Human single-chain variable fragment that specifically targets arthritic cartilage

Abstract

Objective: To demonstrate that posttranslational modification of type II collagen (CII) by reactive oxygen species (ROS), which are known to be present in inflamed arthritic joints, can give rise to epitopes specific to damaged cartilage in rheumatoid arthritis (RA) and osteoarthritis (OA) and to establish a proof of concept that antibodies specific to ROS-modified CII can be used to target therapeutics specifically to inflamed arthritic joints.

Methods: We used a semisynthetic phage display human antibody library to raise single-chain variable fragments (scFv) specific to ROS-modified CII. The specificity of anti-ROS-modified CII scFv to damaged arthritic cartilage was assessed in vitro by immunostaining articular cartilage from RA and OA patients and from normal controls. The in vivo targeting potential was tested using mice with antigen-induced arthritis, in which localization of anti-ROS-modified CII scFv in the joints was determined. The therapeutic effect of anti-ROS-modified CII scFv fused to soluble murine tumor necrosis factor receptor II-Fc fusion protein (mTNFRII-Fc) was also investigated.

Results: The anti-ROS-modified CII scFv bound to damaged arthritic cartilage from patients with RA and OA but not to normal preserved cartilage. When systemically administered to arthritic mice, the anti-ROS-modified CII accumulated selectively at the inflamed joints. Importantly, when fused to mTNFRII-Fc, it significantly reduced inflammation in arthritic mice, as compared with the effects of mTNFRII-Fc alone or of mTNFRII-Fc fused to an irrelevant scFv.

Conclusion: Our findings indicate that biologic therapeutics can be targeted specifically to arthritic joints and suggest a new approach for the development of novel treatments of arthritis.

Figure 1

Binding of 1-11E to reactive oxygen species (ROS)–modified type II collagen (CII). A, An enzyme-linked immunosorbent assay (ELISA) to test binding of the anti–ROS-modified CII single-chain variable fragment (scFv) 1-11E to native and modified CII was performed as described previously (43). Briefly, a microtiter plate was coated with native CII or CII modified by glycation, HOCl, ^.OH, or ONOO⁻. After blocking and incubation with 1-11E, mouse anti–Myc-Tag antibodies were added, followed by horseradish peroxidase (HRP)–conjugated anti-mouse antibodies, to probe bound scFv. In the ELISA, 1-11E bound to most types of modified CII (bars 2–5) but did not bind to native CII (bar 1). B, Western blotting was performed as described previously (16). Briefly, modified and native bovine CII were run on a sodium dodecyl sulfate gel under reducing conditions and then blotted onto a nitrocellulose membrane. After blocking, membranes were incubated with 1-11E, then with mouse anti–Myc-Tag, and then with HRP-conjugated anti-mouse antibodies. Binding of 1-11E to a range of CII α-chain fragments below 100 kd, as well as to aggregates with high molecular weight, was observed. Numbered horizontal bars represent the position of molecular weight markers, in kd. Lanes 1–5 represent native CII and CII modified by glycation, HOCl, ^.OH, or ONOO⁻, respectively. OD = optical density.

Figure 2

Binding of 1-11E to cartilage in patients with rheumatoid arthritis (RA) and osteoarthritis (OA). Antibody 1-11E diffusely stained RA cartilage in all layers (brown). A, Staining of the RA specimen in the superficial area and the middle zone was stronger than that in the deep zone. B, Staining of the RA specimen with Safranin O was weak and localized mainly in the deep zone (red). C, Staining of an OA cartilage sample with extensive erosions and marked surface damage, including the formation of fragments discrete from the parent cartilage, was strong in the most severely damaged area. D, Staining with 1-11E colocalized with an area of weak Safranin O staining in a parallel nonconsecutive OA cartilage section. E and F, Antibody 1-11E staining of cartilage from a patient with mild OA with typical fissuring of the surface of the upper cartilage appeared as a territorial “halo” around the chondrocytes (E), while staining with Safranin O was strong (F). G–I, Staining of the subchondral bone was not observed in any of the samples tested. No staining with 1-11E was detected in histologically normal cartilage (G), which also stained normally with a commercial anti–type II collagen monoclonal antibody (H) and with Safranin O (I). Bars in A and B = 5,000 μm; bars in C–H = 2,000 μm; bar in I = 500 μm.

Figure 3

Binding of 1-11E to cartilage from mice with experimental inflammatory arthritis and posttraumatic osteoarthritis (OA). A and B, Inflamed cartilage was uniformly and strongly stained with 1-11E (A), while no staining was observed using irrelevant anti–hen egg lysozyme (anti-HEL) single-chain variable fragment (scFv) (B). C, Antibody 1-11E did not stain control uninflamed cartilage. D, Strong pericellular staining with 1-11E was observed in the damaged area of cartilage from mice with knee injury that developed posttraumatic OA. E, No binding was observed using control anti-HEL scFv. Bars = 100 μm.

Figure 4

Selective accumulation of 1-11E in the inflamed paw. Mice with antigen-induced arthritis and similar degrees of paw swelling were injected intravenously with Alexa Fluor 680–labeled 1-11E and with control anti–hen egg lysozyme (anti-HEL [C7]) (n = 3 per group). Single-chain variable fragment (scFv) localization analysis was performed using ImageJ software (NIH Image, National Institutes of Health, Bethesda, MD; online at http://rsbweb.nih.gov/ij/); the RGB Split and Image Calculator functions in ImageJ were used to subtract background from signal. The localization index is the product of the area and the mean pixel density, corrected for scale bar variations between images (Excel; Microsoft, Redmond, WA). A, Greater accumulation of 1-11E was observed in inflamed paws than in uninflamed paws after 3 hours. Control anti-HEL scFv exhibited no specific accumulation, and the signal was reduced down to a constant level throughout the experiment. At 1 hour, the level of localized fluorescence-labeled 1-11E in inflamed paws was ∼3 times lower than the level of fluorescence-labeled anti-HEL (P = 0.05) in inflamed paws. At 3 hours, however, the level of localized fluorescence-labeled 1-11E was ∼2 times higher than the level of fluorescence-labeled anti-HEL in inflamed paws (P = 0.05). Values are the mean ± SD. B and C, Scans of mice with an inflamed left paw (B) or knee (C) that were injected with fluorescence-labeled 1-11E revealed specific accumulation of 1-11E in the inflamed tissue.

Figure 5

Fusion of 1-11E to soluble murine tumor necrosis factor receptor II–Fc fusion protein (mTNFRII-Fc). A, Schematic representation of the 1-11E/mTNFRII-Fc construct containing a matrix metalloproteinase (MMP) cleavage site between 1-11E and mTNFRII-Fc. Murine TNFRII-Fc was amplified with forward primer A (5′-GCTAAGCTTATGGCGCCCGCCGCCCTC) and reverse primer B (5′-CTTGAATTCTTTACCCAGAGACCGGGA). After digestion, the polymerase chain reaction (PCR)–amplified fragment was cloned into the Hind III–Eco RI sites of pFastBac1.AH, which was created from pFastBac1 (Invitrogen) containing an MMP-1 cleavage site cloned between Eco RI and Not I. Antibody 1-11E or control anti–hen egg lysozyme (anti-HEL) single-chain variable fragment (scFv) was amplified with forward primer C (5′-CAGGCGGCCGCAATGGCCGAGGTGCAGCTG-3′) and reverse primer D (5′-CTTGGGCCCTCAATGGTGGTGGTGATGGTGTCTAGACCGTTTGATTTCCACCTT-3′) to amplify the scFv and to include a His-Tag between Xba I and Apa I. The PCR-amplified fragment was then digested and cloned into pFastBac1.AH digested with Not I and Apa I. The fusion protein was expressed using the baculovirus expression system. B, Western blot analysis of the fusion protein product with the expected molecular weight band of 75 kd, which was reduced to 25 kd after cleavage with MMP-1, corresponding to the scFv detected by the anti–His-Tag. C, Measurement of the activity of the 1-11E/mTNFRII-Fc fusion protein. The activity of the 1-11E/mTNFRII-Fc fusion protein was similar to that observed with mTNFRII-Fc alone, as measured by inhibition of TNFα-mediated induction of the NF-κB promoter–driven luciferase reporter gene (luc) in HeLa 57A cells. Values are the mean.

Figure 6

Superior therapeutic effect of 1-11E/murine tumor necrosis factor receptor II–Fc fusion protein (mTNFRII-Fc) in the antigen-induced arthritis (AIA) model. We used mice with AIA during the chronic phase of disease that were rechallenged with methylated bovine serum albumin (mBSA), followed by injection of therapeutic protein on day 1 and day 3 (n = 8 for the 1-11E/mTNFRII-Fc and the hen egg lysozyme [HEL]/mTNFRII-Fc treatment groups; n = 7 for the mTNFRII-Fc and phosphate buffered saline [PBS] control groups). Reduction of swelling of inflamed knees was accelerated in mice in the 1-11E/mTNFRII-Fc fusion group as compared with the control anti-HEL/mTNFRII-Fc and mTNFRII-Fc groups (P = 0.0008 by repeated-measures analysis of variance). P values for the post hoc test were calculated using the Newman-Keuls multiple comparison test: for 1-11E/mTNFRII-Fc versus HEL/mTNFRII-Fc, P < 0.001; for 1-11E/mTNFRII-Fc versus mTNFRII-Fc, P < 0.01; and for HEL/mTNFRII-Fc versus mTNFRII-Fc, P > 0.05. These results indicate that 1-11E specifically accelerates the reduction of knee swelling. Values are the mean and SD.