- Email: [email protected]
Expression of fibronectin and its integrin receptor α5β1 in canine mammary tumours
Research in Veterinary Science 1994, 57, 358-364
Expression of fibronectin and its integrin receptor a5131 in canine mammary tumours L. PElqA, A. NIETO, M. D. PEREZ ALENZA, A. RODRIGUEZ, M. A. SANCHEZ, M. CASTAlqO,
Department of Animal Pathology II, Veterinary Faculty, Complutense University of Madrid, 28040 Madrid, Spain
Fibroneetin and its integrin receptor ~t5131 were studied by immunohistochemical methods in five normal canine mammary glands, four dysplastic glands and 18 mammary tumours. The aim of the study was to evaluate the possible changes in the a5131 integrin receptor and its ligand fibronectin in relation to the metastatic capacity of canine mammary neoplasms. The immunostaining of ct5~1 was very uniform in the hyperplastie glands but uneven in the mammary tumours. The expression of 0t5 and 131 was diminished in metastatic tumours but there were some ~5-positive cells with pronounced features of malignancy and immaturity. Stromal fibronectin was increased in most cases and cytoplasmic staining of fibronectin was observed in epithelial and myoepithelial cells in mammary neoplasms but not in normal or dysplastic mammary tissue. There was no relationship between the content of ct5131 and the expression of fibronectin in canine mammary tumours. NEOPLASTIC cells interact with the extracellular matrix (Liotta 1986) during the process of invasion and the interaction is mediated by cell surface receptors of the integrin family. Integrins constitute a family of transmembrane receptors that participate in cell-to-cell and cell-to-matrix interactions. They are heterodimeric proteins formed by the non-covalent association of different tx and 13 subunits (Hynes 1987, Ruoslahti and Pierschbacher 1987). The 131 integrin subfamily, also called VLA and t~ integrins, are ctl31 receptors that mediate adhesion to extracellular matrix proteins such as fibronectin, collagens and laminin (Hemler et al 1987). The classical fibronectin receptor ~t5131
recognises the RGD sequence (Arg-Gly-Asp) of fibronectin. The results of several studies (Terranova et al 1982, Horan et al 1985, Humphries et al 1986, Wewer et al 1987, Castronovo et al 1990) have suggested that integrins are important in tumour invasion and metastasis, and changes in the integrin expression of neoplastic cells have been demonstrated in vivo and in vitro (Albelda et al 1990). There have been studies of integrins in human mammary tumours (Zutter et al 1990, D'Ardenne et al 1991, Pignatelli et al 1991) but the authors have found no reports of studies in canine mammary tumours and there is disagreement between some of the findings in different studies of human turnouts. Fibronectin is a large glycoprotein found in the connective tissue stroma and in the basement membranes of human mammary glands (Natali et al 1984, Christensen et al 1985). It plays an important role in cell adhesion and in the migration of both normal and neoplastic cells (Ruoslahti 1988), but many of the results of studies of fibronectin in neoplasms are either contradictory or differ according to the cells or tissues employed. The synthesis of fibronectin and the expression of its receptor are comodulated (Varani and Chakrabarty 1991) and the purpose o f this investigation was to study the expression of fibronectin and its main receptor ot5~1 in canine mammary tumours by immunohistochemical methods.
Materials and methods
Clinical studies A physical and radiographic examination of each dog's thorax was made at the first clinical examination. Follow-up examinations for the
358
359
Fibronectin and a5 fll integrin in canine mammary tumours
recurrence of the tumour or metastasis were carfled out at intervals of three months for 18 months after surgery. Histological studies
Samples were collected at surgery or post mortem from five normal canine mammary glands, four dysplastic glands and 22 mammary tumours (18 primary tumours and four pulmonary metastases). In most cases the axillary or inguinal lymph nodes were also collected for histological examination for metastases. The tissues were fixed in 10 per cent neutral formalin and paraffin-wax embedded sections were cut at 5 ~tm for the histological and immunohistochemical studies. Sections stained with haematoxylin and eosin were used for the World Health Organization system of classification (Hampe and Misdorp 1974). According to the histological grade of the malignancy and its metastatic capacity, the samples were classified as hyperplastic lesions, benign tumours, non-metastatic malignant tumours - that is, without lymphatic infiltration or, if recurrent, without metastasis - or metastatic malignant tumours - that is, primary tumours with lymphatic infiltration or metastatic turnouts. Immunohistochemical studies
Integrin subunits were detected in the 5 ~tm sections by use of the peroxidase-anti-peroxidase technique (PAP) with the following primary antibodies in phosphate buffered saline (PBS): rabbit anti-human [31 integrin subunit (cytoplasmic domain) (Chemicon International AB1938) at a
working dilution of 1:75; rabbit anti-human ~t5 integrin subunit (cytoplasmic domain) (Chemicon International, AB1928), at a working dilution of 1:75; and rabbit anti-human fibronectin (Chemicon International, AB1940) at a working dilution of 1:100. The second antibody was porcine antirabbit immunoglobulin (Dakopatts Z196) at a working dilution of 1:100. The immunoreaction was revealed by treating the sections with rabbit peroxidase-antiperoxidase complex (Dakopatts Zl13) at a working dilution of 1:64, followed by diaminobenzidine (DAB, Sigma D-5905), 20 mg m1-1 in PBS, and then with 100 ~tl of 30 per cent hydrogen peroxide, after which they were counterstained with haematoxylin. For the detection of fibronectin the sections were digested with 0.1 per cent trypsin (Sigma T-8128) at 37°C, pH 7.8 before the immunohistochemical staining. All the staining procedures included negative control sections incubated in PBS without the primary antibody. The immunohistochemical staining of fibronectin and integrin subunits was assessed semiquantitatively as described by D'Ardenne et al (1991). According to the intensity and extent of the staining the reaction was graded as: (-) absent, (+) staining of a few tumour cells for integrins or very weak staining of stroma for fibronectin, (++) staining of a moderate number of tumour cells or focal staining of the stroma, or moderate diffuse staining of both, or (+++) intense staining of all the tumour cells and of the stroma. Results The results are summarised in Tables 1 and 2.
TABLE 1: Fibronectin and c~5~1 immunostaining in normal canine mammary gland, dysplasias and benign tumours Fibronectin
Cases
Histological type
131
c~5
Stroma
E/MYO
+++
++
+
-/-
Dysplasias
1 2 3 4
Lobular hyperplasia Lobular hyperplasia Hyperplasia + epitheliosis Sclerosing adenosis
+ ++ ++ +++
+ + + +
+ + + +
-/+ -/+ -/-/-
Benign tumours
5 6 7 8 9
Complex adenoma Complex adenoma Benign mixed tumour Benign mixed tumour Benign mixed tumour
++ +++ +++ +++ ++
++ +++ +++ +++ ++
+++ +++ +++ +++ +++
++/+++ +++/++ +++/+++ +++/+++ ++/4-+
Five normal glands
- Absent, + Slight, ++ Moderate, +++ Strong, E Epithelial/MYO Myoepithelial cells
360
L. Pe~a, A. Nieto, M. D. Perez Alenza, A. Rodriguez, M. A. Sanchez, M. Castaho
TABLE 2: Fibronectin and c~5~1 immunostaining in malignant canine mammary tumours Cases Non-metastatic malignant t u m o u r s
Histological t y p e
10
Simple tubular adenocarcinoma C o m p l e x tubular adenocarcinoma Papillary adenocarcinoma Papillary adenocarcinoma S q u a m o u s cell carcinoma Malignant m i x e d tumour
11 12 13 14 15 Metastatic malignant t u m o u r s
16 17 18 19 20 21 22
Papillary adenocarcinoma Malignant m i x e d tumour S q u a m o u s cell carcinoma Solid c a r c i n o m a Solid c a r c i n o m a Solid c a r c i n o m a Spindle cell c a r c i n o m a
L y m p h a t i c emboli 131 or5 FN
P u l m o n a r y metastasis 131 ct5 FN
++
++
++
+
None
None
None
+++
++
++
++
+
+
++
+
+++ + ++
++ + + + ++
++ ++ + + None
+ + + + None
++ + + + None
++ None None + +
+ None None + +
None None +
Fibronectin S t r o m a E/MYO
[31
c~5
+++
++
++
++/++
+++
++
+++
++/+++
++
++
+++
++
+++
++
+++
++/+++
+
-
-
-
+++
+++
+++
++
+++
+
+++
+
+
++ + + ++ +++
+ + + ++ +++
- Absent, + Slight, ++ M o d e r a t e , + + + Strong, FN Fibronectin, E Epithelial/MYO Myoepithelial cells (if both cell types are present)
Normal mammary gland
Staining for 131 was present in the epithelial cells of ducts, ductules and alveoli. In the myoepithelial and basal cells the staining was more intense along a band in contact with the basement membrane. In the stroma, 131 was present in fibroblasts, endothelial cells and smooth muscle cells. The intensity of the reaction of 131 in the smooth muscle cells of arterioles served as an internal control for the immunostaining in all the samples. Staining for the ct5 integrin subunit was moderate in the epithelial cells but more intense in the myoepithelial cells. In the stroma the fibroblasts were positive and both endothelial cells mad smooth muscle cells were slightly positive. Fibronectin was present in very small amounts in the basement membranes but was more abundant and distributed diffusely in the connective tissue stroma. It was not detected in epithelial or myoepithelial cells but was abundant in the cytoplasm of fibroblasts in the stroma. Dysplastic mammary gland
In glands with lobular hyperplasia, there was a decreased but uniform expression of 131 and ct5 in
all the samples (cases 1 to 3). The myoepithelial or basal cells contained more 131 and ct5 than did the epithelial cells. In one gland with lobular hyperplasia (case 3), areas of regular epithelial proliferation ('epitheliosis') within pre-existing structures did not differ in their integrin expression from other hyperplastic areas. In contrast with normal mammary glands, two glands with lobular hyperplasia (cases 1 and 2) contained fibronectin not only in the basement membranes or stroma but also in the cytoplasm of myoepithelial cells. In one gland with simple tubular sclerosing adenosis (case 4), there was weak staining for the a5 subunit, strong staining for the 131 subunit, and no increase in fibronectin in the desmoplastic stroma. Mammary tumours Benign mammary tumours. In the five benign mammary tumours, 131 immunostaining was moderate (++) to intense (+++), as in normal mammary tissue. The immunostaining for c~5 was similar to that in normal mammary gland in two of the tumours (cases 5 and 9) but increased in a complex adenoma (case 6) and in two benign mixed tumours (cases 7 and 8). In four of the five
Fibronectin and a5fll integrin in canine mammary tumours
benign neoplasms (two complex adenomas and two benign mixed tumours, cases 5 to 8) the myoepithelial ceils in basal locations and in areas of proliferation were intensely stained for [31 and ct5. In the benign mixed tumours the chondrocytes were also stained for [31 and Gt5 integrins. Uneven immunostaining for fibronectin was observed in all the benign tumours and was stronger than in the normal or dysplastic mammary glands. There was diffuse staining for fibronectin in the stroma and there were fibronectinpositive membrane fragments in focal areas in which proliferation of myoepithelial cells was evident. Fibronectin was not increased in desmoplastic areas in one benign tumour (case 9). There was staining for fibronectin in the cartilaginous matrix of only one of the three benign mixed tumours (case 8) and in all three tnmours the chondrocytes were negative. In two tumours (cases 5 and 9) some of the epithelial cells were strongly stained for fibronectin but others were negative. In cases 6 to 8 fibronectin was present in the cytoplasm of all the epithelial cells. In all the benign tumours the myoepithelial cells were strongly positive for fibronectin. Non-metastatic malignant tumours. In four of the
six tumours in this group (cases 10 to 13) the amount of integrin did not differ greatly from that in normal mammary gland. In one squamous cell carcinoma (case 14), the staining for 131 and ct5 was weaker and in one mixed malignant tumour (case 15), there was strong staining for a5 in the myoepithelial cells. Chondrocytes in newly formed cartilaginous tissue had small amounts of both 131 and a5 subunits. In two of the six non-metastatic malignant tumours (cases 10 and 11), the staining for integrin subunits in proliferated myoepithelium or in cells in typical basal locations was more evident than in epithelial cells. In three tumours (cases 10, 12 and 13) some cells with obvious characteristics of malignancy (lack of differentiation, large nuclei and multiple nuclei), there was very intense staining for a5 and in case 12 these cells also stained intensely for [31. In one non-metastatic malignant tnmour (case 10) there were non-adherent epithelial cells in the connective tissue without notable features of malignancy but with intense staining for a5. Fibronectin was increased in the stroma and was present in the epithelial and myoepithelial
361
cells of five of the six non-metastatic malignant tumours (cases 10 to 13 and 15). Fibronectin formed membrane fragments in two complex malignant tumours (cases 11 and 12). In two adenocarcinomas (cases 11 and 13), there were thickened fibronectin-positive basement membranes and the most positive cells were the myoepithelial cells and non-cohesive epithelial cells. In three tumours (cases 11 to 13) there was strong staining for fibronectin in the cytoplasmic membrane of the epithelial cells. In one squamous cell carcinoma (case 14) with very weak staining for [31 and a5, fibronectin was almost absent. Metastatic malignant tumours. Four of the seven
tumours (cases 17 to 20) had weak staining for [31 and ct5. In case 16 only ct5 was weakly staining and in case 21 only the [31 subunit was weakly staining. In one spindle cell carcinoma (case 22) both of these subunits were strongly stained. The staining for integrin was more intense in the basal cells of two metastatic malignant neoplasms (cases 18 and 19). In spite of the low levels of or5 in the metastatic tumours, there were three tumours in which obviously malignant cells, especially multinucleated cells and very immature cells, were strongly stained for (~5. Strongly positive cells were also found in some malignant non-metastatic tumours. ct5-positive non-cohesive epithelial ceils were also found in the stroma of all the metastatic tumours and some of them were also positive for [31. The staining for fibronectin in metastatic malignant tumours was very uneven; it was increased in the stroma in most cases (cases 16 to 18 and 22) and all the tumours had neoplastic fibronectin-positive cells. Immunostaining for fibronectin was observed in the epithelial cells of three solid carcinomas (cases 19 to 21) in which the staining for fibronectin in the stroma was weak and there was a decrease in 131 and ~t5. In the spindle cell carcinoma (case 22) in which there was an increase in [31 and ot5, there was moderate staining for flbronectin in the cytoplasm. Lymphatic emboli. There were neoplastic emboli
in the lymphatics of six of the seven metastatic tumours (cases 16 to 21). In all cases, the intensity of staining for [31 and or5 in the metastatic cells was very uneven. In one solid carcinoma (case 20) the cells strongly positive for ct5 were migrating
362
L. Peha, A. Nieto, M. D. Perez Alenza, A. Rodriguez, M. A. Sanchez, M. Casta~o
through the walls of the lymphatic capillaries. In two tumours (cases 17 and 20) neoplastic cells inside the lymphatics and in contact with endothelial cells were arranged in palisades and were very strongly stained for [31 and a5. The presence of fibronectin in metastatic cells within emboli was very uneven. In five cases (cases 16, 17 and 19 to 21) most of the metastatic cells were weakly stained for fibronectin and very few were strongly positive. In the squamous cell carcinoma (case 18), the fibronectin content of the metastatic cells was also distributed unevenly but it was stained intensely in most cells, especially in the cytoplasmic membrane. metastases. In all four pulmonary metastases of mammary carcinomas (cases 17, 18, 21 and 22) the immunostaining for both integrin subunits and for fibronectin was very weak or negative, except in migrating cells in case 17, in which there was strong staining for the or5 subunit. Pulmonary
Discussion Fibronectin and its receptors regulate several cell functions, including cell adhesion, polarity, migration, differentiation (Ruoslahti 1988), morphology, the assembly of extracellular matrix, the formation of bundles of microfilaments (Akiyama et al 1989) and the proliferation of cells (Giancotti and Ruoslaht 1990). The aim of this work was to evaluate the possible changes in integrin ct5131 and its ligand fibronectin in relation to the metastatic capacity of canine mammary neoplasms. In normal canine mammary tissue the staining for 131 and (z5 was similar to that described in human mammary tissue (Zutter et al 1990). The staining for 131 in the epithelial and myoepithelial cells of human mammary tissue (Zutter et al 1990) has been interpreted as indicating a greater expression of integrins in the myoepithelial cells (Pignatelli et al 1991). Smooth muscle cells and fibroblasts in the dermis are strongly positive for 131 (De Strooper et al 1989). D'Ardenne et al (1991) found weak staining for the cz5 subunit in human mammary epithelial cells but there was more intense staining for the (z5 subunit in the canine myoepithelial cells. The presence of fibronectin in the stroma and weak staining for it in basement membranes of normal human breast tissue was reported by Natali et al (1984). The absence of fibronectin from the
cytoplasm of epithelial and myoepithelial cells in the present study is also in agreement with these authors' findings. An interesting difference between the hyperplastic and neoplastic lesions was the uniformity of the immunostaining for [31 and ~t5 in the former and its unevenness in the latter. Unevenness of integrin expression has been reported for oil and or6 subunits in breast cancer (D'Ardenne et al 1991) and in human malignant melanoma (Albelda et al 1990). Zutter et al (1990) found high levels of expression of the collagen receptor ct2131, which is associated with the orderly, regulated proliferation of epithelial cells. Although hyperplasia of the canine mammary gland are regulated proliferations, a decrease in the integrin [31 subunit was observed in the present work. In the benign mammary tumours, complex adenomas and mixed benign neoplasms, there was strong staining for both 131 and or5 in myoepithelial cells and the staining for fibronectin was also increased. D'Ardenne et al (1991) found a strong expression of integrin subunits in the myoepithelial or basal cells and even different subunits expressed in the two cell types. An increase in fibronectin was also found in the stroma of some malignant tumours without myoepitheiial proliferation, and in 17 of the 22 samples of dysplasfic glands and tumours, the epithelial cells stained positively for this glycoprotein. This finding suggests that both the epithelial and the myoepithelial cells in the canine mammary gland secrete fibronectin when they have undergone pre-neoplastic or neoplastic changes. Cytoplasmic staining for fibronecfin has been reported in malignant human mammary tumours (Natali et al 1984, Chfistensen et al 1985), although other authors have reported that the expression of fibronectin is diminished in breast tumours (Hynes and Yamada 1982). In four of the six non-metastatic malignant tumours the expression of integfin was uneven but normal, whereas the expression of at least one integrin subunit was decreased in six of the seven metastatic tumours and both were decreased in four of the seven. This decrease most often involved the c~5 subunit. Pignatelli et al (1991) found a similar reduction in the expression of the 131 subunit in human mammary tumours with a high grade of malignancy. Like the present findings in metastatic tumours, Zutter et al (1990) found that the expression of a5131 was decreased
Fibronectin and a5flI integrin in canine mammary tumours
363
in poorly-differentiated human mammary carcino- have been found in induced fibrocytic tumours mas but only slightly decreased in the well-differ- with a well-organised fibronectin matrix. This entiated lesions. Other authors (D'Ardenne et al suggests that the synthesis of fibronectin and its 1991) did not find visible differences in the cc5 receptor expression are comodulated (Varani and immunostaining of neoplastic mammary epithe- Chakrabarty 1991). In the present work, no relationship was found between the content of ct5131 lium, the staining being generalised and weak. Decreases or increases in the expression of and the expression of fibronectin, because in 14 of other integrin subunits have been reported in 18 of the tumours the fibronectin in the stroma human breast tumours but the results are contra- was increased while the expression of both c~5 and dictory (Horan et al 1985, Zutter et al 1990, [31 was normal or decreased. Fibronectin was almost absent from only two tumours with very D'Ardenne et al 1991). The finding of cells having marked criteria of low levels of ~t5 and 1~1. It is known that the malignancy which also expressed high levels of deposition of matrix is disturbed in malignant cells or5 in six of 13 malignant tumours which other- (Ruoslahti 1988) and most investigators have wise stained weakly for this subunit appears to found stromal fibronectin to be increased in maligcontradict the evidence of a relationship between nant tumours (D'Ardenne and Barnard 1989). The the metastatic phenotype and the loss of ct5, that is intensity of staining of fibronectin in breast that in human pancreatic adenocarcinoma the tumours has been found to be correlated with the expression of ct5131 is inversely correlated with a degree of anaplasia and with the independent growth of cells (Christensen et al 1985). In the malignant phenotype (Rosendhal et al 1992). Other authors have reported a greater expres- canine tumours there was no correlation between sion of ct5[~l in immature cells than in those with histological malignancy and increased amounts of a metastatic phenotype; Federico et al (1992) fibronectin, although there was a strong expresreported a possible relationship between the loss sion of fibronectin in non-cohesive epithelial cells. The expression of integrin in cells passing into of the fibronectin receptor and the acquisition of the mature phenotype in cells of the myeloid line- the vascular system is of special interest. It is age and it has been demonstrated that the terminal limited to malignant cells which interact with the differentiation of keratinocytes involves a loss of vascular endothelium and/or its basement memadhesiveness to fibronectin (and a loss of Gt5131). brane (Kramer et al 1991). In the present work Cardarelli et al (1988) suggested that in thymo- cells penetrating lymphatic capillaries stained cytes the loss of fibronectin receptors is related to positively for the fibronectin receptor but the the functional maturation or programmed death of expression of integrin was lower in cells in metastatic emboli, except in cells migrating from the cells. The exact role of the cc5131 fibronectin receptor lymphatics. It is known that changes can occur in in cell migration is not clear. It has been demon- tumour cells during the metastatic process and strated that an antibody against the chicken phenotypic heterogeneity has been demonstrated fibronectin receptor complex may inhibit the among primary and metastatic lesions, but little is migration of cells (Ruoslahti 1988) but others known about the changes in the expression of intehave reported that the a5131 receptor retards the grin in this process (Kramer et al 1991). migration of cultured fibroblasts (Akiyama et al The reduction in the ot5131 fibronectin receptor 1989). Furthermore, transformed hamster ovary in metastatic canine mammary tumours may be cells with an increased expression of ct5131 migrat- necessary to obtain independence of anchorage ed less than control cells, suggesting that ct5131 has (Giancotti and Ruoslahti 1990). The authors cona role in the pericelhilar assembly of fibronectin sider that the major expression of or5 found in (Giancotti and Ruoslahti 1990). In the present some independent cells and in cells with a high study, malignant cells with high levels of ct5 and degree of malignancy could be related to their intense staining for fibronectin in the cytoplasm immaturity or to an intense secretion of were observed particularly in one metastatic fibronectin, rather than to this receptor's role in tumour, which is consistent with this role of the migration. In the canine mammary tumours a relafibronectin receptor rather than with a migration tionship between the deposition of fibronectin and process. the expression of cc5131 was found only in benign Increased amounts of the fibronectin receptor tumours in which there was a strong expression of
364
L. PeKa, A. Nieto, M. D. Perez Alenza, A. Rodriguez, M. A. Sanchez, M. Castafio
the c~5 subunit and cytoplasmic fibronectin in myoepithelial cells. Studies of possible changes in other integrin subunits in mammary tumours, such as ot4131, may bring further understanding of the results of this study of fibronectin and its receptor ot5131. Acknowledgements The authors thank Dr Bruce Belshaw and Miss Ana Rodriguez for a careful review of the manuscript. References AKIYAMA, S. K., YAMADA, S. S., CHEN, W. T. & YAMADA, K. M. (1989) Analysis of fibronectin receptor function with monoclonal antibodies: roles in cell adhesion, migration, matrix assembly and cytoskeletal organization. Journal of Cell Biology 11}9, 863-874 ALBELDA, S. M., METTE, S. A., ELDER, D. E., STEWART, R., DAMJANOVICH, L., HERLY, M. & BUCK, C. A. (1990) Integrin distribution in malignant melanoma: association of the 133 subunit with tumor progression. Cancer Research 50, 6757-6764 CARDARELLI, P. M., CRISPE, I. N. & PIERSCHBACHER, M. D. (1988) Preferential expression of fibronectin receptors on immature thymocytes. Journal of CeUBiology 11}6,2183-2190 CASTRONOVO, V., COLIN, C., CLAYSMITH, A. P., CHEN, P. H. S., LIFRANGE, E., LAMBOTrE, R., KRUTZSCH, H., LIOTTA, L. A. & SOBEL, M. E. (1990) Immunodetection of the metastasis-associated laminin receptor in human breast cancer cells obtained by fineneedle aspiration biopsy. American Journal of Pathology 137, 1373-1381 CHRISTENSEN, L., NIELSEN, M., HOLUND, B. & CLEMMENSEN, I. (1985) In vivo demonstration of cytoplasmic fibmnectin in human breast carcinomas. VirchowArchivA (PathologyAnatomy) 407, 337-346 D'ARDENNE, A. J. & BARNARD, N. J. (1989) Paucity of fibroneefin in invasive lobular carcinoma of breast. Journal of Pathology 157, 219-224 D'ARDENNE, A. 1., RICHMAN, P. I., HORTON, M. A., MCAULAY, A. E. & JORDAN, S. (1991) Co-ordinate expression of the alpha-6 integrin laminin receptor subunit and laminin in breast cancer. Journal of Pathology 165, 213-220 DE STROOPER, B., VAN DER SCHUEREN, B., JASPERS, M., SAISON, M., SPAEPEN, M., VAN LEUVEN, F., VAN DEN BERGHE, H. & CASSIMAN, J.-J. (1989) Distribution of the I~1 subgroup of the integrins in human cells and tissues. Journal of Histoehemistry and Cytochemistry 37, 299-307 FEDERICO, M. H. H., YAMAMOTO, M., MARIA, D. A., KATAYAMA, M. L. H., SONOHARA, S., ROELA, M. A., KOIKE, M. A. & BRENTANI, M. M. (1992) Fibronectin receptor expression in human leukemic cells. Clinical and Experimental Metastasis 10, Supplement
1,40
GIANCOTI~, F. G. & RUOSLAHTI, E. (1990) Elevated levels of the e~5~l fibronectin receptor suppress the transformed phenotype of Chinese hamster ovary cells. Cell 60, 849-859 HAMPE, J. F. & MISDORP, W. (1974) Tumors and dysplasias of the mammary gland. Bulletin of the World Health Organization 50, 111-133 HEMLER, M. E., HUANG, C. & SCHWARTZ, L. (1987) The VLA protein family. Journal of Biological Chemistry 262, 3300-3309 HORAN, H. P., THOR, A , SCHLOM, J., RAO, C. N. & LIOTTA, L. A. (1985) Expression of laminin receptor in normal and carcinomatous tissues as defined by a monoclonal antibody. Cancer Research 45, 2713-2719 HUMPHRIES, M. J., OLDEN, K. & YAMADA, K. M. (1986) A synthetic peptide from fibronectin inhibits experimental metastasis of murine melanoma cells. Science 233, 467-470 HYNES, R. O. (1987) Integrins: a family of cell surface receptors. Cell 48, 549-554 HYNES, R. O. & YAMADA, K. M. (1982) Fibronectins: multifunctional modular glycoproteins. Journal of Cell Biology 95, 369-377 KRAMER, R. H., ENENSTEIN, 1., RAMOS, D. M., VU, M. P. & CHENG, J. F. (1991) The role of integrin receptors in tumor cell adhesion to the microvasculature. In Microcirculation in Cancer Metastasis. Boca Raton, CRC Press. pp 145-160 LIOTTA, L A. (1986) Tumor invasion and metastasis-role of the extracellular matrix. Cancer Research 46, 1-7 NATALI, P. G., GIACOMINI, P., BIGOTTI, G , NICOTRA, M. R., BELLOCI, M. & DE MARTINO, C. (1984) Heterogeneous disWibution of actin, myosin, fibronectin and basement membrane antigens in primary and metastatic human breast cancer. Virchow Archiv A (Pathology Anatomy) 405, 69-83 PIGNATELLI, M., HANBY, A. M. & STAMP, G. W. H. (1991) Low expression of 131, ~x2 and {x3 subunits of VLA integrins in malignant mammary tumours. Journal of Pathology 165, 25-32 ROSENDHAL, A., WEINEL, R. J., NEUMANN, K., ERB, D. & ROTHMUND, M. R. (1992) VLA-receptors: expression and function in human pancreatic adenoearcinoma and tumour cell lines. Clinical and Experimental Metastasis 10, Supplement 1, 37 RUOSLAHTI, E. (1988) Fibronectin and its receptors. Annual Review of Biochemistry 57, 375-413 RUOSLAHTI, E. & PIERSCHBACHER, M. D. (1987) New perspectives in cell adhesion: RGD and integrins. Science 238, 491-497 TERRANOVA, V. P., LIOTTA, L. A., RUSSO, R. & MARTIN, G. R. (1982) Role of laminin in the attachment and metastasis of routine tumor cell. Cancer Research 42, 2265-2269 VARANI, J. & CHAKRABARTY, S. (1991) Changes in extracellular matrix during transformation and differentiation. In Microcirculation in Cancer Metastasis. Boca Raton, CRC Press. pp 1-22 WEWER, V. M., TARABOLETTI, G., SOBEL, M. E., ALBRECHTSEN, R. & LIOTTA, L. A. (1987) Role of laminin receptor in tumor cell migration. Cancer Research 47, 5691-5698 ZUTTER, M. M., MAZOUJIAN, G. & SANTORO, S. A. (1990) Decreased expression of integrin adhesive protein receptors in adenocarcinoma of the breast. American Journal of Pathology 137, 863-870
Received December 8, 1993 Accepted June 6, 1994
Expression of fibronectin and its integrin receptor a5131 in canine mammary tumours L. PElqA, A. NIETO, M. D. PEREZ ALENZA, A. RODRIGUEZ, M. A. SANCHEZ, M. CASTAlqO,
Department of Animal Pathology II, Veterinary Faculty, Complutense University of Madrid, 28040 Madrid, Spain
Fibroneetin and its integrin receptor ~t5131 were studied by immunohistochemical methods in five normal canine mammary glands, four dysplastic glands and 18 mammary tumours. The aim of the study was to evaluate the possible changes in the a5131 integrin receptor and its ligand fibronectin in relation to the metastatic capacity of canine mammary neoplasms. The immunostaining of ct5~1 was very uniform in the hyperplastie glands but uneven in the mammary tumours. The expression of 0t5 and 131 was diminished in metastatic tumours but there were some ~5-positive cells with pronounced features of malignancy and immaturity. Stromal fibronectin was increased in most cases and cytoplasmic staining of fibronectin was observed in epithelial and myoepithelial cells in mammary neoplasms but not in normal or dysplastic mammary tissue. There was no relationship between the content of ct5131 and the expression of fibronectin in canine mammary tumours. NEOPLASTIC cells interact with the extracellular matrix (Liotta 1986) during the process of invasion and the interaction is mediated by cell surface receptors of the integrin family. Integrins constitute a family of transmembrane receptors that participate in cell-to-cell and cell-to-matrix interactions. They are heterodimeric proteins formed by the non-covalent association of different tx and 13 subunits (Hynes 1987, Ruoslahti and Pierschbacher 1987). The 131 integrin subfamily, also called VLA and t~ integrins, are ctl31 receptors that mediate adhesion to extracellular matrix proteins such as fibronectin, collagens and laminin (Hemler et al 1987). The classical fibronectin receptor ~t5131
recognises the RGD sequence (Arg-Gly-Asp) of fibronectin. The results of several studies (Terranova et al 1982, Horan et al 1985, Humphries et al 1986, Wewer et al 1987, Castronovo et al 1990) have suggested that integrins are important in tumour invasion and metastasis, and changes in the integrin expression of neoplastic cells have been demonstrated in vivo and in vitro (Albelda et al 1990). There have been studies of integrins in human mammary tumours (Zutter et al 1990, D'Ardenne et al 1991, Pignatelli et al 1991) but the authors have found no reports of studies in canine mammary tumours and there is disagreement between some of the findings in different studies of human turnouts. Fibronectin is a large glycoprotein found in the connective tissue stroma and in the basement membranes of human mammary glands (Natali et al 1984, Christensen et al 1985). It plays an important role in cell adhesion and in the migration of both normal and neoplastic cells (Ruoslahti 1988), but many of the results of studies of fibronectin in neoplasms are either contradictory or differ according to the cells or tissues employed. The synthesis of fibronectin and the expression of its receptor are comodulated (Varani and Chakrabarty 1991) and the purpose o f this investigation was to study the expression of fibronectin and its main receptor ot5~1 in canine mammary tumours by immunohistochemical methods.
Materials and methods
Clinical studies A physical and radiographic examination of each dog's thorax was made at the first clinical examination. Follow-up examinations for the
358
359
Fibronectin and a5 fll integrin in canine mammary tumours
recurrence of the tumour or metastasis were carfled out at intervals of three months for 18 months after surgery. Histological studies
Samples were collected at surgery or post mortem from five normal canine mammary glands, four dysplastic glands and 22 mammary tumours (18 primary tumours and four pulmonary metastases). In most cases the axillary or inguinal lymph nodes were also collected for histological examination for metastases. The tissues were fixed in 10 per cent neutral formalin and paraffin-wax embedded sections were cut at 5 ~tm for the histological and immunohistochemical studies. Sections stained with haematoxylin and eosin were used for the World Health Organization system of classification (Hampe and Misdorp 1974). According to the histological grade of the malignancy and its metastatic capacity, the samples were classified as hyperplastic lesions, benign tumours, non-metastatic malignant tumours - that is, without lymphatic infiltration or, if recurrent, without metastasis - or metastatic malignant tumours - that is, primary tumours with lymphatic infiltration or metastatic turnouts. Immunohistochemical studies
Integrin subunits were detected in the 5 ~tm sections by use of the peroxidase-anti-peroxidase technique (PAP) with the following primary antibodies in phosphate buffered saline (PBS): rabbit anti-human [31 integrin subunit (cytoplasmic domain) (Chemicon International AB1938) at a
working dilution of 1:75; rabbit anti-human ~t5 integrin subunit (cytoplasmic domain) (Chemicon International, AB1928), at a working dilution of 1:75; and rabbit anti-human fibronectin (Chemicon International, AB1940) at a working dilution of 1:100. The second antibody was porcine antirabbit immunoglobulin (Dakopatts Z196) at a working dilution of 1:100. The immunoreaction was revealed by treating the sections with rabbit peroxidase-antiperoxidase complex (Dakopatts Zl13) at a working dilution of 1:64, followed by diaminobenzidine (DAB, Sigma D-5905), 20 mg m1-1 in PBS, and then with 100 ~tl of 30 per cent hydrogen peroxide, after which they were counterstained with haematoxylin. For the detection of fibronectin the sections were digested with 0.1 per cent trypsin (Sigma T-8128) at 37°C, pH 7.8 before the immunohistochemical staining. All the staining procedures included negative control sections incubated in PBS without the primary antibody. The immunohistochemical staining of fibronectin and integrin subunits was assessed semiquantitatively as described by D'Ardenne et al (1991). According to the intensity and extent of the staining the reaction was graded as: (-) absent, (+) staining of a few tumour cells for integrins or very weak staining of stroma for fibronectin, (++) staining of a moderate number of tumour cells or focal staining of the stroma, or moderate diffuse staining of both, or (+++) intense staining of all the tumour cells and of the stroma. Results The results are summarised in Tables 1 and 2.
TABLE 1: Fibronectin and c~5~1 immunostaining in normal canine mammary gland, dysplasias and benign tumours Fibronectin
Cases
Histological type
131
c~5
Stroma
E/MYO
+++
++
+
-/-
Dysplasias
1 2 3 4
Lobular hyperplasia Lobular hyperplasia Hyperplasia + epitheliosis Sclerosing adenosis
+ ++ ++ +++
+ + + +
+ + + +
-/+ -/+ -/-/-
Benign tumours
5 6 7 8 9
Complex adenoma Complex adenoma Benign mixed tumour Benign mixed tumour Benign mixed tumour
++ +++ +++ +++ ++
++ +++ +++ +++ ++
+++ +++ +++ +++ +++
++/+++ +++/++ +++/+++ +++/+++ ++/4-+
Five normal glands
- Absent, + Slight, ++ Moderate, +++ Strong, E Epithelial/MYO Myoepithelial cells
360
L. Pe~a, A. Nieto, M. D. Perez Alenza, A. Rodriguez, M. A. Sanchez, M. Castaho
TABLE 2: Fibronectin and c~5~1 immunostaining in malignant canine mammary tumours Cases Non-metastatic malignant t u m o u r s
Histological t y p e
10
Simple tubular adenocarcinoma C o m p l e x tubular adenocarcinoma Papillary adenocarcinoma Papillary adenocarcinoma S q u a m o u s cell carcinoma Malignant m i x e d tumour
11 12 13 14 15 Metastatic malignant t u m o u r s
16 17 18 19 20 21 22
Papillary adenocarcinoma Malignant m i x e d tumour S q u a m o u s cell carcinoma Solid c a r c i n o m a Solid c a r c i n o m a Solid c a r c i n o m a Spindle cell c a r c i n o m a
L y m p h a t i c emboli 131 or5 FN
P u l m o n a r y metastasis 131 ct5 FN
++
++
++
+
None
None
None
+++
++
++
++
+
+
++
+
+++ + ++
++ + + + ++
++ ++ + + None
+ + + + None
++ + + + None
++ None None + +
+ None None + +
None None +
Fibronectin S t r o m a E/MYO
[31
c~5
+++
++
++
++/++
+++
++
+++
++/+++
++
++
+++
++
+++
++
+++
++/+++
+
-
-
-
+++
+++
+++
++
+++
+
+++
+
+
++ + + ++ +++
+ + + ++ +++
- Absent, + Slight, ++ M o d e r a t e , + + + Strong, FN Fibronectin, E Epithelial/MYO Myoepithelial cells (if both cell types are present)
Normal mammary gland
Staining for 131 was present in the epithelial cells of ducts, ductules and alveoli. In the myoepithelial and basal cells the staining was more intense along a band in contact with the basement membrane. In the stroma, 131 was present in fibroblasts, endothelial cells and smooth muscle cells. The intensity of the reaction of 131 in the smooth muscle cells of arterioles served as an internal control for the immunostaining in all the samples. Staining for the ct5 integrin subunit was moderate in the epithelial cells but more intense in the myoepithelial cells. In the stroma the fibroblasts were positive and both endothelial cells mad smooth muscle cells were slightly positive. Fibronectin was present in very small amounts in the basement membranes but was more abundant and distributed diffusely in the connective tissue stroma. It was not detected in epithelial or myoepithelial cells but was abundant in the cytoplasm of fibroblasts in the stroma. Dysplastic mammary gland
In glands with lobular hyperplasia, there was a decreased but uniform expression of 131 and ct5 in
all the samples (cases 1 to 3). The myoepithelial or basal cells contained more 131 and ct5 than did the epithelial cells. In one gland with lobular hyperplasia (case 3), areas of regular epithelial proliferation ('epitheliosis') within pre-existing structures did not differ in their integrin expression from other hyperplastic areas. In contrast with normal mammary glands, two glands with lobular hyperplasia (cases 1 and 2) contained fibronectin not only in the basement membranes or stroma but also in the cytoplasm of myoepithelial cells. In one gland with simple tubular sclerosing adenosis (case 4), there was weak staining for the a5 subunit, strong staining for the 131 subunit, and no increase in fibronectin in the desmoplastic stroma. Mammary tumours Benign mammary tumours. In the five benign mammary tumours, 131 immunostaining was moderate (++) to intense (+++), as in normal mammary tissue. The immunostaining for c~5 was similar to that in normal mammary gland in two of the tumours (cases 5 and 9) but increased in a complex adenoma (case 6) and in two benign mixed tumours (cases 7 and 8). In four of the five
Fibronectin and a5fll integrin in canine mammary tumours
benign neoplasms (two complex adenomas and two benign mixed tumours, cases 5 to 8) the myoepithelial ceils in basal locations and in areas of proliferation were intensely stained for [31 and ct5. In the benign mixed tumours the chondrocytes were also stained for [31 and Gt5 integrins. Uneven immunostaining for fibronectin was observed in all the benign tumours and was stronger than in the normal or dysplastic mammary glands. There was diffuse staining for fibronectin in the stroma and there were fibronectinpositive membrane fragments in focal areas in which proliferation of myoepithelial cells was evident. Fibronectin was not increased in desmoplastic areas in one benign tumour (case 9). There was staining for fibronectin in the cartilaginous matrix of only one of the three benign mixed tumours (case 8) and in all three tnmours the chondrocytes were negative. In two tumours (cases 5 and 9) some of the epithelial cells were strongly stained for fibronectin but others were negative. In cases 6 to 8 fibronectin was present in the cytoplasm of all the epithelial cells. In all the benign tumours the myoepithelial cells were strongly positive for fibronectin. Non-metastatic malignant tumours. In four of the
six tumours in this group (cases 10 to 13) the amount of integrin did not differ greatly from that in normal mammary gland. In one squamous cell carcinoma (case 14), the staining for 131 and ct5 was weaker and in one mixed malignant tumour (case 15), there was strong staining for a5 in the myoepithelial cells. Chondrocytes in newly formed cartilaginous tissue had small amounts of both 131 and a5 subunits. In two of the six non-metastatic malignant tumours (cases 10 and 11), the staining for integrin subunits in proliferated myoepithelium or in cells in typical basal locations was more evident than in epithelial cells. In three tumours (cases 10, 12 and 13) some cells with obvious characteristics of malignancy (lack of differentiation, large nuclei and multiple nuclei), there was very intense staining for a5 and in case 12 these cells also stained intensely for [31. In one non-metastatic malignant tnmour (case 10) there were non-adherent epithelial cells in the connective tissue without notable features of malignancy but with intense staining for a5. Fibronectin was increased in the stroma and was present in the epithelial and myoepithelial
361
cells of five of the six non-metastatic malignant tumours (cases 10 to 13 and 15). Fibronectin formed membrane fragments in two complex malignant tumours (cases 11 and 12). In two adenocarcinomas (cases 11 and 13), there were thickened fibronectin-positive basement membranes and the most positive cells were the myoepithelial cells and non-cohesive epithelial cells. In three tumours (cases 11 to 13) there was strong staining for fibronectin in the cytoplasmic membrane of the epithelial cells. In one squamous cell carcinoma (case 14) with very weak staining for [31 and a5, fibronectin was almost absent. Metastatic malignant tumours. Four of the seven
tumours (cases 17 to 20) had weak staining for [31 and ct5. In case 16 only ct5 was weakly staining and in case 21 only the [31 subunit was weakly staining. In one spindle cell carcinoma (case 22) both of these subunits were strongly stained. The staining for integrin was more intense in the basal cells of two metastatic malignant neoplasms (cases 18 and 19). In spite of the low levels of or5 in the metastatic tumours, there were three tumours in which obviously malignant cells, especially multinucleated cells and very immature cells, were strongly stained for (~5. Strongly positive cells were also found in some malignant non-metastatic tumours. ct5-positive non-cohesive epithelial ceils were also found in the stroma of all the metastatic tumours and some of them were also positive for [31. The staining for fibronectin in metastatic malignant tumours was very uneven; it was increased in the stroma in most cases (cases 16 to 18 and 22) and all the tumours had neoplastic fibronectin-positive cells. Immunostaining for fibronectin was observed in the epithelial cells of three solid carcinomas (cases 19 to 21) in which the staining for fibronectin in the stroma was weak and there was a decrease in 131 and ~t5. In the spindle cell carcinoma (case 22) in which there was an increase in [31 and ot5, there was moderate staining for flbronectin in the cytoplasm. Lymphatic emboli. There were neoplastic emboli
in the lymphatics of six of the seven metastatic tumours (cases 16 to 21). In all cases, the intensity of staining for [31 and or5 in the metastatic cells was very uneven. In one solid carcinoma (case 20) the cells strongly positive for ct5 were migrating
362
L. Peha, A. Nieto, M. D. Perez Alenza, A. Rodriguez, M. A. Sanchez, M. Casta~o
through the walls of the lymphatic capillaries. In two tumours (cases 17 and 20) neoplastic cells inside the lymphatics and in contact with endothelial cells were arranged in palisades and were very strongly stained for [31 and a5. The presence of fibronectin in metastatic cells within emboli was very uneven. In five cases (cases 16, 17 and 19 to 21) most of the metastatic cells were weakly stained for fibronectin and very few were strongly positive. In the squamous cell carcinoma (case 18), the fibronectin content of the metastatic cells was also distributed unevenly but it was stained intensely in most cells, especially in the cytoplasmic membrane. metastases. In all four pulmonary metastases of mammary carcinomas (cases 17, 18, 21 and 22) the immunostaining for both integrin subunits and for fibronectin was very weak or negative, except in migrating cells in case 17, in which there was strong staining for the or5 subunit. Pulmonary
Discussion Fibronectin and its receptors regulate several cell functions, including cell adhesion, polarity, migration, differentiation (Ruoslahti 1988), morphology, the assembly of extracellular matrix, the formation of bundles of microfilaments (Akiyama et al 1989) and the proliferation of cells (Giancotti and Ruoslaht 1990). The aim of this work was to evaluate the possible changes in integrin ct5131 and its ligand fibronectin in relation to the metastatic capacity of canine mammary neoplasms. In normal canine mammary tissue the staining for 131 and (z5 was similar to that described in human mammary tissue (Zutter et al 1990). The staining for 131 in the epithelial and myoepithelial cells of human mammary tissue (Zutter et al 1990) has been interpreted as indicating a greater expression of integrins in the myoepithelial cells (Pignatelli et al 1991). Smooth muscle cells and fibroblasts in the dermis are strongly positive for 131 (De Strooper et al 1989). D'Ardenne et al (1991) found weak staining for the cz5 subunit in human mammary epithelial cells but there was more intense staining for the (z5 subunit in the canine myoepithelial cells. The presence of fibronectin in the stroma and weak staining for it in basement membranes of normal human breast tissue was reported by Natali et al (1984). The absence of fibronectin from the
cytoplasm of epithelial and myoepithelial cells in the present study is also in agreement with these authors' findings. An interesting difference between the hyperplastic and neoplastic lesions was the uniformity of the immunostaining for [31 and ~t5 in the former and its unevenness in the latter. Unevenness of integrin expression has been reported for oil and or6 subunits in breast cancer (D'Ardenne et al 1991) and in human malignant melanoma (Albelda et al 1990). Zutter et al (1990) found high levels of expression of the collagen receptor ct2131, which is associated with the orderly, regulated proliferation of epithelial cells. Although hyperplasia of the canine mammary gland are regulated proliferations, a decrease in the integrin [31 subunit was observed in the present work. In the benign mammary tumours, complex adenomas and mixed benign neoplasms, there was strong staining for both 131 and or5 in myoepithelial cells and the staining for fibronectin was also increased. D'Ardenne et al (1991) found a strong expression of integrin subunits in the myoepithelial or basal cells and even different subunits expressed in the two cell types. An increase in fibronectin was also found in the stroma of some malignant tumours without myoepitheiial proliferation, and in 17 of the 22 samples of dysplasfic glands and tumours, the epithelial cells stained positively for this glycoprotein. This finding suggests that both the epithelial and the myoepithelial cells in the canine mammary gland secrete fibronectin when they have undergone pre-neoplastic or neoplastic changes. Cytoplasmic staining for fibronecfin has been reported in malignant human mammary tumours (Natali et al 1984, Chfistensen et al 1985), although other authors have reported that the expression of fibronectin is diminished in breast tumours (Hynes and Yamada 1982). In four of the six non-metastatic malignant tumours the expression of integfin was uneven but normal, whereas the expression of at least one integrin subunit was decreased in six of the seven metastatic tumours and both were decreased in four of the seven. This decrease most often involved the c~5 subunit. Pignatelli et al (1991) found a similar reduction in the expression of the 131 subunit in human mammary tumours with a high grade of malignancy. Like the present findings in metastatic tumours, Zutter et al (1990) found that the expression of a5131 was decreased
Fibronectin and a5flI integrin in canine mammary tumours
363
in poorly-differentiated human mammary carcino- have been found in induced fibrocytic tumours mas but only slightly decreased in the well-differ- with a well-organised fibronectin matrix. This entiated lesions. Other authors (D'Ardenne et al suggests that the synthesis of fibronectin and its 1991) did not find visible differences in the cc5 receptor expression are comodulated (Varani and immunostaining of neoplastic mammary epithe- Chakrabarty 1991). In the present work, no relationship was found between the content of ct5131 lium, the staining being generalised and weak. Decreases or increases in the expression of and the expression of fibronectin, because in 14 of other integrin subunits have been reported in 18 of the tumours the fibronectin in the stroma human breast tumours but the results are contra- was increased while the expression of both c~5 and dictory (Horan et al 1985, Zutter et al 1990, [31 was normal or decreased. Fibronectin was almost absent from only two tumours with very D'Ardenne et al 1991). The finding of cells having marked criteria of low levels of ~t5 and 1~1. It is known that the malignancy which also expressed high levels of deposition of matrix is disturbed in malignant cells or5 in six of 13 malignant tumours which other- (Ruoslahti 1988) and most investigators have wise stained weakly for this subunit appears to found stromal fibronectin to be increased in maligcontradict the evidence of a relationship between nant tumours (D'Ardenne and Barnard 1989). The the metastatic phenotype and the loss of ct5, that is intensity of staining of fibronectin in breast that in human pancreatic adenocarcinoma the tumours has been found to be correlated with the expression of ct5131 is inversely correlated with a degree of anaplasia and with the independent growth of cells (Christensen et al 1985). In the malignant phenotype (Rosendhal et al 1992). Other authors have reported a greater expres- canine tumours there was no correlation between sion of ct5[~l in immature cells than in those with histological malignancy and increased amounts of a metastatic phenotype; Federico et al (1992) fibronectin, although there was a strong expresreported a possible relationship between the loss sion of fibronectin in non-cohesive epithelial cells. The expression of integrin in cells passing into of the fibronectin receptor and the acquisition of the mature phenotype in cells of the myeloid line- the vascular system is of special interest. It is age and it has been demonstrated that the terminal limited to malignant cells which interact with the differentiation of keratinocytes involves a loss of vascular endothelium and/or its basement memadhesiveness to fibronectin (and a loss of Gt5131). brane (Kramer et al 1991). In the present work Cardarelli et al (1988) suggested that in thymo- cells penetrating lymphatic capillaries stained cytes the loss of fibronectin receptors is related to positively for the fibronectin receptor but the the functional maturation or programmed death of expression of integrin was lower in cells in metastatic emboli, except in cells migrating from the cells. The exact role of the cc5131 fibronectin receptor lymphatics. It is known that changes can occur in in cell migration is not clear. It has been demon- tumour cells during the metastatic process and strated that an antibody against the chicken phenotypic heterogeneity has been demonstrated fibronectin receptor complex may inhibit the among primary and metastatic lesions, but little is migration of cells (Ruoslahti 1988) but others known about the changes in the expression of intehave reported that the a5131 receptor retards the grin in this process (Kramer et al 1991). migration of cultured fibroblasts (Akiyama et al The reduction in the ot5131 fibronectin receptor 1989). Furthermore, transformed hamster ovary in metastatic canine mammary tumours may be cells with an increased expression of ct5131 migrat- necessary to obtain independence of anchorage ed less than control cells, suggesting that ct5131 has (Giancotti and Ruoslahti 1990). The authors cona role in the pericelhilar assembly of fibronectin sider that the major expression of or5 found in (Giancotti and Ruoslahti 1990). In the present some independent cells and in cells with a high study, malignant cells with high levels of ct5 and degree of malignancy could be related to their intense staining for fibronectin in the cytoplasm immaturity or to an intense secretion of were observed particularly in one metastatic fibronectin, rather than to this receptor's role in tumour, which is consistent with this role of the migration. In the canine mammary tumours a relafibronectin receptor rather than with a migration tionship between the deposition of fibronectin and process. the expression of cc5131 was found only in benign Increased amounts of the fibronectin receptor tumours in which there was a strong expression of
364
L. PeKa, A. Nieto, M. D. Perez Alenza, A. Rodriguez, M. A. Sanchez, M. Castafio
the c~5 subunit and cytoplasmic fibronectin in myoepithelial cells. Studies of possible changes in other integrin subunits in mammary tumours, such as ot4131, may bring further understanding of the results of this study of fibronectin and its receptor ot5131. Acknowledgements The authors thank Dr Bruce Belshaw and Miss Ana Rodriguez for a careful review of the manuscript. References AKIYAMA, S. K., YAMADA, S. S., CHEN, W. T. & YAMADA, K. M. (1989) Analysis of fibronectin receptor function with monoclonal antibodies: roles in cell adhesion, migration, matrix assembly and cytoskeletal organization. Journal of Cell Biology 11}9, 863-874 ALBELDA, S. M., METTE, S. A., ELDER, D. E., STEWART, R., DAMJANOVICH, L., HERLY, M. & BUCK, C. A. (1990) Integrin distribution in malignant melanoma: association of the 133 subunit with tumor progression. Cancer Research 50, 6757-6764 CARDARELLI, P. M., CRISPE, I. N. & PIERSCHBACHER, M. D. (1988) Preferential expression of fibronectin receptors on immature thymocytes. Journal of CeUBiology 11}6,2183-2190 CASTRONOVO, V., COLIN, C., CLAYSMITH, A. P., CHEN, P. H. S., LIFRANGE, E., LAMBOTrE, R., KRUTZSCH, H., LIOTTA, L. A. & SOBEL, M. E. (1990) Immunodetection of the metastasis-associated laminin receptor in human breast cancer cells obtained by fineneedle aspiration biopsy. American Journal of Pathology 137, 1373-1381 CHRISTENSEN, L., NIELSEN, M., HOLUND, B. & CLEMMENSEN, I. (1985) In vivo demonstration of cytoplasmic fibmnectin in human breast carcinomas. VirchowArchivA (PathologyAnatomy) 407, 337-346 D'ARDENNE, A. J. & BARNARD, N. J. (1989) Paucity of fibroneefin in invasive lobular carcinoma of breast. Journal of Pathology 157, 219-224 D'ARDENNE, A. 1., RICHMAN, P. I., HORTON, M. A., MCAULAY, A. E. & JORDAN, S. (1991) Co-ordinate expression of the alpha-6 integrin laminin receptor subunit and laminin in breast cancer. Journal of Pathology 165, 213-220 DE STROOPER, B., VAN DER SCHUEREN, B., JASPERS, M., SAISON, M., SPAEPEN, M., VAN LEUVEN, F., VAN DEN BERGHE, H. & CASSIMAN, J.-J. (1989) Distribution of the I~1 subgroup of the integrins in human cells and tissues. Journal of Histoehemistry and Cytochemistry 37, 299-307 FEDERICO, M. H. H., YAMAMOTO, M., MARIA, D. A., KATAYAMA, M. L. H., SONOHARA, S., ROELA, M. A., KOIKE, M. A. & BRENTANI, M. M. (1992) Fibronectin receptor expression in human leukemic cells. Clinical and Experimental Metastasis 10, Supplement
1,40
GIANCOTI~, F. G. & RUOSLAHTI, E. (1990) Elevated levels of the e~5~l fibronectin receptor suppress the transformed phenotype of Chinese hamster ovary cells. Cell 60, 849-859 HAMPE, J. F. & MISDORP, W. (1974) Tumors and dysplasias of the mammary gland. Bulletin of the World Health Organization 50, 111-133 HEMLER, M. E., HUANG, C. & SCHWARTZ, L. (1987) The VLA protein family. Journal of Biological Chemistry 262, 3300-3309 HORAN, H. P., THOR, A , SCHLOM, J., RAO, C. N. & LIOTTA, L. A. (1985) Expression of laminin receptor in normal and carcinomatous tissues as defined by a monoclonal antibody. Cancer Research 45, 2713-2719 HUMPHRIES, M. J., OLDEN, K. & YAMADA, K. M. (1986) A synthetic peptide from fibronectin inhibits experimental metastasis of murine melanoma cells. Science 233, 467-470 HYNES, R. O. (1987) Integrins: a family of cell surface receptors. Cell 48, 549-554 HYNES, R. O. & YAMADA, K. M. (1982) Fibronectins: multifunctional modular glycoproteins. Journal of Cell Biology 95, 369-377 KRAMER, R. H., ENENSTEIN, 1., RAMOS, D. M., VU, M. P. & CHENG, J. F. (1991) The role of integrin receptors in tumor cell adhesion to the microvasculature. In Microcirculation in Cancer Metastasis. Boca Raton, CRC Press. pp 145-160 LIOTTA, L A. (1986) Tumor invasion and metastasis-role of the extracellular matrix. Cancer Research 46, 1-7 NATALI, P. G., GIACOMINI, P., BIGOTTI, G , NICOTRA, M. R., BELLOCI, M. & DE MARTINO, C. (1984) Heterogeneous disWibution of actin, myosin, fibronectin and basement membrane antigens in primary and metastatic human breast cancer. Virchow Archiv A (Pathology Anatomy) 405, 69-83 PIGNATELLI, M., HANBY, A. M. & STAMP, G. W. H. (1991) Low expression of 131, ~x2 and {x3 subunits of VLA integrins in malignant mammary tumours. Journal of Pathology 165, 25-32 ROSENDHAL, A., WEINEL, R. J., NEUMANN, K., ERB, D. & ROTHMUND, M. R. (1992) VLA-receptors: expression and function in human pancreatic adenoearcinoma and tumour cell lines. Clinical and Experimental Metastasis 10, Supplement 1, 37 RUOSLAHTI, E. (1988) Fibronectin and its receptors. Annual Review of Biochemistry 57, 375-413 RUOSLAHTI, E. & PIERSCHBACHER, M. D. (1987) New perspectives in cell adhesion: RGD and integrins. Science 238, 491-497 TERRANOVA, V. P., LIOTTA, L. A., RUSSO, R. & MARTIN, G. R. (1982) Role of laminin in the attachment and metastasis of routine tumor cell. Cancer Research 42, 2265-2269 VARANI, J. & CHAKRABARTY, S. (1991) Changes in extracellular matrix during transformation and differentiation. In Microcirculation in Cancer Metastasis. Boca Raton, CRC Press. pp 1-22 WEWER, V. M., TARABOLETTI, G., SOBEL, M. E., ALBRECHTSEN, R. & LIOTTA, L. A. (1987) Role of laminin receptor in tumor cell migration. Cancer Research 47, 5691-5698 ZUTTER, M. M., MAZOUJIAN, G. & SANTORO, S. A. (1990) Decreased expression of integrin adhesive protein receptors in adenocarcinoma of the breast. American Journal of Pathology 137, 863-870
Received December 8, 1993 Accepted June 6, 1994