Regulation of Cell Adhesion Receptors by ... - Semantic Scholar

THEJOURNAL OF BIOLOGICAL CHEMISTRY

Vol. 264, No. 1, Issue of January 5, pp. 380-388,1989

0 1989 by The American Society for Biochemistry and Molecular Biology, Inc.

Printed in U.S.A.

Regulation of Cell AdhesionReceptors by Transforming Growth Factor-fl CONCOMITANT REGULATION OF INTEGRINS THATSHARE A COMMON 81 SUBUNIT* (Received for publication, August 17, 1988)

Jyrki HeinoS, Ronald A. Ignotzp, Martin E. Hemlerll, Carol Crousen, and JoanMassague From the Department of Biochemistry, University of Massachusetts Medical School, Worcester,Massachusetts 01655 and the TDana-Farber Cancer Institute, Haruard Medical School, Boston,Massachusetts 02115

Cell adhesion to extracellular matrices is mediated cell adhesion receptors may interact with actin filament conby a set of heterodimeric cell surface receptorscalled necting proteins, suggesting that integrins link extracellular integrins that might be the subject of regulation by matrix to cytoskeleton. The basic structure of integrins is a growth and differentiation factors. We have examined heterodimeric complex of one 130-200-kDa a subunit linked the effect of transforming growth factor-81(TGF-81) noncovalently to one 90-130-kDa p subunit. Three different on the expressionof the very late antigens or group p subunits have been identified. They are structurallyrelated of integrins in human cell lines. The six knownmem- to each other, with about 44-46% amino acid sequence idenbers of this familyshare a common 81subunit but have tity (3-6), and they define three families of integrins (1). distinct a subunits that confer selective affinity towardThe prototype integrin is the fibronectin receptor originally type I collagen, fibronectin, laminin, and other as yet purified from human placenta (7) and from human osteosarunknown cell adhesion proteins. Usinga panel of specific antibodies and cDNA probes, we showthat inWI- coma cells (8).It recognizes the Arg-Gly-Asp sequence present 38 lung fibroblasts TGF-81 elevates concomitantly the in the fibronectin molecule or in synthetic peptides (9). Its a expression of a l , aZ,as, ab, and B1 integrin subunits at subunit is divided in two fragments that derive from proteothe protein and/or mRNA level, their assembly into the lytic processing of the translation product but remain disulfide-bonded to one another in the maturepolypeptide (4). corresponding aB1 complexes, and their exposure on Independent studies on T-lymphocyte surface membrane the cell surface. The rate of synthesis of total a subunits relative toDl subunit is higher in TGF-81-treatedcells proteins have revealed a family of heterodimeric cell adhesion receptors, originally termed very late antigens (VLAs)’ (10than in controlcells. The characteristically slow (tn 12). The VLAs share one common subunit, now known as the 10 h) rate of conversionfromprecursorformto mature glycoprotein in untreated cells increases mark- /I1 integrin subunit. The VLA or aP1 integrin family consists edly (to tn 3 h) in response to TGF-81. The results of at least six different members (12). At least four of the suggest that in WI-38 fibroblasts the 81 subunit is members of this family mediate adhesion of cells to extracelsynthesized in excess overa subunits, and assembly of lular matrix proteins. alplintegrin (VLA-1) was firstdesubunits with rate-limiting a subunits is required scribed in activated peripheral blood lymphocytes. However, for transit through the Golgi and exposureof aB1 com- it is absent from most other hematopoietic cells, while it is plex on the cell surface. TGF-81 does not induce the expressed by several other cell types, predominantly fibrosynthesis of integrin subunits that are not expressed in blasts. alp1integrin appears to be a cell adhesion receptor, unstimulated cells, suchas a4and (Ye subunits inWI-38 but its ligand is unknown (10, 12, 13). a& integrin can bind fibroblasts. However, a4 and f f 6 subunits canbe regu- typeI collagen and a& integrin mediates attachmentto lated by TGF-8 in those cells that express them. The laminin, collagen, and fibronectin (14, 15). a& (VLA-4) is results suggest that TGF-8 regulates the expression of widely expressed in leukocytes, but it is usually absent from individual integrin subunits by parallel but independent mechanisms. By modifying the balance of individ- most adherent cell types (16). The a& integrin was independently identified as aVLA-5, but amino acid sequence and ual abl integrins, TGF-81 might modulate those aspects of cell migration, positioning, and development that immunological evidence have indicated that it is the same as the classical fibronectin receptor, initially called also a& are guided by adhesion toextracellular matrices. (17). a& (VLA-6) has been recently identified as a novel integrin in human platelets (18). The function of a& and a& integrins is still unknown. A large bodyof evidence Integrins are afamily of membrane spanning glycoproteins indicates that the aPl integrins play an important role as that mediate cell adhesion to extracellular matrix and base- mediators of cell adhesion and migration, and their function ment membranes, to other cells, andto plasma proteins influences cell proliferation and differentiation (for reviews (reviewed in Refs. 1and 2). The cytoplasmic domains of these see Refs.1and 2).Recent studies have shown that thenormal function of a& integrins is critical in certain differentiation * This work was supported by National Institutes of Health Grants and developmental processes. For example, disruption of fiCA39240 (to J. M.) and CA42368 (to M. H.). The costs of publication bronectin receptor function by addition of specific antibodies of this article were defrayed in part by the payment of page charges. to chick embryo myoblasts alters myogenic differentiation of

-

-

This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solelyto indicate this fact. $Supported by a postdoctoral fellowship from the Academyof Finland. § Supported by a postdoctoral fellowship from the National Institutes of Health.

‘The abbreviations used are: VLA, very late antigen; TGF-01, transforming growth factor-pl; HEPES,4-(2-hydroxyethyl)-l-~iperazineethanesulfonic acid EGTA, [ethylenebis(oxyethylenenitrilo)] tetraacetic acid.

380

Integrins TGF-p Regulates these cells (19). Furthermore, mutations that disrupt the P integrin subunit structural gene in Drosophila lead to a phenotype characterized by a lethal failure to assemble muscular tissue during the fly's embryogenesis (20). In previous studies, we have shown that expression of fibronectin receptors in fibroblasts and epithelial cells from human, rat, and mouse origin is regulated by transforming growth factors-pl and -P2 (TGF-P1, TGF-P2). Increases in the level of fibronectin receptor P1 subunit mRNA and subunit protein caused by TGFs-/3 lead to a higher level of fibronectin receptor on the cell surface, higher fibronectin binding capacity and elevated adhesion of cells to fibronectin matrices (21). It has also been shown that fibroblasts treated with TGF-Pl retain higher levels of fibronectin receptor a5 subunit mRNA (22). TGFs-P are representative of a large family of hormonally active polypeptides that control growth, differentiation, and morphogenesis in cultured cells and organisms from insects to mammals (reviewed in Ref. 23). This family includes members implicated in developmental processes such as embryogenic dorsoventral patterning and imaginal disc formation in Drosophila (24), and morphogenesis in Xenopus early embryo (25). TGFs-/3 themselves can act as inducers of mesodermal myogenesis markers in Xenopus embryos (26, 27). In addition to their ability to regulate expression of fibronectin receptors, TGFs-/3 are potent stimulators of the expression of extracellular matrix proteins including fibronectin (28), various types of collagen (28, 29), matrix proteoglycans (30), as well as inhibitors of matrix-degrading proteases (31, 32). The extensive changes induced by TGFsP in extracellular matrixproteinsare regarded asevents through which these factorsmay control expression of specific phenotypes in vitro and may influence tissue formation and repair in vivo (21, 23, 29). We have explored the possibility that TGFs-P might regulate other members of the integrin family in addition to the fibronectin receptor. This report focuses on the large group of integrins that share the P1 subunit. A companion paper (33) describes the regulation by TGF-p of two other families of integrins, those defined by the presence of P2 and P3 subunits and involved in cell-cell and cell-vitronectin adhesion. Here, we show that TGF-P1 canmodulate the expression of several coexisting aP1 integrins. The results of this study also provide information on the process of assembly and maturation of mammalian aPl integrins.

381

HEPES, pH 7.4. Iodination was carried out on ice by addition of 0.5 mCi of Na"'I (Amersham Corp.; 16 mCi/pg), 0.1 mg of lactoperoxidase, and 0.005% H202with periodic mixing for 10 min. Cells were washed four times in the same buffer before solubilization and immunoprecipitation. Zmmunoprecipitation-Biosynthetically labeled cell monolayers were washed on ice with a buffer containing 150 mM NaCl, 1 mM CaC12, 1 mM MgCl,, and 25 mM Tris-HC1, pH 7.4,followedby detachment of the cells by scraping. Cell pellets obtained by centrifugation at 500 X g for 5 min were solubilized in 0.2 ml of the same buffer containing 100 mM n-octyl-6-D-glycopyranoside(Sigma) on ice for 10 min with occasional vortexing. The insoluble material was removed by centrifugation at 13,000 X g for 4 min at 4 "C. Triton X100 (0.5 v/v %) and bovine serum albumin (0.5 mg/ml) were added to the supernatants, which were then precleared by incubation with 20 ~1 of packed protein A-Sepharose beads (Pharmacia LKBBiotechnology Inc.). The resulting supernatants were immunoprecipitated with antibodies for 12 h at 4 "C. Conditions for quantitative immunoprecipitation of the corresponding cellular components were established for each antibody preparationin preliminary experiments. Rabbit nonimmune serum and monoclonal rat anti-mouse immunoglobulin K chain antibodies were used as negative controls. Immune complexeswere recovered by binding to protein A-Sepharose, washing the beads four times with Tris-buffered saline containing 0.1% Triton X-100 and 1 mg/ml bovine serum albumin, and once with 150 mM NaCl and 25 mM Tris-HC1, pH 7.4. The immunoprecipitates were analyzed by electrophoresis on 7% or 8% polyacrylamide-dodecyl sulfate gels followed by Coomassie-staining, fixing, drying, and exposing the gels to fluorography using Enlightening (DuPont-New England Nuclear). Gel electrophoresis was performed in the absence of reductant unless otherwise indicated. Western Zmmunoblot-WI-38 fibroblasts were detached as described above, counted using a Coulter Counter, lysed into electrophoresis sample buffer, and subjected to dodecyl sulfate-polyacrylamide gel electrophoresis. Proteins were then transferred electrophoretically onto nitrocellulose membranes for 1 2 h at 250 mA and at 4 "C. Membranes were incubated for 2 h with 30 pg/ml of bovine serum albumin in 0.15 M NaC1,25 mM Tris, pH 7.4, andthen overnight with the same buffer containing apolyclonal anti-fibronectin receptor antiserum at 1:lOO dilution. The membranes were washed with the same buffer, incubated for 2 h with 1251-labeledprotein ASepharose (2-10 pCi/pg, DuPont-New England Nuclear) in the same buffer, washed, and subjected to autoradiography. EndoglycosidaseH Treatment-For endoglycosidase H (endo-6-Nacetylglucosaminidase, EC 3.2.1.96; Boehringer Mannheim) digestion, immunoprecipitates bound to protein A-Sepharose were mixed with 300 p1 of0.2 M sodium citrate buffer, pH 5.5, containing 5 milliunits of enzyme, and incubated for 20 h at 37 "C. After incubation, the Sepharose beads were collected by centrifugation heated in sample buffer, and analyzed by electrophoresis. Northern Blot Hybridization-Northern hybridization analysis of integrin mRNA was done using total cellular RNA prepared by the guanidine isothiocyanate, CsC1 centrifugation procedure (39). The EXPERIMENTALPROCEDURES RNA was fractioned on 1%agarose gels containing formaldehyde, Antibodies-Polyclonal anti-P1integrinsubunitantiserum was transferred onto nitrocellulose, and hybridized to cDNA probes speraised in rabbits by immunization with human placenta fibronectin cific for a5 integrin subunit (40), or Dl subunit (4): A cDNA probe receptor affinity-purified according to the method of Pytela et al. specific for glyceraldehyde-3-phosphate dehydrogenase (41) was used (34). The following monoclonal antibodies all against human proteins as a control. The cDNA probes were labeled with 32Pusing the were used A-1A5 specific for mature Dl subunit (lo), TS2/7 specific Multiprime DNA labeling system (Amersham Corp.). for a1 subunit (13), 12F1 specific for a2 subunit (16),5143 specific for 013 subunit (11, 35), B5G10 specific for a4 subunit (36), and GOH3 RESULTS specific for 016 subunit (18). In the immunoprecipitation experiments some of the monoclonal antibodies were used with a secondary Identification of apl Integrin Subunits and Their Biosynantibody: either with rat anti-mouse K chain (mAb 187.1) (37) or with thetic Precursors in WI-38 Cells-The human fibronectin mouse anti-rat K chain (RG7/9) (38). a5 subunit and one P1 subunit held receptor consists of one Cell Culture-WI-38 human lung fibroblasts (American Type Culture Collection) were grown in minimum essential medium supple- together via noncovalent interactions that are stabilized by mented with 10% fetal bovine calf serum. For biosynthetic labeling divalent cations (1, 2). The P1 subunit is present in at least experiments, cells were grown to confluence and labeled with 50 pCi/ five additional human integrins that contain distinct CY subml of [%]methionine (Amersham Corp.; >800 Ci/mmol) or 20 pCi/ units (1, 12). To raise antibodies that would immunoprecipiml of [?3]cysteine (Amersham Corp.; 1000Ci/mmol) in methionine- tate all these integrins, we immunized rabbits with human or cysteine-free minimum essential medium, respectively. placenta fibronectin receptor purified by affinity chromatogCell surface iodinations were carried out with cells in suspension. Cells were detached by incubation for 15 min a t 37 "C with a solution raphy. The resulting polyclonal antibodies recognized P1 subcontaining 125 mM NaCl, 5 mM KCl, 10 mM Tris, 1 mM EDTA, pH unit but not 015 subunit in purified fibronectin receptor prep7.4, and recovered by centrifugation at 500 X g for 5 min. The samples arations as determined by immunoblot analysis (Fig. 1A). were adjusted to contain equal numbers of cells in 0.5 ml of a buffer consisting of 128 mM NaC1, 5 mM KCl, 1.2 mM CaC12, and 25 mM R. Hynes and V. Patel, personal communication.

TGF-p Regulates p, Integrim

382

B

A 1

2

C 1

3 4

2

mw

k Da

3

1 2 3

4

200-

05-

k Da

- 200

150K a-

FIG. 1. Fibroblast integrin subunits recognized by anti-&subunit antibody. A, fibronectin receptor CY^ andsubunits purified from humanplacenta were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresisand were silver-stained (lane I ) or probedwith anti-human fibronectinreceptorpolyclonal antiserum in Western immunohlot assays (lane 2). Western immunoblots were also performed with octylglycoside extracts of WI-38 human lung fibroblasts (lanes 3 and 4, IO6 and 5 X IO5 cells per sample, respectively) using the same antibody preparation. After incubation with antiserum, Western blot membranes were probed with I2'Iprotein A and subjected to autoradiography. Themolecular mass of protein standards run in parallel is indicated in kilodaltons. predl denotes PI subunit biosynthetic precursor. R, confluent cultures of WI-38 cells were incubated for 18 h in serum-free medium without (lanes I and 3 ) or with (lows 2 and 4 ) 250 PM TGF-01. Cells were either metabolically labeled with50 pCi/ml of ["S]methionine in methionine-free medium during the last 6 hof incubation (lanes I and 2 ) or were surface-laheled with l2'1 at the end of the incubation (lanes 3 and 5). Detergent-soluble extracts of the labeled cells were immunoprecipitated with polyclonal antiserum against fibronectin receptor and analyzed hy sodium dodecyl sulfate-polyacrylamide gel electrophoresis and fluorography. The position and molecular mass in kilodaltons ( K )of the integrin subunits areindicated. C, WI-38 fibroblastswere metabolically labeled with [%]methionine in the presence of TGF-P1 asdescribed above. Aliquots of the detergent-soluble extract from those cells were immunoprecipitated with anti-fihronectin receptor polyclonal antiserum (lone I ) or monoclonal antibody specific for mature B1 integrin subunit (lane 2). Immunoprecipitated samples were displayed by gel electrophoresis, followed by 48-h fluorography. Lane 3, a 12-h fluorographic exposure of lane I.

Polyclonal antibodies immunoprecipitated a5 and PI subunits under conditions that preserve integrin complex stability, but they precipitatedonly PI subunit if the calcium chelator EGTAor the ionic detergent sodium dodecyl sulfate were presentduring immunoprecipitation (resultsnot shown). These observations indicated that the polyclonal antibodies recognized predominantly epitopes in the PI subunit. The ability of these antibodies to precipitate as and other a subunits under conditions that stabilize integrin complexes (see below) could be due to either recognition of conformational epitopes in assembled a subunits or, most likely, coprecipitation of a subunits thatwere assembled withPI subunit. These polyclonal antibodies were highly specific as they recognized only PI subunits or their biosynthetic precursors in immunoblots of crude extracts from WI-38 human fibroblasts (Fig.

lA ). Immunoprecipitation of detergent extracts from ["S]methionine-labeled WI-38 cells using this polyclonal antiserum yielded labeled products of 105, 115, 130, 150, and 190 kDa, respectively (Fig. 1B). Only the 130-, 150-, and 190-kDa products could be detected by precipitation of extracts from WI-38 cells surface-labeled with I2'I (Fig. 1B). Based on their properties, these products were tentatively identified as biosynthetic precursors of the PI integrin subunit (the 105- and 115-kDa products), mature PI subunit exposed in part on the cell surface (the 130-kDa product), and at least two a subunits (150- and 190-kDa products) coprecipitated with PI subunits. The identity of these products was confirmed in part by pulse-chasemetaboliclabeling experiments using ["'Ss]cysteine to label selectively the cysteine-rich PI subunit (3, 4). The 105-115-kDa precursors were progressively converted into the 130-kDa mature PI subunit (Fig. 2 A ; see Fig. 6 below

A

chase, minutes 10 20 30 60 360

C

8

Endo H - +

-DTT

I -

k OB -200

FIG. 2. Biosynthesis of 81 integrin subunit in WI-38 fibroblasts. A, confluent culturesof WI-38 fibroblastswere labeled for 10 min in serum-free, cysteine-free medium with 100 pCi/ml of ["SI cysteine. The cells were then placed in complete unlabeled medium and harvested 10-360 min later. Detergent-soluble cell extracts were precipitated with anti-fibronectin receptor polyclonal antiserum and analyzed by gel electrophoresis and fluorography. E , cellular proteins in WI-38 fibroblasts were labeled for 6 h in methionine-free medium with 50 pCi/ml of [''Sjmethionine. Cell extracts were immunoprecipitated with anti-fihronectin receptor polyclonal antiserum. Immunocomplexes hound to proteinA-Sepharose beads were incubated with (+) or without (-) endoglycosidase H (Endo H),and were then analyzed by gel electrophoresis and fluorography. C, WI-38 cells were metabolically labeled with 20 pCi/ml of ["S]cysteine in serum-free and cysteine-free medium for 2 h. Cell extracts were immunoprecipitated with anti-fibronectin receptor antiserum and the precipitants were subjected to gel electrophoresis in the presence (+) or absence (-) of dithiothreitol (DTT).

TGF-p Regulates PI Integrins

383

for longer chase time points). The PI subunit biosynthetic precursor(s) migrated in sodium dodecyl sulfate-polyacrylamide gels as a broad species thatin some experiments kDa resolved well into twobands. Thisheterogeneity was not eliminated by treatment of immunoprecipitated sampleswith - 200 remove immature endoglycosidase H, an enzyme that can (endoplasmic reticulum-type) N-linked oligosaccharide side - 116 chains from glycoproteins (Fig. 2B). However, the heteroge- 94 neous appearance of the PI subunit precursors was eliminated by reduction of the samples before electrophoresis (Fig. 2C). Under reducing conditions, all of the labeled subunit pre- 68 cursor material migratedas a tight 115-kDa band, suggesting that the heterogeneity was due to the presence of different - 45 forms corresponding to progressive stages of disulfide bond formation in the cysteine-rich Dl polypeptide chain and proFIG.3. Effects of TGF-81 on the biosynthesis of individual gressive acquisition of molecular compactness due to inter- integrin subunits in WI-38 fibroblasts. Confluent cultures of chain disulfide bonding. The decrease in electrophoretic mo- WI-38 cells were incubated with or without 250 PM TGF-Dl for 18 h, bility of mature integrin p subunits upon reduction has been the last 6 h in the presence of 50 pCi/ml of [3SS]methioninein methionine-free medium. Cells were harvested and aliquots of deterconsistently noted inprevious reports (1, 2). T o assess further the nature of the products precipitated gent-soluble cell extracts were immunoprecipitated with monoclonal antibodies against the indicated integrin subunits, or with a polyby the polyclonal antibody, samples of metabolically labeled clonal antiserum against fibronectin receptor (poly&). ImmunopreWI-38 cells were precipitated with monoclonal antibody di- cipitated proteins were analyzed by gel electrophoresis and fluorogrected againstthematurehumansubunit.Thisantibody raphy. precipitated the mature PI subunit but not its biosynthetic precursors (Fig. 1C). More importantly, this antibody also 0 4 8 12 Hours precipitated the 150- and 190-kDa a subunits. These results indicated that WI-38 human fibroblasts contain multiple a integrinsubunitsthatare assembled with PI subunitsas functional integrincomplexes and canbe coprecipitated with the PI subunit by anti-pl subunit antibodies. Integrin Subunits Regulated by TGF-&Treatment of WI38 cells with TGF-B1 increased the rate of synthesis of PI 8 r) .* -GAPDH subunit and the level of this subunit on the cell surface. Importantly, TGF-B1 treatment alsoled to a marked increase FIG.4. Regulation of 8, and aa integrin mRNA levels by in the levels of precipitable LY subunits (Fig. 1B). The higher TGF-B1. WI-38 lung fibroblasts were incubated with TGF-81for the level of a subunits in samples immunoprecipitated from TGF- indicated lengths of time. Cells were harvested and total cellular RNA &treated WI-38 fibroblasts could derive from increased co- .was isolated. RNA samples (20 pg) were electrophoresed in agarose precipitationdue to higheravailability of subunit. Alter- gels, transferred onto nitrocellulose membranes, and subjected to Northern blot hybridization using 32P-labeledcDNA probes specific natively, the increase in a subunits could be due to a direct for humanDl and asintegrin subunits, or glyceraldehyde-3-phosphate effect of TGF-Bl on theirexpression. To distinguish between dehydrogenase (GAPDH) as a control. The panel show sections of these two possibilities, extracts from [35S]methionine-labeled autoradiograms corresponding to probe-hybridizing mRNAs of 4.2 cells were precipitated with a panel of anti-integrin mono- kilobases (PI),4.9 kilobases ( @ e ) , and 1.3 kilobases glyceraldehyde-3clonal antibodies, including antibodies against al,a2,a3,a4, phosphate dehydrogenase. and a 6 subunits in addition to monoclonal and polyclonal antibodies against PI subunit. Unstimulated WI-38 cells ex- precipitated with anti-& subunit monoclonal antibodies (Fig. pressed 190-kDa aIsubunits, 150-kDa a2 subunits, and 155- 3). Thus, the use of this panel of antibodies allowed a close kDa a3 subunits as detectedby precipitation with the corre- quantitation of assembled integrin complexes in WI-38 cells. sponding antibodies. Furthermore, treatment of cells with Assuming a 1:l stoichiometry of a and /3 subunits in integrin TGF-Pl markedly increased the level of newly synthesized al, complexes and given the methionine content of a and j3 a ~and , a3 subunits (Fig. 3). Expression of a4 subunit could subunits deduced from the corresponding cDNA sequences (4); the results shown in Fig. 3 suggest that unstimulated not be detected in either control or TGF-pl-treated WI-38 cells after cell labeling with [3sS]methionine (Fig. 3) or [35S] cells synthesized subunits a t a faster rate than a subunits. cysteine (not shown). Anti-as subunit antibodies precipitated Antibodies against a5subunits were not available for these a 165-kDaproduct whose level was slightly increased by TGF- studies, but Northern blot analysis of WI-38 RNA using a5 D l (Fig. 3). This product is larger than the 140-150-kDa 0 6 subunit andDl subunit cDNA probes indicated that treatment subunits previously identified in other cell types (18). The of WI-38 cells with TGF-p1 increased the steady-state level relationship of this 165-kDa product to (Y6 subunit remainsto of a5subunit mRNA in parallel with the increase in subunit mRNA (Fig. 4). The t8,+of this effect was between 4-8 h. be clarified. The effect of TGF-B1 on the biosynthesis and cell surface The amount of mature & subunit coprecipitated with al, expression of multiple integrin subunits occurred in the ab02, and a3 subunits added together was close to the amount sence of a general increase in protein synthesisfor a t least 24 of maturesubunitprecipitatedwithitscorresponding monoclonal or polyclonal antibodies (Fig. 3) suggesting that h (Fig. 5). We note that treatmentof WI-38 cells with TGFthe contributionof other a subunits (e.g. as)to the totalpool p l did not always increase the levels of all integrin subunits in parallel. Fig. 5 illustrates one case in which the strong of a subunits assembled with subunit in WI-38 cellswas relatively small. Reciprocally, the amount of a subunits precipitated with their antibodies was similar to the amount Unpublished results.

384

TGF-/3 Regulates

Integrins

-

A. Anti-81 antibody elevation of aI subunit expressionby TGF-P1 occurred in - TGF-P1 + TGF-81 parallel with a weak response in PI subunit synthesis and no 0 1 3 712 0 1 3 712 significanteffect in m2/a3 (150-kDa a) subunitsynthesis. q7-TGF-P1 increased the synthesis of a l , a2, and a3 integrin subunits in 10 outof 10,7 outof 9, and 6 out of 8 experiments, -"a1 respectively. The average-fold increase induced by TGF-P1 -150Ka was 10-fold for the CY^ subunit, 2.5-fold for the a2subunit, and a P ~ A . . . 0.0s -ApreB1 3-fold for the a3 subunit, relative to control cells. Although the variables that influence the responsiveness of individual integrin subunits to TGF-81 in WI-38 fibroblasts have not 6.Anti-a antibody been defined yet, these results suggested that expression of - TGF-01 + TGF-81 individual integrin subunits mightbe regulated by independ0 1 3 712 0 1 3 712 ent mechanisms. Modulation of Integrin Complex Assembly by TGF-P1: Role ia1 of a Subunit Availability-The biosynthetic maturation and veal assembly of a and /3 integrin subunits and their modulation -Dl by TGF-P1 were examinedin moredetail. Parallel pulsechase metaboliclabeling assays were performed in control and TGF-Dl-treated WI-38 cells, followed by precipitation FIG.6. Effects of TGF-Dl on the rate of and a1 integrin with anti-al subunit antibody and anti-& subunit antibody. The amountof PI subunit precursor synthesized duringa 1-h conversion from precursor forms to mature glycoproteins. Confluent culturesof WI-38 fibroblasts were incubated in serum-free pulse with ["S]methionine was slightly higher in TGF-Dlmedium for 12 h, labeled in methionine-free medium with 100 pCi/ treated cells compared to controlcells (Fig. 6A). More impor- ml of [''S]methionine for 1 h, thenshiftedto unlabeled normal tant,thefate of this newly synthesizedmaterial differed medium for the indicated lengths of time (hours).In some cell cultures markedly between control and TGF-Dl-treatedcells. In con- TGF-Pl (250 PM) was present throughout all incubations. Soluble trol cells, the PI subunit precursors matured a t a very slow cell extracts were immunoprecipitated with the indicated antibodies rate(tHapproximately10h)characteristic of PI subunit and analyzed by gel electrophoresis and fluorography. maturation in other fibroblasts (21,42). Furthermore,a large portion of the PI subunit precursorpool in controlcells turned subunitsand was not markedly altered by TGF-Dl even over beforeit reached maturation (Fig. 6A). The accumulation though this treatment increased aI subunit synthesis rate of mature Dl subunit in TGF-Dl-treated cells was faster (tH (Fig. 6B). Like the mature a1subunit, the precursorcould be approximately 3 h) than in control cells, and the precursor coprecipitated with PI subunits as seen at short chase time was converted into mature subunit with little premature deg- points (1 h, Fig. 6A). The PI subunit coprecipitated with crl radation (Fig. 6A). The appearance of mature PI subunits in subunit was mostly in the matureform (Fig. 6B), buta small amount of PI subunit precursor could be detected at the 1-h both control and TGF-Dl-treated cells correlated with the assemblywith a1 and a2/a3subunitsasindicated by the chase time point after prolonged autoradiography of the gel shown in Fig. 6B (notshown). coprecipitation of these subunits with anti-& subunit antiThe ratio of newly synthesized LY subunits to PI subunits bodies (Fig. 6A). Precipitation of a1subunit with its corresponding antibodywas markedly higher in TGF-Dl-treated cells than in control showed that TGF-Pl caused the characteristically strongel- cells (Figs.1,3, and 6). These results together with the unusual evation in the rateof synthesis of this subunit (Fig. 6B). The kinetics of PI subunit maturation suggested the possibility cells might determine rate of conversion of the 180-kDa al subunit precursor into that availability of CY subunits in WI-38 the 190-kDa mature subunitwas rapid comparedto thatof PI the rateof assembly of integrin complexes. T o obtain further support for this possibility, we examined the biosynthesisof integrins in A431 human epidermoid carcinoma cells whose Anti-& Anti-a1 Total Protelh basal rate of CY subunit relative to PI subunit (Fig. 7A) was Hours 0 4 8 12 24 0 4 8 12 24 0- 4- 8 12 24 higher than in WI-38 cells. In A431 cells, the precursor and mature forms of the PI subunit migrated closely on electrophoresis gels, but could be distinguished from each other by the sensitivityof the precursorform but not the mature form to endoglycosidase H (Fig. 7A). The rate of Dl subunit maturation andcomplex formation was considerablyfaster in A431 cells than in WI-38 cells. Conversely, 3T3-Ll mouse fibroblasts that hada ratio of a to subunit synthesis even lower than WI-38 fibroblastsshowed the slowest rate of PI subunit maturation (Fig. 7B). Thus, an inverse relationshipexists between the availability of newly synthesized CY subunits and the rateof PI subunit maturation. Regulation of cup1 Integrins in Other Cell Lines-Table I FIG.5. Selectivity of the effect of TGF-Bl on integrin subunit expression. WI-38 cells were incubated with TGF-PI (250 PM) summarizes the resultsof a survey of CY& integrin responses for the indicated lengths of time, the last 4 h in methionine-free to TGF-Pl in eight human cell lines in addition to WI-38 medium containing 50 pCi/ml of ["S]methionine. Detergent-soluble fibroblasts. Each of the six CY subunits that belong to theCY& cell extracts were immunoprecipitated with polyclonal anti-fihronec- integrin family was expressed in a t least one of the cell lines tin receptor antiserum (Anti&) or monoclonal antibody specific for examined. The expression of each of thesix CY subunits a,integrin suhunit ( A n t i - n l ) The . immunocomplexes were harvested increased in response to TGF-8 in most but not all cell lines. with protein A-Sepharose. The materialhoundtotheSepharose heads, and small aliquots of the supernatants (Total Protein) were Like in WI-38 fibroblasts, TGF-Bl did not induce the expresanalyzed by gel electrophoresis and fluorography. sion of integrin subunits thatwere not expressed in unstim" 1 " -

-

--- -

" 0

TGF-fi Regulates filIntegrim ulated cells. In thecase of MG-63 osteosarcoma cells, elevated synthesis of a2 and P1 subunits in response to TGF-P1 correlated in the same experiments with an80% reduction in the rate of as subunit synthesis. These results provide further evidence for the ability of TGF-Pl toregulate the expression of all members of the a& integrin family via parallel but independent mechanisms. DISCUSSION

Using a panel of polyclonal and monoclonal antibodies, we have shown in the present studies that WI-38 human lung a& integrins. fibroblastsexpresssimultaneouslyvarious These integrins include a& whose exact function is not yet known, cy2& that functions as a receptor for type I collagen and a& that functions as a broad specificity fibronectin receptor that also binds laminin and type I collagen (14, 15). In addition, the presence of as integrin subunit mRNA sug-

B.

3 T 3 - L l Cells

Anti-

01

chase,hcus 0

1

2

3

5 8

FIG.7. Rate of & integrin conversion from precursor form to matureglycoprotein in A 4 3 1 human epidermoid carcinoma cells and in 3T3-Ll mouse fibroblasts. A431 cells ( A ) or 3T3-Ll cells ( R )were labeled for 6 h in methionine-free medium with 50 pCi/ ml of ["S]methionine. Detergent-soluble extracts from the cells were immunoprecipitated with anti-fibronectin receptor antiserum (81)or withanti-nnmonoclonal antibody (an),and wereanalyzed by gel electrophoresis and fluorography (left panels).In pulse-chase experiments (right panels), cells were labeled for 1h in cysteine-free medium with 50 pCi/ml ["S]cysteine to preferentially label theintegrin subunit. Cells were then shifted to unlabeled normaLmedium for the indicated lengths of time. Soluble cell extracts were immunoprecipitated with anti-fibronectinreceptorantiserum. Immunocomplexes bound to proteinA-Sepharosebeads were incubatedwith (+) or without (-) endoglycosidase H (Endo H)and then analyzed by gel electrophoresis andfluorography. The arrowhead points at thepreB, product of endoglycosidase H digestion.

385

gests that the as& fibronectin receptor is also expressed in these cells. Like other adherent cell types (16), WI-38 fibroblasts do not show detectable expression of a4.A 165-kDa product recognized by anti-a6 subunit antibody is presentin a 6 subunit WI-38 cells, but its relationship to 140-150-kDa the form described in other cell types (18) is not clear yet. Coexpression of multiple integrins that share the same type a of PI subunit is not unique to WI-38 fibroblasts but is general phenomenon. The biosynthesis and assembly of a/31 integrins is a multifactorial process in which distinct a subunits may have to compete for a limited supply of subunit, or PI subunit may be made in excess to support theprocess of found in assembly with a subunits. The latter is the situation WI-38 and other fibroblasts. Information on the basic features of thebiosynthesis of aPI integrinsis neededbefore the regulation of aPI integrin expression by TGF-P1 canbe properly analyzed and interpreted. Thegeneral characteristics of upI integrin biosynthesis and assembly in the WI-38 human cell model system are asfollows. Biosynthesis of Integrins That Share a Common Subunit-The amino acid sequence and electrophoretic mobility of PI integrin subunit have indicated that the extracellular domain of this protein is rich in cysteine residues, many of which are in the oxidized state forming intrachain disulfide bonds (3, 4). In WI-38 human lung fibroblasts, the cysteinerich PI subunit is synthesized asa 115-kDa precursor that is slowly converted to a 105-kDa form. The 105-kDa form can be converted to a 115-kDa form by reduction in uitro, suggesting that it is generated from the 115-kDa precursor by progressive formation of disulfide bonds. The process of disulfide bond formation occurs while the PI subunit is still in precursor form, i.e. when it is probably located in the endoplasmic reticulum. The sensitivity of the 105-kDa O1 form to deglycosylation by endoglycosidase H supports this conclusion. The shift to the mature molecular mass of 130kDa occurs with generation of a PI subunit form whose N-linked glycan chains are no longer sensitive to endoglycosidase H. By analogy with the biosyntheticprocess of other membrane glycoproteins, these changes mostlikely mark the passage of Golgi complex where the N-linked the PI subunit through the carbohydrate chains undergo final conversion and extension. Thereafter, the PI subunit moves to the cell surface where it can be identified by surface labeling. The biosynthetic steps of PI integrin subunit pathway are typical of other cell surface glycoproteins. However, the kinetics of maturation of PI subunit in WI-38cells and in other cell types including 3T3-Ll and NIH3T3 fibroblasts(21, 42) are exceptionally slow. The resultssuggest that this phenomenon might be due tolimited availability of a subunits andis

TABLEI Effects of TGF-PI on expression of integrin subunits in various cell types The expression of individual subunits increased (Up), did not change (Same), decreased or (Down) as measured bv biosynthetic labeling and immunoprecipitation after 12 h of incubation with TGF-B1. Integrin subunit

Cell line

Cell type 81

broblasts Lung WI-38 UP Osteosarcoma MG-63 UP RD Rhabdomyosarcoma UP Same Fibrosarcoma HT-1080 EJ Bladder carcinoma UP UP CCL-228 Colon carcinoma Hep G2 Hepatoma UP leukemia T-cellHPB-MLT UP IDP Normal T-cell line Down Only mRNA level was analyzed; 0, not expressed

a1

UP 0 0 Same 0 0 0 UP 0 Same in this cell

a2

a3

a4

UP UP 0 UP Down N.D. 0 0 UP Same 0 0 UP Same 0 UP UP 0 0 UP 0 0 UP Same 0 0 line; N.D., not determined.

as"

a6

UP Up N.D. N.D. N.D. N.D. N.D. N.D. N.D.

0 N.D. 0 0 0 UP N.D. 0 0

386

Integrins TGF-p Regulates

consistent with a model in which assembly of p1subunits with a subunits is necessary for transit of integrins through the Golgi complex where they mature as glycoproteins. Several lines of evidence support this model. First, the results show that the time of retention of the p1 subunit as an immature precursor presumably located in the endoplasmic reticulum is inversely proportional to the rate of a subunit synthesis. Second, the results (Figs. 3,6, and 7) suggest that in thebasal state, WI-38 and 3T3 fibroblasts synthesize p1 subunits in excessover a subunits. Third, the various a subunits expressed by WI-38 cells mature rapidly from their biosynthetic precursors (possibly assembling with a preexisting pool of unlabeled Dl subunit precursors) while the p1subunit matures slowly. Fourth, the results of coprecipitation experiments with anti-& antibodies show that thea subunits canassemble with p1subunits when they are stillin the precursor form. Labeled Dl subunit precursors cannot be readily detected by coprecipitation with anti-a! subunit antibodies probably due to dilution of the biosynthetically labeled p1 subunit precursors into a larger pool of unlabeled precursors that preexists in the cell. Taken together, these resultssuggest that integrin complexes are assembled when both subunits are stillin precursor form. The retention of unassembled p1 subunit precursor in a locus proximal to the endoplasmic reticulum as suggested by our results is analogous to theretention of unassembled and/ or incorrectly folded immunoglobulin heavy chains (43) or influenza virus hemagglutinin subunits (44). Folding and assembly of subunits precedes the release of multisubunit complexes from the endoplasmic reticulum, at least in thecase of influenza virus hemagglutinin (45). Immunoglobulin heavy chains produced by pre-B lymphocytes that do not express light chains appear to be recognized by an active retention mechanism in the endoplasmic reticulum (43). A similar mechanism is involved in the retention of influenza virus hemagglutinin subunits that are unable to assemble into trimeric complexes (44).These mechanisms may relate to the one that recognizes fibroblast integrin p1 subunit precursors unable to assemble into dimeric integrins due to limited availability of cu subunits. However, we have not detected association of PI subunit precursor with the BiP/gp 78 protein,4 the78-kDa endoplasmic reticulum resident protein that is associated with unfolded immunoglobulin chains and viral hemagglutinin subunits and perhaps intervenes in subunit folding and the oligomer assembly process (43, 44, 46, 47). pl integrin subunits synthesized in excess in WI-38 fibroblasts aredirectly degraded when they are still in the precursor form, as shown by the results (Fig. 6). Thisdegradation might occur by a mechanism similar to that recently reported for the elimination of excess T-cell receptor subunits in a nonlysosomal locus proximal to theendoplasmic reticulum (48). Our model proposing that the availability of a subunits determines the rate of assembly and maturation of subunits is in contrast to thatsuggested by Akiyama and Yamada (42) who reported that the kinetics of maturation of PI integrin subunit are slow in mouse NIH3T3 fibroblasts but not in chick embryo fibroblast. These authors suggested that avian and mammalian fibroblasts may use very different intracellular pathways for the biosynthesis of integrin complexes. They also suggested that maturation of the fibronectin receptor p subunit might function as a regulator of the rate of functional receptor assembly. Our results favor the hypothesis that the maturation process of subunit is a consequence rather than a causal determinant of the process of receptor J. Heino and J. MassaguB, unpublished results obtained with the anti-BiP monoclonal antibody described in Ref. 46 and generously supplied by J. Kearney.

assembly. The different rate of p1 subunitmaturation in NIH3T3 andchick embryo fibroblasts noted by Akiyamaand Yamada (42) could be due to the higher relative rate of a! subunit synthesisseen in the avian cells compared to NIH3T3 cells. Regulation of a& Integrins by TGF-@”he availability of an extensive panel of integrin subunit-specific antibodies and cDNA probes has allowed us to address several issues related to theregulation of fibronectin receptors by TGF-P. It is clear from the present studies that TGF-P1 treatment can elevate the expression of the set of a integrin subunits that form complexes with the p1subunit in WI-38 fibroblasts. TGF-Pl treatment increased severalfold the expression of al, aZ,as, and a5 subunits in parallel with the increase in p1 subunit expression. The increase in integrin subunit expression by TGF-/3 is evident at the mRNA level (for at least a5and Dl), proteinsynthesis,and cell surface levels. All the integrin subunits regulated by TGF-P1 are expressed in the unstimulated cells. WI-38 cells stimulated with TGF-Pl do not show detectable expression of integrin subunits that are not expressed in theunstimulated state. Thus, treatment with TGF/3l increases the level of integrin expression but does not appear to dictate what subsetof a subunits will be expressed by WI-38 fibroblasts. The same general conclusion applies to other cell lines surveyed in this study. Although TGF-Bl can increase the overall rate of protein synthesis in certain fibroblasts (22), the present results show that theeffects on integrin synthesis observed in WI-38 cells can occur in the absence of a significant effect on total protein synthesis. Further evidence for the selective nature of the TGF-Pl effect is provided by analysis of steady-state levels of a5mRNA and p1mRNA both of which undergo a netincrease in the presence of TGF-01 in comparison to the levels of glyceraldehyde-3-phosphate dehydrogenase mRNA used as a control. This result contrastswith the observation of Roberts et al. (22) that in IMR-90 human lung fibroblasts TGF-P1 does not increase the level of a5mRNA but prevents itsdecay under serum-free culture conditions. IMR-90 cells kept inthe absence of serum respond to TGF-Pl with an increase in fibronectin receptor synthesis that is similar to that of the total cellular proteins and therefore is nonselective (22). It is possible that the high sensitivity of IMR-90 cells to serum deprivation reduces their ability to respond to TGF-Pl with a selective increase of integrins and other cell adhesion proteins. The ability of TGF-Pl to increase the otherwise slow rate of maturation of p1 integrin subunit was first noted in our previous study with 3T3-Ll fibroblasts (21). In 3T3-Ll and WI-38 cells, this effect correlates with an increase in the ratio of a to 01 subunit synthesis (21, this report). According to the model for integrin complex assembly described above, the elevated rate of p1subunit maturationwould bea consequence of the higher level of a subunits available for assembly of integrin complexes. However,it is possible that inaddition to this effect, TGF-Dlmight control directly the mechanism for intracellular retention of unassembled subunits. In addition to its effect on the levels of integrin a and p subunits, TGF-Pl elevates the expression of fibronectin, type I collagen andchondroitin/dermatansulfate proteoglycan with similar kinetics in allcases (this report, Refs. 28 and 30, and results not shown). The effect of TGFs-/3 on these cell adhesion proteins and receptors contrasts with the lack of a detectable effect on the expression of basement membrane parallel proteins including laminin and typeIV ~ollagen.~ The effect of TGFs-fi on multiple members of the integrin family and other matrix components suggests the presence of com-

Integrins TGF-p Regulates mon regulatory elements that control the expression of these proteins. TGFs-/3 increase the mRNA levels of multiple cell adhesion proteinsand the available evidence suggests the involvement of transcriptional as well as posttranscriptional events in this response (30, 32,49-52). A nuclear factor 1 (NF1) binding site present in the mouse a2 (I) collagen gene promoter region has been implicated in the transcriptional activation of this gene by TGF-P (50). Putative NF1 binding sites may also be present in other genes regulated by TGF-P, but in one such case, the human fibronectin gene, removal of a NF1 binding site does not block transcriptional activation by TGF-Pl (52): Progress in defining transcription control sequences in TGF-P-regulated genes as well as posttranscriptionalcontrolsites may lead to the identification of the molecular basis for parallel regulation of these genes by TGFP. However, the results of the present study indicate that cellspecific and other as yet undefined variables determine the degree of responsiveness of individual integrin subunits to TGF-P1. In at least one case (MG-63 osteosarcoma cells), the expression of one a integrin subunit decreases in response to TGF-Pl while the expression of other a and P integrin subunits increases. Changes of opposite sign also take place in the levels of aland a3integrin subunits in human fibroblasts when they become quiescent due to serum deprivation (53). The evidence accumulated thus far suggests that regulation of individual integrin subunits and extracellular matrix proteins by TGF-Pl and other factors is accomplished via overlapping but independent mechanisms. The present results provide an explanation for the previously observed enhancement of the ability of TGF-p-treated cells to incorporate type I collagen into their extracellular matrix (21, 28). Two integrins, a& and a&, that are upregulated in response to TGF-P1 mediate cell attachment to type I collagen (14, 15). The involvement of the a& fibronectin receptor in the enhancement of cell adhesion to fibronectin in response to TGF-P1 and TGF-P2 has also been described (21). The functional implications of the marked increase in alp1, a&, and a6P1integrin expression in response to TGF-8 shall become clear once the ligands for these integrins are identified. The coexistence of multiple a subunits that can assemble with a common P1 subunit to form a repertoire of integrin complexes with distinct adhesive specificities, and thefashion in which these complexes are assembled have important implications in theregulation of cell adhesion in uiuo. Alterations in the expression of individual integrins subunits by regulatory factors like TGF-P1 are likely to change the balance of functional integrin complexes expressed on the surface of the cell,modifying asa result the adhesive interactions that determine cell homing and positioning in tissues as well as the expression of cell phenotype during development and morphogenesis.

387 5. 6. 7. 8. 9.

10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29.

30. 31. 32. 33. 34. 35.

Acknowledgments-We thank Drs. A. Albino,K. Pischel, V. Woods, and A. Sonnenberg for monoclonal antibodies and Drs. E. Ruoslahti and R. Hynes for antibodies and cDNA probes used in this study. We thank John Rauthfor dedicated help with cell culture and Catherine English for preparation of the manuscript.

37.

REFERENCES

38.

1. Hynes, R. 0. (1987) Cell 48,549-554 2. Ruoslahti, E.,and Pierschbacher, M. D. (1987) Science 238,491497 3. Tamkun, J. W., DeSimone, D. W., Fonda, D., Patel, R. S., Buck, C., Honvitz, A. F., and Hynes, R. 0. (1986) Cell 46, 271-282 4. Argraves, W. S., Suzuki, S., Arai, H., Thompson, K., Piersch-

39.

D. Dean and S. Bourgeois, personal communication.

36.

40. 41.

bacher, M. D., and Ruoslahti, E. (1987) J. Cell. Biol. 105,11831190 Fitzgerald, L. A., Steiner, B., Rall, S. C., Jr., Lo, S., and Phillips, D. R. (1987) J. Biol. Chem. 262,3936-3939 Kishimoto, T. K., O’Connor, K.,Lee, A., Roberts, T. M., and Springer, T. A. (1987) Cell 4 8 , 681-690 Pytela, R., Pierschbacher, M.D., Ginsberg, M.H., Plow, E. F., and Ruoslahti, E. (1986) Science 2 3 1 , 1559-1562 Pytela, R., Pierschhacher, M.D., and Ruoslahti, E. (1985) Cell 40,191-198 Pierschbacher, M. D., and Ruoslahti, E. (1984) Nature 309,3033 Hemler, M. E., Ware, C.F., and Strominger, J. L. (1983) J. Zmmunol. 131,334-340 Hemler, M. E., Huang, C., and Schwartz, L. (1987) J . Biol. Chem. 262,3300-3309 Hemler, M. E. (1988) Zmmunol. Today 9 , 109-113 Hemler, M. E., Sanchez-Madrid, F., Flotte, T. J., Krensky, A. M., Burakoff, S. J., Bhan, A. K., Springer, T. A., and Strominger, J. L. (1984) J. Zmmunol. 132,3011-3018 Takada, Y., Wayner, E. A., Carter, W. G., and Hemler, M. E. (1988) J. Cell Biochem. 3 7 , 385-389 Wayner, E. A., and Carter, W. G . (1987) J. Cell Biol. 1 0 5 , 18731884 Pischel, K. D., Hemler, M. E., Huang, C., Bluestein, H. G., and Woods, V. L. (1987) J. Zmmunol. 138, 226-233 Takada, Y., Strominger, J. L., and Hemler, M. E. (1987) Proc. Natl. Acud. Sci. U. S. A . 84,3239-3243 Hemler, M. E., Crouse, C., Takada, Y., and Sonnenberg, A. (1988) J. Biol. Chem. 263,7660-7665 Menko, A. S., and Boettiger, D. (1987) Cell 51,51-57 MacKrell, A. J., Blumberg, B., Haynes, S. R., and Fessler, J. H. (1988) Proc. Natl. Acud. Sci. U. S. A . 8 5 , 2633-2637 Ignotz, R. A,, and Massagui., J. (1987) Cell 5 1 , 189-197 Roberts, C. J., Birkenmeier, T. M., McQuillan, J. J., Akiyama, S. K., Yamada, S.S., Chen, W.-T., Yamada, K.M., and McDonald, J. A. (1988) J. Biol. Chem. 263,4586-4592 Massagui., J. (1987) Cell 49, 437-438 Padgett, R. W., St. Johnston, R. D., and Gelbart, W. M. (1987) Nature 325,81-84 Weeks, D. L., and Melton, D.A. (1987) Cell 51,861-867 Kimelman, D., and Kirschner, M. (1987) Cell 51,869-877 Rosa, R., Roberts, A. B., Danielpour, D., Dart, L. L., Sporn, M. B., and David, I. B. (1988) Science 2 3 9 , 783-785 Ignotz, R. A., and Massaguk, J. (1986) J. Bwl. Chem. 261,43374345 Roberts, A. B., Sporn, M. B., Assoian, R. K., Smith, J. M., Roche, N. S., Wakefield, L. M., Heine, U. I., Liotta, L. A., Falanga, V., Kehrl, J. H., and Fauci, A. S. (1986) Proc.Natl.Acud.Sci. U. S. A . 83,4167-4171 Bassols, A., and Massagui., J. (1988) J. Bwl. Chem. 2 6 3 , 30393045 Laiho, M., Saksela, O., Andreasen, P. A., and Keski-Oja, J. (1986) J . Cell Biol. 1 0 3 , 2403-2410 Edwards, D. R., Murphy, G., Reynolds, J. J., Whitham, S. E., Docherty, A. J. P., Angel, P., and Heath, J. K. (1987) EMBO J. 6,1899-1904 Ignotz, R. A., Heino, J., and Massage, J. (1989) J. B i d . Chem. 264,000-000 Pytela, R., Pierschbacher, M.D., Argraves, S., Suzuki, S., and Ruoslahti, E. (1987) Methods Enzymol. 144,475-489 Kantor, R. R. S., Mattes, M. J., Lloyd, K. O., Old, L. J. and Albino, A. P. (1987) J. Biol. Chem, 2 6 2 , 15158-15165 Hemler, M. E., Huang, C., Takada, Y., Schwarz, L., Strominger, J. L., and Clabby, M. L. (1987) J. Bwl. Chem. 2 6 2 , 1147811485 Yelton, D. E., Desaymard, C., and Scharff, M. D. (1981) Hybridoma 1,5-11 Springer, T. A., Bhattacharya, A., Cardoza, J. T., and SanchezMadrid, F. (1982) Hybridoma 1,257-273 Chirgwin, J. M., Przybyla, A. E., MacDonald, R. J., and Rutter, W. J. (1979) Biochemistry 10, 5291-5299 Argraves, W. S., Pytela, R., Suzuki, S., Millan, J. L., Pierschbacher, M.D., and Ruoslahti, E. (1986) J. Biol. Chem. 2 6 1 , 12922-12924 Fort, P., Marty, L., Piechaczyk, M., El Sabrouty, S., Dani, C., Jeanteur, P., and Blanchard, J. M. (1985) Nucleic Acids Res. 13, 1431-1442

388

TGF-8 Regulates Integrins

42. Akiyama, S. K., and Yamada, K. M. (1987) J. Biol. Chern. 2 6 2 , 17536-17542 43. Haas, I. G., and Wabl, M. (1983) Nature 3 0 6 , 387-389 K., &thing, i ~ g"J., t and ~ 44. Kozutsumi, y., Segal, M,, ~ ~ ~ ~ Sambrook, J. (1988) Nature 3 3 2 , 462-464 45. copeland, c. s., Zimmer, K. p,, Wagner, K. R., Healey, G . A.,

Mellman, I., and Helenius, A. (1988) Cell 5 3 , 197-209 46. Bole, D. G., Hendershot, L. M., and KearneY, J. F. (1986) J. Cell Bioi. 102,1558-1566 47. Rothman, J. E. (1987) Cell 50,521-522 48. Lippincott-Schwartz, J., Bonifacino, J. S., Yuan, L. C., and

Klausner, R. D. (1988) Cell 5 4 , 209-220 49. Ignotz, R.A., Endo, T., and Massage, J. (1987) J . Biol. Chern. 262,6443-6446 ~50. ,Rossi, p., Karsenty, G., Roberts, A. B., Roche, N. s.,SPorn, M. B., and de Crombrugghe, B. (1988) Cell 52,405-414 51. Lurid, L.R., Riccio, A., Andreasen, P.A,, Nielsen, L. s.,Kristensen, P., Laiho, M., Saksela, O., Blasi, F., and Dan& K. (1987) EMBO J. 6 , 1281-1286 52. Dean, D. C., Newby, R. F., and Bourgeois, S. (1988) J . Cell Bioi. 106,2159-2170 53. Fingerman, E., and Hemler, M. E. (1988) Erp. Cell. Res. 177, 132-142