Differentiation
Differentiation (1990) 42: 199-207
Ontogeny and Neoplasia 0 Springer-Verlag 1990
Tenascin distribution in the normal human breast is altered during the menstrual cycle and in carcinoma Janice E. Ferguson', Ana M. Schor', Anthony Howell', and Mark W.J. Ferguson2* CRC Department of Medical Oncology, The Christie Hospital and Holt Radium Institute, Wilmslow Road, Withington, Manchester M20 9BX, England Department of Cell & Structural Biology, University of Manchester, Coupland I11 Building, Manchester MI 3 9PL, England
Abstract. Tenascin is a novel extracellular matrix glycoprotein which appears to have a major role in tissue development [6]. Previous studies have stated that tenascin is absent from the normal human 1221, rat [6] and mouse [18] breast, its distribution being restricted to embryonic and malignant mammary tissues [22]. No previous studies have investigated tenascin distribution as a function of the normal menstrual cycle. Therefore this study addresses the cyclical appearance of tenascin in the normal breast and associated changes in distribution in preinvasive cancer (carcinoma-insitu) and invasive infiltrating ductal carcinoma. Tenascin is present in the normal human adult mammary gland, principally in the basement membrane, sub-basement-membrane zone and delimiting layer of fibroblasts around the ductules. Both the distribution and quantity of tenascin change during the menstrual cycle. In carcinoma-in-situ (preinvasive cancer) tenascin is present in the attenuated basement membrane/sub-basement-membranezone around the expanded ductules and in small amounts in the stroma. In infiltrating ductal carcinoma, tenascin is absent from the remnants of the basement membrane and sub-basement-membrane zone but greatly increased in the adjacent intralobular and interlobular stroma. Therefore, if tenascin is used as a basement membrane/sub-basement-membrane marker for distinguishing carcinoma-in-situ from invasive ductal carcinoma, the time of the menstrual cycle is of importance in interpreting the biopsy appearance. This study suggests that the optimal time for biopsy is between weeks 3 and 4 of the cycle, to avoid confusion between the normal low levels of tenascin (due to hormonal status) and those due to microinvasive disease. That tenascin expression in the normal breast is cyclically regulated suggests that this molecule may act as a mediator in the hormonal control of mammary epithelial cell behaviour.
Introduction
Tenascin is a novel extracellular matrix glycoprotein with a disulfide-bonded subunit of 150-320 kDa (5, 16, 211. It is a hexabrachion consisting of two pairs of three arms connected to a central globule and contains 13 epidermal growth factor (EGF)-like repeats [29]. Tenascin has a restricted distribution in the mesenchymal condensations ar-
* To whom offprint requests should be sent
ound budding epithelia of developing organs such as hair follicles, tooth germs, mammary anlagen 151, scleral ossicles [I51 and the palate [14] in addition to the perichondral and tendinous anlagen [5].Tenascin has been found in the mesenchyme adjacent to differentiated gut epithelium in the embryonic mouse and in the stroma adjacent to migration pathways of the continually renewed intestinal villous epithelia in the adult mouse 121. Tenascin also lines the future migratory pathways of neural crest cells in quail, rat, 1241 and amphibian embryos [I I]. Tenascin antagonises the cell binding activity of fibronectin 171 and thus may be important in stimulating cell migration and cell rounding. Tenascin has been found in the mesenchyme beneath the epithelial rudiment in the fetal mammary gland, and in the stroma of both carcinogen-induced mammary carcinomas of the rat, and spontaneous human mammary adenocarcinomas [6, 221. It is apparently absent from benign neoplasms, fibrocystic disease and normal adult mammary tissue in both human and rat [6, 221. Tenascin has not been described in normal human breast in relation to the menstrual cycle, nor in carcinoma-in-situ lesions. The above distribution of tenascin led Chiquet-Ehrismann et al. 161 to suggest that tenascin was an oncofetal molecule, important in the development of the normal mammary gland and in the early stages of mammary carcinogenesis. They also suggested that the appearance of stroma1 tenascin may be a marker of epithelial malignancy and therefore have important diagnostic and therapeutic implications. This study first investigated the distribution of tenascin in the normal human mammary gland at different points in the menstrual cycle. Tenascin distribution in infiltrating ductal carcinoma and carcinoma-in-situ was then studied in order to investigate whether these conditions could be distinguished from each other and whether the pattern and distribution of tenascin observed might suggest a role for tenascin in the critical stages between preinvasive (carcinoma-in-situ) and invasive infiltrating ductal carcinoma. Methods
Nineteen normal mammary biopsies were obtained from the periphery of the breast in young women undergoing operation for benign fibroadenoma. The day of the menstrual cycle was recorded at biopsy (Table 1). Tumour specimens were obtained from surgical mastectomies of infiltrating ductal carcinoma (n = 5 ) and carcino-
200
Table 1. Distribution of specimens used in the study of the normal breast at different stages in the menstrual cycle, including the age of the patient, in years, day of the cycle and the operation during which the normal specimen was removed Age
Day of cycle
Operation
20 24 20 22 25 26 18 21 22 16 19 22 19 22 21 23 19 25 21
2 3 5 6 6 8 9 9 12 17 20 21 22 22 25 26 26 28 28
Fibroadenoma Normal/no lesion Fi broadenoma Fibroadenoma Fi broadcnoma Fibroadenoma Fibroadenoma Fibroadenoma Fibroadenoma Fibroadenoma Fibroadenoma Fibroadenoma Fibroadenoma Fibroadenoma Fibroadenoma Lipoma Fibroadenoma Fibroadenoma Lymph node
ma in situ ( n = 2 ) in older women (aged 37-51 years). The tissue was rapidly frozen in liquid nitrogen and stored at -70" C for up to 1 year. Specimens were embedded in OCT (Miles Inc USA) and sectioned at 5-9 pm using a Leitz Cryostat. The sections were mounted on clean glass slides, air-dried and fixed in cold acetone for 10 min. They are either used immediately or stored for up to 1 month at - 70" C. In order to exclude possible masking of epitopes (in cases of negative results) the sections received one of the following pretreatments: (a) 0.1 M acetic acid, (b) 0.5 M acetic acid, (c) sheep testicular hyaluronidase type I11 (Sigma) 1 mg/50 ml in 1 M sodium acetate buffer adjusted to pH 5, (d) no pretreatment for 30 min at room temperature. They were then incubated in a humidified chamber with polyclonal antiserum raised in rabbit against the chick tenascin molecule, at a dilution of 1 : 100 (for details of antibody preparation and characterization see [6] for 1 h and then washed for three 5-min periods with phosphatebuffered saline (PBS). After applying fluorescein isothiocyanate (FITC) conjugated sheep anti-rabbit as secondary antibody (Serotec) and washing with PBS, sections were mounted in glycerol containing PBS and 1,4 diazobicyclo 2,2,2, octane [19] (DABCO) to inhibit fading. Examination for immunofluorescence was carried out using a Leitz Orthoplan incident epifluorescence microscope (equipped with a Ploemopak filter) and photographs taken using Ektachrome ASA160 colour tungsten film from which black and white prints were made. Control staining was performed, as above, but with the following exceptions: (a) omission of the primary antibody and substitution of PBS; (b) preabsorption of the primary antibody with purified tenascin; (c) use of preimmune serum instead of the primary antisera. The distribution and pattern of tenascin staining were noted. For each time point an estimate was made of the percentage of the ductular circumference which stained positively for tenascin. Such estimates were derived by study
of five to ten ductules (of the particular size) from at least five different sections, of at least four different biopsy specimens. Selected sections were also stained using the above protocol, except with mouse monoclonal vimentin antibody (Lab Systems) and a sheep antimouse-FITC conjugate as secondary antibody (Serotec). Results General
All control sections exhibited no staining (Fig. 2C). Tenascin was present in each of the 19 normal breast specimens examined. Its distribution and appearance in the ductule were heterogeneous and showed marked variability during the menstrual cycle. In mammary blood vessels the presence of tenascin appeared to be a function of vessel location and calibre as well as stage of the menstrual cycle. Pretreatment of sections with 0.1 M or 0.5 M acetic acid or hyaluronidase did not alter the distribution of tenascin from that observed in untreated sections. Previous niorphological studies [28],particularly at the EM level, have revealed three layers surrounding the mammary ductule: the basement membrane proper, the sub-basement-membrane zone (reticular layer) and the delimiting layer of fibroblasts comprising the epithelial-stromal junction. Our associated study of the distribution of laminin and type IV collagen in the human mammary gland (Ferguson 1990, submitted) confirmed the presence of these three layers - namely: (i) a thin brightly staining basement membrane immediately adjacent to the basal surfaces of the epithelial and myoepithelial cells; (ii) a broad palely staining sub-basement-membrane zone; and (iii) an outermost brightly staining thin layer on the surface of the delimiting layer offihroblasts. The following descriptions use the above three terms to describe areas stained by the tenascin antibody. Normal breast. Week 1 of the menstrual cycle (days 1-7)
Tenascin was present as a thin bright discontinuous line in the sub-basement-membrane zone of the ductules, covering 50%-70% of the ductular circumference of larger ductules, less than 10% in some of the smallest ductules, and approximately 30%-50% of the circumference of mediumsized (Fig. 1A) ductules. At this stage the basement membrane was not distinguishable as a separate entity from the broad sub-basement-membrane zone. Tenascin was also apparent in the delimiting layer of fibroblasts as fine streaks amounting to 10%-30% of the total circumference. The intralobular stroma contained a few pale streaks of tenascin in some lobules but none in most lobules. In the interlobular stroma, fine streaks of tenascin were observed on the cell surfaces of fibroblast clusters (Fig. 2A). The intervening stroma was always negative. Large vessels and arterioles in the interlobular stroma had a continuous line of tenascin staining in the intima and fine closely spaced layers of tenascin in the media and adventitia. By contrast small vessels in the intralobular stroma did not contain tenascin. Tenascin was consistently absent from the epithelial cells and intercellular spaces of the ductules, the ductal lumina and the basement membranes of fat cells.
201
Normal breast. Week 2 of the menstrual cycle (days 8-15]
In the 2nd week tenascin appeared as a broad paler and more complete band in the -sub-basement-membiane zone covering 50%-70% of the circumference of most ductules (Fig. 1g).However tenascin had completely disappeared from the delimiting layer of fibroblasts and interlobular stromal fibroblasts. Tenascin remained absent from the in-
tralobular stromal fibroblasts and the small vessels of the intralobular stroma. Normal breast. Week 3 of the menstrual cycle (days 1G 2 2 ) By the 3rd week there was an overall increase in the amount of tenascin present in all compartments. The basement membrane, hitherto indistinguishable from the sub-base-
202
ment-membrane zone, was now seen as a well-delineated bright continuous thin line immediately adjacent to the base of the epithelial cells and covering 50%-70% of the ductule circumference (Fig. 1C). The sub-basement-membrane zone was broad, pale and continuous around the entire ductule.
The delimiting layer of fibroblasts contained a bright line of tenascin, which was continuous in the major ducts. In some ductules tenascin staining was also continuous but in others it was discontinuous covering only 50%-70% of the circumference.
203 Table 2. Summary of the major changes in tenascin distribution, in normal breast, during the 4 weeks of the menstrual cycle. See also Fig. 3 Week
1
Ductal Basement membrane Sub-basement-membrane + zone Delimiting layer of fibroblasts
+
2 -
++ -
3
4
+ +++ ++++ ++ +++ -
22
lntralobular stroma Fibroblasts Vessels
Day 1 ng/ml Progesterone
+-
-
+
-
-
+ +
++ ++
+
++ ++
Day 8
Interlobular stroma Fibroblasts Vessels
+
++
, small amount of tenascin present; ent; + + , large amount present; present; -, no tenascin present
+
++
++
+ + , moderate amount pres+ + + +, abundant tenascin ngfml Estrogen Day 1 5
The intralobular stroma contained a network of palestaining fibrils and the intralobular vessels had pale-staining walls. Tenascin staining reappeared on the surfaces of fibroblast clusters (Fig. 2A, B) in the interlobular stroma. The appearance of major vessels and fat cells was unchanged.
Fig. 3. Diagrammatic cross-section of a mammary ductule summarizing the changes in tenascin distribution in the basement membrane, sub-basement-membrane zone and delimiting layer of fibroblasts, with changing oestrogen and progesterone levels during the menstrual cycle. ---- Basement membrane; Subbasement membrane zone; - - - - Delimiting layer offibroblasts; -- Progesterone; -Estrogen
Normal breast. Week 4 of the menstrual cycle (days 23-30] In the 4th week of the cycle the amount of tenascin reached a maximum. All ductules had a wide, continuous, brightly staining band of tenascin in the sub-basement-membrane zone (Fig. 1D). The delimiting layer of fibroblasts had fine streaks of tenascin covering approximately 70%-100% of the circumference. The intralobular stroma contained large quantities of palely staining fibres, and small vessels in the intralobular stroma had tenascin-positive walls. Interlobular fibroblast clusters (Fig. 2B) were clearly positive with bright streaks of surface tenascin present (Fig. 2 A). Table 2 summarises the findings, and Fig. 3 correlates these in graphic form with the known hormonal profiles at different stages in the cycle. Interestingly, intralobular stromal vessels, but not interlobular stromal vessels, displayed cycle-dependent tenascin expression with positively staining walls in the 3rd and 4th weeks and no staining in weeks 1 and 2. Large vessels in the interlobular stroma expressed a constant amount and distribution of tenascin throughout the cycle. Carcinoma-in-situ The distribution and appearance of tenascin in the adjacent normal ductules and intralobular stroma were completely normal for the stage of the cycle in which the biopsy was removed. Each of the expanded ductules containing a carcinoma-in-situ lesion were surrounded by a thin continuous homogeneous band of tenascin (Fig. 4). The basement membrane and delimiting layer of fibroblasts were not delineated by tenascin staining. An unusual broad lucent gap separated the sub-basement-membrane zone from the intralobular stroma.
Fig. 4. Two carcinoma-in-situ lesions adjacent to normal ductules. Carcinoma-in-situ lesions arc surrounded by a homogeneous band of tenascin. Exterior to this, a lucent gap separates the lesions from the intralobular stroma. Adjacent normal ductules have tenascin present in the sub-basement-membrane zone and delimiting layer of fibroblasts
Infiltrating ductal carcinoma All tumours showed a striking increase in stromal tenascin. The basement membrane of well-differentiated tumours was fragmented and thin (as seen using laminin antibodies) but there was no evidence of any tenascin in the residual tumour
204
basement membrane. The sub-basement-membrane zone and the delimiting layers of fibroblasts were absent (Fig. 5A). Some parts of the epithelial stromal junction stained very intensely so that individual junctional cells appeared to be completely surrounded by tenascin (Fig. 5 B).
The remainder of the stroma contained large fibrils of tenascin surrounding small clusters and cords of unstained malignant cells (Fig. 5D). Many of these fibrils of tenascin appeared to be codistributed with stromal fibroblasts identified using vimentin antibody. Ribbons of tenascin-positive
205
material were occasionally seen between but not within the malignant epithelial cells of the ductules (Fig. 5C). Using antibodies to other basement membrane components (laminin, type IV collagen, heparan sulphate proteoglycan) and to blood clotting factor VIII, these ribbons were identified as stromal remnants, some of which contained tiny vessels (Ferguson 1990, submitted). In diffuse poorly differentiated tumours, ductules and their associated layers are absent. Specimens consist of clusters of cells in a fibrous stroma. Tenascin was heterogeneously distributed in this stroma. Where cells were sparse there was a dense fibrotic stroma, which stained intensely for tenascin. By contrast, areas of high celI density had little stroma, which contained fine streaks of tenascin. Discussion
This study presents the first description of the distribution and appearance of tenascin in the normal breast during the menstrual cycle and in carcinoma-in-situ. The appearance of tenascin in infiltrating ductal carcinoma is also reported. By contrast to the work of Mackie [22] who found that tenascin was restricted to embryonic and malignant mammary tissue, we found tenascin consistently in all normal breast samples examined. Moreover, the expression of tenascin appears to undergo major quantitative changes, in relation to cyclical secretions of ovarian steroids. The basement membrane remained histologically intact throughout the cycle, indicating that changes in tenascin distribution were related to changes in basement membrane composition rather than being secondary to its physical disruption. Previous studies [l, 41 have described laminin and type IV collagen, in the basement membranes of normal ductules as a single homogeneous band. By contrast, tenascin appeared in three distinct but adjacent layers at the base of the ductal epithelial cells, namely in the basement membrane, sub-basement-membrane zone and the delimiting layer of fibroblasts. Tenascin in each layer underwent changes throughout the cycle. Tenascin distribution in all the fibroblast populations (interlobular, intralobular and delimiting) varied according to the stage of the menstrual cycle. By contrast the amount and distribution of tenascin appear to be cyclically regulated only in the small intralobular vessels. The larger interlobular vessels showed no evidence of cyclical changes in tenascin. Thus the distribution and amount of tenascin in vessels appears to be a function of calibre as well as hormonal status. Moreover, tenascin was present in the basement membrane, media and adventitia of medium and large vessels, by contrast to molecules such as laminin and type IV collagen, which are restricted to the basement membrane. Role of tenascin in normal mammary tissue
A number of important events occur in the mammary epithelium at the time of maximal tenascin expression (weeks 4 and 3): (1) the epithelium becomes highly polarised and secretory; (2) there is a peak in epithelial mitosis on day 25(+2) [12]; (3) apoptosis (cell death) occurs on day 28 [12]; and (4) the estrogen levels achieve a second peak and progesterone levels a first peak in week 3 (they rapidly decline in week 4). The poor adhesive properties of tenascin [7] may reduce
cell-to-matrix adhesion and promote cell movement from the bulbous highly proliferative end-bud zones into the adjacent ductules, and permit easy passage of apoptotic cells from their luminal position to the periphery of the ductules where they are phagocytosed. A reduced ratio of fibronectin to tenascin is important in facilitating embryonic neural crest cell migration [I 1J and epithelial migration during wound healing [25].Tenascin is also known to have significant effects on cell shape. It promotes rounding of chick embryonic fibroblasts in culture, and antagonises the cellbinding activity of fibronectin [7]. Cyclical changes in tenascin may therefore not only facilitate ductule cell migration but also alter cell shape, polarization, secretory activity and differentiation in much the same way that tenascin regulates chondrocyte differentiation [23]. Little is known about the control of tenascin synthesis. It is stimulated by the addition of transforming growth factor$, (TGF-a,) to the media of chick embryonic fibroblasts in vitro and by an unknown factor in fetal calf serum [29]. Epidermal growth factor (EGF), TGFa, TGFP,, platelet-derived growth factor (PDGF) and bovine fibroblast growth factor (FGF) either alone, or in various combinations, all stimulate tenascin biosynthesis in embryonic mouse palatal mesenchymal cells in vitro [lo, 13, 14, 32, 321. The target cells for estrogen in the breast appear to be the epithelia, which contain estrogen receptors. Estrogenic effects are thought to be mediated by the production by the epithelium of growth factors e.g. TGFa, TGFP,, which then act in an autocrine or paracrine fashion on adjacent epithelial and stromal cells. We speculate that the first estrogen peak in week 2 and the second estrogen peak and first progesterone peak in week 3 (Fig. 3) alter the growth factor profile, e.g. increase the synthesis of TGFa [9] and so stimulate tenascin synthesis. Moreover, the falling levels of estrogen in the 4th week stimulate TGFP, production [20], which would further stimulate tenascin synthesis. Therefore if in human mammary glands the epithelium induces tenascin production by stromal fibroblasts, as it does in the embryonic mouse [18], this may be mediated by a variety of growth factors. These epithelial-mesenchymal interactions are reciprocal with respect to target tissue and time, i.e. the epithelium produces growth factors (under hormonal control), which affect tenascin production by stromal cells, tenascin in turn affecting epithelial morphology and function. Such temporal reciprocity is a common feature of most epithelial mesenchymal interactions [32] and allows a high degree of local control and integration of events in adjacent cells of a tissue. Tenascin in tumours
The presence of tenascin in the normal mammary gland as well as in malignant tumours discounts tenascin as an all-or-none marker of mammary malignancy [22]. Moreover our demonstration that the quantity and distribution of tenascin are cyclically regulated emphasises the necessity of having accurate data on the period of the menstrual cycle when any biopsy is removed for histopathological examination. Failure to recognise that, at some stages in the cycle, tenascin normally has a patchy distribution with a discontinuous basement membrane layer, may lead to misdiagnosis. Striking differences in the distribution and quantity of tenascin between the normal mammary gland, carcinoma-
206 in-situ, a n d infiltrating ductal carcinomas m a y be of diagnostic value, particularly in distinguishing carcinoma-insitu from microinvasive infiltrating ductal carcinoma i n postmenopausal women. Attention to biopsy time i n premenopausal women, in relationship t o the menstrual cycle, may be important in maximising t h e diagnostic accuracy in distinguishing carcinoma-in-situ from microinvasive infiltrating ductal carcinomas.
Role of tenascin in malignancy
A striking feature of the well-differentiated infiltrating carcinoma was the presence a t t h e epithelial-stromal junction of large a m o u n t s of tenascin, which appeared t o completely surround the invading epithelial cells and adjacent fibroblasts. Malignant mammary epithelium produces excessive quantities of numerous growth factors including T G F a , TGFP [31], insulin like growth factor I ( I G F I ) [I71 a n d PDGF [30]. These may a c t on the adjacent delimiting and intralobular stromal fibroblasts to increase the production of extracellular matrix molecules including tenascin. These in t u r n could exert a mitogenic effect on fibroblasts and tumour cells [6]. The tenascin molecule contains 13 EGFlike repeats in its structure - these could mimic the mitogenic effects of E G F . Increased extracellular matrix production would also contribute t o the desmoplasia so characteristic of breast cancer. T h e excessive tenascin deposits f o u n d a t t h e t u m o u r epithelial-stromal junction may n o t only increase t u m o u r cell proliferation b u t also promote t u m o u r cell rounding and decreased adhesiveness, so facilitating detachment o f tumour cells from the epithelial sheet and invasion of the stroma. T h u s if malignant epithelial cells c a n induce tenascin production by adjacent fibroblasts [18] this may prom o t e extensive invasion and permit metastasis. Many other extracellular matrix molecules, including fibronectin [I] and collagen types I, 111 and V [26, 27, 31 a r e present in greatly increased quantities in malignant stroma. T h e increased production of at least one of these - fibronectin [8] - has been shown t o correlate with low metastatic potential o f the tumour. Tenascin has been found in the stroma of metastases f r o m a malignant mammary cell line GmT-L [18] and may thus be associated with increased metastatic potential. The growth factor profile of some tumours could conceivably alter the tenascin-tofibronectin ratio. A relative excess of tenascin might facilitate tumour invasion, whereas a relative excess of fibronectin might result in t u m o u r containment. Exogenous manipulation of the tenascin-to-fibronectin ratio m a y have therapeutic potential. T h e changing cellular effects of different ratios of extracellular matrix molecules emphasises t h e necessity of documenting the distribution and quantity o f many extracellular matrix molecules i n the s a m e biopsy specimens: such studies are underway using t h e biopsies studied here f o r tenascin. Acknowledgements. We are extremcly grateful to Dr. Ruth ChiquetEhrismann, Friedrich Micscher lnstitut in Basel for generously giving us the polyclonal antibody to tenascin.
References 1. Alon Y, Horowitz AT, Biran S, Weiss DW, Doljanski F (1986)
Immunofluorescent characterization of fibronectin, laminin
and keratin in normal and neoplastic human mammary epithelial cells in culturc and in breast tissue sections. Jnt J Tiss React VIII (5):401-410 2. Aufderheide E, Eckblom P (1988) Tenascin during gut development. Appearance in the mesenchyme, shift in molccular forms and dependance on epithelial-mesenchymal interactions. J Cell Biol 105:599-608 3. Barsky S, Rao CN, Grotendorst C , Liotta LA (1982) Increased content of type V collagen in desmoplasia of human breast carcinoma. Am J Path 108:276-283 4. Barsky S, Siegal GP, Jannotta F, Liotta LA (1983) Loss of basement membrane components by invasive tumours but not by their benign counterparts. Lab Invest 49: 140-147 5. Chiquet M, Fambrough DM (1984) Chick myotendinous antigen. 1. A monoclonal antibody as a marker for tendon and muscle morphogenesis. J Cell Biol 98: 1926-1936 6. Chiquet-Ehrismann R, Mackie EJ, Pearson CA, Sakakura T (1 986) Tenascin: an extracellular matrix protein involved in tissue interactions during fetal development and oncogenesis. Cell 47: 131-139 7. Chiquet-Ehrismann R, Kalla P, Pearson CA, Beck K, Chiquet M (1988) Tenascin interferes with fibronectin action. Cell 531383-390 8. Christensen L, Nielsen M , Andersen J, Clemniensen I (1988) Stromal fibronectin staining pattern and metastasizing ability of human breast carcinoma. Cancer Res 48: 6221-6233 9. Dickson RB, Bates SE, McManaway ME, Lippman ME (1986) Characterization of estrogen responsive transforming activity in human breast cancer cell lines. Cancer Res 46 : 1707-1 71 3 10. Dixon MJ, Ferguson MWJ (1990) The effects of epidermal growth factor, transforming growth factors alpha and beta and platelet-derived growth factor on murine palatal shelves in organ culture. Arch Oral Biol (in press) 11. Epperlein HH, Halfter W, Tucker RP (1988) The distribution of fibronectin and tenascin along migratory pathways of the neural crest in the trunk of amphibian embryos. Development 103r743-756 2. Ferguson DJP, Anderson TJ (1981) Morphological evaluation of cell turnover in relation to the menstrual cycle in the ‘resting’ human breast. Br J Cancer 44:177-181 3. Ferguson MJW (1987) “Palate development”. Mechanisms and malformations (Conway Rcvicw Lecture). Irish J Med Sci 156:309-315 4. Ferguson MJW (1988) Palate development. Devclopment 103 [SUPPI]41-60 15. Fyfe DM, Ferguson MWJ, Chiquet-Ehrismann R (1988) Immunocytochemical localisation of tenascin during the development of scleral papillae and scleral ossicles in the embryonic chick. J Anat 159:117-127 16. Gulcher JR, Nies DE, Marton LS, StePansson K (1989) An alternatively spliced region of the human hexabrachion contains a repeat of potential N-glycosylation sites. Proc Natl Acad Sci USA 86:1588-1592 17. Huff KK, Kaufman D, Gabbay KH, Spencer EM, Lippman ME, Dickson RB (1986) Secretion of an insulin like growth factor-1-related protein by human breast cells. Cancer Res 46 ~4613-4619 18. Inaguma Y, Kusakabe M, Mackie EJ, Pearson CA, ChiquetEhrismann R, Sakdkura T (1988) Epithelial induction of stroma1 tenascin in the mouse mammary gland from embryogenesis to carcinogcnesis. Dev Biol 128:245-255 19. Johnson GD, Davidson RS, McNamee KC, Russell G, Goodwin D, Holborow EJ (1982) Fading of immunofluorescence during microscopy: a study of the phenomenon and its remedy. J Immunol Meth 55:231-242 20. Knabbe C, Lippman M, Wakefield LM, Flanders KC, Kasid A, Derynck R, Dickson R (1987) Evidence that transforming growth factor /Iis a hormonally regulated negative growth factor in human breast cancer cells. Cell 48 :41 7 4 2 8 21. Lightner VA, Gumkowski F, Bigner DD, Erickson HP (1989) Tenascinihexabrachion localisation in human skin : biochemical
207 identification and localisation by light and electron microscopy. J Cell Biol 108 :2483-249 3 22. Mackie EJ, Chiquet-Ehrismann R, Pearson CA, Inaguma Y, Taya K, Kawarada Y, Sakakura T (1987) Tenascin is a stromal marker for epithelial malignancy in the mammary gland. Proc Natl Acad Sci USA 84:4621-4625 23. Mackie EJ, Thesleff I, Chiquet-Ehrismann R (1987) Tenascin is associated with chondrogenic and osteogenic differentiation in vivo and promotes chondrogenesis in vitro. J Cell Biol 105:2569-2579 24. Mackie EJ, Tucker RP, Halfter W, Chiquet-Ehrismann R, Epperlein HH (1988) The distribution of tenascin coincides with pathways of neural crest cell migration. Development 1021237-250 25. Mackie EJ, Halfter W, Liverani D (1988) Induction of tenascin in healing wounds. J Cell Biol 107:2757-2767 26. Mahfouz SM, Chevallier M, Grimaud JA (1987) Distribution of the major connective matrix components of the stromal reaction in breast carcinoma. An immunohistochemical study. Cell Mol Biol 33(4):453-467 27. Minafra P, Luparello C, Sciarrino S, Tomasino RM, Minafra S (1 985) Quantitative determination of collagen types present in the ductal infiltrating carcinoma of human mammary gland. Cell Biol Int Rep 9 No. 3:291-296
28. Ozello L (1970) Epithelial-stromal junction of normal and dysplastic mammary glands. Cancer 25: 586600 29. Pearson CA, Pearson D, Shibahara S, Hofsteenge J, ChiquetEhrismann R (1988) Tenascin: cDNA cloning and induction by TGFB. EMBO J 7(10):2677-2981 30. Rozengurt E, Sinnett-Smith J, Taylor-Papadimitriou J (1985) Production of PDGF-like growth factor by breast cancer cell lines. Int J Canc 36: 147-252 31. Salomon DS, Zwiebel JA, Ban0 M, Losonezy I, Fechnel P, Kidwell WR (1984) Presence of transforming growth factors in human breast cancer cells. Cancer Res 44:4069-4077 32. Sharpe PM, Ferguson MWJ (1988) Mesenchymal influences on epithelial differentiation in developing systems. J Cell Sci [SUPPI]10: 195-230 33. Sharpe PM, Foreman DM, Carette MJM, Schor SL, Ferguson MWJ (1990) The effects of transforming growth factor-beta (TGFP,) on the proliferation and biosynthetic activity of mouse embryonic palatal mesenchyme cells in vitro. J Cell Physiol (in press)
Accepted in revised form October 33, 1989
Differentiation (1990) 42: 199-207
Ontogeny and Neoplasia 0 Springer-Verlag 1990
Tenascin distribution in the normal human breast is altered during the menstrual cycle and in carcinoma Janice E. Ferguson', Ana M. Schor', Anthony Howell', and Mark W.J. Ferguson2* CRC Department of Medical Oncology, The Christie Hospital and Holt Radium Institute, Wilmslow Road, Withington, Manchester M20 9BX, England Department of Cell & Structural Biology, University of Manchester, Coupland I11 Building, Manchester MI 3 9PL, England
Abstract. Tenascin is a novel extracellular matrix glycoprotein which appears to have a major role in tissue development [6]. Previous studies have stated that tenascin is absent from the normal human 1221, rat [6] and mouse [18] breast, its distribution being restricted to embryonic and malignant mammary tissues [22]. No previous studies have investigated tenascin distribution as a function of the normal menstrual cycle. Therefore this study addresses the cyclical appearance of tenascin in the normal breast and associated changes in distribution in preinvasive cancer (carcinoma-insitu) and invasive infiltrating ductal carcinoma. Tenascin is present in the normal human adult mammary gland, principally in the basement membrane, sub-basement-membrane zone and delimiting layer of fibroblasts around the ductules. Both the distribution and quantity of tenascin change during the menstrual cycle. In carcinoma-in-situ (preinvasive cancer) tenascin is present in the attenuated basement membrane/sub-basement-membranezone around the expanded ductules and in small amounts in the stroma. In infiltrating ductal carcinoma, tenascin is absent from the remnants of the basement membrane and sub-basement-membrane zone but greatly increased in the adjacent intralobular and interlobular stroma. Therefore, if tenascin is used as a basement membrane/sub-basement-membrane marker for distinguishing carcinoma-in-situ from invasive ductal carcinoma, the time of the menstrual cycle is of importance in interpreting the biopsy appearance. This study suggests that the optimal time for biopsy is between weeks 3 and 4 of the cycle, to avoid confusion between the normal low levels of tenascin (due to hormonal status) and those due to microinvasive disease. That tenascin expression in the normal breast is cyclically regulated suggests that this molecule may act as a mediator in the hormonal control of mammary epithelial cell behaviour.
Introduction
Tenascin is a novel extracellular matrix glycoprotein with a disulfide-bonded subunit of 150-320 kDa (5, 16, 211. It is a hexabrachion consisting of two pairs of three arms connected to a central globule and contains 13 epidermal growth factor (EGF)-like repeats [29]. Tenascin has a restricted distribution in the mesenchymal condensations ar-
* To whom offprint requests should be sent
ound budding epithelia of developing organs such as hair follicles, tooth germs, mammary anlagen 151, scleral ossicles [I51 and the palate [14] in addition to the perichondral and tendinous anlagen [5].Tenascin has been found in the mesenchyme adjacent to differentiated gut epithelium in the embryonic mouse and in the stroma adjacent to migration pathways of the continually renewed intestinal villous epithelia in the adult mouse 121. Tenascin also lines the future migratory pathways of neural crest cells in quail, rat, 1241 and amphibian embryos [I I]. Tenascin antagonises the cell binding activity of fibronectin 171 and thus may be important in stimulating cell migration and cell rounding. Tenascin has been found in the mesenchyme beneath the epithelial rudiment in the fetal mammary gland, and in the stroma of both carcinogen-induced mammary carcinomas of the rat, and spontaneous human mammary adenocarcinomas [6, 221. It is apparently absent from benign neoplasms, fibrocystic disease and normal adult mammary tissue in both human and rat [6, 221. Tenascin has not been described in normal human breast in relation to the menstrual cycle, nor in carcinoma-in-situ lesions. The above distribution of tenascin led Chiquet-Ehrismann et al. 161 to suggest that tenascin was an oncofetal molecule, important in the development of the normal mammary gland and in the early stages of mammary carcinogenesis. They also suggested that the appearance of stroma1 tenascin may be a marker of epithelial malignancy and therefore have important diagnostic and therapeutic implications. This study first investigated the distribution of tenascin in the normal human mammary gland at different points in the menstrual cycle. Tenascin distribution in infiltrating ductal carcinoma and carcinoma-in-situ was then studied in order to investigate whether these conditions could be distinguished from each other and whether the pattern and distribution of tenascin observed might suggest a role for tenascin in the critical stages between preinvasive (carcinoma-in-situ) and invasive infiltrating ductal carcinoma. Methods
Nineteen normal mammary biopsies were obtained from the periphery of the breast in young women undergoing operation for benign fibroadenoma. The day of the menstrual cycle was recorded at biopsy (Table 1). Tumour specimens were obtained from surgical mastectomies of infiltrating ductal carcinoma (n = 5 ) and carcino-
200
Table 1. Distribution of specimens used in the study of the normal breast at different stages in the menstrual cycle, including the age of the patient, in years, day of the cycle and the operation during which the normal specimen was removed Age
Day of cycle
Operation
20 24 20 22 25 26 18 21 22 16 19 22 19 22 21 23 19 25 21
2 3 5 6 6 8 9 9 12 17 20 21 22 22 25 26 26 28 28
Fibroadenoma Normal/no lesion Fi broadenoma Fibroadenoma Fi broadcnoma Fibroadenoma Fibroadenoma Fibroadenoma Fibroadenoma Fibroadenoma Fibroadenoma Fibroadenoma Fibroadenoma Fibroadenoma Fibroadenoma Lipoma Fibroadenoma Fibroadenoma Lymph node
ma in situ ( n = 2 ) in older women (aged 37-51 years). The tissue was rapidly frozen in liquid nitrogen and stored at -70" C for up to 1 year. Specimens were embedded in OCT (Miles Inc USA) and sectioned at 5-9 pm using a Leitz Cryostat. The sections were mounted on clean glass slides, air-dried and fixed in cold acetone for 10 min. They are either used immediately or stored for up to 1 month at - 70" C. In order to exclude possible masking of epitopes (in cases of negative results) the sections received one of the following pretreatments: (a) 0.1 M acetic acid, (b) 0.5 M acetic acid, (c) sheep testicular hyaluronidase type I11 (Sigma) 1 mg/50 ml in 1 M sodium acetate buffer adjusted to pH 5, (d) no pretreatment for 30 min at room temperature. They were then incubated in a humidified chamber with polyclonal antiserum raised in rabbit against the chick tenascin molecule, at a dilution of 1 : 100 (for details of antibody preparation and characterization see [6] for 1 h and then washed for three 5-min periods with phosphatebuffered saline (PBS). After applying fluorescein isothiocyanate (FITC) conjugated sheep anti-rabbit as secondary antibody (Serotec) and washing with PBS, sections were mounted in glycerol containing PBS and 1,4 diazobicyclo 2,2,2, octane [19] (DABCO) to inhibit fading. Examination for immunofluorescence was carried out using a Leitz Orthoplan incident epifluorescence microscope (equipped with a Ploemopak filter) and photographs taken using Ektachrome ASA160 colour tungsten film from which black and white prints were made. Control staining was performed, as above, but with the following exceptions: (a) omission of the primary antibody and substitution of PBS; (b) preabsorption of the primary antibody with purified tenascin; (c) use of preimmune serum instead of the primary antisera. The distribution and pattern of tenascin staining were noted. For each time point an estimate was made of the percentage of the ductular circumference which stained positively for tenascin. Such estimates were derived by study
of five to ten ductules (of the particular size) from at least five different sections, of at least four different biopsy specimens. Selected sections were also stained using the above protocol, except with mouse monoclonal vimentin antibody (Lab Systems) and a sheep antimouse-FITC conjugate as secondary antibody (Serotec). Results General
All control sections exhibited no staining (Fig. 2C). Tenascin was present in each of the 19 normal breast specimens examined. Its distribution and appearance in the ductule were heterogeneous and showed marked variability during the menstrual cycle. In mammary blood vessels the presence of tenascin appeared to be a function of vessel location and calibre as well as stage of the menstrual cycle. Pretreatment of sections with 0.1 M or 0.5 M acetic acid or hyaluronidase did not alter the distribution of tenascin from that observed in untreated sections. Previous niorphological studies [28],particularly at the EM level, have revealed three layers surrounding the mammary ductule: the basement membrane proper, the sub-basement-membrane zone (reticular layer) and the delimiting layer of fibroblasts comprising the epithelial-stromal junction. Our associated study of the distribution of laminin and type IV collagen in the human mammary gland (Ferguson 1990, submitted) confirmed the presence of these three layers - namely: (i) a thin brightly staining basement membrane immediately adjacent to the basal surfaces of the epithelial and myoepithelial cells; (ii) a broad palely staining sub-basement-membrane zone; and (iii) an outermost brightly staining thin layer on the surface of the delimiting layer offihroblasts. The following descriptions use the above three terms to describe areas stained by the tenascin antibody. Normal breast. Week 1 of the menstrual cycle (days 1-7)
Tenascin was present as a thin bright discontinuous line in the sub-basement-membrane zone of the ductules, covering 50%-70% of the ductular circumference of larger ductules, less than 10% in some of the smallest ductules, and approximately 30%-50% of the circumference of mediumsized (Fig. 1A) ductules. At this stage the basement membrane was not distinguishable as a separate entity from the broad sub-basement-membrane zone. Tenascin was also apparent in the delimiting layer of fibroblasts as fine streaks amounting to 10%-30% of the total circumference. The intralobular stroma contained a few pale streaks of tenascin in some lobules but none in most lobules. In the interlobular stroma, fine streaks of tenascin were observed on the cell surfaces of fibroblast clusters (Fig. 2A). The intervening stroma was always negative. Large vessels and arterioles in the interlobular stroma had a continuous line of tenascin staining in the intima and fine closely spaced layers of tenascin in the media and adventitia. By contrast small vessels in the intralobular stroma did not contain tenascin. Tenascin was consistently absent from the epithelial cells and intercellular spaces of the ductules, the ductal lumina and the basement membranes of fat cells.
201
Normal breast. Week 2 of the menstrual cycle (days 8-15]
In the 2nd week tenascin appeared as a broad paler and more complete band in the -sub-basement-membiane zone covering 50%-70% of the circumference of most ductules (Fig. 1g).However tenascin had completely disappeared from the delimiting layer of fibroblasts and interlobular stromal fibroblasts. Tenascin remained absent from the in-
tralobular stromal fibroblasts and the small vessels of the intralobular stroma. Normal breast. Week 3 of the menstrual cycle (days 1G 2 2 ) By the 3rd week there was an overall increase in the amount of tenascin present in all compartments. The basement membrane, hitherto indistinguishable from the sub-base-
202
ment-membrane zone, was now seen as a well-delineated bright continuous thin line immediately adjacent to the base of the epithelial cells and covering 50%-70% of the ductule circumference (Fig. 1C). The sub-basement-membrane zone was broad, pale and continuous around the entire ductule.
The delimiting layer of fibroblasts contained a bright line of tenascin, which was continuous in the major ducts. In some ductules tenascin staining was also continuous but in others it was discontinuous covering only 50%-70% of the circumference.
203 Table 2. Summary of the major changes in tenascin distribution, in normal breast, during the 4 weeks of the menstrual cycle. See also Fig. 3 Week
1
Ductal Basement membrane Sub-basement-membrane + zone Delimiting layer of fibroblasts
+
2 -
++ -
3
4
+ +++ ++++ ++ +++ -
22
lntralobular stroma Fibroblasts Vessels
Day 1 ng/ml Progesterone
+-
-
+
-
-
+ +
++ ++
+
++ ++
Day 8
Interlobular stroma Fibroblasts Vessels
+
++
, small amount of tenascin present; ent; + + , large amount present; present; -, no tenascin present
+
++
++
+ + , moderate amount pres+ + + +, abundant tenascin ngfml Estrogen Day 1 5
The intralobular stroma contained a network of palestaining fibrils and the intralobular vessels had pale-staining walls. Tenascin staining reappeared on the surfaces of fibroblast clusters (Fig. 2A, B) in the interlobular stroma. The appearance of major vessels and fat cells was unchanged.
Fig. 3. Diagrammatic cross-section of a mammary ductule summarizing the changes in tenascin distribution in the basement membrane, sub-basement-membrane zone and delimiting layer of fibroblasts, with changing oestrogen and progesterone levels during the menstrual cycle. ---- Basement membrane; Subbasement membrane zone; - - - - Delimiting layer offibroblasts; -- Progesterone; -Estrogen
Normal breast. Week 4 of the menstrual cycle (days 23-30] In the 4th week of the cycle the amount of tenascin reached a maximum. All ductules had a wide, continuous, brightly staining band of tenascin in the sub-basement-membrane zone (Fig. 1D). The delimiting layer of fibroblasts had fine streaks of tenascin covering approximately 70%-100% of the circumference. The intralobular stroma contained large quantities of palely staining fibres, and small vessels in the intralobular stroma had tenascin-positive walls. Interlobular fibroblast clusters (Fig. 2B) were clearly positive with bright streaks of surface tenascin present (Fig. 2 A). Table 2 summarises the findings, and Fig. 3 correlates these in graphic form with the known hormonal profiles at different stages in the cycle. Interestingly, intralobular stromal vessels, but not interlobular stromal vessels, displayed cycle-dependent tenascin expression with positively staining walls in the 3rd and 4th weeks and no staining in weeks 1 and 2. Large vessels in the interlobular stroma expressed a constant amount and distribution of tenascin throughout the cycle. Carcinoma-in-situ The distribution and appearance of tenascin in the adjacent normal ductules and intralobular stroma were completely normal for the stage of the cycle in which the biopsy was removed. Each of the expanded ductules containing a carcinoma-in-situ lesion were surrounded by a thin continuous homogeneous band of tenascin (Fig. 4). The basement membrane and delimiting layer of fibroblasts were not delineated by tenascin staining. An unusual broad lucent gap separated the sub-basement-membrane zone from the intralobular stroma.
Fig. 4. Two carcinoma-in-situ lesions adjacent to normal ductules. Carcinoma-in-situ lesions arc surrounded by a homogeneous band of tenascin. Exterior to this, a lucent gap separates the lesions from the intralobular stroma. Adjacent normal ductules have tenascin present in the sub-basement-membrane zone and delimiting layer of fibroblasts
Infiltrating ductal carcinoma All tumours showed a striking increase in stromal tenascin. The basement membrane of well-differentiated tumours was fragmented and thin (as seen using laminin antibodies) but there was no evidence of any tenascin in the residual tumour
204
basement membrane. The sub-basement-membrane zone and the delimiting layers of fibroblasts were absent (Fig. 5A). Some parts of the epithelial stromal junction stained very intensely so that individual junctional cells appeared to be completely surrounded by tenascin (Fig. 5 B).
The remainder of the stroma contained large fibrils of tenascin surrounding small clusters and cords of unstained malignant cells (Fig. 5D). Many of these fibrils of tenascin appeared to be codistributed with stromal fibroblasts identified using vimentin antibody. Ribbons of tenascin-positive
205
material were occasionally seen between but not within the malignant epithelial cells of the ductules (Fig. 5C). Using antibodies to other basement membrane components (laminin, type IV collagen, heparan sulphate proteoglycan) and to blood clotting factor VIII, these ribbons were identified as stromal remnants, some of which contained tiny vessels (Ferguson 1990, submitted). In diffuse poorly differentiated tumours, ductules and their associated layers are absent. Specimens consist of clusters of cells in a fibrous stroma. Tenascin was heterogeneously distributed in this stroma. Where cells were sparse there was a dense fibrotic stroma, which stained intensely for tenascin. By contrast, areas of high celI density had little stroma, which contained fine streaks of tenascin. Discussion
This study presents the first description of the distribution and appearance of tenascin in the normal breast during the menstrual cycle and in carcinoma-in-situ. The appearance of tenascin in infiltrating ductal carcinoma is also reported. By contrast to the work of Mackie [22] who found that tenascin was restricted to embryonic and malignant mammary tissue, we found tenascin consistently in all normal breast samples examined. Moreover, the expression of tenascin appears to undergo major quantitative changes, in relation to cyclical secretions of ovarian steroids. The basement membrane remained histologically intact throughout the cycle, indicating that changes in tenascin distribution were related to changes in basement membrane composition rather than being secondary to its physical disruption. Previous studies [l, 41 have described laminin and type IV collagen, in the basement membranes of normal ductules as a single homogeneous band. By contrast, tenascin appeared in three distinct but adjacent layers at the base of the ductal epithelial cells, namely in the basement membrane, sub-basement-membrane zone and the delimiting layer of fibroblasts. Tenascin in each layer underwent changes throughout the cycle. Tenascin distribution in all the fibroblast populations (interlobular, intralobular and delimiting) varied according to the stage of the menstrual cycle. By contrast the amount and distribution of tenascin appear to be cyclically regulated only in the small intralobular vessels. The larger interlobular vessels showed no evidence of cyclical changes in tenascin. Thus the distribution and amount of tenascin in vessels appears to be a function of calibre as well as hormonal status. Moreover, tenascin was present in the basement membrane, media and adventitia of medium and large vessels, by contrast to molecules such as laminin and type IV collagen, which are restricted to the basement membrane. Role of tenascin in normal mammary tissue
A number of important events occur in the mammary epithelium at the time of maximal tenascin expression (weeks 4 and 3): (1) the epithelium becomes highly polarised and secretory; (2) there is a peak in epithelial mitosis on day 25(+2) [12]; (3) apoptosis (cell death) occurs on day 28 [12]; and (4) the estrogen levels achieve a second peak and progesterone levels a first peak in week 3 (they rapidly decline in week 4). The poor adhesive properties of tenascin [7] may reduce
cell-to-matrix adhesion and promote cell movement from the bulbous highly proliferative end-bud zones into the adjacent ductules, and permit easy passage of apoptotic cells from their luminal position to the periphery of the ductules where they are phagocytosed. A reduced ratio of fibronectin to tenascin is important in facilitating embryonic neural crest cell migration [I 1J and epithelial migration during wound healing [25].Tenascin is also known to have significant effects on cell shape. It promotes rounding of chick embryonic fibroblasts in culture, and antagonises the cellbinding activity of fibronectin [7]. Cyclical changes in tenascin may therefore not only facilitate ductule cell migration but also alter cell shape, polarization, secretory activity and differentiation in much the same way that tenascin regulates chondrocyte differentiation [23]. Little is known about the control of tenascin synthesis. It is stimulated by the addition of transforming growth factor$, (TGF-a,) to the media of chick embryonic fibroblasts in vitro and by an unknown factor in fetal calf serum [29]. Epidermal growth factor (EGF), TGFa, TGFP,, platelet-derived growth factor (PDGF) and bovine fibroblast growth factor (FGF) either alone, or in various combinations, all stimulate tenascin biosynthesis in embryonic mouse palatal mesenchymal cells in vitro [lo, 13, 14, 32, 321. The target cells for estrogen in the breast appear to be the epithelia, which contain estrogen receptors. Estrogenic effects are thought to be mediated by the production by the epithelium of growth factors e.g. TGFa, TGFP,, which then act in an autocrine or paracrine fashion on adjacent epithelial and stromal cells. We speculate that the first estrogen peak in week 2 and the second estrogen peak and first progesterone peak in week 3 (Fig. 3) alter the growth factor profile, e.g. increase the synthesis of TGFa [9] and so stimulate tenascin synthesis. Moreover, the falling levels of estrogen in the 4th week stimulate TGFP, production [20], which would further stimulate tenascin synthesis. Therefore if in human mammary glands the epithelium induces tenascin production by stromal fibroblasts, as it does in the embryonic mouse [18], this may be mediated by a variety of growth factors. These epithelial-mesenchymal interactions are reciprocal with respect to target tissue and time, i.e. the epithelium produces growth factors (under hormonal control), which affect tenascin production by stromal cells, tenascin in turn affecting epithelial morphology and function. Such temporal reciprocity is a common feature of most epithelial mesenchymal interactions [32] and allows a high degree of local control and integration of events in adjacent cells of a tissue. Tenascin in tumours
The presence of tenascin in the normal mammary gland as well as in malignant tumours discounts tenascin as an all-or-none marker of mammary malignancy [22]. Moreover our demonstration that the quantity and distribution of tenascin are cyclically regulated emphasises the necessity of having accurate data on the period of the menstrual cycle when any biopsy is removed for histopathological examination. Failure to recognise that, at some stages in the cycle, tenascin normally has a patchy distribution with a discontinuous basement membrane layer, may lead to misdiagnosis. Striking differences in the distribution and quantity of tenascin between the normal mammary gland, carcinoma-
206 in-situ, a n d infiltrating ductal carcinomas m a y be of diagnostic value, particularly in distinguishing carcinoma-insitu from microinvasive infiltrating ductal carcinoma i n postmenopausal women. Attention to biopsy time i n premenopausal women, in relationship t o the menstrual cycle, may be important in maximising t h e diagnostic accuracy in distinguishing carcinoma-in-situ from microinvasive infiltrating ductal carcinomas.
Role of tenascin in malignancy
A striking feature of the well-differentiated infiltrating carcinoma was the presence a t t h e epithelial-stromal junction of large a m o u n t s of tenascin, which appeared t o completely surround the invading epithelial cells and adjacent fibroblasts. Malignant mammary epithelium produces excessive quantities of numerous growth factors including T G F a , TGFP [31], insulin like growth factor I ( I G F I ) [I71 a n d PDGF [30]. These may a c t on the adjacent delimiting and intralobular stromal fibroblasts to increase the production of extracellular matrix molecules including tenascin. These in t u r n could exert a mitogenic effect on fibroblasts and tumour cells [6]. The tenascin molecule contains 13 EGFlike repeats in its structure - these could mimic the mitogenic effects of E G F . Increased extracellular matrix production would also contribute t o the desmoplasia so characteristic of breast cancer. T h e excessive tenascin deposits f o u n d a t t h e t u m o u r epithelial-stromal junction may n o t only increase t u m o u r cell proliferation b u t also promote t u m o u r cell rounding and decreased adhesiveness, so facilitating detachment o f tumour cells from the epithelial sheet and invasion of the stroma. T h u s if malignant epithelial cells c a n induce tenascin production by adjacent fibroblasts [18] this may prom o t e extensive invasion and permit metastasis. Many other extracellular matrix molecules, including fibronectin [I] and collagen types I, 111 and V [26, 27, 31 a r e present in greatly increased quantities in malignant stroma. T h e increased production of at least one of these - fibronectin [8] - has been shown t o correlate with low metastatic potential o f the tumour. Tenascin has been found in the stroma of metastases f r o m a malignant mammary cell line GmT-L [18] and may thus be associated with increased metastatic potential. The growth factor profile of some tumours could conceivably alter the tenascin-tofibronectin ratio. A relative excess of tenascin might facilitate tumour invasion, whereas a relative excess of fibronectin might result in t u m o u r containment. Exogenous manipulation of the tenascin-to-fibronectin ratio m a y have therapeutic potential. T h e changing cellular effects of different ratios of extracellular matrix molecules emphasises t h e necessity of documenting the distribution and quantity o f many extracellular matrix molecules i n the s a m e biopsy specimens: such studies are underway using t h e biopsies studied here f o r tenascin. Acknowledgements. We are extremcly grateful to Dr. Ruth ChiquetEhrismann, Friedrich Micscher lnstitut in Basel for generously giving us the polyclonal antibody to tenascin.
References 1. Alon Y, Horowitz AT, Biran S, Weiss DW, Doljanski F (1986)
Immunofluorescent characterization of fibronectin, laminin
and keratin in normal and neoplastic human mammary epithelial cells in culturc and in breast tissue sections. Jnt J Tiss React VIII (5):401-410 2. Aufderheide E, Eckblom P (1988) Tenascin during gut development. Appearance in the mesenchyme, shift in molccular forms and dependance on epithelial-mesenchymal interactions. J Cell Biol 105:599-608 3. Barsky S, Rao CN, Grotendorst C , Liotta LA (1982) Increased content of type V collagen in desmoplasia of human breast carcinoma. Am J Path 108:276-283 4. Barsky S, Siegal GP, Jannotta F, Liotta LA (1983) Loss of basement membrane components by invasive tumours but not by their benign counterparts. Lab Invest 49: 140-147 5. Chiquet M, Fambrough DM (1984) Chick myotendinous antigen. 1. A monoclonal antibody as a marker for tendon and muscle morphogenesis. J Cell Biol 98: 1926-1936 6. Chiquet-Ehrismann R, Mackie EJ, Pearson CA, Sakakura T (1 986) Tenascin: an extracellular matrix protein involved in tissue interactions during fetal development and oncogenesis. Cell 47: 131-139 7. Chiquet-Ehrismann R, Kalla P, Pearson CA, Beck K, Chiquet M (1988) Tenascin interferes with fibronectin action. Cell 531383-390 8. Christensen L, Nielsen M , Andersen J, Clemniensen I (1988) Stromal fibronectin staining pattern and metastasizing ability of human breast carcinoma. Cancer Res 48: 6221-6233 9. Dickson RB, Bates SE, McManaway ME, Lippman ME (1986) Characterization of estrogen responsive transforming activity in human breast cancer cell lines. Cancer Res 46 : 1707-1 71 3 10. Dixon MJ, Ferguson MWJ (1990) The effects of epidermal growth factor, transforming growth factors alpha and beta and platelet-derived growth factor on murine palatal shelves in organ culture. Arch Oral Biol (in press) 11. Epperlein HH, Halfter W, Tucker RP (1988) The distribution of fibronectin and tenascin along migratory pathways of the neural crest in the trunk of amphibian embryos. Development 103r743-756 2. Ferguson DJP, Anderson TJ (1981) Morphological evaluation of cell turnover in relation to the menstrual cycle in the ‘resting’ human breast. Br J Cancer 44:177-181 3. Ferguson MJW (1987) “Palate development”. Mechanisms and malformations (Conway Rcvicw Lecture). Irish J Med Sci 156:309-315 4. Ferguson MJW (1988) Palate development. Devclopment 103 [SUPPI]41-60 15. Fyfe DM, Ferguson MWJ, Chiquet-Ehrismann R (1988) Immunocytochemical localisation of tenascin during the development of scleral papillae and scleral ossicles in the embryonic chick. J Anat 159:117-127 16. Gulcher JR, Nies DE, Marton LS, StePansson K (1989) An alternatively spliced region of the human hexabrachion contains a repeat of potential N-glycosylation sites. Proc Natl Acad Sci USA 86:1588-1592 17. Huff KK, Kaufman D, Gabbay KH, Spencer EM, Lippman ME, Dickson RB (1986) Secretion of an insulin like growth factor-1-related protein by human breast cells. Cancer Res 46 ~4613-4619 18. Inaguma Y, Kusakabe M, Mackie EJ, Pearson CA, ChiquetEhrismann R, Sakdkura T (1988) Epithelial induction of stroma1 tenascin in the mouse mammary gland from embryogenesis to carcinogcnesis. Dev Biol 128:245-255 19. Johnson GD, Davidson RS, McNamee KC, Russell G, Goodwin D, Holborow EJ (1982) Fading of immunofluorescence during microscopy: a study of the phenomenon and its remedy. J Immunol Meth 55:231-242 20. Knabbe C, Lippman M, Wakefield LM, Flanders KC, Kasid A, Derynck R, Dickson R (1987) Evidence that transforming growth factor /Iis a hormonally regulated negative growth factor in human breast cancer cells. Cell 48 :41 7 4 2 8 21. Lightner VA, Gumkowski F, Bigner DD, Erickson HP (1989) Tenascinihexabrachion localisation in human skin : biochemical
207 identification and localisation by light and electron microscopy. J Cell Biol 108 :2483-249 3 22. Mackie EJ, Chiquet-Ehrismann R, Pearson CA, Inaguma Y, Taya K, Kawarada Y, Sakakura T (1987) Tenascin is a stromal marker for epithelial malignancy in the mammary gland. Proc Natl Acad Sci USA 84:4621-4625 23. Mackie EJ, Thesleff I, Chiquet-Ehrismann R (1987) Tenascin is associated with chondrogenic and osteogenic differentiation in vivo and promotes chondrogenesis in vitro. J Cell Biol 105:2569-2579 24. Mackie EJ, Tucker RP, Halfter W, Chiquet-Ehrismann R, Epperlein HH (1988) The distribution of tenascin coincides with pathways of neural crest cell migration. Development 1021237-250 25. Mackie EJ, Halfter W, Liverani D (1988) Induction of tenascin in healing wounds. J Cell Biol 107:2757-2767 26. Mahfouz SM, Chevallier M, Grimaud JA (1987) Distribution of the major connective matrix components of the stromal reaction in breast carcinoma. An immunohistochemical study. Cell Mol Biol 33(4):453-467 27. Minafra P, Luparello C, Sciarrino S, Tomasino RM, Minafra S (1 985) Quantitative determination of collagen types present in the ductal infiltrating carcinoma of human mammary gland. Cell Biol Int Rep 9 No. 3:291-296
28. Ozello L (1970) Epithelial-stromal junction of normal and dysplastic mammary glands. Cancer 25: 586600 29. Pearson CA, Pearson D, Shibahara S, Hofsteenge J, ChiquetEhrismann R (1988) Tenascin: cDNA cloning and induction by TGFB. EMBO J 7(10):2677-2981 30. Rozengurt E, Sinnett-Smith J, Taylor-Papadimitriou J (1985) Production of PDGF-like growth factor by breast cancer cell lines. Int J Canc 36: 147-252 31. Salomon DS, Zwiebel JA, Ban0 M, Losonezy I, Fechnel P, Kidwell WR (1984) Presence of transforming growth factors in human breast cancer cells. Cancer Res 44:4069-4077 32. Sharpe PM, Ferguson MWJ (1988) Mesenchymal influences on epithelial differentiation in developing systems. J Cell Sci [SUPPI]10: 195-230 33. Sharpe PM, Foreman DM, Carette MJM, Schor SL, Ferguson MWJ (1990) The effects of transforming growth factor-beta (TGFP,) on the proliferation and biosynthetic activity of mouse embryonic palatal mesenchyme cells in vitro. J Cell Physiol (in press)
Accepted in revised form October 33, 1989
