expression of muc1 and muc2 mucins in epithelial ovarian tumours

, . 183: 311–317 (1997)

EXPRESSION OF MUC1 AND MUC2 MUCINS INEPITHELIAL OVARIAN TUMOURS

1, . 2, . 3, . 3, 1, . 4 . c1*

1Department of Obstetrics and Gynaecology, University of Queensland, Royal Brisbane Hospital, Queensland, Australia2Department of Surgery, University of Queensland, Royal Brisbane Hospital, Queensland, Australia3Department of Pathology, University of Queensland, Royal Brisbane Hospital, Queensland, Australia

4Queensland Cancer Fund Laboratories, Queensland Institute of Medical Research, Herston, 4029 Queensland, Australia

SUMMARY

This is the first study to describe the association between expression of MUC1 and MUC2 mucins and prognosis in ovarian cancer.Paraffin sections of epithelial ovarian tumours (n=182: 29 benign, 21 low malignant potential, and 132 invasive tumours) were analysedimmunohistochemically for expression of MUC1 and MUC2 mucin core proteins. Most benign, low malignant potential, and invasivetumours showed high MUC1 expression in the cytoplasm. Low cytoplasmic expression of MUC1 was a predictor for good prognosis,particularly within stage III tumours. A minority of benign epithelial tumours, but most low malignant potential and invasivenon-mucinous tumours, showed high MUC1 expression on the cell membrane. High apical MUC1 reactivity was associated withnon-mucinous tumours. Low expression of MUC1 in the apical membrane was associated with early stage and good outcome for invasivetumours. Most benign and low malignant potential tumours, but only a minority of invasive tumours, showed MUC2 expression. MUC2was found in non-mucinous as well as in mucinous tumours. The presence of MUC2 was inversely associated with high tumour grade butwas not associated with altered survival. These results support experimental evidence that MUC1 influences the metastatic ability ofovarian cancer. ? 1997 John Wiley & Sons, Ltd.

J. Pathol. 183: 311–317, 1997.No. of Figures: 6. No. of Tables: 0. No. of References: 34.

KEY WORDS—MUC1; MUC2; mucins; ovarian carcinoma; prognosis; immunohistochemistry

INTRODUCTION

Mucins are a family of highly glycosylated, highmolecular weight (>200 kD) glycoproteins consisting ofa core protein with mainly O-linked carbohydrate side-chains.1 Eight distinct mucin genes have been cloned inhumans, including MUC1 to MUC7.2,3 These polymor-phic genes encode proteins which all contain a majorheavily O-glycosylated domain consisting of a variablenumber of tandem amino acid repeats (VNTR). Syn-thesis and secretion of mucin are common features ofnormal epithelial cells, and increased expression ofmucins has been shown in epithelial tumours.1,4–6 Inaddition, cancer-associated mucins show alterations instructure including aberrant or incomplete glycosylationresulting in accumulation of core oligosaccharidestructures and exposure of the core protein.3Most glandular epithelial cells express the membrane-

bound mucin encoded by the MUC1 gene at their apicalcell surface. Expression of MUC1 is increased consider-ably in the majority of carcinomas.4 Many monoclonalantibodies (MAbs) reacting with the MUC1 VNTRhave been reported, and because MUC1 is shed by

tumours, serum assays incorporating some of theseMAbs have been shown to be particularly useful inmonitoring patients with breast cancer and ovariancancer.7 The MAb BC2, one of the antibodies whichbind to an epitope on the VNTR of MUC1, recognizesan epitope that is highly expressed in breast cancer.8,9In addition, the BC2 antibody is reactive with mostepithelial ovarian carcinomas and appears to be a usefultool for the detection of micrometastases.10The MAb 4F1 recognizes a peptide epitope on the

VNTR of MUC2.11 The MUC2 mucin is a majorcomponent of normal colonic mucus. This protein isalso produced by colon and lung cancers,12 and isproduced in copious amounts by many organs in cysticfibrosis.12,13 Preliminary studies using anti-MUC2 coreprotein reactive MAbs and polyclonal antibodiesshowed the presence of MUC2 in tumours of thelung, breast, bladder, and ovary by immunohisto-chemistry.11,14–18 Previously we have shown that thepresence of MUC2 correlated with a shorter disease-freeinterval in breast cancer patients.17Mucin expression by ovarian cancers has not been

studied in detail. Ovarian mucinous tumours consist ofendocervical, mixed endocervical–intestinal, and intesti-nal cell types.19,20 It has been shown that the malignantmucinous tumours of the ovary more often expressintestinal mucin than benign and borderline mucinoustumours.20 In this study, the expression of MUC1detected with the MAb BC2 and MUC2 detected withthe MAb 4F1 was examined in epithelial ovarian

*Correspondence to: Dr Michael A. McGuckin, Departmentof Obstetrics and Gynaecology, Clinical Sciences Building, RoyalBrisbane Hospital, Herston, 4029 Queensland, Australia. E-mail:[email protected]

Contract grant sponsor: National Health and Medical ResearchCouncil, Australia.

CCC 0022–3417/97/110311–07 $17.50 Received 13 January 1997? 1997 John Wiley & Sons, Ltd. Accepted 12 May 1997

tumours and compared with clinicopathological par-ameters to evaluate the association between the expres-sion of these mucins and the biological characteristics ofthe tumours.

MATERIALS AND METHODS

Patients and clinical data collectionPatients for this study (n=178: 29 benign, 21 low

malignant potential, and 128 invasive tumours) wereselected randomly and retrospectively from cases ofepithelial ovarian tumours treated at the RoyalBrisbane, Mater and Wesley Hospitals, Brisbane. Thesepatients underwent maximal surgical debulking as partof their treatment. Clinical data for the patients werecollected from hospital records and treatment notesfrom the Queensland Radium Institute, and entered intoa database for analysis. Patients were aged 18–89 yearsat diagnosis and cancer patients were followed for aminimum of 3 years, unless they died of disease sooner.Two pathologists (RGW and MCC) reviewed all histo-logical material to confirm the histological diagnosisprior to patient recruitment. Disease staging was inaccordance with the International Federation of Gynae-cologists and Obstetricians (FIGO) staging system, andthe histological type and grade were assigned accordingto the WHO classification.

Immunohistochemistry

Representative paraffin blocks containing tumourfrom each case were sectioned at 4 ìm, affixed to slides,and dried overnight at 37)C. Sections were dewaxed inxylene and rehydrated through descending graded alco-hols to Tris-buffered saline, pH 7·4 (TBS). After washingthoroughly in three changes of TBS, sections weretreated for 10 min with 3 per cent (v/v) H2O2, 18 per cent(v/v) methanol in TBS to inhibit endogenous peroxidaseactivity. Non-specific antibody binding was reduced byincubating the sections with 4 per cent (w/v) commercialnon-fat skim milk powder in TBS for 15 min. Thesections were covered with 10 per cent (v/v) normal goatserum (Zymed Corporation, San Francisco, CA,U.S.A.) for 20 min. After decanting excess serum, sec-tions were incubated with BC2 mouse ascites (1/2000dilution) for 45 min or 4F1 mouse ascites (1/1000dilution) for 1 h at room temperature. The sections weresubsequently incubated with pre-diluted biotinylatedgoat anti-mouse immunoglobulin (Zymed) for 30 min,followed by pre-diluted streptavidin–horseradish peroxi-dase conjugate (Zymed) for 15 min. After washing,peroxidase activity was detected with a 5 min incubationin 0·05 per cent (w/v) 3,3*-diaminobenzidine in Trisphosphate buffer (33·3 m Tris in 10 m phosphatebuffer, 150 m NaCl, pH 7·6) as chromogen with 0·02per cent (v/v) H2O2 as substrate. The sections werecounterstained for 1–2 min with Harris’s haematoxylin,dehydrated, cleared in xylene, and mounted in DePeX.All sections were examined by four observers (YD,

MDW, MAM, and MCC) using a conference micro-scope. The sections were scored according to consensus

of the four observers regarding the overall proportion ofpositively stained tumour cells for both MAbs BC2 and4F1, and were categorized using the following division:0=negative; 1=1–10 per cent; 2=11–25 per cent; 3=26–50 per cent; 4=51–75 per cent; 5=76–100 per cent. Thesame categories were used for scoring the staining ofapical membrane, cytoplasm, circumferential mem-brane, and vacuoles. Staining intensity was scored on athree-point scale.

Statistical analyses

The results for antibody reactivity with regard to theproportion, intensity, and localization of staining withintumour cells were compared with pathological grade andclinical stage. Frequency distribution analysis wasapplied to determine the statistical significance of associ-ations between the expression of MUC1 and MUC2 andother variables. Both univariate (Kaplan–Meier) andmultivariate (Cox proportional hazards regression)analyses were used to assess survival differences.21 Allstatistical calculations were performed using SPSS forWindows version 6.0.

RESULTS

Expression of MUC1 in epithelial ovarian tumoursExamples of MUC1 expression, as detected by BC2,

in ovarian tumours are shown in Figs 1A and 1B. Forthe purposes of the statistical analysis, we have definedhigh MUC1 expression as greater than 25 per cent ofcells positive for BC2. Most benign (76 per cent), lowmalignant potential (95 per cent), and invasive (92 percent) epithelial ovarian tumours showed high MUC1reactivity. No correlation was found between the pro-portion of positive cells and tumour grade or stage. Inmalignant tumours, a trend was found associating highBC2 reactivity and non-mucinous histological subtypes(P=0·06). However, both non-mucinous and mucinousbenign and low malignant potential tumours showedcases of high MUC1 expression. Most high-expressingtumours showed MUC1 localized concurrently to theapical membranes, cytoplasm, vacuoles, and extra-cellular secretions. The frequency of simultaneousexpression at all these subcellular locations increasedfrom benign (23 per cent), through to low malignantpotential (40 per cent) and malignant (68 per cent)tumours.A majority of benign (55 per cent), low malignant

potential (67 per cent), and invasive (71 per cent)tumours had more than 25 per cent of cells showingMUC1 in the cytoplasm (P=0·03). The cytoplasmicpresence of MUC1 was not associated with histologicaltype (Fig. 2A), tumour grade (Fig. 2B), or stage (Fig.2C). Univariate survival analysis showed that a lowproportion (1–25 per cent) of cells with MUC1 accumu-lation in the cytoplasm was a predictor for good prog-nosis (P=0·02, Fig. 3A). However, when all parameterswere combined in a multivariate analysis, the indepen-dent predictors for poor survival were late stage(P<0·001) and size of residual tumour after surgery

312 Y. DONG ET AL.

? 1997 John Wiley & Sons, Ltd. , . 183: 311–317 (1997)

(P=0·002), but not cytoplasmic MUC1 expression.Analysis of MUC1 expression was also undertakenwithin the largest stage group (stage III). Within stageIII tumours, low cytoplasmic BC2 reactivity was alsoassociated with long survival (P=0·02, Fig. 3B), but wasnot associated with any particular histological type,tumour grade, or residual tumour. However, multivari-ate analysis again demonstrated that cytoplasmic BC2reactivity was not an independent predictor for survival(P=0·3) of stage III tumours.A minority of benign epithelial tumours (34 per cent),

but most low malignant potential (86 per cent) andinvasive (81 per cent) tumours, had high MUC1 expres-sion in the apical cell membranes where luminal surfaceswere identified (P<0·0001, see examples in Fig. 1).Because of the generally high expression of MUC1 onapical membranes, we have defined high expression asmore than 50 per cent of cells showing positivity. Highapical MUC1 expression was associated with non-mucinous tumours (P=0·001, Fig. 4A). A minority ofmucinous benign tumours (16 per cent), but most lowmalignant potential (67 per cent) and invasive (53 percent) mucinous tumours, had high apical MUC1 expres-sion, in contrast to more than 75 per cent of non-mucinous tumours which showed strong apical MUC1reactivity. In addition, high apical expression was associ-ated with advanced stage (P=0·0001, Fig. 4C) and bulkyresidual disease after surgery (P=0·004), but not with

tumour differentiation (Fig. 4B). Univariate analysisshowed that a low expression of MUC1 in the apicalmembrane was associated with good outcome for in-vasive tumours (P=0·004, Fig. 5), probably due to itsassociation with early stage (Fig. 4C). Six additionalcases with no luminal surfaces present were of poordifferentiation and late stage, and showed very poorsurvival.

Expression of MUC2 in epithelial ovarian tumours

Examples of MUC2 expression are shown in Figs 1Cand 1D. More than half of benign (52 per cent) and lowmalignant potential (52 per cent) tumours, but only 31per cent of invasive tumours, expressed MUC2 assessedby 4F1 reactivity. In benign tumours, MUC2 was local-ized mainly in the cytoplasm (20 per cent) or in both thecytoplasm and vacuoles (40 per cent). In a limitednumber of benign cases, it was also present in the apicalmembrane (7 per cent, 1/15), basolateral membrane(14 per cent, 2/15), or vacuoles alone (20 per cent, 3/15).Low malignant potential tumours showed MUC2expression only in the cytoplasm (64 per cent) or boththe cytoplasm and vacuoles (36 per cent), and malignanttumours showed MUC2 localization in the cytoplasm(68 per cent) or both cytoplasm and vacuoles (24 percent). Most mucinous benign (68 per cent), low malig-nant potential (70 per cent), and invasive (68 per cent,

Fig. 1—Expression of MUC1 and MUC2 mucins in ovarian tumour biopsies. (A) Benign serous cystadenoma of ovary showing MUC1expression, as detected by MAb BC2, on the apical cell membrane. (B) Poorly differentiated serous ovarian cancer showing apical membraneexpression (small arrow-head) and focal intense cytoplasmic expression (large arrow) of MUC1, as detected by MAb BC2. (C) Well-differentiated mucinous ovarian cancer showing MUC2 expression, as detected by MAb 4F1, as granular staining in the perinuclear cytoplasm(arrows). (D) Well-differentiated serous ovarian cancer showing diffuse cytoplasmic MUC2 expression, as detected by MAb 4F1, in almost allcancer cells. Magnification for A–D#125

313EXPRESSION OF MUCINS IN OVARIAN TUMOURS


P=0·003, Fig. 6A) tumours showed MUC2 expression.On the other hand, most non-mucinous tumours includ-ing benign (90 per cent), low malignant potential (64 percent), and invasive (72 per cent) tumours were MUC2-negative (Fig. 6A). In addition, the presence of the 4F1epitope was inversely associated with high tumour grade(P=0·01, Fig. 6B). However, no association was foundbetween 4F1 reactivity and stage (Fig. 6C), residualtumour or prognosis.

DISCUSSION

In agreement with previous reports,10,22,23 most ovar-ian carcinomas showed strong expression of the trans-membrane mucin MUC1. The MUC1 peptide epitopereactive with BC2 antibody was expressed more univer-sally and more strongly in low malignant potential andmalignant tumours than in benign tumours. In addition,our data showed that in invasive tumours, low expres-sion of MUC1 in the apical membrane was associatedwith good prognosis, and low cytoplasmic MUC1expression was also associated with good prognosis even

Fig. 2—Frequency analysis of detection by MAb BC2 of cytoplasmicMUC1 in epithelial ovarian tumours. (A) Histological types: Ser,serous; Muc, mucinous; End, endometrioid; Ccc, clear cell; Mix, mixedcarcinomas. (B) Differentiation status: LMP, low malignant potential;Well, well differentiated; Mod, moderately differentiated; Poor, poorlydifferentiated. (C) FIGO stage. Number of patients is given inparentheses

Fig. 3—Relationship between MUC1 detection in cytoplasm andsurvival of patients with invasive ovarian cancer. (A) All ovariancancer patients were divided into two groups: low cytoplasmic expres-sion (0–25 per cent of positive cells); high cytoplasmic expression (>25per cent of positive cells). (B) Stage III ovarian cancer patientswere divided into low and high cytoplasmic expression. For eachgroup, the number of patients in that group followed by the number ofuncensored patients at 100 months is shown in parentheses

314 Y. DONG ET AL.


within stage III tumours. We have also described theexpression of the gel-forming mucin, MUC2, in bothmucinous and non-mucinous ovarian tumours, andshown no association between expression of MUC2 andprognosis.MUC1 reactivity with the BC2 antibody is influenced

by glycosylation,24 almost certainly because there areglycosylation sites either side of the APDTR epitoperecognized by BC2.25 Increase in mucin productionand/or altered mucin glycosylation occurs in carcino-mas, generally resulting in greater exposure of VNTRepitopes in cancer cells than in normal epithelial cells,although this appears heterogeneous.24 This may explainthe high immunoreactivity to BC2 in low malignantpotential and invasive ovarian tumours. It is possiblethat in some tumours we have failed to detect MUC1, orunderestimated its expression, due to heavy glycosyla-tion of the core protein. Although glycosidases areavailable and their use is compatible with immunohisto-chemistry, these enzymes will remove only some selected

mucin carbohydrate structures. Unfortunately, chemicaldeglycosylation techniques which completely removeall mucin carbohydrate structures are incompatiblewith immunohistochemistry and preservation of mor-phology. Our results suggest that high expression and/oraltered glycosylation of MUC1 may be associated withthe progression of ovarian cancer.The MUC1 protein (polymorphic epithelial mucin) is

a membrane-bound glycoprotein that can be releasedfrom the cell surface, and possibly also by direct secre-tion, into the circulation. It may be interesting tocompare the different cellular localizations and expres-sion levels of MUC1 found in this study with serumconcentrations of MUC1 which are of clinical signifi-cance. We have previously shown markedly hetero-geneous secretion rates of MUC1 from primary culturesof ovarian cancer cells.26 High cytoplasmic expressionappears to be correlated with high secretion rates. Arecent study has shown that secreted MUC1 inducesapoptosis in activated T cells.27 The association betweenhigh cytoplasmic expression and poor survival in ourstudy may be related to this immunosuppression.The MUC1 protein can act as an anti-adhesion mol-

ecule, as its overexpression on the cell surface reducescell–cell and cell–extracellular matrix adhesion, prob-ably because its elongated structure interferes withinteractions between adhesion molecules and theirligands.28,29 Transfection of MUC1 into melanoma cellshas resulted in increased metastasis of these cells in nudemice.30 MUC1 may also cause immunosuppression byinhibiting lysis of tumour cells with high cell surfaceexpression by cytotoxic lymphocytes.31 Recent datasuggest that MUC1 may also act as a receptor-likemolecule participating in signal transduction.32,33 As a

Fig. 4—Frequency analysis of detection by MAb BC2 of apical MUC1in epithelial ovarian tumours. (A) Histological types: Ser, serous; Muc,mucinous; End, endometrioid; Ccc, clear cell; Mix, mixed carcinomas.(B) Differentiation status: LMP, low malignant potential; Well, welldifferentiated; Mod, moderately differentiated; Poor, poorly differen-tiated. (C) FIGO stage. Number of patients is given in parentheses

Fig. 5—Relationship between MUC1 expression in the apical mem-brane and survival of patients with invasive ovarian cancer. Allovarian cancer patients were divided into two groups: low apicalmembrane staining (1–50 per cent of positive cells) and high apicalmembrane expression (>50 per cent of positive cells). The number ofpatients followed by the number of uncensored patients at 100 monthsis shown in parentheses



consequence, MUC1 may not simply be a marker ofdifferentiation but may also influence metastasis, con-sistent with our finding that invasive carcinomas withhigh MUC1 expression had a poor prognosis.Confirming our preliminary study,11 both mucinous

and non-mucinous ovarian carcinomas were found toexpress the MUC2 protein, although the proportion ofpositive cells was greater in mucinous tumours. Theseresults contrast with a previous smaller study which didnot find MUC2 expression in serous carcinomas.15 In

contrast to the very high expression of MUC1, theMUC2 mucin was found in only about 30 per cent ofovarian cancers. MUC2 mucin is characteristicallysecreted by both normal and malignant gastrointestinalcells.34 The 4F1 antibody reacts with an epitope that ismasked when the mucin is fully glycosylated. In normalcolonic cells, the MAb 4F1 reacts with the protein coreonly in the endoplasmic reticulum prior to the comple-tion of glycosylation, rather than with the maturemucin.11 Approximately 90 per cent of ovarian muci-nous tumours have intestinal differentiation,19,20explaining the MUC2 expression in these tumours. Inaddition, the low apical expression of MUC1 in muci-nous tumours is probably attributable to the differentglycosylation from non-mucinous sub-types rather thandecreased synthesis, consistent with the high degreeof MUC1 glycosylation in non-malignant colonic epi-thelium. The inverse correlation between MUC2 expres-sion and tumour grade may reflect decreased synthesisand/or altered glycosylation of the epitope in poorlydifferentiated ovarian cancers. Interestingly, this gel-forming mucin was also produced by many non-mucinous tumours, but was not associated withprognosis.In conclusion, our study has detailed the expression of

MUC1 and MUC2 in epithelial ovarian tumours andthe results support experimental evidence suggestingthat MUC1 influences metastasis.

ACKNOWLEDGEMENTS

Ying Dong was supported by the Australian Inter-national Development Assistance Bureau. The projectwas funded by the National Health and MedicalResearch Council.

REFERENCES1. Devine PL, McKenzie IFC. Mucins: structure, function, and associations

with malignancy. Bioessays 1992; 14: 619–625.2. Gum JR. Mucin genes and the proteins they encode: structure, diversity,

and regulation (Review). Am J Respir Cell Mol Biol 1992; 7: 557–564.3. Gendler SJ, Spicer AP. Epithelial mucin genes (Review). Annu Rev Physiol

1995; 57: 607–634.4. Zotter S, Hageman PC, Lossnitzer A, Mooi WJ, Higers J. Tissue and tumor

distribution of human polymorphic epithelial mucin. Cancer Rev 1988; 11:55–101.

5. Hay FG, Leonard RCF. Expression of human milk fat globule antigensHMFG1 and HMFG2 on ovarian tumours and lung tumours. Dis Markers1988; 6: 29–39.

6. Girling A, Bartkova J, Burchell J, Gendler S, Gillett N, Taylor-Papadimitriou J. A core protein epitope of the polymorphic epithelial mucindetected by the monoclonal antibody SM-3 is selectively exposed in a rangeof primary carcinomas. Int J Cancer 1989; 43: 1072–1076.

7. Devine PL, McGuckin MA, Ward BG. Circulating mucins as tumormarkers in ovarian cancer (Review). Anticancer Res 1992; 12: 709–717.

8. Xing PX, Reynolds K, Tjandra JJ, Tang ZL, McKenzie IFC. Syntheticpeptides reactive with anti-human mild fat globule monoclonal antibodies.Cancer Res 1990; 50: 89–96.

9. Xing PX, Tjandra JJ, Stacker SA, et al. Monoclonal antibodies reactivewith mucin expressed in breast cancer. Immunol Cell Biol 1989; 67: 183–195.

10. McGuckin MA, Wright G, Ward BG. Expression of a polymorphicepithelial mucin antigen defined by the monoclonal antibody BC2 inovarian carcinoma. Use of the BC2 antibody for the detection of micro-metastases. Am J Clin Pathol 1991; 96: 46–52.

11. Devine PL, McGuckin MA, Birrell GW, et al. Second generation mono-clonal antibodies reacting with the MUC2 mucin core protein. Br J Cancer1993; 67: 1182–1188.

12. Jany BH, Gallup MW, Yan PS, Gum JR, Kim YS, Basbaum CB. Humanbronchus and intestine express the same mucin gene. J Clin Invest 1991; 87:77–82.

Fig. 6—Analysis of MUC2 detection by MAb 4F1 in epithelialovarian tumours. (A) Histological types: Ser, serous; Muc, mucinous;End, endometrioid; Ccc, clear cell; Mix, mixed carcinomas. (B)Differentiation status: LMP, low malignant potential; Well, welldifferentiated; Mod, moderately differentiated; Poor, poorly differen-tiated. (C) FIGO stage. Number of patients is given in parentheses

316 Y. DONG ET AL.


13. Gerard C, Eddy RL, Shows TB. The core polypeptide of cystic fibrosistracheal mucin contains a tandem repeat structure—evidence for a commonmucin in airway and gastro-intestinal tissue. J Clin Invest 1990; 86:1921–1927.

14. Xing PX, Prezoska J, Layton GT, Devine PL, McKenzie IFC. Second-generation monoclonal antibodies to intestinal MUC2 peptide reactive withcolon cancer. J Natl Cancer Inst 1992; 84: 699–703.

15. Tashiro Y, Yonezawa S, Kim YS, Sato E. Immunohistochemical study ofmucin carbohydrates and core proteins in human ovarian tumors. HumPathol 1994; 25: 364–372.

16. Price MR, Sekowski M, Ladanyi A, et al. Immune recognition of humancolonic-tumour-associated MUC-2 mucins using an anti-peptide antibody.Int J Cancer 1993; 55: 753–759.

17. Walsh MD, McGuckin MA, Devine PL, Hohn BG, Wright RG. Expressionof MUC2 epithelial mucin in breast carcinoma. J Clin Pathol 1993; 46:822–925.

18. Walsh MD, Hohn BG, Thong W, Devine PL, Samaratunga MLTH,McGuckin MA. Mucin expression by transitional cell carcinomas of thebladder. Br J Urol 1994; 73: 256–262.

19. Ball NJ, Robertson DI, Duggan MA, Snider DD. Intestinal differentiationin ovarian mucinous tumours. Virchows Arch A [Pathol Anat Histopathol]1990; 417: 197–201.

20. Tenti P, Aguzzi A, Riva C, et al. Ovarian mucinous tumors frequentlyexpress markers of gastric, intestinal, and pancreatobiliary epithelial cells.Cancer 1992; 69: 2131–2142.

21. Kaplan EL, Meier P. Nonparametric estimation from incomplete observa-tions. J Am Stat Assoc 1958; 53: 457–481.

22. Croghan GA, Wingate MB, Gamarra M, et al. Reactivity of monoclonalantibody F36/22 with human ovarian adenocarcinomas. Cancer Res 1984;44: 1954–1962.

23. Ward BG, Lowe DG, Shepherd JG. Patterns of expression of a tumorassociated antigen, defined by the monoclonal antibody HMFG2, in humanepithelial ovarian carcinoma. Comparison with expression of the HMFG1,AUA1 and F36/22 antigens. Cancer 1987; 60: 787–793.

24. Devine PL, Warren JA, Ward BG, McKenzie IFC, Layton GT. Glycosyla-tion and the exposure of tumour-associated epitopes on mucins. J TumorMarker Oncol 1990; 11–26.

25. Nishimori I, Perini F, Mountjoy KP, et al. N-Acetylgalactosamine glyco-sylation of MUC1 tandem repeat peptides by pancreatic tumor cell extracts.Cancer Res 1994; 54: 3738–3744.

26. McGuckin MA, Hurst TG, Ward BG. Heterogeneity in production, secre-tion and glycosylation of MUC1 epithelial mucin by primary cultures ofovarian carcinoma. Int J Cancer 1991; 63: 412–418.

27. Gimmi CD, Morrison BW, Mainprice BA, et al. Breast cancer-associatedantigen, DF3/MUC1, induces apoptosis of activated human T cells. NatureMed 1996; 2: 1367–1370.

28. Wesseling J, van der Valk SW, Vos HL, Sonnenberg A, Hilkens J. Episialin(MUC1) overexpression inhibits integrin-mediated cell adhesion to extra-cellular matrix components. J Cell Biol 1995; 129: 255–265.

29. Wesseling J, van der Valk SW, Hilkens J. A mechanism for inhibition ofE-cadherin-mediated cell–cell adhesion by the membrane-associated mucinepisialin/MUC1. Mol Biol Cell 1996; 7: 565–577.

30. Hilkens J, Vos HL, Wesseling J, et al. Is episialin/MUC1 involved in breastcancer progression? Cancer Lett 1995; 90: 27–33.

31. van de Wiel-van Kemenade E, Ligtenberg MJL, de Boer AJ, et al. Episialin(MUC1) inhibits cytotoxic lymphocyte–target cell interaction. J Immunol1993; 151: 767–776.

32. Wreschner DH, Zrihan Licht S, Baruch A, et al. Does a novel form of thebreast cancer marker protein, MUC1, act as a receptor molecule thatmodulates signal transduction? Adv Exp Med Biol 1994; 353: 17–26.

33. Pandey P, Kharbanda S, Kufe D. Association of the DF3/MUC1 breastcancer antigen with Grb2 and the Sos/Ras exchange protein. Cancer Res1995; 55: 4000–4003.

34. Yan PS, Ho SB, Itzkowitz SH, Byrd JC, Siddiqui B, Kim YS. Expression ofnative and deglycosylated colon cancer mucin antigens in normal andmalignant epithelial tissues. Lab Invest 1990; 63: 698–706.



expression of muc1 and muc2 mucins in epithelial ovarian tumours

Documents