During tumor cell invasion, certain extracellular matrix (ECM) components such as hyaluronan (HA) are degraded into small oligosaccharides, which are detected in patients. We previously reported that such HA oligosaccharides induce the proteolytic cleavage of an ECM-binding molecule CD44 from tumor cells and promote tumor cell migration in a CD44-dependent manner. Here, we report that chondroitin sulfate E (CSE), another component of the tumor ECM, strongly enhances CD44 cleavage and tumor cell motility when degraded into oligosaccharides. CSE and its degradation products were detected in pancreatic ductal adenocarcinoma. In CD44-expressing pancreatic tumor cells, degraded forms of CSE but not intact CSE enhanced CD44 cleavage; enzymatic digestion of such low-molecular weight CSE (LMW-CSE) abrogated this enhancement. Among the LMW-CSE preparations examined, 3-kDa CSE most potently induced CD44 cleavage. Nuclear magnetic resonance analysis showed that the 3-kDa-CSE bound to CD44, and that blocking such binding abrogated the CD44 cleavage induction. LMW-CSE also induced prominent filopodia formation and cytoskeletal changes in tumor cells; these effects were also abrogated by blocking the LMW-CSE binding to CD44. Chemically synthesized CSE hexasaccharides also enhanced the CD44 cleavage and tumor cell motility in a CD44-dependent manner. We conclude that the degraded forms of CSE modulate cell adhesion and migration by interacting with tumor-cell CD44, suggesting that the degradation products of tumor-associated ECMs that interact with CD44 play a significant role in CD44-mediated tumor progression. [Cancer Res 2008;68(17):7191–9]
- Chondroitin Sulfate E fragments
- tumor cell
Chondroitin sulfate (CS) is an important component of the extracellular matrix (ECM) in normal as well as tumor tissues. CS consists of repeating units of GalNAc-GlcA disaccharides polymerized into long chains with an average size of 20 kDa under physiologic conditions. CS has been found in large amounts in melanomas ( 1, 2) and other types of cancer ( 3, 4). CS seems to play an important role in tumor growth, given that the treatment of tumors with chondroitinase AC results in reduced tumor growth and invasion ( 5) and CS-targeting liposomes have been used to target drugs to metastases in mice ( 6).
There are at least five different types of CSs, as defined by the sulfation pattern of the repeating disaccharide moieties. One CS, CSE, which has sulfate groups at the four and six positions of the GalNAc residue, was reported to bind a variety of biological molecules, such as heparin-binding factors ( 7) including midkine ( 8), L- and P-selectin and CD44 ( 9), and chemokines ( 9), and possesses important physiologic functions especially in the central nervous system ( 10, 11). Recently, it has been shown that CSE is strongly expressed in the ECM of ovarian adenocarcinomas but not in normal ovaries or cystadenomas ( 12), although its exact roles in tumor biology remain to be elucidated.
Tumor cell adhesion and degradation of the ECM proteins surrounding the cells are tightly linked processes in vivo. Tumor cells interact with ECM components, including hyaluronan (HA; ref. 13) and CS ( 9), via the multifunctional transmembrane protein CD44, and such interactions have been suggested to promote ECM degradation by activating ECM-degrading enzymes in the tumor tissue ( 14). The ECM-degrading enzymes promote tumor progression by dynamic remodeling of the tumor stroma to create better conditions for growth and invasion, and interestingly, the resulting ECM degradation products have often been shown to contribute to tumor progression ( 15– 18). We reported previously that HA 6 to 36-mers interact with CD44 to induce up-regulation of the proteolytic cleavage of cell surface CD44 and also the up-regulation of cell motility as well as the invasion of tumor cells in vitro ( 19), and that the HA fragments that are capable of inducing CD44 cleavage are produced by the tumor cells themselves ( 16). Like HA, CS oligosaccharides are expected to be generated under certain physiologic and pathologic conditions in vivo, given that bacterial hyaluronidases ( 20) and certain mammalian HA–degrading enzymes ( 21) have been shown to degrade CS in vitro. However, the exact degradation mechanism for CS in vivo remains unclear, let alone the possible functions of CS oligosaccharides in the various biological situations where they are likely to be produced.
The proteolytic cleavage of CD44 has been widely observed in various human tumors in vivo ( 22). The level of circulating soluble CD44, which is determined by the amount of cleavage, has been reported to correlate with tumor burden in gastric and colon cancer patients ( 23), indicating that CD44 cleavage plays an important role in tumor progression.
Given these observations, we hypothesized that non-HA degradation products of ECM that interacts with CD44 may also induce CD44 cleavage and thus be involved in tumor progression. Here, we show that CSE and its degradation products are present in pancreatic ductal adenocarcinoma (PDAC), and that low molecular weight (LMW)-CSE (LMW-CSE) that mimic those in size found in PDACs indeed induced the proteolytic cleavage of CD44 from tumor cells and promoted tumor-cell migration. These results strengthen the hypothesis that tumor cells and the surrounding ECM act on each other reciprocally, to promote tumor progression ( 15, 17). They also indicate that tumor cell–CD44 plays a crucial role in these interactions by recognizing a non-HA ECM degradation product, LMW-CSE, directly implicating LMW-CSE in CD44-mediated tumor progression.
Materials and Methods
Mice. Male BALB/c nu/nu mice (4- to 5-wk-old) were purchased from Harlan Sprague-Dawley, Inc., and were maintained in accordance with the University of California, Santa Barbara, institutional guidelines governing the care of laboratory mice. The animal experimentation was approved by Animal Research Committees at University of California, Santa Barbara.
Cell culture. The human pancreatic carcinoma cell line MIA PaCa-2 was obtained from the Cell Resource Center for Biomedical Research, Institute of Development, Aging and Cancer, Tohoku University. The cells were grown as described previously ( 19).
Reagents. The ScFv antibody GD3G7 was constructed as described previously ( 12). A rabbit polyclonal antibody (pAb) directed against the cytoplasmic domain of CD44, anti-CD44cyto pAb, ( 24) was kindly provided by Dr. Hideyuki Saya (Division of Gene Regulation, Institute for Advanced Medical Research, School of Medicine, Keio University, Tokyo, Japan). An anti-human CD44 mAb, BRIC235, was purchased from the International Blood Group Reference Laboratory, and the Fab fragment of BRIC235 was prepared as described previously ( 10). Anti–β-tubulin monoclonal antibody (mAb) and mouse IgG were purchased from Calbiochem and Sigma-Aldrich Co., respectively. Horseradish peroxidase (HRP)-conjugated anti-rabbit IgG and HRP-conjugated anti-mouse IgG were purchased from American Qualex. A biotin conjugated vesicular stomatitis virus (VSV)-G tag antibody was from Rockland Immunochemicals, Inc. Vectastain Elite ABC-Peroxidase kit and 3,3′-diaminobenzidine (DAB) peroxidase substrate kit were from Vector laboratories, Inc. CSE preparations and the HA oligosaccharide were provided by Seikagaku Kogyo Co. ( 10) and Taisho Pharmaceutical Co., Ltd. Synthetic 6-mer CSE was prepared as described previously ( 25, 26). Human umbilical cord HA, which mainly consists of 1,000-kDa HA, was purchased from Sigma-Aldrich Co. Chondroitinase ABC was purchased from Seikagaku Kogyo Co. Carbobenzoxyl-leucinyl-leucinyl-leucinal (MG132) was from the Peptide Institute, and phorbol myristate acetate (PMA) was from Sigma-Aldrich Co.
Generation of tumor samples and immunohistochemistry. BALB/c nu/nu mice were injected with 1 million MIA PaCa-2 cells into the pancreas. The tumors were harvested after 6 wk and frozen on dry ice. Frozen PDAC samples from 10-wk-old Pdx1-Cre LSL-KrasG12D Ink4a/Arflox/lox mice ( 27) and normal pancreas samples from wild-type mice with the same background were kindly provided by Dr. Douglas Hanahan (Department of Biochemistry and Biophysics, Comprehensive Cancer Center and Diabetes Center, University of California at San Francisco, San Francisco, CA). Paraffin-embedded human tumor tissue arrays were generously provided by Applied Phenomics LLC. Immunohistochemistry was performed as described previously ( 28). For cryosections, 5-μm-thick tumor sections were fixed with acetone/methanol (1:1) and rehydrated with water. The sections were treated with 0.5% hydrogen peroxidase for 10 min, blocked in 1% bovine serum albumin/10% rabbit serum in TBS for 1 h, treated with avidin/biotin blocking kit following manufacturer's instructions (Vector laboratories, Inc.), and incubated with the ScFv antibody GD3G7 (diluted 10-fold) for overnight at 4°C. After incubation with a biotin-conjugated VSV-G tag antibody (diluted 1,000-fold) for 1 h at room temperature, the sections were treated with Vectastain ABC solution (diluted 100-fold) for 30 min, and colored with a DAB peroxidase substrate kit according to the manufacturer's instructions. Finally, the sections were treated with a series of ethanol solutions and xylene, and mounted with a xylene-based mounting medium. For human tumor tissue arrays, 6-μm-thick sections were deparaffinized in a series of xylene and ethanol, and were processed for staining as described above. In some cases, the sections were treated with 5 mU per section of intact or boiled chondroitinase ABC protease free to confirm the specificity of the staining with GD3G7 ( 12, 28).
CSE degradation (chondroitinase ABC treatment). One milligram of 3-kDa CSE was treated with 5 or 200 mU/mL of chondroitinase ABC ( 28– 30) at 37°C for 4 h and filtered through a 0.22-μm pore filter, then dialyzed against PBS. In some experiments, chondroitinase ABC was heat inactivated before use ( 19, 28, 30).
Detection of CSE fragments in tumor samples (gel filtration chromatography and ELISA). Size profiles of the CSE present in the tumor extracts of PDACs were determined by a gel filtration chromatography combined with an ELISA ( 16). Fifteen PDAC tumors collected from 10-wk-old Pdx1-Cre LSL-KrasG12D Ink4a/Arflox/lox mice were combined and homogenized in PBS with an electric homogenizer. The homogenized samples were treated with 0.25% trypsin (Sigma-Aldrich Co.) for 4 h at 37°C followed by 200 μg/mL Pronase (Sigma-Aldrich Co.) for 18 h at 37°C. After centrifugation (15,000 × g, 4°C, 120 min), the supernatant was collected, passed through a 0.22-μm pore filter, and was applied to a Sepharose CL-6B column (1.5 × 120 cm; Amersham Biosciences) that was calibrated with 75-, 10-, and 4-kDa CSE. Samples were eluted in PBS at 0.875 mL/min, and 3.5-mL fractions were collected. Subsequently, the concentrations of the CSE fragments in the fractions were determined by an ELISA. The ELISA works on a competitive binding principle ( 16). The CSE present in the samples competes with CSE coated on microtiter wells to bind GD3G7. Ninety-six-well plates coated with 500 μg/mL 75-kDa CSE were incubated with the gel filtrated samples of the tumor extracts and GD3G7 (diluted 20-fold). After incubation at 37 °C for 2 h, the wells were washed in PBS containing 0.05% Tween 20, and incubated with a biotin-conjugated VSV-G tag Ab (diluted 2,000-fold) for 1 h at 37°C. After being washed, the wells were analyzed quantitatively using an avidin-biotin detection system. A standard curve was drawn by plotting the absorbance at 490 nm against the concentration of 4- and 75-kDa CSE, and using the curves, the HA concentration in each sample was calculated.
Binding of CSE fragments to CD44—nuclear magnetic resonance measurements. Uniformly 15N-labeled CD44 HA-binding domain (HABD; residues 21–178) was prepared as described previously ( 31, 32). Nuclear magnetic resonance (NMR) measurements were carried out at 25°C on a Bruker Avance 600 spectrometer. The initial sample for CSE titration experiments consisted of 0.3 mmol/L 15N-labeled CD44 HABD, 50 mmol/L sodium phosphate (pH 6.7), 150 mmol/L NaCl, and 1 mmol/L NaN3 in H2O/D2O (9/1) buffer. A stock solution of unlabeled CSE was concentrated to 10 mmol/L in the same buffer. Aliquots of the CSE stock solution were then added to the 15N-labeled CD44 HABD in the titration experiments, and the 1H-15N HSQC spectra of CD44 HABD were compared before and after the addition of CSE.
CD44 cleavage assay. The CD44 cleavage assay was performed as reported previously ( 16, 19). Briefly, MIA PaCa-2 cells (5 × 104 cells per well) were cultured overnight at 37°C, and incubated with 10 μmol/L MG132 for 30 min at 37°C to inhibit the secondary cleavage of the CD44 intracellular domain ( 19, 33). The cells were then incubated with one of the CSE preparations or 6.9-kDa HA for 1 h, or with PMA for 30 min at 37°C in the presence of 10 μmol/L MG132. In some assays, the cells were pretreated with 10 μg/mL BRIC235 Fab fragment 30 min before and during the MG132 treatment. The cells were lysed with SDS sample buffer and subjected to Western blotting using anti-CD44cyto pAb or anti–β-tubulin mAb.
Cell morphology analyses (Immunofluorescence Microscopy). The immunofluorescence assay was performed as reported previously ( 16, 19) except that MIA PaCa-2 cells were treated with CSE preparations or 6.9-kDa HA at 50 μg/mL for 1 h. In some assays, the cells were pretreated with 10 μg/mL BRIC235 Fab fragment 30 min before and during the MG132 treatment. For quantitative analyses, the number of spindle-shaped cells in five defined fields (×100) was counted under a light microscope following the same treatment as described above, and the average was determined. Each assay was performed thrice.
Migration assay. Cell migration was analyzed using 12-well Costar Transwell chambers (Corning, Inc.) with 12-μm pore size filters as described previously ( 16, 19). Briefly, both sides of the filter were coated with 500 μg/mL 1,000-kDa HA. MIA PaCa-2 cells (1 × 105 cells/mL) were added to the upper compartment and incubated at 37°C for 3 h, then incubated with or without 10 μg/mL BRIC235 or mouse IgG, 20 min before and during the migration assay. Finally, one of the CSE preparations or 6.9-kDa HA was added to the upper compartment of the wells at a final concentration of 50 μg/mL. The chambers were subsequently incubated at 37°C for 15 h, and the number of cells that migrated to the lower side of the membranes was counted in 10 defined high-power fields (×200), and the average was determined. Each assay was performed five times.
CSE is expressed in various tumors. CS is overexpressed in various tumors ( 1– 4), and the change in its sulfation pattern was reported in several types of cancers ( 34, 35). However, little is known about the precise expression patterns of the CS units in cancers, partially due to the lack of antibodies that specifically recognize each disaccharide unit. This issue was overcome by GD3G7, an antibody recently established to be recognizing CSE specifically ( 28). Using GD3G7, we examined the expression of CSE in pancreatic cancers that are one of the most aggressive human tumors ( 27). A strong CSE expression was detected in spontaneous PDAC tumors from Pdx1-Cre LSL-KrasG12D Ink4a/Arflox/lox mice ( 27), especially in the ECM surrounding the tumor ducts ( Fig. 1Ab, arrowheads ) but minimally in the areas that seemed to be histologically normal ( Fig. 1Ab, *) and in the pancreas sections from normal donors ( Fig. 1A, e and f). The specificity of the signals was confirmed by pretreatment of the tissue sections with intact chondroitinase ABC that abolished the signals ( Fig. 1A, c) and with boiled chondroitinase ABC that did not ( Fig. 1A, d). Overexpression of CSE was similarly observed in an orthotopic xenograft tumor of MIA PaCa-2, a human pancreatic carcinoma cell line, in nude mice ( Fig. 1A, h); the specificity of the signals in the tumor was confirmed with chondroitinase ABC pretreatment of the sections ( Fig. 1A, i and j). In addition, strong CSE signals were detected in a s.c. xenograft of MIA PaCa-2 (data not shown). Furthermore, strong CSE signals were detected in the ECM of human PDACs ( Fig. 1B, a–e), which were abolished by chondroitinase ABC digestion ( Fig. 1B, f–j). Such signals were often intense in the ECM surrounding the tumor ducts ( Fig. 1B, a, arrowheads). In addition, strong expression of CSE was also detected in the ECM of a wide variety of human tumor microarray samples (data not shown).
Fragmented but not intact CSE enhances CD44 cleavage in tumor cells. Degradation products of the tumor ECM components, such as HA, are often found in the tumor microenvironment ( 15– 17). This was also the case with CSE. We found that the tumors contained various size populations of CSE ranging from those smaller than 4-kDa to those much larger than 75-kDa ( Fig. 2A, f1–f3 ). In addition, small continuous peaks were observed between f1 and f2 ( Fig. 2A, arrowheads).
We next asked whether such LMW-CSE is capable of inducing CD44 cleavage in MIA PaCa-2 cells that abundantly express CD44H, the 90-kDa standard form of CD44 ( 19); the cleavage was assessed by measuring the membrane-associated 25-kDa CD44 semiquantitatively in Western blotting assays ( 16, 19, 22, 24). As shown in the left of Fig. 2B, enzymatically degraded CSE fragments, such as 10- and 5-kDa CSE, enhanced CD44 cleavage, as did PMA or 6.9-kDa HA, whereas 75-kDa CSE induced little CD44 cleavage in MIA PaCa-2 cells.
Because 5-kDa CSE enhanced CD44 cleavage most efficiently, we compared different LMW-CSE preparations ranging from 1- to 6-kDa in terms of their ability to induce CD44 cleavage. As shown in Fig. 2B (right), 3-kDa CSE was the most potent in enhancing CD44 cleavage of the LMW-CSE preparations we tested. Further analysis showed that LMW-CSE such as 3-kDa CSE ( Fig. 2C, left), as well as 1-, 2-, and 4-kDa CSE preparations (data not shown), enhanced CD44 cleavage in a concentration-dependent manner, whereas the 75-kDa CSE preparation failed to induce cleavage significantly at any concentration tested ( Fig. 2C, right), indicating that LMW-CSE but not high molecular weight CSE (HMW-CSE) can induce CD44 cleavage efficiently. Essentially identical results were obtained in a human glioblastoma cell line, U251MG (data not shown).
Complete digestion of CSE abolishes the LMW-CSE–induced CD44 cleavage. We next examined whether the enhanced CD44 cleavage by LMW-CSE requires the repeating disaccharide structure of CSE. As shown in Fig. 3 , when extensively digested by 200 mU/mL chondroitinase ABC, the 3-kDa CSE preparation failed to up-regulate CD44 cleavage. Digestion by chondroitinase ABC at a concentration that selectively degrades CS (5 mU/mL; refs. 28– 30) also strongly reduced the ability of CSE to enhance the cleavage of CD44, whereas digestion by heat-inactivated chondroitinase ABC reduced it only marginally. These results indicate that the CD44 cleavage was induced by the repeating disaccharide structure contained in LMW-CSE but not by non–CSE molecules such as HA fragments or growth factors that might have been present, if any, in the CSE preparation.
CSE fragments enhance CD44 cleavage by binding to CD44. We next investigated whether degraded CSE binds to CD44 by examining the binding of LMW-CSE to the recombinant CD44 HABD using NMR. As shown in Fig. 4A , comparison of the two-dimensional 1H-15N HSQC spectra of uniformly 15N-labeled CD44 HABD in the unbound and LMW-CSE–bound states clearly revealed the presence of chemical shift differences for certain residues of CD44 HABD (D23, A31, E48, L52, Y105, L 107, V132, V148, and R162), indicating that LMW-CSE binds to CD44 HABD. Because the CD44 cleavage enhanced by the 3-kDa CSE was completely inhibited by an anti-CD44 neutralizing antibody BRIC235 ( Fig. 4B; refs. 19, 29), this binding of LMW-CSE to CD44 seems to be crucial for the up-regulation of CD44 cleavage.
The 3-kDa CSE fragments promote migration of MIA PaCa-2 cells. It has been reported that CD44 cleavage plays a critical role in tumor cell migration ( 36) and that HA fragments that can induce CD44 cleavage promote CD44-dependent tumor cell migration ( 19). Accordingly, we examined whether CSE affects the CD44-dependent motility of cancer cells by observing the morphology and migration of CSE-treated cells on materials coated with 1,000-kDa HA, which was previously shown to be useful as a CD44-dependent foothold that does not affect the motility of the cells ( 19). As shown in Fig. 5 , the LMW-CSE that induced CD44 cleavage also exhibited a potent ability to promote tumor cell motility, inducing marked filopodia and actin filament remodeling in MIA PaCa-2 cells ( Fig. 5A, d–f) at a level comparable with that seen with 6.9-kDa HA ( Fig. 5A, m–o), whereas 75-kDa CSE induced only minimal changes in cell morphology ( Fig. 5A, j–l). The anti-CD44 antibody BRIC235, which blocks the binding of CD44 to CSE ( 29), completely blocked the ability of LMW-CSE to induce morphologic changes in MIA PaCa-2 tumor cells ( Fig. 5, g–i), indicating that the interaction between CD44 and LMW-CSE was responsible for the observed morphologic changes. Quantitative analyses further supported these findings. LMW-CSE dramatically increased the filopodia-forming cell population as well as 6.9-kDa HA, whereas 75-kDa CSE did not ( Fig. 5B). This increase was completely inhibited by BRIC235.
We next examined whether LMW-CSE could increase tumor cell motility, using a Boyden chamber assay. Although 75-kDa CSE did not affect the migration of MIA PaCa-2 cells across an HA-coated filter, 3-kDa CSE induced tumor cell migration to the same extent as did 6.9-kDa HA ( Fig. 5C; ref. 19). The addition of the anti-CD44 BRIC235 ( 29) but not mouse IgG completely blocked the LMW-CSE–enhanced migration ( Fig. 5C), indicating that the ability of LMW-CSEfs to enhance cell migration was CD44 dependent, compatible with the results of the morphologic analysis. In addition, the enhanced motility was not observed on substrates other than 1,000-kDa HA, such as fibronectin or Matrigel (data not shown), confirming that the migration was CD44 dependent. Collectively, these results suggest that LMW-CSE enhances not only CD44 cleavage but also CD44-dependent cell migration in tumor cells, in a manner very similar to that seen with LWM-HA.
Synthetic LMW-CSE also enhances CD44 cleavage and cell migration in MIA PaCa-2 cells. To verify that the CSE structure (GlcA[β1-3]GalNAc[4,6-O-disulfate]) is critical for the ability of LMW-CSE to mediate the up-regulation of CD44 cleavage and cell migration in tumor cells, and to further exclude the possibility that any contamination in the LMW-CSE preparation caused the effects, we investigated whether pure, synthetic hexasaccharide CSE could also induce the changes seen with the 3-kDa CSE fragments in MIA PaCa-2 cells. As shown in Fig. 6A , the synthetic CSE enhanced the CD44 cleavage as efficiently as 3-kDa CSE in MIA PaCa-2 cells, and this enhancement was blocked by the Fab fragment of BRIC235, suggesting that the cleavage was induced by the CD44-CSE interaction ( Fig. 6A). The synthetic CSE also induced motility changes in MIA PaCa-2 cells that were very similar to those seen with 3-kDa CSE; i.e., it induced prominent filopodia formation and actin filament remodeling ( Fig. 6B and C), enhanced CD44-dependent tumor cell migration ( Fig. 6D), and these effects were blocked by BRIC235. These results are fully compatible with the idea that the CSE structure is critical for the induction of CD44 cleavage and for the CD44-dependent cell migration observed in MIA PaCa-2 cells.
In the present study, we report novel biological activities of LMW-CSE in tumor cells. We showed that degradation products of CSE that mimic one of the populations of the CSE fragments found in PDACs potently enhanced CD44 cleavage and cell motility in tumor cells. LMW-CSE seemed to exert its effects by binding to CD44 on the tumor-cell surface, and the repeating disaccharide units in the LMW-CSE preparations were essential for these abilities.
Among various types of CSs, CSE, which has the characteristic E unit (GlcUA-GalNAc[4S,6S]), has been reported to be abundantly expressed in the ECM of ovarian adenocarcinomas but not in normal ovaries or cystadenomas ( 12). In agreement with this report, we found that CSE is highly expressed in human and mouse PDACs, especially in the ECM surrounding the tumor ducts ( Fig. 1). We also found that CSE ranging from ∼ 4 kDa or smaller to a much larger size than 75 kDa were present in the mouse PDAC extracts ( Fig. 2A, f1–f3 and arrowheads) and that human and mouse PDACs showed similar CSE expression patterns. We thus think it likely that such CSE fragments are also present in human pancreatic carcinomas.
The exact mechanism of CSE fragment production in vivo remains unclear. In vitro, CSE fragments can be generated by enzymatic digestion of CSE with chondroitinases and hyaluronidases ( 37). Hyaluronidases have been strongly implicated in tumor growth and or invasion ( 15– 18, 38) and are often highly expressed in certain malignant tumors ( 39) where CS has also been reported to be expressed abundantly ( 1, 3, 4). Because it has been suggested that some of the hyaluronidases can digest CS ( 21), it is tempting to speculate that hyaluronidases, may exhibit protumoral functions also by generating LMW-CSE in the ECM. Identification of the enzymes that regulate the metabolism of CSE in mammals and those that are responsible for the degradation of CSE in tumor tissues requires further study. In addition, because CSE often exists as side chains of proteoglycans ( 10, 11), it is possible that specific CSE-proteoglycan(s) may be up-regulated in these areas within the tumors, which also merits future investigation.
It was intriguing that LMW-CSEs but not HMW-CSEs exhibited the enhancement of CD44 cleavage and cell motility in tumor cells. It is unlikely that any contamination in the CSE fragment preparations caused such effects for the following reasons: (a) the CSE fragments were prepared from the highly purified “superspecial grade” 75-kDa CSE ( 40); (b) chondroitinase ABC pretreatment strongly reduced the enhancement of CD44 cleavage in CSE fragment-treated cells ( Fig. 3); (c) a pure synthetic CSE hexasaccharide enhanced CD44 cleavage and tumor cell motility ( Fig. 6).
Although the E unit is the crucial component within the LMW-CSE preparations that triggered the enhancement of CD44 cleavage and tumor cell motility, such ability does not seem to arise from the change in the proportion of the E unit within the CSE preparations. The LMW-CSE preparations were generated by treating the 75-kDa CSE with sheep testicular hyaluronidase that produces saturated CSE fragments that contain E units ( 40). Indeed, all of the CSE fragments contained the E unit, and the proportion of the E unit within the fragments ranged from ∼55% to 77% (e.g., 3-kDa CSE, 76.4%), which was similar to that of the intact 75-kDa CSE, 60.8% (analytic data provided by Seikagaku Kogyo, Co.). Besides the E unit, it is possible that other components such as GlcUA-GalNAc, GlcUA-GalNAc(4S), and GlcUA-GalNAc(6S) present in the CSE fragments were also partially responsible for the effects because such CS are able to bind to CD44 ( 29). Preliminary results indicated that synthetic chondroitin hexasaccharides that lacked the sulfate residues also up-regulated CD44 cleavage in MIA PaCa-2 cells, albeit much more weakly than the synthetic oversulfated chondroitin hexasaccharides. 6 This finding raises the possibility that other types of CS might as well show similar effects to LMW-CSE, and that the effects are affected by the sulfation patterns of the CS chains. Because various CS epitopes are differentially expressed in tumor cells and their surrounding tissues ( 1, 2, 12, 28, 34, 35), functions of the breakdown products of other CSs also merit future investigation.
Although the binding of LMW-CSE to CD44 is likely to be essential for the induction of CD44 cleavage, sheer occupation of the binding site on CD44 or the crosslinking of CD44 seems insufficient because HMW-CSE (75-kDa; Fig. 2) and several anti-CD44 monoclonal antibodies (data not shown), all of which would be expected to induce significant binding and crosslinking of CD44, were unable to induce CD44 cleavage. This may indicate that a postbinding event(s), rather than the simple binding of CSE to CD44, is crucial for the LMW-CSE–induced up-regulation of CD44 cleavage. Considering that LMW-CSE requires CD44 to exert its effects, LMW-CSE as well as HA ( 41) may be endocytosed by cells via CD44, and CSE of different sizes may differentially initiate intracellular signaling. Recent investigations showed that cell-matrix interaction via CD44 induces activation of the small GTPase Rac ( 42) and stimuli such as PMA and ionomycin activate a disintegrin and metalloproteinase (ADAM)17 and ADAM10 ( 43, 44), respectively, which then results in the cleavage of the CD44 ectodomain ( 43, 44) and lamellipodia formation of the CD44-expressing cells ( 44, 45). In addition, it was shown that LMW-HA interaction with CD44 triggers structural changes in CD44, resulting in an increased susceptibility of the CD44 molecule to proteolytic cleavage by trypsin ( 46). Therefore, a similar scenario can be envisaged for LMW-CSE that it initiates intracellular signaling pathways that activate proteases that cleave CD44, which became structurally susceptible to proteolysis, and also promote tumor cell motility by activating small GTPases. Finally, although it is also possible that LMW-CSE interacts with non-CD44 molecules as well, the observation that the functional blocking of CD44 by a neutralizing antibody alone was sufficient to abrogate the ability of LMW-CSE ability to promote CD44 cleavage, and cell motility indicates that the interaction of LMW-CSE with non-CD44 molecules is probably not a critical event, if it occurs.
In conclusion, these results reinforce the idea that tumor cells and their surrounding ECM components, particularly glycosaminoglycan chains, interact with each other, mediating dynamic and critical interactions between the tumor and its stroma. Given that cancer stem cells highly express CD44 ( 47), further investigation of the LWM ECM breakdown products that can interact with CD44 should provide new insights into the ways tumor growth and invasion are regulated from the viewpoint of ECM metabolism, which may lead to the development of new therapeutic regimens that can control tumor progression.
Disclosure of Potential Conflicts of Interest
M. Miyasaka: Grant, Taisho Pharmaceutical Co. Ltd. The other authors disclosed no potential conflicts of interest.
Grant support: Grant-in-aid from the Ministry of Education, Culture, Sports, Science and Technology of Japan, and Taisho Pharmaceutical Company.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Dr. Hideyuki Saya (Division of Gene Regulation, Institute for Advanced Medical Research, School of Medicine, Keio University, Tokyo, Japan) for providing the anti-CD44cyto pAb; Dr. Douglas Hanahan (Department of Biochemistry and Biophysics, Comprehensive Cancer Center, and Diabetes Center, University of California at San Francisco, San Francisco, CA) for the PDAC and wild-type pancreas samples; Applied Phenomics LLC (Tartu, Estonia) for the tumor tissue microarrays; Seikagaku Kogyo, Co. and Taisho Pharmaceutical Co., Ltd. for the CSE and HA preparations; Kazuyo Tomiyama, Tamae Kondo, and Masaaki Miyo for their technical assistance; Dr. Yukihiko Ebisuno for critically reading the manuscript; and Drs. Erkki Ruoslahti and Tambet Teesalu for their generous support.
Note: Present addresses for K.N. Sugahara: Vascular Mapping Center, Burnham Institute for Medical Research at UCSB, Santa Barbara, California; T. Tanaka: Laboratory of Immunobiology, School of Pharmacy, Hyogo University of Health Sciences, Kobe, Japan; M. Takeda: Structural Biology Center, Graduate School of Science, Nagoya University, Aichi, Japan; and H. Terasawa: Department of Structure-Function Physical Chemistry, Faculty of Medical and Pharmaceutical Sciences, Kumamoto University, Kumamoto, Japan.
↵6 K.N. Sugahara, et al., unpublished data.
- Received November 14, 2007.
- Revision received May 5, 2008.
- Accepted June 3, 2008.
- ©2008 American Association for Cancer Research.