The synovial fluid (SF) cells of rheumatoid arthritis (RA) patients express a specific CD44 variant designated CD44vRA. Using a cellular model of this autoimmune disease, we show in this study that the mammalian lectin, galectin-8 (gal-8), is a novel high-affinity ligand of CD44vRA. By affinity chromatography, flow cytometry, and surface plasmon resonance, we demonstrate that gal-8 interacts with a high affinity (Kd, 6 × 10−9 M) with CD44vRA. We further demonstrate that SF cells from RA patients express and secrete gal-8, to a concentration of 25–65 nM, well within the concentration of gal-8 required to induce apoptosis of SF cells. We further show that not all gal-8 remains freely soluble in the SF and at least part forms triple complexes with CD44 and fibrinogen that can be detected, after fibrinogen immunoprecipitation, with Abs against fibrinogen, gal-8 and CD44. These triple complexes may therefore increase the inflammatory reaction by sequestering the soluble gal-8, thereby reducing its ability to induce apoptosis in the inflammatory cells. Our findings not only shed light on the receptor-ligand relationships between CD44 and gal-8, but also underline the biological significance of these interactions, which may affect the extent of the autoimmune inflammatory response in the SF of RA patients.

Using an autoimmune model of a joint inflammation, we have previously shown (1) that synovial fluid (SF)3 cells derived from the joints of rheumatoid arthritis (RA) patients express a specific CD44 variant designated CD44vRA. This variant is not expressed on peripheral blood leukocytes from the same patient or on SF cells of osteoarthritis patients. The CD44vRA variant of CD44 is an important modulator of inflammatory autoimmune responses. For example, CD44vRA presents, in an exclusive orientation, fibroblast growth factor (FGF)-2 to proinflammatory synovium cells of RA patients bearing the cognate FGF receptor-1. This allows enhanced proliferation of FGF-2-stimulated cells, thus aggravating the inflammatory cascade in these patients (1). In this study, we suggest an additional proinflammatory function for CD44vRA, related to its ability to ligate and sequester galectin-8 (gal-8), an anti-inflammatory proapoptotic protein.

The cell surface receptor CD44 has a wide spectrum of ligands involved in many physiological and pathological processes (2). The structural nature of human CD44 is related to alternative splicing of nine variant exons inserted in different combinations between two C regions, each containing five constant exons (2, 3, 4, 5). Theoretically, >800 CD44 variants (CD44v) can be generated by differential use of the variant exons (6), of which several dozen have already been detected (3). For example, splicing and insertion of exons v3 to v10 between the two constant exons, generates the molecule CD44v3-10, which is preferentially expressed on normal keratinocytes (7) and SF cells from RA patients (1). A comparison of the transcript sequence of CD44v3-10 derived from these two cell types revealed that the CD44 variant of SF cells from the majority of RA patients includes an extra trinucleotide, CAG, which is transcribed from the end of the intron flanking exon v5. This CD44 variant was designated CD44vRA, whereas CD44v3-10 is the wild-type molecule (1). Direct splicing from constant exon 5 to constant exon 16, which skips all the variant exons, generates the shortest and most common isoform-standard CD44 (CD44s), which is preferentially expressed on hemopoietic cells (3).

The variability in CD44 structure is further increased due to its N- and O-glycosylation as well as attachments of glycosaminoglycans (e.g., heparan sulfate, chondroitin sulfate) (2, 3, 4, 5). The multistructural nature of CD44 is presumably associated with its multifunctionality, which includes cell-cell and cell matrix interactions; support of cell migration; presentation of growth factors, cytokines, chemokines, and enzymes to other cells or to the surrounding tissues as well as signal delivery from the cell’s surface to its interior, leading to programmed cell death or cell survival and proliferation (2). The rich isoform repertoire of CD44 may explain its ability to bind a considerable number of different ligands, some of which are, presumably, yet to be discovered. The list of CD44 ligands is continuously expanding and so far includes hyaluronic acid, which is the principal ligand, as well as collagen, fibronectin, fibrinogen, laminin, chondroitin sulfate, mucosal vascular adhesion, serglycin/gp600, osteopontin, the MHC class II invariant chain (li), L-selectin, and E-selectin (2).

The interaction between the CD44 receptor and its ligands dictates the outcoming signaling, leading to defined physiological or pathological activities. Therefore, it is important not only to illuminate the signal transduction of specific ligand-CD44 interactions, but also to identify novel CD44 ligands that may be engaged in pathological functions and, consequently, might be used as therapeutic targets. In this study, we provide evidence that one such novel ligand is gal-8.

Gal-8 is a member of the galectin (animal lectin) family of molecules characterized by one or two carbohydrate recognition domains that interact with β-galactose-containing sugars (8). Galectins are externalized by an atypical secretory mechanism (9) to regulate cell growth, cell transformation, embryogenesis, and apoptosis (10, 11). In accordance with their proposed functions, galectins enhance or inhibit cell-matrix interactions (12, 13). Gal-8 contains two homologous carbohydrate recognition domains linked by a hinge peptide (14). Alternative splicing generates several isoforms of this molecule (15) that are expressed in a number of tissues (13). Upon secretion, gal-8 acts as a matrix protein equipotent to fibronectin in promoting cell adhesion by ligation and clustering of a selective subset of cell surface integrins (16, 17). Complex formation between gal-8 and integrins involves sugar-protein interactions and triggers integrin-mediated signaling cascades such as Tyr phosphorylation of focal adhesion kinase and paxillin, and a robust and sustained activation of the MAPK and PI3K pathways (17, 18). In contrast, when present as a soluble ligand, gal-8 negatively regulates cell adhesion, induces growth arrest and apoptosis. Such a mechanism allows local signals emitted by secreted gal-8 to specify territories available for cell adhesion and migration (13, 16, 17, 19).

Knowing that the interaction between gal-8 and integrins is mediated by the sugar moieties of the adhesion receptors and that this interaction can subsequently lead to apoptotic signaling, we predicted that gal-8 could also be ligated to CD44. This assumption was based on the fact that, like integrins, CD44 is a glycosylated cell surface receptor, involved in cell adhesion that can deliver death signals (2). Challenging this prediction experimentally, we not only confirmed the receptor-ligand relationship between CD44 and gal-8, but also used the inflammatory SF cells of RA patients as a model to demonstrate how high-affinity complex formation between gal-8 and CD44 may influence the extent of the autoimmune inflammation, which destroys essential components of the tissue.

Protein G-agarose beads and protein A-agarose beads were purchased from Sigma-Aldrich. Affinity-purified mAbs against gal-8 (106.1) were generated as described (17). Goat antiserum to human fibrinogen and human fibrinogen were purchased from MD Biomedicals. mAbs against p21/Cip1 and phosphotyrosine (PY-20) were obtained from BD Transduction Laboratories. Polyclonal JNK Abs were obtained from Santa Cruz Biotechnology. HRP-conjugated F(ab′)2 of goat anti-mouse IgG (H + L) was obtained from Jackson ImmunoResearch Laboratories. Polyclonal protein kinase B (PKB) (Akt) Abs and monoclonal phosphospecific JNK Abs were obtained from Sigma-Aldrich. Polyclonal phosphospecific (Ser473) PKB Abs were purchased from New England Biolabs. Hermes-3 was derived from American Type Culture Collection hybridomas and mAbs were purified in our laboratory. Anti-pan-CD44 mAb (IgG2b) was obtained from Serotec; anti-CD44v6 mAb (IgG1) was obtained from Bender MedSystem; 106 anti-gal-8 mAb (IgG) is described in Refs. 17 and 20 . Coomassie blue was purchased from Sigma-Aldrich.

SF was obtained from RA patients by arthrocentesis, aseptically aspirated, and transferred into tubes. At the same time, blood samples were obtained from the cubital vein. The patients fulfilled the RA definition of the American College of Rheumatology 1987 criteria. Collected blood (after separation by Lymphoprep) and SF samples were centrifuged at 800 × g for 15 min, divided into aliquots, and stored at −20°C until use. The protocol was approved by the institution’s ethics review committee.

The cloning and transfection of human CD44 cDNA and its transfection to Namalwa cells was described in our previous paper (1). The transfected Namalwa cells were designated Namalwa-CD44vRA, Namalwa-CD44v3-10, Namalwa-CD44s, and Namalwa-null (or just Namalwa) when the cells were transfected with empty vector pcDNA3.1 (Invitrogen Life Technologies).

Soluble CD44v3-10 cDNA was isolated from the total RNA of primary human keratinocytes by RT-PCR (PTC-100 Programmed Thermal Controller) and amplified, using two primers assigned from the published CD44 sequence: Ex1s, 5′-TATCTAGAGCCGCCACCATGGACAAGTTTTGGTGG-3′; Ex16/17as, 5′-TATCTAGAGCCATTCTGGAATTTGGGGTGT-3′.

Both primers contained an Xbal recognition site. The PCR products were digested with Xbal (New England Biolabs) and the pCXFc zeo-vector was digested with NheI (Boehringer-Mannheim). After digestion, the PCR products were ligated into the pCXFc zeovector to generate the CD44v3-10-Ig-Fc recombinant. Using the same protocol, soluble CD44vRA and soluble CD44s cDNAs were cloned from synovial cells of a RA patient. A quantity of 3 μg of each of the above Fc-containing plasmids was incubated for 20 min with 12 μl of FuGene (Roche Diagnostics). The mixture was added to 15-cm-diameter cell plates containing 70% confluent 293T (American Type Culture Collection) cells and the supernatant was collected after 48 and 72 h. The CD44-IgFc proteoglycans were purified on a protein-G column and their validity was assessed by sequencing (to exclude mutations), SDS-PAGE and immunoblotting with an Ab recognizing the constant epitope shared by all CD44 isoforms, designated anti- pan-CD44 (or anti-CD44s) mAb (Hermes-3; American Type Culture Collection).

The binding of these proteins to GST-gal-8 was performed as previously described in Refs. 14 and 17 .

The Western blot and flow cytometry analysis were performed as described in Ref. 1 .

The kinetic and equilibrium constants of the interaction between gal-8 and soluble CD44 were analyzed, using galectins or soluble CD44 isoforms (interactants) immobilized on sensorchips in a Biacore Biosensor system (Biacore 3000; Biacore). The sensorchip surface was activated with N′-hydroxysuccinimide and N-ethyl-N-(3-dimethylaminopropyl)carbodiimide hydrochloride (NHS/EDC), according to the manufacturer’s instructions. Interactants in coupling buffer (10 mM acetate (pH 4)) were injected over the sensorchip surface to obtain immobilization levels of 10,000 resonance units (RU) with gal-8 (1 mg/ml), and of 1000 RU with CD44vRA attached onto protein A (0.7 mg/ml). Deactivation of the remaining activated groups was performed using 100 mM ethanolamine. The immobilized interactant surfaces were washed two times with 10 mM NaOH for 1 min to eliminate proteins that were not covalently bound. In all Biacore experiments, HEPES-buffered saline supplemented with 0.005% surfactant p20 served as the running buffer. The tested proteins were passed at various concentrations over the sensorchip surface at a flow rate of 20 μl/min. The interaction was monitored for changes in surface plasmon resonance response at 25°C. Galectin-immobilized surfaces were regenerated for a new binding analysis by injecting 100 mM lactose. The reference channel was produced by activating and deactivating the chip surface in the absence of coupling protein. The primary data were analyzed using BIAevaluation 3.1 software and applying a Langmuir binding model (stoichiometry, 1:1) to calculate Kon (association rate constant, M−1 s−1), Koff (dissociation rate constant, s−1), and Kd (equilibrium constant) for the interaction of gal-8 with different CD44 isoforms. Identical conditions were used for measuring the affinity of fibrinogen to gal-8.

RNA was extracted from SF cells of RA patients, using the RNA-BEE reagent (RNA isolation solvent; Tel-Test) according to the manufacturer’s instructions. Reverse transcription was performed using 5 U of avian myeloblastosis virus reverse transcriptase (Promega) in a 20 μl reaction volume containing 50 mM Tris-HCl (pH 8.3), 50 mM KCl, 10 mM MgCl2, 10 mM DTT, 20 U of RNasin (Promega), 500 ng of RNA, and 100 ng of oligo(dT) primer (Promega). After incubating the reaction samples for 1 h at 41°C, the reverse transcriptase was inactivated by heating the mixture for 10 min at 65°C. Amplification was performed in a microprocessor (PTC-100 programmable Thermal Controller), using 0.5 μl of the reverse transcriptase reaction product (cDNA) in a final volume of 50 mM KCl, 1.5 mM MgCl2, 10 mM Tris-HCl (pH 9), 250 μM dNTPs, and 2.5 U of TaqDNA polymerase (Promega). To amplify the gal-8 cDNA, the following primers were added to the reaction mix: 5′-sense, AAAGTTGGCACCATTCCCGATCAG; 5′-antisense, AAACCAGCTCCTTACTTCCAGTAA. Gal-8 amplification was conducted for 30 cycles with denaturation at 94°C for 1 min, annealing at 53°C for 1 min, and extension at 68°C for 2 min, followed by 10-min final extension at 72°C. The amplified products were resolved on 1.5% agarose gel. Determination of the cellular gal-8 transcript was based on the position of the band in relation to the marker ladder, and on the expected base pair size.

SF from the joints of RA patients was incubated for 4 h at 4°C with 100 μg of Hermes-3 anti-CD44 mAb or isotype-matched control mAb (IgG2a) directed against MHC class I (derived from the W6/32 American Type Culture Collection clone) followed by an additional incubation for 16 h at 4°C with 150 μl of protein A-agarose beads. Samples of the same SF were also directly incubated for 16 h at 4°C with 150 μl of lactosyl sepharose (LS) (to trap gal-8). The complexes were washed four times with Net gel buffer (50 mM HEPES, 150 mM NaCl, 0.3% Triton X-100, and 10% glycerol (pH 7.5)), and mixed with 150 μl of Laemmli sample buffer, boiled for 15 min, resolved on 10% SDS-PAGE, and immunoblotted with anti-gal-8 or Hermes-3 anti-pan-CD44 mAb as described above (see Western blot analysis and flow cytometry). Fibrinogen/r-gal-8 coprecipitation was performed as follows: a total of 100 μg of anti-fibrinogen Ab was incubated with 150 μl of protein G-agarose beads for 2 h at 4°C. After extensive washing, protein G-agarose beads coupled to anti-fibrinogen Ab were incubated with 50 μg of fibrinogen or PBS (as a negative control) for 2 h at 4°C. After extensive washing, protein G-agarose beads bound to fibrinogen through anti-fibrinogen Ab were incubated with 50 μg of r-gal-8 for 16 h at 4°C. The complexes were treated as described above and immunoblotted with anti-gal-8. Fibrinogen, gal-8, and CD44 coprecipitation was performed as follows: after extensive washing, protein G-agarose beads coupled to anti-fibrinogen Ab, generated as shown above, were incubated with 5 ml of SFs of RA patients for 16 h at 4°C. Samples of the same SFs were also directly incubated for 16 h at 4°C with 150 μl of protein G-agarose beads (as negative control). The complexes were treated as described above and immunoblotted with anti-gal-8, Hermes-3 anti-pan-CD44 mAb, or with anti-fibrinogen Ab.

Apoptosis was induced by culturing 2 × 106 SF cells from RA patients, with different concentrations of gal-8 diluted in serum-free RPMI 1640 for 16 h at 37°C in a CO2 incubator. RA SF cells incubated with 5 μg/ml doxorubicin served as positive control. A portion of the cells was coincubated with 5 μg of soluble CD44vRA, as shown in Results. The cells were washed once in PBS and resuspended in 100 μl of binding buffer (BD Pharmingen). A 5-μl volume of FITC-conjugated annexin V (BD Pharmingen) and 5 μg/ml propidium iodide (Sigma-Aldrich) were added to the cells. The cells were mixed gently and incubated for 15 min at room temperature in the dark, and then diluted with 400 μl of binding buffer and analyzed by two-color flow cytometry (BD Immunocytometry Systems). Apoptotic cells were distinguished from normal cells by labeling with annexin V and propidium iodide. Events accumulated in the lower right quadrate of the panel are considered to represent cells in early apoptosis, those accumulated in the upper right quadrate are considered to represent cells in late apoptosis, and those in the lower left quadrate represent survived cells.

The analysis of signal transduction was performed as described in Ref. 19 .

CD44-negative Namalwa Burkitt lymphoma cells were transfected with CD44vRA, CD44v3-10, and CD44s cDNAs, as well as empty vector (pcDNA3.1; CD44-null cells). CD44vRA, CD44v3-10, and CD44s were expressed almost equally as shown by flow cytometry (1). To evaluate the ability of the expressed CD44 to interact with gal-8, cell extracts from these cell lines were incubated with immobilized GST (control) or immobilized GST conjugated to gal-8 (GST-gal-8), in the presence or absence of thiodigalactoside (TDG), an inhibitor of galectin-sugar interaction. The proteins, which remained bound to GST-gal-8, were analyzed by Western blot, using Hermes-3 anti-CD44 mAb. As shown in Fig. 1, cell-derived CD44s, CD44v3-10, and CD44vRA interacted with gal-8, in contrast to CD44-null cell extract. This interaction was markedly reduced in the presence of TDG, which binds to gal-8 and interferes with CD44 ligation.

FIGURE 1.

Affinity chromatography reveals binding of gal-8 to cell surface CD44. Cell extracts from the indicated Namalwa transfectants were incubated with immobilized GST (lane 2) or GST-gal-8 in the presence (lane 4) or absence (lane 3) of TDG. The eluted bound proteins were resolved on 7.5% SDS-PAGE, transferred to nitrocellulose, and immunoblotted with Hermes-3 anti-CD44 mAb. Lane 1, Total protein; GST (lane 2); GST-gal-8 (lane 3); GST-gal-8 plus TDG (lane 4).

FIGURE 1.

Affinity chromatography reveals binding of gal-8 to cell surface CD44. Cell extracts from the indicated Namalwa transfectants were incubated with immobilized GST (lane 2) or GST-gal-8 in the presence (lane 4) or absence (lane 3) of TDG. The eluted bound proteins were resolved on 7.5% SDS-PAGE, transferred to nitrocellulose, and immunoblotted with Hermes-3 anti-CD44 mAb. Lane 1, Total protein; GST (lane 2); GST-gal-8 (lane 3); GST-gal-8 plus TDG (lane 4).

Close modal

Using surface plasmon resonance, we analyzed the binding and affinity of interaction of the different CD44 isoforms with immobilized gal-8. All soluble CD44 isoforms bound to gal-8 in a dose-dependent manner. The binding affinity of CD44vRA was the highest (Kd, 5.8 × 10−9 M) and was 5- and 170-fold greater than that of CD44v3-10 (Kd, 2.7 × 10−8 M) and CD44s (Kd, 10−6 M), respectively (Fig. 2, A–C). Therefore, only CD44vRA was used in all subsequent experiments involving soluble CD44. The binding of gal-8 to CD44 was specific because other galectins, i.e., galectin-1 and galectin-3, exhibited only negligible binding to CD44 (Fig. 2,D). The same was true for the plant lectin Con A, which did not interact, even at a concentration of 1 mg/ml, with any of the three CD44 isoforms (Fig. 2, insets). These results suggest that CD44vRA serves as a high-affinity receptor for a selected subset of galectins, such as gal-8. It should be noted that the differences in gal-8 affinity for the various CD44 isoforms shown in Fig. 2 cannot be reflected in the results presented in Fig. 1, because in the experiment described in the latter figure the analyzed proteins were in excess.

FIGURE 2.

Sensograms recording the interactions of galectins with soluble CD44 isoforms in the Biacore system. The indicated concentrations of CD44vRA (A), CD44v3-10 (B), and CD44s (C) were continuously injected over sensorchip surfaces coupled to gal-8, or conversely (D) gal-8, galectin-1, and galectin-3 at a concentration of 1 mg/ml were injected over sensorchip surfaces coupled to CD44vRA. Running buffer was injected and the response was recorded in RU as a function of time. Control, Injection of CD44 isoforms over Con A surface (insets, A–C). The binding affinity for gal-8 of each soluble CD44 protein, calculated as described in Materials and Methods, is shown below the upper panels.

FIGURE 2.

Sensograms recording the interactions of galectins with soluble CD44 isoforms in the Biacore system. The indicated concentrations of CD44vRA (A), CD44v3-10 (B), and CD44s (C) were continuously injected over sensorchip surfaces coupled to gal-8, or conversely (D) gal-8, galectin-1, and galectin-3 at a concentration of 1 mg/ml were injected over sensorchip surfaces coupled to CD44vRA. Running buffer was injected and the response was recorded in RU as a function of time. Control, Injection of CD44 isoforms over Con A surface (insets, A–C). The binding affinity for gal-8 of each soluble CD44 protein, calculated as described in Materials and Methods, is shown below the upper panels.

Close modal

To further assess the significance of CD44-gal-8 interaction, this event was analyzed within the context of intact cells (Fig. 3). The ability of gal-8 to bind to other types of cell surface glycoproteins (e.g., integrins) (13) made it difficult to identify the selective binding of this lectin to cellular CD44 isoforms. To overcome this obstacle, we assessed the ability of gal-8 to inhibit the binding of anti-CD44 mAbs to Namalwa cells expressing different CD44 isoforms. Flow cytometry revealed that gal-8 blocked the binding of anti-CD44v6 mAb (which recognizes epitopes included in v6) to Namalwa-CD44v3-10 and Namalwa-CD44vRA cells. In contrast, gal-8 did not inhibit the binding of anti-CD44v5 mAb or the binding of anti-pan-CD44 mAb (which recognizes a constant CD44 epitope shared by all CD44 isoforms) to all transfectants (Fig. 3,A). These findings suggest that the binding of gal-8 to a v6 exon product of CD44 causes a conformational change of this cell surface epitope or masks its presentation to the anti-v6 mAb. In contrast, gal-8 binding does not mask the v5 or the constant exon products, allowing binding of their corresponding Abs. The ability of gal-8 to inhibit binding of the anti-CD44v6 Abs was markedly reduced in the presence of lactose, but not glucose (Fig. 3 B), indicating that gal-8 is bound to sugar moieties of the CD44 glycoprotein.

FIGURE 3.

Gal-8 binds to cell surface CD44: flow cytometry analysis. A, Gal-8 inhibits the binding of anti-CD44v6 mAb to cell surface CD44. The indicated Namalwa (Nam) transfectants (top) were preincubated with gal-8 to test its ability to disrupt the cell surface binding of the indicated mAbs (ordinate). Abs bound to the cell surface in the absence (red histograms) or presence (blue histograms) of gal-8, were detected with anti-mouse Ig conjugated to fluorescein. The first histogram in each panel (green histograms), Second Ab only. The shift of the blue histograms to the left in the bottom panels indicates that gal-8 interfered with the binding of anti-CD44v6 to Namalwa-CD44v3-10 and Namalwa-CD44vRA cells. x-axis, Mean fluorescence intensity (logarithmic scale); y-axis, cell number. B, Gal-8 binds to cell surface CD44 via sugar moieties. Namalwa-CD44vRA transfectants were preincubated without gal-8 (red) or with this lectin in the presence of medium (blue), lactose (orange), or glucose (purple), and the ability of anti-CD44v6 mAb to bind to the cells was analyzed by flow cytometry, as described in A.

FIGURE 3.

Gal-8 binds to cell surface CD44: flow cytometry analysis. A, Gal-8 inhibits the binding of anti-CD44v6 mAb to cell surface CD44. The indicated Namalwa (Nam) transfectants (top) were preincubated with gal-8 to test its ability to disrupt the cell surface binding of the indicated mAbs (ordinate). Abs bound to the cell surface in the absence (red histograms) or presence (blue histograms) of gal-8, were detected with anti-mouse Ig conjugated to fluorescein. The first histogram in each panel (green histograms), Second Ab only. The shift of the blue histograms to the left in the bottom panels indicates that gal-8 interfered with the binding of anti-CD44v6 to Namalwa-CD44v3-10 and Namalwa-CD44vRA cells. x-axis, Mean fluorescence intensity (logarithmic scale); y-axis, cell number. B, Gal-8 binds to cell surface CD44 via sugar moieties. Namalwa-CD44vRA transfectants were preincubated without gal-8 (red) or with this lectin in the presence of medium (blue), lactose (orange), or glucose (purple), and the ability of anti-CD44v6 mAb to bind to the cells was analyzed by flow cytometry, as described in A.

Close modal

Flow cytometry analysis documented the expression of endogenous gal-8 on SF cells from RA patients, because anti-gal-8 mAb showed considerable binding to these cells (Fig. 4,A), with much less binding to blood leukocytes (Fig. 4,B). An even greater binding intensity of anti-gal-8 Ab was revealed when the SF cells were preincubated with gal-8 (Fig. 4,A), indicating that the gal-8 binding sites were not saturated by the endogenous lectin. Binding of exogenous gal-8 to blood leukocytes was also detected after their preincubation with anti-gal-8 Ab (Fig. 4,B). SF cells from a different RA patient did not bind anti-gal-8 Abs, indicating that the inflammatory cells of this specific patient do not contain endogenous gal-8. However, these SF cells did (like those of the first patient) bind exogenous gal-8, detected with anti-gal-8 Ab. Furthermore, an excess of soluble CD44vRA, which was preincubated with gal-8, reduced the gal-8 binding to the SF cells by almost one order of magnitude (Fig. 4,C). The possibility that anti-gal-8 Ab bound to soluble CD44vRA, nonspecifically adsorbed on the SF cells, was ruled out (Fig. 4 C, inset). These results show that soluble CD44vRA interacts with gal-8, thus reducing its ability to bind to the SF cells.

FIGURE 4.

Analysis of gal-8 attachment to SF cells and blood leukocytes from an RA patient. SF cells (A) and blood leukocytes (B) from an RA patient were preincubated with gal-8 (black line) or medium (gray line) for 45 min at 4°C, and then incubated with anti-gal-8 Ab (A and B). Binding of the Ab to the cells was detected, by flow cytometry, using fluorescein-labeled anti-mouse Ig (second Ab; faint line). Representative data of seven patients are shown. In a second experiment (C), the gal-8 (0.03 mg/ml) was preincubated with medium or with 0.18 mg/ml soluble (sol) CD44vRA for 15 min at 4°C, and then its binding to the SF cells was assayed with anti-gal-8 Ab, as mentioned above. The staining with anti-gal-8 Ab is matched with staining with the second Ab alone. The possibility that anti-gal-8 Ab bound to soluble CD44vRA, which may nonspecifically adsorbed on the SF cells, was ruled out because the anti-gal-8 Ab (dark line) histogram is matched with the “second Ab” (faint line) histogram (inset), indicating lack of binding.

FIGURE 4.

Analysis of gal-8 attachment to SF cells and blood leukocytes from an RA patient. SF cells (A) and blood leukocytes (B) from an RA patient were preincubated with gal-8 (black line) or medium (gray line) for 45 min at 4°C, and then incubated with anti-gal-8 Ab (A and B). Binding of the Ab to the cells was detected, by flow cytometry, using fluorescein-labeled anti-mouse Ig (second Ab; faint line). Representative data of seven patients are shown. In a second experiment (C), the gal-8 (0.03 mg/ml) was preincubated with medium or with 0.18 mg/ml soluble (sol) CD44vRA for 15 min at 4°C, and then its binding to the SF cells was assayed with anti-gal-8 Ab, as mentioned above. The staining with anti-gal-8 Ab is matched with staining with the second Ab alone. The possibility that anti-gal-8 Ab bound to soluble CD44vRA, which may nonspecifically adsorbed on the SF cells, was ruled out because the anti-gal-8 Ab (dark line) histogram is matched with the “second Ab” (faint line) histogram (inset), indicating lack of binding.

Close modal

The flow cytometry results were confirmed by RT-PCR of total RNA extracted from SF cells of five RA patients. In the example depicted in Fig. 5,A, using gal-8-specific primers, we detected an expected band of 894 bp, indicating that these cells contain the gal-8 transcript, corresponding to the human gal-8 isoform Po66 (15). These findings were confirmed by direct nucleotide sequencing of cDNA derived from RNA extracted from SF cells of additional three RA patients (data not shown). However, in subsequent studies (our unpublished studies), we have found that SF cells from different RA patients contain additional gal-8 isoforms. The presence of secreted gal-8 in the SF of RA patients was further documented following its adsorption onto LS beads. A gal-8 isoform of 69 kDa, in variable amounts, was detected by Western blot in the SF of nine patients tested (Fig. 5 B). This finding is restricted to RA patients because the joint of normal individuals contains neither SF nor SF cells, excluding a normal control in this analysis. By comparing the amount of gal-8 present in the SF of RA patients to that of known amounts of recombinant gal-8, we calculated that the concentration of gal-8 in the SF of RA patients is 25–65 nM.

FIGURE 5.

SF cells and SF of RA patients contain gal-8. A, Analysis at the transcript level. A representative RT-PCR analysis of total RNA extract from SF cells. Gal-8-specific primers amplified a band at 894 bp. A similar band was observed in samples of five RA patients. L, Ladder; C, negative control; P, patient. B, Analysis at the protein level. Western blot of SF from nine RA patients, using anti-gal-8 Ab after concentration of gal-8 by LS. C, Gal-8 forms a complex with CD44 in the SF of RA patients. The SF of an RA patient was incubated with LS (to trap gal-8), anti-CD44 mAb, or isotype-matched (IM) control Ab. After boiling, the proteins were resolved on 10% SDS-PAGE, transferred to a nitrocellulose membrane, and immunoblotted with anti-gal-8 Ab or Hermes-3 anti-CD44 mAb. Representative data of samples from four patients are shown. Pre, Precipitation.

FIGURE 5.

SF cells and SF of RA patients contain gal-8. A, Analysis at the transcript level. A representative RT-PCR analysis of total RNA extract from SF cells. Gal-8-specific primers amplified a band at 894 bp. A similar band was observed in samples of five RA patients. L, Ladder; C, negative control; P, patient. B, Analysis at the protein level. Western blot of SF from nine RA patients, using anti-gal-8 Ab after concentration of gal-8 by LS. C, Gal-8 forms a complex with CD44 in the SF of RA patients. The SF of an RA patient was incubated with LS (to trap gal-8), anti-CD44 mAb, or isotype-matched (IM) control Ab. After boiling, the proteins were resolved on 10% SDS-PAGE, transferred to a nitrocellulose membrane, and immunoblotted with anti-gal-8 Ab or Hermes-3 anti-CD44 mAb. Representative data of samples from four patients are shown. Pre, Precipitation.

Close modal

Furthermore, coimmunoprecipitation analysis (Fig. 5,C) revealed the formation of the Po66 gal-8-CD44 complexes in SF of RA patients. This was evident by the fact that proteins immunoprecipitaed with anti-CD44 mAb (but not with isotype-matched control mAb) contained gal-8. Similarly, proteins retained by LS beads (which traps gal-8) included CD44 that could be detected with anti-CD44 mAb as a diffused band (Fig. 5 C, right panel) of molecular mass >80 kDa, somewhat higher than the expected molecular mass of soluble CD44 which ranges between 50 and 80 kDa (21, 22, 23, 24, 25). These findings suggest that the CD44 complexed to gal-8 in the SF of RA patients contains CD44 variants of a higher molecular mass (e.g., CD44v3-10; CD44vRA). Proteinases from the extracellular matrix presumably cleave these soluble CD44 proteins into smaller fragments (26), further accounting for the wide range of molecular masses.

To determine whether soluble CD44 is the major glycoprotein bound to the endogenous gal-8 present in the SF of RA patients, the fluid was adsorbed onto LS beads to trap the endogenous gal-8 and the proteins in complex with it (Fig. 6,A, middle lane). It should be stressed that gal-8, having two binding sites can simultaneously interact with both gal-8-binding proteins (e.g., CD44) as well as with LS beads. To better identify all potential gal-8 binding-proteins present in the SF of RA patients, the fluid was also adsorbed onto LS loaded with exogenous gal-8 as a carrier (Fig. 6,A, left lane). Both total proteins present in the SF (Fig. 6,A, right lane) as well as the proteins bound to the endogenous plus exogenous gal-8 were subjected to SDS-PAGE and were stained with Coomassie blue (Fig. 6,A). The predominant bands that represent protein binding to the endogenous gal-8, as well as binding that was further enhanced in the presence of carrier gal-8 (marked as 1, 2, and 3 in Fig. 6,A), were excised from the gel and subjected to protein sequence analysis by mass spectrometry. The analysis revealed that all three bands were different fragments of fibrinogen, suggesting that gal-8 and fibrinogen form complexes in the SF of RA patients. These results further indicated that CD44 is not a major gal-8-binding protein (because, as opposed to fibrinogen, it is not detectable by Coomassie blue staining), although it is readily detected in complex with gal-8, as indicated by immunoblotting of the LS-eluted binding proteins with anti-CD44 mAb (Fig. 6 B).

FIGURE 6.

Fibrinogen forms a complex with gal-8 and CD44 in SF of RA patients. A, I, Coomassie blue staining of electrophoresed (12% SDS-PAGE) proteins bound to a mixture of exogenous and endogenous (derived from the SF of an RA patient) gal-8 adsorbed on LS beads. II, Coomassie blue staining of electrophoresed proteins bound to endogenous gal-8 adsorbed on LS beads. III, Coomassie blue staining of total proteins included in the SF of the same RA patient. Similar staining was observed in an additional four patients (data not shown). Bands marked 1, 2, and 3 (lane I) were sent for spectrometric analysis (see Results). B, Gal-8 forms a complex with CD44 in SF of RA patients. The SF of an RA patient were incubated with LS beads (to trap gal-8) or left unbound (total protein) as described in A. Following boiling, the proteins were subjected to SDS-PAGE and immunoblotted with Hermes-3 anti-CD44 mAb. C, Sensograms recording the interaction of gal-8 with fibrinogen in a Biocore system. The indicated concentrations of human fibrinogen were continuously injected over sensorchip surfaces coupled to gal-8. Running buffer was injected, and the response was recorded in RU as a function of time. The binding affinity for gal-8, 1.2 × 10−6 to 6.8 × 10−6, was calculated as described in Materials and Methods. Inset, r-gal-8 forms a complex with fibrinogen. Protein G-coupled anti-fibrinogen Ab-fibrinogen complex was incubated with r-gal-8. After boiling, the proteins were subjected to SDS-PAGE and immunoblotted with 106.1 anti-gal-8 mAb. D, Gal-8 forms a complex with CD44 and fibrinogen in the SF of RA patients. The SF of an RA patient was incubated with anti-fibrinogen Ab (to trap fibrinogen). After boiling, the proteins were subjected to SDS-PAGE and immunoblotted with anti-Gal-8 Ab, Hermes-3 anti-CD44 mAb, or anti-fibrinogen Ab. Representative data of samples from two patients are shown.

FIGURE 6.

Fibrinogen forms a complex with gal-8 and CD44 in SF of RA patients. A, I, Coomassie blue staining of electrophoresed (12% SDS-PAGE) proteins bound to a mixture of exogenous and endogenous (derived from the SF of an RA patient) gal-8 adsorbed on LS beads. II, Coomassie blue staining of electrophoresed proteins bound to endogenous gal-8 adsorbed on LS beads. III, Coomassie blue staining of total proteins included in the SF of the same RA patient. Similar staining was observed in an additional four patients (data not shown). Bands marked 1, 2, and 3 (lane I) were sent for spectrometric analysis (see Results). B, Gal-8 forms a complex with CD44 in SF of RA patients. The SF of an RA patient were incubated with LS beads (to trap gal-8) or left unbound (total protein) as described in A. Following boiling, the proteins were subjected to SDS-PAGE and immunoblotted with Hermes-3 anti-CD44 mAb. C, Sensograms recording the interaction of gal-8 with fibrinogen in a Biocore system. The indicated concentrations of human fibrinogen were continuously injected over sensorchip surfaces coupled to gal-8. Running buffer was injected, and the response was recorded in RU as a function of time. The binding affinity for gal-8, 1.2 × 10−6 to 6.8 × 10−6, was calculated as described in Materials and Methods. Inset, r-gal-8 forms a complex with fibrinogen. Protein G-coupled anti-fibrinogen Ab-fibrinogen complex was incubated with r-gal-8. After boiling, the proteins were subjected to SDS-PAGE and immunoblotted with 106.1 anti-gal-8 mAb. D, Gal-8 forms a complex with CD44 and fibrinogen in the SF of RA patients. The SF of an RA patient was incubated with anti-fibrinogen Ab (to trap fibrinogen). After boiling, the proteins were subjected to SDS-PAGE and immunoblotted with anti-Gal-8 Ab, Hermes-3 anti-CD44 mAb, or anti-fibrinogen Ab. Representative data of samples from two patients are shown.

Close modal

Having found that fibrinogen forms complexes with gal-8, we next measured the dissociation constant of this interaction, using surface plasmon resonance. We found that gal-8 binds to soluble fibrinogen with an affinity similar to the binding of gal-8 to CD44s, i.e., Kd (M) = 1.2 × 10−6 to 6.8 × 10−6 (Fig. 6,C). To ensure that fibrinogen indeed forms direct complexes with gal-8, the r-gal-8 was immunoprecipitated by a fibrinogen/anti-fibrinogen Ab complex conjugated to protein G. A single band having the expected molecular mass of 34 kDa was detected when the precipitated proteins were subjected to Western blot analysis using gal-8 Abs (Fig. 6 C, inset). These findings support our hypothesis that gal-8 can form direct complexes with fibrinogen.

Because gal-8 and CD44vRA form complexes in the SF of RA patients and because fibrinogen is also a ligand of CD44 (27), we sought direct evidence for the formation of a triple complex between fibrinogen, CD44, and gal-8 in the SF of RA patients. Fibrinogen proteins from the SF of an RA patient were immunoprecipitated with anti-fibrinogen Ab bound to a protein-G column. The bound proteins were eluted and subjected to SDS-PAGE, and then immunoblotted with anti-gal-8 Ab, anti-CD44 mAb, or anti-fibrinogen Ab. The proteins precipitated by the anti-fibrinogen Ab reacted with anti-gal-8 and anti-CD44 Abs (Fig. 6,D), implying that fibrinogen forms a complex with gal-8 and several CD44 isoforms (see multiple bends, Fig. 6,D) in vivo. Of note, fibrinogen is not a major constituent of the SF of RA patients. This conclusion is based on mass spectrometric analysis of the major protein bands present in the SF from an RA patient that were subjected to SDS-PAGE and Coomassie blue staining (Fig. 6 A, lane III, and data not shown). Those bands (data not shown) were identified as transferrin, albumin, IgG1, hepatoglobin, Ag binding fragment from anti-Fas Ab and the Ig rearranged γ-chain of the V-J-C region. These findings support our conclusion that, although fibrinogen, like CD44, is not a major constituent of the SF of RA patients, it is the principal protein that binds gal-8 and CD44 in this tissue.

We next posed the question: Is the formation of the triple complex between gal-8, fibrinogen, and soluble CD44 of any biological significance? To this end, we tested the ability of CD44 to modulate apoptosis induced by soluble gal-8. As shown in Fig. 7,A, increasing concentrations of gal-8 (up to 0.1 μM) that are included in the range of SF physiological concentrations (25–65 nM) markedly enhanced the apoptosis of RA SF cells. Furthermore, the addition of soluble CD44vRA (which interferes with the binding of gal-8 to SF cells, as shown in Fig. 4,C) reduced the extent of gal-8-induced apoptosis over a wide range of gal-8 concentrations (Fig. 7, A and C). The doxorubicin-induced apoptosis and the basal apoptosis were not influenced by the presence of soluble CD44vRA. In contrast, increasing gal-8 concentrations reduced the percentage of surviving cells (Fig. 6,B, representing the percentage of cells at the lower left quadrate of each panel in Fig. 7,C), whereas, when soluble CD44vRA was added, a higher percentage of cells was rescued from apoptosis (Fig. 7, B and C, 7–9 vs 2–4, respectively). These findings suggest that sequestration of gal-8 by soluble CD44 reduces the apoptotic potential of this lectin, resulting in enhanced inflammation.

FIGURE 7.

CD44vRA antagonizes gal-8-induced apoptosis in RA SF cells. A, Soluble CD44vRA inhibits gal-8-induced apoptosis. SF cells from RA patients were incubated with the indicated concentrations of gal-8 (x-axis) or doxorubicin in the presence (▪; gray bar) or absence (♦; black bar) of 5 μg of soluble CD44vRA. The percentage of apoptotic cells (y-axis) was determined by their staining with annexin V and propidium iodide, and subsequent flow cytometry analysis. B, Protection of RA SF cells from apoptosis by soluble CD44vRA. RA SF cells were incubated with different concentrations of gal-8 in the presence (▪) or absence (♦) of soluble CD44vRA. The percentage of unstained surviving cells is shown in the lower left quadrate of each panel in C. C, Two-dimensional flow cytometry (graphically depicted in B) of cells incubated with gal-8 (concentrations shown above the panels) in the presence (bottom panels) or absence (top panels) of soluble CD44vRA. Controls: 1) Cells incubated without gal-8 in the presence (6) or absence (1) of soluble CD44vRA. 2) Doxorubicin: cells incubated with doxorubicin in the presence (10) or absence (5) of soluble CD44vRA.

FIGURE 7.

CD44vRA antagonizes gal-8-induced apoptosis in RA SF cells. A, Soluble CD44vRA inhibits gal-8-induced apoptosis. SF cells from RA patients were incubated with the indicated concentrations of gal-8 (x-axis) or doxorubicin in the presence (▪; gray bar) or absence (♦; black bar) of 5 μg of soluble CD44vRA. The percentage of apoptotic cells (y-axis) was determined by their staining with annexin V and propidium iodide, and subsequent flow cytometry analysis. B, Protection of RA SF cells from apoptosis by soluble CD44vRA. RA SF cells were incubated with different concentrations of gal-8 in the presence (▪) or absence (♦) of soluble CD44vRA. The percentage of unstained surviving cells is shown in the lower left quadrate of each panel in C. C, Two-dimensional flow cytometry (graphically depicted in B) of cells incubated with gal-8 (concentrations shown above the panels) in the presence (bottom panels) or absence (top panels) of soluble CD44vRA. Controls: 1) Cells incubated without gal-8 in the presence (6) or absence (1) of soluble CD44vRA. 2) Doxorubicin: cells incubated with doxorubicin in the presence (10) or absence (5) of soluble CD44vRA.

Close modal

Although the above results clearly demonstrated the in vivo interaction among gal-8, soluble CD44, and fibrinogen, we still sought to evaluate the contribution of cell surface CD44 to the apoptosis induced by gal-8. For this purpose, RA SF cells were incubated with anti-CD44 mAb. However, such incubation did not influence apoptosis induced by gal-8 (data not shown), suggesting that cell surface CD44 is not involved to a significant extent in transmission of the apoptotic signals induced by this lectin, which are presumably mediated by other cell surface glycoconjugates. Using flow cytometry, we found that Namalwa cells transfected with vector only (i.e., Namalwa CD44−/−) bind gal-8 slightly, presumably insignificantly, less than Namalwa cells expressing CD44 isoforms (including CD44vRA) (data not shown). However, Namalwa cells transfected with empty vector and Namalwa cells transfected with vectors containing CD44 cDNAs exhibited similar levels of gal-8-induced apoptosis. Similar levels of gal-8-induced apoptosis were detected also in spleen cells derived from wild-type and CD44-knockout NOD and DBA/1 mice (data not shown). These findings stress the fact that cell surface-bound CD44 is not a major transmitter of apoptotic signals induced by gal-8. Although surface-bound CD44 (unlike its soluble counterpart) did not regulate, to a significant extent, gal-8-mediated apoptosis, we studied the ability of this receptor to modulate selected signaling pathways induced by gal-8 (18, 19). For this purpose we used CD44-transfected Namalwa cells rather than primary SF cells, because the latter express multiple CD44 isoforms, excluding proper analysis of signaling. Furthermore, the primary cells constitute a heterogonous population.

Namalwa cells transfected with CD44s, CD44v3-10, or CD44vRA cDNAs were incubated with gal-8, and its ability to activate PKB and JNK was evaluated. PKB was activated (phosphorylated) to virtually the same extent in Namalwa-pc3.1, Namalwa-CD44s, Namalwa-CD44v3-10, and Namalwa-CD44vRA cells, indicating that CD44 is not involved in the activation of PKB triggered by gal-8 (data not shown). In contrast, gal-8-induced activation of JNK was affected by the nature of the CD44 isoform expressed on the surface of the cells. As shown in Fig. 8, gal-8 induced JNK phosphorylation in Namalwa-CD44s cells and Namalwa-CD44vRA cells, but considerably less in Namalwa-CD44v3-10 cells. The JNK response of sham-transfected Namalwa cells (i.e., Namalwa transfected with empty vector) was similar to that of Namalwa-CD44v3-10 (data not shown). This finding suggests that the enhancement of the JNK signaling, mediated by cell surface CD44, is largely abolished when CD44v3-10 is expressed instead of the standard isoform of this receptor. Furthermore, replacement of CD44v3-10 with the CD44vRA isoform, which contains a single additional alanine residue, restores its ability to transmit a robust signal leading to JNK activation.

FIGURE 8.

Gal-8 activates JNK. A, Western blot analysis of pJNK. Namalwa-CD44s, Namalwa-CD44v3-10, and Namalwa-CD44vRA were seeded in plastic dishes, starved for 18 h, and incubated with (+) or without (−) gal-8. At the end of the incubation, cell lysates were subjected to SDS-PAGE, and then immunoblotted with anti-pJNK and anti-JNK (total protein control) Abs. The results of two experiments (of three), conducted in triplicate, are shown. B, Densocytometrical analysis of Western blot. The y-axis represents the ratio between pJNK and JNK band intensities of the indicated treatment. The x-axis represents the different treatments. The results are the mean ± SEM of three experiments conducted in triplicate.

FIGURE 8.

Gal-8 activates JNK. A, Western blot analysis of pJNK. Namalwa-CD44s, Namalwa-CD44v3-10, and Namalwa-CD44vRA were seeded in plastic dishes, starved for 18 h, and incubated with (+) or without (−) gal-8. At the end of the incubation, cell lysates were subjected to SDS-PAGE, and then immunoblotted with anti-pJNK and anti-JNK (total protein control) Abs. The results of two experiments (of three), conducted in triplicate, are shown. B, Densocytometrical analysis of Western blot. The y-axis represents the ratio between pJNK and JNK band intensities of the indicated treatment. The x-axis represents the different treatments. The results are the mean ± SEM of three experiments conducted in triplicate.

Close modal

Autoimmune inflammation is a complicated process, which includes multiple elements of innate and adaptive immunity. This work presents our efforts to elucidate at least part of the mechanism, which regulates this elusive biological cascade. Using the autoimmune model of RA, we provide evidence that gal-8 serves as a novel CD44 high-affinity ligand with the highest affinity for the CD44vRA variant. Binding of gal-8 to cell surface CD44 selectively activates JNK to promote CD44 signaling pathways. We further demonstrate that SF cells from RA patients express and secrete gal-8 to a concentration of 25–65 nM, well within the concentration required to induce apoptosis of SF cells. However, soluble gal-8 and soluble CD44, present in the SF of RA patients, bind to fibrinogen to form triple complexes. We hypothesized, therefore, that sequestration of soluble gal-8, upon formation of these complexes has a functional outcome—to reduce the proapoptotic and anti-inflammatory activity of gal-8 in the inflamed joint (Fig. 9). Several lines of evidence agree with this model.

FIGURE 9.

The balance between cell surface glycoconjugates (including cell surface CD44) and soluble CD44 determine the severity of joint inflammation (a suggested model). A, Binding of gal-8 to cell surface glycoconjugates (including CD44) induces apoptosis in joint inflammatory cells resulting in the resolution of inflammation. B, Fibrinogen accumulated in the inflamed joint traps gal-8 and soluble CD44, allowing the binding of gal-8 to soluble CD44 and consequently its sequestering from the cell surface glycoconjugates of joint inflammatory cells. C, The rescued joint inflammatory cells are free to destroy the bone and cartilage, resulting in aggravated arthritis.

FIGURE 9.

The balance between cell surface glycoconjugates (including cell surface CD44) and soluble CD44 determine the severity of joint inflammation (a suggested model). A, Binding of gal-8 to cell surface glycoconjugates (including CD44) induces apoptosis in joint inflammatory cells resulting in the resolution of inflammation. B, Fibrinogen accumulated in the inflamed joint traps gal-8 and soluble CD44, allowing the binding of gal-8 to soluble CD44 and consequently its sequestering from the cell surface glycoconjugates of joint inflammatory cells. C, The rescued joint inflammatory cells are free to destroy the bone and cartilage, resulting in aggravated arthritis.

Close modal

First, we found that gal-8 forms different affinity interactions with CD44, which are dependent on the nature of the CD44 isoform. These receptor-ligand interactions were documented by affinity chromatography of the cell extracts, surface plasmon resonance of recombinant proteins, and flow cytometry of intact cells. The lowest affinity (Kd, 10−6) is for the CD44s, whereas the CD44 variant CD44vRA, isolated from SF cells of RA patients (1), binds gal-8 at a substantially higher affinity (Kd, 5.8 × 10−9), which even exceeds its binding to the wild-type variant CD44v3-10 (Kd, 2.7 × 10−8). The priority of soluble CD44vRA over other CD44 isoforms in gal-8 binding is possibly related to changes in cell surface CD44 configuration due to an extra CAG insertion in the CD44 transcript, allowing translation of an extra alanine without disrupting the reading frame (1).

The high affinity of CD44vRA for gal-8 may be attributed to a unique sugar or protein configuration selectively expressed by CD44vRA that significantly increases its affinity to gal-8. Still, we cannot rule out the possibility that protein-protein interactions might also contribute to the interactions of gal-8 with CD44, in addition to the expected contribution of protein-carbohydrate interactions.

Both gal-8 (this study) and galectin-3 (28, 29) as well as CD44 (30) are present in the synovial cells (70% polymorphonuclears and macrophages; the remainder, T and B lymphocytes, fibroblasts, and pluripotent stem/progenitor cells) (1, 31, 32) and SF of RA patients. Furthermore, all the leukocyte subpopulations of the RA SF express CD44vRA (33). Still, the fact that unlike gal-8, galectin-1 and galectin-3 bind only marginally, if at all, to CD44 isoforms makes this finding even more striking and further suggests a specific biological function for CD44-gal-8 interactions. Furthermore, because both gal-8 and CD44 are detected in the SF of RA patients, we envisage that this inflammatory disease could be a good model for examining the possible function of the interaction between the two components, focusing on the mechanistic rather than the clinical aspect of this event.

We (this study and Ref. 1) as well as others (30) found that the SF cells of RA patients express different CD44 isoforms, including CD44s and CD44vRA. The soluble CD44 forms are presumably released from the SF cells (21) by enzymatic cleavage, and then further fragmented by local extracellular matrix proteinases (26) to generate a mixture of several CD44 ectodomains, derived from the standard and variant CD44 molecules. We show that at least some of these fragments (e.g., those derived from CD44vRA) are capable of interacting with gal-8 and form biologically significant complexes, which may have a regulatory function in the inflamed joints of RA patients. The various binding affinities of gal-8 for the discrete CD44 variants, combined with the presence of different-length ectodomains in the SF of the RA patients suggest that there is a wide spectrum of CD44-gal-8 complexes with different affinities in this fluid whose binding interactions are dictated by the nature of the CD44 variant.

However, there is a third partner to the gal-8-CD44 complex: (fibrin)ogen (fibrin is a coagulated fibrinogen). The accumulation of bloodborne fibrinogen in the joints of RA patients is an important pathological factor in this disease. In the joint of RA patients, the extravascular homeostasis between the formation and the degradation of fibrin is tilted in favor of fibrin accumulation. Coagulated fibrin can activate intimal layer synoviocytes (including fibroblasts) to release enzymes that destroy the bone and cartilage (34). Furthermore, matrix fibrinogen retains growth factors such as vascular endothelial growth factor and FGF, which further perpetuate the inflammatory process (35). In addition, fibrinogen can enhance the expression of proinflammatory cytokines (IL-1β, IL-6), chemokines (IL-8), and adhesion molecules (ICAM-1) (36) in the SF cells. Our study suggests an additional pathogenic role for fibrinogen, which is engaged in complex formation with gal-8 and soluble CD44. The relatively high quantities of fibrinogen in the SF can bind substantial amounts of the gal-8, albeit at relatively low affinity (Kd, 10−6). This reduces its availability as a free ligand, which can trigger apoptosis in the inflammatory cells, resulting in relief from inflammation. Although the low binding affinity may allow some dissociation of gal-8 from fibrin(ogen), this effect may ultimately be compensated by the binding of the released lectin to high-affinity soluble CD44, which is also included in the complex. Alternatively, or in addition, fibrin(ogen) may be used as a scaffold protein for gal-8 and CD44, allowing the concentration of these interacting molecules on its backbone and thereby reducing the availability of gal-8 to induce apoptosis.

Ligation of gal-8 in this triple complex appears to have an important pathological outcome, because it sequesters soluble gal-8 preventing it from exerting its cytostatic and proapoptotic activity (19), thus inhibiting the anti-inflammatory activity of this lectin. Indeed, we show in this study that gal-8 induces apoptosis in destructive inflammatory synovium-derived cells and that this process may be attenuated in the presence of soluble CD44 and fibrin(ogen). We suggest that binding of gal-8 to cell surface glycoproteins, including CD44, delivers an apoptotic signal, resulting in the programmed cell death of inflammatory cells and the possible relief of joint inflammation. In contrast, soluble CD44 released from activated inflammatory cells forms a complex with gal-8, which in conjunction with fibrin(ogen) reduces the ability of gal-8 to deliver death signals. Consequently, according to this model, more inflammatory cells survive and the inflammatory process is at least partly enhanced (Fig. 9). The blocking effect of soluble CD44, released from activated cells, has been previously reported (23). Therefore, the balance between the anti-inflammatory effect of gal-8 and the proinflammatory effect of fibrin(ogen) and soluble CD44 determines the severity of joint inflammation. Neutralizing the antiapoptotic soluble CD44 by injection of anti-CD44 Ab should shift the regulatory balance in favor of inflammation resolution. Indeed, we (37, 38) and others (39, 40) have previously shown that injection of anti-CD44 mAbs reduced the inflammation in animal models of arthritis and type 1 diabetes.

The ability of gal-8 to modulate the immune response of SF cells by inducing their apoptosis is in accordance with recent evidence implicating galectins and their ligands as master regulators of immune cell homeostasis (11). Whereas some members of this family, such as galectin-3, behave as amplifiers of the inflammatory cascade, others, such as galectin-1, trigger homeostatic signals to shut off T cell effector functions. In this respect, gal-8 seems to exert effects similar to those of galectin-1, which was shown to induce apoptosis in activated T cells (8, 41), and to ameliorate collagen-induced arthritis in DBA/1 mice as well (41).

We have previously shown (19) that gal-8 inhibits cell growth by promoting the accumulation of the cyclin-dependent kinase inhibitor p21. This process involves activation of JNK, thereby inducing NF-κB and leading to p21 gene transcription (42). A second, somewhat minor signaling pathway used by gal-8 to induce the accumulation of p21 is the activation of PKB. Activated JNK is both a prominent proapoptotic protein (43) as well as a downstream effecter of cell surface CD44 (2). In view of these findings, gal-8-induced apoptosis could, in principle, be mediated upon gal-8 high-affinity binding to CD44. However, we found no difference in the level of gal-8-induced apoptosis, when CD44-positive and CD44-negative Namalwa cells were compared for programmed cell death. This suggests that CD44 plays only a minor role in mediating gal-8-induced apoptosis, whereas other cell surface receptors have a more prominent function. Nonetheless, we were able to show that expression of different CD44 variants affects the ability of gal-8 to activate the JNK pathway, while having no effect on the activation of PKB by this lectin. JNK phosphorylation was stronger in Namalwa-CD44s and Namalwa-CD44vRA cells than in Namalwa cells transfected with vector alone or in Namalwa-CD44v3-10 cells. These results indicate that the different CD44 transfectants, when triggered by gal-8, can differently affect the magnitude of JNK activation. Because gal-8 does not interfere with the binding of the mAb, which recognizes a constant epitope on cell surface standard CD44 (Fig. 3), we assume that this Ab and gal-8 interact with different epitopes located on the same or distinct molecules. Indeed, our finding reconciles with the view that the apoptotic cascade can be triggered by gal-8-CD44 interaction or by gal-8-CD44 cooperation. However, this effect cannot be detected owing to the strong apoptotic “noise,” generated by non-CD44 cell surface glycoproteins, such as members of the integrin family. These receptors could then cooperate with CD44 in signal transmission. It should be further noted that the transfected Namalwa cells and the RA SF cells express similar levels of CD44 (1), indicating that the differential signaling within the Namalwa cells cannot be attributed to different levels of CD44 expression.

It is of interest that JNK activation, transduced via cell surface standard CD44, was partially attenuated when the V region v3 to v10 was introduced into the CD44 molecule. Furthermore, maximal JNK activation could be restored when the wild-type CD44v3-10 molecule was replaced by the CD44vRA variant, which contains an alanine insertion in the CD44v3-10 between exons v4 and v5 (1). This disparity could not be accounted for by differences in the level of expression of the distinct CD44 variants, because similar amounts of these receptors were expressed in all cell lines tested (1). It appears, therefore, that the presence of the additional extracellular fragment encoded by v3-10, increases on the one hand the binding affinity of gal-8 for CD44, but at the same time alters the transmembranal conformation, which elicits the activation of JNK by this receptor. Although the molecular nature of this alteration is unclear, it could involve binding of gal-8 to a putative ligand-inhibitory site on the extracellular domain of the CD44v3-10 receptor. Along this line, one must assume that the Ala insertion in CD44v3-10, converting it into the CD44vRA variant, relieves this inhibition, rendering CD44vRA as active as the CD44s, which lacks the additional extracellular variable domain. Further studies are required to elucidate the molecular basis for the different signaling capabilities of CD44v3-10 vs CD44vRA.

In conclusion, the data presented in this study prove that gal-8 serves as a novel, high-affinity ligand of CD44, with preferential, high-affinity binding to CD44vRA, the variant found in the SF of RA patients. (Figs. 1–3). We further show that gal-8 is arrested in a ternary complex that includes also soluble CD44 and fibrin(ogen) (Fig. 6). Hence, binding of soluble fibrin(ogen) and CD44 to gal-8 may masks the gal-8 ability to induce a maximal proapoptotic effect in inflammatory RA SF cells (Fig. 7). The combined three findings suggest the working hypothesis shown in Fig. 9. Thus, the severity of autoimmune inflammation, exemplified by the model, is influenced by several antagonistic factors that shift the balance toward stronger (e.g., cell surface CD44vRA/FGF-2 or gal-8/soluble CD44vRA/fibrin(ogen) complexes) or weaker (e.g., nonarrested gal-8) inflammatory responses. However, this reflects only part of a complicated, highly regulated, autoimmune inflammatory network, which includes many additional proinflammatory (IL-12, IL-18, and IFN-γ) or anti-inflammatory (e.g., IL-4 and IL-13) cytokines. Changes in this cytokine balance influence the inflammatory cascade in arthritic joints (44). Additional step-by-step studies, including experimental approaches designed to prove (or disprove) the model discussed in this study (Fig. 9) should help to elucidate this network. Obviously, significant progress in this direction can be achieved mainly in animal models of joint inflammation.

We thank Dr. Alexandra Mahler for editorial assistance.

The authors have no financial conflict of interest.

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.

1

This work was supported by Concern Foundation (Los Angeles, CA), Foundation for Research into Disease of Aging (Bala Cynwyd, PA), and Women Health Research Center (Weizmann Institute).

3

Abbreviations used in this paper: SF, synovial fluid; RA, rheumatoid arthritis; FGF, fibroblast growth factor; gal-8, galectin-8; CD44v, CD44 variant; CD44s, CD44 standard; h, human; LS, lactosyl sepharose; TDG, thiodigalactoside; PKB, protein kinase B; RU, resonance unit.

1
Nedvetzki, S., I. Golan, N. Assayag, E. Gonen, D. Caspi, M. Gladnikoff, A. Yayon, D. Naor.
2003
. A mutation in a CD44 variant of inflammatory cells enhances the mitogenic interaction of FGF with its receptor.
J. Clin. Invest.
111
:
1211
-1220.
2
Naor, D., S. Nedvetzki, I. Golan, L. Melnik, Y. Faitelson.
2002
. CD44 in cancer.
Crit. Rev. Clin. Lab. Sci.
39
:
527
-579.
3
Lesley, J., R. Hyman, P. W. Kincade.
1993
. CD44 and its interaction with extracellular matrix.
Adv. Immunol.
54
:
271
-335.
4
Naor, D., R. Vogt-Sionov, D. Ish-Shalom.
1997
. CD44: structure, function, and association with the malignant process.
Adv. Cancer Res.
71
:
241
-319.
5
Ponta, H., L. Sherman, P. A. Herrlich.
2003
. CD44: from adhesion molecules to signalling regulators.
Nat. Rev. Mol. Cell Biol.
4
:
33
-45.
6
van Weering, D. H. J., P. D. Baas, J. L. Bos.
1993
. A PCR-based method for the analysis of human CD44 splice products.
PCR Methods Appl.
3
:
100
-106.
7
Hofmann, M., W. Rudy, M. Zoller, C. Tolg, H. Ponta, P. Herrlich, U. Gunthert.
1991
. CD44 splice variants confer metastatic behavior in rats: homologous sequences are expressed in human tumor cell lines.
Cancer Res.
51
:
5292
-5297.
8
Liu, F.-T..
2000
. Galectins: a new family of regulators of inflammation.
Clin. Immunol.
97
:
79
-88.
9
Hughes, R. C..
1999
. Secretion of the galectin family of mammalian carbohydrate-binding proteins.
Biochim. Biophys. Acta
1473
:
172
-185.
10
Perillo, N. L., K. E. Pace, J. J. Seilhamer, L.G. Baum.
1995
. Apoptosis of T cells mediated by galectin-1.
Nature
378
:
736
-739.
11
Rabinovich, G. A., N. Rubinstein, L. Fainboim.
2002
. Unlocking the secrets of galectins: a challenge at the frontier of glyco-immunology.
J. Leukocyte Biol.
71
:
741
-752.
12
Hughes, R. C..
2001
. Galectins as modulators of cell adhesion.
Biochimie
83
:
667
-676.
13
Zick, Y., M. Eisenstein, R. A. Goren, Y. R. Hadari, Y. Levy, D. Ronen.
2004
. Role of Galectin-8 as a mediator of cell adhesion and cell growth.
Glycoconj. J.
19
:
517
-526.
14
Hadari, Y. R., K. Paz, R. Dekel, T. Mestrovic, D. Accili, Y. Zick.
1995
. Galectin-8: a new rat lectin, related to galectin-4.
J. Biol. Chem.
270
:
3447
-3453.
15
Bidon, N., F. Brichory, P. Bourguet, J. P. Le Pennec, L. Dazord.
2001
. Galectin-8: a complex sub-family of galectins (review).
Int. J. Mol. Med.
8
:
245
-250.
16
Hadari, Y. R., R. Arbel-Goren, Y. Levy, A. Amsterdam, R. Alon, R. Zakut, Y. Zick.
2000
. Galectin-8 binding to integrins inhibits cell adhesion and induces apoptosis.
J. Cell Sci.
113
:
2385
-2397.
17
Levy, Y., R. Arbel-Goren, Y. R. Hadari, S. Eshhar, D. Ronen, E. Elhanany, B. Geiger, Y. Zick.
2001
. Galectin-8 functions as a matricellular modulator of cell adhesion.
J. Biol. Chem.
276
:
31285
-31295.
18
Levy, Y., D. Ronen, A. D. Bershadsky, Y. Zick.
2003
. Sustained induction of ERK, protein kinase B, and p70 S6 kinase regulates cell spreading and formation of F-actin microspikes upon ligation of integrins by galectin-8, a mammalian lectin.
J. Biol. Chem.
278
:
14533
-14542.
19
Arbel-Goren, R., Y. Levy, D. Ronen, Y. Zick.
2005
. Cyclin-dependent kinase inhibitors and JNK act as molecular switches, regulating the choice between growth arrest and apoptosis induced by galectin-8.
J. Biol. Chem.
280
:
19105
-19114.
20
Eshhar, Z..
1985
. Monoclonal antibody strategy and techniques. T. Springer, ed.
Hybridoma Technology in the Biosciences and Medicine
1
-41. Plenum, New York.
21
Bazil, V., V. Horejsi.
1992
. Shedding of the CD44 adhesion molecule from leukocytes induced by anti-CD44 monoclonal antibody simulating the effect of a natural receptor ligand.
J. Immunol.
149
:
747
-753.
22
Yang, H., R. M. Binns.
1993
. Isolation and characterization of the soluble and membrane-bound porcine CD44 molecules.
Immunology
78
:
547
-554.
23
Ahrens, T., J. P. Sleeman, C. M. Schempp, N. Howells, M. Hofmann, H. Ponta, P. Herrlich, J. C. Simon.
2001
. Soluble CD44 inhibits melanoma tumor growth by blocking cell surface CD44 binding to hyaluronic acid.
Oncogene
20
:
3399
-3408.
24
Kajita, M., Y. Itoh, T. Chiba, H. Mori, A. Okada, H. Kinoh, M. Seiki.
2001
. Membrane-type 1 matrix metalloproteinase cleaves CD44 and promotes cell migration.
J. Cell Biol.
153
:
893
-904.
25
Nakamura, H., N. Suenaga, K. Taniwaki, H. Matsuki, K. Yonezawa, M. Fujii, Y. Okada, M. Seiki.
2004
. Constitutive and induced CD44 shedding by ADAM-like proteases and membrane-type 1 matrix metalloproteinase.
Cancer Res.
64
:
876
-882.
26
Cichy, J., E. Pure.
2003
. The liberation of CD44.
J. Cell Biol.
161
:
839
-843.
27
Henke, C. A., U. Roongta, D. J. Mickelson, J. R. Knutson, J. B. McCarthy.
1996
. CD44-related chondroitin sulfate proteoglycan, a cell surface receptor implicated with tumor cell invasion, mediates endothelial cell migration on fibrinogen and invasion into a fibrin matrix.
J. Clin. Invest.
97
:
2541
-2552.
28
Neidhart, M., C. A. Seemayer, K. M. Hummel, B. A. Michel, R. E. Gay, S. Gay.
2003
. Functional characterization of adherent synovial fluid cells in rheumatoid arthritis: destructive potential in vitro and in vivo.
Arthritis Rheum.
48
:
1873
-1880.
29
Ohshima, S., S. Kuchen, C. A. Seemayer, D. Kyburz, A. Hirt, S. Klinzing, B. A. Michel, R. E. Gay, F. T. Liu, S. Gay, M. Neidhart.
2003
. Galectin 3 and its binding protein in rheumatoid arthritis.
Arthritis Rheum.
48
:
2788
-2795.
30
Naor, D., S. Nedvetzki.
2003
. CD44 in rheumatoid arthritis.
Arthritis Res. Ther.
5
:
105
-115.
31
Kiener, H. P., R. Hofbauer, M. Tohidast-Akrad, S. Walchshofer, K. Redlich, P. Bitzan, S. Kapiotis, G. Steiner, J. S. Smolen, P. Valent.
2000
. Tumor necrosis factor α promotes the expression of stem cell factor in synovial fibroblasts and their capacity to induce mast cell chemotaxis.
Arthritis Rheum.
43
:
164
-174.
32
Jones, E. A., A. English, K. Henshaw, S. E. Kinsey, A. F. Markham, P. Emery, D. McGonagle.
2004
. Enumeration and phenotypic characterization of synovial fluid multipotential mesenchymal progenitor cells in inflammatory and degenerative arthritis.
Arthritis Rheum.
50
:
817
-827.
33
Golan, I., S. Nedvetzki, I. Golan, L. Eshkar-Sebban, D. Levartovsky, O. Elkayam, D. Caspi, S. Aamar, A. Howard, A. Rubinov, D. Naor.
2007
. Expression of extra trinucleotide in CD44 variant of rheumatoid arthritis patients allows generation of disease-specific monoclonal antibody.
J. Autoimmun.
28
:
99
-113.
34
Sanchez-Pernaute, O., R. Largo, E. Calvo, M. A. Alvarez-Soria, J. Egido, G. Herrero-Beaumont.
2003
. A fibrin based model for rheumatoid synovitis.
Ann. Rheum. Dis.
62
:
1135
-1138.
35
Simpson-Haidaris, P. J., B. Rybarczyk.
2001
. Tumors and fibrinogen: the role of fibrinogen as an extracellular matrix protein.
Ann. NY Acad. Sci.
936
:
406
-425.
36
Busso, N., J. A. Hamilton.
2002
. Extravascular coagulation and the plasminogen activator/plasmin system in rheumatoid arthritis.
Arthritis Rheum.
46
:
2268
-2279.
37
Nedvetzki, S., M. Walmsley, E. Alpert, R. O. Williams, M. Feldmann, D. Naor.
1999
. CD44 involvement in experimental collagen-induced arthritis (CIA).
J. Autoimmun.
13
:
39
-47.
38
Weiss, L., S. Slavin, S. Reich, P. Cohen, S. Shuster, R. Stern, E. Kaganovsky, E. Okon, A. M. Rubinstein, D. Naor.
2000
. Induction of resistance to diabetes in non-obese diabetic mice by targeting CD44 with a specific monoclonal antibody.
Proc. Natl. Acad. Sci. USA
97
:
285
-290.
39
Mikecz, K., F. R. Brennan, J. H. Kim, T. T. Glant.
1995
. Anti-CD44 treatment abrogates tissue oedema and leukocyte infiltration in murine arthritis.
Nat. Med.
1
:
558
-563.
40
Verdrengh, M., R. Holmdahl, A. Tarkowski.
1995
. Administration of antibodies to hyaluronan receptor (CD44) delays the start and ameliorates the severity of collagen II arthritis.
Scand. J. Immunol.
42
:
353
-358.
41
Rabinovich, G. A., G. Daly, H. Dreja, H. Tailor, C. M. Riera, J. Hirabayashi, Y. Chernajovsky.
1999
. Recombinant galectin-1 and its delivery suppress collagen-induced arthritis via T cell apoptosis.
J. Exp. Med.
190
:
385
-397.
42
Kobayashi, K., I. Tsukamoto.
2001
. Prolonged Jun N-terminal kinase (JNK) activation and the upregulation of p53 and p21(WAF1/CIP1) preceded apoptosis in hepatocytes after partial hepatectomy and cisplatin.
Biochim. Biophys. Acta
1537
:
79
-88.
43
Liu, J., A. Lin.
2005
. Role of JNK activation in apoptosis: a double-edged sword.
Cell Res.
15
:
36
-42.
44
Schulze-Koops, H., J. R. Kalden.
2001
. The balance of Th1/Th2 cytokines in rheumatoid arthritis.
Best Pract. Res. Clin. Rheumatol.
15
:
677
-691.