t-PA t-PA PAI-I (IO) (100)(l00)
O 0
0
+
+
FIG. 4. Effect of antibody to t-PA, u-PA, and PAI- on HME cell migration was determined using the Boyden chamber assay. (A) Goat antibody against t-PA (10 and 100 rglml) or PAI- (100 pg/ml) and nomimmune goat IgG as control (Ab, 0 pglml); (B) goat antibody against t-PA (IO and 100 pg/ml) or PAI(100 &g/ml) and nonimmune goat IgG as control (Ab; 0 pg/ml; EGF; 10 rig/ml); and (C) mouse antibody against u-PA (100 fig/ml) with or without 10 rig/ml EGF or nonimmune mouse IgG as control with or without 10 rig/ml EGF (Ab, 0 pg/ml in the absence and presence of EGF) was incubated in the lower chamber, and then 2.5 X lo5 cells/ml of HME cells were placed in the upper chamber for 24 h. The chemotactic activity was quantitated by counting the number of cells within high-power field (X100 magnification). These data are presented as mean number of cells from determination of cell number of 16 different fields in four chambers. Columns, mean; bars, SD. In (A), **, significantly (P < 0.05) different from the value for control in the absence of t-PA antibody. In (B), *, significantly (P < 0.01) different from the value for control in the presence of 10 rig/ml EGF alone. In (C), there appeared to be no significant (P > 0.05) difference between the value in the absence and presence of U-PA antibody when EGF was present.
motactic activity of HME cells treated with EGF was also significantly reduced to about 30% when the antibody was present (Fig. 5B). Both wound migration and Boyden chamber assays again suggest that the HME cell migration may be correlated with the cellular levels of t-PA activity. Expression
of t-PA and Its Inhibitor,
0lrl-
ii
n II; EGF”
0
0
I-PAAb
0
+
+ 0
t +
EGF
0
0
+
+
I-PAAb
0
+
0
+
FIG. 5. Inhibition of migration of HME cells by t-PA antibody. (A) The wound migration assay was carried out. Rabbit antibody to t-PA (100 *g/ml) or nonimmune rabbit serum for the control with and without 10 rig/ml EGF was incubated for 24 h. The results are presented as mean number of cells per field, and each mean value was deduced from determination of cell number of six different fields in two 35.mm plates. Columns, mean; bars, SD; *, significantly (P < 0.01) difIerent from the value for control in the presence of EGF and nonimmune IgG. (B) The Boyden chamber assay was carried out. Rabbit antibody against t-PA (100 pg/ml) or nonimmune rabbit IgG with or without 10 rig/ml EGF was incubated in the lower chamber, and then 2.5 X lo5 cells/ml of HME cells were placed in the upper chamber for 24 h. These data are presented as mean number of cells from the determination of cell number of 16 different fields in four (P < 0.01) differchambers. Column, mean; bars, SD; *, significantly ent from the value for control in the presence of EGF and nonimmune serum.
100 U/ml TNF alone, and at 6 h t-PA was decreased to about 50% of the initial activity. Coadministration of EGF and TNF caused a l&fold increase at 3 h, but t-PA production was thereafter decreased and the t-PA was about 50 to 30% of the initial control level at 12 to 24 h after the addition of both factors (Fig. 6).
PAI-
We examined in more detail the expression of t-PA in HME cells treated with EGF and/or TNF. The time course for t-PA production in the absence or presence of EGF and/or TNF was determined with antibody to tPA (Fig. 6). HME cells without added factor produced the same amount of t-PA by 24 h. Production of t-PA showed a maximum increase of three- to four-fold at 6 and 12 h after the addition of 10 rig/ml EGF. The production of t-PA declined after 12 h incubation with EGF and then returned to the basal level by 24 h. By contrast, TNF inhibited the production of t-PA. Secretion of tPA was greatly inhibited immediately after addition of
hr FIG. 6. Kinetics of t-PA production from HME cells treated with EGF and/or TNF. Confluent HME cells were incubated with 10 ng/ ml EGF (0) or 100 U/ml TNF (A) or 10 rig/ml EGF plus 100 U/ml TNF (0) for the indicated time. After washing, they were incubated in serum-free media for 6 h. The conditioned media were assayed by ELISA. Each point was the average values in the duplicate determinations that varied less than 10% of the average values.
578
MAWATARI
0
C
B
A Time(h)
ET AL.
3
6
9
12
Time(h)
0
6
9
12
rRNA-
0
3
6 9 12
4
4 t-PA-
Time(h)
t-PA-
4 t-PA-
4
4
4
rRNA-
rRNA-
FIG. 7. Time kinetics of t-PA and PAImRNA synthesis in HME cells treated with EGF or/and TNF. Confluent HME cells were incubated for indicated times (0,X, 6,9, 12 hl with 10 rig/ml EGF (Al or 100 IT/ml TNF (I31 or EGF (10 rig/ml) plus TNF (100 U/ml) (C) and harvested. Total RNA (15 pgl was fractionated on a 1% agarose gel and transferred to a Nytran filter. Northern blot hybridization was performed with s*P-labeled t-PA cDNA and PAI- cDNA. Arrowheads on the right side indicate rRNAs of 28 S and 18 S. 28 S rRNAs which were loaded on the gels under various conditions were presented after staining with ethidium bromide. The relative intensities of the bands for t-PA PAI- mRNA were analyzed with a densitometer.
We determined whether changes of t-PA activity in EGF- and/or TNF-treated HME cells were due to altered expression of the steady-state level of t-PA mRNA. Furthermore t-PA activity is modulated by inhibitors (PAIs), especially PAI- which is mainly produced from the endothelial cells [ 191 and binds to PAS. To examine the PAI- production by HME cells, we also determined the steady-state level of PAImRNA. First, we examined time kinetics of t-PA and PAImRNA synthesis in HME cells treated with EGF and/or TNF (Fig. 7). Northern blot analysis with t-PA cDNA probe demonstrated the expression of t-PA mRNA of 2.6 kb in HME cells (Fig. 7), consistent with our published paper on the t-PA mRNA sizes appearing in human breast cancer cells [15]. EGF stimulated the expression of t-PA mRNA in a time-dependent manner. Expression of t-PA mRNA increased two- to threefold over the initial level at time 0 at 9 h after treatment with EGF. By contrast, TNF decreased the t-PA mRNA level to less than 50% of the control in a time-dependent manner. In the presence of both EGF and TNF, the expression of t-PA mRNA was increased at 3 h and then decreased. Northern blot analysis with PAZ-1 cDNA probe demonstrated two PAI- mRNAs of 2.3 and 3.2 kb in HME cells (Fig. 7), consistent with the previous report [20]. The steady-state level of PAImRNA increased in the presence of either EGF or TNF in a timedependent manner, and increased about two- to threefold at 9 h. Coadministration of EGF and TNF apparently additively increased PAImRNA levels. Figure 8 demonstrates the effect of various doses of
EGF or TNF on t-PA mRNA or PAI- mRNA synthesis in HME cells. EGF stimulated the expression of both t-PA mRNA and PAImRNA in a dose-dependent manner. Expression of PAI- gene was stimulated and peaked at 10 rig/ml EGF. By contrast, TNF decreased the t-PA mRNA levels in a dose-dependent manner.
0
A lose
(ngiml)
0
06252.5
IO 40
Dose
(U/ml)
10 1625(251100
00 .
I-PA
t-PA-
.
PAI-
-
FIG. 8. Effects of various doses of EGF or TNF on t-PA or PAImRNA synthesis in HME cells. Confluent HME cells were incubated for 12 h with indicated doses of EGF (0,0.625,2.5,10,40 rig/ml) (A) or TNF (0, 6.25, 25, 100, 400 IJ/ml) (B) and harvested. Total RNA (15 pg) was fractionated on a 1% agarose gel and transferred to a Nytran filter. Northern blot hybridization was performed with “‘P-labeled t-PA and PAI- cDNA. Arrowheads on the right side indicate rRNAs of 28 S and 18 S. 28 S rRNAs which were loaded on the gels under various conditions for (A) and (B) were presented after staining with ethidium bromide.
EFFECT
OF TNF
ON ENDOTHELIAL
Our present data demonstrate that TNF inhibits cell migration accompanying decreased production of t-PA and increased production of PAI-1, whereas EGF stimulates cell migration accompanying a concomitant increase of t-PA production. DISCUSSION PAS are synthesized by cultured endothelial cells derived from large vessels and capillaries [ 191. Plasminogen is converted to a serine proteinase, plasmin, by PAS resulting in activation of the latent collagenase. Plasmin degrades fibronectin and laminin [4] and activates some growth factors [al]. One can thus expect that activation of the PA-plasmin system is required for cell migration. Sato and Rifkin [8] have reported that migration of bovine capillary endothelial cells was enhanced by bFGF with a concomitant increase of u-PA. Transforming growth factor-a (TGF-LU) is also angiogenic in viva [22, 231 as well as in vitro [21] and it stimulates the migration of bovine capillary endothelial cells in. vitro [24, 251. We have previously shown that EGF is angiogenic for cultured HME cells and induces collagenase [14]. Our present study demonstrates that EGF enhances HME cell migration and t-PA production. Pepper et al. [6] have reported the involvement of u-PA in the migration of bovine capillary endothelial cells. In our present assay systems for HME cells, antibody to u-PA can not affect the HME cell migration. We could not detect uPA activity in HME cells and the conditioned medium when assayed by zymography (data not shown). In contrast, the migration of HME cells in the presence of EGF is inhibited by the addition of anti-t-PA antibody to the culture medium. The t-PA-plasmin system might be involved in cell migration of the human endothelial cells. In either cultured bovine capillary endothelial cells or human umbilical vein endothelial cells, production of t-PA is not stimulated by EGF [26, 271. In our present culture system with HME cells, TGF-cu stimulated the production of t-PA at similar levels as EGF, but bFGF, TGF-/3, or insulin-like growth factor showed much less activity than EGF (unpublished data), suggesting a rather specific effect of EGF on t-PA production in HME cells. In comparison with the distance migrated with bovine aortic endothelial cells [8], cultured HME cells migrated about one-third to one-fourth. The differential effects of these growt,h factors and the difference in migration distance appear to depend upon the species and origins of the endothelial cells. TNF is known to induce hemorrhagic necrosis when injected into mice with transplantable tumors [12]. In addition, Watanabe et al. [28] have reported extravasation and thrombosis in tumor vessels in mice with transplantable tumors when treated with TNF. Furthermore, Sato et al. [ 131 have demonstrated morphological
CELL
MIGRATION
579
changes of human umbilical vein endothelial cells in culture by TNF. The cobblestone-like cell morphology changes to a fibroblastic morphology observed after treatment of HME cells with both TNF and EGF [14]. Our present study indicates that TNF apparently antagonizes the stimulatory effect of EGF on the migration of HME cells in two migration assay systems, the Boyden chamber and the wound migration systems. One may argue how TNF inhibits the migration of the HME cells. TNF appears to modulate t-PA activity by dual pathways: It affects synthesis of both t-PA and its inhibitor, PAI-1. EGF stimulates the production of both t-PA and PAI- in HME cells. By contrast, TNF alone decreases t-PA production and antagonizes the enhancement of t-PA production by EGF. TNF also increases the production of PAI- in HME cells, and the stimulatory effect of PAIproduction by TNF is magnified when EGF is present. In our cultured HME cells in the presence of EGF, TNF inhibits the cell migration of HME cells with concomitant increase in PAIproduction and decrease in t-PA production. A relevant report by Sawdey et al. [29] shows enhanced expression of the PAI- gene in cultured bovine aortic endothelial cells by TNF. The increased expression of PAImRNA by TNF is thought to be due to enhanced transcription of the inhibitor gene [29]. In our HME cells, TNF as well as EGF increases the steady-state levels of PAImRNA about two- to threefold over the control, suggesting a mechanism similar to that reported in the bovine endothelial cells. Our present study favors the idea that TNF may have an inhibitory action on angiogenesis. Sato and Rifkin [30] have recently reported that TGF-/3 inhibits the migration and u-PA production of bovine capillary endothelial cells. This action of TGF-fl in the bovine endothelial cells appears to mimic that of TNF in the HME cells. Consistent with our present study, TNF has more inhibitory effect on angiogenesis of bovine capillary endothelial cells in the presence of an angiogenic factor, bFGF, than in its absence [31]. Other studies have, however, reported angiogenic activity by TNF alone on bovine capillary endothelial cells in vitro [32] as well as on either rabbit cornea or chick chorioallantoic membrane in uiuo [33]. It remains to be studied why TNF is angiogenie for some cells and anti-angiogenic for others. The angiogenic effect of TNF per se might be due to the enhanced production of an angiogenic factor(s) from TNF-treated endothelial cells since TNF induces various growth factors and cytokines in endothelial cells (IL-l, IL-6, PGE,, GM-CSF, etc.) [14,34-361. This possibility has not been verified. Further study should be required to understand the underlying molecular mechanism for modulation of proteolysis by TNF and/or EGF in relation to cell migration of the human endothelial cells.
580
MAWATARI
We thank Dr. H. K. Kleinman (NIDR, NIH, MD) for critically reading of this manuscript and invaluable comments. We also thank Drs. M. Kobayashi, Y. Uchida, and their colleagues (Oita Medical School) for supplying us with omental adipose tissue.
REFERENCES Moscatelli, D. A., Presta, M., and R&in, D. B. (1986) Proc. Natl. Acad. Sci. USA 83, 2091. 2. Mignatti, P., Tsuboi, R., Robbins, E., and Rifkin, D. B. (1989) J. Cell Biol. 108, 671. 3. Presta, M., Moscatelli, D., Joseph-Silverstein, J., and Rifkin, D. B. (1986) Mol. Cell. Biol. 6, 4060. 4. Werb, Z., Banda, M. J., and Jones, P. A. (1980) J. Exp. Med.
ET AL. 18 19
10. 11. 12. 13.
Loskutoff, D. J., Linders, M., Keijer, J., Veerman, H., Van Heerikhuizen, H., and H. Pannekoek. (1987) Biochemistry 26,3763. Sprengers, E. D., Printer, H. M., Kooistra, T., and Van Hinsbergh, V. W. (1985) J. Lab. Clin. Med. 105, 751. Carswell, E. A., Old, L. J., Kassel, R. L., Green, S., Fiore, N., and Williamson, B. (1975) Proc. Natl. Acad. Sci. USA 72, 3666. Sato, N., Goto, T., Haranaka, K., Satomi, N., Nariuchi, H., Mano-Hirano, Y., and Sawasaki, Y. (1986) J. N&l. Cancer Inst.
76,1113. 14. Mawatari,
M., Kohno, K., Mizoguchi, H., Matsuda, T., Asoh, K., Damme, J. V., Welgus, H. G., and Kuwano, M. (1989) J. Immunol. 143, 1619. 15. Mizoguchi, H., Uchiumi, T., Ono, M., Kohno, K., and Kuwano, M. (1990) Biochim. Biophys. Acta 1052, 475. 16. Chomczynski, P., and Sacchi, N. (1987) Anal. Biochrm. 162,
156. 17. Asoh, K., Watanabe, Y., Mizoguchi, Kohno, K., and Kuwano,
H., Mawatari, M., Ono, M., M. (1989) Biochem. Biophys. Res. Com-
mun. 162,794. Received June 6, 1990 Revised version received September
24, 1990
Moscatelli,
D., and Rifkin,
G. D., and Moses, H. L.
D. B. (1988) Biochim.
Biophys. Acta
948,67. 20. Ginsburg,
D., Zeheb, Z., Yang, A., Rotterty, M., Andreaseu, P. A., Nielsen, L., Dan, K., Lebo, V., and Gelehrter, T. D. (1986) J. Clin. Inucst. 78, 1673.
1.
152,134O. 5. Mullins, D. E., and Rohrlich, S. T. (1983) Biochim. B&&s. Acta 695,177. 6. Pepper, M. S., Vassalli, J. D., Montesano, R., and Orci, I,. (1987) J. Cell Biol. 105, 2535. 7. Saksela, 0. (1985) Biochim. Biophys. Acta 823, 35. 8. Sato, Y., and R&in, D. B. (1988) J. Cell Biol. 107, 1199. 9. Emeis, J. J., and Kooistra, T. (1986) J. Exp. Med. 163, 1260.
Coffey, R. J., Jr., Leaf, E. B., Shipley, (1987) J. Cell. Physiol. 132, 143.
21.
Montesano, R., Vassalli, J-D., Baird, A., Gillemin, I,. (1986) Proc. Nat/. Acad. Sci. lJSA 83, 5968.
R., and Orci,
22. Hayek, A., Culler, F. I,., Beattie, G. M., Lopez, A. D., Cuevas, P., and Baird, A. (1987) Biochem. Biophys. Res. Commun. 147,876. 23. Schreiber, A. B., Winkler, M. E., and Derynck, R. (1986) Science 232,125O. 24. Mano-Hirano, Y., Sato, N., Sawasaki, Y., and Goto, T. (1986) Proc. Jpn. Acad. 62, 235.
25. Zetter, B. R. (1980) Nature (London) 285, 41. 26. Gross, J. L., Moscatelli, D., and Rifkin, D. B. (1983) Proc. Natl. Acad. Sci. USA 80, 2623.
27. Moscatelli, D. A., Rifkin, D. B., and Jaffe, F. A. (1985) Exp. Cell. RPS. 156,379. 28. Watanabe, N., Niitsu, Y., IJmeno, H., Kuriyama, H., Neda, H., Yamauchi, 48, 2179. 29. 30. 31.
N., Maeda, M., and IJrushizaki,
I. (1988) Cancer Res.
Sawdey, M., Podor, T. J., and Loskutoff, D. J. (1989) J. Biol. Chem. 264, 10,396. Sato, Y., and Rifkin, D. B. (1989) J. Cell Biol. 109, 309. Schweigerer, L., Malerstein, B., and Gospodarowicz, B&hem. Bioph-ys. Res. Commun. 143, 997.
D. (1987)
32. Frater-Schroder, Bohlen,
M., Risau, W., Hallmann, R., Gautschi, P., and P. (1987) Proc. Natl. Acad. Sci. USA 84, 5277.
33. Leibovich,
S. J., Polverini, P. J., Shepard, H. M., Wiseman, D. M., Shively, V., and Nuseir, N. (1987) Nature (London) 329,
630. 34. Bachwich,
P. R., Chensue, S. W., Larrick, J., and Kunkel, (1986) B&hem. Biophys. Res. Commun. 136, 94.
S. L.
35. Broudy, V. C., Kaushansky, Adamson,
K., Segal, G. M., Harlan, J. M., and J. W. (1986) Proc. Natl. Acad. Sci. 1JSA 83, 7467.
36. Dayer, ,J. M., Beutler, 162.2163.
B., and Cerami,
A. (1985) J. Exp. Med.
CELL
RESEARCH
192,
574-580
(1991)
Tumor Necrosis Factor and Epidermal Growth Factor Modulate Migration of Human Microvascular Endothelial Cells and Production of Tissue-Type Plasminogen Activator and Its Inhibitor MASASUMI
MAWATARI,*
KAZUKI OKAMURA,* TAKAO MATSUDA,* RYOJI HAMANAKA,* HIROMOTO KANJI HIGASHIO,? KIMITOSHI KOHNO,* AND MICHIHIKO KUWANO*,’
*Department
of Riochcmistry,
Oita Medical School, Oita X79-.56, Japan; and tfficwarch Institute, Snow Brand Milk Products Company. 7’ochigi 329-0.5. Japan
INTRODUCTION
The production of proteases and cell migration are essential during angiogenesis. Proteases, such as plas-
0014.4827191
Copyright All
rights
$3.00 (c 1991 by of reproduction
and reprint
requests
should
’ Ahhreviations used: PA, plasminogen activator; t-PA, tissue-type plasminogen activator; u-PA, urokinase-type plasminogen activitor; PAI-1, plasminogen activator inhibitor-l, EGF, epidermal growth factor; TNF, tumor necrosis f’actor; hFGF, basic fibroblast growth factor; HME cell, human microvascular endothelial cell.
be ad-
574
Academic Press, Inc. in any
form
reserved.
of Life Sciwzw,
minogen activator (PA)’ and collagenase, are produced from endothelial cells when angiogenic factors are present [l-3]. In addition, PA can convert the latent collagenase to active collagenase through converting plasminogen to plasmin. Plasmin itself is also active in de[4]. The grading extracellular matrix components production of PA may be regulated in concert with cell migration [S-7], since basic fibroblast growth factor (bFGF) stimulates cell migration of bovine aortic endothelial cells and causes a concomitant increase of tissuetype plasminogen activator (t-PA) induction [S]. An inhibitor of PA, plasminogen activator inhibitor-l (PAIl), is also synthesized and secreted from endothelial cells [9, 101 and hepatocytes [ll]. PAI- rapidly binds with PA, and thus can modulate PA activity. Tumor necrosis factor (TNF), which is secreted from activated macrophages and monocytes, causes hemorrhagic necrosis in mice with transplantable tumors [ 121. TNF shows cytotoxic or cytostatic effects and induces morphologic changes in endothelial cells [ 131. We have previously reported that epidermal growth factor (EGF) induces the vessel-like structures of human microvascular endothelial (HME) cells in a type I collagen gel, but TNF abrogates this induction of vessel-like structures [ 141. Coadministration of EGF and TNF synergistically increases the production of tissue inhibitor of metalloproteinases and interleukin-6 [ 141. The mechanisms for the effects of EGF and TNF on endothelial cells are not understood but could be related to cell migration and the production of PA. EGF has been shown to promote cell migration [ 151, but its activity as an autocrine motility factor has not been demonstrated. In this study, we have examined the effect of TNF on expression of PA and its inhibitor, PAI-1, in HME cells and also on migration in the presence or absence of EGF.
Epidermal growth factor (EGF) induces tubular formation of cultured human microvascular endothelial (HME) cells in the gel matrix containing collagen, and tumor necrosis factor (TNF) disrupts the tubular formation (Mawatari et al. (1989) J. Immunol. 143, 1619-1627). Here we studied the effects of EGF and TNF on endothelial cell migration and on the production of proteases. Confluent HME cells, when wounded with a razor blade, moved into the denuded space. This migration was stimulated by EGF and inhibited by TNF in this assay and in the Boyden chamber assay. Antibody against tissue-type plasminogen activator (t-PA) inhibited the EGF-stimulated cell migration in both assays by approximately 70%, but antibody against urokinase-type plasminogen activator (u-PA) could not inhibit its migration. Quantitative immunoreactive assays showed an approximately threeto fourfold increase of t-PA at 6 to 12 h after EGF addition, and TNF inhibited the production of t-PA by 50%. Northern blot analysis showed increased expression of t-PA mRNA by EGF alone in a time- and dose-dependent manner, whereas TNF alone inhibited its expression in a time- and dose-dependent manner. Northern blot analysis showed a significant increase of plasminogen activator inhibitor-l (PAI-1) mRNA when EGF or TNF was present. Stimulation by EGF of cell migration of HME cells and its inhibition by TNF appear to be closely correlated with the cellular modulation of t-PA and 6’ 1991 Academic Press, Inc. PAIactivities.
1 To whom correspondence dressed. Fax: 0975-49-1706.
MIZOGUCHI,*
EFFECT
MATERIALS
AND
OF TNF
ON ENDOTHELIAL
METHODS
Cells. HME cells were isolated from omental tissue and were cultured in Medium 199 (M-199) supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 pg/ml kanamycin as previously reported (14). The cells at passages (3-5) were used. Wound migration assay. The wound migration assay was performed according to Sato and Rifkin [8]. Confluent HME cells on a 35.mm plate were scraped with a razor blade, washed twice with phosphate-buffered saline (PBS) and incubated with M-199 containing 1% FBS and either 10 rig/ml of EGF or 100 U/ml of TNF. To assay the effect of t-PA antibody on HME cell migration, 100 fig/ml of rabbit anti-t-PA antibody or nonimmune rabbit IgG was incubated in the absence or presence of 10 rig/ml of EGF. After 24 h, cells were fixed with methanol and stained with Giemsa. We did not detect any cell division in the presence of a reduced amount of serum (1%). The numbers of cell nuclei crossing the starting line marked on the plate were counted in a 100 X lo-pm square. Several different parts of the plate were counted and the mean values were determined. Chemota.& of HME cells. Chemotactic assays were performed in blind well chambers (Ieda Boeki Co., Tokyo, Japan). M-199 containing 1% FBS and various doses of either EGF or TNF was placed in the lower chambers which were covered with nucleopore filters (5-wcm pore size) and then 200 ~1 of 2.5 X lo5 cells/ml HME cells was placed in the upper chambers. To assay the effect of t-PA, u-PA, and PAIL1 antibody on the chemotaxis of HME cells, 100 fig/ml of rabbit or goat anti-t-PA antibody or 100 pg/ml of goat anti-PAIantibody or nonimmune goat IgG, or 100 pg/ml of mouse anti-u-PA antibody or nonimmune mouse IgG was incubated with or without 10 rig/ml EGF in the lower chambers. After 24 h incubation at 37”C, media in the upper chambers were aspirated and cells on the upper surface of the filter were removed with a cotton swab. Cells on the lower surface were fixed with methanol and were stained with Giemsa. The number of stained nuclei was counted. Four high-power fields (X100) per each chamber were counted, and average values were determined from assays with four chambers. t-PA antigen assn~. Enzyme immunoassays with anti-t-PA antibody were carried out according to the procedure published by Mizoguchi et al. [15]. In brief, each well of a 96.well microplate was coated with 100 ~1 of anti-t-PA antibody F(ah), per well at 4°C for overnight. After samples of 100 ~1 were incubated at 37°C for 3 h, the plates were washed with BT-PBS (0.1% BSA and 0.05% Tween 20 in PBS) three times. The microplates were further incubated with 100 ~1 of peroxidase-labeled antibody diluted 1:lOO with BT-PBS at 37°C for 3 h, washed with BT-PBS three times, and then incubated with 100 ~1 of solution containing 0.4 mg/ml o-phenylenediamine and 0.006% H,O, in EIA buffer (0.1 M citrate-O.2 M sodium phosphate, pH 4.5) for 30 min at 37°C. The reaction was terminated by the addition of50 ~1 of 6 N H,SO, and absorbance was measured at 492 nm. Northern blot analysis. Northern blot analysis was carried out as described previously [15]. Total RNAs were isolated from HME cells incubated with or without EGF and/or TNF as described by Chomczynski and Sacchi [16]. Briefly, harvested cells were suspended in 4 ml of 4 M guanidinium thiocyanate, 25 mM sodium citrate, pH 7.0, 0.5% sarkosyl, and 0.1 M mercaptoethanol. Four hundred microliters of 2 M sodium acetate, pH 4.0, 4 ml of water-saturated phenol, and 800 ~1 of chloroform were added successively to the sample. After mixing vigorously, the mixture was left on ice for 20 min, and then samples were centrifuged at 10,OOOg for 30 min. The pelleted RNA was dissolved in sterile water. Fifteen micrograms of total RNA was fractionated on 1% agarose gels containing 2.2 M formaldehyde and then transferred to a Nytran filter (Schleicher & Schuell, Inc., Keene, NH). The filter-bound RNA was hybridized to 32P-labeled probes. Hybridization was carried out in Hybrisol 1 (Oncoruw.) for 24 h at 42°C. The filter was washed at room temperature in 2~ SSC andO.l% SDS and then washed in 0.2~ SSC and 0.1% SDS. Autoradiography was carried out using Kodak XAR film [14].
CELL
575
MIGRATION
DNA probes, antibodies, and other materials. Human t-PA cDNA was obtained from Dr. W-D. Schleuning (Schering Aktiengeselshaft Pharma Forshung, Berlin, West Germany) and a PAI- cDNA probe consisting of 3 kb PAI- human cDNA inserted in the EcoRl site of plasmid pGEM, was obtained from Dr. D. J. Loskutoff (Research Institute of Scripps Clinic, La Jolla, CA). Anti-human t-PA antibody and immunoassay (ELISA) kits of t-PA were used for t-PA assays. Poiyclonal rabbit IgG to human t-PA, which recognizes t-PA and t-PA-PAIcomplex, was prepared in Snow Brand Milk Products Co. Polyclonal goat IgG to t-PA or PAI- was purchased from Biopool AB, Sweden. Monoclonal mouse IgG to u-PA was purchased from Immunoteck Lab., France. An ELISA kit for PAI- was purchased from American Diagnostica Inc., New York. EGF (Toyobo Co., Osaka, Japan) and human recombinant TNF-n (TNF) 114,171 (Asahi Kasei Chemical lnd. Co., Tokyo, Japan) were used.
RESULTS
Effect of TNF andlor EGF on Migration Cells
of Endothelial
The migration of endothelial cells is a critical process in angiogenesis. Our previous study showed that EGF increased and TNF inhibited the formation of vessellike structures by HME cells in collagen gels [14]. Here we examined the effect of EGF and/or TNF on cell migration in a migration assay [8]. ConAuent monolayers of HME cells were wounded with a razor blade and further incubated in M-199 containing 1% FBS and either EGF (10 rig/ml) or TNF (100 U/ml) or both for 24 h at 37°C. HME cells that migrated from the edge of the wound were fixed and stained (Fig. 1). HME cell migration was stimulated over control 2.2-fold in the presence of EGF and was inhibited in the presence of TNF (Fig. 2). HME cell migration was reduced by 43% in the presence of both EGF and TNF over that observed with EGF. Consistent with a previous report [ 141, coadministration of EGF and TNF caused morphological changes of HME cells. Suramin, a polyanionic trypanocidal drug, which inhibits binding of growth factors to their receptor [ 181, almost completely inhibited the migration of HME cells in the presence of EGF (data not shown). Almost identical data for HME cell migration as seen in Fig. 2 were observed when DNA synthesis was inhibited by mitomycin (0.1 pg/ml) by more than 90% (data not shown). We concluded that EGF stimulates the migration of HME cells in the absence of active cell division. We next examined the chemotactic activity of HME cells using Boyden chambers. As seen in Fig. 3,10 rig/ml EGF stimulated the chemotactic activity of HME cells 4.3-fold higher than control in the presence of serum alone while 5 rig/ml EGF showed only a slight effect. In our assay system, EGF at lo-20 rig/ml showed a maximal effect on chemotactic activity as well as on wound migration activity (unpublished data). Addition of various doses of TNF alone did not stimulate the chemotactic activity (Fig. 3). The chemotactic activity of HME cells was enhanced by 10 rig/ml EGF, whereas the activity was inhibited by coadministration of 100 U/ml
576
MAWATARI
ET AL.
FIG. 1. Wound migration assay. Confluent HME cells were scraped with a razor blade and then washed with PBS. They were incubated for 24 h in the absence of added factor (A) or in the presence of 10 rig/ml EGF (B), 100 U/ml TNF (C), or EGF (10 rig/ml) plus TNF (100 U/ml) (D). They were fixed with methanol, stained with Giemsa, and photographed. The arrow points to the wound edge.
TNF by 60%. TNF alone had little effect on the HME cell migration and chemotactic activity, but TNF inhibited EGF-induced stimulation of HME cell migration. of Cell Migration
Inhibition
by Anti-t-PA
Antibody
Cell invasion [2] and endothelial cell migration [S] are often correlated with cellular levels of PA activities. To examine whether HME cell migration is correlated with PAS, we assayed the effect of anti-PA antibodies or anti-PAIantibody on the cell migration. In the Boyden chamber assay system, EGF at 10 rig/ml stimulated about four-fold the HME cell migration over the control, and the HME cell migration with EGF was significantly reduced in the presence of 100 pug/ml goat anti-tPA antibody, but not in the presence of 10 pg/ml t-PA antibody (Figs. 4A and 4B). By contrast, the HME cell migration in the presence of EGF was not inhibited
when 100 wg/ml mouse anti-u-PA antibody was present (Fig. 4C). In the presence of goat antibody against PAI1, the HME cell migration was stimulated by about 140% over that in the presence of EGF alone (Fig. 4B), but the difference was found to be statistically insignificant. These results suggest that the HME cell migration may be correlated with the expression of t-PA, rather than that of u-PA. Moreover, we further examined the possible involvement of t-PA in cell migration by using antibody against t-PA in both wound migration and Boyden chamber assay systems (Fig. 5). In these assay systems, we used anti-t-PA antibody developed in rabbit. In the wound migration assay system, incubation of HME cells with EGF and rabbit antibody to t-PA significantly reduced the HME cell migration to less than 40% of that in the presence of EGF alone (Fig. 5A). The che-
t ( I I .
50
= 2
40 30 20 10 0
C
A
TNF(Ulml 0
20
40
60
en DlSTANCE
100
0
20
40
60
80
(sum)
FIG. 2. Quantitation of EGF and/or TNF on wound migration activity by HME cells. Confluent HME cells were scraped with a razor blade and then washed with PBS. They were incubated for 24 h in the absence of added factor (A) or in the presence of 10 rig/ml EGF (B), 100 U/ml TNF (C), or EGF (10 rig/ml) plus TNF (100 U/ml) (D). Migration was quantitated by counting the number of cells within a 100 X lo-+rn area in six fields. The results are presented as mean number of cells per field and each result varied less than 10% of the mean values.
1
0
0
0
0
0
50
100 500
0
L
0
100
100
FIG. 3. Effect of EGF and/or TNF on chemotactic activity of HME cells in a Boyden chamber. EGF (0,5, 10, 50 rig/ml) (A), TNF (0, 50, 100, 500 U/ml) (B), and EGF (10 rig/ml) and/or TNF (100 U/ml) (C) was incubated in the lower chamber, then 2.5 X lo5 cells/ml of HME cells were placed in the upper chamber for 24 h. The chemotactic activity was quantitated by counting the number of cells within a high-power field (X100 magnification). These data are presented as mean number of cells from determination of cell number of 16 different fields in four chambers. Columns, mean; bars, SD. In (A), *, significantly (P < 0.01) different from the value in the absence of EGF. In (P < 0.01) different from the value in the presence ((3, *> significantly of 10 rig/ml EGF and 100 U/ml TNF by Student’s f test.
EFFECT
OF TNF
ON ENDOTHELIAL
CELL
E
577
MIGRATION
A
300
i
t
I
1
ii ”
x 100
T u
Ab (*lglmll
O
I-PA (IO)
t-PA (100)
PAl-1 (100)
O
EGF
0
0
0
0
++++
t-PA t-PA PAI-I (IO) (100)(l00)
O 0
0
+
+
FIG. 4. Effect of antibody to t-PA, u-PA, and PAI- on HME cell migration was determined using the Boyden chamber assay. (A) Goat antibody against t-PA (10 and 100 rglml) or PAI- (100 pg/ml) and nomimmune goat IgG as control (Ab, 0 pglml); (B) goat antibody against t-PA (IO and 100 pg/ml) or PAI(100 &g/ml) and nonimmune goat IgG as control (Ab; 0 pg/ml; EGF; 10 rig/ml); and (C) mouse antibody against u-PA (100 fig/ml) with or without 10 rig/ml EGF or nonimmune mouse IgG as control with or without 10 rig/ml EGF (Ab, 0 pg/ml in the absence and presence of EGF) was incubated in the lower chamber, and then 2.5 X lo5 cells/ml of HME cells were placed in the upper chamber for 24 h. The chemotactic activity was quantitated by counting the number of cells within high-power field (X100 magnification). These data are presented as mean number of cells from determination of cell number of 16 different fields in four chambers. Columns, mean; bars, SD. In (A), **, significantly (P < 0.05) different from the value for control in the absence of t-PA antibody. In (B), *, significantly (P < 0.01) different from the value for control in the presence of 10 rig/ml EGF alone. In (C), there appeared to be no significant (P > 0.05) difference between the value in the absence and presence of U-PA antibody when EGF was present.
motactic activity of HME cells treated with EGF was also significantly reduced to about 30% when the antibody was present (Fig. 5B). Both wound migration and Boyden chamber assays again suggest that the HME cell migration may be correlated with the cellular levels of t-PA activity. Expression
of t-PA and Its Inhibitor,
0lrl-
ii
n II; EGF”
0
0
I-PAAb
0
+
+ 0
t +
EGF
0
0
+
+
I-PAAb
0
+
0
+
FIG. 5. Inhibition of migration of HME cells by t-PA antibody. (A) The wound migration assay was carried out. Rabbit antibody to t-PA (100 *g/ml) or nonimmune rabbit serum for the control with and without 10 rig/ml EGF was incubated for 24 h. The results are presented as mean number of cells per field, and each mean value was deduced from determination of cell number of six different fields in two 35.mm plates. Columns, mean; bars, SD; *, significantly (P < 0.01) difIerent from the value for control in the presence of EGF and nonimmune IgG. (B) The Boyden chamber assay was carried out. Rabbit antibody against t-PA (100 pg/ml) or nonimmune rabbit IgG with or without 10 rig/ml EGF was incubated in the lower chamber, and then 2.5 X lo5 cells/ml of HME cells were placed in the upper chamber for 24 h. These data are presented as mean number of cells from the determination of cell number of 16 different fields in four (P < 0.01) differchambers. Column, mean; bars, SD; *, significantly ent from the value for control in the presence of EGF and nonimmune serum.
100 U/ml TNF alone, and at 6 h t-PA was decreased to about 50% of the initial activity. Coadministration of EGF and TNF caused a l&fold increase at 3 h, but t-PA production was thereafter decreased and the t-PA was about 50 to 30% of the initial control level at 12 to 24 h after the addition of both factors (Fig. 6).
PAI-
We examined in more detail the expression of t-PA in HME cells treated with EGF and/or TNF. The time course for t-PA production in the absence or presence of EGF and/or TNF was determined with antibody to tPA (Fig. 6). HME cells without added factor produced the same amount of t-PA by 24 h. Production of t-PA showed a maximum increase of three- to four-fold at 6 and 12 h after the addition of 10 rig/ml EGF. The production of t-PA declined after 12 h incubation with EGF and then returned to the basal level by 24 h. By contrast, TNF inhibited the production of t-PA. Secretion of tPA was greatly inhibited immediately after addition of
hr FIG. 6. Kinetics of t-PA production from HME cells treated with EGF and/or TNF. Confluent HME cells were incubated with 10 ng/ ml EGF (0) or 100 U/ml TNF (A) or 10 rig/ml EGF plus 100 U/ml TNF (0) for the indicated time. After washing, they were incubated in serum-free media for 6 h. The conditioned media were assayed by ELISA. Each point was the average values in the duplicate determinations that varied less than 10% of the average values.
578
MAWATARI
0
C
B
A Time(h)
ET AL.
3
6
9
12
Time(h)
0
6
9
12
rRNA-
0
3
6 9 12
4
4 t-PA-
Time(h)
t-PA-
4 t-PA-
4
4
4
rRNA-
rRNA-
FIG. 7. Time kinetics of t-PA and PAImRNA synthesis in HME cells treated with EGF or/and TNF. Confluent HME cells were incubated for indicated times (0,X, 6,9, 12 hl with 10 rig/ml EGF (Al or 100 IT/ml TNF (I31 or EGF (10 rig/ml) plus TNF (100 U/ml) (C) and harvested. Total RNA (15 pgl was fractionated on a 1% agarose gel and transferred to a Nytran filter. Northern blot hybridization was performed with s*P-labeled t-PA cDNA and PAI- cDNA. Arrowheads on the right side indicate rRNAs of 28 S and 18 S. 28 S rRNAs which were loaded on the gels under various conditions were presented after staining with ethidium bromide. The relative intensities of the bands for t-PA PAI- mRNA were analyzed with a densitometer.
We determined whether changes of t-PA activity in EGF- and/or TNF-treated HME cells were due to altered expression of the steady-state level of t-PA mRNA. Furthermore t-PA activity is modulated by inhibitors (PAIs), especially PAI- which is mainly produced from the endothelial cells [ 191 and binds to PAS. To examine the PAI- production by HME cells, we also determined the steady-state level of PAImRNA. First, we examined time kinetics of t-PA and PAImRNA synthesis in HME cells treated with EGF and/or TNF (Fig. 7). Northern blot analysis with t-PA cDNA probe demonstrated the expression of t-PA mRNA of 2.6 kb in HME cells (Fig. 7), consistent with our published paper on the t-PA mRNA sizes appearing in human breast cancer cells [15]. EGF stimulated the expression of t-PA mRNA in a time-dependent manner. Expression of t-PA mRNA increased two- to threefold over the initial level at time 0 at 9 h after treatment with EGF. By contrast, TNF decreased the t-PA mRNA level to less than 50% of the control in a time-dependent manner. In the presence of both EGF and TNF, the expression of t-PA mRNA was increased at 3 h and then decreased. Northern blot analysis with PAZ-1 cDNA probe demonstrated two PAI- mRNAs of 2.3 and 3.2 kb in HME cells (Fig. 7), consistent with the previous report [20]. The steady-state level of PAImRNA increased in the presence of either EGF or TNF in a timedependent manner, and increased about two- to threefold at 9 h. Coadministration of EGF and TNF apparently additively increased PAImRNA levels. Figure 8 demonstrates the effect of various doses of
EGF or TNF on t-PA mRNA or PAI- mRNA synthesis in HME cells. EGF stimulated the expression of both t-PA mRNA and PAImRNA in a dose-dependent manner. Expression of PAI- gene was stimulated and peaked at 10 rig/ml EGF. By contrast, TNF decreased the t-PA mRNA levels in a dose-dependent manner.
0
A lose
(ngiml)
0
06252.5
IO 40
Dose
(U/ml)
10 1625(251100
00 .
I-PA
t-PA-
.
PAI-
-
FIG. 8. Effects of various doses of EGF or TNF on t-PA or PAImRNA synthesis in HME cells. Confluent HME cells were incubated for 12 h with indicated doses of EGF (0,0.625,2.5,10,40 rig/ml) (A) or TNF (0, 6.25, 25, 100, 400 IJ/ml) (B) and harvested. Total RNA (15 pg) was fractionated on a 1% agarose gel and transferred to a Nytran filter. Northern blot hybridization was performed with “‘P-labeled t-PA and PAI- cDNA. Arrowheads on the right side indicate rRNAs of 28 S and 18 S. 28 S rRNAs which were loaded on the gels under various conditions for (A) and (B) were presented after staining with ethidium bromide.
EFFECT
OF TNF
ON ENDOTHELIAL
Our present data demonstrate that TNF inhibits cell migration accompanying decreased production of t-PA and increased production of PAI-1, whereas EGF stimulates cell migration accompanying a concomitant increase of t-PA production. DISCUSSION PAS are synthesized by cultured endothelial cells derived from large vessels and capillaries [ 191. Plasminogen is converted to a serine proteinase, plasmin, by PAS resulting in activation of the latent collagenase. Plasmin degrades fibronectin and laminin [4] and activates some growth factors [al]. One can thus expect that activation of the PA-plasmin system is required for cell migration. Sato and Rifkin [8] have reported that migration of bovine capillary endothelial cells was enhanced by bFGF with a concomitant increase of u-PA. Transforming growth factor-a (TGF-LU) is also angiogenic in viva [22, 231 as well as in vitro [21] and it stimulates the migration of bovine capillary endothelial cells in. vitro [24, 251. We have previously shown that EGF is angiogenic for cultured HME cells and induces collagenase [14]. Our present study demonstrates that EGF enhances HME cell migration and t-PA production. Pepper et al. [6] have reported the involvement of u-PA in the migration of bovine capillary endothelial cells. In our present assay systems for HME cells, antibody to u-PA can not affect the HME cell migration. We could not detect uPA activity in HME cells and the conditioned medium when assayed by zymography (data not shown). In contrast, the migration of HME cells in the presence of EGF is inhibited by the addition of anti-t-PA antibody to the culture medium. The t-PA-plasmin system might be involved in cell migration of the human endothelial cells. In either cultured bovine capillary endothelial cells or human umbilical vein endothelial cells, production of t-PA is not stimulated by EGF [26, 271. In our present culture system with HME cells, TGF-cu stimulated the production of t-PA at similar levels as EGF, but bFGF, TGF-/3, or insulin-like growth factor showed much less activity than EGF (unpublished data), suggesting a rather specific effect of EGF on t-PA production in HME cells. In comparison with the distance migrated with bovine aortic endothelial cells [8], cultured HME cells migrated about one-third to one-fourth. The differential effects of these growt,h factors and the difference in migration distance appear to depend upon the species and origins of the endothelial cells. TNF is known to induce hemorrhagic necrosis when injected into mice with transplantable tumors [12]. In addition, Watanabe et al. [28] have reported extravasation and thrombosis in tumor vessels in mice with transplantable tumors when treated with TNF. Furthermore, Sato et al. [ 131 have demonstrated morphological
CELL
MIGRATION
579
changes of human umbilical vein endothelial cells in culture by TNF. The cobblestone-like cell morphology changes to a fibroblastic morphology observed after treatment of HME cells with both TNF and EGF [14]. Our present study indicates that TNF apparently antagonizes the stimulatory effect of EGF on the migration of HME cells in two migration assay systems, the Boyden chamber and the wound migration systems. One may argue how TNF inhibits the migration of the HME cells. TNF appears to modulate t-PA activity by dual pathways: It affects synthesis of both t-PA and its inhibitor, PAI-1. EGF stimulates the production of both t-PA and PAI- in HME cells. By contrast, TNF alone decreases t-PA production and antagonizes the enhancement of t-PA production by EGF. TNF also increases the production of PAI- in HME cells, and the stimulatory effect of PAIproduction by TNF is magnified when EGF is present. In our cultured HME cells in the presence of EGF, TNF inhibits the cell migration of HME cells with concomitant increase in PAIproduction and decrease in t-PA production. A relevant report by Sawdey et al. [29] shows enhanced expression of the PAI- gene in cultured bovine aortic endothelial cells by TNF. The increased expression of PAImRNA by TNF is thought to be due to enhanced transcription of the inhibitor gene [29]. In our HME cells, TNF as well as EGF increases the steady-state levels of PAImRNA about two- to threefold over the control, suggesting a mechanism similar to that reported in the bovine endothelial cells. Our present study favors the idea that TNF may have an inhibitory action on angiogenesis. Sato and Rifkin [30] have recently reported that TGF-/3 inhibits the migration and u-PA production of bovine capillary endothelial cells. This action of TGF-fl in the bovine endothelial cells appears to mimic that of TNF in the HME cells. Consistent with our present study, TNF has more inhibitory effect on angiogenesis of bovine capillary endothelial cells in the presence of an angiogenic factor, bFGF, than in its absence [31]. Other studies have, however, reported angiogenic activity by TNF alone on bovine capillary endothelial cells in vitro [32] as well as on either rabbit cornea or chick chorioallantoic membrane in uiuo [33]. It remains to be studied why TNF is angiogenie for some cells and anti-angiogenic for others. The angiogenic effect of TNF per se might be due to the enhanced production of an angiogenic factor(s) from TNF-treated endothelial cells since TNF induces various growth factors and cytokines in endothelial cells (IL-l, IL-6, PGE,, GM-CSF, etc.) [14,34-361. This possibility has not been verified. Further study should be required to understand the underlying molecular mechanism for modulation of proteolysis by TNF and/or EGF in relation to cell migration of the human endothelial cells.
580
MAWATARI
We thank Dr. H. K. Kleinman (NIDR, NIH, MD) for critically reading of this manuscript and invaluable comments. We also thank Drs. M. Kobayashi, Y. Uchida, and their colleagues (Oita Medical School) for supplying us with omental adipose tissue.
REFERENCES Moscatelli, D. A., Presta, M., and R&in, D. B. (1986) Proc. Natl. Acad. Sci. USA 83, 2091. 2. Mignatti, P., Tsuboi, R., Robbins, E., and Rifkin, D. B. (1989) J. Cell Biol. 108, 671. 3. Presta, M., Moscatelli, D., Joseph-Silverstein, J., and Rifkin, D. B. (1986) Mol. Cell. Biol. 6, 4060. 4. Werb, Z., Banda, M. J., and Jones, P. A. (1980) J. Exp. Med.
ET AL. 18 19
10. 11. 12. 13.
Loskutoff, D. J., Linders, M., Keijer, J., Veerman, H., Van Heerikhuizen, H., and H. Pannekoek. (1987) Biochemistry 26,3763. Sprengers, E. D., Printer, H. M., Kooistra, T., and Van Hinsbergh, V. W. (1985) J. Lab. Clin. Med. 105, 751. Carswell, E. A., Old, L. J., Kassel, R. L., Green, S., Fiore, N., and Williamson, B. (1975) Proc. Natl. Acad. Sci. USA 72, 3666. Sato, N., Goto, T., Haranaka, K., Satomi, N., Nariuchi, H., Mano-Hirano, Y., and Sawasaki, Y. (1986) J. N&l. Cancer Inst.
76,1113. 14. Mawatari,
M., Kohno, K., Mizoguchi, H., Matsuda, T., Asoh, K., Damme, J. V., Welgus, H. G., and Kuwano, M. (1989) J. Immunol. 143, 1619. 15. Mizoguchi, H., Uchiumi, T., Ono, M., Kohno, K., and Kuwano, M. (1990) Biochim. Biophys. Acta 1052, 475. 16. Chomczynski, P., and Sacchi, N. (1987) Anal. Biochrm. 162,
156. 17. Asoh, K., Watanabe, Y., Mizoguchi, Kohno, K., and Kuwano,
H., Mawatari, M., Ono, M., M. (1989) Biochem. Biophys. Res. Com-
mun. 162,794. Received June 6, 1990 Revised version received September
24, 1990
Moscatelli,
D., and Rifkin,
G. D., and Moses, H. L.
D. B. (1988) Biochim.
Biophys. Acta
948,67. 20. Ginsburg,
D., Zeheb, Z., Yang, A., Rotterty, M., Andreaseu, P. A., Nielsen, L., Dan, K., Lebo, V., and Gelehrter, T. D. (1986) J. Clin. Inucst. 78, 1673.
1.
152,134O. 5. Mullins, D. E., and Rohrlich, S. T. (1983) Biochim. B&&s. Acta 695,177. 6. Pepper, M. S., Vassalli, J. D., Montesano, R., and Orci, I,. (1987) J. Cell Biol. 105, 2535. 7. Saksela, 0. (1985) Biochim. Biophys. Acta 823, 35. 8. Sato, Y., and R&in, D. B. (1988) J. Cell Biol. 107, 1199. 9. Emeis, J. J., and Kooistra, T. (1986) J. Exp. Med. 163, 1260.
Coffey, R. J., Jr., Leaf, E. B., Shipley, (1987) J. Cell. Physiol. 132, 143.
21.
Montesano, R., Vassalli, J-D., Baird, A., Gillemin, I,. (1986) Proc. Nat/. Acad. Sci. lJSA 83, 5968.
R., and Orci,
22. Hayek, A., Culler, F. I,., Beattie, G. M., Lopez, A. D., Cuevas, P., and Baird, A. (1987) Biochem. Biophys. Res. Commun. 147,876. 23. Schreiber, A. B., Winkler, M. E., and Derynck, R. (1986) Science 232,125O. 24. Mano-Hirano, Y., Sato, N., Sawasaki, Y., and Goto, T. (1986) Proc. Jpn. Acad. 62, 235.
25. Zetter, B. R. (1980) Nature (London) 285, 41. 26. Gross, J. L., Moscatelli, D., and Rifkin, D. B. (1983) Proc. Natl. Acad. Sci. USA 80, 2623.
27. Moscatelli, D. A., Rifkin, D. B., and Jaffe, F. A. (1985) Exp. Cell. RPS. 156,379. 28. Watanabe, N., Niitsu, Y., IJmeno, H., Kuriyama, H., Neda, H., Yamauchi, 48, 2179. 29. 30. 31.
N., Maeda, M., and IJrushizaki,
I. (1988) Cancer Res.
Sawdey, M., Podor, T. J., and Loskutoff, D. J. (1989) J. Biol. Chem. 264, 10,396. Sato, Y., and Rifkin, D. B. (1989) J. Cell Biol. 109, 309. Schweigerer, L., Malerstein, B., and Gospodarowicz, B&hem. Bioph-ys. Res. Commun. 143, 997.
D. (1987)
32. Frater-Schroder, Bohlen,
M., Risau, W., Hallmann, R., Gautschi, P., and P. (1987) Proc. Natl. Acad. Sci. USA 84, 5277.
33. Leibovich,
S. J., Polverini, P. J., Shepard, H. M., Wiseman, D. M., Shively, V., and Nuseir, N. (1987) Nature (London) 329,
630. 34. Bachwich,
P. R., Chensue, S. W., Larrick, J., and Kunkel, (1986) B&hem. Biophys. Res. Commun. 136, 94.
S. L.
35. Broudy, V. C., Kaushansky, Adamson,
K., Segal, G. M., Harlan, J. M., and J. W. (1986) Proc. Natl. Acad. Sci. 1JSA 83, 7467.
36. Dayer, ,J. M., Beutler, 162.2163.
B., and Cerami,
A. (1985) J. Exp. Med.
