American Journal of Pathology, Vol. 138, No. 4, Apil 1991 Copyright © American Association of Pathologis
Rapid Communication Cingulin, a Specific Protein Component of Tight Junctions, Is Expressed in Normal and Neoplastic Human Epithelial Tissues Sandra Citi,*tt Andrea Amorosi,t Flavia Franconi, Alberto Giotti, Giancarlo Zampit From the Department of Cell Biology and Anatomy, * Cornell University Medical College, New York, New York; and Istituto di Anatomia Patologicat and Dipartimento di
Farmacologia Preclinica e Clinica, Universita' di Firenze, Firenze, Italy
Cingulin is a 140-kd protein localized on the cytoplasmic face of avian tightjunctions. The expression of cingulin in human normal and neoplastic colonic tissue has been investigated with an antiserum against chicken cingulin. Human cingulin shares its apparent molecular mass and localization with avian cingulin In normal colonic epithelium, villous adenomas, and differentiated adenocarcinomas cingulin staining is observed in the junctional region of the polarized cells lining the surface, the crypts, and the glandular lumina In poorly differentiated adenocarcinomas, labeling also is observed at the interface between cancer tissue and stromag or in clumps of malignant cells; forming a pattern that highlights the presence of small; compressed lumina The cingulin content offour adenocarcinomas, estimated by immunoblotting and densitometry, was higher than that of the normal tissue (150% to 230%). Cingulin was detected in a metastasis from a colon adenocarcinoma but not in nonepithelial tissues and neoplasias, suggesting that cingulin may be a useful marker in the characterization of colonic and probably other epithelial neoplasias. (Am J Pathol 1991, 138:781-789)
Tight junctions are typical specializations of the apicolateral domain of epithelial cell membranes and are absent from tissues of mesenchymal origin.1 In polarized epithe-
lial cells, they separate the endoluminal from the basolateral compartments of the extracellular space (for reviews, see Staehelin2 and Gumbiner3), playing a crucial role in the permeability barrier function of epithelial tissues.4 In addition, tight junctions may be involved in cel-cell adhesion1 and in maintaining cell polarity.5 Freezefractured tight junctions show a network of cross-linked fibrils and complementary grooves, corresponding to the sites of apparent membrane fusion, visible when junctions are analyzed by transmission electron microscopy.6 Recently progress was made in elucidating the biochemical composition of tight junctions, with the discovery of the tight junction specific proteins ZO-17 and cingulin.8 Despite the importance of tight junctions in the function of epithelia, little is known about their distribution and patterns of expression in human disease. The observation that in several solid tumors intercellular adhesiveness is decreased,9 and cel-cell contacts and electrotonic coupling are modified lead to the hypothesis that alterations in intercellular junctions may contribute to cell sloughing and metastasis.1012 In malignant tumors, altered morphology and function of desmosomes, adherens, and gap junctions have been described (reviewed in Weinstein13 and in Weinstein and Pauli14). On the other hand, studies of tight junctions have been hindered by the lack of biochemical markers, electron microscopy being, until recently, the only experimental tool to evaluate tight junction structure. For example, loss or attenuation of tight junctions was reported in virus-transformed cultured murine embryonic cells.15 In contrast, apparently normal tight junctions were described in cultured and metastatic mammary tumor cells16'17 and in several lines of human Supported by Consiglio Nazionale delle Ricerche (Contributo n.88.03127.26), Istituto Scientifico Roussel, MPI, National Institutes of Health (BRSG 2S07 RR05396-28). Sandra Citi is a Cornell Scholar. Accepted for publication January 11, 1991. Address reprint requests to Dr. Sandra Citi, Department of Cell Biology and Anatomy, Cornell University Medical College, 1300 York Ave., New York, NY 10021.
781
782
Citi et al
AJP April 1991, Vol. 138, No. 4
colon carcinoma cells.18 The studies documenting the pattern of expression in situ of tight junctions in cancer are few and inconclusive. Transmission electron microscopy and freeze-fracture analysis indicated reduction or loss of junctional strands in mouse mammary adenocarcinomas, rat skin squamous carcinomas,19 and bladder transitional cell carcinoma.20'21 However, in other solid epithelial tumors, including mammary and uterine adenocarcinomas, tight junctions appeared to be normal.13 In this paper, we describe an immunocytochemical approach to study tight junction distribution in human neoplasias, using an antiserum against cingulin. Cingulin is localized in the cytoplasmic domain of tight junctions8 and shows a characteristic distribution in polarized chicken epithelia.22 Purified cingulin is a heat-stable, elongated dimer,22 and partial cDNA sequencing reveals a a-helical, coiled-coil structure.23 Here we demonstrate that human epithelial tissues contain a protein with similar apparent molecular mass, intracellular localization, and pattern of tissue distribution as those of chicken cingulin. In addition, we show that cingulin expression is maintained in primary and metastatic colon adenocarcinomas, and that cingulin levels are increased in four colon adenocarcinomas. Finally cingulin was not detected in nonepithelial tissues and tumors.
Paul Matsudaira) was used at a dilution of 1:500 for immunoblotting.
Ge/ Electrophoresis Gel electrophoresis was performed using 5% to 20% polyacrylammide gradient, sodium dodecyl sulfate (SDS)-containing gels.26 Freshly excised biopsies or surgical material (50 to 200 mg wet weight) were homogenized in four volumes of ice-cold 0.3 mol/l (molar) NaCI, 25 mmol/l (millimolar) sodium phosphate, pH 7.0, 0.1 mmol/I NaN3, 1 mmol/I dithiothreitol, containing a mixture of protease inhibitors27 and mixed 1:1 with SDS sample buffer, or directly solubilized in sample buffer. The samples were incubated at 1000C for 3 minutes and used immediately for electrophoresis or stored at - 20°C. Protein concentration was determined on duplicate samples of the homogenates.28 CaCo-2 and UVEC cells were collected by centrifugation, washed with phosphatebuffered saline (PBS) and solubilized in SDS sample buffer. Gels were calibrated with Sigma molecular weight markers (MW-SDS-BLUE), and stained with 0.1% Brilliant Blue G (Sigma Chemical Co., St. Louis, MO).
Immunoblotting Experiments Materials and Methods Tissues and Cells Samples of human normal and neoplastic tissues were obtained after biopsy (liver, gastric mucosa, and granulation tissue) or surgical resection. For colon adenocarcinomas, samples of tumor tissue (central and peripheral areas) and normal mucosal scrapes were taken from each specimen. Specimens were washed with cold saline and transported on ice to the laboratories for immediate processing. Grading and staging of colorectal neoplasms were performed on routinely processed tissue specimens.24'25 CaCo-2 and umbilical vein endothelial cells were a gift of Dr. Andre Le Bivic (Cornell University, Medical College) and Dr. Marina Ziche (University of Florence), respectively.
Antibodies Polyclonal rabbit antiserum against purified intestinal chicken cingulin8 was used at a dilution of 1:500 for immunostaining of frozen sections, and at a dilution of 1:2000 for immunoblotting. Anti-villin antiserum (gift of Dr.
Gels were transferred onto nitrocellulose (Schleicher & Schuell, 0.1 ,um pore size) for 12 hours at 40C at 0.18 Amp. Nitrocellulose sheets were incubated with anticingulin antiserum, followed by 1251-Protein A (0.1 ,uCi/ml)'o3or by alkaline-phosphatase (AP)-conjugated anti-rabbit antibodies (Promega, Madison, WI). Incubations were for 2 hours at room temperature (RT) or 16 hours at 40C. For autoradiograms, dried gels were exposed for 14 to 18 hours at - 70°C using Dupont-Cronex Quantalll intensifying screens (Dupont, Wilmington, DE) and Kodak XAR film (Eastman-Kodak, Rochester, NY). Blots incubated with AP anti-rabbit were developed with the bromochloroindolyl phosphate/nitroblue tetrazolium
substrate. For semiquantitative analysis of cingulin content, two loadings of at least three separate dilutions of tissue samples (containing between 25 and 300 p.g total protein per lane) were examined. Purified cingulin was run as a control (0.05 to 2 ,ug). Densitometry of autoradiograms was performed on a GS 300 densitometer using GS 360 software (Hoefer Scientific Instruments, San Francisco, CA). The integrated areas of the densitometry peaks corresponding to cingulin (Mr, 140 kd) were a linear function of total protein content within the range above. Comparison of cingulin content in normal versus cancer tissue was carried out on samples containing the same amount of total protein. Cingulin content of the cancer tissue was
Cingulin in Human Tissues and Neoplasias
783
AJP April 1991, Vol. 138, No. 4
expressed as percentage of that of the normal tissue from the same specimen, taken as 100%. To evaluate the contribution of epithelial cells in normal versus tumor samples, duplicate blots were stained with anti-villin antibodies followed by 1251-Protein A.
Immunohistochemical Procedures Cryostat sections (6 to 10 ,um) were cut from freshly fromaterial, dried on poly-(L-lysine)-coated slides, and permeabilized and fixed in acetone for 10 minutes at RT. Immunolabeling of sections with anti-cingulin antiserum was performed with the peroxidase-antiperoxidase method (Cambridge Research Laboratory, Cambridge, MA) following the protocol recommended by the manufacturers. The sections were counterstained in hematoxylin and mounted in Permount. Negative controls included substitution of the primary antibody by a similar antibody of the same species and subtype, or substitution of the primary antibody by PBS alone. Positive controls consisted of frozen sections of chicken intestine. zen
Results
Cingulin Expression in Normal Colonic Epithelium and Other Tissues In chicken epithelia, the anti-cingulin antibodies allow the specific localization of tight junctions in frozen sections of tissues and in cultured cells.22 When cryostat sections of normal human colonic mucosa were examined by the immunoperoxidase technique with anti-cingulin antiserum, clear labeling was observed along the apical membrane of polarized crypt and surface epithelial cells (Figure 1 A and B). In longitudinally cut sections, the immunostaining typically appeared as a thin bar across the terminal web of each columnar absorptive cell, generating a linear pattern along the crypt and intestinal lumen (Figure 1). When cell surfaces were viewed from the top, the staining was distributed clearly along peripheral belts in the areas of contact between cells, producing the characteristic honeycomblike network (Figure 1A). Labeling also was noted in the areas of contact between absorptive and goblet cells, confirming previous electron microscopic observations of chicken intestine.22 No significant reactivity was detected in the cytoplasm, the basolateral membrane domain of the epithelial cells, or in the connective tissue. A junctional pattern of labeling was observed along the borders of epithelial cells in cryostat sections of human liver, stomach, and small intestine (Table 1).
To characterize further the expression of cingulin in normal human tissues, we carried out sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting (Figure 2). In chicken tissues, anticingulin antibodies label two polypeptides of Mr 108 and 140 kd, the lower molecular mass species probably being derived from proteolysis of cingulin-1 40.8 Purified cingulin migrates with an apparent Mr of 108 kd.8 When crude samples of human colonic cells were analyzed, a polypeptide of Mr 140 kd was specifically labeled (Figure 2A, lane 1). The distribution and electrophoretic mobility of cingulin from mucosal samples of patients affected by inflammatory bowel disease was similar to normal tissue (Table 1). When samples from stomach mucosa, liver, and small intestine were analyzed, the Mr 140 kd and the Mr 108 kd cingulin polypeptides were specifically labeled (Figure 2, Table 1). When homogenates of human liver were subjected to heat treatment, the antigen recognized by the anti-cingulin antiserum remained soluble (not shown), as shown for chicken cingulin.22 No labeling of the cingulin polypeptides was detected in samples of epidermis, endothelial cells, granulation tissue, and skeletal muscle (Figure 2B, Table 1), indicating that immunoreactive cingulin is not expressed by nonepithelial tissues, endothelial cells and stratified, keratinizing epithelia.
Cingulin Distribution in Human Colon Adenomas and Adenocarcinomas The expression of cingulin in two villous adenomas with mild to moderate dysplasia and 12 adenocarcinomas with varying degrees of differentiation was studied using the immunoperoxidase technique. The localization of cingulin in these tumors was overall similar to that of normal colon tissue, in the peripheral junctional region of the cell. In the adenomas, the papillae were lined by a tightly packed layer of polarized cells. When these cells were cut longitudinally, they exhibited positive cingulin staining along their junctional boundaries, as in normal tissue, generating a line bordering the surface of the papillae (Figure 3a). In the glands of well-differentiated carcinomas, cell polarity was well preserved, based on conventional hystologic examination. In these polarized cells, cingulin labeling was distributed along the apical borders of the cells lining the glandular lumina of the tumor (Figure 3b). Close examination of these sections showed that sometimes cingulin staining appeared discontinuous, probably because of the presence of mucus-secreting cells, and because the orientation of cells prevented the plane of section from including all apical surfaces of the polarized cells. Similar labeling was observed in the glandular
784 Citi et al AJP il 1991, Vol. 138, No. 4
Figure 1. Cingulin localization in normal human epithelium. 7he surface epithelium (A and A') and cryptae (B and B') are shown. A and B: PAP- hematoxylin with anti-cingulin antiserum; A' and B': control sections (no primary antibody). Asterisks indicate the lumina of the cayptae. Inflammatory cells show positivity because of endogeneous peroxidase activity, which was not blocked. Arrowheads indicate the polygonal pattern of cingulin labeling obserued when cells are viewed en face (original magnification, x 112).
structures of moderately differentiated adenocarcinomas, where glandular lumina often appeared more irregular in their shape and distribution (Figure 3c). No labeling was observed in the basolateral domains of the cells or in the stroma. In poorly differentiated adenocarcinomas, most of the tumor is formed by solid clumps of malignant cells, but small glandular lumina sometimes were seen. In the glands, although most of the cells did not display a clear cell polarity, positive reaction was observed on the surface of the cells bordering the lumina (Figure 3d). In some cases, even when glandular spaces were not recogniz-
able in tumor cell aggregates by conventional histologic examination, it was possible to detect positive labeling distributed in irregular patterns, eg, oval, roundish, or elongated shapes (Figure 3e). Such an array of cingulincontaining structures might reflect a complex architectural pattern of the tumor, with small and compressed glandular lumina scattered in the apparently solid tissue (also see Discussion). Furthermore in poorly differentiated adenocarcinomas immunostaining was observed at the interface between islets of solid tumor tissue and the surrounding neoplastic stroma (Figure 3d, f), presumably associated with the peripheral membrane of the neoplas-
Cingulin in Human Tissues and Neoplasias
785 AJP Apl 1991, Vol. 138, No. 4
Table 1. Digstribution of Cingulin (Mr 108 and 140 kd) in Human Tissues and Cultured CellLines WB IC Colonic epithelium Normal (14) Ulcerative colitis (2) Crohn's disease (2) Villous adenoma (2) Primary adenocarcinoma (12) Metastatic adenocarcinoma (1) CaCo2 cell line Small intestine mucosa (3) Liver (2) Gastric mucosa (3) Epidermis (1) Pectoralis muscle (2) Granulation tissue (2) Umbilical vein endothelial cells Neurinoma (1)
Rhabdomyosarcoma (1) Metastatic squamous carcinoma (uterine cervix) (1)
+ + + + + + + +
+ + -
A
+ + + + + + + + + + ND ND ND
-
-
ND ND ND
I
Tissues were analyzed by immunoblotting (WB) and/or immunocytochemistry (IC) using rabbit anti-cingulin antiserum, as described in Materials and Methods. (+) or (-) indicate positive or negative immunoreactivity. ND, Not determined. The number of cases examined is shown in parentheses.
F
-
tic epithelial cells, which exhibited little or no polarity. This localization is unusual because tight junctions are typically absent from the domains of epithelial cells facing the connective tissue in situ. The staining pattern of normal colonic mucosa adjacent to carcinomas (so-called transitional mucosa) was identical to that of normal mucosa far away from the tumor.
Immunoblotting Analysis of Cingulin from Human Neoplasias
...
4' 5' t7'8' 1 40- -': .108w-,-. i'.....:.'.. ...
:.
1.
'.
-
:,
-.1.
L '.'
.
9'10'
B
.,
C
normi
-H C
,0I
I
100
0 %
When samples of colon adenocarcinoma tissues and of the cell line CaCo2 were analyzed by SDS-PAGE and immunoblotting, a polypeptide of M, 140 kd was specifically labeled (Figure 2A), similarly to what shown in normal colonic tissue. In addition, positive reaction was observed with a metastasis from a colon adenocarcinoma (Figure 2B, lane 9'). Junctions morphologically similar to tight junctions have been described in nonepithelial tumors,31 thus we analyzed by immunoblot samples from a neurinoma and a rhabdomyosarcoma, as well as a metastatic squamous carcinoma. No specific reactivity was observed in these tissues (Figure 2B, lanes 4', 6', and 10'), suggesting that cingulin expression may be limited to gland-forming epithelial neoplasias. To evaluate the level of expression of cingulin in colon adenocarcinomas, we carried out a semiquantitative study by immunolabeling blots with anti-cingulin antiserum and [1251]-Protein A, and densitometry-scan the corresponding autoradiograms (Figure 20). The contribution
i
.1 ...
200
cing;ul in content
300
Figure 2. A and B: Immunodetection of cingulin in human cells and tissues. Lanes 1 to 3: SDS-PAGE (5% to 20% polyaoylamide gradient gels). Lanes 1' to 10': immunobloting analysi with anticingulin ansrum,followedby '25[I-Protein-A (1' to 3') oralkalinepbospate-conjugated anti-rabbit antbody (4' to 10') (only the representative area of the blot is shown in lanes 4' to 10'). Lanes 1-1': puifled chicbken cingulin (Mr 108 kd), at 1 mg/ml (loadings were2 p1 and 0.2 p1, respectively). Lanes 2-2': Caco-2 cells. Lanes 3-3': normal colon tissue. Lane 4': metastatic squamous carcinoma (Drimary localization: uterine cervix). Lane 5': gastrc mucosa. Lane 6': granulation tisue (base ofpeptic ulcer). Lane 7': epidermis. Lane 8': neurinoma. Lane9': metaatic adenocarcinoma (yrima,y localization: colon). Lane 10': rhabdomyosarcoma. For tissue sarples, 75 to 150 pg of total protein was loaded The apparent molecular weights of punfied cingulin (MR 108 kd) and cingulin in tissue samples (Mr 108 and 140 kd) are indicated by 108 and 140. C: Analysis of the cingulin content of colon adenocarcinomas by immunoblowang and scanning densitomeiy. The ratios betveen the integrated areas of the cingulin (Mr 140kdpolypeptide)peaksin tumortissue versus owrmal tissue (taken as 1X00%) are shown. Case 1:DukesB adenocarcinoma of the sigma (staged as Gl); case 2: Dukes B adenocarcinoma of the rectum (Gl); case 3: Dukes B adenocarcinoma of the rigbt colon (G3); case 4: Dukes A carcinoma of the sigma (Gi). Enror bars represent standard error on three or four separate measuements. For each measurment, the values were calulated from two or three dilutions ofeach sample. Valuesfor cases 2 and 3 represents the mean of two measurements.
786 Citi et al AJP APril1991, Vol. 138, No. 4
rs
rW.
....
Iup
VW
_
6,-._
s
..
"WI
Figure 3. Cingulin localization in human adenoma and adenocarcinomas. Agalley ofselected micrographsshowing A: villousadenoma; B: well-differentiated adenocarcinoma; C, D, E, and F: moderately to poorly differentiated adenocarcinomas. A ruptured lumen isseen on the left of the cellular aggregate. Arrows indicate cingulin labeling at the interface between tumor tissue and surrounding stroma. Micrographs E and F were taken using a Blue Kodak Wratten gelatin filter 80a to enhance clarity of labeling. The punctate labeling visible in E is due to necrotic cells. 7be tumors were normally vascularized and anti-cingulin antibodies did not react with blood vessel endothelia (original magnifwation, X90 (A) and x112 [B, C, D, E, F]).
of epithelial cells in normal and tumor samples was similar, based on morphologic analysis and expression of the protein villin by immunoblotting (not shown). The cingulin content of the neoplastic mucosa from four distinct cases of adenocarcinomas was higher than that of the normal tissue. The levels of cingulin ranged from approximately 150% to approximately 230% of normal content, with no apparent correlation with the degree of differen-
tiation of the tumor, established by conventional hystologic examination.
Discussion In the present paper we show that epithelia of the human gastroenteric tract contain cingulin, a specific tight junc-
Cingulin in Human Tissues and Neoplasias
787
AJP April 1991, Vol. 138, No. 4
tion protein first identified in avian tissues. We also describe the localization of cingulin in normal and neoplastic colon tissues, providing the first immunocytochemical analysis of tight junctions in human colon adenocarcinomas, one of the major causes of death from cancer in developed countries.32 Finally we show that cingulin is absent from nonepithelial human tissues and neoplasias but is expressed in metastatic colon neoplasms and in inflammatory bowel disease. The notion that the antigen cross-reacting with the anti-cingulin antibodies is human cingulin is supported by a number of observations. Human brush borders share with avian brush borders all the major protein components.33 The junctional localization of cingulin in human colon (Figures 1 and 3), gastric, liver, and small intestinal epithelia, and on the colonic cancer cell line CaCo-2 (not shown), is identical to that described in native and cultured chicken epithelia.22 The human antigen shares with chicken cingulin its apparent molecular mass (Figure 3), and heat stability properties. In neoplastic cells that retained morphologic polarity and formed glands cingulin was localized in the apical, junctional belt, as in normal cells. In nests of solid tumor tissue, the distribution of cingulin indicated a glandlike pattern of arrangement of the cells, with small, deformed, and compressed lumina (Figure 3e), consistent with three-dimensional reconstruction studies of gastrointestinal epithelial tumors, showing that the major part of poorly differentiated adenocarcinomas may retain a porous network.34 In poorly differentiated adenocarcinomas, the anti-cingulin antiserum sometimes labeled cells at the interface between tumor cell aggregates and stroma, in areas that could not be cleary identified as apical or basolateral (Figure 3f). Such apparent heterogeneity of labeling may reflect subtle morphogenetic changes, similar to those observed in tight junctions in various lines of colon carcinoma in vitro.18,35 Immunoblot analysis showed that cingulin is expressed in metastatic colon adenocarcinomas, but is absent from nonepithelial tissues and neoplasias, and from a metastatic squamous carcinoma. Taken together, these data suggest that anticingulin antibodies may be an useful tool, along with antibodies to other cytoskeletal markers,36-38 in differentiating epithelial from nonepithelial neoplasias, classifying different types of epithelial neoplasias, and establishing the origin of metastatic carcinomas. A systematic analysis of cingulin expression in a wider range of human tumors will, however, be necessary to establish the general validity of this hypothesis. Because of the different techniques and tissues used, it is difficult to compare our observations with previous electron microscopy studies, suggesting attenuation or loss of tight junctions in some malignant epithelial tumors.13,15,19 Evaluation of junctions in tumors by the
freeze-fracture technique focuses on the ultrastructural integrity of the junctions (number and shape of fibrils), rather than their biochemical composition. Using the freeze-fracture technique, results can be affected by sampling problems, especially when the tissues are highly heterogeneous, or by the incomplete assembly of junctions, for example as a result of different functional states.39 In fact, it was suggested that in transitional cell carcinomas, defects in the freeze-fracture behavior of tight junctions could be related to the ineffective assembly of junctional components.40 The levels of the Mr kd cingulin polypeptide per total protein in four adenocarcinomas were found to be higher than in normal tissue (Figure 2B). Because normal tissues (mucosal scrapes) had a similar content in the epithelial protein villin as the tumor samples, it is unlikely that the increased cingulin content is due to variations in the epithelial/stromal cell ratio in the normal versus tumor samples. The increase in cingulin therefore could be due to a higher concentration of cellular protein, but also to a reduction in cell size and/or an increase in the number and extension of tight junctions. However, in a recent study describing supranormal levels of fodrin in human colon adenocarcinomas,41 an increase in fodrin content by immunoblotting correlated with increased pools of cellular fodrin, as quantitated by digital confocal microscopy. What could be the significance of cingulin increase in tumor tissues? As proposed for fodrin,41 it is possible that a higher content of cingulin may signal a reaction of the cell cytoskeleton to pathologic stress, and thus may be useful as a general marker of neoplastic activity. Little is known about the relationship between junction physiology and the expression of junctional proteins in neoplasia. Studies on the function of gap junctions in transformed cells indicate a correlation between transformed phenotype, metastatic potential, and loss of intercellular communication.12.42 On the other hand, the localization and amounts of the tight junction protein ZO-1 are identical in two strains of Madine-Darby canine kidney cells, which differ in transepithelial resistance.' Our results and the ultrastructural analysis of several types of cultured tumor cells18 indicate that loss of tight junctions or tight junctional protein(s) may be a relatively rare maturation defect in neoplastic epithelial cells. Alterations in tight junctions might affect intercellular adhesiveness and the barrier function of epithelia; however the hypothesis that such alterations may contribute to the invasiveness of tumors in vivo is largely speculative. In fact, junctional deficiencies have been described in tumors at the preinvasive stage14 and some metastatic tumors display apparently normal junctions.17 Interestingly, in cultured kidney cells, tumor promoters cause rapid changes in tight junctional permeability, suggesting that neoplastic process primarily may affect junction physiology.4445 Cul-
788
Citi et al
AP APil 1991, Vol. 138, No. 4
tured epithelial cells will be useful model systems to correlate changes in tight junction permeability induced by tumor promoters with the expression and function of junctional proteins.
Acknowledgments The authors thank Dr. John Kendrick-Jones and Dr. David Stone for helpful discussions and support.
References 1. Farquhar MG, Palade G: Junctional complexes in various epithelia. J Cell Biol 1963, 17:375-412 2. Staehelin LA: Structure and function of intercellular junctions. Int Rev Cytol 1974, 39:191-283 3. Gumbiner B: The structure, biochemistry and assembly of epithelial tight junctions. Am J Physiol 1987, 253:C749-
C758 4. Madara JL, Hecht G: Tight (occluding) junctions in cultured (and native) epithelial cells. In Malin KS, Valentich JD, eds. Functional Epithelial Cells in Culture. New York, Alan R. Liss, pp 131-163 5. van Meer G, Simons K: The function of tight junctions in maintaining differences in lipid composition between the apical and the basolateral cell surface domains of MDCK cells. EMBO J 1986, 5:1455-1464 6. Goodenough DA, Revel JP: A fine structural analysis of intercellular junctions in mouse liver. J Cell Biol 1970, 45:272-
290 7. Stevenson BR, Siliciano JD, Mooseker MS, Goodenough DA: Identification of ZO-1: A high molecular weight polypeptide associated with the tight junction (zonula occludens) in a variety of epithelia. J Cell Biol 1986, 103:755-766. 8. Citi S, Sabanay H, Jakes R, Geiger B, Kendrick-Jones J: Cingulin, a new peripheral component of tight junctions. Nature 1988, 333:272-276 9. Coman DR: Decreased mutual adhesiveness: A property of cells from squamous cell carcinomas. Cancer Res 1944;
4:625-629 10. Abercrombie M, Ambrose EJ: The surface properties of cancer cells. Cancer Res 1962; 22:525 11. Weinstein R, Zel G, Merk F: Quantitation of occludens, adherens, and nexus cell junctions in human tumors. In Schultz J, Block RE, eds. Membrane Transformations in Neoplasia. New York, Academic Press, 1974, pp 127-150 12. Loewenstein WR: Junctional intercellular communication and the control of growth. Biochim Biophys Acta 1979, 560:1-75. 13. Weinstein RS, Merk FB, Alroy J: The structure and functions of intercellular junctions in cancer. Adv Cancer Res 1976,
23:23-89 14. Weinstein RS, Pauli BU: Cell junctions and the biological behaviour of cancer. In Junctional Complexes of Epithelial
Cells. Ciba Found. Symp. 1987, 125. Chichester, Wiley, 1987, pp 240-260 15. Martinez-Palomo A, Braislovsky C, Bernhard W: Ultrastructural modifications of the cell surface and intercellular contacts of some transformed cell strains. Cancer Res 1969, 29:925-937 16. Pickett PB, Pitelka DR, Hamamoto ST, Misfeldt DS: Occluding junctions and cell behaviour in primary cultures of normal and neoplastic mammary gland cells. J Cell Biol 1975, 66:31 6-332 17. Pitelka DR, Hamamoto ST, Taggart BN: Epithelial cell junctions in primary and metastatic mammary tumors of mice. Cancer Res 1980; 40:1588-1599 18. Chantret I, Barbat A, Dussaulx E, Brattain MG, Zweibaum A: Epithelial polarity, villin expression, and enterocytic differentiation of cultured human colon carcinoma cells: A survey of twenty cell lines. Cancer Res 1988, 48:1936-1942 19. Martinez-Palomo A: Ultrastructural modifications of intercellular junctions in some epithelial tumors. Lab Invest 1970; 22:605-614 20. Fulker MJ, Cooper EH, Tanaka T: Proliferation and ultrastructure of papillary transitional cell carcinoma of the human bladder. Cancer 1971, 27:71-82 21. Merk FB, Pauli BU, Jacobs JB, Alroy A, Friedel GH, Weinstein RS: Malignant transformation of urinary bladder in humans and in FANFT-exposed Fischer rats. Ultrastructure of the major components of the permeability barrier. Cancer Res 1977, 37:2843-2853 22. Citi S, Sabanay H, Kendrick-Jones J, Geiger B: Cingulin: Characterization and localization. J Cell Sci 1989, 93:107122 23. Citi S, Kendrick-Jones J, Shore D: Cloning of cingulin cDNA: evidence for a tight junction protein with a coiled-coil structure. 30th Annual Meeting of the ASCB, 9-13 Dec. 1990, San Diego, CA, USA. J Cell Biol 1990,111 :409a 24. Dukes CE: The classification of cancer of the rectum. J Pathol Bacteriol 1932, 35:323-332 25. Dukes CE, Bussey HJR: The spread of rectal cancer and its effect on prognosis. Br J Cancer 1985, 12:309-320 26. Matsudaira PT, Burgess DR: SDS microslab linear gradient polyacrylamide gel electrophoresis. Anal Biochem 1978, 87:386-396 27. Citi S, Kendrick-Jones J: Regulation in vitro of brush border myosin by light chain phosphorylation. J Mol Biol 1986,
188:369-382 28. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ: Protein measurement with the Folin reagent. J Biol Chem 1985, 193:265-275 29. Burnette WN: Western blotting: Electrophoretic transfer of proteins from sodium dodecyl sulfate-polyacrylamide gels to unmodified nitrocellulose and radiographic detection with antibody and radioiodinated Protein A. Anal Biochem 1981,
112:195-203 30. Citi S, Kendrick-Jones J: Studies on the structure and conformation of brush border myosin using monoclonal antibodies. Eur J Biochem 1987, 165:315-325 31. Quinonez G, Simon GT: Cellular junctions in a spectrum of human malignant tumors. Ultrastr Pathol 1988, 12:389-405
Cingulin in Human Tissues and Neoplasias
789
AJP Apnil 1991, Vol. 138, No. 4
32. Muir C, Waterhouse J: Cancer incidence in five continents. V. IARC Scientific Publications, no. 88, 1988 33. Carboni JM, Howe CL, West AB, Barwick KW, Mooseker MS, Morrow JS: Characterization of intestinal brush border cytoskeletal proteins of normal and neoplastic human epithelial cells: A comparison with the avian brush border. Am J Pathol 1987,129:589-600 34. Takahashi T, Iwama N: Architectural pattern of gastric adenocarcinoma. A 3-dimensional reconstruction study. Virchows Arch (Pathol Anat) 1984, 403:127-134 35. Fukamachi H, Mizuno T, Kim YS: Gland formation in human colon cancer cells combined with foetal rat mesenchyme. J Cell Science 1987, 87:615-621 36. Moll R, Cowin P, Kapprell HP, Franke WW: Desmosomal proteins: new markers for identification and classification of tumors. Lab Invest 1986, 54:4-25 37. Robine S, Huet C, Moll R, Sahuquillo-Merino C, Coudrier E, Zweibaum A, Louvard D: Can villin be used to identify malignant and undifferentiated normal digestive epithelial cells? Proc Natl Acad Sci USA 1985, 82:8488-8492 38. West AB, Isaac CA, Carboni JM, Morrow JS, Mooseker MS, Barwick KW: Localization of villin, a cytoskeletal protein specific to microvilli, in human ileum and colon and in colonic neoplasms. Gastroenterology 1988, 94:343-352
39. Friend DS, Gilula NB: Variations in tight and gap junctions in mammalian tissues. J Cell Biol 1972, 53:758-776 40. Pauli BU, Weinstein RS: Correlations between cell surface protease activities and abnormalities in occludens junctions in rat bladder carcinoma in vitro. Cancer Res 1982, 42:2289-2297 41. Younes M, Harris AS, Morrow JS: Fodrin as a differentiation marker. Redistributions in colonic neoplasia. Am J Pathol 1989,135:1197-1212 42. Nicolson GL, Dulski KM, Trosko JE: Loss of intercellular junctional communication correlates with metastatic potential in mammary adenocarcinoma cells. Proc Natl Acad Sci USA 1988, 85:473-476 43. Stevenson BR, Anderson JM, Goodenough DA, Mooseker MS: Tight junction structure and ZO-1 content are identical in two strains of Madin-Darby canine kidney cells which differ in transepithelial resistance. J Cell Biol 1988, 107:24012408 44. Ojakian GK: Tumor promoter-induced changes in the permeability of epithelial tight junctions. Cell 1981, 23:95-103 45. Mullin JM, O'Brien TG: Effects of tumor promoters on LLCPK, renal epithelial tight junctions and transepithelial fluxes. Am J Physiol 1986, 251 :C597-C602
Rapid Communication Cingulin, a Specific Protein Component of Tight Junctions, Is Expressed in Normal and Neoplastic Human Epithelial Tissues Sandra Citi,*tt Andrea Amorosi,t Flavia Franconi, Alberto Giotti, Giancarlo Zampit From the Department of Cell Biology and Anatomy, * Cornell University Medical College, New York, New York; and Istituto di Anatomia Patologicat and Dipartimento di
Farmacologia Preclinica e Clinica, Universita' di Firenze, Firenze, Italy
Cingulin is a 140-kd protein localized on the cytoplasmic face of avian tightjunctions. The expression of cingulin in human normal and neoplastic colonic tissue has been investigated with an antiserum against chicken cingulin. Human cingulin shares its apparent molecular mass and localization with avian cingulin In normal colonic epithelium, villous adenomas, and differentiated adenocarcinomas cingulin staining is observed in the junctional region of the polarized cells lining the surface, the crypts, and the glandular lumina In poorly differentiated adenocarcinomas, labeling also is observed at the interface between cancer tissue and stromag or in clumps of malignant cells; forming a pattern that highlights the presence of small; compressed lumina The cingulin content offour adenocarcinomas, estimated by immunoblotting and densitometry, was higher than that of the normal tissue (150% to 230%). Cingulin was detected in a metastasis from a colon adenocarcinoma but not in nonepithelial tissues and neoplasias, suggesting that cingulin may be a useful marker in the characterization of colonic and probably other epithelial neoplasias. (Am J Pathol 1991, 138:781-789)
Tight junctions are typical specializations of the apicolateral domain of epithelial cell membranes and are absent from tissues of mesenchymal origin.1 In polarized epithe-
lial cells, they separate the endoluminal from the basolateral compartments of the extracellular space (for reviews, see Staehelin2 and Gumbiner3), playing a crucial role in the permeability barrier function of epithelial tissues.4 In addition, tight junctions may be involved in cel-cell adhesion1 and in maintaining cell polarity.5 Freezefractured tight junctions show a network of cross-linked fibrils and complementary grooves, corresponding to the sites of apparent membrane fusion, visible when junctions are analyzed by transmission electron microscopy.6 Recently progress was made in elucidating the biochemical composition of tight junctions, with the discovery of the tight junction specific proteins ZO-17 and cingulin.8 Despite the importance of tight junctions in the function of epithelia, little is known about their distribution and patterns of expression in human disease. The observation that in several solid tumors intercellular adhesiveness is decreased,9 and cel-cell contacts and electrotonic coupling are modified lead to the hypothesis that alterations in intercellular junctions may contribute to cell sloughing and metastasis.1012 In malignant tumors, altered morphology and function of desmosomes, adherens, and gap junctions have been described (reviewed in Weinstein13 and in Weinstein and Pauli14). On the other hand, studies of tight junctions have been hindered by the lack of biochemical markers, electron microscopy being, until recently, the only experimental tool to evaluate tight junction structure. For example, loss or attenuation of tight junctions was reported in virus-transformed cultured murine embryonic cells.15 In contrast, apparently normal tight junctions were described in cultured and metastatic mammary tumor cells16'17 and in several lines of human Supported by Consiglio Nazionale delle Ricerche (Contributo n.88.03127.26), Istituto Scientifico Roussel, MPI, National Institutes of Health (BRSG 2S07 RR05396-28). Sandra Citi is a Cornell Scholar. Accepted for publication January 11, 1991. Address reprint requests to Dr. Sandra Citi, Department of Cell Biology and Anatomy, Cornell University Medical College, 1300 York Ave., New York, NY 10021.
781
782
Citi et al
AJP April 1991, Vol. 138, No. 4
colon carcinoma cells.18 The studies documenting the pattern of expression in situ of tight junctions in cancer are few and inconclusive. Transmission electron microscopy and freeze-fracture analysis indicated reduction or loss of junctional strands in mouse mammary adenocarcinomas, rat skin squamous carcinomas,19 and bladder transitional cell carcinoma.20'21 However, in other solid epithelial tumors, including mammary and uterine adenocarcinomas, tight junctions appeared to be normal.13 In this paper, we describe an immunocytochemical approach to study tight junction distribution in human neoplasias, using an antiserum against cingulin. Cingulin is localized in the cytoplasmic domain of tight junctions8 and shows a characteristic distribution in polarized chicken epithelia.22 Purified cingulin is a heat-stable, elongated dimer,22 and partial cDNA sequencing reveals a a-helical, coiled-coil structure.23 Here we demonstrate that human epithelial tissues contain a protein with similar apparent molecular mass, intracellular localization, and pattern of tissue distribution as those of chicken cingulin. In addition, we show that cingulin expression is maintained in primary and metastatic colon adenocarcinomas, and that cingulin levels are increased in four colon adenocarcinomas. Finally cingulin was not detected in nonepithelial tissues and tumors.
Paul Matsudaira) was used at a dilution of 1:500 for immunoblotting.
Ge/ Electrophoresis Gel electrophoresis was performed using 5% to 20% polyacrylammide gradient, sodium dodecyl sulfate (SDS)-containing gels.26 Freshly excised biopsies or surgical material (50 to 200 mg wet weight) were homogenized in four volumes of ice-cold 0.3 mol/l (molar) NaCI, 25 mmol/l (millimolar) sodium phosphate, pH 7.0, 0.1 mmol/I NaN3, 1 mmol/I dithiothreitol, containing a mixture of protease inhibitors27 and mixed 1:1 with SDS sample buffer, or directly solubilized in sample buffer. The samples were incubated at 1000C for 3 minutes and used immediately for electrophoresis or stored at - 20°C. Protein concentration was determined on duplicate samples of the homogenates.28 CaCo-2 and UVEC cells were collected by centrifugation, washed with phosphatebuffered saline (PBS) and solubilized in SDS sample buffer. Gels were calibrated with Sigma molecular weight markers (MW-SDS-BLUE), and stained with 0.1% Brilliant Blue G (Sigma Chemical Co., St. Louis, MO).
Immunoblotting Experiments Materials and Methods Tissues and Cells Samples of human normal and neoplastic tissues were obtained after biopsy (liver, gastric mucosa, and granulation tissue) or surgical resection. For colon adenocarcinomas, samples of tumor tissue (central and peripheral areas) and normal mucosal scrapes were taken from each specimen. Specimens were washed with cold saline and transported on ice to the laboratories for immediate processing. Grading and staging of colorectal neoplasms were performed on routinely processed tissue specimens.24'25 CaCo-2 and umbilical vein endothelial cells were a gift of Dr. Andre Le Bivic (Cornell University, Medical College) and Dr. Marina Ziche (University of Florence), respectively.
Antibodies Polyclonal rabbit antiserum against purified intestinal chicken cingulin8 was used at a dilution of 1:500 for immunostaining of frozen sections, and at a dilution of 1:2000 for immunoblotting. Anti-villin antiserum (gift of Dr.
Gels were transferred onto nitrocellulose (Schleicher & Schuell, 0.1 ,um pore size) for 12 hours at 40C at 0.18 Amp. Nitrocellulose sheets were incubated with anticingulin antiserum, followed by 1251-Protein A (0.1 ,uCi/ml)'o3or by alkaline-phosphatase (AP)-conjugated anti-rabbit antibodies (Promega, Madison, WI). Incubations were for 2 hours at room temperature (RT) or 16 hours at 40C. For autoradiograms, dried gels were exposed for 14 to 18 hours at - 70°C using Dupont-Cronex Quantalll intensifying screens (Dupont, Wilmington, DE) and Kodak XAR film (Eastman-Kodak, Rochester, NY). Blots incubated with AP anti-rabbit were developed with the bromochloroindolyl phosphate/nitroblue tetrazolium
substrate. For semiquantitative analysis of cingulin content, two loadings of at least three separate dilutions of tissue samples (containing between 25 and 300 p.g total protein per lane) were examined. Purified cingulin was run as a control (0.05 to 2 ,ug). Densitometry of autoradiograms was performed on a GS 300 densitometer using GS 360 software (Hoefer Scientific Instruments, San Francisco, CA). The integrated areas of the densitometry peaks corresponding to cingulin (Mr, 140 kd) were a linear function of total protein content within the range above. Comparison of cingulin content in normal versus cancer tissue was carried out on samples containing the same amount of total protein. Cingulin content of the cancer tissue was
Cingulin in Human Tissues and Neoplasias
783
AJP April 1991, Vol. 138, No. 4
expressed as percentage of that of the normal tissue from the same specimen, taken as 100%. To evaluate the contribution of epithelial cells in normal versus tumor samples, duplicate blots were stained with anti-villin antibodies followed by 1251-Protein A.
Immunohistochemical Procedures Cryostat sections (6 to 10 ,um) were cut from freshly fromaterial, dried on poly-(L-lysine)-coated slides, and permeabilized and fixed in acetone for 10 minutes at RT. Immunolabeling of sections with anti-cingulin antiserum was performed with the peroxidase-antiperoxidase method (Cambridge Research Laboratory, Cambridge, MA) following the protocol recommended by the manufacturers. The sections were counterstained in hematoxylin and mounted in Permount. Negative controls included substitution of the primary antibody by a similar antibody of the same species and subtype, or substitution of the primary antibody by PBS alone. Positive controls consisted of frozen sections of chicken intestine. zen
Results
Cingulin Expression in Normal Colonic Epithelium and Other Tissues In chicken epithelia, the anti-cingulin antibodies allow the specific localization of tight junctions in frozen sections of tissues and in cultured cells.22 When cryostat sections of normal human colonic mucosa were examined by the immunoperoxidase technique with anti-cingulin antiserum, clear labeling was observed along the apical membrane of polarized crypt and surface epithelial cells (Figure 1 A and B). In longitudinally cut sections, the immunostaining typically appeared as a thin bar across the terminal web of each columnar absorptive cell, generating a linear pattern along the crypt and intestinal lumen (Figure 1). When cell surfaces were viewed from the top, the staining was distributed clearly along peripheral belts in the areas of contact between cells, producing the characteristic honeycomblike network (Figure 1A). Labeling also was noted in the areas of contact between absorptive and goblet cells, confirming previous electron microscopic observations of chicken intestine.22 No significant reactivity was detected in the cytoplasm, the basolateral membrane domain of the epithelial cells, or in the connective tissue. A junctional pattern of labeling was observed along the borders of epithelial cells in cryostat sections of human liver, stomach, and small intestine (Table 1).
To characterize further the expression of cingulin in normal human tissues, we carried out sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting (Figure 2). In chicken tissues, anticingulin antibodies label two polypeptides of Mr 108 and 140 kd, the lower molecular mass species probably being derived from proteolysis of cingulin-1 40.8 Purified cingulin migrates with an apparent Mr of 108 kd.8 When crude samples of human colonic cells were analyzed, a polypeptide of Mr 140 kd was specifically labeled (Figure 2A, lane 1). The distribution and electrophoretic mobility of cingulin from mucosal samples of patients affected by inflammatory bowel disease was similar to normal tissue (Table 1). When samples from stomach mucosa, liver, and small intestine were analyzed, the Mr 140 kd and the Mr 108 kd cingulin polypeptides were specifically labeled (Figure 2, Table 1). When homogenates of human liver were subjected to heat treatment, the antigen recognized by the anti-cingulin antiserum remained soluble (not shown), as shown for chicken cingulin.22 No labeling of the cingulin polypeptides was detected in samples of epidermis, endothelial cells, granulation tissue, and skeletal muscle (Figure 2B, Table 1), indicating that immunoreactive cingulin is not expressed by nonepithelial tissues, endothelial cells and stratified, keratinizing epithelia.
Cingulin Distribution in Human Colon Adenomas and Adenocarcinomas The expression of cingulin in two villous adenomas with mild to moderate dysplasia and 12 adenocarcinomas with varying degrees of differentiation was studied using the immunoperoxidase technique. The localization of cingulin in these tumors was overall similar to that of normal colon tissue, in the peripheral junctional region of the cell. In the adenomas, the papillae were lined by a tightly packed layer of polarized cells. When these cells were cut longitudinally, they exhibited positive cingulin staining along their junctional boundaries, as in normal tissue, generating a line bordering the surface of the papillae (Figure 3a). In the glands of well-differentiated carcinomas, cell polarity was well preserved, based on conventional hystologic examination. In these polarized cells, cingulin labeling was distributed along the apical borders of the cells lining the glandular lumina of the tumor (Figure 3b). Close examination of these sections showed that sometimes cingulin staining appeared discontinuous, probably because of the presence of mucus-secreting cells, and because the orientation of cells prevented the plane of section from including all apical surfaces of the polarized cells. Similar labeling was observed in the glandular
784 Citi et al AJP il 1991, Vol. 138, No. 4
Figure 1. Cingulin localization in normal human epithelium. 7he surface epithelium (A and A') and cryptae (B and B') are shown. A and B: PAP- hematoxylin with anti-cingulin antiserum; A' and B': control sections (no primary antibody). Asterisks indicate the lumina of the cayptae. Inflammatory cells show positivity because of endogeneous peroxidase activity, which was not blocked. Arrowheads indicate the polygonal pattern of cingulin labeling obserued when cells are viewed en face (original magnification, x 112).
structures of moderately differentiated adenocarcinomas, where glandular lumina often appeared more irregular in their shape and distribution (Figure 3c). No labeling was observed in the basolateral domains of the cells or in the stroma. In poorly differentiated adenocarcinomas, most of the tumor is formed by solid clumps of malignant cells, but small glandular lumina sometimes were seen. In the glands, although most of the cells did not display a clear cell polarity, positive reaction was observed on the surface of the cells bordering the lumina (Figure 3d). In some cases, even when glandular spaces were not recogniz-
able in tumor cell aggregates by conventional histologic examination, it was possible to detect positive labeling distributed in irregular patterns, eg, oval, roundish, or elongated shapes (Figure 3e). Such an array of cingulincontaining structures might reflect a complex architectural pattern of the tumor, with small and compressed glandular lumina scattered in the apparently solid tissue (also see Discussion). Furthermore in poorly differentiated adenocarcinomas immunostaining was observed at the interface between islets of solid tumor tissue and the surrounding neoplastic stroma (Figure 3d, f), presumably associated with the peripheral membrane of the neoplas-
Cingulin in Human Tissues and Neoplasias
785 AJP Apl 1991, Vol. 138, No. 4
Table 1. Digstribution of Cingulin (Mr 108 and 140 kd) in Human Tissues and Cultured CellLines WB IC Colonic epithelium Normal (14) Ulcerative colitis (2) Crohn's disease (2) Villous adenoma (2) Primary adenocarcinoma (12) Metastatic adenocarcinoma (1) CaCo2 cell line Small intestine mucosa (3) Liver (2) Gastric mucosa (3) Epidermis (1) Pectoralis muscle (2) Granulation tissue (2) Umbilical vein endothelial cells Neurinoma (1)
Rhabdomyosarcoma (1) Metastatic squamous carcinoma (uterine cervix) (1)
+ + + + + + + +
+ + -
A
+ + + + + + + + + + ND ND ND
-
-
ND ND ND
I
Tissues were analyzed by immunoblotting (WB) and/or immunocytochemistry (IC) using rabbit anti-cingulin antiserum, as described in Materials and Methods. (+) or (-) indicate positive or negative immunoreactivity. ND, Not determined. The number of cases examined is shown in parentheses.
F
-
tic epithelial cells, which exhibited little or no polarity. This localization is unusual because tight junctions are typically absent from the domains of epithelial cells facing the connective tissue in situ. The staining pattern of normal colonic mucosa adjacent to carcinomas (so-called transitional mucosa) was identical to that of normal mucosa far away from the tumor.
Immunoblotting Analysis of Cingulin from Human Neoplasias
...
4' 5' t7'8' 1 40- -': .108w-,-. i'.....:.'.. ...
:.
1.
'.
-
:,
-.1.
L '.'
.
9'10'
B
.,
C
normi
-H C
,0I
I
100
0 %
When samples of colon adenocarcinoma tissues and of the cell line CaCo2 were analyzed by SDS-PAGE and immunoblotting, a polypeptide of M, 140 kd was specifically labeled (Figure 2A), similarly to what shown in normal colonic tissue. In addition, positive reaction was observed with a metastasis from a colon adenocarcinoma (Figure 2B, lane 9'). Junctions morphologically similar to tight junctions have been described in nonepithelial tumors,31 thus we analyzed by immunoblot samples from a neurinoma and a rhabdomyosarcoma, as well as a metastatic squamous carcinoma. No specific reactivity was observed in these tissues (Figure 2B, lanes 4', 6', and 10'), suggesting that cingulin expression may be limited to gland-forming epithelial neoplasias. To evaluate the level of expression of cingulin in colon adenocarcinomas, we carried out a semiquantitative study by immunolabeling blots with anti-cingulin antiserum and [1251]-Protein A, and densitometry-scan the corresponding autoradiograms (Figure 20). The contribution
i
.1 ...
200
cing;ul in content
300
Figure 2. A and B: Immunodetection of cingulin in human cells and tissues. Lanes 1 to 3: SDS-PAGE (5% to 20% polyaoylamide gradient gels). Lanes 1' to 10': immunobloting analysi with anticingulin ansrum,followedby '25[I-Protein-A (1' to 3') oralkalinepbospate-conjugated anti-rabbit antbody (4' to 10') (only the representative area of the blot is shown in lanes 4' to 10'). Lanes 1-1': puifled chicbken cingulin (Mr 108 kd), at 1 mg/ml (loadings were2 p1 and 0.2 p1, respectively). Lanes 2-2': Caco-2 cells. Lanes 3-3': normal colon tissue. Lane 4': metastatic squamous carcinoma (Drimary localization: uterine cervix). Lane 5': gastrc mucosa. Lane 6': granulation tisue (base ofpeptic ulcer). Lane 7': epidermis. Lane 8': neurinoma. Lane9': metaatic adenocarcinoma (yrima,y localization: colon). Lane 10': rhabdomyosarcoma. For tissue sarples, 75 to 150 pg of total protein was loaded The apparent molecular weights of punfied cingulin (MR 108 kd) and cingulin in tissue samples (Mr 108 and 140 kd) are indicated by 108 and 140. C: Analysis of the cingulin content of colon adenocarcinomas by immunoblowang and scanning densitomeiy. The ratios betveen the integrated areas of the cingulin (Mr 140kdpolypeptide)peaksin tumortissue versus owrmal tissue (taken as 1X00%) are shown. Case 1:DukesB adenocarcinoma of the sigma (staged as Gl); case 2: Dukes B adenocarcinoma of the rectum (Gl); case 3: Dukes B adenocarcinoma of the rigbt colon (G3); case 4: Dukes A carcinoma of the sigma (Gi). Enror bars represent standard error on three or four separate measuements. For each measurment, the values were calulated from two or three dilutions ofeach sample. Valuesfor cases 2 and 3 represents the mean of two measurements.
786 Citi et al AJP APril1991, Vol. 138, No. 4
rs
rW.
....
Iup
VW
_
6,-._
s
..
"WI
Figure 3. Cingulin localization in human adenoma and adenocarcinomas. Agalley ofselected micrographsshowing A: villousadenoma; B: well-differentiated adenocarcinoma; C, D, E, and F: moderately to poorly differentiated adenocarcinomas. A ruptured lumen isseen on the left of the cellular aggregate. Arrows indicate cingulin labeling at the interface between tumor tissue and surrounding stroma. Micrographs E and F were taken using a Blue Kodak Wratten gelatin filter 80a to enhance clarity of labeling. The punctate labeling visible in E is due to necrotic cells. 7be tumors were normally vascularized and anti-cingulin antibodies did not react with blood vessel endothelia (original magnifwation, X90 (A) and x112 [B, C, D, E, F]).
of epithelial cells in normal and tumor samples was similar, based on morphologic analysis and expression of the protein villin by immunoblotting (not shown). The cingulin content of the neoplastic mucosa from four distinct cases of adenocarcinomas was higher than that of the normal tissue. The levels of cingulin ranged from approximately 150% to approximately 230% of normal content, with no apparent correlation with the degree of differen-
tiation of the tumor, established by conventional hystologic examination.
Discussion In the present paper we show that epithelia of the human gastroenteric tract contain cingulin, a specific tight junc-
Cingulin in Human Tissues and Neoplasias
787
AJP April 1991, Vol. 138, No. 4
tion protein first identified in avian tissues. We also describe the localization of cingulin in normal and neoplastic colon tissues, providing the first immunocytochemical analysis of tight junctions in human colon adenocarcinomas, one of the major causes of death from cancer in developed countries.32 Finally we show that cingulin is absent from nonepithelial human tissues and neoplasias but is expressed in metastatic colon neoplasms and in inflammatory bowel disease. The notion that the antigen cross-reacting with the anti-cingulin antibodies is human cingulin is supported by a number of observations. Human brush borders share with avian brush borders all the major protein components.33 The junctional localization of cingulin in human colon (Figures 1 and 3), gastric, liver, and small intestinal epithelia, and on the colonic cancer cell line CaCo-2 (not shown), is identical to that described in native and cultured chicken epithelia.22 The human antigen shares with chicken cingulin its apparent molecular mass (Figure 3), and heat stability properties. In neoplastic cells that retained morphologic polarity and formed glands cingulin was localized in the apical, junctional belt, as in normal cells. In nests of solid tumor tissue, the distribution of cingulin indicated a glandlike pattern of arrangement of the cells, with small, deformed, and compressed lumina (Figure 3e), consistent with three-dimensional reconstruction studies of gastrointestinal epithelial tumors, showing that the major part of poorly differentiated adenocarcinomas may retain a porous network.34 In poorly differentiated adenocarcinomas, the anti-cingulin antiserum sometimes labeled cells at the interface between tumor cell aggregates and stroma, in areas that could not be cleary identified as apical or basolateral (Figure 3f). Such apparent heterogeneity of labeling may reflect subtle morphogenetic changes, similar to those observed in tight junctions in various lines of colon carcinoma in vitro.18,35 Immunoblot analysis showed that cingulin is expressed in metastatic colon adenocarcinomas, but is absent from nonepithelial tissues and neoplasias, and from a metastatic squamous carcinoma. Taken together, these data suggest that anticingulin antibodies may be an useful tool, along with antibodies to other cytoskeletal markers,36-38 in differentiating epithelial from nonepithelial neoplasias, classifying different types of epithelial neoplasias, and establishing the origin of metastatic carcinomas. A systematic analysis of cingulin expression in a wider range of human tumors will, however, be necessary to establish the general validity of this hypothesis. Because of the different techniques and tissues used, it is difficult to compare our observations with previous electron microscopy studies, suggesting attenuation or loss of tight junctions in some malignant epithelial tumors.13,15,19 Evaluation of junctions in tumors by the
freeze-fracture technique focuses on the ultrastructural integrity of the junctions (number and shape of fibrils), rather than their biochemical composition. Using the freeze-fracture technique, results can be affected by sampling problems, especially when the tissues are highly heterogeneous, or by the incomplete assembly of junctions, for example as a result of different functional states.39 In fact, it was suggested that in transitional cell carcinomas, defects in the freeze-fracture behavior of tight junctions could be related to the ineffective assembly of junctional components.40 The levels of the Mr kd cingulin polypeptide per total protein in four adenocarcinomas were found to be higher than in normal tissue (Figure 2B). Because normal tissues (mucosal scrapes) had a similar content in the epithelial protein villin as the tumor samples, it is unlikely that the increased cingulin content is due to variations in the epithelial/stromal cell ratio in the normal versus tumor samples. The increase in cingulin therefore could be due to a higher concentration of cellular protein, but also to a reduction in cell size and/or an increase in the number and extension of tight junctions. However, in a recent study describing supranormal levels of fodrin in human colon adenocarcinomas,41 an increase in fodrin content by immunoblotting correlated with increased pools of cellular fodrin, as quantitated by digital confocal microscopy. What could be the significance of cingulin increase in tumor tissues? As proposed for fodrin,41 it is possible that a higher content of cingulin may signal a reaction of the cell cytoskeleton to pathologic stress, and thus may be useful as a general marker of neoplastic activity. Little is known about the relationship between junction physiology and the expression of junctional proteins in neoplasia. Studies on the function of gap junctions in transformed cells indicate a correlation between transformed phenotype, metastatic potential, and loss of intercellular communication.12.42 On the other hand, the localization and amounts of the tight junction protein ZO-1 are identical in two strains of Madine-Darby canine kidney cells, which differ in transepithelial resistance.' Our results and the ultrastructural analysis of several types of cultured tumor cells18 indicate that loss of tight junctions or tight junctional protein(s) may be a relatively rare maturation defect in neoplastic epithelial cells. Alterations in tight junctions might affect intercellular adhesiveness and the barrier function of epithelia; however the hypothesis that such alterations may contribute to the invasiveness of tumors in vivo is largely speculative. In fact, junctional deficiencies have been described in tumors at the preinvasive stage14 and some metastatic tumors display apparently normal junctions.17 Interestingly, in cultured kidney cells, tumor promoters cause rapid changes in tight junctional permeability, suggesting that neoplastic process primarily may affect junction physiology.4445 Cul-
788
Citi et al
AP APil 1991, Vol. 138, No. 4
tured epithelial cells will be useful model systems to correlate changes in tight junction permeability induced by tumor promoters with the expression and function of junctional proteins.
Acknowledgments The authors thank Dr. John Kendrick-Jones and Dr. David Stone for helpful discussions and support.
References 1. Farquhar MG, Palade G: Junctional complexes in various epithelia. J Cell Biol 1963, 17:375-412 2. Staehelin LA: Structure and function of intercellular junctions. Int Rev Cytol 1974, 39:191-283 3. Gumbiner B: The structure, biochemistry and assembly of epithelial tight junctions. Am J Physiol 1987, 253:C749-
C758 4. Madara JL, Hecht G: Tight (occluding) junctions in cultured (and native) epithelial cells. In Malin KS, Valentich JD, eds. Functional Epithelial Cells in Culture. New York, Alan R. Liss, pp 131-163 5. van Meer G, Simons K: The function of tight junctions in maintaining differences in lipid composition between the apical and the basolateral cell surface domains of MDCK cells. EMBO J 1986, 5:1455-1464 6. Goodenough DA, Revel JP: A fine structural analysis of intercellular junctions in mouse liver. J Cell Biol 1970, 45:272-
290 7. Stevenson BR, Siliciano JD, Mooseker MS, Goodenough DA: Identification of ZO-1: A high molecular weight polypeptide associated with the tight junction (zonula occludens) in a variety of epithelia. J Cell Biol 1986, 103:755-766. 8. Citi S, Sabanay H, Jakes R, Geiger B, Kendrick-Jones J: Cingulin, a new peripheral component of tight junctions. Nature 1988, 333:272-276 9. Coman DR: Decreased mutual adhesiveness: A property of cells from squamous cell carcinomas. Cancer Res 1944;
4:625-629 10. Abercrombie M, Ambrose EJ: The surface properties of cancer cells. Cancer Res 1962; 22:525 11. Weinstein R, Zel G, Merk F: Quantitation of occludens, adherens, and nexus cell junctions in human tumors. In Schultz J, Block RE, eds. Membrane Transformations in Neoplasia. New York, Academic Press, 1974, pp 127-150 12. Loewenstein WR: Junctional intercellular communication and the control of growth. Biochim Biophys Acta 1979, 560:1-75. 13. Weinstein RS, Merk FB, Alroy J: The structure and functions of intercellular junctions in cancer. Adv Cancer Res 1976,
23:23-89 14. Weinstein RS, Pauli BU: Cell junctions and the biological behaviour of cancer. In Junctional Complexes of Epithelial
Cells. Ciba Found. Symp. 1987, 125. Chichester, Wiley, 1987, pp 240-260 15. Martinez-Palomo A, Braislovsky C, Bernhard W: Ultrastructural modifications of the cell surface and intercellular contacts of some transformed cell strains. Cancer Res 1969, 29:925-937 16. Pickett PB, Pitelka DR, Hamamoto ST, Misfeldt DS: Occluding junctions and cell behaviour in primary cultures of normal and neoplastic mammary gland cells. J Cell Biol 1975, 66:31 6-332 17. Pitelka DR, Hamamoto ST, Taggart BN: Epithelial cell junctions in primary and metastatic mammary tumors of mice. Cancer Res 1980; 40:1588-1599 18. Chantret I, Barbat A, Dussaulx E, Brattain MG, Zweibaum A: Epithelial polarity, villin expression, and enterocytic differentiation of cultured human colon carcinoma cells: A survey of twenty cell lines. Cancer Res 1988, 48:1936-1942 19. Martinez-Palomo A: Ultrastructural modifications of intercellular junctions in some epithelial tumors. Lab Invest 1970; 22:605-614 20. Fulker MJ, Cooper EH, Tanaka T: Proliferation and ultrastructure of papillary transitional cell carcinoma of the human bladder. Cancer 1971, 27:71-82 21. Merk FB, Pauli BU, Jacobs JB, Alroy A, Friedel GH, Weinstein RS: Malignant transformation of urinary bladder in humans and in FANFT-exposed Fischer rats. Ultrastructure of the major components of the permeability barrier. Cancer Res 1977, 37:2843-2853 22. Citi S, Sabanay H, Kendrick-Jones J, Geiger B: Cingulin: Characterization and localization. J Cell Sci 1989, 93:107122 23. Citi S, Kendrick-Jones J, Shore D: Cloning of cingulin cDNA: evidence for a tight junction protein with a coiled-coil structure. 30th Annual Meeting of the ASCB, 9-13 Dec. 1990, San Diego, CA, USA. J Cell Biol 1990,111 :409a 24. Dukes CE: The classification of cancer of the rectum. J Pathol Bacteriol 1932, 35:323-332 25. Dukes CE, Bussey HJR: The spread of rectal cancer and its effect on prognosis. Br J Cancer 1985, 12:309-320 26. Matsudaira PT, Burgess DR: SDS microslab linear gradient polyacrylamide gel electrophoresis. Anal Biochem 1978, 87:386-396 27. Citi S, Kendrick-Jones J: Regulation in vitro of brush border myosin by light chain phosphorylation. J Mol Biol 1986,
188:369-382 28. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ: Protein measurement with the Folin reagent. J Biol Chem 1985, 193:265-275 29. Burnette WN: Western blotting: Electrophoretic transfer of proteins from sodium dodecyl sulfate-polyacrylamide gels to unmodified nitrocellulose and radiographic detection with antibody and radioiodinated Protein A. Anal Biochem 1981,
112:195-203 30. Citi S, Kendrick-Jones J: Studies on the structure and conformation of brush border myosin using monoclonal antibodies. Eur J Biochem 1987, 165:315-325 31. Quinonez G, Simon GT: Cellular junctions in a spectrum of human malignant tumors. Ultrastr Pathol 1988, 12:389-405
Cingulin in Human Tissues and Neoplasias
789
AJP Apnil 1991, Vol. 138, No. 4
32. Muir C, Waterhouse J: Cancer incidence in five continents. V. IARC Scientific Publications, no. 88, 1988 33. Carboni JM, Howe CL, West AB, Barwick KW, Mooseker MS, Morrow JS: Characterization of intestinal brush border cytoskeletal proteins of normal and neoplastic human epithelial cells: A comparison with the avian brush border. Am J Pathol 1987,129:589-600 34. Takahashi T, Iwama N: Architectural pattern of gastric adenocarcinoma. A 3-dimensional reconstruction study. Virchows Arch (Pathol Anat) 1984, 403:127-134 35. Fukamachi H, Mizuno T, Kim YS: Gland formation in human colon cancer cells combined with foetal rat mesenchyme. J Cell Science 1987, 87:615-621 36. Moll R, Cowin P, Kapprell HP, Franke WW: Desmosomal proteins: new markers for identification and classification of tumors. Lab Invest 1986, 54:4-25 37. Robine S, Huet C, Moll R, Sahuquillo-Merino C, Coudrier E, Zweibaum A, Louvard D: Can villin be used to identify malignant and undifferentiated normal digestive epithelial cells? Proc Natl Acad Sci USA 1985, 82:8488-8492 38. West AB, Isaac CA, Carboni JM, Morrow JS, Mooseker MS, Barwick KW: Localization of villin, a cytoskeletal protein specific to microvilli, in human ileum and colon and in colonic neoplasms. Gastroenterology 1988, 94:343-352
39. Friend DS, Gilula NB: Variations in tight and gap junctions in mammalian tissues. J Cell Biol 1972, 53:758-776 40. Pauli BU, Weinstein RS: Correlations between cell surface protease activities and abnormalities in occludens junctions in rat bladder carcinoma in vitro. Cancer Res 1982, 42:2289-2297 41. Younes M, Harris AS, Morrow JS: Fodrin as a differentiation marker. Redistributions in colonic neoplasia. Am J Pathol 1989,135:1197-1212 42. Nicolson GL, Dulski KM, Trosko JE: Loss of intercellular junctional communication correlates with metastatic potential in mammary adenocarcinoma cells. Proc Natl Acad Sci USA 1988, 85:473-476 43. Stevenson BR, Anderson JM, Goodenough DA, Mooseker MS: Tight junction structure and ZO-1 content are identical in two strains of Madin-Darby canine kidney cells which differ in transepithelial resistance. J Cell Biol 1988, 107:24012408 44. Ojakian GK: Tumor promoter-induced changes in the permeability of epithelial tight junctions. Cell 1981, 23:95-103 45. Mullin JM, O'Brien TG: Effects of tumor promoters on LLCPK, renal epithelial tight junctions and transepithelial fluxes. Am J Physiol 1986, 251 :C597-C602
