Identification of a Major Carbohydrate Capping Group of the L-selectin Ligand on High Endothelial Venules in Human Lymph Nodes as 6-Sulfo Sialyl Lewis X*

We investigated the molecular species of sulfated sialyl Lewis X determinants, the putative L-selectin ligand, expressed on high endothelial venules (HEV) in human lymph nodes. Comparison of the reactivity pattern of HEV with the reactivity of the pure 6-sulfo, 6′-sulfo, or 6,6′-bissulfo sialyl Lewis X determinant with hitherto known anti-sialyl Lewis X antibodies strongly suggested 6-sulfo sialyl Lewis X to be the best candidate for the major sulfated sialyl Lewis X determinant on HEV, followed by 6,6′-bissulfo sialyl Lewis X, whereas 6′-sulfo sialyl Lewis X was unlikely. We newly generated monoclonal antibodies (mAbs) G152 and G72 directed against 6-sulfo sialyl Lewis X, which intensely labeled HEV in immunohistochemical examination and inhibited binding of recombinant L-selectin-IgG to HEV, suggesting that the determinant serves as a ligand for L-selectin. To test the concomitant expression of 6,6′-bissulfo sialyl Lewis X, specific mAbs (G2706, G27011, G27037, and G27039) were generated, but all antibodies failed to react to HEV. Next, we established mAbs (AG97 and AG273) directed against 6-sulfo Lewis X, the asialo form of 6-sulfo sialyl Lewis X. The antibodies were not reactive to untreated HEV, but strongly reacted to sialidase-treated HEV. This indicated the predominance of the sialylated form of 6-sulfo sialyl Lewis X and minimal expression of its asialo form, corroborating that it was synthesized by fucosyltransferase VII, the isoenzyme that preferentially produces the sialylated form of the determinant.

lymph nodes are thought to express a carbohydrate ligand that binds to L-selectin. However, the nature of the L-selectin ligand expressed at the surface of HEV endothelial cells is not yet clear. Some sulfated carbohydrate determinants are known to bind to L-selectin or inhibit L-selectin-mediated cell adhesion. On the other hand, it was suggested that sialyl Lewis X and sialyl Lewis A have a significant capacity to bind L-selectin, despite the lack of a sulfate residue in their carbohydrate structures (4,5), right after they had first been shown to serve as ligands for E-selectin (6 -10). However, most of the known anti-sialyl Lewis X antibodies as well as anti-sialyl Lewis A antibodies did not react efficiently to HEV endothelial cells. In an earlier report, we showed that one of the anti-sialyl Lewis X antibodies (2H5) specifically detected a sialyl Lewis X-like determinant on human HEV, and the antigenic determinant served as a ligand for L-selectin since the antibody inhibited the binding of lymphoid cells to human HEV endothelial cells in Stamper-Woodruff assays (11). We suggested that the antigenic determinant detected by the antibody was closely related, but not exactly identical, to genuine sialyl Lewis X and called it a complex-type sialyl Lewis X determinant, which was defined by the 2H5 antibody at that time (11).
In studying the specificity of the known anti-sialyl Lewis X antibodies, we noticed that they can be classified into two groups according to their cross-reactivity to sialyl Lewis X-related structures. One group consists of antibodies that recognize mainly sialic acid and fucose residues, but do not recognize the type 2 chain core Gal␤134GlcNAc␤, whereas the other group contains antibodies that unerringly recognize sialic acid residues and type 2 chain core Gal␤134GlcNAc␤ structure, but only weakly recognize fucose residues (12). The anti-sialyl Lewis X antibodies that reacted to HEV were all found to belong to the former group, and the antibodies in the latter group lacked reactivity to HEV endothelial cells. These findings led us to assume that the complex sialyl Lewis X on HEV could be a sialyl Lewis X determinant that was somehow modified at the type 2 chain core structure. This modification would not affect the reactivity of the antibodies in the former group, but would seriously affect the reactivity of the antibodies in the latter group. The sulfated sialyl Lewis X determinants, proposed by Hemmerich et al. (13,14) to be expressed on HEV in murine lymph nodes, were the sialyl Lewis X determinants modified by sulfate residues at the type 2 chain core structure. We thought one of these might be the complex sialyl Lewis X determinant that had been detected by our antibody on human HEV endothelial cells.
In a previous study, we investigated the reactivity of a few anti-sialyl Lewis X antibodies against human HEV and compared it with the reactivity pattern of the antibodies against a series of pure sulfated sialyl Lewis X determinants (15). Our unexpected conclusion was that the best candidate for the molecular species carried by HEV is 6-sulfo sialyl Lewis X (15), a conclusion quite different from those of preceding researchers, who suggested the 6Ј-sulfo sialyl Lewis X species to be a major determinant (13,14). In this paper, we expanded our studies and generated a series of new monoclonal antibodies directed against 6-sulfo sialyl Lewis X, 6,6Ј-bissulfo sialyl Lewis X, and 6-sulfo Lewis X. By using these new reagents, we tried to establish 6-sulfo sialyl Lewis X as the major sialyl Lewis X-like carbohydrate determinant carried by HEV endothelial cells, and the ligand carrying this capping structure serves as a ligand for L-selectin in human HEV.

EXPERIMENTAL PROCEDURES
Monoclonal Antibodies Directed against Sialyl Lewis X-related Carbohydrate Determinants-The anti-sialyl Lewis X antibodies 2H5 (established against a mixture of complex sialyl Lewis X-active determinants prepared from human cancer cells) and 2F3 (established against the synthetic sialyl Lewis X glycolipid) were prepared as described previously (11,16). The antibody HECA-452 was a kind gift from Dr. T. K. Kishimoto (Department of Pathology, Stanford University School of Medicine, Stanford, CA) (17). SNH-3 and FH-6 were gifts from Dr. S. Hakomori (Pacific Northwest Research Foundation, Seattle, WA) (7,18,19). CSLEX-1 was obtained from American Type Culture Collection (Bethesda, MD) (20). The anti-Lewis X antibody LeuM1 was purchased from Becton Dickinson Labware (Franklin Lakes, NJ); FH-2 was also a gift from Dr. S. Hakomori; and 73-30 and Tx-17 were established in our laboratory. The antibody SU59 directed against 3Ј-sulfo Lewis X was a gift from Dr. Akihiro Hino (Nisshin Shokuhin Co. Ltd., Otsu, Japan). These antibodies are all murine IgM, except for HECA-452, which is rat IgM.
New hybridoma cells that secrete monoclonal antibodies directed against sulfated sialyl Lewis X and sulfated Lewis X determinants were generated according to the method described by Köhler and Milstein (21) and subsequently adopted for anti-carbohydrate antibodies (22). Briefly, the synthetic 6-sulfo sialyl Lewis X, 6-sulfo Lewis X, or 6,6Јbissulfo sialyl Lewis X glycolipid was adsorbed to Salmonella minnesota strain R595 and used for repeated intraperitoneal immunizations of BALB/c mice on day 0 (5 g of glycolipid), day 3 (10 g), day 7 (15 g), day 12 (20 g), day 17 (25 g), and day 31 (35 g). Three days after the final immunization, the spleen cells were harvested and fused with mouse myeloma P3/X63-Ag8U1. The same glycolipid was used as the antigen in solid-phase enzyme immunoassays of hybridoma culture supernatants for the cloning procedures. Two hybridoma clones (G152 and G72) were obtained by immunization of the 6-sulfo sialyl Lewis X glycolipid, and four hybridoma clones (G2706, G27011, G27037, and G27039) were obtained by immunization of the 6,6Ј-bissulfo sialyl Lewis X glycolipid. The antibodies AG97 and AG273 were obtained by immunization of the 6-sulfo Lewis X glycolipid. Monoclonal antibodies obtained in this study were all of the IgM class.
Preparation of Carbohydrate Determinants, ELISA, and TLC Immunostaining-ELISA was performed using glycolipid antigens immobilized at the bottom of 96-well culture plates by a standard method described previously (23). Peroxidase-conjugated rabbit anti-rat IgM (-chain-specific; Zymed Laboratories Inc., South San Francisco, CA) was used for HECA-452, and peroxidase-conjugated goat anti-mouse IgM (-chain-specific; Cappel Inc., Malvern, PA) was used for the other antibodies. TLC immunostaining was performed according to the method described previously (23,24), except that peroxidase-conjugated antibody was used as the second antibody, and ECL Western blotting detection reagents (Amersham International, Buckinghamshire, United Kingdom) were used to detect positive reactions instead of 125 I-protein A.
The oligosaccharide SL2L4 was a generous gift from Dr. Keiichi Yoshida (Seikagaku Kogyo, Tokyo, Japan) and was prepared from shark cartilage keratan sulfate by digestion with keratanase II (Bacil-lus sp.; Seikagaku Kogyo), followed by anion-exchange chromatography and gel filtration. The oligosaccharide was coupled to cholesterylaniline through reductive amination (16) by adding the aqueous solution of the oligosaccharide (200 l of 5 mg/ml) to 2 ml of cholesterylaniline in pyridine (180 mg/ml) and incubating for 1 h at 80°C in the presence of 1.0 mM AlCl 3 , followed by additional incubation for 1 h at 80°C after adding 300 mg of dimethylamine-borane and 1 ml of acetic acid. AlCl 3 was removed by Dowex 50W-X8 (100 -200 mesh, H ϩ form) column chromatography from the product SL2L4-cholesterylaniline. Cholesterylaniline had been synthesized by adding 7.2 g of 1,4-p-phenylenediamine (Sigma) in 54 ml of Me 2 SO supplemented with 320 l of pyridine to 1.8 g of cholesteryl chloroformate (Sigma) in 100 ml of 1,2-dichloroethane and incubating for 1 h at room temperature. Cholesterylaniline was recovered from the lower layer after the addition of 46 ml of methanol, 100 ml of 1,2-dichloroethane, and 200 of ml H 2 O.
Immunohistochemical Examination and Binding Assay of L-selectin-IgG Chimera on Frozen Lymph Node Sections-Consecutive frozen sections of 10-m thickness were prepared from human axillary and mesenteric lymph nodes and other tissues and organs for immunohistochemical examination using anti-sialyl Lewis X and newly generated anti-sulfo sialyl Lewis X antibodies. The avidin-biotin complex technique for immunohistochemical study was performed as described in the instructions for the kits (Vectastain) provided by Vector Labs, Inc. (Burlingame, CA). When indicated, the sections were pretreated with sialidase from A. ureafaciens.
For the binding assay of recombinant L-selectin-IgG chimera, frozen sections prepared from freshly obtained human lymph nodes were fixed with 0.5% paraformaldehyde at 4°C for 10 min. The sections were pretreated for 30 min at room temperature with phosphate-buffered saline containing 20% goat serum. To this, a mixture (1:1:1, v/v/v) of the recombinant human L-selectin-IgG chimera (20 g/ml; kindly provided by Drs. H. Kondo and Y. Inoue, New Drug Research Labs, Kanebo Ltd., Osaka, Japan), affinity-purified biotinylated goat anti-human IgG antibody (30 g/ml), and peroxidase-labeled avidin (30 g/ml) was overlaid and incubated for 60 min at 4°C. The reaction was visualized with 3,3Ј-diaminobenzidine tetrahydrochloride, followed by after-staining with 1% methyl green. For the inhibition study, sections were incubated with anti-carbohydrate antibodies (50 -100 g/ml) for 120 min at 4°C prior to the addition of the recombinant human L-selectin-IgG chimera/ peroxidase mixture.
Adhesion of L-selectin-transfected Cells to 6-Sulfo Sialyl Lewis X-A 96-well plate was coated with 6-sulfo sialyl Lewis X by adding ethanol solution containing glycolipid (200 ng/well), egg yolk phosphatidylcholine (1 g/well), and cholesterol (500 ng/well), followed by air-drying at 37°C. After blocking with 5% bovine serum albumin in phosphatebuffered saline overnight at 4°C, the plate was incubated with appropriate inhibitor antibodies (40 g/well) for 30 min at room temperature. BCECF-AM (Molecular Probes, Inc., Eugene, OR)-labeled HULT-7 cells (2 ϫ 10 5 /well), the human L-selectin-transfected cells (15), were added and incubated for 60 min at 37°C. The number of adherent cells was estimated by measuring fluorescence at 490 nm excitation and 530 nm emission on a fluorometer (Fluoroscan II, Labosystems, Tokyo).

Detection of the Sialyl Lewis X-like Determinant on HEV Endothelial Cells Using Anti-sialyl Lewis X and Anti-Lewis X Antibodies-Endothelial cells of HEV in human peripheral
lymph nodes were reactive to only a part of anti-sialyl Lewis X antibodies as shown in Fig. 1 (a-f). The antibodies 2F3 and HECA-452 were strongly reactive to the HEV endothelial cells as well as the 2H5 antibody as reported previously (11). The same pattern of staining was observed consistently with 14 different lymph node samples prepared from different individuals. This indicated that there are two groups of anti-sialyl Lewis X antibodies: one group (2F3, 2H5, and HECA-452) of antibodies that stain HEV endothelial cells and another group (SNH-3, FH-6, and CSLEX-1) that are unable to do so. Reactivity of the 2F3, 2H5, and HECA-452 antibodies was eliminated when the sections were pretreated with sialidase (data not shown). These results suggest that the carbohydrate determinant(s) carried by HEV endothelial cells have sialylated structures closely related to sialyl Lewis X, but not genuine sialyl Lewis X, which would have been readily detected by all six anti-sialyl Lewis X antibodies.

6-Sulfo Sialyl Le X as L-selectin Ligand in Lymph Nodes
HEV endothelial cells were not reactive to four anti-Lewis X antibodies, including LeuM1, FH-2, 73-30 and Tx-17 (in Fig. 1  (g and h), the results obtained with 73-30 and Tx-17 were omitted since they were essentially the same as those obtained with FH-2 and LeuM1), before and after sialidase treatment. The only cells positively stained by all four antibodies were intravascular granulocytes. Staining of monocytes was prominent only with LeuM1. Upon sialidase treatment, the genuine sialyl Lewis X determinant ordinarily yields the Lewis X determinant, which is readily detected by these conventional anti-Lewis X antibodies. This finding again confirmed that the sialyl Lewis X-like determinant on HEV was not genuine sialyl Lewis X. The anti-3Ј-sulfo Lewis X antibody SU59 failed to stain HEV endothelial cells before and after sialidase treatment.
Reactivity of Anti-sialyl Lewis X Antibodies to Synthetic Sulfated Sialyl Lewis X Determinants and Other Related Compounds-The six anti-sialyl Lewis X antibodies reacted equally well to the genuine sialyl Lewis X glycolipid in ELISA (Fig. 2a), and we thought the difference in the reactivity of anti-sialyl Lewis X antibodies against HEV endothelial cells stemmed from the different cross-reactivity of the antibodies. To deter- The antibodies 2H5, 2F3, and HECA-452 did not discriminate the difference between genuine sialyl Lewis X, 6-sulfo sialyl Lewis X, 6Ј-sulfo sialyl Lewis X, and 6,6Ј-bissulfo sialyl Lewis X.
c Antibodies in parentheses indicate that the observed reaction was due to cross-reactivity. The cross-reaction was generally weak except that of the G72 antibody against SL2L4-cholesterylaniline. d SU59 is reactive to both 3Ј-sulfo Lewis X and 3Ј-sulfo Lewis A.

6-Sulfo Sialyl Le X as L-selectin Ligand in Lymph Nodes
mine whether the determinants detected by these anti-sialyl Lewis X antibodies were sulfated sialyl Lewis X determinants, which Hemmerich et al. (13,14) proposed were carried by GlyCAM-1 in murine lymph nodes, we tested the reactivity of the six anti-sialyl Lewis X antibodies against synthetic sulfo sialyl Lewis X determinants. As shown in Fig. 2b, the sialyl Lewis X sulfated at ␤-Gal (6Ј-sulfo sialyl Lewis X) was reactive to an unexpectedly wide variety of antibodies, including CSLEX-1, SNH-3, 2F3, 2H5, and HECA-452. Only the antibody FH-6 lost its binding activity upon 6-sulfation at the ␤-Gal moiety of the sialyl Lewis X structure. On the other hand, sialyl Lewis X sulfated at the ␤-GlcNAc moiety (6-sulfo sialyl Lewis X) was detected by antibodies 2F3, 2H5, and HECA-452, but not by antibodies CSLEX-1, SNH-3, and FH-6 ( Fig. 2c). This reaction profile was exactly the same as that of human HEV endothelial cells against these six antibodies, as indicated in Fig. 1, and suggested that 6-sulfo sialyl Lewis X would be a good candidate for the sialyl Lewis X-like determinant expressed on HEV endothelial cells.
The reactivity of disulfated sialyl Lewis X (6,6Ј-bissulfo sialyl Lewis X) was essentially the same as that of GlcNAc-sulfated sialyl Lewis X (6-sulfo sialyl Lewis X). The antibodies 2F3, 2H5, and HECA-452 were reactive to 6,6Ј-bissulfo sialyl Lewis X, whereas FH-6 and CSLEX-1 were not (Fig. 2d). This led us to assume that 6,6Ј-bissulfo sialyl Lewis X is another good candidate for the determinant carried by human HEV endothelial cells. The only difference between 6-sulfo sialyl Lewis X and 6,6Ј-bissulfo sialyl Lewis X was that the latter determinant was weakly reactive to SNH-3.
Next, we tested the reactivity of four anti-Lewis X antibodies, all of which recognize the Lewis X determinant (Fig. 2e), against the asialo derivatives of the above sulfated sialyl Lewis X determinants (Fig. 2, f-h). Three out of the four anti-Lewis X antibodies were found to react to 6Ј-sulfo Lewis X (Fig. 2f), whereas all of them failed to react to 6-sulfo Lewis X (Fig. 2g). Since sialidasetreated HEV endothelial cells were not stained with any of the anti-Lewis X antibodies in immunohistochemical examination, these findings again suggest that 6-sulfo sialyl Lewis X is actually the sialyl Lewis X-like determinant expressed on HEV endothelial cells, but 6Ј-sulfo sialyl Lewis X is not. Only one anti-Lewis X antibody (FH-2) showed a weak cross-reactivity, and the others were not reactive to 6,6Ј-bissulfo Lewis X (Fig. 2h). None of the 10 antibodies tested was reactive to the 3Ј-sulfo Lewis X determinant, which had been reported to be present in mucin side chains of human meconium (28) and later found to bind to recombinant L-selectin (27).
Generation of Monoclonal Antibodies Specific to 6-Sulfo Sialyl Lewis X-Since analysis of the reactivity pattern of known anti-sialyl Lewis X and anti-Lewis X antibodies strongly suggested 6-sulfo sialyl Lewis X is most probably the sulfated sialyl Lewis X determinant carried by HEV endothelial cells, we generated two monoclonal antibodies (G152 and G72) by immunizing synthetic 6-sulfo sialyl Lewis X. The specificity of these new antibodies against pure carbohydrate determinants in ELISA is shown in Fig. 3. Both antibodies were strongly reactive  6-sulfo sialyl Lewis X, but were not reactive to other related structures, including genuine sialyl Lewis X and 6Ј-sulfo sialyl Lewis X, and were only slightly cross-reactive to 6,6Ј-bissulfo sialyl Lewis X in ELISA (Fig. 3). They did not react to asialo derivatives such as 6-sulfo Lewis X, 6Ј-sulfo Lewis X, 6,6Ј-bissulfo Lewis X, and 3Ј-sulfo Lewis X. G152 was not at all reactive, whereas G72 showed significant reactivity against the SL2L4 determinant, which does not contain a fucose residue (Fig. 3).
The results of TLC immunostaining (Fig. 4) also confirmed the reactivity of the two antibodies, indicating that G152 de-tects only 6-sulfo sialyl Lewis X, whereas G72 detects 6-sulfo sialyl Lewis X and SL2L4. Cross-reactivity to 6,6Ј-bissulfo sialyl Lewis X was not detected by TLC immunostaining (data not shown), suggesting that this was only a weak and negligible cross-reactivity.
These results indicate that the G152 antibody is highly specific to 6-sulfo sialyl Lewis X; it requires the presence of sialic acid and fucose residues and 6-sulfo modification at GlcNAc in the antigenic determinant for binding. The specificity of the G72 antibody was very similar to the specificity of G152, except FIG. 3. Specificity of newly generated anti-6-sulfo sialyl Lewis X antibodies as ascertained by ELISA. a and b show reactivity of the G152 and G72 antibodies, respectively, against carbohydrate determinants (q, 6-sulfo sialyl Lewis X; OE, 6Ј-sulfo sialyl Lewis X; f, 6,6Ј-bissulfo sialyl Lewis X; E, 6-sulfo Lewis X; ‚, SL2L4-cholesterylaniline; Ⅺ, sialyl Lewis X, 6Ј-sulfo Lewis X, 6,6Јbissulfo Lewis X, or 3Ј-sulfo Lewis X). Serial dilution of immobilized carbohydrate determinants started from 40 ng/well. For structures of the carbohydrate determinants, see Table I.  Table I. FIG. 5. Immunohistochemical detection of 6-sulfo sialyl Lewis X on HEV endothelial cells using newly generated monoclonal antibodies. a (low-power field) and b (high-power field) show immunohistochemical staining of HEV endothelial cells in human peripheral lymph nodes with anti-6-sulfo sialyl Lewis X antibody (G152). c shows staining of a human tonsillar lymphoid tissue with another anti-6-sulfo sialyl Lewis X antibody (G72). d shows G152 staining of Peyer's patches, showing a weaker expression of 6-sulfo sialyl Lewis X compared with HEV in peripheral lymph nodes. In d, arrows indicate antigen-positive HEV, and arrowheads indicate antigen-negative HEV.

6-Sulfo Sialyl Le X as L-selectin Ligand in Lymph Nodes
that recognition of the fucose residue by this antibody seemed weaker than that by G152.
Characterization of the 6-Sulfo Sialyl Lewis X Determinant on HEV Endothelial Cells-Both G152 and G72 antibodies strongly stained HEV endothelial cells in human peripheral lymph nodes in immunohistochemical examination (Fig. 5). Frozen sections had to be used in the immunohistochemical detection of the antigen by these two antibodies, and virtually no positive staining was obtained when Formalin-fixed sections were used. As summarized in Table II, these antibodies stained HEV endothelial cells in peripheral lymph nodes, tonsils, Peyer's patches, and appendices. The staining of HEV in Peyer's patches and appendices was significantly weaker than in lymph nodes and tonsils. No blood vessels in other organs including the spleen, thymus, gastrointestinal tract, lung, and kidney were stained by the two antibodies, which is in line with the distribution of L-selectin ligands. This distribution of the antigen also coincides with that detected by the 2H5 antibody as reported earlier (11).
A human lymphoid leukemia cell line transfected with human L-selectin (HULT-7 cells) underwent significant adhesion to a 96-well plate coated with the synthetic 6-sulfo sialyl Lewis X glycolipid (Fig. 6a). The HULT-7 cells adhered also to 6Ј-sulfo sialyl Lewis X and 6,6Ј-bissulfo sialyl Lewis X as well as to genuine sialyl Lewis X. A 50 -60% increase in binding was observed with 6-sulfo sialyl Lewis X and 6,6Ј-bissulfo sialyl Lewis X compared with the binding to genuine sialyl Lewis X. Adhesion to the 6-sulfo sialyl Lewis X glycolipid was strongly inhibited by the addition of the G152 or G72 antibody in vitro, as shown in Fig. 6b. The antibody CSLEX-1, which does not have a binding activity against the carbohydrate determinant, failed to inhibit the adhesion.
The recombinant human L-selectin-IgG chimera bound to HEV endothelial cells in frozen sections prepared from human lymph nodes (Fig. 7a), and this was inhibited by sialidase treatment of frozen sections (Fig. 7b), indicating that a sialic acid-containing carbohydrate determinant is involved in the binding. The G152 and G72 antibodies also clearly inhibited the binding of recombinant human L-selectin-IgG chimera to HEV endothelial cells (Fig. 7, c and d). Although the possibility of steric hindrance cannot be fully excluded, these results strongly suggest that 6-sulfo sialyl Lewis X constitutes a major carbohydrate capping group of the L-selectin ligand on human HEV endothelial cells. The antibody 2H5, the anti-sialyl Lewis X antibody having a heavy cross-reactivity to 6-sulfo sialyl Lewis X, also inhibited the binding (Fig. 7e), whereas CSLEX-1, the anti-sialyl Lewis X antibody with no reactivity to 6-sulfo sialyl Lewis X, did not (Fig. 7f).
Detection of 6,6Ј-Bissulfo sialyl Lewis X Using Newly Generated Specific Monoclonal Antibodies-Next, we generated antibodies directed against 6,6Ј-bissulfo sialyl Lewis X since the determinant seemed to be another candidate for the sulfated L-selectin ligand expressed in human lymph nodes. Four monoclonal antibodies (G2706, G27011, G27037, and G27039) were obtained by the immunization of the synthetic 6,6Ј-bissulfo sialyl Lewis X glycolipid. These antibodies showed strong reactivity to the immunogen 6,6Ј-bissulfo sialyl Lewis X and were moderately cross-reactive to its desialylated structure, 6,6Јbissulfo Lewis X, in ELISA (Fig. 8, a-d). None of these antibodies reacted to 6-sulfo sialyl Lewis X, genuine sialyl Lewis X, or other related structures, including SL2L4-cholesterylaniline, but some of them showed a weak cross-reactivity to 6Јsulfo sialyl Lewis X.
In TLC immunostaining, the antibodies appeared to be quite  c Only the findings on endothelial cells are counted. Epithelial cells in these organs sometimes showed positive reactions against some antibodies.
FIG. 6. Inhibition of adhesion of L-selectin-transfected human lymphoid cells to 6-sulfo sialyl Lewis X-coated plates by the G152 and G72 antibodies. a shows adhesion of BCECF-AM-labeled HULT-7 cells expressing human L-selectin to 96-well plates coated with the synthetic genuine sialyl Lewis X, 6Ј-sulfo sialyl Lewis X, 6-sulfo sialyl Lewis X, or 6,6Ј-bissulfo sialyl Lewis X glycolipid. In b, adhesion of BCECF-AM-labeled HULT-7 cells to 96-well plates coated with 6-sulfo sialyl Lewis X was almost completely inhibited by the presence of the G152 or G72 antibody. Adherent cells are expressed as a percentage of the number of cells added to the wells. Statistical significance was calculated according to Student's t test. For experimental details, see "Experimental Procedures." N.S., not significant. specific to 6,6Ј-bissulfo sialyl Lewis X (Fig. 9a), and crossreactivity to other structures weakly detected by ELISA did not show up on TLC immunostaining. For reasons yet unknown, the G27037 antibody showed multiple positive bands in 6,6Јbissulfo sialyl Lewis X (lane 6).
When applied to immunohistochemical examination of HEV endothelial cells in human peripheral lymph nodes, these four antibodies all failed to react to HEV endothelial cells (Fig. 10, e-h), in clear contrast to the strong staining with the G152 and G72 antibodies observed on the serial frozen sections prepared from the same lymph node (Fig. 10, a and b). These findings suggest that the major molecular species of the sulfated sialyl Lewis X determinant on HEV endothelial cells is 6-sulfo sialyl Lewis X and that expression of the 6,6Ј-bissulfo sialyl Lewis X determinant is negligible.
Detection of 6-Sulfo Lewis X in Sialidase-treated HEV Endothelial Cells-By immunizing sialidase-treated 6-sulfo sialyl Lewis X, we further established two monoclonal antibodies directed against the corresponding asialo compound, 6-sulfo Lewis X. The antibodies (AG97 and AG273) thus generated had a strong reactivity to 6-sulfo Lewis X in ELISA, but were not reactive to other related asialo structures, including Lewis X, 6Ј-sulfo Lewis X, and 3Ј-sulfo Lewis X (Fig. 8, e and f). Of course, these antibodies lacked reactivity to the sialylated determinants such as 6-sulfo, 6Ј-sulfo, 6,6Ј-bissulfo, and genuine sialyl Lewis X. Weak reactivity was noted against 6,6Ј-bissulfo Lewis X in ELISA, but this reactivity was not detected in TLC immunostaining (Fig. 9b, lane 4), suggesting that it was only a weak cross-reactivity.
These two antibodies did not stain intact HEV endothelial cells in human lymph nodes in immunohistochemical examination, but strongly labeled the cells when the sections were pretreated with sialidase ( Fig. 10, i-l). This was in a striking contrast to the strong staining of intact HEV by G152 and G72 antibodies, which was nearly completely eliminated after sialidase treatment (Fig. 10, a-d). These findings indicate that 6-sulfo sialyl Lewis X was abundantly present, whereas there was little or none of its asialo form (6-sulfo Lewis X) on intact HEV endothelial cells. The latter determinant was produced only when 6-sulfo sialyl Lewis X in the frozen sections was hydrolyzed with sialidase.

DISCUSSION
The aberrant reaction of anti-sialyl Lewis X antibodies against HEV endothelial cells in human lymph nodes has long puzzled researchers (11,29). Some anti-sialyl Lewis X antibodies react to HEV and inhibit the binding of L-selectin to HEV, yet others do not. Our preliminary study (Ref. 15 and this study), using a series of hitherto known anti-sialyl Lewis X and anti-Lewis X antibodies, suggested that 6-sulfo sialyl Lewis X and 6,6Ј-bissulfo sialyl Lewis X, but not 6Ј-sulfo sialyl Lewis X, are possible candidates for the complex sialyl Lewis X-like determinants present on HEV endothelial cells. 6-Sulfo sialyl Lewis X seemed to be the most predominant molecular species of sulfated sialyl Lewis X-related determinants on human HEV endothelial cells, judging from their reactivity pattern to hith- FIG. 7. Inhibition of binding of the recombinant human L-selectin-IgG chimera to HEV endothelial cells in peripheral lymph nodes by anti-6-sulfo sialyl Lewis X antibodies. Binding of the recombinant human L-selectin-IgG chimera to HEV endothelial cells was studied using frozen sections of human peripheral lymph nodes. a shows that the recombinant human L-selectin-IgG chimera significantly bound to HEV endothelial cells, which was eliminated when the sections were sialidase-treated as in b. c and d show that the binding was almost completely inhibited when the sections were pretreated with the anti-6-sulfo sialyl Lewis X antibodies G152 and G72, respectively. A similar inhibition of binding was observed when the sections were pretreated with the 2H5 antibody (e), but not with CSLEX-1 (f). For experimental details, see "Experimental Procedures." FIG. 8. Specificity of newly generated 6,6-bissulfo sialyl Lewis X and anti-6-sulfo Lewis X antibodies as ascertained by ELISA. a-d show reactivity of the anti-6,6Ј-bissulfo sialyl Lewis X antibodies G2706, G27011, G27037, and G27039, respectively, and e and f show reactivity of the anti-6-sulfo Lewis X antibodies AG97 and AG273, respectively, against carbohydrate determinants (q, 6-sulfo sialyl Lewis X; OE, 6Ј-sulfo sialyl Lewis X; f, 6,6Ј-bissulfo sialyl Lewis X; E, 6-sulfo Lewis X; ‚, 6Ј-sulfo Lewis X; Ⅺ, 6,6Ј-bissulfo Lewis X; छ, genuine sialyl Lewis X, Lewis X, or 3Ј-sulfo Lewis X). Serial dilution of immobilized carbohydrate determinants started from 40 ng/well. For structures of the carbohydrate determinants, see Table I. erto known anti-sialyl Lewis X and anti-Lewis X antibodies. The results of this study using a series of new monoclonal antibodies directed against the respective sulfated determi-nants provided strong evidence corroborating the predominant expression of 6-sulfo sialyl Lewis X on HEV endothelial cells. The anti-6-sulfo sialyl Lewis X antibodies G152 and G72 strongly labeled HEV in immunochemical examination, and the anti-6-sulfo Lewis X antibodies AG97 and AG273 intensely stained HEV after sialidase treatment.
If a significant amount of conventional sialyl Lewis X were expressed on HEV, the reactivity of HEV against anti-sialyl Lewis X antibodies should have been CSLEX-1 ϩ SNH-3 ϩ FH-6 ϩ 2H5 ϩ 2F3 ϩ HECA-452 ϩ instead of CSLEX-1 Ϫ SNH-3 Ϫ FH-6 Ϫ 2H5 ϩ 2F3 ϩ HECA-452 ϩ . Moreover, the Lewis X determinant should have been detected by all conventional anti-Lewis X antibodies after an appropriate sialidase treatment, but it was not. Similarly, if a considerable amount of 6Ј-sulfo sialyl Lewis X were present on HEV, the reactivity of HEV against antisialyl Lewis X antibodies should have been CSLEX-1 ϩ SNH-3 ϩ FH-6 Ϫ 2H5 ϩ 2F3 ϩ HECA-452 ϩ , and its asialo form determinant (6Ј-sulfo Lewis X), which is detectable by a majority of conventional anti-Lewis X antibodies (Fig. 2), should have been detected after sialidase treatment. Concerning the disulfated determinant, we could not obtain any positive evidence to support the presence of 6,6Ј-bissulfo sialyl Lewis X in HEV of human peripheral lymph nodes, although we generated as many as four monoclonal antibodies specific to the determinant.
Our results also indicated that 6-sulfo sialyl Lewis X accounts for the majority of the carbohydrate capping group of the Lselectin ligand on HEV endothelial cells as far as human HEVs are concerned, judging from the nearly complete inhibition of recombinant L-selectin binding to HEV endothelial cells. These results are at variance with the previous notion that nearly equal amounts of 6Ј-sulfo and 6-sulfo sialyl Lewis X are expressed on murine HEV (13,14). This might be partly due to the species difference since we could not detect any significant reactivity of our anti-6-sulfo sialyl Lewis X antibodies (G152 and G72) against HEV endothelial cells in murine and rat lymph nodes. 2 It is well known that sialyl Lewis X, the putative ligand for E-and Pselectins, is barely detectable on murine and rat leukocytes using most of the conventional anti-sialyl Lewis X antibodies that readily detect the determinant on human leukocytes, and the carbohydrate ligand on HEV for L-selectin in mice and rats could well be different from that in humans.
An isoenzyme of ␣133-fucosyltransferase (fucosyltrans- FIG. 9. Specificity of newly generated anti-6,6-bissulfo sialyl Lewis X and anti-6-sulfo Lewis X antibodies as ascertained by TLC immunostaining. a shows reactivity of the anti-6,6Ј-bissulfo sialyl Lewis X antibodies G2706, G27011, G27037, and G27039, respectively, against carbohydrate determinants (lane 1, sialyl Lewis X; lane 2, 6Ј-sulfo sialyl Lewis X; lane 3, 6Ј-sulfo Lewis X; lane 4, 6-sulfo sialyl Lewis X; lane 5, 6-sulfo Lewis X; lane 6, 6,6Ј-bissulfo sialyl Lewis X; lane 7, 6,6Ј-bissulfo Lewis X). b shows reactivity of the anti-6-sulfo Lewis X antibodies AG97 and AG273 against carbohydrate determinants (lane 1, 6-sulfo sialyl Lewis X; lane 2, 6-sulfo Lewis X; lane 3, 6,6Ј-bissulfo sialyl Lewis X; lane 4, 6,6Ј-bissulfo Lewis X). The orcinol panels show results from TLC plates developed with orcinol/H 2 SO 4 reagent, indicating the TLC mobility of each glycolipid. For structures of the carbohydrate determinants, see Table I. FIG. 10. Immunohistochemical detection of 6,6-bissulfo sialyl Lewis X and 6-sulfo Lewis X on HEV endothelial cells using newly generated monoclonal antibodies. a-d show immunohistochemical staining with the anti-6-sulfo sialyl Lewis X antibodies G152 (a and c) and G72 (b and d) of untreated (a and b) and sialidase-treated (c and d) frozen sections of human peripheral lymph nodes, serving as staining controls. e-h show immunohistochemical staining with the anti-6,6Ј-bissulfo sialyl Lewis X antibodies G2706, G27011, G27037, and G27039, respectively, of untreated frozen sections of peripheral lymph nodes. i-l show immunohistochemical staining with the anti-6-sulfo Lewis X antibodies AG97 (i and k) and AG273 (j and l) of untreated (i and j) and sialidase-treated (k and l) frozen sections of lymph nodes. ferase VII) has been shown to be present on HEV and is suggested to be mainly involved in the synthesis of the Lselectin ligand on HEV (30). In line with this, we found only a disproportionally small amount of the asialo species of the 6-sulfo sialyl Lewis X determinant (i.e. 6-sulfo Lewis X) on human HEV in this study. This implies the contribution of fucosyltransferase VII in the synthesis of the determinant since fucosyltransferase VII is the only ␣133-fucosyltransferase isoenzyme that lacks the synthetic ability of the asialo species of the corresponding determinant (31,32). Malý et al. (33) studied the substrate specificity of fucosyltransferase VII and concluded that the enzyme has an ability to transfer fucose residues onto 6-sulfated lactosamine, but not onto 6Ј-sulfated lactosamine. Habuchi et al. (34) recently studied the substrate specificity of a cloned 6Ј-sulfotransferase and reported that the enzyme preferred Gal␤ moieties in partially sulfated polylactosamine much more than those in non-sulfated polylactosamine as an acceptor of sulfate residues. On the other hand, 6-sulfation has recently been suggested to occur on the terminal GlcNAc␤ moiety even before the transfer of the Gal␤ moiety, and 6-sulfolactosamine has been suggested to be formed afterward by the action of ␤134-galactosyltransferase on 6-sulfo-GlcNAc␤ (35). These findings imply that sulfation at the 6Ј-position occurs mainly after the formation of 6-sulfolactosamine and may not proceed without 6-sulfo-GlcNAc␤ during the synthesis of sulfated substrates for fucosyltransferases. This could explain our results showing that 6-sulfo sialyl Lewis X is strongly expressed, whereas the 6Ј-sulfo species is virtually undetectable in human HEV endothelial cells.
P-selectin glycoprotein ligand-1 (PSGL-1) has been reported to be necessary for the high affinity binding of sialyl Lewis X-positive leukocytes not only to P-selectin, but also to Lselectin (36 -39). This finding was reproducible in our laboratory regarding the leukocyte-leukocyte interaction. The sulfated tyrosine residue near the N terminus of PSGL-1 was proposed to be responsible for the observed high affinity binding to L-selectin. The problem with the lymphocyte-HEV interaction was that we could not detect PSGL-1 in human HEV endothelial cells using well characterized specific antibodies in our preliminary experiments. When tested in the binding assay using recombinant soluble human L-selectin (40) and in the cell adhesion assay using human L-selectin-transfected cells (Fig.  6a), the synthetic 6-sulfo sialyl Lewis X glycolipid exhibited a 30 -60% increase in L-selectin binding when compared with the genuine sialyl Lewis X glycolipid. It is not clear if this increase is comparable to the enhancing effect of PSGL-1 on L-selectin-mediated cell adhesion. Sanders et al. (41) recently studied the binding activity of several synthetic sulfated Lewis X determinants and suggested that even a subtle difference in the binding activity of the 6-sulfo species observed in the monovalent binding assay would possibly exert a profound effect on the polyvalent binding in intercellular adhesion.
Although this would provide one possible explanation, the sulfate residue at the penultimate GlcNAc in 6-sulfo sialyl Lewis X nevertheless may not substitute for the sulfate residue near the N terminus of PSGL-1 in the high affinity binding to L-selectin. An alternative explanation would be the presence of an amount of PSGL-1 too small to be detected by immunohistochemical examination or the presence of PSGL-1-like molecule(s) on HEV endothelial cells. Although we failed to detect it on normal human HEV endothelial cells, PSGL-1 is reportedly expressed on microvascular endothelial cells in some pathologic tissues (42).