Involvement of Highly Sulfated Chondroitin Sulfate in the Metastasis of the Lewis Lung Carcinoma Cells*

The altered expression of cell surface chondroitin sulfate (CS) and dermatan sulfate (DS) in cancer cells has been demonstrated to play a key role in malignant transformation and tumor metastasis. However, the functional highly sulfated structures in CS/DS chains and their involvement in the process have not been well documented. In the present study, a structural analysis of CS/DS from two mouse Lewis lung carcinoma (3LL)-derived cell lines with different metastatic potentials revealed a higher proportion of Δ4,5HexUA-GalNAc(4,6-O-disulfate) generated from E-units (GlcUA-GalNAc(4, 6-O-disulfate)) in highly metastatic LM66-H11 cells than in low metastatic P29 cells, although much less CS/DS is expressed by LM66-H11 than P29 cells. This key finding prompted us to study the role of CS-E-like structures in experimental lung metastasis. The metastasis of LM66-H11 cells to lungs was effectively inhibited by enzymatic removal of tumor cell surface CS or by preadministration of CS-E rich in E-units in a dose-dependent manner. In addition, immunocytochemical analysis showed that LM66-H11 rather than P29 cells expressed more strongly the CS-E epitope, which was specifically recognized by the phage display antibody GD3G7. More importantly, this antibody and a CS-E decasaccharide fraction, the minimal structure recognized by GD3G7, strongly inhibited the metastasis of LM66-H11 cells probably by modifying the proliferative and invading behavior of the metastatic tumor cells. These results suggest that the E-unit-containing epitopes are involved in the metastatic process and a potential target for the diagnosis and treatment of malignant tumors.

charide fraction, the minimal structure recognized by GD3G7, strongly inhibited the metastasis of LM66-H11 cells probably by modifying the proliferative and invading behavior of the metastatic tumor cells. These results suggest that the E-unit-containing epitopes are involved in the metastatic process and a potential target for the diagnosis and treatment of malignant tumors.
The poor prognosis of cancer is mainly due to the metastasis of malignant cells from the primary neoplasm. Metastasis is a selective process involving invasion, embolization, survival in the circulation, arrest in distant capillary beds, and extravasation into and multiplication within the target organ parenchyma (1,2). In the process of metastasis, tumor cells are involved in a series of interactions with surrounding extracellular matrix (ECM) 5 components and nontumor cells such as platelets and endothelial cells. Proteoglycans (PGs), which bear heparan sulfate (HS) or chondroitin sulfate (CS)/dermatan sulfate (DS) side chains and are widely expressed on the cell surface and in the ECM, are important in modulating these interactions (3)(4)(5). It has been well documented that PGs with HS side chains play important roles in metastasis (6 -8). In addition to HS-PGs, CS/DS-PGs have also been implicated in normal biological processes such as neuronal development, morphogenesis, growth factor binding, and cell signaling (9, 10) but also play a crucial role in the metastatic process (4).
Human melanoma CS/DS-PG and its homologue NG2 in rats are involved in matrix adhesion, migration, and invasion, a role that is abolished by treatment with antibodies against CS/DS-PG (11)(12)(13). Moreover, CD44-related CS/DS-PG on the cell surface is essential in the invasion/migration of normal endothelial and melanoma cells (14,15). The interaction between CS/DS-PG and membrane type 3 matrix metalloproteinase is important for the invasion of melanoma cells (16). In addition, treatment with chondroitinase (CSase) AC (specifically cleaving CS) or CSase B (specifically cleaving DS) signifi-cantly inhibited the proliferation and invasion of melanoma cells (17). Most recently, cell surface CS/DS was shown to participate in basic fibroblast growth factor-induced proliferation of human metastatic melanoma cell lines (18), the membrane type 3 matrix metalloproteinase-mediated activation of promatrix metalloproteinase-2 (19), and P-selectin-mediated metastasis of breast cancer cell lines (20). These studies point to the fact that CS/DS side chains play crucial roles through binding to various ligands, although the core protein also has ligand binding capabilities (21).
CS chains are composed of repeating disaccharide units of GlcUA-GalNAc, where GlcUA and GalNAc represent D-glucuronic acid and N-acetyl-D-galactosamine, respectively, whereas DS is a stereoisomer of CS chains formed from precursor CS chains through the action of glucuronyl C5 epimerase (22,23). CS and DS chains are often found as co-polymeric structures (CS/DS) that tend to exhibit a periodic distribution of GlcUAcontaining disaccharide repeats and L-iduronic acid-containing disaccharide repeats in a cell/tissue-specific manner (9,24). CS/DS chains are further modified by the differential sulfation pattern of specific sulfotransferases at C-2 of GlcUA/L-iduronic acid and/or C-4 and/or C-6 of GalNAc to yield enormous structural diversity (25). Hence, functional structures of CS/DS chains may be generated in tumor cells because of the differential expression of the individual modification enzymes during tumor progression, which may have a direct/indirect link with metastatic potential. Thus, identification of these altered functional structures would be a significant step in understanding the mechanism of involvement of CS/DS-PGs in metastasis and enable us to address the diagnosis and treatment of malignant tumors.
Highly sulfated disaccharide units like E-unit, GlcUA␤1-3GalNAc(4S,6S) (26), where 4S and 6S stand for 4-O-and 6-Osulfate, respectively, are rare, and E-unit-rich CS preparations show remarkable biological activities like the promotion of neurite outgrowth and high affinity binding to growth factors (9,10,27). Increasing evidence suggests that the expression pattern of cell surface CS/DS-PGs is related to metastatic potential (3,12,13). However, the involvement of highly sulfated structures of CS/DS chains in the metastatic process remains obscure. In this study, E-unit-containing epitopes specifically recognized by the antibody GD3G7 (28,29) were identified as crucial for the highly metastatic potential of the Lewis lung carcinoma-derived LM66-H11 clone.

EXPERIMENTAL PROCEDURES
Materials-CSase ABC (EC 4.2.2.4), highly purified CSase ABC (protease-free CSase ABC), standard unsaturated disaccharides, CS-A from whale cartilage, CS-B from porcine skin, CS-C and CS-D from shark cartilage, and CS-E from squid cartilage were purchased from Seikagaku Corp. (Tokyo, Japan). The single chain antibody GD3G7 was selected for reactivity with rat embryo-derived GAGs by the phage display technique (28). The monoclonal anti-vesicular stomatitis virus tag antibody P5D4 and porcine intestinal mucosal heparin were from Sigma. Alexa Fluor 488-conjugated goat anti-mouse IgG (HϩL) was obtained from Invitrogen. Actinase E was from Kaken Pharmaceutical Co. (Tokyo, Japan). The Diff-Quick solution was purchased from International Reagent Corp. (Kobe, Japan). 2-Aminobenzamide (2AB) was purchased from Nacalai Tesque (Kyoto, Japan). Sodium cyanoborohydride (NaBH 3 CN) was from Aldrich. All other chemicals and reagents were of the highest quality available. Size-defined even-numbered CS-E oligosaccharides were prepared by enzymatic fragmentation of a commercial squid cartilage CS-E with sheep testicular hyaluronidase (Sigma), followed by fractionation using gel filtration column chromatography on Bio-Gel P-10 as described previously (30). For RNA extraction, a QuickPrep TM total RNA extraction kit was purchased from GE Healthcare. RNA-qualified RNase-free DNase, RNasin ribonuclease inhibitor, and Moloney murine leukemia virus reverse transcriptase were obtained from Promega (Madison, WI). A Platinum SYBR Green qPCR Supermix-UDG kit was purchased from Invitrogen.
Animals and Cell Lines-Six-week-old male C57BL/6 mice and 9-week-old female C3H/HeN mice were obtained from Japan SLC (Hamamatsu, Japan) and kept in standard housing. All of the experiments were performed under the experimental protocol approved by the local animal care committee of Hokkaido University. The low metastatic P29 and high metastatic LM66-H11 cells cloned from a murine Lewis lung carcinoma 3LL were prepared as reported (31) and cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (v/v) fetal bovine serum (Thermo Trace, Melbourne, Australia), streptomycin (100 g/ml), and penicillin (100 units/ml) at 37°C in a humidified 5% CO 2 atmosphere. The cells were harvested after incubation with 2 mM EDTA in phosphate-buffered saline (EDTA/PBS) for 10 min at 37°C by gentle flushing with a pipette and subcultured twice a week.
Extraction of GAGs from P29 and LM66-H11 Cells-The cells were dehydrated and delipidated by extraction with acetone, air-dried, and used for extraction of GAGs essentially as described previously (32) with some modifications. Briefly, the dried acetone powder was digested with heat-activated (60°C, 30 min) actinase E in 200 l of 0.1 M sodium borate, pH 8.0, containing 10 mM calcium acetate at 60°C for 48 h. Following incubation, each sample was treated with 5% trichloroacetic acid, and the resultant precipitate was removed by centrifugation. The supernatant was extracted with ether to remove trichloroacetic acid. After neutralization with 1.0 M sodium carbonate, the aqueous phase was adjusted to contain 80% ethanol and 1% sodium acetate and kept at 4°C overnight. The precipitated crude GAGs were recovered by centrifugation, desalted on a PD-10 column (GE Healthcare) using 50 mM pyridine acetate, pH 5.0, as an eluent, and evaporated dry.
Disaccharide Composition Analysis of CS Chains-The disaccharide composition of the GAG preparations from both Lewis lung carcinoma cell lines was determined as previously described (32). Briefly, the sample was dissolved in water, and an aliquot was digested with CSase ABC (33), the digest was labeled with 2AB (34), and excess 2AB was removed by extraction with chloroform (35). The 2AB-labeled digest was analyzed by anion exchange HPLC on a PA-03 silica column (YMC-Pack PA, Kyoto, Japan). Identification and quantification of the resulting disaccharides were achieved by compari-son with the elution positions of CS-derived authentic unsaturated disaccharides (Ref. 34 and Table 1).
Immunocytochemistry-E-unit-containing epitopes on the cell surface of P29 and LM66-H11 cells were stained using a single chain phage display antibody, GD3G7 (28,29). Briefly, the carcinoma cells (5 ϫ 10 4 cells/well) were plated on 8-well Lab-Tech chamber slides (Nalge Nunc International), cultured in DMEM supplemented with 10% fetal bovine serum for 24 h, and fixed with the Diff-Quick reagent A. After being blocked with PBS containing 3% bovine serum albumin for 1 h at room temperature, the fixed cells were incubated with 100 l of the primary antibody GD3G7 (diluted 1:100 (10 g/ml) in 0.1% bovine serum albumin/PBS) for 1 h at room temperature and washed with PBS containing 0.5% Tween 20 three times. In a control experiment, GD3G7 was preincubated with CS-E (5 or 10 g) at 4°C for 30 min to demonstrate the specificity of GD3G7 toward CS-E. After being incubated with anti-vesicular stomatitis virus antibody (diluted 1:5,000 in 0.1% bovine serum albumin/PBS) for 1 h at room temperature, the cells were washed with PBS containing 0.5% Tween 20. To detect the antivesicular stomatitis virus antibodies, the cells were stained with the Alexa Fluor 488-conjugated third antibody (diluted 1:500 in 0.1% bovine serum albumin/PBS) and visualized with a laserscanning confocal microscope, FLUOVIEW (Olympus, Tokyo, Japan). To evaluate the specificity of the antibody, the cells were treated with CSase ABC in the presence of a protease inhibitor mixture (Nacalai Tesque, Inc., Kyoto, Japan) and then processed for immunostaining as described above.
Assays for Lung Metastasis-To prepare LM66-H11 cells for injection, the cells were harvested by brief exposure to EDTA/ PBS, and the cell viability in single-cell suspensions was determined by trypan blue exclusion. A total of 4 ϫ 10 5 cells (Ͼ90% viability) suspended in 200 l of DMEM were injected into a lateral tail vein of C57BL/6 or C3H/HeN mice. Although the cell lines LM66-H11 and P29 were established from the C57BL/6 mouse, the C3H/HeN mouse strain was also used for metastasis experiments because of the better handling advantage, and the results obtained were similar to those obtained with a syngeneic C57BL/6 mouse. After 3 weeks, the animals were sacrificed, and the number of visible tumor cell parietal nodules in the lung was counted by two observers in a blinded fashion. The inoculation procedure described above was used for the subsequent experiments.
To investigate the involvement of cell surface CS/DS in metastasis, LM66-H11 cells were pretreated with either DMEM (control) or DMEM containing protease-free CSase ABC (20 mIU/ml) for 30 min at 37°C before being injected into C3H/HeN mice. In a separate series of experiments, C57BL/6 or C3H/HeN mice received the indicated amount (100 g/mouse) of commercial CS/DS preparations (CS-A, CS-B, CS-C, CS-D, and CS-E) or heparin 30 min before the tumor cell injection. CS-E digested with CSase ABC was also used as a control. For dose-dependent inhibition experiments and for inhibition experiments using size-defined CS-E oligosaccharides, various doses of CS-E (10, 50, 100, 200, and 300 g/mouse), and hexa-, octa-, deca-, or dodecasaccharides (25 g/mouse) were administered into C3H/HeN mice 30 min before the tumor cell injection. In other instances, to investigate the involvement of the antibody GD3G7 epitope in metastasis, LM66-H11 cells were preincubated with 200 l of serially diluted GD3G7 (0.5, 0.25, and 0.125 g/ml) or the irrelevant antibody MPB49V (0.5 g/ml) for 30 min at 37°C and used for metastasis experiments in both C57BL/6 and C3H/HeN mice. Aliquots of the cell suspension were assessed for cell viability before the injection.
Dynamic Monitoring of LM66-H11 Cell Proliferation Using the RT-CES TM System-The cell proliferation assay was done using the RT-CES TM system (ACEA Biosciences, San Diego, CA). Briefly, ACEA 96X microtiter plates (e-plate TM ) were coated with laminin (25 g/ml) at 37°C for 30 min, and LM66-H11 cells (2 ϫ 10 4 /well) were seeded in 100 l of medium. In the first experiment, cell attachment was monitored up to 20 min, at which point various inhibitors, CS-A (100 g/ml), CS-E (100 g/ml), or GD3G7 (2 g/ml) in 150 l of DMEM were individually added. The spreading of the cells was continuously monitored up to 60 min using the RT-CES TM system. In other instances, ϳ24 h after seeding, when the cells were in a log growth phase, serially diluted GD3G7 (2, 0.5, and 0.2 g/ml) in 150 l of DMEM was added into the corresponding wells. The proliferation was monitored for a period of 64 h and expressed as a cell index as per the manufacturer's instructions. The cell index is a quantitative measure of the spreading and/or proliferative status of the cells in an electrode-containing well.
Cell Migration and Invasion Assays in Vitro-The ability of LM66-H11 and P29 cells to migrate and invade was assessed using the BD BioCoat TM chamber with or without Matrigel (BD Biosciences) in vitro. In some instances, single cell suspensions of LM66-H11 (5 ϫ 10 4 cells/ml) were prepared by detaching and resuspending in serum-free DMEM. Before being added to the upper chamber (8 m PET pores), LM66-H11cells were preincubated with CS-A (100 g/ml), CS-E (100 g/ml), or GD3G7 (2 g/ml) in 500 l DMEM for 30 min at 37°C in a CO 2 incubator. The lower chambers were filled with DMEM containing 10% fetal bovine serum. After incubation for 24 h, the cells that had migrated and invaded through the membrane alone or the Matrigel-coated membrane remained bound to the underside of the membranes. These cells were stained

composition of CS/DS chains in Lewis lung carcinoma (3LL)-derived low metastatic P29 and high-metastatic LM66-H11 cells
The GAG preparation from each carcinoma cell line was digested with chondroitinase ABC and analyzed by anion exchange HPLC after labeling with a fluorophore 2AB as detailed under "Experimental Procedures." with the Diff-Quik staining kit and counted in five random microscopic fields/filter.

3
H-Labeled CS-E (4 ϫ 10 5 cpm) was prepared as described previously (27) and injected into C3H/HeN mice through a lateral tail vein. At 1, 1.5, and 2.5 h post-injection, the blood samples were collected from anesthetized animals via cardiac puncture. The animals were then sacrificed, and the organs including liver, lungs, kidneys, spleen, and brain were dissected out, quickly rinsed, and weighed. The blood and each tissue from all the animals were pooled and homogenized with 1% Triton X-100/PBS in a Polytron homogenizer while being cooled on ice. Aliquots of the tissue homogenates and the plasma were freeze-thawed (five times) using liquid nitrogen bath and treated with 5% trichloroacetic acid. The resultant supernatants were recovered by centrifugation at 15,000 rpm for 10 min, and the radioactivity was measured using a multi-purpose liquid scintillation counter (Beckman coulter LS6500). The counts were converted to disintegrations on the basis of a standard quench correction curve, and the distribution of 3 H-CS-E in the plasma and tissues was expressed as dpm/ml and dpm/g, respectively.
Relative Quantification of Gene Expression of Sulfotransferases and Epimerase-Total RNA was extracted from P29 and LM66-H11 cells in 100-mm culture plates using a QuickPrep total RNA extraction kit according to the manufacturer's instructions, and then each extract was treated with RNase-free DNase for 30 min at 37°C. The cDNA was synthesized from ϳ1 g of the total RNA using Moloney murine leukemia virus reverse transcriptase and an oligo(dT) 16 primer (Hokkaido System Science, Sapporo, Japan) and then purified with a MinElute PCR purification kit (Qiagen). Quantitative real time reverse transcription-PCR was performed using a Platinum SYBR Green qPCR Supermix-UDG kit in the iCyclear iQ TM (Bio-Rad). Briefly, the reaction mixture (25 l) contained a SYBR Green Supermix, each primer set at 0.2 M (as shown in Table 2), and each template cDNA. PCR was carried out for 40 cycles at 95°C for 15 s and 60°C for 1 min, and the amplified products were measured by the iCycler iQ real time PCR ana-lyzing system (Bio-Rad). The expression level of each sulfotransferase and epimerase mRNA was normalized to that of the glyceraldehyde-3-phosphate dehydrogenase transcript.

Comparison of Amount and Composition of CS/DS between P29 and LM66-H11 Cells-
To investigate the possible alteration in the expression of CS/DS in the carcinoma clones with different metastatic potentials, GAGs were extracted from P29 and LM66-H11 cells, as described under "Experimental Procedures," and the amount and composition of CS/DS in both GAG preparations were determined by digestion with CSase ABC followed by anion exchange HPLC. The results are detailed in Table 1. The amount of CS/DS expressed in the low metastatic P29 cells was 10 times that in the LM66-H11 cells with a high metastatic potential (Table 1) (Table 1). In addition, the proportion of ⌬ 4,5 HexUA␣1-3GalNAc(4S) was significantly decreased from 94.5% in CS/DS from P29 cells to 65.6% in CS/DS from LM66-H11 cells. However, it is interesting to note that both the amount and proportion of ⌬ 4,5 HexUA␣1-3GalNAc(4S,6S) were substantially increased in highly metastatic LM66-H11 cells.
The unique highly sulfated structure GlcUA␤1-3GalNAc(4S,6S) in CS/DS chains has been shown to be very important for the interactions of the chains with various functional proteins (27,35). The increasing proportion of ⌬ 4,5 HexUA␣1-3GalNAc(4S,6S) units in the CS/DS preparation from LM66-H11 cells may be a key factor in metastasis. To confirm the high expression of E-units on the surface of LM66-H11 cells, a phage display antibody, GD3G7, which specifically recognizes E-unit-containing CS/DS (28,29), was used for immunostaining both the cell lines. The results showed that LM66-H11 cells were more strongly stained by GD3G7 than P29 cells (Fig. 1, A and C), and the staining was completely abolished by pretreatment with CSase ABC (Fig. 1, B and D). Preincubation of GD3G7 with CS-E significantly suppressed the staining of LM66-H11 cells by GD3G7, supporting the specificity of the antibody (supplemental Fig. S1). Taken together, these results support the high expression of E-units on the surface of LM66-H11 cells.
Involvement of Cell Surface CS/DS in the Metastasis of LM66-H11 Cells-To examine the functions of CS/DS in metastasis, LM66-H11 cells were treated with protease-free CSase ABC to remove the cell surface CS/DS before being injected into mice. After 3 weeks, the mice were sacrificed, and pulmonary metastasis was analyzed by counting tumor foci on the lung surface. The treatment with CSase ABC significantly reduced the metastasis of LM66-H11 cells compared with a control treated with DMEM (Fig. 2), suggesting an important role for cell sur- face CS/DS in the metastatic process. The involvement of the cell surface CS/DS in the metastasis prompted us to explore the possibility that the administration of CS/DS preparations inhibits the metastatic process, as in the case of heparin/HS in the previous work by us (36) and others (37).
Characterization of Anti-metastatic Activity of CS-E-To investigate this possibility, various commercial CS/DS preparations (100 g/mouse) were individually preinjected into the mice 30 min before the injection of LM66-H11 cells. All of the CS/DS preparations tested showed anti-metastatic activity to some degree, except CS-C from shark cartilage (Fig. 3). Remarkably, CS-E from squid cartilage, characterized by a high proportion (62%) of E-units (30), was not only the strongest inhibitor among the CS/DS preparations but also more potent than heparin, well known for its anti-metastatic activity (37), suggesting the importance of E-units in the metastasis. In the next experiments, dose-dependent inhibition was observed for CS-E against the LM66-H11 cell metastasis in the low dose range (10 -100 g/mouse), whereas the inhibition gradually decreased at higher doses (Ͼ100 g/mouse) (Fig. 4A). The seemingly conflicting results at low and high doses may be due to interactions with various proteins at different concentrations and remain to be investigated. The anti-metastatic activity of CS-E was almost completely abolished by digestion with CSase ABC (Fig. 4B), suggesting that the inhibitory activity of the CS-E preparation is due to CS-E itself rather than impurities.
Anti-metastatic Activity of Antibody GD3G7-Higher expression of E-units on the surface of LM66-H11 cells and the potent anti-metastatic activity of CS-E led us to hypothesize that the E-unit-containing CS/DS chains on the tumor cell surface may be involved in the metastatic process. To investigate this assumption, the antibody GD3G7, which recognizes E-unit-containing CS/DS chains, was used for anti-metastasis assays. The preincubation of LM66-H11 cells with GD3G7 (0.5, 0.25, and 0.125 g/ml) for 30 min of preinjection strongly reduced their lung metastasis in a dose-dependent manner, in contrast to preincubation with the irrelevant antibody MPB49V (negative control single-chain antibody) (Fig. 5), suggesting that the epitopes of the antigen for the antibody GD3G7 play a key role in the metastasis of LM66-H11 cells to the lung.
The Minimal Anti-metastatic Structure of CS-E Recognized by GD3G7-The strong anti-metastatic activity of the antibody GD3G7 suggests that the epitope sequences of antigen act as functional domains in metastasis. Most recently, we found that a CS-E decasaccharide fraction was the critical minimal structure needed for recognition by GD3G7, and at least four structurally defined decasaccharides from this fraction were identified as epitopic sequences (29). In the present study, we aimed FIGURE 1. Immunocytological detection of the GD3G7 epitope on the cell surface of Lewis lung carcinoma (3LL)-derived P29 and LM66-H11 clones. LM66-H11 (A) and P29 (C) cells were seeded separately on chamber slides, cultured for 24 h, and fixed with Diff-Quick reagent A. They were incubated with the antibody GD3G7 for 1 h, and bound GD3G7 was detected with an anti-vesicular stomatitis virus glycoprotein antibody followed by an Alexa-conjugated third antibody and visualized by confocal microscopy. In control experiments, LM66-H11 cells (B) and P29 cells (D) were treated with CSase ABC in the presence of a protease inhibitor mixture and then processed for immunostaining as described above. Scale bar, 100 m. to determine the minimal size of the functional domains of CS/DS chains involved in the metastasis. The decasaccharide fraction and three other size-defined oligosaccharide fractions, all of which were prepared by partial digestion of CS-E with sheep testicular hyaluronidase, were used for inhibition assays. The results showed that the deca-and dodecasaccharides were substantial inhibitors of the metastasis of LM66-H11 cells compared with the hexa-and octasaccharides (Fig. 6), suggesting that E-unit-containing decasaccharides recognized by the antibody GD3G7 act as minimal sized functional domains of CS/DS chains in the metastatic process.
Effects of CS-E and GD3G7 on the Proliferation, Migration and Invasion of LM66-H11 Cells-Although our findings clearly show CS-E and GD3G7 to have anti-metastatic properties, the mechanism involved is unknown. We further extended our study to include cell proliferation. Inhibitory effects of CS-A (100 g/ml), CS-E (100 g/ml), MPB49V, and GD3G7 (2 g/ml) on the proliferation of LM66-H11 cells on laminincoated plates were tested using the RT-CES TM system. As shown in Fig. 7A, CS-E and GD3G7 strongly suppressed the spreading, whereas CS-A did not. No cytotoxicity of CS-E and GD3G7 was observed, as verified by staining for viability using trypan blue. Therefore, in subsequent experiments, LM66-H11 cells were treated with various dilutions of GD3G7 (2, 0.5, and 0.2 g/ml), which showed that the 2 g/ml was the most effective for the inhibition of the proliferation of LM66-H11 cells (Fig. 7B). The migration and invasion studies clearly showed that both the migration and invasion of highly metastatic LM66-H11 cells were greater than those of low metastatic P29 cells (Fig. 8A). The effects of CS-A (100 g/ml), CS-E (100 g/ml), or GD3G7 (2 g/ml) on the migration and invasion of LM66-H11 cells were examined next. Although the migration was not significantly diminished by any of the inhibitors tested when compared with the control (Fig. 8B), the invasion of LM66-H11 cells through Matrigel was significantly suppressed by CS-E and GD3G7 as compared with the control (Fig. 8C), suggesting that CS-E and GD3G7 interfere with the metastasis via anti-proliferative and anti-invasive means.
Efficient Accumulation of CS-E in the Lungs-Cell/organspecific accumulation of anti-cancer drugs is essential for the success of drug targeting in vivo. To examine the biodistribution of CS-E, radiolabeled CS-E ( 3 H-CS-E) was administered intravenously, and the distribution of the radioactivity was

. Dose-dependent effects of CS-E and effects of digestion with CSase ABC on the anti-metastatic activity of CS-E.
A, the anti-metastatic activity of various doses (10,50,100,200, and 300 g/mouse) of CS-E was investigated as described in the legend to Fig. 3. B, to confirm the anti-metastatic activity, CS-E was digested with CSase ABC. After heat-inactivating the enzyme, the digest (100 g/mouse), undigested CS-E (100 g/mouse as positive control), and DMEM (200 l/mouse) were used for the analysis, as described above. Six C3H/HeN mice were used per group. The data represent the mean values Ϯ S.D. for two independent experiments. *, p Ͻ 0.05; **, p Ͻ 0.01; and ***, p Ͻ 0.001 versus control, Mann-Whitney U test. DECEMBER 5, 2008 • VOLUME 283 • NUMBER 49 determined after 1, 1.5, and 2.5 h. As shown in Fig. 9, 3 H-CS-E accumulated preferentially in the lungs, where LM66-H11 cells form tumor nodules. 3 H-CS-E accumulated quickly in the lungs and may explain the lower levels of this compound in the blood and other organs. We also noted an increase in the level of 3 H-CS-E in the kidneys, which could be another indication of the rapid clearance.

Expression of the N-Acetyl-D-galactosamine-4-O-sulfate 6-O-Sulfotransferase Gene in Highly Metastatic LM66-H11 Cells-
Quantitative real time reverse transcription-PCR revealed that the expression level of the N-acetyl-D-galactosamine-4-O-sulfate 6-O-sulfotransferase gene, which encodes the enzyme essential for the biosynthesis of E-units, was 5.5-fold higher in LM66-H11 cells than in P29 cells (Fig. 10). This observation was consistent with the higher proportion of E-units in the disaccharide analysis of CS chains (Table 1). In contrast, the expression level of the chondroitin 4-O-sulfotransferase-1 gene, which is involved in the biosynthesis of GlcUA␤1-3GalNAc(4S) units, a precursor for an E-unit (10), was lower  . Anti-metastatic activity of size-defined oligosaccharides of CS-E. A series of size-defined CS-E oligosaccharides (hexa-, octa-, deca-, and dodecasaccharides, 5 g/mouse), which were produced by digestion with sheep testicular hyaluronidase, were used for the anti-metastasis assay, respectively, as described in the legend to Fig. 3. Six C3H/HeN mice were used per group. The data represent the mean values Ϯ S.D. for two independent experiments. **, p Ͻ 0.01 versus control, Mann-Whitney U test.

FIGURE 7. Inhibitory effects of CS-E and the anti-CS-E phage display antibody GD3G7 on the proliferation of highly metastatic LM66-H11 cells.
A, effects of CS-A, CS-E, MPB49V, and GD3G7 (4 g) on the proliferation of highly metastatic H11 cells. LM66-H11 cells (f) were seeded on laminin-coated plates at a density of 2 ϫ 10 4 /well, and the effects of various inhibitors, CS-A (100 g/ml, ϫ), CS-E (100 g/ml, OE), GD3G7 (2 g/ml, E), and MPB49V (2 g/ml, ࡗ) were observed for 60 min as described under "Experimental Procedures." The arrow indicates the time of the addition of inhibitors. The cell index is a quantitative measure of the spreading and/or proliferative status of the cells in an electrode-containing well. B, dose-dependent effects of GD3G7 on the proliferation of H11 cells. H11 cells (ࡗ) were seeded as described for A and continuously monitored up to 24 h, at which point serially diluted GD3G7 (2 g/ml, Ⅺ; 0.5 g/ml, ‚; 0.2 g/ml, ϫ) and negative control MPB49V (2 g/ml, E) were added into the corresponding wells (indicated by arrow). The cells were monitored for 64 h and expressed as cell index. The data represent the mean values Ϯ S.D. for triplicate wells.

DISCUSSION
In this study, we demonstrated that administered CS-E accumulated in the mouse lungs, where LM66-H11 cells highly expressing E-units colonized to form tumor nodules. This accumulation of CS-E in the lungs of mice may be important for the inhibition of metastasis, particularly the proliferation and invasion of tumor cells. The therapeutic index of a drug would be increased if it enables better transport to the target cell/ organ. We also showed that a single administration of CS-E or pretreatment of LM66-H11 cells with the phage display antibody GD3G7, which specifically recognizes E-unit-containing epitopes, suppressed the metastasis.
During tumor progression, the cell surface expression of PGs changes as shown at genetic and structural levels and is thought to play a Janus-faced role (3). PGs and GAGs are often overexpressed in tumor stroma and tumor fibrotic tissues compared with surrounding normal tissues (3). However, accumulating   DECEMBER 5, 2008 • VOLUME 283 • NUMBER 49 evidence suggests that cell surface PGs inhibit metastasis by promoting tight cell-cell and cell-extracellular matrix adhesion (6). Low levels of cell surface HS correlate with high metastatic activity of many tumors (38 -41), although there are some exceptions (42). Increasing evidence shows that the expression of some HS-PGs such as syndecan-1 on various tumor cells decreases with increasing metastatic potential (43)(44)(45)(46)(47)(48). Our recent studies revealed that the expression of syndecan-2 in highly metastatic LM66-H11 cells was significantly lower than that in low metastatic P29 cells from the same parental line, leading to a failure of LM66-H11 cells to form stress fibers on the fibronectin substratum (49).

Chondroitin Sulfate in Metastasis
In the present study, LM66-H11 cells were shown to have a much smaller amount of CS/DS on the cell surface than low metastatic P29 cells, suggesting the concomitant expression of certain particular CS/DS-PGs, which may be involved in cell adhesion and facilitate metastasis, together with decreased syndecan-2. Therefore, identification of specific structures of cell surface CS/DS-PGs is important for understanding details of the malignant progression of Lewis lung carcinoma in further studies. Other than a decrease in the overall amount of CS/DS, the disaccharide composition of CS/DS of LM66-H11 cells was significantly different from that of P29 cells (Table 1). ⌬ 4,5 HexUA␣1-3GalNAc(4S) was the major disaccharide in both cases, but its relative proportion was markedly decreased in the CS/DS preparation from LM66-H11 cells, and correspondingly the proportion of nonsulfated disaccharide was dramatically increased up to 27.4% from 3.7% in P29 cells, resulting in undersulfation (S/unit: 0.79) of CS/DS in the highly metastatic LM66-H11 cells. Similarly, the disaccharide composition of perlecan HS has been shown to be altered specifically during the transition to malignancy (8), especially from colon adenoma to carcinoma, and in transformed mammary epithelial cells (50,51), and undersulfation of HS is another substantial change that facilitates metastasis by weakening tumor cell adhesion to extracellular matrix molecules (52,53). Taken together, the lower content and sulfation of CS/DS appear to endow LM66-H11 cells with a greater capacity to escape from the primary tumor surrounded by various extracellular matrixes. Moreover, the changes may mimic the CS/DS level on the normal cell surface to avoid monitoring by the immune system.
An intriguing finding of this study was a substantial increase in the amount of ⌬ 4,5 HexUA␣1-3GalNAc(4S,6S) in CS/DS from LM66-H11 cells compared with that from P29 cells. CS/DS chains rich in GlcUA␤1-3GalNAc(4S,6S) (E-unit), a parental structure of ⌬ 4,5 HexUA␣1-3GalNAc(4S,6S), have been implicated in various biological events, such as the development of the central nervous system (10) and viral attachment (54 -56). In particular, E-units are present in the CS/DS moiety of versican, a key component for the specific interaction with Land P-selectin, as well as various chemokines (35). Most recently, CS chains on metastatic breast cancer cell lines have been demonstrated to be a major P-selectin ligand involved in the prometastatic heterotypic adhesion to platelets and endothelium (20). Therefore, we speculate that up-regulated expression of E-units further endows LM66-H11 cells with metastatic potential through the specific interaction of the E-unit-containing CS/DS chains with adhesion molecules such as selectins expressed on the surfaces of platelets and the vascular endothelium. This speculation was supported by our observations that experimental lung metastasis of LM66-H11 cells was strongly inhibited by the pretreatment of cells with CSase ABC or the preadministration of CS-E to mice. A putative receptor in the lung for CS-E remains to be identified.
Our recent studies showed that a single chain antibody, GD3G7, strongly reacts with E-unit-containing CS/DS, and CS-E-derived decasaccharides with three consecutive E-units were identified as possible minimal epitope structures for GD3G7 (29). The expression of the epitope recognized by GD3G7 is strongly up-regulated in ovarian carcinomas, and its expression is undetectable in normal ovary. Moreover, expression of this epitope is found in fenestrated and tumor blood vessels, both of which are vascular endothelial growth factor-dependent (28). Consistent with these findings, LM66-H11 cells rather than P29 cells are specifically immunostained by GD3G7, suggesting that the expression of such E-unit-containing epitopes is also upregulated with the increase in metastatic potential. The strong inhibitory activity of the antibody GD3G7 against experimental lung metastasis of LM66-H11 cells provides powerful evidence for the involvement of the GD3G7 epitope in the metastasis. Furthermore, our results showed that CS-E decasaccharides were minimal oligosaccharides with substantial inhibitory activity against the metastasis of LM66-H11 cells, which is consistent with the notion that CS-E-derived decasaccharides were the minimal size required for the binding of the antibody GD3G7 (29). A few studies have addressed that CS-PGs play a major role in cancer cell behavior (57) and that alterations in their expression and function during transformation facilitate tumor progression (57). However, the expression of specific CS/DS structures has not been correlated to tumor aggressiveness.
Several studies on blood-borne metastasis have shown that tumor cells in the circulation generally get arrested in the microcirculation of secondary organs and may extravagate with high efficiency (58). Because tumor cell migration, invasion, and survival are also crucial components of metastasis, here we asked whether the up-regulated expression of E-units plays a role in the metastasis of Lewis lung carcinoma cells. Therefore, we compared the migratory and invasive activities of LM66-H11 cells highly expressing E-units versus P29 cells expressing fewer E-units. As expected, the migration and invasion were significantly higher for LM66-H11 cells than for the low metastatic P29 cells (Fig. 8A), which may be correlated with the ECM-degradation process. These phenomena prompted us to see at which step of the metastatic cascade these inhibitors (CS-E and GD3G7) might interfere in vivo. Our studies showed that none of the CS variants or GD3G7 affected the migration of these cells in vitro. In contrast, CS-E and GD3G7 markedly suppressed the invasion of LM66-H11 cells. Next, we tested the anti-proliferative effect of these inhibitors, because intravascular proliferation of attached tumor cells is also essential to the metastatic process (59). CS-E and GD3G7 significantly inhibited the LM66-H11 cell proliferation on laminin-coated plates. Other major findings of this study are structural changes of CS chains at the surface of highly metastatic LM66-H11 cells and differences in the invasion and proliferation of LM66-H11 cells as compared with those of low metastatic P29 cells. These changes may modulate the behavior of these tumor cells.
In support of our disaccharide and immunocytochemical analyses showing that LM66-H11 cells highly express E-unitcontaining structures, gene expression studies confirmed the up-regulated expression of N-acetyl-D-galactosamine-4-O-sulfate 6-O-sulfotransferase responsible for the synthesis of E-units in high metastatic LM66-H11 cells compared with low metastatic P29 cells. In addition, the expression of chondroitin 4-O-sulfotransferase-1 involved in the synthesis of GlcUA␤1-3GalNAc(4S) units, as well as glucuronyl C5 epimerase, dermatan 4-O-sulfotransferase-1, and chondroitin 4-O-sulfotransferase-2 involved in the synthesis of DS corroborated the results from the disaccharide analysis (Table 1) and is consistent with our previous results that ovarian cancer cells overexpressing N-acetyl-D-galactosamine-4-O-sulfate 6-O-sulfotransferase showed higher binding of GD3G7 (28). These data further strengthen the notion that CS/DS chains containing E-units are involved in the metastatic potential of LM66-H11 cells. Notably, increased expression of N-acetyl-D-galactosamine-4-Osulfate 6-O-sulfotransferase was recently reported for colorectal cancer cells (60). It has recently been reported that the CS-E epitope recognized by GD3G7 is strongly expressed in the ECM of a wide variety of human tumors including pancreatic ductal adenocarcinomas (61). The epitope is also strongly expressed in the ECM of human ovarian adenocarcinomas but not in normal ovaries or cystadenomas and to variable degrees in human ovarian carcinoma cell lines (28). Although strong staining has been found in the tumor stroma in lymph node metastasis of serous adenocarcinomas (28), the role of the GD3G7 epitope in human tumor metastasis remains to be investigated.
In conclusion, the E-unit-containing epitope recognized by the antibody GD3G7 is highly expressed in the high metastatic LM66-H11 cells, involved in tumor metastasis, and provides a target for the diagnosis and treatment of tumors. CS-E and the phage display antibody GD3G7 interfere with the proliferation and invasion of at least Lewis lung carcinoma cells during metastasis. These inhibitors may also suppress the expression and/or activation of matrix metalloproteinase-2 and thereby prevent invasion and metastasis as in the case of HS (7). This study shows the cell surface E-unit-rich CS/DS produced by highly metastatic lung carcinoma cells to be involved in cell proliferation, survival, and invasion, and so attenuating the several biological processes of E-unit-rich CS/DS on the surface of tumor cells may provide a new therapeutic approach for tumors.