Modified heparin inhibits P-selectin-mediated cell adhesion of human colon carcinoma cells to immobilized platelets under dynamic flow conditions.

Accumulating evidence indicates that the formation of tumor cell platelet emboli complexes in the blood stream is a very important step during metastases and that the anti-metastasis effects of heparin are partially due to a blockade of P-selectin on platelets. In this study, heparin and chemically modified heparins were tested as inhibitors of three human colon carcinoma cell lines (COLO320, LS174T, and CW-2) binding to P-selectin, adhering to CHO cells expressing a transfected human P-selectin cDNA, and adhering to surface-anchored platelets expressing P-selectin under static and flow conditions. The aim was to screen for heparin derivatives with high anti-adhesion activity but negligible anticoagulant activity. In this study, four modified heparins with high anti-adhesion activity were identified including RO-heparin, CR-heparin, 2/3ODS-heparin, and N/2/3DS-heparin. NMR analysis proved the reliability of structure of the four modified heparins. Our findings suggested that the 6-O-sulfate group of glucosamine units in heparin is critical for the inhibition of P-selectin-mediated tumor cell adhesion. Heparan sulfate-like proteoglycans on these tumor cell surfaces are implicated in adhesion of the tumor cells to P-selectin. Some chemically modified heparins with low anticoagulant activities, such as 2/3ODS-heparin, may have potential value as therapeutic agents that block P-selectin-mediated cell adhesion and prevent tumor metastasis.

Hematogenous metastasis is a highly regulated and dynamic process in which the cancerous cells separate from the primary tumor, migrate across blood vessel walls into the bloodstream, and disperse throughout the body to generate new colonies. Several lines of experiments have demonstrated that platelets can facilitate hematogenous dissemination of tumor cells (1). In this process, platelets act mainly through emboli complexes formed by the interaction of tumor cells and platelets. The formation of tumor cell-platelet emboli complexes in the blood stream not only provides a protective shield that masks them from the cytotoxic activity of natural killer cells, but also favors tumor cell adhesion to vascular endothelial cells (2)(3)(4).
The interaction between tumor cells and platelets may involve several kinds of adhesion molecules. Selectins are a family of intercellular adhesion molecules that mediate the initial adhesion of leukocytes to the endothelia of blood vessels during inflammation (5)(6)(7)(8)(9). The family includes E-, P-, and L-selectin, and all three selectins can bind to sialylated, fucosylated, or, in some cases, sulfated glycans on glycoproteins, glycolipids, or proteoglycans (10). The tetrasaccharides [Neu5Ac␣2,3Gal␤1,4-(Fuc␣1,3)GlcNAc] (sLe x ) 1 and [Neu5Ac␣2,3Gal␤1,3(Fuc␣1,4)-GlcNAc] (sLe a ) have been identified as the minimal ligands for all three types of selectins. Recently, accumulating evidence indicates that P-selectin plays a crucial role during hematogenous metastasis, and P-selectin has been shown to bind to several human cancers and human cancer-derived cell lines, such as colon cancer, lung cancer, breast cancer, malignant melanoma, gastric cancer, tongue squamous cancer, and neuroblastoma (11)(12)(13)(14)(15)(16)(17)(18). The role of P-selectin on both endothelium and platelets is to mediate the tumor cell adhesion to vascular endothelial cells and the interaction of activated platelets and tumor cells during metastasis. According to recent reports, P-selectin plays roles mainly in the initial process of tumor cell adhesion to platelets (19). Therefore, blocking P-selectin on platelets with antagonists can effectively inhibit the formation of tumor cell-platelet emboli complexes and successfully prevent hematogenous metastasis.
Various P-selectin antagonists have been discovered at present, such as mAbs, sLe x , sLe a , and its mimetics, as well as recombinant PSGL-1 (20). As these selectin inhibitors have various drawbacks, such as narrow cross-reactivity, weak affinity, relatively low selectivity among the selectins, short circulating half-life, great expense to generate, potential antigenicity, and a very limited track record as intravenous therapeutic agents, the development of these compounds into effective drugs for clinical use has been greatly limited. Heparin has been used clinically as an anticoagulant for over 60 years. Heparin and heparan sulfate are linear polysaccharides consisting of uronic acid-(134)-D-glucosamine-repeating disaccharide subunits. In addition to its effects on anticoagulant, heparin exhibits many other activities (21). Several studies have demonstrated that heparin and heparan sulfate are li-gands for P-selectin and can block its binding to carbohydrate ligand (22). Therefore, one of the proposed mechanisms by which heparin can inhibit metastasis is that heparin may block the P-selectin-based interaction of platelets and tumor cells during metastasis. It is reasonable that heparin should be a candidate for a clinical trial in the immediate future. However, the kind of native human tumors that actually use the Pselectin-based ligands to interact with platelets is unknown, and little information is available about the specific oligosaccharide structures in heparin that are recognized by P-selectin. In addition, heparin's significant anticoagulant properties and its potential for bleeding complications may contraindicate its use as an anti-metastasis compound.
The anticoagulant activity of heparin depends on its unique antithrombin binding pentasaccharide sequence (23). Chemical modifications, which are directed against the unique structure, can effectively reduce the anticoagulant activities of heparin. In this study, we prepared eight chemically modified heparins and analyzed whether they could still effectively reduce tumor cell adhesion. The results demonstrated that heparin and four modified heparins could significantly inhibit P-selectin-mediated tumor cell adhesion. The sulfate groups at C6 on the glucosamine residues in heparin play a critical role in the inhibition. Our results also showed that three human colon carcinoma cell lines strongly expressed heparan sulfate proteoglycans, which are involved in adhesion of these tumor cells as a ligand of P-selectin. Our data suggest that some chemically modified heparins with low anticoagulant activities, such as 2/3ODS-heparin, have potential value for preventing tumor hematogenous metastasis.
The anticoagulant activity of heparin and modified heparinoids was analyzed by aPTT assay. Clotting times were determined using an ACL200 Automated Coagulation Laboratory (Japan) and lyophilized silica aPTT kit.
Protein and Antibodies-Recombinant human P-selectin/Fc chimera protein (P-Fc), 9E1 (a leukocyte adhesion blocking IgG mAb against P-selectin), and P1 (a leukocyte adhesion nonblocking IgG mAb against P-selectin) were purchased from R&D Systems, Inc. CSLEX-1, an IgM mAb to sLe x , was a gift from O. Spertini. 10E4, an IgM mAb to the native heparan sulfate chains of proteoglycans, was generously provided by Dr. Guido David. KPL1, an IgG mAb to PSGL-1 was purchased from Santa Cruz Biotechnology, Inc. Mouse IgM isotype control and FITC-conjugated Ab against mouse IgM were purchased from Sigma-Aldrich Inc. Human IgG and mouse IgG isotype control, FITC-conjugated Ab against human and mouse IgG were purchased from Jackson ImmunoResearch Laboratories, Inc.
Cells-The human colon carcinoma cells (COLO320, LS174T, and CW-2), CHO cells, and U937 cells were purchased from the Cell Bank of Type Culture Collection of Chinese Academy of Science (Shanghai, China). They were cultured in IMDM (Invitrogen) supplemented with 10% heat-inactivated fetal bovine serum (FBS) at 37°C in the presence of 5% CO 2 . COLO320 and CHO cells were harvested with 0.02% EDTA. LS174T and CW-2 cells were detached by mild trypsinization (0.25% trypsin/EDTA for 2 min at 37°C; Invitrogen), and subsequently incubated at 37°C for 2 h to regenerate surface glycoproteins, as previously described (11). Subsequently, tumor cells were washed once with PBS (10 mM HEPES, pH 7.4), resuspended in serum-free media containing 0.1% bovine serum albumin (BSA). CHO cells expressing a transfected human P-selectin cDNA (CHO-P cells) were generated by transient transfection of CHO cells, which was performed using the Poly-Fect transfection reagent (Qiagen) according to instructions from the manufacturer.
Immobilization of Platelet Layers on Glass Slides-To provide a substrate that readily binds platelets, circular glass slides were coated with 3-aminopropyltriethoxysilane (APES; Sigma). Human blood was drawn by venipuncture from healthy volunteers into sodium citrate (0.38% w/v) anticoagulant. Platelet-rich plasma (PRP) was prepared by centrifugation of whole blood at 200 ϫ g for 15 min. The PRP count was adjusted to 2 ϫ 10 8 /ml before being bound to APES-treated circular glass slides for 60 min (30). Nonspecific binding was blocked with 0.1% BSA for 10 min at 37°C. Under these conditions, a confluent layer of platelets was formed as evaluated by light microscopy for each experiment. The density and confluency of platelet layers were not affected during the flow experiment.
Flow Cytometric Assays-All cells were washed once and resuspended in PBS/FBS (5 ϫ 10 6 cells/ml). For the cell surface P-selectin binding assay, each aliquot (0.1 ml) of cells was incubated with 0.3 g of human IgG or P-Fc at 22°C for 20 min, followed by a FITC-conjugated Ab against human IgG at 22°C for 1 h with end-to-end rotation. Cells were sedimented at 1500 rpm for 5 min in a tabletop centrifuge. Each aliquot was then resuspended in 0.5 ml of PBS/FBS for immediate flow cytometric analysis (FACScan, Beckman-Counter). For antibody or heparin derivative inhibition experiments, 0.3 g of P-Fc were preincubated with 1.0 g of 9E1, P1, or heparin derivatives of various concentrations at 22°C for 30 min. The following experiments were as above. For cell surface P-selectin ligand experiments, each aliquot (0.1 ml) of cells was incubated with 1ϳ2 g of mouse IgM, mouse IgG, CSLEX-1 mAb, 10E4 mAb, or KPL1 mAb at 22°C for 30 min, followed by treatment with a FITC-conjugated Ab against mouse IgM or mouse IgG at 22°C for 1 h with end-over-end rotation.
Laminar Flow Adhesion Assays-Tumor cell adhesion to immobilized platelets (or CHO-P cells) was quantitated under flow conditions using a parallel plate flow chamber (GlycoTech, Rockville, MD). All cells were washed once and resuspended to 1 ϫ 10 6 cells/ml in serum-free media containing 0.1% BSA. A platelet-coated circular glass slide (or a CHO-P cell-coated 35-mm tissue culture dish) was assembled in a flow chamber and mounted on the stage of an inverted microscope (Olympus Optical, Tokyo, Japan) equipped with a camera (Panasonic, Yokohama, Japan) and connected to a VCR and a TV monitor.
Surface-adhered platelets were then incubated with either 1 unit/ml thrombin or Dulbecco's PBS, 0.1% BSA for 10 min at 37°C. After washing the platelet layer with Dulbecco' s PBS, 0.1% BSA for ϳ2 min, tumor cells were perfused through the chamber for 3 min at the appropriate flow rates to obtain wall shear stresses of 0.3-1.2 dyn/cm 2 at 22°C using a syringe pump (Cole-Parmer Instrument Co.), thereby mimicking the fluid mechanical environment of the microcirculation and postcapillary venules. Interactions between tumor cells and surface-adherent platelets (or CHO-P cells) were visualized in real time by phase-contrast videomicroscopy. The numbers of bound cells were quantified from videotape recordings of 10 -20 fields of view obtained (3 min after flowing cells through the chamber) while scanning the lower plate of the flow chamber using a ϫ10 objective lens. Their number was manually determined by reviewing the videotapes. For some inhibition experiments, immobilized platelets (or CHO-P cells) were preincubated with mAb (20 g/ml) or heparin derivatives with various concentrations at 22°C for 30 min.
Statistics-Data were expressed as the mean Ϯ S.D. Statistical significance of differences between means was determined by analysis of variance. If means were shown to be significantly different, multiple comparisons by pairs were performed with the Tukey test. Probability values of p Ͻ .05 or p Ͻ .01 were selected to be statistically significant.

COLO320 Cells Express P-selectin Ligand-P-selectin
has been shown to bind to several human cancers and human cancer-derived cell lines, although which kind of native human tumors use P-selectin-based mechanisms to successfully metastasize is still not clear. In this assay, COLO320, a human colon carcinoma cell line, was first examined for the interaction with P-Fc using a static cell surface binding assay, and a FITC-conjugated Ab to human IgG was used to reveal the binding of P-Fc to tumor cells by flow cytometry. As shown in Fig. 1, P-Fc, but not human IgG, bound to COLO320 cells. The percentage of positive cells binding to P-selectins was 90.3%. Preincubation of P-Fc with 9E1 (a leukocyte adhesion-blocking mAb to P-selectin), but not P1 (a leukocyte adhesion nonblocking mAb to P-selectin), inhibited this binding, confirming the binding specificity of P-selectin to COLO320 cells.
Modified Heparins Inhibit P-selectin Binding to COLO320 Cells under Static Conditions-Using various well established methods, we prepared a series of chemically modified heparins, including RO-heparin, CR-heparin, CRS-heparin, 2/3ODS-heparin, NDS-heparin, N/2/3DS-heparin, 6ODS-heparin, and N/6DS-heparin. Representative saccharide units of each type of heparinoid are given in Fig. 2. The Sepharose CL-6B gel chromatographic analysis and the ion chromatographic analysis indicated that the integrity and the molecular weights (data not show) of the main structures of the eight chemically modified heparins remained whereas the sulfate content changed as expected (Table I).
To determine the in vitro anticoagulant activities of these chemically modified heparins, we measured the aPTT values. As shown in Table II, all modified heparins had reduced their anticoagulant activities to some extent when measured with fresh human plasmas. The eight heparin derivatives can be ranked according to their anticoagulant activities as: heparin Ͼ RO-heparin Ͼ 2/3ODS-heparin Ͼ NDS-heparin Ͼ CRheparin Ͼ CRS-heparin Ͼ 6ODS-heparin Ͼ N/2/3DS-heparin Ͼ N/6DS-heparin. These results indicate that the more the heparins were modified, the more their anticoagulant activities were reduced.
As discussed earlier, heparin can function as a ligand for P-selectin and can block P-selectin binding to its native carbohydrate ligands (22). In this assay, therefore, we examined whether modified heparins could inhibit the P-selectin binding under static conditions. First, we used heparin and modified heparins with a high concentration of 5.0 mg/ml to conduct the experiment. Fig. 3A shows that heparin, RO-heparin, CR-heparin, 2/3ODS-heparin, and N/2/3DS-heparin, but not CRS-heparin, NDS-heparin, 6ODS-heparin, and N/6DS-heparin, could significantly abolish the binding of P-selectin to COLO320 cells. Compared with the positive control, the preincubation of heparin, RO-heparin, CR-heparin, 2/3ODS-heparin, and N/2/ 3DS-heparin reduces the percentage of COLO320 cells binding to P-selectin by 92, 81, 80, 80, and 71%, respectively. In the following experiments, we tested the above four modified heparins with low concentrations of 1.0 mg/ml and 0.1 mg/ml to determine their ability to inhibit tumor cells binding to Pselectin. The results showed that the four modified heparins at concentrations of 1.0 mg/ml, to some extent, had an inhibiting effect (Fig. 3B). However, at the concentration of 0.1 mg/ml, the inhibition was faint (Fig. 3C). In brief, the results indicated that some modified heparins remain able to inhibit P-selectin binding to tumor cells, and the ability is proportionally related to their concentration.
Modified Heparins Inhibit COLO320 Cells Adhering to CHO-P Cells under Flow Conditions-The previous experiment  proved that human colon carcinoma COLO320 cells can bind to P-selectin under static conditions, and the binding can be inhibited by heparin and the four modified heparins. However, in the experiment above, recombinant human P-selectin/Fc chimera protein (P-Fc) was used instead of P-selectin, and the experiment was conducted under static conditions. Therefore, these experiments cannot completely reflect the binding of tumor cells to P-selectin under physiological conditions. To corroborate the above findings, CHO cells expressing a transfected human P-selectin cDNA (CHO-P) was used as the carrier of P-selectin, and the flow adhesion assay, mimicking the fluid mechanical environment of the microcirculation and postcapillary venules in vitro, was performed.
In this assay, tumor cells were perfused through a parallel  3. Inhibition of P-selectin binding to COLO320 by heparin and modified heparins. A, P-Fc was preincubated with heparin or modified heparins (5.0 mg/ml). B, P-Fc was preincubated with heparin or modified heparins (1.0 mg/ml). C, P-Fc was preincubated with heparin or modified heparins (0.1 mg/ml). Then tumor cells were incubated with blocked or unblocked P-Fc, followed by a FITC-conjugated Ab against human IgG. The binding events were analyzed by flow cytometry. Results are presented as histograms of the log fluorescence intensities from 10 4 cells. The proportion of positive cells is indicated in each histogram. plate flow chamber whose lower plate was coated with a layer of CHO-P cells at the appropriate flow rates to obtain wall shear stresses of 0.3 dyn/cm 2 . The interaction of tumor cells and CHO-P cells can be observed when the tumor cells run through the surface of CHO-P cells. As shown in Fig. 4, COLO320 cells bound to CHO-P cells, but not to CHO cells. Preincubation of CHO-P cells with 9E1, but not P1, inhibited this binding, indicating the binding specificity of COLO320 cells to CHO-P cells. To investigate whether heparin and the four modified heparins could inhibit the binding of COLO320 cells to CHO-P cells, we conducted the following experiments.

Modified Heparins Inhibit COLO320 Cells Adhering to Surface-immobilized Platelets under Flow Conditions-The interaction between platelets and the tumor cells is crucial to successful metastasis. Previous studies have shown that platelets
interact with a number of tumor cell lines, but the cell adhesion experiments were conducted exclusively under static conditions (11,12,31). To determine whether COLO320 cells also adhere to platelets under flow conditions and to determine the role of P-selectin on platelets in the process of cell adhesion, tumor cells were perfused through a parallel plate flow chamber whose lower plate was coated with a layer of platelets. Fig.  5A shows that immobilized platelets supported COLO320 cell adhesion in a shear stress-dependent manner. No interaction was observed at 1.2 dyn/cm 2 . In the following experiments, using mAb against P-selectin, we examined the potential role of platelet P-selectin in the adhesive interaction under flow conditions. The results showed that incubation of the platelets layer with 9E1 alone resulted in an ϳ50 -58% reduction of adhesion at 0.3 dyn/cm 2 or 0.6 dyn/cm 2 , and P1 nearly had no effect on the adhesion (Fig. 5, B and C). The results above indicated that COLO320 cells could adhere to platelets under flow, and P-selectin, which expresses on the surface of platelets, plays an important role in the process of the adhesion of tumor cells to platelets.
In the following step, we examined the effect of heparin and the four modified heparins on inhibiting the adhesion of COLO320 cells to surface-immobilized platelets under flow conditions. The results showed that heparin and the four modified heparins could significantly inhibit the adhesion of COLO320 cells to surface-immobilized platelets. At all concentrations of modified heparins used in the experiment, 5.0, 1.0, and 0.1 mg/ml, were dramatically effective for blocking the adhesion of the tumor cells to the surface-adhered platelets under the condition of the wall shear stress of 0.3 dyn/cm 2 or 0.6 dyn/cm 2 (Fig. 5, B and C). Even at a concentration of 0.1 mg/ml, which is the lowest concentration tested, heparin and the four modified heparins could still significantly inhibit the adhesion of COLO320 cells to surface-immobilized platelets.
Adhesion of COLO320 Cells to P-selectin Involves Heparan Sulfate-like Proteoglycans-We next attempted to identify the counter receptor for platelet P-selectin on COLO320 cell surface using oligosaccharide-specific mAbs. In this assay, we used U937 cells as a control because U937 cells express PSGL-1, a natural ligand for P-selectin, and it is well known that under flow conditions, U937 cells can roll on and adhere to immobilized P-selectin. Fig. 6A shows that CSLEX-1 (a mAb to sLe x ) did not bind to U937 cells, and COLO320 cells had no apparent expression of sLe x . 10E4 (a mAb to heparan sulfates) recognized COLO320 cells, but did not recognize U937 cells. These results suggested that COLO320 cells strongly expressed heparan sulfate-like proteoglycans on their cell surfaces. Prior studies have demonstrated that the major ligand on leukocytes involved in the recruitment to P-selectin is PSGL-1. Therefore, its contribution to COLO32O cell adhesion to surface-adhered platelets was examined. Flow cytometric analysis of COLO320 adhesion receptor expression suggested that PSGL-1 was not expressed on the tumor cell surface (Fig. 6B). These results are in agreement with recent reports that suggest that a variety of cell lines from colon carcinomas interact with P-selectin substrates in a PSGL-1-independent manner (14,19). Our results imply that heparan sulfate-like proteoglycans expressing on the COLO32O cells may mediate the binding of tumor cells to P-selection. Heparin as the ligand of P-selectin can inhibit P-selectin-mediated tumor cell adhesion, because heparin has the similar structure with heparan sulfate proteoglycans.
To corroborate the above inference, we digested COLO320 cells with a combination of three heparinases (heparinase I, II, and III), which remove heparan sulfate-like proteoglycans from cell surfaces. As shown in Fig. 6C, treatment with heparinases reduced the binding of P-selectin to COLO320 cells, and also reduced the binding of 10E4 to COLO320 cells, suggesting that heparan sulfate-like proteoglycans of cell surface were involved in the binding of P-selectin to COLO320 cells.
The Molecular Mechanism of P-selectin-mediated Human Colon Carcinoma Cell Adhesion May Exist Broadly-To validate that modified heparins are capable of inhibiting adhesion of colon cancer cells, in addition to COLO320, to immobilized platelets under flow conditions, we checked the other human colon carcinoma cell lines, LS174T and CW-2. In these experiments, we obtained similar results with COLO320 cells. In the static cell surface binding assay and the flow adhesion assay, we found that both LS174T cells and CW-2 cells could bind to P-selectin, and heparin and the four modified heparins (RO-heparin, CR-heparin, 2/3ODS-heparin, and N/2/3DS-heparin) could effectively in- hibit LS174T cells or CW-2 cells binding to P-selectin, adhesion to CHO-P cells and adhesion to immobilized platelets under static or flow conditions (data not shown). The 6-O-sulfate group of glucosamine units in heparin is crucial for the inhibition. Mean-while, we found that LS174T cells and CW-2 cells strongly expressed heparan sulfate proteoglycans on their cell surface, and the treatment of these tumor cells with three heparinases can also reduce the binding of P-selectin to these tumor cells (data not FIG. 5. Inhibition of COLO320 cell adhesion to surface-adhered platelets by heparin and modified heparins under shear flow conditions. A, effect of wall shear stress on COLO320 cell adhesion to surface-adhered platelets. After washing the platelet layer with PBS/0.1% BSA for ϳ2 min, tumor cells (10 6 /ml) were perfused over the platelet layer for 3 min at different shear stress. B, inhibition of COLO320 cell adhesion to surface-adhered platelets by heparin and modified heparins at 0.3 dyn/cm 2 . C, inhibition of COLO320 cell adhesion to surface-adhered platelets by heparin and modified heparins at 0.6 dyn/cm 2 . Negative control was designated as Ϫ, positive control was designated as ϩ. For mAb inhibition experiments, immobilized platelets were preincubated with 9E1 or P1. Adhesion of tumor cells to surface-adhered platelets was measured by videomicroscopy. All values were expressed as the mean number (Ϯ S.D.) of adherent tumor cells in 10 -20 fields of view using a ϫ10 objective lens. Controls were run before and after each assay. All results are representative of 3-6 separate experiments. **, p Ͻ 0.01 with respect to positive control. #, values are percentage of inhibition of tumor cell adhesion to surface-adhered platelets preincubated to indicated heparin and modified heparins compared with the adhesion to surface-adhered platelets with no heparin and modified heparins.
shown). Taken together, these data suggested that our findings that modified heparins can inhibit P-selectin-mediated cell adhesion of tumor cells to surface-immobilized platelets under flow conditions were applicable to other human colon carcinoma cell lines, such as LS174T cells and CW-2 cells. DISCUSSION Modified heparins were first applied to the study of inhibition of P-selectin-mediated tumor cell adhesion. In this article, we demonstrated that heparin could significantly inhibit the adhesion of COLO320, a human colon carcinoma cell line, to FIG. 6. Cell surface expression of heparan sulfate proteoglycan and effects of heparinases treatment on P-selectin binding. A, detection of sLe x and heparan sulfate on the tumor cell lines. Tumor cells were incubated with mouse IgM, CSLEX-1 (a mAb to sLe x) , or 10E4 (a mAb to heparan sulfate proteoglycan) followed by an FITC-conjugated Ab against mouse IgM. B, detection of PSGL-1 on the tumor cell lines. Tumor cells were incubated with mouse IgG or KPL1 (a mAb to PSGL-1) followed by an FITC-conjugated Ab against mouse IgG. C, effects of heparinases on P-selectin binding. After digestion with heparinases I, II, and III, COLO320 cells were incubated with human IgG, P-Fc, mouse IgM, or 10E4 followed by an FITC-conjugated Ab against human IgG or mouse IgM. The binding events were analyzed by flow cytometry. Results are presented as histograms of the log fluorescence intensities from 10 4 cells. The proportion of positive cells is indicated in each histogram. the surface-bound platelets under flow conditions, that the adhesion primarily depended on P-selectin, and that the 6-Osulfate group of the glucosamine unit in heparin is critical for inhibition of P-selectin-mediated tumor cell adhesion. Additionally, four modified heparins with high anti-adhesion activity but low anticoagulant activity were identified, including RO-heparin, CR-heparin, 2/3ODS-heparin, and N/2/3DS-heparin. Among the four modified heparins, 2/3ODS-heparin was best because of its appropriate anti-adhesion activity and its convenient preparation. Further, we found similar results with other human colon carcinoma cell lines, LS174T and CW-2, suggesting that the molecular mechanism of P-selectin-mediated human colon carcinoma cell adhesion may be widespread. These findings suggest that chemically modified heparins may have potential value as therapeutic agents to block P-selectinmediated cell adhesion and prevent tumor metastasis.
Previous work has suggested that one of the mechanisms that heparin uses to inhibit metastasis is that heparin may block the P-selectin-based interaction of platelets with tumor cells (22,32). In our experiments, on the basis of the static cell surface binding of tumor cells with P-selectin and the cell adhesion of tumor cells on CHO-P cells, we made a further observation of the inhibiting effect of various modified heparins on the adhesion of tumor cells on platelets under flow conditions. We found that the four modified heparins, RO-heparin, CR-heparin, 2/3ODS-heparin, and N/2/3DS-heparin, showed excellent ability to block P-selectin-mediated tumor cell adhesion. The elimination of the 6-O-sulfate group of the glucosamine unit in heparin reduced the inhibiting activity dramatically, indicating that the 6-O-sulfate group of the glucosamine unit in heparin is crucial to the inhibiting activity. The NMR structure analysis proves sufficiently that the four modified heparins, RO-heparin, CR-heparin, 2/3ODS-heparin, and N/2/ 3DS-heparin, had different structures, as anticipated. According to recent literature (20), the anti-inflammatory effects of heparin require glucosamine 6-O-sulfation, and the effects are mediated by blockade of P-and L-selectin. A reasonable hypothesis is that this mechanism should be suitable to heparin's other activities mediated by blockade of P-and L-selectins. Our experiments proved this deduction; that is, the glucosamine 6-O-sulfate group is needed in the inhibition of P-selectin-based tumor cell adhesion.
We also attempted to identify the P-selectin ligand on the COLO320 cell surface using specific mAbs. Prior work indicated that PSGL-1, expressed on the neutrophil surface, is the primary ligand for P-selectin on platelet under flow conditions (33). Using flow cytometry, we found that PSGL-1 was not expressed on the COLO320 cell surface. This result is in accordance with previously published data demonstrating that a variety of tumor cell lines bind to P-selectin through novel ligands that are structurally and functionally distinct from PSGL-1 (13,14,19,34). Additionally, we found that sLe x , the minimal ligand for P-selectin, may be marginally expressed on the COLO320 cell surface. Instead, the COLO320 cells strongly expressed heparan sulfate proteoglycans on their cell surface, and heparinase treatment appeared to reduce the adhesion of P-selectin to COLO320 cells. Further, we found similar results for human colon carcinoma cell lines LS174T cells and CW-2 cells. These findings suggest that the P-selectin ligand on these tumor cell surfaces possess heparan sulfate-like proteoglycan structure and that this molecular mechanism may exist broadly in other tumor cell lines that interact with P-selectin. It has been proved that P-selectin could bind to sialylated, fucosylated, or, in some cases, sulfated glycans on glycoproteins, glycolipids, or proteoglycans, indicating that the type of P-selectin ligand is very complicated. Our experiments have demonstrated that P-selectin ligand on the surface of COLO320 cells have the structure of heparan sulfate-like proteoglycans. But other studies showed that a mucin-type glycoprotein is responsible for COLO320 cell adhesin to P-selectin (11). It seems that the P-selectin ligand on the surface of tumor cell is not consistent. We conjecture that in the process of carcinoma progression, increased expression and altered glycosylation of mucins on the tumor cell surface may provide several types of ligands for the interaction of tumor cells with P-selectin. Another possibility is that the P-selectin ligand on the surface of tumor cells has both sialylated and heparan sulfate-like proteoglycan structure. Recently, Koenig et al. (35) suggested that the heparan sulfate and sialylated ligands might be both recognized in the regions close to the lectin-binding site. How can two such disparate carbohydrate structures as mucins and heparan sulfate chains be recognized by the same or similar binding sites? One possibility that Koenig et al. suggested is that the recognition involves not a linear defined oligosaccharide sequence but a "clustered saccharide patch" that can be generated in multiple different ways. Thus, a clustered patch with the appropriate arrangement of carboxyl groups, sulfate esters, and hydroxyl groups might be generated either by the closely packed oligosaccharides in the mucins, or by the densely modified heparan sulfate proteoglycans chain. The present data support, but do not prove, this hypothesis.
As we know, heparan sulfate proteoglycans are expressed and secreted by most, if not all, mammalian cells and are strategically located on cell surfaces and in the extracellular matrix (21,36). Assuming some tumor cells use heparan sulfate as P-selectin ligand, then the structure of heparan sulfate on the surface of tumor cells is different from that on the surface of normal cells. According to Jayson et al. (37), heparan sulfate underwent specific structural changes during the progression from human colon adenoma to carcinoma in vitro, and the progression to malignancy was associated with specific structural changes, including a reduction in 2-O-sulfated iduronic acid units and an increase in 6-O-sulfated glucosamine units. This change is exactly the same as the change of heparan sulfate expressed on the surface of the human colon carcinoma CaCo-2 cells in the process of cell differentiation (38). These findings suggest that heparan sulfate on the surface of the normal cells is not the P-selectin ligand. In the process of carcinoma progression, heparan sulfate gets the characteristic of P-selectin ligand because of the change of the heparan sulfate structure on the surface of the malignant cells. In our study, we have shown that the 6-O-sulfate group of glucosamine units in heparin, not the 2-O-sulfate, is crucial to the recognition of P-selectin-mediated tumor cell adhesion. Our results are consistent with the above conclusions about specific structural changes of heparan sulfate during the progression to malignancy.
In the flow adhesion assays testing inhibition of the adhesion of tumor cells to platelets, we found that preincubation of 9E1 (a leukocyte adhesion-blocking mAb to P-selectin) with platelets reduced the percentage of COLO320 cell adhesion to platelets by 50%, implicating that P-selectin on the platelets is not the only molecule mediating the tumor cell adhesion to platelets, and other adhesion molecules may be involved in this process. Platelets have been reported to bind to small cell lung and colon carcinoma via P-selectin (11,12,31,39). Alternatively, platelets may attach to certain melanoma, breast, and colon cancer cells through the integrin receptor ␣ IIb ␤ 3 on the platelet surface (4, 40 -43). In the study of LS174T cells, Mc-Carty et al. (19) proved that platelet P-selectin and integrin ␣ IIb ␤ 3 cooperate to mediate adhesive interactions between platelets and LS174T cells. Platelet P-selectin supports LS174T cell initiating adhesion, subsequent platelet ␣ IIb ␤ 3 involvement contributes to conjugate stabilization by converting LS174T cell to firm arrest. In addition, based on the blockade of heparin to P-selectin on platelets, we showed that heparin could inhibit tumor cell adhesion to platelets under flow conditions. We noticed that heparin and its derivatives, even at lower concentrations (0.1 mg/ml), inhibit the adhesion of tumor cells to platelets better than 9E1. This result indicates that heparin and its derivatives can not only inhibit the P-selectin on platelets, but also influence the adhesion of tumor cells to platelets through other approaches. That is, heparin may block other kinds of adhesion molecules involved in the process, such as integrin ␣ IIb ␤ 3 . Recently, Sobel et al. (44) reported that heparin can bind to ␣ IIb ␤ 3 of platelets either in vitro or in vivo, and ␣ IIb ␤ 3 is the main molecule that heparin binds to on platelets. Therefore, heparin inhibits tumor cell adhesion to platelets not only by blocking P-selectin, but also by a broader mechanism under flow conditions.
Heparin may have potential value as a therapeutic agent to prevent metastasis. As an anticoagulant, however, heparin has the possibility of causing bleeding, which restricts the application of heparin in preventing metastasis. In our study, various types of chemical modification, which are directed against the unique antithrombin-binding pentasaccharide sequence in heparin, effectively reduced its anticoagulant activities. Then, we found that some modified heparins still showed strong anti-adhesion activities. Our findings suggest that some chemical modifications of heparin, especially 2-O, 3-O-desulfation of heparin, may generate a potent, non-anticoagulant, anti-adherent, anti-metastatic therapeutic agent to block unwanted P-selectin-dependent reactions.