[CANCER RESEARCH 49, 4550-4556. August 15.

1989]

ABH Blood Group Antigen Expression, Synthesis, and Degradation in Human Colonie Adenocarcinoma Cell Lines1
Rajvir Dahiya,2 Steven H. Itzkowitz,3 James C. Byrd, and Young S. Kim
Gastrointestinal Research Laboratory, San Francisco, California 94121 Veterans Administration Medical Center, and Department of Medicine, University of California Medical Center,

ABSTRACT
The synthesis of blood group ABH antigens is under genetic control, where the primary gene products are glycosyltransferases. Several studies have demonstrated cancer-associated alterations in ABH antigen expres sion in human colon cancer tissues. However, the mechanism(s) respon sible for these alterations has not been elucidated. Therefore, experiments were conducted using nine established colon cancer cell lines (four type O, three type A, and two type B) to examine ABH antigen expression by immunocytochemistry and correlate this with activities of ABH biosynthetic (glycosyltransferase) and degradative (glycosidase) enzymes. The products of the glycosyltransferase enzymes were characterized by high performance liquid chromatography and paper chromatography, and substrate affinities (apparent A,,, values) of the cancer cell-derived gly cosyltransferases were analyzed. The present data demonstrate: (a) all cell lines except H-498 (blood type A) expressed the appropriate ABH glycosyltransferase as well as all three glycusiduses; (b) product characterization and substrate depen dence experiments suggested that the cancer cell-derived ABH glycosyl transferase enzymes had properties that were similar to those of the ABH enzymes in human serum; (c) H-498 cells exhibited A antigen deletion with accumulation of H precursor substance, most likely due to insufficient A transferase activity; (rf) SW1417 cells (blood type B) demonstrated B antigen deletion without precursor accumulation, despite adequate levels of B transferase and low a-galactosidase activity; and (c) weak incompatible A antigen expression occurred in LoVo (type B) and SW1116 (type O) cells, and weak incompatible B antigen expression occurred in H-498 (type A) and SW1116 cells. However, since these cells lacked incompatible A or B transferase activity, these incompatible antigens are probably not the true A or B antigens. Thus, the colon cancer cell lines used in this study exhibit all of the ABH alterations previously described in colon cancer tissues and appear to be useful experimental models for studying the molecular events involved in cancer-associated ABH expression.

H antigen is synthesized by the addition of L-fucose to the terminal galactosyl residues of blood group-type oligosaccharides on glycolipids or glycoproteins, the action of a-2-L-fucosyltransferase enzyme. The H substance can then serve as the immediate precursor for the A and B antigens. The structural al-eles,A and B at the ABO locus, encode, respectively, an aW-acetyl-D-galactosaminyltransferase and an a-o-galactosyltransferase, which catalyze the addition of jV-acetyl-D-galactosamine or o-galactose to H antigen and form the A and B determinants (1,3, 7). The blood group ABH antigens are expressed by colonocytes throughout the entire colon during fetal life (8, 9) but only in the proximal colon of the normal adult (4, 8-11). Because adenomas and cancers of the distal colon frequently reexpress the fetal phenotype, ABH antigens in this location are consid ered as oncofetal tumor-associated antigens (9, 11-14). Studies of colon cancer tissues have shown the deletion of ABH antigens (9, 15, 16) and the accumulation of precursor determinants, such as H and li substance (17), and of incompatible blood group antigens, i.e., A antigen in patients with blood type O or B (9, 18). The mechanisms underlying these alterations in developmental and cancer-associated ABH antigen expression are not clear. Cultured cells potentially offer a good experimen tal model for studying regulation of expression of ABH anti gens. There have been few reports of ABH antigen expression in cultured colon cancer cell lines (19, 20). However, in those studies, the blood group ABH-specific glycosyltransferases were not assayed. Therefore, the present experiments were under taken to investigate the expression of ABH antigens as well as their synthetic and degradative enzymes in human colonie adenocarcinoma cell lines. MATERIALS AND METHODS

INTRODUCTION Blood group ABH substances are carbohydrate structures, which are formed by the sequential addition of specific monosaccharides to the carbohydrate side chains of glycolipids and glycoproteins (1, 2). They are expressed not only in blood cells (3) but also in a variety of epithelial cells (4). Their expression is under genetic control, where the primary gene products are glycosyltransferases (5). Classically, ABH blood group deter minants reside on either a type 1 (Gal/81>3GlcNAc-R)4 type or 2 (Gall>4GlcNAc-R) chain (5), although recently type 3 (Gal/31-3GalNAca-R) and type 4 (Gal/3l3GalNAc/3-R) chains have also been described (6).
Received 10/4/88; revised 3/27/89; accepted 5/9/89. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1Supported by grants from the National Cancer Institute (CA42981 to S. H. I. and CA47551 to Y. S. K.), a VA Research Associate award (to S. H. I.), a VA Medical Investigator award (to Y. S. K.). and the Veterans Administration Medical Research Service. ! To whom requests for reprints should be addressed, at GI Research Lab (I51M2), VA Medical Center. 4150 Clement St., San Francisco, CA 94121. 3 Present address: Division of Gastroenterology, Box 1069, Mount Sinai Medical Center, 1 Gustave Levy Place, New York, NY 10029. * The abbreviations used are: Gal, galactose; GlcNAc, /V-acetylglucosamine; GalNAc, A'-acetylgalactosamine; Fuc. fucose; PBS. phosphate buffered saline; HPLC, high performance liquid chromatography.

Materials. UDP-D-[3H]galactose (specific activity, 17 Ci/mmol) was purchased from New England Nuclear (Boston, MA), UDP-JV[l4C]Acetyl-D-galactosamine (specific activity, 23 mCi/mmol) was pur chased from ICN (Irvine, CA), and GDP-L-[14C]fucose (specific activity, 308 mCi/mmol) was obtained from Amersham (Arlington Heights, IL). Phenyl /3-galactoside, 2'-fucosyllactose,/-nitrophenyl galactoside, p-nitrophenyl a-fucoside, /-nitrophenyl-a-A'-acetylgalactosaminide, and biotinylated isolectin I-B4 from Bandeiraea simplicifolia were pur chased from Sigma Chemical Co. (St. Louis, MO). Dowex 1-X8formate (100-200 mesh) was obtained from Bio-Rad Laboratories (Richmond, CA). Histostain-S.P. kit was purchased from Zymed Lab oratories (South San Francisco, CA). Cell Culture. Human colonie adenocarcinoma cell lines LS174T, SW1417, LoVo, SW1116, HT29, and SW480 were obtained from the American Type Culture Collection (Bethesda, MD). The HM7 and LM12 cell lines are high- and low-mucin variant clones established from the LS174T cell line in our laboratory (21). The H498 cell line was recently established from a blood type A individual and was generously provided by Dr. A. Gazdar, National Cancer Institute. All cell lines were grown in Dulbecco's modified Eagle's medium, with high glucose, penicillin (50 units/ml), streptomycin (5 Mg/ml), and 10% heat-inactivated fetal bovine serum, as described (18). All of the cell lines were tested for Mycoplasma contamination using the Mycoplasma T.C. II rapid detection system (Gen-Probe, Inc., San Diego, CA) and were found to be negative. Cell cultures were incubated at 37Cin a

4550

ABH ANTIGENS IN ADENOCARCINOMA

CELL LINES

humidified atmosphere of 5% CO2 in air. Heat treatment (56"C for 1 h) of the fetal bovine serum destroyed almost all the glycosyltransferase activity of the fetal bovine serum. After the cells were confluent, the medium was decanted and the cells were washed 3 times with PBS and scraped from plates into 8 ml of PBS with a rubber policeman. The cells were centrifuged at 3000 rpm for 10 min and the cell pellet was resuspended in 2 ml of 20 HIM cacodylate buffer (pH 6.8) and sonicated for 2 min. Aliquots (1 ml) were frozen (70C) immediately, until further use. Protein content was determined by the method of Lowry et al. (22). Glycosyltransferase Assays. The details for glycosyltransferase assays have been described earlier (23, 24). The assay mixture for al,3GalNAc transferase (A transferase) contained 410 IM 2'-fucosyllactose, 2.79 /-M UDP-Ar-['4C]acetyl-D-galactosamine (30,000-50,000 cpm), reaction cocktail (0.1 M sodium cacodylate-HCl buffer, pH 6.8; 200 pM ATP; 20 mM MnCl2; 0.1% Triton X-100), and cell homogenate (25-100 pg). The assay mixture for a 1,3-galactosyltransferase (B transferase) con tained 410 MM 2'-fucosyllactose, 17-32 nM UDP-['H]galactose (50,000-100,000 cpm), reaction cocktail, and cell homogenate (25-75 pg). The reaction mixture for al,2-fucosyltransferase (H transferase) contained 39 mM phenyl-/3-galactoside, 1-1.8 pM GDP-['4C]fucose (30,000-40,000 cpm), reaction cocktail, and cell extract (100-200 pg). Final volumes of all assays were made up to 100 pi. All glycosyltrans ferase assays were performed within the linear range of both the amount of enzyme (cell homogenate) and incubation time. Results shown are the median of triplicate tubes for A transferase and B transferase and single assays for H transferase. Incubations were carried out at 37C for 90 min. The reactions were stopped by being placed on ice. The reaction mixtures were then applied to Dowex 1-X8 (100-200 mesh) formate columns (0.5 x 2 cm) and the products were separated by washing each column with 2 ml water. The product of a 1,2-L-fucosyltransferase enzyme was further separated by descending paper chromatography (24). An aliquot of each eluate was mixed with ACS II scintillation fluid and counted in a scintillation counter. The values obtained without acceptor were taken as endogenous enzyme activity and were subtracted from the total incorporation to give the acceptordependent activity. Characterization of A and B Tetrasaccharides. For characterization of A tetrasaccharide, cell homogenate (HT29 cell line), as well as normal A blood type serum, was incubated with UDP-A'-['4C]acetyl-Dgalactosamine in the presence or absence of 2.2 m\i 2'-fucosyllactose and the reaction mixtures were passed through Dowex 1-X8 formate columns (0.5 x 2 cm) as for the standard assay. The unbound fractions from Dowex 1-X8 formate columns were evaporated, chromatographed on Sephadex G-15 columns (0.9 x 59 cm), and eluted with water at a flow rate of 37 ml/h. In each case, there was a peak of I4C that eluted at the position of stachyose standard which was absent in the controls (without 2'-fucosyllactose). This peak was pooled, evaporated, and examined by amine-phase HPLC. Two columns of 0.4- x 30-cm AX10 (Varian) were used in series with a reverse phase guard column. The column was equilibrated with acetonitrile and 15 mM potassium phos phate, pH 5.2 (in the ratio of 85:15, v/v), and eluted with an acetonitrile gradient of 85 to 0% over the course of 170 min. Fractions of 0.6 ml were collected and counted. For characterization of B tetrasaccharide, cell homogenate (SW1417 cell line), as well as B blood type serum, were incubated with UDP-['H]galactose in the presence or absence of 2.2 HIM2'-fucosylIactose. The reaction mixtures were passed through Dowex 1-X8 formate columns and then chromatographed on Sephadex G-15 as described for the A tetrasaccharide. The peak of apparent tetrasaccharide size was pooled, evaporated, and analyzed by aminephase HPLC as described above. Characterization of H Product. The product of 1.2-L-fucosyltransferase enzyme was characterized by using descending paper chromatography (24). The unbound Dowex 1-X8 column fractions were dried under nitrogen and then spotted on Whatman No. 3 paper. Paper chromatography was performed by using the solvent system of ethylacetate:pyridine:water (10:4:3, v/v) for 5 h. The paper strips were then cut into 1.5-cm fractions and counted in a scintillation counter. The radioactivity incorporated into phenylgalactoside acceptor was ex pressed as a percentage of the total counts in the original reaction mixture. The product was well separated from GDP-fucose and free L-

['4C]fucose, as judged by using reference sugars, which were detected with silver nitrate reagent (25). Glycosidase Assay. All glycosidase assays were performed as de scribed previously, with 2 mM /7-nitrophenylglycoside in 0. l M acetate, pH 4.0 (23). These reactions measure the ability of cell homogenates to release any a-GalNAc, a-galactose, or a-fucose residue, including those of the blood group A, B, and H substances, using p-nitrophenyla-glycoside substrates. Glycosidases were also assayed under the con ditions of the glycosyltransferase assays (pH 6.8) with 2 mM p-nitrophenylglycosides but without nucleotide sugars. Immunocytochemistry. The monoclonal antibodies used in this study have been used by us previously (9) and were generously provided by Dr. R. Murray Ratcliffe, Chembiomed, Ltd. (Edmonton, Alberta, Can ada). The anti-A and anti-B monoclonal antibodies recognize their respective epitopes on either type 1 or type 2 chains, whereas the antiH monoclonal antibody recognizes only H type 2 antigen. Cells were seeded onto chamber slides (Intermountain Scientific Co., Bountiful, UT) at a density of 3 x IO5 cells/chamber and allowed to attach overnight. The cells were then washed 3 times with buffer (PBS/ 1% bovine serum albumin), fixed in acid-ethanol for 20 min at -20C, and washed again 3 times with buffer. In pilot experiments, the acidethanol fixation was found to give lower background staining than formaldehyde or glutaraldehyde fixation. Endogenous peroxidase activ ity was blocked by incubating cells with 1% hydrogen peroxide in methanol for 10 min. followed by three washes with buffer. Normal rabbit serum (10%) was applied for 10 min and blotted off. Primary antibody (2 pg/ml) was added for 60 min, the cells were washed 3 times with buffer, and biotinylated rabbit anti-mouse IgG + IgA + IgM (10 Pg/ml) was applied for 20 min. The cells were then washed, reacted with streptavidin-peroxidase conjugate (10 pg/ml) for 30 min, washed again, and incubated with diaminobenzidine for 10 min. Slides were counterstained with 1% methyl green, rinsed in tap water, dehydrated, and mounted. For lectin binding the primary antibody was omitted and the secondary antibody was replaced by biotinylated isolectin I-B4 of A. simplicifolia. Negative controls consisted of substituting normal mouse IgM for primary antibody and PBS for the primary and secondary antibodies. In all cases, when primary antibody was omitted, a completely negative staining pattern was obtained. Cells were scored for antigen expression based on percentage of positive cells as well as staining intensity: weak (+), moderate (++), strong (+++).

RESULTS Table 1 shows the expression of A antigen and a-/V-acetyl-Dgalactosaminyltransferase (A transferase) and aWV-acetylgalactosaminidase activities in the nine colonie adenocarcinoma cell lines examined. Of the three cell lines derived from blood type A individuals, only two, HT-29 and SW480, showed substantial A transferase activity as well as A antigen expression (Fig. 1, a and b). When unlabeled UDP-GalNAc (40 MM) was added, there was a 2- to 3-fold decrease in net I4C incorporation for HT-29 and SW480 and A transferase was still undetectable in
Table 1 Expression of A antigen, a 1,3-N-acetylgalactosaminyltransferase (A transferase). and N-acetylgalactosaminidase in human colonie adenocarcinoma cell lines A antigen exprso-W-AcetylgalactosaBlood sion (% of positive A transferase minidasc (nmol/mg/ Cell line type cells) (pmol/mg/90 min) 16h) 10.711.0 LS174THM-7LM-12SW1116HT-29SW-480H-498LoVoSW14170OoO I0(+)rA (++)A 60 100(+++)AB

8.74.5 2.725.2 15.1421. 32.8418.5 7 37.83.9 3.935.9 (+)B19.0 5 15.70546103464337332420380850*46282636197568 70426 27 " Mean SE, five experiments except for H-498 (four experiments). * Mean four experiments. SE, c Numbers in parentheses, staining intensity scored as weak (+). moderate (++). and strong (+++).

4551

ABH ANTIGENS IN ADENOCARCINOMA

CELL LINES

the other cell lines. Interestingly, the H-498 cell line did not express any A antigen, thereby representing an example of blood group A antigen deletion (Fig. le). This antigenic deletion most likely results from an absence of A transferase activity in these cells (Table 1). Very weak expression of A antigen was also found in SW1116 (blood type O) and LoVo (blood type B), representing incompatible A antigen. However, since there was very little or no activity of A transferase in any of the O or B blood type cell lines, it is likely that this weak incompatible A antigen represents an A-like or cross-reacting substance, ayV-Acetylgalactosaminidase activity measured at pH 4.0 was similar in all of the cell lines, regardless of blood type (Table

1). Under the conditions of the A transferase assay (pH 6.8), a-yV-acetylgalactosaminidase was undetectable. Using HT-29 cells as a prototype blood type A cell line, the enzyme kinetics of A transferase were examined with different concentrations of the acceptor, 2'-fucosyllactose, and the do nor, UDP-GalNAc. The apparent Km for 2'-fucosyllactose was 0.050 HIM (Fig. 2b), which is comparable to the Km of the A enzyme in human serum (Km = 0.03-0.16 mivi) (26-28). The fucosyl moiety of 2'-fucosyllactose was required for acceptor activity. When the nine cell lines were tested for transfer of A'acetylgalactosamine to lactose, there was less than 0.4 pmol/ mg/90 min in each. The apparent Km for UDP-GalNAc of the HT-29 A transferase was 0.032 mM (Fig. 2a), which is also similar to that of the serum enzyme (Km = 0.05-0.09 HIM)(2628). The apparent Km of the SW480 enzyme for UDP-GalNAc was 0.024 mivi (data not shown). The A transferase of HT-29 cells was inhibited by UDP-galactose, with 65% inhibition at 50 /iM UDP-galactose in the presence of 64 /M UDP-[I4C]GalNAc. However, there was no detectable transfer of galactose to 2'-fucosyllactose by the A transferase of HT-29 cells.

v* *' ****
*^H/*-*i.i
.*

Ar

.l- >

In order to establish the product of the A transferase, reaction mixtures were passed through Dowex 1-X8 columns, deionized, and separated by gel nitration on Sephadex G-15 columns. There was a peak of radioactivity corresponding to tetrasaccharide size which was formed only in the presence of 2'fucosyllactose. This peak was further examined by amine-phase HPLC (Fig. 3). The tetrasaccharide-size peak was chromatographically similar to the product formed by the A transferase of human serum, which has been previously characterized as GalNAcal,3[Fucal,2]Gal/31,4Glc (29). Table 2 shows the expression of B antigen and 1,3-galactosyltransferase (B transferase) and galactosidase activities in human colonie adenocarcinoma cell lines. Of the two blood group B cell lines, only LoVo cells showed B antigen expression (Fig. 4a) as well as B transferase activity. When unlabeled UDP-galactose (2 /M) added, there was a 2-fold decrease was in net incorporation of 'H in SW1417 and LoVo cells, and B

transferase was still undetectable in the other cell lines. Inter estingly, the SW1417 cell line had a comparable level of B

*-:.v
E O

100-

50-

20 -

1c
Fig. I. Blood group A antigen expression in blood type A cell lines, a, SW480. All of the cells express the antigen strongly on cell membranes and in cytoplasm, x 62. 6. HT29. Antigen expression is localized more in the cytoplasm with a granular staining pattern. Some cells are negative, x 62. e, H498. None of the cells express the A antigen, x 50.

0.2

0.3

0.4

[Fuca2Galp4Glc],

mM

Fig. 2. Substrate dependence of A transferase. A homogenate of HT-29 (20 Mgprotein) was incubated with varying concentrations of UDP-[l4C]GalNAc (A) and 2'-fucosyllactose (B). Linear regression analysis of V versus V'/S plots (insets) was used to determine apparent Kmand I7., values.

4552

12 f
TT T

ABU ANTIGENS [N ADENOCARCINOMA CELL LINES

4b), but there was also strong staining of LS174T, HM-7, SW480, and H-498 (Fig. 4d). Galactosidase activity at pH 4.0 was present in all of the cell lines, regardless of blood type antigen expression, but no -galactosidase could be detected at pH 6.8. The enzyme kinetics of the SW1417 B transferase was ex amined with different concentrations of 2'-fucosyllactose and UDP-galactose. The apparent K, 2'-fucosyIlactose was 0.24 for
80
number

20

40

60
fraction

100

120

mM (Fig. 56), which is comparable to the serum B transferase (Km = 0.05 HIM) (31). As for the A transferase, the fucosyl moiety of 2'-fucosyllactose was required for B transferase ac ceptor activity. The nine cell lines all showed at most 2 pmol/ mg/90 min transfer of galactose to lactose. Using UDP-galac tose, the B transferase of SW1417 colon cancer cells showed an apparent A",,,of 0.0021 mM (Fig. 5a), which is lower than that of the serum enzyme (K, 0.01 mM) (31). The B trans = ferase of LoVo cells had an apparent K,n for UDP-galactose of 0.0010 mM (data not shown). The B transferase of SW1417 cells was inhibited by UDP-GalNAc, with 65% inhibition at 10 MMUDP-GalNAc in the presence of 5 //M UDP-['H]galactose. However, there was no detectable transfer of 7V-acetylgalactosamine to 2'-fucosyllactose by the B transferase of SW1417 cells. The B-specific tetrasaccharide product was characterized in a manner similar to that of the A tetrasaccharide (Fig. 3). A single peak of tetrasaccharide size containing ['Hjgalactose was formed only in the presence of 2'-fucosyllactose. This tetrasac charide product was chromatographically similar to that formed by the serum B transferase, which has previously been charac terized as Gall-3[Fucl-2]Gal/Jl-4Glc (32). Expression of H antigen and 1,2-fucosyltransferase (H transferase) and fucosidase activities in colonie adenocarcinoma cell lines is shown in Table 3. All of the blood group O cell lines expressed H antigen as well as H transferase activity. The H antigen expression was less prevalent in LS174T and its derivatives (HM-7, LM-12) than in SW1116 cells (Fig. 6). Of the A and B cell lines, only H-498 reacted with anti-H antibody. This represents accumulation of the precursor H substance in this cell line, which is also deficient in A transferase activity. There were significant amounts of H transferase activity in all

Fig. 3. Characteri/ation of A transferase and B transferasc products. Homogenates of HT-29, incubated with UDP-|MC]GalNAc and 2'-fucosyllactose (). and SW 1417. incubated with UDP-|'H[galactose and 2'-fucosyllactose (D), were deionizcd on Dowex 1 formate, subjected to gel filtration on Sephadex G-l 5. and then Mii.ih/><!HPLC as described in "Materials and Methods." Arrows, left to by right, show the elution positions of /V-acetylgalactosamine, galactose, and 2'fucosyllactose. Table 2 Expression of H antigen, <\l,3-%alactosyltrunsferase (B transferasef, and ga/actosidase in human colonie adenocarcinonta cell lines antigenMAb" transferase Lectin60(++)40 min)1.4 lineLS174THM Cell typeOOooAAABBB (pmol/mg/90 1.1*0.2 (nmol/mg/16 h)177 13.1'256.6

7LM-12SWI116HT-29SW-480H-498LoVoSW1417Blood (++)NT30 0.10.2 14146+ 0.10.3 + 19.3114+ 15.669.6 (+)40 0.30.3 0.10.3 + 9.8120.6 (+)10(+) 11.6150.6 0.30.2 80(++)70(++) 0.18.8 9.8141.6+ 17.970.6 90(+++)5(+)B 1.99.4+ 1.5(i-Galactosidase12.5 " MAb, monoclonal antibody; NT, not tested. * Mean SE, six experiments except for LoVo (five experiments). ' Mean SE, four experiments.

r>

jrpr.

4a

VT*.

4c

4d

Fig. 4. Blood group B antigen expression, a, LoVo cells stained with mono clonal anti-B antibody. Cells that react are stained strongly in the cytoplasm but also on cell membranes, x 62. b, LoVo cells stained with B. simplicifolia I-B4 lectin. Staining is more intense and more prevalent than with monoclonal anti-B antibody. X 62. c, H-498 (blood type A cell line) stained with monoclonal anti-B antibody. There is faint, granular, cytoplasmic staining. X 62. d, H-498 cells stained with B. simplicifolia-l lectin. Intense granular staining is shown. X 62.

10
[UDP-Gal],

is
MM

20

transferase activity but did not react with anti-B antibody (Fig. 2),thereby demonstrating B antigen deletion. This antigenic deletion is not apparently due to excess degradation because, if anything, a-galactosidase activity in SW1417 cells was lower than in the other cell lines. Weak incompatible B antigen expression was noted in SW1116 and H498 cell lines (Fig. 4c). However, since there was little or no B transferase activity in the blood type O and A cell lines, this incompatible antigen expression must represent a B-like or cross-reacting substance. Using B. simplicifolia isolectin I-B4, which is specific for agalactosyl residues (30), LoVo was very strongly stained (Fig.

1.0-

V/S

'

I 0.2

0.4

0.6

0.8
mM

[Fuco2Glp4Glc],

Fig. 5. Substrate dependence of B transferase. A homogenate of SW 1417 (20 Mgprotein) was incubated with varying concentrations of UDP-('H]galactose (A) and 2'-fucosyllactose (B). Linear regression analysis of V versus VjS plots (insets) was used to determine apparent A'mand Fm values.

4553

ABH ANTIGENS IN ADENOCARC1NOMA CELL LINES

Table 3 Expression of H antigen, al,2-fucosyltransferase (H transferase), and 600fucosidase in human colonie adenocarcinoma cell lines transferase a-Fucosidase H antigen lineLSI74THM-7LM-12SW1116HT-29SW-480H-498LoVoSW1417Blood Cell type expressionO (pmol/mg/90 min) (nmol/mg/16 h)492.9452.4414.1241.8372.3211.9513.1399.1855.6248454046059177699865739881536345"4505059477442181101439 10(++)O 10(+)O 5(+)O 80(+)AAA 20(-t-+)BBH 202 ean M SE, four experiments.
CGDP-Fuc], MM
400-

2OO -

40

80

[phnyl--Gl], mM

Fig. 7. Substrate dependence of H transferase. A homogenate of LS 174T (50 jig protein) was incubated with varying concentrations of GDP-|'"C]fucose (A) and phenyl-|8-galactoside (B). Linear regression analysis of V versus V/S plots (insets) was used to determine apparent K,,,and I ',. values.

Fig. 6. Blood group H antigen expression. I SI 741 cells stained with mono clonal anti-H antibody. Cells are focally positive, with staining of cell membranes. X50.

of the cell lines, irrespective of the blood type. a-Fucosidase activity at pH 4.0 did not correlate with H antigen expression but was higher than a-galactosidase or a-/V-acetylgalactosaminidase in all the cell lines. The activities of a-fucosidase at pH 6.8 were 1 to 3% of the activities at pH 4.0 (10 to 38 nmol/ mg/16 h). The enzyme kinetics of H transferase was examined with respect to different concentrations of acceptor, phenyl-/3-galactoside, and donor, GDP-fucose. The apparent Km for phenyl-/3galactoside was 8.65 mM (Fig. 7), which is similar to the Km of the human serum enzyme (33). The apparent Km for GDPfucose was 39.7 M. is lower than the GDP-fucose Km of This the human trachal nzyme (34). Because it has been suggested e that ABH antigen expression is growth dependent (35), the doubling times of seven of the cell lines were measured (21). The doubling times of the two cell lines that failed to express the expected antigen (H-498, 42 h; SW1417, 28 h) were within the range of those that did express the appropriate antigen (LS174T, 25 h; HT-29, 30 h; LoVo, 35 h; SW480, 38 h; SW1116, 74 h). In order to establish the product of H transferase, the reaction mixtures were passed through Dowex 1-X8 columns and then separated by descending paper chromatography. As shown in Fig. 8, the reaction product from the fucosyltransferase reaction was clearly separated from free fucose and phenyl--galactoside. This product was chromatographically similar to the product formed by the H fucosyltransferase of human serum, which has been previously characterized as Fucal-2Gal/3-phenyl (24). DISCUSSION Our findings can be summarized as follows, (a) Except for the blood type A H-498 cell line, colon cancer cell lines derived

io

20

10 dittine, cm

I 20

Fig. 8. Characterization of H transferase product. A homogenate of LS 174T was incubated with GDP-[14C]fucose with (A) or without (B) 0.04 M phenyl-fgalactoside. The reaction mixtures, after passing through Dowex 1 formate, were subjected to paper chromatography. The first peak (a) has the mobility of free fucose. The second peak (b) assumed to be Fuc-2Gal/3-phenyl, is taken as the H transferase product.

from blood type A or B individuals expressed the appropriate A or B gene-encoded glycosyltransferase. This was associated with expression of the relevant antigen by each cell line, except H-498 and SW1417, which manifested deletion of the A and B antigens, respectively, (b) All cell lines, regardless of ABO type, expressed the H gene-encoded fucosyltransferase. The H anti gen was found on all blood type O cell lines. However, H-498 cells also expressed H antigen, thereby representing accumula tion of precursor substance in a cell line which is deficient in A transferase activity, (c) In general, the activity of the three aglycosidases did not correlate with blood group antigen expres sion. (</) Demonstration that the measured A, B, and H glyco syltransferase activities represented the action of the true geneencoded A, B, and H transferases was confirmed by HPLC, revealing the predicted tetrasaccharide from the A and B trans ferase reactions, and by paper chromatography, revealing the expected disaccharide from the H transferase reaction. Fur thermore, substrate affinities of the colon cancer-derived A, B,

4554

ABH ANTIGENS IN ADENOCARC1NOMA CELL LINES

and H glycosyltransferases for the saccharide acceptors and nucleotide sugar donors were similar to those of human serum A, B, and H transferases. Immunohistochemical studies have demonstrated that nor mal human colonocytes express A, B, and H antigens compat ible with the host's blood type in the proximal but not distal colon (4, 8-11). However, during fetal life, colonocytes along the entire length of the colon bear the appropriate ABH antigen (8, 9). Thus, there is a tight control of ABH expression that occurs with development of this organ, with an apparent loos ening of this control in malignancy, as manifested by ABH reappearance in distal cancers, ABH deletion of proximal can cers, and incompatible A and B expression (9, 11-18). Very little is known about the mechanisms responsible for these alterations in colonie ABH antigen expression. A few studies have addressed this issue by measuring glycosyltransferases and glycosidases in homogenates of normal and cancerous colonie tissues (36-39). While such an approach is advantageous for comparing normal and cancerous tissues from the same indi vidual, it is limited by the fact that nonepithelial elements, such as RBC, endothelial cells, and glycosidase-producing bacteria might be contributing to the observed results. Therefore, to obtain a better understanding of the mechanisms contributing to ABH expression in malignant colonie epithelial cells, we conducted the present study using well characterized colon cancer cell lines established from individuals of known blood type. No study to our knowledge has investigated mechanisms of ABH antigen biosynthesis in isolated colon cancer cells. In the present study, there were two examples of blood group antigen deletion. Because it is not known whether these cell lines were established from cancers of the proximal or distal colon, it is not clear whether the lack of expression of A and B antigens in H-498 and SW1417 is analogous to the developmental loss of antigen in normal distal colon or to the cancer-associated deletion of antigen in the proximal colon. The H-498 cell line failed to express the A antigen, most likely because of absent A transferase activity. The deletion of A antigen was accompanied by precursor accumulation of H substance in this case. The SW1417 cell line failed to express the B antigen but, unlike in H-498 cells, there appeared to be sufficient levels of the biosynthetic B transferase activity. Enhanced B-degrading enzyme activity did not seem to explain the deletion. Interestingly, in this cell line, there was no accumulation of precursor H antigen, despite very high levels of H transferase activity. This suggests that, despite adequate levels of B and H transferases, precursor synthesis might be impaired. This could occur, for example, if other glycosyltransferases, such as sialyltransferase or a 1,3fucosyltransferase, successfully competed with the H gene a 1,2fucosyltransferase for the type 1 or type 2 oligosaccharide side chain. An alternate explanation for B antigen deletion in these cells could be masking of the immunodeterminant in some way as to make it inaccessible to bind antibody. In terms of incompatible antigen expression, Drewinko and Lichtiger (19) reported incompatible A and B antigen expres sion in all seven colon cancer cell lines examined. However, that study used human polyclonal anti-A and anti-B antisera applied to cell suspensions analyzed by indirect immunofluorescence, and neither glycosyltransferase nor glycosidase activ ities were determined. In the present study, we detected weak binding of monoclonal anti-A to SW1116 (type O) and LoVo (type B) cells, and weak monoclonal anti-B binding to SW1116 and H498 (type A) cells. However, since there was little or no A or B gene-encoded glycosyltransferase activity in these cell lines, this suggests that the antigens being detected are not the
4555

products of the true ABO gene-encoded enzymes. Although the mechanism(s) responsible for incompatible A or B antigen expression is not known, a few possibilities exist. First, the true A or B gene-encoded glycosyltransferases might become activated in cancer cells. While this concept goes against our understanding of classical immunogenetics of blood group substances, one study did find A transferase activity in a colon cancer from a blood type O individual whose tumor tissue expressed incompatible A antigen (18). A second hypothesis to explain incompatible A or B antigen expression might be a loosening of substrate specificity of the A or B transferases. For example, under certain in vitro experimental conditions, the serum A gene-encoded transferase can synthesize the B antigen (48), and the B transferase can synthesize the A antigen (29, 32, 40). A third hypothesis is that the incompatible antigens are really A-like and B-like structures and not the products of the geneencoded transferases. For example, substances with terminal GalNAc residues, such as Forssman antigen (41-43) or Tn antigen (44), behave as A-like structures. However, our mono clonal anti-A antibody does not react with Forssman-positive sheep RBC (9) and has a different staining pattern, compared to specific anti-Tn reagents.5 A-like oligosaccharides have been identified on glycoproteins from human rectal carcinoma (45). Indeed, monoclonal antibodies raised against HT-29 cells react not only with true A substances but also with A-like substances (46). In other studies, naturally occurring antibodies which recognize A-like determinants were found in the sera of two healthy sisters belonging to a kindred at high risk for colon cancer (47). The A antigen can be converted into an "acquired" B antigen by deacetylation of the terminal a-A'-acetylgalactosamine, an uncommon event that has been reported for RBC (48). Alter natively, B-like structures with terminal a-galactose residues, such as fucose-less B antigen, might also account for incompat ible B expression. The B4 isolectin of B. simplicifolia I has greater affinity for B-like Galal,3Gal structures than for the blood group B tetrasaccharide (30). In previous studies we found that purified mucin from LS174T cells reacted strongly with B. simplicifolia I (isolectins A4 and B4) (49) and, by cytochemistry, all of the cell lines used in the present study, except HT-29 and SW1116, bound the B4 isolectin of B. simplicifolia I. This suggests the existence of B-like substances in our cell lines, although we have not been able to demonstrate transfer of galactose to lactose to form a fucosyl B antigen in these same cell lines. The A, B, and H transferases of human colon cancer cells have not previously been characterized with respect to substrate affinity. Our results demonstrate that the Km values for the saccharide acceptors and nucleotide sugar donors of colon cancer cell ABH enzymes were very similar to those reported for these enzymes in human serum. Furthermore, using a sterile cell culture system, our data illustrate the ABH-degrading gly cosidases in the colon, which have been characterized as prod ucts of enteric bacteria (50), may also be produced by colono cytes.

ACKNOWLEDGMENTS
The authors wish to thank Annabelle Friera, Frank Fearney, Robert Lagace, and Dr. Sam Ho for technical assistance and Rita Burns for manuscript editing and preparation.
' Unpublished observations.

ABH ANTIGENS IN ADENOCARC1NOMA CELL LINES

REFERENCES
1. Watkins, M. W. Genetics and biochemistry of some human blood groups. Proc. R. Soc. Lond. B Biol. Sci., 202: 31-53. 1978. 2. Sharon, N. Blood group substances. In: Complex Carbohydrates, Their Chemistry, Biosynthesis, and Functions, N. Sharon (d.), 215-257. MA: pp. Addison-Wesley, 1975. 3. Hakomori, S. Blood group ABH and li antigen of human erythrocytes: chemistry, polymorphism and their developmental changes. Semin. Hematol., 18: 39-62, 1981. 4. Szulman. A. E. The histological distribution of blood group substances A and B in man. J. Exp. Med., ///: 785-800, 1960. 5. Ginsburg, V. Enzymatic basis for blood groups in man. Adv. Enzymol.. 36: 131-148, 1972. 6. Clausen, H., Levery, S. B., Dabelsteen. E., and Hakomori, S. Blood group ABH antigens: a new series of blood group A-associated structures (genetic regulation and tissue distribution). Transplant. Proc., 19: 4408-4412, 1987. 7. Watkins, W. M., and Greenwell, P. Biosynthesis of blood group ABH antigens: genetic regulation and tissue distribution. Transplant. Proc., 19: 4413-4415, 1987. 8. Szulman, A. E. The histological distribution of the blood group substances in man as disclosed by immunofluorescence. III. The A, B, and H antigens in embryos and fetuses from 18 mm in length. J. Exp. Med., 119: 503-515, 1964. 9. Yuan, M., Itzkowitz, S. H., Palekar, A., Shamsuddin, A. M., Phelps, P. C, Trump, B. F., and Kim, Y. S. Distribution of blood group antigens A, B, H, Lewis", and Lewisb in human normal, fetal, and malignant colonie tissue. Cancer Res., 45: 4499-4511, 1985. 10. Szulman, A. E. The histological distribution of blood group substances in man as disclosed by immunofluorescence. II. The H antigen and its relation to A and B antigens. J. Exp. Med., 115: 977-1005, 1962. 11. Dabelsteen. E., Graem. N., Clausen, H., and Hakomori, S. Structural varialions of blood group A antigens in human normal colon and carcinomas. Cancer Res., 48: 181-187, 1988. 12. Cooper, H. S., Cox, J., and Patehefsky, A. S. Immunohistologic study of blood group substances in polyps of the distal colon. Am. J. Clin. Pathol., 73: 345-350, 1980. 13. Cooper, H. S., and Haesler, W. E. Blood group substances as tumor antigens in the distal colon. Am. J. Clin. Pathol., 69: 594-598, 1978. 14. Itzkowitz, S. H., Yuan, M., Ferrell, L. D., Palekar, A., and Kim, Y. S. Cancer-associated alteration of blood group antigen expression in human colorela! olyps. Cancer Res., 46: 5976-5981, 1986. p 15. Ernst, C., Thurin, J., Atkinson, B., Wurzel, H., Herlyn, M., Stromberg, N., Civin, C., and Koprowski, H. Monoclonal antibody localization of A and B isoantigens in normal and malignant fixed human tissues. Am. J. Pathol., 117: 451-461, 1984. 16. Schoentag, R., Primus, F. J., and Kuhns, W. ABH and Lewis blood group expression in colorectal carcinoma. Cancer Res., 47:1695-1700, 1987. 17. Feizi, T., Turberville, C., and Westwood, J. H. Blood-group precursors and cancer-related antigens. Lancet, 2: 391-393, 1975. 18. Clausen, H., Hakomori, S., Graem, N., and Dabelsteen, E. Incompatible A antigen expressed in tumors of blood group O individuals: immunochemical, immunohistologic, and enzymatic characterization. J. Immunol., 136: 326330. 1986. 19. Drewinko, B., and Lichtiger, B. Full expression of blood group-related, transplantation-related, and carcinoembryonic antigens in human colorectal cancer cells with different degrees of phenotypic differentiation. Cancer Res.,
45: 1560-1564, 1985.

27. 28. 29. 39 31

32. 33 34. 35 35 37 3g 39

49

41

42.

43 44

20. Chantret, I., Chevalier, G., Dussaulx, E., and Zweibaum, A. A and H blood group antigens as markers of sucrase-isomaltase from the enterocyte-Iike differentiated human colon carcinoma cell lines HT-29 and CaCo-2. Cancer Res., 47: 1426-1433, 1987. 21. Kuan, S-F., Byrd, J. C., Basbaum. C. B., and Kim, Y. S. Characterization of quantitative muchi variants from a human colon cancer cell line. Cancer Res., 47:5715-5724, 1987. 22. Lowry, O. H., Rosebrough, N. J., Fair, A. L., and Randall, R. J. Protein measurement with the Folin phenol reagent. J. Biol. Chem., 293: 265-275, 1951. 23. Whitehead, J. S., Fearney, F. J., and Kim, Y. S. Glycosyltransferase and glycosidase activities in cultured human fetal and colonie adenocarcinoma cell lines. Cancer Res., 39: 1259-1263, 1979. 24. Chester, M. A., Yates, A. D., and Watkins, W. M. Phenyl 0-D-galactopyranoside as an acceptor substrate for the blood group II gene-associated guanosine diphosphate L-fucose: 0-D-galactosyl o-2-L-fucosyltransferase. Eur. J. Biochem., 69: 583-592, 1976. 25. Trevelyan, W. E., Procter, D. P., and Harrison, J. S. Detection of sugars on paper chromatograms. Nature (Lond.), 166:444-445, 1950. 26. Schachter, H., Michaels, M. A., Tilley, C. A., Crookston, M. C., and Crookston, J. H. Qualitative differences in the /V-acetyl-n-galactosaminyl-

45 45.

47. 4g. 49. 50.

transferases produced by human A1 and A! genes. Proc. Nati. Acad. Sci. USA, 70:220-224. 1973. Whitehead, J. S., Bella, A., and Kim, Y. S. An Y-acetylgalactosaminyltransferase from human blood group A plasma: purification and agarose binding properties. J. Biol. Chem.. 249: 3442-3447, 1974. Kobata, A., and Ginsburg, V. Uridine diphosphate iV-acetyl-n-galactosamine: D-galactose -3-A'-acetyl-D-galactosaminyltransferase, a product of the gene that determines blood type A in man. J. Biol. Chem., 245:1484-1490.1970. Yates. A. D., Feeney, J., Donald. A. R.. and Watkins. W. M. Characterization of a blood group A active tetrasaccharide synthesized by a blood group B gene-specified glycosyltransferase. Carbohydrate Res., 130: 251-260, 1984. Wood. C., Kabat. E. A., Murphy, L. A., and Goldstein, I. J. Immunochemical studies of the combining sites of two isolectins, V, and H.,. isolated from Bandeiraea simplicifolia. Arch. Biochem. Biophys., 198: 1-11, 1979. Carne, L. R., and Watkins, W. M. Human blood group B gene-specified 3-galactosyltransferase: purification of the enzyme in serum by biospecific adsorption onto blood group O erythrocyte membranes. Biochem. Biophys. Res. Commun., 77: 700-707. 1977. Yates, A. D., and Watkins, W. M. The biosynthesis of blood group B determinants by the blood group A gene-specified n-3-A'-acetyl-D-galactosaminyl transferase. Biochem. Biophys. Res. Commun., 109: 958-965, 1982. Kumazaki, T., and Yoshida, A. Biochemical evidence that secretor gene. Se. is a structural gene encoding a specific fucosyltransferase. Proc. Nati. Acad. Sci. USA, 81:4193-4197, 1984. Cheng, P. W., and DeVries, A. Mucin biosynthesis: enzymic properties of human trachal pithelial GDP-i.-fucose:/3-D-galactoside -( e 1,2)-L-fucosyltransferase. Carbohydrate Res., 149: 253-261, 1986. Pann, C.. and Kuhns, W. J. Regulation of glycoconjugate expression on cloned mammalian cells. Exp. Cell Res., 98: 73-78. 1976. Kim, Y. S., Isaacs, R., and Perdomo, J. M. Alterations of membrane glycopeptides in human colonie adenocarcinoma. Proc. Nati. Acad. Sci. USA, 7: 4869-4873, 1973. Kim, Y. S., and Isaacs, R. Glycoprotein metabolism in inflammatory' and neoplastic diseases of the human colon. Cancer Res., 35: 2092-2097, 1975. Kiang, D. T., Kennedy, B. J., and Edstrom, R. D. Levels of blood group synthetic enzymes in human colonie carcinoma. Proc. Soc. Exp. Biol. Med.. 157:411-414, 1978. Yonezawa, S., Matsushita, Y., Muramatsu, H., Arila, Y., Nakamura. T.. Muramatsu, T., and Sato, E. Correlation between glycoconjugate expression and fucosyltransferase activity in human colorectal carcinoma. Acta Pathol. Jpn., 37: 1249-1261, 1987. Greenwell, P., Yates, A. D., and Watkins, W. M. UDP-A'-acetyl-o-galactosamine as a donor substrate for the glycosyltransferase encoded by the B gene at the human blood group ABO locus. Carbohydr. Res., 149:149-170,1986. Hakomori, S., Wang, S. M., and Young, W. W., Jr. Isoantigenic expression of Forssman glycolipid in human gastric and colonie mucosa: its possible identity with "A-like antigen" in human cancer. Proc. Nati. Acad. Sci. USA, 74:3023-3027, 1977. Hattori, H., Uemura, K., Ogata. H.. Katsuyama, T., Taketomi, T.. and Kanfer, J. N. Characterization of glycolipids from the gastric cancer of a patient of p.O, Le(a~,b*) blood type: presence of incompatible blood group antigens in tumor tissues. Cancer Res., 47: 1968-1972, 1987. Breimer. M. E. Adaptation of mass spectrometry for the analysis of tumor antigens as applied to blood group glycolipids of a human gastric carcinoma. Cancer Res., 40: 897-908, 1980. Hirohashi, S,, Clausen, H., Yamada, T., Shimosato, Y., and Hakomori, S. Blood group A cross-reacting epitope defined by monoclonal antibodies NCC-LU-35 and -81 expressed in cancer of blood group O or B individuals: its identification on Tn antigen. Proc. Nati. Acad. Sci. USA, 82: 7039-7043, 1985. Kurosaka, A.. Nakajima, H., Funakoshi, I., Matsuyama, M., Nagayo, T., and Yamashina, I. Structures of the major oligosaccharides from a human rectal adenocarcinoma glycoprotein. J. Biol. Chem., 25:11594-11598, 1983. Gooi, H. C., Hounsell, E. F., Picard, J. K.. Lowe, A. D., Voak, D., Lennox, E. S., and Feizi, T. Differing reactions of monoclonal anti-A antibodies with oligosaccharides related to blood group A. J. Biol. Chem., 260: 1321813224, 1985. Knuth, P., Lloyd, K. O., Lipkin, M., Oettgen. H. F., and Old. L. J. Natural antibodies in human sera directed against blood-group-related determinants expressed on colon cancer cells. Int. J. Cancer, 32: 199-204, 1983. Stayboldt, C., Rearden, A., and Lane, T. A. B antigen acquired by normal A! red cells exposed to a patient's serum. Transfusion (Phila.), 27:41-44, 1987. Byrd, J. C., Nardelli, J., Siddiqui, B., and Kim, Y. S. Isolation and charac terization of colon cancer mucin from xenografts of LS174T cells. Cancer Res., 48: 6678-6685, 1988. Hoskins, L. C., Agustines, M., McKee, W. B., Boulding, E. T., Kriaris, M., and Niedermeyer, G. Mucin degradation in human colon ecosystems: isola tion and properties of fecal strains that degrade ABH blood group antigens and oligosaccharides from mucin glycoproteins. J. Clin. Invest., 75: 944953, 1985.

4556