Transforming growth factor-beta inhibition of epithelial cell proliferation linked to the expression of a 53-kDa membrane receptor.

Cells whose proliferation is blocked by transforming growth factor-beta (TGF-beta) express three distinct surface glycoproteins of 53, 73, and 300 kDa that bind TGF-beta with high affinity, but whose function is unknown. We have isolated two classes of chemically-induced Mv1Lu epithelial cell mutants resistant to growth inhibition by TGF-beta. Class R mutants have selectively lost expression of the 53-kDa (type I) TGF-beta-binding protein. They have also lost the ability to respond to TGF-beta with elevated fibronectin expression and cell flattening. Class S mutants bind normally but do not respond to TGF-beta. TGF-beta-resistant mutants retain a contact inhibited, nontransformed phenotype. The properties of S mutants suggest that they are defective in the TGF-beta signal transduction mechanism, while the results with R mutants identify the type I TGF-beta-binding protein as the receptor involved in mediating TGF-beta actions on cell adhesion and proliferation.

The concept of negative growth regulation has materialized in recent years with the identification of diffusible polypeptides that inhibit mammalian cell proliferation. Transforming growth factors @l and @2 (TGF-@'1 and TGF-@2) are two prototypic growth inhibitors whose antimitogenic action in vitro affects cells of epithelial, vascular endothelial, lymphocytic, hematopoietic, and neuroectodermal lineages (1,2). Due to their multifunctional nature, they control cell differentiation as well as proliferation, the TGFs-@ may function in vivo as important effectors in processes of morphogenesis, tissue development, and repair (3)(4)(5)(6)(7). At the cellular level, the action of TGFs-@ is characterized by changes in the expression of growth regulatory genes (8,9) and genes whose products mediate cell adhesion (3,10,11). By analogy with the mode of action of other polypeptide hormones, TGFs-@ may act by binding to cell surface receptors that initiate the transduction of antimitogenic signals and other gene regulatory information to the nucleus. However, the biochemical components of this pathway are virtually unknown.
This problem derives in part from the fact that the TGF-@ receptor-signaling mechanism appears to be different from those of other hormone and growth factor receptors. The antiproliferative action of TGFs-@ can take place without interference with the biochemical events that immediately follow cell stimulation by mitogens. For example, TGF-@1 * The abbreviations used are: TGF-0, transforming growth factor-@; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid.
blocks the proliferative response of epithelial cells to epidermal growth factor without altering the binding or signaling capacity of EGF receptors (12,13). Hence, efforts to identify TGF-@ receptors on the basis of their signaling activity have been thus far unsuccessful. However, putative TGF-@ receptors have been described based on their ability to bind TGFs-@ with high affinity and specificity (14)(15)(16). Three distinct types of TGF-@ binding components or "receptors" have been identified on the surface of almost all cell types examined. They have been operationally defined as TGF-@ receptor types I, 11, and 111, respectively (4,17). Two of these receptors are glycoproteins of 53 kDa (type I) and 73-95 kDa (type 11) that display higher affinity for TGF-@l than TGF-@2 (18). The third receptor type has the structure of a 300-kDa chondroitin/heparan sulfate membrane proteoglycan with a core protein of 110 kDa that binds TGF-@l and TGF-@2 with similar affinity (18)(19)(20)(21).
The functional role of these cell surface components is not known. In principle, they could participate in tasks other than signal transduction, including various modes of TGF-@ retention and transport into and across the cell. Preliminary evidence to define which one of these binding proteins mediates the actions of TGF-@ is limited to correlative arguments linking the differential binding of TGF-Pl and TGF-@P to the putative receptors and the ligand's differential potency in biological assays. In one case, the higher affinity of TGF-@1 for the type I receptor in mouse hematopoietic progenitor cells correlates with the higher potency of TGF-@l to inhibit growth of these cells when compared with TGF-@2 (22). Similar lines of correlative evidence have suggested the involvement of the type I11 receptor in actions that TGF-@l and TGF-@2 elicit with similar potency, such as inhibition of epithelial cell proliferation (18), regulation of the expression of certain phenotypes, and elevated expression of cell adhesion proteins (11,23). The three TGF-@ receptor types are ubiquitously distributed in cell lines and primary tissues. Their presence as well as the presence of growth inhibitory responses to TGF-@ in normal human retina cells but not in retinoblastoma cell lines has suggested that one or more of these TGF-@-binding proteins might mediate growth inhibition, the loss of receptors leading to loss of growth control in retinoblastoma cells (24).
These findings have left unsolved the issue of which if any of the TGF-/3 binding proteins identified to date is directly involved in mediating the action of TGF-@. Here we report that selection of lung epithelial cell mutants defective in growth inhibitory responses to TGF-@ results with high frequency in the isolation of cell clones that have selectively lost expression of functional type I TGF-@ receptors, a finding that strongly implicates this receptor type as the mediator of TGF-P action.  (31. Cultures were grown at 37 "C in a humidified atmosphere containing 5% co,. Production of TGF-P-insensitiue Mutants-Pilot studies were done to determine the appropriate cell density for maximal inhibition of MvlLu cell proliferation by TGF-P1 and concentrations of ethyl methanesulfonate necessary to achieve 70-90% cell killing. Cells were seeded in 100-mm plastic tissue culture dishes at a density of 1.3 X lo5 cells/cm2 in 10 ml of growth medium. The next day, cultures were treated with ethyl methanesulfonate (Sigma) at either 0.1 pl/ml or 0.5 pl/ml of media for 24 h. Culture media was changed and 48 h later, 100 PM human TGF-01 was added to the culture media. Cultures were fed with growth media supplemented with 100 pM TGF-01 every 3 days. When macroscopically visible colonies had arisen, colonies were ring cloned, transferred into individual culture vessels, and propagated in the presence of 100 PM TGF-P1. A total of 26 TGF-8resistant colonies were cloned from 18 culture dishes. Colonies isolated from plates treated with 0.5 pl/ml ethyl methanesulfonate have a "5" as the second character in their designation and colonies isolated from plates treated with 0.1 pl/ml ethyl methanesulfonate have a "1." Cell lines which demonstrated stable resistance to TGF-0 were subcloned by limiting dilution and all subsequent experiments used clonal lines.
Affinity Labeling of Mutant Lines-Subconfluent monolayers of various cell lines were affinity labeled as previously described (16). Briefly, cultures were incubated with '251-TGF-01 in the presence or absence of competing TGF-(31 or TGF-02 as indicated for 3.5 h at 4 "C. Cultures were washed extensively and cross-linked using the bifunctional reagent disuccinimidyl suberate (Pierce Chemical Co.). Cells were then solubilized in a buffer containing 1% Triton X-100, 1 mM EDTA, 10 mM Tris, pH 7.4, and protease inhibitors. Gel electrophoresis of the Triton-solubilized receptors reduced with 50 mM dithiothreitol was performed according to the procedure of Laemmli (25) on 6% polyacrylamide gels. Gels were stained for protein with Coomassie Blue, dried, and exposed to Kodak XAR film with enhancing screens (Du Pont Lightning Plus).
Receptor Binding Assays-Binding of 'z51-TGF-01 to cell monolayers was performed as described previously (16). Near confluent monolayers of various lines were incubated in binding buffer (128 mM NaCl, 5 mM KCl, 5 mM MgSO,, 1.3 mM CaC12, 50 mM Hepes, pH 7.4, containing 5 mg/ml of bovine serum albumin) for 30 min, at 37 "C to wash off serum and endogenous proteins. Cultures were subsequently incubated with varying concentrations of lZ5I-TGF-pl and unlabeled TGF-8 for 3.5 h at 4 "C. Cultures were washed four times with binding buffer and solubilized with 1% Triton X-100, 1 mM EDTA, 10 mM Tris, pH 7.4. Nonspecific binding was measured in the presence of 10 nM unlabeled TGF-P and subtracted from total binding to determine specific binding.
Assay of Thymidine Incorporation-Subconfluent cultures of various cell lines were incubated with 100 p~ TGF-01 for 30 h. Cultures were incubated with 1 pCi/ml of [3H]thymidine (Amersham Corp.) for 3 h. Radioactivity incorporated into cellular trichloroacetic acidprecipitable material was assayed as previously described (12).
Assay of Fibronectin Production by Mutant Lines-The effect of TGF-8 on production of fibronectin in various cell lines was assayed as previously described (10). Subconfluent monolayers of various cell lines were treated with 100 p~ TGF-01 for 15 h and subsequently biosynthetically labeled in cysteine-free, serum-free medium containing 10 pCi/ml of [35S]cysteine in the presence or absence of 100 PM TGF-P1 for 3 h. The resulting labeled medium was incubated with gelatin-Sepharose beads (Pharmacia LKB Biotechnology Inc.) for 16 h at 4 'C to isolate fibronectin. The beads were washed, heated in sample buffer, and analyzed by electrophoresis on 7% polyacrylamidesodium dodecyl sulfate gels and fluorography using Enlightning fluid (Du Pont-New England Nuclear).
Anchorage-independent Growth Assays-Cell lines were assayed for anchorage-independent growth as previously described (26). Approximately 1000 cells from various mutant and parental lines were plated into 35-mm tissue culture wells in a suspension of 1.5 ml of Dulbecco's modified Eagle's medium with 10% fetal calf serum and 0.4% agar on a base layer of the same medium containing 0.5% agar. Dishes were incubated at 37 "C in a humidified 5% COz atmosphere for 3 weeks. Cultures were assayed by microscopic examination.
Tumorigenesis Assays-Subconfluent cultures of various cell lines were dissociated from culture plates with physiological saline, 5 mM EDTA. Cells were pelleted and resuspended in physiological saline. 1 X IO7 cells in a volume of 0.2 ml were injected subcutaneously into the lower abdomen of nude mice. Mice were examined every 3-5 days until tumors developed or for 10 weeks if no tumors developed.

RESULTS
Isolation of Mutant Cell Lines Resistant to TGF-@-The strategy to isolate cell mutants resistant to growth inhibition by TGF-@ was based on maintaining sparse cell cultures in the continuous presence of TGF-B1 and isolating cell colonies that grow under these conditions (27). For these studies, we chose the diploid MvlLu mink lung epithelial cell line whose proliferation is arrested by TGF-@l or TGF-P2 acting at picomolar concentrations (15,18). The spontaneous colony forming efficiency of the MvlLu parental cell line maintained in the presence of TGF-B1 was -1/106. Ten such colonies were picked and individually propagated for several weeks. They all eventually reverted to the wild type, TGF-@-inhibited phenotype. The colony forming efficiency of MvlLu cells mutagenized with ethyl methanesulfonate and maintained in the presence of TGF-@l was also -1/106 cells. Most of these colonies and their subclones were significantly growth inhibited by TGF-@ upon expansion of the cultures, as determined by [3H]thymidine incorporation assays. With time in culture, these clones also regained full sensitivity to TGF-@. These clones probably represented the small subpopulation of MvlLu cells that becomes spontaneously resistant to TGF-@ for a limited number of generations. However, of 26 colonies originally isolated from 2 x lo7 mutagenized MvlLu cells, six independently isolated colonies as well as all their subclones have remained completely resistant to TGF-@ for over 6 months in culture, and have been characterized in detail.
TGF-Pl inhibited incorporation of [3H]thymidine into DNA almost completely in the parental MvlLu cell line, but did not decrease this parameter in any of the six mutant cell lines ( Fig. 1). A concentration TGF-Pl 1000-fold higher than that required for a significant inhibitory effect in the parental  cell line was without inhibitory effect on any of the mutant cell lines. The mutants resistant to TGF-Pl were also resistant to TGF-P2 (Fig. 1, and data not shown). The proliferation rate of the six TGF-P-resistant mutant cell lines was the same in the presence or absence of TGF-P (not shown).

TGF-p Receptor Identification
The mutant phenotype in these cell lines is remarkably stable. Using a replica filter assay to screen for cell colonies that do not incorporate [3H]thymidine in the presence of TGF-P,2 no spontaneous revertants to the wild type phenotype could be detected in stock cultures of mutants propagated for 3 months in the absence of TGF-P. Furthermore, no revertant clones could be isolated after exposing mutant clones (R1B and R5L) to cytosine arabinoside in the presence of TGF-P for four consecutive cycles to selectively kill the TGF-Presistant population. Thus, mutational events in the six MvlLu clones are the probable cause of the stable loss of cell functions required for the growth inhibitory response to TGF-P l and TGF-P2.
Mutations Block Multiple Responses to TGF-@-Two other responses of MvlLu cells to TGF-P are increased production of extracellular matrix proteins and morphological changes that lead to a flat and grossly enlarged cell morphology. The elevation of fibronectin expression is a response to TGF-P seen in many other cell lines of epithelial and mesenchymal origin and is commonly accompanied by marked changes in the expression of other matrix proteins and cell adhesion receptors (3,10,11,28). TGF-P1 increased the production of extracellular fibronectin %fold over basal levels in the parental MvlLu cell line (Fig. 2). This increase contrasted with the lack of a detectable change in fibronectin production by the -TGF-Ol +TGF- Ol   FIG. 3. Effect of TGF-@ on cell morphology. Parental MvlLu cells, R1B mutants, and S1B mutants were incubated with medium alone or containing 100 PM TGF-61. Phase-contrast photomicrographs were taken after 3 days in culture. The mutant cells failed to respond to TGF-01 with an enlarged and flattened morphology. All photomicrographs are shown at the same magnification.
six MvlLu cell mutants in response to TGF-P1. The intensity of the fibronectin response to TGF-P1 can be affected by the density of the cell culture^.^ However, the MvlLu cell mutants failed to respond to TGF-/3 at sparse as well as confluent cell density (not shown).
Cell flattening and enlarged morphology are typically observed in association with the strong growth inhibitory response of epithelial and vascular endothelial cells to TGF-P (9,12). Fig. 3 illustrates this phenomenon in parental MvlLu cells and the lack of it in two representative mutant cell lines, S1B and R1B. The other four mutant cell lines also failed to respond to TGF-/3 with morphological changes.
Inhibition of DNA replication, elevation of fibronectin synthesis, and changes in morphology represent separate aspects of a pleiotropic response to TGF-P. The failure of the six MvlLu mutant cell lines to display any of these responses sdggested that the mutation(s) in each of these cell lines affected a central component in the TGF-/3 signal transduction mechanism. Therefore, we examined the ability of these cells to bind TGF-8.
Selective Loss of Type I Receptor in TGF-P-resistant Mutants-Scatchard analysis of '251-TGF-/31 equilibrium binding to parental MvlLu cells indicated the presence of -9,000 surface-binding sites/cell. The curvilinear nature of the Scatchard plot (Fig. 4) indicated the presence of multiple classes of binding sites defined by K d values ranging between 25-300 PM according to the two limit slopes of the curve. These results agree well with the presence of three types of TGF-Pbinding proteins in MvlLu cells, and their respective K d values deduced from receptor affinity labeled experiments (18; and see Figs. 5 and 6 below). Binding of '251-TGF-P1 to monolayers of the six TGF-P-resistant mutants was not grossly different from the parental cells, as seen in Fig. 4 for four of the six mutant cell lines. However, alteration of a given subpopulation of binding sites might not be detected by  measurement of total TGF-@ binding but might become apparent upon visualization of individual receptor types by affinity labeling with '251-TGF-@l.
Like other cell lines, parental MvlLu cells express on their surface TGF-@-binding components of 53 kDa (type I), 73 kDa (type 11), and 300 kDa (type 111) that form affinitylabeled complexes of 65, 85, and 300 kDa, respectively, by cross-linking to '"I-TGF-@l via disuccinimidyl suberate (Fig.  5, and Ref. 17). Analysis of two TGF-@-resistant mutants, A . The wavy line that runs across all the lanes of this figure between the 45-and 68-kDa marks is due to an anomaly of the electrophoresis running front. S1A and SlB, and their subclones yielded a pattern of affinitylabeled TGF-@ components very similar to the pattern in parental MvlLu cells. However, mutant cell lines RlB, R5C, R5D, and R5L and all of their subclones lacked detectable type I TGF-@ receptors (Fig. 5). This loss contrasted with the normal appearance of the type I1 and I11 TGF-@-binding proteins in all mutant cell lines.

6.
To determine whether the absence of type I TGF-@ receptors in four of the six TGF-@-resistant mutants was a random event due to clonal variability, we analyzed various clones isolated from the parental cell line and the mutagenized MvlLu cell population. Of 30 clones randomly isolated from the parental cell line, and seven TGF-@-sensitive clones isolated from the same mutagenized MvlLu population that yieled the six mutants described here, none showed loss of TGF-@ receptors. The profiles of surface components affinitylabeled with '251-TGF-@1 in these 37 clones were the same as the profile in the parental MvlLu cell population (Table I).
Thus, the selective loss of type I TGF-@ receptor in four of six TGF-@-resistant mutants was a nonrandom event, and occurred with high frequency only in these mutants.

Unimpaired TGF-@ Binding to Receptor Types ZZ and ZZZ in
Class R Mutants-To detect possible alterations in the ligandbinding properties of TGF-@ receptor types I1 and I11 that might derive from the loss of type I receptors, we examined one of the clones, RlB, in more detail. A feature that distinguishes receptor types I and I1 from receptor type I11 is their ability to discriminate between various forms of TGF-@. Receptor types I and I1 have 10-30 times higher affinity for TGF-@1 than TGF-@2 (18,22), a property that can be visualized by the higher potency of TGF-@1 to inhibit affinitylabeling of these two receptor types in parental MvlLu cells (Fig. 6A). The type I1 receptor present in R1B cells had an unaltered ability to discriminate between TGF-@1 and TGF-  @2 and showed the same affihity for each of these two ligands as type 11 receptors in parental cells (Fig. 6A). The type I11 receptor that displays similar affilnity for TGF-B1 and TGF-j32 in the parental cell line also retained this property in R1B cells.
Some differences were observed between the apparent molecular weight of type I11 receptors from parental cells and some of the TGF-8 resistant cell lines (Figs. 5 and 6A). Similar changes occur in the type I11 receptor from the parental cell line with time in culture and are due to changes in the composition of chondroitin/heparan sulfate glycosaminoglycan side chains present in the type I11 receptor of MvlLu cells (23).
The stability of the structural and ligand-binding properties of receptor types I1 and I11 contrasted with the loss of functional type I receptors. Type I receptors could not be detected even when mutant cells were affiiity-labeled with a high (100 PM) concentration of 1251-TGF-/31 (Fig. 6B). Under these conditions it should have been possible to detect residual type I receptors if they had suffered as much as a 90% reduction in receptor copy number or affiity for TGF-81 (Fig. 6B).
Mutants Retain a Contact-inhibited, Nontransformed Phenotype-The basal proliferation rate of the TGF-@-resistant mutants was similar to or slower than that of the parental cell line ( Table 11). The proliferation of all six mutants was arrested when cells reached the confluent state, the number of cells/dish remaining essentially constant thereafter. The saturation density number was similar to or lower than that of parental cells. These growth properties contrasted with the higher growth rate and lack of contact inhibition observed in 62M cells, a clone of MvlLu cells transformed with the v-fms oncogene (29) used here as a positive control for oncogenic transformation. The morphology of the parental cell line and six mutant lines was normal (Table II; see also Fig. 3), although mutant lines S1A and R5C exhibited a spindly or elongated morphology. Only R5C and 62M cells formed slowgrowing colonies that contained 10-20 cells after 3 weeks of culture in soft agar. Subcutaneous injection of R5C cells in nude mice led to the formation of tumors at the site of injection in three of three animals. These tumors reached up to 18 mm in length 6 weeks after injection. 62M cells generated tumors that reached up to 30 mm in length after 6 weeks in all animals injected. Other TGF-&resistant mutants tested, R5D and SlA, did not produce tumors in nude mice after 10 weeks.

Two
Classes of TGF-@-resistant Mutants-Isolation of MvlLu cell clones that grow in the presence of TGF-8 yields two general classes of stable mutants, those that have lost type I TGF-/3 receptors (class R mutants) and those that retain these receptors but fail to respond to TGF-8 (class S mutants).
Two-thirds of the mutants that we have obtained correspond to the R class. The selective loss of type I TGF-/3 receptor in class R mutants is significant because it occurs frequently in TGF-&resistant mutants but is a very rare event otherwise. Thus, the type I TGF-@ receptor is present in each one of 37 parental or mutagenized MvlLu clones that are sensitive to TGF-8. This receptor is also present in the majority of the 96 cell types and tissues screened in this laboratory. Cell lines that lack detectable type I receptors include various human retinoblastoma lines, and PC12 rat pheochromocytoma cells, none of which show any detectable response to : The sensitivity of class R mutants to growth inhibition by TGF-Bl is at least lo00 times lower than parental cells. Given that parental MvlLu cells have a low number of type I TGF-8 receptor copies on their surface estimated at -1000/cell, it is likely that the loss of functional types I receptors in class R mutants is complete. This loss could be due to lack of receptor expression or expression of a nonfunctional receptor protein.
The relatively low frequency of isolation of class S mutants was unexpected given the presumed large size of the target for mutations that could generate this phenotype (the entire postreceptor pathway that leads to growth inhibition by TGFj 3 ) compared with the size of the target that is mutated in class R mutants (the type I receptor gene, or the genes that may specifically regulate its expression). Furthermore, class S mutants express type I TGF-8 receptors but are remarkably similar to class R mutants in their failure to display a range of proliferative, biochemical and morphological responses to TGF-8. The phenotype of class S mutants suggests that they are defective in a central component(s) of the TGF-&signaling transduction mechanism. The similarities between the two classes of mutants suggests that the defects in S mutants might be in early events in the type I TGF-p receptor signaling pathway. We do not yet know whether any of the class R and class S mutants belong to the same complementation group, a finding that would suggest the presence of mutations in the type I receptor that block signaling without affecting the ability to bind TGF-8.
Expression of extracellular matrix fibronectin is a major target for regulation by TGF-@ (10, 30) and is probably involved in generating the characteristic morphology of TGF-@-treated cells. The loss of fibronectin and morphological responses in all mutants resistant to growth inhibition by TGF-(3 is of interest. The only other stable TGF-@-resistant mutant cell line previously described (clone 4, Ref. 27) also shows the concomitant loss of growth inhibitory response and morphological response to TGF-@. The concomitant loss of multiple responses to TGF-@ suggests that they are either causally linked to each other or they are mediated by separate pathways that diverge at a point downstream from the defective component(s) in class S mutants.
TGF-8 Resistance, Growth Inhibition, and the Transformed Phenotype-The TGF-@-resistant cell mutants isolated in this study show that the growth inhibitory response of MvlLu epithelial cells to TGF-@ is neither required for contactinhibition of growth, nor critical for the maintenance of a normal, nontransformed phenotype. TGF-@ was originally identified as a secretory product from dense cultures of BSC-1 simian kidney epithelial cells that inhibited the growth of sparse cultures of BSC-1 and other epithelial cell lines (15,31). However, MvlLu cell mutants that have lost growth inhibitory responses to TGF-@ become spontaneously growth arrested when they reach a cell density similar to or even lower than the saturation density of parental MvlLu cells.
We have previously proposed that the loss of TGF-@ receptors and growth inhibitory responses seen in many human retinoblastoma cell lines might represent an important step in the acquisition of the tumorigenic potential by these cells (24). The hyperproliferative phenotype of cell lines derived from human breast epithelial carcinoma also correlates with their inability to respond to the growth inhibitory action of TGF-@ (32). However, the present results show that loss of type I TGF-@ receptors and growth inhibitory responses in MvlLu cells is not sufficient to induce a transformed phenotype. The morphology and growth properties (generation time, saturation density, and anchorage dependence) of several TGF-@-resistant cell lines are essentially the same as those of the parental line. Of all TGF-&resistant mutants, only M C cells can grow small colonies in soft agar culture and small, noninvasive tumors in nude mice. The clear lack of correlation between TGF-@ resistance and phenotypic transformation in MvlLu cells suggest that the loss of TGF-j3 receptors or responses is not sufficient to induce the transformed phenotype. However, loss of TGF-@ growth inhibitory responses might increase significantly the rate at which additional events lead to an openly tumorigenic phenotype. In this regard, it should be noted that defects in a specific genetic locus (33-35) as well as lack of normal TGF-8 receptor types I1 and I11 accompany the lack of type I TGF-8 receptors in all human retinoblastoma cell lines examined (24).
The Type I TGF-@ Receptor as the Mediator of Multiple TGF-@ Actions-The present study identifies which one among multiple cell surface TGF-@-binding proteins is directly involved in mediating TGF-@ action. Identification of the type I receptor as the mediator of TGF-@ action is based on the frequent and selective loss of this receptor in TGF-@resistant mutants, an event that contrasts with the highly conserved expression of this receptor in the vast majority of cell lines examined. The loss of functional type I receptors occurs without any detectable change in the properties of TGF-@ receptor types I1 and 111, suggesting that the different receptors are products of separate genes and contain func-tionally independent ligand-binding sites.
The link between the pleiotropic growth inhibitory response of MvlLu cells to TGF-8 and the expression of the type I TGF-@ receptor was unexpected. TGF-@1 and TGF-82 display similar growth inhibitory potency in MvlLu cells, yet TGF-81 has higher affinity than TGF-@2 for the type I receptor.
Of the three types of cell surface TGF-@-binding proteins, only the type I11 receptor has an affinity for TGF-m that is as high as its affinity for TGF-@l. The type I receptor has 10-30-fold higher affinity for TGF-@l than for TGF-82 (18, 21; this report) independent of the temperature or cell density in the assays.
How can the ligand-binding properties of the type I receptor be reconciled with the evidence provided by class R mutants that this receptor type is the mediator of TGF-@ action in MvlLu cells? There are several potential explanations for the discrepancy. An obvious possibility is that TGF-82 might be more effective than TGF-@l in activating the signaling function of the receptor upon binding to it. This possibility is essentially untestable until early signaling events of the type I receptor have been identified. A second possibility is that TGF-@2 might be more resistant than TGF-@l to degradation in the biological assays. However, MvlLu cells degrade TGF-81 and TGF-@2 at similar rates.' A third possibility is that the type I11 TGF-@ receptor, which structurally is a membrane heparan/chondroitin sulfate proteoglycan (20,21), might modulate the activity of the type I receptor. Extracellularly, type 111 receptors might modulate the response of cells by acting as a nonsignaling binding protein capable of concentrating stores of TGFs-@ near the cell surface. As precedents, certain proteoglycans act as reservoirs of growth factors (36), and biologically inactive analogues of polypeptide hormones can display biological activity by inducing the release of active hormone from nonsignaling receptors into the pericellular medium (37). Intracellularly, activation of the type I11 receptor might modulate rate-limiting steps in the type I receptor signaling pathway. Evidence for the hypothesis that the type I11 receptor may facilitate the action of TGF-82 comes from the observation that TGF-@2 is as potent as TGF-Bl on cells that express type I11 receptors together with type I receptors, such as MvlLu cells (18,21), but is much less potent than TGF-@l in cells that display only type I receptor (22). Direct testing of this hypothesis is underway.
The results obtained with class R and class S mutants indicate that multiple biochemical, proliferative, and morphological responses of epithelial cells to TGF-@ originate from activation of a common type I receptor. It is possible that the other receptor types may mediate effects of TGF-@ not assayed in this study, or modulate cell responsiveness to different forms of TGF-@. Investigation of other cell types with the approach that we have used here and characterization of isolated TGF-@-binding proteins and their genes will be needed to extend the emerging concepts on the mechanism of TGF-@ action.