EXPERIMENTAL

CELL

RESEARCH

202,316-325

(19%)

Effects of TGFPI on the Proliferation and Differentiation of an Immortalized Astrocyte Cell Line: Relationship with Extracellular Matrix DANIELETORU-DELBAUFFE,~DENISEBAGHDASSARIAN,DOMINIQUEBOTH,ROZENNBERNARD,* PIERREROUGET,*ANDMICHELPIERRE Unite’ 96 INSERM, 80, rue du G&&al Leclerc, 94276 Le Kremlin-Bicetre, Cedex, France; and *Laboratoire de Biologie MoGculaire Diff&enciation, Uniuersite’ Paris VI et CollZge de France, 11, Place Marcelin Berth&t, 75231 Paris Cedex 05, France

and in cells of mesenchymal origin. It is considered to be an extremely potent inhibitor of proliferation for all other cell types [ 2-51. The stimulatory effect of TGFP is complex; it occurs only at low concentrations and is often correlated with the induction of PDGF synthesis [5]. It has recently been suggested that the mitogenic signal of TGFP is coupled through G-proteins [6]. The cellular mechanism of TGFP inhibition is not yet fully understood and is now the subject of numerous investigations. TGF/3 also influences cell differentiation, positively or negatively, depending on the cell type [l-3]. With astrocyte primary culture, we have previously shown that TGFPl inhibits the expression of glutamine synthetase, which plays an important role in glutamate/glutamine homeostasis [7]. It also delays the induction of glutamine synthetase activity by bFGF in these cells [8]. Lastly, TGF/3 stimulates the expression of matrix components and provokes important changes in the composition and structure of the ECM. It slows the degradation of the extracellular matrix by inducing inhibitors of proteolytic enzymes and increases the production of cell adhesion molecule receptors [l-3]. The multifunctional nature of TGFP raises the question of its function in vivo as an important effector in processes of cell proliferation and differentiation as well as of morphogenesis, tissue development, and repair. ECM molecules were first studied in terms of their role in maintaining tissue structure. More recent studies in a variety of systems suggest that they are implicated in the control of biological functions [9, lo]. Many ECM components are present during the development of the mammalian nervous system. Some of them persist into adulthood, whereas others, such as laminin and fibronectin, appear onIy transiently during the embryogenesis of the central nervous system. Certain matrix components could influence neural development by promoting neuronal migration, axonal growth, and glial differentiation [ 11,121. Astroglial cells have been shown to be major substrates for axon growth during development, and interactions between neuronal and glial cells depend on membrane-associated cell adhesion molecules and secreted ECM constituents [ 12-161. Laminin

The astrocyte cell line (C.LT.T. 1.1.)) which is immortalized and has retained a normal density-dependent regulation of growth, is a suitable model for studying the relationships between proliferation, differentiation, and the production of extracellular matrix. The growth factor TGFBl was used to modulate these processes. When added to proliferative cells, it inhibited growth and caused morphological changes. It also suppressed the growth arrest at confluence, so that the cells formed multilayers of parallel spindle-shaped cells. Whereas untreated control cells expressed progressively the glial fibrillary acidic protein (GFAP) after arrest of multiplication, the addition of TGF&l to proliferative cells prevented GFAP expression and accumulation of its mRNA. Concomitantly, it increased the amounts of laminin, fibronectin, and collagens synthesized during the growth phase and greatly altered the composition and the structure of the matrix deposited at confluence. In contrast, when added after cell differentiation had begun, TGF@l did not alter the appearance of the matrix whereas it still stimulated, but to a lesser extent, extracellular matrix components production. The results show that TGFPl prevents the transition from the proliferating to the differentiating state and correlatively alters the composition and 0 1992 Academic structure of the extracellular matrix. Press,

Inc.

INTRODUCTION The members of the TGF p family are now known to be multifunctional factors [l-4]. So far, five forms have been identified by protein isolation or cDNA cloning. These forms often display similar activities, particularly in mammals. Their effects on cells depend on the environment, cell type, other associated growth factors, and the extracellular matrix. TGFP is known to be mitogenic in certain cell lines 1 To whom correspondence and reprint requests should dressed at the above address. Fax: (33-l) 49-59-85-40. 0014.4827/92

Copyright All

rights

$5.00 0 1992 by Academic Press, of reproduction in any form

be ad-

316 Inc. reserved.

et

TGFPl

ON IMMORTALIZED

is suspected to control the guidance of axons during the development of the central nervous system [ll, 17, 181. It supports neurite outgrowth in the peripheral nervous system, where it is a permanent component of basal laminae [19] and guides axonal regeneration [ll]. ECM molecules can also influence the motility and adhesion of a variety of cell types [lo, 11, 20-221. Several authors have suggested that the cell matrix is one of the elements that participate in the regulation of cell growth and differentiation 123-271. We have used an immortalized astrocytic clone (C.LT.T.l.l.) [28] to examine this hypothesis. These cells divide until they form a confluent monolayer, where they cease to proliferate and begin to differentiate. The properties of these cells lead us to consider them a model of astrocyte precursors capable of precursor-product transitions. Indeed, this line constitutes an homogeneous, clonal, and unlimited cell material with a clear-cut control of proliferation and differentiation [28]. In contrast, multiplication and differentiation occur at the same time with cultured primary astrocytes. In this study, we show that C.LT.T.l.l. cells, cultured in the presence of serum that contains an inactive form of TGFP [29], express high levels of laminin and fibronectin, which are strongly increased by the active form ofTGFfi1. Thus, C.LT.T.l.l. cells can be considered an appropriate tool for analyzing the relationship between ECM and proliferating or differentiating cells. In addition, the present study shows that TGFPl prevents transition from the proliferating to the differentiating state and that this effect is accompanied by changes in matrix composition and structure. These results suggest that the incorporation of certain ECM components into an integrated structure could contribute to promote the growth arrest and induce the differentiation processes. MATERIALS

AND

METHODS

Cell culture. The C.LT.T.l.l. cell line was derived-from transgenic mice carrying the polyoma virus large T gene [28]. Unless otherwise stated, the cells were plated at an initial density of lo4 cells per 35-mm dish and maintained in Dulbecco’s modified Eagle’s medium (DMEM) from GIBCO supplemented with 10% fetal calf serum (FCS) (Seromed, Berlin) with two medium changes per week. They were incubated with 100 pM human platelet TGFPl (prepared by Dr. D. Barritault, University Paris XII) for the times indicated in the legends to figures. The medium of cultures treated with TGFPl was renewed every day. Cell multiplication was estimated by measuring the DNA content per dish. [3~ ~hymidine incorporation. Confluent cells cultured in loo-mm dishes were maintained for 3 days at this stage, released by trypsinization, and plated at low density into 35-mm dishes. After 8 h, the cells were washed three times with DMEM and grown in DMEM supplemented.with 10% FCS. They were subjected (or not) to TGFPl, for the times indicated in the figure legends. [3H]Thymidine (2 &iI ml) was added to the medium for the last 5 h of incubation. The medium was then removed and [Wlthymidine incorporated into DNA was measured as previously described [7]. [‘%]Proline incorporation and collagenuse treatment. Cultures at the indicated times in figure legends were washed twice with proline-

ASTROCYTES

317

free DMEM and incubated at 37°C for 16 h in ? ml proline-free DMEM/ascorbate 50 pg/ml; containing 2.5 pCi/ml L-[i*C]proline (280 mCi/mmol from NEN Research Products) with and without TGF/?l. Labeled medium was harvested and protease inhibitors were immediately added (see Immunoblotting). Before analysis, the medium was centrifuged and adjusted to 3 mM &Cl,. The cells were rinsed twice with DMEM, scraped off in 500 ~1 of buffer (10 mM Tris-MCI, pH 7.4, 1 m&f EDTA, 10% hepsrin, 3 mM CaCl,), and homogenized. Aliquots of medium (500 ~1) and cell extracts (cells + cell-associated matrix) (250 ~1) were digested with 100 U/ml bacterial collagenase (type III, Calbiochem) for 2 h at 37°C. Digestion was stopped by precipitation with 10% trichloroacetic acid and the mixture was centrifuged. The pellets were treated by SDS sample buffer, boiled for 5 min, and volumes corresponding to equivalent quantities of cell-associated DNA were applied to SDS-PAGE (6%). The samples were electrophoresed under reducing conditions and the gels were impregnated with ENHANCE and analyzed by fluorography. Immunocytochemistry. The cells were seeded on dishes containing glass coverslips and grown in DMEM supplemented with 10% FCS. At various times, the cells on the coverslips were fixed with 4% paraformaldehyde in PBS for 30 min at room temperature and then permeabilized or not with 1% Triton X-100 in PBS for 45 a. Tie coverslips were then rinsed with PBS and incubated with a single primary antibody, or a mixture of monoclonal (mouse) and polyclonal (rabbit) antibodies for double labeling, for 1 h. The cultures were washed extensively and incubated with the appropriate labeled secondary antibodies. This step was repeated twice when two polycional primary antibodies were used for double labeling. The primary antibodies were rabbit anti-rat laminin antiserum (diluted 150, Calbio&em), rabbit anti-rat fibronectin antiserum (diluted l:lO@, Calbio&em), and mouse monoclonal anti-GFAP IgGl (diluted 1200, ICN). The secondary antibodies were Auorescein isotbiocyanate (FITC)conjugated goat anti-rabbit Ig (diluted 1:20, Amen&am), Texas red (TR)-conjugated goat anti-rabbit Ig (diluted 1:40, Amersham), and FIT&l-goat anti-mouse Ig (diluted 120; Amersham). Coverslips were mounted in Mowiol mounting medium and examined under the epifluorescence microscope. Immunoblotting. Cells were scraped 08 in 200 ~1,buffer (10 miw Tris-HCl, pH 7.4, E mM EDTA, 10% heparin) containing protease inhibitors (10 pg/ml leupeptin, 10 pglml antipain, 100 Kg/ml benzamidin, 50 fig/ml aprotinin, 1 pglml pepstatin, 100 &ml trypsin inhibitor, 1 rnM pbeny~met~y~s~lfo~yl fluoride) and sonically disrupted at 4°C by 5 s treatments with an MSH sonic oscillator. The proteins were separated on SDS-PAGE (5-42.5% acrylamide gradient). Volumes corresponding to equivalent quantities of cell-associated DNA were applied in each lane. Proteins were then transferred onto nitrocellulose membranes. The blots were incubated for 2 h at room temperature with the antib’odies previously used for immunochemistry: anti-laminin (diluted l/1000), anti-fibronectin. (diluted l/2000), antiGFAP (diluted l/1000). Laminin extracted from mouse F9 embryocarcinoma cells [30] and fibronectin from human plasma /Sigma) were used as a control. Immunoreactive protein bands were visualized by incubation with alkaline phosphatase-conjugated anti-rabbit or anti-mouse Ig (diluted l/1500; Sigma, Biosys) and subsequent color development. RArA blots. RNA were extracted from cells with guanidine thiocyanate/phenoi [31]. RNA (20 @g/lane) was fractionated by electrophoresis on glyoxal/dimethyl sulfoxide/l% agarose gels and transferred to Gene Screen (NEN) ny!on membranes under vacuum [32J. A cDNA probe specific to GFAP, kindly provided by J. de Vellis and N. Cowan [33], was labeled (4 X 10’ cpm//rg of DNA) by oligonucleotide-primed extension (Pharmacia kit). The membranes were prehybridized for 4 h then hybridized with the cDNA probe (i-2 X 10” cpm/ml) for 15 h at 42°C in 50% (v/v) formamide/5x SSC/ZX Denhardt’s reagent/O.l% SDS/200 pg/ml denatured DNA salmon sperm/l% dextran sulfate. Membranes were then treated as described by Sambrook et al. [32].

318

TORU-DELBAUFFE

ET AL.

0 FIG. 1,. Effect of TGFPI on C,.LT.T.UL.. ceU proliferatiun Cells were cultured in DMEM supplemented With lO% FCS without (0) or with (m) 100 pM TGFPl from the day of seeding. The D%\TAeontents of dishes were measured at the indicated times. The results shown are from a typical experiment from a group of similar experiments. The maximum deviations were less than 10%.

RESULTS Effixts of TGFP 3 on Cell Proliferation The growthcurve ofC.LT.T.I.l. cells (Fig. 1) shows a logarithmic growth phase followed by growth arrest at the confluence. From this stage, the cells began to differentiate. 1. When starting on the seeding -time, a daily treatm.ent with TGF@l produces two modifications in cell proliferation: a. The rate of proliferation was slower during the logarithmic growth phase. The .average doubling ti.me of cultures exposed to TGF@l was 36 h, instead of 24 h for control cultures (Pig. 1) This was likely due to inhibition of DNA synthesis, The kinetics of [3H]thymidine incorporation (Fig. 2) showed that TGl?@l caused a marked decrease of thymidine incorporation rate and/ or reduced the proportion of cells synthesizing DNA. It should be noted that TGFol did not delay the onset of DNA synthesis. b. The inhibition of growth by saturation density was suppressed. In the presence of TGFPl, the cells continued to divide after confluence, thus producing a dense multilayered culture. 2. The proliferative resp.onse to TGFPl depended on the growth stage at which the factor was added to the cells. Treatm.ent was started at various times postseeding, corresponding to daily intervals betwe.en the beginning of the exponential phase and postconfluence. TGF/31 was added 5 days before the cell number determination (Fig. 3). When the treatment started in the early growth phase and stopped before confluence, it inhibited proliferation as described above. This inhibition was not observed when TGF@l was added just before confluence. In this case, proliferation continued even

10

20

The

30

(hours)

F1.G. 2. Effect of TGFPl on ~[‘H]thymidine incarporation by C.LT.T.l.l. cells. Confluent cells were maintained 3 days in DMEM plus 10% FCS, trypsinized, and plated into growth medium without (0) and with (m) 100 p&f TGF@l. [3H]Thymidine (2 &iJml) was added 5 h before the indicatedtimes.

when confluence was reached. When added at confluence, TGFPl caused less cell proliferation, while the cells were not responsive to Tel?/31 if it ~was added at late confluence and they maintained their growth arrest. These results .demonstrated that the confluence was a critical point in the proliferatiwe cell response to TGF@l. Effects of TGFP 1 ,on Cell Differentiation

Mor@m!ogy The appearance the phase-contrast

2

of C.LT.T.l.l. cells observed with microscope during the logarithmic

6

10

14

18

Time (days) FIG. 3. Influence of growth stage on the proliferative cell response to TGFPl. CLT.T.1,1. cells were cultured in DMEM plus 10% FCS without (0) and with (m) 100 pM TGFfil added 5 days preceeding the indicated times, and the DNA content of the dishes was measured.

TGFPl

ON IMMORTALIZED

ASTROCYTES

319

4. Effect of TGFPl on the morphology of C.LT.T.l.l. cells, as shown by phase-contrast microscopy. TGFfll-untreated cells: A, Suent cells; B, confluent cells. TGFPl-treated cells: C, subconfluent cells after 1 day TGF,Bl exposure; D, confluent cells exposed to TGFPl for 1 day before confluence. TGFPl concentration was 100 PM. A, C and B, D same magnification, respectively. Bar, 35 ,~m. FIG.

SUbCOIl!

growth phase varied. The cells exhibited a variety of randomly oriented morphologies and displayed thin processes (Fig. 4A). At confluence, they became flat and formed a monolayer (Fig. 4 Exposure of sparsely plated cells to TGFPl for a few hours involved tbe appearance of larger and more flattened cells (Fig. 4C). If treatment was started about 1 day before confluence, the cells underwent a dramatic morphological change at confluence, becoming long and spindle-shaped. All the cells acquired a bipolar character and were arranged in parallel arrays over large areas of the dish (Fig. 4D). Proliferation continued, producing multilayers of eriss-crossed cells. In contrast, exposure of cells to TGFPl after confluence caused no morphological changes as compared to untreated cells. Production

of the Cytoskeletal Protein,

the exponential phase until press GFAP, provid daily with TGF@l.

postconfluence

Indeed, the level of GFAP m confluent cells were exposed to

was

did not ex-

not altered when

ies showed that if the treatment was stopped by transferring cells to TGFBB-free rn~di~~~ G reappeared progressively in the cell processes (not shown).

GFAP

Tbe astrocyte cytoskeletal protein, GFAP, began to be detected in C.LT.T.l.l. cells after confluence. Its expression progressively increased with time, and almost all cells contained GFAP around the nucleus and in the cell processes by about 20 days after confluence (Fig. 5A). As shown by immunofluorescence (Fig. 5B) and immunoblotting (Fig. 6A) cells cultured with TGFPl from

EfTects of -IXXflI

on EG

Protein

Procbxtisn

Collagen. The effects of TGF@l on colla sessed by [i4C]proline incorpora cell extracts (cells + cell-associ matrix). Amounts of proteins corresponding to equivalent quantities of cell-associated DNA from cultures were compared. T

320

TORU-DELBAUFFE

ET AL.

multiplication produced collagen at approximately the same rate at all culture times examined (Fig. 7). A new lower M,, collagenase-sensitive band corresponding to 125 kDa was present at each culture time. Fibronectin

and laminin.

These glycoproteins were detected by immunofluorescence and immunoblotting methods. The cells were double-immunostained with antibodies against native laminin and fibronectin. Inside the cells, before confluence, both glycoproteins were colocalized in a punctate and perinuclear pattern that then spread toward the cell processes (Figs. 8A and 8E). Only in areas of high cell density, laminin and fibronectin began to be deposited in an irregular, fibrillar matrix under the cell layer. When the confluence was attained, both fibronectin and laminin immunoreactivities became organized into a regular and fibrillar ECM which masked the cells

KDa

FIG. 5. Effect of TGFPl on GFAP expression, visualized by immunofluorescence. A, untreated cells; and B, cells treated with TGF@l (100 PM) from the replicative phase. All cells were analyzed at 28 days postseeding. Bar, 25 pm.

newly synthesized collagen was determined by quantifying the collagenase-sensitive proteins. Little [14C]proline was found in the collagenase-sensitive proteins in media or control cell extracts (not shown). Cultures incubated with TGFPl for 48 h showed large changes in the labeling intensity of electrophoretic bands degraded by collagenase. At least three collagenase-sensitive bands (185, 170, 160 kDa) were identified in the media (Fig. 7) and one band (185 kDa) in cell extracts. A single addition of TGF@l caused a strong increase in labeling. The TGFfll-stimulated increase in labeling of collagenase-sensitive proteins from cell extracts was much smaller than that of proteins secreted in media. Consequently, only the collagen secretion into the culture media was further investigated. The stimulation of secreted newly synthesized collagen promoted by 48 h TGFPl treatment decreased as the cells underwent terminal differentiation. In contrast, chronically treated cells which proceeded their

Kb

-2.4

TGf=R

_ + _ + _ + u uuu 2 12

3u

FIG. 6. Effect of culture time on GFAP and GFAP mRNA expression by C.LT.T.l.l. cells treated with TGFfll. (A) Western blot analysis: the proteins corresponding to 15 pg DNA per dish were electrophoresed and transferred to nitrocellulose as described under Materials and Methods. Lanes 1,2, and 3: short TGFPl treatment, corresponding to the immunoblots of GFAP from cells cultured for 7, 12, and 26 days. TGFj31 was added 4 days before the indicated times. Lanes 4 and 5: chronic TGFPl treatment. Cells were cultured for 12 and 20 days. (B) Northern blot analysis: RNA corresponding to 20 pg per lane was fractionated by electrophoresis and transferred onto nylon membranes. Lanes 1 and 2: chronic TGFPl treatment. GFAP mRNA prepared from cells cultured for 18 and 14 days. Lane 3: short TGFPl treatment. GFAP mRNA prepared from cells cultured for 14 days. TGFPl was added 4 days preceeding the indicated times. Lane labeled GFAP corresponds to adult mouse brain GFAP mRNA.

TGFPl 1

TGFBCoil.

ASTRGCUTES

2 -

+

ON IMMORTALIZED

-

+

+"-

-3+

+'

--++--++---+

20.5.

FIG. 7. Effects of time in culture and of TGFPl on collagenase-sensitive protein production. Samples of medium (corresponding to 25 fig of cell-associated DNA) from C.LT.T.l.l. cells labeled with [‘4C]proline in the presence or absence of TGFPl (100 pM) were electrophoresed without or after incubation with collagenase (Coil.). Fluorograms of the dried gels: (A) Proteins from cells cultured for 6 (lane l), 11 (lane 2), and 20 days (lane 3). TGFPl was added 2 days before the indicated times. (B) Lanes 1,2, and 3, the same culture times as in A, but TGFPl was added on the day of seeding.

(Figs. 8C and 86). We have observed that the appearance of fibronectin in the matrix preceded that of laminin, so that the distributions of laminin and fibronectin were very different at the beginning of ECM formation (Figs. 8B and 8F). After 16 to 20 h, laminin incorporated into ECM. Confluent cultures that had been replated at low density after being scraped off and examined by immunofluorescence some hours later showed the distribution of laminin and GFAP. Portions of the confluent monolayers were covered by ECM as indicated by strong laminin staining. The cells adjacent to these aggregates, which contained GFAP, did not have intracellular laminin immunostaining. The cells separated from the edges of aggregates, and that had migrated into the celldepleted space, accumulated intracellular and punctate laminin, but GFAP was not expressed. Extracellular deposits of laminin were associated with differentiated cells which did not divide, whereas intracellular laminin was seen in undifferentiated cells (Fig. 9) which were dividing [ 281. Short treatments with TGFPl (48 h) at different culture times did not alter the localization of laminin and fibronectin as compared to that in control cells. However, both glycoproteins were more abundant in treated than in untreated cultures, as shown by immunoblotting analysis (Fig. 10). When TGFPl was added before

the confluence and maintained a long time, the differentiation was prevented and co~~~rn~~a~tly the cells became elongated and organized in parallel arrays. As soon as the cells reached confluence, fibronectin was deposited in thick fibers oriented Iike the cells. trast, laminin was seen inside the cells a (Figs. 8D and 8H). At any of the cultu these cells contained no GFAP. However, Iaminin was not completely absent from the matrix because cultures stained without permeabilization with Triton showed some deposits of laminin. Thus, when cell proliferation continued after confluence under the influence of TGFPB, the structure and protein composition of the matrix was very different from that of differentiated cells and the cell morphology was also altered. Laminin and fibronectin wer by immunublotting in cell extracts and culture TGFfi1 added to cells for the final 48 h caused increase in their production. Anti-rat fibroneetin or ~arn~~~~ antibodies did not react with proteins from If serum (not shown). TGFPl effects on laminin and of cell extracts were further investigated. Cell extracts contained more fibronectin than laminin in both control and treated cells (Fig. 10). Fibronectin and laminin expression was high during proliferation. C~~~e~ue~tly, the TGFpl-induced accumulation of both glyco

322

TORU-DELBAUFFE

ET AL.

E‘IG. 8. Immunofluorescence staining for laminin and fibronectin expression. A-C and E-G are control cells; D and H are TGFpl-tre ated staining. E-H are anti-fibronectin staining. A and E, before confluence. B and F, distribution of laminin and cell s. A-D are anti-laminin fibr ,onectin, when the confluence happened. C and G, ECM after confluence. D and H, confluent cells exposed to TGFPl from subconflue ‘rice. TG Ffll concentration was 100 PM. Bar, 25 pm.

TGFPl

FIG. 9. Distribution of GFAP and laminin later with (A) anti-laminin and (B) anti-GFAP.

into the matrix was maximal creased progressively once the noteworthy that, like primary C.LT.T.1.1. cells produced only [14, 16, 341. 1 TGFR

‘-

2 +“-

3 +“-+“-

ON IMMORTALIZED

in control cells. Confluent Bar, 50 Km.

at confluence and decells differentiated. It is cultures of astrocytes, the B chains of laminin

4

5

F9 LN

A

FN FIG. 10. Effects of time in culture and of TGFPl on laminin and fibronectin expression in extracts of C.LT.T.l.l. cells. Western blot analysis: (A) Laminin; cell extracts corresponding to 0.5 pg DNA per lane were blotted witb antibodies to native rat laminin. The arrowbead indicates the position of laminin B chains from F9 cells (F9 LN). (B) Fibronectin; cell extracts corresponding to 0.3 pg DNA per lane were blotted with antibodies to native rat fibronectin. Lanes l-5: cells cultured for 5, 10,12,15, and 17 days. TGFPl was added 2 days prior the indicated times. The arrowhead indicates the position of human plasma fibronectin (FN).

ASTROCYTES

cells were replated

323

at low density, and immunostained

some hours

We show here that TGFpl is a potent modulator of certain astroeytie cell line functions, IS inhibits seruminduced proliferation before con of the important astrocytic marker of ~~~~rentia~ion, GFAP, after confluence. It suppresses confluence and causes ~~a~t~tat~~e changes in the EC Its inhibitory effect on cell mult fluence is associated with a sharp thesis rate. This fall could be due to a slackening of DNA replication rate and/or the fact that fewer cells are engaged in DNA synthesis. This corroborates our r sults obtained on astrocytes in primary cul.ture [7], an is compatible with reports w demonstrate that TGFPl is generally a potent inhibitor of cell proliferation for many kinds of cells [2, 31. Tb this inhibition is not yet well un Indeed, TGFPl has a complex on C.ET.T.1.P. cell proliferation Added before ence, it rapidly inhibits cell proliferation, In contrast, if’ it is added at the same stage, but for a longer ti induces a progressive increase in cell liferation does n esult from an n2c DNA synthesis a of cell division; it is a~~orn~a~ied by the formation of cell multilayers, demonstrating that growth arrest at confluence is overcome under these conditions. These results raise the uestions of how TGF/ll suppresses growth arrest an why the type of cell response to TGF~YL de ends an the growth phase.

324

TORU-DELBAUFFE

We cannot presently answer these questions. However, in NRK cells, the suppression of contact inhibition produced by TGFPl has been associated with an increase in the extent of intercellular communication by van Zoelen and Tertoolen [35]. Confluent cultures of C.LT.T.l.l. cells produce an extensive organized matrix with a regular laminin and fibronectin architecture. In contrast to laminin, fibronectin is deposited in the matrix at the beginning of ECM formation. There is a delay of 20 h before the two glycoproteins are colocalized in a network. It should be noted that Ard and Bunge [13] also found in cultured astrocytes another matrix component, heparan sulfate proteoglycan, that was deposited in the ECM prior to laminin. This delay may suggest that the incorporation of laminin in cell matrix is related to mechanisms promoting growth arrest. This hypothesis is supported by the fact that cells which have been treated with TGFPl before confluence, thus before matrix formation, deposit fibronectin but not laminin into the ECM in thick fibers oriented parallely to the cells, when these become confluent. Correlatively, proliferation continues and the cells become spindle-shaped and organized at first in parallel arrays and then in dense multilayers. In contrast, TGFPl no longer affects cell multiplication and matrix organization after growth arrest at confluence. Thus, there is a good correlation between the absence of laminin in the matrix and the maintenance of proliferative status in C.LT.T.l.l. cells. Some authors have identified components of ECM involved in regulating cell proliferation [20,26,36-421. Moreover, there is an association between inhibition of growth and altered morphology in C.LT.T.l.l. cells, in primary cultures of astrocytes [7] and in epithelial and endothelial cells [3] treated with TGFPl. Several studies suggest that ECM components regulate cell proliferation by altering the shape of cells [27,43,44]. Further investigations will be required to establish the mechanism of the correlation between cell proliferation, morphology, and ECM in C.LT.T.l.l. cells. TGFPl also affects C.LT.T.l.l. cell differentiation. The results reported here show that it inhibits the expression of GFAP, a cytoskeletal marker of astrocytic differentiation. This inhibition is reversible once TGFPl is removed. Similar reversions have been observed in adipogenic and myogenic differentiation [3]. TGFPl is effective in C.LT.T.l.l. cells only if it is added to cells before the confluence, thus before the cells undergo terminal differentiation. A hypothesis could be that TGFPl might act in. uiuo to delay the terminal differentiation of astrocytes. In contrast, if the treatment starts after the confluence, once the cells have began to differentiate, they become refractory to the action of TGF@. The mechanism of this refractoriness is unknown, but in 3T3-Ll cells and myoblasts this phenomenon is not due to a loss of TGFP receptors [3]. Massague

ET AL.

[3] proposes that TGFPl inhibition of myoblast differentiation occurs via a first mechanism that blocks the expression of a myogenic differentiation gene and a second one that probably involves changes in cell adhesion and matrix production. Other authors have correlated TGFP-induced changes in cell differentiation and ECM expression [l, 451. More generally, some results contribute to the thought that ECM is a major factor influencing cell differentiation [25, 26, 461. Ard and Bunge [13] have reported that astrocytes subcultured for 5 days in defined medium bear processes and produce GFAP, but not ECM. TGFPl very effectively stimulates laminin, fibronectin, and collagens expression by C.LT.T.l.l. cells, but this effect decreases as the cells differentiate. In addition, Heino and Massague [47] showed that tibrillar collagen completely blocks the biochemical differentiation of L6E9 myoblasts. Adams and Watt [48] showed that fibronectin inhibits the terminal differentiation of human keratinocytes. Several authors also believe that in regulating the expression of cytoskeletal genes, ECM influences mechanisms that modulate tissue-specific gene expression [49-511. The present study shows that elongated C.LT.T.l.l. cells which express GFAP in their processes no longer accumulate laminin in their cytoplasm. Conversely, retention of laminin in cells is correlated with a lack of GFAP. We suggest that laminin deposition into matrix could participate in growth arrest and influence cell differentiation. The results obtained with this astrocyte precursor model support the view that the structural and spatial organization of ECM components play a critical role in the transmission of information orienting the cells toward a proliferating or differentiating status. Obviously, further investigation is required to clarify the interactions between ECM components and the other elements that regulate cell fate. We thank Mr. Bahloul for help with photography and Miss M. Sorgue and Mrs. M. P. Decaudin for the preparation of this manuscript. This study was supported by grants from CNRS, the Association pour la Recherche contre le Cancer, and the University Paris XI.

REFERENCES 1. Rizzino, A. (1988) Den Biol. 130,411-422. 2. Roberts, A. B., Flanders, K. C., Kondaiah, P., Thompson,

N. L., Van Obberghen-Schilling, E., Wakefield, L., Rossi, P., De Crombrugghe, B., Heine, U., and Sporn, M. B. (1988). Rec. Prog. Horm. Res. 44,157-197. 3. Massague, J. (1990) Annu. Reu. Cell Biol. 6, 597-641. 4. Lyons, R. M., and Moses, H. L. (1990) Eur. J. Biochem. 187,

467-473. 5. Moses, H. L., Yang, E. Y., and Pietenpol, J. A. (1990) Cell 63, 245-247. 6. Howe, P. H., Cunningham, M. R., and Leof, E. B. (1990) J. Cell. Physiol. 142, 39-45. 7.

Toru-Delbauffe,

D., Baghdassarian-Chalaye,

D., Gavaret,

J. M.,

TGFPl Courtin, F., Pomerance, hem. 54, 1056-1061. 8. 9. 10. 11. 12.

13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28.

29. 30. 31.

M., and Pierre,

ON IMMORTALIZED

M. (1990) J. Neuro-

Labourdette, G., Janet, T., Laeng, P., Perraud, F., Lawrence, and Pettman, B. (1990) J. Cell. Physiol. 144, 473-484.

D.,

Ingber, D. E., and Folkman, J. (1989) Cell 58, 803-805. Zetter, B. R., and Brighman, S. E. (1990) Curr. Opinion CeZZBioZ. 2,850-856. Sanes, J. R. (1989) Annu. Rev Neurosci. 12,491-516. Reichardt, L. F., Bixby, J. L., Hall, D. E., Ignatius, M. J., Neugebauer, K. M., and Tomaselli, K. J. (1989) Deu. Neurosci. 11, 332-347. Ard, M. D., and Bunge, R. P. (1988) J. Neurosci. 8, 2844-2858. Liesi, P., and Risteli,

L. (1989) Exp. Neurol.

105, 86-92.

Rousselet, A., Autillo-Touati, A., Araud, D., and Prochiantz, A. (1990)Dev.Biol. 137, 33-45. Wujek, J. R., Haleem-Smith, II., Yamada, Y., Lipsky, R., TzeLan, Y., and Freese, E. (1990) Deu. Brain Res. 55, 237-247. Liesi, P., and Silver, J. (1988) Dev. Biol. 130, 774-785. Lander, A. D. (1990) Curr. Opinion Cell Biol. 2,907-913. Sephel, 6. C., Tashiro, K., Sasaki, M., Kandel, S., Yamada, Y., and Kleinman, H. K. (1989) Dev. Biol. 135, 172-181. Ruosiahti, E. (1989) J. Biol. Chem. 264, 13369-13372. Chiquet-Erisbmann, R., Kaila, P., and Pearson, C. A. (1989). Cancer Res. 49,4322-4325. Yamada, K. M. (1991) J. Biol. Chem. 266, 12809-12812. Hynes, R. 0. (1987) Cell 48,549-554. Ruoshlati, E. (1988) Annu. Rev. Biochem. 57,375-413. Reh, T. A., and Radke, K. (1988) Dev. Biol. 129, 283-293. Ingber, D. E., and Folkman, J. (1989) J. Cell Biol. 109,317-330. Ingber, D. E. (1990) Proc. Natl. Acad. Sci. USA 87, 3579-3583. Galiana, E., Borde, I., Marin, P., Rassoulzadegan, M., Cuzin, F., Gros, F., Rouget, P., and Evrard, C. (1990) J. Neurosci. Res. 26, 269-277. O’Connor-Mecourt, M. D., and Wakefield, L. M. (1987) J. Biol. Chem. 262, 14090-14092. Nicolas, J. F., Avener, P., Gaillard, J., Guenet, J. L., Jakob, H., and Jacob, F. (1976) Cancer Res. 361,4224-4231. Chomczynski, P., and Sacchi, N. (1987) Anal. Biochem. 162, 156-159.

Received March 17, 1992 Revised version received June 9, 1992

32.

33. 34. 35. 36. 37.

ASTROCYTES

325

Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989). Analysis of RNA in Molecular Cloning, 2nd ed., Vol. 1. pp 7.39-7.52, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Lewis, S. A. (i984) Proc. Natl. Acad. Sci. USA 81, 2743-2746. Chin, A. Y., Espinosa de Los Monteros, A., Cole, R. A., Loera, S., and De vellis, J. (1991) GZia 4, II-24 Van Zoelen, E. J. J., and Tertoolen, L. G. J. (1930) J. Biol. Chem. 266, 12075-1208:. Geotschy, J. F., Ulrich, G., Aunis, D., and Ciesielski-Treska J. (1987). Pnt. J. Dev. Neurosci. 5, 63-70. Madri, J. A., Pratt, B. M., and Tucker, A. M. (1988) J. Cell Biol.

106,1375-1384. 38.

39. 40.

Wright, T. C., Jr., Castellot, J. J.: Jr., Petitou, NE., Lormeau, J. C., Choay, J., and Karnovsky, M. J. (1989) J’. Biol. Chem. 264, 1534-1542. Nugent, M. A., and Newman, M. J. (1989) j. 61ioI. Chem. 264, 18060-18067. Hamati, H. F., B&ton, E. L., and Carey, D. Jo (3989) j. CellBiol.

108,2495-2505. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51.

Gatchalian, C. L., Schachner, M., and Sanes, J. R. (1989) J. Cell Bioi. 108,1873-1890. G. D., Roberts, L. A., Liotta, L. A., and Giavazzi, R. Taraboletti, (1990) J. Gel!: Biol. 111, 765-772. Boyd, F. T., and Massague, J. (1989) ri. Bioi. C&m. 264,22722278. Newton, L. K., Yung, W. K. A., Pettigrew, L. C., and Steck, P. A. (1990). Exp. Cell Res. 196, 127-132. Williams, C. A., and Allen-HofFmann, B. L. (1990) J. Biol. Cfr,em. 265,6467-6472. Aratani, Y., and Kitagawa, Y. (1988) J. Bioi. Chem. 263,1616316169. Heino, J., and Massague, J. (1990) J. Bio2. &em. 265, 1018110184. Adams, J. C., and Watt, F. M. (1989) Nature 346, 307-309. Machie, E. J., Thesleff, I., and Chiquet-Esishmann, E. (1987) J. Cell Biol. 165, 256992579. Ben-Ze’ev, A., Robinson, 6. S., Bucher, N. L. R., and Farmer, S. R. (1988) Prac. Natl. Acad. Sci. USA 85, 2169-2165. Chiquet-Erishmann, R., Kalla, P., Pearson, C. A., Beck, K., and Chiquet, M. (1988) CeEl53, 3833390.