This article is mentioned in:

Introduction

Transformation-associated alteration of the carbohydrate structures in cellular glycoconjugates, including glycolipids and glycoproteins, occurs frequently in various types of cancer, mainly due to the aberrant expression of glycosyltransferases (1). Detection of these structures, including sialyl lacto-N-fucopentaose (CA19.9, sialyl Lewis a), in sera was successfully applied for the clinical diagnosis of epithelial cancer in gynecologic tissues and the digestive tract (2). However, in comparison to CA125, whose frequency in ovarian carcinomas is higher than that of CA19.9, and the level of which is used for preoperative surgical counseling and planning, the clinical usefulness and cell biological properties of ovarian cancers with CA19.9-carbohydrates have not been clearly elucidated yet (3). Since a number of carbohydrate structures were shown to play a role in the ligands of animal lectins, such as NeuAcα2-6Galβ1-4GlcNAc for CD22 (4), sialyl LeX for P-selectin (5) and sialyl 6-sulfo-LeX for L-selectin (6), the expression of these structures and their modifications may affect the lectin-mediated adhesion related to the invasion and metastasis of cancer cells. Transfection of the α1,2-fucosyltransferase gene into RMG-1 cells resulted in increases in LeY and H-1 glycolipids, and a concomitant decrease in sialylated glycolipids. The transfectants exhibited increased adhesion with mesothelial cells and resistance against an anticancer drug, 5-fluorouracil, in comparison to those of RMG-1 cells (7,8). In addition, significant changes in glycolipids including Lewis-active ones were observed in ovarian carcinoma-derived KF28 cells exhibiting anticancer drug-resistance to cisplatin and taxol, probably due to an alteration of the activities of transporter proteins in regard to the excretion of drugs in glycolipid-rich membrane rafts (9,10). These findings showed that the expression of fucosylated glycolipids exhibiting Lewis- and H-antigenecities is closely correlated to the malignancy of cancer cells, including increased dissemination, metastatic potential and anticancer drug-resistance. However, since Lewis-active glycolipids are constructed of more than five carbohydrates, synthesis occurs through more than five glycosyltransferase reactions, whose activities are regulated by various epigenetic factors, including the concentrations of sugar nucleotides and acceptor glycolipids, pH and divalent cations. Notably, glycolipids in each step of the sequential multi-step reaction serve as substrates for the following step, suggesting that glycosyltransferase reactions determine the overall profile of glycolipids, including cancer-associated ones. Accordingly, the glycolipids in tissues from patients with ovarian carcinomas and cell lines derived from them were quantitatively determined in order to clarify their histologic classification-associated alterations, including Lewis-active ones, and to apply them as molecular markers for determining the malignancy of ovarian carcinomas, similar to those for colorectal carcinomas (11,12).

Materials and methods

Tissue specimens

Histologic classification of the ovarian cancers was performed using criteria defined by the World Health Organization. The serous (7 cases) and mucinous (6 cases) cystadenocarcinoma, clear cell adenocarcinoma (3 cases) and endometrioid carcinoma (3 cases) tissues were obtained from the National Saitama Hospital. Written informed consent to use the specimens in this study was obtained from the patients, and the experimental protocol was approved by the local ethics committee.

Cell lines derived from ovarian cancers

The cell lines used in this experiment were obtained from patients with the following ovarian cancers: HAC-2 and RMG-1 from clear cell adenocarcinomas, 2008 and KF28 from serous cystadenocarcinomas, HMKOA from mucinous cystadenocarcinomas and HNOA from endometrioid carcinomas. The cell lines were cultured in Dulbecco’s modified Eagle’s medium (Nissui, Tokyo, Japan) supplemented with 10% fetal bovine serum (Nichirei Biosciences Inc., Tokyo, Japan) under a humidified atmosphere containing 5% CO2 at 37°C.

Materials

The glycolipids used in this experiment were purified from various sources in our laboratory: GM3 and IV3NeuAcα-nLc4Cer from human erythrocytes, GD3 from bovine brain, LeX from human fetal brain and Lc4Cer, Leb and IV6NeuAcα-nLc4Cer from human meconium (9,10).

Antisera

Monoclonal antibodies, Y916 and 5h6, were prepared by the immunization of mice with gangliosides from bovine milk and IV3NeuAcα-nLc4Cer from human erythrocytes, respectively, and the hybridization of lymphocytes with myeloma P3-X63-Ag8.653 in our laboratory (13,14). As shown in Fig. 1, Y916 reacted with GD3 and IV6NeuAcα-nLc4Cer, and 5h6 with GM3, GD3 and IV3NeuAcα-nLc4Cer, indicating that structural isomers, IV6NeuAcα- and IV3NeuAcα-nLc4Cer, are identified with Y916 and 5h6. Moreover, GM3 and GD3 were identified by their mobility on thin layer chromatograpy (TLC) and their reactions with Y916 and 5h6. The following monoclonal antibodies were kindly donated: NCC-LU-279 for LeX and NCC-ST-433 for LeY by Dr S. Hirohashi, National Cancer Center, Tokyo, Japan, MSN-1 for Leb by Dr S. Nozawa, Keio University, Tokyo, Japan, and 3C11 for sialyl Lea by Dr K. Matsumoto, Mikuri Immunol. Lab., Kyoto, Japan.

Figure 1 Thin layer chromatography (TLC) and TLC-immunostaining of glycolipids with monoclonal antibodies Y916 and 5h6. (1) GM3, (2) IV3NeuAcα-nLc4Cer, (3) GD3 and (4) IV6NeuAcα-nLc4Cer were developed on plastic-coated TLC plates with chloroform/methanol/0.5% CaCl2 in water (55:45:10, by volume). Spots were visualized with orcinol-H2SO4 reagent and monoclonal antibodies Y916 and 5h6.

Analysis of lipids

The neutral and acidic glycolipids derived from the tissues and cells were examined by TLC and TLC-immunostaining with the development solvents, chloroform/methanol/water (65:35:8, by volume) for neutral glycolipids and chloroform/methanol/0.5% CaCl2 (55:45:10, by volume) for gangliosides, as previously described (10,13). Known amounts (0.1–1.5 μg) of glycolipids, such as GM3, Lc4Cer, IV3NeuAcα-nLc4Cer, IV6NeuAcα-nLc4Cer and GD3, were developed on the same TLC plates for the preparation of standard curves. The densities of spots on TLC plates were determined by image analysis using NIH image.

α2,3- and α2,6-sialyltransferases

The cancer tissues were homogenized using a homogenizer (Polytron; Kinematica, Luzern, Switzerland) in 0.25 M sucrose, and the microsomal fractions were prepared by centrifugation as previously described (15). The standard assay mixture for microsomal α2,3- and α2,6-sialyltransferases comprised 7.6 nmol nLc4Cer, 10 mM MgCl2, 5 mM CaCl2, 10 mM CMP-sialic acid, 0.3% Triton CF54, 50 mM 4-morpholinoethane sulfonic acid-NaOH buffer (pH 6.4), and 50 μg enzyme protein, in a final volume of 50 μl (16,17). After incubation at 37°C for 3 h, the reaction was terminated with 100 μl of ethanol. Then, 50 μl aliquots of the solution were developed on two TLC plates with chloroform/methanol/0.5% CaCl2 in water (55:45:10, by volume), detection being performed with 5h6 for one and with Y916 for the other. Known amounts of IV3NeuAcα-nLc4Cer and IV6NeuAcα-nLc4Cer (5–100 ng) were stained on the same plates, and the densities of the spots were determined by image analysis using NIH image. The amounts of endogenous gangliosides in the microsomes were subtracted from the values after the enzyme reactions.

RT-PCR analysis

Total RNA extracted from the tissues with Isogen (Nippongene, Toyama, Japan) was reverse-transcribed to cDNA with reverse transcriptase (M-MuLV; Takara, Kyoto, Japan) and oligo dT-primers, and then subjected to PCR with 0.5 units of Taq DNA polymerase (GoTaq; Promega, Kyoto, Japan) under the following conditions: LacCer sialyltransferase (GM3 synthase, AB018356), sense primer, atttgagcacaggtatagc, antisense primer, gatgtcaaaggcagtctct; GM3 sialyltransferase (GD3 synthase, D26360), sense primer, acaaatggaagactgctgcga, antisense primer, tggctctgt tcctgtcttcat; α2,6-sialyltransferase (BC031476), sense primer, tgcgtcctggtctttcttct, antisense primer, tctgcactgaacttgatgcc; α2,3-sialyltransferase (BC010645), sense primer, atctcccg ggaagacaggta, antisense primer, ccatgaagaaggggttgaga; and α1,3-fucosyltransferase 3 (FUT3, NM1097640), sense primer, tggtggctgtgtgtttcttc, antisense primer, ggctccaagttgaaccagat; 35 cycles of 95°C for 15 sec, 54–64°C for 30 sec and 72°C for 40 sec. The primers for glyceraldehyde 3-phosphate dehydrogenase (GAPDH) were used as controls. The resulting PCR products were electrophoresed on a 1.5% agarose gel, stained with ethidium bromide, and examined using a UV trans-illuminator (15).

Results

Glycolipids in ovarian carcinoma tissues

Fig. 2 shows TLC-immunostaining of lipids from a number of ovarian carcinoma and uterine endometrial carcinoma tissues. In agreement with our previous results (7), Leb in mucinous cystadenocarcinomas was present in higher amounts than those in the other carcinomas (Table I). Among the Lewis glycolipids examined, Leb, LeY, LeX and sialyl Lea, whether one or a number of them, were detected in all tissues other than ovarian serous cystadenocarcinoma ones, in which they were not present or only in a trace amount. Alternatively, serous cystadenocarcinomas contained IV6NeuAcα-nLc4Cer in significantly higher amounts than in the other carcinomas (Fig. 2). The amounts of IV6NeuAcα-nLc4Cer in serous cystadenocarcinomas were >9 times higher than those in the other carcinomas, while no significant differences were observed in the amounts of IV3NeuAcα-nLc4Cer among the various types of ovarian carcinomas (Fig. 2 and Table I). The expression of IV6NeuAcα-nLc4Cer and Leb was further examined in an additional 5 cases of ovarian serous and mucinous cystadenocarcinomas, respectively. As shown in Fig. 3 and Table II, IV6NeuAcα-nLc4Cer was present in 3/5 serous cystadenocarcinomas in amounts >0.1 μg per mg of dry weight, but not in mucinous cystadenocarcinomas. Conversely, Leb was detected in 4/5 mucinous cystadenocarcinomas and in one of the serous cystadenocarcinomas. Thus, the frequencies of expression of IV6NeuAcα-nLc4Cer and Lewis-active glycolipids were significantly high in serous cystadenocarcinomas and the other ovarian carcinoma tissues, respectively.

Figure 2 TLC-immunostaining of glycolipids from human ovarian carcinoma tissues. Glycolipids, corresponding to 0.2 mg dry weight, were developed with chloroform/methanol/0.5% CaCl2 in water (55:45:10, by volume). Spots were visualized by TLC-immunostaining with anti-Leb, anti-LeY, anti-LeX, anti-sialyl Lea, and Y916 and 5h6 antibodies. The specimen numbers are the same as those shown in Table I. Standard glycolipids for Y916 was a mixture of GD3 and IV6NeuAcα-nLc4Cer.

Figure 3 TLC-immunostaining of glycolipids from human ovarian mucinous and serous cystadenocarcinoma tissues. Glycolipids were obtained from an additional five specimens, respectively, i.e., other than those in Fig. 2 and Table I. The glycolipids, corresponding to 0.1 mg dry weight, were used for TLC-immunostaining with monoclonal Y916 and anti-Leb antibodies as described in Materials and methods. Standard glycolipids for Y916 was a mixture of GD3 and IV6NeuAcα-nLc4Cer.

Table I Glycolipids in human ovarian carcinoma tissues.

Table I

Glycolipids in human ovarian carcinoma tissues.

No.	Histological classification	Specimen Case	LeX	LeY	Leb	Sialyl Lea	GM3	GD3	α2,6	α2,3
1	Ovarian serous cystadenocarcinoma	1	–	–	–	–	1.19	0.11	0.45	0.04
2		2	–	–	–	–	1.25	0.08	0.45	0.06
3	Ovarian mucinous cystadenocarcinoma	1	–	0.01	0.53	0.04	0.54	–	0.01	–
4	Endometrial adenocarcinoma	1	–	0.30	0.06	0.04	0.48	0.08	0.02	0.02
5	Ovarian endometrioid carcinoma	1	0.09	0.09	0.02	–	1.33	–	0.04	0.03
6		2	–	tr	0.03	0.02	0.50	0.01	0.05	0.02
7		3	–	–	0.02	0.04	0.90	0.07	0.01	tr
8	Ovarian clear cell adenocarcinoma	1	–	0.01	0.01	–	0.37	–	–	tr
9		2	–	tr	0.01	0.01	0.15	tr	0.02	0.01
10		3	–	0.20	tr	–	0.22	0.02	0.02	0.01

[i] The amounts of glycolipids are expressed as μg per mg dry tissue and are the means of three determinations by TLC-densitometry with the respective standard glycolipids. α2,6, IV⁶NeuAcα-nLc₄Cer ; α2,3, IV³NeuAcα-nLc₄Cer; tr, trace amount. Sample numbers correspond to those in Fig. 2.

Table II Amounts of IV6NeuAcα-nLc4Cer and Leb in ovarian serous and mucinous cystadenocarcinomas.

Table II

Amounts of IV6NeuAcα-nLc4Cer and Leb in ovarian serous and mucinous cystadenocarcinomas.

No.	Histological classification	Specimen Case	α2,6	Leb
1	Ovarian serous cystadenocarcinoma	3	0.06	0.01
2		4	0.23	—
3		5	0.37	—
4		6	0.11	—
5		7	0.01	0.13
6	Ovarian mucinous cystadenocarcinoma	2	—	0.01
7		3	—	0.12
8		4	—	0.02
9		5	0.01	0.08
10		6	—	0.48

[i] The amounts of glycolipids are expressed as μg per mg dry tissue and are the means of three determinations by TLC-densitometry with the respective standard glycolipids. Sample numbers correspond to those in Fig. 3.

Glycolipids in ovarian carcinoma-derived cells

The expression of IV6NeuAcα-nLc4Cer and Lewis-active glycolipids was examined in cell lines established from various types of ovarian carcinomas. Although IV6NeuAcα-nLc4Cer was not detected in any cell line, Lewis-active glycolipids were present in clear cell adenocarcinoma-derived RMG-1 and HAC-2, mucinous cystadenocarcinoma-derived HMKOA and endometrioid carcinoma-derived HNOA, in amounts higher than those in serous cystadenocarcinoma-derived 2008 and KF28. Their presence shows that relatively low and high amounts of Lewis-active glycolipids in serous cystadenocarcinomas and the other ovarian carcinoma tissues, respectively, are retained in the respective cell lines (Fig. 4).

Figure 4 TLC-immunostaining of glycolipids from ovarian carcinoma-derived cells. Glycolipids, corresponding to 0.2 mg dry weight, were used for TLC-immunostaining with anti-Leb and anti-LeY antibodies. Lanes: 1, RMG-1; 2, HAC-2; 3, HMKOA; 4, HNOA; 5, 2008 and 6, KF28.

Enzyme activities and gene expression of α2,3- and α2,6-sialyltransferases in ovarian carcinoma tissues

To examine the enzymatic and genetic backgrounds of the expression of IV6NeuAcα-nLc4Cer and Lewis-active glycolipids, the specific activities of α2,3- and α2,6-sialyltransferases were determined by detection of the products, IV3NeuAcα-nLc4Cer and IV6NeuAcα-nLc4Cer, and the gene expression by RT-PCR.

As shown in Fig. 5 and Table III, although the specific activities of α2,3-sialyltransferase with nLc4Cer as the substrate in the tissues were not correlated with the amounts of IV3NeuAcα-nLc4Cer, or with the histological classification, those of α2,6-sialyltransferase were positively correlated with the relative amounts of IV6NeuAcα-nLc4Cer in the tissues, indicating that the high amounts of IV6NeuAcα-nLc4Cer in serous cystadenocarcinomas are due to the higher specific activity of α2,6-sialyltransferase. However, the α2,3- and α2,6-sialyltransferase genes were ubiquitously expressed in all of the tissues examined and their relative intensities were not positively correlated with the enzymatic activities, or with the amounts of glycolipids (Fig. 6). Similarly, expression of the FUT3 gene encoding an α1,3/4 fucosyltransferase responsible for the synthesis of Lewis antigen was not correlated with the amounts of Leb, LeX, LeY and sialyl Lea. In contrast to the gene expression of sugar transferases for neolacto-series glycolipids, the relative intensities of the GM3 and GD3 synthase genes were positively correlated with the amounts of GM3 and GD3 (Fig. 6).

Figure 5 TLC-immunostaining of products following reactions of α2,3- and α2,6-sialyltransferases (SAT). The products following the reactions with 50 μg of enzyme proteins were developed with chloroform/methanol/0.5% CaCl2 in water (55:45:10, by volume), and detected by TLC-immunostaining with monoclonal antibodies 5h6 for IV3NeuAcα-nLc4Cer (α2,3SAT) and Y916 for IV6NeuAcα-nLc4Cer (α2,6SAT), respectively. Standard glycolipids IV3NeuAcα-nLc4Cer for α2,3SAT and IV6NeuAcα-nLc4Cer for α2,6SAT, respectively. The specimen numbers are the same as those in Fig. 2 and Table I.

Figure 6 RT-PCR analysis of the glycosyltransferase genes in human ovarian carcinoma tissues. A, GAPDH; B, GM3 synthase; C, GD3 synthase; D, α2,3-sialyltransferase; E, α2,6-sialyltransferase and F, α1,3/4-fucosyltransferase 3. The specimen numbers are the same as those in Fig. 2 and Table I.

Table III Specific activities of α2,3- and α2,6-sialyltransferases with nLc4Cer as the substrate.

Table III

Specific activities of α2,3- and α2,6-sialyltransferases with nLc4Cer as the substrate.

No.	α2,3-sialyltransferase	α2,6-sialyltransferase
1	7.4	57.2
2	8.5	105.2
3	1.0	21.7
4	11.6	12.4
5	12.0	7.5
6	5.5	10.9
7	5.3	nd
8	5.2	nd
9	3.9	nd
10	11.0	16.1

[i] Specific activities (pmol/100 μg of enzyme protein/h) are calculated from the amounts of products by TLC-immunostaining as shown in Fig. 5. Microsomes that served as the enzyme source were prepared from the same tissues as those in Table I and Fig. 2. nd, not detected.

Discussion

Among ovarian carcinoma tissues with different histologic classifications, serous cystadenocarcinomas have been shown to express IV6NeuAcα-nLc4Cer at a higher frequency than the other carcinomas, and in compensation, the expression of Lewis-related glycolipids in serous cystadenocarcinomas was lower than in the other carcinomas. As shown in Fig. 7, since the syntheses of IV6NeuAcα-nLc4Cer and Lewis-active glycolipids occur with the same substrate at the branch of the lacto-series pathway, our findings suggest that the synthesis of lacto-series glycolipids is influenced by the availability of substrate glycolipids in individual steps of glycosyltransferase reactions. As reported in our previous paper (8), when the activity of α1,2-fucosyltransferase in ovarian carcinoma-derived RMG-1 cells increased to 20–30 fold that in the original cells on transfection with the α1,2-fucosyltransferase gene, the amount of LeY increased to 10-fold of the original level. On the other hand, sialylated glycolipids in the original cells, including sialyl LeX and IV3NeuAcα-nLc4Cer, were absent in the transfectants, suggesting that the enhanced fucosylation of LeX and nLc4Cer at the terminal step of glycosylation inhibits their sialylation through deprivation of the substrate. Similarly, the high specific activity of α2,6-sialyltransferase in serous cystadenocarcinomas was considered to cause the increased amount of IV6NeuAcα-nLc4Cer (Lc4Cer) and the absence of the Lewis antigen due to the consumption of nLc4Cer (Lc4Cer) for the syntheses of LeX (Lea) by α1,3/4-fucosyltransferase (FUT3), blood group H-glycolipid by α1,2-fucosyltransferase (FUT2) and IV6NeuAcα-nLc4Cer by α2,3-sialyltransferase (α2,3SAT), whose mRNAs were expressed in all of the tissues examined (18). Therefore, the epigenetic regulation of enzymatic activities in the individual steps of the neolacto- and lacto-series pathways may be involved in the determination of the mode of expression of sialylated and fucosylated glycolipids, including Lewis antigens, irrespective of the expression of glycosyltransferase mRNA. In contrast, the amounts of GM3 with shorter carbohydrate chains were closely correlated to the relative intensities of GM3-and GD3-synthase mRNAs, whose expression may directly lead to the active syntheses of GM3 and GD3 with a sufficient supply of substrate LacCer in proportion to their enzymatic activities. However, among ovarian carcinoma-derived cells, Lewis-related glycolipids were present in mucinous cystadenocarcinoma-, clear cell adenocarcinoma- and endometrioid carcinoma-derived cells in significantly higher amounts than in serous cystadenocarcinoma-derived cells, suggesting that the synthetic potential as regards to Lewis glycolipids in the tissues is maintained in the cultured cell lines. On the other hand, IV6NeuAcα-nLc4Cer was present in 5/7 serous cystadenocarcinoma tissues in amounts of more than 0.1 μg per mg dried tissue, while the amounts in the other carcinomas, if present, were less than 0.05 μg per mg dry weight, indicating that the expression of IV6NeuAcα-nLc4Cer in serous cystadenocarcinomas occurs at a higher frequency than in the other carcinomas. In agreement with our results, the frequency of detection of sialyl Lea (CA19.9) in sera of patients with ovarian serous cystadenocarcinomas was reported to be low in comparison to those in the other carcinomas (3). In the case of a murine lymphoblastoid cell line, cells with IV6NeuAcα-nLc4Cer were shown to exhibit a low metastatic potential in comparison to those without it, probably due to the attenuated expression of adhesion-related Lewis structures, due to enhanced α2,6-sialyltransferase activity (19). Transfection of the α2,6-sialyltransferase gene into cell lines with Lewis glycolipids is currently under investigation to demonstrate the modification of the Lewis glycolipid expression. In addition, since serous cystadenocarcinomas generally show a favorable prognosis, it can be suggested that IV6NeuAcα-nLc4Cer is a useful marker for the benign properties of cancer cells. To demonstrate the value of screening for IV6NeuAcα-nLc4Cer and Lewis-active glycolipids for the diagnosis of ovarian carcinomas, amounts of these glycolipids in sera of patients with serous cystadenocarcinomas are now being determined in comparison to those in tissues in our laboratory.

Figure 7 Metabolic pathways for the syntheses of IV6NeuAcα-nLc4Cer and Lewis-active glycolipids. SAT, sialyltransferase; FUT2, α1,2-fucosyltrans-ferase; FUT3, α1,3/4-fucosyltransferase; H, blood group H antigen.

References