Novel Fucolipids of Human Adenocarcinoma: Monoclonal Antibody

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1986 hy The American Society of Biological Chemists, Inc.

Vol. 261, No. 24, Issue of August 25, pp. 11254-11258 1986 Printed in i l ~ . ~ .

Novel Fucolipids of Human Adenocarcinoma: Monoclonal Antibody Specific for Trifucosyl LeY (II13FucV3FucV12FucnLc6) and a Possible Three-dimensional Epitope Structure*

(Received for publication, February 3, 1986)

Tokio KaizuSg, Steven B. LeveryS, Edward NudelmanS, Ronald E. Stenkampll, and Sen-itiroh HakomoriS From the $Program of Biochemical Oncology/Membrane Research, Fred Hutchinson Cancer Research Center, and the Departments of Pathobiology, Microbiology, and Immunology, University of Washington, Seattle, Washington 98104 and the TDepartment of Biological Structure, University of Washington, Seattle, Washington 98195

Immunization of mice with Ley-active trifucosylnonaosylceramide (II13FucV3FucV12FucnLc6) isolated from human colonic adenocarcinoma (Nudelman, E., Levery, S . B., Kaizu, T., and Hakomori, S . (1986) J. Biol. Chem. 261, 11247-11253) followed by selection of hybridoma by positive reaction with this antigen and negative reaction with two other LeY antigens (II13FucIV2FucnLc, and V3FucV12FucnLc6) resulted in successful isolation of the hybridoma producing IgM antibody, termed KH1, specific to Ley-active trifuco- syinonaosylceramide, which does not cross-react with Ley-active hexaosyl- or octaosylceramides (II13Fuc- IV2FucnLc4 and V3FucV12FucnLc6) without internal fucosyl substitution. The three-dimensional structure of the trifucosylnonaosylceramide was simulated based on previously published glycosidic torsion angles for fucosyl type 2 chain (Le’ and LeY) and for GlcNAc@l+ 3Gal@ as predicted by hard sphere exo-anomeric calculations (Th~gersen, H., Lemieux, R. U., Bock, K., and Meyer, B. (1982) Can. J. Chem. 60, 44-57). The picture thus constructed showed a broad nonpolar area consisting of the hydrophobic surface of the pyranosyl ring and acetamido group of N-acetylglucosamine and three CH3 groups of L-fucose; this hydrophobic area is adjacent to a hydrophilic area. In analogy to the de- tailed structure of Leb or LeY involved in their interactions with antibodies and lectins (Spohr, U., Hinds- gaul, O., and Lemieux, R. U. (1985) Can. J. Chem. 63, 2644-2652), such a wide hydrophobic area adjacent to a hydrophilic region could be recognized by the antibody KH1, as shown in the model illustrated in the text. Since the axis of ceramide, which is inserted in the lipid bilayer, is perpendicular to the plane of type 2 chain, the epitope recognized by the antibody KHl is located at the external nonpolar surface of the carbohydrate chain that is overlaid on the lipid bilayer.

* This investigation was supported by a research grant from the Otsuka Research Foundation. This is Paper VI in the series “Novel Fucolipids of Human Adenocarcinoma.” Ref. 1 is Part V of this series. 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.

This paper is dedicated to Professor Luis L. Leloir on the occasion of his 80th birthday.

fj Supported by and on leave from the Japan Immunoresearch Laboratory, 17-5 Sakae, Takasaki, Gumma 370 Japan (present ad- dress).

A novel glycolipid called LeY trifucosylnonaosylceramide (II13FucV3FucV12FucnLc6)1 was isolated and characterized from human colonic adenocarcinoma (1). This glycolipid is present consistently in all colonic cancers tested; LeY hexaosylceramide (II13FucIV2FucnLc4) is present in trace quantities in some tumor tissues, whereas LeY octaosylceramide (V3FucV12FucnLc6) and other extended LeY antigens are also present in normal cells and tissues. The possibility of estab- lishing a monoclonal antibody that would react specifically with Ley trifucosylnonaosylceramide, but not with LeY hexaosylceramide (II13FucIV2FucnLc,) or octaosylceramide (V3- FucV12FucnLc6), has been explored since such an antibody could have higher specificity for human adenocarcinoma than those antibodies reacting with LeY determinant irrespective of its carrier structure. An IgM antibody, termed KH1, has been established after immunization of mice with LeY trifucosylnonaosylceramide antigen and selection of hybridoma by positive reactivity with the immunogen and negative reactivity with LeY hexaosyl- and octaosylceramide. An approximate three-dimensional picture of II13FucV3FucV12FucnLc6 has been constructed, and a possible epitope for KH1 antibody is discussed.

MATERIALS AND METHODS

Ley trifucosylnonaosylceramide (II13FucV3FucV12FucnLc6) and Ley octaosylceramide (VSFucVIZFucnLc6) were isolated from human adenocarcinoma as described previously (1). LeY hexaosylceramide (II13FucIV2FucnLc4) and Leb hexaosylceramide (II14FucIV2FucnLc4) were prepared originally by Dr. John McKibbin (Department of Biochemistry, University of Alabama, Birmingham, AL) from dog and human intestine, respectively (2,3), and were further purified by preparative high performance thin-layer chromatography (HPTLC’). Dimeric Le’ (difucosyl yz, II13V3FuCznLC6) was prepared from human colonic adenocarcinoma as previously described (1, 4). Le’ heptaosylceramide (y2, V3FUCnLC6) was prepared from colonic adenocarcinoma (4) and human whole blood cell membranes (5).

The II13FucV3FucV12FucnLc6 antigen was mixed with Salmonella minnesota to immunize BALB/c mice as previously described (6). Approximately 20 pg of glycolipid solution in ethanol (50 pl) was mixed with 800 p1 of phosphate-buffered saline, pH 7.4, and the solution was further mixed with 250 p1 of a suspension containing 250 pg of S. minnesota. The whole mixture was incubated at 50 “C. A suspension containing 10 pg of glycolipid was intravenously injected in Day 1. Subsequently, a suspension containing 4 pg of glycolipid was injected every 4 days for a total of four times. Host spleen cells were harvested 3 days after the last immunization. The fused cells were placed in 96-well plates (Dynatech Immunolon,

Glycolipid designations and definition of Le” and LeY are the same

The abbreviations used are: HPTLC, high performance thin-layer as in the accompanying paper (1).

chromatography; HSEA, hard sphere exo-anomeric.

11254

Monoclonal Antibody Specific for Trifucosyl LeY 11255

TABLE I Literature values ($SqH) (15, 16) for preferred glycosidic torsion angles of sugars comprising glycolipids including

di- ann' trifucosylnor~xaosylceramides reacting with monoclonal antibody KHI, estimated by HSEA (1 7, 18) calculations on dimeric, trimeric, ann' tetrameric structures

Glycolipid Fucal + 2Gal131 + 4(Fucal "t 3)GlcNAcPl + 3Galpl -+ 4(Fucal + 3)GkNAcpl + 3Galpl -P 4Glc

nLc6 55", 5" a 60", -10' 55", 5" a 60", -10" 55", 0" a V12FucnLc6(H2) 50", 15" 55", 0" 60". -10" 55', 5' a 60", -10" 55", 0" a

II13V3Fuc2nLc6 55", 10" 55", 25" 60", -10" 55", 10" 55", 25" 60", -10' 55", 0"" 1II3V3VI2Fuc3nLcfi 50", 10' 55", 10' 50", 25" 60", -10"" 55", 10" 55", 25" 60", -10' 55", 0"

Ref. 16. Ref. 15.

Dynatech Laboratories, Inc., Alexandria, VA). The antibody-producing hybridomas were selected and cloned on a few 96-well Plates by limited dilution. The cells producing antibody with positive reactivity with II13FucV3FucV12FucnLcfi and negative reactivity with V3FucV12FucnLcfi were isolated and further propagated and recloned by limited dilution method. The reactivity of antibody with glycolipids was determined by the method described previously by Kannagi et al. (7). The specificity of the KH1 antibody was further confirmed by inhibition of antibody binding to the solid-phase antigen by various glycolipid antigens (see Fig. 3 legend). The class and subclass of the antibody were determined by class- and subclass-specific antibodies (Cappel Laboratories, Cochranville, PA). The IgM antibody was purified after precipitation with 2% boric acid followed by gel filtra- tion on Sephacryl 200 (8). Identification of the antibody was further confirmed by sodium dodecyl sulfate-polyacrylamide gel electropho- resis with Western blot technique (9). Immunostaining of glycolipids separated on HPTLC plates (J. T. Baker Chemical Co., Phillipsburg, NJ) was performed according to a modification (5) of a method originally described by Magnani et al. (10).

A three-dimensional structural depiction of Ley trifucosylnonaosylceramide was assembled from a computer library of monosaccharide crystal coordinates (11-14) using the previously published glycosidic torsion angles (&:HI - C1 - 0, - Cx and J/H:C1 - O1 - CX - Hx) for Le' trisaccharide (15), Le-" tetrasaccharide (15), and the GlcNAcp1-3Galfl disaccharide (16), as listed in Table I, which are predicted by hard sphere exo-anomeric (HSEA) calculations (17, 18).

RESULTS

Establishment of Hybridoma Secreting Antibody KHl -At the initial stage of cloning, hybridomas in eight wells showed positive reactivity with LeY trifucosylnonaosylceramide. Of these hybridomas, those secreting antibodies showing reactivity with L@ hexaosylceramide (II13FucIV2FucnLc4) and LeY octaosylceramide (V3FucV12FucnLc6) were excluded. The re- mainder were recloned, and finally a single hybridoma showing an exclusive reactivity with L@ trifucosylnonaosylceramide was isolated and termed KH1. The antibody was identified as IgM.

Specificity of Antibody KHI-The KH1 antibody reacted specifically with LeY trifucosylnonaosylceramide, but did not react with LeY octaosylceramide (V3FucV12FucnLc6), LeY hexaosylceramide (II13FucIV2FucnLc4), and Leb hexaosylceramide (II14FucIV2FucLc4) tested on solid-phase radio- immunoassay with antigen dilution (Fig. L 4 ) as well as antibody dilution (Fig. 2). The KH1 antibody showed, however, a weak but definitive cross-reactivity with dimeric Le" (II13FucV3FucnLc6) when the antibody was tested on an antigen dilution basis (Fig. L4). Nevertheless, the antibody FH4, which recognizes dimeric Lex, does not react with the KH1 antigen (Ley trifucosylnonaosylceramide) on solid-phase binding assay (Fig. 1B) and on HPTLC immunostaining (data not shown). Only LeY trifucosylnonaosylceramide, but not other glycolipids, was capable of inhibiting the binding of KH1 antibody on solid-phase Ley trifucosylnonaosylceramide (Fig. 3).

The reactivity of KH1 antibody on HPTLC immunostaining with various glycolipids having structures analogous to LeY trifucosylnonaosylceramide and with the neutral glyco-

Glycolipid (ng/well) Glycolipid (ng/well)

FIG. 1. Reactivities of KH1 and FH4 antibodies with various concentrations of glycolipids on solid-phase radioimmu- noassay. Various concentrations of glycolipids (indicated on the abscissa) mixed with five times the weight of lecithin and three times the weight of cholesterol were coated on vinyl strip wells (Costar, Cambridge, MA). The antibody binding assay was performed with 1:500 diluted KH1 ascites and undiluted FH4 culture supernatant according to the method as previously described (7). A shows the reactivity of KH1 ascites. Similar results were obtained with undiluted KH1 supernatant. M, LeY trifucosylnonaosylceramide; A-A, dimeric Le' (difucosyl y2); c."., LeY hexaosylceramide (II13FucIV2FucnLc4); M, LeY octaosylceramide (V3FucVIZFucnLc6); a " 0 , Le' heptaosylceramide (yz glycolipid, V3FucnLc6). B shows the reactivity of FH4 antibody. U, LeY trifucosylnonaosylceramide; A-A, dimeric Le'.

2

1 4 16 64 256 1024 4096 Reciprocal of Antibody Dilution

2

1 4 16 64 256 1024 4096 Reciprocal of Antibody Dilution

FIG. 2. Reactivity of various glycolipids with different concentrations of KH1 antibody on solid-phase radioimmunoas- say. Various glycolipids with five times the weight of lecithin and three times the weight of cholesterol were coated on vinyl strip wells as for Fig. 1. The quantity of glycolipids added was 10 ng/well. Each well was reacted with a different concentration of antibody as indicated in the abscissa. Undiluted antibody is the undiluted culture medium of KH1 hybridoma. The analysis was made as described previously (7). M, LeY trifucosylnonaosylceramide; M, LeY octaosylceramide; a " 1 7 , Ley hexaosylceramide; U, Leb hexaosylceramide (II14FucIVZFucLc4).

lipid fraction prepared from various normal and tumor tissues is shown in Fig. 4. As shown in Fig. 4C, a major band corresponding to Ley trifucosylnonaosylceramide (band e in A and B ) , a weak slow migrating band, and a band at the origin were stained by KH1 antibody. As clearly indicated in this

11256 Monoclonal Antibody Specific for Trifucosyl Ley

100 -

00 - 0 c .- .- .- 2 60- .c c H Y- 40- 0

8 20 -

0 4000 1000 100 10 1

Glycolipid ( nq/well) FIG. 3. Inhibition of KH1 antibody binding to solid-phase

Ley trifucosylnonaosylceramide by various glycolipid antigens. 10 ng of LeY trifucosylnonaosylceramide antigen with 50 ng of lecithin and 30 ng of cholesterol was placed in each vinyl strip well. The antigen-coated well was treated with 5% bovine serum albumin for blocking, and each well was incubated with 25 pl of undiluted culture supernatant of KH1 hybridoma and 25 pl of glycolipid solution containing different amounts of glycolipids as indicated on the abscissa. The highest concentration of glycolipid added in a well was 4000 ng. The inhibition of binding activity is expressed as the per- centage of antibody binding activity to a solid-phase glycolipid on each well. Inhibitor glycolipids are identified as follows: U, LeY trifucosylnonaosylceramide; W, L e y octaosylceramide; Cl-AJ, LeY hexaosylceramide; U, L e b hexaosylceramide (II14FucIV2- FucLc~).

reference run (lane 8 of A-C), the bands corresponding to LeY hexaosylceramide and LeY octaosylceramide were not stained by KHl, in contrast to the strong staining of these bands by AH6 (lane 8 of B). The AH6 antibody reacts equally well with all glycolipids bearing LeY determinant, and AH6-reactive glycolipids appear to be present in greater quantity in all eight cases of colinic adenocarcinoma (lanes 9-1 7 of B) than in normal blood cells and normal tissues (lanes 1-6 of B) , except for pancreas, which has a larger quantity of Ley-active glycolipids (lane 7 of B). In striking contrast, only a band corresponding to II13FucV3FucV12FucnLc6 was stained by antibody KH1, and this reactive band was completely absent in glycolipids from normal blood cells, normal liver, normal kidney, and normal colon and intestine (lanes 1-6 of C), except for a weak band that was observed in glycolipids extracted from pancreas (lune 7 of C). All glycolipid samples prepared from colonic adenocarcinoma tissues contained this KH1-defined band (lanes 9-17 of C ) .

Conformational Structure of LeY-actiue Trifucosylnonaosyl- ceramide-Based on the literature values for the preferred glycosidic torsion angles, shown in Table I, the conformational structure of Ley-active trifucosylnonaosylceramide has been assembled by computer, as shown in Fig. 5. A space- filling model of the KH1-binding region was produced using standard van der Waals radii for the atoms involved (Fig. 6).

DISCUSSION

Specificity of KHl Antibody-The antibody KH1 showed a novel specificity directed to LeY trifucosylnonaosylceramide and did not cross-react with LeY hexaosylceramide or LeY octaosylceramide on solid-phase antibody binding and on HPTLC immunostaining. A weak cross-reactivity with dimeric Le" (II13V3Fuc2nLc6) was observed only on antibody binding on glycolipid with lecithin and cholesterol solid phase, but not on HPTLC immunostaining. Often a preparation of dimeric Le" is contaminated with LeY trifucosylnonaosylceramide; hence, a false reactivity can be observed. Extensively purified dimeric Le", however, does not give any reactivity on

A - - Hex2Cer.

Gb4, nLc4,

H1. no-

Ab- H2-

H3-

B

1 2 3 4 5 6 7 8 9 1011 121314I51617 - ~~

"

. .

" "

I 2 3 4 5 6 7 8 9 IO I I 12 13 14 15 16 17

FIG. 4. Immunostaining of polar neutral glycolipids from normal blood cells, normal tissues, and various colonic adenocarcinoma by KH1 antibody as compared with that by AH6 antibody. The glycolipid samples placed on each lane in A-C are identical and represent the polar neutral fraction obtained from Folch's upper phase prepared as previously described (4). The quantity of glycolipids placed in each well was derived from approximately the same quantity of tissues and cell protein. The yield of total polar glycolipids from colonic adenocarcioma was 300-500 pg/lOO mg of protein, whereas that from normal cell membranes and tissues was 80-250 pg/lOO mg of protein. Each aliquot of glycolipid placed in each HPTLC lane was from approximately 1 mg of tissue protein. A, orcinol-sulfuric acid staining; B, AH6 staining; C, KH1 staining. Lanes 8 and 14, a mixture of reference glycolipids, II13FucIV2FucnLc4 (upper band), V3FucV12FucnLc6 (middle band), II13FucV3FucnLc6 (lower band), and II13FucV3FucV12FucnLcs (lower band e); lane 1, A blood cell membranes; lane 2,O blood cell membranes; lane 3, B blood cell membranes; lane 4, normal liver; lane 5, normal colon; lane 6, normal kidney; lane 7, normal pancreas; lanes 9-13 and 15-1 7, colonic adenocarcinomas TG118 ( l a n e 9), TG105 ( l a n e lo), TG038 ( l a n e 111,

( l a n e 15), and TG115 ( l a n e 17). The faster mobility of the KH1- stained band in lane 17 may be due to differences in fatty acid moiety. The positions of bands are identified in the margin. Chromatograms were run on J. T. Baker HPTLC plates developed in chloroform/ methanol/water (55405, v/v/v).

FT75-620 ( l a n e 12), TG067 ( l a n e 13), FT75-845 ( l a n e 15), TG126

TLC immunostaining. The antibody FH4, which recognizes difucosyl residues linked at 1113-GlcNAc and V3GlcNAc of type 2 chain (19), does not react with Ley trifucosylnonaosylceramide. This reactivity profile indicates that the antibody KH1 may not only be directed to the terminal LeY structure (Fuccwl~2Gal~l-*4(Fuca1+3)GlcNAc), but may also be directed to the internal difucosyl structure.

As expected, KH1 antibody shows a higher specificity for human colonic adenocarcinoma than do other anti-LeY anti-

Monoclonal Antibody Specific for Trifucosyl LE? 11257

GlcNAcVVAc lU3,FucCH3 h

YI'Fuc

FIG. 5. Computer simulation (ball and stick) of the three- dimensional structure of II13FucV3FucV12FucnLc6. The oligosaccharide structure was assembled using a library of monosaccharide crystal coordinates (11-14) and the published (15, 16) preferred glycosidic torsion angles, listed in Table I, derived from HSEA calculations (17, 18) on di-, tri-, and tetrassaccharides. The axis of ceramide is oriented upwards and behind the plane of the carbohydrate chain (indicated by the arrow).

V2 Fuc

FIG. 6. Space-filling model of oligosaccharide comprising the proposed binding site of KH1 antibody. The orientation is identical to that described in the legend to Fig. 5. The large area enclodes by dashed lines indicates a hydrophobic surface, part or all of which might contribute to binding to a complementary surface on the antibody. The shorter dashes delineate the region (including parts of Gal VI, GlcNAc V, V3Fuc, and II13Fuc) still providing some reactivity in the absence of VI'Fuc (see "Results"). The cluster of hydroxy groups, which are marked A (V3Fuc OH-4), B (Gal VI OH- 4), C (Gal VI OH-3), and D (VI'Fuc OH-2), could provide a key polar interaction to initiate binding in accordance with the mechanism proposed by Lemieux and co-workers (15, 29-33) for carbohydrate- protein interactions.

bodies that react promiscuously with all carbohydrate chains containing LeY. Interestingly, the KH1 antibody defines a single glycolipid band, present in all eight colonic cancer tissue extracts examined, in contrast to AH6, which detects multiple bands. The KH1-defined band was absent in the glycolipid fraction extracted from various normal cells and tissues, except the pancreas, which showed a band with slightly faster mobility than that in tumor tissues. The KH1-defined band should have the common determinant Fucal+2Gal/31+ 4(Fucal~3)GlcNAc(31-3Gal~1-4(Fuccul+3)GlcNAc~1+ R, although the TLC mobility of the KH1-reactive band from tumor tissues showed a great deal of variation. Such variation

should depend on ceramide composition. On immunohistolog- ical examination, AH6 antibody stained the proximal region of normal colon and crypt areas, whereas KH1 antibody did not stain normal colonic epithelia, either in the proximal or distal region, even in the crypt areas. In contrast, KH1 stained colonic adenocarcinomas with 80-85% positive incidence irrespective of the location and stage of the cancer?

Conformational Analysis of KHl Antigen-In order to vis- ualize the structural features of the KH1 antigen, the approximate model shown in Fig. 5 was constructed using previously published glycosidic torsion angles predicted by HSEA calculations (17, 18). The validity of such calculations, within reasonable limitations, has been repeatedly demonstrated (15, 17, 18, 20-22). Although a global energy minimization of the structure depicted may be desirable, it is expected in the present case that the result might not deviate significantly from this first approximation. The helical nature of the poly- N-acetyllactosamine core has been shown several times by x- ray diffraction studies on polysaccharides configured in a similar alternating (B1-3 eq p1-A eq), arrangement (23,24). In the present case, this allows Fucal+3 units linked to alternating @-GlcNAc residues to be close enough to be part of the same epitope (FH4, for example (19)) without severe steric interaction between the Fucal-3 and N-acetyl sub- stituents. The addition of Fucal-3 units to the po1y-N- acetyllactosamine core may have a modifying effect on the helix shape since the calculated preferred torsion angles for the Galpl-AGlcNAc linkage change slightly from 4, J / 55", 5" to 55", 10" (15, 16). The addition of a terminal Fucal+2 unit affects the calculated orientation of the nearest Fucal-, 3 by a 5" increment (see Table I), but has no further effect on the Galpl-AGlcNAc linkage (15).

The model thus assembled is consistent with immunological and 'H NMR data obtained for this and similar structures. For all structures examined by 'H NMR containing the Le" trisaccharide, the resonance for Fuccul+3 H-1 is found at the same frequency at a given temperature regardless of the position on the core, number, or terminal substituent (4, 25, 26). In addition, the resonance for CH3 of Fucal+3 units is found (in the absence of terminal substitution) at only two positions separated by a small but measurable increment (A6 = 0.005 ppm); for units attached to any GlcNAc except the penultimate one, the Fuc CH, is shifted upfield by this value. This can be rationalized as an effect of the anisotropic GlcNAc carbonyl, tending to suppc$ the model, which shows the Fuc CH, protons within 3-4 A, and positioned in the shielding zone, of that group. Addition of the Fucal-2 unit causes a downfield shift of CH, on the nearest Fucal-3 (A6 = 0.051 f 0.002 ppm) (25); inspection of the model shows that this could be a deshielding effect of 0-5 of the Fucal-2 unit as noted previously for Leb tetrasaccharide (17).

The resonance position of H-5 of Fucal-3 appears to vary somewhat with the number and position of these units, as well as with terminal substitution (25). Since it is believed that the extreme downfield position of these resonances in Le" and LeY structures is due to deshielding by proximity to GlcNAc 0-4 and the vicinally linked Gal 0-5 (27), it might be expected that this proton would be sensitive to small deformations of the Galpl-sIGlcNAc linkage. Finally, it has been noted that the addition of a Fucal+3 unit causes a measurable deshielding (A6 = 0.010-0.018 ppm) of H-1 of the next GlcNAc down the chain (25). This could be due to a small reorientation in the equilibrium position of the NAc

3Y. S. Kim, M. Yuan, S. H. Itzkowitz, Q. Sun, A. Palekar, T. Kaizu, and S. Hakomori, manuscript in preparation.

11258 Monoclonal Antibody Specific for Trifucosyl Ley

group of GlcNAc, which is known to influence the shift of that proton (28).

KHI-binding Site-Having established a three-dimensional picture of the KH1 oligosaccharide antigen, it now becomes possible to discuss a hypothetical binding site on the molecule in light of current theory of protein-carbohydrate interactions (15, 29-33). Fig. 6 illustrates a hydrophobic region involving II13Fuc and V3Fuc, the residues which provide a minimum structure for cross-reactivity with KH1. The additional hydrophobic area provided by the Fucal+2 unit is shown as well. This extended region is shared by the binding site demonstrated for Griffonia simplicifolia lectin IV (29). The increased binding of the KH1 antigen including the Fucal- 2 unit may be attributed to one of two possible effects: (i) the additional Fuccul+2 may cause a reorientation of sugar units in a previously existing epitope to provide a better fit of the hydrophobic topography or (ii) the additional Fuccul+2 may extend the hydrophobic region to provide a larger area complementary to the KH1 antibody. Lemieux’s proposal, that hydrophobic interactions between complementary sites are the requisite for avid protein-carbohydrate binding (15), has recently been supported by considerable experimental evi- dence (29-33). However, it is also apparent from these exper- iments that adjacent polar sites are crucial to the binding mechanism. These are thought to act as a key to initiate binding, but not to affect appreciably the stability of the final protein-carbohydrate complex (29-32). The key polar group- ing that was demonstrated for binding to G. simplicifolia lectin IV is indicated in Fig. 6. An additional hydroxy group of Fucal-+2 (OH-2) (D in Fig. 6) was demonstrated to be im- portant in the binding of a hybridoma antibody to Leb (31), which shares a similar topography in the terminal sugars. Delineating these features is not meant to imply that the precise sites of interaction with KH1 antibody are therefore known, but only that such features are available in this structure in an arrangement that has already been demonstrated (29,31,32) to fit into a convincing scheme for binding. Investigation of the binding of KH1 with an epitope that may cover parts of as many as 6 or 7 saccharide residues might provide further insight into the general applicability of this scheme.

Acknowledgment-We thank Drs. Keith Watenpaugh and Lyle Jensen of the Department of Biological Structure, University of Washington, for their help in construction of molecular models based on computer-calculated bond angles.

REFERENCES

1. Nudelman, E., Levery, S. B., Kaizu, T., and Hakomori, S. (1986) J. Biol. Chem. 261, 11247-11253

2. McKibbin, J . M., Spencer, W. A., Smith, E. L., Mansson, J.-E., Karlsson, K.-A,, Samuelsson, B. E., Li, Y.-T., and Li, S.-C. (1982) J. Biol. Chem. 2 5 7 , 755-760

3. Smith, E. L., McKibbin, J. M., Karlsson, K.-A., Pascher, I., and Samuelsson, B. E, (1975) Biochim. Biophys. Acta 388, 171- 179

4. Hakomori, S., Nudelman, E., Levery, S. B., and Kannagi, R. (1984) J. Biol. Chem. 259,4672-4680

5. Kannagi, R., Nudelman, E., Levery, S. B., and Hakomori, S.

6. Young, W. W., Jr., MacDonald, E. M. S., Nowinski, R. C., and

7. Kannagi, R., Stroup, R., Cochran, N. A., Urdal, D. L., Young, W.

8. Badin, J., and Levesque, H. (1961) Rev. Rhum. Mal. Osteo-artic.

9. Towbin, H., Staehelin, T., and Gordon, J. (1979) Proc. Natl. Acad. Sei. U. S. A. 76,4350-4354

10. Magnani, J. L., Smith, D. F., and Ginsburg, V. (1980) Anal. Biochem. 109,399-402

11. Fries, D. C., Rao, S. T., and Sundaralingam, M. (1971) Acta Crystallogr. Sect. B Struct. Crystallogr. Cryst. Chem. 2 7 , 994- 1005

12. Mo, F. (1979) Acta Chem. Scad. Ser. A. Phys. Znorg. Chem. 33,

13. Longchambon, P. F., Ohannesian, J., Avenel, D., and Neumann, A. (1975) Acta Crystallogr. Sect. B Struct. Crystallogr. Cryst. Chem. 31,2623-2627

14. Sheldrick, B. (1976) Acta Crystallogr. Sect. B Struct. Crystallogr. Cryst. Chem. 32,1016-1020

15. Lemieux, R. U. (1982) in ZUPAC Frontiers in Chemistry (Laidler, K. J., ed) pp. 3-24, Pergamon Press, New York

16. Bock, K., Breimer, M. E., Brignole, A., Hansson, G. C., Karlsson, K.-A., Larson, G., Leffler, H., Samuelsson, B. E., Stromberg, N., Svanborg E d h , C., and Thurin, J. (1985) J. Biol. Chem.

17. Lemieux, R. U., Bock, K., Delbaere, L. T. J., Koto, S., and Rao,

18. Thrdgersen, H., Lemieux, R. U., Bock, K., and Meyer, B. (1982)

19. Fukushi, Y., Hakomori, S., Nudelman, E., and Cochran, N. (1984)

20. Bock, K. (1983) Pure Appl. Chem. 55,605-622 21. Bock, K., and Lemieux, R. U. (1982) Carbohydr. Res. 100 , 63-

22. Sabesan, S., Bock, K., and Lemieux, R. U. (1984) Can. J. Chem.

23. Atkins, E. D. T., Isaac, D. H., Nieduszynski, I. A., Phelps, C. F., and Sheehan, J.-K. (1974) Polymer 15, 263-271

24. Rees, D. A. (1975) in Biochemistry of Carbohydrates: International Review of Science (Whelan, W. J., ed.) Series One, Vol. 5, pp. 1-42, Butterworth & Co., Ltd., London

25. Levery, S. B., Nudelman, E., Andersen, N. H., and Hakomori, S. (1986) Carbohydr. Res., in press

26. Levery, S. B., Kannagi, R., Nudelman, E., and Hakomori, S. (1983) in Proceedings of the VZZth International Symposium on Glycoconjugates (Davidson, E. A., Williams, J. C., and Di- Ferrante, N. M., eds) pp. 258-259, Praeger Scientific, New York

27. Hindsgaul, O., Norberg, T., Le Pendu, J., and Lemieux, R. U. (1982) Carbohydr. Res. 109 , 109-142

28. Dabrowski, J., Hanfland, P., and Egge, H. (1980) Biochemistry

29. Spohr, U., Hindsgaul, O., and Lemieux, R. U. (1985) Can. J. Chem. 63,2644-2652

30. Hindsgaul, O., Khare, D. P., Bach, M., and Lemieux, R. U. (1985) Can. J. Chem. 63,2653-2658

31. Spohr, U., Morishima, N., Hindsgaul, O., and Lemieux, R. U. (1985) Can. J. Chem. 6 3 , 2659-2663

32. Lemieux, R. U., Venot, A. P., Spohr, U., Bird, P., Mandal, G., Morishima, N., and Hindsgaul, 0. (1985) Can. J. Chem. 63,

33. Lemieux, R. U., Wong, T.-C., Liao, J., and Kabat, E. A. (1984)

(1982) J. Biol. Chem. 2 5 7 , 14865-14874

Hakomori, S. (1979) J. Exp. Med. 150 , 1008-1019

W., Jr., and Hakomori, S. (1983) Cancer Res. 43 , 4997-5005

28 , 101-107

207-218

260,8545-8551

V. S. (1980) Can. J. Chem. 58,631-653

Can. J. Chem. 60,44-57

J. Biol. Chem. 259,4681-4685

74

62,1034-1045

19,5652-5658

2664-2668

Mol. Zmmunol. 2 1,751-759

Documents

Novel Fucolipids of Human Adenocarcinoma: Monoclonal Antibody