GAGs from mouse paws/ankles and knees/elbows bind GPI with different affinity. Gel mobility shift assay with 50 nm ¹²⁵I-GPI in each lane is shown. A, each lane contained 50 × (0, 10,000, 1000, 100, 10, 1) nm of galactosamine-equivalent GAGs, respectively, from left to right. B, graphic analysis of binding affinity between GPI and mouse paws/ankles GAG based on the data presented in A. The binding affinity between GAG and GPI was estimated by the method developed by Lee and Lander (12) where a retardation coefficient R (R = (Mo - M)/Mo; where M is migration distance) was introduced to measure affinity for the gel mobility assay. The dissociation constant K_d was derived from the R versus R/[GalN] plot shown. Paw/ankle GAG had an average chain length of 21 saccharides, which equals 8.5 disaccharides plus a linkage tetrasaccharide. Thus, a K_D of 83 nm for paw/ankle GAGs (706 nm per disaccharide divided by 8.5 disaccharides) was derived. C, each lane contained 50 × (0, 250, 25, 2.5) nm of galactosamine-equivalent GAGs. Left panel, GAGs from paws. Right panel, GAGs from knees and elbows. The experiments were repeated three times. Each time, independent GAG preparations from paw/ankle and knee/elbow joints of BALB/c mice and two additional murine strains, (B6xNOD)F1 and K/BxN, were used. The observation of the superior potency of paw/ankle GAGs to knee/elbow GAGs was replicated in every instance.

CS was the major GPI receptor on cartilage surface. Ankle and knee joint sections were deparaffinized and air-dried. Sections were fixed with 50% acetone in PBS and incubated with PBS containing 2 m NaCl to remove GAG-binding proteins. Sections were treated with either chondroitinase ABC or β-glucuronidase at 37 °C for 2 h to digest selective GAG structures. Subsequently, sections were incubated with 20 nm recombinant mouse GPI, followed by incubation with 1:50 diluted sera of arthritic K/BxN mice. After washing, sections were incubated with Alexa-488-conjugated goat anti-mouse secondary antibody. Negative controls include 1:50 diluted sera of arthritic K/BxN mice plus goat anti-mouse Alexa-488 antibody without GPI preincubation. Nuclear was stained by To-Pro-3. Sections were examined by confocal fluorescence microscopy. The experiments were repeated twice with similar results.

GAG nonreducing end GlcA contributed to high affinity GPI/GAG interaction. Paw/ankle and knee/elbow GAGs (right panel) digested with β-glucuronidase were used in a gel mobility shift assay to determine their effects on GPI binding. The data shown here are representative of three independent experiments. Left panel, control treated with digestion buffer; right panel, β-glucuronidase-treated samples. Each lane contained 50 nm ¹²⁵I-GPI and 50 × (100, 25, 6.25) nm of galactosamine equivalent GAGs. When GAGs were at low concentration or absent, some ¹²⁵I-GPI did not enter the gel and was lost.

GAGs from paw/ankle or knee/elbow were partial inhibitors of enzymatic activity of GPI. Each well contained 4.2 units/ml glucose-6-phosphate dehydrogenase, 1.0 mm NADP, 0.4 μg/ml (6.3 μm) recombinant mouse GPI, and inhibitors at various concentrations in 50 mm Tris, 88 mm KCl, pH 8.0. After equilibration at 37 °C, the reactions were initiated by the addition of 20 μl of fructose 6-phosphate for a final concentration of 2 mm. Absorbances were read every 3 s during an 8-min run. V_max values were estimated using Softmax Pro 4.6, the application software for collecting and analyzing SpectraMax M2 data. The experiment was repeated three times.

GAG sequences are not directly encoded by genes but are assembled in the Golgi by enzymes encoded by over 40 genes. Because of the vast expression repertoire of the GAG assembly enzymes and existence of GAG structural modification enzymes in extracellular matrix, GAGs display a sulfation pattern, chain length, and fine structure unique to each cell and tissue (8). GlcA is structurally similar to the natural substrate of GPI, glucose 6-phosphate (Structure 1). We hypothesize that GPI binds to GlcA residues of GAGs on the cartilage surface via its active site, providing a joint-specific target for anti-GPI Ab.

Flow cytometric analysis of GPI binding to CHO cells. Binding of GPI to CHO-745, CHO-HS and CS (CHO-HS&CS), and CHO-CS cells was performed with 1 μg of recombinant mouse GPI on ice. GPI binding was detected with 1 μg of anti-GPI monoclonal antibody followed by goat anti-mouse secondary antibody labeled with Alexa-488. Negative controls include goat anti-mouse Alexa-488 antibody alone (+2nd Ab) and αGPI monoclonal antibody 147 plus goat anti-mouse Alexa-488 antibody (1st+2nd Ab) without GPI preincubation. The experiments were repeated twice with similar results.

Capillary HPLC-coupled MS analysis of CS GAG structures from mouse paws/ankles and knees/elbows. GAGs were digested by chondroitinase ABC and analyzed by capillary HPLC-coupled MS. A, UV absorbance chromatograms (A_232, dashed line) and MS TIC chromatograms (solid line) from paw/ankle (left) or knee/elbow (right). B, XIC of disaccharides. Peak 1 contains ΔUA-GalNAc ([M - H]^- 378.1). Peak 2 contains ΔUA-GalNAc 1S ([M - H]^- 458.1). Peak 3 contains ΔUA-GalNAc 2S ([M - H]^- 268.5). Peak 4 contains ΔUA-GalNAc3S ([M - H]^- 747.0). C, left, XIC of linkage hexasaccharides. Hexa1S =ΔUA-GalNAc-GlcA-Gal-Gal-Xylol 1 sulfate (z2 545.6); Hexa2S =ΔUA-GalNAc-GlcA-Gal-Gal-Xylol 2 sulfates (z2 585.6); Hexa3S =ΔUA-GalNAc-GlcA-Gal-Gal-Xylol 3 sulfates. C, right, XIC of GalNAcS and GlcA-GalNAcS terminated structures. The experiments were repeated three times. Each time, independent GAG preparations from paw/ankle and knee/elbow joints of BALB/c mice and two additional murine strains, (B6xNOD)F1 and K/BxN, were used. The reported structural differences between paw/ankle GAGs to knee/elbow GAGs were constant in every instance.

A GPI substrate, two different GPI inhibitors, and a GlcA containing disaccharide interfere with GPI/GAG interactions. Across the top of each row, the concentrations of small molecules (left column) tested for their ability to disrupt GPI/GAG interaction are listed. Each lane contained 50 nm ¹²⁵I-GPI and 50 × (100, 50, 0) nm of galactosamine equivalent GAGs from paws/ankles. Each compound was assayed at least twice.