THEJOURNALOF BIOLOGICAL CHEMISTRY (c)
Vol. 266, No. 14, Issue of May 15, pp. 9108-9112,1991 Printed in U.S.A.
1991 by The American Society for Biochemistry and Molecular Biology, Inc
Responsiveness to Transforming GrowthFactor-@(TGF-@) Restored by Genetic Complementation between Cells Defective in TGF-P Receptors I and 11” (Received for publication, December 20, 1990)
Marikki LaihoS, Frances M. B. Weis$, Frederick T. Boyd,Ronald A. Ignotzll,and Joan Massague$(I From the Cell i3iolog.y and Genetics Program and SHoward Hughes Medical Institute, Memorial Soan-Kettering Cancer Center, New York, New Yo;k 10021
Selection of mutantMvlLu mink lung epithelial cells resistant to growth inhibition by transforming growth factor-@(TGF-@)has led to the isolation of cell clones with distinct alterations in type I and I1 TGF-@receptors. Certain mutant clones present a decreased number or complete loss of detectable type I receptor. Other clones show a loss and/or altered electrophoretic mobility of the type I1 receptor, with concomitant loss of the type I receptor. Using somatic cell hybridization analysis we demonstrate the recessive nature of these mutants with respect to the wild-type phenotype and define various mutant complementation groups. Among these, hybrids between cells that express only type I1 receptor (R mutants) and cells that express neither receptor type (DRa mutants) rescue wild-type expression of type I receptors. Moreover, these hybrids regain full responsiveness to TGF-81, as measured by inhibition of DNA synthesis as well as stimulation of fibronectin and plasminogen activator inhibitor- 1 production. These results provide evidence for an interaction between TGF-@receptor components I and I1 and show that, in MvlLucells, expression of both receptor types is required for mediation of biological responses to TGF-Bl.
but in low numbers, in most normal and transformed mammalian and avian cells (3) and display ligand binding properties that markedly discriminate between the various TGF0isoforms. Receptor types I and I1 appear as separate products based on differences in the peptide maps of their ligand binding domains (4) and differences in their ability to bind various TGF-P isoforms ( 5 ) . Betaglycan is a 200-400-kDa proteoglycan with chondroitin sulfate andheparan sulfate chains linked to a 110-130-kDa N-glycosylated core protein that also contains the TGF-0 binding site (6-8). The tissue distribution of betaglycan ismore restricted than that of receptors I and I1 and is absent from myoblasts, hematopoietic cells, and endothelial cells which, nevertheless, respond to TGF-p (9-11). No evidence is presently available for a direct role of betaglycan in signal transduction. Its proteoglycan nature, secretion as soluble forms, and presence in extracellular matrices (12) have suggested that betaglycan may functionasa TGF-(3 storage protein or as a regulator of the availability of bioactive ligand to thecell. One of the most remarkable effects of TGF-0 is inhibition of cell growth (13). This effect is observed in cells of many lineages and leads to a complete growth arrest in certain cell lines like, for example, MvlLu mink lung epithelial cells and mouse keratinocytes (14). In MvlLu cells, TGF-P prevents phosphorylation of the retinoblastoma susceptibility gene The transforming growth factor-0 (TGF-0)’family includes product RB retaining this protein in its growth-suppressive several homologous polypeptides that actin a paracrine fash- state and arresting the cell cycle in late G1 (15). MvlLu cells ion to regulate proliferation, differentiation, and other func- have also been useful in identifying cell surface receptors that tions of the cell (1,2). Themammalian homodimeric isoforms mediate TGF-8 action. This has been accomplished by isolaTGF-01, -02, and $33, and the heterodimer TGF-01.2 bind tion of a panel of mutant MvlLu clones resistant to the with high affinity to a setof cell surface proteins that include growth inhibitory action of TGF-pl and -02 (16, 17). These the type I receptor, the type I1 receptor, and betaglycan (2). mutants were isolated by selection of chemically mutagenized Receptor types I and I1 are cell surface glycoproteins of 53 cell clones that wouldgrow in the continuous presence of and 70 kDa, respectively. They are expressed ubiquitously, TGF-P (18).Characterization of these mutants showed that they had lost all known responses to TGF-/3 including stim* This work was supported in partby National Institutesof Health ulation of fibronectin and plasminogen activator inhibitor-1 Grant CA-34610. The costs of publication of this articlewere defrayed (PAI-1) production and morphological changes (16, 17). in part by the payment of page charges. This article must therefore Analysis of cell surface TGF-P-binding proteinsin these cells be hereby marked “aduertisement” inaccordancewith 18 U.S.C. indicates that themutations affected with high frequency the Section 1734 solely to indicate this fact. $ Supported by the Academy of Finland. expression of TGF-P receptors I and I1 but not betaglycan. (i Present address: Dept. of Cell Biology and Anatomy, University Some mutant clones had lost expression of type I receptors of Massachusetts Medical School, Worcester, MA 01655. (16, 17). Others showed concomitant loss of type I and I1 (1 A HowardHughes Medical InstituteInvestigator. To whom correspondence shouldbe addressed Memorial Sloan-Kettering Can- receptors with evidence of structural alterations in the type cer Center,Box 116,1275 York Ave., New York, NY 10021. Fax: 212- I1 receptor, as determined by receptor affinity-labeling assays 717-3298. (17). ’ The abbreviations used are: TGF-0; transforming growth factor- This evidence has linked receptors I and I1 to mediation of 8; FBS, fetal bovine serum; NEAA, nonessential amino acids; PAI-1, the pleiotropic action of TGF-p in MvlLu cells. Moreover, plasminogen activatorinhibitortype 1; MEM,minimalessential medium; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; the frequent concomitant loss of both receptors in these cell IdUrd, iododeoxyuridine; SDS-PAGE, sodium dodecyl sulfate-poly- mutants has raised the possibility that thetwo receptor components interact. In the present study, we have generated acrylamide gel electrophoresis.
9108
TGF-/I Receptor Components somatic cell hybrids between MvlLu cell clones representing multiple mutant phenotypes. We demonstrate that all known responses to TGF-/3 can be regained when expression of receptors I and I1 is restored in the hybrids. These results indicate that in MvlLu epithelial cells both receptor components function together to initiate intracellular signal transduction. EXPERIMENTALPROCEDURES
Cells, Cell Transfections, and Somatic Cell Fusions-Cell cultures were maintained in minimal essential medium (MEM) supplemented with 1 X nonessential amino acids (NEAA) and 10% fetal bovine serum (FBS) in a humidified (5% co2/95% air) atmosphere. The MvlLu mutant cell clones studied here were generated and characterized in previous studies (16, 17). Cells were transfected with plasmids pSV2neo or pSV2his (19) carrying G418 and histidinol resistance genes, respectively. Cellswere transfected with plasmid DNA (20 pg) by the calcium phosphate-precipitation method (20). Cells stably expressing either plasmid were selected in MEM containing 1X NEAA, 10% FBS, and 1mg/ml G418 (GIBCO) for pSV2neotransfected cells or in histidine-free MEM containing 1 X NEAA, 10% dialyzed FBS (GIBCO) and 0.5 mM histidinol (Sigma) for pSV2his-transfected cells.Coloniesfrom each transfection were pooled and maintained in the selection medium. For cell fusions, 2 X lo6 cells of a G418-resistant clone and 2 X lo6 cells of histidinol-resistant clones were plated in 60-mm dishes. 1day later, the cultures were briefly covered with 2-3 ml of 45% polyethylene glycol (M, 1300-1600, American Type Culture Collection) in MEM, 10 mM Hepes, with a final pH 7.3. The polyethylene glycol/ MEM solution was immediately aspirated and the cultures were incubated for 15-20 min at 37 "C followed by three washes with MEM and two washes with MEM containing 1 X NEAA and 10% FBS. All washes with media were at 37 "C. On the next day the cultures were trypsinized and plated into 150-mm dishes. Hybrid cell clones were selected in histidine-free MEM containing 1 X NEAA, 10% dialyzed FBS, 0.5 mM histidinol, and 800 pg/ml G418 for 2-3weeks.Cell clones were then ring-cloned and propagated in medium containing both histidinol and G418. DNA Synthesis Assays-The effect of TGF-pl on DNA synthesis of the cell clones was determined on subconfluent cultures in24-well cluster plates. Cultures were treated with the indicated concentrations of bovine bone TGF-01 for 22-28 h in MEM containing 1 X NEAA and 0.2% FBS. At the end of the incubation, cultures were labeled for 4 h with 0.25 pCi/ml ['251]iododeoxyuridine(['251]IdUrd,specific activity 1 0 0 Ci/mmol; Du Pont-New England Nuclear). Incorporation of ['*'II]IdUrd into the nuclei was determined as described (15). The assays were carried out in either duplicate or triplicate with results diverging from the mean by 10% or less. TGF-P Receptor Affinity Labeling-TGF-81 and $2 were purified Dom bovine bone as described (21). TGF-01 was iodinated using the chloramine T protocol as described (22). '251-TGF-plaffinity labeling was carried out with confluent cultures using 100 PM 'z51-TGF-pland 150 p~ disuccinidimyl suberate (Pierce Chemical Co.) as previously described (6). Detergent extracts from affinity-labeled cells were separated by SDS-PAGE on 5-8% gradient polyacrylamide gels under reducing conditions, followed by autoradiography to detect affinitylabeled products. Detection of Extracellular Matrix Proteins-Confluent cell cultures were treated with or without TGF-P1 (250 pM) for the indicated time in serum-free MEM. Cultures were labeled with [35S]methionine(40 pCi/ml, specific activity 1100 Ci/mmol; Du Pont-New England Nuclear) in methionine-free MEM for the last 2 h of the incubation. The medium wasaspirated and theextracellular matrix proteinswere prepared as described (23) and analyzed by SDS-PAGE under reducfluorography using ing conditions on 6 or 9% gels,followedby EnlightningTM(Du Pont-New England Nuclear).
9109
Wild Type Receptor Profile
R -Type -Type
11, 70 kDa S1 A
I, 53 kDa
, LR
R FIG. 1. Summary of the TGF-@receptor profiles in seven TGF-@-resistantmutant phenotypes derived from the MvlLu cell line and reconstitution of TGF-@responsiveness in somatic cell hybrids between these mutants. The wild-type and each mutant TGF-p receptor phenotype are depicted with a box containing two bands that represent the affinity-labeled TGF-P receptors I and I1 as visualized by SDS-PAGE (see Fig. 3 below and Boyd and Massague (16) and Laiho et al. (17) for actual data). The designation of each mutant phenotype is shown with the corresponding receptor defects, which include low receptor levels (stippled band), undetectable receptor (no band), or anomalously fast receptor migration (lowered band position). Lines linking the various phenotypes indicate recovery (continuous line) or no recovery (broken line) of growth-inhibitory TGF-0 response in cell hybrids between these phenotypes. See text, Fig. 2, and Table I for details and data.
showed anomalies in the expression of TGF-@receptor types I or I1 in many of the TGF-P-resistant clones with no loss of betaglycan in any of the clones (16, 17). Fig. 1 summarizes the previously described profile of labeled receptors I and I1 in parental cells and in the seven mutant phenotypes. In parental MvlLu cells, receptors I and I1 migrate on SDSPAGE as glycoproteins of 53 and 70 kDa cross-linked to a 12kDa lZSI-TGF-filmonomer (3,24). Clone S1A presents awildtype receptor phenotype (16). LR and R1B clones present, respectively, low or undetectable levels of type I receptor; the type I1 receptor in these clones appears normal (16, 17). The various DR phenotypes present various degrees of type I1 receptor loss paralleled by similar losses of type I receptor (17). Thus, DRa and DRbclones present complete or partial loss, respectively, of both receptor types. DRc clones present partial loss of both receptor types and anomalously fast migration of the type I1 receptor on SDS-PAGE, an anomaly which is also observed in the deglycosylated receptor protein (17). Clone S1B presents normal levels of both receptors with the type I1 receptor running slightly faster on SDS-PAGE. This small anomaly escaped detection in earlier experiments (17) but is consistently observed.2 Recessive Character of the TGF-@-resistantMutants-We generated somatic cell hybrids of mutant cell clones with wildtype MvlLu cells and with each other. Wild-type MvlLu cells and clones representing each mutant phenotype were transfected with cDNAswhose expression confers resistance to either G418 or histidinol. Clones expressing these markers were selected, and retention of their original phenotype was verified by DNA synthesis assays and TGF-P receptor affinity-labeling assays. Cell fusions between partners expressing RESULTS the two selectable markers were carried out usingpolyethylene Seven classes of mutant MvlLu cell clones resistant to the glycol. Cell hybrids of interest were selected and expanded in growth-inhibitory action of TGF-/3 have been obtained by the presence of both agents. Analysis of hybrids between wildprolonged culture of chemically mutagenized cells in thepres- type cells and thedifferent mutants indicated that in all cases ence of TGF-@1or TGF-@2(16, 17). The TGF-P receptor these hybrids maintained full responsiveness to TGF-Pl as phenotype of parental and mutant MvlLu cells was determined by receptor cross-linking with '
[email protected] M. Laiho, F. T. Boyd, and J. Massague, unpublished observations.
Components TGF-p Receptor
9110
TABLE I Somatic cell hybridizations of TGF-8 receptor mutants Somatic cell hybrids were created between the clones listed in the table stably carrying eithera histidinol or a G418 resistance gene. Fused cell clones resistant to both agents were isolated and theirresponsiveness to TGF-P1mediated inhibition of DNA synthesis was determined. Numbers indicate how many hybrid clones responded to TGF-P1 over how many tested. Inhibition of ["'I]IdUrd incorporation after 24 h with TGF-P1 (75pM) was 2 70% relative to control cultures not treated with TGF-P1, except in clones (*) in which inhibition ranged between 30 and 65%. WT
DRc SlA
RIB
26-1 10-1
ND
W414 T S1A RIB LR S1B
616
516 515
515 ND
313 0111 013
C 0
DRa
DRb
SIB
LR
NS" 3*/3 114 15/15 616
9-127-1 37-1
515 2*/5 717 10111
33-1
111 3*/3
111 4*/7 515 313 016
111 NS 014
516 2'12 313
617 0110 414
114
113
818
NS 013 NS, no successful fusions were obtained; ND, not determined; WT, wild-type MvlLu cells.
016
ND
o/ 1
018
111
100
.-
F3
F12
F24
CI
2 80 0
0
-c =)
E
Ln
60
i"
N r
2
40
il
.,..( . . \
20]
A 26-1
-
W RX26 F12 A RX26 F24
a? 0 0
1 0 01
10
-
FIG. 3. Rescue of TGF-j3 receptor expression insomatic cell hybrids between R1B and DRa mutant clones. Cell surface receptors were affinity-labeled by cross-linking with "'I-TGF-P1. Cell FIG. 2. Reconstitution of growth-inhibitoryresponseto proteins were separated by SDS-PAGE under reducing conditions, TGF-j31 in somaticcell hybrids between R1B and DRa mutant clones. Subconfluent cell cultures were incubated with the indicated and the labeled proteins were detected by autoradiography. Labeled concentrations of TGF-P1 for 24 h. The cultures were labeled for the bands corresponding to TGF-8 receptorsI and I1 and betaglycan are last 4 h with ["'II]IdUrd, followed by determination of radioactive indicated. label incorporated into DNA. Assayswere done in triplicate with variations 5 10%. Data are expressed as the percentage decrease in No complementation was obtained with three groups of ["'I]IdUrd incorporation relative to cultures that received no TGF- crosses. These were LR x RlB, S1A X SlB, and S1Bcrossed ljl. TGF 01 (PM)
to any of the DR clones. The total number of hybrids unremeasured by inhibition of ['25]IdUrd incorporation (Table I) sponsive to TGF-Dl were 11, 10, and 24, respectively (Table and had normal TGF-P receptor profiles (data not shown). I). Fig. 1 summarizes thepresence or absence of complementation between all mutantphenotypes. This demonstrated that the mutations causing TGF-P resistance were all recessive with respect to the parental phenotype. Rescue of TGF-/3 Receptors and of Multiple TGF-P ReGenetic Complementation between TGF-P-resistant Mu- sponses-Receptor cross-linking to '2sI-TGF-Pl showed that tants-Cell hybrids derivedfrom fusions between various both the type I and the type I1 TGF-/3 receptor components mutant cell clones were analyzed for responsiveness to TGF- were expressed on the surface of those hybrid clones that 3fl, as determinedby the ability of this factor to inhibit DNA regained responsiveness to TGF-Dl. This was most clearly synthesis. Hybrids between certain mutants regained nearly observed in the cross between clone R1B which lacks type I fullresponsiveness toTGF-P1 which inhibited ['2s]IdUrd receptor andclone DRa/26-1 which lacks both receptor types (Fig. 3). Hybrids resulting from this cross not only expressed incorporation into DNA by over 70% (Table I). These inrestored exprescluded the hybrids derived from either R1B or LRcrossed to normal levels of type I1 receptor, but had also all other mutant clones (Table I), with the exception of the sion of normal levels of type I receptor (Fig. 3). On the other reR1B x DRc/9-1 cross. The TGF-P1EDsofor growth inhibition hand, hybrid clones which did not reconstitute TGF-P was 2-5 PM TGF-P1 in these hybrids as compared to2.5 PM sponsiveness had the altered receptor profiles of one or the other parental clones used in fusionas determinedby receptor in parental MvlLu cells, indicating that the hybrids had affinity labeling (data not shown). regained full sensitivity toTGF-P1 (Fig. 2 anddatanot Although the mutants studied here were originally isolated shown). Hybridsbetween S1A and thevarious DR cloneswere characterized by a clear but limited (30-65% inhibition of on the basisof their resistance to growth inhibition by TGFDNA synthesis)growth inhibitory response to TGF-Dl (TableP, they were also lacking of other responses to TGF-/3 normallyobserved inparentalMvlLu cells. These responses I).
TGF-/3 Receptor Components include elevated expression of fibronectin (23, 25) and plasminogen activator inhibitor-1 (PAI-1) (23). These responses wererescuedin the RIB X DRa hybrid clones which had regained expression of normal TGF-D receptors I and I1 and growth inhibitory response to thisfactor. Stimulation of deposition of fibronectin and PAI-1 into the extracellular matrices by TGF-Dl in these hybrids was similar to the stimulation observed in parental MvlLucells (Fig. 4).
9111
acterization of clones R1B and LR/36-2 indicated that they have an apparently normal typeI1 receptor. However, type I receptor expression in these clones is altered; itis undetectable in R1B and present at low levels in LR (16, 17). These features suggest that both clones are primarily defective in the type I receptor. Accordingly, clones RIB and LR do not complement each other but they complement with clonesS1A and SIBwhich express functional type I receptor. “Functional receptor expression” here means that the receptor must not DISCUSSION only be at thecell surface but it must also be able to bind ”.”IChemical mutagenesis of MvlLu cells and selection of cells TGF-D as measured by receptor affinity labeling. The lack of cells could that have lost responsiveness to TGF-Dl and $2 has led to detectable TGF-P receptor components in mutant the isolation of several mutant cell clones which frequently be due toa lack of synthesis of the receptor protein, inability show alterations in the type I and I1 TGF-P receptors. By to reach the cell surface, or inability to bindligand once it is unlikely, however, that the defectsin TGF-0 creating somaticcell hybrids between these mutants,we show there.Itis receptors are caused by gross anomalies in general mechahere that biological responses to TGF-Dl can beregained when wild-typeexpression of bothtype I and I1 TGF-@ nisms of membrane protein synthesis since such anomalies receptors is recovered in thehybrids. Reconstitution of recep- would probably be incompatible with cell survival. Indeed,the tor expression restores all the TGF-/3 responses measured, expression of another membraneTGF-/3-binding protein, beincluding inhibition of DNA synthesis, stimulation of fibro- taglycan, is normal inall mutant clones. nectin production, and PAI-1 production. These results supThe results with the R1B X DRa crosses (Figs. 3and 4) are port the hypothesis that initiation of TGF-D signal transduc- of particular interest. DR clones show low or undetectable tion in MvlLu epithelial cells requires the function of both levels of receptors I and 11. How could they complement R1B TGF-/3 receptor components. and LRwhich are defective in typeI receptor? In interpreting Several of the cell crosses studied here showfull biological these results, it should be noted that therelative frequency of responsiveness to TGF-Dl. Theseinclude crosses of wild-type isolation of DR mutants (22 DR mutants out of a total 36 MvlLu cells to all of the mutantclones, indicating that these receptor mutants, and outof a total of 1.6 X 10’ mutagenized mutants arerecessive with respectto thewild-type phenotype. cells (17)) renders thepossibility of a double mutation in two We analyzed more than one hybrid of each cross whenever separate receptor genes very unlikely. Furthermore, all mupossible. Furthermore, we created hybrids with two clones tants with low expression of thetype I1 receptor have a each of every DR mutant phenotype. This was necessary concomitant loss, to the samedegree, of type I receptor (17). because several crosses repeatedly failed to yield stable hy- Thus, it appears that while the typeI1 receptor canbind TGFbrids(TableI), whereas others yielded hybridsthatrep in the absenceof type I receptor, as in R1Bcells, the ability sponded to TGF-Dl and hybrids that did not (see for example of type I receptors to reach the cell surface and/or bindligand LR X DRc/9-1 and R1BX DRc/9-1 in Table I). In such cases, in MvlLu cells may depend on normalexpression of the type however, recovery of TGF-01 responsiveness in some of the I1 receptor. DR mutantsmay, therefore, encode a normal type hybrids was interpreted asevidence of complementation; lack I receptor whose functional expression is rescued by the of TGF-fi responsiveness in the other hybrids from the same presence of a normal type I1 receptor provided by R1B or LR cross could be attributedtoaccidental escape fromdrug in the hybrids. selection, spontaneous chromosomalloss, or instability of the Clone S1A expresses apparently normal TGF-/3 receptors, hybrids due to other causes. Several if not all of the TGF-D and has been postulated to present a defect a t some level of receptor-defective mutants studied here have remarkably sta-the TGF-/3 signal transduction couplingmechanism(16). blephenotypes, with spontaneousrevertants never having Accordingly, this clone should complement any mutant defecbeen isolated even when they were sought.’ Therefore, it is tive in receptors I or 11, as it does with RlB, LR, and DR unlikely that recovery of TGF-P responses in these hybrids clones. Surprisingly, however, S1A x DR hybrids show only was due to mutant reversion. a limited growth-inhibitoryresponse to TGF-Dl. More unexHybrids of LR or R1B clonescrossed with SlA, SlB, and pected was the finding that S1A does not complement S1B. DR clones regain full TGF-@responsiveness. Previous char- S I B cells express a type I1 receptor that migrates slightly faster on SDS-PAGE. This, and the inability of clone S1B to R I B2 6 - 1 R1Bx26-I clones complement DR clones suggest that the type I1 receptor in S1B cells might be defective. Indeed, fusion with R1B or LR clones which provide a normal type I1 receptor compensates I for the defectinclone S1B. We donot know why then, hybrids between SIB andS1A do notregain responseto TGFPAI-14 D. Of note, the inabilityof S1A to complement clones defective in type I1 receptor, suchasS1BandDR clones, is FN 4 proportional to thelevel of defective type I1 receptor found in these clones. It is conceivable that a defective receptor protein % Increase 21 8 5 19 82 138 100 FIG. 4. Concomitant recovery of multiple TGF-8 responses in S1B and DR mutants might inactivate or sequester an endogenous signal transduction component and this compoin hybrids between mutant cells. Confluent cell cultures were incubated in serum-free medium for 4 h ( A ) or 24 h ( R )with TGF- nent is the siteof the inactivating mutationsin S1A mutants. /jl (250p ~ )The . cultures were labeled with [‘‘“Slmethionine for the This hypothetical situationwould render clones S I B and DR last 2 h followed by selective extraction of extracellular matrix pro- dominant suppressors in hybrids with clone SlA, thus exteins using sodium deoxycholate. Matrix proteins were detected by SDS-PAGE on 9 ( A ) or 6% ( R )gels followed by fluorography. The plaining the deserved results. Some TGF-D responsive cell lineshave no detectable (A) relevant portion of the fluorograms containing the labeled PAI-1 expression of type I1 receptor. These include several mouse or fihronectin ( R )bands is shown. I
I*
Components TGF-0 Receptor
9112
hematopoietic progenitor cell lines (10). It is possible that these cells express a form of the type I TGF-/3receptor that can function in the absence of type I1 receptor. Alternatively, type I1 receptor might be present in these cell lines at levels below the detection limit of receptor affinity labeling assays since the level of total TGF-/3receptors in these cell lines is very low (110-240 binding sites per cell (3,lO)). While significant cell-specific differences may exist in TGF-P receptors (5), the present data strongly favors the possibility that functional expression of receptors I and I1 is required for biological responsiveness to TGF-/3in MvlLu cells. Acknowledgment-We thank H.Hernandez for assistance with the tissue culture. REFERENCES 1. Massagui., J. (1990) Annu. Reu. Cell Biol. 6, 597-641 2. Roberts, A. B., and Sporn, M. B. (1990) in Peptide Growth Factors and Their Receptors (Sporn, M. B., and Roberts, A. B., eds) pp. 419-472, Springer-Verlag, Heidelberg, Germany 3. Massagui, J., Boyd, F. T., Andres, J. L., and Cheifetz, S. (1990) Ann. N. Y. Acad. Sci. 593,59-72 4. Cheifetz, S., Like, B., and MassaguB, J. (1986) J. Biol. Chem. 261,9972-9978 5. Cheifetz, S., Hernandez, H., Laiho, M., ten Dijke, P., Iwata, K. K., and Massagui, J. (1990) J. Biol. Chem. 265,20533-20538 6. Massagui., J. (1985) J. Biol. Chem. 260, 7059-7066 7. Segarini, P. R., and Seyedin, S. M. (1988) J . Biol. Chem. 263, 8366-8370 8. Cheifetz, S., Andres, J. L., and Massagu6, J. (1988) J . Biol. Chem. 263,16984-16991 9. Massagui., J., Cheifetz, S., Endo, T., and Nadal-Ginard, B. (1986)
Proc. Natl. Acad. Sci U. S. A. 83, 8206-8210 10. Ohta, M., Greenberger, J. S., Anklesaria, P., Bassols, A., and Massagui, J. (1987) Nature 329,539-541 11. Segarini, P. R., Rosen, D. M., and Seyedin, S. M. (1989) Mol. Endocrinol. 3, 261-272 12. Andres, J. L., Stanley, K., Cheifetz, S., and MassaguB, J. (1989) J. Cell Biol. 109, 3137-3145 13. Tucker, R. F., Shipley, G. D., Moses, H. L., and Holley, R. W. (1984) Science 226, 705-707 14. Shipley, G. D., Pittelkow, M. R., Wille, J. J. Jr., Scott, R. E., and Moses, H. L. (1986) Cancer Res. 46,2068-2071 15. Laiho, M., DeCaprio, J. A., Ludlow, J. W., Livingston, D. M., and MassaguB, J. (1990) Cell 62,175-185 16. Boyd, F. T., and Massagui, J. (1989) J. Biol. Chern. 2 6 4 , 22722278 17. Laiho, M., Weis, F. M. B., and Massaguir, J. (1990) J. Biol. Chem. 265,18518-18524 18. Chinkers, M. (1987) J. Cell. Physiol. 130, 1-5 19. Hartman, S. C., and Mulligan, R. C. (1988) Proc. Natl. Acad. Sci. U. S. A. 85,8047-8051 20. Wigler, M., Silverstein, S., Lee, L.-S., Pellicer, A., Cheng, Y.-C., and Axel, R. (1977) Cell 1 1 , 223-232 21. Seyedin, S., Thomas, T. C., Thopmson, A. Y., Rosen, D. M., and Piez, K. A. (1985) Proc. Natl. Acad. Sci. U. S. A. 8 2 , 22672272 22. Cheifetz, S., Bassols, A., Stanley, K., Ohta, M., Greenberger, J., and Massagui., J. (1988) J. Biol. Chem. 263,10783-10789 23. Laiho, M., Ronnstrand, L., Heino, J., DeCaprio, J. A., Ludlow, J. W., Livingston, D. M., and MassaguB, J. (1991) Mol. Cell. Biol. 11,972-978 24. Cheifetz, S., Weatherbee, J. A., Tsang, M. L. S., Anderson, J. K., Mole, J. E., Lucas, R., and MassaguB, J. (1987) Cell 48, 409415 25. Ignotz, R. A., and MassaguB, J. (1986) J. Biol. Chem. 261,43374345
