Heterodimeric Transforming Growth Factor @ BIOLOGICAL PROPERTIES AND INTERACTION WITH THREE TYPES OF CELL SURFACE RECEPTORS*

Type beta transforming growth factors (TGF) are disulfide-linked homo- and heterodimers of two related polypeptide chains, beta 1 and beta 2. The homodimers TGF-beta 1 and TGF-beta 2 are widely distributed, but the heterodimer TGF-beta 1.2 has been found only in porcine platelets (Cheifetz, S., Weatherbee, J.A., Tsang, M.L.-S., Anderson, J.K., Mole, J.E., Lucas, R., and Massagué, J. (1987) Cell 48, 409-415). Here we characterize the receptor binding and biological properties of TGF-beta 1.2 and compare them with those of TGF-beta 1 and TGF-beta 2. Three types of cell surface receptors previously identified by affinity labeling with 125I-TGF-beta 1 are available for binding to TGF-beta 1.2. These three types of receptors are detected as 65-kDa (type I), 85-95-kDa (type II), and 250-350-kDa (type III) affinity-labeled receptor complexes on electrophoresis gels. They co-exist in many cell types, have high affinity for TGF-beta 1, and varying degrees of affinity for TGF-beta 2. Of the 11 cell lines screened in the present study none showed evidence for additional receptor types that would bind TGF-beta 2 but not TGF-beta 1. In receptor competition studies, TGF-beta 1, TGF-beta 1.2, and TGF-beta 2 competed for binding to type I and type II receptors with a relative order of potencies of 16:5:1 and 12:3:1, respectively, whereas all three forms of TGF-beta were equipotent as ligands for the type III receptors. The three forms of TGF-beta were equally potent at stimulating the biosynthesis of extracellular sulfated proteoglycan in BRL-3A rat liver epithelial cells, a response that presumably involves the type III receptor present in these cells. In contrast, the ability of the three ligands to inhibit the growth of B6SUt-A multipotential hematopoietic progenitor cells which display only type I receptors decreased in the order TGF-beta 1, TGF-beta 1.2, and TGF-beta 2 with a relative potency of 100:30:1. The results indicate that the presence of one beta 1 chain in TGF-beta 1.2 increases (with respect to TGF-beta 2) the biological potency and binding affinity toward receptor types I and II, but the presence of a second beta 1 chain in the dimer is required for full potency.

J. (1987) Cell 48,[409][410][411][412][413][414][415]. Here we characterize the receptor binding and biological properties of TGF-81.2 and compare them with those of TGF-81 and TGF-82. Three types of cell surface receptors previously identified by affinity labeling with '261-TGF-/31 are available for binding to TGF-81.2. These three types of receptors are detected as 65-kDa (type I), 85-95-kDa (type 11), and 250-350-kDa (type 111) affinity-labeled receptor complexes on electrophoresis gels. They coexist in many cell types, have high affinity for TGF-81, and varying degrees of affinity for TGF-82. Of the 11 cell lines screened in the present study none showed evidence for additional receptor types that would bind TGF-82 but not TGF-81. In receptor competition studies, TGF-81, TGF-81.2, and TGF-82 competed for binding to type I and type I1 receptors with a relative order of potencies of 16:5:1 and 12:3:1, respectively, whereas all three forms of TGF-8 were equipotent as ligands for the type I11 receptors. The three forms of TGF-8 were equally potent at stimulating the biosynthesis of extracellular sulfated proteoglycan in BRL-3A rat liver epithelial cells, a response that presumably involves the type I11 receptor present in these cells. In contrast, the ability of the three ligands to inhibit the growth of B6SUt-A multipotential hematopoietic progenitor cells which display only type I receptors decreased in the order TGF-81, TGF-81.2, and TGF-82 with a relative potency of 100:30:1. The results indicate that the presence of one 81 chain in  increases (with respect to TGF-82) the biological POtency and binding affinity toward receptor types I and 11, but the presence of a second 81 chain in the dimer is required for full potency.
The TGF-B' family of polypeptides that regulate cell growth and phenotype includes numerous factors found in insects * This work was supported by National Institutes of Health Grant CA34610. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.  . We have speculated that the expression of multiple forms of TGF-P that have varying degrees of affinity for different types of cell surface receptors may provide a means to finely tune the response of multicellular systems to these polypeptides (5, 9). Cell types that respond differently to TGF-pl and TGF-@2 have indeed been found including various lines of multipotential hematopoietic progenitor cells (9). The existence of the heterodimer TGF-B1.2 increases the regulatory potential of the TGF-P system, but the limited availability of this product had not allowed the study of its properties until now. TGF-B1 and TGF-B2 bind with high affinity to three distinct components on the surface of target cells. These three components display affinity constants for TGF-B1 in the picomolar range, are not recognized by unrelated polypeptide hormones, and have properties of integral membrane glycoproteins. Based on these characteristics, they have been operationally defined as TGF-@ receptors and have been classified according to their structural and functional properties (reviewed in Ref. 10). Two of the three types of TGF-8 receptors, types I and 11, have higher affinity for TGF-P1 than 9). These two receptor types can be distinguished by the proteolytic peptide maps of their affinity-labeled domains (11). Type I TGF-B receptors are identified as affinitylabeled complexes of 65 kDa.' Type I1 receptors yield affinitylabeled complexes of 85 kDa in rat, mouse, and mink cells, 95 kDa in human and monkey cells, and 110 kDa in chick fibroblasts. A third TGF-P receptor type, type 111, is a 250-350-kDa glycoprotein with high affinity for both TGF-B1 and TGF-P2. In some cell lines type I11 receptors exist as part of a larger complex stabilized by disulfide bonds (12). With the exception of retinoblastoma and pheochromocytoma cells that lack TGF-@ receptors (13), at least one and most frequently all three TGF-P receptor types coexist in nearly 100 cell lines * The molecular weight values given for the affinity-labeled receptor complexes are estimates based on the migration of these complexes weight values include 12 kDa that correspond to one '*'I-TGF-@l on electrophoresis gels run against standards. The receptor molecular monomer chain cross-linked to the receptor protein in dithiothreitolreduced samples. and tissues that we have screened ( 11-15).3

10783
In the present studies we have characterized the biological activity of TGF-81.2 and its interaction with the three receptor types. The results provide new information on the contribution of individual TGF-8 chains to the receptor binding and biological properties of the TGFs-8.

Receptor Binding and Biological Properties of TGF8-1.2-
The ability of TGF-81.2 to compete with 1251-TGF-j31 for S. Cheifetz, F. T. Boyd, and J. MassaguB, unpublished work. ' Portions of this paper (including "Experimental Procedures,"part of "Results," and Figs. 1-4 and 6) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press. binding to and affinity labeling of cell surface receptors was compared with that of human platelet TGF-81 (which has the same amino acid sequence as porcine TGF-81 (21)) and that of porcine platelet TGF-82 in affinity-labeling competition experiments (Fig. 5). BRL-3A rat liver epithelial cells were selected for these experiments because they exhibit relative levels of affinity-labeled receptor types I, 11, and I11 that facilitate comparison between labeled species in the same autoradiogram.
The potency of TGF-81.2 in inhibiting the labeling of receptor types I and I1 was intermediate between those of TGF-81 and TGF-82 (Fig. 5). Densitometric scans of the autoradiograms revealed reproducible differences between the binding to these two receptor types. Thus, the approximate order of potencies of TGF-81, TGF-81.2, and TGF-82 to compete for type I receptors was 16:5:1. At low concentrations  After chemical cross-linking, electrophoresis, and autoradiography, a single affinitylabeled band was observed in these cells that corresponds to type I TGF-/3 receptors, as previously described (9). The results of densitometric analysis of this band are plotted as described in Fig. 5 TGF-P1.2 was similar to TGF-Pl and TGF-P2 in its ability to inhibit the labeling of type I11 receptors; all three ligands caused a half-maximal inhibition of labeling at about 25 pM concentration (Fig. 5). These results indicate that the three ligands have an equivalent affinity for type I11 receptors. Table I summarizes results of experiments comparing the ability of TGF-P1 and TGF-P2 to induce various responses in cultured cells whose TGF-@ receptor composition is known. In some cases noted in Table I, the relative ability of TGF-D l and TGF-82 to modulate these parameters has already been reported and is included in the table for purposes of comparison and summary.
The biological potency of TGF-81 and TGF-82 was similar in all cases except for the growth inhibitory response of hematopoietic progenitor cells which was induced much more potently by TGF-(31 than TGF-P2. B6SUt-A and 32DCL3 mouse hematopoietic progenitors exhibit only type I TGF-P receptors (9). Fig. 7A shows that TGF-81.2 competed with "51-TGF-/31 for binding to the type I receptors in B6SUt-A cells with a potency intermediate between that of TGF-P1 and TGF-P2. The response of these cells to TGF-P1.2 was examined in order to investigate further the correlation between receptor occupancy and induction of a biological response by the TGFs-@. Fig. 7B shows that [3H]thymidine incorporation into DNA in B6SUt-A cells responding to interleukin 3 (multicolony stimulating factor) was inhibited by TGF-81.2 with a potency intermediate between that of TGF-Pl and TGF-P2. The order of potencies of TGF-Dl, TGF-(31.2, and TGF-P2 to compete for type I receptors and to inhibit B6SUt-A proliferation was 16:4:1 and 100:30:1, respectively. Thus, although the differences in potencies between the various TGF-P forms seemed to be amplified at the postreceptor level in BGSUt-A cells, a good correlation exists between the ability of three forms of TGF-/3 to bind to type I TGF-P receptors and their biological potency in these cells.

DISCUSSION
The results of the present study provide information on three aspects of the interaction of TGF-Ps with receptors in target cells. First, the results show that in a sample of 11 cell lines the only receptors with which TGF-P2 can detectably interact are the three types of cell surface components previously identified as receptors for TGF-P1. Second, the results define biological and receptor binding properties of the heterodimer, TGF-P1.2. Third, the results with TGF-61.2 extend preliminary evidence correlating induction of particular cellular responses with occupancy of different types of TGF-(3 receptors. The results of binding experiments show that all the binding of Iz5I-TGF-P2 to cells can be competed as effectively by TGF-Pl as by TGF-p2. In reverse experiments, TGF-P2 inhibits strongly the labeling of type I11 receptors by lZ5I-TGF-(31 but inhibits more poorly the labeling of receptor types I and 11. This is consistent with the observation that TGF-P2 is less potent than TGF-P1 in competing with lZ5I-TGF-P1 for binding to intact cells and that '251-TGF-P2 affinity labels well the type I11 receptor but much less effectively the receptor types I and 11. Preincubation of cells with TGF-Pl to saturate binding sites for this ligand failed to uncover residual sites that would preferentially bind TGF-P2. In contrast to the ubiquitous and easily detectable occurrence of TGF-8 receptor types I, 11, and 111, the results indicate that if additional types of receptors exist that bind preferentially TGF-P2 over TGF-Pl they would have to be present at levels below the detectability limit of our assays (<300 copies/cell, Ref. 9) or be restricted to cell types not screened in this study.
The approach used here and in our previous work to identify TGF-P binding proteins has been used recently by Segarini et al. (26). Their results confirm ours in identifying three affinity-labeled receptor types of 65, 85, and 250 kDa, but with one significant difference. This difference lies in the inability of TGF-Pl to compete in their studies with lZ5I-TGF-P2 for binding to and labeling of all three receptor components in NRK-49F and Swiss 3T3 cells. Based on this result, Segarini et al. (26) have proposed that each one of the three receptor types is in turn a mixture of various classes of binding sites including some that bind preferentially TGF-P1 and some that bind preferentially TGF-P2. Our results are in disagreement with the observations of Segarini et al. (26). The reason for the discrepancy is not clear because the source of TGF-P1 and TGF-P2, and the source and passage number of the NRK-49F cells used in both studies were the same. However, differences in the results might arise from differences in lZ5I-labeled ligand preparations. We use milder oxidizing conditions than those previously described for iodination of TGF-P (18).
The availability of the heterodimer TGF-P1.2 has allowed a better characterization of the interaction of the TGF-(3s with their cell surface receptors. The results show that the presence of one Pl chain in TGF-P1.2 makes it a better ligand for receptor types I and I1 and a more potent agonist than TGF-P2. However, only the TGF-P1 homodimer displays full potency. It is of interest that the relative potencies of TGF-Pl, TGF-P1.2, and TGF-P2 for binding to type I receptors are 16:5:1, but the differences between these three ligands are amplified at the level of the biological response in B6SUt-A hematopoietic progenitor cells reaching an order of potencies of 100:30:1 (see Fig. 7). This observation is reproducible and is not due to a faster rate of degradation of TGF-P2 by B6SUt-A cells (9) or a wider divergence in the receptor affinities of TGF-Pl and TGF-P2 at 37 "C than at 4 "C (data not shown).
The basis for this phenomenon is not known. It is possible that each of the two chains in a TGF-P molecule binds one receptor molecule, ligand-induced receptor dimerization being necessary for receptor activation and cell stimulation. In this case, the relative potency for receptor activation by a given form of TGF-(3 would be closer to the product of the relative binding potencies of the individual TGF-P chains.
Comparisons between the biological potency and the receptor binding affinity represent a first approximation to the question of which receptors may be involved in mediating individual actions of the TGFs-P. The simplest interpretation of the available data suggests that occupancy of type I receptors may be involved in inhibition of proliferation in hematopoietic progenitor cells, while all the other responses measured including regulation of cell adhesion proteins and their receptors, and inhibition of adipogenic differentiation, myogenic differentiation, and epithelial cell proliferation may be mediated by type I11 receptors. Alternative approaches to probe more directly the biological properties of the various TGF-P receptors may uncover a scheme more complex than the one outlined here. In this context, it is intriguing that growth inhibition by TGFs-0 correlates with interaction of these polypeptides with type I receptors in B6SUt-A cells, but with type I11 receptors in MvlLu epithelial cells even though this cell line also has type I receptors. It is possible that growth inhibition is the end point of pathways that are different in these two cell types. However, cooperation between the signals of the two receptors in epithelial cells or