Thrombin Cleavage of Osteopontin Disrupts a Pro-chemotactic Sequence for Dendritic Cells, Which Is Compensated by the Release of Its Pro-chemotactic C-terminal Fragment*

Background: Osteopontin is a chemotactic protein for dendritic cells without a clear definition of its functional domains. Results: A pro-chemotactic sequence, 168RSKSKKFRR176, spanning the thrombin cleavage site at Arg168 was identified. The C-terminal fragment released by thrombin cleavage assumes a new conformation. Conclusion: Thrombin cleavage of osteopontin disrupts the pro-chemotactic sequence, RSKSKKFRR. Significance: Thrombin cleavage of osteopontin modulates dendritic cell migration. Thrombin cleavage alters the function of osteopontin (OPN) by exposing an integrin binding site and releasing a chemotactic C-terminal fragment. Here, we examined thrombin cleavage of OPN in the context of dendritic cell (DC) migration to define its functional domains. Full-length OPN (OPN-FL), thrombin-cleaved N-terminal fragment (OPN-R), thrombin- and carboxypeptidase B2-double-cleaved N-terminal fragment (OPN-L), and C-terminal fragment (OPN-CTF) did not have intrinsic chemotactic activity, but all potentiated CCL21-induced DC migration. OPN-FL possessed the highest potency, whereas OPNRAA-FL had substantially less activity, indicating the importance of RGD. We identified a conserved 168RSKSKKFRR176 sequence on OPN-FL that spans the thrombin cleavage site, and it demonstrated potent pro-chemotactic effects on CCL21-induced DC migration. OPN-FLR168A had reduced activity, and the double mutant OPNRAA-FLR168A had even lower activity, indicating that these functional domains accounted for most of the pro-chemotactic activity of OPN-FL. OPN-CTF also possessed substantial pro-chemotactic activity, which was fully expressed upon thrombin cleavage and its release from the intact protein, because OPN-CTF was substantially more active than OPNRAA-FLR168A containing the OPN-CTF sequence within the intact protein. OPN-R and OPN-L possessed similar potency, indicating that the newly exposed C-terminal SVVYGLR sequence in OPN-R was not involved in the pro-chemotactic effect. OPN-FL and OPN-CTF did not directly bind to the CD44 standard form or CD44v6. In conclusion, thrombin cleavage of OPN disrupts a pro-chemotactic sequence in intact OPN, and its loss of pro-chemotactic activity is compensated by the release of OPN-CTF, which assumes a new conformation and possesses substantial activity in enhancing chemokine-induced migration of DCs.

cellular signaling. Adjacent to this RGD domain is a thrombin cleavage site at Arg 168 2Ser 169 . Thrombin cleavage of fulllength OPN (OPN-FL) leads to OPN-Arg (OPN-R), with a newly exposed C-terminal sequence, 162 SVVYGLR 168 , that binds to ␣4␤1 and ␣9␤1 integrins, allowing RGD-independent cell adhesion (8 -10) and a survival advantage to cells (11). Carboxypeptidase N (CPN) or B2 (CPB2) can remove Arg 168 from OPN-R, converting it to OPN-Leu (OPN-L), inactivating the binding site for integrin ␣ 9 ␤ 1 (12). In addition to these two binding sites, a putative CD44 binding domain is located within the last 18 amino acids at the highly conserved C terminus (13). Another highly conserved sequence that spans the thrombin cleavage site, 165 YGLRSKSKKFRR 174 , binds to the heparan sulfate on syndecan-4, protecting OPN-FL from cleavage by thrombin (14).
Both thrombin generation and OPN expression are enhanced during inflammation. Thrombin cleavage alters OPN function, allowing better accessibility to RGD by integrin ␣5␤3 (15) and to SVVYGLR by ␣9␤1 and ␣4␤1 integrins (10). However, the integrity of OPN appears to be necessary for certain functions of OPN. Phosphorylation of OPN at the C-terminal region inhibits RGD-mediated cell adhesion to OPN-FL (16), and antibody blocking of C-terminal Lys 298 -Asn 314 also inhibits OPN-FL binding to integrin-expressing cells (17). Furthermore, following thrombin cleavage, the OPN C-terminal fragment (OPN-CTF), Ser 169 -Asn 314 , is chemotactic for macrophages (18,19). Similarly, a 5-kDa fragment (Leu 167 -Asp 210 ) released by matrix metalloprotease (MMP)-9 cleavage, which overlaps the N-terminal sequence of OPN-CTF, induces migration of hepatocellular carcinoma cells (20). Although the OPN C-terminal half sequence is well conserved between species, the biochemical properties, pathophysiological role, and fate of OPN-CTF after thrombin cleavage have not been fully defined.
Upon antigen uptake and activation, DCs migrate to lymph nodes, a process primarily driven by engagement of chemokines CCL19 and CCL21 with their common receptor CCR7 (21). Because both DCs (6,22) and T cells (23) are sources of OPN that may play a role in DC migration, in this study, we investigated the impact of thrombin cleavage of OPN on DC migration to identify the functional domains of OPN. We identified a new pro-chemotactic domain in OPN-FL spanning its thrombin cleavage site. Thrombin cleavage disrupts its integrity, leading to a significant reduction in its pro-chemotactic function, but this is compensated by the released OPN-CTF that may assume a new conformation and possesses substantial pro-chemotactic activity.
Mice-OPN Ϫ/Ϫ mice on a C57BL/6J background were obtained from Jackson Laboratories and bred at the Stanford University School of Medicine Veterinary Service Center. Wild type C57BL/6J mice were purchased from the Jackson Laboratory (Sacramento, CA). All mouse experiments were performed under protocols approved by the Stanford University School of Medicine Committee of Animal Research and were in accordance with National Institutes of Health guidelines.
Production of Recombinant Osteopontin-Recombinant human OPN-FL, OPN-R, and OPN-L and RGD-mutated OPN-FL, denoted as OPN RAA -FL, were generated as described previously (12). The cDNA encoding mature human OPN-FL and its RGD-mutated counterpart in which Arg 168 was replaced by alanine (OPN-FL R168A and OPN RAA -FL R168A , respectively) and OPN-CTF (amino acids Ser 169 -Asn 314 ) were inserted as C-terminal in-frame fusions to GST using the Escherichia coli expression vector pGEX-6P-3 (GE Healthcare). The mutated sequences were generated using site-directed mutagenesis (Agilent Technologies, Santa Clara CA). The cDNA encoding OPN-CTF was inserted into E. coli expression vector pET-24D (EMD Millipore, Billerica, MA) with a free N terminus and with a His 6 tag fused at the C terminus. The plasmids pOPN-FL R168A , pOPN RAA -FL R168A , pOPN-CTF, and pOPN-CTF His6 were transformed into One Shot Top10 Chemically Competent E. coli (Invitrogen) for plasmid expansion, purified with a Qiagen plasmid Midi kit (Germantown, MD), and sequenced by ELIM Biopharmaceuticals to verify the mutation (Hayward, CA). The purified plasmids were then transformed into One Shot BL21(DE3) pLysS Chemically Competent E. coli (Invitrogen). The procedures for large scale production and purification of OPN-FL R168A and OPN-CTF were similar to those used previously to produce other OPN forms (12). His 6 -tagged OPN-CTF His6 was purified from lysate of E. coli that overexpressed OPN-CTF His6 using a HisTrap FF column (GE Healthcare). Lysate was loaded in 40 mM imidazole in phosphate buffer (20 mM sodium phosphate, 0.5 M NaCl, pH 7.4), and bound protein was eluted with 500 mM imidazole in phosphate buffer (20 mM sodium phosphate, 0.5 M NaCl, pH 7.4). The N-terminal sequence of the OPN-CTF His6 was verified by Edman degradation. The schematic structure and sequences of recombinant OPN and OPN fragments used in this study are illustrated in Fig. 1A. The recombinant proteins were greater than 95% pure as judged by SDS-PAGE followed by Coomassie Blue staining (Fig. 1B). The endotoxin levels in all protein samples were less than 2 EU/ml, as determined by the ToxinSensor Chromogenic LAL endotoxin assay kit (GenScript, Piscataway, NJ).
Preparation of Human and Mouse DCs-DCs were prepared from human peripheral blood mononuclear cells in buffy coats from 500 ml of whole blood obtained from the Stanford Blood Center. Briefly, 30 -35 ml of the buffy coat was layered over 15 ml of Ficoll-Paque PREMIUM separating solution (GE Healthcare). Mononuclear cells were separated by gradient centrifugation at 300 ϫ g for 30 min at room temperature. Peripheral blood mononuclear cells were collected and washed with HBSS before incubation with FcR-blocking reagent (Miltenyi Biotec, Auburn, CA) and anti-CD14-conjugated magnetic beads (Miltenyi Biotec). The CD14 ϩ cells were selected using an LS column on a MidiMACS separator. The CD14 ϩ cell fraction was cultured in 6-well or T-75 tissue culture-treated culture plates for 6 days at a density of 10 6 /ml in RPMI 1640 medium supplemented with 10% FBS, 100 ng/ml recombinant human GM-CSF (PeproTech, Rocky Hill, NJ), and 100 ng/ml recombinant human IL-4 (PeproTech) to obtain immature DCs. Immature DCs were matured by adding 100 ng/ml LPS from E. coli 0128:B12 (Sigma-Aldrich) to the cell culture on day 7, and mature DCs were harvested 18 h later. Mouse bone marrow cells from 10 -12-week-old C57BL/6 or OPN Ϫ/Ϫ male mice were grown in RPMI 1640 with 10% FBS and 100 ng/ml recombinant murine GM-CSF (PeproTech) for 6 days to derive bone marrow-derived DCs as described previously (25) before maturation with 100 ng/ml LPS for 18 h.
DC Migration Assay-Mature or immature DCs were harvested, washed three times with HBSS, and resuspended at 2 ϫ 10 6 /ml in RPMI 1640 medium with no phenol red, and 2 ϫ 10 5 cells in 100 l of RPMI 1640 were added to the upper chamber of a Transwell plate (Corning, Inc.; 6.5-mm diameter, 5.0-m pore size). The chemoattractant being tested was added to the lower chamber in 600 l of RPMI 1640. The cells were incubated at 37°C, 5% CO 2 for 4 h. The cells that migrated to the lower chamber of the Transwell chamber were determined , and thrombincleaved OPN C-terminal fragment with free N terminus and His 6 tag at its C terminus (OPN-CTF His6 ). B, recombinant osteopontin proteins were purified by affinity chromatography using a glutathione-Sepharose column. A total of 2 g of purified proteins were separated by SDS-PAGE and stained by Coomassie Brilliant Blue.
using a cell counting kit (Dojindo, Japan), and absorbance was read at 450 nm. The signals from medium alone were averaged and subtracted as background. The chemotactic index, a normalized measure of the chemotaxis, was calculated as (number of migrated cells)/(number of cells that migrated to CCL21 or CCL19).
Surface Plasmon Resonance-Binding of MAB197P, CD44-Fc, and CD44v6 with OPN was measured using the Bio-Rad ProteOn XPR36 system (Bio-Rad) and GLM chips. The chip was activated with 40 mM 1-ethyl-3-(3-dimethyl aminopropyl) carbodiimide and 10 mM N-hydroxysuccinimide injected at a flow rate of 30 l/min for 120 s immediately after mixing. The antibody MAB197P, CD44-Fc, or CD44v6 protein molecules were immobilized onto a GLM chip through amine coupling to the carboxyl groups on the chip. Briefly, 10 -25 g/ml MAB197P and 10 g/ml CD44-Fc and CD44v6 in 10 mM acetate (pH 4.0) were injected onto the chip surface at a flow rate of 30 l/min for 60 s. The chip surface was then deactivated with 1 M ethanolamine HCl. OPN-FL, OPN-CTF, OPN RAA -FL R168A , or OPN-CTF His6 in PBS containing 0.005% Tween 20 (pH 7.4) or buffer containing 2 mM Ca 2ϩ , 2 mM Mg 2ϩ , 50 mM Tris, and 100 mM NaCl (pH 7.4) at concentrations from 10 to 1000 nM was passed over the chip surface at a rate of 50 l/min for 180 s. The antibody IM7 (pan-CD44) at concentrations from 0.62 to 50 nM was injected to measure its binding to immobilized CD44-Fc or CD44v6 as a positive control. All data were corrected for both an interspot reference where samples were passed over a chip surface without any immobilized protein and a row reference where PBS-Tween buffer was passed over a chip surface with immobilized MAB197P, CD44-Fc, or CD44v6. Binding data were analyzed using ProteOn Manager software.
Measurement of Calcium Transients-DCs were incubated in HBSS (Ca 2ϩ , Mg 2ϩ ) with 5 g/ml Fluo-8/AM (AAT Bioquest, Sunnyvale, CA), 2.5 mM probenecid (AAT Bioquest), 0.04% pluronic F127 (AAT Bioquest) at 37°C for 30 min to load the calcium indicator, Fluo-8. Washed cells were then suspended in HBSS (Ca 2ϩ , Mg 2ϩ ), and 100 l was plated in Nunclon 96-well microwell plates (Thermo Scientific) at a density of 1.25 ϫ 10 5 /well. Baseline green fluorescence (excitation, 485 nm; emission, 538) was recorded, followed by the addition of CCL21 or premixed CCL21-OPN. The green fluorescence was recorded over 3 min immediately to monitor the calcium transient. The amplitude of calcium transient, denoted as ⌬E 538 nm , was calculated by subtracting the baseline fluorescence from the maximum values of the green fluorescence.
Statistical Analysis-Data were analyzed with Prism version 6.02 for Windows. Student's t test was used to compare two groups, and analysis of variance with post hoc Dunnet multiple comparison was used for comparison of three or more groups. Error bars show 95% confidence interval (CI). p Ͻ 0.05 was considered statistically significant. The sigmoidal dose response model in Prism version 6.02 was used for curve fitting in experiments to determine dose dependence. EC 50 values are presented as mean with CI in parenthesis.

Thrombin Cleavage of Osteopontin Modulates Chemotaxis
Neither OPN nor Its Fragments Induce Migration of Mature DCs-Because OPN had previously been shown to induce mouse DC (3) and macrophage (19) chemotaxis, we first tested the potency of OPN in inducing human DC migration in comparison with the lymphatic chemokine CCL21 in a transmigration assay. To our surprise, mature DCs did not migrate toward OPN-FL, OPN RAA -FL, OPN-R, and OPN-L (Fig. 3A). To exclude the possibility that OPN caused DCs to adhere to the polycarbonate membrane and not transmigrate into the lower wells, we used 0.025% trypsin EDTA to detach any DCs from the lower surface of the insert membrane but did not observe any cells that became detached. In contrast, DCs readily migrated to the lower chamber when it contained CCL21 (4.2 nM), with ϳ10% of the input cells in the lower chamber at 4 h.
Enhancement of CCL21-induced DC Migration Involves the RGD Domain-Although OPN and its fragments by themselves are not chemotactic for DCs, OPN-FL could augment the effect of CCL21 on DC migration, with an EC 50 of 358 nM (CI, 248 -518 nM) (Fig. 3B). Based on that EC 50 , OPN fragments were tested at 100 nM for their effects on CCL21-induced DC migration. Compared with OPN-FL, OPN RAA -FL had a significantly reduced potentiation effect (p Ͻ 0.0001) (Fig. 3C), indicating the importance of the RGD domain. Similarly, OPN-R and OPN-L also augmented CCL21-induced DC migration (Fig.  3C), but the extent of augmentation was substantially less than that of OPN-FL. There was no significant difference between OPN-R and OPN-L, suggesting that the newly exposed C-terminal SVVYGLR binding site in OPN-R does not play a role in this potentiation of chemotaxis. The reduction in the potency of OPN-R versus OPN-FL suggests that in OPN-FL, the augmentation might also be mediated via the C-terminal fragment. To demonstrate that the potentiating effect was not CCL21specific, we tested CCL19, another lymphatic chemokine that binds to CCR7, and found that CCL19-induced DC migration was also augmented by OPN and its fragments in a pattern similar to that observed with CCL21 (Fig. 3D).
OPN-CTF Enhances CCL21-induced DC Migration-To investigate the role of OPN-CTF in CCL21-induced DC migration, recombinant OPN-CTF (Ser 169 -Asn 314 ) was expressed FIGURE 3. OPN-FL and thrombin-cleaved N-terminal fragments enhance DC migration toward lymphatic chemokines CCL21 and CCL19. DCs were added to the upper chamber of a Transwell plate. CCL21 (50 ng/ml or 4.2 nM) and/or OPN (100 nM) was added into serum-free culture medium in the lower chamber to create a concentration gradient. A, DC migration to OPN or thrombin-cleaved OPN fragments was compared with migration of DCs to CCL21 (n ϭ 9 -24; data pooled from seven independent experiments; *, p Ͻ 0.0001, CCL21 versus each OPN). B, DC migration to CCL21 in the presence of various concentrations of OPN-FL (n ϭ 3-6). The level of migration observed in CCL21 alone was subtracted for curve fitting. C, DC migration to 50 ng/ml CCL21 in the presence of 100 nM OPN was compared with migration to CCL21 alone (n ϭ 6 -20; data pooled from six independent experiments). D, DC migration to 25 ng/ml CCL19 in the presence of 100 nM OPN was compared with migration to CCL19 alone (n ϭ 6; data are representative of three independent experiments). In all experiments, the value from medium alone was subtracted as background. Migrated cells in the lower chamber of the Transwell were normalized relative to cells migrating to CCL21 or CCL19. Data are shown as mean with CI (error bars). and purified (Fig. 1B). Similar to OPN-FL, OPN-CTF by itself did not induce DC migration but potentiated the CCL21-induced chemotaxis with an EC 50 of 326 nM (CI, 139 -778 nM) (Fig. 4, A and B). The potency is in the same range as that of OPN-FL. In the presence of 100 nM OPN-CTF, the CCL21 dose-response curve was shifted to the left without increasing the maximum effect of CCL21 (Fig. 4C), suggesting that OPN-CTF increased the potency of CCL21 but not its efficacy. Interestingly, when both OPN-CTF and OPN-R were added at the same molar concentration to the lower wells, no additive effect was observed (Fig. 4A). These data strongly suggested that the integrity of OPN-FL is necessary for its maximum effect. In order to identify the functional domain(s) on OPN-CTF that is responsible for its potentiating effect, we synthesized eight overlapping peptides based on the OPN-CTF sequence that are least 25 amino acids long and have at least a 5-amino acid overlap with the next peptide (Table 1). These were tested for their ability to antagonize DC migration in the presence of OPN-CTF and CCL21 as well as to promote DC migration to CCL21. In the antagonism experiment, none of the eight peptides inhibited the potentiating effect of OPN-CTF on CCL21-induced DC migration (data not shown). Next we tested whether any of the peptides could augment CCL21-induced migration of DCs. Peptide P0 (Ser 162 -Arg 176 ), which overlapped the thrombin cleavage site and contained SVVYGLR and the first 8 N-terminal amino acids of the OPN-CTF, markedly enhanced CCL21-induced cell migration. P1 (Ser 169 -Met 193 ) and P7 (Pro 289 -Asn 314 ), located at the N and C termini of OPN-CTF, respectively, also augmented CCL21-induced migration of DCs, although they were much less potent than P0 (Fig. 4D).
SVVYGLRKSKKFRR Is a Pro-chemotactic Sequence within OPN-FL-Because of the markedly higher potency of P0 (Ser 162 -Arg 176 ) than P1 (Ser 169 -Met 193 ), we hypothesized that the sequence spanning the thrombin cleavage site,  Table 1 for peptide sequences; n ϭ 6 -17; data pooled from three independent experiments). Data are shown as mean with CI (error bars). *, p Ͻ 0.05; **, p Ͻ 0.01; ****, p Ͻ 0.0001.  5A). Similarly peptide P0 is significantly more active than the mutant peptide P0 R168A containing the same substitution of Ala for Arg 168 (Fig. 5B). These data confirmed that Arg 168 is a key residue for the augmentation of CCL21-induced DC migration by OPN-FL. We further examined other basic amino acids present in the peptide P0 by testing two more peptides, P0 K170A and P0 K172A/K173A , in which basic amino acids Lys 170 , Lys 172 , and Lys 173 were replaced by alanine. Both mutants demonstrated significantly less potency than P0, as shown in a dosedependent study of DC migration in response to CCL21 (Fig.  5B). Additionally, P0 reverse , which had the sequence of P0 reversed but possesses the same charge, was significantly less potent than P0, indicating that the effect of P0 was sequencespecific and was not just due to its high pI. None of the synthetic peptides induced DC migration by themselves (data not shown). Taken together, these results indicate that R168A plays a key role in the potentiating effect of OPN-FL and P0.
Thrombin Cleavage Reduces the Pro-chemotactic Activity of OPN-FL by Disrupting the Sequence RSKSKKFRR-Because thrombin cleavage of OPN disrupted the sequence of P0, 152 SVVYGLRSKSKKFRR 176 , we designed additional peptides to study the consequence of thrombin cleavage. Peptide P S169-T185 , corresponding to the first 17 N-terminal amino acids in thrombin-cleaved OPN-CTF, had a modest potentiating effect on CCL21-induced DC migration compared with its scrambled sequence, whereas P R168-T185 demonstrated the same potency as P0, as shown in a dose-dependent study (Fig. 5C). These data confirmed the key role of Arg 168 in this pro-chemotactic sequence and excluded the involvement of SVVYGLR.
OPN-CTF Manifests Its Full Pro-chemotactic Activity upon Thrombin Cleavage and Release from OPN-FL-To assess the relative role of RGD, Arg 168 , and the C-terminal fragment (Ser 169 -Asn 314 ), we produced a double mutant protein, OPN RAA -FL R168A . The double mutant OPN had markedly reduced pro-chemotactic activity, and strikingly, on an equimolar basis, it was much less active than OPN-CTF (Fig.  6A), suggesting that the full pro-chemotactic activity of the C-terminal fragment may be expressed only upon thrombin cleavage and its release from the intact protein, resulting in an alteration of its conformation. That the pro-chemotactic activity of OPN-CTF is conformation-dependent is also suggested by the finding that the addition of a His 6 tag to the C terminus of OPN-CTF markedly reduced its activity (Fig. 6B). To further demonstrate this, we measured the binding of an OPN-CTFspecific monoclonal antibody, MAB197P, to OPN-FL, OPN-CTF, OPN RAA -FL R168A , and OPN-CTF His6 , which all contain the OPN-CTF sequence but differ in their C-or N-terminal modifications. The association rate constants (K a ) for MAB197P binding to all four recombinant OPN proteins were similar ( Fig. 6C and Table 2). However, the disassociation rate constants (K d ) and equilibrium state constants (K D ) of MAP197P binding to OPN-CTF were about 6 times lower than its binding to OPN-FL and OPN-CTF His6 . Interestingly, binding of MAB197P to OPN RAA -FL R168A was essentially irreversible because there was a Ͼ10-order of magnitude decrease in K d and K D when compared with its binding to the other forms of OPN. The difference in binding kinetics of MAB197P to these different OPN proteins revealed that OPN-CTF exists in different conformations, depending on its adjacent N-or C-terminal sequences.

OPN-FL and OPN-CTF Exert Maximum Effect When Present in Physical
Proximity to CCL21-To test whether the effect of OPN-FL and OPN-CTF requires a concentration gradient, similar to CC chemokines, we assessed CCL21-induced DC migration when OPN was present in the upper well, lower well, or both in the cell migration assay. When OPN-FL or OPN-CTF was present in the upper well, no enhancement of CCL21induced migration was observed relative to CCL21 alone in the lower well (Fig. 7, A and B). When OPN-FL or OPN-CTF was present both in upper and lower wells, the enhancement was similar to that of CCL21 plus OPN-FL or OPN-CTF in the lower well alone (Fig. 7, A and B). These data suggest that a  Table 1 for peptide sequences; n ϭ 10; data are representative of three independent experiments). C, DC migration to 50 ng/ml CCL21 was measured in the presence of various concentrations of P0, P R168-T185 , P S169-T185 , and P S168-T185-scrambled (see Table 1 for peptide sequences; n ϭ 9 -10; data are representative of two independent experiments). Data are shown as mean with CI (error bars).
concentration gradient is not required, and both OPN-FL and OPN-CTF need to be present in physical proximity to CCL21 to enhance chemotaxis. We then assessed the kinetics of OPN-FL and OPN-CTF on CCL21-induced DC migration. Although pretreatment with OPN-FL or OPN-CTF slightly enhanced CCL21-induced DC migration, the degree of enhancement was markedly less than that observed if OPN was constantly present in the lower wells during the experiment (Fig. 7, C and D).

OPN-FL and OPN-CTF Do Not Alter the Amplitude of CCL21-induced Calcium Transients in
DCs-Activation of CCR7, a G i -coupled G protein-coupled receptor, induces an intracellular calcium transient, whose amplitude is associated with the concentration of the ligand. Therefore, we assessed whether OPN-FL or OPN-CTF alters the amplitude of the CCL21-induced calcium transient in DCs. CCL21 induced intracellular calcium transients in a dose-dependent fashion with an EC 50 of 3.56 nM (CI, 2.35-5.40 nM) (Fig. 8A). We then   (Fig. 8B).

OPN-FL and OPN-CTF Do Not
Bind to CD44s and CD44v6 -The standard form of CD44 (CD44s) and its variant form CD44v6 have been implicated as OPN receptors in experiments in which the effects of OPN interactions with cells were inhibited by anti-CD44 antibodies (3,18,19,27). We tried to  block DC migration using anti-CD44 antibodies. We noticed, however, that anti-CD44s or anti-CD44v6 antibodies antagonized CCL21-induced DC migration in the absence of OPN. The anti-pan-CD44 antibody clone IM7 completely abolished DC motility because the migration observed in its presence was significantly less than that found in medium alone (Fig. 9). These results suggest that the anti-CD44 antibodies had a direct inhibitory effect on DC migration, which is not mediated via blocking of CD44 interaction with OPN.
To further investigate whether OPN or its thrombin-cleaved fragments interacted with one of the CD44 variants, we measured the binding of OPN-FL and OPN-CTF to CD44s by an ELISA binding assay in which protein binding to immobilized CD44s was measured using anti-GST. No binding was detected between CD44s and OPN either in the presence or absence of divalent cations (data not shown). Using surface plasmon resonance, we also did not detect any measurable binding of OPN-FL and OPN-CTF to immobilized CD44s-His 6 or CD44v6 ectodomain in the presence or absence of divalent cations, whereas the control consisting of anti-CD44 (IM7) bind-ing to the immobilized CD44 proteins was positive (data not shown).
OPN Augments Chemokine-induced Chemotaxis of Murine DCs-Because the thrombin cleavage site is conserved between species (11), we tested whether the augmentation of chemokine-induced DC migration also applies to other species and chemokines. To do so, we examined the effect of human OPN and its fragments on augmenting mouse bone-marrow-derived DCs (BMDCs) migrating to CCL21 and immature BMDC migration to CCL3. We observed a similar pattern of potentiating effects by OPN in BMDCs because OPN-FL and its fragments by themselves did not induce mouse DC migration. Instead, OPN-FL augmented both CCL21-and CCL3-induced DC migration with a significant reduction when OPN-R or OPN-L was tested (Fig. 10, A and B). This is consistent with the suggestion that OPN enhances all chemokine-induced DC migration. Mature mouse BMDCs from OPN-deficient mice responded to exogenous added OPN in a similar pattern (Fig.  10C), indicating that the augmentation was independent of DC production of OPN and intracellular OPN.

DISCUSSION
In this study, we demonstrated the role of OPN as an enhancer of CC-chemokine-induced DC migration while not being a chemoattractant by itself. When DCs are activated, a rapid up-regulation of chemokine receptor CCR7 and downregulation of other chemokine receptors dictate its migration exclusively to draining lymph nodes. The CCR7 ligands CCL19 and CCL21 are exclusively expressed by stromal cells in lymph nodes and the high endothelial venules of the afferent lymphatics, where they contribute to the chemokine gradient. OPN is present at a high level at sites of inflammation where antigen uptake and DC activation occur. Our study suggests that although OPN by itself is insufficient to induce directional migration of DCs to lymph nodes, its presence in the milieu provides an amplification mechanism for chemokine-induced G protein-coupled receptor-mediated migration. The same properties were found in DCs derived from both wild type and OPN-deficient mice, showing that production of endogenous OPN by the DCs was not necessary for the enhancement of FIGURE 9. Anti-CD44 and CD44v6 antibodies blocked 50 ng/ml CCL21induced migration in the absence of OPN. The cells were treated with 10 g/ml of antibody for 15 min before adding them to the upper chamber. Antibody (10 g/ml) was added to the lower chamber of the Transwell plate (n ϭ 8 -14; data pooled from two independent experiments). Data are shown as mean with CI (error bars). *, p Ͻ 0.05; ****, p Ͻ 0.0001.  8). B, WT immature BMDC migration to CCL3 (50 ng/ml or 6.4 nM) supplemented with different OPN forms was compared with migration to CCL3 alone (n ϭ 3). C, mature OPN Ϫ/Ϫ BMDC migration to 100 ng/ml CCL21 supplemented with different OPN forms was compared with migration to CCL21 alone (n ϭ 3). Data are shown as mean with CI (error bars). *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001; ****, p Ͻ 0.0001. migration. The phenomenon was also observed in immature DCs responding to chemokines, such as CCL3, suggesting that OPN signaling may enhance leukocyte migration induced by multiple CC chemokines.
We identified three functional domains involved in the potentiation of DC migration to CCL21 by OPN: the well defined RGD domain present in OPN-FL that can potentially bind to integrin ␣ v ␤ 1 , ␣ v ␤ 3 , ␣ v ␤ 5 , ␣ 5 ␤ 1 , and ␣ 8 ␤ 1 ; a new prochemotactic sequence, 168 RSKSKKFRR 176 , that spans the thrombin cleavage site Arg 168 2Ser 169 in OPN-FL; and a third functional domain within the OPN-CTF (Ser 169 -Asn 314 ). There was no difference in potency between OPN-R and OPN-L, and thus SVVYGLR is not required for the enhancement of chemotaxis, thereby excluding the involvement of ␣ 4 ␤ 1 and ␣ 9 ␤ 1 integrins in this process. Importantly, OPN-FL is the most active protein in potentiating DC migration to CCL21, indicating that its integrity is critical for the maximal prochemotactic effect because both OPN-R and OPN-CTF are less potent than OPN-FL. The presence of both OPN-R and OPN-CTF in the medium, mimicking the results of thrombin cleavage, did not recover the full potency of OPN-FL, highlighting the importance of the integrity of the newly identified sequence, 168 RSKSKKFRR 176 , that spans the thrombin cleavage site. Within this sequence, Arg 168 is critical for its full activity because the peptide, P R168-T185, demonstrated the same potency as P0 (Ser 162 -Arg 176 ), whereas P S169-T185 had markedly reduced activity. MMP-3, -7, and -9 cleave OPN at Gly 166 2Leu 167 (20,28), with the release of a 5-kDa fragment (Leu 167 -Asp 210 ) that has been reported to induce chemotaxis of hepatocellular carcinoma cells (20). This fragment contains the newly described pro-chemotactic site. However, a shorter sequence was not identified within this 5-kDa fragment. The receptor site for 168 RSKSK-KFRR 176 on DC remains to be defined.
OPN-CTF (Ser 169 -Asn 314 ), released from OPN-FL upon thrombin cleavage, clearly enhances chemokine-induced DC migration. Peptide P1 (Ser 169 -Met 193 ) has a modest potentiating effect compared with P0 (Ser 162 -Arg 176 ), which is almost certainly due to the loss of the critical Arg 168 . P7 (Pro 289 -Asn 314 ) also has a modest enhancement of chemokine-induced DC migration, but it is far less potent than OPN-CTF. The activity of OPN-CTF appears to be dependent on its conformation because it needs to be released from OPN-FL, and thus it is likely that its full activity is not captured by the individual peptides. This is supported by the evidence that, on an equimolar basis, OPN-CTF is significantly more potent than OPN RAA -FL R168A , which contains intact OPN-CTF sequence but differs from OPN-CTF in the N-terminal residues. Moreover, the C-terminal His 6 tag significantly reduced the potency of OPN-CTF. In that context, positively charged His 6 tag may bind to a negatively charged domain (P3) within OPN-CTF, thereby distorting the requisite conformation required for the optimal expression of its pro-chemotactic activity. Moreover, MAB197B demonstrates drastically different kinetics when binding to OPN-FL, OPN-CTF, OPN RAA -FL-R 168A , and OPN-CTF His6 , consistent with the notion that OPN-CTF may exist in different conformations, depending on its adjacent N-terminal sequence or C-terminal modifications. The precise pro-che-motactic domain and mechanism of the OPN-CTF pro-chemotactic activity remain to be further defined.
The enhancement of CCL21-induced migrations of DCs by OPN-FL or OPN-CTF does not require a concentration gradient of OPN but requires its presence in physical proximity to CCL21 during migration. This confirms that CCL21 is the main driver for DC migration in this experimental setting. Chemokine-induced DC migration is driven by asymmetrical signaling through its G i -coupled G protein-coupled receptor at the leading edge, resulting in polarized cytoskeleton reorganization. In this context, if OPN serves as an amplifier to chemokine signaling, a localized OPN signaling through an unknown receptor in proximity to the leading edge may be required. Alternatively, OPN and CCL21 may form a complex, which then facilitates the engagement of CCL21 to CCR7. However, in the presence of OPN, CCL21 does not induce a stronger calcium transient than CCL21 alone. The exact mechanism(s) by which OPN enhances chemokine-induced DC migration remains to be clarified in future studies.
It is interesting that thrombin cleavage of OPN, leading to the generation of OPN-R and its subsequent cleavage by CPB2 to OPN-L, has been previously shown to modulate cell adhesion in Jurkat cells (12), cultured synoviocytes (2,29), and some glioblastoma cell lines (11). These effects are mediated by the exposure of the SVVYGLR sequence in OPN-R and the subsequent removal of the C-terminal Arg 168 in OPN-L. Elevated OPN-R and OPN-L levels have been found in inflammatory diseases (29). In contrast, thrombin cleavage leads to a diminished, rather than enhanced, augmentation effect of OPN on chemokine-induced DC migration, due to disruption of the sequence 168 RSKSKKFRR 176 , which spans the cleavage site, and the critical role of Arg 168 . In this regard, the sequence spanning the thrombin cleavage site, 165 YGLRSKSKKFRR 174 , has been shown to bind to heparan sulfate in syndecan-4, which then protects it from enzymatic cleavage (15). It is possible that at the initiation of inflammation, while DC is being activated, cell matrix proteoglycan-bound OPN will maintain its full-length form and exert its maximal potentiating effect on chemokineinduced DC migration despite the presence of thrombin. With further progression of inflammation and local generation of thrombin and CPB2, proteolytic cleavages of OPN will occur that modify its cellular inflammatory properties on DCs, monocytes, and macrophages. Although the DC chemotactic activity in OPN-FL is disrupted by thrombin cleavage, this is compensated by the release of OPN-CTF that assumes a new conformation and has substantial pro-chemotactic activity. Thus, the net effect of proteolytic processing of OPN and its fragments will depend on a complex interplay between the specific cell types involved and their surrounding extracellular milieu (Fig.  11).
CD44 and its variants have been identified as OPN receptors by the use of anti-CD44 antibodies (3,18,19,27), but here we did not detect any binding of OPN to CD44 or CD44v6 by two independent techniques. We also showed that anti-CD44 antibody had an inhibitory effect on cell migration by itself, irrespective of the type of chemokine or presence of OPN. Indeed, certain anti-CD44 antibodies, such as IM7, strongly inhibited baseline cell motility, as shown by a decreased non-directional migration to medium alone. The inhibitory effect of anti-CD44 antibodies on cell motility is consistent with the observation that injection of anti-CD44 and CD44v antibodies into an ex vivo ear specimen inhibits emigration of Langerhans cells to the medium (6). In contrast, cells transfected with CD44v6 or/and CD44v7 ectodomain bind to OPN as well as to its thrombincleaved N-and C-terminal fragments in a ␤1 integrin-dependent manner (18). It is possible that binding of the CD44-transfected cells to OPN may be controlled by an inside-out signaling pathway initiated from transfected CD44v6, leading to activation of ␤1 integrin (30). These results suggest that the observed binding of OPN might be mediated via activated integrins rather than CD44. In other experiments, similar to our data, no binding of OPN to CD44H (CD44, standard form), CD44E (CD44v8 -10), CD44v3, v8-v10 (CD44v3 plus v8 -10), or CD44v3 was detected (31). Additionally, CD44v3-6-transfected A3 cells do not bind to recombinant OPN either (31). The rat CD44v6 ectodomain used in our study has low homology to its human counterpart, but it has been shown to be active in regulating EGFR-2 in the human system (24). It is possible that the sequence differences in certain residues between the human and rat CD44v6 domain account for the rat CD44v6 not binding to human OPN in our experiments.
Previous studies on OPN and leukocyte migration utilizing murine macrophages suggested that OPN functions as a chemoattractant (19), in contrast to our results showing that OPN is not a chemoattractant itself for DCs. This does not rule out OPN being a direct chemoattractant for macrophages or the possibility that there is an indirect interaction of OPN with CD44 and its variant forms. Discrepancies between macrophages and DCs may be due to intrinsic differences between the two cell types or to the different experimental settings.
In summary, we found that OPN and its fragments did not induce DC migration by themselves but significantly potentiated CCL21-induced DC migration via the well defined RGD and a new pro-chemotactic sequence, 168 RSKSKKFRR 176 , that spans the thrombin cleavage site. Thrombin cleavage disrupts this site, but this loss of pro-chemotactic activity is compensated by the release of OPN-CTF that assumes a new conformation and possesses substantial activity in enhancing chemokine-induced migration of DCs. FIGURE 11. Thrombin cleavage of OPN modulates chemokine-induced DC migration. At the initiation of inflammation, while DC is being activated, cell matrix proteoglycan-bound OPN will maintain its full-length form and exert its maximal potentiating effect on chemokine-induced DC migration, mediated by the 159 RGD 161 and 168 RSKSKFRR 176 sequences, despite the presence of thrombin. With further progression of inflammation and local generation of thrombin and CPB2, thrombin cleavage of OPN occurs, which will increase cell adhesion by exposing SVVYGLR and allowing better access to RGD in OPN-R while reducing its potentiating effect on DC migration. Further cleavage of OPN-R by CPB2 (or CPN) inactivates SVVYGLR, converting it to OPN-L, and thus decreases cell adhesion mediated by ␣ 4 ␤ 1 and ␣ 9 ␤ 1 integrins. While thrombin cleavage disrupts the RSKSKKFRR pro-chemotactic domain in intact OPN, it releases OPN-CTF (Ser 169 -Asn 314 ), which assumes a new conformation and possesses substantial pro-chemotactic activity, compensating for the loss of the pro-chemotactic activity in OPN-R and OPN-L.