The Early History of TGF-beta

by Michael B. Sporn
Department of Pharmacology
Dartmouth Medical School

This brief review is a retrospective on the original discovery of TGF-ß, and the development of the field of TGF-ß research in its earliest years, before 1990. Although such a commentary might be ancient history to some, perhaps a revisiting of the context of the original discovery of the molecule and of some of the assumptions and conclusions, both correct and incorrect, can still provide useful insights for the contemporary investigator at the laboratory bench.

In the late 1970s, it had become apparent that the growth of normal cells is largely controlled by the interplay between several polypeptide hormones and hormone-like growth factors that are present in tissue fluids. Many new polypeptide growth factors had recently been identified in blood, serum, tissue fluids, and cellular extracts. Moreover, it was known that malignant cells were not subject to all the same growth controls as were normal cells. In general, it had been shown that malignant cells required less of these exogenous growth factors than did their normal counterparts for optimal growth and multiplication, and the Nobel laureate, Robert Holley, had suggested that "transformed or malignant cells escape from normal growth controls by requiring less of such hormones or growth factors."1

Figure 1. Original description of autocrine secretion.2 Regulatory chemical messengers appear in latent form within the cell. They are secreted and may act on a remote cell by entering the circulation (endocrine), on a neighboring cell (paracrine), or on the same cell (autocrine).

It was in that setting that a new hypothesis of "autocrine secretion" was formulated in 1980.2 This hypothesis suggested a new explanation for the mechanism whereby a polypeptide growth factor might cause the malignant transformation of cells. The hypothesis suggested the following novel properties, namely: 1) the transforming polypeptide should be produced by the putative transformed cell itself, and 2) the putative transformed cell should have its own functional cellular receptors for this polypeptide, allowing phenotypic expression of the pep-tide by the same cell that produced it (Figure 1). With this model of autocrine secretion, the classic "lesser requirement of malignant cells for exogenous growth factors" could be simply explained: the endogenous production of growth-promoting polypeptides by a transformed cell would lessen its own requirement for an exogenous supply of similar growth factors.

Shortly before the autocrine hypothesis was formulated, Todaro and De Larco3 had made the important discovery that virally transformed cells produced a "factor," which would cause phenotypic transformation of a "normal" reader cell, such as a normal rat kidney (NRK) fibroblast. They called this factor "sarcoma growth factor" (SGF) and they assayed its transforming activity by the ability of cells to grow in soft agar. The soft agar assay was considered to be important, because it was generally believed that growth in soft agar was a property of malignant cells that was not shared by normal ones. My own laboratory, at the time, was interested in mechanisms whereby retinoic acid and other retinoids might influence malignant transformation, and we set out to characterize SGF. In keeping with the autocrine hypothesis, it had been postulated that SGF might be an autocrine factor, and so our initial experiments, designed to purify SGF, were conducted on acid-ethanol extracts of murine sarcoma cells transformed by Moloney sarcoma virus.

At this point an important discovery was made in our laboratory by Anita Roberts and her colleagues,4 namely that essentially all the "transforming" activity of crude preparations of SGF (as measured by enhancement of the growth of NRK cells in soft agar) was lost when SGF extracts were subjected to the gentle techniques of gel filtration analysis on columns. However, all the transforming activity could be recovered when two separate fractions from the columns, both inactive individually, were reconstituted. Thus, it was shown for the first time that SGF was not a single substance, but rather a mixture of at least two constituents. One of the two fractions induced formation of only a few small colonies of NRK cells in soft agar, and was named "transforming growth factor-a" (TGF-a).5 It competed strongly with radio-iodine-labeled epidermal growth factor (EGF) in a receptor binding assay, and when later purified to homogeneity was found to have strong sequence similarity to EGF itself. Indeed, TGF-a is probably the embryonic form of the EGF that is found in the adult salivary gland. It was later found that, in the classic in vivo assay for accelerating newborn mouse eyelid opening (the assay used by Stanley Cohen to monitor the original purification of EGF), recombinant TGF-a has potency equivalent to that of EGF.6

However, the second fraction derived from gel filtration analysis of SGF showed no competition with EGF in a receptor binding assay, but had the remarkable property of inducing the growth of many very large colonies of NRK cells in the soft agar assay but only if small amounts of EGF were present as well. The protein in this second fraction was named "transforming growth factor-ß," (TGF-ß)5 and a major effort was undertaken in our laboratory to purify this material to homogeneity in the early 1980s. By this time, it was already clear that the assumption that TGFs were cancer-specific was erroneous, since when acid-ethanol extracts were made of numerous normal tissues, including liver, heart, and kidney, one could obtain peptides whose activity profile resembled that of TGF-ß derived from sarcoma cells.7 Since it was desired to have very large amounts of initial starting material for purification to homogeneity, sarcoma cells were then abandoned as a primary source, and the further purification of TGF-ß was pursued with acid-ethanol extracts from normal tissues that were readily available. Two human sources were selected, namely placenta8 and blood platelets,9 and since it was desired to see if similar proteins were found in a different species, purification from bovine kidney10 was also undertaken. The choice of platelets was particularly fortuitous, although this was not pure serendipity. The presence of the growth-promoting activity we have just described, in so many normal tissues, had suggested that the putative TGF-ß had a normal function, which was hypothesized to be that of an agent involved in tissue repair. Indeed, it had long been hypothesized that there was a functional connection between tumors and the process of wound healing, as manifested in Haddow's famous dictum that "the wound is a tumor that heals itself"11 (later inverted by Dvorak to "tumors are wounds that do not heal"12).

Figure 2. System originally used for first demonstration that TGF-ß could be used in vivo for enhancement of wound healing.13 Wound chambers made of stainless steel wire mesh were implanted in the backs of rats, and preparations of TGF-ß were injected into chambers A, B, and C, while chambers D, E, and F served for control injections. Five days after initiation of treatment, contents of the chambers were evaluated for collagen, and by histological analysis.

Eventually, the original hypothesis that one of the functions of TGF-ß in normal tissues was to participate in wound healing was confirmed with in vivo studies, which showed that administration of exogenous TGF-ß could stimulate the formation of collagen and the vascularized connective tissue found in healing wounds13,14 (Figure 2). Moreover, it had been shown that extracts of blood platelets had the ability to promote growth of cells other than NRK fibroblasts in soft agar,15 so the purification and characterization of TGF-ß from platelets was pursued with intensive effort.

The three tissue sources just described, after further purification (extensive in the case of placenta and kidney, and relatively easy, with hindsight, from platelets), all eventually yielded homogeneous preparations of TGF-ß, as assessed by the demonstration of a single band on silver-stained gels. The apparent molecular weight of this band from all three sources was identical (approximately 25 kDa), and moreover, all three preparations yielded a single band of half that molecular weight upon reduction with mercaptoethanol.8-10 The preparation from human platelets described by Assoian was particularly elegant, since it involved only 2 steps after acid-ethanol extraction, namely sequential gel filtration on Bio-Gel® P-60 columns, first in the absence and then in the presence of urea.9 Although skeptics had rumored that this preparation of TGF-ß from platelets was, in reality, nothing more than platelet-derived growth factor (PDGF), this clearly was not the case, since its amino acid composition was distinctly different than that of PDGF. Most importantly, the demonstration that TGF-ß could be purified relatively easily from platelets then set the stage for R&D Systems to undertake the first commercial preparations of TGF-ß. Using techniques adapted from the Assoian procedure, R&D Systems set out to purify TGF-ß from porcine platelets. Porcine platelets were found to contain a second form of TGF-ß, known as TGF-ß2, which was not found in human platelets.

Once homogeneous preparations of TGF-ß were available, many advances ensued. Most notably, in 1984, Moses and colleagues made the important discovery that TGF-ß could also inhibit cell growth,16 if an appropriate reader cell, such as a mink lung epithelial cell (CCL-64), was used. Moreover, the first receptor binding assay was published in the same year.17,18 Another major advance came when TGF-ß1 was cloned in 1985, by Derynck and colleagues at Genentech.19 Even further complexity developed in 1985, when it was shown that TGF-ß could be multifunctional in the very same cells in which it was assayed, depending on the context of the assay.20 Thus, in the presence of PDGF, TGF-ß caused stimulation of growth of reader cells, while in the presence of EGF, TGF-ß functioned as a growth inhibitor (Figure 3). TGF-ß research was changed dramatically by these demonstrations of multifunctionality. The simplistic notion of TGF-ß as an agent merely promoting cell growth was no longer tenable.

Figure 3. Original demonstration of the bifunctional activity of TGF-ß.20 Effects of TGF-ß on the anchorage-independent growth of Myc-1 cells. In the presence of PDGF, TGF-ß stimulates growth (A) and in the presence of EGF, TGF-ß inhibits growth (B).

Studies on the role of TGF-ß as a regulator of the immune and inflammatory processes rapidly led to a much more sophisticated appreciation of its role as a multifunctional regulatory molecule. Although the first descriptions of the activity of TGF-ß on both T and B lymphocytes emphasized its inhibitory effects,21,22 it was soon found that TGF-ß could act as both a stimulator and inhibitor of IgA production in B lymphocytes.23 At the same time both activating and deactivating effects of TGF-ß on macrophages were also found.24,25 At present, we have a much more sophisticated understanding of the multifunctional actions of TGF-ß in both inflammation and the immune response, summarized as "TGF-ß: the good, the bad, and the ugly."26 It is clear that to understand all these various activities, we must relate them to the context in which cells and tissues find themselves. Ultimately, in the world of cellular physiology, TGF-ß should not be regarded as a "thing," but rather as an element of a complex biological signaling language, which is used for both intercellular and intracellular communication.27

Like a symbol or a letter of an alphabet in a code or a language, the meaning of TGF-ß always needs to be considered in the context of all the other signals present. We have suggested that TGF-ß should be regarded as a cellular switch, and that its true function is to provide a mechanism for coupling a cell to its environment, so that the cell has the necessary plasticity to respond appropriately to changes in its environment, or even within its own state. One may summarize this concept of TGF-ß as a switch, as follows: if something in the cell is "off," TGF-ß may turn it "on," while if something in the cell is "on," TGF-ß may turn it "off."27 Furthermore, the discovery of high concentrations of TGF-ß within mitochondria28 of several types of cells suggested that TGF-ß may even act as a linker between the energetics of the cell and the cell's other activities. Recent studies on the regulation of the structure and function of the mitochondrion by TGF-ß appear to confirm the original suggestions (A. Roberts, personal communication).

Yet another major advance was the original development of epitope-specific antibodies by Flanders,29 which opened the way to the first studies on the role of TGF-ß in many conditions in vivo, but particularly as a regulatory molecule during normal embryogenesis,30,31 with no reference whatsoever to the phenomenon of "malignant" transformation. This is now a field of immense interest.

From these early beginnings, the field of TGF-ß research has increased dramatically, so that there are now thousands of publications on this topic. The cloning and characterization of the three receptors for TGF-ß, the development of the entire field of Smad signaling, and the present attempts to develop specific antagonists of TGF-ß (either soluble receptors or inhibitors of the signal transduction pathway) represent major fields of current interest. There are many possibilities for drugs that either enhance or suppress the expression or activity of TGF-ß in clinical medicine, in areas as diverse as cancer, cardiovascular disease, neurological disease, inflammatory bowel disease, and diseases of connective tissue. In many of these areas, the ability of TGF-ß to modulate the activity of the immune system is of central importance.

Perhaps the biggest lesson that we can learn from this historical retrospective is that it is essential to be flexible in one's scientific outlook. Many of the original hypotheses and assumptions about the nature of TGF-ß have turned out to be erroneous. Nevertheless, these hypotheses have led to important experimental work, which continually generates new data and modifies our understanding of this very complex, multifunctional molecule, particularly its impact on the regulatory mechanisms of cells and tissues. The next big advance should come at the level of the whole organism, in which enhancement or suppression of the activity of TGF-ß will be used for human benefit.


  1. Holley, R. (1975) Nature 258:487.
  2. Sporn, M. & G. Todaro (1980) N. Engl. J. Med.303:878.
  3. De Larco, J. & G. Todaro (1978) Proc. Natl. Acad. Sci. USA 75:4001.
  4. Roberts, A. et al. (1981) Proc. Natl. Acad. Sci. USA 78:5339.
  5. Anzano, M. et al. (1982) Cancer Res. 42:4776.
  6. Smith, J. et al. (1985) Nature 315:515.
  7. Roberts, A. et al. (1983) Fed. Proc. 42:2621.
  8. Frolik, C. et al. (1983) Proc. Natl. Acad. Sci. USA 80:3676.
  9. Assoian, R. et al. (1983) J. Biol. Chem. 258:7155.
  10. Roberts, A. et al. (1983) Biochemistry 22:5692.
  11. Haddow, A. (1972) Adv. Cancer Res. 16:181.
  12. Dvorak, H. (1986) N. Engl. J. Med. 315:1600.
  13. Sporn, M. et al. (1983) Science 219:1329.
  14. Roberts, A. et al. (1986) Proc. Natl. Acad. Sci. USA 83:4167.
  15. Childs, C. et al. (1982) Proc. Natl. Acad. Sci. USA 79:5312.
  16. Tucker, R. et al. (1984) Science 226:705.
  17. Frolik, C. et al. (1984) J. Biol. Chem. 259:10995.
  18. Tucker, R. et al. (1984) Proc. Natl. Acad. Sci. USA 81:6757.
  19. Derynck, R. et al. (1985) Nature 316:701.
  20. Roberts, A. et al. (1985) Proc. Natl. Acad. Sci. USA 82:119.
  21. Kehrl, J. et al. (1986) J. Exp. Med. 163:1037.
  22. Kehrl, J. et al. (1986) J. Immunol. 137:3855.
  23. Coffman, R. et al. (1989) J. Exp. Med. 170:1039.
  24. Wahl, S. et al. (1987) Proc. Natl. Acad. Sci. USA 84:5788.
  25. Tsunawaki, S. et al. (1988) Nature 334:260.
  26. Wahl, S. (1994). J. Exp. Med. 180:1587.
  27. Sporn, M. & A. Roberts (1990) Cell Regul. 1:875.
  28. Heine, U. et al. (1991) Cell Regul. 2:467.
  29. Flanders, K. et al. (1988) Biochemistry 27:739.
  30. Flanders, K. et al. (1989) J. Cell Biol. 108:653.
  31. Flanders, K. et al. (1991) Development 113:183.
Bio-Gel is a registered trademark of Bio-Rad Laboratories, Inc.