More than 20 years ago it was reported that serum IgG molecules isolated from patients with rheumatoid arthritis (RA) carry an oligosaccharide chain lacking a terminal galactose residue at the conserved N-glycosylation site of the Fc domain, compared to the IgG oligosaccharide chain from healthy individuals.1 This glycoform of IgG is termed agalactosyl IgG. A marked increase in agalactosylated glycoforms of IgG molecules has also been detected in other pathologies, such as in patients with spondyloarthropathy (SpA), active juvenile chronic arthritis, systemic lupus erythematosus, Sjögren syndrome and Crohn disease, suggesting that this defect is a common feature of chronic inflammatory diseases.2^–6 In RA, the prevalence of IgG-associated agalactosyl structures is strongly correlated with the disease activity and the disease course.1 3 7 The opposite phenomenon is associated with remission and pregnancy, a condition known to have a positive effect on the phenotypical disease features of RA.8 9 In addition, follow-up studies have demonstrated that the amount of agalactosylated IgG molecules correlated with the disease activity in patients with early arthritis, and that it can serve as a predictive value for a more progressive disease course.10

The animal model best studied for RA is the mouse collagen-induced arthritis (CIA) model, induced by immunisation of DBA/1 mice with heterologous collagen type II (CII). Rademacher et al demonstrated the pathophysiological involvement of agalactosylated IgG molecules in the development of arthritis as enrichment of agalactosyl glycoforms in a pool of anti-CII IgG serum induces aggravation of arthritic disease upon passive transfer in DBA/1 mice.11 Other serological markers that can be detected in arthritic mice include cartilage oligomeric matrix protein (COMP), osteoprotegerin (OPG) and acute phase proteins. However, these markers only provide information on certain aspects of the disease and therefore the use of a serological marker indicative of overall disease activity is of considerable interest, especially in the context of accurate evaluation of therapy.

As no data are available on the use of agalactosyl Ig as a serological marker in analysing disease progression upon treatment during longitudinal CIA studies, we monitored the changes of galactosylation upon antigen-dependent and antigen-independent treatment regimens during the development and progression of CIA. To do this, we used our improved DNA-sequencer aided fluorophore-assisted carbohydrate electrophoresis (DSA-FACE) technology, which is one of the most sensitive and high-throughput methods for the analysis of protein-linked N-glycans.12^–14 In the past, several methods were developed to routinely determine serum galactosylation ratios, such as chromatographic methods, lectin binding assays and FACE.5 6 15^–19 Unfortunately, the use of these methods in clinical practice is hampered by the lack of sensitivity and/or high throughput. Finally, using the DSA-FACE method, we also show that blockade of tumour necrosis factor (TNF)α in patients with RA and SpA resulted in increased Ig galactosylation levels.

MATERIALS AND METHODS

Animals

DBA/1 male mice, 8 to 12 weeks old, were purchased from Janvier (Le Genest Saint Isle, France) and were treated and used in agreement with the institutional guidelines. All animal procedures were approved by the Institutional Animal Care and Ethics Committee.

Induction of CIA

Chicken CII (Morwell Diagnostics GmbH, Zürich, Switzerland) was dissolved in 0.1 M acetic acid before being emulsified with complete Freund adjuvant (CFA) (Difco, Weybridge, Surrey, UK); 100 μl of this emulsion was then subcutaneously injected at the base of the tail of DBA/1 mice. At 21 days later, the animals were challenged with a secondary injection of CII emulsified in incomplete Freund adjuvant. This secondary immunisation was not performed in the experiments in which CII tolerance was induced. For the longitudinal study, mice were bled at various time points before and after onset of arthritis. Blood sampling was obtained from retro-orbital bleeding, which was performed by well trained personnel.

Induction of CII tolerance

Chicken CII was dissolved in 0.1 M acetic acid and dialysed against phosphate-buffered saline (PBS); 100 μg of this dialysed CII was intravenously injected at day 7 before primary immunisation. Mice were immunised as described above.

Evaluation of CIA

Clinical severity was graded from 0 (normal paws) to 3 (pronounced oedema and erythaema leading to incapacitated limb).20 All clinical evaluations were performed by an investigator unaware of mouse identity. The average arthritic score was obtained by summing the score recorded for each limb of individual mice divided by the number of mice in the group.

Semitherapeutic treatment of CIA

DBA/1 mice were immunised with CII as described above. On day 14 after primary immunisation, mice were daily injected with 10 μg of the glucocorticoid dexamethasone (DEX) (Aacidexam, Organon Europe BV, Oss, The Netherlands) for 2 weeks. As a control, mice were similarly treated with lipopolysaccharide (LPS)-free PBS.

Detection of anti-CII IgG antibodies and COMP level in mouse serum

Serum levels of anti-CII-specific IgG antibodies and COMP were determined by standard ELISA (Chondrex, Redmond, Washington, USA) and by a competitive COMP ELISA (MD Biosciences, Zürich, Switzerland), respectively. Both ELISAs were performed according to the manufacturer’s instructions.

Patients

Group 1 consisted of 20 patients with RA (mean (SD) age of 53.7 (12.2) years) who fulfilled the American College of Rheumatology (ACR) criteria for RA.21 These patients were treated with infliximab at a dosage of 3 mg/kg, in weeks 0, 2 and 6, and every 8 weeks thereafter. Serum samples were taken in weeks 0 and 30. Group 2 included 20 patients with SpA (mean (SD) age of 48.7 (8.7) years) who were treated with infliximab at 5 mg/kg in weeks 0, 2, 6, 20 and 34. Serum samples were taken in weeks 0 and 34. All procedures were approved by the institutional ethics committee.

Isolation of serum Ig using protein-L agarose and improved DSA-FACE

A detailed protocol for the analysis of protein-linked glycans using the DSA-FACE technology is described in detail by Laroy et al.14 Briefly, immunoglobulins were isolated from the serum samples using protein-L agarose. Protein-L binds the κ light chain of all antibodies from various species, such as human and mice. In addition, it interacts with a wider range of Ig molecules than protein-A or protein-G, which binds to the Fc portion. Therefore, protein L is more effective in immunoglobulin purification. Although it is believed that N-glycosylation does not influence binding of protein-A or protein-G to IgGs, we chose protein-L to minimise the risk of missing any immunoglobulin fraction. Subsequently, proteins were treated with peptide N-glycanase F to release the N-glycans from the proteins. The glycans are fluorescently labelled and desialylated.12 22 For partial oligosaccharide sequencing, the N-glycan pools were incubated with different exoglycosidases and combinations thereof. DSA-FACE was used to profile and analyse the labelled glycans. For this study, we defined the undergalactosylation status (UGS) as the log10-transformed ratio between the amount of fully non-galactosylated and the fully galactosylated fucosylated biantennary N-glycans present on the immunoglobulins. This corresponds to the log10-transformed ratio between the heights of peaks 3 and 9, or the ratio of the areas under them (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Glycan profiling of protein-L bound serum fractions by the improved DNA-sequencer aided fluorophore-assisted carbohydrate electrophoresis (DSA-FACE) technique. Representative N-glycan profiles of control and arthritic samples are generated as described in the methods section. In the arthritic sample, fully non-galactosylated core fucosylated biantennary N-glycan (peak 3) predominates over fully galactosylated core fucosylated biantennary N-glycan (peak 9). The opposite situation applies to control samples (A). The proposed N-glycan structures, deduced from partial exoglycosidase sequencing (not shown). The numbers represent the numbered peaks in (A). Squares: N-acetylglucosamine (GlcNAc), diamonds: galactose, circle: mannose, triangle: fucose (B).

Statistical analysis

The Mann–Whitney U and Fisher exact test were used to analyse clinical results. Non-parametric Spearman correlation was calculated to assess association of UGS values with clinical scores. Means were compared using the Student t test. Analysis of diagnostic performance was done by calculation of the area under the receiver operating characteristic curve (AUROC).

RESULTS

Profiling and structural analysis of immunoglobulin N-glycans

In this study, we used the DSA-FACE technology for profiling and studying the N-glycans present on immunoglobulins in autoimmune arthritis. This improved technology allows a highly sensitive and high-throughput method to analyse Ig galactosylation.12 13

Representative electropherograms of immunoglobulins are shown in fig 1A. Partial exoglycosidase sequencing of two profiles (data not included), with different relative intensities, and comparison with commercially available N-glycan standards revealed most of the structures behind the major peaks (fig 1B).

Serum agalactosyl IgG content efficiently discriminates between non-arthritic and arthritic mice

We initially chose to analyse the agalactosyl content of the Ig fraction in serum derived from a large cohort of mice suffering from CIA with well documented clinical disease scores.20 Therefore, mice were bled at day 5 after the primary immunisation (ie, prior onset of arthritis) as well as at day 10 after onset of arthritic disease. As can be seen in fig 2, the UGS score is significantly correlated with the clinical arthritis score (Spearman rho = 0.830, p<0.001, including arthritic as well as non-arthritic animals). Interestingly, arthritic animals can be distinguished from non-arthritic animals with a remarkably good separation of the two groups.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Undergalactosylation status (UGS) values are significantly correlated with clinical arthritis scores. UGS values were analysed in the serum from arthritic mice, which were bled on day 10 after onset of arthritis (n = 34) as well as from non-arthritic mice on day 5 after immunisation with collagen type II (CII)/complete Freund adjuvant (CFA). The figure shows the corresponding values of UGS scores and clinical scores. Data were pooled from two independent experiments. (Spearman rho = 0.830; p<0.001; Spearman non-parametric correlation, including arthritic and non-arthritic animals).

Induction of antigen-specific peripheral tolerance prevents onset of CIA and is not associated with increased serum levels of agalactosyl Ig

It has previously been shown that changes in Ig galactosylation are also apparent during non-pathological immune responses, as undergalactosylation of antibodies towards an immunogen is observed in mice immunised with bovine serum albumin in incomplete Freund adjuvant.23 Therefore, we examined whether induction of antigen-specific tolerance was also associated with changes in Ig agactosylation compared to non-tolerised mice. Three analogous, independent experiments were performed in which peripheral CII tolerance was induced. In two of these experiments, the CII-tolerised mice developed no arthritis, while in one experiment, 50% of the mice developed mild clinical symptoms (table 1). Subsequently, all animals were bled on average 43 days after primary immunisation and UGS scores were determined. Table 1 shows that if PBS and CII-tolerised treatment groups are subdivided in arthritic versus non-arthritic animals, the UGS levels are remarkably distinct between the non-arthritic animals versus the arthritic mice. Additionally, UGS scores are significantly correlated with clinical arthritis scores (Spearman rho = 0.698; p<0.001, including arthritic as well as non-arthritic animals), confirming the results of the first cohort. These observations demonstrate that increased serum agalactosyl Ig content is associated with development of arthritis, rather than being a consequence of immunisation with an adjuvant, and that successful antigen-specific tolerance induction prevents serum Ig agalactosylation.

Increased serum agalactosyl Ig levels correlate with CIA progression and are reversed by successful therapeutic treatment

The clinical scoring data of the longitudinal study indicate that semitherapeutic treatment of CIA with DEX efficiently ameliorates arthritic disease symptoms (fig 3A). Upon withdrawal from DEX therapy these mice rapidly develop arthritis, as illustrated by a steep increase of the mean arthritis index curve starting from day 30 (fig 3A). Results from UGS determination show that the appearance of agalactosyl Ig in the serum coincides with the development of clinical symptoms (fig 3B), the induction of an anti-CII-specific humoral response (fig 3C) as well as with decreased serum levels of COMP which is indicative for the degree of cartilage breakdown (fig 3D). Moreover, the efficacy of DEX treatment in this study is reflected by significantly suppressed UGS values at days 21, 28 and 34 as compared to PBS animals. At day 41, in contrast, withdrawal from DEX therapy resulted in an increased mean UGS score nearly equalling the PBS control group. Agalactosyl Ig content was shown to detect arthritic animals with high sensitivity and specificity on day 28 (AUROC = 0.80), day 34 (AUROC = 0.95) and day 41 (AUROC = 0.95). Furthermore, the efficacy of DEX treatment is also reflected by reduced anti-CII IgG and COMP levels from day 28 after immunisation (fig 3C,D) compared to the PBS group. We also observed significantly lower levels of the acute phase protein, serum amyloid A, in the serum of DEX-treated mice compared to PBS controls at day 28 after immunisation (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Undergalactosylation status (UGS) values correlate with progression of collagen-induced arthritis (CIA) and are indicative of successful therapeutic treatment. At 14 days after primary immunisation, mice were daily treated with either 10 μg dexamethasone (DEX) or phosphate-buffered saline (PBS) for 2 weeks until day 28 (arrow). An additional group was included containing non-immunised, age-matched DBA/1 mice (n = 12) as a control group. Each paw was monitored for the presence of clinical signs of arthritis (***p<0.001, **p<0.01 by Mann–Whitney U test, compared with PBS control group) (A). The three groups of mice were bled on the indicated time points and the obtained sera was used to analyse the UGS values (*p<0.05 by Mann–Whitney U test, compared with PBS control group) (B), the concentration of anti-collagen type II (CII) IgG antibodies (*p<0.05, **p<0.001 by Mann–Whitney U test, compared with PBS control group) (C) as well as the release of cartilage oligomeric matrix protein (COMP) (*p<0.001 by Mann–Whitney U test, compared with PBS control group) (D).

The UGS response is suppressed upon anti-TNFα treatment in RA and SpA

Finally, in an effort to confirm and extend our findings in human subjects, we determined the UGS scores in the serum of patients with RA (n = 20) as well as of patients with SpA (n = 20) who have been treated with infliximab, as described in the patient section. These patients showed significant clinical improvement following this therapy. Our data show that treatment of patients with RA (fig 4A) and SpA (fig 4B) with infliximab resulted in a significant improvement of the galactosylation status of the Ig fraction (paired t test: p<0.001 and p = 0.019, respectively).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

The effect of anti-tumour necrosis factor (TNF0α treatment on undergalactosylation status (UGS) in patients with rheumatoid arthritis (RA) and spondyloarthropathy (SpA). In total, 20 patients with RA (A) and 20 SpA (B) were treated with infliximab as described in the Materials and methods section. Serum UGS scores were evaluated at baseline, week 30 or week 34 after initiation of therapy in patients with RA and SpA, respectively. Means in both groups are significantly different (*p<0.001 and **p = 0.019 by paired t test).

DISCUSSION

In the present study, we used optimised DSA-FACE technology12 to monitor Ig galactosylation changes during the course and treatment of autoimmune arthritis. We initially evaluated the N-glycan profiling of Ig molecules in the serum of arthritic mice. We demonstrated that UGS values correlate with disease activity as measured by independent clinical scoring. In addition, serum UGS values are remarkably able to efficiently distinguish arthritic from non-inflammatory animals. Therefore, UGS determination can be a useful aid in the overall assessment of disease activity in CIA mice.

Next we observed that agalactosylation of serum Ig is not apparent in CII-tolerised mice compared to arthritic PBS control mice, and that unsuccessfully tolerised mice developed elevated levels of agalactosyl Ig. The good correlation between the clinical score and the UGS values was also confirmed in these CIA experiments. Thus, these findings suggest that the occurrence of agalactosyl Ig is related to the pathogenesis of CIA, rather than being a consequence of immunisation with an adjuvant, and that successful antigen-specific tolerisation prevents agalacosylation of Ig molecules.

In the longitudinal CIA study we confirmed that an increased UGS score correlates with the clinical severity during the course of disease, and that it coincides with the appearance of anti-CII specific IgG as well as with the degree of cartilage breakdown. A more clinical point of interest is the question whether agalactosylation of Ig molecules could be prevented upon efficient semitherapeutic treatment. In this study we approached this question and demonstrated that the progressive increase of agalactosylated Ig molecules is suppressed upon semitherapeutic treatment with DEX. The observation that UGS normalises significantly after DEX therapy is in agreement with the suppressed humoral IgG response and COMP levels. In contrast with these decreased anti-CII IgG serum levels, an increase in antigen-specific IgA antibodies could be possible, as antigen-specific IgG and IgA serum levels are known to be differentially affected after DEX treatment.24 Based on our data, UGS may serve as an indicator to measure clinical response in experimental arthritis, and is an additional candidate to evaluate the benefit of therapy. The tendency of decreased UGS levels together with the opposite phenomenon whereby the UGS values increase upon withdrawal of DEX treatment, suggest that therapy not only modulates the symptoms of disease, but also interferes with the biochemical process leading to UGS level alteration. This observation is further strengthened by the fact that anti-TNFα therapy in patients with RA and SpA resulted in decreased UGS values. Importantly, the evolution of UGS levels under infliximab therapy shows the reversibility of alterations in immunoglobulin glycosylation status in chronic arthritis, which is an important pathophysiological finding. To our knowledge, this is the first study describing the effect of anti-TNFα therapy on the UGS score in patients with SpA, thereby confirming and extending earlier observations showing that clinical improvement of patients with RA upon treatment is associated with reversal of IgG agalactosylation levels.25^–27

It has clearly been shown that specific glycosylation of the Fc region is a prerequisite for proper conformation and biological function of immunoglobulin molecules,28^–31 such as elimination of IgGs from the circulation.32 Our longitudinal study demonstrates that there is no drop in anti-CII IgG levels, which may be explained by failure of deglycosylated IgG molecules to be eliminated rapidly. In addition, age-dependent agalactosylation of immunoglobulin molecules might also influence these glycosylated-related functions.33 34 Taking into consideration that deglycosylation may have a profound impact on the activation of the immune system, Malhotra et al proposed an immunological mechanism which may explain the linkage between deglycosylation and the pathogenesis of RA.35 More specifically, the increased amount of agalactosylated IgG molecules observed in arthritis may arise from the decreased B cell glycosylation machinery, such as a decreased galactosyltransferase activity (GTase), resulting in a glycoform of IgG with a terminal N-acetylglucosamine (GlcNAc) residue. This terminal GlcNAc residue can activate the complement cascade through interactions with the mannose-binding protein, resulting in the activation of cellular destructive mechanisms. This is concordant with the observation that agalactosylated IgGs are involved in the pathogenesis of CIA.11 Therefore, as agalactosylated IgG molecules are prevalent in the serum of CIA mice, patients with RA and SpA, it can be hypothesised that these glycoforms may contribute to the pathway of chronic inflammation in rheumatoid autoimmune diseases.

In conclusion, we propose that determination of serum agalactosyl Ig ratios using the optimised DSA-FACE technology can be efficiently used to monitor the disease course in the experimental CIA model. As antigen-specific tolerisation induction, as well as treatment with DEX in CIA or anti-TNFα therapy in patients with RA and SpA, shows the reversibility of Ig galactosylation, this may open new perspectives for monitoring therapeutic approaches.