Syndecan-4 protects against osteopontin-mediated acute hepatic injury by masking functional domains of osteopontin

We initially examined whether heparin binds to OPN. The binding of biotinylated heparin to the full-length form (OPN full/glutathione S-transferase [GST]), the amino-terminal half of thrombin-cleaved form (OPN N half/GST), and the carboxy-terminal half of thrombin-cleaved form (OPN C half/GST) of human OPN was tested. Heparin bound to Escherichia coli–derived OPN full/GST (Fig. 1 A). In contrast, the same heparin preparation failed to bind to both OPN N half/GST and OPN C half/GST. OPN is heavily glycosylated in vivo (17) and E. coli–derived OPN lacks glycosylation. To exclude that the lack of sugar moiety on OPN affects binding of OPN to heparin, we prepared the glycosylated form of recombinant OPN from Chinese hamster ovary (CHO) cells. Again, heparin bound only to the full-length form of OPN in a dose-dependent manner (Fig. S2). Note that the amino acid sequences that correspond to two putative HBDs are ¹⁶⁵YGLRSKSKKF¹⁷⁴, which contains thrombin cleavage site ¹⁶⁸R/S¹⁶⁹, and ²⁹⁸DKHLKFRI³⁰⁵, which exists in OPN C half/GST (Fig. S1). Heparin binds to YGLRSKSKKF and VVYGLRSKSKKFRR, but not to KSKEEDKHLKFRISHE and IPVKQADSGSSEEKQ (Fig. 1 B). Thus, heparin binds only to one of two putative HBDs of OPN, which contains the thrombin cleavage site. Other than heparin within mast cells, a major form of heparin-like GAG exists as HS in vivo, and syndecan is one of the major transmembrane molecules that possesses HS (1, 21). Therefore, we examined whether syndecan-4 binds to OPN in vitro. To do this, we prepared a soluble form of human syndecan-4 in which the extracellular portion of syndecan-4 is fused to the Fc portion of human IgG1 (Syn4Ig). The structural relationship between HS moiety and Syn4Ig is schematically depicted in Fig. S3. Syn4Ig bound to the full-length form but not to the thrombin-cleaved form of human OPN (Fig. 1 C). It should be remembered that the majority of GAG chains added to the core proteins of syndecan are of the HS type; however, syndecan-4 is also modified by CS chains (22). Basic fibroblast growth factor bound only to the HS chains of syndecan, whereas midkine bound to both the HS and CS chains (23–25). Therefore, we asked whether the binding of syndecan-4 to OPN is dependent on HS or CS chains. Syn4Ig was digested with either heparitinase (HSase) and/or chondroitinase ABC (CSase). HSase-treated, but not CSase-treated, Syn4Ig lost its binding to OPN, indicating that the HS chain of syndecan-4 is involved in its binding to OPN. Note that Syn4Ig bound not only to human but also to mouse OPN (Fig. 1 C).

We further tested whether the interaction of OPN and syndecan-4 occurs under physiological conditions. We prepared two human cell lines, OPN/Namalwa and NRC-12, that produce either OPN alone or OPN and syndecan-4, respectively. NRC-12 culture supernatants were applied to the affinity column coupled with either anti–syndecan-4 antibody or control IgG. OPN and syndecan-4 proteins were immunoblotted in the bound fraction from the anti–syndecan-4 antibody–coupled column but not from the control column, indicating the presence of association between OPN and syndecan-4. Culture supernatants were also applied to the affinity column coupled with anti-OPN antibody. The syndecan-4 was immunoblotted in the bound fraction derived from NRC-12 cells but not from OPN/Namalwa cells (Fig. 1 D). In addition, we confirmed the association of OPN and syndecan-4 by using genetically manipulated CHO cells transiently expressing human OPN, syndecan-4, or both OPN and syndecan-4 (Fig. S4). We also found that, like heparin, the binding of Syn4Ig was specific to the one of two putative HBDs of OPN that overlaps the thrombin cleavage site (Fig. S5). Thus, the physiological association of HBDs of OPN and HS moiety of syndecan-4 exists, and the biological significance of this association remains to be elucidated.

We thus examined whether the association of OPN and syndecan-4 modulates the availability of OPN to thrombin and/or its receptors. OPN proteins were mixed with either human IgG or Syn4Ig. Addition of thrombin resulted in the significant reduction of full-length OPN levels in OPN preparation when mixed with human IgG. However, when OPN was mixed with Syn4Ig, OPN was resistant to thrombin digestion (Fig. 2 A). Because the binding domain of syndecan-4 to OPN overlaps the thrombin cleavage site, the thrombin resistance may be achieved through the masking of the thrombin-cleaved site by syndecan-4. Syndecan-4 was originally detected in microvascular endothelial cells as an antithrombin binding molecule (19, 26). Antithrombin is a plasma serine protease inhibitor and, thus, inhibits thrombin activity by forming a covalent complex with thrombin in a 1:1 ratio. Importantly, after binding to syndecan-4 in vivo, antithrombin exhibits a 500-fold increase of its thrombin inhibitory activity (27). Therefore, it is likely that in vivo syndecan-4 regulates formation of thrombin-cleaved OPN by masking the thrombin cleavage site of OPN and/or inhibiting thrombin activity via association with the antithrombin molecule.

OPN can bind to integrin receptors in at least two ways. First, a classical integrin binding domain, ¹⁵⁹RGD¹⁶¹, within the OPN molecule exists close to the consensus thrombin cleavage site, ¹⁶⁸R/S¹⁶⁹ (17). Therefore, the recognition of OPN by an RGD-recognizing integrin receptor such as αvβ3 can be blocked when OPN is bound by syndecan-4. The binding of CHO cells to OPN was specifically inhibited by GRGDS peptides, confirming that this binding is RGD dependent. The binding of CHO cells to OPN was significantly inhibited when OPN was mixed with syndecan-4 but not with human IgG (Fig. 2 B). Note that OPN binds to the plates comparably well in the presence or absence of Syn4Ig in cell adhesion assays (Fig. S6), indicating that Syn4Ig competes for CHO cell adhesion to plate-coated OPN rather than preventing OPN from binding to the plates. Second, the ¹⁶²SVVYGLR¹⁶⁸ domain, which can be recognized by α4β1 integrin (12, 14), exists also very close to the thrombin cleavage site (17). We therefore tested whether the binding of CHO cells expressing α4β1 integrin (α4/CHO) to OPN can be inhibited by Syn4Ig. Because CHO cells express an RGD-recognizing receptor, a binding assay was performed in the presence of GRGDS peptide, thus allowing the binding of α4/CHO cells to OPN to be α4 integrin dependent. As expected, the binding of α4/CHO cells to OPN was specifically inhibited not only by anti–α4 integrin antibody but also by syndecan-4 (Fig. 2 C). It is known that OPN, αvβ3, and α4β1 (also known as very late antigen 4) integrins are involved in recruitment and cell adhesion of inflammatory cells (7, 10, 12, 14, 15, 17), and it is likely that the association of syndecan-4 and OPN may result in the inhibition of integrin-mediated inflammation. It should be remembered that OPN also putatively binds and mediates cell migration through CD44 (7, 10, 17). Thus, it is of interest in future studies to test whether syndecan-4 regulates the interaction of OPN and CD44.

To investigate whether the described scenario operates in vivo, we examined the role of syndecan-4 in ConA-induced hepatic injury in mice. As expected, both plasma OPN and alanine aminotransferase (ALT) levels were significantly elevated after ConA injection (Fig. 3 A). Because antibodies against mouse syndecan-4 and HS are commercially available, we constructed a sandwich ELISA to measure mouse syndecan-4. Plasma syndecan-4 levels were elevated after ConA injection (Fig. 3 B). We also noted that syndecan-4 gene expression in the liver was significantly up-regulated at 2 h and persisted up to 12 h after ConA injection (Fig. 3 C).

If syndecan-4 plays a critical role in the regulation of OPN cleavage by thrombin in vivo, the syndecan-4 deficiency should result in the exacerbation of hepatic injury after ConA injection. Under the condition in which control mice developed mild hepatic injury after ConA injection, syndecan-4–deficient (Syn4KO) mice developed significantly severe hepatic injury, as reflected by the elevation of ALT and aspartate aminotransferase (AST) levels (Fig. 4 A). This finding correlated well with the histology in which control mice showed minor hepatic necrosis, whereas there was massive necrosis in Syn4KO mice (Fig. 4 B). Consistent to these findings, the survival rate of mice treated with ConA was significantly reduced in Syn4KO mice as compared with control mice. Note that plasma levels of the full-length form of OPN were not different between wild-type and Syn4KO mice after ConA injection, indicating that the absence or presence of syndecan-4 did not influence the induction of OPN expression itself. However, importantly, thrombin-cleaved OPN levels were elevated in Syn4KO mice after ConA injection as compared with those in control mice (Fig. 4 C), indicating that levels of the thrombin-cleaved form of OPN correlate well with the degree of inflammation and tissue injury as previously reported (28). The specificity of ELISA used for the detection of the thrombin-cleaved form of mouse OPN is shown in Fig. S7 .

To further test whether OPN is really responsible for the exacerbation of inflammation after ConA injection in Syn4KO mice, mice were treated with a neutralizing antibody raised against the cleaved form of OPN (M5) (12, 14). Syn4KO mice treated with anti-OPN (M5) antibody showed significant amelioration of ConA-induced hepatic injury as judged by the level of plasma ALT and liver histology (Fig. 4 D).

The final issue we addressed was whether the exogenous purified syndecan-4 can protect mice from ConA-induced hepatic injury. Syndecan-4–injected mice were protected from hepatic injury, as reflected by reduced levels of ALT and IFN-γ after ConA injection compared with those in control mice (Fig. 5 A). It has been previously shown that OPN is involved in cell adhesion, cell migration, and inflammation via integrin receptors (7, 10, 12, 14, 15, 17). Therefore, we tested whether administration of syndecan-4 would inhibit OPN-mediated inflammatory responses by flow cytometry. The infiltration of inflammatory cells into liver tissues in ConA-injected mice was significantly inhibited by the administration of exogenous syndecan-4 (Fig. 5 B). Infiltrated leukocytes and macrophages/Kupffer cells were defined by the expression of CD45 and F4/80, respectively. Nevertheless, hepatic necrosis in histology was significantly reduced by the administration of syndecan-4 (Fig. 5 C).

Thus, our report reveals a novel role of HSPG in protection against OPN-mediated hepatic injury. One of the major HSPGs, syndecan-4 is a critical intrinsic regulator of inflammatory reactions in vivo by regulating various OPN functions. We identified the primary amino acid sequence within a putative HBD within OPN, which interacts with the HS moiety of syndecan-4. The putative in vivo role of syndecan-4 in OPN-mediated inflammatory reactions is schematically shown in Fig. 5 D. Initial inflammation insult induces the up-regulation of OPN and syndecan-4 expression, and activation of thrombin at the site of insult. OPN induced the migration and cell adhesion of inflammatory cells through interaction with integrins. Thrombin cleaves OPN specifically and augments the production of the thrombin-cleaved form of OPN, which favors the exacerbation of inflammatory reaction and tissue injury. Meanwhile, syndecan-4 acts as a negative regulator of inflammatory responses by binding to HBDs of OPN, thus interfering with integrin-mediated cell adhesion and migration and/or masking the thrombin cleavage site of OPN, leading to the inhibited formation of the thrombin-cleaved form of OPN. Further research is needed to determine whether OPN function is also regulated by other HSPGs such as syndecan-1, -2, and -3 and glypican. HSPG deserves further investigation as a novel therapeutic tool for various inflammatory diseases.

The three human OPN complementary DNAs (cDNAs) were inserted into pGEX-4T vector (GE Healthcare) in the same reading frame as the carrier gene GST and were transformed in E. coli JM109 cells. Thus, three unglycosylated human OPN/GST fusion proteins were produced: OPN full/GST (M1-N314), OPN N half/GST (M1-R168), and OPN C half/GST (K170-N314). These proteins were purified as described previously (29). To prepare the glycosylated forms of OPN (OPN/CHO and OPN N half/CHO), the full-length human and mouse OPN and OPN N half cDNAs were inserted into pcDNA3 (Invitrogen) and transfected to CHO-K1 cells with Lipofectamine 2000 (Invitrogen). The glycosylated form of human OPN was purified with a formyl-cellulofine column (Seikagaku Kogyo) coupled with anti-OPN (O-17) antibody (Immuno-Biological Laboratories Co., Ltd. [IBL]), as described previously (30). Mouse OPN was purified with a formyl-cellulofine column coupled with rabbit anti–mouse OPN (M3) antibody raised against ¹⁵⁵KSRSFQVSDEQYPDATDE¹⁷² (IBL). The cDNA for human syndecan-4 and the extracellular domain of human syndecan-4 (M1-E145) were PCR amplified from NRC-12 cells, and the cDNA for the extracellular domain of mouse syndecan-4 (M1-R143) was from B16-BL6 mouse melanoma cells using the following primers: human syndecan-4, 5′-TGCGGATCCGGTGCCATGGCCCCCGCCCGT-3′ (sense) and 5′-TGGCTCGAGTCACGCGTAGAACTCATTGGT-3′ (antisense); human syndecan-4 (M1-E145), 5′-TGCAAGCTTGGTGCCATGGCCCCCGCCCGT-3′ (sense) and 5′-TTGGGATCCTCCGTTCTCTCAAAGATGTT-3′ (antisense); and mouse syndecan-4 (M1-R143), 5′-AAAGAATTCGAAGCCATGGCGCCTGCCTGC-3′ (sense) and 5′-GTCGGATCCCTCTCAAAGATGTTGCTG-3′ (antisense). The amplified product of syndecan-4 was cloned into pcDNA3.1. The products of the extracellular domain of syndecan-4 were fused with a Fc portion of human IgG1 (Syn4Ig), cloned into pcDNA3.1, and transfected to CHO-K1 suspension culture cells with Lipofectamine 2000. Syn4Ig was purified by using protein A column chromatography.

Thrombin and PMA were obtained from Sigma-Aldrich. Human renal cell carcinoma cell lines (NRC-12; IBL) and human B lymphocyte lines of Burkitt lymphoma origin (Namalwa; American Type Culture Collection) were cultured in TIL medium (IBL) supplemented with 10% FCS. The Namalwa cells were stably transfected with OPN cDNA and were referred to as OPN/Namalwa. CHO cells were also transfected with OPN alone, syndecan-4 alone, or both OPN and syndecan-4 cDNA. CHO-K1 (Cell Resource Center for Biomedical Research, Tohoku University) and CHO-K1 suspension culture (RIKEN Cell Bank) cells were cultured in DMEM/nutrient mixture F-12 (Wako) supplemented with 5% FCS. NIH3T3 and B16-BL6 (Cell Resource Center for Biomedical Research, Tohoku University) were cultured in DMEM supplemented with 5% FCS. Antibodies used to detect human OPN (1B20) and syndecan-4 in the Western blot analysis were purchased from IBL. Anti–α4 integrin (P1H4) antibody was obtained from Chemicon. Rabbit antibody (M5) specifically recognizing the cryptic epitope within mouse OPN molecules, which is exposed by thrombin cleavage, was obtained from IBL and used to neutralize the function of the thrombin-cleaved form of OPN (12, 14).

6–8-wk-old C57BL/6 mice were obtained from Japan SLC. Syndecan-4 KO mice (3), backcrossed >10 times into the C57BL/6 background, were obtained from the Center for Animal Resources and Development. These animals were maintained in specific pathogen-free conditions in the animal facility of the Laboratory of Animal Experiment for Disease Model (Institute for Genetic Medicine, Hokkaido University). All animal experiments were in accordance with the guidelines of an institutional committee at Hokkaido University.

100 μg Syn4Ig was dialyzed against a 1:1 mixture of 0.1 M sodium acetate buffer, pH 7, and 10 mM calcium acetate, and digested with 20 mU HSase (Seikagaku Kogyo), with 85 mU CSase (Sigma-Aldrich), or with both enzymes at 37°C for 15 h.

OPN proteins or synthetic peptides were coated onto a 96-well plate at various concentrations at 37°C for 1 h, then blocked with 0.1% BSA in Tris-buffered saline (TBS) containing 0.05% NaN3 at 37°C for at least 1 h The plates were washed two times with TBS and incubated with either 10 μg/ml of biotinylated heparin or 10 μg/ml Syn4Ig at 37°C for 1 h After a further three washes, a 1:5,000 dilution of peroxidase-conjugated streptavidin (Jackson ImmunoResearch Laboratories) for biotinylated heparin or peroxidase-conjugated anti–human IgG (Jackson ImmunoResearch Laboratories) for Syn4Ig were added to each well at room temperature for 30 min. Bound protein was quantified by a colorimetric assay using 3,3,5,5-tetramethylbenzidine solution (Thermo Fisher Scientific) as a substrate for 15 min at room temperature. Plates were read at a wavelength of 450 nm. To evaluate endogenous binding of OPN to syndecan-4, NRC-12 cells and OPN/Namalwa cells, stimulated with PMA for 30 min in serum-free medium, were cultured for an additional 48 h without PMA. The supernatant of NRC-12 cells was applied to an anti–syndecan-4 antibody–coupled formyl-cellulofine column and washed extensively. A rabbit IgG–coupled formyl-cellulofine column was used as a control column. Elute fraction with 0.2 M glycine-HCl, pH 2.5, was immediately neutralized and electrophoresed through 12% SDS-PAGE gel and probed with anti-OPN (1B20) or anti–syndecan-4 antibody. The supernatants of NRC-12 and OPN/Namalwa cells were also applied to an anti-OPN antibody (O-17; IBL)–coupled formyl-cellulofine column and Western blotted. The supernatants obtained from CHO cells, transiently transfected with OPN alone, syndecan-4 alone, or both OPN and syndecan-4 cDNA, were also tested for the presence of association between OPN and syndecan-4.

50 μl of 3 μg/ml hOPN/CHO protein was mixed with either 50 μl of human IgG1 (hIgG) or 6 μg/ml Syn4Ig and incubated for 1 h at 37°C, then digested with 2 U of thrombin for 30 min.

The 96-well plates were precoated with 10 μg/ml OPN/CHO protein in the presence or absence of 20 μg/ml Syn4Ig overnight at 4°C, followed by treatment with 0.5% BSA in TBS for 1 h at room temperature. Cells were suspended in DMEM containing 0.25% BSA, and 200 μl of cell suspension (at a cell density of 5 × 10⁴ cells per well) was applied to 96-well plates and incubated for 1 h at 37°C. The medium was removed from the plates, and all wells were washed twice. The adherent cells were fixed and stained by 0.5% crystal violet in 20% methanol for 30 min. All wells were rinsed three times with water, and adherent cells were then lysed with 20% acetic acid. The resulting supernatants from each well were analyzed by an immunoreader (Immuno Mini NJ-2300; Nolge Nunc International), and the absorbance at 590 nm was measured to determine the relative number of cells adhered to wells. The binding of cells to OPN was expressed as 100%.

OPN, human syndecan-4 (IBL), and IFN-γ (BD Biosciences) concentrations were measured by using ELISA kits as specified by the manufacturers. The plasma level of mouse syndecan-4 was measured by using a ELISA system, which was established using 10 μg/ml of rabbit anti–mouse syndecan-4 antibody (IBL) for capture antibody and 3 μg/ml of biotinylated anti-HS antibody (Seikagaku Kogyo) for detection antibody. Purified mouse syndecan-4 Ig was used for standard. The thrombin-cleaved form of mouse OPN was detected by ELISA obtained from IBL. The detailed information on this ELISA for the thrombin-cleaved form of OPN is shown in Fig. S7.

C57BL/6 mice were injected intravenously with 10 mg ConA (Sigma-Aldrich) per kilogram of body weight, dissolved in pyrogen-free PBS. In some experiments, 150 μg Syn4Ig or human IgG was administered to mice intraperitoneally 15 h before ConA challenge. Liver damage was evaluated by measuring the serum activity of ALT and AST by using a standard clinical autoanalyzer (DRI-CHEM 5500V; Fujifilm).

Livers were harvested at various times after ConA injection. All specimens were fixed in 10% buffered formalin and embedded in paraffin. Sections were cut and stained with hematoxylin and eosin (H-E). Light microscopy was performed to assess liver injury. Necrotic areas were measured in each section by using ImageJ (version 1.37; National Institutes of Health), followed by calculation of the necrotic area per section. For each tissue, data were obtained using at least three high power fields (×100).

Liver-infiltrating leukocytes were isolated as previously described (14). In brief, livers were minced after a few minutes of perfusion, pressed through a stainless steel mesh, and suspended in PBS. After washing, cells were resuspended in 33% Percoll solution and centrifuged at 2,000 rpm for 15 min to remove liver parenchymal cells. The pellet was treated with red blood cell lysis solution, washed with PBS, and resuspended in 10% FCS-DMEM. The numbers of leukocytes and macrophages were determined by flow cytometry using monoclonal antibodies reacting to CD45 (BD Biosciences) and F4/80 (Serotec), respectively.

Total RNA from livers was extracted by TRIzol (Invitrogen). The following primers were used: G3PDH, 5′-ACCACAGTCCATGCCATCAC-3′ (sense) and 5′-TCCACCACCCTGTTGCTGTA-3′ (antisense); and syndecan-4, 5′-ATTCGAGAGACAGAGGTCATC-3′ (sense) and 5′-CTCTGAGGGGACACGGATGCC-3′ (antisense). Quantitative real-time PCR analysis of mRNA expression was also performed with LightCycler Fast Start DNA Master SYBR Green I Systems (Roche). The expression of mRNA was calculated by LightCycler software (version 3; Roche). Data were standardized by G3PDH.

Data are presented as means ± SEM and are representative of at least three independent experiments. Significant differences between experimental groups in the adhesion test, thrombin-resistance test, and ConA-induced hepatitis model were analyzed using the Student's t test. Differences were considered to be significant when P < 0.05 (*) or 0.005 (**).

Fig. S1 shows the structure of OPN. Fig. S2 shows the binding of heparin to the full-length form of OPN. Fig. S3 shows the structure of syndecan-4 and Syn4Ig. Fig. S4 shows the association between OPN and syndecan-4 in the supernatant of CHO transfectant cells. Fig. S5 shows the binding of Syn4Ig to the HBD of OPN. Fig. S6 shows the binding of OPN to plates in the presence or absence of Syn4Ig. Fig. S7 shows the specificity of the ELISA system for detection of the thrombin-cleaved form of mouse OPN.

This study was supported by a grant-in-aid from the Ministry of Education, Culture, Sports, Science and Technology of Japan to T. Uede (grant 16209014).

The authors have no conflicting financial interests.