Immunol Cell Biol (1990) 68. 253-262
Production and characterization of a new monoclonal antibody to colorectal carcinoma Jin-Ghee Teh, Christopher H. Thompson and Ian F. C. McKenzie Research Centre for Cancer and Transplantation. Department of Pathology. University of Melbourne. Parkville. Vic. 3052. Australia (Submitted
5 February 1990. Accepted for publication
16 July
1990.)
Summary This study describes a new murine monoclonal antibody (MoAb) 5C1 raised against human colorectal carcinoma, which gave a differential reaction on formalin-fixed sections ofthe gastrointestinal tract. The MoAb 5CI of immunoglobulin M (IgM) isotype reacted with both the cytoplasm and membrane of all normal colonic cpithelia. and with all benign colonic polyps and all premalignant colonic lesions. However, there was a decreased expression ofthe 5C1 antigen in most cases of colonic malignancy and it was this feature that makes Mo,Ab 5CI unique. The distribution of the5CI epitope in normal gastrointestinal tract is limited to a few epithelial cells in the mid-portion of the small intestine but this distribution increased progressively down the digestive tract until it was found on > 90% of normal epithelial cells (in membrane and cytoplasm) ofthe colon. In addition, the 5CI epitope was present on mucin secreting cells from normal organs of the gastrointestinal, reproductive and pulmonary tract and benign amd malignant tissues ofthe colon. On Western blots, MoAb 5C1 was found to detect a heterogeneous population of molecules with molecular weights >IO0 kDa with the strongest staining bands found between 230 and 300 kDa. MoAb 5Ci does not detect carcino-embryonic antigens (CEA), human milk fat globules (HMFG). human lymphocyte antigens (HLA) or ABO blood group antigens. The combination of its presence in mucin secreting cells and its broad molecular weight bands suggest that the antigen detected is a mucin.
INTRODUCTION Most monoclonal antibodies (MoAb) raised against tumours recognize epitopes which are common to both normal and malignant cells (1), and to date no tumour 'specific" MoAb have been discovered for solid tumours. Despite this. these antibodies have been very useful. For example, MoAb raised against carcino-embryonic antigens (CEA) can identify tumour micrometastases in lymph nodes which could not be detected using routine haematoxylin and eosin (H&E)stains(2)yet CEA epitopes are present on some normal cells as normal cross-reactCorrespondence: J-G, Teh. Research Centre for Cancer and Transplantation, Department of Pathology. University of Melbourne, Parkville, Vic. 3052. Australia. Abbreviations u.sed in this paper: BSA, bovine serum albumin: CEA, carcino-embryonic antigen; H&E, haematoxylin and eosin; HMFG. human milk fat globules; MoAb, monoclonal antibody; NCA, normal cross-reacting antigen; PAS. periodic acid Schiff; PBS. phosphate buffered saline; PMN. polymorphonuclcar leucocytes; RIA. radioimmunoassay: i.p,, intraperitoneally: IgM, immunoglobulin M; HLA, human lymphocyte antigens.
ing antigens {NCA) and are not restricted to malignant tissues. Other investigators were able to identify malignant ceils in effusions (3-5) and in cerebrospinal fluid (6). It has also been possible to accurately classify some tumours into different subtypes (7-9). In this paper, we describe a new MoAb with reactivity to colorectal carcinoma, which could be used in the future to study the differentiation of colonic epithelium from the normal state to malignancy. The MoAb. 5CI recognizes an antigen which is strongly expressed on normal colon, but during transition to the malignant state its expression decreases and is in some cases lost completely. Due to its preferential reactivity with tissues of the lower gastrointestinal tract, it may also provide a useful means of distinguishing epithella from different parts of the intestine.
MATERIALS AND METHODS Cell tinesAll cell lines were grown as previously described (10). Adherent human cell lines used for immunoperoxidase studies were grown on sterile cover slips. washed, air dried, fixed in 10% formalin overnight and
254
}'G.TBH ETAL.
treated wilh 0-5% hydrogen peroxide in phosphate buffered saline (PBS) prior to use. Fresh tumour cell preparation A fresh primary colonic tumour was dissected from adjoining connective tissue and passed through a fine mclal sieve, resuspended in PBS. and passed through needles to obtain a single cell suspension: this was freeze-thawed several times and the crude preparation used for immunisation. Immunization and fu.sion procedure A (CBAXBALB/C)FI mouse was immunized twice intrapcritoneally (i.p.) with 10^ cells ofthe colon carcinoma cell line COLO .^97 (Dr George Moore. Denver. Colorado) and then with 2x10'' cells of the fresh tumour preparation given intravenously (twice) and spleen cells fused with the P3-NS-!-Ag4-I murine myeloma cells (II). Hybrids were screened 14 days after fusion by immunoperoxidase tests on sections of coionic carcinoma (frozen and formalin-fixed) before cloning by the limiting dilution method. MoAb 5C1 was found to be an igMi^ (Ouchterlony gel diffusion). Peripheral white blood celt preparation Whole blood was centrifugcd at 350i' for 10 min. the 'buffy coat" carefully removed, washed, resuspended in 0-83% ammonium chloride at 37°C to lysc erythroeytes and resuspended in PBS before use. Immunoperoxidase testing on tissue sections Tissue sections were tested by the immunoperoxidase technique (10). For each tissue examined by immunoperoxidase and each eell line tested by the rosetting assay, the following MoAb were used. MoAb 5Ci.MoAb 17C4 (positive control for CEA and NCA expressing tissues). MoAb 30'6 {negative control for all formalin-fixed human tissues; II) MoAb 3E1-2 (positive control MoAb for some tissues. 12) and a monomorphic anti-human lymphocyte antigen (anliHLA) MoAb (positive control for all human cell lines in rosetting assay). The percentage of cell reactivity was assessed by a single investigator (JGT) by determining the number of cells stained by MoAb 5C1 divided by the total number of tumour cells present. Serological assays A rosetting assay on large! cells was performed (13) using the same antibodies as used in the immunoperoxidase tests. Flow cytometry was also performed: 5x 10^ cells were incubated with 2 mL of MoAb 5C1. 17C4. anti-/3i microglobulin MoAb {R.L. Sparrow. unpublished data) and anti-Ly-2'l murine anligen MoAb (14) supernatant for 30 min at room temperature, washed three times; 100 ^L of 1/50 dilution of fluorescein conjugated sheep anti-mouse antibody in PBS was added (Silenus. Australia), and incubated for 30 min in the dark; after washing, the cells were resuspended in PBS and examined for fluorescence by a Flow Cytometer.
CEA test The reactivity of MoAh with CEA was tested using an assay (15) employing a commercial kit (Pharmacia. Uppsala, Sweden). One hundred microlitres of 70 ng CEA/mL were incubated with 100 ^L of '--''I labelled rabbit anti-CEA antiscra for 3 h at ambient temperature. Then 100 |JL of sheep anti-CEA antisera were added for 30 min. followed by 2 mL o^ horse antisheep antisera coupled to beads (solutions were supplied in the test kit) for 30 min. The beads of all assays were washed twice with PB.S and counted fur radioactivity. Thereactivity of 5C1 with CEA was determined by substituting the sheep anti-CEA and horse antisheep antisera coupled to beads with 100 JJL aliquots of packed CnBr-activated Sepharose beads coupled with MoAb 5C1 (2'4 mg MoAb/mL packed swollen beads). After 1 h incubation, the beads were washed twice with PBS and counted. MoAb 30-6(11) coupled to beads (2-4 mg MoAb/mL packed swollen beads) was used as a negative control and anti-CEA MoAh I 7C4. 0-1,1-1 and JGT-13 as positive controls (produced in our laboratory: JGT unpublished data). HMFG as.say Fifty microlitrcs/wcll of "crude' HMFG {10/jg/mL) (16) were coated to a 96-well PVC plate with carbonate-bicarbonate bufl'er pH 9-6 and 50 /iL/well 0-1% glutaraldehyde for 1 h at 37°C. After washing with 0-05"/!) Tween 20 in PBS and blocking with bovine serum albumin (BSA). 50 /jL hybridoma supernatant was added for 3h at 37°C. washed and 50/iL 1:200 dilution of sheep anti-mouse antibody conjugated to horseradish pcroxidase (HRP-SAMG. Amersham. UK)wasaddedfor2 hat 37°C After washing. 50/iLof 0-03% 2.2-azino-di-(3-ethylbenzlhiazoline sulfate) (ABTS. Amersham. UK) in pH 4 citrate buffer was added and the optical density measured with a Titretek multiscan automated plate reader. Electrophoresis and Western htots Polyacryiamide gel electrophoresis and immunoblott ing were performed as described (17.18). A crude membrane preparation of a fresh colorcctal carcinoma tissue specimen was obtained by homogenizing it in 0'25 moi/L sucrose. 25 mmol/L Tris and I mmol/L EDTA (pH 7-4) solution and this was used as the antigen in Western blots with 10% sodium dodecyt sulfate-potyacrylamide gel eleetrophoresis (SDSPAGE). Serum test An optimal concentration of purified MoAb 5C1 (determined from checkerboard type assays) was incubated with doubling dilutions (1/2 to 1/64) of 12 sera obtained from colonic cancer patients and 12 normal human sera in a BSA blocked plate for 3 h. After transferring the contents of each well to a plate coated with a membrane extract of a coiorectal carcinoma tissue, the plate was washed and developed with horseradish peroxidase conjugated to sheep anti-mouse antibody and ABTS as in the HMFO assay.
MONOCLONAL ANTIBODY TO COLONIC CANCER
RESULTS
2SS
I n the process of producing MoAb to coiorectal carcinomas a new antigenic determinant expressed by the tumour and normal cells was found. As will be shown, the importance of this antigenic detcrtninant was in its distribution on normal and non-malignant mucinous colonic epilhclial cells compared with malignant colonic tumour and premalignant colonic lesions.
specimens tested {Table I). It is important to note that >90%ofall normal coionic epithelial cells showed both cytoplasmic and membrane staining in the 70 tissues studied {Fig. I). MoAb 5C1 also reacted with the respiratory tract {mucous alveoli ofthe pharynx and trachea, and pseudostratified epithelium ofthe bronchioles) and with the reprodtictive system {cndotnetrial cells of the uterus and pseudostratified epithelium of the prostate ducts: Table 2).
Im munopero.xidasc lest on normal gastrointestinal tissttes A large panel of normal tissues was examined by the immunoperoxidase technique for their reaction with MoAb 5CI {Tables 1. 2). MoAb 5CI reacted with a subpopulation of mucin secreting cells in the gastrointestinal tract (mucous alveoli ofsali vary glands, epithelium ofthe gall hiaddcr. appendix, colon and rectum) and also in the cpitheliutn ofthe lower half of the small intestine, where the number of 5C1 positive cells increased from lO*!^) in the middle ofthe intestine to 100%at the ileocaecaljunction. From hereto the end ofthe colon and roctutn. normal epithelial cells were stained by MoAb 5CI in all 70
Immunopero.xida.se staining of henign and premalignant lesions ofthe colon The MoAb 5C1 reacted with all benign lesions of the colon (>90% of the cells were positive) where both the membrane and cytoplasm were stained {Table 3). These polyps were diagnosed as non-malignant by pathologists as the epithclia retained homogeneity and there were no signs of malignancy such as mitotic bodies, dysplasia. atypia. nuclei depolarization, or pieomorphism. in contrast to the coiorectal adenomas examined where two of six eases showed > 90% staining of the cytoplasm and three of six showed >90% membrane reactivity {Table 4). The important point to note was that there was decreased
Table I. Reactivity of MoAb 5C1 with the epithelium ofthe smail and large intestine by immunoperoxidase test. Distance (metres from MoAb 5C1* Distance (metres from MoAb 5CI* stomach) % cells staining Site ileocaecal region) % eells st;iining Site Small mtestinc
I 2 3 4 5
6 7
0 0 0
Appendix Colon
+ {IO%) + {40%) + {75%) + (>90%t
Vh 3
*Both cytoplasmic and membrane staining noted with all cells. Table 2. Reactivity of 5CI with normal human tissues using the immunoperoxidase and rosetting technique. Negative tissues^ (No. of individuals tested)
Positive tissues* {No. of individuals tested)
Mo.Ab 5C} Erythroeytes {10). lymphocytes (13). monoeytes (10), polymorphonuclear cells {10). liver (5). lymph nodes (4). spleen (3). kidney (5). pancreas (3). heart {3), urinary bladder {I), adrenal glands {5). parathyroid (5). skin (4). thyroid (6), breast (I), ovary {1). brain (4). bone (2). cartilage (2). umbilical cord (I), alveoli ( I I ) . stomach {2). skeletal muscle {I)
Lung {bronchioles. 6). nasopharynx {mucous glands, I), prostate (ducts 2). uterus (endometrium. 6), trachea (mucous glands. I), tongue (mucous alveoli. 2). oesophagus (mucous glands, 2), gall bladder {epithelium. 3). colon (epithelium. 70). appendix (epithelium 3), small intestine* {epithelium 4/7)**
* The rosetting assay and fluorescence activated cell sorting was used on erythroeytes, lymphocytes, monoeytes and polymorphs while the immunoperoxidase test was used on all other tissues. ^Non-reactivity in all component tissues (epithelium, cndothelium. connective tissues). ^Non-reactivity in all component tissues except for those stated in parentheses. ^Rcfcr to Table I for details. •"Denotes number of tissues positive/number of tissues tested.
}-G. TEH ETAL.
256
Table 4. Reactivities of MoAb 5C1 on premalignant colorectal tissues^
Tissue Colonic tubulovillous adenoma *sample 1 Colonic tubulovillous adenoma *sample 2 Colonic tubulovilious adenoma 'sample 3 Colonic tubulo adenoma Colonic villous adenoma Rectal villous papillonia
Reactivity (%of cells staining) Membrane Cytoplasm iOO
95
10
5
95
50 80 80 95
80 80 100
*Three diflereni individuals were examined.
^Tissues used were formalin-fixed.
expression ofthe 5CI antigen (as compared with normal and benign colonic growths) in four of six cases of colorectal adenomas with 90% of eells are stained). In the colon and rectum, >90% of epithelial celis are detected. The MoAb ID3 (19) is somewhat similar to MoAb 5C1 as it reacts with ail normal colonic and caeeal epithelia and somecoloreetal carcinomas and mucin secreting carcinomas of other organs; however, unlike 5C1, it does not react with any normal tissues other than colonic and caeeal epithelia, whereas 5C1 detects most mucin secreting cells of other organs. From Westem blots, it was found that the 5C1 antigen detects a heterogeneous population of molecules > 100 kDa with the strongest staining bands between 260 and 300 kDa. The broad range of molecular weights ofthe 5C1 antigen combined with its reactivity with mucin secreting cells strongly suggest that it detects a mucinassociated epitope. Other investigators (20) have also found that most intestinal mucin do not penetrate the SDS-PAGE gel before reduc-
tion and on reduction results in a broad smear as seen in Fig. 3. On the other hand, periodate oxidation and borohydride reduction of formalinfixed tissue sections, which abolished periodic acid Sehiff (PAS) mucin staining through removal of 0-linked sugars (21) has no etlect on the 5CI staining (results not shown). In separate experiments, formalin-fixed sections of colorectal carcinoma, salivary glands of the tongue and submucosal glands of the trachea were first stained with MoAb 5C1, and then followed by the mucin stains of PAS, Aleian Blue and Muciearmine. There were cells in each tissue reactive with MoAb 5C1 but not with any ofthe mucin stains. The 5C1 reactive epitope is not present on CEA or HMFG molecules nor on HLA or ABO blood group antigens. The lack of reactivity of 5C1 with any colonic cancer cell lines by the rosetting and immunoperoxidase techniques suggests that expression ofthe 5C1 antigen is decreased and lost in neoplasia. This was first observed when 12 of 50 colorectal carcinomas were not stained in the cytoplasm and six of 50 were not stained on the cytoplasm or membrane, while all normal, benign and premalignant colonic tissues were positive at the membrane and cytoplasm (79 of 79) with MoAb 5C1. It is further apparent as all (70 of 70) normal colonic epithelia showed
MONOCLONAL ANTIBODY TO COLONIC CANCER
>90% Staining whereas only 12 of 50 coiorectal carcinomas exhibited such reactivity, that is 38 of 50 had 90% reactivity, compared with only two of six premalignant colonic adenomas with MoAb 5C1. These adenomas showed signs of atypia, dysplasia, nuclei depolarization, pieomorphism heterogeneity and hyperchromatism associated with tumours. This is further substantiated by its lack of reactivity with colon cell lines. As a result of finding the antigen in extracellular secretions of colonic tumours by the immunoperoxidase assay, a serum assay was developed to detect its presence in the circulation of patients with colonic tumours as compared with normals. It was found that MoAb 5Cl was unable to discriminate between the two groups.
261
possibly due to secretion ofthe 5CI antigen into the circulation by normal organs such as the gastrointestinal, reproductive and pulmonary tract, or its degradation by enzymes in the circulation. The 5CI antigen which is normally present in colonic epithelial cells had decreased expression in malignancies and premalignancies, and in some cases, was completely absent. This has been shown for other tumours, for example melanomas, as other investigators have described tumour associated antigens with decreased expression in the course of local and systemic tumour growth as well as those which increase in proportion to invasiveness and metastatic properties (22). Acknowledgements The authors acktiowledge the support ofthe Medical Engineering Research Association and the assistance of Ms Toula Athanasiadis.
REFERENCES 1. Teh. J. G.. Stacker, S. A.. Thompson. C. H. and McKenzie, I. F. C. 1985. The diagnosis of human tumors with monoclonal antibodies. Cancer Surveys 4: 149-184. 2. Hockey. M. S., Stokes. H. J., Thompson, H., Woodhouse, C. S., Maedonald, F.. Fielding. J. W. L. and Ford, C. H. J. 1984. Carcinoemhryonie antigen (CEA) expression atid heterogeneity in primary and aulologous metastatic gastric tumours demonstrated by a monoclonal antibody. Brit. J. Cancer 49: 129-133. 3. Szapak. C. A., Johnston, W. W., Lotich, S. C , Kufe, D., Thor. A. and Schlom, J. 1984. Patterns of reactivity of four novel monoclonal antibodies (B72.3, DF3, Bl.l and B6.2) with cells in human malignant and benign fusions. Acta Cytologica 28: 356-367. 4. Epenetos, A. A.. Canti, G.. Taylor-Papadimitriou, J., Curling, M. and Bodmer, W. F. 1982. Use of two epithelium specific monoclonal antibodies for diagnosis of malignaney in serious effusions. Lancet ii: 1004-1006. 5. Woods,J.C.,Spriggs. A. 1., Harris, H. and MeGee. J. O'D. 1982. immunocytochemical detection of malignant cells in serous lluids with the CA-I antibody. Lancet ii: 512-515. 6. Coakham. H. B., Garson, J. A., Brownell. B., Allan. P. M.. Harper. E. I. and Lane. E. B. 1984. Use of monticlonal antibody panel to identify malignant eells in eerebrospinal fluid, Lancet i: 1095-1098. 7. Gaiter. K. C . Abdulaziz. Z., Beverley, P. et al. 1982. Use of monoclonal antibodies for the histopathologieal diagnosis of human malignancy./ Clin. Pathol. 35: 1253-1267. 8. Warnke.R. A..Gatter, K. CFalini.B.crfl/. 1983. Diagnosis of human lymphoma with monoelonal
9.
10.
11.
12.
13.
14.
15.
16.
17.
anti-leukocyte antibodies. N. Engl. J. Med. 309: 1275-1281. Battifora. H. and Trowbridge, I. S. 1983. A monoclonal antibody useful for the differential diagnosis between malignant lymphoma and non-hematopoietic neoplasms. Cancer 51; 816-821. Thompson, C. H., Jones, S. L.. PihI. E. and McKenzie, I. F. C. 1983. A human breast tissue associated antigen detected by a monoclonal antibody. ./. Natl Cancer Inst. 70: 409-419. Thompson, C. H., Jones. S. L.. Pihl, E. and MeKcnzie, I. F. C. 1983. Monoclonal antibodies to human eolon and eolorectal eareinoma. Brit. J. Cancer 47: 595-605. Stacker. S. A., Thompson. C. H.. Riglar. C. and MeKenzie, I. F. C. 1985. A new breast carcinoma antigen defined by a monoclonal antibody. J. Natl Cancer Inst. 75: 801-809. Parish, C. R. and McKenzie, I. F. C. 1978. A sensitive rosetting tnethod for detecting subpopulations of lymphoeytes whieh react with alloantisera. / Immunol. Methods 20: 173-18.1. Hogarth, P. M., Edwards, J., McKenzie, I. F. C , Goding. J. W. and Liew. F. J. 1982. Monoclonal antibodies to murine cell surface antigens. II. Ly-2'l. Immunology 46: 135-144. Teh, J. G., Thompson, C. H. and MeKenzie, I, F. C. 1988. Produetion of monoclonal antibodies to serum antigens in eoiorectal carcinoma. / Immunol. Methods 110: 101-109. Jarasch, E. D., Bruder, G.. Keenan, T. W. and Franke. W. W. 1977. Redox constituents in milk fat globule membranes of membranes of rough endoplasmie reticulum from laetating mammary gland. J. Cell Biol. 73: 223-241. Laemmli, U. K. 1970. Cleavage of structural
262
J-G. TEH ETAL.
proteins during the assembly of the head of bacteriophage T4. Nature 227: 680-685. 18. Towbin, H.. Staehelin, Y. and Gordon, J. 1979. Elcctrophorctic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: proeedure and some applications. Proc. Nail.4cad. Sci. USA 76: 4350-4354. 19. Gangopadhyay. A.. Bhattacharya, M., Chatterjee, S. K.. Barlow. J. J. and Tsukada. Y. 1985. Immunopcroxidase localization of a high molecular weight mucin recognized by monoclonal antibody ID3. Cancer Res. 45: 1744-1752. 20. Mantle, M., Forstner, G. G. and Forstner. J. F.
1984, Biochemical characterization of the component parts of intestinal mucin from patients with cystic fibrosis. Biochem. J. 114: 345-354. 21. Culling, C. F. A., Reid, P. E.. Burton. J. O. and Dunn, W, L. 1975. A histochemical method of differentiating lower gastrointestinal tract mucin from other mucins in primary or metastatic tumours../. Clin. Pathol. 28: 656-658. 22. Brocker, E. B., Suter, L., Bruggcn J . , Ruiter, D. J.. Macher, E. and Sorg, C. 1985. Phenotypic dynamics of tumor progression in human malignant melanoma. Int. J. Cancer^: 29-35.
Production and characterization of a new monoclonal antibody to colorectal carcinoma Jin-Ghee Teh, Christopher H. Thompson and Ian F. C. McKenzie Research Centre for Cancer and Transplantation. Department of Pathology. University of Melbourne. Parkville. Vic. 3052. Australia (Submitted
5 February 1990. Accepted for publication
16 July
1990.)
Summary This study describes a new murine monoclonal antibody (MoAb) 5C1 raised against human colorectal carcinoma, which gave a differential reaction on formalin-fixed sections ofthe gastrointestinal tract. The MoAb 5CI of immunoglobulin M (IgM) isotype reacted with both the cytoplasm and membrane of all normal colonic cpithelia. and with all benign colonic polyps and all premalignant colonic lesions. However, there was a decreased expression ofthe 5C1 antigen in most cases of colonic malignancy and it was this feature that makes Mo,Ab 5CI unique. The distribution of the5CI epitope in normal gastrointestinal tract is limited to a few epithelial cells in the mid-portion of the small intestine but this distribution increased progressively down the digestive tract until it was found on > 90% of normal epithelial cells (in membrane and cytoplasm) ofthe colon. In addition, the 5CI epitope was present on mucin secreting cells from normal organs of the gastrointestinal, reproductive and pulmonary tract and benign amd malignant tissues ofthe colon. On Western blots, MoAb 5C1 was found to detect a heterogeneous population of molecules with molecular weights >IO0 kDa with the strongest staining bands found between 230 and 300 kDa. MoAb 5Ci does not detect carcino-embryonic antigens (CEA), human milk fat globules (HMFG). human lymphocyte antigens (HLA) or ABO blood group antigens. The combination of its presence in mucin secreting cells and its broad molecular weight bands suggest that the antigen detected is a mucin.
INTRODUCTION Most monoclonal antibodies (MoAb) raised against tumours recognize epitopes which are common to both normal and malignant cells (1), and to date no tumour 'specific" MoAb have been discovered for solid tumours. Despite this. these antibodies have been very useful. For example, MoAb raised against carcino-embryonic antigens (CEA) can identify tumour micrometastases in lymph nodes which could not be detected using routine haematoxylin and eosin (H&E)stains(2)yet CEA epitopes are present on some normal cells as normal cross-reactCorrespondence: J-G, Teh. Research Centre for Cancer and Transplantation, Department of Pathology. University of Melbourne, Parkville, Vic. 3052. Australia. Abbreviations u.sed in this paper: BSA, bovine serum albumin: CEA, carcino-embryonic antigen; H&E, haematoxylin and eosin; HMFG. human milk fat globules; MoAb, monoclonal antibody; NCA, normal cross-reacting antigen; PAS. periodic acid Schiff; PBS. phosphate buffered saline; PMN. polymorphonuclcar leucocytes; RIA. radioimmunoassay: i.p,, intraperitoneally: IgM, immunoglobulin M; HLA, human lymphocyte antigens.
ing antigens {NCA) and are not restricted to malignant tissues. Other investigators were able to identify malignant ceils in effusions (3-5) and in cerebrospinal fluid (6). It has also been possible to accurately classify some tumours into different subtypes (7-9). In this paper, we describe a new MoAb with reactivity to colorectal carcinoma, which could be used in the future to study the differentiation of colonic epithelium from the normal state to malignancy. The MoAb. 5CI recognizes an antigen which is strongly expressed on normal colon, but during transition to the malignant state its expression decreases and is in some cases lost completely. Due to its preferential reactivity with tissues of the lower gastrointestinal tract, it may also provide a useful means of distinguishing epithella from different parts of the intestine.
MATERIALS AND METHODS Cell tinesAll cell lines were grown as previously described (10). Adherent human cell lines used for immunoperoxidase studies were grown on sterile cover slips. washed, air dried, fixed in 10% formalin overnight and
254
}'G.TBH ETAL.
treated wilh 0-5% hydrogen peroxide in phosphate buffered saline (PBS) prior to use. Fresh tumour cell preparation A fresh primary colonic tumour was dissected from adjoining connective tissue and passed through a fine mclal sieve, resuspended in PBS. and passed through needles to obtain a single cell suspension: this was freeze-thawed several times and the crude preparation used for immunisation. Immunization and fu.sion procedure A (CBAXBALB/C)FI mouse was immunized twice intrapcritoneally (i.p.) with 10^ cells ofthe colon carcinoma cell line COLO .^97 (Dr George Moore. Denver. Colorado) and then with 2x10'' cells of the fresh tumour preparation given intravenously (twice) and spleen cells fused with the P3-NS-!-Ag4-I murine myeloma cells (II). Hybrids were screened 14 days after fusion by immunoperoxidase tests on sections of coionic carcinoma (frozen and formalin-fixed) before cloning by the limiting dilution method. MoAb 5C1 was found to be an igMi^ (Ouchterlony gel diffusion). Peripheral white blood celt preparation Whole blood was centrifugcd at 350i' for 10 min. the 'buffy coat" carefully removed, washed, resuspended in 0-83% ammonium chloride at 37°C to lysc erythroeytes and resuspended in PBS before use. Immunoperoxidase testing on tissue sections Tissue sections were tested by the immunoperoxidase technique (10). For each tissue examined by immunoperoxidase and each eell line tested by the rosetting assay, the following MoAb were used. MoAb 5Ci.MoAb 17C4 (positive control for CEA and NCA expressing tissues). MoAb 30'6 {negative control for all formalin-fixed human tissues; II) MoAb 3E1-2 (positive control MoAb for some tissues. 12) and a monomorphic anti-human lymphocyte antigen (anliHLA) MoAb (positive control for all human cell lines in rosetting assay). The percentage of cell reactivity was assessed by a single investigator (JGT) by determining the number of cells stained by MoAb 5C1 divided by the total number of tumour cells present. Serological assays A rosetting assay on large! cells was performed (13) using the same antibodies as used in the immunoperoxidase tests. Flow cytometry was also performed: 5x 10^ cells were incubated with 2 mL of MoAb 5C1. 17C4. anti-/3i microglobulin MoAb {R.L. Sparrow. unpublished data) and anti-Ly-2'l murine anligen MoAb (14) supernatant for 30 min at room temperature, washed three times; 100 ^L of 1/50 dilution of fluorescein conjugated sheep anti-mouse antibody in PBS was added (Silenus. Australia), and incubated for 30 min in the dark; after washing, the cells were resuspended in PBS and examined for fluorescence by a Flow Cytometer.
CEA test The reactivity of MoAh with CEA was tested using an assay (15) employing a commercial kit (Pharmacia. Uppsala, Sweden). One hundred microlitres of 70 ng CEA/mL were incubated with 100 ^L of '--''I labelled rabbit anti-CEA antiscra for 3 h at ambient temperature. Then 100 |JL of sheep anti-CEA antisera were added for 30 min. followed by 2 mL o^ horse antisheep antisera coupled to beads (solutions were supplied in the test kit) for 30 min. The beads of all assays were washed twice with PB.S and counted fur radioactivity. Thereactivity of 5C1 with CEA was determined by substituting the sheep anti-CEA and horse antisheep antisera coupled to beads with 100 JJL aliquots of packed CnBr-activated Sepharose beads coupled with MoAb 5C1 (2'4 mg MoAb/mL packed swollen beads). After 1 h incubation, the beads were washed twice with PBS and counted. MoAb 30-6(11) coupled to beads (2-4 mg MoAb/mL packed swollen beads) was used as a negative control and anti-CEA MoAh I 7C4. 0-1,1-1 and JGT-13 as positive controls (produced in our laboratory: JGT unpublished data). HMFG as.say Fifty microlitrcs/wcll of "crude' HMFG {10/jg/mL) (16) were coated to a 96-well PVC plate with carbonate-bicarbonate bufl'er pH 9-6 and 50 /iL/well 0-1% glutaraldehyde for 1 h at 37°C. After washing with 0-05"/!) Tween 20 in PBS and blocking with bovine serum albumin (BSA). 50 /jL hybridoma supernatant was added for 3h at 37°C. washed and 50/iL 1:200 dilution of sheep anti-mouse antibody conjugated to horseradish pcroxidase (HRP-SAMG. Amersham. UK)wasaddedfor2 hat 37°C After washing. 50/iLof 0-03% 2.2-azino-di-(3-ethylbenzlhiazoline sulfate) (ABTS. Amersham. UK) in pH 4 citrate buffer was added and the optical density measured with a Titretek multiscan automated plate reader. Electrophoresis and Western htots Polyacryiamide gel electrophoresis and immunoblott ing were performed as described (17.18). A crude membrane preparation of a fresh colorcctal carcinoma tissue specimen was obtained by homogenizing it in 0'25 moi/L sucrose. 25 mmol/L Tris and I mmol/L EDTA (pH 7-4) solution and this was used as the antigen in Western blots with 10% sodium dodecyt sulfate-potyacrylamide gel eleetrophoresis (SDSPAGE). Serum test An optimal concentration of purified MoAb 5C1 (determined from checkerboard type assays) was incubated with doubling dilutions (1/2 to 1/64) of 12 sera obtained from colonic cancer patients and 12 normal human sera in a BSA blocked plate for 3 h. After transferring the contents of each well to a plate coated with a membrane extract of a coiorectal carcinoma tissue, the plate was washed and developed with horseradish peroxidase conjugated to sheep anti-mouse antibody and ABTS as in the HMFO assay.
MONOCLONAL ANTIBODY TO COLONIC CANCER
RESULTS
2SS
I n the process of producing MoAb to coiorectal carcinomas a new antigenic determinant expressed by the tumour and normal cells was found. As will be shown, the importance of this antigenic detcrtninant was in its distribution on normal and non-malignant mucinous colonic epilhclial cells compared with malignant colonic tumour and premalignant colonic lesions.
specimens tested {Table I). It is important to note that >90%ofall normal coionic epithelial cells showed both cytoplasmic and membrane staining in the 70 tissues studied {Fig. I). MoAb 5C1 also reacted with the respiratory tract {mucous alveoli ofthe pharynx and trachea, and pseudostratified epithelium ofthe bronchioles) and with the reprodtictive system {cndotnetrial cells of the uterus and pseudostratified epithelium of the prostate ducts: Table 2).
Im munopero.xidasc lest on normal gastrointestinal tissttes A large panel of normal tissues was examined by the immunoperoxidase technique for their reaction with MoAb 5CI {Tables 1. 2). MoAb 5CI reacted with a subpopulation of mucin secreting cells in the gastrointestinal tract (mucous alveoli ofsali vary glands, epithelium ofthe gall hiaddcr. appendix, colon and rectum) and also in the cpitheliutn ofthe lower half of the small intestine, where the number of 5C1 positive cells increased from lO*!^) in the middle ofthe intestine to 100%at the ileocaecaljunction. From hereto the end ofthe colon and roctutn. normal epithelial cells were stained by MoAb 5CI in all 70
Immunopero.xida.se staining of henign and premalignant lesions ofthe colon The MoAb 5C1 reacted with all benign lesions of the colon (>90% of the cells were positive) where both the membrane and cytoplasm were stained {Table 3). These polyps were diagnosed as non-malignant by pathologists as the epithclia retained homogeneity and there were no signs of malignancy such as mitotic bodies, dysplasia. atypia. nuclei depolarization, or pieomorphism. in contrast to the coiorectal adenomas examined where two of six eases showed > 90% staining of the cytoplasm and three of six showed >90% membrane reactivity {Table 4). The important point to note was that there was decreased
Table I. Reactivity of MoAb 5C1 with the epithelium ofthe smail and large intestine by immunoperoxidase test. Distance (metres from MoAb 5C1* Distance (metres from MoAb 5CI* stomach) % cells staining Site ileocaecal region) % eells st;iining Site Small mtestinc
I 2 3 4 5
6 7
0 0 0
Appendix Colon
+ {IO%) + {40%) + {75%) + (>90%t
Vh 3
*Both cytoplasmic and membrane staining noted with all cells. Table 2. Reactivity of 5CI with normal human tissues using the immunoperoxidase and rosetting technique. Negative tissues^ (No. of individuals tested)
Positive tissues* {No. of individuals tested)
Mo.Ab 5C} Erythroeytes {10). lymphocytes (13). monoeytes (10), polymorphonuclear cells {10). liver (5). lymph nodes (4). spleen (3). kidney (5). pancreas (3). heart {3), urinary bladder {I), adrenal glands {5). parathyroid (5). skin (4). thyroid (6), breast (I), ovary {1). brain (4). bone (2). cartilage (2). umbilical cord (I), alveoli ( I I ) . stomach {2). skeletal muscle {I)
Lung {bronchioles. 6). nasopharynx {mucous glands, I), prostate (ducts 2). uterus (endometrium. 6), trachea (mucous glands. I), tongue (mucous alveoli. 2). oesophagus (mucous glands, 2), gall bladder {epithelium. 3). colon (epithelium. 70). appendix (epithelium 3), small intestine* {epithelium 4/7)**
* The rosetting assay and fluorescence activated cell sorting was used on erythroeytes, lymphocytes, monoeytes and polymorphs while the immunoperoxidase test was used on all other tissues. ^Non-reactivity in all component tissues (epithelium, cndothelium. connective tissues). ^Non-reactivity in all component tissues except for those stated in parentheses. ^Rcfcr to Table I for details. •"Denotes number of tissues positive/number of tissues tested.
}-G. TEH ETAL.
256
Table 4. Reactivities of MoAb 5C1 on premalignant colorectal tissues^
Tissue Colonic tubulovillous adenoma *sample 1 Colonic tubulovillous adenoma *sample 2 Colonic tubulovilious adenoma 'sample 3 Colonic tubulo adenoma Colonic villous adenoma Rectal villous papillonia
Reactivity (%of cells staining) Membrane Cytoplasm iOO
95
10
5
95
50 80 80 95
80 80 100
*Three diflereni individuals were examined.
^Tissues used were formalin-fixed.
expression ofthe 5CI antigen (as compared with normal and benign colonic growths) in four of six cases of colorectal adenomas with 90% of eells are stained). In the colon and rectum, >90% of epithelial celis are detected. The MoAb ID3 (19) is somewhat similar to MoAb 5C1 as it reacts with ail normal colonic and caeeal epithelia and somecoloreetal carcinomas and mucin secreting carcinomas of other organs; however, unlike 5C1, it does not react with any normal tissues other than colonic and caeeal epithelia, whereas 5C1 detects most mucin secreting cells of other organs. From Westem blots, it was found that the 5C1 antigen detects a heterogeneous population of molecules > 100 kDa with the strongest staining bands between 260 and 300 kDa. The broad range of molecular weights ofthe 5C1 antigen combined with its reactivity with mucin secreting cells strongly suggest that it detects a mucinassociated epitope. Other investigators (20) have also found that most intestinal mucin do not penetrate the SDS-PAGE gel before reduc-
tion and on reduction results in a broad smear as seen in Fig. 3. On the other hand, periodate oxidation and borohydride reduction of formalinfixed tissue sections, which abolished periodic acid Sehiff (PAS) mucin staining through removal of 0-linked sugars (21) has no etlect on the 5CI staining (results not shown). In separate experiments, formalin-fixed sections of colorectal carcinoma, salivary glands of the tongue and submucosal glands of the trachea were first stained with MoAb 5C1, and then followed by the mucin stains of PAS, Aleian Blue and Muciearmine. There were cells in each tissue reactive with MoAb 5C1 but not with any ofthe mucin stains. The 5C1 reactive epitope is not present on CEA or HMFG molecules nor on HLA or ABO blood group antigens. The lack of reactivity of 5C1 with any colonic cancer cell lines by the rosetting and immunoperoxidase techniques suggests that expression ofthe 5C1 antigen is decreased and lost in neoplasia. This was first observed when 12 of 50 colorectal carcinomas were not stained in the cytoplasm and six of 50 were not stained on the cytoplasm or membrane, while all normal, benign and premalignant colonic tissues were positive at the membrane and cytoplasm (79 of 79) with MoAb 5C1. It is further apparent as all (70 of 70) normal colonic epithelia showed
MONOCLONAL ANTIBODY TO COLONIC CANCER
>90% Staining whereas only 12 of 50 coiorectal carcinomas exhibited such reactivity, that is 38 of 50 had 90% reactivity, compared with only two of six premalignant colonic adenomas with MoAb 5C1. These adenomas showed signs of atypia, dysplasia, nuclei depolarization, pieomorphism heterogeneity and hyperchromatism associated with tumours. This is further substantiated by its lack of reactivity with colon cell lines. As a result of finding the antigen in extracellular secretions of colonic tumours by the immunoperoxidase assay, a serum assay was developed to detect its presence in the circulation of patients with colonic tumours as compared with normals. It was found that MoAb 5Cl was unable to discriminate between the two groups.
261
possibly due to secretion ofthe 5CI antigen into the circulation by normal organs such as the gastrointestinal, reproductive and pulmonary tract, or its degradation by enzymes in the circulation. The 5CI antigen which is normally present in colonic epithelial cells had decreased expression in malignancies and premalignancies, and in some cases, was completely absent. This has been shown for other tumours, for example melanomas, as other investigators have described tumour associated antigens with decreased expression in the course of local and systemic tumour growth as well as those which increase in proportion to invasiveness and metastatic properties (22). Acknowledgements The authors acktiowledge the support ofthe Medical Engineering Research Association and the assistance of Ms Toula Athanasiadis.
REFERENCES 1. Teh. J. G.. Stacker, S. A.. Thompson. C. H. and McKenzie, I. F. C. 1985. The diagnosis of human tumors with monoclonal antibodies. Cancer Surveys 4: 149-184. 2. Hockey. M. S., Stokes. H. J., Thompson, H., Woodhouse, C. S., Maedonald, F.. Fielding. J. W. L. and Ford, C. H. J. 1984. Carcinoemhryonie antigen (CEA) expression atid heterogeneity in primary and aulologous metastatic gastric tumours demonstrated by a monoclonal antibody. Brit. J. Cancer 49: 129-133. 3. Szapak. C. A., Johnston, W. W., Lotich, S. C , Kufe, D., Thor. A. and Schlom, J. 1984. Patterns of reactivity of four novel monoclonal antibodies (B72.3, DF3, Bl.l and B6.2) with cells in human malignant and benign fusions. Acta Cytologica 28: 356-367. 4. Epenetos, A. A.. Canti, G.. Taylor-Papadimitriou, J., Curling, M. and Bodmer, W. F. 1982. Use of two epithelium specific monoclonal antibodies for diagnosis of malignaney in serious effusions. Lancet ii: 1004-1006. 5. Woods,J.C.,Spriggs. A. 1., Harris, H. and MeGee. J. O'D. 1982. immunocytochemical detection of malignant cells in serous lluids with the CA-I antibody. Lancet ii: 512-515. 6. Coakham. H. B., Garson, J. A., Brownell. B., Allan. P. M.. Harper. E. I. and Lane. E. B. 1984. Use of monticlonal antibody panel to identify malignant eells in eerebrospinal fluid, Lancet i: 1095-1098. 7. Gaiter. K. C . Abdulaziz. Z., Beverley, P. et al. 1982. Use of monoclonal antibodies for the histopathologieal diagnosis of human malignancy./ Clin. Pathol. 35: 1253-1267. 8. Warnke.R. A..Gatter, K. CFalini.B.crfl/. 1983. Diagnosis of human lymphoma with monoelonal
9.
10.
11.
12.
13.
14.
15.
16.
17.
anti-leukocyte antibodies. N. Engl. J. Med. 309: 1275-1281. Battifora. H. and Trowbridge, I. S. 1983. A monoclonal antibody useful for the differential diagnosis between malignant lymphoma and non-hematopoietic neoplasms. Cancer 51; 816-821. Thompson, C. H., Jones, S. L.. PihI. E. and McKenzie, I. F. C. 1983. A human breast tissue associated antigen detected by a monoclonal antibody. ./. Natl Cancer Inst. 70: 409-419. Thompson, C. H., Jones. S. L.. Pihl, E. and MeKcnzie, I. F. C. 1983. Monoclonal antibodies to human eolon and eolorectal eareinoma. Brit. J. Cancer 47: 595-605. Stacker. S. A., Thompson. C. H.. Riglar. C. and MeKenzie, I. F. C. 1985. A new breast carcinoma antigen defined by a monoclonal antibody. J. Natl Cancer Inst. 75: 801-809. Parish, C. R. and McKenzie, I. F. C. 1978. A sensitive rosetting tnethod for detecting subpopulations of lymphoeytes whieh react with alloantisera. / Immunol. Methods 20: 173-18.1. Hogarth, P. M., Edwards, J., McKenzie, I. F. C , Goding. J. W. and Liew. F. J. 1982. Monoclonal antibodies to murine cell surface antigens. II. Ly-2'l. Immunology 46: 135-144. Teh, J. G., Thompson, C. H. and MeKenzie, I, F. C. 1988. Produetion of monoclonal antibodies to serum antigens in eoiorectal carcinoma. / Immunol. Methods 110: 101-109. Jarasch, E. D., Bruder, G.. Keenan, T. W. and Franke. W. W. 1977. Redox constituents in milk fat globule membranes of membranes of rough endoplasmie reticulum from laetating mammary gland. J. Cell Biol. 73: 223-241. Laemmli, U. K. 1970. Cleavage of structural
262
J-G. TEH ETAL.
proteins during the assembly of the head of bacteriophage T4. Nature 227: 680-685. 18. Towbin, H.. Staehelin, Y. and Gordon, J. 1979. Elcctrophorctic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: proeedure and some applications. Proc. Nail.4cad. Sci. USA 76: 4350-4354. 19. Gangopadhyay. A.. Bhattacharya, M., Chatterjee, S. K.. Barlow. J. J. and Tsukada. Y. 1985. Immunopcroxidase localization of a high molecular weight mucin recognized by monoclonal antibody ID3. Cancer Res. 45: 1744-1752. 20. Mantle, M., Forstner, G. G. and Forstner. J. F.
1984, Biochemical characterization of the component parts of intestinal mucin from patients with cystic fibrosis. Biochem. J. 114: 345-354. 21. Culling, C. F. A., Reid, P. E.. Burton. J. O. and Dunn, W, L. 1975. A histochemical method of differentiating lower gastrointestinal tract mucin from other mucins in primary or metastatic tumours../. Clin. Pathol. 28: 656-658. 22. Brocker, E. B., Suter, L., Bruggcn J . , Ruiter, D. J.. Macher, E. and Sorg, C. 1985. Phenotypic dynamics of tumor progression in human malignant melanoma. Int. J. Cancer^: 29-35.
