WO1989004841A1 - Gastrointestinal tumor-associated glycoprotein expressing the ca 19-9 antigenic determinant - Google Patents

Abstract

A human epithelial cell glycoprotein population with an average molecular weight of about 210 kDa and a relatively homogeneous size distribution, each glycoprotein in the population containing the CA 19-9 antigenic determinant. The population can be generated from larger molecular entities by exposing them, in the presence of a chaotropic agent, to a reducing agent and an alkylating agent.

Description

GASTROINTESTINAL TUMOR-ASSOCIATED GLYCOPROTEIN EXPRESSING THE CA 19-9 ANTIGENIC DETERMINANT

Background

The molecular antigenic determinant, CA 19-9, is the basis of an important diagnostic tool as its concentration level is usually elevated in the serum of patients with colorectal and/or pancreatic cancers. As a result, measurements of CA 19-9 serum levels are useful in the differential diagnosis of patients suspected of having one of these cancers. Measurement of serum CA 19-9 levels is achieved by using antbodies that specifically bind to it. These antibodies are produced in the laboratory by hybridomas, cloned cell lines in which each cell is derived from a hybrid of a mouse myeloma cell and a spleen cell from a mouse that had been immunized with human colon carcinoma cell line, that produces the CA 19-9 determinant. If, however, a purified molecular species carrying the CA 19-9 determinant could be isolated, then one would have the option of immunizing mice with that species instead of with entire cells. Furthermore, one could use that molecular species as a standard in the immunoassay used to detect the CA 19-9 determinant in sera. Indeed, one could also use it as a probe in a test that is the converse of the one for CA 19-9 antigen : a test for the presence of anti-CA 19-9 antibody in humans .

It has , in fact , been established by others that human carcinoma cells shed relatively large molecular entities , with a size distribution in the range of about 200 , 000 daltons ( 200 kDa) to 2000 kDa . These entities were shown to contain both protein and the carbohydrate group , sialylated lacto-N-fucopentaose II , whose sial ic acid and fucose moieties are both required for antibody binding . Whether these 200 to 2000 kDa entities were aggregates that could be dissociated into smaller ones that carr ied the CA 19 - 9 de terminant was not known , however . Simi larly , it was not known whether such smaller unaggregated entities , if producible , would all be the same or vary as regards their structure. We have discovered that these high molecular weight entities are indeed aggregates that can be dissociated to release a relatively homogeneous population of unaggregated glycoproteins with an average molecular weight of about 210 kDa . Af ter dissociation of the high molecular weight entities , the glycoproteins within the 210 kDa population are apparently the only molecular species to carry the CA 19-9 determinant .

Of historical interest is the fact that the information on the structure of other tumor cell high molecular weight entities is sparse . Ovarian epithelial tumor-associated high molecular entities can be dissociated into one antigenic determinantcarrying glycoprotein species of 200 kDa (H. M. Davis et al . Cancer Research. 46 , 6143-6148 , 1986) . That appears to be the situation most similar to the present invention. Even in that case , however , there are apparently fundamental differences in the structures of the aggregates, as compared to the present case : For one thing , the ovarian antigenic determinant is protein only. Additionally, in the present case , destabilization o f natural ly o c cur ing aggegates to produce the 2 10 kDa glycoprotein in either an unaggregated state or as a detergentdissociable aggregate requires reduction and carboxymethylation in the presence of a chaotropic agent. In the case of the ovarian tumor, destabilization of the naturally occuring aggregates could be achieved by exposing them to a chaotropic agent without concomitant reduction and carboxymethylation. Indeed, although reduction and alkylation in the presence of a chaotropic agent had been reported previously, it had not been reported as a means of producing a tumor-associated antigen.

Summary of the Invention

The glycoprotein population of the present invention is a human epithelial glycoprotein population in which:

(i) each glycoprotein contains the CA 19-9 antigenic determinant; (ii) as measured by sodium dodecyl sulfate electrophoresis, the average mobility is that of a protein of about 210 kDa and the mobilities vary from that of a protein of about 150 kDa to one of about 250 kDa;

(iii) on the average, is about 85% carbohydrate and about 15% protein by weight;

(iv) has an amino acid composition, quantitated as phenyl isothiocyanate derivatives in mole % in parentheses, as follows: asp (5.1), glu (9.8), ser (19.7). Gly (8.4), his (1.5), arg (5.5) thr(8.8), ala (7.5), pro (7.3), tyr (2.5), val (5.1), met (0.9), ile (3.4), leu (5.6), phe (2.7), lys (4.1); and

(v) has a carbohydrate composition, quantitated as trimethylsilyl derivatives in molar ratios in parentheses, as follows: fucose (4.1), mannose (1.0), galactose (11.8), N-acetylgalactosamine (2.5) , N-acetylglucosamine (4.9) , N-acetylneuraminic acid (5.1); and each glycoprotein is either in an unaggregated state or a zwitterionic detergent dissociable state.

In another aspect, the invention is a process for facilitating the production of a glycoprotein population of the present invention which comprises the steps of:

1) using human epithelial cells to produce entities with an average molecular mass greater than that of said population and containing the CA 19-9 determinant;

2) dissociating said entities by treating them with a sulfide-sulfide bond reducing agent in the presence of a chaotropic agent.

The glycoprotein population of the present invention is useful as a reagent for immunizing mice in order to generate mouse spleen cells that can be used to make hybridomas that produce antibodies against the CA 19-9 determinant. Furthermore, one can use it as a standard in the immunoassay used to detect the CA 19-9 determinant in suspected colorectal and pancreatic cancer patients. In addition, it can be used to test for the presence of anti-CA 19-9 antibodies in humans.

Brief Description of the Drawings

Figure l. Sizing of glycoproteins expressing CA 19-9 activity by gel filtration on Sepharose 4B-CL. The perchloric acid soluble fraction of SW1116 supernatants was chroraatographed in 50 mM Tris-HCl, pH 8.0, containing 10 mM dithiothreitol, 0.1% SDS, 150 mM sodium chloride, and 25 mM EDTA. CA 19-9 activity as determined by the CA 19-9 IRMA (solid circles) and optical density at 280 nm (open circles) are presented for each column fraction. Pooled fractions are indicated by Roman numerals I, II, and III. The exclusion volume of the column ([A] blue dextran) and the elution positions of murine immunoglobulins M ([B] 800 kDa) and G ([C] 160 kDa) and the inclusion volume (D) in the same column buffer but without dithiothreitol are indicated by arrow.

Figure 2. Monoclonal antibody 19-9 immunoblotting analysis of pooled fractions I (lane A), II (lane B), and III (lane C) after Sepharose 4B-CL separation of 19-9cGPs from SW1116 cell culture supernatants. Pooled fractions were run on a 3-12% gradient SDS polyacrylamide gel after reduction and heating in SDS. The western blot of the gel was incubated with radioiodinated Mab 19-9 and autoradiographed (see Materials and Methods). The molecular mass markers are myosin heavy chain (200 kDa) , beta-galactosidase (116 kDa), phosphorylase b (92 kDa), albumin (66 kDa), and ovalbumin (45 kDa).

Figure 3. Autoradiograph of an SDS polyacrylamide 3-12% gradient gel of purified radioiodinated 19-9cGP antigen (lane B) after exhaustive digestion with pronase (lane A) or neuraminidase (lane C). Radioiodinated 19-9cGP isolated from 40 ml of serum from a patient with colorectal cancer by the same procedure as 19-9cGP in lane A is shown in lane D. Approximately 5000 dpm of radiolabelled glycoproteins were applied to each lane. Sample treatment and molecular mass markers as described in Figure 2.

Figure 4. Sizing of purified radioiodinated 19-9cGP (solid line) and asialo-19-9cGP (dotted line) by size-exclusion chromatography on a TSK-4000 SW HPLC column equilibrated with 0.2 M phosphate, pH 6.8. The exclusion volume of the column ([A] blue dextran) and elution times of thyroglobulin ([B] 660 kDa), beta-galactosidase ([B] 464 kDa), immunoglobulln G ([C] 160 kDa), ovalbumin ([E] 45 kDa), and salts (F) are indicated by arrows.

Detailed Description

Abbreviations

Mab 19-9, monoclonal antibody 1116NS-19-9; CA 19-9, the carbohyhdrate determinant recognized by Mab 19-9; 19-9cGP, CA 19-9 containing glycoprotein; 19-9cGP (Y kDa) , a 19-9 cGP of molecular mass Y kDa; IRMA, immunoradiometric assay; Gdn-HCl, guanidine hydr ochloride; TMS, trimethylsilyl ; PITC, phenylisothiocyanate; Fuc, fucose; Man, mannose; Gal, galactose; GalNAc, N-acetylgalactosamine; GlcNAc, N-acetylglucosamine; NeuAc, N-acetylneuraminic acid. All sugars are assumed to be in the D-configuration, except fucose which is L.

Materials

Murine hybridoma 1116NS-19-9 (H. Koprowski et al. Somatic Cell Genet., 5, 957-972, 1979; ATCC No. HB 8059, American Type Culture Collection, Rockville, Maryland) was obtained from Dr. Zenon Steplewski of the Wistar Institute, Philadelphia, PA. Sepharose 4B-CL and Protein A-Sepharose 4B were purchased from Pharmacia. Trimethylsilylating reagent (TriSilZ) and amino acid standards (Standard H) were purchased from Pierce. The fused silica cross-linked methyl gas chromatography column (Ultra-1) was purchased from Hewlett Packard. Highly purified carbohydrate reference standards were obtained from Pfanstiehl, Waukegan, IL. The Micropak TSK 4000 SW HPLC column was from Varian Instruments. Fish gelatin was purchased from Norland Industries, Inc., New Brunswick, NJ. All other reagents were from standard supply houses, and were of the highest purity commercially available. Neuraminidase was purchased from Calbiochem, alpha-fucosidase from Sigma Chemical Co.

Production of Tumor-associated Glycoproteins

Glycoproteins expressing the CA 19-9 determinant were obtained from the cell culture supernatants of the SW1116 human colorectal carcinoma cell line (A. Leibovitz et al, Cancer Res., 36, 4562-4569, 1976; ATCC No. CCL 233, American Type Culture Collection, Rockville, Maryland). SW1116 cells were grown as previously described (10) and culture supernatants were collected seven to ten days after cells had reached confluence. Pooled cell supernatants were centrifuged at 10,000 X g, filtered through a Sartorius 0.2 micron pore-size cascade filter capsule, and concentrated to one-tenth the original volume with an Amicon DC-2 hollow fiber apparatus and filter cartridge (HP 100-20) with a molecular mass cutoff of 100 kDa. The concentrates were stored frozen at -20 C under which conditions the CA 19-9 activity was stable for at least 12 months.

Immunoradlometric Assay for CA 19-9 Activity

Solid phase immunoradlometric assays (IRMAs) for the detection of CA 19-9 activity were performed as described previously (R.E. Ritts et al, Int. J. Cancer, 33, 339-345, 1984) using kits manufactured by Centocor, Inc., Malvern, Pennsylvania. Briefly, the IRMA utilizes a two-step homogeneous "sandwich" configuration in which purified Mab 19-9 antibody adsorbed to polystyrene beads binds the glycoprotein antigen complex in the first step, and the amount of antigen bound to the bead is determined by binding of radioiodinated Mab 19-9 in the second incubation step. This particular assay configuration is effective since more than one CA 19-9 determinant is associated with each antigen molecule or antigen complex. The amount of antigen is expressed in units/ml or units/mg of protein as determined by reference to a CA 19-9 solution stored at -80 °C. The value of one unit was arbitrary but the same for all experiments reported here, thereby allowing an accurate index of the amounts of purification achieved by the procedures described. Reactivity of purified 19-9cGP with polyclonal anti-carcinoembryonic antigen (anti-CEA) antibodies was determined using a radioimmunoassay to CEA (Abbot Labs, N. Chicago, IL).

Preparation of MAB 19-9 Affinity Column

Immunoglobulin 1116NS 19-9 was purified from ascites by affinity chromatography on Protein A-Sepharose 4B (P. L. Ey et al, Immunochemistry. 15, 429-436, 1978). Purified antibody was bound and covalently cross-linked to a Protein A-Sepharose 4B column by the procedure of C. Schneider et al ( J. Biol . Chem.. 257, 10766-10769, 1982) using 20 mM dimethylpemilimidate as the bifunctional cross-linker. Approximately 4 mg of 19-9 antibody bound per ml of Protein A-Sepharose 4B. This method of conjugating Mab 19-9 to a solid matrix gave us the highest binding capacity for 19-9cGP, but more importantly, Mab 19-9 coupled by this procedure remained coupled to the solid phase. With other methods of conjugation, such as cyanogen bromide activated-Sepharose or carbodiimide coupling, small amounts of contaminating Mab 19-9 continually leached from the column.

Isolation of Tumor-associated Antigen 19-9cGP

Ten-times concentrated SW1116 tissue culture supernatant was extracted with an equal volume of cold (4 C) perchloric acid, the acid soluble fraction was collected after centrifugation at 10,000 x g for 20 minutes at 4 C, and the extract was neutralized with cold 1 .2 M potassium hydroxide (J. Krupey et al, J. Exp. Med . , 128 , 387-398 , 1968 ) . The supernatant recovered after removal of the insoluble potassium perchlorate salt by centrifugation was exhaustively diaf iltered against cold distilled water with the Amicon DC-2 hollow fiber apparatus, and finally concentrated to 1/20 the original volume and stored frozen at -20 C until use. Typical individual lots of pooled SW1116 supernatants contained one to eight liters. For some experiments , the perchloric acid soluble material was fractionated by gel chromatography on a Sepharose 4B-CL column equilibrated with 50 mM Tris , 150 mM sodium chloride, 10 mM dithiothreitol , 0.1% SDS, at pH 8.0.

Human cells of epithelial origin are a likely source for the CA 19-9 antigen . The antigen may be found in normal and carcinoma cells of pancreatic, colorectal, stomach, bile duct or lung origin. There is at least a 10-fold greater expression of the antigen in the cancer cells.

Before affinity chromatography, the concentrated perchloric acid soluble fraction from 1-2 liters of SW1116 supernatants was made 50 mM in sodium citrate, pH 4.5 , 500 mM in sodium chloride, and Nonidet P-40 nonionic detergent was added to a final concentration of 0.5% (w/v) . This solution, containing as much as 6 x 10⁶ units of CA 19-9 activity, was then applied to a 1.5 x 2.7 cm column of Mab 19-9 Protein A-Sepharose equilibrated with the same buffer. Bound antigen complexes were washed and eluted from the Mab 19-9 affinity column by the procedure of Schneider, et al. (C. Schneider et al, J. Blol . Chem., 257, 10766-10769, 1982) with the following modifications:

1) The immobilized antibody was used in a column, not batchwise

2) the buffer for immunoadsorption and washing of the column was 50 mM sodium citrate, pH 4.5, the pH optimal for Mab 19-9 affinity

3) taurodeoxycholate was substituted for deoxycholate in the column washing buffers

4) the affinity column was washed with alternating cycles of acidic buffer (0.1 M sodium acetate with 0.5 M sodium chloride, pH 4.0), and basic buffer (0.1 M sodium bicarbonate with 0.5 M sodium chloride, pH 10.0) prior to each affinity procedure to remove nonspecifically bound proteins from the Mab 19-9 affinity column.

Step. #4 proved to be essential for obtaining pure 19-9cGP. All affinity steps were done at room temperature (20-25 °C).

After exhaustive dialysis against distilled water, the proteins in the first Mab 19-9 affinity eluate were reduced with 50 mM dithiothreitol in the presence of 6M guanidine hydrochloride (6 M Gdn-HCl) for 4 h at 45 C and subsequently carboxymethylated with 75 mM iodoacetic acid. The alkylated proteins were exhaustively dialyzed against cold water, and 19-9cGP was repurified by a second Mab 19-9 affinity chromatography procedure as described above. The diethylamine/taurodeoxycholate eluate was neutralized and dialyzed against 20 mM Tris-HCl, pH 6.8. The dialysate was then applied to a 2.0 x 1.0 cm column of DEAE TrisAcryl column equilibrated with 20 mM Tris-HCl, pH 6.8, and the column was eluted sequentially with ten column volumes each of 0, 50, 100, 150, 200, 300 and 1000 mM sodium chloride in 20 nM Tris-HCl, pH 6.8. The majority of the CA 19-9 activity, usually greater than 90S. of the total activity, eluted with 100 and 150 mM chloride. The purity of the isolated 19-9cGP was determined by SDS polyacrylamide gel electrophoresis (SDS:PAGE) and autoradiography after radloiodination of the antigen by the Bolton-Hunter method (A. E. Bolton et al, Biochem. J., 133, 529-533, 1973) to a specific activity of 1-2 uCi/ug. We chose this method of extrinsic radiolabelling for judging purity because the majority of non-antigen proteins likely to appear as contaminants were from the culture media, impurities which intrinsic labelling of the SW1116 cells would not have revealed. If a particular lot of 19-9cGp was- not suitably pure at this point, it was purified further by a repeat of the Mab 19-9 affinity procedure, and purity reassessed by the above method.

Treatment of 19-9cGP with Enzymes

Exoglycosidase digestions were performed in 0.1 M acetate buffer, ph 4.5, for 24 hours at 37 C. Unit values of exoglycosidases were chosen in order to ensure complete digestion of appropriate oligosaccharide substrates as monitored by thin layer chromatography. Pronase digestion of 19-9cGP was done in 0.1 M Tris-HCl, 10 mM calcium chloride, 2% (w/v) pronase (solution pretreated at 60 C for 1 h) , pH 8.0, for 72 h at 60 C. Pronase digestions were terminated by heating the solution at 100 C for three minutes.

SDS Polyacrylamide Gel Electrophoresis

SDS polyacrylamide slab gels were run using the buffer system of Laemmli (U. K. Laemmli. Nature. 227. 680-685, 1970) with 3-12% gradient gels run to their pore-size limit (2000 volt-hours). Gels for autoradiography were dried and exposed to X-Omat AR film with a Cronex (Dupont) Quanta-Ill intensifying screen for 12-24 hours at -80 C. Monoclonal antibody immunoblotting analysis of 19-9cGPs after separation by electrophoresis was performed by the procedure of H. Towbin et al (Proc. Natl. Acad. Scl. U.S.A.. 76, 4350-4354, 1979). Western blots to nitrocellulose of antigens separated by gel electrophoresis were developed using radioiodinated Mab 19-9. Nitrocellulose transfer sheets were reacted with 4 x 10⁶ cpm of radiolabelled Mab 19-9 (9-13 uC_i/ug) in 25 mis of 100 mM sodium citrate, pH 4.5, containing 2% (v/v) fish gelatin and 0.1% (v/v) Non-Idet 40 (a polymer of octyl phenol ethylene oxide purchased from Sigma Chemical Co.) for six to eight hours at room temperature.

High Performance Liquid Chromatography

HPLC was used to determine the apparent purity and molecular weight of the purified protein under non-reducing, non-denaturing conditions. Molecular weight estimates were obtained by sizeexclusion chromatography on a system consisting of a Waters SM-6000 pump connected to a 7.5 x 300 mm Micropak TSK 4000 SW column equilibrated with 0.2 M potassium phosphate, pH 6.8. The molecular weight was estimated by comparing the retention time of radiolabelled proteins to the retention times of molecular weight standards which were monitored by optical density at 214 nm.

Amino Acid and Carbohydrate Analysis

Purified glycoprotein was dissolved in 6 N HCl containing 0.1% phenol, sealed under vacuum, and multiple samples were hydrolyzed for 24, 36, 48 and 72 hours at 110 C. Amino acids in the hydrolyzed sample were derivatlzed with phenylisothiocyanate (PITC), and PITC-amino acids separated and analyzed by HPLC using the Waters Pico-Tag column and elution conditions (B. A. Bidlingmeyer et al, J . Chromatoqr.. 336. 93-104, 1984). Cysteine was determined as S-(carboxymethyl) cysteine. Amino acids were quantitated by reference to a standard amino acid mixture of known composition treated in the same manner as the samples of unknown composition. The amount of serine in 19-9cGP was determined by extrapolation of picomoles of serine to 0 hours hydrolysis time with normalization to an internal standard of 250 picomoles of aminobutyric acid. Carbohydrate composition was determined from gas chromatographic analysis of triraethylisilyl (TMS) derivatives after HCl-methanolysis (R. A. Laine et al. Methods Enzymol., 28, 159-167, 1972). The TMS-methylglycosides were separated and identified using a Hewlett-Packard Model 5790 gas chromatograph interfaced to a Hewlett Packard Model 5970 mass selective detector (MSD) equipped with a 25 m Ultra-1 capillary column. The column was developed at 3 °C/min from 125 to 160 °C, then 10 °C/min from 160 to 235 °C. The flow rate was 2.91 ml/min; split ratio, 1:14; carrier gas. Helium.

Purification of 19-9cGP

We initially attempted to fractionate the perchloric acid soluble 19-9cGP by gel chromatography on Sepharose 4B-CL in the presence of SDS and a sulfhydryl reducing agent. The results of such studies are illustrated in Figure 1. There are essentially three broad fractions associated with CA 19-9 activity: the highest molecular weight fraction (Fraction I) elutes coincident with the void volume indicating a molecular mass of greater than 2000 kDa; a second small broad peak of CA 19-9 activity (Fraction II) has an apparent molecular mass of greater than 900 kDa; and a predominant peak of CA 19-9 immunoreactivity elutes between approximately 800 and 200 kDa (Fraction III). The relative proportions of these peaks varied slightly for separate lots of SW1116 supernatants. The column Fractions I, II, and III were pooled as indicated in Figure 1 and concentrated. The apparent molecular weights of the glycoproteins associated with CA 19-9 immunoreactivity in Fractions I-III were then determined on SDS polyacrylamide gels with reduction in the presence of SDS. This was done by SDS:PAGE, western blotting to nitrocellulose, reacting transferred glycoprotein with radiolabelled monoclonal antibody 19-9, and autoradiography. While the molecular weights of fractions I-III determined by SDS: PAGE were qualitatively consistent with results obtained from column chromatography, electrophoresis gave lower apparent molecular masses of greater than 500 kDa, 300-400 kDa, and 200-250 kDa (Figure 2, Lanes A, B, and C, respectively). It was apparent from such studies that either more than one species of glycoprotein expressed the CA 19-9 carbohydrate determinant, or that a single glycoprotein with the CA 19-9 determinant existed in multiple, stable glycoprotein complexes, possibly aggregates, refractory to treatment with SDS, heat, and reducing agents. An effort was made to disrupt the highest molecular weight CA 19-9 containing complex (Fraction I, Chart 1) by a series of chemical treatments. Chemical treatments included incubation with chaotropic agents such as 8 M urea or 6 M Gdn-HCl alone, or with 8 M urea or 6 M Gdn-HCl along with εulfhydryl reducing agents, and treatment with ionic and non-ionic detergents alone or in combination with heat and reducing agents. Of these, only reduction with dithiothreitol in the presence of a chaotropic agent such as 8 M urea or 6 M Gdn-HCl was able to disrupt the antigen-containing aggregates, allowing eventual immunoaffinity purification of a homogeneous 19-9cGP antigen preparation of 210 kDa (Figure 3, lane B). [Preferred chaotropic agents (in decreasing order of preference) are guanidine hydrochloride, urea, alkaline salts of bromine (e.g., LiBr}, ammonium thiocyanate, perchlorates, sodium or ammonium sulfate and polyethylene glycols.] All other treatments had no effect on the aggregated material in Fraction I, Figure 1 and hence the results of these negative experiments are not shown. This glycoprotein is apparently identical to that found in Fraction III, Figure 1, and identified in Figure 2, Lane C.

The protocol eventually developed to obtain pure 19-9cGP is illustrated in Table I. With the initial concentration step, Table I

Protocol Used for the Purification of Gastrointestinal Tumor-Associated Antigen 19-9cGP (210 kDa)

The specific activity (units/ml) and fold-purification (units/mg) are indicated for each fractionation step using 1.2 liter of SW1116 colorectal cancer cell supernatants as starting material.

Total Total Protein^a Units^b Units/mg Fold- Fraction mg) (CA 19-9) Purification^c

Supernatant 1.8 x 10⁴ 3.6 X 10⁶ 2.0 x 10² 1.0

10x Concentrate 4.4 x 10³ 3.6 X 10⁶ 8.1 x 10² 4.1

Perchloric Acid Soluble Fraction 2.9 x 10² 3.3 X 10⁶ 1.2 x 10⁴ 59.1

First Mab 19-9 Affinity Step 0.21 1.5 X 10⁶ 7.1 x 10⁶ 3.4 X 10⁴

Second Mab 19-9 Affinity Step 0.14 1.1 x 10⁶ 7.5 x 10⁶ 3.8 X 10⁴

Ion-exchange (DEAE) 0.10 0.8 X 10⁶ 8.0 X 10⁶ 4.0 X 10⁴

^aProtein determined from amino acid analysis (see Materials and Methods)

^DCA19-9 activity determined by the CA 19-9 immunoradlometric assay. cFold-purification relative to starting material in units CA 19-9/mg. 100 percent of the CA 19-9 activity and 33 percent of the protein is retained relative to the starting material. The majority of the fetal calf serum proteins in the initial tissue culture media passes through the 100 kDa cutoff hollow fiber filter. The perchloric acid extraction step results in less than 5 percent loss of CA 19-9 activity, but if fold-purification is expressed as CA 19-9 units/mg of protein, removal of acid insoluble proteins results in a net 59-fold purification. As expected the Mab 19-9 affinity steps result in the greatest fold-purification and are primarily responsible for the eventual production of homogeneous 19-9CGP antigen preparations. The initial antibody affinity step results in an apparent 50 percent loss of CA 19-9 activity, but gives an antigen preparation with 34,000 times the specific activity of the starting material. The loss in activity could not be attributed to incomplete binding of loaded antigen, as all affinity fractions were assayed for CA 19-9 activity, and no activity was ever observed in any of the column flow-throughs or washings. This apparent loss in activity is probably due to a decrease in antigen valency due to disaggregation. The reduction and alkylation step in the presence of 6 M Gdn-HCl was found to be indispenslble in the purification scheme. Without this step, CA 19-9 activity remained partially associated with high molecular weight aggregates after Mab 19-9 affinity chromatography, and radiolabelled material subsequently demonstrated many contaminating low molecular weight proteins by SDS:PAGE after reduction in the presence of 6 M Gdn-HCl (data not shown). Reduction, alkylation, and Mab 19-9 reaffinity results in a further 40 percent loss of CA 19-9 activity, but gives a substantially purer antigen preparation of 7.5 x 10⁶ units/mg. A final DEAE ion-exchange step serves to remove some persistent non-immunoreactive low molecular weight contaminants.

Reduction alone under the appropriate conditions will also cause disaggregation. However, reducton of sulfhydryls is a reversible state such that, when the reducing agent is removed, free sulfhydryls will rejoin and with spontaneous oxidation again form sulfhydryl-sulfhydryl bonds (S-S bonds). Alkylation of the -SH bonds prevents this association.

The reduction conditions are more critical than the alkylation conditions. Reduction must be done at temperatures greater than room temperature if complete reduction is to take place within a reasonable time, i.e., less than 8 hours. Temperatures greater than 60 °C may give side reactions harming protein. The higher the temperature, the shorter the time needed for reduction. For a range of times, 1-8 hours, the range of temperatures is 30 to 60°C. Most organic reducing agents would be somewhat effective, but only dithiothreitol (DTT) and beta-mercaptoethanol (BME) would be able to completely disrupt the aggregates at a reasonable concentration of reducing agent.

The alkylation conditions are not important within broad limits: for times of 0.5 to 2 hours, temperatures of 15 to 40 °C are used. The concentration of the alkylating agent should be kept as low as possible, however, to avoid agent reactions with other side chains on the protein (methionines); Use molar excess (1.5 x the reducing agent concentration). The nature of the alkylating agent is probably not important within broad limits. Any radical alkylating substituent capable of blocking the the sulfhydryl group would work. A preferably small group was used in order to avoid changing the biochemical behavior beyond blocking the sulfhydryls. Larger alkylating moities such as carbobenzoxy- radicals would increase the hydrophobic (non-polar) nature of the antigen.

After the final DEAE ion exchange step, the protein was estimated to be greater than 95% pure as determined by SDS: PAGE and autoradiography of radiolabelled 19-9cGP, an indication that the preparation was substantially free of other proteins.

The purified proteins were radiodinated and analyzed by gradient SDS: PAGE. The efficacy of the purification scheme can be seen in Figure 3, lane B, in which only a single band with an average molecular mass of 210 kDa is visible. As evidenced by its narrowness, that band contains a relatively homogeneous population of glycoproteins, within a size range of about 150 kDa to 250 kDa. The size distribution within the population was probably due to variations in the amount of carbohydrate per glycoprotein molecule; the protein moiety was probably the same in all molecules within the population. It is important to point out that this glycoprotein antigen, in our experience, iodinated very poorly; thus, contaminant proteins showed up readily by this type of analysis. The complete absence of any extraneous bands indicates that the preparation is very pure. However, this does not eliminate the possibility that there may be contaminants with the same electrophoretic mobility as 19-9cGP. The effects of treatment of radiolabelled 19-9cGP with the enzymes pronase or neuraminidase are demonstrated in lanes A and C, respectively. Even after extensive digestion with pronase, some glycopeptides of 45-200 kDa still exist indicating that 19-9cGP is resistant to proteases. Asialo-19-9cGP demonstrates an anomalous behavior in that its apparent molecular mass increased to 220-300 kDa although nearly 14% of its mass should have been removed. The observations that the apparent mass of the asialo-19-9-Gcp increased from 210 kDa to 220-300 kDa was confirmed by gradient gels run to their pore-limit. This eliminated the possibility that the apparent increase in the size of the antigen was due to changes in the electrophoretic mobilities, not to real changes in the size or hydrodynamic volume of the asialo-19-9cGp since such gradient systems when at equilibrium separate by size alone. The great specificity of this isolation procedure is indicated by its capacity to purify a 19-9cGP antigen of 180 kDa from 40 ml of serum from a patient with colorectal cancer (Figure 3, lane D). The lower molecular weight contaminants could be removed by repeated Mab 19-9 affinity chromatography.

In an effort to obtain another estimate of the molecular size of the purified antigen, the radioiodinated protein was analyzed by HPLC on a TSK 4000 sizing column equilibrated with 0.2 M phosphate, pH 6.8 (Figure 4). Only one rather broad peak of radioactivity with a mean retention time of 8.87 minutes is found for the intact antigen. The high mean molecular mass (approximately 900 kDa) and breadth of the peak on HPLC, however, suggests that the antigen may be self-aggregating under these conditions. Under similar conditions, asialo-19-9cGP also elutes as a broad peak, but in contrast to the results seen with gel electrophoresis (Figure 3, lane C), has a retention time of 9.30 minutes indicating a lower molecular mass of approximately 800 kDa. Apparently SDS or some similar zwitterionic detergent is needed to keep the antigen unaggregated. Othe detergents that may be used include deoxycholate or its derivatives and CHAPS.

Biochemical Characterization of Purified 19-9cGP

To better understand the 19-9cGP on a molecular basis, carbohydrate and amino acid analyses were performed on the purified antigen. The carbohydrate composition is shown in Table

Table II Carbohydrate Composition³ of 19-9cGP (210 kDa)

Sugar Molar Ratio^b

Fuc 4.1

Man 1.0

Gal 11.8

GalNAc 2.5

GlcNAc 4.9

NeuAC 5.1

Composition determined on TMS derivatives. bA11 values are molar ratios with Man set to 1.0. II. In the intact antigen, sugars were found in the ratios of Fuc: Man: Gal: GalNAc: GlcNAc: NeuAc as 4.1: 1.0: 11.8: 2.5: 4.9: 5.1. The presence of both Man and GalNAc suggest that both N-and O-linked oligosaccharides are found on 19-9cGP. The amino acid composition of purified 19-9cGP is presented in Table III.

Table III Amino Acid Composition^a of 19-9cGP (210 kDa)

Residue Mole % Residue Mole %

Asp 5.1 Pro 7.3

Glu 9.8 Tyr 2.5 cmCys^b 0.0 Val 5.1

Ser 19.7 Met 0.9

Gly 8.4 Ile 3.4

His 1.5 Leu 5.6

Arg 5.5 Phe 2.7

Thr 8.8 Lys 4.1

Ala 7.5

aAmino acids quantitated as PTC-derivatives. bBelow limits of detectability.

The most notable aspect of the composition was that serine , proline , and threonine together accounted for greater than 35% of the amino acids which is consistent with a mucin-like protein structure . The extremely high mole percent of serine ( 19.7% ) , however , is unusual . Moreover , the relatively high amount of acidic and acid amide amino acids and no detectable cysteine in 19-9cGP is not typical of a mucin. Combined carbohydrate and amino acid analyses of the same antigen preparation indicated that the antigen was as much as 85% carbohydrate by weight . Treatment of the purified 19-9cGP with neuraminidase and afucosidase resulted in complete loss of CA 19-9 activity. The purified 19-9cGP also did not react with polyclonal antisera to CEA (data not shown) .

We have not yet established whether the 210 kDa glycoprotein is itself involved in disulfide stabilized aggregates or merely non- covalently associated with other proteins linked by intra- or interchain disul f ide bonds which stabil ize the glycoprotein aggregates . Alternatively, however , the demonstrated effect of reduction in decreasing the size of the aggregates might be due to activation of intrinsic proteolytic activities .

Claims

WHAT IS CLAIMED IS:

1. A human epithelial cell glycoprotein population in which:

(i) each glycoprotein contains the CA 19-9 antigenic determinant;

(ii) as measured by sodium dodecyl sulfate electrophoresis, the average mobility is that of a protein of about 210 kDa and the mobilities vary from that of a protein of about 150 kDa to one of about 250 kDa;

(ill) on the average, is about 85% carbohydrate and about 15% protein by weight;

(iv) has an amino acid composition, quantitated as phenyl isothiocyanate derivatives in mole % in parentheses, as follows: asp (5.1), glu (9.8), ser (19.7). gly (8.4), his (1.5), arg (5.5) thr(8:8), ala (7.5), pro (7.3), tyr (2.5), val (5.1), met (0.9), ile (3.4), leu (5.6), phe (2.7), lys (4.1); and

(v) has a carbohydrate composition, quantitated as trimethylsilyl derivatives in molar ratios in parentheses, as follows: fucose (4.1), mannose (1.0), galactose (11.8), N-acetylgalactosamine (2.5), N-acetylglucosamlne (4.9), N-acetylneuraminic acid (5.1); and each glycoprotein is either in an unaggregated state or a zwitterionic detergent dissociable state.

2. A glycoprotein population of Claim 1, substantially free of other proteins.

3 . A glycoprotein population of Claim 2, wherein each glycoprotein is in the unaggregated state.

4. A glycoprotein population of Claim 3, wherein each glycoprotein is in a zwitterionic detergent dissociable state.

5. A glycoprotein population of claim 2 prepared by a method which comprises using human epithelial cells to produce entities from which the glycoprotein population can be isolated .

6 . A glycoprotein population of Claim 5 , wherein the human epi thel ial cells were carcinoma cel ls of either pancreat ic , colorectal , stomach, bile duct or lung origin.

7 . A glycoprotein population of Claim 6 , wherein the human epithelial cells were colorectal carcinoma cells .

8 . A glycoprotein population of Claim 7 , wherein the human epithelial cells were SW1116 cells .

9. A glycoprotein population of Claim 8 wherein each glycoprotein is in the aggregated state .

10 . A g lycopro te i n population of Claim 8 wherein each glycoprotein is in a zwitterionic detergent dissociable aggregate.

11. A process for facilitating the production of a glycoprotein population of claim 1 which comprises the steps of :

1 ) using human epithelial cells to produce entities with an average molecular mass greater than that of said population and containing the CA 19-9 determinant;

2) dissociating said entities by treating them with a sulfidesulfide bond reducing agent in the presence of a chaotropic agent .

12 . The process of Claim 11 in which the treatment with a reducing agent is followed by a step :

3) alkylating the components of the dissociated entities in the presence of a chaotropic agent.

13. The process of Claim 11 in which the human epithelial cells are SW1116 human colorectal carcinoma cells , the reducing agent is dithiothrei tol , and the chaotropic agent is guanidine hydrochloride.

14. The process of Claim 12 in which the human epithelial cells are SW1116 human colorectal carcinoma cells, the reducing agent is dithiothreitol , the alkylating agent is iodoacetic acid, and the chaotropic agent is guanidine hydrochloride.