JOURNAL OF BONE AND MINERAL RESEARCH Volume 7, Number 12, 1992 Mary Ann Lieberl, Inc., Publishers
The Effect of Transforming Growth Factor p on the Plasminogen Activator Activity of Normal Human Osteoblast-like Cells and a Human Osteosarcoma Cell Line MG-63 FIONA W. FAWTHROP,’ BABATUNDE 0. OYAJOBI,’ ROWENA A.D. BUNNING,2 and R. GRAHAM G. RUSSELL’
ABSTRACT Transforming growth (TGF-0) has been proposed to have a role in bone remodeling by affecting the differentiation and activity of osteoblasts and osteoclasts and by inhibiting the production of proteinases, such as plasminogen activators (PAS). Studies on PAS have largely been based on data from nonhuman and fetal cell lines, however. The purpose of this study was to investigate the effect of TGF-0 on the PA activity of normal human osteoblast-like cells and to compare this with its action on the human osteosarcoma cell line MG-63. The action of interleukin-10 (IL-10) was also assessed because it has been shown to increase PA activity in other connective tissue cell types. Normal osteoblast-like cells had low to undetectable basal urokinase (uPA) and tissue plasminogen activator (tPA) activity, which was significantly stimulated by TGF-0,. This action was shown to be dependent on transcription and new protein synthesis. TGF-0, had a similar action. IL-10 did not stimulate PA activity. In contrast, the MG-63 cell line had high basal tPA and uPA activities. TGF-0, decreased basal PA activity, the effect being most marked for uPA activity. IL-10 stimulated uPA and tPA activity. TGF-0, inhibited IL-10-stimulated uPA activity, but the effect on tPA was more variable. This study has shown that TGF-0 has opposite effects on the PA activity of the two osteoblast-like cell types studied. Care must therefore be used before extrapolating data from one cell type to another. It is suggested that under certain circumstances TGF-0 may be involved in bone resorption by modulating PA activity.
INTRODUCTION HE STRUCTURE OF THE SKELETON is maintained by a continuous process of remodeling of the calcified matrix by the activity of bone cells. Bone remodeling begins with a phase of bone resorption mediated by osteoclasts, which is then followed by bone formation carried out by osteoblasts.(l.*L In normal bone these two events are closely linked or “coupled.” The exact mechanism of this coupling has not yet been fully established, but it may involve control by locally derived cytokines and growth factors, such as transforming growth factor fi.
T
Transforming growth factors PI and fi, (TGF-fi, and TGF-fi,) are pluripotent cytokines that have been isolated in large amounts from b ~ n e . ‘ ~ . ‘TGF-P, ’ is the dominant form, TGF-P, representing only about 20% of the total. Both are present as high-molecular-weight complexes that are inactive.‘2) TGF-fi is activated at low This may be important in the region of the ruffled border of the osteoclast, where the pH is thought to be very low.(” In addition, TGF-fi is activated by the direct action of the osteoclast.(8’ TGF-fi has been proposed to have a role in bone remodeling in several ways. TGF-fi may have stimulatory or inhibitory effects on bone resorption depending
‘Department of Human Metabolism and Clinical I3iochemistry. University of Sheffield Medical School, England. ’Division of Biomedical Sciences, Sheffield City P’olytechnic, England.
1363
FAWTHROP ET AL.
1364 on the dose and tissue used. TGF-8 may inhibit osteoclast recruitment. It can also act as a mitogen for osteoblasts and stimulate the production of extracellular matrix and new bone in vitro. TGF-8 may also affect bone remodeling by inhibiting the release of proteinases, such as plasminogen activators, increasing their naturally occurring inhibitors, and by stimulating the production of the extracellular matrix.‘’’ However, these ideas have largely been based on work carried out on either fetal or transformed cells and nonhuman cell lines. Plasminogen activators (PAS) are serine proteinases that convert the zymogen plasminogen to plasmin. They may also have a direct role in the degradation of extracellular matrix gly~oprotein.(~) Two forms of PAS are found in humans: urokinase (uPA) and tissue plasminogen activator (tPA). These are products of different genes and have different immunologic reactivity and molecular weights. ( l o ) The activation of tPA, but not uPA, is fibrin dependent, and the former is thought to be important in fibrinolysis. ( L 1 , 1 2 )Plasmin is a broad-spectrum serine proteinase with a similar range of substrate specificities to trypsin. ( l o ) It is known to degrade proteoglycans and extracellular matrix glycoproteins, is able to activate latent collagenase,(ls) and can activate the inactive form of PAS.‘”) Plasmin has been shown to activate 50-60% of the acid-activatable pool of latent TGF-8 secreted by fibroblasts and latent TGF-8 from human platelets. (I6) Further evidence for this action comes from work showing that, during coculture of bovine endothelial cells with pericytes, the activation of TGF-8 can be diminished by the addition of a specific plasmin inhibitor to the coculture system. (I7) The purpose of this study was to investigate the effect of TGF-8 on the PA activity of normal human osteoblast-like cells and to compare this with its action on the human osteosarcoma cell line MG-63. It was shown previously that normal human bone cell cultures similar to those referred to in this paper express markers of the osteoblast phenotype, such as alkaline phosphatase and osteocalcin, in response to 1,25-dihydroxyvitamin D, [ 1,25-(OH),D3]. After treatment with lo-’ M 1,25-(OH),D, for 48 h, about 55-6570 of the cells in each culture express alkaline phosphatase, osteocalcin, or both antigens simultaneously, the percentage of cells positive for either or both of these proteins varying from one donor to another (Thavarajah, personal communication). MG-63 cells have been reported to synthesize type I collagen, express tissue-nonspecific alkaline phosphatase activity, which is further enhanced by 1,25-(OH)’D3, and synthesize osteocalcin in response to 1,25-(OH),D3.(18-20) All of these indicate that this cell line represents osteoblasts with a mature phenotype. By using the former cell type it was hoped to gain information that would be more relevant to the mechanisms controlling bone turnover in the normal human adult skeleton.
MATERIALS AND METHODS The following tissue culture materials were obtained from GIBCO (Paisley, UK): Eagle’s minimum essential medium with Earle’s salts, Dulbecco’s modified essential
medium, penicillin, streptomycin, glutamine, and Dulbecco’s Ca2+- and Mg2+-free phosphate-buffered saline. Fetal calf serum was obtained from Northumbria Biologicals (Cramlington, UK). Falcon vented tissue culture dishes (9 cm), 75 cm2 flasks, and 24-well plates were from Becton Dickinson UK (Oxford, UK); 96-well (flat-bottomed) plates were obtained from Corning (Stone, UK). tPA (Bowes melanoma 2 chain), uPA (high molecular weight), and cyanogen bromide fibrinogen fragments from American Diagnostica were purchased from Ortho Diagnostic Systems (High Wycombe, UK). Human plasminogen from Kabi Diagnostics was obtained from Quadratech (Epsom, UK). Casein (Hammarsten) and N, N’-methylenebisacrylamide (Electron) were purchased from BDH Chemicals (Poole, UK). Sodium dodecyl sulfate and acrylamide, both electrophoretic grade, were from Bio-Rad Laboratories (Watford, UK). High-molecular-weight protein standard mixture (SDSdH), 5-5’-dithiobis-(2-nitrobenzoic acid), and n-a-CBz-L-lysine thiobenzyl ester hydrochloride were obtained from Sigma Chemical Company (Poole, UK). Recombinant human TGF-0, (rhTGF-8,) was a gift from Dr. E. Amento, Genentech, Inc. (San Francisco, CA), recombinant human interleukin-18 (rhIL18) was given by Dr. A. Shaw, Glaxo Group Research, and rhTGF-(3, was a gift from Sandoz AG (Basel, Switzerland). All other chemicals were from BDH Chemicals or Sigma Chemicals and were of the purest grades available.
Tissue culture techniques Human Osteoblast-like Cells: Human trabecular bone was obtained from the tibia1 and femoral condyles of above-knee amputation specimens. Bone explants were cleaned of any surrounding soft tissue and marrow and then transferred to 9 cm tissue culture dishes as previously described. I l l ) Primary monolayer cultures were grown to confluence in Eagle’s minimum essential medium (EMEM) containing 10% fetal calf serum, 100 U/ml of penicillin, 100 pg/ml of streptomycin, and 2 mM glutamine in a humidified atmosphere of 5% CO, and 95% air at 37°C. For experimental purposes cells were passaged into 24-well plates (surface area per well, 200 mm’) and cultured for 3-4 days before use. After washing the cells three times with phosphate-buffered saline (PBS), test substances were added in serum-free medium containing 0.1% bovine serum albumin. For each experiment, osteoblasts from a single patient were used, control and test conditions being set up in quadruplicate wells. Each set of experiments were carried out on first passage cultures from at least three different donors.
Human Osteosarcoma Cell Line MG-63: MG-63 cells were cultured in 75 cm’ flasks with Dulbecco’s modified minimum essential medium containing 10% fetal calf serum, 2 mM glutamine, and antibiotics (see earlier). For experimental purposes cells were prepared as for normal osteoblasts.
TGF-8 AND PLASMINOGEN ACTIVATOR
TGFnr i M l
1365
TGFRI iMi IL-lR IOU rnl
i Bi
at 4°C until use (within 6 h). Plasminogen activator activity assay was carried out essentially according to the method of Leprince et al.'zz) In outline, this is a two-stage colorimetric assay during the first stage of which plasminogen activators in the sample convert plasminogen to plasmin. Plasmin activity generated in step 1 is then determined by the color change of the chromogenic substrate N-a-CBz-L-lysine thiobenzyl ester hydrochloride. Different incubation times were used depending upon the amount of PA activity thought to be present. For levels between 0.1 and 1 mIU per 5 pl, step 1 incubation time was 1 h and step 2 up to 4 h (long incubation assay); for levels between 10 and 100 mIU per 5 pl they were 15 minutes and up to 1 h, respectively (short incubation). tPA and uPA activity can be differentiated by the preincubation of samples with fibrin fragments, giving a measure of total PA activity, and subtracting from this the PA activity determined without fibrin fragments, that is, the uPA activity.
Characterization of plasminogen activators by sodium dodecyl sulfate-polyacrylamide gel electrophoresis
IL I n
lOUrnl
FIG. 1. The effect of rhTGF-0, (10-7-10-9M) on tPA (A) and uPA (B) activity as measured in conditioned media and cell lysates, respectively, of human osteoblast-like cells and its interaction with rhIL-10. Human osteoblast-like cells were passaged into 24-well plates and grown to confluence in Eagle's minimum essential medium with Earle's salts containing 100 U/ml of penicillin, 100 pg/rnl of streptomycin, 2 mM glutamine, and 10% fetal calf serum. The cells were washed three times with phosphate-buffered saline and 500 pl incubation medium consisting of dilutions of rhTGF-0, in EMEM containing 0.1% bovine serum albumin with and without rhIL-10 (10 U/ml) added to each well. After a 24 h incubation period, the conditioned medium was removed, cells washed three times with PBS, 100 pi Triton and PBS added to each well, and cell layers scraped into this. Conditioned media and cell lysates were assayed for PA activity. The results for PA activity are expressed as mIU per 5 p1 sample, mean f standard error of the mean (n = 4). **Significantly different from control, p c 0.01, Scheffe's multiple-range analysis.
Plasminogen activator assay Bone cells were incubated for 24 h in the presence of rhTGF-0, or rhTGF-& alone and in combination with rhIL-10. The conditioned media were removed; aliquots were stored at 4°C and then assayed within 6 h. The cell layers were thoroughly washed with cold phosphate-buffered saline. Phosphate-buffered saline containing 0.1 Yo (vol/vol) of Triton X-100 (100 pl) was added to each well and the cells scraped into this solution. Lysates were stored
Bone cells were incubated for 24 h in the presence of rhTGF-0, or rhIL-10 alone and in combination. The conditioned media were removed; aliquots were stored at 4°C and then used for gels within 24 h. The cell layers were washed three times with cold phosphate-buffered saline. Phosphate-buffered saline containing 0.1 % (vol/vol) of Triton X-100 (100 pl) was then added to each well and the cells scraped into this solution. Lysates were stored as for culture media and used within 24 h. Sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis was carried out essentially according to the method of Roche et a1.Iz3) A 4% (wt/vol) acrylamide stacking gel and 9% (wt/vol) resolving gel were used. Two gels were run in parallel, one of the resolving gels containing casein (2 mg/ml) and human plasminogen (6 pg/ml) and the other containing casein only, to show that the activity seen was plasminogen dependent. Samples for electrophoresis in 0.0625 M TrisHCI (pH 6.8) containing 2.5% (wt/vol), 8% SDS (vol/ vol), glycerol, and bromophenol blue were applied to gels without boiling. The high-molecular-weight standards included phosphorylase b (97,400), bovine serum albumin (66,000), ovalbumin (45,000), and carbonic anhydrase (29,000). Gels were run overnight and then washed in 2.5% (vol/vol) aqueous Triton X-100 solution for I h to remove the SDS. For the development of plasminogen activator bands, the gels were incubated at 37°C for 24 h in 0.1 M Tris-HC1 (pH 8.1). The gels were stained with Coomassie brilliant blue R250. Plasminogen activator activity appeared as clear bands against a dark background where plasminogen activators had converted plasminogen to plasmin and the plasmin had degraded the casein in the gel.
Statistical analysis For a given experiment the mean and standard error of the mean were calculated for each group. Analysis of vari-
FAWTHROP ET AL.
1366
ance was performed on all data, followed by Scheffe’s multiple-range analysis. Calculations were made using the Statview 512+ application on an Apple Macintosh computer.
Protein assay Total proteins were assayed using a modification of the method of Oyama and Eagle.”‘’
RESULTS Effect of rhTGF-0, on the PA activity of normal human osteoblast-like cells Human osteoblast-like cells produce low or undetectable, that is, ~ 0 . 1 mIU per 5 pl, levels of both tPA and uPA activity when cultured under serum-free conditions. rhlL-l@ had no stimulatory effect on tPA activity and stimulated uPA activity in only one experiment in which basal levels of uPA were high. rhTGF-PI, when used at concentrations 10-9-10-7M, caused a significant increase in both tPA and uPA activity (Fig. 1). This was reproducible using cells from several different patient sources. Higher levels of uPA activity were detected in cell lysates than conditioned media, and vice versa for tPA activity. The maximal stimulation recorded for each was around 1 mIU per 5 PI. There was no obvious difference in the response to doses of rhTGF-@,between and lo-’ M. The stimulation of uPA activity by rhTGF-@, was confirmed by SDS-polyacrylamide gel electrophoresis (Fig. 2); however, tPA activity was seen only as a very faint band. t P A does not appear clearly in the zymograms, possibly because
within the system there are no molecules that bring about its activation. When rhTGF-@, was used at lower concentrations (Fig. 3) the significant stimulatory effect on both M but maintained uPA and tPA was lost at a dose of at a dose of lo-” M, the response being comparable to that of M rhTGF-@,. The effect of rhTGF-@, and IL-I@in combination o n osteoblast P A was variable. For t P A activity the response to the rhTGF-PI and IL-10 was not significantly different from that of the corresponding dose of rhTGF-@, when used alone. For uPA activity, in two of four experiments the combination caused a significant increase in activity compared to rhTGF-@, alone, differences from controls not being significant in the other cases. Experiments were carried out to assess the effect of indomethacin ( M), cycloheximide (1 pg/ml), and actinomycin D (lo+’ and M) on the basal and rhTGF-@,stimulated P A activity of human osteoblast-like cells (results for t P A activity are shown in Table 1). In no case did indomethacin, cycloheximide, or actinomycin D alone stimulate P A activity. Indomethacin did not significantly affect the action of rhTGF-@,, suggesting that it is not mediated via the prodcution of prostaglandin E. Actinomycin D blocked the stimulatory action of rhTGF-@, on PA activity. However, the inhibitory effect of cycloheximide was not always consistent, possibly a result of the variation in penetration of cycloheximide into the cells.
Effect of rhTGF-0, on the PA activity of normal human osteoblast-like cells Recombinant human TGF-@, stimulated the production of both tPA and uPA by human osteoblast-like cells (Fig. 4). Unlike experiments carried out with rhTGF-@,, the
FIG. 2. SDS-polyacrylamide gel containing plasminogen and casein showing the plasminogen activator species present in the cell lysates of human osteoblast-like cells treated with rhTGF-@, or rhIL-l@ alone or in combination. Human osteoblast-like cells were passaged into 24-well plates and grown to confluence in Eagle’s minimum essential medium with Earle’s salts containing 100 U/ml of penicillin, 100 pg/ml of streptomycin 2 mM glutamine, and 10% fetal calf serum. The cells were washed three times with phosphate-buffered saline and 500 pl incubation medium consisting of dilutions of rhTGF-@,in EMEM containing 0.1% bovine serum albumin with and without rhIL-l@(10 U/ml) added to each well. After a 24 h incubation period, the medium was removed, cells washed three times with PBS, 100 pl Triton and PBS added to each well, and cell layers scraped into this. Plasminogen activator species present in the cell lysates were assessed by zymography. Lane 1, tPA (Bowes melanoma 2 chain); lane 2, uPA (high molecular weight); lane 3, control; lane 4, M; lane 7, rhTGF-@, M , + rhIL-l@(10 U/ml); M; lane 6 , rhTGF-@, rhIL-l@(10 U/ml); lane 5 , rhTGF-@, and lane 8, rhTGF-0, M, + rhIL-10 (10 U/ml). N o bands of activity developed in gels without plasminogen (not shown).
1367
TGF-0 AND PLASMINOGEN ACTIVATOR
In'
"I 0,
01
03 0: 01
OD . l l 0 U ,"I/
TGFOiMi
. T G F O IMi
rhIL-10 consistently stimulated tPA and uPA activity, and this was statistically significant in most cases. As shown in Fig. 5 , rhTGF-0, decreased levels of P A activity below those of the control. This was a consistent response and reached statistical significance in three of four cases for uPA, but only in one experiment at one dose for tPA. This suggests that rhTGF-0, has a greater inhibitory effect on the basal uPA activity than the tPA activity of these cells. The inhibitory effect of rhTGF-P, on uPA activity was confirmed by zymography (Fig. 6). There was no difference in the effect of the varying concentrations of rhTGF0, used from lo-' to M (data for the lowest doses not shown). rhTGF-0, significantly inhibited rhIL-10-stimulated uPA activity in three of four experiments to levels below those of the controls. In the fourth it decreased the activity, but this did not reach statistical significance. Its effect on rhll-10-stimulated tPA activity was variable, sometimes stimulating and sometimes decreasing tPA activity.
DISCUSSION
rhll-lO , I O U m1
FIG. 3. The effect of rhTGF-0, used at low concentrations on the tPA (A) and uPA (B) activity as measured in the conditioned media and cell lysates, respectively, of human osteoblast-like cells and its interaction with rhIL-10. Human osteoblast-like cells were passaged into 24-well plates and grown to confluence in Eagle's minimum essential medium with Earle's salts containing 100 Uiml of penicillin, 100 pg/ml of streptomycin, 2 mM glutamine, and 10% fetal calf serum. The cells were washed three times with phosphate-buffered saline and 500 pl incubation medium consisting of dilutions of rhTGF-PI, in E,MEM containing 0.1070 bovine serum albumin with and without rhIL-10 (10 U/ml) added to each well. After a 24 h incubation period, the conditioned medium was removed, cells were washed three times with PBS, 100 pI Triton and PBS added to each well, and cell layers scraped into this. Conditioned media and cell lysates were assayed for PA activity. The results for PA activity are expressed as mlU per 5 pl sample, mean f standard error of the mean ( n = 4). **Significantly different from control, p < 0.01, Scheffe's multiple-range analysis. level of stimulation did not reach statistical significance in all cases. As for rhTGF-P,, the effect of rhTGF-0, and rhIL-10 in combination on osteoblast PA activity was variable.
Effect of rhTGF-0, on the PA activity of the human osteosarcoma cell line MG-63 These human osteosarcoma cells produced high basal levels of both uPA and tPA, in excess of 100-fold that recorded for normal human osteoblast-like cells. Higher levels of t P A activity were detected in the conditioned medium than in cell lysates, and vice versa for uPA activity.
Results from this study show that normal adult human osteoblast-like cells have low basal levels of PA activity and are capable of producing both u P A and tPA. Low P A levels may reflect the maturity of the cells, since it has been shown in fetal rat osteoblasts that the expression of P A activity tends to fall with It is of interest that both types of PAS were detected. This contrasts with several other osteoblast-like cell types, including UMR-201 and UMR-106-01, in which the PA activity has been shown to be exclusively due to tPA, with no detectable uPA by zymography or with specific antisera.'261Such a discrepancy probably reflects phenotypic differences between these cell types rather than variation in sensitivities and specificities of the assays used or may reflect heterogeneity of the cultures. uPA was detected primarily in the cell lysates and to a much lesser extent in the conditioned medium. This is not surprising because many cell lines have cell surface receptors for UPA.(27-29'The proenzyme is bound to the receptor and is converted to the active molecule, which is then able to bind the zymogen plasminogen before activating it to pla~min.'~O) rhTGF-0, significantly stimulated both the uPA and tPA activity of normal adult human osteoblast-like cells when used over a concentration range of 10-7-10-12M. However, this effect was lost at M. In this colorimetric assay the net change in PA activity is measured. It is conceivable that rhTGF-0, caused an increase in plasminogen activator inhibitor (PAI) activity, but obviously the major effect of rhTGF-0, was an increase in P A activity. Indomethacin did not inhibit the stimulatory effect of rhTGF-PI, suggesting that its mechanism of action is not mediated via prostaglandin synthesis. The ability of actinomycin D and cycloheximide to inhibit P A activity indicates that both transcription and new protein synthesis, respectively, must occur in response to rhTGF-0,. The increased activity cannot solely be attributed to activation of the proenzyme. New protein synthesis is involved in the action of
FAWTHROP ET AL.
1 O U ml
TGFR 2
(MI
TGFO 2 ( M i iL-in lOUml
IL 1R I O U rnl
10 9
!O TGFOl !MI
10
10 3
10
10
TGFOI (MI it-in IOU rnl
"1
TGFO 2 !MI
TGFR *(MI IL-in
!OU ",I
The effect of rhTGF-0, on tPA (A) and uPA (B) activity as measured in conditioned media and cell lysates, respectively, of human osteoblast-like cells and its interaction with rhIL-l@. Human osteoblast-like cells were passaged into 24-well plates and grown to confluence in Eagle's minimum essential medium with Earle's salts containing 100 U/ml of penicillin, 100 pg/ml of streptomycin, 2 mM glutamine, and 10% fetal calf serum. The cells were washed three times with phosphate-buffered saline and 500 11 incubation medium coinsisting of dilutions of rhTGF-@, in EMEM containing 0.1 Vo bovine serum albumin with and without rhIL-10 (10 U/ml) added to each well. After a 24 h incubation period, the conditioned medium was removed, cells washed three times with PBS, 100 pl Triton and PBS added to each well, and cell layers scraped into this. Conditioned media and cell lysates were assayed for PA activity. Results for PA activity are expressed as mIU standard error of the mean (n = per 5 p1 sample, mean 4). **Significantly different from control, p < 0.01, Scheffe's multiple-range analysis.
*
rhTGF-@, on other cell types, such as skin fibroblasts, in which the increase in uPA activity is inhibited by cycloheximide. C3') The action of rhTGF-0, was similar to that of rhTGFp,. TGF-p, and TGF-0, share 70% homology and in most systems have a common action.(3z)As of yet no major difference has been reported in their action on bone. (s.6,33) That the stimulatory effects of TGF-@, and TGF-@, were
FIG. 5. The effect of rhTGF-@,on tPA (A) and uPA (B) activity as measured in conditioned media and cell lysates, respectively, of the human osteosarcoma cell line MG-63 and its interaction with rhIL-l@. Human osteosarcoma cells were passaged into 24-well plates and grown to confluence in Dulbecco's modified essential medium (DMEM) containing 100 U/ml of penicillin, 100 pg/ml of streptomycin 2 mM glutamine, and 10% fetal calf serum. The cells were washed three times with phosphate-buffered saline and 500 p1 incubation medium consisting of dilutions of rhTGF-@, in DMEM containing 0.1 Vo bovine serum albumin, with and without rhIL-10 added to each well. After a 24 h incubation period, the conditioned medium was removed, cell layers washed three times with PBS, 0.1% Triton and PBS (100 pl) added to each well, and the cells scraped into this. Conditioned media and cell lysates were then assayed for PA activity. Results for PA activity are expressed as mIU per 5 p1 sample, mean f standard error of the mean (n = 4). **Significantly different from control, p < 0.01, Scheffe's multiple-range analysis.
due to TGF-@per se was indicated by similar effects of the two different sources of TGF-@. Recombinant human IL-10 did not stimulate the PA activity of normal human osteoblast-like cells. This finding is in accordance with that of Evans,(34)who showed that IL10 can only stimulate the PA activity of human osteoblastlike cells in the presence of serum. Under serum-free conditions a small reduction in PA activity was demonstrated.
TGF-0 AND PLASMINOGEN ACTIVATOR
1369
TABLE1. EFFECTOF INDOMETHACIN, CYCLOHEXIMIDE, AND ACTINOMYCIN D ON THE TPA ACTNITY (MEASURED IN CONDITIONED MEDIA)OF HUMANOSTEOBLAST-LIKE CELLSSTIMULATED AND UNSTIMULATED BY TGF-/3,a tPA activity (mIUper 5 PI)
Experimental conditions ~~~
Control
TGF-0(10-7 M) Indomethacin (10 -6 M) TGF-0 M) + indomethacin (10-6 M) Cycloheximide (1 pg/ml) TGF-0 M) + cycloheximide (1 a d m u Actinomycin D M 10-7 M TGF-0 (lo-’ M) + actinomycin D (10-7 M)
~
0.184 (0.077) 1 .ooo(0) 0.056 (0.056) 0.!2 (0.1)
The Effect of Transforming Growth Factor p on the Plasminogen Activator Activity of Normal Human Osteoblast-like Cells and a Human Osteosarcoma Cell Line MG-63 FIONA W. FAWTHROP,’ BABATUNDE 0. OYAJOBI,’ ROWENA A.D. BUNNING,2 and R. GRAHAM G. RUSSELL’
ABSTRACT Transforming growth (TGF-0) has been proposed to have a role in bone remodeling by affecting the differentiation and activity of osteoblasts and osteoclasts and by inhibiting the production of proteinases, such as plasminogen activators (PAS). Studies on PAS have largely been based on data from nonhuman and fetal cell lines, however. The purpose of this study was to investigate the effect of TGF-0 on the PA activity of normal human osteoblast-like cells and to compare this with its action on the human osteosarcoma cell line MG-63. The action of interleukin-10 (IL-10) was also assessed because it has been shown to increase PA activity in other connective tissue cell types. Normal osteoblast-like cells had low to undetectable basal urokinase (uPA) and tissue plasminogen activator (tPA) activity, which was significantly stimulated by TGF-0,. This action was shown to be dependent on transcription and new protein synthesis. TGF-0, had a similar action. IL-10 did not stimulate PA activity. In contrast, the MG-63 cell line had high basal tPA and uPA activities. TGF-0, decreased basal PA activity, the effect being most marked for uPA activity. IL-10 stimulated uPA and tPA activity. TGF-0, inhibited IL-10-stimulated uPA activity, but the effect on tPA was more variable. This study has shown that TGF-0 has opposite effects on the PA activity of the two osteoblast-like cell types studied. Care must therefore be used before extrapolating data from one cell type to another. It is suggested that under certain circumstances TGF-0 may be involved in bone resorption by modulating PA activity.
INTRODUCTION HE STRUCTURE OF THE SKELETON is maintained by a continuous process of remodeling of the calcified matrix by the activity of bone cells. Bone remodeling begins with a phase of bone resorption mediated by osteoclasts, which is then followed by bone formation carried out by osteoblasts.(l.*L In normal bone these two events are closely linked or “coupled.” The exact mechanism of this coupling has not yet been fully established, but it may involve control by locally derived cytokines and growth factors, such as transforming growth factor fi.
T
Transforming growth factors PI and fi, (TGF-fi, and TGF-fi,) are pluripotent cytokines that have been isolated in large amounts from b ~ n e . ‘ ~ . ‘TGF-P, ’ is the dominant form, TGF-P, representing only about 20% of the total. Both are present as high-molecular-weight complexes that are inactive.‘2) TGF-fi is activated at low This may be important in the region of the ruffled border of the osteoclast, where the pH is thought to be very low.(” In addition, TGF-fi is activated by the direct action of the osteoclast.(8’ TGF-fi has been proposed to have a role in bone remodeling in several ways. TGF-fi may have stimulatory or inhibitory effects on bone resorption depending
‘Department of Human Metabolism and Clinical I3iochemistry. University of Sheffield Medical School, England. ’Division of Biomedical Sciences, Sheffield City P’olytechnic, England.
1363
FAWTHROP ET AL.
1364 on the dose and tissue used. TGF-8 may inhibit osteoclast recruitment. It can also act as a mitogen for osteoblasts and stimulate the production of extracellular matrix and new bone in vitro. TGF-8 may also affect bone remodeling by inhibiting the release of proteinases, such as plasminogen activators, increasing their naturally occurring inhibitors, and by stimulating the production of the extracellular matrix.‘’’ However, these ideas have largely been based on work carried out on either fetal or transformed cells and nonhuman cell lines. Plasminogen activators (PAS) are serine proteinases that convert the zymogen plasminogen to plasmin. They may also have a direct role in the degradation of extracellular matrix gly~oprotein.(~) Two forms of PAS are found in humans: urokinase (uPA) and tissue plasminogen activator (tPA). These are products of different genes and have different immunologic reactivity and molecular weights. ( l o ) The activation of tPA, but not uPA, is fibrin dependent, and the former is thought to be important in fibrinolysis. ( L 1 , 1 2 )Plasmin is a broad-spectrum serine proteinase with a similar range of substrate specificities to trypsin. ( l o ) It is known to degrade proteoglycans and extracellular matrix glycoproteins, is able to activate latent collagenase,(ls) and can activate the inactive form of PAS.‘”) Plasmin has been shown to activate 50-60% of the acid-activatable pool of latent TGF-8 secreted by fibroblasts and latent TGF-8 from human platelets. (I6) Further evidence for this action comes from work showing that, during coculture of bovine endothelial cells with pericytes, the activation of TGF-8 can be diminished by the addition of a specific plasmin inhibitor to the coculture system. (I7) The purpose of this study was to investigate the effect of TGF-8 on the PA activity of normal human osteoblast-like cells and to compare this with its action on the human osteosarcoma cell line MG-63. It was shown previously that normal human bone cell cultures similar to those referred to in this paper express markers of the osteoblast phenotype, such as alkaline phosphatase and osteocalcin, in response to 1,25-dihydroxyvitamin D, [ 1,25-(OH),D3]. After treatment with lo-’ M 1,25-(OH),D, for 48 h, about 55-6570 of the cells in each culture express alkaline phosphatase, osteocalcin, or both antigens simultaneously, the percentage of cells positive for either or both of these proteins varying from one donor to another (Thavarajah, personal communication). MG-63 cells have been reported to synthesize type I collagen, express tissue-nonspecific alkaline phosphatase activity, which is further enhanced by 1,25-(OH)’D3, and synthesize osteocalcin in response to 1,25-(OH),D3.(18-20) All of these indicate that this cell line represents osteoblasts with a mature phenotype. By using the former cell type it was hoped to gain information that would be more relevant to the mechanisms controlling bone turnover in the normal human adult skeleton.
MATERIALS AND METHODS The following tissue culture materials were obtained from GIBCO (Paisley, UK): Eagle’s minimum essential medium with Earle’s salts, Dulbecco’s modified essential
medium, penicillin, streptomycin, glutamine, and Dulbecco’s Ca2+- and Mg2+-free phosphate-buffered saline. Fetal calf serum was obtained from Northumbria Biologicals (Cramlington, UK). Falcon vented tissue culture dishes (9 cm), 75 cm2 flasks, and 24-well plates were from Becton Dickinson UK (Oxford, UK); 96-well (flat-bottomed) plates were obtained from Corning (Stone, UK). tPA (Bowes melanoma 2 chain), uPA (high molecular weight), and cyanogen bromide fibrinogen fragments from American Diagnostica were purchased from Ortho Diagnostic Systems (High Wycombe, UK). Human plasminogen from Kabi Diagnostics was obtained from Quadratech (Epsom, UK). Casein (Hammarsten) and N, N’-methylenebisacrylamide (Electron) were purchased from BDH Chemicals (Poole, UK). Sodium dodecyl sulfate and acrylamide, both electrophoretic grade, were from Bio-Rad Laboratories (Watford, UK). High-molecular-weight protein standard mixture (SDSdH), 5-5’-dithiobis-(2-nitrobenzoic acid), and n-a-CBz-L-lysine thiobenzyl ester hydrochloride were obtained from Sigma Chemical Company (Poole, UK). Recombinant human TGF-0, (rhTGF-8,) was a gift from Dr. E. Amento, Genentech, Inc. (San Francisco, CA), recombinant human interleukin-18 (rhIL18) was given by Dr. A. Shaw, Glaxo Group Research, and rhTGF-(3, was a gift from Sandoz AG (Basel, Switzerland). All other chemicals were from BDH Chemicals or Sigma Chemicals and were of the purest grades available.
Tissue culture techniques Human Osteoblast-like Cells: Human trabecular bone was obtained from the tibia1 and femoral condyles of above-knee amputation specimens. Bone explants were cleaned of any surrounding soft tissue and marrow and then transferred to 9 cm tissue culture dishes as previously described. I l l ) Primary monolayer cultures were grown to confluence in Eagle’s minimum essential medium (EMEM) containing 10% fetal calf serum, 100 U/ml of penicillin, 100 pg/ml of streptomycin, and 2 mM glutamine in a humidified atmosphere of 5% CO, and 95% air at 37°C. For experimental purposes cells were passaged into 24-well plates (surface area per well, 200 mm’) and cultured for 3-4 days before use. After washing the cells three times with phosphate-buffered saline (PBS), test substances were added in serum-free medium containing 0.1% bovine serum albumin. For each experiment, osteoblasts from a single patient were used, control and test conditions being set up in quadruplicate wells. Each set of experiments were carried out on first passage cultures from at least three different donors.
Human Osteosarcoma Cell Line MG-63: MG-63 cells were cultured in 75 cm’ flasks with Dulbecco’s modified minimum essential medium containing 10% fetal calf serum, 2 mM glutamine, and antibiotics (see earlier). For experimental purposes cells were prepared as for normal osteoblasts.
TGF-8 AND PLASMINOGEN ACTIVATOR
TGFnr i M l
1365
TGFRI iMi IL-lR IOU rnl
i Bi
at 4°C until use (within 6 h). Plasminogen activator activity assay was carried out essentially according to the method of Leprince et al.'zz) In outline, this is a two-stage colorimetric assay during the first stage of which plasminogen activators in the sample convert plasminogen to plasmin. Plasmin activity generated in step 1 is then determined by the color change of the chromogenic substrate N-a-CBz-L-lysine thiobenzyl ester hydrochloride. Different incubation times were used depending upon the amount of PA activity thought to be present. For levels between 0.1 and 1 mIU per 5 pl, step 1 incubation time was 1 h and step 2 up to 4 h (long incubation assay); for levels between 10 and 100 mIU per 5 pl they were 15 minutes and up to 1 h, respectively (short incubation). tPA and uPA activity can be differentiated by the preincubation of samples with fibrin fragments, giving a measure of total PA activity, and subtracting from this the PA activity determined without fibrin fragments, that is, the uPA activity.
Characterization of plasminogen activators by sodium dodecyl sulfate-polyacrylamide gel electrophoresis
IL I n
lOUrnl
FIG. 1. The effect of rhTGF-0, (10-7-10-9M) on tPA (A) and uPA (B) activity as measured in conditioned media and cell lysates, respectively, of human osteoblast-like cells and its interaction with rhIL-10. Human osteoblast-like cells were passaged into 24-well plates and grown to confluence in Eagle's minimum essential medium with Earle's salts containing 100 U/ml of penicillin, 100 pg/rnl of streptomycin, 2 mM glutamine, and 10% fetal calf serum. The cells were washed three times with phosphate-buffered saline and 500 pl incubation medium consisting of dilutions of rhTGF-0, in EMEM containing 0.1% bovine serum albumin with and without rhIL-10 (10 U/ml) added to each well. After a 24 h incubation period, the conditioned medium was removed, cells washed three times with PBS, 100 pi Triton and PBS added to each well, and cell layers scraped into this. Conditioned media and cell lysates were assayed for PA activity. The results for PA activity are expressed as mIU per 5 p1 sample, mean f standard error of the mean (n = 4). **Significantly different from control, p c 0.01, Scheffe's multiple-range analysis.
Plasminogen activator assay Bone cells were incubated for 24 h in the presence of rhTGF-0, or rhTGF-& alone and in combination with rhIL-10. The conditioned media were removed; aliquots were stored at 4°C and then assayed within 6 h. The cell layers were thoroughly washed with cold phosphate-buffered saline. Phosphate-buffered saline containing 0.1 Yo (vol/vol) of Triton X-100 (100 pl) was added to each well and the cells scraped into this solution. Lysates were stored
Bone cells were incubated for 24 h in the presence of rhTGF-0, or rhIL-10 alone and in combination. The conditioned media were removed; aliquots were stored at 4°C and then used for gels within 24 h. The cell layers were washed three times with cold phosphate-buffered saline. Phosphate-buffered saline containing 0.1 % (vol/vol) of Triton X-100 (100 pl) was then added to each well and the cells scraped into this solution. Lysates were stored as for culture media and used within 24 h. Sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis was carried out essentially according to the method of Roche et a1.Iz3) A 4% (wt/vol) acrylamide stacking gel and 9% (wt/vol) resolving gel were used. Two gels were run in parallel, one of the resolving gels containing casein (2 mg/ml) and human plasminogen (6 pg/ml) and the other containing casein only, to show that the activity seen was plasminogen dependent. Samples for electrophoresis in 0.0625 M TrisHCI (pH 6.8) containing 2.5% (wt/vol), 8% SDS (vol/ vol), glycerol, and bromophenol blue were applied to gels without boiling. The high-molecular-weight standards included phosphorylase b (97,400), bovine serum albumin (66,000), ovalbumin (45,000), and carbonic anhydrase (29,000). Gels were run overnight and then washed in 2.5% (vol/vol) aqueous Triton X-100 solution for I h to remove the SDS. For the development of plasminogen activator bands, the gels were incubated at 37°C for 24 h in 0.1 M Tris-HC1 (pH 8.1). The gels were stained with Coomassie brilliant blue R250. Plasminogen activator activity appeared as clear bands against a dark background where plasminogen activators had converted plasminogen to plasmin and the plasmin had degraded the casein in the gel.
Statistical analysis For a given experiment the mean and standard error of the mean were calculated for each group. Analysis of vari-
FAWTHROP ET AL.
1366
ance was performed on all data, followed by Scheffe’s multiple-range analysis. Calculations were made using the Statview 512+ application on an Apple Macintosh computer.
Protein assay Total proteins were assayed using a modification of the method of Oyama and Eagle.”‘’
RESULTS Effect of rhTGF-0, on the PA activity of normal human osteoblast-like cells Human osteoblast-like cells produce low or undetectable, that is, ~ 0 . 1 mIU per 5 pl, levels of both tPA and uPA activity when cultured under serum-free conditions. rhlL-l@ had no stimulatory effect on tPA activity and stimulated uPA activity in only one experiment in which basal levels of uPA were high. rhTGF-PI, when used at concentrations 10-9-10-7M, caused a significant increase in both tPA and uPA activity (Fig. 1). This was reproducible using cells from several different patient sources. Higher levels of uPA activity were detected in cell lysates than conditioned media, and vice versa for tPA activity. The maximal stimulation recorded for each was around 1 mIU per 5 PI. There was no obvious difference in the response to doses of rhTGF-@,between and lo-’ M. The stimulation of uPA activity by rhTGF-@, was confirmed by SDS-polyacrylamide gel electrophoresis (Fig. 2); however, tPA activity was seen only as a very faint band. t P A does not appear clearly in the zymograms, possibly because
within the system there are no molecules that bring about its activation. When rhTGF-@, was used at lower concentrations (Fig. 3) the significant stimulatory effect on both M but maintained uPA and tPA was lost at a dose of at a dose of lo-” M, the response being comparable to that of M rhTGF-@,. The effect of rhTGF-@, and IL-I@in combination o n osteoblast P A was variable. For t P A activity the response to the rhTGF-PI and IL-10 was not significantly different from that of the corresponding dose of rhTGF-@, when used alone. For uPA activity, in two of four experiments the combination caused a significant increase in activity compared to rhTGF-@, alone, differences from controls not being significant in the other cases. Experiments were carried out to assess the effect of indomethacin ( M), cycloheximide (1 pg/ml), and actinomycin D (lo+’ and M) on the basal and rhTGF-@,stimulated P A activity of human osteoblast-like cells (results for t P A activity are shown in Table 1). In no case did indomethacin, cycloheximide, or actinomycin D alone stimulate P A activity. Indomethacin did not significantly affect the action of rhTGF-@,, suggesting that it is not mediated via the prodcution of prostaglandin E. Actinomycin D blocked the stimulatory action of rhTGF-@, on PA activity. However, the inhibitory effect of cycloheximide was not always consistent, possibly a result of the variation in penetration of cycloheximide into the cells.
Effect of rhTGF-0, on the PA activity of normal human osteoblast-like cells Recombinant human TGF-@, stimulated the production of both tPA and uPA by human osteoblast-like cells (Fig. 4). Unlike experiments carried out with rhTGF-@,, the
FIG. 2. SDS-polyacrylamide gel containing plasminogen and casein showing the plasminogen activator species present in the cell lysates of human osteoblast-like cells treated with rhTGF-@, or rhIL-l@ alone or in combination. Human osteoblast-like cells were passaged into 24-well plates and grown to confluence in Eagle’s minimum essential medium with Earle’s salts containing 100 U/ml of penicillin, 100 pg/ml of streptomycin 2 mM glutamine, and 10% fetal calf serum. The cells were washed three times with phosphate-buffered saline and 500 pl incubation medium consisting of dilutions of rhTGF-@,in EMEM containing 0.1% bovine serum albumin with and without rhIL-l@(10 U/ml) added to each well. After a 24 h incubation period, the medium was removed, cells washed three times with PBS, 100 pl Triton and PBS added to each well, and cell layers scraped into this. Plasminogen activator species present in the cell lysates were assessed by zymography. Lane 1, tPA (Bowes melanoma 2 chain); lane 2, uPA (high molecular weight); lane 3, control; lane 4, M; lane 7, rhTGF-@, M , + rhIL-l@(10 U/ml); M; lane 6 , rhTGF-@, rhIL-l@(10 U/ml); lane 5 , rhTGF-@, and lane 8, rhTGF-0, M, + rhIL-10 (10 U/ml). N o bands of activity developed in gels without plasminogen (not shown).
1367
TGF-0 AND PLASMINOGEN ACTIVATOR
In'
"I 0,
01
03 0: 01
OD . l l 0 U ,"I/
TGFOiMi
. T G F O IMi
rhIL-10 consistently stimulated tPA and uPA activity, and this was statistically significant in most cases. As shown in Fig. 5 , rhTGF-0, decreased levels of P A activity below those of the control. This was a consistent response and reached statistical significance in three of four cases for uPA, but only in one experiment at one dose for tPA. This suggests that rhTGF-0, has a greater inhibitory effect on the basal uPA activity than the tPA activity of these cells. The inhibitory effect of rhTGF-P, on uPA activity was confirmed by zymography (Fig. 6). There was no difference in the effect of the varying concentrations of rhTGF0, used from lo-' to M (data for the lowest doses not shown). rhTGF-0, significantly inhibited rhIL-10-stimulated uPA activity in three of four experiments to levels below those of the controls. In the fourth it decreased the activity, but this did not reach statistical significance. Its effect on rhll-10-stimulated tPA activity was variable, sometimes stimulating and sometimes decreasing tPA activity.
DISCUSSION
rhll-lO , I O U m1
FIG. 3. The effect of rhTGF-0, used at low concentrations on the tPA (A) and uPA (B) activity as measured in the conditioned media and cell lysates, respectively, of human osteoblast-like cells and its interaction with rhIL-10. Human osteoblast-like cells were passaged into 24-well plates and grown to confluence in Eagle's minimum essential medium with Earle's salts containing 100 Uiml of penicillin, 100 pg/ml of streptomycin, 2 mM glutamine, and 10% fetal calf serum. The cells were washed three times with phosphate-buffered saline and 500 pl incubation medium consisting of dilutions of rhTGF-PI, in E,MEM containing 0.1070 bovine serum albumin with and without rhIL-10 (10 U/ml) added to each well. After a 24 h incubation period, the conditioned medium was removed, cells were washed three times with PBS, 100 pI Triton and PBS added to each well, and cell layers scraped into this. Conditioned media and cell lysates were assayed for PA activity. The results for PA activity are expressed as mlU per 5 pl sample, mean f standard error of the mean ( n = 4). **Significantly different from control, p < 0.01, Scheffe's multiple-range analysis. level of stimulation did not reach statistical significance in all cases. As for rhTGF-P,, the effect of rhTGF-0, and rhIL-10 in combination on osteoblast PA activity was variable.
Effect of rhTGF-0, on the PA activity of the human osteosarcoma cell line MG-63 These human osteosarcoma cells produced high basal levels of both uPA and tPA, in excess of 100-fold that recorded for normal human osteoblast-like cells. Higher levels of t P A activity were detected in the conditioned medium than in cell lysates, and vice versa for uPA activity.
Results from this study show that normal adult human osteoblast-like cells have low basal levels of PA activity and are capable of producing both u P A and tPA. Low P A levels may reflect the maturity of the cells, since it has been shown in fetal rat osteoblasts that the expression of P A activity tends to fall with It is of interest that both types of PAS were detected. This contrasts with several other osteoblast-like cell types, including UMR-201 and UMR-106-01, in which the PA activity has been shown to be exclusively due to tPA, with no detectable uPA by zymography or with specific antisera.'261Such a discrepancy probably reflects phenotypic differences between these cell types rather than variation in sensitivities and specificities of the assays used or may reflect heterogeneity of the cultures. uPA was detected primarily in the cell lysates and to a much lesser extent in the conditioned medium. This is not surprising because many cell lines have cell surface receptors for UPA.(27-29'The proenzyme is bound to the receptor and is converted to the active molecule, which is then able to bind the zymogen plasminogen before activating it to pla~min.'~O) rhTGF-0, significantly stimulated both the uPA and tPA activity of normal adult human osteoblast-like cells when used over a concentration range of 10-7-10-12M. However, this effect was lost at M. In this colorimetric assay the net change in PA activity is measured. It is conceivable that rhTGF-0, caused an increase in plasminogen activator inhibitor (PAI) activity, but obviously the major effect of rhTGF-0, was an increase in P A activity. Indomethacin did not inhibit the stimulatory effect of rhTGF-PI, suggesting that its mechanism of action is not mediated via prostaglandin synthesis. The ability of actinomycin D and cycloheximide to inhibit P A activity indicates that both transcription and new protein synthesis, respectively, must occur in response to rhTGF-0,. The increased activity cannot solely be attributed to activation of the proenzyme. New protein synthesis is involved in the action of
FAWTHROP ET AL.
1 O U ml
TGFR 2
(MI
TGFO 2 ( M i iL-in lOUml
IL 1R I O U rnl
10 9
!O TGFOl !MI
10
10 3
10
10
TGFOI (MI it-in IOU rnl
"1
TGFO 2 !MI
TGFR *(MI IL-in
!OU ",I
The effect of rhTGF-0, on tPA (A) and uPA (B) activity as measured in conditioned media and cell lysates, respectively, of human osteoblast-like cells and its interaction with rhIL-l@. Human osteoblast-like cells were passaged into 24-well plates and grown to confluence in Eagle's minimum essential medium with Earle's salts containing 100 U/ml of penicillin, 100 pg/ml of streptomycin, 2 mM glutamine, and 10% fetal calf serum. The cells were washed three times with phosphate-buffered saline and 500 11 incubation medium coinsisting of dilutions of rhTGF-@, in EMEM containing 0.1 Vo bovine serum albumin with and without rhIL-10 (10 U/ml) added to each well. After a 24 h incubation period, the conditioned medium was removed, cells washed three times with PBS, 100 pl Triton and PBS added to each well, and cell layers scraped into this. Conditioned media and cell lysates were assayed for PA activity. Results for PA activity are expressed as mIU standard error of the mean (n = per 5 p1 sample, mean 4). **Significantly different from control, p < 0.01, Scheffe's multiple-range analysis.
*
rhTGF-@, on other cell types, such as skin fibroblasts, in which the increase in uPA activity is inhibited by cycloheximide. C3') The action of rhTGF-0, was similar to that of rhTGFp,. TGF-p, and TGF-0, share 70% homology and in most systems have a common action.(3z)As of yet no major difference has been reported in their action on bone. (s.6,33) That the stimulatory effects of TGF-@, and TGF-@, were
FIG. 5. The effect of rhTGF-@,on tPA (A) and uPA (B) activity as measured in conditioned media and cell lysates, respectively, of the human osteosarcoma cell line MG-63 and its interaction with rhIL-l@. Human osteosarcoma cells were passaged into 24-well plates and grown to confluence in Dulbecco's modified essential medium (DMEM) containing 100 U/ml of penicillin, 100 pg/ml of streptomycin 2 mM glutamine, and 10% fetal calf serum. The cells were washed three times with phosphate-buffered saline and 500 p1 incubation medium consisting of dilutions of rhTGF-@, in DMEM containing 0.1 Vo bovine serum albumin, with and without rhIL-10 added to each well. After a 24 h incubation period, the conditioned medium was removed, cell layers washed three times with PBS, 0.1% Triton and PBS (100 pl) added to each well, and the cells scraped into this. Conditioned media and cell lysates were then assayed for PA activity. Results for PA activity are expressed as mIU per 5 p1 sample, mean f standard error of the mean (n = 4). **Significantly different from control, p < 0.01, Scheffe's multiple-range analysis.
due to TGF-@per se was indicated by similar effects of the two different sources of TGF-@. Recombinant human IL-10 did not stimulate the PA activity of normal human osteoblast-like cells. This finding is in accordance with that of Evans,(34)who showed that IL10 can only stimulate the PA activity of human osteoblastlike cells in the presence of serum. Under serum-free conditions a small reduction in PA activity was demonstrated.
TGF-0 AND PLASMINOGEN ACTIVATOR
1369
TABLE1. EFFECTOF INDOMETHACIN, CYCLOHEXIMIDE, AND ACTINOMYCIN D ON THE TPA ACTNITY (MEASURED IN CONDITIONED MEDIA)OF HUMANOSTEOBLAST-LIKE CELLSSTIMULATED AND UNSTIMULATED BY TGF-/3,a tPA activity (mIUper 5 PI)
Experimental conditions ~~~
Control
TGF-0(10-7 M) Indomethacin (10 -6 M) TGF-0 M) + indomethacin (10-6 M) Cycloheximide (1 pg/ml) TGF-0 M) + cycloheximide (1 a d m u Actinomycin D M 10-7 M TGF-0 (lo-’ M) + actinomycin D (10-7 M)
~
0.184 (0.077) 1 .ooo(0) 0.056 (0.056) 0.!2 (0.1)
