Int. J. Cancer: 51,692-697 (1992) 0 1992 Wiley-Liss, Inc.
Publication of the International Union Against Cancer Publication de I'Union Internationale Contre le Cancer
STUDY OF FIBRONECTIN mRNA IN HUMAN LARYNGEAL AND ECTOCERVICAL CARCINOMAS BY IN SITU HYBRIDIZATION AND IMAGE ANALYSIS Laura MORO', Marina COLOMBI',Maria Pia MOLINARI TO SAT TI^ and Sergio BARLAT11,3 'Division of Biology and Genetics and 2Divisionof Histology, Depaifment of Biomedical Sciences and Biotechnology, University of Brescia, Itab. The expression of fibronectin (FN) mRNA was studied in histological sections of surgical biopsies from human laryngeal and ectocervical invasive carcinomas of different grading stages by in sku hybridization and image analysis. This approach made it possible to identify the cell types synthesizing FN mRNA in the tissue sections and to compare semi-quantitatively the FN mRNA levels expressed in the different specimens. The carcinoma cells synthesized low levels of FN mRNA, comparable to those detected in control epithelia and connective-tissue fibroblasts. Well-differentiated (G I) laryngealand ectocervical carcinomas induced the synthesis of FN mRNA-to levels 7 to 13 times higher than in control connective tissues-in the stromal fibroblasts surrounding the tumors. In carcinoma samples analysed, the amount of FN mRNA detected in the stroma decreased in relation to tumor grading (from GI to G3) and the stromal destruction. FN mRNA was not detectable in the endothelial cells of venules while it was present in large amounts in those surrounding the capillaries present in the stroma. These data indicate that FN, usually observed around carcinomas, is produced by stromal fibroblasts, which are induced to express FN mRNA, presumably in response to diffusible factors produced by the tumor cells, and/or by endothelial cells of the infiltrating capillary vessels. The induction of FN mRNA, inversely proportional to the tumor grading, may be useful in evaluatingthe invasion potential of the tumor.
o 1992 Wiley-Liss, Inc.
Fibronectins (FNs) are high-molecular-weight, adhesive glycoproteins particularly abundant in body fluids, loose connective tissue, granulation tissue, basement membranes and embryonic structures (Stenman and Vaheri, 1978; Ruoslahti et al., 1981; ThiCry et al., 1989). Tissue FN interacts with many components of the extracellular matrices (ECM) (i.e. collagens and glycosaminoglycans) (Yamada, 1983) as well as with cell-surface receptors (Ruoslahti, 1988; Fogerty et a l , 1990) and may therefore provide a scaffold for ECM assembly and organization. FNs also play a role in tissue remodelling and cell migration that occur during embryonic development and wound healing (Grinnell, 1984; ThiCry et al., 1989). Much evidence has been obtained regarding the role of ECMadhesive glycoproteins in tumor growth and invasion (Alitalo and Vaheri, 1982; McCarthy et aL, 1985; van den Hooff, 1988; Vaheri et aZ., 1989; Humphries et al., 1989). FN is greatly reduced, or absent, on the surface of cultured tumor-derived or in vitro-transformed cells (Hynes, 1973; Vaheri and Ruoslahti, 1974). Immunohistochemical analysis of sections from human solid tumors of several types revealed that FN was not expressed by carcinoma and melanoma cells, while it was synthesized by individual cells in some sarcomas (Stenman and Vaheri, 1981). The synthesis of FN by tumor cells was therefore related to their embryonic derivation. The stroma surrounding advanced carcinomas was, in contrast, strongly positive for FN (Stenman and Vaheri, 1981). Previous immunohistochemical studies on laryngeal (Antonelli et aL, 1991) and ectocervical carcinoma sections (Molinari-Tosatti et al., 1992) showed that carcinoma cells lacked intra- or pericellular FN, while fibroblasts surrounding tumors expressed consistent amounts of FN. Stromal FN may be synthesized by fibroblasts or by endothelial cells of blood
vessels; it may also partly stem from serum as an inflammatory exudation. In order to verify the source of the stromal FN we performed in situ hybridization with a radiolabelled FN cDNA, combined with image analysis of sections from control human tissues and from laryngeal and ectocervical invasive carcinomas. Semiquantitative evaluation of FN mRNA expressed by the different cell types present in the sections of tumors at different stages was also made possible by this approach. MATERIAL AND METHODS
Chemicals and reagents Bio-plast was obtained from Bioptica, Milan, Italy; proteinase K, paraformaldehyde, triethanolamine, acetic anhydride and formamide from BDH Italia, Milan; Sephadex G-50, Ficoll and polyvinyl pyrrolidone from Pharmacia L I B , Uppsala, Sweden; dextran sulphate, BSA and glycerol-gelatin from Sigma, St. Louis, MO, and glycine from BioRad, Richmond, CA. Preparation of tissue sections Tissue specimens were obtained from patients whose clinical diagnosis was confirmed by histopathological examination. From each patient bearing laryngeal carcinoma, 2 surgical biopsies were collected, one from areas of the tumor showing no signs of necrosis, the other from apparently normal laryngeal mucosa. Ectocervical carcinoma specimens were collected according to the same criteria, while the normal tissue was obtained from patients who had undergone surgical operations for non-neoplastic diseases. At least 2 different tumors were examined for each grading. Tissue specimens were fixed in 4% paraformaldehyde in PBS, dehydrated in rising ethanol series, cleared in xylene and embedded in bio-plast using routine histological procedures. Sections 5 pm thick were cut from the tissue blocks, spread on microscope slides and incubated overnight at 37°C. Before hybridization, the slides were washed for 2 x 10 min in xylene, 1 x 10 min in 100% ethanol and 1 x 10 min in 70% ethanol. Fibronectin cDNA probe and labellingprocedures The probe used was a HindIII/BamHI fragment of 2.1 kb (FN 1-1) obtained from the pFHl recombinant plasmid which contains the 3'-terminal portion of FN cDNA (Kornblihtt et al., 1984). For in situ hybridization, 500 ng of FN cDNA were labelled by nick translation, according to the supplier's instructions (Gibco-BRL, Gaithersburg, MD) in the presence of 6 pM each of 3H-TTP (103 Ci/mM), 3H-dCTP (53 Ci/mM), 3H-dATP (65 Ci/mM) (Amersham, Aylesbury, UK) and 6 KM unlabelled dGTP. The DNA probe was labelled to a specific activity of 3.5 x lo7 cpm/kg for hybridization of larynx sections and of 2.7 x lo7 cpm/yg for hybridization of ectocervical sections. The probe was separated from the unincorporated 3Towhom correspondence and reprint requests should be sent.
Received: December 30,1991.
693
FIBRONECTIN mRNA IN HUMAN CARCINOMAS
nucleotides on a Sephadex G-50 column, precipitated, dried and re-dissolved in the hybridization buffer. In situ hybridization Slides were subjected to in situ hybridization according to Moro et al. (1990). The slides, hydrated in PBS for 10 min, were sequentially immersed in 0.2N HCI for 20 min, in PBS for 2 x 3 min, in 1 pgiml proteinase K, dissolved in 10 mM Tris-HC1, 2 mM CaC12, pH 7.4, at 37°C for 15 rnin and in 2 pg/ml glycine in PBS for 2 x 3 min. The preparations were acetylated in freshly prepared 0.25% acetic anhydride in 0.1 M triethanolamine-HC1 pH 8.0 for 10 rnin (Hayashi et al., 1978), washed in distilled water, dehydrated in 50%, 75%, 95% and 100% ethanol 5 rnin for each passage and air-dried. The hybridization was performed in a mixture containing 1 pg/ml 3H-labelled probe, 50% deionized formamide, 10% dextran sulphate, 2 x SSC, 40 mM Na2HP04, 0.1% SDS, 1 x Denhardt's solution and 1,000-fold excess of salmon sperm
DNA as carrier. Before hybridization, the mixture was heated to 70°C for 10 min and cooled in ice. Aliquots of 15-25 p1were applied onto each section, overlaid with a coverslip and incubated at 42°C in a 50% formamide/2 x SSC saturated environment for 16-18 hr. Slides were rinsed 4 times for 10 min in 50% formamide/2 x SSC at 39"C, 3 times for 10 min in 2 x SSC at 3 9 T , 3 times for 10 min in 2 x SSC at room temperature and 2 times for 1 hr in 1 x SSC at room temperature, then dehydrated through an ethanol series (50%, 75%, 95%, 100%) 10 rnin for each passage and air-dried. Autoradiography was performed by dipping the slides into Kodak NTB2 emulsion diluted 1:1 with 2% glycerol at 42"C, air-drying for 2 hr in the dark and storing for a week in sealed boxes at 4°C. The slides were developed in Kodak D19 developer, fixed in Kodak fixer, rinsed in water, air-dried, stained with hematoxylin-eosin and mounted with glycerolgelatin.
B
A GREV IMAGE CAPTURE RND STORRGE
I I
I
I
t
I
D E F I N l T l D N OF C E L L O U T L I N E RND P R O C E S S I N G B Y L R P L R C E RND RRNK D I G I T R L F I L T E R S
S I L U E R G R R I H S ' THRESHOLD i n i o B I N R R V IMAGE
I
I
2
3
I S U B T R A C T I O N OF RRER OF NUCLEUS FROM THE TOTRL
I
1
I
I
4
I
I
COLOR CONTRRST OF THE PROCESSED IMRGE RND COMPRRISON U I T H THE STORED GREV IMAGE
I
MERSUREIIENT OF RRER OF G R R I N S RHO OF C E L L COMPONENTS
5
I
L L - 6
RNOTHER C E L L 7
I
DRTR F I L E C R E A T I O N
I
F~GURE 1 - (a) Flow chart of irz situ hybridization analysis developed from IN SITU software available from Magiscan Image Analysis System. (b) Selected hybridized section during analysis.
694
MORO ET AL.
Image analysis Quantitative evaluation of the hybridization signals was performed using the Magiscan Image Analysis System (Joyce Loebl, Gateshead, UK). A task-list file for the analysis of hybridized sections was developed from IN SITU software, available on the Magiscan, that could be executed in a interactive mode on different images (Fig. 1).
One tissue image was input (at a time) via a TV camera mounted on a Nikon light microscope, digitalized on the high-resolution monitor (using 64 density levels on a grid of 512 x 512 pixels) and stored within the Magiscan’s image memory as “grey image”. The following step was the “light pen” definition of a field, in “grey image”, in which the area of the grains had to be measured. The image was processed by the application of a Laplace linear filter and a Rank non-linear filter to enhance image boundary definition and remove noise. The image was then segmented to form a binary image corresponding to silver grains. Regions of interest (i.e. epithelium or tumor), within the selected field, could be extracted by drawing the boundaries with the “light pen”. The sequence of options in the task-list was repeated until a sufficient number of different fields on a slide was measured, then a data file containing the measurements was created for analysis by the “Results” program. Between the available measurements we considered: the field area, the area of extracted regions and the area of silver grains associated with each compartment (i.e. epithelium and connective tissue or tumor and stroma). RESULTS
FIGURE 2 -In situ hybridizationon G1 laryngeal (a) and ectocervical (b) carcinoma sections using an FN cDNA-radiolabelled probe. S, stroma; T, tumor mass; V, venules; C, capillary vessels.
In order to verify whether stromal FN originates from the serum and/or is produced by the stromal fibroblasts or by endothelial cells of blood vessels, we performed in situ hybridization with a cDNA probe specific for human FN mRNA on sections of control tissues and of laryngeal and ectocervical carcinomas of different gradings. The results obtained after autoradiography (Fig. 2) showed that the FN mRNAproducing cells were the stromal fibroblasts and the endothelial cells of capillary vessels present in the peritumoral stroma. The cells inside the tumor masses, as well as endothelial cells surrounding venules, did not express significant amounts of FN mRNA. Semi-quantitative evaluation of FN mRNA expressed in control and tumor tissue sections was performed by image analysis (Fig. 3). Ten fields of comparable areas in control (i.e. epithelium and connective tissue) and in tumor sections (ie. tumor masses and peritumoral stroma) were evaluated; the
A
B
FIGURE3 - Detection of F N mRNA by in situ hybridization and image analysis on normal larynx and laryngeal carcinomas (G1 to G3) and on normal ectocervix and ectocervical carcinomas (G1 to G3) (b). Epithelia and tumors are in pink; connective tissues and stromas are in red. (a)
695
FIBRONECTIN mRNA I N HUMAN CARCINOMAS TABLE I - QUANTITATIVE EVALUATION OF FN mRNA IN SECTIONS FROM NORMAL LARYNX AND FROM CARCINOMAS OF LARYNX BY IN SITU HYBRIDIZATION AND IMAGE ANALYSIS
TABLE I11 - FIBRONECTIN mRNA PER UNIT O F AREA IN LARYNGEAL AND ECTOCERVICAL CARCINOMAS BY IN SITU HYBRIDIZATION AND IMAGE ANALYSIS
Normal larvnx Epithelium
n1 x2 u3
cv4
6111742331 0.8 0.4 46.4
X NS
1.0
Laryngeal carcinomas Connective tissue
15601758289 2.1 0.9 44.0
EN
Stage
Tumor
Stimulation1
Stroma
Stimulation
1.0
G1 G2 G3
206 194 147
2.5 2.3 1.8
1390 1083 45 1
6.8 5.3 2.2
Stage
Tumor
Stimulation*
Stroma
Stimulation
G1 G2 G3
189 77 91
2.1 0.8 1.0
1234 738 500
12.3 7.4 5.0
Ectocewical carcinomas
Laryngeal carcinomas Stage
G1
Tumor
nX U
G2
G3
cv n
nX U
cv
14351815024 1.8 0.7 40.6 16711865574
11151762792 1.4 0.7 47.2
XN
2.2
1.8
Stroma
106241762637 14.0 4.3 30.4 63851588195
20001442488 4.2 3.2 76.8
XN
7.0
'Stimulation factor obtained by normalizing in relation to control larynx epithelium (83) and connective tissue (205).2Stimulation factor obtained by normalizing in relation to control ectocervical epithelium (91) and connective tissue (100).
'H area of grains in pixels18 detected areas out of 10 randomly chosen fields.-h avera es out of 10 detected fields x103.3Standard deviation x 10?-4Coeficient of variation = u/E. 100.-sX N, F normalized in relation to normal tissues.
B
A
2.0
W E
a
E
5
1000
3
a W
& -I Y)
W
I
500
&
n X
U
CV
8541940899 0.9 0.1 12.0
9811983790
1.0
1.0
1.0
0.3
N
Ectocewical carcinomas Staee
G1
G2
Tumor
n
nX U
G3
cv n X
U
cv
IN
62
63
N
61
62
63
Stroma
XN
FIGURE4 - FN mRNA levels in the stroma surrounding laryngeal (a) and ectocervical (b) carcinomas expressed per unit of area (105 pixels).
9886 / 803971
16601879483
6361826833 0.8 0.4 50.0 7491846372 0.8 0.4 51.4
61
34.n
0.9
0.9
5430/738600 7.4 3.9 53.0 18971381127 5.7 2.5 44.7
7.4
5.7
For explanation of symbols, see footnotes to Table I. averages of the ratios between the area of the grains and the selected surfaces in tumor sections were determined and normalized vs. control tissues. Tables I and I1 report the quantitative evaluation of the hybridization signals obtained in laryngeal and ectocervical sections, respectively, excluding from the analysed area the visible capillaries. Laryngeal carcinoma cells exprcss levels of F N mRNA which are 2 or 3 times higher than the amount measured in control epithelium. The peritumoral stromal fibroblasts around G 1 laryngeal carcinomas express levels of FN mRNA which are 7 times higher than those detected in the cells of the connective tissue. Stromal cells surrounding G2 and G 3 laryngeal carcinomas produce levels of F N mRNA which are
respectively 5 and 2 times higher than those expressed by connective tissue fibroblasts (Table I). The cells in the ectocervical carcinoma nests express low amounts of FN mRNA, similar to those measured in control epithelium. The levels of FN mRNA measured in the fibroblasts surrounding G 1 ectocervical tumors are 12.6 times higher than those expressed by control tissue; this value drops to 7.4 and 5.7 in G2 and G 3 tumor sections, respectively (Table 11). The differences in levels of FN mRNA expressed by laryngeal and ectocervical stromal fibroblasts could be due to differences between the connective tissue specimens taken as controls: while the ectocervical normal tissue was from nonneoplastic cerviccs, the laryngeal control tissue was obtained from peritumoral, histologically normal larynx. The basal levels of F N mRNA measured in the laryngeal connective tissue fibroblasts might be increased as a consequence of tumor-derived diffusible factors. Table I11 and Figure 4 report the levels of FN mRNA expressed by control and tumor sections cxpressed per units of area.
DISCUSSION
A number of studies have been focused on the altered distribution of extracellular matrix and basement membrane
696
MORO ETAL.
components in malignant conditions (Alitalo and Vaheri, 1982; van den Hooff, 1988; Erickson and Bourdon, 1989). In several tumor types, FN deposition has been observed in the stroma surrounding the tumors (Stenman and Vaheri, 1981). So far, it has not been ascertained whether the peritumoral FN is produced by stromal fibroblasts, or by endothelial cells of blood vessels; FN might also stem from blood vessels. In this work we have studied FN mRNA expression in invasive laryngeal and ectocervical carcinomas by means of in situ hybridization and image analysis. The results show that laryngeal and ectocervical carcinoma cells, as well as cells from carcinomas localized in other organs (Vaheri et al., 1989), contain low levels of FN mRNA in agreement with the protein data. We also observed that high levels of FN mRNA are expressed by the stromal fibroblasts and by endothelial cells of the capillary vessels surrounding the tumors. In situ hybridization analysis indicates that peritumoral FN should be synthesized mainly by stromal fibroblasts. Furthermore, in the analysed specimens, FN mRNA expressed by peritumoral fibroblasts is a function of the tumor stage: stromal fibroblasts surrounding well-differentiated carcinomas produce the highest levels of FN mRNA, which diminish in moderately and poorly differentiated carcinomas. These data indicate that a progressive modification in the normal distribution of FN mRNA patterns occurs in laryngeal and ectocervical carcinomas in relation to tumor growth and invasion. The expression of FN probably represents a host reaction antagonizing tumor growth and invasion. The role of FN in maintaining the normal cell phenotype has been reported (Yamada et al., 1976). Besides, the synthesis of excessive ECM, i.e. desmoplasia, frequently accompanying invasive carcinomas of colon, breast and prostate (Meissner and Didamandopoulos, 1977) has been reported to temporarily antagonize tumor growth by inhibiting angiogenesis (Barsky et al., 1987).
The pattern of FN mRNA expression in the peritumoral stroma could be explained assuming that the reaction of stromal fibroblasts to G1 tumors (i.e. synthesis of FN) is overcome by activities favoring tumor growth and invasion, released and/or induced by the tumor cells. It is known that tumors release enhanced levels of serine-proteases and of metalloproteases necessary for the degradation of peritumoral tissues and tumor growth (Dana et al., 1985; Matrisian, 1990). These proteases are induced in stromal fibroblasts by the tumor cells: breast adenocarcinomas induced, in the stroma, the synthesis of stromelysin-3 mRNA according to a gradient (Basset et aZ., 1990) and colon carcinomas induced the synthesis of mRNA encoding urokinase in the surrounding fibroblasts (Dano et aZ., 1985; Pyke et aZ., 1991). In conclusion, tumors (ix.carcinomas) induce in the peritumoral stroma not only proteolytic activities favoring their growth and invasion, but also ECM glycoproteins (i.e. fibronectin) which may antagonize these processes. Tumor growth and tissue invasion, at early tumor stages, could depend on the balance between the levels of proteins maintaining or degrading the peritumoral tissues.
ACKNOWLEDGEMENTS
We thank Drs. A.R. Antonelli and U. Bianchi for providing laryngeal and ectocervical specimens, respectively, Drs. P.G. Grigolato, M. Favret, S. Parolini and D. Rosa for histopathological examination and Ms. M.R. Nava for secretarial assistance. This work was supported by AIRC, CNR Target Projects “Biotechnology and Bioinstrumentation”, “Genetic Engineering”, “Clinical Applications of Oncological Research” and grant 9002063CTII.
REFERENCES ALITALO, K. and VAHERI, A,, Pericellular matrix in malignant transfor- KORNBLIHTT, A.R., VIBE-PEDERSEN, K. and BARALLE, F.E., Human mation.Advanc. Cancer Res., 37,111-158 (1982). fibronectin: cell specific alternative mRNA splicing generates polypepANTONELLI, A.R., NICOLAI, P., CAPPIELLO, J., PERETTI,G., MOLINARI tide chains differing in the number of internal repeats. Nucl. Acids P.G., FAVRET, M. and MAROC- Res., 12,5853-5868 (1984). TOSATTI, M.P., ROSA,D., GRIGOLATO, COLO,D., Basement membrane components in normal, dysplastic, MATRISIAN, L.M., Metalloproteinases and their inhibitors in matrix neoplastic laryngeal tissue and metastatic lymph nodes. Acta Otolalyn- remodeling. Trends Genet., 6,121-125 (1990). gol. (Stockh.), 11,437-443 (1991). MCCARTHY, J.B., BASARA, M.L., PALM,S.L., SAS,D.F. and FURCHT, BARSKY, S.M., NELSON,L.L. and LEVY,V.A., Tumor desmoplasia L.. The role of cell adhesion uroteins-laminin and fibronectin-in the inhibits angiogenesis. Lancet, 11,1336-1337 (1987). movement of malignant metastatic cells. Cancer Mefusf. Rev., 4, BASSET, P., BELLOCQ, J.P., WOLF,C., STOLL,I., HUTIN,P., LIMACHER, 125-152 (1985). J.M., PODHAJCER, O.L., CHENARD, M.P., Rro, C.M. and CHAMBON, P., W.A. and DIDAMANDOPOULOS. G.. Neoolasia. In: W.A.D. A novel metalloproteinase gene specifically expressed in stromal cells MEISSNER. Anderson’and J.M. Kissane (eds.), Puthohgy,’7th id., p. 665, Mosby, of breast carcinomas. Nature (Lond.), 248,699-704 (1990). St. Louis (1977). DAN^, K., ANDREASEN, P.A., GRONDAHL-HANSEN, J., KRISTENSEN, P., S., ROSA, D., BONARDI,F., L., Plasminogen activators, tissue degrada- MOLINARITOSATTI,M.P., PAROLINI, NIELSEN,L.S. and SKRIVER, FAVRET,M., RAMPINELLI, F. and BIANCHI, U., Basement membrane tion and cancer. Advanc. Cancer Res., 44,139-266 (1985). components in normal, dysplastic, neoplastic cervix tissues. Cervix ERICKSON, H.P. and BOURDON,M.A., Tenascin: an extracellular (1992). (In press). matrix protein prominent in specialized embryonic tissues and tumors. MORO, L., COLOMBI, M., CRESPI,S., DI LERNIA,R. and BARLATI, S., Ann. Rev. CellBiol., 5,71-92 (1989). Study of fibronectin expression in tumor cells by dot-blot and in situ FOGERTY, F.J., AKIYAMA, S.K., YAMADA, K.M. and MOSHER,D.F., Inhibition of binding of fibronectin to matrix assembly sites by hybridization: quantitative evaluation by image analysis. Cell Bid. int. Rep., 14,701-714 (1990). p l ) antibodies. J. CellBiol., 111,699-708 (1990). anti-integrin (d, P., RALFKIAER, E., GRONDAHL-HANSEN, J., GRINNELL, F., Fibronectin and wound healing. J. Cell Biochem., 26, PYKE,C., KRISTENSEN, ERIKSEN, J., BLASI,F. and DAN0, K., Urokinase-type plasminogen 107-116 (1984). activator is expressed in stromal cells and its receptor in cancer cells at HAYASHI, S., GILLAM, I.C., DELANEY, A.D. and TENER,G.M., Acetyla- invasive foci in human adenocarcinomas. Amer. J. Pathol., 138, tion of chromosome squashes of Drosophila melunogasfer decreases the 1059-1067 (1991). background in autoradiographs from hybridization with [1251]-labeled RUOSLAHTI, E., Fibronectin and its receptors. Ann. Rev. Biochem., 57, RNA. J. Histochem. Cytochem., 26,677-679 (1978). 375413 (1988). HUMPHRIES, M.J., OBARA,M., OLDEN,K. and YAMADA, K.M., Role of E., ENGVALL, E. and HAYMAN, G., Fibronectin: current fibronectin in adhesion, migration and metastasis. Cancer Invest., 7, RUOSLAHTI, concepts of its structure and functions. Coll. Rex, 1,95-128 (1981). 373-393 (1989). HYNES,R.O., Alteration of cell surface proteins by viral transforma- STENMAN, S. and VAHERI,A,. Distribution of a maior connective tissue tion and by proteolysis. Proc. nut. Acad. Sci. (Wash.), 74, 2855-2859 protein, fibronectin, in normal human tissue. -J. exp. Med., 147, (1973). 1054-1064 (1978).
FIBRONECTIN mRNA IN HUMAN CARCINOMAS
STENMAN, S. and VAHERI, A., Fibronectin in human solid tumors. Int. J. Cancer, 27,427-435 (1981). THIERY,J.P., DUBAND,J.L., DUFOUR,S., SAVAGNER, P. and IMHOF, B.A., Roles of fibronectins in embrvogenesis. In: D.F. Mosher (ed.). \ , Fibronectin, Academic Press, San Dikgo (1989). VAHERI, A. KESKI-OJA, J. and TAPIO,V., Fibronectin and malignant transformation. In: D.F. Mosher (ed.), Fibronectin, Academic Press, San Diego (1989). VAHERI,A. and RUOSLAHTI, E., Disappearance of major cell-type-
697
specific surface glycoprotein antigen (SF) after transformation of fibroblasts by Rous sarcoma virus. Int. J. Cancer, 13,579-586 (1974). VAN DEN HOOFF, A,, Stromal i~vOlvement in malignant growth. Advunc. Cancer R e s . ~ 159-196 (1988). YAMADA,K.M., Cell surface interactions with extra-cellular materials. Ann. Rev.Biochem., 52,761-799 (1983). YAMADA7 s. and PASTAN, I., surface protein partially restores morphology, adhesiveness and contact inhibition of movement to transformed fibroblasts. Proc. naf.Acad. Sci. Wash. ). 73. 1217-1221 (1976). K.M.7
Publication of the International Union Against Cancer Publication de I'Union Internationale Contre le Cancer
STUDY OF FIBRONECTIN mRNA IN HUMAN LARYNGEAL AND ECTOCERVICAL CARCINOMAS BY IN SITU HYBRIDIZATION AND IMAGE ANALYSIS Laura MORO', Marina COLOMBI',Maria Pia MOLINARI TO SAT TI^ and Sergio BARLAT11,3 'Division of Biology and Genetics and 2Divisionof Histology, Depaifment of Biomedical Sciences and Biotechnology, University of Brescia, Itab. The expression of fibronectin (FN) mRNA was studied in histological sections of surgical biopsies from human laryngeal and ectocervical invasive carcinomas of different grading stages by in sku hybridization and image analysis. This approach made it possible to identify the cell types synthesizing FN mRNA in the tissue sections and to compare semi-quantitatively the FN mRNA levels expressed in the different specimens. The carcinoma cells synthesized low levels of FN mRNA, comparable to those detected in control epithelia and connective-tissue fibroblasts. Well-differentiated (G I) laryngealand ectocervical carcinomas induced the synthesis of FN mRNA-to levels 7 to 13 times higher than in control connective tissues-in the stromal fibroblasts surrounding the tumors. In carcinoma samples analysed, the amount of FN mRNA detected in the stroma decreased in relation to tumor grading (from GI to G3) and the stromal destruction. FN mRNA was not detectable in the endothelial cells of venules while it was present in large amounts in those surrounding the capillaries present in the stroma. These data indicate that FN, usually observed around carcinomas, is produced by stromal fibroblasts, which are induced to express FN mRNA, presumably in response to diffusible factors produced by the tumor cells, and/or by endothelial cells of the infiltrating capillary vessels. The induction of FN mRNA, inversely proportional to the tumor grading, may be useful in evaluatingthe invasion potential of the tumor.
o 1992 Wiley-Liss, Inc.
Fibronectins (FNs) are high-molecular-weight, adhesive glycoproteins particularly abundant in body fluids, loose connective tissue, granulation tissue, basement membranes and embryonic structures (Stenman and Vaheri, 1978; Ruoslahti et al., 1981; ThiCry et al., 1989). Tissue FN interacts with many components of the extracellular matrices (ECM) (i.e. collagens and glycosaminoglycans) (Yamada, 1983) as well as with cell-surface receptors (Ruoslahti, 1988; Fogerty et a l , 1990) and may therefore provide a scaffold for ECM assembly and organization. FNs also play a role in tissue remodelling and cell migration that occur during embryonic development and wound healing (Grinnell, 1984; ThiCry et al., 1989). Much evidence has been obtained regarding the role of ECMadhesive glycoproteins in tumor growth and invasion (Alitalo and Vaheri, 1982; McCarthy et aL, 1985; van den Hooff, 1988; Vaheri et aZ., 1989; Humphries et al., 1989). FN is greatly reduced, or absent, on the surface of cultured tumor-derived or in vitro-transformed cells (Hynes, 1973; Vaheri and Ruoslahti, 1974). Immunohistochemical analysis of sections from human solid tumors of several types revealed that FN was not expressed by carcinoma and melanoma cells, while it was synthesized by individual cells in some sarcomas (Stenman and Vaheri, 1981). The synthesis of FN by tumor cells was therefore related to their embryonic derivation. The stroma surrounding advanced carcinomas was, in contrast, strongly positive for FN (Stenman and Vaheri, 1981). Previous immunohistochemical studies on laryngeal (Antonelli et aL, 1991) and ectocervical carcinoma sections (Molinari-Tosatti et al., 1992) showed that carcinoma cells lacked intra- or pericellular FN, while fibroblasts surrounding tumors expressed consistent amounts of FN. Stromal FN may be synthesized by fibroblasts or by endothelial cells of blood
vessels; it may also partly stem from serum as an inflammatory exudation. In order to verify the source of the stromal FN we performed in situ hybridization with a radiolabelled FN cDNA, combined with image analysis of sections from control human tissues and from laryngeal and ectocervical invasive carcinomas. Semiquantitative evaluation of FN mRNA expressed by the different cell types present in the sections of tumors at different stages was also made possible by this approach. MATERIAL AND METHODS
Chemicals and reagents Bio-plast was obtained from Bioptica, Milan, Italy; proteinase K, paraformaldehyde, triethanolamine, acetic anhydride and formamide from BDH Italia, Milan; Sephadex G-50, Ficoll and polyvinyl pyrrolidone from Pharmacia L I B , Uppsala, Sweden; dextran sulphate, BSA and glycerol-gelatin from Sigma, St. Louis, MO, and glycine from BioRad, Richmond, CA. Preparation of tissue sections Tissue specimens were obtained from patients whose clinical diagnosis was confirmed by histopathological examination. From each patient bearing laryngeal carcinoma, 2 surgical biopsies were collected, one from areas of the tumor showing no signs of necrosis, the other from apparently normal laryngeal mucosa. Ectocervical carcinoma specimens were collected according to the same criteria, while the normal tissue was obtained from patients who had undergone surgical operations for non-neoplastic diseases. At least 2 different tumors were examined for each grading. Tissue specimens were fixed in 4% paraformaldehyde in PBS, dehydrated in rising ethanol series, cleared in xylene and embedded in bio-plast using routine histological procedures. Sections 5 pm thick were cut from the tissue blocks, spread on microscope slides and incubated overnight at 37°C. Before hybridization, the slides were washed for 2 x 10 min in xylene, 1 x 10 min in 100% ethanol and 1 x 10 min in 70% ethanol. Fibronectin cDNA probe and labellingprocedures The probe used was a HindIII/BamHI fragment of 2.1 kb (FN 1-1) obtained from the pFHl recombinant plasmid which contains the 3'-terminal portion of FN cDNA (Kornblihtt et al., 1984). For in situ hybridization, 500 ng of FN cDNA were labelled by nick translation, according to the supplier's instructions (Gibco-BRL, Gaithersburg, MD) in the presence of 6 pM each of 3H-TTP (103 Ci/mM), 3H-dCTP (53 Ci/mM), 3H-dATP (65 Ci/mM) (Amersham, Aylesbury, UK) and 6 KM unlabelled dGTP. The DNA probe was labelled to a specific activity of 3.5 x lo7 cpm/kg for hybridization of larynx sections and of 2.7 x lo7 cpm/yg for hybridization of ectocervical sections. The probe was separated from the unincorporated 3Towhom correspondence and reprint requests should be sent.
Received: December 30,1991.
693
FIBRONECTIN mRNA IN HUMAN CARCINOMAS
nucleotides on a Sephadex G-50 column, precipitated, dried and re-dissolved in the hybridization buffer. In situ hybridization Slides were subjected to in situ hybridization according to Moro et al. (1990). The slides, hydrated in PBS for 10 min, were sequentially immersed in 0.2N HCI for 20 min, in PBS for 2 x 3 min, in 1 pgiml proteinase K, dissolved in 10 mM Tris-HC1, 2 mM CaC12, pH 7.4, at 37°C for 15 rnin and in 2 pg/ml glycine in PBS for 2 x 3 min. The preparations were acetylated in freshly prepared 0.25% acetic anhydride in 0.1 M triethanolamine-HC1 pH 8.0 for 10 rnin (Hayashi et al., 1978), washed in distilled water, dehydrated in 50%, 75%, 95% and 100% ethanol 5 rnin for each passage and air-dried. The hybridization was performed in a mixture containing 1 pg/ml 3H-labelled probe, 50% deionized formamide, 10% dextran sulphate, 2 x SSC, 40 mM Na2HP04, 0.1% SDS, 1 x Denhardt's solution and 1,000-fold excess of salmon sperm
DNA as carrier. Before hybridization, the mixture was heated to 70°C for 10 min and cooled in ice. Aliquots of 15-25 p1were applied onto each section, overlaid with a coverslip and incubated at 42°C in a 50% formamide/2 x SSC saturated environment for 16-18 hr. Slides were rinsed 4 times for 10 min in 50% formamide/2 x SSC at 39"C, 3 times for 10 min in 2 x SSC at 3 9 T , 3 times for 10 min in 2 x SSC at room temperature and 2 times for 1 hr in 1 x SSC at room temperature, then dehydrated through an ethanol series (50%, 75%, 95%, 100%) 10 rnin for each passage and air-dried. Autoradiography was performed by dipping the slides into Kodak NTB2 emulsion diluted 1:1 with 2% glycerol at 42"C, air-drying for 2 hr in the dark and storing for a week in sealed boxes at 4°C. The slides were developed in Kodak D19 developer, fixed in Kodak fixer, rinsed in water, air-dried, stained with hematoxylin-eosin and mounted with glycerolgelatin.
B
A GREV IMAGE CAPTURE RND STORRGE
I I
I
I
t
I
D E F I N l T l D N OF C E L L O U T L I N E RND P R O C E S S I N G B Y L R P L R C E RND RRNK D I G I T R L F I L T E R S
S I L U E R G R R I H S ' THRESHOLD i n i o B I N R R V IMAGE
I
I
2
3
I S U B T R A C T I O N OF RRER OF NUCLEUS FROM THE TOTRL
I
1
I
I
4
I
I
COLOR CONTRRST OF THE PROCESSED IMRGE RND COMPRRISON U I T H THE STORED GREV IMAGE
I
MERSUREIIENT OF RRER OF G R R I N S RHO OF C E L L COMPONENTS
5
I
L L - 6
RNOTHER C E L L 7
I
DRTR F I L E C R E A T I O N
I
F~GURE 1 - (a) Flow chart of irz situ hybridization analysis developed from IN SITU software available from Magiscan Image Analysis System. (b) Selected hybridized section during analysis.
694
MORO ET AL.
Image analysis Quantitative evaluation of the hybridization signals was performed using the Magiscan Image Analysis System (Joyce Loebl, Gateshead, UK). A task-list file for the analysis of hybridized sections was developed from IN SITU software, available on the Magiscan, that could be executed in a interactive mode on different images (Fig. 1).
One tissue image was input (at a time) via a TV camera mounted on a Nikon light microscope, digitalized on the high-resolution monitor (using 64 density levels on a grid of 512 x 512 pixels) and stored within the Magiscan’s image memory as “grey image”. The following step was the “light pen” definition of a field, in “grey image”, in which the area of the grains had to be measured. The image was processed by the application of a Laplace linear filter and a Rank non-linear filter to enhance image boundary definition and remove noise. The image was then segmented to form a binary image corresponding to silver grains. Regions of interest (i.e. epithelium or tumor), within the selected field, could be extracted by drawing the boundaries with the “light pen”. The sequence of options in the task-list was repeated until a sufficient number of different fields on a slide was measured, then a data file containing the measurements was created for analysis by the “Results” program. Between the available measurements we considered: the field area, the area of extracted regions and the area of silver grains associated with each compartment (i.e. epithelium and connective tissue or tumor and stroma). RESULTS
FIGURE 2 -In situ hybridizationon G1 laryngeal (a) and ectocervical (b) carcinoma sections using an FN cDNA-radiolabelled probe. S, stroma; T, tumor mass; V, venules; C, capillary vessels.
In order to verify whether stromal FN originates from the serum and/or is produced by the stromal fibroblasts or by endothelial cells of blood vessels, we performed in situ hybridization with a cDNA probe specific for human FN mRNA on sections of control tissues and of laryngeal and ectocervical carcinomas of different gradings. The results obtained after autoradiography (Fig. 2) showed that the FN mRNAproducing cells were the stromal fibroblasts and the endothelial cells of capillary vessels present in the peritumoral stroma. The cells inside the tumor masses, as well as endothelial cells surrounding venules, did not express significant amounts of FN mRNA. Semi-quantitative evaluation of FN mRNA expressed in control and tumor tissue sections was performed by image analysis (Fig. 3). Ten fields of comparable areas in control (i.e. epithelium and connective tissue) and in tumor sections (ie. tumor masses and peritumoral stroma) were evaluated; the
A
B
FIGURE3 - Detection of F N mRNA by in situ hybridization and image analysis on normal larynx and laryngeal carcinomas (G1 to G3) and on normal ectocervix and ectocervical carcinomas (G1 to G3) (b). Epithelia and tumors are in pink; connective tissues and stromas are in red. (a)
695
FIBRONECTIN mRNA I N HUMAN CARCINOMAS TABLE I - QUANTITATIVE EVALUATION OF FN mRNA IN SECTIONS FROM NORMAL LARYNX AND FROM CARCINOMAS OF LARYNX BY IN SITU HYBRIDIZATION AND IMAGE ANALYSIS
TABLE I11 - FIBRONECTIN mRNA PER UNIT O F AREA IN LARYNGEAL AND ECTOCERVICAL CARCINOMAS BY IN SITU HYBRIDIZATION AND IMAGE ANALYSIS
Normal larvnx Epithelium
n1 x2 u3
cv4
6111742331 0.8 0.4 46.4
X NS
1.0
Laryngeal carcinomas Connective tissue
15601758289 2.1 0.9 44.0
EN
Stage
Tumor
Stimulation1
Stroma
Stimulation
1.0
G1 G2 G3
206 194 147
2.5 2.3 1.8
1390 1083 45 1
6.8 5.3 2.2
Stage
Tumor
Stimulation*
Stroma
Stimulation
G1 G2 G3
189 77 91
2.1 0.8 1.0
1234 738 500
12.3 7.4 5.0
Ectocewical carcinomas
Laryngeal carcinomas Stage
G1
Tumor
nX U
G2
G3
cv n
nX U
cv
14351815024 1.8 0.7 40.6 16711865574
11151762792 1.4 0.7 47.2
XN
2.2
1.8
Stroma
106241762637 14.0 4.3 30.4 63851588195
20001442488 4.2 3.2 76.8
XN
7.0
'Stimulation factor obtained by normalizing in relation to control larynx epithelium (83) and connective tissue (205).2Stimulation factor obtained by normalizing in relation to control ectocervical epithelium (91) and connective tissue (100).
'H area of grains in pixels18 detected areas out of 10 randomly chosen fields.-h avera es out of 10 detected fields x103.3Standard deviation x 10?-4Coeficient of variation = u/E. 100.-sX N, F normalized in relation to normal tissues.
B
A
2.0
W E
a
E
5
1000
3
a W
& -I Y)
W
I
500
&
n X
U
CV
8541940899 0.9 0.1 12.0
9811983790
1.0
1.0
1.0
0.3
N
Ectocewical carcinomas Staee
G1
G2
Tumor
n
nX U
G3
cv n X
U
cv
IN
62
63
N
61
62
63
Stroma
XN
FIGURE4 - FN mRNA levels in the stroma surrounding laryngeal (a) and ectocervical (b) carcinomas expressed per unit of area (105 pixels).
9886 / 803971
16601879483
6361826833 0.8 0.4 50.0 7491846372 0.8 0.4 51.4
61
34.n
0.9
0.9
5430/738600 7.4 3.9 53.0 18971381127 5.7 2.5 44.7
7.4
5.7
For explanation of symbols, see footnotes to Table I. averages of the ratios between the area of the grains and the selected surfaces in tumor sections were determined and normalized vs. control tissues. Tables I and I1 report the quantitative evaluation of the hybridization signals obtained in laryngeal and ectocervical sections, respectively, excluding from the analysed area the visible capillaries. Laryngeal carcinoma cells exprcss levels of F N mRNA which are 2 or 3 times higher than the amount measured in control epithelium. The peritumoral stromal fibroblasts around G 1 laryngeal carcinomas express levels of FN mRNA which are 7 times higher than those detected in the cells of the connective tissue. Stromal cells surrounding G2 and G 3 laryngeal carcinomas produce levels of F N mRNA which are
respectively 5 and 2 times higher than those expressed by connective tissue fibroblasts (Table I). The cells in the ectocervical carcinoma nests express low amounts of FN mRNA, similar to those measured in control epithelium. The levels of FN mRNA measured in the fibroblasts surrounding G 1 ectocervical tumors are 12.6 times higher than those expressed by control tissue; this value drops to 7.4 and 5.7 in G2 and G 3 tumor sections, respectively (Table 11). The differences in levels of FN mRNA expressed by laryngeal and ectocervical stromal fibroblasts could be due to differences between the connective tissue specimens taken as controls: while the ectocervical normal tissue was from nonneoplastic cerviccs, the laryngeal control tissue was obtained from peritumoral, histologically normal larynx. The basal levels of F N mRNA measured in the laryngeal connective tissue fibroblasts might be increased as a consequence of tumor-derived diffusible factors. Table I11 and Figure 4 report the levels of FN mRNA expressed by control and tumor sections cxpressed per units of area.
DISCUSSION
A number of studies have been focused on the altered distribution of extracellular matrix and basement membrane
696
MORO ETAL.
components in malignant conditions (Alitalo and Vaheri, 1982; van den Hooff, 1988; Erickson and Bourdon, 1989). In several tumor types, FN deposition has been observed in the stroma surrounding the tumors (Stenman and Vaheri, 1981). So far, it has not been ascertained whether the peritumoral FN is produced by stromal fibroblasts, or by endothelial cells of blood vessels; FN might also stem from blood vessels. In this work we have studied FN mRNA expression in invasive laryngeal and ectocervical carcinomas by means of in situ hybridization and image analysis. The results show that laryngeal and ectocervical carcinoma cells, as well as cells from carcinomas localized in other organs (Vaheri et al., 1989), contain low levels of FN mRNA in agreement with the protein data. We also observed that high levels of FN mRNA are expressed by the stromal fibroblasts and by endothelial cells of the capillary vessels surrounding the tumors. In situ hybridization analysis indicates that peritumoral FN should be synthesized mainly by stromal fibroblasts. Furthermore, in the analysed specimens, FN mRNA expressed by peritumoral fibroblasts is a function of the tumor stage: stromal fibroblasts surrounding well-differentiated carcinomas produce the highest levels of FN mRNA, which diminish in moderately and poorly differentiated carcinomas. These data indicate that a progressive modification in the normal distribution of FN mRNA patterns occurs in laryngeal and ectocervical carcinomas in relation to tumor growth and invasion. The expression of FN probably represents a host reaction antagonizing tumor growth and invasion. The role of FN in maintaining the normal cell phenotype has been reported (Yamada et al., 1976). Besides, the synthesis of excessive ECM, i.e. desmoplasia, frequently accompanying invasive carcinomas of colon, breast and prostate (Meissner and Didamandopoulos, 1977) has been reported to temporarily antagonize tumor growth by inhibiting angiogenesis (Barsky et al., 1987).
The pattern of FN mRNA expression in the peritumoral stroma could be explained assuming that the reaction of stromal fibroblasts to G1 tumors (i.e. synthesis of FN) is overcome by activities favoring tumor growth and invasion, released and/or induced by the tumor cells. It is known that tumors release enhanced levels of serine-proteases and of metalloproteases necessary for the degradation of peritumoral tissues and tumor growth (Dana et al., 1985; Matrisian, 1990). These proteases are induced in stromal fibroblasts by the tumor cells: breast adenocarcinomas induced, in the stroma, the synthesis of stromelysin-3 mRNA according to a gradient (Basset et aZ., 1990) and colon carcinomas induced the synthesis of mRNA encoding urokinase in the surrounding fibroblasts (Dano et aZ., 1985; Pyke et aZ., 1991). In conclusion, tumors (ix.carcinomas) induce in the peritumoral stroma not only proteolytic activities favoring their growth and invasion, but also ECM glycoproteins (i.e. fibronectin) which may antagonize these processes. Tumor growth and tissue invasion, at early tumor stages, could depend on the balance between the levels of proteins maintaining or degrading the peritumoral tissues.
ACKNOWLEDGEMENTS
We thank Drs. A.R. Antonelli and U. Bianchi for providing laryngeal and ectocervical specimens, respectively, Drs. P.G. Grigolato, M. Favret, S. Parolini and D. Rosa for histopathological examination and Ms. M.R. Nava for secretarial assistance. This work was supported by AIRC, CNR Target Projects “Biotechnology and Bioinstrumentation”, “Genetic Engineering”, “Clinical Applications of Oncological Research” and grant 9002063CTII.
REFERENCES ALITALO, K. and VAHERI, A,, Pericellular matrix in malignant transfor- KORNBLIHTT, A.R., VIBE-PEDERSEN, K. and BARALLE, F.E., Human mation.Advanc. Cancer Res., 37,111-158 (1982). fibronectin: cell specific alternative mRNA splicing generates polypepANTONELLI, A.R., NICOLAI, P., CAPPIELLO, J., PERETTI,G., MOLINARI tide chains differing in the number of internal repeats. Nucl. Acids P.G., FAVRET, M. and MAROC- Res., 12,5853-5868 (1984). TOSATTI, M.P., ROSA,D., GRIGOLATO, COLO,D., Basement membrane components in normal, dysplastic, MATRISIAN, L.M., Metalloproteinases and their inhibitors in matrix neoplastic laryngeal tissue and metastatic lymph nodes. Acta Otolalyn- remodeling. Trends Genet., 6,121-125 (1990). gol. (Stockh.), 11,437-443 (1991). MCCARTHY, J.B., BASARA, M.L., PALM,S.L., SAS,D.F. and FURCHT, BARSKY, S.M., NELSON,L.L. and LEVY,V.A., Tumor desmoplasia L.. The role of cell adhesion uroteins-laminin and fibronectin-in the inhibits angiogenesis. Lancet, 11,1336-1337 (1987). movement of malignant metastatic cells. Cancer Mefusf. Rev., 4, BASSET, P., BELLOCQ, J.P., WOLF,C., STOLL,I., HUTIN,P., LIMACHER, 125-152 (1985). J.M., PODHAJCER, O.L., CHENARD, M.P., Rro, C.M. and CHAMBON, P., W.A. and DIDAMANDOPOULOS. G.. Neoolasia. In: W.A.D. A novel metalloproteinase gene specifically expressed in stromal cells MEISSNER. Anderson’and J.M. Kissane (eds.), Puthohgy,’7th id., p. 665, Mosby, of breast carcinomas. Nature (Lond.), 248,699-704 (1990). St. Louis (1977). DAN^, K., ANDREASEN, P.A., GRONDAHL-HANSEN, J., KRISTENSEN, P., S., ROSA, D., BONARDI,F., L., Plasminogen activators, tissue degrada- MOLINARITOSATTI,M.P., PAROLINI, NIELSEN,L.S. and SKRIVER, FAVRET,M., RAMPINELLI, F. and BIANCHI, U., Basement membrane tion and cancer. Advanc. Cancer Res., 44,139-266 (1985). components in normal, dysplastic, neoplastic cervix tissues. Cervix ERICKSON, H.P. and BOURDON,M.A., Tenascin: an extracellular (1992). (In press). matrix protein prominent in specialized embryonic tissues and tumors. MORO, L., COLOMBI, M., CRESPI,S., DI LERNIA,R. and BARLATI, S., Ann. Rev. CellBiol., 5,71-92 (1989). Study of fibronectin expression in tumor cells by dot-blot and in situ FOGERTY, F.J., AKIYAMA, S.K., YAMADA, K.M. and MOSHER,D.F., Inhibition of binding of fibronectin to matrix assembly sites by hybridization: quantitative evaluation by image analysis. Cell Bid. int. Rep., 14,701-714 (1990). p l ) antibodies. J. CellBiol., 111,699-708 (1990). anti-integrin (d, P., RALFKIAER, E., GRONDAHL-HANSEN, J., GRINNELL, F., Fibronectin and wound healing. J. Cell Biochem., 26, PYKE,C., KRISTENSEN, ERIKSEN, J., BLASI,F. and DAN0, K., Urokinase-type plasminogen 107-116 (1984). activator is expressed in stromal cells and its receptor in cancer cells at HAYASHI, S., GILLAM, I.C., DELANEY, A.D. and TENER,G.M., Acetyla- invasive foci in human adenocarcinomas. Amer. J. Pathol., 138, tion of chromosome squashes of Drosophila melunogasfer decreases the 1059-1067 (1991). background in autoradiographs from hybridization with [1251]-labeled RUOSLAHTI, E., Fibronectin and its receptors. Ann. Rev. Biochem., 57, RNA. J. Histochem. Cytochem., 26,677-679 (1978). 375413 (1988). HUMPHRIES, M.J., OBARA,M., OLDEN,K. and YAMADA, K.M., Role of E., ENGVALL, E. and HAYMAN, G., Fibronectin: current fibronectin in adhesion, migration and metastasis. Cancer Invest., 7, RUOSLAHTI, concepts of its structure and functions. Coll. Rex, 1,95-128 (1981). 373-393 (1989). HYNES,R.O., Alteration of cell surface proteins by viral transforma- STENMAN, S. and VAHERI,A,. Distribution of a maior connective tissue tion and by proteolysis. Proc. nut. Acad. Sci. (Wash.), 74, 2855-2859 protein, fibronectin, in normal human tissue. -J. exp. Med., 147, (1973). 1054-1064 (1978).
FIBRONECTIN mRNA IN HUMAN CARCINOMAS
STENMAN, S. and VAHERI, A., Fibronectin in human solid tumors. Int. J. Cancer, 27,427-435 (1981). THIERY,J.P., DUBAND,J.L., DUFOUR,S., SAVAGNER, P. and IMHOF, B.A., Roles of fibronectins in embrvogenesis. In: D.F. Mosher (ed.). \ , Fibronectin, Academic Press, San Dikgo (1989). VAHERI, A. KESKI-OJA, J. and TAPIO,V., Fibronectin and malignant transformation. In: D.F. Mosher (ed.), Fibronectin, Academic Press, San Diego (1989). VAHERI,A. and RUOSLAHTI, E., Disappearance of major cell-type-
697
specific surface glycoprotein antigen (SF) after transformation of fibroblasts by Rous sarcoma virus. Int. J. Cancer, 13,579-586 (1974). VAN DEN HOOFF, A,, Stromal i~vOlvement in malignant growth. Advunc. Cancer R e s . ~ 159-196 (1988). YAMADA,K.M., Cell surface interactions with extra-cellular materials. Ann. Rev.Biochem., 52,761-799 (1983). YAMADA7 s. and PASTAN, I., surface protein partially restores morphology, adhesiveness and contact inhibition of movement to transformed fibroblasts. Proc. naf.Acad. Sci. Wash. ). 73. 1217-1221 (1976). K.M.7
