antibody-dependent cellular cytotoxicity
collagen antibody-induced arthritis
collagen type II
endo-β-N-acetylglucosaminidase from S. pyogenes
Lens culinaris agglutinin
surface plasmon resonance
The glycosylation status of IgG has been implicated in the pathology of rheumatoid arthritis. Earlier, we reported the identification of a novel secreted endo-β-N-acetylglucosaminidase (EndoS), secreted by Streptococcus pyogenes that specifically hydrolyzes the β-1,4-di-N-acetylchitobiose core of the asparagine-linked glycan of human IgG. Here, we analyzed the arthritogenicity of EndoS-treated collagen type II (CII)-specific mouse mAb in vivo. Endoglycosidase treatment of the antibodies inhibited the induction of arthritis in (BALB/c × B10.Q) F1 mice and induced a milder arthritis in B10.RIII mice as compared with the severe arthritis induced by non-treated antibodies. Furthermore, EndoS treatment did not affect the binding of IgG to CII and their ability to activate complement, but it resulted in reduced IgG binding to FcγR and disturbed the formation of stable immune complexes. Hence, the asparagine-linked glycan on IgG plays a crucial role in the development of arthritis.
The impact of glycosylation, one of the most important post-translational modifications, on the structure and biological properties of glycoproteins has been well documented 1, 2. IgG molecules are mainly glycosylated at Asn-297 of the CH2 domain within the Fc region 3, 4, with variable galactosylation but limited sialylation. The remaining glycosylation occurs in the hypervariable regions of the Fab region, with the position and frequency of occurrence being dependent on the presence of the consensus sequence Asn-Xaa-Thr/Ser for N-glycosylation, and is characterized by a high incidence of sialylated structures. Murine IgG contain 2.3 asparagine-linked (N-linked) biantennary oligosaccharide chains per molecule 5, and human IgG contain 2.8 6. The minimal oligosaccharide structure is a hexasaccharide (GlcNAc2Man3GlcNAc) with variable sugar residues attached, which results in the generation of the multiple glycoforms. About 30 variants of biantennary chains occur, resulting in many different glycoforms of IgG 6. X-Ray crystal electron density maps of the IgG-Fc revealed that the N297-linked glycan is sequestered within the internal space enclosed by the CH2 domains. There are extensive non-covalent interactions between the carbohydrate and the protein moiety, resulting in reciprocal influences on conformation 3.
These complex biantennary-type oligosaccharides attached to IgG have been shown to be essential for effector functions mediated through Fc receptors (FcR) and complement C1q 7, 8. Furthermore, the Fc glycans of IgG are critically involved in the structural integrity of the antibody (Ab) 9–11. Modifications in these oligosaccharides affect susceptibility to proteolytic degradation, clearance rate, Ab-dependent cellular cytotoxicity (ADCC), as well as complement activation apart from binding to FcR, monocytes, protein G and C1q/C1 12–17. However, Fc glycosylation is not required for either protein A binding 18 or recognition of antigen 19–22. Recently, de-fucosylation on the N297-linked glycan in the Fc part of the Ab has been shown to increase ADCC activity, indicating the importance of glyco-engineering of Ab for improved clinical efficacy 23. On the other hand, Fab N-linked glycosylation in the hypervariable regions, while occurring much less frequently, has been reported to influence the binding affinity of antigens 24–26 and may also be involved in IgG self-association, aggregation, and cryo-precipitation 27.
Glycoside hydrolases (EC 3.2.1.-) are a group of enzymes found in all types of organisms including bacteria and mammals. They hydrolyze the glycosidic bond between two or more carbohydrates or between a carbohydrate and a non-carbohydrate moiety. Earlier, we reported a novel secreted endo-β-N-acetylglucosaminidase, a member of the glycosyl hydrolases of family 18 (FGH18) in S. pyogenes (EndoS), which specifically hydrolyzes the β-1,4-di-N-acetylchitobiose core of the asparagine-linked glycan of human IgG 28. EndoS has similarities to endo-β-N-acetylglucosaminidases that cleave the β1–4 linkage between the two N-acetylglucosamines found in the core of the N-linked glycan of IgG. EndoS exclusively hydrolyzes the complex-type biantennary glycan on the heavy γ-chain of nof native IgG 28. Endoglycosidase activity on the IgG molecule by EndoS altered its function through impaired FcγR binding as well as decreased activation of the classical pathway of complement, which ultimately led to increased bacterial survival in human blood 29.
Ab to self antigens [collagen type II (CII), citrullinated antigens, rheumatoid factors etc.] are prevalent in rheumatoid arthritis (RA) patients and may play a role in arthritis. CII-specific mAb induce an acute form of arthritis in mice, the collagen Ab-induced arthritis (CAIA) 30–33 that resembles the effector phase of collagen-induced arthritis (CIA) without involving a priming phase. This Ab-mediated arthritis is dependent on complement components 34, FcγR 35, 36, effector cytokines TNF-α and IL-1β 37, and on the cellular players, neutrophils and macrophages 32. We used CAIA in the present study to understand the importance of glycosylation of the IgG molecule by EndoS treatment. Removal of the N-linked glycan on the IgG antibodies rendered them less arthritogenic. Although the endoglycosidase activity did not affect the binding of IgG to CII and complement activation, it reduced IgG binding to FcγR and formation of stable immune complexes.
EndoS removes carbohydrate moieties from CII-specific mouse mAb
EndoS specifically hydrolyzes the β-1,4-di-N-acetylchitobiose core of the N-linked glycan of IgG, which can be visualized by a size difference (4 kDa) on SDS-PAGE and lectin blot analysis using Lens culinaris agglutinin (LCA) (Fig. 1). LCA recognizes sequences containing α-linked mannose residues and is enhanced in its affinity binding by the core fucose attached to the GlcNAc closest to the asparagine on the protein backbone. LCA lectin blot analysis of the samples revealed a significantly reduced signal when incubated with EndoS. Loss of lectin signal has previously been shown to correspond well to the complete digestion of the chitobiose core of the glycan on N297 of human γ-chains as determined by mass spectroscopy 38. In contrast, the γ-chains were easily recognized by the LCA lectin when incubated in the absence of EndoS. These data support the hypothesis that EndoS has the ability to remove structures containing α-1,3 mannose from the γ-chains of mouse IgG.
EndoS-treated Ab bind to cartilage in vivo
To understand whether the removal of carbohydrate moieties from CII-specific IgG mAb affected their capacity to bind joint cartilage in vivo, we injected mAb into neonatal DA rats. Paw samples collected 24 h after mAb injection were sectioned and stained with anti-kappa Ab. There was no difference in the binding pattern of EndoS-treated and untreated Ab to the joint cartilage in vivo, demonstrating that the removal of carbohydrates by EndoS did not affect the antigen binding capacity of the mAb (Fig. 2). Similarly, we did not find any difference in the sera levels of these injected Ab, i.e. the clearance of normal and EndoS-treated Ab in vivo (data not shown).
Arthritogenicity of anti-CII mAb is abrogated by endoglycosidase treatment
A cocktail of two CII-specific mAb [M2139 (IgG2b) binding to the J1 epitope (551–564; GERGAAGIAGPK) and CIIC1 (IgG2a) binding to the C1I epitope (359–363; ARGLT)] induced an acute arthritis in mice (CAIA) that resembled the effector phase of arthritis 39. To understand whether the removal of carbohydrate side chains affects the arthritis-inducing capacity of pathogenic mAb, we injected a cocktail of mAb treated or untreated with EndoS. As shown in the Fig. 3, there was a profound inhibition of clinical arthritis in (BALB/c × B10.Q) F1 and B10.RIII mice that had received EndoS-treated Ab. These strains of mice were earlier shown to be highly susceptible to CAIA 32. There was a massive infiltration of cells, as well as cartilage and bone erosion, in the joints from mice injected with normally glycosylated Ab. In contrast, mouse paws showed only minor bone erosion and no significant cell infiltration after the transfer of EndoS-treated mAb. Joint articular cartilage from these mice looked normal (Fig. 3). Thus, these results clearly indicate that the removal of the N-linked glycan of IgG by EndoS abrogates their arthritis-inducing capacity on two different genetic backgrounds, and the lesser arthritogenic capacity of the EndoS-treated Ab, as demonstrated earlier, was not due to the inability of these Ab to bind to the target antigen.
EndoS-treated mAb bind C1q and activate complement
The IgG heterosaccharides are known to be involved not only in the stabilization of the Fc region binding sites for Clq 19, 40 but also in the structural properties of the Fc region 41. Therefore, we tested in vitro whether the removal of carbohydrate from γ-chains of IgG by EndoS reduced or abolished the ability of mAb to activate complement (Fig. 4). We found that there was no difference in the first complement component C1q deposition on mAb ± EndoS treatment bound to CII (Fig. 4A) or coated directly onto a plastic surface (Fig. 4B). Similarly, there was no difference between treated and non-treated Ab with regard to the deposition of C3b (Fig. 4C, D). Interestingly, CIIC1 mAb (± EndoS treatment) bound negligible amounts of C1q and did not cause any activation of complement, as has been previously reported 34. However, when the Ab were directly coated onto plates, CIIC1 was able to bind C1q and cause deposition of C3b.
EndoS-treated Ab bind less efficiently to FcγR
Since, there was no difference in the complement activation as well as in the ability to bind antigens between fully glycosylated and EndoS-treated Ab, we next asked the question why the removal of carbohydrate from γ-chains of IgG reduced/abolished the clinical arthritis. Using surface plasmon resonance (SPR) (Biacore) kinetic analysis, we found that EndoS-treated Ab have lower affinity to recombinant FcγI, FcγIIb, FcγIII and FcγIV proteins (Fig. 5, Table 1). These findings provide one possible explanation for the loss of arthritogenicity of EndoS-treated mAb because arthritis in CAIA involves FcγR systems apart from activation of complement components.
|CIIC1 (IgG2a)||4.32 × 107||0.24 × 106||0.29 × 106||1.86 × 107|
|CIIC1D||3.71 × 106||× ∼104||<<104||0.9 × 106|
|M2139 (IgG2b)||ND||1.87 × 106||0.95 × 106||2.59 × 107|
|M2139D||ND||0.05 × 106||0.06 × 106||0.54 × 106|
IgG glycosylation status affects stable immune complex formation
It is most likely that an early step in the initial triggering event in the CAIA model is the binding of the Ab to CII in the cartilage matrix and the formation of collagen–IgG immune complexes 42, 43. The CII epitopes recognized by Ab are located in a repetitive structure formed by CII molecules within the matrix and on the surface of the cartilage 42, 44. Hence, it is possible that the two different Ab can form multimeric complexes favoring arthritogenicity either by optimal complement activation or binding to FcγR-bearing cells. Similarly, immune complex formation precipitating on the joint surface was found to be absolutely required for arthritis induction in the anti-G6PI serum transfer-induced arthritis 45. Furthermore, Fc–Fc interactions are found to be important for immune complex formation 46–48, and carbohydrates present in the CH2 domain of IgG might have an important role in this process 19. To determine whether removal of sugar moieties from the CH2 domain of the Fc part of IgG could affect the formation of stable immune complexes, single immunodiffusion assays were performed on CII-impregnated agarose gels. As shown in Fig. 6 and Table 2, EndoS-treated Ab did not form stable immune complexes compared to glycosylated mAb. The inability to form stable immune complexes could be yet another reason for the loss of arthritogenicity of the treated Ab.
|mAb||EndoS treatment||Volume (×104 intensity units/mm2)||Circle area (×102 mm2)||Circle width (mm)||Pattern|
Carbohydrates present in the CH2 domain of IgG have an important role in its effector functions 19, 40 and also in the structural properties of Fc regions 41. In the present study, we found that removal of the N-linked glycan in the CH2 domain of the CII-specific IgG mAb rendered them less arthritogenic in the CAIA mouse model. Although endoglycosidase treatment did not affect the binding of IgG to CII, clearance of Ab in vivo, oxidative burst by neutrophils and macrophages in vitro (data not shown) and complement activation, it reduced IgG binding to FcγR and formation of stable immune complexes.
Earlier studies suggest a pathogenic role for the agalactosyl form of IgG in arthritis. Serum IgG from patients with RA and a small number of other rheumatic diseases contains the same set of N-linked biantennary oligosaccharides found in normal individuals, although in very different and characteristic amounts 6, 49. The incidence of structures lacking galactose is dramatically increased in arthritis. Interestingly, elevated levels of agalactosyl glycoforms were found in female RA patients, while decreased levels were correlated with disease remission during gestation followed by postpartum recurrence 50. The glycoform distribution of serum IgG was shown to change with age 51, 52. It has been suggested that low galactosylation of IgG may have a critical role in the pathology of autoimmune disorders such as RA and SLE 53–55, but can also occur through aging 56. Functional differences have been recognized between these glycoforms. Recently, Nimmerjahn et al.57 have demonstrated that agalactosyl IgG mediates its activity by binding to FcR but not complement.
On the other hand, endoglycosidase activity on IgG compromises the recognition by all three cellular FcR 21, 58, 59. Thermal stability and functionality of the CH2 domains of IgG are progressively reduced with successive removal of outer-arm sugar residues 60. Aglycosylated IgG fails to activate complement 19, is more liable to proteolytic attack 18 and is not recognized by cells expressing FcγRI and II receptors 61. Furthermore, removal of the complete carbohydrate moiety abolished the ability to activate complement and ADCC of a human IgG1 mAb, Campath-1H, but left the antigen and protein A binding activities intact, whereas removal of terminal sialic acid residues through glycopeptidase-F digestion did not affect any of the tested IgG activities 13. Moreover, sialylated IgG autoantibodies remained poorly pathogenic because of the limited Fc-associated effector functions and loss of cryoglobulin activity 62, 63. These contradictory observations might be explained by the length and nature of residual carbohydrate structures that remained after cleavage from the IgG molecule.
Furthermore, we found no difference in the C1q and C3b deposition on mAb (± EndoS treatment) bound to CII or a plastic surface. Interestingly, CII-bound CIIC1 (both EndoS treated and untreated) mAb initiated negligible amounts of C1q binding and also did not cause any complement activation as measured by C3b deposition. However, when the Ab were directly coated onto plates, CIIC1 was able to bind C1q and cause deposition of C3b, implying that it is the orientation of these Ab when bound to collagen that precludes binding of C1q and activation of complement. We observed a similar phenomenon with different mAb directed against C1III but not other CII epitopes (data not shown).
Thus, it will be interesting to further analyze the carbohydrate structure-function relationship of IgG molecules using recombinant EndoS for future analytic and therapeutic applications in autoimmune diseases.
Materials and methods
The founders of our B10.Q and B10.RIII mice originate from Prof. Jan Klein's (Tübingen, Germany) stock and has since more than 20 years been maintained in our laboratory. Thus, these strains are named B10.Q/Hd and B10.RIII/Hd. BALB/c mice were obtained from The Jackson Laboratories (Bar Harbor, ME). (BALB/c × B10.Q) F1 mice, short-named QB, and DA/Han rats were bred in the Medical Inflammation Research animal house facility in Lund. Male mice 4–6 months old were used in all the experiments. All the animals were kept in a conventional but barrier animal facility with a climate-controlled environment having 12-h light/dark cycles, in polystyrene cages containing wood shavings; they were fed standard rodent chow and water ad libitum. Local animal welfare authorities permitted the animal experiments.
Purification of CII-specific mAb
The CII-specific hybridomas were generated and characterized as described in detail elsewhere 64, 65. The anti-CII Ab-producing hybridomas, M2139 and CIIC1, were cultured in ultra-low bovine IgG-containing DMEM Glutamax-I culture medium (Gibco BRL, Invitrogen AB, Sweden) with 100 mg/L of Kanamycin monosulfate (Sigma, St. Louis, MO). mAb were generated in large scale as culture supernatant using Integra cell line 1000 (CL-1000) flasks (Integra Biosciences, Switzerland). Ab were purified using γ-bind plus affinity gel matrix (GE Healthcare, Sweden) and an Äkta purification system (GE Healthcare, Sweden). Briefly, culture supernatants were centrifuged at 12 500 rpm for 30 min, filtered and degassed before applying to the gel matrix. The gel was washed extensively and the Ab were eluted using acetic acid buffer at pH 3.0 and neutralized with 1 M Tris-HCl, pH 9.0. The peak fractions were pooled and dialyzed extensively against PBS, pH 7.0 with or without azide. The IgG content was determined by freeze-drying. The Ab solutions were filter-sterilized using 0.2-µm syringe filters (Dynagard; Spectrum Laboratories, CA), aliquoted and stored at –70°C until used. The amounts of endotoxin in the Ab solutions prepared were found to be negligible, and the Ab induced arthritis in the TLR4-deficient (LPS non-responder) mice 32.
EndoS treatment of mAb
mAb C1 and M2139 were hydrolyzed with recombinant EndoS fused to GST (GST-EndoS) purified as previously described 66. The enzyme/substrate molar ratio was 1 : 400 in PBS, and samples were incubated for 24 h at 37°C. GST-EndoS was removed from the samples by passing three times over a glutathione-Sepharose column (Amersham Biosciences, Uppsala, Sweden). Of treated or untreated IgG, 1 µg was separated by 10% SDS-PAGE followed by staining with Coomassie Blue. For lectin blot analysis, 100 ng of IgG was separated as above and blotted onto Immobilon-P PVDF membranes (Millipore, Bedford, MA). The membranes were blocked with 10 mM HEPES (pH 7.5) with 0.15 M NaCl, 0.01 mM MnCl2, 0.1 mM CaCl2, and 0.1% Tween-20 (HBST) and incubated with 1 µg/mL biotinylated LCA lectin (Vector Laboratories, Burlingame, CA). After washing in HBST, membranes were incubated with 50 ng/mL peroxidase-labeled streptavidin (Vector Laboratories). After washing, membranes were developed using the Super Signal West Pico Chemiluminescent Substrate (Pierce, Rockford, IL) and developed further using a ChemiDoc XRS imaging system (Bio-Rad, Hercules, CA).
All incubation steps were made with 50 µL solution for 1 h and at room temperature, except when stated otherwise. Every step was followed by extensive washing with 50 mM Tris-HCl, 150 mM NaCl, 0.1% Tween-20; pH 7.5. Microtiter plates (Maxisorp; Nunc, Roskilde, Denmark) were coated overnight at 4°C with either CII (10 µg/mL) or directly with the mAb diluted in 75 mM Na-carbonate, pH 9.6. The wells were blocked for 2 h with 200 mL 3% fish gelatine in washing buffer (blocking buffer). The plates coated with CII were incubated with 10 µg/mL of each Ab diluted in the blocking buffer and washed. Dilutions of (BALB/c × B10.Q) F1 serum in DGVB++ (2.5 mM veronal buffer pH 7.3, 70 mM NaCl, 140 mM glucose, 0.1% gelatine, 1 mM MgCl2 and 0.15 mM CaCl2) were added to the plates and incubated for 1 h at 37°C, followed by incubation with specific digoxigenin-labeled rat polyclonal Ab against mouse C1q (generous gift of Prof. Mohamed Daha and Dr. Leendert Trouw, Leiden University) or FITC-labeled goat anti-mouse C3 Ab (ICN Biomedicals/Cappel, Aurora, OH), both diluted 1 : 1000 in blocking solution. Horseradish peroxidase (HRP)-labeled secondary Ab against goat Ig (Dako, Glostrup, Denmark) or digoxigenin (Roche Applied Science, Indianapolis, USA) were then allowed to bind (both diluted 1 : 1000 in the blocking buffer). Bound enzyme was assayed using 1,2-phenylenediamine dihydrochloride (OPD) tablets (Dako) and absorbance was measured at 490 nm.
SPR analysis was performed as described 67. Briefly, EndoS-treated and untreated mAb were immobilized on the surface of CM5 sensor chips. Soluble FcγR were injected at five different concentrations through flow cells at room temperature in HBS-EP running buffer (10 mM HEPES, pH 7.4, 150 mM NaCl, 3.4 mM EDTA, and 0.005% surfactant P20) at a flow rate of 30 mL/min. Soluble FcR were injected for 3 min and dissociation of bound molecules was observed for 7 min. Background binding to control flow cells was subtracted automatically. Control experiments were performed to exclude mass transport limitations. Affinity constants were derived from sensorgram data using simultaneous fitting to the association and dissociation phases and global fitting to all curves in the set. A 1 : 1 Langmuir binding model closely fitted the observed sensorgram data and was used in all experiments.
Collagen Ab-induced arthritis
The two-arthritogenic mAb combination described earlier 32 was used in this study with or without EndoS treatment: M2139 (γ2b) and CIIC1 (γ2a) binding to J1 (triple helical MP*GERGAAGIAGPK – P* indicates hydroxyproline) and C1I (triple helical GARGLT) epitopes. The cocktail of the mAb (9 mg per mouse) was prepared by mixing equal concentrations of each of the sterile-filtered Ab solutions. Mice were injected i.v. with 250–500 µL solution. As internal controls, mice received equal volumes of PBS. On day 5, all the mice received LPS (50 µg/mouse, i.p.). None of the control mice receiving PBS with or without LPS developed arthritis.
Clinical evaluation of arthritis
Mice were examined daily for the arthritis development for a minimum of 21 days or until the inflammation subsided. Scoring of animals was done blindly using a scoring system based on the number of inflamed joints in each paw, with inflammation being defined by swelling and redness as described 68. Scoring was recorded in the phalangeal joints (maximum of 1 point per digit, 5 points per paw), the metacarpus or metatarsus (5 points), and in the wrist and ankle joints (5 points). Thus, the maximum score was 15/paw resulting in a peak of 60 for the total joint count.
Paws were dissected on the indicated days from each group of mice (three to four mice per group), fixed in 4% phosphate-buffered paraformaldehyde solution (pH 7.0) for 24 h, decalcified for 3–4 wk in a solution containing EDTA, polyvinylpyrrolidone and Tris-HCl, pH 6.95, followed by dehydration and embedding in paraffin. Sections of 6 µm were stained with hematoxylin-eosin to determine cellular infiltration and bone and cartilage morphology. For immunohistochemistry, paws were immediately frozen in OCT compound using isopentane on dry ice. The samples were stored at –70°C until cryosectioned at 10 µm at –30°C. Biotinylated rat anti-mouse Igκ mAb (clone 187.1), streptavidin peroxidase and diaminobenzidine were used for the detection of cartilage-bound anti-CII Ab.
All the mice were included for calculation of arthritis susceptibility and severity. The severity of arthritis was analyzed by Mann Whitney U test and the incidence by Chi Square test using Statview (version 5.0.1).
We thank Margareta Svejme and Emma Mondoc for performing histology, Carlos Palestro for taking care of the animals and Mohamed Daha as well as Leendert Trouw for rat polyclonal Ab against mouse C1q. The study was supported by grants from King Gustaf V's 80 years and Professor Nanna Svartz foundations, Swedish Rheumatism Association (to K.S.N.), Swedish Research Council, Swedish Foundation for Strategic Research (to A.M.B.), the EU [MUGEN LSHG-CT-2005–005203 and the EU project LSHB-CT-2006–018661 (AUTOCURE) (to R.H.)], Swedish Research Council (project 2005–4791), the foundations of Crafoord, Jeansson, Zoéga, Bergvall, Österlund, Groschinsky, the Swedish Society for Medical Research, the Swedish Society of Medicine, the Royal Physiografic Society, and the Medical Faculty at Lund University (to M.C.). M.C. is the recipient of an Assistant Professorship from the Swedish Research Council.