Journal of Cell Science 102, 315-322 (1992)Printed in Great Britain © The Company of Biologists Limited 1992

315

Regulation of fibroblast-mediated collagen gel contraction by platelet-

derived growth factor, interleukin-1 a and transforming growth factor-/^

ANDERS TINGSTROM1-2, CARL-HENRIK HELDIN2 and KRISTOFER RUBIN1*1Department of Medical and Physiological Chemistry, University of Uppsala, BMC, Box 575, S-751 23 Uppsala, Sweden2Ludwig Institute of Cancer Research, Uppsala Branch, BMC, Box 595, S-751 24 Uppsala, Sweden

•Author for correspondence

Summary

We have examined the effects of three macrophage-derived cytokines, platelet-derived growth factor(PDGF), transforming growth factor-01 (TGF-01) andinterleukin-1 a (IL-la) on the contraction of collagentype I gels populated by human foreskin fibroblasts.Contraction was quantified as loss in gel weight. BothPDGF-AA and PDGF-BB were found to induce a rapidcollagen-gel contraction. TGF-/J1 also stimulated gelcontraction but with a delayed onset and at a slower ratethan the PDGF-stimulated contraction. Rabbit poly-clonal IgGs recognizing PDGF-AA and PDGF-BB,respectively, specifically inhibited the effects of thecorresponding PDGF Lsoforms. However, the stimula-tory effect of TGF-/S1 was not affected by any of the anti-PDGF antibodies. The ability of PDGF to stimulatecontraction became less pronounced in collagen gelcultures grown in the absence of growth factors overperiods of several days. Under the same conditions, thestimulatory effect of TGF-/J1 was not reduced. Thereduced response to PDGF may be due to reducedtension on fibroblasts growing in collagen gels, sincefibroblasts on free-floating gels showed a marked

reduction in PDGF-BB-induced PDGF ^-receptor ag-gregates when compared to fibroblasts on attachedcollagen gels. LL-1 a inhibited initial collagen gel contrac-tion, and at later stages induced a visible degradation ofthe collagen gels, presumably due to the generation ofcollagenase activity. The combination of IL-la andPDGF-BB stimulated initial collagen gel contraction,although less effectively than PDGF-BB alone. At laterstages, collagen gel degradation was stimulated by thiscombination of cytokines. In contrast, the combinationof IL-la and TGF-/51 did not stimulate collagen gelcontraction, or any visible collagen gel degradation. Ourdata suggest that fibroblast-mediated collagen gel con-traction can be modulated by cytokines via differentmechanisms. Our data are of importance in theunderstanding of the modulatory roles of cytokines inconnective tissue cell activities in inflammatory pro-cesses, such as wound healing.

Key words: collagen, interleukin-1, fibroblasts, platelet-derived growth factor, transforming growth factor-̂ 3.

Introduction

Fibroblasts grown within floating three-dimensionalcollagen gels exhibit a different phenotype fromfibroblasts grown on plastic dishes. The phenotypeaquired in collagen gels resembles in many respects thatdisplayed by fibroblast-like stromal cells in normal adulttissues (Elsdale and Bard, 1972; Tomasek et al., 1982;Nusgens et al., 1984; Mauch et al., 1988). Whentransferred to collagen gels, fibroblasts are able toreorganize the collagen fibres and contract the collagengel, a process preceding the changes in phenotype (Bellet al., 1979; Harris et al., 1981; Guidry and Grinnell,1985). Collagen gel contraction is regarded as an in vitromodel for wound contraction (Bell et al., 1979; Ehrlichand Wyler, 1983), and has been studied from severalaspects. Thus, the characteristics of the contraction

reaction depend on the type and concentration of thecollagen in the lattice (Bell et al., 1979; Ehrlich, 1988)and on the cell type (Steinberg et al., 1980). Collagengel contraction can be inhibited by the addition of drugsaffecting cytoskeletal organization (Bell et al., 1979;Guidry and Grinnell, 1985; Tomasek and Hay, 1984).

Fibroblast-driven contraction of collagen gels isstimulated by serum (Steinberg et al., 1980; Buttle andEhrlich, 1983). The presence of contraction stimulatoryactivity has been demonstrated in endothelial cellculture supernatants (Guidry et al., 1990). Fibronectinhas been reported to stimulate collagen gel contraction(Gillery et al., 1986), but in other systems no such rolefor fibronectin was observed (Gullberg et al., 1990).Inhibitors of collagen gel contraction, e.g. granulomasupernatants, have likewise been described (Ehrlichand Wyler, 1983). The inhibitory acivity in the granu-

316 A. Tingstrom and others

loma supernatants was linked to prostaglandins, sincethe effect was abrogated by the addition of indometha-cin, and prostaglandin E2 alone inhibited contraction(Ehrlich and Wyler, 1983).

Studies using defined cytokines have demonstratedthat transforming growth factor-/? (TGF-/3) (Montesanoand Orci, 1988) promotes collagen gel contraction,whereas interleukin-1 (IL-1) has been reported toinhibit this process (Gillery et al., 1989). We and othershave demonstrated that platelet-derived growth factor(PDGF) potently stimulates collagen gel contractionmediated by rat heart and human dermal fibroblasts(Clark et al., 1989; Gullberg et al., 1990). Yet otherpolypeptide growth factors, for example epidermalgrowth factor, have no effect on collagen gel contrac-tion (Montesano and Orci, 1988), indicating a speci-ficity of the contraction reaction. Furthermore, it wasfound that the PDGF-stimulated collagen gel contrac-tion was mediated by collagen-binding ft integrins(Gullberg et al., 1990). These collagen-binding inte-grins are independent of the Arg-Gly-Asp sequence,which is recognized by several other integrin hetero-dimeric matrix receptors (Ruoslahti and Pierschbacher,1987; Gullberg et al., 1989). A direct role for thecollagen-binding integrin a^fii in fibroblast-mediatedcollagen gel contraction has recently been demon-strated (Klein et al., 1991; Schiro et al., 1991).

PDGF is a 30 kDa connective tissue cell mitogenconsisting of two disulfide-linked polypeptide chains,denoted A and B, that can be assembled as homo- orheterodimers (for a review on PDGF, see Heldin andWestermark, 1990). PDGF exerts its cellular effects bybinding to two different but structurally similar receptortypes: a-receptors bind all PDGF isoforms with highaffinity, whereas preceptors bind only PDGF-BB withhigh affinity. TGF-/J is a family of closely related factorsthat stimulate or inhibit the growth of cells, dependingon cell type and culture conditions (for a review onTGF-/3, see Roberts and Sporn, 1990).

To understand better the mechanisms involved in theregulation of collagen gel contraction, which may berelated to wound contraction in vivo, it is important tocharacterize the effects of factors produced by cellsinvolved in the inflammatory reaction on this process.In the present study, we have investigated the effects onfibroblast-mediated collagen gel contraction of differ-ent cytokines that are produced by macrophages, i.e.PDGF isoforms, TGF-/31 and IL-1 or. We report that thedifferent cytokines have different effects on the con-traction process.

Materials and methods

ReagentsCalf skin collagen type I (Vitrogen 100) was obtained fromCollagen Corporation (Palo Alto, CA). Recombinant PDGF-AA and PDGF-BB were from transfected yeast cells (Ostmanet al., 1989). Recombinant human IL-lo-was purchased fromGenzyme Co. (Boston, Mass.) and recombinant TGF-/31 waskindly donated by Dr. Kohei Miyazono, Ludwig Institute forCancer Research, Uppsala. Dulbecco's modified Eagle's

medium (DMEM) and MCDB 105 medium for serum-freecell culture were obtained from the National VeterinaryInstitute, Uppsala, Sweden, and the Department of Pathol-ogy, Uppsala University, Sweden, respectively. Fetal bovineserum (FBS) was obtained from Row Laboratories (Irvine,UK). Trypsin (Cat. no. 043-05300 H) was from GIBCO(Paisley, UK). Soybean trypsin-inhibitor and bovine serumalbumin, Fraction V (BSA) were obtained from SigmaChemical Co. (St. Louis, MO). Rabbit polyclonal antibodiesto recombinant PDGF-AA and PDGF-BB have been de-scribed previously (Thyberg et al., 1990).

CellsStock cultures of human foreskin fibroblasts, AG 1518(Genetic Mutant Cell Repository, Camden, NJ), were grownin 175 cm2 cell culture flasks (Costar, Cambridge, MA) inDMEM supplemented with 10% FBS, 0.5 ^g/ml strepto-mycin, 0.5 /Jg/ml neomycin and 0.6 /ig/ml penicillin G. Cellswere passaged once a week and fed with fresh medium twice aweek. All experiments were performed with cells at passagenumbers 15 to 25. The cells were detached by 2 rinses with0.13 M NaCl and 0.01 M sodium phosphate, pH 7.4 (PBS),containing 10 mM EDTA, and subsequent incubation for 3 to5 minutes with 600 pA 0.05% trypsin and 0.02% EDTA inPuck's saline A (GIBCO). After the cells had becomedetached, the remaining trypsin activity was neutralized bythe addition of 10 ml PBS containing 1 mg soybean trypsininhibitor. Finally, the cells were centrifuged, resuspended inice-cold MCDB 105 medium and counted.

Preparation of collagen gelsTissue culture plates (24-well, Costar) were coated overnightwith 2% BSA in PBS and then washed 3 times with PBS.Neutral collagen solution was prepared by mixing 2 timesconcentrated MCDB 105 medium, 0.2 M HEPES, pH 8.0,and Vitrogen 100, in the relation 5:1:4 (by vol.). Suspendedfibroblasts (500,000 cells/ml in MCDB 105 medium) wereadded in the relation 1:9 (v/v) to the neutral collagen solution.One ml of the collagen/cell suspension was added per well toprecoated 24-well plates and the collagen was polymerized byincubating the suspension for 1 hour at 37°C. If thecell/collagen suspensions were ice-cold at the time of additionto the 24-well plate, most of the cells sedimented duringpolymerization of the gel and appeared in the lower part ofthe polymerized gel. To prevent this, the neutral collagensolution was warmed up to room temperature prior to theaddition of cells. It was then necessary to observe the collagensolution carefully in order to add the cells before thepolymerization process started. After polymerization, thecollagen gels were floated by adding 1 ml of MCDB 105medium containing the desired cytokines.

Measurement of gel contractionCollagen gels were fixed at the desired points of time byaddition of 200 ̂ 1 of a 37% formalin solution to the wells andsubsequent incubation for several hours at room temperatureor overnight at 4CC. The fixed gels were washed once withTris-buffered saline with 0.2% Tween-20. Detergent waspresent to reduce surface tension. The gels were then weighedon an analytical digital scale with a precision of 0.1 mg.

Detection of PDGF-BB-induced PDGF preceptoraggregates in floating versus attached collagen gelsA 1 ml sample of neutral collagen solution prepared in MCDB105 medium as described above was allowed to polymerize for1 hour at 37°C. Human foreskin fibroblasts in 1 ml MCDB 105medium were seeded on top of the preformed collagen gels at

a density of 200,000 cells/dish. After 2 hours, when the cellshad attached to the collagen support, 2 ml of MCDB 105medium was added to the dishes. Half of the collagen gelswere then detached from the culture dishes with a spatula.The cells were incubated further for 3 days and then treatedfor 1 hour with 50 ng/ml of PDGF-BB to induce formation ofPDGF /J-receptor aggregates. Finally, the gels were fixed bythe addition of 200 jil 20% paraformaldehyde in PBS. The gelswere left overnight at 4°C and were then crysectioned.Immunofluorescence staining of the sections was done withPDGFR-B2 igG (1 j/g/ml) as described (Reuterdahl et al.,1991). The fluorescent sections were photographed using aZeiss inverted microscope equipped with epi-illumination.

Results

Quantification of collagen gel contractionSerum-free collagen gels populated by human foreskinfibroblasts (AG 1518) were prepared as described inMaterials and methods. After polymerization of thecollagen gels for 1 hour at 37°C, culture mediumcontaining the cytokines to be tested was added.

Collagen gel contraction is conventionally deter-mined by measuring the reduction of collagen geldiameter. A more accurate way of measuring collagengel contraction is to determine decrease in collagen gelvolume. One method of doing this is based on themeasurement of tritiated water retained in the collagengels (van Bockxmeer et al., 1984). In the present studywe have developed and used a novel assay for thequantification of collagen gel contraction. In this assaycollagen gels were fixed in formaldehyde and weighed.The fixation made the gels sufficiently rigid to allow fortheir removal from the wells and transfer to a scalewithout loss of fluid from the gels. This method turnedout to give more accurate readings than those achievedwhen measuring gel diameter, especially at later stagesof the contraction process, when changes in geldiameter are too small to be accurately measured. Fig. 1shows a comparison between the measurement of gelweight and gel diameter in an experiment in whichPDGF-AA was used to stimulate collagen gel contrac-tion (see further below).

Effect of PDGF, TGF-fil and lL-laon collagen gelcontractionFibroblast-populated collagen gels that had receivedPDGF-AA or PDGF-BB contracted at a faster rate,and to a greater extent than did control gels that had notreceived PDGF (Fig. 2A and B). PDGF-AA- andPDGF-BB-stimulated collagen gel contraction oc-curred rapidly and could be detected after a few hours(Fig. 2A). TGF-/J1 also stimulated collagen gel contrac-tion, but more weakly and with a slower onset thanPDGF. Thus after 24 hours no, or very little, stimula-tory effect of TGF-/31 was evident (Fig. 2A and B),whereas after 96 hours TGF-/21 had caused a significantcontraction of the gels (Fig. 2A and B). Under theconditions chosen, the collagen gels showed a markedcontraction in the absence of stimulators also (Fig. 2A);however, the stimulatory effects on contraction by

Cytokine effects on collagen gel contraction

300

317

10

PDGF-AA (nM)

Fig. 1. Method for quantification of collagen gelcontraction. Fibroblast-populated collagen gels werestimulated with different concentrations of PDGF-AA,ranging from 0 to 8.3 nM, for 24 hours. The gels werefixed and their weight (•) and diameter (O) weredetermined as described in Material and methods. Valuesare means of duplicate samples and bars represent totalspread of values. A close correlation was seen between thetwo methods.

TGF-/J1 and PDGF were evident as measured with thepresent method. The decrease in collagen gel weightwith time, given as percentage of initial gel weight, isshown in Fig. 2A, and gives a more accurate measure ofthe stimulatory effects of PDGF and TGF-/31. In whatfollows the effects of the cytokines on contraction arerelated to the contraction in control gels incubated inthe absence of stimulators (compare Fig. 2A and B).

The effects of PDGF-AA and PDGF-BB, as well asthose of TGF-/31, were concentration-dependent (Fig.2C and D). At both time points studied, 8 and 96 hours,PDGF-BB was more potent than PDGF-AA in stimu-lating collagen gel contraction. The PDGF concen-trations required to give half-maximal responses incollagen gel contraction were similar to those requiredto give half-maximal stimulation of mitosis with thesame type of cells (Heldin et al., 1988). Similarly, thedifference in collagen gel contraction-stimulatory ac-tivity between PDGF-AA and PDGF-BB, respectively,is comparable to their different abilities to stimulatemitosis in the same cell type (Heldin et al., 1988). Thestimulatory effect of TGF-/S1 was also concentration-dependent (Fig. 2D).

The presence of IL-la- (5 units/ml) inhibited collagengel contraction at early time points; at later stages, IL-la induced a visible degradation of the collagen gel(Fig. 3, and data not shown). The combination ofPDGF-BB (0.17 nM) and IL-lo-(5 units/ml) stimulatedcollagen gel contraction at early time points, but afterprolonged incubation periods this combination ofcytokines had induced a visible degradation of thecollagen gels (Fig. 3, and data not shown). In thepresence of TGF-/31 (0.2 nM) the inhibitory effect oncollagen gel contraction by IL-la was less marked andno visible degradation of the collagen gels had occurred

318 A. Tingstrom and others

'(1)

'E"o

20 40 60 80

Time (hours)

100 20 40 60 80

Time (hours)

100

Concentration (nM)

0.001 0.01 0.1 1

Concentration (nM)

10

Fig. 2. Time and concentration dependency of collagen gel contraction in response to TGF-/31, PDGF-AA and PDGF-BB.(A and B) Fibroblast-populated collagen gels were stimulated with 1 nM TGF-/31 (O O); 1.7 nM PDGF-AA ( • • ) ;1.7 nM PDGF-BB (A A); or medium without additions ( • • ) . Gel weights were determined after 8, 24 and 96hours. The results are plotted relative to the initial gel weight (A), or as the relative decrease in gel weights compared toweights, at each time point, of control gels that had not recieved cytokines (B). (C and D) Fibroblast-populated collagengels were stimulated with PDGF-AA ( • • ) or PDGF-BB (A A) at concentrations ranging from 0 to 1.7 nM, orwith TGF-/J1 (O O) at concentrations ranging from 0 to 1 nM. Gel weights was determined after 8 (C) and 96 (D)hours.

(Fig. 3). TNF-o' affected collagen gel contraction in aqualitatively similar way to IL-lar (data not shown).

Effect of anti-PDGF-AA IgG and anti-PDGF-BBIgGThe stimulatory effects on collagen gel contraction byPDGF-AA and PDGF-BB were inhibited by IgGsspecifically recognizing the respective PDGF isoforms(Fig. 4). The effect of TGF-£1 on collagen gelcontraction was not affected by any of these IgGs,indicating that the stimulatory effect of TGF-/31 was notmediated by production of PDGF (Fig. 4). IgG derivedfrom a non-immune serum had no effect on thecontraction-stimulating capabilities of any of thesecytokines (Fig. 4).

Effects of time in culture on fibroblast susceptibility toPDGF and TGF-filPDGF-AA or PDGF-BB added immediately after theformation of fibroblast-populated collagen gels,potently stimulated collagen gel contraction whenmeasured after 48 hours (Figs 2A, B and 6), whereasTGF-/31 under the same conditions had a less-markedstimulatory effect (Fig. 2A and B). PDGF-AA, PDGF-BB or TGF-/31 added to floating collagen gels that had

been cultured for 48 hours in MCDB 105 mediumwithout additions induced an equally strong collagengel contraction over the next 48 hours (Fig. 5). Whenthe contraction stimulators were added to floatingcollagen gels that had been precultured for 96 hours inMCDB 105 medium without additions, and the contrac-tion-stimulatory effect was measured after anotherincubation period of 48 hours, it was found that PDGF-AA and PDGF-BB were less potent than TGF-/31 (Fig.5). Thus, the contraction-stimulatory effects of PDGF-AA and PDGF-BB became less marked with time whencells were cultured in floating collagen gels. In contrast,the effect of TGF-/S1 was slightly increased with time.

Expression of PDGF ^-receptors on free-floatingversus attached collagen gelsThe finding that fibroblasts grown in collagen gelsgradually became less responsive to PDGF encouragedus to investigate the expression of PDGF /3-receptors onsuch cells. It has previously been demonstrated thatfibroblasts cultured inside free-floating collagen gels,when compared with fibroblasts cultured inside at-tached gels, show a decreased responsivness to PDGFstimulation, measured as increase in DNA synthesis(Nakagawa et al., 1989). It seems likely that this

Cytokine effects on collagen gel contraction 319

decrease in sensitivity to PDGF is not the result of cellinteractions with collagen, but is due rather to thereduction in tension forces acting on the cells (Naka-gawa et al., 1989; Fukamizu and Grinnell, 1990). PDGF/J-receptors can be demonstrated if fibroblasts aretreated with PDGF-BB for 1 hour prior to fixation and

80 100 120

Time (hours)

Fig. 3. Effects of combinations of IL-la- and TGF-01 orPDGF-BB. Fibroblast-populated collagen gels werestimulated with IL-la (5 units/ml), ( • D); IL-la- (5units/ml) + PDGF-BB (0.17 nM), (A A); IL-la- (5units/ml) + TGF-£ (0.2 nM) ( • • ) ; or PDGF-BB(0.17 nM), (A A). Gel weights were determined aftervarious times of incubation. IL-la, or the combination ofIL-la' and PDGF-BB induced a degradation of the collagengels, seen as loss of gel weight (indicated by broken linesin the figure). The combination of TGF-/51 and IL-la-didnot induce a reduction in gel weight.

SZ

0)

lo

12 "

10"

8"

6"

4 "

PDGF-AA PDGF-BB TGF-B1 Control

Fig. 4. Effects of antibodies to PDGF-AA and PDGF-BB.Fibroblast-populated collagen gels were grown in thepresence of anti-PDGF-AA IgG (50 jig/ml); open bars,anti-PDGF-BB IgG (50 /Jg/ml); hatched bars or controlIgG (50 }j%Jm\); filled bars, plus PDGF-AA (0.17 nM),PDGF-BB (0.17 nM), TGF-)31 (0.2 nM), or no addition.Gel weights were determined after 4 days. Means ofduplicate incubations are given; bars represent totalspreading between the values. Anti-PDGF-AA inhibitedthe effect of PDGF-AA, and anti-PDGF-BB inhibited theeffect of PDGF-BB. None of these antibodies inhibited theeffect of TGF-01. The control IgG did not affect thecontraction process.

*- 100-D)

0)D5

Coo

80-

60-

40-

20-

0-48 48-96

Time (hours)

96-144

Fig. 5. Effects of preculturing of the fibroblast-populatedcollagen gels. Fibroblasts were grown in collagen gels in a24-well plate. The gels were floated by the addition of 1 mlof medium. PDGF-AA (0.17 nM; hatched bars), PDGF-BB (0.17 nM; open bars) or TGF-£1 (0.2 nM; filled bars)were added to the first set of gels; at 48 hours the first gelswere fixed with paraformaldehyde and the gel weightsdetermined. At 48 hours the second set of gels receivedcytokines, and were fixed and weighed after another 48hours; finally at 96 hours, the third set of gels recievedcytokines, and were fixed and weighed after another 48hours. Values are means of triplicate incubations and barsrepresent total spread of values.

staining (Tingstrom et al., 1992). Owing to problemswith impaired diffusion, this technique is, however, notreadily applicable to cells cultured inside collagen gels.We therefore chose to compare the expression ofPDGF /3-receptors on fibroblasts grown on top ofattached gels with cells grown on top of free-floatingcollagen gels. Applying the technology described aboveit was found that the expression of PDGF ^-receptorswas clearly reduced on cells grown on top of free-floating collagen gels, when compared with cells grownon top of attached collagen gels (Fig. 6).

Discussion

PDGF is the major mitogen in serum for culturedfibroblasts, glial cells and smooth muscle cells (Raineset al., 1990; Heldin and Westermark, 1990). Inaddition, PDGF acts as a chemotactic factor forfibroblasts (Seppa et al., 1982), smooth muscle cells(Grotendorst et al., 1982), monocytes and granulocytes(Deuel et al., 1982; Siegbahn et al., 1990). Further-more, PDGF induces actin filament reorganization, i.e.formation of circular membrane ruffles (Mellstrom etal., 1988; Hammacher et al., 1989) and causes vasocon-striction of rat aortic strips (Berk et al., 1986). In allthese effects of PDGF, cytoskeletal elements areinvolved either directly or indirectly. It is possible thatthe stimulation of fibroblast-driven collagen gel con-traction by PDGF is also related to effects of PDGF onthe cytoskeleton, since collagen gel contraction isdependent on cytoskeletal proteins (Bell et al., 1979;Guidry and Grinnell, 1985; Tomasek and Hay, 1984).

320 A. Tingstrom and others

Fig. 6. Visualization of PDGF ^-receptors on fibroblasts grown on collagen gels. Cells were grown for 4 days under serum-free conditions on collagen gels attached to the culture dish (A) or collagen gels free-floating in the culture medium (B).PDGF /J-receptor aggregates were induced by PDGF-BB treatment and then visualized as described in Materials andmethods.

The effects of the PDGF isoforms and TGF-/31 on thefibroblast-driven collagen gel contraction displayedseveral differences. The PDGF-stimulated contractionwas rapid, and the effect was evident after a few hours,whereas the effect of TGF-/31 was not evident until after24-48 hours (Fig. 2). This is in agreement with what hasbeen reported by others (Montesano and Orci, 1988;Gark et al., 1989), and with our own previous work(Gullberg et al., 1990). These findings, however, arecompatible with the possibility that TGF-/J1 exerts itsstimulatory effect via production of PDGF, a possibilitythat is supported by the observation that TGF-/31induces the synthesis of PDGF-AA in human foreskinfibroblasts (Paulsson et al., 1988). However, antibodiesspecifically recognizing PDGF-AA or PDGF-BB hadno effect on TGF-/81-stimulated collagen gel contrac-tion (Fig. 4). In addition, fibroblasts became lessresponsive to stimulation by PDGF, but not tostimulation by TGF-/J1 after the cells had grown incollagen gels for several days in the absence ofcontraction stimulators (Fig. 5). Taken together, thedata indicate that PDGF and TGF-01 stimulate col-lagen gel contraction via different mechanisms, and thatthe stimulatory effect of TGF-/J1 is not due to inductionof PDGF synthesis.

The observed reduction in ability of PDGF tostimulate collagen gel contraction in fibroblasts culturedfor longer time periods in free-floating gels is inagreement with the observation that PDGF does notstimulate mitosis in such fibroblasts (Nakagawa et al.,1989). In the present report we show that this reducedresponsiveness to PDGF by cells grown on free-floatingcollagen gels can be explained by a decrease inexpression of PDGF /J-receptors (Fig. 6). Since thiseffect was not seen on fibroblasts grown on attached

collagen gels, it is reasonable to suggest that expressionof PDGF jS-receptors is influenced by cell shape. Sincethe response to FGF and EGF, measured as increase inDNA synthesis, is also reduced in fibroblasts grownunder low tension conditions (Nakagawa et al., 1989), itis possible that the expression of receptors for thesegrowth factors is also influenced by the cell shape.

TGF-/8 has been shown to be chemotactic forinflammatory cells (Wahl et al., 1987) and for fibro-blasts (Postlethwaite et al., 1987), and the effect ofTGF-/J on collagen gel contraction may be linked to itsmigration-stimulatory activity. Several other effects ofTGF-/8 may also be of importance in collagen gelcontraction, including stimulation of the synthesis ofextracellular matrix (ECM) components, such as fibro-nectin, as well as an inhibition of the release of enzymescapable of degrading ECM componentes (reviewed byRoberts and Sporn, 1990). Furthermore, TGF-/31increase the expression of ECM receptors belonging tothe pi subfamily of integrins (Heino et al., 1989). Therelative importance of these various TGF-/3 activitiesfor the stimulation of collagen gel contraction remainsto be elucidated.

The finding that PDGF-AA stimulates collagen gelcontraction suggests that the PDGF a-receptor canmediate this cellular response. Clark et al. (1989) havereported that PDGF-BB and PDGF-AB, but notPDGF-AA, stimulate collagen gel contraction me-diated by fibroblasts. It is possible that the discrepancybetween the results reported here and those of Clark etal. (1989) is due to differences in the numbers of PDGFa-receptors on the cell types studied.

IL-la inhibited initial collagen gel contraction, and atlater stages induced degradation of the collagen gels(Fig. 3). IL-la is known to induce the production of

Cytokine effects on collagen gel contraction 321

prostaglandins as well as of collagenase (Postlethwaiteet al., 1983; Dejana et al., 1987), and prostaglandin E2has been reported to inhibit collagen gel contraction(Ehrlich and Wyler, 1983). However, the effects of IL-1 a observed in the present study were not abrogated bythe presence of indomethacin (data not shown),suggesting that the effect of IL-laron contraction is notmediated via the production of prostaglandins. Thecombination of TGF-/S1 and IL-la did not stimulatecollagen gel contraction when compared with controlcultures without added cytokines (Fig. 3). The presenceof TGF-/31 did, however, inhibit any visible IL-la-induced degradation of the collagen gels. The latter is ingood agreement with the reported inhibitory effects ofTGF-/3 on the release of metalloproteinases capable ofdegrading collagen (see Roberts and Sporn, 1990).

In summary, we show here that the process ofcollagen gel contraction is differentially regulated bycytokines that can be produced by activated macro-phages, such as PDGF-AA, PDGF-BB, TGF-/31 andIL-lar. This regulation appears complex, and mayinvolve synthesis and degradation of matrix com-ponents, regulation of integrin expression and avidity,as well as modulation of the tension generated in thecytoskeleton. Thus, to assess the likely roles of aparticular cytokine, such as PDGF, in e.g. inflamma-tory reactions, it is necessary to take into accountpossible modulatory effects of combinations of cytok-ines.

This study was supported by grants from The SwedishCancer Society, The Lundberg Foundation, Konung GustafV:s 80-arsfond and Anna-Greta Crafords forskningsfond.

References

Bell, E., Ivarsson, B. and Merrill, C. (1979). Production of a tissue-like structure by contraction of collagen lattices by humanfibroblasts of different proliferative potential in vitro. Proc. Nat.Acad. Sci. U.S.A. 76, 1274-1278.

Berk, B. C , Alexander, R. W., Brock, T. A., Gimbrone, M. A. Jr andWebb, R. C. (1986). Vasoconstriction: A new activity for platelet-derived growth factor. Science 232, 87-90.

Buttle, D. J. and Ehrlich, H. P. (1983). Comparative studies ofcollagen lattice contraction utilizing a normal and a transformedcell line. J. Cell. Physiol. 116, 159-166.

Clark, R. A. F., Folkvord, J. M., Hart, C. E., Murray, M. J. andMcPherson, J. M. (1989). Platelet isoforms of platelet-derivedgrowth factor stimulate fibroblasts to contract collagen matrices. J.Clin. Invest. 84, 1036-1040.

Dejana, E., Brevario, F., Erroi, A., Bussolino, F., Mussoni, L.,Gramse, M., Pintucci, G., Casali, B., Dinarcllo, C. A., van Damme,J. and Montovani, A. (1987). Modulation of endothelial cellfunction by different molecular species of IL-]. Blood 69, 695-699.

Deuel, T. F., Senior, R. M., Huang, J. S. and Griffin, G. L. (1982).Chemotaxis of monocytes and neutrophiles to platelet-derivedgrowth factor. J. Clin. Invest. 60, 1046-1049.

Ehrlich, H. P. (1988). The modulation of contraction of fibroblastpopulated collagen lattices by types I, II and III collagen. Tissue &Cell 20, 47-50.

Ehrlich, H. P. and Wyler, D. J. (1983). Fibroblast contraction ofcollagen lattices in vitro: inhibition by chronic inflammatory cellmediators. J. Cell. Physiol. 116, 345-351.

Elsdale, T. R. and Bard, J. B. L. (1972). Collagen substrata forstudies on cell behaviour. J. Cell Biol. 54, 626-637.

Fukamizu, H. and GrinneU, F. (1990). Spatial organization of

extracellular matrix and fibroblast activity: Effects of semm,transforming growth factor-/? and fibronectin. Exp. Cell Res. 190,276-282.

Gillery, P., Coustry, F., Pujol, J.-P. and Borel, J.-P. (1989).Inhibition of collagen gel synthesis by interleukin-1 in three-dimensional collagen lattice cultures of fibroblasts. Expenentia 45,98-101.

GUlery, P., Maquart, F.-C. and Borel, J.-P. (1986). Fibronectindependence of the contraction of collagen lattices by human skinfibroblasts. Exp. Cell Res. 167, 29-37.

Grotendorst, G. R., Chang, T., Seppa, H. E. J., Kleinman, H. K. andMartin, G. R. (1982). Platelet-derived growth factor is achemoattractant for vascular smooth muscle cells. J. Cell. Physiol.113, 261-266.

Guidry, C. and GrinneU, F. (1985). Studies on the mechanism ofhydrated collagen gels reorganization by human skin fibroblasts. J.Cell Sci. 79, 67-81.

Guidry, C , Hohn, S. and Hook, M. (1990). Endothelial cells secrete afactor that promotes fibroblast contraction of hydrated collagengels. J. Cell Biol. 110, 519-528.

Gullberg, D., Terracio, L., Borg, T. K. and Rubin, K. (1989).Identification of integrin like matrix receptors with affinity forinterstitial collagens. J. Biol. Chem. 264, 12,686-12,694.

Gullberg, D., Tingstrom, A., Thuresson, A.-C., Olsson, L., Terrack),L., Borg, T. K. and Rubin, K. (1990). /?, Integrin-ediated collagengel contraction is stimulated by PDGF. Exp. Cell Res 186,264-272.

Hamraacher, A., Mellstrttm, K., Heldin, C.-H. and Westermark, B.(1989). Isoform-specific induction of actin reorganization byplatelet-denved growth factor suggests that the functionally activereceptor is a dimer. EMBO J 8, 2489-2495.

Harris, A. K., Stopak, D. and WUd, P. (1981). Fibroblast traction as amechanism for collagen morphogenesis. Nature 290, 249-253

Helno, J., Ignotz, R. A., Hemler, M. E., Crouse, C. and Massagu£, J.(1989). Regulation of cell adhesion receptors by transforminggrowth factor-/?. Concomitant regulation of integrins that share acommon /3, subunit. J. Biol. Chem. 264, 380-388.

Heldin, C.-H., Hackstrom, G., Ostman, A., Hammacher, A.,Ronnstrand, L., Rubin, K., Niste"r, M. and Westermark, B. (1988).Binding of different dimeric forms of PDGF to human fibroblasts:evidence for two separate receptor types. EMBO J. 7, 1387-1393.

Heldin, C.-H. and Westermark, B. (1990). Platelet-derived growthfactor: mechanism of action and possible in vivo function. CellRegul. 1, 555-566.

Klein, C. E., Dressel, D., Steinmayer, T., Mauch, C , Eckes, B.,Krieg, T., Bankert, R. B. and Weber, L. (1991). Integrin arft isupregulated in fibroblasts and highly aggressive melanoma cells inthree-dimensional collagen lattices and mediates the reorganizationof collagen I fibrils. J. Cell Biol 115, 1427-1436.

Mauch, C , Hatamochi, A., ScharfTetter, K. and Krieg, T. (1988).Regulation of collagen synthesis in fibroblasts within a three-dimensional collagen gel. Exp. Cell Res. 178, 493-503.

Mellstrom, K., Hoglund, A. S., Niste>, M., Heldin, C.-H.,Westermark, B. and Lindberg, U. (1983). The effect of platelet-derived growth factor on morphology and motility of human gliacells. J. Muscle Res. Cell Mold. 4, 589-609.

Montesano, R. and Orci, L. (1988). Transforming growth factor-/?stimulates collagen-matrix contraction by fibroblasts: implicationsfor wound healing. Proc. Nat. Acad. Sci. U.S.A. 85, 4894-4897.

Nakagawa, S., Pawelek, P. and GrinneU, F. (1989). Extracellularmatrix organization modulates fibroblast growth and growth factorresponsiveness. Exp. Cell Res. 182, 572-582.

Nusgens, B., MerriU, C , Lapiere, C. and BeU, E. (1984). Collagenbiosynthesis by cells in a tissue equivalent matrix in vitro. CollagenRel. Res. (Matrix) 4, 351-364.

Ostman, A., BSckstrom, G., Fong, N., Betsholtz, C , Wernstedt, C ,Hellman, U., Westermark, B., Valenzuela, P. and Heldin, C.-H.(1989). Expression of three recombinant homodimeric isoforms ofPDGF in Saccharomyces Cerevisiae: Evidence for differences inreceptor binding and functional activities. Growth Factors 1, 271-281.

Paulsson, Y., Beckmann, P. M., Westermark, B. and Heldin, C.-H.(1988). Density-dependent inhibition of cell growth bytransforming growth factor-/31 in normal human fibroblasts.Growth Factors 1, 19-27.

322 A. Tingstrom and others

Postlethwaite, A. E., Kcski-Oja, J., Moses, H. L. and Kang, A. H.(1987). Stimulation of the chemotatctic migration of humanfibroblasts by transforming growth factor beta. J. Exp. Med. 165,251-256.

Raines, E. W., Bowen-Pope, D. F. and Ross, R. (1990). Platelet-derived growth factor. In Handbook of ExperimentalPharmacology. Peptide Growth Factors and Their Receptors, vol.95, part I (ed. Sporn, M.B. and Roberts, A.B.), pp. 173-262.Heidelberg: Spnnger-Verlag.

Reuterdahl, C , Tingstrdm, A., Terracio, L., Fuua, K., Heldin, C.-H.and Rubin, K. (1991). Characterization of platelet-derived growthfactor /3-receptor expressing cells in the vasculature of humanrheumatoid synovium. Lab. Invest. 64, 321-329.

Roberts, A. B. and Sporn, M. B. (1990). Transforming growth factor-fk. In Handbook of Experimental Pharmacology. Peptide GrowthFactors and Their Receptors, vol. 95, part 1 (ed. Sporn, M.B. andRoberts, A.B.), pp. 419-458. Heidelberg: Springer-Verlag.

Ruoslahti, E. and Pierschbacher, M. D. (1987). New perspectives incell adhesion: RGD and integrins. Science 238, 491-497.

Schiro, J. A., Chan, B. M. C , Roswit, W. T., Kassner, P. D.,PenUand, A. P., Hemler, M. E., Eisen, A. Z. and Kupper, T. S.(1991). Integrin ai/5, (VLA-2) mediates reorganization andcontraction of collagen matrices by human cells. Cell 67, 403-410.

Scppii, H., Grotendorst, G., SeppS, S., Schlffmann, E. and Martin,G. R. (1982). Platelet-derived growth factor is chemotactic forfibroblasts. J. Cell Biol. 92, 584-588.

Siegbahn, A., Hammacher, A., Westermark, B. and Heldin, C.-H.(1990). Different effect of the various isoforms of platelet- derivedgrowth factor on chemotaxis of fibroblasts, monocytes andgranulocytes. J. Clin Invest. 85, 916-920.

Steinberg, B., Smith, K., Colozzo, M. and Pollack, R. (1980).Establishment and transformation diminish the ability of

fibroblasts to contract a native collagen gel. J. Cell Biol. 87, 304-308.

Thyberg, J., Ostman, A., Buckstrom, G., Westermark, B. andHeldin, C.-H. (1990). Localization of platelet-derived growth factor(PDGF) in CHO cells transfected with PDGF A- or B-chaincDNA: retention of PDGF-BB in the endoplasmatic reticulum andGolgi complex. J. Cell Sci. 97, 219-229.

Tingstrom, A., Reuterdahl, C , Lindahl, P., Heldin, C.-H. and Rubin,K. (1992). Expression of PDGF /S-receptors on human fibroblasts:Regulation by recombinant PDGF-BB, interleukin-1 and tumornecrosis factor-a\ J. Immunol. 148, 546-554.

Tomasek, J. J. and Hay, E. D. (1984). Analysis of the role ofmicrofilaments and microtubules in acquisition of bipolarity andelongation of fibroblasts in hydrtaed collagen gels. J. Cell Biol. 99,536-549.

Tomasek, J. J., Hay, E. D. and Fujiwara, K. (1982). Collagenmodulates cell shape and cytoskeleton of embryonic corneal andfibroma fibroblasts: Distribution of actin, o^actin, and myosin.Develop. Biol. 92, 107-122.

van Bockxmeer, F. M., Martin, C. E. and Constable, I. J. (1984).Effect of cyclic AM P on cellular contractility and DN A synthesis inchorioretinal fibroblasts maintained in collagen matrices. Exp. CellRes. 155, 413-421.

Wahl, S. M., Hunt, D. A., Wakefield, L. M., McCartney-Francis, N.,Wanl, L. M., Roberts, A. B. and Sporn, M. B. (1987).Transforming growth factor beta (TGF-beta) induces monocytechemotaxis and growth factor production. Proc. Nat. Acad. Sci.U.S.A. 84, 5788-5792.

(Received 20 December 1991 - Accepted, in revised form,6 March 1992)