Galectin-3 Binds to Lubricin and Reinforces the Lubricating Boundary Layer of Articular Cartilage

Lubricin is a mucinous, synovial fluid glycoprotein that enables near frictionless joint motion via adsorption to the surface of articular cartilage and its lubricating properties in solution. Extensive O-linked glycosylation within lubricin’s mucin-rich domain is critical for its boundary lubricating function; however, it is unknown exactly how glycosylation facilitates cartilage lubrication. Here, we find that the lubricin glycome is enriched with terminal β-galactosides, known binding partners for a family of multivalent lectins called galectins. Of the galectin family members present in synovial fluid, we find that galectin-3 is a specific, high-affinity binding partner for lubricin. Considering the known ability of galectin-3 to crosslink glycoproteins, we hypothesized that galectins could augment lubrication via biomechanical stabilization of the lubricin boundary layer. We find that competitive inhibition of galectin binding results in lubricin loss from the cartilage surface, and addition of multimeric galectin-3 enhances cartilage lubrication. We also find that galectin-3 has low affinity for the surface layer of osteoarthritic cartilage and has reduced affinity for sialylated O-glycans, a glycophenotype associated with inflammatory conditions. Together, our results suggest that galectin-3 reinforces the lubricin boundary layer; which, in turn, enhances cartilage lubrication and may delay the onset and progression of arthritis.

polypropylene conical tubes for clarification via high-speed centrifugation at 10,000xg for 1 h. The synovial fluid supernatants were transferred to fresh polypropylene tubes and stored at -80°C. Clarified synovial fluid was thawed, pooled and purified using previously described methodology, 1  Lubricin was eluted from the DEAE column after rinsing with 50mM NaAc using a 200mM step-wise elution gradient from 200mM to 1M NaCl. Each 200mM fraction was loaded onto a 4-20% tris-glycine gradient gel (Bio-Rad, Hercules, CA) for reducing and non-reducing sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE), followed by Coomassie staining (Teknova, Hollister, CA). The 400-600mM fraction with a single prominent band at >250kD was dialyzed against phosphate buffered saline (PBS) with proteolytic inhibitors at 4°C for 36 h, with 3 exchanges of PBS. Immunoblotting of all FPLC-purified fractions was performed using a polyclonal antibody raised in rabbit and directed against the synthetic peptide CLPNIRKQPDGYDYYAFSKDQ corresponding to amino acids 1356-1374 of the C-terminus of human lubricin (ab28484, Lot: GR116636-3, Abcam®, Cambridge, UK), in addition to the monoclonal antibody 9G3 raised in mouse and directed against the mucin domain of human lubricin (MABT401; EMD Millipore), which is highly conserved across species. Both antibodies were predicted to cross-react with equine lubricin based on sequence homology of the C-terminus and mucin-rich region of lubricin, and antibodies were validated in equine using synovial fluid as a positive control and serum as a negative control. The final protein concentration of lubricin fractions was quantitated using a bicinchoninic acid protein assay (Thermo Scientific, Rockford, IL) using bovine serum albumin as the standard and read out on a microplate reader (Tecan Safire, Mannedorf, Switzerland) at 562nm absorbance.

Lubricin
For lubricin glycosylation analysis, synovial fluid was aspirated in similar fashion from the carpal joints of a healthy 5-year old horse and clarified by centrifugation.
The >100kD synovial fluid retentate was digested at with 1U/mL of Streptomyces hyaluronidase in 50mM NaAc buffer, pH 5.5 with proteolytic inhibitors overnight at 4°C. Lubricin was purified using DEAE anion-exchange resin, rinsing with 350mM NaCl and eluting in 1M NaCl. The purified lubricin was de-salted overnight using a 50kD tube dialyzer (G-Biosciences, St. Louis, MO), followed by removal of fibronectin via gelatin separopore® 4B purification (Bioworld, Dublin, OH) and elution in ultrapure water. For mass spectrometry, the lubricin samples were dried and subjected to beta-elimination to cleave O-linked glycans. Olinked glycans were cleaved with NaOH and serine-linked GalNAc was reduced with NaBH4. The reaction was neutralized with acetic acid, desalted, and cleaned of borates prior to permethylation. O-linked glycans were permethylated for structural characterization by mass spectrometry, dissolved with methanol and crystallized with α-dihydroxybenzoic acid, 20mg/mL, in 50% methanol: water.

Lubricin and Galectin Cartilage Immunohistochemistry
Osteochondral blocks were harvested from the carpal bones of the middle carpal joint of a healthy 4-year old horse, fixed in 4% paraformaldehyde, de-calcified in 10% EDTA for three weeks and embedded in paraffin. Thin (5 μm) sagittal sections were obtained for immunohistochemical staining of lubricin and galectins. Sections were deparaffinized, re-hydrated in three changes of xylene and serial alcohol, followed by treatment with 1% hyaluronidase (Sigma-Aldrich, St. Louis, MO) solution in 20mM sodium acetate at 37°C for 30 min.
Endogenous peroxidase activity was quenched by treatment with 3% hydrogen peroxide for 30 min. After blocking in normal goat or rabbit serum, a mouse monoclonal antibody against lubricin (MABT401; EMD Millipore) or a goat polyclonal antibody against galectin-3 (sc-19280; Santa Cruz) was added at 1:200 or 1:100 dilutions and incubated at room temperature for 1 hr. After three washes in PBS-T (0.1%), the sections were incubated with a biotinylated goat anti-mouse IgG (Vectastain, Vector Labs) or a biotinylated rabbit anti-goat IgG (Vectastain, Vector Labs). After 3 rinses in PBS, immunodetection was performed using the Vectastain ABC Kit and ImmPACT DAB reagent (Vector Laboratories). The sections were rinsed in PBS, counterstained using haematoxylin (Fisher) and imaged with a 20x objective using an Olympus DP80 (Olympus) camera and an Olympus IX73 microscope (Olympus). β-lactose to extract endogenous galectins. Following re-equilibration in PBS, each explant was incubated in 10ug/mL of A647-labeled galectin-1 or galectin-3 in PBS for 1 hour at room temperature. As a control for non-specific binding, explants were also incubated with A647-labeled galectin-1 or galectin-3 in the presence of 300mM β-lactose. For lubricin staining, explants were incubated in 10ug/mL of A568-labeled anti-lubricin mAb (MABT401, EMD Millipore). To compare lubricin staining following β-lactose removal of galectins, explants were incubated in either PBS or PBS + 100mM β-lactose at 4°C for 12 hours, followed by 3 rinses in PBS, incubation in 10ug/mL of A568-labeled anti-lubricin mAb for 1 hour, and Hoechst staining. Explants were rinsed in PBS, and z-stacks were obtained for four regions of interest on five separate explants for quantitation. All experiments were performed in triplicate.
Recombinant human galectin-1 and galectin-3 constructs were obtained from C.
Bertozzi and, along with the galectin-3C mutant, were expressed in XL1-Blue competent E. coli. Recombinant galectins were purified using β-lactosyl sepharose affinity chromatography similar to previously described methods 2 .
Briefly, Sepharose® 6B (Sigma-Aldrich, St. Louis, MO) was activated with 40% divinyl sulfone prepared in 0.5M Na2CO3, pH 11.0, followed by washing with distilled, deionized water and 0.5 M Na2CO3, pH 10.0, and conjugation with 40%  groove of young bovine stifles and stored in PBS at -20°C prior to tribological testing. After thawing to room temperature, cylindrical samples (6 mm diameter by 2 mm high) were incubated in sterile PBS or recombinant human galectin-1, galectin-3, or galectin-3C at a concentration of 50μg/mL for 1 h. Cartilage samples were submerged in a bath of PBS or galectin for tribological testing using a custom friction apparatus. This testing method has been previously described 3 . Briefly, the friction apparatus linearly oscillated each sample against a glass counterface three times at a controlled speed of 0.32mm/sec after allowing the cartilage sample to relax for 1 hour under an applied normal strain of 30%. Custom Matlab code (The Mathworks, Natick, MA) was used to calculate the mean equilibrium frictional coefficient (μeq) based on the final two oscillations.

Biotinylation of equine galectins
Tribometry data were compared using 2-sample student's t-tests with p < 0.05 considered significant, and all values are reported as mean ± SEM.