MOLECULAR REPRODUCTION AND DEVELOPMENT 32:145-151 (1992)
Regulation of Mammary Growth and Function by TGF-P CHARLES W. DANIEL AND STEPHEN D. ROBINSON Department of Biology, University of California, Santa Cruz, California
ABSTRACT We have previously shown that TGF-p1 rapidly and reversibly inhibits ductal growth in vivo when administered by miniature slow-release plastic implants. A possible role for endogenous TGF-61 was suggested by the observation that the normal gland displayed substantial, developmentally regulated levels of TGF-pl transcripts and protein. These studies have now been extended to include the other two mammalian TGF-p isoforms. When tested with slow-release plastic implants, TGF-P2 and TGF-p3 also caused disappearance of the proliferating mammary stem cell layer, with rapid involution of ductal end buds and cessation of glandular growth. None of the isoforms was active in inhibiting alveolar morphogenesis. We conclude that under the conditions of these tests, the three mammaliian isoforms are functionally equivalent. However, striking differences in patterns of gene expression and in the distribution of immunoreactive peptides suggest that TGF-p2 was expressed only at low levels, and mainly during pregnancy. TGF-(33 was expressed in ductal stroma and epithelium, and was the only isoform detected in myoepithelial cells. Developing alveolar tissue and its associated ducts displayed striking TGF-p3 gene expression and immunostaining, which were greatly reduced during lactation. We are now investigating the possibility that the observed high levels of TGF-(3 expression in pregnancy, particularly of TGF-(33, and the absence of substantial expression of any isoform during lactation, may indicate a role for the TGF-p in regulating functional differentiation or the onset of milk secretion. 0 1992 Wiley-Liss, Inc.
Key Words: Transforming growth factor-(3, Mammary, Growth regulation
INTRODUCTION In general, we know a great deal more about what stimulates the growth of organs than we know about what regulates their growth. This is particularly true in the case of the mammary gland, in which the systemic mitogens, the mammogenic hormones which drive growth, have been defined and examined by generations of endocrinologists. However, we do not know what processes or what factors limit the final size of the mammary glands, even in the continued presence of these powerful mitogens. Nor do we know what defines the process of pattern formation during ductal morphogenesis or what stabilizes the pattern that is made. We
0 1992 WILEY-LISS, INC.
will discuss these aspects of the mammary gland using the mouse as the experimental animal. Mammary biology is often regarded as a specialized subject, but its branching morphogenesis is reminiscent of developmental patterns in other structures such as the lung, kidney, exocrine organs such as the salivary glands, and various glands of the reproductive tract. Studies of the mammary gland therefore have general applicability with respect to other tissues and to the general process of organogenesis in vertebrates. An important advantage of mammary studies is that most mammary growth and morphogenesis occurs after birth and occurs in an animal that is large enough to work on conveniently and in which the glands are in an accessible location, so that unlike the lung, kidney, or other internal organs, manipulations in vivo can readily be carried out. Growth of the mammary gland begins in utero. After birth the mammary ducts grow slowly, but at the time of puberty and sexual maturation ductal growth begins in earnest. The mammary tree, a system of branching ducts, is created, which then undergoes another and quite different growth process during pregnancy, when fine branching and secretory alveoli are formed. Alveolar development eventually culminates in the formation of secretory lobules that become engorged with milk during lactation. Secretion continues until weaning, when, in the absence of lactogenic hormones, the mass of secretory tissue degenerates and the gland reverts to a pattern similar to that of the virgin. The period of ductal morphogenesis (see Daniel and Silberstein, 1987, for review), the stage of mammary development in which we have been particularly interested, occurs in the mouse a t around 5-8 weeks of age, and is driven by the ovarian and pituitary hormones associated with sexual maturation. Ductal elongation takes place as a result of intense mitotic activity in end buds, the bulbous structures at the tips of ducts. As the end buds penetrate the surrounding adipose stroma, they leave behind a tree-like pattern of differentiated ducts that are mitotically inactive. Consequently, in a growing gland only about 5%of the total tissue is actually engaged in growth, while the remainder of the
Address reprint requests to Dr. Charles W. Daniel, Department of Biology, University of California, Thimann Laboratories, Santa Cruz, CA 95064.
146
C.W. DANIEL AND S.D. ROBINSON
trunk and branches of the ductal tree are in static phase. The mammary ducts are cloaked in a mantle of fibrous tissue, termed the periductal matrix, which consists of type I collagen and a variety of other extracellular matrix (ECM) elements. In contrast, the tips of growth-active buds are not surrounded by matrix and the basal lamina of the epithelial mass may be in direct contact with adipocytes of the stroma (Williams and Daniel, 1983). There is a great difference, both structurally and biologically, between the mature duct and the proliferating buds. There are three types of buds in the mammary gland: the end buds, which are responsible for ductal elongation; the lateral buds, which develop along mature ducts from time to time, grow a short distance, and then stop because of the presence of competing ductal elements; and alveolar buds, which in most mouse strains develop only in response to the hormones of pregnancy. In the virgin gland, branching morphogenesis creates a pattern in which the interductal space measures about .25mm. The adipose tissue in this interductal region is nonpermissive for growth of lateral buds. Our evidence suggests that TGF-(3is a t least in part responsible for restricting the formation and growth of lateral buds, and functions normally to maintain the open pattern of branching that is required for alveolar development during pregnancy. TGF-fi does not, on the other hand, restrict the development of alveolar structures (Daniel et al., 1989). We have investigated the question of what regulates mammary gland patterning. Why do the ductal buds not fill the stroma between ducts-that is, why does the mammary gland look like a tree in winter instead of a plate of spaghetti? One possible explanation is that the ducts are terminally differentiated and no longer have the capacity to form new growth points. That this is not the case was shown several years ago by Faulkin and DeOme (1960).Mammary cells taken from any location in the gland will form new buds when transplanted into a mammary gland-free “cleared fat pad” and regenerate a new gland that is identical in all respects with the parent tissue except that it no longer makes connection with a nipple. By contrast, mammary epithelial cells transplanted into a fat pad containing ducts fail to grow and remain in healthy but static condition for the life of the host. These results indicate that the mammary gland is held in a dynamic balance between proliferation and stasis, pulled toward mitosis by powerful systemic mitogens, but inhibited by the presence of surrounding tissue. We have examined the possibility that TGF-p serves a regulatory role in the mammary gland by asking several questions. Is TGF-P present in the gland and if so, is it present in developmentally interesting places? What are the biological effects of TGF-p? What is the comparative biology of the TGF-(3isoforms?
TGF-P GENE EXPRESSION IN THE MAMMARY GLAND A possible role for endogenous TGF-(31was suggested by the observation that the normal gland displayed
TGFPl
-2.4
-6.0 TGFP2 4-5.0 -4.0 4-3.5
TGFP3
-3.8
18s
Fig. 1. Northern analysis of TGF-p isoform expression during mammary gland development. “Cleared” indicates glands that have been surgically cleared of mammary epithelium. Arrows on right indicate sizes of TGF-P transcripts (four transcripts are characteristically seen for TGF-PX). The lower frame has been probed for 185 ribosomal RNA, as a loading control.
substantial, developmentally regulated levels of TGF-Pl transcripts. These studies have now been extended to include the other two mammalian isoforms, TGF-P2 and TGF-p3 (Robinson et al., 1991). Striking differences in patterns of gene expression indicate a high level of developmental regulation (Fig. 1).TGF-p1 was expressed in all stages except lactation. TGF-pB
REGULATION OF MAMMARY GROWTH AND FUNCTION BY TGF-p
147
Fig. 2. End-bud expression of TGF-PI mRNA. Bright-field and corresponding dark-ground photomicrographs of adjacent sections through a bifurcating end bud. The riboprobe was 35S-UTP-labeled and consisted of a fragment from the coding region for the mature (active form) peptide. Controls were hybridized to probes to irrelevant transcripts (not shown). TGF-61 expression is seen over the end bud epithelium with heaviest labeling at the tips (arrows). (Reproduced from Robinson et al., 1991, with permission of The Company of Biologists Limited.)
was expressed in immature glands only at low levels, and somewhat increased mRNA amounts were found in pregnancy. TGF-P3 was expressed in ductal stroma and epithelium, and its transcripts were strikingly elevated in pregnancy. We are now investigating the possibility that the observed high levels of TGF-@3expression in pregnancy, and to a lesser extent TGF-P1 and -P2, taken together with the absence of substantial expression of any isoform during lactation, may indicate a role for the TGF-ps in regulating functional differentiation or the onset of milk secretion.
DISTRIBUTION OF TGF-@TRANSCRIPTS AND PEPTIDES IN THE MAMMARY GLAND To investigate the question of when and where TGF-P is expressed we studied TGF-P1 transcripts in mammary glands from the 5 to 7-week-old mouse, the mature gland, during pregnancy, and in lactation, using Northern and in situ hybridization to riboprobes (Robinson et al., 1991). Northern analysis showed that isoform transcripts were expressed in a stage- and isoform-specific manner during all developmental stages except lactation, where transcripts abruptly disappeared (Fig. 1).In situ hybridization showed high levels of TGF-P transcripts in mammary epithelial tissues and a lower level of transcripts in fat cells and fibrous stroma of connective tissues (Fig. 2). The in situ results were consistent with data from Northerns. Developing alveolar tissue and its associated ducts displayed strik-
ing TGF-P3 gene expression in mammary epithelium, which was absent during lactation. Using immunostaining for both precursor and active forms of the growth factors, we investigated mammary ducts and spontaneously occurring lateral buds. Mature TGF-@was found in the periductal matrix along the sides of the bud, but not a t the tip of growing buds. The ductal epithelium was rich in intracellular TGF-@, whereas fibroblasts, which synthesize various matrix components, failed to stain for intracellular TGF-P. We conclude that the matrix-associated TGF-P was synthesized and secreted basally by mammary epithelium. It is possible that there is a stromal contribution as well, since adipocytes are rich in TGF-P. Focal depletion at the growth points provides support for the hypothesis that the extracellular matrix acts as a reservoir for growth factors, perhaps bound to specific matrix components, which can be released at morphogenetically active sites (Rouslahti and Yamaguchi, 1991). As a result of these studies we have developed the concept that interductal spacing is maintained by TGF-@accumulation in the periductal matrix, and focal loss of the growth factor is associated with budding and growth.
LOCALLY IMPLANTED TGF-P INHIBITS MAMMARY DUCTAL GROWTH To observe the effect of exogenous TGF-@on ductal growth we implanted slow-release plastic implants con-
148
C.W. DANIEL AND S.D. ROBINSON
Fig. 3. Photomicrograph illustrating the effect of TGF-PI on ductal growth. Five-week-old C57 female mice were treated for 4 days with slow-release plastic implants (star) containing carrier protein (A) or TGF-61 (B). End buds (arrows) have regressed in the treated gland. (Reproduced from Silberstein and Daniel, 1987b, with permission of Science. Copyright 1987 by the AAAS.)
taining TGF-p into specific sites in the gland, where they released the growth factor slowly over a period of time (Silberstein and Daniel, 1982,1987a). Control implants placed in contralateral glands in the same animal served to indicate whether the observed effects were systemic or local. Figure 3 shows the inhibitory effects of TGF-P1 48 h r after implantation; the growth points of the end buds were completely inhibited. Similar results were obtained with all three isoforms (Robinson et al., 1991) causing disappearance of the proliferating mammary stem cell layer, rapid involution of buds, and cessation of glandular growth. None of the isoforms were active in inhibiting alveolar morphogenesis or DNA synthesis, indicating considerable developmental specificity. We conclude that under the conditions of these tests, the three mammalian isoforms are functionally equivalent, a result that is consistent with their observed cross reactivity with receptors (Roberts and Sporn, 1989).
THE NATURE OF DUCTAL GROWTH INHIBITION BY TGF-P To determine whether the effect of TGF-p was reversible and whether the tissue became refractory to the
sustained release of TGF-p, we implanted TGF-p1 in the mammary gland and left the implant until the charge of TGF-pl was exhausted. After a period of about a week the gland resumed its normal growth (Fig. 41, indicating reversibility. We also found that if we maintained a continuous supply of TGF-p1 in the region of the implants, the gland remained chronically suppressed and there was no indication t h a t it became refractory to the continued presence of TGF-6. This result is consistent with the observation that TGF-p receptors are not down-regulated. If we allowed the chronically suppressed gland to exhaust its charge of TGF-p, growth again resumed. Thus, the effect of TGF-8 was fully reversible and continued as long a s the factor was present, a characteristic that is required if TGF-P is to qualify as a chronic inhibitor. We used transplantation procedures to determine whether TGF-p was capable of preventing the formation of buds, as well as inhibiting the growth of those previously formed (Daniel et al., 1989). When the implant contained TGF-p, the formation of new buds in transplanted tissue was suppressed. The DNA synthetic labeling index in buds in control glands was extremely high (about 2 0 3 0 % ) and decreased to basal
REGULATION OF MAMMARY GROWTH AND FUNCTION BY TGF-P 1 9 , IL
I
I
0
4
8
12
16
20
24
28 30
Days
Fig. 4. Effects of multiple TGF-p1 treatment on ductal growth. Treatment-recouery-treatmentcycle (0):A sample of mice in the experimental group were sacrificed at the time points indicated by circles. Open arrows indicate re-implantation. Timing was adjusted so that implants exhausted their TGF-p1 charge, permitting the end buds to become re-established to levels of the control group (shaded band) and indicating reversibility of the inhibitory effect. Prolonged exposure ( 0 ) : Glands were re-implanted at short intervals to maintain relatively constant level of exposure to TGF-p1. Within the time period tested, the gland did not become refractory to inhibition. (Reproduced from Daniel et al., 1989, with permission of Academic Press.)
levels with TGF-p treatment. The basal level of DNA synthesis in mammary ducts was low ( 2 4 % ) but far from negligible, which we interpret as a maintenance level reflecting normal cell loss and replacement. The lack of a TGF-p effect on this maintenance DNA synthesis indicates that stem cells, which are capable of bud formation, are the specific targets of TGF-p action on mammary epithelium.
149
chyme inductive processes in the embryo (see also Heine et al., 1987; Bissell and Hall, 1987).
EFFECT OF TGF-P ON MILK PROTEINS We have examined the effects of TGF-p on the production of milk proteins by primary explants of pregnant gland, cultured in a medium containing lactogenic hormones. In these studies we found that TGF-p inhibited the release of newly synthesized milk proteins into the medium. TGF-P3, whose transcripts were prominent during pregnancy, was inhibitory at picomolar levels. Proteins were identified in a western blot using a pan-antibody against milk. Autoradiograph of radiolabeled proteins released into the medium by explants incubated with and without TGF-p showed that there was no effect of TGF-P on other proteins, indicating that the effect of TGF-p was specific. These data suggest that the TGF-ps, particularly TGF-P3, may have a role in negatively regulating the secretion of milk proteins from the mammary gland during pregnancy, a role historically ascribed to progestins. This suggestion is consistent with the elevated levels of TGF-p3 peptides and mRNA in pregnancy and their rapid decline at birth. We are exploring the possibility that TGF-p acts as an intermediate in hormone action, negatively regulating milk protein secretion as mammary cells undergo functional differentiation. CONCLUSIONS The data suggest that TGF-p functions as a physiological regulator of both mammary growth and function. In the immature gland, TGF-P appears to be secreted by mammary epithelium and perhaps adipocytes, where it becomes associated with the extracellular matrix surrounding mammary ducts. Matrixassociated TGF-P was never observed at growing buds, and became depleted in a highly focal manner at locations where new growth points were forming. Exogenous TGF-P, administered by slow-release plastic implants, inhibited existing buds and prevented the formation of new ones by suppressing stem cell DNA synthesis. Recent experiments suggest an additional role for TGF-p in regulating the secretion of milk caseins during pregnancy.
ROLE OF THE EXTRACELLULAR MATRIX When TGF-(3-inhibited end buds were examined by in situ hybridization using a probe for collagen type I, the buds located near an implanted source of TGF-P became surrounded by a cap of cells with greatly eleACKNOWLEDGMENTS vated levels of collagen mRNA, and the end bud became surrounded by newly synthesized fibrous matrix (SilWe acknowledge the collaboration of Kathleen berstein et al., 1990). This was not found around end Flanders, Michael Sporn, and Anita Roberts at the Labbuds from untreated animals or with control implants. oratory of Chemoprevention, NIH. The research was Notably, we did not see an increased level of collagen funded by PHS grant HD 27845. mRNA in fibroblasts immediately around the implant, nor was increased matrix protein synthesis around the REFERENCES implant ever observed. Instead, increased levels of gene Bissell MJ, Hall HG (1987): Form and function in the mammary transcription and increased synthesis of matrix compogland the role of extracellular matrix. In MC Neville, CW Daniel (eds): “The Mammary Gland: Development, Regulation, and Funcnents was found only at the surface of the buds, indicattion.” New York: Plenum Press, pp 97-146. ing the epithelium-dependent nature of the TGF-p efCW, Silberstein GB, VanHorn K, Strickland P, Robinson S fect on matrix. TGF-P appears to modify the Daniel (1989): TGF-beta-l-induced inhibition of mouse mammary ductal epithelium-stroma interaction. I am reminded of Dr. growth: Developmental specificity and characterization. Dev Biol Akhurst’s comments (this issue) about the possible im13520-30. portance of TGF-p in mediating epithelium-mesen- Daniel CW, Silberstein, GB (1987): “Postnatal development of the
150
C.W. DANIEL AND S.D. ROBINSON
mammary gland.” In MC Neville, CW Daniel (eds): “The Mammary Gland: Development, Regulation, and Function.” New York: Plenum Press, pp 3-36. Faulkin U Jr, DeOme KB (1960): Regulation of growth and spacing of gland elements in the mammary fat pad of the C3H mouse. JNCJ 24:953-968. Heine UI, Munoz EF, Flanders KC, Ellingsworth LR, Lam PH, Thompson NL, Roberts AB, Sporn ME (1987): Role of transforming growth factor beta in the development of the mouse embryo. J Cell Biol 105:2861-2876. Roberts AB, Sporn ME (1989): The transforming growth factor-betas. In ME Sporn, AB Roberts (eds): “Peptide Growth Factors and Their Receptors.” Heidelberg: Springer-Verlag. Robinson SD, Silberstein GB, Roberts AB, Flanders KC, Daniel CW (1991):Regulated expression and growth inhibitory effects of transforming growth factor-beta isoforms in mouse mammary gland development. Development 113:867-878. Rouslahti E, Yamaguchi Y (1991): Proteoglycans as modulators of growth factor activities. Cell 64567-869. Silberstein GB, Daniel CW (1982): Elvax 40P implants: Sustained local release of bioactive molecules influencing mammary ductal development. Dev Biol93:215-222. Silberstein GB, Daniel CW (1987a):Investigation of mouse mammary ductal growth regulation using slow-release plastic implants. J Dairy Sci 70:1981-1990. Silberstein GB, Daniel C (1987b): Reversible inhibition of mammary gland growth by transforming growth factor+. Science 237:291293. Silberstein GB, Daniel CW, Coleman S, Strickland P (1990): Epithelium-dependent induction of mouse mammary gland extracellular matrix by TGF-beta-1. J Cell Biol 110:2209-2219. Williams JM, Daniel CW (1983): Mammary ductal elongation: Differentiation of myoepithelium during branching morphogenesis. Dev Biol97:27&290.
QUESTIONS AND ANSWERS Q: Do transplanted tissues from mammary gland arborize in fat pads other than of the mammary gland? A: This has been a subject of considerable debate, but people have always reported that it does. In heterologous locations such as the interscapular fat pad or the fat pad above the kidney, the amount of growth is extremely restricted. There is nothing like the luxurious growth that you see in the mammary fat pad, so I think there is a considerable, although not a n absolute specificity, in the stromal requirements for mammary ductal growth. Q: Could you speculate on how matrix-bound TGF-@ is dissipated? A: I think that matrix active proteases are probably involved. We do have evidence for the presence of PAI, stromelysin, and other proteases in these ducts. We don’t yet know whether the enzymes are responsible for the focal loss of TGF-P. This is just pushing the question back another step, because the question really is, what initiates this process of budding. We really have no information on that at all. TGF-P seems to be uniformly distributed down the duct. There doesn’t seem to be any sort of pattern to its distribution. Q: What’s the distribution of the extracellular matrix and TGF-P in precancerous nodules? A: All we know is that growth in the precancerous nodules is not inhibited by TGF-p. The question is whether the lack of response means that the precancer-
ous cells are not inhibited because of their particular growth state or whether they might have lost receptors and become insensitive or refractory. It is a n interesting question and we are just getting started in studying that. Q: You said that EGF was also inhibitory. Is there any role for endogenous EGF? A: Yes, we’ve published that. There is endogenous EGF in the gland. When you apply EGF exogenously to the gland it down-regulates the EGF receptors in the mammary gland to a n undetectible level. We think this is why EGF inhibits proliferation. There’s a very definite dose-dependent effect. It is low levels chronically administered over a period of time that gives rise to this inhibitory effect of EGF. We think that EGF (or perhaps TGF-alpha)is a prominent endogenous mitogen in the mammary gland. Q: Have you compared the effect of adding latent TGF-p as opposed to the mature protein? A: That’s a n experiment that we’ve thought of doing many times, but have been stopped by the lack of a good supply of latent TGF-P. I do think that it is a potentially very interesting experiment because we don’t know anything about the activation of latent TGF-P in vivo. And it would be a n opportunity to combine Elvax implants containing latent TGF-p with specific agents such as enzymes which might modify the environment and activate the growth factor. Q: Has anyone done any work with antisense oligodeoxynucleotides with the plastic implants? A: Not as far as I know, but Barbara Vonderhaar a t the NIH is doing work in this direction. It seems like a reasonable thing to do, and I agreed that it has potential. Q: You observed a n increase in collagen close to the implant. One of the very strong strong responses to TGF-P is the induction of the plasminogen activator inhibitor. Its release by TGF-@-stimulatedcells could achieve the same sort of effect of accumulation of extracellular matrix. Have you looked at the expression of PAI-1 in cells near the implant and could PAI-1 be the factor that’s causing extracellular matrix accumulation by inhibiting its degradation? A: I don’t think we’ve immunostained for PAI-1. But, I don’t see how that would explain the epithelium dependence of the process in which the only place that you get matrix synthesis is next to the epithelium. Only right at that point. You never get it anywhere else around in the sphere of influence of the pellet. Q: Have you characterized the gradient of TGF-P from the implant? How far way can you detect it? A: Yes, we’ve done that. It looks a lot like a n Ouchterlony dish where you have diffusion through agar. It’s really quite surprising that we found such uniform diffusion of the material in a tissue which is permeated with capillaries. There is the expected nonlinear gradient decreasing from the center of the source. It extends for a considerable distance of 2 or 3 mm away from the implant. It is actually possible to quantitate the
REGULATION OF MAMMARY GROWTH AND FUNCTION BY TGF-P
151
amount of material released by this technique, by doing is that the epitopes are characteristically distributed autoradiographs and looking at densities. for each of these antibodies. For example, if you don’t Q: Is it correct to assume that the antisera you used to see the intracellular form in fibroblasts, that doesn’t detect TGF-f3 could detect both latent and active TGF-p? mean it isn’t there, but what it does mean is that if A: I would make no claims about being able to distin- TGF-P is present, the epitope is in a different configuraguish between latent and active TGF-P. All one can say tion because it’s no longer available to the antibody.
Regulation of Mammary Growth and Function by TGF-P CHARLES W. DANIEL AND STEPHEN D. ROBINSON Department of Biology, University of California, Santa Cruz, California
ABSTRACT We have previously shown that TGF-p1 rapidly and reversibly inhibits ductal growth in vivo when administered by miniature slow-release plastic implants. A possible role for endogenous TGF-61 was suggested by the observation that the normal gland displayed substantial, developmentally regulated levels of TGF-pl transcripts and protein. These studies have now been extended to include the other two mammalian TGF-p isoforms. When tested with slow-release plastic implants, TGF-P2 and TGF-p3 also caused disappearance of the proliferating mammary stem cell layer, with rapid involution of ductal end buds and cessation of glandular growth. None of the isoforms was active in inhibiting alveolar morphogenesis. We conclude that under the conditions of these tests, the three mammaliian isoforms are functionally equivalent. However, striking differences in patterns of gene expression and in the distribution of immunoreactive peptides suggest that TGF-p2 was expressed only at low levels, and mainly during pregnancy. TGF-(33 was expressed in ductal stroma and epithelium, and was the only isoform detected in myoepithelial cells. Developing alveolar tissue and its associated ducts displayed striking TGF-p3 gene expression and immunostaining, which were greatly reduced during lactation. We are now investigating the possibility that the observed high levels of TGF-(3 expression in pregnancy, particularly of TGF-(33, and the absence of substantial expression of any isoform during lactation, may indicate a role for the TGF-p in regulating functional differentiation or the onset of milk secretion. 0 1992 Wiley-Liss, Inc.
Key Words: Transforming growth factor-(3, Mammary, Growth regulation
INTRODUCTION In general, we know a great deal more about what stimulates the growth of organs than we know about what regulates their growth. This is particularly true in the case of the mammary gland, in which the systemic mitogens, the mammogenic hormones which drive growth, have been defined and examined by generations of endocrinologists. However, we do not know what processes or what factors limit the final size of the mammary glands, even in the continued presence of these powerful mitogens. Nor do we know what defines the process of pattern formation during ductal morphogenesis or what stabilizes the pattern that is made. We
0 1992 WILEY-LISS, INC.
will discuss these aspects of the mammary gland using the mouse as the experimental animal. Mammary biology is often regarded as a specialized subject, but its branching morphogenesis is reminiscent of developmental patterns in other structures such as the lung, kidney, exocrine organs such as the salivary glands, and various glands of the reproductive tract. Studies of the mammary gland therefore have general applicability with respect to other tissues and to the general process of organogenesis in vertebrates. An important advantage of mammary studies is that most mammary growth and morphogenesis occurs after birth and occurs in an animal that is large enough to work on conveniently and in which the glands are in an accessible location, so that unlike the lung, kidney, or other internal organs, manipulations in vivo can readily be carried out. Growth of the mammary gland begins in utero. After birth the mammary ducts grow slowly, but at the time of puberty and sexual maturation ductal growth begins in earnest. The mammary tree, a system of branching ducts, is created, which then undergoes another and quite different growth process during pregnancy, when fine branching and secretory alveoli are formed. Alveolar development eventually culminates in the formation of secretory lobules that become engorged with milk during lactation. Secretion continues until weaning, when, in the absence of lactogenic hormones, the mass of secretory tissue degenerates and the gland reverts to a pattern similar to that of the virgin. The period of ductal morphogenesis (see Daniel and Silberstein, 1987, for review), the stage of mammary development in which we have been particularly interested, occurs in the mouse a t around 5-8 weeks of age, and is driven by the ovarian and pituitary hormones associated with sexual maturation. Ductal elongation takes place as a result of intense mitotic activity in end buds, the bulbous structures at the tips of ducts. As the end buds penetrate the surrounding adipose stroma, they leave behind a tree-like pattern of differentiated ducts that are mitotically inactive. Consequently, in a growing gland only about 5%of the total tissue is actually engaged in growth, while the remainder of the
Address reprint requests to Dr. Charles W. Daniel, Department of Biology, University of California, Thimann Laboratories, Santa Cruz, CA 95064.
146
C.W. DANIEL AND S.D. ROBINSON
trunk and branches of the ductal tree are in static phase. The mammary ducts are cloaked in a mantle of fibrous tissue, termed the periductal matrix, which consists of type I collagen and a variety of other extracellular matrix (ECM) elements. In contrast, the tips of growth-active buds are not surrounded by matrix and the basal lamina of the epithelial mass may be in direct contact with adipocytes of the stroma (Williams and Daniel, 1983). There is a great difference, both structurally and biologically, between the mature duct and the proliferating buds. There are three types of buds in the mammary gland: the end buds, which are responsible for ductal elongation; the lateral buds, which develop along mature ducts from time to time, grow a short distance, and then stop because of the presence of competing ductal elements; and alveolar buds, which in most mouse strains develop only in response to the hormones of pregnancy. In the virgin gland, branching morphogenesis creates a pattern in which the interductal space measures about .25mm. The adipose tissue in this interductal region is nonpermissive for growth of lateral buds. Our evidence suggests that TGF-(3is a t least in part responsible for restricting the formation and growth of lateral buds, and functions normally to maintain the open pattern of branching that is required for alveolar development during pregnancy. TGF-fi does not, on the other hand, restrict the development of alveolar structures (Daniel et al., 1989). We have investigated the question of what regulates mammary gland patterning. Why do the ductal buds not fill the stroma between ducts-that is, why does the mammary gland look like a tree in winter instead of a plate of spaghetti? One possible explanation is that the ducts are terminally differentiated and no longer have the capacity to form new growth points. That this is not the case was shown several years ago by Faulkin and DeOme (1960).Mammary cells taken from any location in the gland will form new buds when transplanted into a mammary gland-free “cleared fat pad” and regenerate a new gland that is identical in all respects with the parent tissue except that it no longer makes connection with a nipple. By contrast, mammary epithelial cells transplanted into a fat pad containing ducts fail to grow and remain in healthy but static condition for the life of the host. These results indicate that the mammary gland is held in a dynamic balance between proliferation and stasis, pulled toward mitosis by powerful systemic mitogens, but inhibited by the presence of surrounding tissue. We have examined the possibility that TGF-p serves a regulatory role in the mammary gland by asking several questions. Is TGF-P present in the gland and if so, is it present in developmentally interesting places? What are the biological effects of TGF-p? What is the comparative biology of the TGF-(3isoforms?
TGF-P GENE EXPRESSION IN THE MAMMARY GLAND A possible role for endogenous TGF-(31was suggested by the observation that the normal gland displayed
TGFPl
-2.4
-6.0 TGFP2 4-5.0 -4.0 4-3.5
TGFP3
-3.8
18s
Fig. 1. Northern analysis of TGF-p isoform expression during mammary gland development. “Cleared” indicates glands that have been surgically cleared of mammary epithelium. Arrows on right indicate sizes of TGF-P transcripts (four transcripts are characteristically seen for TGF-PX). The lower frame has been probed for 185 ribosomal RNA, as a loading control.
substantial, developmentally regulated levels of TGF-Pl transcripts. These studies have now been extended to include the other two mammalian isoforms, TGF-P2 and TGF-p3 (Robinson et al., 1991). Striking differences in patterns of gene expression indicate a high level of developmental regulation (Fig. 1).TGF-p1 was expressed in all stages except lactation. TGF-pB
REGULATION OF MAMMARY GROWTH AND FUNCTION BY TGF-p
147
Fig. 2. End-bud expression of TGF-PI mRNA. Bright-field and corresponding dark-ground photomicrographs of adjacent sections through a bifurcating end bud. The riboprobe was 35S-UTP-labeled and consisted of a fragment from the coding region for the mature (active form) peptide. Controls were hybridized to probes to irrelevant transcripts (not shown). TGF-61 expression is seen over the end bud epithelium with heaviest labeling at the tips (arrows). (Reproduced from Robinson et al., 1991, with permission of The Company of Biologists Limited.)
was expressed in immature glands only at low levels, and somewhat increased mRNA amounts were found in pregnancy. TGF-P3 was expressed in ductal stroma and epithelium, and its transcripts were strikingly elevated in pregnancy. We are now investigating the possibility that the observed high levels of TGF-@3expression in pregnancy, and to a lesser extent TGF-P1 and -P2, taken together with the absence of substantial expression of any isoform during lactation, may indicate a role for the TGF-ps in regulating functional differentiation or the onset of milk secretion.
DISTRIBUTION OF TGF-@TRANSCRIPTS AND PEPTIDES IN THE MAMMARY GLAND To investigate the question of when and where TGF-P is expressed we studied TGF-P1 transcripts in mammary glands from the 5 to 7-week-old mouse, the mature gland, during pregnancy, and in lactation, using Northern and in situ hybridization to riboprobes (Robinson et al., 1991). Northern analysis showed that isoform transcripts were expressed in a stage- and isoform-specific manner during all developmental stages except lactation, where transcripts abruptly disappeared (Fig. 1).In situ hybridization showed high levels of TGF-P transcripts in mammary epithelial tissues and a lower level of transcripts in fat cells and fibrous stroma of connective tissues (Fig. 2). The in situ results were consistent with data from Northerns. Developing alveolar tissue and its associated ducts displayed strik-
ing TGF-P3 gene expression in mammary epithelium, which was absent during lactation. Using immunostaining for both precursor and active forms of the growth factors, we investigated mammary ducts and spontaneously occurring lateral buds. Mature TGF-@was found in the periductal matrix along the sides of the bud, but not a t the tip of growing buds. The ductal epithelium was rich in intracellular TGF-@, whereas fibroblasts, which synthesize various matrix components, failed to stain for intracellular TGF-P. We conclude that the matrix-associated TGF-P was synthesized and secreted basally by mammary epithelium. It is possible that there is a stromal contribution as well, since adipocytes are rich in TGF-P. Focal depletion at the growth points provides support for the hypothesis that the extracellular matrix acts as a reservoir for growth factors, perhaps bound to specific matrix components, which can be released at morphogenetically active sites (Rouslahti and Yamaguchi, 1991). As a result of these studies we have developed the concept that interductal spacing is maintained by TGF-@accumulation in the periductal matrix, and focal loss of the growth factor is associated with budding and growth.
LOCALLY IMPLANTED TGF-P INHIBITS MAMMARY DUCTAL GROWTH To observe the effect of exogenous TGF-@on ductal growth we implanted slow-release plastic implants con-
148
C.W. DANIEL AND S.D. ROBINSON
Fig. 3. Photomicrograph illustrating the effect of TGF-PI on ductal growth. Five-week-old C57 female mice were treated for 4 days with slow-release plastic implants (star) containing carrier protein (A) or TGF-61 (B). End buds (arrows) have regressed in the treated gland. (Reproduced from Silberstein and Daniel, 1987b, with permission of Science. Copyright 1987 by the AAAS.)
taining TGF-p into specific sites in the gland, where they released the growth factor slowly over a period of time (Silberstein and Daniel, 1982,1987a). Control implants placed in contralateral glands in the same animal served to indicate whether the observed effects were systemic or local. Figure 3 shows the inhibitory effects of TGF-P1 48 h r after implantation; the growth points of the end buds were completely inhibited. Similar results were obtained with all three isoforms (Robinson et al., 1991) causing disappearance of the proliferating mammary stem cell layer, rapid involution of buds, and cessation of glandular growth. None of the isoforms were active in inhibiting alveolar morphogenesis or DNA synthesis, indicating considerable developmental specificity. We conclude that under the conditions of these tests, the three mammalian isoforms are functionally equivalent, a result that is consistent with their observed cross reactivity with receptors (Roberts and Sporn, 1989).
THE NATURE OF DUCTAL GROWTH INHIBITION BY TGF-P To determine whether the effect of TGF-p was reversible and whether the tissue became refractory to the
sustained release of TGF-p, we implanted TGF-p1 in the mammary gland and left the implant until the charge of TGF-pl was exhausted. After a period of about a week the gland resumed its normal growth (Fig. 41, indicating reversibility. We also found that if we maintained a continuous supply of TGF-p1 in the region of the implants, the gland remained chronically suppressed and there was no indication t h a t it became refractory to the continued presence of TGF-6. This result is consistent with the observation that TGF-p receptors are not down-regulated. If we allowed the chronically suppressed gland to exhaust its charge of TGF-p, growth again resumed. Thus, the effect of TGF-8 was fully reversible and continued as long a s the factor was present, a characteristic that is required if TGF-P is to qualify as a chronic inhibitor. We used transplantation procedures to determine whether TGF-p was capable of preventing the formation of buds, as well as inhibiting the growth of those previously formed (Daniel et al., 1989). When the implant contained TGF-p, the formation of new buds in transplanted tissue was suppressed. The DNA synthetic labeling index in buds in control glands was extremely high (about 2 0 3 0 % ) and decreased to basal
REGULATION OF MAMMARY GROWTH AND FUNCTION BY TGF-P 1 9 , IL
I
I
0
4
8
12
16
20
24
28 30
Days
Fig. 4. Effects of multiple TGF-p1 treatment on ductal growth. Treatment-recouery-treatmentcycle (0):A sample of mice in the experimental group were sacrificed at the time points indicated by circles. Open arrows indicate re-implantation. Timing was adjusted so that implants exhausted their TGF-p1 charge, permitting the end buds to become re-established to levels of the control group (shaded band) and indicating reversibility of the inhibitory effect. Prolonged exposure ( 0 ) : Glands were re-implanted at short intervals to maintain relatively constant level of exposure to TGF-p1. Within the time period tested, the gland did not become refractory to inhibition. (Reproduced from Daniel et al., 1989, with permission of Academic Press.)
levels with TGF-p treatment. The basal level of DNA synthesis in mammary ducts was low ( 2 4 % ) but far from negligible, which we interpret as a maintenance level reflecting normal cell loss and replacement. The lack of a TGF-p effect on this maintenance DNA synthesis indicates that stem cells, which are capable of bud formation, are the specific targets of TGF-p action on mammary epithelium.
149
chyme inductive processes in the embryo (see also Heine et al., 1987; Bissell and Hall, 1987).
EFFECT OF TGF-P ON MILK PROTEINS We have examined the effects of TGF-p on the production of milk proteins by primary explants of pregnant gland, cultured in a medium containing lactogenic hormones. In these studies we found that TGF-p inhibited the release of newly synthesized milk proteins into the medium. TGF-P3, whose transcripts were prominent during pregnancy, was inhibitory at picomolar levels. Proteins were identified in a western blot using a pan-antibody against milk. Autoradiograph of radiolabeled proteins released into the medium by explants incubated with and without TGF-p showed that there was no effect of TGF-P on other proteins, indicating that the effect of TGF-p was specific. These data suggest that the TGF-ps, particularly TGF-P3, may have a role in negatively regulating the secretion of milk proteins from the mammary gland during pregnancy, a role historically ascribed to progestins. This suggestion is consistent with the elevated levels of TGF-p3 peptides and mRNA in pregnancy and their rapid decline at birth. We are exploring the possibility that TGF-p acts as an intermediate in hormone action, negatively regulating milk protein secretion as mammary cells undergo functional differentiation. CONCLUSIONS The data suggest that TGF-p functions as a physiological regulator of both mammary growth and function. In the immature gland, TGF-P appears to be secreted by mammary epithelium and perhaps adipocytes, where it becomes associated with the extracellular matrix surrounding mammary ducts. Matrixassociated TGF-P was never observed at growing buds, and became depleted in a highly focal manner at locations where new growth points were forming. Exogenous TGF-P, administered by slow-release plastic implants, inhibited existing buds and prevented the formation of new ones by suppressing stem cell DNA synthesis. Recent experiments suggest an additional role for TGF-p in regulating the secretion of milk caseins during pregnancy.
ROLE OF THE EXTRACELLULAR MATRIX When TGF-(3-inhibited end buds were examined by in situ hybridization using a probe for collagen type I, the buds located near an implanted source of TGF-P became surrounded by a cap of cells with greatly eleACKNOWLEDGMENTS vated levels of collagen mRNA, and the end bud became surrounded by newly synthesized fibrous matrix (SilWe acknowledge the collaboration of Kathleen berstein et al., 1990). This was not found around end Flanders, Michael Sporn, and Anita Roberts at the Labbuds from untreated animals or with control implants. oratory of Chemoprevention, NIH. The research was Notably, we did not see an increased level of collagen funded by PHS grant HD 27845. mRNA in fibroblasts immediately around the implant, nor was increased matrix protein synthesis around the REFERENCES implant ever observed. Instead, increased levels of gene Bissell MJ, Hall HG (1987): Form and function in the mammary transcription and increased synthesis of matrix compogland the role of extracellular matrix. In MC Neville, CW Daniel (eds): “The Mammary Gland: Development, Regulation, and Funcnents was found only at the surface of the buds, indicattion.” New York: Plenum Press, pp 97-146. ing the epithelium-dependent nature of the TGF-p efCW, Silberstein GB, VanHorn K, Strickland P, Robinson S fect on matrix. TGF-P appears to modify the Daniel (1989): TGF-beta-l-induced inhibition of mouse mammary ductal epithelium-stroma interaction. I am reminded of Dr. growth: Developmental specificity and characterization. Dev Biol Akhurst’s comments (this issue) about the possible im13520-30. portance of TGF-p in mediating epithelium-mesen- Daniel CW, Silberstein, GB (1987): “Postnatal development of the
150
C.W. DANIEL AND S.D. ROBINSON
mammary gland.” In MC Neville, CW Daniel (eds): “The Mammary Gland: Development, Regulation, and Function.” New York: Plenum Press, pp 3-36. Faulkin U Jr, DeOme KB (1960): Regulation of growth and spacing of gland elements in the mammary fat pad of the C3H mouse. JNCJ 24:953-968. Heine UI, Munoz EF, Flanders KC, Ellingsworth LR, Lam PH, Thompson NL, Roberts AB, Sporn ME (1987): Role of transforming growth factor beta in the development of the mouse embryo. J Cell Biol 105:2861-2876. Roberts AB, Sporn ME (1989): The transforming growth factor-betas. In ME Sporn, AB Roberts (eds): “Peptide Growth Factors and Their Receptors.” Heidelberg: Springer-Verlag. Robinson SD, Silberstein GB, Roberts AB, Flanders KC, Daniel CW (1991):Regulated expression and growth inhibitory effects of transforming growth factor-beta isoforms in mouse mammary gland development. Development 113:867-878. Rouslahti E, Yamaguchi Y (1991): Proteoglycans as modulators of growth factor activities. Cell 64567-869. Silberstein GB, Daniel CW (1982): Elvax 40P implants: Sustained local release of bioactive molecules influencing mammary ductal development. Dev Biol93:215-222. Silberstein GB, Daniel CW (1987a):Investigation of mouse mammary ductal growth regulation using slow-release plastic implants. J Dairy Sci 70:1981-1990. Silberstein GB, Daniel C (1987b): Reversible inhibition of mammary gland growth by transforming growth factor+. Science 237:291293. Silberstein GB, Daniel CW, Coleman S, Strickland P (1990): Epithelium-dependent induction of mouse mammary gland extracellular matrix by TGF-beta-1. J Cell Biol 110:2209-2219. Williams JM, Daniel CW (1983): Mammary ductal elongation: Differentiation of myoepithelium during branching morphogenesis. Dev Biol97:27&290.
QUESTIONS AND ANSWERS Q: Do transplanted tissues from mammary gland arborize in fat pads other than of the mammary gland? A: This has been a subject of considerable debate, but people have always reported that it does. In heterologous locations such as the interscapular fat pad or the fat pad above the kidney, the amount of growth is extremely restricted. There is nothing like the luxurious growth that you see in the mammary fat pad, so I think there is a considerable, although not a n absolute specificity, in the stromal requirements for mammary ductal growth. Q: Could you speculate on how matrix-bound TGF-@ is dissipated? A: I think that matrix active proteases are probably involved. We do have evidence for the presence of PAI, stromelysin, and other proteases in these ducts. We don’t yet know whether the enzymes are responsible for the focal loss of TGF-P. This is just pushing the question back another step, because the question really is, what initiates this process of budding. We really have no information on that at all. TGF-P seems to be uniformly distributed down the duct. There doesn’t seem to be any sort of pattern to its distribution. Q: What’s the distribution of the extracellular matrix and TGF-P in precancerous nodules? A: All we know is that growth in the precancerous nodules is not inhibited by TGF-p. The question is whether the lack of response means that the precancer-
ous cells are not inhibited because of their particular growth state or whether they might have lost receptors and become insensitive or refractory. It is a n interesting question and we are just getting started in studying that. Q: You said that EGF was also inhibitory. Is there any role for endogenous EGF? A: Yes, we’ve published that. There is endogenous EGF in the gland. When you apply EGF exogenously to the gland it down-regulates the EGF receptors in the mammary gland to a n undetectible level. We think this is why EGF inhibits proliferation. There’s a very definite dose-dependent effect. It is low levels chronically administered over a period of time that gives rise to this inhibitory effect of EGF. We think that EGF (or perhaps TGF-alpha)is a prominent endogenous mitogen in the mammary gland. Q: Have you compared the effect of adding latent TGF-p as opposed to the mature protein? A: That’s a n experiment that we’ve thought of doing many times, but have been stopped by the lack of a good supply of latent TGF-P. I do think that it is a potentially very interesting experiment because we don’t know anything about the activation of latent TGF-P in vivo. And it would be a n opportunity to combine Elvax implants containing latent TGF-p with specific agents such as enzymes which might modify the environment and activate the growth factor. Q: Has anyone done any work with antisense oligodeoxynucleotides with the plastic implants? A: Not as far as I know, but Barbara Vonderhaar a t the NIH is doing work in this direction. It seems like a reasonable thing to do, and I agreed that it has potential. Q: You observed a n increase in collagen close to the implant. One of the very strong strong responses to TGF-P is the induction of the plasminogen activator inhibitor. Its release by TGF-@-stimulatedcells could achieve the same sort of effect of accumulation of extracellular matrix. Have you looked at the expression of PAI-1 in cells near the implant and could PAI-1 be the factor that’s causing extracellular matrix accumulation by inhibiting its degradation? A: I don’t think we’ve immunostained for PAI-1. But, I don’t see how that would explain the epithelium dependence of the process in which the only place that you get matrix synthesis is next to the epithelium. Only right at that point. You never get it anywhere else around in the sphere of influence of the pellet. Q: Have you characterized the gradient of TGF-P from the implant? How far way can you detect it? A: Yes, we’ve done that. It looks a lot like a n Ouchterlony dish where you have diffusion through agar. It’s really quite surprising that we found such uniform diffusion of the material in a tissue which is permeated with capillaries. There is the expected nonlinear gradient decreasing from the center of the source. It extends for a considerable distance of 2 or 3 mm away from the implant. It is actually possible to quantitate the
REGULATION OF MAMMARY GROWTH AND FUNCTION BY TGF-P
151
amount of material released by this technique, by doing is that the epitopes are characteristically distributed autoradiographs and looking at densities. for each of these antibodies. For example, if you don’t Q: Is it correct to assume that the antisera you used to see the intracellular form in fibroblasts, that doesn’t detect TGF-f3 could detect both latent and active TGF-p? mean it isn’t there, but what it does mean is that if A: I would make no claims about being able to distin- TGF-P is present, the epitope is in a different configuraguish between latent and active TGF-P. All one can say tion because it’s no longer available to the antibody.
