Histochemistry(1991) 95:427-433 030155649100029A
Histochemistry © Springer-Verlag 1991
Differences in tetranectin immunoreactivity between benign and malignant breast tissue L. Christensen 1 and I. Clemmensen 2
1 Departments of Pathology,Rigshospitalet,Copenhagenand FrederiksbergHospital, 2 Dakopatts Ltd., Glostrup, and the 2 Department of Clinical Microbiology,Statens Serum Institut at Rigshospitalet,Copenhagen,Denmark ReceivedJune 16, 1990 / Accepted November24, 1990
Summary. Tetranectin (TN) is a human, plasminogen kringle 4 binding plasma protein with ubiquitous cellular distribution and lectin-like characteristics. By means of the peroxidase-antiperoxidase staining technique a polyclonal and a monoclonal antibody were used to demonstrate TN within the intracellular as well as the extracellular compartment of invasive breast carcinoma. Whereas cell associated TN was universal showing only quantitative differences depending of the growth pattern of the tumor, 78 of 133 tumors displayed TN extracellularly as well. The occurrence of this stromal TN immunoreactivity was closely associated with desmoplasia, recognized morphologically by an increase in fibroblastic cells and immunohistochemically by an intense staining for the connective tissue glycoprotein fibronectin (FN). Benign breast tissue displayed a universal, intense cytoplasmic but no extracellular reaction for TN, with the exception of rare loci of granulation tissue and around dilated cysts. Functional studies have shown that human embryonal lung fibroblasts increase their release of TN to the growth medium upon stimulation. The presence of TN extracellularly within fibroblast-rich foci of desmoplasia (and granulation tissue) suggests that a similar increased release of the protein takes place in vivo during active states. Desmoplasia has been found to have a protective effect on tumor cell propagation and metastasis in a murine model. The molecular interactions, which are responsible for this effect, are undoubtedly complex. However, TN may, by its specific binding to kringle 4 of plasminogen and its high affinity for sulphated polysaccharides, add to the understanding of how plasminogen activation is modulated at the local extracellular level.
unpublished observations), which binds to kringle 4 of plasminogen, Ca --+ (Clemmensen et al. 1986) and sulphated polysaccharides (Clemmensen 1989). TN has been demonstrated within the cytoplasm of almost all normal cell types with maximal intensity in endocrine epithelia (Christensen et al. 1987) and exocrine glandular cells with a high turn-over or transport capacity (Christensen et al. 1989). The biological function of the protein is not known. Its primary structure reveals amino acid sequence homologies to various membrane receptors (Fuhlendorff et al. 1987), and the plasma level of TN has been found to be reduced during various physiological and pathological conditions (Kluft et al. 1989). This is also the case in cancer, including breast cancer, where a further significant reduction in plasma TN has been observed in metastatic disease (Jensen 1988). Normal breast tissue displays TN immunoreactivity within all nucleated cell types, whereas the extracellular compartment is consistently unreactive (Christensen et al. 1989). However, recent studies have shown that human embryonal lung fibroblasts in culture (WI-38) deposit TN within the extracellular matrix produced by the cells and release an increased amount of TN to the growth medium upon stimulation (Clemmensen et al., in press). The present study describes immunolocalization of TN both intra- and extracellularly within 78 of 133 invasive breast carcinomas. This combined TN immunoreactivity is closely related to the occurrence of fibroblastic proliferation of the connective tissue surrounding the tumor cells known as desmoplasia and alterations in the TN expression of the tumor cells.
Materials and methods Introduction
Tetranectin (TN) is a human, tetrameric protein of plasma (Clemmensen et al. 1986) and other body fluids (own Offprint requests to." L. Christensen, Frederik V's vej 11, DK-2100
Copenhagen O, Denmark
Tissues
Formalin fixed, paraffin embeddedbreast samples from 56 benign breast lesions, 15 in situ carcinomasand 133 invasivebreast carcinomas (IBC) were obtained from the files of the Departments of
428 Pathology, Frederiksberg Hospital and Rigshospitalet, University of Copenhagen, Denmark. Serial sections were cut at 5 Ixm and stained routinely with H&E, orcein, Van Gieson/Alcian blue and Gordon and Sweets reticulin stain. Neighbor sections were stained for TN and FN, and additional sections were used for controls (vide infra). Frozen samples of benign breast tissue (adenosis, fibroadenoma, involution, cysts and fibrosis) and 15 invasive breast carcinomas covering the different subtypes represented in the paraffin embedded material were stained for TN using a murine monoclonal antibody (vide infra).
Proteolytic digestion Prior to immunostaining deparaffinized sections were pretreated with 1% testicular hyaluronidase (Hyalase) (Leo, Helsingborg, Sweden) in McIlvain buffer, pH 7, for 30 min at 37° C (Holund et al. 1982). Sections stained for TN and control sections were subsequently incubated with 0.02% trypsin (Sigma, St. Louis Mo., USA, type no. III, code no. 8253) in 0.1% CaClz adjusted to pH 7,4 with 0.1 M NaOH at 37° C for 35 min. Sections stained for FN were incubated with 0.1% protease (Sigma, VII 5255) in Tris buffered saline (TBS) (0.1 M Tris (5%) + 0.1 M NaC1, pH 7.6) at room temperature for 5 rain.
Immunoperoxidase staining for TN Conventional peroxidase-antiperoxidase staining techniques for polyclonal and monoclonal antibodies, respectively, were used (Sternberger 1979). When staining frozen, acetone-fixed sections neither proteolytic digestion nor blocking of endogenous peroxidase were carried out. The primary antibodies were 1) a rabbit polyclonal IgG against human tetranectin (Dakopatts Ltd., Glostrup, Denmark, code no. A 317), which had been tested for any impurity due to FN immunoreactivity by immunoblotting using purified plasma FN (Clemmensen et al. 1982), and 2) a mouse monoclonal antibody against human TN, the specificity of which had been controlled by immunoblotting. The protein concentrations for both antibodies were 25 rag/l, and the incubation time 24 h at 4° C followed by 24 h at room temperature. Incubation with the primary specific antibody was for P01yclonals preceded by preincubation for 10 rain with 5% bovine albumin (Behringwerke, Marburg/Lahn, FRG) in TBS. As chromogen was used 3-amino-9-ethylcarbazole, and nuclei were counterstained with Mayer's hematoxylin.
Control stainings To test the specificity of the staining product each type of benign lesion and each subtype of in situ and invasive carcinoma were control stained by a modification of the CLONO-GLAD procedure (Scopsi et al. 1986, vide infra). Using the same polyclonal staining protocol as for TN and the same protein concentration, five tumors with foci of desmoplasia were also stained with a polyclonal rabbit antibody against cow wide spectrum keratin in order to establish the spatial precision of the chromogen distribution. In another 15 control sections 0.3 M NaC1 was added to the primary specific antibody solution in order to test the binding stability of the TN antigen-antibody complex. As negative control was used anti-TN absorbed with purified TN, as controlled by Western blotting.
CLONO-GLAD procedure In a modified version of the gold-labelled antigen detection method (CLONO-GLAD) (Scopsi et al. 1986) incubation with the poly-
clonal antibody against TN was succeeded by the addition of purified TN antigen (2 ~tg/1 in TBS) for 1 h at room temperature. As a third link was used the murine monoclonal antibody against human TN (25 mg/1) in phosphate buffered saline/1% bovine albumin for i h at room temperature. For visualization of the reaction product a conventional alkaline phosphatase/anti-alkaline phosphatase staining procedure for monoclonal antibodies was used (Cordell et al. 1984).
Immunostaining for FN The peroxidase-antiperoxidase staining protocol for FN has been described previously (Christensen et al. 1985). The primary antibody was rabbit polyclonal IgG against human FN (Dakopatts Ltd., code no. A 245) used at a protein concentration of 50 rag/1.
Immunoperoxidase staining with two polyclonal antibodies Double staining for TN and FN by a combined indirect immunoperoxidase/alkaline phosphatase staining technique was carried out as follows: 1) deparaffinization and hydration, 2) proteolytic digestion with trypsin and hyaluronidase as described above, 3) blocking of endogenous peroxidase activity with H202 in methanol, 4) incubation with rabbit anti-TN as described, 5) incubation with an affinity purified goat polyclonal antibody to human plasma FN (Accurate Chemical and Scientific Corporation, New York, USA, code no. YKP 011024), 2 mg/1 for 30 rain, 6) incubation with an alkaline phosphatase-conjugated swine-antirabbit immunoglobulin (Dakopatts Ltd., code no. D 306), 1 : 20 for 60 rain, 7) incubation with a peroxidase-eonjugated rabbit-antigoat immunoglobulin (Dakopatts Lts., code no. P 160), 1:20 for 30rain, 8) staining with 3-amino-9-ethylcarbazole, 9) staining with 45 gl 5-bromo-4chloro-3-indolyl phosphate diluted in 100% dimethylformamide (50 mg/ml) + 45 ~tlnitro blue tetrazolium diluted in 70% dimethylformamide (75 mg/ml) in 10 ml 0.1 M Tris/HC1, 0.05 M MgC12, pH 9.0 for 20 rain. Levamisol, I raM, was added to block endogenous alkaline phosphatase activity, 10) Mayer's hematoxylin (nuclear counterstaining) and mounting in Glycergel (Dakopatts Ltd, Denmark).
Immunofluorescence with two polyclonal antibodies Double staining for TN and FN by a modification of the combined indirect immunofluorescence staining technique (Wang et al. 1985) using two different fluorochromes (FITC and TRITC, respectively) was carried out as follows: 1) deparaffinization and hydration followed by proteolytic digestion with trypsin and hyaluronidase as described above, 2) preincubation with 5% bovine albumin for 10 rain followed by incubation with anti-TN as described, 3) incubation with goat anti-FN as described above, 4) incubation with rhodamin (TRITC) conjugated swine immunoglobulins to rabbit immunoglobulins (Dakopatts Ltd., code no R 156), 1:200 for 30 rain, 5) incubation with fluorescein (FITC) conjugated rabbit immunoglobulinsto goat immunoglobulins(Dakopatts Ltd., code no F 250) 1:40 for 30 min. Sections were mounted in Glycergel, Dakopatts Ltd, Denmark) and observed in a Leitz'orthoplan fluorescence microscope (Leitz, Wetzlar, FRG) equipped for epiilumination with selective filters for excitation at 490 nm (FITC) and 546 nm (TRITC). If not otherwise notified all immunological reactions were carried out at room temperature. The monoclonal anti-TN was diluted in 1% bovine albumin in TBS and all polyclonal antibodies were diluted in 5% bovine albumin in TBS. Whenever there was a risk of cross-reactivity, 50 gl. of serum from the cross-reacting species was added per ml antibody solution. Sections were washed 10 rain x 3 with Triton X-100 in TBS between all steps except after preincubation and after staining with the chromogens, where running tap water for 15 min was used.
429
Diagnosis The original H&E stained slides were reviewed and diagnosed morphologically prior to immunostaining. Benign breast lesions were classified according to the criteria of Azzopardi (1979) and Haagensen (1986), and the IBCs were classified and subclassified by following the guidelines of the World Health Organization (WHO) (Sobin 1981) and Bloom and Richardson (1957), respectively.
Evaluation of staining results A TN positive staining reaction was recorded for both epithelial and mesenchymal cells and the extracellular matrix. The cytoplasmic staining was estimated as intense, moderate, and weak. Extracellular staining was registered, if a distinct, linear or finely granular staining of individual connective tissue fibres was observed. Desmoplasia was defined morphologically on H&E sections as a significant increase in fibroblastic cells and extracellular matrix density, and immunohistochemically by an intense, coarse fiberrelated reactivity for the connective tissue glycoprotein fibronectin.
Table 1. Extracellular TN [and FN] staining pattern of 56 benign proliferative breast lesions Histology
TN staining ÷
Cysts Duct ectasia Adenosis Sclerosing adenosis Adenosis tumor Fibroadenoma Lactating adenoma Tubular adenoma Epitheliosis Solitary papilloma
3 2 0 0 0 0 0 0 0 3
5 3 8 6 4 5 5 3 6 3
Total
8
48
Table 2. Extracellular TN staining in relation to histological type of 133 invasive breast carcinomas
Results
Histological type
+
Normal and benign breast tissue (Table 1) The T N staining pattern of all benign lesions was similar to the one previously recorded for normal breast tissue (Christensen et al. 1989): Between 80% and 90% of the secretory cells showed a distinct, finely granular reaction, which tended to be apically oriented. The majority of the myoepithelial cells were intensely stained as well, whereas stromal cells such as fibroblasts and inflammatory cells showed a weak to moderate reaction. When the secretory cells appeared larger than normal as in apocrine metaplasia and lactation the T N positive granules were either spread out with an even distribution throughout the cytoplasm (apocrine metaplasia) or concentrated around secretory globules, which were negative (lactation). The extracellular compartment was generally negative. The only exceptions to this were dense connective tissue around cysts and dilated ducts (Fig. 1 a) and granulation tissue, e.g. around ruptured cysts or corresponding to the connective tissue core of solitary intraductal papillomas (Table 1). F N was similarly increased in such lesions (Fig. 1 b [Table 1]).
In situ carcinomas Fifteen tumors were pure in situ carcinomas (11 ductal and 4 lobular), and another 78 of the IBCs had in situ lesions associated. All cells from the lobular and most cells from the ductal in situ elements were stained for T N like their benign counterparts. However, a reduced cytoplasmic reactivity ( + ) was occasionally observed, which was contrasted by a few, often peripherally locat•ed, intensely stained tumor cells ( + + + ) (Fig. 2 a). Central necroses of comedo elements were negative (Fig. 2b). Stroma cells were stained like in the benign lesions, and the extracellular matrix was negative, apart
TN staining
Duct carcinoma (IDC) Grade 1 Grade 2-: Grade 3 Papillary carcinoma Tubular carcinoma Colloid carcinoma Medullary carcinoma Invasive lobular c. (ILC) Sigillocellular carcinoma IDC Grade 1/ILC IDC Grade 2/ILC IDC Grade 3/ILC
15 33 15 3 3 0 3 2 1 1 0 2
7 8 6 1 2 3 1 24 0 1 1 1
Total
78
55
from a fine ring of T N positive connective tissue fibres surrounding some of the intraductal elements.
Infiltrating carcinomas (Table 2) An overview of the different histological subtypes of IBC and their extracellular T N immunoreactivity is presented in Table 2. The tumors showed considerable heterogeneity in staining patterns coinciding with the occurrence of desmoplasia (Fig. 2c) and an intense immunoreactivity for FN. Extracellular T N was inconspicious in 55 of the tumors (Fig. 3 a and b), the dominating feature in another 26 (Fig. 4a) and present just focally, predominantly towards the centre, in the last 52 tumors (Table 2). Also epithelial, cytoplasmic staining for T N varied considerably among the same tumor types as well as within the same tumor. In general, the intensity was similar to ( + + ) or slightly increased ( + + + ) compared to the one observed in benign lesions (Figs. 3 and 4a). However, some carcinomas tended to have intensely
Fig. 1 a, b. Cyst wall displaying intense cytoplasmic staining for TN and a fine rim of TN positive fibres just beneath the flattened epithelium (a). A similar linear staining for F N is seen in b. Magnification, x 96. Bar = 50 micron Fig. 2a--e. Intraductal carcinoma of comedotype stained for TN. The tumor cells may be of clear cell type with the exception of an occasional rim of intensely stained cells peripherally (a) (arrows), or they may show an intense, universal immunoreactivity, only contrasted by necroses, which are unreactive (b). Frozen section of an intraductal carcinoma with transition into an infiltrating carcinoma stained with a monoclonal antibody to TN is shown in e. Desmoplastic connective tissue surrounding the infiltrating tumor cells displays an intense immunoreactivity (arrows). Magnification, × 110. Bar = 60 micron
Figs. 1 4 : Immunostaining for TN (and FN) counterstained with Mayer's hemalun
Fig. 3a, b. Two Grade 2 duct carcinomas showing a moderate to intense cytoplasmic and no stromal staining for TN. The cytoplasmic immunoreactivity is evenly distributed in (a) and accentuated peripherally within tumor cell clusters in (b). Magnification x 110). Bar = 60 micron
Fig. 4a, b. Grade 2 duct carcinoma adjacent to breast tissue with cysts and intraductal carcinoma of cribriform and clinging types. The infiltrating tumor cells are weakly to moderately stained for TN, and the extracellular connective tissue is positive as well (a). Also F N immunostaining is intense within the desmoplastic connective tissue stroma, which surrounds the tumor islands (b). Magnification, × 110. Bar=60 micron
Fig. 5 a-c. Close-up photomicrograph of a solidly growing duct
,---
b
carcinoma with TN immunoreactivity concentrated peripherally in the tumor cell clusters and no stromal staining (arrows) (a). In b another tumor shows an intense staining for TN of connective tissue fibres, whereas the tumor cells appear empty. The same tumor stained by the C L O N O - G L A D procedure, is presented in c. Note the intensely stained normal duct (left). Magnification, x 150. Bar = 30 micron Fig. 6. Grade 2 duct carcinoma showing double labeling for TN and F N using alkaline phosphatase with a black chromogen and peroxidase with a brownish-red chromogen, respectively (a). In b the same tumor is shown in a double immunofluorescence staining. The red color represents TN (TRITC) and the green F N (FITC). Magnification, × 96. Bar = 50 micron
432 stained cells only peripherally in the tumor cell clusters (Figs. 3b, 5a), and in cases of desmoplasia or solid growth pattern of the tumor, the overall cytoplasmic immunoreactivity was frequently reduced (Fig. 5 b).
Control staining Sections stained by the CLONO-GLAD procedure and frozen sections stained directly by the monoclonal antibody produced results similar to the ones obtained with the polyclonal antibody against TN (Figs. 2 a-c and 5 b c). Using the same staining procedure and the same or double the protein concentration of the one used for anti-TN, staining for wide spectrum keratin showed an expected positivity corresponding to the tumor cells but no extracellular staining. The absorbed antibody failed to stain tumor cells as well as extracellular matrix. The addition of 0.3 M NaC1 to the rabbit polyclonal anti-TN solution did not influence intensity nor localization of the staining product.
Double staining procedures Double staining studies of TN and FN were performed on the same tumors, which were used for control stainings, in order to rule out the possibility that some of the TN immunoreactivity demonstrated extracellularly originated from contaminating anti-FN antibody in the TN antibody solution. With a simple 2-layer immunological labeling method using antibodies from different species, different marker enzymes and chromogens or fluorochromes, respectively, 2 different, double-stained sections were produced, which were complementary. Tissue morphology was superior in the enzyme-stained sections, whereas localization of the staining product was optimal in the immunofluoroscence stained sections (Fig. 6a and b). TN and FN stained different fibres, confirming separate sublocalizations of the proteins extracellularly (Fig. 6 a and b).
Discussion
TN was originally discovered in human plasma as a tetrameric protein, which bound to the kringle-4 structure of plasminogen (Clemmensen et al. 1986). Up to the present the protein has been discovered in almost all human tissues (Christensen et al. 1987 and 1989) and body fluids (own unpublished observations), suggesting a function in fundamental, biological processes. Because of its close relationship with plasminogen and its stimulating effect on t-PA catalysed plasminogen activation in vitro (Clemmensen et al. 1986) TN is thought to participate in processes, where plasminogen activators are involved (e.g. tissue degradation, involution, extracellular proteolysis and cell migration) (Dano et al. 1985). This idea is further supported by the observation that TN binds strongly to sulphated polysaccharides (Clemmensen et al. 1989), of which fucoidan has been found
to enhance plasminogen activation catalysed by both u-PA and t-PA (Brunner et al. 1988). Studies of TN at the cellular level have shown that the protein is expelled from granulocytes upon stimulation (Borregaard et al. 1990). Furthermore, cultured hu'man, embryonal lung fibroblasts (WI-38) have been found to produce TN to an extracellular matrix produced by the cells and release increased amounts of TN to the growth medium after stimulation (Clemmensen et al., in press). The finding in this study of TN extracellularly within tissues with a high fibroblast concentration (e.g. desmoplasia and granulation tissue) may reflect a similar increased cellular release of TN in vivo. Desmoplasia is thought to exert a protective effect on tumor spread. This has most convincingly been shown in a routine system, where inhibition of the desmoplastic response elicited by certain melanoma cells increased tumor invasion and metastasis (Barsky et al. 1987). A number of components other than TN are present in desmoplasia. The connective tissue glycoprotein FN is significantly increased in such lesions (Lagace et al. 1985; Christensen et al. 1988), and so is elastin, vitronectin (Loridon-Rosa et al. 1988), collagen type 3 and 5 (Barsky et al. 1982; Lagace et al. 1985) and certain proteoglycans (Takeuchi et al. 1976). FN binds, like TN, to plasminogen in vitro (Salonen et al. 1985), and because of its high concentration in plasma, from which the TN used for immunization has been purified, double staining experiments for TN and FN were carried out. These showed different extracellular sublocalizations of the 2 proteins, minimizing the possibility that the TN immunoreactivity was due to a contaminating titer against FN in the TN antibody solution (Fig. 6). The differences found in cellular distribution between benign secretory cells of the breast and their malignant counterparts and the heterogeneity in staining pattern within the same tumor, have been described for other epithelial markers such as epithelial membrane antigen (Sloane et al. 1981), carcinoembryonic antigen (Kuhajda et al. 1983) and the keratins (Gusterson et al. 1982). However, in contrast to TN, none of these have been demonstrated within the extracellular compartment. The constantly growing information about the interactions of TN in vitro has so far allocated the protein as being a possible participant in the extravascular proteolytic process, which includes conversion of plasminogen to plasmin by plasminogen activators. The importance of plasmin for tissue degradation in cancer growth and spread has been stressed before (Dano et al. 1985). The identification of TN within not only the cells but also within the extracellular matrix of IBCs corresponding to the desmoplastic connective tissue response, combined with the fact that TN as the only known, naturally occurring ligand binds to kringle 4 of plasminogen, may add to the understanding of how plasminogen activation is modulated at the extracellular level.
Acknowledgements. This study was supported by grants from The Danish Cancer Societyand The Danish Medical Research Council.
433 The authors wish to thank laboratory assistants Mrs. A. Pedersen, A. Christensen, I. Bech and A. Ryberg for valuable technical assistance, and photographer Mr. B. Borgesen is gratefully acknowledged for excellent photographical help.
References Azzopardi JG (1979) Problems in breast pathology, vol 2. H.B. Saunders, London Barsky SH, Rao CN, Grotendorst GR, Liotta LA (1982) Increased content of type V collagen in desmoplasia of human breast carcinoma. Am J Pathol 108:276-283 Barsky SH, Gopalakrishna R (:1987) Increased invasion and spontaneous metastasis of BL6 melanoma with inhibition of the desmoplastic response in C57 BL/6 mice. Cancer Res 47 : 1663 1667 Bloom HJG, Richardson WW (1957) Histological grading and prognosis in breast cancer. Br J Cancer l 1:359-365 Borregaard N, Christensen L, Bjerrum OW, Birgens HS, Clemmensen I (1990) Secretory granules. Identification of a highly mobilizable subset of human neutrophil granules, which contains tetranectin and latent alkaline phosphatase. J Clin Invest 85: 408-417 Brunner G, Schirrmacher V (1988) Stimulation of plasminogen activation by sulphated glycosaminoglycans: a possible mechanism for extracellular matrix penetration by tumor cells. Clin Exp Met 6 [Suppl 1] :43 (abstract) Christensen L, Strange L (1987) Universal immunoperoxidase staining protocol to optimize the use of polyclonal and monoclonal antibodies. J Histotechnol 10:11-15 Christensen L, Johansen N, Jensen BA, Clemmensen I (1987) Immunohistochemical localization of a novel, human plasma protein, tetranectin, in human endocrine tissues. Histochemistry 87:195 199 Christensen L, Nielsen M, Andersen J, Clemmensen I (:1988) Stromal fibronectin staining pattern and metastasizing ability of human breast carcinoma. Cancer Res 48:6227-6233 Christensen L, Clemmensen I (1989) Tetranectin immunoreactivity in normal human tissues. An immunohistochemical study of exocrine epithelia and mesenchyme. Histochemistry 92:29-35 Clemmensen I, Andersen RB (:1982) Different molecular forms of fibronectin in rheumatoid synovial fluid. Arthritis Rheum 25:25-31 Clemmensen I, Petersen L, Kluft C (1986) Purification and characterization of a novel, oligomeric, plasminogen kringle 4 binding protein from human plasma. Eur J Biochem 156 : 327-333 Clemmensen I (1989) Interaction of tetranectin with sulphated polysaccharides and trypan blue. Scand J Clin Lab Invest 49 : 7:19-725 Clemmensen I, Lurid L, Christensen L, Andreasen P (1990) Tetranectin: production and extracellular matrix deposition by human embryonal fibroblasts. Eur J Biochem (in press) Cordell JL, Falini B, Erber W, Ghosh A, Abdulaziz Z, MacDonald S, Pulford KAF, Stein H, Mason DY (:1984) Immunoenzymatic labelling of monoclonal antibodies using immune complexes of alkaline phosphatase and monoclonal anti-alkaline phosphatase (APAAP complexes). J Histochem Cytochem 32:219-229 Dano K, Andreasen PA, Grondahl-Hansen J, Kristensen P, Niel-
sen LS, Skriver L (1985) Plasminogen activators, tissue degradation and cancer. Adv Cancer Res 44:139-266 Fuhlendorff J, Clemmensen I, Magnusson S (1987) The primary structure of a kringle 4 binding protein: tetranectin. Structure homology with asialoglycoprotein receptors and a proteoglycan core protein. Biochemistry 26:6757-6764 Gusterson BA, Warburton M J, Mitchell D, Ellison M, Neville AM, Rudland PS (1982) Distribution of myoepithelial cells and basement membrane proteins in the normal breast and in benign and malignant breast disease. Cancer Res 42:4763-4770 Haagensen CD (1986) Diseases of the breast, 3rd edn. W.B. Saunders, Philadelphia London Toronto Holund B, Clausen PP, Clemmensen I (1981) The influence of fixation and tissue preparation on the immunohistochemical demonstration of fibronectin in human tissues. Histochemistry 72 : 291-299 Holund B, Clemmensen I (1982) The value of hyaluronidase treatment of different tissues before demonstration of fibronectin by the indirect immunoperoxidase technique. Histochemistry 76: 517-525 Jensen BA, Clernmensen I (1988) Plasma tetranectin is reduced in cancer and related metastasia. Cancer 62:3222-3227 Kluft C, Los P, Clemmensen I, Brommer EJP, Leuven JAG, Boks AL, Vellenga E (I 989) Quantitation of plasma levels of tetranectin effects of oral contraceptives, pregnancy, treatment with L-asparaginase and liver cirrhosis. Thromb Haemostas 62: 792796 Kuhajda FP, Offutt LE, Mendelsohn G (1983) The distribution of carcinoembryonic antigen in breast carcinoma. Cancer 52:1257-1264 Lagace R, Grimaud J-A, Schwurch W, Seemayer TA (1985) Myofibroblastic stromal reaction in carcinoma of the breast: variations of collagenous matrix and structural glycoproteins. Virchows Arch [A] 408 : 49-59 Loridon-Rosa B, Vielh P, Cuadrado C, Burtin P (1988) Comparative distribution of fibronectin and vitronectin in human breast and colon carcinomas. An immunofluorescence study. Am J Clin Pathol 90: 7-16 Salonen EM, Saksela O, Vartio T, Vaheri A, Nielsen LS, Zeuthen J (1985) Plasminogen and tissue type plasminogen activator bind to immobilized fibronectin. J Biol Chem 260:12302-12307 Saunders S, Bernfiels M (1988) Cell surface proteoglycan binds mouse mammary epithelial cells to fibronectin and behaves as a receptor for interstitial matrix. J Cell Biol 106:423-430 Scopsi L, Bock E, Larsson L-I (1986) Monoclonal antibody immunocytochemistry: novel method extending usefulness of monoclonat antibodies for antigen visualization. Eur J Cell Biol 41:97-10:1 Sloane JP, Ormerod MG (1981) Distribution of epithelial membrane antigen in normal and neoplastic tissues and its value in diagnostic tumor pathology. Cancer 47:1786-1795 Sobin L (1981) Histological typing of breast tumors, 2nd edn. WHO, Geneva Sternberger LA (1979) Immunocytochemistry, 2nd edn. John Wiley, New York Takeuchi J, Sobue M, Sato E, Shamoto M, Miura K, Nakagaki S (1976) Variation in glycosaminoglycan components of breast tumors. Cancer Res 36 : 2133-2139 Wang B-L, Larsson L-I (:1985) Simultaneous demonstration of multiple antigens by indirect immunofluorescence or immunogold staining. Histochemistry 83 : 47-56
Histochemistry © Springer-Verlag 1991
Differences in tetranectin immunoreactivity between benign and malignant breast tissue L. Christensen 1 and I. Clemmensen 2
1 Departments of Pathology,Rigshospitalet,Copenhagenand FrederiksbergHospital, 2 Dakopatts Ltd., Glostrup, and the 2 Department of Clinical Microbiology,Statens Serum Institut at Rigshospitalet,Copenhagen,Denmark ReceivedJune 16, 1990 / Accepted November24, 1990
Summary. Tetranectin (TN) is a human, plasminogen kringle 4 binding plasma protein with ubiquitous cellular distribution and lectin-like characteristics. By means of the peroxidase-antiperoxidase staining technique a polyclonal and a monoclonal antibody were used to demonstrate TN within the intracellular as well as the extracellular compartment of invasive breast carcinoma. Whereas cell associated TN was universal showing only quantitative differences depending of the growth pattern of the tumor, 78 of 133 tumors displayed TN extracellularly as well. The occurrence of this stromal TN immunoreactivity was closely associated with desmoplasia, recognized morphologically by an increase in fibroblastic cells and immunohistochemically by an intense staining for the connective tissue glycoprotein fibronectin (FN). Benign breast tissue displayed a universal, intense cytoplasmic but no extracellular reaction for TN, with the exception of rare loci of granulation tissue and around dilated cysts. Functional studies have shown that human embryonal lung fibroblasts increase their release of TN to the growth medium upon stimulation. The presence of TN extracellularly within fibroblast-rich foci of desmoplasia (and granulation tissue) suggests that a similar increased release of the protein takes place in vivo during active states. Desmoplasia has been found to have a protective effect on tumor cell propagation and metastasis in a murine model. The molecular interactions, which are responsible for this effect, are undoubtedly complex. However, TN may, by its specific binding to kringle 4 of plasminogen and its high affinity for sulphated polysaccharides, add to the understanding of how plasminogen activation is modulated at the local extracellular level.
unpublished observations), which binds to kringle 4 of plasminogen, Ca --+ (Clemmensen et al. 1986) and sulphated polysaccharides (Clemmensen 1989). TN has been demonstrated within the cytoplasm of almost all normal cell types with maximal intensity in endocrine epithelia (Christensen et al. 1987) and exocrine glandular cells with a high turn-over or transport capacity (Christensen et al. 1989). The biological function of the protein is not known. Its primary structure reveals amino acid sequence homologies to various membrane receptors (Fuhlendorff et al. 1987), and the plasma level of TN has been found to be reduced during various physiological and pathological conditions (Kluft et al. 1989). This is also the case in cancer, including breast cancer, where a further significant reduction in plasma TN has been observed in metastatic disease (Jensen 1988). Normal breast tissue displays TN immunoreactivity within all nucleated cell types, whereas the extracellular compartment is consistently unreactive (Christensen et al. 1989). However, recent studies have shown that human embryonal lung fibroblasts in culture (WI-38) deposit TN within the extracellular matrix produced by the cells and release an increased amount of TN to the growth medium upon stimulation (Clemmensen et al., in press). The present study describes immunolocalization of TN both intra- and extracellularly within 78 of 133 invasive breast carcinomas. This combined TN immunoreactivity is closely related to the occurrence of fibroblastic proliferation of the connective tissue surrounding the tumor cells known as desmoplasia and alterations in the TN expression of the tumor cells.
Materials and methods Introduction
Tetranectin (TN) is a human, tetrameric protein of plasma (Clemmensen et al. 1986) and other body fluids (own Offprint requests to." L. Christensen, Frederik V's vej 11, DK-2100
Copenhagen O, Denmark
Tissues
Formalin fixed, paraffin embeddedbreast samples from 56 benign breast lesions, 15 in situ carcinomasand 133 invasivebreast carcinomas (IBC) were obtained from the files of the Departments of
428 Pathology, Frederiksberg Hospital and Rigshospitalet, University of Copenhagen, Denmark. Serial sections were cut at 5 Ixm and stained routinely with H&E, orcein, Van Gieson/Alcian blue and Gordon and Sweets reticulin stain. Neighbor sections were stained for TN and FN, and additional sections were used for controls (vide infra). Frozen samples of benign breast tissue (adenosis, fibroadenoma, involution, cysts and fibrosis) and 15 invasive breast carcinomas covering the different subtypes represented in the paraffin embedded material were stained for TN using a murine monoclonal antibody (vide infra).
Proteolytic digestion Prior to immunostaining deparaffinized sections were pretreated with 1% testicular hyaluronidase (Hyalase) (Leo, Helsingborg, Sweden) in McIlvain buffer, pH 7, for 30 min at 37° C (Holund et al. 1982). Sections stained for TN and control sections were subsequently incubated with 0.02% trypsin (Sigma, St. Louis Mo., USA, type no. III, code no. 8253) in 0.1% CaClz adjusted to pH 7,4 with 0.1 M NaOH at 37° C for 35 min. Sections stained for FN were incubated with 0.1% protease (Sigma, VII 5255) in Tris buffered saline (TBS) (0.1 M Tris (5%) + 0.1 M NaC1, pH 7.6) at room temperature for 5 rain.
Immunoperoxidase staining for TN Conventional peroxidase-antiperoxidase staining techniques for polyclonal and monoclonal antibodies, respectively, were used (Sternberger 1979). When staining frozen, acetone-fixed sections neither proteolytic digestion nor blocking of endogenous peroxidase were carried out. The primary antibodies were 1) a rabbit polyclonal IgG against human tetranectin (Dakopatts Ltd., Glostrup, Denmark, code no. A 317), which had been tested for any impurity due to FN immunoreactivity by immunoblotting using purified plasma FN (Clemmensen et al. 1982), and 2) a mouse monoclonal antibody against human TN, the specificity of which had been controlled by immunoblotting. The protein concentrations for both antibodies were 25 rag/l, and the incubation time 24 h at 4° C followed by 24 h at room temperature. Incubation with the primary specific antibody was for P01yclonals preceded by preincubation for 10 rain with 5% bovine albumin (Behringwerke, Marburg/Lahn, FRG) in TBS. As chromogen was used 3-amino-9-ethylcarbazole, and nuclei were counterstained with Mayer's hematoxylin.
Control stainings To test the specificity of the staining product each type of benign lesion and each subtype of in situ and invasive carcinoma were control stained by a modification of the CLONO-GLAD procedure (Scopsi et al. 1986, vide infra). Using the same polyclonal staining protocol as for TN and the same protein concentration, five tumors with foci of desmoplasia were also stained with a polyclonal rabbit antibody against cow wide spectrum keratin in order to establish the spatial precision of the chromogen distribution. In another 15 control sections 0.3 M NaC1 was added to the primary specific antibody solution in order to test the binding stability of the TN antigen-antibody complex. As negative control was used anti-TN absorbed with purified TN, as controlled by Western blotting.
CLONO-GLAD procedure In a modified version of the gold-labelled antigen detection method (CLONO-GLAD) (Scopsi et al. 1986) incubation with the poly-
clonal antibody against TN was succeeded by the addition of purified TN antigen (2 ~tg/1 in TBS) for 1 h at room temperature. As a third link was used the murine monoclonal antibody against human TN (25 mg/1) in phosphate buffered saline/1% bovine albumin for i h at room temperature. For visualization of the reaction product a conventional alkaline phosphatase/anti-alkaline phosphatase staining procedure for monoclonal antibodies was used (Cordell et al. 1984).
Immunostaining for FN The peroxidase-antiperoxidase staining protocol for FN has been described previously (Christensen et al. 1985). The primary antibody was rabbit polyclonal IgG against human FN (Dakopatts Ltd., code no. A 245) used at a protein concentration of 50 rag/1.
Immunoperoxidase staining with two polyclonal antibodies Double staining for TN and FN by a combined indirect immunoperoxidase/alkaline phosphatase staining technique was carried out as follows: 1) deparaffinization and hydration, 2) proteolytic digestion with trypsin and hyaluronidase as described above, 3) blocking of endogenous peroxidase activity with H202 in methanol, 4) incubation with rabbit anti-TN as described, 5) incubation with an affinity purified goat polyclonal antibody to human plasma FN (Accurate Chemical and Scientific Corporation, New York, USA, code no. YKP 011024), 2 mg/1 for 30 rain, 6) incubation with an alkaline phosphatase-conjugated swine-antirabbit immunoglobulin (Dakopatts Ltd., code no. D 306), 1 : 20 for 60 rain, 7) incubation with a peroxidase-eonjugated rabbit-antigoat immunoglobulin (Dakopatts Lts., code no. P 160), 1:20 for 30rain, 8) staining with 3-amino-9-ethylcarbazole, 9) staining with 45 gl 5-bromo-4chloro-3-indolyl phosphate diluted in 100% dimethylformamide (50 mg/ml) + 45 ~tlnitro blue tetrazolium diluted in 70% dimethylformamide (75 mg/ml) in 10 ml 0.1 M Tris/HC1, 0.05 M MgC12, pH 9.0 for 20 rain. Levamisol, I raM, was added to block endogenous alkaline phosphatase activity, 10) Mayer's hematoxylin (nuclear counterstaining) and mounting in Glycergel (Dakopatts Ltd, Denmark).
Immunofluorescence with two polyclonal antibodies Double staining for TN and FN by a modification of the combined indirect immunofluorescence staining technique (Wang et al. 1985) using two different fluorochromes (FITC and TRITC, respectively) was carried out as follows: 1) deparaffinization and hydration followed by proteolytic digestion with trypsin and hyaluronidase as described above, 2) preincubation with 5% bovine albumin for 10 rain followed by incubation with anti-TN as described, 3) incubation with goat anti-FN as described above, 4) incubation with rhodamin (TRITC) conjugated swine immunoglobulins to rabbit immunoglobulins (Dakopatts Ltd., code no R 156), 1:200 for 30 rain, 5) incubation with fluorescein (FITC) conjugated rabbit immunoglobulinsto goat immunoglobulins(Dakopatts Ltd., code no F 250) 1:40 for 30 min. Sections were mounted in Glycergel, Dakopatts Ltd, Denmark) and observed in a Leitz'orthoplan fluorescence microscope (Leitz, Wetzlar, FRG) equipped for epiilumination with selective filters for excitation at 490 nm (FITC) and 546 nm (TRITC). If not otherwise notified all immunological reactions were carried out at room temperature. The monoclonal anti-TN was diluted in 1% bovine albumin in TBS and all polyclonal antibodies were diluted in 5% bovine albumin in TBS. Whenever there was a risk of cross-reactivity, 50 gl. of serum from the cross-reacting species was added per ml antibody solution. Sections were washed 10 rain x 3 with Triton X-100 in TBS between all steps except after preincubation and after staining with the chromogens, where running tap water for 15 min was used.
429
Diagnosis The original H&E stained slides were reviewed and diagnosed morphologically prior to immunostaining. Benign breast lesions were classified according to the criteria of Azzopardi (1979) and Haagensen (1986), and the IBCs were classified and subclassified by following the guidelines of the World Health Organization (WHO) (Sobin 1981) and Bloom and Richardson (1957), respectively.
Evaluation of staining results A TN positive staining reaction was recorded for both epithelial and mesenchymal cells and the extracellular matrix. The cytoplasmic staining was estimated as intense, moderate, and weak. Extracellular staining was registered, if a distinct, linear or finely granular staining of individual connective tissue fibres was observed. Desmoplasia was defined morphologically on H&E sections as a significant increase in fibroblastic cells and extracellular matrix density, and immunohistochemically by an intense, coarse fiberrelated reactivity for the connective tissue glycoprotein fibronectin.
Table 1. Extracellular TN [and FN] staining pattern of 56 benign proliferative breast lesions Histology
TN staining ÷
Cysts Duct ectasia Adenosis Sclerosing adenosis Adenosis tumor Fibroadenoma Lactating adenoma Tubular adenoma Epitheliosis Solitary papilloma
3 2 0 0 0 0 0 0 0 3
5 3 8 6 4 5 5 3 6 3
Total
8
48
Table 2. Extracellular TN staining in relation to histological type of 133 invasive breast carcinomas
Results
Histological type
+
Normal and benign breast tissue (Table 1) The T N staining pattern of all benign lesions was similar to the one previously recorded for normal breast tissue (Christensen et al. 1989): Between 80% and 90% of the secretory cells showed a distinct, finely granular reaction, which tended to be apically oriented. The majority of the myoepithelial cells were intensely stained as well, whereas stromal cells such as fibroblasts and inflammatory cells showed a weak to moderate reaction. When the secretory cells appeared larger than normal as in apocrine metaplasia and lactation the T N positive granules were either spread out with an even distribution throughout the cytoplasm (apocrine metaplasia) or concentrated around secretory globules, which were negative (lactation). The extracellular compartment was generally negative. The only exceptions to this were dense connective tissue around cysts and dilated ducts (Fig. 1 a) and granulation tissue, e.g. around ruptured cysts or corresponding to the connective tissue core of solitary intraductal papillomas (Table 1). F N was similarly increased in such lesions (Fig. 1 b [Table 1]).
In situ carcinomas Fifteen tumors were pure in situ carcinomas (11 ductal and 4 lobular), and another 78 of the IBCs had in situ lesions associated. All cells from the lobular and most cells from the ductal in situ elements were stained for T N like their benign counterparts. However, a reduced cytoplasmic reactivity ( + ) was occasionally observed, which was contrasted by a few, often peripherally locat•ed, intensely stained tumor cells ( + + + ) (Fig. 2 a). Central necroses of comedo elements were negative (Fig. 2b). Stroma cells were stained like in the benign lesions, and the extracellular matrix was negative, apart
TN staining
Duct carcinoma (IDC) Grade 1 Grade 2-: Grade 3 Papillary carcinoma Tubular carcinoma Colloid carcinoma Medullary carcinoma Invasive lobular c. (ILC) Sigillocellular carcinoma IDC Grade 1/ILC IDC Grade 2/ILC IDC Grade 3/ILC
15 33 15 3 3 0 3 2 1 1 0 2
7 8 6 1 2 3 1 24 0 1 1 1
Total
78
55
from a fine ring of T N positive connective tissue fibres surrounding some of the intraductal elements.
Infiltrating carcinomas (Table 2) An overview of the different histological subtypes of IBC and their extracellular T N immunoreactivity is presented in Table 2. The tumors showed considerable heterogeneity in staining patterns coinciding with the occurrence of desmoplasia (Fig. 2c) and an intense immunoreactivity for FN. Extracellular T N was inconspicious in 55 of the tumors (Fig. 3 a and b), the dominating feature in another 26 (Fig. 4a) and present just focally, predominantly towards the centre, in the last 52 tumors (Table 2). Also epithelial, cytoplasmic staining for T N varied considerably among the same tumor types as well as within the same tumor. In general, the intensity was similar to ( + + ) or slightly increased ( + + + ) compared to the one observed in benign lesions (Figs. 3 and 4a). However, some carcinomas tended to have intensely
Fig. 1 a, b. Cyst wall displaying intense cytoplasmic staining for TN and a fine rim of TN positive fibres just beneath the flattened epithelium (a). A similar linear staining for F N is seen in b. Magnification, x 96. Bar = 50 micron Fig. 2a--e. Intraductal carcinoma of comedotype stained for TN. The tumor cells may be of clear cell type with the exception of an occasional rim of intensely stained cells peripherally (a) (arrows), or they may show an intense, universal immunoreactivity, only contrasted by necroses, which are unreactive (b). Frozen section of an intraductal carcinoma with transition into an infiltrating carcinoma stained with a monoclonal antibody to TN is shown in e. Desmoplastic connective tissue surrounding the infiltrating tumor cells displays an intense immunoreactivity (arrows). Magnification, × 110. Bar = 60 micron
Figs. 1 4 : Immunostaining for TN (and FN) counterstained with Mayer's hemalun
Fig. 3a, b. Two Grade 2 duct carcinomas showing a moderate to intense cytoplasmic and no stromal staining for TN. The cytoplasmic immunoreactivity is evenly distributed in (a) and accentuated peripherally within tumor cell clusters in (b). Magnification x 110). Bar = 60 micron
Fig. 4a, b. Grade 2 duct carcinoma adjacent to breast tissue with cysts and intraductal carcinoma of cribriform and clinging types. The infiltrating tumor cells are weakly to moderately stained for TN, and the extracellular connective tissue is positive as well (a). Also F N immunostaining is intense within the desmoplastic connective tissue stroma, which surrounds the tumor islands (b). Magnification, × 110. Bar=60 micron
Fig. 5 a-c. Close-up photomicrograph of a solidly growing duct
,---
b
carcinoma with TN immunoreactivity concentrated peripherally in the tumor cell clusters and no stromal staining (arrows) (a). In b another tumor shows an intense staining for TN of connective tissue fibres, whereas the tumor cells appear empty. The same tumor stained by the C L O N O - G L A D procedure, is presented in c. Note the intensely stained normal duct (left). Magnification, x 150. Bar = 30 micron Fig. 6. Grade 2 duct carcinoma showing double labeling for TN and F N using alkaline phosphatase with a black chromogen and peroxidase with a brownish-red chromogen, respectively (a). In b the same tumor is shown in a double immunofluorescence staining. The red color represents TN (TRITC) and the green F N (FITC). Magnification, × 96. Bar = 50 micron
432 stained cells only peripherally in the tumor cell clusters (Figs. 3b, 5a), and in cases of desmoplasia or solid growth pattern of the tumor, the overall cytoplasmic immunoreactivity was frequently reduced (Fig. 5 b).
Control staining Sections stained by the CLONO-GLAD procedure and frozen sections stained directly by the monoclonal antibody produced results similar to the ones obtained with the polyclonal antibody against TN (Figs. 2 a-c and 5 b c). Using the same staining procedure and the same or double the protein concentration of the one used for anti-TN, staining for wide spectrum keratin showed an expected positivity corresponding to the tumor cells but no extracellular staining. The absorbed antibody failed to stain tumor cells as well as extracellular matrix. The addition of 0.3 M NaC1 to the rabbit polyclonal anti-TN solution did not influence intensity nor localization of the staining product.
Double staining procedures Double staining studies of TN and FN were performed on the same tumors, which were used for control stainings, in order to rule out the possibility that some of the TN immunoreactivity demonstrated extracellularly originated from contaminating anti-FN antibody in the TN antibody solution. With a simple 2-layer immunological labeling method using antibodies from different species, different marker enzymes and chromogens or fluorochromes, respectively, 2 different, double-stained sections were produced, which were complementary. Tissue morphology was superior in the enzyme-stained sections, whereas localization of the staining product was optimal in the immunofluoroscence stained sections (Fig. 6a and b). TN and FN stained different fibres, confirming separate sublocalizations of the proteins extracellularly (Fig. 6 a and b).
Discussion
TN was originally discovered in human plasma as a tetrameric protein, which bound to the kringle-4 structure of plasminogen (Clemmensen et al. 1986). Up to the present the protein has been discovered in almost all human tissues (Christensen et al. 1987 and 1989) and body fluids (own unpublished observations), suggesting a function in fundamental, biological processes. Because of its close relationship with plasminogen and its stimulating effect on t-PA catalysed plasminogen activation in vitro (Clemmensen et al. 1986) TN is thought to participate in processes, where plasminogen activators are involved (e.g. tissue degradation, involution, extracellular proteolysis and cell migration) (Dano et al. 1985). This idea is further supported by the observation that TN binds strongly to sulphated polysaccharides (Clemmensen et al. 1989), of which fucoidan has been found
to enhance plasminogen activation catalysed by both u-PA and t-PA (Brunner et al. 1988). Studies of TN at the cellular level have shown that the protein is expelled from granulocytes upon stimulation (Borregaard et al. 1990). Furthermore, cultured hu'man, embryonal lung fibroblasts (WI-38) have been found to produce TN to an extracellular matrix produced by the cells and release increased amounts of TN to the growth medium after stimulation (Clemmensen et al., in press). The finding in this study of TN extracellularly within tissues with a high fibroblast concentration (e.g. desmoplasia and granulation tissue) may reflect a similar increased cellular release of TN in vivo. Desmoplasia is thought to exert a protective effect on tumor spread. This has most convincingly been shown in a routine system, where inhibition of the desmoplastic response elicited by certain melanoma cells increased tumor invasion and metastasis (Barsky et al. 1987). A number of components other than TN are present in desmoplasia. The connective tissue glycoprotein FN is significantly increased in such lesions (Lagace et al. 1985; Christensen et al. 1988), and so is elastin, vitronectin (Loridon-Rosa et al. 1988), collagen type 3 and 5 (Barsky et al. 1982; Lagace et al. 1985) and certain proteoglycans (Takeuchi et al. 1976). FN binds, like TN, to plasminogen in vitro (Salonen et al. 1985), and because of its high concentration in plasma, from which the TN used for immunization has been purified, double staining experiments for TN and FN were carried out. These showed different extracellular sublocalizations of the 2 proteins, minimizing the possibility that the TN immunoreactivity was due to a contaminating titer against FN in the TN antibody solution (Fig. 6). The differences found in cellular distribution between benign secretory cells of the breast and their malignant counterparts and the heterogeneity in staining pattern within the same tumor, have been described for other epithelial markers such as epithelial membrane antigen (Sloane et al. 1981), carcinoembryonic antigen (Kuhajda et al. 1983) and the keratins (Gusterson et al. 1982). However, in contrast to TN, none of these have been demonstrated within the extracellular compartment. The constantly growing information about the interactions of TN in vitro has so far allocated the protein as being a possible participant in the extravascular proteolytic process, which includes conversion of plasminogen to plasmin by plasminogen activators. The importance of plasmin for tissue degradation in cancer growth and spread has been stressed before (Dano et al. 1985). The identification of TN within not only the cells but also within the extracellular matrix of IBCs corresponding to the desmoplastic connective tissue response, combined with the fact that TN as the only known, naturally occurring ligand binds to kringle 4 of plasminogen, may add to the understanding of how plasminogen activation is modulated at the extracellular level.
Acknowledgements. This study was supported by grants from The Danish Cancer Societyand The Danish Medical Research Council.
433 The authors wish to thank laboratory assistants Mrs. A. Pedersen, A. Christensen, I. Bech and A. Ryberg for valuable technical assistance, and photographer Mr. B. Borgesen is gratefully acknowledged for excellent photographical help.
References Azzopardi JG (1979) Problems in breast pathology, vol 2. H.B. Saunders, London Barsky SH, Rao CN, Grotendorst GR, Liotta LA (1982) Increased content of type V collagen in desmoplasia of human breast carcinoma. Am J Pathol 108:276-283 Barsky SH, Gopalakrishna R (:1987) Increased invasion and spontaneous metastasis of BL6 melanoma with inhibition of the desmoplastic response in C57 BL/6 mice. Cancer Res 47 : 1663 1667 Bloom HJG, Richardson WW (1957) Histological grading and prognosis in breast cancer. Br J Cancer l 1:359-365 Borregaard N, Christensen L, Bjerrum OW, Birgens HS, Clemmensen I (1990) Secretory granules. Identification of a highly mobilizable subset of human neutrophil granules, which contains tetranectin and latent alkaline phosphatase. J Clin Invest 85: 408-417 Brunner G, Schirrmacher V (1988) Stimulation of plasminogen activation by sulphated glycosaminoglycans: a possible mechanism for extracellular matrix penetration by tumor cells. Clin Exp Met 6 [Suppl 1] :43 (abstract) Christensen L, Strange L (1987) Universal immunoperoxidase staining protocol to optimize the use of polyclonal and monoclonal antibodies. J Histotechnol 10:11-15 Christensen L, Johansen N, Jensen BA, Clemmensen I (1987) Immunohistochemical localization of a novel, human plasma protein, tetranectin, in human endocrine tissues. Histochemistry 87:195 199 Christensen L, Nielsen M, Andersen J, Clemmensen I (:1988) Stromal fibronectin staining pattern and metastasizing ability of human breast carcinoma. Cancer Res 48:6227-6233 Christensen L, Clemmensen I (1989) Tetranectin immunoreactivity in normal human tissues. An immunohistochemical study of exocrine epithelia and mesenchyme. Histochemistry 92:29-35 Clemmensen I, Andersen RB (:1982) Different molecular forms of fibronectin in rheumatoid synovial fluid. Arthritis Rheum 25:25-31 Clemmensen I, Petersen L, Kluft C (1986) Purification and characterization of a novel, oligomeric, plasminogen kringle 4 binding protein from human plasma. Eur J Biochem 156 : 327-333 Clemmensen I (1989) Interaction of tetranectin with sulphated polysaccharides and trypan blue. Scand J Clin Lab Invest 49 : 7:19-725 Clemmensen I, Lurid L, Christensen L, Andreasen P (1990) Tetranectin: production and extracellular matrix deposition by human embryonal fibroblasts. Eur J Biochem (in press) Cordell JL, Falini B, Erber W, Ghosh A, Abdulaziz Z, MacDonald S, Pulford KAF, Stein H, Mason DY (:1984) Immunoenzymatic labelling of monoclonal antibodies using immune complexes of alkaline phosphatase and monoclonal anti-alkaline phosphatase (APAAP complexes). J Histochem Cytochem 32:219-229 Dano K, Andreasen PA, Grondahl-Hansen J, Kristensen P, Niel-
sen LS, Skriver L (1985) Plasminogen activators, tissue degradation and cancer. Adv Cancer Res 44:139-266 Fuhlendorff J, Clemmensen I, Magnusson S (1987) The primary structure of a kringle 4 binding protein: tetranectin. Structure homology with asialoglycoprotein receptors and a proteoglycan core protein. Biochemistry 26:6757-6764 Gusterson BA, Warburton M J, Mitchell D, Ellison M, Neville AM, Rudland PS (1982) Distribution of myoepithelial cells and basement membrane proteins in the normal breast and in benign and malignant breast disease. Cancer Res 42:4763-4770 Haagensen CD (1986) Diseases of the breast, 3rd edn. W.B. Saunders, Philadelphia London Toronto Holund B, Clausen PP, Clemmensen I (1981) The influence of fixation and tissue preparation on the immunohistochemical demonstration of fibronectin in human tissues. Histochemistry 72 : 291-299 Holund B, Clemmensen I (1982) The value of hyaluronidase treatment of different tissues before demonstration of fibronectin by the indirect immunoperoxidase technique. Histochemistry 76: 517-525 Jensen BA, Clernmensen I (1988) Plasma tetranectin is reduced in cancer and related metastasia. Cancer 62:3222-3227 Kluft C, Los P, Clemmensen I, Brommer EJP, Leuven JAG, Boks AL, Vellenga E (I 989) Quantitation of plasma levels of tetranectin effects of oral contraceptives, pregnancy, treatment with L-asparaginase and liver cirrhosis. Thromb Haemostas 62: 792796 Kuhajda FP, Offutt LE, Mendelsohn G (1983) The distribution of carcinoembryonic antigen in breast carcinoma. Cancer 52:1257-1264 Lagace R, Grimaud J-A, Schwurch W, Seemayer TA (1985) Myofibroblastic stromal reaction in carcinoma of the breast: variations of collagenous matrix and structural glycoproteins. Virchows Arch [A] 408 : 49-59 Loridon-Rosa B, Vielh P, Cuadrado C, Burtin P (1988) Comparative distribution of fibronectin and vitronectin in human breast and colon carcinomas. An immunofluorescence study. Am J Clin Pathol 90: 7-16 Salonen EM, Saksela O, Vartio T, Vaheri A, Nielsen LS, Zeuthen J (1985) Plasminogen and tissue type plasminogen activator bind to immobilized fibronectin. J Biol Chem 260:12302-12307 Saunders S, Bernfiels M (1988) Cell surface proteoglycan binds mouse mammary epithelial cells to fibronectin and behaves as a receptor for interstitial matrix. J Cell Biol 106:423-430 Scopsi L, Bock E, Larsson L-I (1986) Monoclonal antibody immunocytochemistry: novel method extending usefulness of monoclonat antibodies for antigen visualization. Eur J Cell Biol 41:97-10:1 Sloane JP, Ormerod MG (1981) Distribution of epithelial membrane antigen in normal and neoplastic tissues and its value in diagnostic tumor pathology. Cancer 47:1786-1795 Sobin L (1981) Histological typing of breast tumors, 2nd edn. WHO, Geneva Sternberger LA (1979) Immunocytochemistry, 2nd edn. John Wiley, New York Takeuchi J, Sobue M, Sato E, Shamoto M, Miura K, Nakagaki S (1976) Variation in glycosaminoglycan components of breast tumors. Cancer Res 36 : 2133-2139 Wang B-L, Larsson L-I (:1985) Simultaneous demonstration of multiple antigens by indirect immunofluorescence or immunogold staining. Histochemistry 83 : 47-56
