Structural changes induced in the sugar chains of glycoproteins by malignant transformation of producing cells and their clinical application

Biochimie 70 (1988) 1575-1585 © Soci~t~ de Chimie biologique/Elsevier, Paris

1575

Structural changes induced in the sugar chains of glycoproteins by malignant transformation of producing cells and their clinical application* Akira KOBATA

Department of Biochemistry, Institute of Medical Science, University of Tokyo, 4-6-1 Shirokanedai, Minato-ku, Tokyo 108, Japan (Received 24-12-1987, "accepted 21-3-1988)

Summary - - Altered glycosylation is widely observed in glycoproteins produced by tumors. One of the most consistently observed alterations is the increase of larger asparagine-linked sugar chains in the plasma membrane glycoproteins. This phenomenon is brought about by the increase of N-acetylglucosaminyltransferase V, which is responsible for the formation of the GlcNAc/31--,,6Mant~-l--~6 group. The enrichment of the complex-type sugar chains containing the -GicNAc/31--->6(-GlcNAc/31-->2)Manal--->6 group is correlated with tumorigenicity and metastasic potential of tumor cells. Comparative study of the sugar chains of human chorionic gonadotropin isolated from the urine of pregnant women and of patients with trophoblastic diseases including choriocarcinoma revealed that many new oligosaccharides are included in the tumor hCG. The altered glycosylation of hCG is brought about by the ectopic expression of N-acetyiglucosaminyltransferase IV. With use of this altered glycosylation, a novel method useful for the diagnosis of choriocarcinoma was established. oligosaccharide / glycoprotein / tumor / metastasis / diagnosis Introduction Most proteins in mammals are glycosylated and are called glycoproteins. The carbohydrate moieties of many glycoproteins play important roles in cell - cell recognition, which is essential in fertilization [1], morphogenesis [2] and the maintenance of the ordered social life of each cell in mature multicellular organisms [2]. Cells are thought to recognize, through specific carbohydrate-binding proteins in their plasma membrane, the carbohydrates of glycoproteins on adjacent cells or in the extracellular matrices. Accordingly, alterations in the carbohydrate structures of glycoproteins found in various tumors are considered to be the basis of the abnormal social behavior of tumors cells, such as

invasion into the surrounding tissues and metastasis. Therefore, comparative studies of sugar chains of glycoproteins produced by malignant cells versus normal cells will provide useful information concerning the pathology, diagnosis, prognosis as well as immunotherapy of tumors. Usually, the glycoproteins produced by normal and malignant cells can be obtained only in small amounts. Because of this situation, studies on their carbohydrate moieties were performed using rather indirect methods, such as monosaccharide analysis, comparative study of the molecular sizes of the glycopeptides obtained by exhaustive pronase digestion and reactivities with several lectins. However, the recent development of a variety of sensitive methods to analyze the oligosaccharide structures [3] has enabled

*This work has been supported by Grant-in-Aid for Special Project Research (Cancer-Bioscience) from the Ministry of Education, Science and Culture of Japan. Abbreviations: NeuAc. N-acetylneuraminic acid; Gal: galactose; GIcNA¢: N-acetylglucosamine; GalNAe: N-acetylgalactosamine; Fue: fucose.

1576

A. Kobata

the elucidation of the exact structural alterations of sugar chains in several tumor glycoproteins to begin. This review will introduce two recent topics out of our currently available knowledge on this line of study to help readers consider the importance of this new field of cancer research.

A

GI¢

Vo

~

20O

Structural alteration of the sugar chains of

plasma membrane glycoprotems produced by malignant transformation of fiBroblasts Warren- Glick phenomenon One of the most consistently observed alterations induced by neoplastic transformation is a shift towards the synthesis and expression of larger asparagine-linked oligosaccharides in the plasma membrane glycoproteins. The phenomenon was reported for the first time by Meezan et al., in 1969 [4]. They cultured mouse 3T3 fibroblasts and their Simian virus-transformants in a medium containing radioactive glucosamine. When the plasma membrane glycoproteins, whose sugar chains were metabolically labeled, were exhaustively digested with pronase and analyzed by a Sephadex G - 5 0 column, the major glycopeptides from transformed cells were larger than those from non-transformed cells. This interesting observation propelled many investigators to analyze giycopeptides in other cell systems. Warren's group [5, 6] examined BHK2t/C13 cells, a hamster cell line, and the cells transformed with an RNA tumor virus, Rous sarcoma virus (RSV) and two DNA tumor viruses, polyoma virus (Py) and SV-40. They also analyzed mouse 3T3 cells, chick embryo fibroblasts and their RSV transformants [6]. In every case, large glycopeptides from the surface of transformed cells were highly increased. Since Warren and Glick energetically studied and published many papers on this phenomenon, it is nosy widely known by the name of Warren(;licV: phenomenon. A; shown in Fig. 1 as a typical example, the difference was also observed when radioactive fucose was used for metabolic labeling of glycoproteins. This phenomenon was observed not only in virus-transformed cells but in chemically and spontaneously transformed cells [7, 8]. The cells transformed with a temperature-sensitive mutant of RSV showed the presence of the large glycopeptides at permissive temperature, where

I

I

I

~0

50

60

I

I

z0

80

I

go

I "-

too

"~[

'" __

ira

FRACTION NUMBER

Fig. 1. Gel-permeation chromatography of radioactive glycopeptide fractions obtained from fucose-labeled cell surface glycoproteins by exhaustive pronase digestion, o - - ~ : Py-BHK cells; o - - - o : BHK cells.

the transformed phenotype is expressed; shift of the cells to a non-permissive temperature resulted in the disappearance of the transformed phenotype and the large glycopeptides [9]. The same observation was extended to non-fibrohlastic cells, including rat hepatoma cells [10], human leukemic cells [11], lymphoma cells [11] and neuroblastoma cells [12]. Therefore, it is strongly suggested that the increment of the large glycopeptides is a general phenomenon detected in transformed and tumorigenic cells, regardless of ~--" u species and ,h,. •.,, manner of transformation.

Structural basis of phenomenon

the

Warren-Glick

Extensive analyses had been performed on the change in the elution profiles of glycopeptides in gel-filtration, but the structural basis for the change remained unclear. Many people examined the possible involvement of sialylation in the appearance of the large glycopeptides. In fact, the size difference between glycopeptides of normal and transformed cells almost vanished when the glycopeptides were exhaustively digested with sialidase before gel-permeation chromatography. Based on this evidence, Warren et al. [13] suggested that the difference between the large glycopeptides from transformed cells and the small glycopeptides from normal counterparts should be ascribed only to the amount of sialic acid residues included in their sugar chains,

1577

Altered glycosylation in tumors

and proposed that sialylation is a key step in the inducement of the alteration of cell surface sugar chains by malignant transformation. The levels of sialyltransferase activities in normal and transformed cells were comparatively studied by several investigators for the purpose of confirming Warren's hypothesis. However, the results were not favorable: in some experiments, the enzyme level was higher, whereas in others it was lower in transformed cells compared to normal counterparts [13, 14]. Through studies of behavior of various oligosaccharides and glycopeptides of known structures in a concan~valin A (Con A)-Sepharose column, we found that column chromatography can be used as a valuable tool to estimate the structures of complex-type asparagine-linked sugar chains [15]. Comparative study of the glycopeptides (Fig. 1) obtained from metabolically labeled Py-BHK and BHK cells by a Con A Sepharose column [16] revealed that the structural change in the surface sugar chains of transformed cells are not that simple as suggested by Warren et al. When the fraction GP-Is in Fig. 1 obtained from normal and transformed cells were subjected to Con A-Sepharose column chromatography, most of them bound to the column and were eluted with 0.1 M a-methylmannopyranoside solution (Fig. 2a, b). On the

a

"3

.....

other hand, all of the fraction GP-II from PyBHK cells passed through the column without •retardation (Fig. 2c). The elution profile did not change even after sialidase digestion (Fig. 2d, dotted line), indicating that the difference between GP-I and GP-II cannot be ascribed only to the difference in the amount of sialic acid residues. When the asialo GP-II was further digest ~d with jack bean/3-galactosidase and/3-N-acetylhexosaminidase and then analyzed on a Con A Sepharose column, most of the glycopeptides bound to the column (Fig. 2d, solid line). That the glycopeptide at this stage has the structure: Mana 1--*6(Manod-*3)ManBl-~4GlcNAc431-,4(Fucal-,6)GlcNAc--~Asn, was confirmed by its conversion to Fucal--~6GlcNAc--,Asn after digestion with endo-B-N-acetylglucosaminidase D. Based on the binding specificity of a Con A Sepharose column [15], the results indicated that the structure of the major oligosaccharide in fraction GP-I was a biantennary complex-type sugar chain and fraction GP-II was probably a mixture of higher antennary complex-type sugar chains. This estimation was structurally proven by our study performed in 1984 [17]. Oligosaccharides released from Py-BHK cells by hydrazinolysis were found to be more enriched in the complex-type sugar chains, containing the 2,6branched outer chain as well as the elongated

b

....

o

1

MM

^

I

0

o

l°2 10

20 Fraction number

10

20

Fig. 2. Con A - Sepharose column chromatography of radioactive glycopeptide fractions shown in Fig. 1. The column was first eluted with 0.01 M Tris--HCI buffer, containing 0.1 M NaCI, pH 7.5, and then with the same buffer containing 0.1 M a-methylmannopyranoside (MM). a. Fraction GP-I from BHK ceils, b. GP-I from Py-BHK ceils, e. GP-II from Py-BHK ceils, d. GP-II treated with sialidase (dotted line) and then with a mixture of jack bean/3-galactosidase and/3-N-acetylhexosaminidase (solid line).

A. Kobata

1578

GnT V GnT II

GIcN Ac 13IN~ c 172Manc~ 1,~ GIcNAcS" 6

GnT III (bisecting GIcNAc) . . . .

+ Fucc~l 6

GIcNAc131-~4Man131-,4C, IcNAc~l-~qGIcNAc

GnT IV

G IcNAc 13l ' , ~ U a n a 1 / 3

OnT I

GIcNAct31 ' j 2

IFig.3. N-Acetylglucosaminyltransferases(GnTs)responsiblefor the formationofcomplex-typeasparagine-linkedsugarchains.

outer chains with N-acetyllactosamine repeats, than those from BHK cells.

Enzymatic basis of the alteration of surface sugar chains in transformed cells So far, 5 different N-acetylglucosaminyltransferases (GnTs), shown in Fig. 3, have been found to transfer g-N-acetylglucosaminyl residues to the different po:dtions of the trimannosyl core commonly found in the complex-type sugar chains [18]. By comparative study of the levels of the 5 GnTs in the homogenates of Py-BHK and BHK cells, it was found that GnT V activity was increased in Py-BHK cells compared to BHK cells [19]. In contrast, same levels of GnTs I, II and IV were detected in both cell lines. GnT III was not detected at all in either cell type. The last evidence agrees with the fact that no bisected sugar chain was detected in the surface glycoproteins of both cell lines [17]. Recently, Van den Eijnden and Schiphorst [20] reported that another N-acetylglucosaminyltransferase, responsible for the outer chain elongation by adding /3-N-acetylglucosaminyl residue to the Gal/31---~4GIcNAc group, works most favorably on the sugar chains with the Gal/31--->4GIcNAc/3F->6(Gall31---~4GIcNAc/31---~2)Manal---~6 group. Therefore, the increase of GnT V in Py-BHK cells explains and enrichment of tri- and tetraantennary complex-type sugar chains, as well as those with elongated outer chains in Py-BHK cells as compared to BHK cells. Later on, increase of the complex-type sugar chains with a 2,6-branched outer chain and of those with the elongated outer chains was found

to occur in RSV-BHK cells by Pierce and Arango [21]. They also reported that the GnT V level was increased in RSV-BHK cells as compared to BHK cells [22]. Therefore, the increase of GnT V might well be considered as the enzymatic basis of the Warren-Glick phenomenon.

Correlation of the Warren- Glick phenomenon with tumorigenicity Although a high correlation between the glycopeptide changes and tumorigenicity has been found [23-25], there are a few reports indicating that some transformed cells do not possess the large glycopeptide in spite of their showing the phenotype of tumor cells as judged by in vitro criteria [25, 26]. We have also found that the Pytransformant of NIH 3T3 cells are not enriched in the complex-type oligosaccharide with a 2,6branched outer chain (unpublished results). These cells have the characteristics of malignancy, such as growth in soft agar and high saturation density in tissue culture. More detailed study of variant cells of Py-NIH 3T3 revealed that the appearance of the large glycopeptides seems to be highly correlated with tumorigenicity of the malignant cell lines (Kobata et al:, manuscript in preparation). Dennis and coworkers recently reported that enrichment of the 2,6-branch is strongly correlated with the acquisition of metastatic potential of murine mammary carcinoma [27]. Therefore, elucidation of the mechanism that increases the level of GnT V in malignant cells may provide the key to understanding the tumor progression caused by increased metastatic potential.

Altered glycosylation in tumors Table !. Classification of tumor markers. Group

Major markers

Oncofetai proteins a-fetoprotein carcinoembryonic antigen pancreatic oneofetal antigen basic fetoprotein Tumor-related antigens

myeloma proteins isofen'itin CA 19-9 CA 125

Hormones

hCG ACTH MSH

Isoenzyme.~

alkaline phosphatase 7-glutamyltranspeptidase aldolase prostatic acid phosphatase amylase neuron-specific enolase

Altered gly.cosylation of human chorionic gonadotropm Bioactive materials which are produced in abno~ally large amounts or in an ectopic manner in tumor cells have been used clinically as diagnostic markers of tumors. Table I summarizes the well-known markers so far reported. Although these markers are giycoproteins, their sugar moieties have been ignored for many years because no structural information was available. However, our recent studies on the sugar chains of y-glutamyltranspeptidase and human chorionic gonadotropin (hCG) revealed that the structural alteration of their sugar chains in tumors can be used effectively to increase the diagnostic value of these tumors markers. Since the story of y-glutamyltranspeptidase has already been introduced elsewhere [28], the research on hCG will be described in this review.

Structures o f the sugar chains o f hCG hCG is a glycoprotein hormone produced by trophoblasts in placenta. It is composed of 2 subunits with apparent molecular weights of 16 000 (t~) and 30 000 (/3). Both subunits contain 2 as-

1579

paragine-linked sugar chains [29, 30]. The B-subunit also contains 4 mucin-type sugar chains close to its C-terminal portion [30]. The whole structures of the asparagine-linked sugar chains of hCG were elucidated, as shown in Fig. 4, based on the structural analysis of oligosaccharides released from urinary hCG by hydrazinolysis [31]. Shortly after our report, Kessler et al. [32] confirmed the occurrence of A1 in urinary hCG. They also elucidated the structure of the mucin-type sugar chain of hCG, as A4 shown in Fig. 5 [33]. Recently, Cole et al. [34] reported that the mucin-type sugar chains of hCG are more complicated, containing sialyl derivatives of the GalB1---> 4GIcNAcB1---~6(GalB1---~3)GalNAc as minor components (Fig. 6). Structurally, the 4 oligosaccharides, A 2 - A 5 in Fig. 4 can be considered as incomplete biosynthetic products of A1. Unlike nucleic acids and proteins, no template is included in the biosynthesis of the sugar chains of glycoproteins. Even a very complicated sugar chain is formed only by the concerted action of glycosyltransferases. In this biosynthetic mechanism, the structure of the sugar chain formed is determined by the substrate specificity of each glycosyltransferase for a particular sugar nucleotide and for a particular glycose acceptor, and by its ability to synthesize a particular type of linkage [35]. Therefore, many factors, such as the shortage of certain sugar nucleotides or change in the relative a~ivi*ies of glycosyltransferases, can induce changes in the oligosaccharide pattern of a glycoprotein. Accordingly, microheterogeneity has long been considered an inherent c~ aracteristic of the sugar chains of glycoproteins. However, a study, which will be described below, revealed that this is not the case in the asparagine-linked sugar chains of hCG.

Site-specific sugar chain structures of hCG By sialidase treatment, the 5 acidic oligosaccharides in Fig. 4 are converted into the 3 neutral oligosaccharides shown in Fig. 6. Actually, the asiaiooligosaccharide fraction from urinary hCG was separated into the 3 peaks shown in Fig. 7A. Analysis of the oligosaccl'~aride fractions obtained from or- and B-subunits of hCG revealed that the a-subunit contains equal amounts of oligosaccharides N-2 and N-3 (Fig. 7B) and the Bsubunit contains N-1 and N-2 in a 1:1 molar ratio (Fig. 7C) [36]. These results indicated that oligosaccharides N-2 and N-3 should not be consider-

A. Kobata

1580 A-1

Fucks1 NeuAc~. 2-* 3Gal ~ 1--4G IcN A c B 1-* 2Manc~ 1~,~ 6 ; M a n f~1~4G I c N A c B 1 - * 4 G I c N A c N e u A c c~2-* 3Ga1131 -* 4G I c N A c ~ 1-* 2Ma n ot I j

A-2

Fucc~l Gal I~1 ~ 4 G I c N A c (31-, 2Manc~ 1hr. 1 -). 4 G I c N A c

6 1 -'~"4 G ICN A c

N eu Acc~ 2-+ 3Gal B 1-* 4GIcN A c !31-* 2Man c~173Man'~b

A-3 N e u A c ~ 2-* 3Gall3 I ~ 4 G I c N Ac!31~ 2Manc~ 1,~6 3Man L~1 -*4G I c N A c 131-* 4G I c N A c N euAc::~ 2-* 3Gal [31 -*4G I c N A c 31 -* 2Man c~1, J

A-4 Gal g 1-*4GIcN A c !31-÷2Mano~ 1,h~6 ~Man !31 -* 4G IcN AC [31 -* 4G ICN A c NeuAcc~ 2~ 3Gal [31 -* 4GIcN A c I31 -* 2Man L~1j ~

A-5 Man~l

_ 'h~n Ma ~ NeuAcc~ 2-* 3Cai [31-. 4GIcN A c [31-* 2Man~ 1 f ~

{31 -*qG I c N A c 131-*4G I c N A c

Fig. 4. Structures of the asparagine-linked sugar chains of hCG purified from the urine of pregnant women,

ed as incomplete biosynthetic products of N-1, but are the final products at particular asparagine loci of hCG. The mechanism of the site-specific formation of different oligosaccharides cannot be explained by our current knowledge of sugar chain biosynthesis. However, it is estimated that a control mechanism that involves steric effects from the surrounding polypeptide moiety plays a role in the maturation of the asparagine-linked sugar chains of this hormone.

Comparative study of the asparagine-linked sugar chains of hCG purified from the urine o] patients with trophoblastic diseases High levels of hCG are detected not only in the urine of pregnant women, but also in the urine of patients with trophoblastic diseases, such as hydatidiform mole and choriocarcinoma.

Because the formation of the 4 asparagine-linked sugar chains of hCG molecule is highly controlled, it was of interest to see if the sugar chains of hCG from choriocarcinoma patients showed any structural change. Study of an hCG sample purified from the urine of a choriocarcinoma patient gave several exciting results [37]. The sugar chains were totally free of sialic acid residues. Analysis of the oligosaccharide fraction by a Bio-Gel P-4 column gave a fractionation pattern quite different (Fig. 7D) from those of urinary hCGs obtained from normal pregnant women. Structural study of the oligosaccharides in each peak revealed that the 8 oligosaccharides shown in Fig. 8 were included in the choriocarcinoma hCG. Further studies of the sugar chains of hCGs isolated from the urine of 3 other choriocarcinoma patients and 3 patients with hydatidiforrn mole revealed many interesting aspects of the

A l t e r e d g l y c o s y l a t i o n in t u m o r s N:

Gall,1 ~4GIcNAc;~ 1,~6 Gall~ 1,f3 GalNAc

G a l ~ l ~ 3GalNAc GalNAc AI:

f G a l ~1 * 4 G I c N A c ~1 ,~ N euAc r~2-* 3.~ 6 Ga I N Ac L Gal ~o1f 3

A2:

NeuAccz2~6 Gal~31,f3 GalNAC NeuAc.~2*3GalB1 * 3GalNAc NeuAc= ~2 ~*6GalNAc

A3:

NeuAcc~2*3Gal f~1.4GIcNAc~ 1~6 N euAcc~ 2÷ 3Gal B 1f 3 G a l N A c

A4:

NeuAcc~2~6 3GalNAc N euAcc~ 2 *3Gal ~3~,"~

Fig. 5. Structures of the mucin-type sugar chains hCG purified from the urine of pregnant women.

1581

.

altered glycosylation of hCG in choriocarcinoma [38]. First of all, the deletion of sialic acid residues in the sugar chains was not commonly detected in the hCGs produced by choriocarcinoma. However, the 8 neutral oligosaccharides in Fig. 8 were detected in all desialylated fractions, although ~he molar ratio of each oligosaccharide varied from one sample to another, In contrast, the 3 hCG samples from hydatidiform mole patients have exactly the same neutral portion of oligosaccharides as normal hCG, and the molar ratio of oligosaccharides E, F and H was always 1:2:1. Therefore, the appearence of the 5 oligosaccharides, A, B, C, D and G could be considered as a specific characteristic of choriocarcinoma hCG. Hydatidiform mole is a benign lesion, althou.gh the rate of incidence of choriocarcinoma in this disease is much higher than in normal pregnancy. Therefore, it was important to determine at what stage of the pathological process leading to choriocarcinoma, the altered glycosylation of hCG starts. Some of the hydatidiform moles show more malignant characteristics, such as invasion into the surrounding tissues and metastasis, and are called invasive moles. It has recently been shown that the 2 hCGs purified from the urine of patients with invasive mole contain oligosaccharides A, B, E, F, G and H in Fig. 8 but not C and D [39]. Therefore, a part

N- !

Fucc~ i +

Gal B 1-~4G I c N A c ~ 1-*2Mana l ~ 6 M a n 131..4GicNAc B

6 1-*4GIcNAc

Ga1131 -~-4GIcN Ac 131-+2Mano~ 1,,~t

N-2 Ga1131 ÷4G I c N A c ~ I ÷ 2Man o~I "~6 I -~4GIcNAc

1 ~4GIcNAc

Gal I~I -*4G I c N A c 131-* 2Man c~I f 3 M a n

N-3 Manc~ 1 '~6. Ga1131-~4G I c N A c ~31÷ 2Man c~1 7 3 M a n B 1*4G I c N A c 131* 4 G I c N A c Fig. 6. Structures of the neutral portions of the asp~ragine-linked sugar chains of normal hCG.

A. Kobata

1582

191817 16 15 14 13 12 11

10

A

B

i

I

state. Absence of oligosaccharides C and D in Fig. 8 indicated that the newly expressed GnT IV in invasive mole transfers N-acetylglucosamine residue to biantennary complex-type sugar chains but not to monoantennary sugar chains. This substrate specificity is the same as that of GnT IV in normal tissues which can produce the triantennary sugar chains, because oligosaccharides C and D are not detected in the glycoproteins produced by normal cells. In choriocarcinoma, the ectopically expressed GnT IV might further be modified to have a wider specificity towards acceptor sugar chains because oligosaccharides C and D are formed.

Clinical application of the altered glycosylation of hCG in trophoblastic diseases

>, =m

g

o

O 10 t~ Iv"

r

400

300

Elution

volume

(ml)

Fig. 7. Fractionation of the desialylated radioactive oligosaccharide fractions by a Bio-Gei P-4 (under 400 mesh) column. Arrows indicate the elution positions of glucose oligomers (numbers indicate the glucose units). A, Oligosaccharide fraction from normal hCG. B, Fraction from c~subunit of normal hCG. 12. Fraction from/3-subunit of normal hCG. D. Fraction from choriocarcinoma hCG.

but not all of the structural abnormalities detccted in neutral portion of the asparagine-linked sugar chains of choriocarcinoma hCG is induced m the sugar chains of invasive mole hCGs. The analytical data described so far indicated that the biochemical events leading to the altered glycosylation of hCG in choriocarcinoma can be considered as follows. Because hCG produced by normal trophoblasts does not contain triantennary sugar chains, GnT IV in Fig. 3 may not be expressed in these cells. Detection of triantennary sugar chains A and B in invasive mole hCG indicated that ectopic expression of GnT IV occurs in this lesion. Therefore, invasive mole should be considered as a precancerous

The results described so far indicated that develpment of any method to discriminate between invasive mole hCG and choriocarcinoma hCG, and normal and mole hCGs is important for the diagnosis of malignant trophoblastic diseases. We have inve~,tigated the behavior of oligosaccharides in an immobilized Datura stramonium agglutinin (DSA) column, and found that complex-type oligosaccharides can be separated into 3 groups: a pass through, a retarded and a bound fraction [40]. Oligosaccharides which have the Gal/31-+4GIcNAc/31--~4(Gal/31-~4GIcNAc/31~ 2)Man group in totally unsubstituted form were recovered in the retarded fraction and those which have either the Gal/31---~4GlcNAc/31--.6(Gal/31--~4GlcNAc/31-~2)Man group or the Gal/31--->4GlcNAc/31---~3Gal/31--~4GlcNAc group in unsubstituted form were recovered in the bound fraction. Since normal and hydatidiform mole hCGs do not contain any interacting oligosaccharides but invasive mole and choriocarcinoma hCGs contain oligosaccharides A and B, which should weakly interact with a DSA-Sepharose column, we expected that these hCGs might be discriminated by such column. Urine from a pregnant woman and from a choriocarcinoma patient were treated with sialidase and passed through a DSA-Sepharose column. When the effluent was monitored for hCG using an ELISA-method, almost all hCG from the pregnant woman passed through the column without interaction. In contrast, almost all hCG from a choriocarcinoma patient bound to the adsorbent. The latter hCG was not eluted with buffer, not even with an N-acetylglucosamine oligomer solution, but eluted with 0.1 N acetic

Altered glycosylation in tumors N-I

A:

B:

!583

Fuc~, 1 4 G a l ? 1 ~-qG I c N A c . - 1 -* 2Man, ~ 6 Ga I ~, 1-* 4 G I c N A c ~:',1,h~q l ' ~ 6 M a n Z, 1-~ 4G ICNAC ? 1-, 4G IcN A c ~ M a n , ~ l .~'~ G a l ~ 1 -* 4GIc N A c £~,1. f z

Gal '~ 1 -*4GICN A c f.~1 ~ 2 M a n , , lx~. Gal ~ 1 -~qG IcN A c ~ 1,~4 b~Man ~ 1 ~4G ICN AC f.~1 -*/.IG I c N A c _Man,-~ 1 ,.I~ G a l ~ 1 ÷4GICN AC ~ 1'p'Z

N-II

Fuc~( 1

C:

D:

E:

Man~l,~_ 6 Ga1131 -+ 4 G I c N A c ~ 1,~,~4 ~bMan ~1 -*4G ICNACi31 -*qG I c N A c ~ M a n ~ 1' f ~ Ga1131 -+q G I c N Acl31 , f z

Manc~l,,~_ Ga1131 -+4G IcN A c 131.~q - - ~L$ M a n ~ 1 ~ qG Ic N AC 131-*4G IcN A c ~Manctl,, 't Ga1131 -*4G IcN AC ~ 1f z Fucctl Gall31 ~ 4 G I c N A c I ~ 1 -* 2Manct 1N~6 6 3 M a n 131-*qG IcN A c B 1 -*@G IcN A c G a l / , 1--*4GICNACLR 1 2Manet1, f

Ga1131 -* 4G IcN Acl31 -* 2Manet l'N~_ F:

1 -*4G IcN A c

1- * 4 G I c N A c

Ga1131 ÷4GICN A c 131 -* 2Manc( 1, f 3bM a n

N-Ill

G: GalI31-+qGIcNAc21

H:

F u c ,~t1 ¢ Manc~l,~_ 6 M a n ~ 1 - * q G I c N A c , ~ 1 -~4G IcN A c ; 2Man,-,.l - J

Man,:t 1,~6Ma n 131~ 4GICNAC i~ 1 ~4G I c N A c G a l ~ I ~ 4 G I C N A c ~ 1 -* 2Man~, 1j a

Fig. 8. Structures of the asparagine-linked sugar chains of a choriocarcinoma hCG sample.

acid. This unexpectedly strong binding of a choriocarcinoma hCG might be explained by the fact that the tumor glycohormone contains more than one sugar chain interacting with DSA. Then, we analyzed urinary hCGs from several of each trophoblastic disease and obtained the results summarized in Fig. g [41]. Even after sialidase digestion, all urinary hCGs from pregnant

women and from hydatidiform mole patients did not bind to a DSA-Sepharose column (Fig. 9A and B). In contrast, a large proportion of urinary hCGs from invasive mole and choriocarcinoma patients bound to the column (Fig. 9C and D). Although the analysis of many urine samples is necessary to determine the practical cut off value of diagnosis, these preliminary data indicate that

A. Kobata

1584

DSA column chromatography will become a very useful method for the diagnosis of trophoblastic diseases. Concluding

remarks

Comparative study of the plasma membrane glycoproteins of fibroblasts and their malignant transformants revealed the increased expression of GnT V in malignant cells. Study of hCG revealed that ectopic'expression of GnT IV is the key to induce the alteration of its sugar chains in choriocarcinoma. Our previous studies of 7-glutamyltranspeptidase revealed that ectopic expression of GnT III is responsible for the alteration of its sugar chains in rat hepatoma [28]. These results indicated that the changes in the sugar chains of glycoproteins during oncogenesis are quite diverse. Disorder in the controlled expression of genes in tumor cells may result in a variety of changes in the sugar chain structures. However, the results discussed in this review also indicate that the alteration of sugar chains of a particular g~ycoprotein produced by a tumor cell is rather constant. Therefore, we can expect

many clinical applications of the altered sugar chains of glycoproteins in tumors. As is well documented in the studies of choriocarcinoma hCG, abnormal sugar chains produced by cancer cells can include those which have never been detected in the glycoproteins produced by normal cells. As discussed already, these sugar chains might be produced because tumor glycosyltransferases, acquired wider substrate specificity towards acceptor sugar chains than the enzymes of normal cells. That tumor glycosyltransferases might have wider or looser substrate specificities has also been suggested by comparative study of glycolipids in normal and tumor cells. Therefore, elucidation of the biochemical basis of such modification of glycosyltransferases in tumor cells is one of the important problems to be solved in the future. As is exemplified by the study of plasma membrane glycoproteins of fibroblasts and their malignant transformants, increase of GnT V activity might be correlated with the metastatic behavior of the malignant cells. Since many cell lines with different metastatic potentials have already been established, study of this area will develop quickly. Acknowledgment

1°°I 80

B

C

I

DO

I

=

E 0

o

~E

'60

Q

-o 40 co

m

20

!

Fig. 9. Percent molar ratio of urinary hCGs bound to a D S A - Sepharose column before ( o ) and after ( • ) sialidase treatment [41]. A. Urine samples from normal pregnant women. B. Specimens from hydatidiform mole patients. C. Specimens from invasive mole patients. D. Specimens from choriocarcinoma patients.

The author would like to express his sincere gratitude to Miss Yumiko Kimizuka for her skilled secretarial assistance. Regerences

1 Florman H. M. & Wasserman P. M. (1985) Cell 41,313-329 2 Tamura G. (1982) Tunicamycin. Jpn. Sci. Soc. Press, Tokyo, pp. 220 3 Kobata A. (1984) in: Biology of Carbohydrates (Ginsburg V. & Robbins P.W., eds.), John Wiley & Sons, New York, vol. 2, pp. 87-161 4 Meezan E., Wu H. C., Black P. H. & Robbins P. W. (1969) Biochemistry 8, 2518-2524 5 Buck C. A., Glick M. C. & Warren L. (1970) Biochemistry 9, 4567-4576 6 Buck C. A., Glick M. C. & Warren L. (1971) Science 172, 169-171 7 Smets L. A., Van Beck W. P. & Van Nie R. (1977) Cancer Len. 3, 133-138 8 Von Nest G. & Grimes W. J. (1977) Biochemistry 16, 2902-2908 9 Warren L., Critchley D. & Macpherson I. (1972) Nature 235,275- 278 10 Van Beek W. P., Smets L. A. & Emmelot P. (1973) Cancer Res. 33, 2913-2922 11 Van Beck W. P., Smets L. A. & Emmelot P. (1975) Nature 253,457-460

1585

Altered glycosylation in tumors

12 Glick M. C., Schlesinger H. & Hummeler K. (1976) Cancer Res. 36, 4520-4524 13 Warren L., Fuhrer J. P. & Buck C. A. (1972) Proc. Natl. Acad. Sci. USA 69, 1838-1842 14 Grimes W. J. (1973) Biochemistry 12,990-996 15 Ogata S., Muramatsu T. & Kobata A. (1975) J. Biochem. Tokyo 78, 687-696 16 Ogata S., Muramatsu T. & Kobata A. (1976) Nature 259, 580-582 17 Yamashita K., Ohkura T., Tachibana Y., Takasaki S. & Kobata A. (1984) J. BioL Chem. 259, 10834-10840 18 Schachter H., Narasimhan S., Gleeson P. & Vella G. (1983) in: Gann Monograph on Cancer Research (Makita A., Tsuiki S., Fujii S. & Warren L., eds.), No. 29, Plenum Press, New York, pp. 177-195 19 Yamashita K., Tachibana Y., Ohkura T. & Kobata A. (1985)J. Biol. Chem. 260, 3963-3969 20 Van den Eijnden D. H. & Schiphorst W. E. C. M. (1983) in: Glycoconjugate (Chester M. A., Heinegard D., Lundblad A. & Svensson S., eds.), Rahms i Lund, Lund, pp. 760-761 21 Pierce M. & Arango J. (1986) J. Biol. Chem. 261, 10772-10777 22 Arango J., Shoreibah M. & Pierce M. (1987) Proc. IXth Int. Syrup. Glycoconjugates (Montreuil J., Verbert A., Spik G. & Fournet B., eds.), abstr. E 114 23 Glick M. C., Rabinowitz Z. & Sachs L. (1973) Biochemistry 12, 4864-4869 24 Glick M. C., Rabinowitz Z. & Sachs L. (1974) Virology 13,967-974 25 Smets L. A., Van Beek W. P. & Van Rooij H. (1976) Int. J. Cancer 18, 462-468 9~ /",~et, a r l n l C

[107~

72, 2687-2690

IOr,a r

Marl

Arad

.grl II.gA

27 Dennis J. W., Lafert6 S., Waghorne C., Breitman M.L. & Kerbel R. S. (1987) Science 236, 582-586 28 Kobata A. & Yamashita K. (1984) Pure Applied Chem. 56, 821-832 29 Bellisario R., Carlsen P. B. & Bahl O. P. (1973) J. Biol. Chem. 248, 6796-6809 30 Carlsen R. B., Bahl O. P. & Swaminathan N. (1973) J. Biol. Chem. 248, 6810-6825 31 Endo Y., Yamashita K., Tachibana Y., Tojo S. & Kobata A. (1979) J. Biochem. Tokyo 85, 669-679 32 Kessler M. J., Reddy M. S., Shah R. H. & Bahl O. P. (1979) J. Biol. Chem. 254, 7901-7908 33 Kessler M. J., Mise T., Ghai R. D. & Bahl O. P. (1979) J. Biol. Chem. 254, 7909-7914 34 Cole L. A., Birken S. & Perini F. (1985) Biochem. Biophys. Res. Commun. 126, 333-339 35 Ginsburg V. & Kobata A. (1971) in: Structure and Function of Biological Membranes (Rothfield L., ed.), Academic Press, New York, pp. 439-459 36 Mizuochi T. & Kobata A. (1980) Biochem. Biophys. Res. Commun. 97,772-778 37 I'Aizuochi T., Nishimura R., Derappe C., Taniguchi T., Hamamoto T., Mochizuki M. & Kobata A. (1983)J. Biol. Chem. 258, 14126-14129 38 Mizuochi T., Nishimura R., Taniguchi T., Utsunomiya T., Mochizuki M., Derappe C. & Kobata A. (1985) Jpn. J. Cancer Res. 76, 752-759 39 Endo T., Nishimura R., Kawano T., Mocl~'.'.'.'.'.'.'.'.~M. & Kobata A. (1987) Cancer Res. 47, 5242:5245 40 Yamashita K., Totani K., Ohkura T., Goldstein I. J. & Kobata A. (1987) J. Biol. Chem. 262, 1602-1607 A1

l~nrlaT

,Tin,~g~

~,~7~w~

|i7nl~aR

~Knhata

A. (1988) Jpn. J. Cancer Res., 79, 160-164