Differential Binding of Transforming Growth Factor-/?1, -02, and -03 by Fibroblasts and Epithelial Cells Measured by Affinity Cross-Linking of Cell Surface Receptors

Russette M. Lyons*, Duncan A. Millerf, Jeannette L. Graycar, Harold L. Moses, and Rik Derynck Department of Cell Biology Vanderbilt University School of Medicine (R.M.L., D.A.M., H.L.M.) Nashville, Tennessee 37232 Department of Developmental Biology Genentech, Inc. (J.L.G., R.D.) South San Francisco, California 94080

INTRODUCTION

A murine fibroblast cell line (AKR-2B clone 84A) and an epithelial cell line (BALB/MK) were compared for their ability to bind different transforming growth factor-/? (TGF/3) species. The results of competitive binding assays indicated that the epithelial cells had a higher affinity for TGF/3 than the fibroblasts. This difference may be the basis for the sensitivity of epithelial cells to much lower concentrations of TGF/S than fibroblasts. Affinity cross-linking studies showed that both cell types express the three cell surface TGF#-binding molecules that have been previously described for a variety of cell types. The complexity of these cell surface binding proteins was further evaluated using all possible combinations of radiolabeled ligands in competition with each of the three unlabeled TGF0 species. Differences in the ability of specific TGF/9 types to compete with radiolabeled TGF02 for binding to the type I and II receptors were observed, with TGF01 being more potent for epithelial cells, and TGF/32 being more potent for fibroblasts. In addition, a difference in the ability of different TGF/3 species to compete the [125I]TGF03 from epithelial cell surface receptors was apparent. TGF/92 was not able to compete with [125l]TGF/?3 for binding to the type II receptor at any concentration tested, while TGF/31 and TGF/S3 were about equally potent in competition for this receptor type. These differences in cell surface receptor binding of structurally and biologically similar molecules may reflect different functions for these molecules. (Molecular Endocrinology 5: 1887-1896, 1991)

Originally described as the most potent polypeptide growth inhibitor isolated from natural sources (1, 2), transforming growth factor-/? (TGF/3) may better be described as being pleiotropic in effect and function (for review, see Ref. 3). This protein, initially purified from human platelets, has been isolated from a variety of cells and tissues, both normal and transformed (2, 4). Biologically active TGF/3 is a 25-kDa molecule comprised of two identical subunits. This ubiquitous growth factor has a variety of in vitro effects which may depend on both cell type and culture conditions. Some of these biological activities include stimulation of fibroblast proliferation in both soft agar (5) and monolayer (6), increased production and deposition of extracellular matrix components (7, 8), chemotaxis of fibroblasts and macrophages (9, 10), inhibition of differentiation of some cell types (11,12), and inhibition of proliferation of nonmesenchymal cells (1, 2,13). In addition to the numerous and diverse activities that have been demonstrated for this polypeptide growth factor, it is now clear that TGF/3 belongs to a large family of both closely and distantly related proteins. The first TGF/3 isolated and characterized is now designated TGF/31. Closely related proteins have been identified using cDNA-cloning techniques. These include TGF/32 (14,15), TGF/33 (16-18), TGF04 (19), and TGF/35 (20). While TGF/32 has been purified from porcine platelets, naturally occurring TGF/33 has not been isolated. Recombinant TGF/?3 has been expressed by Chinese hamster ovary (CHO) cells transfected with the fulllength TGF/33 cDNA and has been partially characterized (21, 22). TGF/34 has been cloned from a chicken cDNA library (19), and TGF/35 has been cloned from a frog cDNA library (20); however, the mammalian homologs have yet to be discovered.

0888-8809/91/1887-1896$03.00/0 Molecular Endocrinology Copyright © 1991 by The Endocrine Society

1887

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 00:43 For personal use only. No other uses without permission. . All rights reserved.

Vol5No. 12

MOL ENDO-1991 1888

Distantly related members of the TGF/3 superfamily have been reported by various investigators and include the Xenopus Vg-1 (23), the Drosophila decapentaplegic gene complex (24), the mammalian inhibins/activins (25), Mullerian inhibiting substance (26), bone morphogenetic proteins (27, 31), and VgM/BMP-6 (28, 31), OP-1/BMP-7 (29, 31), and GDF-1 (30) polypeptides. Whereas most of the structural information has been derived from cDNA sequencing, only a few of these have been identified and isolated as mature proteins. While the in vivo role of these molecules is not clear, many of these proteins and genes have been implicated in the regulation of cell growth and developmental processes. Specific, high affinity TGF/3 receptors have been described on a variety of cell types (14, 32-39). Dissociation constants ranging from 25-140 pM and receptor numbers ranging from 2,000-40,000/cell have been reported. TGF/3 has been shown to interact with three distinct cell surface molecules by affinity labeling of cells. These are the glycosylated type I (55-65 kDa) and type II (70-85 kDa) receptors and the high mol wt proteoglycan type III (200-280 kDa) receptor. The contribution of each of these cell surface binding molecules to the TGF/3-induced specific response remains unclear. However, studies by both Boyd and Massague (40) and Segarini et al. (41) indicate that the low mol wt type I receptor is required for TGF/31 sensitivity. The TGF0 receptors differ from other well characterized growth factor receptors in that there have been no reports of kinase activity, intracellular Ca2+ fluxes, or inositol phosphate involvement associated with these cell surface binding molecules. Interestingly, however, the possible involvement of a GTP-binding protein has been reported (42) and further substantiated by Howe and Leof (43). Of particular interest in the study of the different TGF/3 species and TGF/3 receptors is our current knowledge of the biological activities of the three mammalian TGF/3 species. The limited data on the comparisons of the biological activities of TGF/31, -/32, and -03 in vitro suggest that the three species often, but not always, exert similar or identical activities (21,22). On the other hand, several recent studies have shown some major differences. For example, the regulation of mRNA expression of the TGF/3 genes is differentially regulated by the different TGF0 proteins themselves (44,45), and there are differences in the antiproliferative effects of the different TGF0 forms (46, 47). In addition, TGF/32 and -/33 are potent inducers of mesoderm formation in Xenopus, while TGF/31 is ineffective (48). Thus, the possibility that these molecules have different in vivo functions should not be disregarded. Several studies have evaluated the interactions of different TGF/3 species (14,36-39) and have established that TGF01, -/32, and -j83 have the ability to interact with all three receptor types. Competition studies in which unlabeled TGF/3 ligand competes with 125l-labeled ligand for binding to the receptors have been mainly restricted to a comparison of TGF/31 and -/32, but have revealed some differences in their interactions (14, 36, 37, 39). In addition,

TGF/33 has recently been compared with the two other TGF/3 forms for receptor binding (38). However, all of these studies taken together do not add up to a complete data set in which the binding characteristics of the three TGF/3 species to all receptor types are established on the basis of competition experiments with all possible combinations of radiolabeled and unlabeled ligands. In addition, whereas most studies (14, 37, 38) were performed with the Mv1 Lu epithelial cells, we do not have a complete and coherent set of data for different cell types. Therefore, we have in the current report evaluated the ligand-receptor interactions of the three TGF/3 isoforms in competition experiments using all different combinations of radiolabeled and unlabeled TGF/3 species, followed by electrophoretic analysis of the cross-linked receptor-ligand complexes. Such a systematic study was carried out using cells of two types: the AKR-2B fibroblast cell line and the BALB/ MK keratinocyte cell line. We observed unambiguous differences in the binding of the TGF/3 forms to the three receptor types in both cell lines. In addition, there were reproducible differences between the two cell types studied, fibroblast and epithelial. These differences between cell types and the binding of distinct TGF/3 species to cell surface receptors may provide an important in vivo mechanism that allows discrimination at the level of the responding cell or tissue type between the closely related TGF/3 molecules. Thus, such a systematic study of ligands and receptors further demonstrates the complexity of the TGF0 family of growth regulatory molecules.

RESULTS Biological Activity of Labeled Proteins The three TGF/3 species were radiolabeled with 125I using a modified chloramine-T method (33) and then subjected to a variety of biological assays. These assays included growth of fibroblasts in soft agar and monolayer, inhibition of cultured epithelial cells, and receptor assays. All radiolabeled TGF/3 types were functional in every biological assay performed (data not shown), although a slight decrease in potency was occasionally observed. Such minor differences may, however, be due to overestimations of protein recovery after iodination. In all cases, the same maximum level of biological activity could be achieved with iodinated molecules compared with unlabeled molecules. The 125l-labeled TGF/3 preparations used in the experiments discussed in this report were all tested for activity in standard TGFjfi RRAs and met the following criteria: 1) labeled molecules retained the ability to bind to the indicator cell surface, and 2) displacement of 50% binding of a particular iodinated TGF/J molecule from the cell surface could be achieved with 1 unlabeled molecule.

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 00:43 For personal use only. No other uses without permission. . All rights reserved.

1889

Binding of TGF01, -02, and -03

Responses of Epithelial and Fibroblast Cells to TGF/31 100 Preliminary experiments were performed to determine the half-maximal concentration of TGF/31 necessary to achieve a specific response from each cell type. Epithelial cells (BALB/MK) stimulated with epidermal growth factor (EGF) were exposed to increasing concentrations of unlabeled TGF/31 for 24 h, followed by a 2-h pulse of [3H]thymidine. A dose-dependent decrease in DNA synthesis was observed, with half-maximal inhibition occurring at 0.1 ng/ml (data not shown). The stimulation of AKR-2B fibroblast colony formation in soft agar indicated that the half-maximal concentration for these cells was 1 ng/ml (data not shown). These values are in agreement with previous reports (43) and indicate that epithelial cells are approximately 10-fold more sensitive to TGF/3 than are fibroblasts. These experiments also provided the necessary information for selecting a range of TGF/3 concentrations for the subsequent affinity labeling studies.

0.01

100 -,

0.1 1 10 TGF-0 (ng/ml)

100

0.1 1 10 TGF-0 (ng/ml)

100

0.1 1 10 TGF-/S (ng/ml)

100

B

80 -

Specificity of Cell Surface Binding of TGF/31, -02, and -03

S I 60 H N

RRAs were performed to determine the ability of each of the three TGF/? types to displace 125l-labeled TGF/3 species from cell surface receptors (Fig. 1). Each TGF/3 species was able to compete for receptor binding, although small differences in the concentrations needed to obtain 50% inhibition of binding were observed. The concentration necessary for half-maximal inhibition of binding ranged from 0.3 ng/ml for TGF01 and -/32 to 1.0 ng/ml for TGF/33. These small differences could in principle be due to variability in protein concentration or differences in binding to the three receptor types on the cell surface. This latter possibility prompted a careful analysis of the receptor-binding patterns of all three TGF/3 species using affinity cross-linking techniques to compare two cell types, AKR-2B fibroblasts and BALB/ MK epithelial cells, which exhibit different biological responses to TGFjS.

3

Z 40 -

20 0

0.01

100 -,

Affinity Cross-Linking Studies AKR-2B (clone 84A) and BALB/MK cells were affinity labeled with radioiodinated TGF01 (Fig. 2). This analysis revealed that both cell lines express the previously described pattern of three different TGF/3 receptor types, termed type I (55-65 kDa), type II (70-85 kDa), and type III proteoglycan (200-280 kDa). A lower mol wt form of the type III receptor is often detected after cross-linking. This form, which is often seen as a doublet (Figs. 2-4), presumably corresponds to unglycosylated polypeptides, as previously suggested (36-39). For simplicity and consistency, this nomenclature for the receptors will be used throughout the remainder of this study. When AKR-2B fibroblasts were affinity labeled with [125I]TGF/?1 (Fig. 2A), unlabeled TGF/31 was the most

0.01

Fig. 1. TGF0 RRA [125I]TGF/31 (A), [125I]TGF02 (B), or [125I]TGF03 (C) was used in competition with each of the unlabeled ligands, using AKR2B cells. The competition curves for the unlabeled ligands are represented in each panel by the following symbols; TGF/31, • ; TGF02, • ; and TGF03, A. All experiments were performed in triplicate, and the individual values were within 5% error of the mean values shown.

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 00:43 For personal use only. No other uses without permission. . All rights reserved.

MOL ENDO-1991 1890

Vol5No. 12

8 TGF/31 TGF/32 TGF/33 £ I II II I O 100 30 10 3 100 30 10 3 100 30 10 3 (ng/ml) — 200

B I

TGF/31

O 10 3

TGF/32

TGF/33

1 0.3 10 3 1 0.3 10 3 1 0.3 (ng/ml) — 200

Fig. 2. Affinity Cross-Linking of [125I]TGF01 AKR-2B fibroblasts (A) and BALB/MK epithelial cells (B) were subjected to affinity cross-linking in the presence of [125I] TGF/31 and the indicated concentrations of unlabeled TGF/31, -/32, and -/33. The positions of mol wt markers are indicated. The type I, type II, and type III receptors are also indicated. A doublet, presumed to be an unglycosylated form of the type III receptor, is marked (*).

effective TGF/3 species in competition for the three types of cell surface binding proteins. A greater than 50% reduction in band intensity on the cross-linking gels was achieved for all receptor types with 3 ng/ml competing unlabeled TGF/31. In contrast, about 10-30 ng/ml TGF/32 were needed to obtain a similar degree of competition with [125I]TGF/31 for receptor binding in similar experiments. Thus, TGF/32 did not compete equally for the three cell surface binding proteins. At the highest concentration of TGF/32 used (100 ng/ml), [125I]TGF/31 binding to the type II and III receptors was still evident, whereas type I receptor binding was completely abolished. The ability of unlabeled TGF/33 to compete with [125I]TGF/31 for receptor binding appears to be intermediate, being less effective than TGF/31 and more effective than TGF/32. There were no obvious differences in the binding of TGF/33 to different receptor types on the basis of this analysis, as seen by equal competition for the three cell surface binding proteins at any given concentration of unlabeled TGF/33. The results of affinity cross-linking of BALB/MK epithelial cells with [125I]TGF/31 (Fig. 2B) were similar to

the results obtained with AKR-2B fibroblasts described above. Unlabeled TGF/31 was the most effective in competing for receptor binding, followed by TGF/33, with TGF/32 being the least effective. The ability of unlabeled TGF/32 to compete for receptor binding did not differ among the three receptor types, as observed with fibroblasts (Fig. 2A). The estimated concentration of unlabeled TGF/31 required to achieve a 50% decrease in band intensity was approximately 10-fold less for BALB/MK cells compared to AKR-2B cells. This may correlate with the biological assays performed in this study as well as those previously reported (50), which demonstrated that BALB/MK cells are more sensitive to TGF/31 than are AKR-2B cells. Affinity Cross-Linking Studies Using [12SI]TGF£2 and -03 The affinity labeling of cell surface receptors using [125I] TGF/32 gave results very similar to those obtained with radiolabeled TGF/31 (Fig. 3). The three receptor types on both fibroblasts and epithelial cells were indeed affinity labeled with iodinated TGF/32. This result is in agreement with those of other investigators (14, 3 6 39, 41) and supports the hypothesis that TGF/31 and /32 bind to the same cell surface receptors. However, differences in the ability of unlabeled TGF/3 species to compete with the radiolabeled TGF/32 for receptor binding were readily apparent. Affinity labeling of fibroblasts with [ 125 l]TGF/32 followed by competition for receptor binding with unlabeled ligands resulted in a striking pattern (Fig. 3A). Unlabeled TGF/32 was extremely effective in competition for binding to all three receptor types. In contrast, unlabeled TGF/31 and -/33 were considerably less effective than TGF/32 in their ability to compete for binding to the type I and II receptors. Interestingly, TGF/31 and -/33 did not compete for type III receptor binding. The latter finding represents a discrepancy with Fig. 1B, which showed a relatively efficient total inhibition at 100 ng/ml TGF/31 and -/33. We do not understand the basis for this discrepancy, although it should be pointed out that the procedures in both types of experiments are very different. In addition to the different receptor types, a doublet of 120-130 kDa was affinity labeled with [125I] TGFj82. This doublet, observed by other investigators, may represent an unglycosylated form of the type III receptor (39, 40), which is supported by the binding and similar competition of [ 125 I]TGF02 for the type III receptor and the lower mol w t values. An analysis of [ 125 l]TGF/32 affinity labeling of the epithelial cells resulted in similar, but less striking, patterns (Fig. 3B). Again, unlabeled TGF/32 was most effective in competing for binding to all receptor types. TGF/31 was less effective, and TGF/33 was least effective in competition with [ 125 l]TGF/32 for receptor binding. As seen above with the AKR-2B fibroblasts, TGF/31 and -/33 were very ineffective in their competition for binding to the type III receptor and the 120- to 130-kDa doublet.

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 00:43 For personal use only. No other uses without permission. . All rights reserved.

1891

Binding of TGF/31, -/32, and -/33

A

1

TGF/31

TGF/32

TGF/31

TGF03

O 100 30 10 3 100 30 10 3 100 30 10

TGF/32

TGF/33

O 100 30 10 3 100 30 10 3 100 30 10 3

(ng/ml)

(ng/ml)

— 200 — 200

I B

TGF/32 TGF/33 TGF/31 II II I £ I O 10 3 1 0.3 10 3 1 0.3 10 3 1 0.3 (ng/ml)

B

I

— 68

1 TGF/31 TGF/32 TGF/33 § I || || | O 10 3 1 0.3 10 3 1 0.3 10 3 1 0.3 (ng/ml)

— 200

— 200

*

-96

*

•

— 96 — 68

Fig. 3. Affinity Cross-Linking of [125l]TGF/?2 AKR-2B fibroblasts (A) and BALB/MK epithelial cells (B) were subjected to affinity cross-linking in the presence of [125I] TGF/32 and the indicated concentrations of unlabeled TGF/31, -/32, and -03. The positions of mol wt markers are indicated. The type I, type II, and type III receptors are also indicated. A doublet, presumed to be an unglycosylated form of the type III receptor, is marked (*).

Affinity labeling of both cell types with 125Mabeled TGF/33 was also performed (Fig. 4) and demonstrated that TGF/33 binds to the three receptor types identified. In the case of AKR-2B fibroblasts (Fig. 4A), unlabeled TGF/31 and -/33 were equivalent in their ability to compete for receptor binding. TGF/32, however, was unable to compete with [125l]TGF/33 for binding to the type II receptor at any concentration. TGF/32 was able to compete for binding to both type I and III receptors as effectively as other TGF/3 species. A similar, yet somewhat different, pattern emerged when BALB/MK cells were affinity labeled with [125I] TGF/33 (Fig. 4B). Both TGF01 and -03 competed equally well for binding to all receptor species. TGF/32 was able to compete with [125l]TGF/33 for binding to the type III

Fig. 4. Affinity Cross-Linking of [125l]TGF/33 AKR-2B fibroblasts (A) and BALB/MK epithelial cells (B) were subjected to affinity cross-linking in the presence of [125I] TGF/33 and the indicated concentrations of unlabeled TGF/31, -/32, and -(83. The positions of mol wt markers are indicated. The type I, type II, and type III receptors are also indicated. A doublet, presumed to be an unglycosylated form of the type III receptor, is marked (*).

receptor; however, this TGF/3 species was ineffective at competing for either type I or type II receptor binding. In general, a 50% reduction in band intensities could be achieved at lower concentrations of unlabeled TGF/3 ligand in BALB/MK cells compared with AKR-2B cells when any TGF/3 species was analyzed. The differences in receptor affinity observed in these studies is reminiscent of differences in half-maximal concentrations of TGF/3 needed for a biological response of each cell type. Thus, receptor affinity may be a determining factor in the biological response of a particular cell type. DISCUSSION The experiments presented in this report demonstrate differences in the affinity of the TGF/3 receptors for the

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 00:43 For personal use only. No other uses without permission. . All rights reserved.

MOL ENDO-1991 1892

closely related TGF/3 species. These differences in affinity varied between the fibroblast and epithelial cell lines examined. Previous studies have indicated that the binding of TGF/31 to the three identified TGF/3 cell surface receptors displays different affinities depending on cell type and passage number (36) and that there are differences in the binding of the different TGF/3 forms to the different receptor types (14, 36-39). However, the interaction of TGF/31, TGF/32, and TGF/33 with the different TGF/3 receptors has not been systematically examined. We, thus, evaluated the interaction of the three mammalian TGFjS species with the individual receptor types. The radiolabeled ligands used in this study had retained full biological activity. Affinity labeling of cells using radioiodinated TGF/31 showed the typical TGF/3 receptor pattern; the type I, type II, and type III receptors were present on both fibroblasts and epithelial cells. The affinity of the TGF/3 receptors for TGF/31 was approximately 3- to 10-fold higher on epithelial cells than on fibroblasts, as determined by a 50% decrease in band intensity. A higher affinity of the receptors in BALB/MK cells is also seen with TGF/32 and -/33. It is interesting to note that BALB/ MK cells are also more sensitive to exogenous TGF/31 than are AKR-2B cells (50). The most striking difference between the TGF/3 forms was that TGF/32 did not compete well with [125I]TGF/31 for binding to the type II and III receptors in AKR-2B cells and for all three receptor types in BALB/MK cells. Previous results (37, 37) also established a weaker ability of TGF/32 to compete with TGF/31 for binding to the type I and II receptors in other cell lines. Our current data using AKR 2B and BALB/MK cells are, thus, in general agreement with these results, although we found a more efficient competition of TGF/32 to the type I receptors in AKR2B cells. The ability of TGF/33, intermediate between those of TGF/31 and TGF/32, is in agreement with the data obtained by Cheiftez et al. (38). Affinity labeling of cells with [125l]TGF/32 yielded a pattern distinct from that of [125I]TGF/31. Unlabeled TGF/32 was much more effective in competing radiolabeled TGF/32 from all receptor types on AKR-2B fibroblasts. In contrast, TGF/31 and -/33 were relatively ineffective in competing with radiolabeled TGF/32 for binding to the type III proteoglycan receptor species, as previously observed (36-39). TGF/31 and -/33 were also less effective than TGF/32 in displacing radiolabeled TGF/32 from the type III receptors, with TGF/31 being the least effective. Cells were also affinity labeled using iodinated TGF/33. While all three TGF/3 receptor types were affinity labeled using this molecule, unlabeled TGF/32 was not able to compete for type I or type II receptor binding at any concentration on BALB/MK cells or for type II receptor binding on AKR-2B cells, thus indicating major differences in affinities. The results presented in this report have revealed that the different TGF/3 species interact in different ways with the three TGF/3 receptor types in two cell lines representative of different cell types. These differences

Vol5No. 12

in interactions are presumably due to differences in affinities in some instances, but could also reflect differences in steric configurations of the ligand-receptor complexes. This latter possibility has been suggested in the case of interactions of TGFa and EGF with their common receptor (51). The complexity of the different TGF/3 cell surface receptor types with their own differential affinities for the TGF/3 species stresses the inadequacy of the commonly used RRAs. Not only do such assays provide misleading information about the interactions of TGF/3 with the cell surface receptors, but they may even fail to reveal differences in specific receptor-ligand interactions. The apparent differences between the interactions of the different TGF/3 species with each of the receptor types may be of major relevance for the biological activities of the three TGF/3 species and may provide a mechanism for cell- and tissue-dependent specificity, depending on the presence of the type of ligands and receptors involved. At present there is very little information about the biological relevance of the different receptor types. It has been suggested that the type III receptors provide sensitivity to TGF/3 (52), but more recent and more convincing data indicate that the type I receptor may be responsible for the TGF/3-induced effects on cell proliferation and extracellular matrix formation (34, 35). This, in turn, would suggest that the type II and III receptors may not be signal transducing receptors, but merely cell surface binding proteins. The proteoglycan nature of the type III receptor suggests that this receptor could be involved in depositing TGF/3 in the extracellular matrix. Molecular characterization of the receptors together with an in-depth understanding of the signal transduction and biological effects following interaction of TGF/3 with the cell will be needed to understand the individual role of each receptor type. It would certainly not be surprising if each of these receptor types mediates TGF/3 signals of some sort. Thus, the differences between the TGF/3 species in their interactions with the different receptor types may modulate the TGF/3-induced response of the cell. The existence of at least three different TGF/3 species produced by mammalian cells has led many investigators to evaluate the similarities and differences in their biological activities. The different TGF/3 species often exert biological activities in vitro that are similar, but some major quantitative differences have been recognized. Besides these similarities and differences, which may be due at least in part to differential interactions of the TGF/3 forms with the TGF/3 receptor types, there are some other major differences among the three different mammalian TGF/3 proteins. Histological localizations of the sites of synthesis of TGF/3 have indicated major differences in the spatial and developmental expression patterns of TGF/31, -/32, and -/33 (53-56) and, thus, suggest different functions in development. Several other recent studies have indicated that expression of the different TGF/3 mRNA molecules is differentially modulated (e.g. by retinoic acid-induced differentiation and estrogen) (57, 58). Finally, it is also pos-

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 00:43 For personal use only. No other uses without permission. . All rights reserved.

1893

Binding of TGF/31, -/32, and -/33

sible that the proteolytic activation of latent TGF/3 complexes may differ considerably between the different TGF/3 forms. It has been demonstrated that in the case of TGF/31, the amino-terminal segment of proTGF/31 remains noncovalently associated with the mature peptide, resulting in a latent or inactive TGF/31 complex (59, 60) and that the latent TGF/3 complex can be activated under normal physiological conditions by plasmin (60-63). Considering the secretion of all three TGF/3 forms as latent complexes and the structural differences among the three TGF/3 precursor segments, it is conceivable that the proteolytic activation of the latent TGF/3 complexes in vivo is quite distinct in nature and may be subject to different types of regulation. Thus, the different TGF/3 forms may have very distinct functions in vivo as a result of differential regulation of transcription and translation, different spatial and temporal localization of their expression, differences in activation of the latent complexes, and, as illustrated in this report, different interactions with the receptor types.

MATERIALS AND METHODS Cell Lines and Culture The mouse embryo fibroblast cell line AKR-2B (clone 84A) was used for biological assays, which included TGF/3 RRAs and soft agar growth studies. Growth inhibition was analyzed using the mink lung epithelial cell line CCL-64. Affinity cross-linking of cell surface receptors was performed using a mouse fibroblast cell line, AKR-2B (clone 84A), and a mouse epithelial cell line, BALB/MK. AKR-2B and CCL-64 cell cultures were maintained in McCoy's 5a medium supplemented with 5% fetal calf serum (FCS). BALB/MK cells, which are normal mouse skin keratinocytes, were cultured as previously described (49). Cells were trypsinized briefly and seeded in plastic tissue culture dishes, as described below for particular studies. Radioiodination of TGF/J Purified porcine TGF/31 and porcine TGF/32 were obtained from R&D Systems (Minneapolis, MN). Recombinant TGF03 was produced as previously described (21). All TGF/3 proteins had essentially the same biological activities, with only minor differences in potency, as previously reported (21). Both natural and recombinant proteins were iodinated identically using a modified chloramine-T method described by Frolik et al. (33), which was slightly altered. Briefly, 1 ng lyophilized protein was hydrated in 10 M' 30% acetonitrile with 0.1% trifluoroacetic acid and vortexed vigorously for several hours before use. The pH was adjusted by the addition of 5 n\ 0.15 M NaPOv Na125l (0.5 mCi; IMS 30, Amersham, Arlington Heights, IL) was added to the solubilized protein, followed by chloramine-T. The chloramine-T (0.01 mg/ml for TGF/31; 0.1 mg/ml for TGF/32 and /33) was added sequentially in three 5-^1 aliquots and incubated for 2,1.5, and 1 min. The reaction was stopped after the final chloramine-T addition with 20 n\ 20 mM A/-acetyl-tyrosine, followed by 200 n\ 60 mM p\ and 200 fi\ 8 M urea-1 M acetic acid. The reaction mixture was incubated for 3-5 min at 22 C and then fractionated on a prepacked G-25 column (Pharmacia, Piscataway, NJ) that had been prepared by washing with 5 column vol elution buffer (4 mM HCI and 100 mM NaCI, pH 7.4, with 0.1 mg/ml BSA). Fractions (500 M') were collected, and radioactivity was measured in a radioisotope calibrator (Capintech, Ramsey, NJ). Radiolabeled TGF/3 eluted in the

void volume, and three or four fractions were pooled. Iodinated TGF/3 was diluted to a final concentration of 0.25 ng/10 n\ based on 25% protein recovery, as determined from mock iodinations. The specific activities of the proteins were: TGF/31, 75-125 Ci/g; TGF/32, 200-250 Ci/g; and TGF/33, 80-100 Ci/ g. All radioiodinated proteins were aliquoted and stored at - 2 0 C for no longer than 30 days before use. Stimulation of Growth in Soft Agar AKR-2B (clone 84A) cells were seeded at a density of 7.5 x 103 in 1 ml McCoy's 5a medium containing 10% FCS and 0.4% agar in 35-mm plastic tissue culture dishes. This upper layer containing cells was placed on a lower layer of 1 ml McCoy's 5a medium supplemented with 10% FCS and 0.8% agar. Separate additions of unlabeled TGF/31, TGF/32, or TGF/33 or radioiodinated TGF/31, TGF/32, or TGF/33 were made to the upper cell-containing layer before overlaying. Colonies larger than 50 ^m were counted after 10 days of culture, using an Omnicon image analyzer (Rochester, NY). RRA AKR-2B cells were seeded in six-well plastic tissue culture dishes at a density of 2 x 105 cells/well in McCoy's 5a medium containing 5% FCS. After 24 h of incubation, the cells were washed two times in PBS, pH 7.4, containing 0.1% BSA. One milliliter of binding buffer (128 mM NaCI, 5 mM KCI, 1.2 mM CaCI2, 5 mM MgSO4, and 50 mM HEPES, pH 7.4) with 2 mg/ ml BSA was added to each well, and cells were incubated with continual rocking for 1 h at 22 C. After incubation, the buffer was replaced with 1 ml fresh binding buffer and radioiodinated TGF/31, TGF/32, or TGF/33 (0.25 ng/ml). Nonspecific binding was determined by the addition of 30 ng unlabeled TGF/3 corresponding to the labeled molecule being studied. Incubation was continued for 2 h at 22 C. The cells were then washed three times in PBS with 0.1% BSA. After the final wash, 1 ml PBS with 1 % Triton X-100 was added to each well. After a 10-min incubation at 22 C, the lysed cells were collected, and membrane-associated radioactivity was measured using a Beckman 5500 7-counter (Palo Alto, CA). Inhibition of Monolayer Growth Mink lung epithelial cells (CCL-64) were plated in six-well plastic tissue culture dishes at a density of 1 x 105 cells/well in McCoy's 5a medium containing 5% FCS. The medium was replaced with fresh medium 24 h later, at which time separate additions of either unlabeled TGF/31, TGF/32, or TGF/33 or radioiodinated TGF/31, TGF/32, or TGF/33 were made at the indicated concentrations. Cell number per well was determined after 3 and 5 days of growth, using a hemocytometer. Affinity Cross-Linking Fibroblast (AKR-2B) and epithelial (BALB/MK) cell lines were plated in 100-mm plastic tissue culture dishes and maintained in the appropriate growth medium until confluent. Cells were chilled on ice briefly and rinsed once with 5 ml ice-cold binding buffer. Cells were then incubated in 5 ml binding buffer for 1 h at 4 C with continual rocking. The binding buffer was then removed and replaced with 2 ml fresh binding buffer and iodinated TGF/3 (1 ng/ml). The affinity of each radiolabeled protein was determined using increasing concentrations of unlabeled proteins (0.03-100 ng/ml). Incubation was continued for 4 h at 4 C on a platform rocker. Cells were then washed once in BSA-free binding buffer, and 2 ml BSA-free binding buffer were added to each plate. Disuccinimidyl suberate (Pierce, Rockford, IL) solubilized in dimethylsulfoxide was added to each plate at a final concentration of 100 ^M . Incubation continued on ice for 15 min, at which time the cells were washed rapidly in 5 ml detachment buffer (0.25 M SU-

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 00:43 For personal use only. No other uses without permission. . All rights reserved.

Vol5No. 12

MOL ENDO-1991 1894

crose, 10 rriM Tris, 1 ITIM EDTA, and 1 ITIM phenylmethylsulfonylfluoride, pH 7.4). The cells were scraped from the dishes using a rubber policemen in 750 /J detachment buffer, followed by an additional 500 n\ detachment buffer. The cells were centrifuged for 15 min at 12,000 x g, the supernatant was discarded, and 50 fi\ solubilization buffer (10 ITIM Tris and 1 mM EDTA, pH 7.4, with 1% Triton) were added to each pellet. Protease inhibitors (1 mM phenylmethylsulfonylfluoride, 1 fig/ ml leupeptin, 0.55 U/ml aprotinin, 0.2 fig/m\ pepstatin, and 0.5 Mg/ml bestatin) were added at this time to prevent degradation of affinity cross-linked complexes. Cell pellets were vortexed vigorously and rocked at 4 C for 30-40 min. Insoluble material was pelleted at 12,000 x g for 15 min, and the supernatant was transferred to a clean microfuge tube. Laemlli sample buffer (40 /il) was added to each sample and boiled for 5 min before electrophoresis. Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis Affinity-cross-linked samples were analyzed on discontinuous polyacrylamide gels. 14C-Labeled mol wt protein markers (Amersham) were also used. Linear gradient gels (5-7.5%) were used in combination with a 4% stacking gel. Samples were electrophoresed at 30 V for 16 h, followed by 2-3 h at 30 mamp/gel constant current. Gels were then fixed in 50% methanol-10% acetic acid for 1 h, rinsed two times in water, and soaked in Amplify (Amersham) for 1-2 h. Gels were dried and analyzed by autoradiography after 3-10 days of exposure to x-ray film.

6.

7. 8.

9.

10.

11.

12.

Acknowledgments

13.

The authors wish to thank Nancy Thomas for excellent photographic assistance.

14.

Received May 30, 1991. Revision received August 17, 1991. Accepted September 6,1991. Address requests for reprints to: Dr. Rik Derynck, Departments of Growth, Development, and Anatomy, Program in Cell Biology, University of California, San Francisco, California 94143-0640. This work was supported by Grant CA42572. * Present address: Genetic Therapy, Inc., 19 Firstfield Road, Gaithersburg, Maryland 20878. t Present address: Unigene Laboratories, Inc., 110 Little Falls Road, Fairfield, New Jersey 07004.

15.

16.

17.

18.

REFERENCES

1. Tucker RF, Shipley GD, Moses HL, Holley RW 1984 Growth inhibitor from BSC-1 cells closely related to type /? transforming growth factor. Science 226:705-707 2. Moses HL, Tucker RF, Leof EB, Coffey Jr RJ, Halper J, Shipley GD 1985 Type /? transforming growth factor is a growth stimulator and a growth inhibitor. In: Feramisco J, Ozanne B, Stiles C (eds) Cancer Cells 3. Cold Spring Harbor Laboratory, Cold Spring Harbor, pp 67-71 3. Lyons RM, Moses HL 1990 Transforming growth factors and the regulation of cell proliferation. Eur J Biochem 187:467-473 4. Roberts AB, Anzano MA, Lamb LC, Smith JM, Sporn MB 1981 New class of transforming growth factors potentiated by epidermal growth factor: isolation from nonneoplastic tissues. Proc Natl Acad Sci USA 78:53395343 5. Moses HL, Branum EB, Proper JA, Robinson RA 1981

19.

20.

21.

22.

Transforming growth factor production by chemically transformed cells. Cancer Res 41:2842-2848 Shipley GD, Tucker RF, Moses HL 1985 Type /? transforming growth factor/growth inhibitor stimulates entry of monolayer cultures of AKR-2B cells into S phase after a prolonged prereplicative interval. Proc Natl Acad Sci USA 82:4147-4151 Ignotz RA, Endo R, Massague J 1987 Regulation of fibronectin and type I collagen mRNA levels by transforming growth factor-/? action. J Biol Chem 262:6443-6446 Raghow R, Postlethwaite AE, Keski-Oja J, Moses HL, Kang AH 1987 Transforming growth factor-/? increases steady state levels of type I procollagen and fibronectin messenger IRMA posttranscriptionally in cultured human dermal fibroblasts. J Clin Invest 79:1285-1288 Postlethwaite AE, Keski-Oja J, Moses HL, Kang AH 1987 Stimulation of the chemotactic migration of human fibroblasts by transforming growth factor-/?. J Exp Med 165:251-256 Wahl SM, Hunt DA, Wakefield LM, McCartney-Francis M, Wahl LM, Roberts AB, Sporn MB 1987 Transforming growth factor-/3 (TGF-/?) induces monocyte chemotaxis and growth factor production. Proc Natl Acad Sci USA 84:5788-5792 Ignotz RA, Massague J 1985 Transforming growth factor/? stimulates the expression of fibronectin and collagen and their incorporation into the extracellular matrix. Proc Natl Acad Sci 82:8530-8534 Massague J, Cheifetz S, Endo T, Nadal-Ginard B 1986 Type 0 transforming growth factor is an inhibitor of myogenic differentiation. Proc Natl Acad Sci USA 83:82068210 Moses HL, Leof EB 1986 Transforming growth factor-/?. In: Kahn P, Graf T (eds) Oncogenes and Growth Control. Springer-Verlag, Berlin, pp 51-57 Cheifetz S, Weatherbee JA, Tsang ML-S, Anderson JK, Mole JE, Lucas R, Massague J 1987 The transforming growth factor-/? system, a complex pattern of crossreactive ligands and receptors. Cell 48:409-415 Seyedin SM, Thompson AY, Bentz H, Rosen DM, McPherson JM, Conti A, Siegel NR, Galluppi GR, Piez KA 1986 Cartilage-inducing factor A: apparent identity to transforming growth factor-/?. J Biol Chem 261:56935695 Derynck R, Lindquist PB, Lee A, Wen D, Tamm J, Graycar JL, Rhee L, Mason AJ, Miller DA, Coffey Jr RJ, Moses HL, Chen EY 1988 A new type of transforming growth factor-0, TGF-/?3. EMBO J 7:3737-3743 ten Dijke P, Hansen P, Iwata KK, Pieler C, Foulkes JG 1988 Identification of another member of the transforming growth factor-/? gene family. Proc Natl Acad Sci USA 85:4715-4719 Jakowlew SB, Kondaiah P, Dillard PJ, Sporn MB, Roberts AB 1988 Complementary deoxyribonucleic acid cloning of a novel transforming growth factor-/? messenger ribonucleic acid from chick embryo chondrocytes. Mol Endocrinol 2:747-755 Jakowlew SB, Dillard PJ, Sporn MB, Roberts AB 1988 Complementary ribonucleic acid cloning of a messenger ribonucleic acid encoding transforming growth factor-/?4 from chick embryo chondrocytes. Mol Endocrinol 2:11861195 Kondaiah P, Sands MJ, Smith JM, Fields A, Roberts AB, Sporn MB, Melton DA 1990 Identification of a novel transforming growth factor-/? (TGF-/35) mRNA in Xenopus laevis. J Biol Chem 265:1089-1093 Graycar JL, Miller DA, Arrick BA, Lyons RM, Moses HL, Derynck R 1989 Human transforming growth factor-/?3: recombinant expression, purification and biological activities in comparison with transforming growth factors-/?1 and -(82. Mol Endocrinol 3:1977-1986 ten Dijke P, Iwata KK, Goddard C, Pieler C, Canalis E, McCarthy TL, Centrella M 1990 Recombinant transforming growth factor type /?3: biological activities and recep-

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 00:43 For personal use only. No other uses without permission. . All rights reserved.

Binding of TGF01, -02, and -03

23. 24.

25.

26.

27.

28.

29.

30. 31.

32.

33.

34.

35.

36.

37.

38.

39.

tor-binding properties in isolated bone cells. Mol Cell Biol 10:4473-4479 Weeks DL, Melton DA 1987 A maternal mRNA localized to the vegetal hemisphere in Xenopus eggs codes for a growth factor related to TGF-0. Cell 51:861-867 Padgett RW, St Johnston RD, Gelbart WM 1987 A transcript from a Drosophila pattern gene predicts a protein homologous to the transforming growth factor-/} gene family. Nature 325:81-84 Mason AJ, Hayflick JS, Ling N, Esch F, Ueno N, Ying Y, Guillemin R, Niall H, Seeburg PH 1985 Complimentary DNA sequences of ovarian follicular fluid inhibin show precursor structure and homology with transforming growth factor-/?. Nature 318:659-663 Cate RL, Mattaliano RJ, Hession C, Tizard R, Farber NM, Cheung A, Ninfa EG, Frey AZ, Gash DJ, Chow EP, Fisher RA, Bertonis JM, Torres G, Wallner BP, Ramachandran KL, Ragin RC, Managanaro TF, MacLauchlin DT, Donahoe PK 1986 Isolation of the bovine and human genes for Mullerian inhibiting substance and expression of the human gene in animal cells. Cell 45:685-698 Wozney JM, Rosen V, Celeste AJ, Mitsock LM, Whitter MJ, Kriz RW, Hewick RM, Wang EA 1988 Novel regulators of bone formation: molecular clones and activities. Science 242:1528-1534 Lyons K, Graycar JL, Lee A, Hashmi S, Lindquist P, Chen EY, Hogan BLM, Derynck R 1989 Vgr-1, a mammalian gene related to Xenopus vg-1 and a new member of the TGF-0 superfamily. Proc Natl Acad Sci USA 86:45544558 Ozkaynak E, Rueger DC, Drier EA, Corbett C, Ridge RJ, Sampath TK, Oppermann H 1990 OP-I cDNA encodes an osteogenic protein in the TGF-0 family. EMBO J 9:20852093 Lee S-J 1990 Identification of a novel member of the transforming growth factor-0 superfamily. Mol Endocrinol 4:1034-1040 Celeste AJ, lannazzi JA, Taylor RC, Hewick RM, Rosen V, Wang EA, Wozney JM 1990 Identification of transforming growth factor 0 family members present in boneinductive protein purified from bovine bone. Proc Natl Acad Sci USA 87:9843-9847 Tucker RF, Branum EL, Shipley GD, Ryan RJ, Moses HL 1984 Specific binding to cultured cells of 125l-labeled type 0 transforming growth factor from human platelets. Proc Natl Acad Sci USA 81:6757-6761 Frolik DA, Wakefield LM, Smith DM, Sporn MB 1984 Characterization of a membrane receptor for transforming growth factor-0 in normal rat kidney fibroblasts. J Biol Chem 259:10995-11000 Massague J, Like B 1984 Cellular receptors for type 0 transforming growth factor. Ligand binding and affinity labeling in human and rodent cell lines. J Biol Chem 259:2636-2645 Segarini PR, Seyedin SM 1988 The high molecular weight receptor to transforming growth factor 0 contains glycosaminoglycan chains. J Biol Chem 263:8366-8370 Segarini PR, Roberts AB, Rosen DM, Seyedin SM 1987 Membrane binding characteristics of two forms of transforming growth factor-0. J Biol Chem 262:14655-14662 Cheifetz S, Bassols A, Stanley K, Ohta M, Greenberger J, Massague J 1988 Heterodimeric transforming growth factor-/?. Biological properties and interaction with three types of cell surface receptors. J Biol Chem 263:1078310789 Cheifetz S, Hernandez H, Laiho M, ten Dijke P, Iwata K, Massague J 1990 Distinct transforming growth factor-/? (TGF-0) receptor subsets as determinants of cellular responsiveness to three TGF-0 isoforms. J Biol Chem 265:20533-20538 Mitchell EJ, O'Connor-McCourt MD 1991 A transforming growth factor 0 (TGF-j8) receptor from human placenta exhibits a greater affinity for TGF-02 than for TGF-01. Biochemistry 30:4350-4356

1895

40. Boyd FT, Massague J 1989 Transforming growth factor 0 inhibition of epithelial cell proliferation linked to the expression of a 53 kDa membrane receptor. J Biol Chem 264:2272-2278 41. Segarini PR, Rosen DM, Seyedin SM 1989 Binding of transforming growth factor-0 to cell surface receptors varies with cell type. Mol Endocrinol 3:261 -272 42. Murthy US, Anzano MA, Stadel JM, Greig R 1988 Coupling of TGF-0 induced mitogenesis to G-protein activation in AKR-2B cells. Biochem Biophys Res Commun 152:1228-1235 43. Howe PH, Leof EB 1989 Transforming growth factor-01 treatment of AKR-2B cells is coupled through a pertussistoxin-sensitive G protein(s). Biochem J 261:879-886 44. Van Obberghen-Schilling E, Roche NS, Flanders KC, Sporn MB, Roberts AB 1988 Transforming growth factorial positively regulates its own expression in normal and transformed cells. J Biol Chem 263:7741-7746 45. Bascom Jr CC, Wolfshol J, Coffey Jr RJ, Madisen L, Webb NR, Purchio AR, Derynck R, Moses HL 1989 Complex regulation of transforming growth factor 01, 02 and -03 mRNA expression in mouse fibroblasts and keratinocytes by transforming growth factors 01 and 02. Mol Cell Biol 9:5508-5515 46. Heimark RL, Twardzik DR, Schwartz SM 1986 Inhibition of endothelial regeneration by type 0 transforming growth factor from platelets. Science 233:1078-1080 47. Jennings JC, Mohan S, Linkhart TA, Widstorm M, Baylink DJ 1988 Comparison of the biological activities of TGFbeta 1 and beta 2: differential activity in endothelial cells. J Cell Physiol 137:167-172 48. Rosa F, Roberts AB, Danielpour D, Dart LL, Sporn MB, David IB 1988. Mesoderm induction in amphibians: the role of TGF-02-like factors. Science 239:783-785 49. Weissmann BE, Aaronson SA 1983 BALB and Kirsten murine sarcoma viruses alter growth and differentiation of EGF-dependent BALB/c mouse epidermal keratinocyte lines. Cell 32:599-606 50. Coffey Jr RJ, Sipes NJ, Bascom CC, Graves-Deal R, Pennington CY, Weissman BE, Moses HL 1988 Growth modulation of mouse keratinocytes by transforming growth factors. Cancer Res 48:1596-1602 51. Ebner R, Derynck R 1991 Epidermal growth factor and transforming growth factor-a: differential intracellular routing and processing of ligand-receptor complexes. Cell Regul 2:599-612 52. Ohta MS, Greenberg JS, Anklesaria P, Bassols A, Massague J 1987 Two forms of transforming growth factor0 distinguished by multipotential heatopoietic progenitor cells. Nature 329;539-541 53. Heine Ul, Munoz EF, Flanders KC, Ellingsworth LR, Lam H-YP, Thompson NL, Roberts AB, Sporn MB 1987 Role of transforming growth factor-0 in the development of the mouse embryo. J Cell Biol 105:2861-2876 54. Lehnert SA, Akhurst RJ 1988 Embryonic expression pattern of transforming growth factor 0 type IIRMA suggests both paracrine and autocrine mechanisms of action. Development 104:263-273 55. Wilcox JN, Derynck R 1988 Developmental expression of transforming growth factors alpha and beta in mouse fetus. Mol Cell Biol 8:3415-3422 56. Pelton RW, Nomura A, Moses HL, Hogan BLM 1989 Expression of transforming growth factor-02 IRMA during murine embryogenesis. Development 106:759-768 57. Glick AB, Flanders KC, Danielpour D, Yuspa SH, Sporn MB 1989 Retinoic acid induces transforming growth factor-02 in cultured keratinocytes and mouse epidermis. Cell Regul 1:87-97 58. Arrick BA, Korc M, Derynck R 1990 Differential regulation of expression of three transforming growth factor 0 species in human breast cancer cell lines by estradiol. Cancer Res 50:299-303 59. Gentry LE, Webb NR, Lim GJ, Brunner AM, Ranchalis JE, Twardzik DR, Lioubin MN, Marquardt H, Purchio AF1987

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 00:43 For personal use only. No other uses without permission. . All rights reserved.

MOL ENDO-1991 1896

Molecular events in the processing of recombinant preprotransforming growth factor beta to the mature polypeptide. Mol Cell Biol 8:4162-4168 60. Lyons RM, Gentry LE, Purchio AE, Moses HL 1990 Mechanism of activation of latent transforming growth factor-/31 by plasmin. J Cell Biol 110:1361-1367 61. Lyons RM, Keski-Oja J, Moses HL 1988 Proteolytic activation of latent transforming growth factor-/? from fibroblast conditioned medium. J Cell Biol 106:1659-1665

Vol5No. 12

62. Sato Y, Rifkin DB 1989 Induction of endothelial cell movement by pericytes and smooth muscle cells: activation of a latent transforming growth factor /31-like molecule by plasmin during co-culture. J Cell Biol 109:309-315 63. Sato Y, Tsuboi R, Lyons RM, Moses HL, Rifkin DB 1990 Characterization of the activation of latent TGF-/? by cultures of endothelial cells and pericytes or smooth muscle cells: a self-regulating system. J Cell Biol 111:757-763

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 00:43 For personal use only. No other uses without permission. . All rights reserved.