Int. .I. Cancer: 48, 355-363 (1991) 0 1991 Wiley-Liss, Inc.

Publication of the International Union Against Cancer Publication de I’Union lnternationale Contre 1e Cancer

ONCOFETAL EXPRESSION OF THE HUMAN INTESTINAL MUCIN GLYCOPROTEIN ANTIGENS IN GASTROINTESTINAL EPITHELIUM DEFINED BY MONOCLONAL ANTIBODIES Paul J . HERTzoG’.~,H. Clem. ROBIN SON^, Jeng MA’, Ian R. MACKAY’ and Anthony W. LINNANE’ ‘Centre for Molecular Biology and Medicine and 2Department of Biochemistry, Monash University, Clayton, Victoria 3168. Australia. A mucin preparation from a colonic adenocarcinoma was used t o prepare monoclonal antibodies (MAbs) that reacted specifically either with normal adult small-intestine mucin antigen(s) (SIMA), or normal adult large-intestine mucin antigen@) (LIMA). Both SlMA and LIMA show a unique oncofetal pattern of expression. Thus SlMA was expressed in early fetal stomach, large and small intestines but thereafter only in the normal small intestine. SlMA expression was detected immunohistochemically in cancers of the colorectum (8211 12) and stomach (48/86). LIMA was detected in the stomach of the early fetus but thereafter only in the normal large intestine. LIMA expression was detected in 61/86 cancers of the stomach. Moreover, both SlMA and LIMA were expressed inappropriately in mucosa adjacent t o tumors, indicative of the detection of possible pre-malignant epithelium. We used a sandwich ELISA and biochemical procedures t o show that the SlMA and LIMA molecules were large extensively glycosylated multi-unit mucin glycoproteins that differed markedly from each other. SIMA, whether extracted from normal small-intestine or colonic cancers, had a molecular weight above 1.000 kDa, a mean buoyant density 1.33 g/ml and s value of 4.8. LIMA had a molecular weight above 10.000 kDa, a mean buoyant density 1.45 g/ml and an s value 9.5. The SlMA and LIMA epitopes were judged t o be carbohydrate in nature by reason of their resistancet o harsh physical chemical treatments or protease digestion, and sensitivity t o periodate oxidation, neuraminidase or p elimination. Only the SlMA epitope was sensitive t o neuraminidase. In conclusion, MAbs to carbohydrate-dependent epitopes on SlMA and LIMA identify the oncofetal pattern of expression of these distinct intestinal mucin glycoproteins in colonic and gastric carcinoma. These MAbs will be useful in further studies of the significance of oncofetal mucin expression during carcinogenesis.

Tumor antigens defined by polyclonal and more recently monoclonal antibodies often show oncofetal or differentiationassociated expression; that is, while not detected in the antecedent cells of the corresponding organ in the healthy adult, they are detected in the less differentiated cells of that organ in the fetus (Hakomori and Kannagi, 1983). In the gastrointestinal tract, differentiation-associated tumor antigens have been identified by antibodies raised using tumor cells (Nocera et a l . , 1987), normal cells (Hughes et al., 1986), cell lines (Koprowski et al., 1979), mucosal extracts (Richman and Bodmer, 1987), tumor extracts (Pant et al., 1986) or cell products such as much glycoproteins (Ma et al., 1980; Bara et a l . , 1980) as immunogens. Mucin glycoproteins, which represent a major secretory product of epithelial cells, are extensively glycosylated (6040% carbohydrate) high-molecular-weight (>1.OW kDa) multi-unit structures (Allen, 1983) contributing to the rheological properties of the mucus layer which protects the underlying epithelium. Previous immunohistochemical studies using polyclonal antisera to mucins have identified antigens in the small intestine, the large intestine or stomach, and have shown aberrant expression of these antigens in colorectal cancers (Bara et a l . , 1980; Ma et al., 1980). In view of the low specificity associated with the use of polyclonal antisera and the complexity of mucin glycoprotein structure, we have raised MAbs to human intestinal mucins expressed in gastrointestinal cancers with the aim of obtaining

more specific probes for the identification and characterization of different much glycoprotein antigens. The MAb-defined pattern of mucin antigen expression described herein appears unique, in that epitopes localized on mucins of the normal human adult small intestine, SIMA, show an oncofetal pattern of expression with respect to the gastric and colorectal mucosa. The expression of SIMA in colorectal cancers represents a “new” antigen for the colon, distinct from the norma1 mucin, LIMA. By contrast, epitopes localized on the mucins of the normal adult large intestine, LIMA, show an oncofetal pattern of expression in the gastric mucosa. MATERIAL AND METHODS

Extraction and purijication of mucins from tissues Tissue (approximately 5 g wet weight) from a resected cancer, diagnosed histologically as a partly mucinous adenocarcinoma of the sigmoid colon (designated specimen 1946) was cut into small pieces, homogenized in 10 volumes of 4 M guanidine hydrochloride in PBS (10 mM Na,HPO,, 1.5 m~ KH,PO,, 150 m~ NaC1, 3 mM KCl, PH 7.5) containing protease inhibitors (10 m~ N-ethylmaleimide, 1 mM benzamidine hydrochloride, 24 mM EDTA), then centrifuged at 20,000 g for 20 min. The supernatant was adjusted to a density of 1.35 g/ml with CsCl and centrifuged at 105,000 g for 64 hr. The resulting gradient was separated into 6 fractions of equal volume ranging in density from approximately 1.25 to 1.55 g/ml. Each fraction was dialysed for 48 hr against distilled H,O, then assayed for proteins (Lowry et a l . , 1951), and hexoses (Trevelyan and Harrison, 1952). Prior to the availability of MAbs, each fraction was tested for mucin antigenicity by inhibition of immunofluorescent staining of intestinal mucins by polyclonal antisera (Ma et al., 1980). Peak fractions with densities ranging from 1.35 to 1.50 g/ml were pooled for use in the production of MAbs. For characterization of mucin antigens, extracts were also prepared, as described above, from scrapings of normal small- and large-intestinal mucosa obtained at autopsy, 5 to 10 hr post-mortem, and peak fractions were pooled and subjected to 2 further CsCl density-gradient ultracentrifugation steps as described above. Immunization schedule and hybridoma production Female BALB/c mice were immunized with a total of 50 Fg protein per mouse of CsCl gradient-fractionated cancer mucin preparation, emulsified in Freund’s adjuvant, and the mouse with the highest serum antibody titer as determined by indirect ELISA (see below) was boosted and the spleen removed for cell fusion and selection (Hertzog et a l . , 1982). Hybridoma culture supernatants were screened for specific MAb production by indirect ELISA and immunohistochemistry (see below). Hybridomas which secreted MAbs reactive with mucins

3To whom correspondence and reprint requests should be addressed. Received: November 12, 1990 and in revised form February 12, 1991.

356

HERTZOG ET AL.

were cloned twice by limiting dilution. Immunoglobulins were purified from culture supernatants or ascites fluid by protein-A Sepharose (Pharmacia, Uppsala, Sweden) affinity chromatography and the sub-classes of each MAb determined by an isotyping kit (Misotest, Commonwealth Serum Laboratories, Melbourne, Australia). ELISA Indirect ELISA. Polystyrene microtiter plate wells (Nunc immunoplates, Rosklide, Denmark) were coated with CsClpurified cancer mucin preparation (10 pg protein/ml) in 10 mM bicarbonate buffer, PH 9.6, for 3 hr at 37°C. Wells were washed 4 times with PBS containing 0.05% Tween 20 following each of the following incubations at 37°C: (i) with 1% BSA in PBS for 30 rnin to block non-specific binding sites; (ii) with mouse antibodies (hybridoma culture supernatants or mouse serum) for 1 hr; (iii) with alkaline phosphatase-conjugated sheep anti-mouse immunoglobulin for I hr; and finally (iv) with substrate (1 mg/ml p-nitrophenyl phosphate in 10 m~ bicarbonate buffer, PH 9.8) for 1 hr. The absorbances of the solutions in the microtiter plates were read at 405 nm in a Titertek Multiskan (Flow, Irvine, UK). Samples which gave absorbance values greater than twice the background values obtained in the absence of coating mucin antigen were considered positive. Competitive ELISA. These assays were developed for the measurement of mucin antigenicity. Wells of polystyrene microtiter plates were coated with mucin and blocked with BSA as described above. In the next step, the coated wells were incubated with a suitable dilution of MAb without or with several concentrations of competing antigen for 1 hr at 37°C. The remainder of the assay was as described above. In order to determine the relative reactivities of the MAbs produced in this study, each antibody was tested in a competitive ELISA with a range of competing antigen concentrations. Sandwich ELISA. These assays were developed to facilitate the quantitation of LIMA or SIMA epitopes using the same MAb (2C3 in the case LIMA, or 4D3 in the case of SIMA), as the capture antibody coated to the microtiter plates, and as the detector MAb coupled to alkaline phosphatase (Voller et al., 1976). Optimal concentrations for the coating MAb and for the alkaline-phosphatase-conjugatedMAb were determined by checkerboard titration for each assay. The final assay conditions were as follows: protein-A-purified MAb (2 pg/ml in 10 m~ bicarbonate PH 9.6) was coated to polystyrene microtiter plates by incubation for 2 hr at 37°C. The plates were then washed with PBS/Tween and non-specific binding sites blocked by incubation with 1% BSA in PBS for 30 rnin at 37°C. Mucin antigen was then added at various dilutions and incubated for 1 hr at 37°C. After washing, the MAbalkaline-phosphatase conjugate was incubated at a concentration of 2 pgiml in PBS for 1 hr at 37°C. The remainder of the assay was performed as described for the indirect ELISA. An aliquot of SIMA and LIMA preparations, extracted from a mucinous colonic adenocarcinoma and normal adult colon, respectively and purified by 3 successive CsCl density-gradient ultracentrifugation steps, was set aside as a reference standard. These laboratory standards of SIMA and LIMA were each assigned an arbitrary concentration in “units” per ml. Standard curves were constructed using a range of standard antigen concentrations from 0 to 100 MAb-reactive units per ml (see “Results”) and were included in each assay. Units of SIMA, 4D3-reactive units, are not equivalent to units of LIMA, 2C3reactive units, since the antibody reactivity is dependent upon the epitope density on the mucin glycoprotein and the MAb avidity. Competition of MAb for muck epitopes. The ELISA method of Friguet et al., (1983) was used to determine whether the

MAbs competed for the same or closely related epitopes. In this procedure, limiting amounts of cancer mucin antigen are applied to microtiter plates, then an excess of a particular MAb is added to saturate all available epitopes. Following addition of a second MAb, an increase in signal indicated that the MAbs did not compete for the same or overlapping epitopes, whereas no increase in signal after addition of the second antibody indicated that the binding of the first MAb inhibited binding of the second MAb. Immunohistochemistry Both indirect immunofluorescent and indirect immunoperoxidase techniques were employed in the study. Positive hybridoma supernatants identified by first-round screening using ELISA were tested by indirect immunofluorescent staining of normal Gi mucosa from adult and fetal tissues and cancers of the gastrointestinal tract obtained at the time of surgical resection. Normal tissues, from specimens free of Gi pathology, were obtained at autopsy (7 cases) or biopsies and resections (stomach and small intestine from 3 cases, large intestine from 18 cases). Serial dilutions of culture supernatants were incubated for 20 rnin at 20°C with 5-p tissue sections from formalin-fixed, paraffin-embedded specimens mounted on glass slides. The sections were dewaxed, then washed twice with PBS, incubated with fluorescein-isothiocyanate-conjugated sheep anti-mouse immunoglobulin F(ab’), fragments (Silenus, Melbourne, Australia), washed again and mounted for examination by fluorescence microscopy using transmitted illumination with narrow-band blue (495 nm) excitation. For indirect immunoperoxidase staining, the dewaxed sections were blocked with 5% BSA in PBS for 5 min, drained and incubated with the diluted mouse antibodies for 20 min. After two 5-min washings with PBS, the sections were covered with horseradish-peroxidase-labelled rabbit anti-mouse immunoglobulin (Dakopatts, Copenhagen, Denmark), washed twice for 5 min with PBS, then the color developed with the substrate, 3-amino-9-ethylcarbazole in 50 mM acetate buffer, pH 5.0, for 4 min, counterstained with Mayer’s hemotoxylin for 6 min and mounted in 3% gelatin. Sections from all blocks were also stained with hematoxylin-eosin and Alcian blue (PH 2.5)/ periodic-acid-Schiff‘s-reagent to assess morphologic features and the presence of mucin glycoprotein. Antigen characterization The following analyses were performed on extracts of normal small-intestine, large-intestine or colorectal cancer specimens. Buoyant densities were derived from the fractions of CsCl density gradient ultra-centrifugation of tissue extracts, which contained the peaks of antigenic activity. Gel filtration chromatography was performed on columns (1.5 cm X 60 cm) of Sepharose 6B, 4B or 2B and Sephacryl 500 and 1000, equilibrated with 10 m~ phosphate buffer, PH 7.45. Fractions were collected at a flow rate of 5 ml/min and assayed for protein, hexose and antigenicity by sandwich ELISA as described above. Sedimentation coefficients of CsCl gradientpurified mucins were measured by ultracentrifugation on a glycerol gradient (10% to 20%) at 35,000 g for 15 hr or 105,000 g for 5 hr at 4°C. The antigenicity of SIMA and LIMA preparation was tested by sandwich ELISA after different physical, chemical and enzymic treatments. Aliquots of the mucin preparation were either boiled at 100°C for 5, 10, 20 and 30 min, or incubated in reducing conditions in 10 mM dithiothreitol (DTT), 6 M guanidine HC1, 0.5 M Tris, PH 8.1, 2 mM EDTA at 50°C for 2, 4 or 16 hr, then alkylated with iodo-acetic acid at 4°C for 16 hr. p elimination was performed by incubation in 0.5 M KOH with IM NaBH, at 4°C for 16 hr. Mucin preparations were also incubated with 0.01 to 10 mg/ml of pepsin (Calbiochem, La Jolla, CA) in 0.2 M acetate buffer, PH 2.0; 1 mg ml of

ONCOFETAL EXPRESSION OF INTESTINAL MUCIN ANTIGENS

clostripain (Calbiochem) in 0.1 M Tris buffer, PH 7.3, with 30 m~ dithiothreitol and 0.2 M CaCl,; 0.1 or 1.O mg/ml of papain (Sigma, St. Louis, MO) in acetate buffer, PH 5.5, containing 1 mgfml cysteine and 5 mM EDTA; and 1 mg/ml of neuraminidase (Sigma) in 50 mM sodium acetate buffer, PH 6.0, with 10 m~ NaCl for 2, 4, and 16 hr at 37°C. LIMA and SIMA samples were boiled for 5 min (which did not affect antigenicity) to destroy enzyme activity prior to immunoassay. In the case of pepsin digestion, where the enzyme incubation was performed at PH 2.5, samples were neutralized prior to immunoassay or, in the reduction experiment, diluted to 0.05 M guanidinium chloride. Control experiments showed that none of the denatured enzymes or the digestion buffers affected the immunoassay itself. The periodate oxidation method of Woodward et al. (1985) was used to provide evidence for the carbohydrate nature of the epitopes. Microtiter plates were coated with 100 units per ml of SIMA or LIMA and blocked as described above for the indirect ELISA, then treated with 0.1 to 50 m~ sodium metaperiodate (Sigma) and sodium borohydride. Reactivity with serial dilutions of 2C3 or 4D3 alkaline-phosphatase conjugates was tested as described in the ELISA section above. Coated antigens treated with acetate buffer and sodium borohydride served as controls. Expression of results Data are presented as mean and standard deviation with the number of experiments in parentheses. RESULTS

Characteristics of monoclonal antibodies Of 288 hybridomas resulting from the fusion experiment, 10 produced MAbs reacting with the cancer much preparation (specimen 1946) as determined by the indirect ELISA. Immunofluorescence staining of sections from the normal adult GI tract showed that 5 MAbs, 4D3, 4C2, 4D1, 2A1 and 3C5, reacted specifically with much of the small intestine and were accordingly designated anti-SIMA. Five other MAbs, 2C3, 2D3, 3C3,3D4 and 3B4, reacted specifically with mucin of the large intestine and were accordingly designated anti-LIMA. All but one of the MAbs were found to be of the IgG, subclass; 4D3 was IgG,. Competition of MAbs for mucin epitopes (Friguet et al., 1983; see “Material and Methods”) clearly demonstrated that the anti-LIMA MAbs recognized separate epitopes from the anti-SIMA MAbs in a much preparation from a colorectal cancer. Each anti-LIMA MAb gave an additive response, with each anti-SIMA MAb indicating the recognition of different epitopes or of different antigens by the 2 groups of MAbs. However, none of the anti-LIMA MAbs gave an additive response with other anti-LIMA MAbs, nor did any anti-SIMA MAb give an additive response with other anti-SIMA MAbs. Competitive ELISA were carried out within each group to establish the relative avidity of each MAb for antigen. The concentration of LIMA required for 50% inhibition of color development differed over a 10-fold range for the anti-LIMA MAbs; the binding of 2C3 showed the highest relative avidity. In contrast, the anti-SIMA MAbs differed only over a 4-fold range in antigen avidity. In view of the apparent similarity of the reactivity of the antibodies within each group and the instability of some of the original hybridomas, the anti-LIMA MAb 2C3 and the anti-SIMA MAb 4D3 were selected for characterization. Neither MAb was positive in red-blood-cell hemagglutination tests using red blood cells carrying ABH, Lewisa or Lewisb structures. Sandwich ELISA for WMA and SIMA Sandwich ELISA were developed for SIMA using the 4D3

357

MAb and 4D3-alkaline-phosphatase conjugate and for LIMA using the 2C3 MAb and 2C3-alkaline-phosphatase (Fig. 1, a and b respectively). For these assays, the reference standard preparations of SIMA and LIMA contained 50 p g of proteinl ml, and were assigned arbitrary levels of activity of 5,000 4D3-reactive unitdm1 and 500 2C3-reactive unitdm1 respectively. Each standard curve was constructed using mucin at concentrations ranging from 0 to 100 MAb-reactive units/ml, and gave reproducible results in the range 2 to 100 units/ml; 20 unitdm1 in the SIMA (4D3) sandwich ELISA gave an absorbance reading of about 0.5 and in the LIMA (2C3) sandwich ELISA an absorbance of about 0.7. Physicochemical characterization of LIMA and SIMA molecules Buoyant density. Equilibrium ultracentrifugation of tissue extracts in a CsCl density gradient was used to isolate LIMA reactive to MAb 2C3 from contaminating protein and hexosepositive material (Fig. 2a). In the example given, the peak LIMA fraction contained 5,650 antigen units/ml, 85 pg of proteidml (specific activity: 67 units per pg protein) and 125 pg of hexose/ml. Repeated ultracentrifugation in CsCl density gradients removed more contaminating protein and yielded peak LIMA fractions with specific activities of 250 units/pg of protein after a second centrifugation and 1,580 units/p,g of protein after a third. Chemical analysis of several preparations of LIMA purified on CsCl gradients from different tissue specimens revealed considerable variation in activity and composition: ELISA reactivity, 555 k 312 (n = 6) units per mg dry weight; hexose content, 9.2 4 8.6 (n = 6) and protein content 18.6 ? 25.6% (n = 6). CsC1-gradient centrifugation of extracts of specimens of normal large intestine and sandwich ELISA with MAb 2C3 demonstrated that LIMA banded in the density range 1.36 to 1.52 g/ml (mean 1.45 k0.04, n = 31). LIMA extracted from 9 colorectal cancer specimens and centrifuged in the same way showed a similar average peak density of 1.40 ? 0.03 g/ml. As with the LIMA fraction, CsC1-gradient ultracentrifugation separated the 4D3-reactive SIMA fraction from contaminating protein and hexose-positive material in crude tissue extracts. The peak SIMA fraction shown in Figure 2b contained 58,000 antigenic unitstml, 150 pgiml protein (specific activity: 387 units per p g protein) and 160 pg/ml hexose. Repeated ultracentrifugation in CsCl density gradients yielded further purification of SIMA, with specific activities of 490 unitdpg after a second centrifugation and 650 units/kg after a third. However, CsC1-gradient centrifugation of extracts of specimens of normal small intestine and analysis by sandwich ELISA with MAb 4D3 demonstrated that SIMA banded at a mean density of 1.33 2 0.02 g/ml (n = lo), lower than that for LIMA. Furthermore, SIMA extracted from colorectal cancer specimens banded at a mean density of 1.34 ? 0.02 g/ml (n = 7), also clearly different from the LIMA activity in the same cases. Gel-jiltrationchromatography. LIMA eluted at the void volume during gel filtration on columns of Sepharose 6B, 4B or 2B. Gel filtration of LIMA on a column of Sephacryl lo00 indicated that the majority of the antigenic activity was included in this matrix; it eluted as a broad peak, characteristic of a polydisperse mucin preparation (Fig. 3). By contrast, gel filtration on Sepharose 2B showed that SIMA was included in the gel and exhibited polydispersity, eluting as a number of peaks all of smaller molecular size than LIMA which eluted in the void volume. SIMA was also more retarded than LIMA on Sephacryl 1000 (Fig. 3). Sedimentation coefficients. Ultracentrifugation of purified mucin preparations on 10 to 20% glycerol gradients yielded a mean sedimentation coefficient for LIMA of 9.5 -+ 1.5 s. The

358

HERTZOG ET AL.

z

c

4

4 U

0

g

2000’

200’

0

v z

8

100.

4

r

X

W I

a FRACTION NUMBER

0

2

5

10

20

50

--I

; --

L

100

80009

II

E

*

\

:.

E

m

50000’

3

-z \

D

SIMA CONCENTRATION ( 4 D 3 - r o a c t l v o Unlts/rnl)

0 c

2 400003OOP 2

4

p

c

?I 0 g

200’

:

100.

0 W

I

30000’

2 ?I 2 0 0 0 0 ‘ 0 gv 10000’ 4:

3

b FRACTION NUMBER

FIGURE2 - Cesium chloride density gradient ultracentrifugation profile of (a) LIMA extracted from a specimen of normal adult colon, and (b) SIMA extracted from a specimen of normal adult jejunum. Fractions were measured as described in “Material and Methods” for density, protein (---), hexose (-) and antigen (-) concentration by sandwich ELISA.

I-

U

LIMA CONCENTRATION ( 2 C 3 - r e a c t i v e Units/ml)

FIGURE1 - Sandwich ELISA for SIMA (a) and LIMA (b). SIMA reference standard was assayed in a sandwich ELISA at concentrations ranging from 0-100 units/ml using microtiter plates coated with MAb 4D3 to immobilize the antigen and MAb 4D3-alkaline-phosphatase conjugate with p-nitro-phenyl phosphate for detection. The results are shown as mean and standard error of triplicate determinations in a typical assay. A similar assay using anti-LIMA MAb 2C3 and a LIMA reference standard is shown in (b).

4

z VO

F R A C T I O N NUMBER

Vt

mean sedimentation coefficient determined for a number of FIGURE3 - Gel filtration on Sephacryl 1000 of a sample of LIMA SIMA preparations on 10% to 20% glycerol gradients was from a specimen of normal adult colon (-) and SIMA from a specimen of normal adult jejunum (----), partially purified on a CsCl gra4.8 i 1.4 s. Physical, chemical and enzyme sensitivity of epitopes. The dient and tested for antigenicity in a sandwich ELISA using 2C3 or respectively. The column (1.5 X 60 cm) was eluted with 10 mM antigenicity of LIMA in a sandwich or competitive assay with 4D3 phosphate buffer, PH 7.45, at 5 ml per min. MAb 2C3 was not eliminated by heating at 100°C for 10 min (60% recovery of activity), treatment with 10 mM dithiothreitol, PH 7.5, at 50°C for 16 hr (98% recovery). Exhaustive ery). In order to demonstrate the carbohydrate nature of the digestion with various enzymes failed to produce a marked 2C3 epitope, P elimination of LIMA was performed. As a reduction of LIMA reactivity. These included pepsin (95% result, antigenicity in the sandwich ELISA was lost, whereas recovery), pronase (>90% recovery), papain (70% recovery), reactivity in a competitive ELISA was retained. This result is P-galactosidase (86% recovery) or neuraminidase (98% recov- consistent with the release, during P elimination, of monova-

359

ONCOFETAL EXPRESSION OF INTESTINAL MUCIN ANTIGENS

lent oligosaccharides containing the 2C3 epitope. Treatment of LIMA with 0.1, 1.0 or 5 mM sodium periodate resulted in 90%, 97% and 99.7% loss of 2C3 reactivity, respectively. The antigenicity of SIMA assayed with MAb 4D3 was also found to be stable to heating at 100°C (94% recovery) and resistant to exhaustive digestion with enzymes including pronase (>90% recovery) and papain (85% recovery). In contrast to LIMA, the neuraminidase digestion of SIMA reduced its reactivity with MAb 4D3 to less than 0.1% of the original activity. Treatment of SIMA with 0.1, 1.0 or 5 m~ sodium periodate resulted in a 38%, 59% and 84% loss of reactivity, respectively. Reactivity of normul adult tissues

The anti-LIMA MAb 2C3, and the anti-SIMA MAb 4D3 both showed strong and specific reactivity with mucin in sections of formalin-fixed paraffin-embedded tissues from the large and small intestine respectively. The anti-LIMA MAb 2C3 reacted only with mucin in the large intestine, except for occasional goblet cells in the terminal ileum, trachea and endocervix (Table I). The staining pattern in the large intestine was localized to mucin within goblet cells at all levels of the crypt and secreted mucin, both in the lumen of the crypts and the intestinal lumen (Fig. 4). Sections from different parts of the colon and rectum showed a similar even staining pattern except for the ascending colon, in which the mucin in occasional goblet cells was not stained. No 2C3 reactivity was observed in the normal stomach, duodenum, jejunum or proximal ileum. The anti-SIMA MAb 4D3 reacted almost exclusively with much throughout in the normal small intestine except for occasional goblet cells in the proximal cecum, trachea and endocervix. Both goblet-cell and secreted mucin were stained, most intensely at the base of the crypts (Fig. 5). No anti-SIMA 4D3 reactivity could be detected in the normal large intestine or stomach. No typical blood-group-type (i.e., red blood cell or endothelial cell) reactivity of 4D3 or 2C3 was observed in any normal tissue specimens.

FIGURE4 - Immunoperoxidase staining of normal left colon mucosa with LIMA MAb, 2C3, showing staining of goblet cells (arrow) and secreted much (arrowhead). Scale bar: 50 pm.

Reactivity of fetal tissue

Tissues from 10 fetuses were examined for reactivity with 4D3 and 2C3 MAbs (Table I). Anti-SIMA MAb 4D3, in contrast to its staining pattern of normal adult tissues, showed a broader staining range of mucin in different parts of the fetal Gi tract. Sections from only 4 specimens contained a portion of fetal stomach with goblet cells, and in all 4 the mucin reacted with 4D3 (Fig. 6a). SIMA was also detected using MAb 4D3 in the goblet cells of the large intestine of the 2 specimens from fetuses of less than 13 weeks (Fig. 6b), but was not detected in the 8 specimens of large intestine from fetuses more than 13 weeks old. Similarly to that of the adult, the fetal small intestine at both early and late stages of development contained SIMA reactive with 4D3. Similarly to SIMA, LIMA was de-

tected by using MAb 2C3 in the 4 specimens of fetal stomach. However, in contrast to SIMA, MAb-2C3-reactive LIMA was absent from the fetal small intestine, and from the large intestine in fetuses less than 13 weeks old; but it was detected in the large intestine from fetuses of more than 13 weeks, at which stage SIMA was no longer detected (Table I). Reactivity of cancers

Cancer tissues were surveyed for reactivity with MAbs 2C3 and 4D3 by immunoperoxidase staining (Table 11). Many gastric, colorectal, gall bladder and pancreatic carcinomas were positive for LIMA production. With the exception of colorectal

TABLE I - REACTIVITY OF ANTI-LIMA MAb 2C3 AND ANTI-SIMA MAb 4D3 IN NORMAL ADULT AND FETAL TISSUES

Number of cases

Number of cases Normal adult tissues’

Stomach Duodenum Jejunum Ileum Colon and rectum

Fetal tissues’

Positive to:

Total

7c1 ---

10 10 10 10 25

0 0 0 0

25

4n-4

0 10 10 10 0

Stomach Small intestine Large intestine 8-12 weeks 13-40 weeks

Total

Positive to: 2C3

4D3

4 10

4 0

4 10

2 8

0 8

2 0

‘There was no reactivity with anti-LIMA MAb (ZC3) or with anti-SIMA MAb (4D3) in sections of normal adult esophagus (3), gall bladder (3,pancreas (3,liver (5). lung (5). breast (9), ovary and fallopian tube (7). ecto-cervix (4), prostate (lo), parotid (4). skin (6),brain (2), spleen ( Z ) , lymph node (4). peripheral blwd leucocytes (12), kidney (2), bladder (4). thyroid (2) and heart (2). Occasional goblet cells in the terminal ileum were positive for the anti-LIMA MAb 2C3, in the proximal cecumfor anti-SIMA MAb 4D3 and in the endo-cervix and tracbea for both MAhs.-’Other fetal tissues observed to be negative for reactivity with MAh 2C3 or 4D3 were lung (lo), kidney (lo), liver (lo), pancreas (lo), skin (10) and brain (10).

360

HERTZOG ET A L .

LIMA and SIMA were frequently detected using the 2C3 and 4D3 MAbs respectively. Reactivity of gastrointestinal cell lines

Since mucin production is a property of specialized goblet cells in the normal gastrointestinal epithelium, yet also a feature of a high proportion of cancers, it was of interest to determine whether continuous cell lines derived from colorectal cancers produced SIMA or LIMA. The cell lines HT29, colo 320, colo 394, colo 229 (Jones et al., 1983) produced no detectable SIMA or LIMA. However, supernatants from LIM 1215 (Whitehead et a l . , 1985) contained about 15 to 30 2C3reactive units per ml of LIMA and supernatants of LIM 1863 (Whitehead et al., 1987) contained between 50 and 150 2C3reactive units per ml; neither produced SIMA detectable in sandwich ELISA using 4D3. Immunoperoxidase staining of cell smears or pellets fixed in 10% neutral formalin indicated that only LIM 1215 and LIM 1863 stained with 2C3, and no cells stained with 4D3. DISCUSSION

FIGURE5 - Immunoperoxidase staining of normal jejunum with SIMA MAb 4D3 showing staining of goblet cells (arrow) and secreted mucin (arrowhead). Scale bar: 50 Fm.

cancer, this LIMA production was tissue-inappropriate, since LIMA is normally detected only in the colon and rectum. Also many colorectal (Fig. 7), gastric (Fig. 8), gall bladder and pancreatic cancers were positive for SIMA and this occurrence was again tissue-inappropriate. The histological types of GI cancers included classical mucinous cancers, signet-ring-cell carcinoma, as well as moderately and poorly differentiated cancers. Some specimens from all types of cancers showed inappropriate mucin antigen expression, but a higher proportion of the mucinous cancers were positive. The staining pattern in cancers included diffuse cytoplasmic (Fig. 8) and apical membrane-associated (Fig. 7) staining, also extracellular staining present in the lumen of the glands or occurring as so-called mucinous lakes (Figs. 7 and 8). Examination of some non-Gi cancer specimens revealed that some breast and ovarian carcinoma also reacted with MAbs 2C3 or 4D3 (Table 11). Transition mucosa

In the mucosa immediately adjacent to colorectal cancers, the “transitional mucosa”, there was a change in the mucin detected in the goblet cells from 2C3-reactive LIMA, to 4D3reactive SIMA. This change in the mucin occurred most markedly at the base of crypts and at times was observed in the entire length of the crypt, particularly close to the tumor. Similarly, intestinal metaplasia and dysplasia were frequently observed in the mucosa adjacent to gastric cancers, and both

Monoclonal antibodies were produced by immunizing mice with mucin derived from a colonic carcinoma, and classified according to their immunohistochemical reaction with much occurring in the normal adult small intestine (MAbs denoted anti-SIMA) or large intestine (MAbs denoted anti-LIMA). One MAb was selected from each group for more extensive characterization, the anti-SIMA MAb 4D3 and the anti-LIMA MAb 2C3. A striking oncofetal or differentiation-associated pattern of SIMA and LIMA expression was defined by these MAbs. In the 8-to-12-week-old fetus, the SIMA epitope identified by 4D3 was expressed in the stomach and the small and large intestines, yet after 13 weeks of fetal development, also in the normal adult, the epitope was present only in the small intestine. However, it was re-expressed in gastric and colorectal cancers. In the early fetus, the LIMA epitope detected by 2C3 was expressed only in the stomach, but during later development and in the adult its expression was restricted to the colon and rectum; but again, it was expressed in some gastric cancers. These mAbs used in ELISA assays, together with biochemical analyses, permitted the antigens to be characterized as mucins and demonstrated that SIMA production in colonic cancers represented production of a new molecule distinct from the normal colonic mucin, LIMA. These anti-SIMA and anti-LIMA MAbs identify useful “tumor marker” antigens, since immunohistochemistry revealed aberrant expression of SIMA and LIMA in a high proportion of cancers of the stomach (56% 4D3-positive; 7 1% 2C3-positive) and colon (82% 4D3-positive; 79% 2C3-positive). It is noteworthy that, in contrast to their localization in normal tissues, the SIMA and LIMA epitopes were not confined exclusively to GI cancers but were also commonly detected in carcinomas of the gall bladder and pancreas and in mucin-producing tumors of the lung, breast and ovary. Similarly, wide patterns of reactivity have been reported for other oncofetal antigens of stomach (Bara et al., 1980), breast (Griffiths et al., 1987) and ovary (Pant et al.. 1986). Comparison of the biochemical characteristics and tissue distribution of LIMA and SIMA with published data for other antigens associated with tumors of the Gi tract demonstrates that MAb 4D3 and MAb 2C3 recognize novel epitopes (Itzkowitz et al., 1986; Schoentag et al., 1987; Kimet al., 1986). Thus the anti-LIMA MAb 2C3 epitope is evenly distributed throughout the colon, whereas the ABH and Leb antigens are localized to the proximal colon (Schoentag et al., 1987). Also,

ONCOFETAL EXPRESSION OF INTESTINAL MUCIN ANTIGENS

361

FIGURE6 - Immunoperoxidase staining of 10-week-old fetal colon (a) and stomach (6) using the anti-SIMA MAb 4D3, showing strong reactivity with goblet cells (arrow). Scale bar: 50 p,m. TABLE I1 - REACTIVITY OF ANTI-LIMA MAb 2C3 AND ANTI-SIMA MAb 4D3 IN CARCINOMA Carcinoma

Number of cases

Stomach Colon and rectum Gall bladder Pancreas ovary Breast

86 112 7 4 10 5

Mucin reactivity LIMA-2C3

SIMA-4D3

61 88 3

48 82 6 3 5 2

2 5 4

Summary of the number of cases of various carcinomas showing positive reactivity with MAbs 2C3 and 4D3 by indirect immunoperoxidase staining.

in contrast to the intense staining of goblet-cell mucin by 2C3, antibodies to Lex, Ley or Lea antigens rarely stain goblet-cell mucin (Itzkowitz et al., 1986; Kim et al., 1986). Using MAb 2C3 or 4D3, we could not detect blood-group reactivity in hemagglutination assays, or blood-cell staining or endothelialcell staining by immunohistochemistry. In contrast to SIMA and LIMA expression, the MI and M2 mucin epitopes are expressed in normal adult stomach tissues, whereas M3 is expressed in both small and large intestines (Bara et al., 1980). Other laboratories have produced MAb to carbohydrate epitopes on colonic mucin (which may have similarities to LIMA but not SIMA), but the detection of these epitopes during organ development or in gastrointestinal cancers has not been described (Podolsky et al., 1986). Partial characterization of LIMA and SIMA was obtained by physicochemical analyses and degradation studies combined

with immunoassays of mucin glycoproteins isolated from normal adult large- or small-intestine or colorectal cancers. Gelfiltration chromatography, buoyant density and the sedimentation coefficients were all consistent with the recognition by both MAbs of epitopes on extensively glycosylated, much glycoproteins of high m.w. (above 1.000 kDa). These analyses indicated that LIMA was a larger macromolecule and of higher density (possibly reflecting relative carbohydrate-to-protein composition) than SIMA. The relative stability of the epitopes to harsh treatment such as boiling, disulphide reduction, alkaline hydrolysis, protease digestion, formalin fixation and paraffin embedding of tissues is consistent with both 4D3 and 2C3 MAbs recognizing conformation-independent epitopes probably composed of, or protected by, carbohydrate moieties. Furthermore, the reactivity of LIMA with MAb 2C3 was sensitive to p elimination, which released monovalent (probably oligosaccharide) antigenic structures. The reactivity of SIMA with MAb 4D3, in contrast to that of LIMA with MAb 2C3, was neuraminidase-sensitive, indicating that the epitope contained N-acetyl neuraminic acid, or that the conformation or stability of the epitope required N-acetyl neuraminic acid (Sadler et al., 1979). Other colorectal tumor antigens which contain N-acetylneuraminic acid include CA 19-9, a sialylated Lewisa antigen, and sialylated Lewisx-related structures (Gong et al., 1985), but these differ from SIMA in their patterns of reactivity with normal small and large intestine. We also note similarities in the staining of colorectal carcinoma and transitional mucosa by the presently reported anti-SIMA MAb 4D3, while B72.3 MAbs (Xu etal., 1989) and TKH2 (Orntoff etal., 1990), both of which recognize the sialosyl T, epitope, could not be detected in the normal small intestine. Moreover, we find that the reactivity of MAbs 4D3 and TKH2 with the nor-

362

HERTZOG ET A L .

FIGURE7 - Immunoperoxidase staining of partially mucinous adenocarcinoma of the colon with MAb 4D3, showing staining of the secreted mucin, mucinous lakes (arrow), and apical borders of the cells within glands (arrowhead). Scale bar: 100 pm.

FIGURE8 - Immunoperoxidase staining of mucinous adenocarcinoma of the stomach with MAb 4D3 showing the staining in mucinous lakes (arrow) and of intracellular material (arrowhead). Scale bar: 50 p m .

ma1 Gi tract are clearly different (data not shown). The reactivity of 4D3 was also sensitive to periodate oxidation of SIMA, consistent with a carbohydrate epitope. The fact that the SIMA epitope is more sensitive than the LIMA epitope to periodate oxidation could be explained by the different sensitivity of the residues which comprise these epitopes. The mucin glycoproteins SIMA and LIMA differ from other tumor-marker antigens, including cell-surface glycoproteins

and glycolipids, because they are more extensively glycosylated, and are naturally secreted molecules. Although they are produced in various organs, we nonetheless expect them to be useful tumor markers, since they are normally secreted only into the Gi lumen, while in cancers they become demonstrable in atypical sites and may enter the blood stream. Indeed, SIMA and LIMA are demonstrable in serum of patients with cancer, according to preliminary data using ELISA.

REFERENCES

ALLEN,A., Mucus-a protective secretion of complexity. TIBS, 169-173 (1983). R. and BURTIN,P., Antigens of gastnc and intesBARA,J., LOISILLIER, tinal mucous cells in human colonic tumours. Brit. J . Cancer, 41, 209-221 (1980). FRICUET,B., DJAVADI-OHANIANCE, L., PAGES,J , BUSSARO, A. and GOLDBERG, M., A convenient enzyme-linked immunosorbent assay for testing whether monoclonal antibodies recognize the same antigenic site. Application to hybridomas specific for the @,-subunitof E. coli tryptophan synthase. J . irnmunol. Methods, 60, 351-358 (1983). S., SHIMOSATO, Y . , WATANABE, M., INO, Y., GONG,E., HIROHASHI, S. and KODAIRA, S., Expression of carbohydrate antigen 19-9 TESHIMA, and stage-specific embryonic antigen 1 in non-tumorous and tumorous epithelia of the human colon and rectum. J. nut. Cancer. Inst., 75,44745 (1985). J., GENDLER, S., LEWIS,A., BLIGHT,K., GRIFFITHS,A.B., BURCHELL, TILLY,R. and TAYLOR-PAPADIMITRIOU, J., Immunological analysis of much molecules expressed by normal and malignant mammary epithelial cells. Int. J . Cancer, 46, 319-327 (1987).

HAKOMORI, S.-I. and KANNAGI, R., Glycosphingolipids as tumorassociated and differentiation markers. J . nut. Cancer Inst., 71, 231-248 (1983). HERTZOG, P.J., SHAW,A. and GARNER, R.C., Improved conditions for the production of monoclonal antibodies to carcinogen-modifiedDNA for use in enzyme-linked immunosorbent assay (ELISA). J . irnrnunol. Methods, 62, 49-58 (1982). HUGHES,N.R., WALLS,R.W.,NEWLAND, R.C. and PAYNE,J.E., Antigen expression in normal and neoplastic colonic mucosa: three tissuespecific antigens using monoclonal antibodies to isolated colonic glands. Cancer Res., 46, 2164-2171 (1986). Y., PALEKAR, A,, PHELPS,P.C., ITZKOWITZ, S.H., YUAN,M., FUKUSHI, SHAMSUDDIN, A.M., TRUMP,B.F., HAKOMORI, S.-I. and KIM, Y.S., Lewisx- and sialylated Lewisx-relatedantigen expression in human malignant and non-malignant colonic tissues. Cancer Res.. 46, 2627-2632 (1986). P.J. and NAIRN,R.C., JONES,S .L., RHL,E., NIND,A.P.P., CHALMERS, Antibody-dependent cellular cytotoxicity (ADCC) in colorectal carcinoma. 11. Blood-group association. J . surg. Oncol., 24, 229-232 (1983).

ONCOFETAL EXPRESSION OF INTESTINAL MUCIN ANTIGENS

KIM,Y.S.,YUAN,M., ITZKOWTZ,S.H., SUN,Q., KAIZU,T., PALEKAR, S.-I., Expression of Ley and extended A,, TRUMP,B.F. and HAKOMORI, L e y blood-group-related antigens in human malignant, pre-malignant and nonmalignant colonic tissues. Cancer Res., 46, 5985-5992 (1986). KOPROWSKI, H., STEPLEWSKI, Z., MITCHELL, K., HERLYN,M., HERLYN, P., Colorectal carcinoma antigens detected by hybridoma D. and FUHRER, antibodies. Somatic Cell Genet., 5 , 957-972 (1979). LOWRY,0.J., ROSEBROUGH, N. J., FARR,A.L. and RANDALL, R. J., Protein measurement with Folin phenol reagent. J . biol. Chem., 193,265-275 (195 1). MA, J., DE BOER,W.G.R.M., WARD,H.A. and NAIRN,R.C., Another oncofoetal antigen in colonic carcinoma. Brit. J . Cancer, 41, 325-328 (1980). NOCERA,M., SHOCHAT, D., PRIMAS,F.J., KRUPEY,J., JESPERSEN, D.L. and GOLDENBERG, D.M., Representation of epitopes on colon-specific antigen-p defined monoclonal antibodies. J. nut. Cancer Inst., 79, 943948 (1987). ORNTOFT,T.F., HARVING,N. and LANGKILDE, N.C., 0-linked mucintype glycoproteins in normal and malignant colon mucosa: lack of Tantigen expression and accumulation of T, and sialosyl-T, antigens in carcinomas. Inr. J. Cancer, 45, 666672 (1990). PANT,K.D., STEWART, T.A., BERRY,C.O.A. and RHODES,B.A., Production of monoclonal antibody SP-21 to colon-ovarian tumour antigen, COTA. Hybridoma, 5 , 129-135 (1986). PODOLSKY, D.K., FOURNIER,D.A. and LYNCH,K.E., Human colonic goblet cells. Demonstration of distinct subpopulations defined by mucinspecific monoclonal antibodies. J. clin. Invest., 77, 1263-1271 (1986). RICHMAN, P.I. and BODMER,W.F., Monoclonal antibodies to human co-

363

lorectal epithelium. Markers for differentiation and tumour characterization. Znr. J . Cancer, 49, 317-328 (1987). SADLER,J.E., PAULSON, J.C. and HILL,R.L., The role of sialic acid in the expression of human MN blood group antigens. J. b i d . Chem., 254, 2112-2119 (1979). SCHOENTAG, R., PRIMUS,F.J. and KUHNS,W., ABH and Lewis blood group expression in colorectal carcinoma. Cancer Res.. 47, 1695-1700 (1987). TREVELYAN, W.E. and HARRISON, J.S., Studies on yeast metabolism. I. Fractionation and microdetermination of cell carbohydrates. Eiochem. J . , 50, 298-303 (1952). VOLLER,A., BIDWELL,D.E. and BARTLETT,A., Enzyme immunoassays in diagnostic medicine: theory and practice. Bull. WHO, 53, 55 (1976). WHITEHEAD, R.H., JONES,J.K., GABRIEL,A. and LUKIES,R.E., A new carcinoma cell line (LIM 1863) that grows as organoids with spontaneous differentiation into crypt-like structures in virro. Cancer Res., 47, 26832689 (1987). WHITEHEAD,R.H., MACRAE,F.A., ST. JOHN,D.J.B. and MA, J., A colon-cancer cell line (LIM1215) derived from a patient with inherited non-polyposis colorectal cancer. J. nut. Cancer Inst., 74,759-765 (1985). WOODWARD, M.P., YOUNG,JR., W.W. and BLOODGOOD, R.A., Detection of monoclonal antibodies specific for carbohydrate epitopes using periodate oxidation. J. immunol. Methods, 78, 143-153 (1985). Xu, M., REAL,F.X., WELT,S . , SCHUSSLER, M.H., OETTGEN,H.F. and OLD, L.J., Expression of TAG-72 in normal colon, transitional mucosa and colon cancer. Int. J. Cancer, 44, 985-989 (1989).

Oncofetal expression of the human intestinal mucin glycoprotein antigens in gastrointestinal epithelium defined by monoclonal antibodies.

A mucin preparation from a colonic adenocarcinoma was used to prepare monoclonal antibodies (MAbs) that reacted specifically either with normal adult ...
1MB Sizes 0 Downloads 0 Views