Loss of basement membrane deposits and development of invasive potential by virally-transformed rat mammary cells are independent of collagenase production

The myoepithelial-type cell line, Rama 712, derived from a normal rat mammary gland, deposits an extracellular matrix containing type-IV collagen and other basement membrane proteins round its cellular periphery. After transformation with a temperature-sensitive mutant of Rous sarcoma virus (tsRSV) the cells fail to deposit an extracellular matrix at the permissive temperature (35 degrees C), but retain the capacity to do so at the non-permissive temperature (41 degrees C). The synthesis of type-IV collagen is not affected by the temperature shift. Rama 712 cells fail to form tumours in syngeneic rats. However, Rama 712-tsRSV cells form tumours that are locally invasive but fail to metastasize. In histological sections, the tumour cells stain with an antibody to type-IV collagen, but do not deposit any extracellular type-IV collagen. Cells isolated from the tumours (Rama 712T) remain temperature-sensitive for the extracellular deposition of type-IV collagen when grown in vitro. Rama 712, Rama 712-tsRSV and Rama 712T fail to produce any detectable type-I or type-IV collagenase at either 35 degrees C or 41 degrees C. These results show that in this system extracellular deposits of basement membrane proteins are lost from invasive tumours produced by myoepithelial-type cells by mechanisms other than those due to the production of collagenolytic enzymes.     

Cellular behavior of metastatic B16 melanoma in experimental blood-borne implantation and cerebral invasion. An electron microscopic study

Using B16 melanoma variant subline B16-B14b, which was selected in vivo for enhanced brain surface colonization abilities in C57BL/6 mice, we have examined ultrastructurally the cellular behavior of B16 melanoma cells during the process of cerebral invasion. Within 14 days after the injection of viable tumor cells (10(5) cells/0.2 ml) into the common carotid artery, gross nodular tumors were identifiable in the cerebral cortex. Tumor cells in the boundary area of the cerebral cortex and leptomeninges breached the limiting glial membrane with numerous small cytoplasmic protrusions and invaded into the brain parenchyma by direct extension or along blood vessels. Direct invasion into the brain parenchyma was characterized by two ultrastructural features of the tumor cells: the production of many elongated cytoplasmic protrusions at zones of invasion which separated, fragmented, and eventually engulfed nervous tissue; and the formation of tumor cell attachments to the dendrites of nerve cells via tight junction-like structures. Another mechanism of brain parenchyma infiltration by the melanoma cells was invasion and migration along blood vessel walls of the cerebral capillaries. In this case B16 melanoma cells were found attached directly to the basolateral side of the blood vessel basal lamina. The tumor cells appeared to achieve their basal lamina location by pushing aside the perivascular astrocytes.     

Müllerian adenosarcoma of the uterus: an ultrastructural study of four cases.

Four cases of uterine müllerian adenosarcoma, a distinctive form of mixed müllerian tumor, were studied by light and electron microscopy. All tumors showed the characteristic histologic pattern of benign neoplastic glands within sarcomatous stroma. Ultrastructurally, both mesenchymal and epithelial cells were seen. The mesenchymal cells showed some features of endometrial stromal cells, including the presence of intracytoplasmic collagen fibers. The epithelial cells formed glands, which resembled benign endometrial glands and were separated from the stroma by a well-defined basal lamina. No transitional cells between the epithelial and mesenchymal cells were seen. The ultrastructural features of these tumors suggest that the sarcomatous portion is of endometrial stromal origin. The glandular portion may arise, along with the stroma, from multipotential stem cells, or the glands may be non-neoplastic entrapped endometrial glands stimulated by the stroma and thus appearing to form an integral part of the tumor.     

Inappropriate production of collagen and prolyl hydroxylase by human breast cancer cells in vivo.

Thirty-two scirrhous cancers of breast have been examined to determine the origin of the collagen stroma in these tumours. Employing two immunohistochemical techniques it has been shown that the malignant epithelial cells in 30 of these tumours contain not only collagen but also prolyl hydroxylase, a key enzyme in collagen biosynthesis. Neither this enzyme nor collagen was detectable in the spindle cells in the stroma of these tumours. Neither the epithelium in normal breast, that in fibrocystic disease and in fibroadenomata, nor the malignant epithelium in two medullary cancers of breast contained either collagen or prolyl hydroxylase. These results strongly suggest that the malignant epithelium of scirrhous breast cancers produces its own collagen stroma and that the scirrhous reaction in these tumours is not a host response to tumour invasion. The production of collagen and prolyl hydroxylase by breast cancer cells (of the scirrhous type) therefore represents another example of inappropriate protein production by a human tumour.     

Stromal accumulation of chondroitin sulphate in mammary tumours of dogs.

To contribute to the investigation of the composition of the extracellular matrix in epithelial tumours, mammary gland tissues of dogs (including tumours, hyperplasias and normal tissue as well as metastatic lesions in lymph nodes and lung) were studied histochemically and immunohistochemically for distribution of sulphated glycosaminoglycans (s-GAGs). The formaline-fixed tissue was stained by alcian blue at pH 5.8, using the 'critical electrolyte concentration' to study the degree of sulphation of s-GAGs. s-GAGs were characterized by degradation with enzymes and nitrous acid and by immunohistochemistry with two anti-chondroitin sulphate monoclonal antibodies. The light microscopic investigation of s-GAG deposits revealed a limited number of patterns of their distribution. The main s-GAGs found in the mammary gland tumours of dogs and in metastatic lesions were chondroitin sulphate (CS) and heparin/heparan sulphate (HEP/HS). CS accumulated in diffuse structures between epithelial cells as well as around clusters of tumour cells. The latter pattern, possibly representing a mesenchymal reaction to the tumour, was present in 74% of the tumours, and in 67% of these, highly sulphated CS was present. A diffuse accumulation of CS was present almost exclusively in complex and mixed tumours; because of the expression of the 3B3 epitope for CS in immature cartilage the spindle cells of complex tumours are argued to be the precursors of the cartilage in mixed tumours. HEP/HS was stored mainly in mast cells that were found in increased numbers in hyperplasias and tumours. By pretreatment of microscopic slides with chondroitinase AC or ABC immunostaining of fibronectin could be made possible in areas in which CS was abundantly present, suggesting that CS may mask fibronectin epitopes. It is concluded that CS with different degrees of sulphation is the most important s-GAG in the extracellular matrix of mammary tumours of dogs. CS and other s-GAGs accumulate at different sites and may have a different pathogenetic significance.     

Alterations of the basement membrane and connective tissue antigens in human metastatic lymph nodes

Antigens of the basement membrane (type-IV collagen and laminin) and the connective tissue (type-Ill collagen and fibronectin) were studied by immunofluorescence in 16 lymph nodes draining colorectal carcinomas and 6 lymph nodes draining breast carcinomas. A comparison was also made between 7 primary colorectal carcinomas and 9 lymph nodes draining these tumors. Anti-type-IV collagen and anti-laminin rarely stained the basement membrane of metastatic tumors. In contrast, we detected type-IV collagen in the peritumoral stroma, although similar images were rarely seen in primary tumors. When tumoral cells were in the vicinity of lymphoid cells, they were occasionaly separated by a barrier stained by the four antisera, or only by antifibronectin and anti-type-lll collagen. In other cases no barrier was observed between both types of cells which were in close contact. On the whole the above alterations were more marked in the lymph nodes draining breast carcinomas, in comparison to those draining colorectal carcinomas. Tumor cells were never stained by antitype-IV collagen or antilaminin serum. Some cells found either in the lymphoid or in the tumor area of metastatic lymph nodes were stained not only by these antisera, but also by a monoclonal antibody against Willebrand Factor, which is a marker of endothelial cells. Thus the labelled cells were characterized as being derived from the capillary wall.     

Distribution of laminin and collagen type-IV in benign and malignant lesions of melanocytic origin

Using antisera for two specific basement membrane (b.m.) antigens such as laminin and collagen type-IV, together with electron microscopy, we have shown that fully antigenic b.m. and morphologically typical basal lamina (b.l.) are associated with normal and transformed cells of the melanocyte lineage in different ways. Thus, while b.m. and b.l. surround individual choroidal melanocytes, intradermal nevus cells and cells of blue nevi are not detectable at the periphery of resting and proliferating epidermal melanocytes. They have a low degree of expression with a heterogeneous pattern of distribution in primary and metastatic melanoma. This heterogeneity is present within single metastases and among autologous metastases. These findings indicate that the presence of b.m. can be an additional marker for cells of the melanocyte lineage and should be considered when applying serological means for the detection and control of neoplasms of melanocytic origin.     

Immunohistochemical detection of the expression of the cell adhesion molecules E-cadherin, desmoglein-2, beta4-integrin, ICAM-1 and HCAM (CD44s) in Warthin's tumour of the parotid gland

The study of the expression of cell adhesion molecules (CAMs), E-cadherin, desmoglein-2, beta4-integrin, HCAM (CD44s) and ICAM-1 in Warthin's tumours. Twenty formalin--fixed, paraffin--embedded parotid Warthin's tumours were studied using an Envision/HRP immunohistochemical technique. Beta4-integrin was strongly expressed in all cell-basement membrane and intercellular contacts of the epithelium, E-cadherin and desmoglein-2 in cell-cell contacts, but not in basal cell-basement membrane connections and on columnar cells' luminal surfaces, HCAM (CD44s) in intercellular contacts of both luminal (mainly), basal cells and also in the periphery of monocytic-lymphocytic stroma, and ICAM-1 was weak to moderate expressed in both luminal and basal epithelial cells and strongly in the germinal lymphocytic centres. CAM expression suggests a bilayered excretory ductal structure of the neoplastic epithelium in Warthin's tumour, as a result of hyperplastic process of the glandular epithelium that interacts with the excessive lymphoid tissue of the stroma.     

Ultrastructural tumour differentiation and organ specificity in high and low metastatic lines from a mouse lung carcinoma

A tissue culture cell line CMT64 was established from a spontaneous alveolar lung carcinoma of a C57BL female mouse (Franks et al., 1976). Subcutaneous inoculation of these cells produced a local tumour and a small number of lung metastases. Four sublines CMT167, 170, 175 and 181 with increased metastatic ability were selected, as described in the accompanying paper (Layton & Franks, 1984). The tissue culture cells and the tumours produced by all the lines are well differentiated and produce laminated surfactant-like bodies as well as basal lamina, even in metastases. No ultrastructural differences were found that might correlate with metastatic behaviour in vivo. Metastases, after subcutaneous inoculation and tumour colonies after intravenous inoculation of all cell lines are only found in the lung, but after inoculation of cells into the arterial system via the left ventricle of the heart, extravascular tumour colonies were found in many organs.     

Invasion of lung tissue by bronchogenic squamous-cell carcinomas: interaction of tumor cells and lung parenchyma in the tumor periphery

We collected multiple samples from 43 human bronchogenic squamous-cell carcinomas and studied the ultrastructure of the tumor-cell-lung parenchyma interaction in the tumor periphery. Since, in the periphery of a growing tumor, the surrounding tissue has only recently been reached by the tumor cells, it is the initial stage of interaction which can be observed here. The main findings were: The tumor cells in the tumor periphery always penetrated the lung parenchyma along the epithelial side of the alveolar basal lamina. The non-neoplastic alveolar epithelial cells were either detached from their basal lamina or overgrown by the tumor cells, without being visibly damaged or destroyed. When the alveolar epithelial cells were overgrown by the tumor, they retracted and formed extra- or intracellular lumina much smaller than the original alveoli. The contact between tumor cells and alveolar epithelial cells resulted in the formation of common desmosomes and complete junctional complexes and the common lining of lumina. Although the tumor cells extended small pseudopods through the basal lamina, they virtually never migrated through it to reach the interstitial compartment. These results indicate that the initial invasion of lung parenchyma is characterized by a smooth integration of tumor cells and elements of the preexisting tissue, leading to orderly associations of neoplastic and non-neoplastic cells, rather than by tissue destruction by the tumor cells.     

Detection by in situ hybridization of messenger RNAs of collagen types I and IV in murine mammary cancer

Ten breast carcinomas developing in C3H/Bi female mice were studied by an in situ hybridization technique using cDNA probes encoding alpha-I chains of collagens of types I and IV. The results obtained are compared to histochemical data on antibodies to collagens of types I and IV and ultrastructural observations on these tumors. Immunohistochemistry has revealed the presence of type-IV collagen in basement membranes lining intraductal and well-differentiated cancer nests. Type-I collagen was detected in the stroma surrounding these cells. There was no cellular labelling when these antisera were used. In situ hybridization has shown that type-IV collagen mRNA is detected in non-invasive intraductal and well-differentiated tumor cells and in endothelial cells in the stroma. Good correlation between detection of type-IV collagen lining these cells and evidence of mRNA by in situ hybridization was thus observed. Invasive cancer cells did not express hybridization grains with the type-IV collagen probe. Type-I collagen mRNA was visualized in stromal cells, probably fibroblasts and myofibroblasts as shown by electron microscopy. The most active cells were localized close to non-invasive areas. Our data indicate persistent in-vivo biosynthesis of type-IV collagen by some cancer cells that produce their own limitative environment; they suggest that type-IV collagen is not produced by invasive tumor cells and that stromal cells lining the non-invasive regions have a peculiar behavior.     

Different expression of lysosome-associated membrane protein-1 in human melanomas and benign melanocytic lesions

Lysosome-associated membrane protein-1 is a protein with a significant content of beta1,6-branched N-glycans. It is thought that enhanced expression of lysosome-associated membrane protein-1 in tumour cells may promote invasion by influencing both adhesion to extracellular matrix and perhaps also binding to endothelial cells. The present study was aimed at examining levels of lysosome-associated membrane protein-1 in human melanomas and benign pigmented lesions to evaluate whether this protein might be considered a potential molecular marker of melanoma progression. The expression of lysosome-associated membrane protein-1 was for the first time determined immunohistochemically in formalin-fixed paraffin-embedded specimens comprising 42 primary cutaneous melanomas, 15 lymph node melanoma metastases (11 correlated with primary tumours), three melanoma recurrences (correlated with both primary and metastatic melanomas), 27 nevi and four epithelial tumours (two seborrhoeic keratoses and two basal cell carcinomas). Our results demonstrate that development and progression of melanoma are associated with changes of the lysosome-associated membrane protein-1 level. The expression was strongest in melanoma recurrences and lymph node metastases, weaker in primary cutaneous melanomas and not detectable in melanocytes of pigmented nevi. Nodular melanomas expressed lysosome-associated membrane protein-1 at higher level than superficially spreading melanomas.     

Cathepsin S in tumours, regional lymph nodes and sera of patients with lung cancer: relation to prognosis.

Cysteine proteinase cathepsin S (Cat S) is expressed mainly in lymphatic tissues and has been characterised as a key enzyme in major histocompatibility complex class II (MHC-II) mediated antigen presentation. Cat S has been measured in tissue cytosols of lung parenchyma, lung tumours and lymph nodes and in sera of patients with lung tumours and of healthy controls, by specific enzyme-linked immunosorbent assay (ELISA). A difference in Cat S level was found between tumour and adjacent control tissue cytosols of 60 lung cancer patients (median 4.3 vs. 2.8 ng mg(-1) protein). In lymph nodes obtained from 24 patients of the same group, the level of Cat S was significantly higher than in tumours or lung parenchyma (P< 0.001). Additionally, significantly higher levels were found in non-infiltrated than in infiltrated lymph nodes (median 16.6 vs 7.5 ng mg(-1) protein). Patients with low levels of Cat S in tumours and lung parenchyma exhibited a significantly higher risk of death than those with high levels of Cat S (P = 0.025 - tumours; P = 0.02 - parenchyma). Immunohistochemical analysis (IHA) of lung parenchyma revealed a staining reaction in alveolar type II cells, macrophages and bronchial epithelial cells. In regional lymph node tissue, strong staining of Cat S was found in lymphocytes and histiocytes. Nevertheless, Cat S was detected also in tumour cells, independently of their origin. Our results provide evidence that Cat S may be involved in malignant progression. Its role, however, differs from that of the related Cats B and L and could be associated with the immune response rather than with remodelling of extracellular matrix.     

Hyalinizing clear cell carcinoma of salivary glands: a study of extracellular matrix

Hyalinizing clear cell carcinoma (HCCC) is a recently described low-grade carcinoma of the salivary glands presenting two main histological features: clear neoplastic cells and prominent hyalinized stroma. Our aim was to characterize the composition of the hyalinized stroma in three cases of HCCC. An immuno-histochemical study using laminin and type I, III and IV collagens, fibronectin and tenascin antibodies was performed. In one case, electron microscopy study was also done. Collagen I and fibronectin were always present in tumoral stroma. Collagen III was documented in two cases and tenascin at the invasive front of the neoplasm. Collagen IV and laminin were found around neoplastic cells but not at the interstitial stroma. Foci of basal lamina reduplication and large deposits of long space collagen were observed in EM. This study demonstrates that the hyalinized stroma of HCCC is not formed by the accumulation of basement membrane-type material since it is mostly composed by collagen I and fibronectin.     

The relationship of collagenolytic activity to stage of human colorectal carcinoma

Collagenolytic enzymes produced by tumor cells are believed to play a significant role in the destruction of surrounding normal tissue and, in certain experimental animal systems, the ability of tumor cells to degrade type-IV collagen (basement membrane collagen) correlates positively with those cells' metastatic capacity. We measured collagenolytic activity levels of extracts from freshly excised colorectal carcinoma tissues and of conditioned media from primary organ culture (total of 114 tissues from 53 patients) by using purified radiolabelled type-I (rat tail) and type-IV (mouse Engelbreth-Holm-Swarm [EHS] sarcoma) collagens. Both type-IV and type-I collagenolytic activity levels of extracts from tumor and adjacent mucosa ranged from less than 1 to 80 ng/hr/mg wet tissue, and no significant differences between mucosa and carcinoma tissues were observed. In conditioned media, the type-IV collagenolytic activity was low for normal mucosa and benign tumors and slightly higher for carcinoma than for mucosa. In 5 of 32 primary tumors, collagenolytic activity levels were 2-5 times higher than in the rest of the tumors and mucosal tissues. There were no significant differences in collagenolytic activity levels of conditioned media and tissue extract from colorectal carcinoma of different Dukes' stages. Deep and superficial areas of primary tumors released similar type-IV collagenolytic activity levels, suggesting that there was little intratumoral heterogeneity in the release of this enzyme.     

Primary melanoma tumour regression associated with an immune response to the tumour-associated antigen melan-A/MART-1.

A prediction of the theory of immunologic surveillance is that tumour antigens can be recognised by cell-mediated immunity during early development of the primary tumour by formation of tumour antigen-specific cytotoxic lymphocytes (CTLs) and that such recognition leads to destruction of those tumour cells (tumour regression) with subsequent appearance of tumour antigen-loss variants. However, this has never been shown in nonviral-induced experimental animal models of primary malignancy or in human primary cancer. We examined 2 groups of human melanoma patients where primary tumour regression was observed. Twenty-three patients with multiple (>/=3) primary melanoma showed significant histologic regression of their last tumour (median tumour regression 33%) compared to matched tumours from patients with a single primary melanoma (median 0%) (p = 0.008) or compared to their first primary tumour (median 0%) (p = 0.001). This increased regression is consistent with an "immunisation effect" seen in murine tumour transplantation studies where innoculation with >/=3 asynchronous tumours induces transplantation rejection on subsequent challenge. A significant decrease in MART-1-positive stained tumour area in the last primary tumour from multiple melanoma subjects (median 8%) vs. matched single melanoma patients (median 79%) (p = 0.004) and in the last vs. first tumour (median 76%) in multiple primary subjects was found (p = 0.008). Metastatic tumours from 17 patients whose primary skin melanomas had completely regressed (occult primary melanoma) also showed significant MART-1 tumour-loss variants (median 0% MART-1-positive tumour) compared to matched metastatic tumours from patients with nonregressing primary tumours (median 51%) (p = 0.001). A correlation with the presence of peripheral blood MART-1-specific CTLs (MHC class I-restricted IFN-gamma producing T lymphocytes) and MART-1 tumour antigen-loss variants was found (p = 0.001). Thus, in 2 groups of human melanoma subjects, we provide evidence of tumour regression associated with Melan A/MART-1 tumour antigen-loss variants correlating with formation of specific Melan A/MART-1 CTLs.     

Transforming growth factor-beta1 increases survival of human melanoma through stroma remodeling.

Transforming growth factor (TGF)-beta is growth inhibitory for normal epithelial cells and melanocytes but can stimulate mesenchymal cells. Resistance to its inhibitory effects is characteristic of human melanoma, the growth of which may instead be promoted by TGF-beta, because its production is increased with melanoma progression. Whether TGF-beta has an autocrine function for melanoma cells or is important for paracrine stimulation of the tumor stroma is not known. In this study, TGF-beta1 was expressed in melanoma cells via adenoviral gene transfer, and tumor growth was analyzed in vitro, in human skin grafts, and in mixtures with fibroblasts that were injected s.c. into immunodeficient mice. The TGF-beta1 produced by the melanoma cells activated the fibroblasts to produce matrix within and around the tumor mass, whereas control tumors showed less stroma and more cell death. High expression of collagen, fibronectin, tenascin, and alpha2 integrin was detected in the TGF-beta1-expressing tumors by immunohistochemistry. Number and size of lung metastases were significantly increased. cDNA expression array analysis of TGF-beta1-transduced fibroblasts embedded in type I collagen and of TGF-beta1-transduced melanoma cells demonstrated induction of types XV, XVIII, and VI collagens, tenascin, plasminogen activator inhibitor-I, vascular endothelial growth factor, cysteine-rich fibroblast growth factor receptor-1, and platelet-derived growth factor receptor-beta, which could be linked to promotion of growth and survival in melanoma. These data suggest that remodeling of the neighboring stroma, which provides a supporting scaffolding and a positive feedback stimulation of tumor growth, is an important function of TGF-beta1 in melanoma.     

Adhesion characteristics of human melanoma cell lines of varying metastatic potential

The adhesive behaviour of a series of human melanoma cell lines, of varying metastatic potential, to basement membrane and stromal components was investigated in vitro. Experimental metastatic propensity was assessed from the number of pulmonary nodules formed after i.v. injection of cells into BALB/c nude mice. All cell lines showed similar kinetics of attachment when tested on plastic, type-I collagen films, type-I collagen hydrated gels, fibronectin, laminin type-IV collagen substrates and bovine aortic endothelial monolayers. Fibronectin-coated plastic compared to plastic alone produced increased cell attachment and spreading to the same extent in all the cell lines. The melanoma lines attached preferentially to cryostat sections of lung compared to other organs reflecting the pattern of organ involvement of metastasis in vivo. However, no significant quantitative differences in attachment to lung sections were seen between melanoma variants of differing metastatic capacities. Cells labelled with [125I]iododeoxyuridine to determine their initial organ distribution following i.v. injection showed that tumour-cell arrest was not significantly changed enough to explain the differing metastatic capacities. Thus it appears that adhesive properties of these melanoma cells are not correlated with their capacity to form metastases in vivo.     

Expression of individual lamins in basal cell carcinomas of the skin

In this study we used a unique collection of type specific anti-lamin antibodies to study lamin expression patterns in normal human skin and in skin derived from patients with basal cell carcinomas (BCCs). Lamin expression in serial sections from frozen tissue samples was investigated by single and double indirect immunofluorescence. In normal skin, lamin A was expressed in dermal fibroblasts and in suprabasal epithelial cells but was absent from all basal epithelial cells. Lamin C was expressed in dermal fibroblasts, suprabasal epithelial cells and a majority of basal epithelial cells. However, lamin C was not expressed in quiescent basal epithelial cells. Lamin B1 was expressed in all epithelial cells but was not expressed in dermal fibroblasts. Finally, lamin B2 was expressed in all epithelial cells but was not expressed in dermal fibroblasts. Finally, lamin B2 was expressed in all cell types in normal skin. Lamin expression was also investigated in a collection of 16 BCCs taken from a variety of body sites. Based upon patterns of lamin expression the BCCs were classified into four groups: A-negative (10/16 tumours), C-negative (5/16 tumours), A/C-negative (1/16 tumours) and A/B2-negative (1/16 tumours). Lamin expression was also compared to cell proliferation index by staining serial sections with the proliferation marker Ki67. 9/10 of the lamin A negative tumours were highly proliferative, whereas 4/5 of the lamin C negative tumours were slow growing. Thus as a general rule absence of lamin A was correlated with rapid growth within the tumour, while absence of lamin C was correlated with slow growth within the tumour. Our data supports the hypothesis that lamin A has a negative influence on cell proliferation and its down regulation may be a requisite of tumour progression.