Article Text
PDF
Abstract
Objectives To understand the molecular features distinguishing anti-citrullinated protein antibodies (ACPA) from ‘conventional’ antibodies in rheumatoid arthritis (RA).
Methods Serum of ACPA-positive RA patients was fractionated by size exclusion chromatography and analysed for the presence of ACPA-IgG by ELISA. ACPA-IgG and non-citrulline-specific IgG were affinity purified from serum, plasma and/or synovial fluid and analysed by gel electrophoresis. Electrophoresis bands were excised, enzymatically digested and analysed by mass spectrometry. Binding affinity to citrullinated antigens was measured by ELISA and imaging surface plasmon resonance using recombinant monoclonal ACPA with molecular modifications.
Results In all donor samples studied (n=24), ACPA-IgG exhibited a 10–20 kDa higher molecular weight compared with non-autoreactive IgG. This feature also distinguished ACPA-IgG from antibodies against recall antigens or other disease-specific autoantibodies. Structural analysis revealed that a high frequency of N-glycans in the (hyper)variable domains of ACPA is responsible for this observation. In line with their localisation, these N-glycans were found to modulate binding avidity of ACPA to citrullinated antigens.
Conclusions The vast majority of ACPA-IgG harbour N-glycans in their variable domains. As N-linked glycosylation requires glycosylation consensus sites in the protein sequence and as these are lacking in the ‘germline-counterparts’ of identified variable domains, our data indicate that the N-glycosylation sites in ACPA variable domains have been introduced by somatic hypermutation. This finding also suggests that ACPA-hyperglycosylation confers a selective advantage to ACPA-producing B cells. This unique and completely novel feature of the citrulline-specific immune response in RA elucidates our understanding of the underlying B cell response.
- Rheumatoid Arthritis
- Autoantibodies
- Ant-CCP
Statistics from Altmetric.com
Introduction
Rheumatoid arthritis (RA) is a severely destructive inflammatory disorder affecting ∼1% of the population.1 The majority of RA patients (60–70%) harbour anti-citrullinated protein autoantibodies (ACPA). ACPA represent highly disease-specific biomarkers of important diagnostic and prognostic value, with ACPA-positive patients being at higher risk for rapidly progressive, destructive and systemic disease.2 ,3 ACPA strongly associate with polymorphisms in the human leukocyte antigen (HLA) region, which indicates a role for antigen-specific T cells in the formation and/or the evolution of this immune response.4 Initially, the ACPA response generates polyclonal antibodies at low level and can be present for many years in the absence of clinical symptoms.5 Upon a putative trigger, the ACPA epitope recognition repertoire broadens, more isotypes are being used and ACPA serum levels rise.6 ,7 Although this expansion of the immune response occurs well before the onset of clinically detectable arthritis, subclinical synovitis and bone loss may already be present at the pre-disease stage.8 Thus, the event that initiates a broadening of the citrulline-specific immune response could mark a crucial moment upon which the inflammatory process becomes self-perpetuating and, potentially, irreversible. Therefore, it is of great relevance to understand the events leading to the generation/perpetuation of the ACPA response at the molecular level.
A growing body of evidence has suggested mechanisms by which ACPA could be involved in driving the pathogenic processes found in RA. These include activation of immune effector cells,9–12 triggering of complement pathways13 and direct effects on bone metabolism.14 Most of these effects relate to in vitro assays, however, and many aspects of the in vivo ACPA response remain yet to be defined. For instance, it has been a puzzling observation that the citrulline-specific immune response, despite signs of extensive isotype switching and somatic hypermutation, generates a pool of antibodies of remarkably low avidity.15 As this and other aspects are poorly understood, we set out to characterise ACPA molecules in detail to better understand ACPA-mediated biological effects, to gain insight into ACPA structure–function relationships and to perhaps be able to conclude on the underlying B cell response.
We found that in all patient samples analysed (n=24), the vast majority of ACPA-IgG molecules exhibit a higher molecular weight in comparison with other autoantibodies or non-autoreactive IgG. Structural analyses demonstrated that the increase in weight is due to the presence of N-glycans in the ACPA variable domains. Finally, these variable domain N-glycans were found to modulate the binding of ACPA to citrullinated antigens.
Methods
Additional information on material and methods is available as online supplementary file.
Patient samples
Serum, plasma and synovial fluid samples from ACPA-positive RA patients were collected at the outpatient clinic of the rheumatology department at Leiden University Medical Center (LUMC). All RA patients fulfilled the American College of Rheumatology 1987 revised criteria for the classification of RA and gave written informed consent.16 Treatment included disease-modifying antirheumatic drugs, biological agents and glucocorticoids. For detailed RA patient characteristics, see online supplementary table S1. Sera from patients suffering from systemic lupus erythematosus (n=4), Sjögren's syndrome (n=3), coeliac disease (n=3) or myasthenia gravis (n=3) served as controls. Permission for conduct of the study was obtained from the LUMC ethical review board.
Gel filtration chromatography and ELISA
Gel filtration was performed by fast protein liquid chromatography (ÄKTA-FPLC equipped with a Hiload Superdex 200 (GE Healthcare), see online supplementary file). Chromatography fractions were analysed by ELISA to detect total IgG (Bethyl Laboratories), anti-CCP2-IgG (Immunoscan RA Mark 2; Eurodiagnostica), anti-dsDNA-IgG (Anti-dsDNA DIASTAT, Eurodiagnostica) and anti-SSA-IgG (SS-A p200 WIESLAB, Eurodiagnostica), according to the manufacturer's instructions. Anti-citrullinated fibrinogen (Fib), anti-citrullinated myelin basic protein (MBP), anti-tetanus toxoid, anti-muscle-specific tyrosine kinase (MuSK) and anti-transglutaminase (TGA) ELISA were performed using in-house protocols, as described previously.15 ,17–20
Structural analysis
(ACPA)-IgG (isotypes 1, 2 and 4) were isolated from RA patient samples by affinity purification using FPLC (ÄKTA, GE Healthcare) as described (see online supplementary file and refs. 21 ,22). Antibody F(ab′)2 and Fc fragments were generated by antibody digestion with ideS (FabRICATOR; Genovis) and purification of the resulting fragments on CaptureSelect affinity beads (Thermo Fisher). Following sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), heavy and light chains (HC/LC) of (ACPA)-IgG were digested in-gel by PNGase F to release N-glycans. Glycans were labelled with 2-aminobenzoic acid (2-AA), purified by hydrophilic interaction chromatography solid-phase extraction (HILIC-SPE) and characterised by matrix assisted laser desorption/ionisation-time of flight mass spectrometry (MALDI-TOF-MS).23 ,24 N-glycosylation sites within ACPA were identified by in-gel PNGase F-assisted 18O-labelling of N-glycosylation sites,25 followed by in-gel trypsin digestion and nanoHPLC MS/MS analysis of the digested peptides (see online supplementary file).26 18O-peptide sequences identified were confirmed by matching tandem mass spectra of gel-eluted peptides with those of their synthetic counterparts.
Antigen-binding assay
Monoclonal ACPA (mACPA) were produced in human embryonic kidney (HEK) 293T cells (see online supplementary file). Binding of mACPA to CCP2 was analysed by ELISA using undiluted and diluted (1/10 to 1/250) HEK cell culture supernatants. Background signal corresponding to binding of mACPA to the arginine variant of CCP2 was subtracted when appropriate. Imaging surface plasmon resonance (iSPR) was performed as described in the online supplementary methods.27
Results
ACPA-IgG exhibit an increased molecular weight
To study characteristics of the ACPA molecules, we fractionated serum of ACPA-positive RA patients (n=7) using size exclusion chromatography. Unexpectedly, ACPA-IgG eluted from the chromatography column in fractions preceding those containing most other IgG molecules, indicating an increased molecular weight (figure 1A). This was observed when CCP-2 (figure 1A), citrullinated fibrinogen (cit-Fib) or citrullinated MBP (cit MBP) antigens (figure 1B) were used to detect the presence of ACPA. In contrast, IgG against recall antigens from the same individuals, such as tetanus (figure 1A), Escherichia coli and diphtheria toxin, co-eluted in fractions corresponding to ‘conventional’ IgG. Moreover, no elution shift was observed for a number of other autoantibodies (anti-double-stranded DNA (anti-dsDNA), anti-SSA, anti-TGA and anti-MuSK) derived from serum of patients with systemic lupus erythematosus (n=4), Sjögren's syndrome (n=3), coeliac disease (n=3) or myasthenia gravis (n=3), respectively (figure 1C). The particular elution pattern of ACPA was observed for all sera tested, regardless of rheumatoid factor status (figure 1D). This intriguing observation indicated that ACPA-IgG are either larger than ‘conventional’ IgG or are bound by a factor causing the elution shift in size exclusion chromatography. To confirm our findings and to differentiate between these two possibilities, we affinity-purified ACPA-IgG (IgG1, 2 and 4 subclasses) and non-citrulline-specific IgG (depleted of ACPA) from serum (n=7), plasma (n=7) or synovial fluid (n=3) of RA patients followed by SDS-PAGE. Under non-reducing conditions, we again observed a higher molecular weight (corresponding to an additional 10–20 kDa) for purified ACPA-IgG compared with ACPA-depleted IgG from the same donors (figure 2A).
Anti-citrullinated protein antibodies (ACPA)-IgG exhibit a higher molecular weight than other IgG molecules. (A) Rheumatoid arthritis (RA) patient (RA1) serum fractionation by gel filtration chromatography, followed by ELISA detection, shows that ACPA-IgG (anti-CCP2 IgG antibodies) elute earlier than other IgG molecules, including antibodies against recall antigens (ie, anti-tetanus). (B) The earlier elution was also observed for other ACPA-IgG antibodies directed against citrullinated fibrinogen (cit-Fib) and citrullinated myelin basic protein (MBP). (C) Other IgG autoantibodies (Anti-SSA, Anti-dsDNA, Anti-TGA and Anti-MuSK) do not show this shift and co-elute with total IgG. Anti-SSA, anti-SSA p200 peptide; Anti-dsDNA, anti-double-stranded DNA; Anti-TGA, anti-transglutaminase; Anti-MuSK, anti-muscle-specific tyrosine kinase. (D) The higher molecular weight of ACPA-IgG is independent of rheumatoid factor (RF) serology. Both ACPA-IgG from serum of RF+ and RF− RA patients have a higher molecular weight than other IgG molecules.
The higher molecular weight of anti-citrullinated protein antibodies (ACPA) is conserved in the F(ab′)2 fragments, as well as in the light chain (LC) and in the heavy chain (HC). (A) SDS-PAGE analysis under non-reducing conditions confirms that native ACPA-IgG or their F(ab′)2 fragments present a 10–20 kDa higher molecular weight relative to other IgG molecules isolated from three rheumatoid arthritis (RA) patients (RA8, 9 and 10). (B) Under reducing conditions, ACPA-IgG exhibit one to three light chains at 26, 30 and 33 kDa (LC1 to LC3) and one to three heavy chains (HC1 to HC3) at 55, 60 and 65 kDa, respectively. In contrast, reduced IgG are mainly represented by one light chain at 26 kDa (LC1) and one heavy chain at 55 kDa (HC1).
The higher molecular weight of ACPA-IgG is due to extensive Fab-linked N-glycosylation
To locate and identify possible post-translational modifications that could be responsible for the increased molecular weight of ACPA, we analysed ACPA-IgG F(ab′)2 fragments and Fc fragments by SDS-PAGE. While no mass difference was observed for the Fc fragments, the F(ab′)2 portion of ACPA-IgG showed a higher molecular weight than ‘conventional’ IgG (figure 2A). Under reducing conditions, this observation was maintained, with both HC and LC displaying several bands of increased molecular weight, thereby excluding non-covalent linkage of one or more additional proteins to the ACPA molecule (figure 2B). Of note, the presence of one or more HC and LC bands was found in ACPA-IgG of all donor samples analysed (n=17, figure 2B and see online supplementary figure S1) and indicated that the increase in molecular weight could be due to the variable addition of one or more defined structures to the ACPA-IgG molecule. As such shifts in the electrophoretic profile of antibodies have previously been attributed to the presence of Fab-linked N-glycans,28–31 we hypothesised that glycans could be responsible for the mass shift of ACPA-IgG. To test this, purified (ACPA)-IgG molecules were treated with PNGase F and compared with untreated antibodies by SDS-PAGE under reducing conditions (figure 3A). As expected, PNGase F digestion of non-ACPA IgG decreased the apparent molecular weight of the HC due to release of the constitutive Fc-glycans, while no effect was observed on the electrophoretic mobility of the LC. In contrast, PNGase F treatment completely abolished the presence of additional electrophoretic HC and LC bands of ACPA-IgG, indicating that the higher molecular weight observed for ACPA-IgG is indeed due to additional N-glycans in the Fab-portion. To confirm and substantiate this finding, electrophoretic bands of ACPA-IgG and ACPA-depleted IgG were excised and N-glycans were released by in-gel PNGase F digest followed by MALDI-TOF-MS (figure 3B and see online supplementary figure S2). As expected, analysis of the HC of ‘conventional’ IgG showed the typical profile of Fc-linked glycans, whereas no glycans were found on the IgG LC (figure 3B).32 ,33 In contrast, the additional LC electrophoretic band of ACPA-IgG (LC 2) showed the presence of typical Fab-linked N-glycans.34 Glycans released from ACPA-IgG HC bands displayed either only the Fc-glycan (HC1) or a mixture of Fc-linked and Fab-linked glycans (HC2 and 3). To further confirm their presence and structure, these glycans were characterised by MS/MS fragmentation (data not shown and see online supplementary figure S2). Based on the SDS-PAGE data presented in figure 2A and online supplementary figure S1 and on the quantification of total, Fc-linked and F(ab′)2-linked N-glycans (data not shown), we estimate that at least 80% of ACPA-IgG molecules contain Fab-glycans.
The higher molecular weight of anti-citrullinated protein antibodies (ACPA) results from extensive N-glycosylation. (A) SDS-PAGE of ACPA-IgG and IgG under reducing conditions. Following the release of N-glycans by PNGase F, ACPA-IgG show a similar electrophoretic profile on SDS-PAGE than other IgG molecules, indicating that the additional ACPA-IgG heavy (HC) and light chains (LC) are N-glycosylated. (B) After in-gel PNGase F digestion, the released N-glycans were labelled with the 2-aminobenzoic acid (2-AA) fluorescent tag, purified by hydrophilic interaction chromatography (HILIC) and analysed by matrix assisted laser desorption/ionisation-time of flight mass spectrometry (MALDI-TOF-MS). The spectra clearly show the presence of monosialyted and disialylated N-glycans in the additional light (LC2) and heavy chains (HC2 and HC3) of ACPA-IgG, which is not found in conventional IgG. A detailed description of N-glycan structures and their respective masses is depicted in online supplementary figure S2.
ACPA F(ab′)2-glycans are linked to the immunoglobulin variable domain
N-glycosidic linkage requires asparagine residues in the protein backbone that are part of a defined consensus sequence (Asn-X-Ser/Thr, where X is not a proline). As no consensus sequence exists in the first constant domain of the IgG HC or LC (CH1/CL1), Fab-linked N-glycans are usually found in the antibody variable region (VH and VL). However, few reports have also described non-consensus glycosylation sites (ie, Ser/Thr-X-Asn) in the CH1 domain of monoclonal IgG and human serum IgG, albeit in very low frequency.35 ,36 To determine the exact localisation of ACPA Fab-glycans, electrophoretic bands corresponding to either the entire ACPA molecule, the F(ab′)2-domain, and/or the LC and HC were digested by PNGase F in the presence of H218O. This reaction allows labelling of the N-glycosylation site through the incorporation of an 18O atom into the Asp-X-Ser/Thr sequence.25 Following tryptic digestion, 18O-labelled peptides were identified by liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS). When applied to the entire ACPA-IgG molecule, this procedure reliably identified the conserved Fc N-glycosylation site at position 297 (Asn297) of the CH2-domain, thereby validating this strategy for identification of N-glycosylation sites (data not shown). LC-MS/MS analysis and peptide sequencing of the ACPA-IgG Fab-region identified six additional N-glycosylation sites in ACPA molecules of seven RA patients. The sequence of these peptides was confirmed by tandem MS of the synthetically prepared counterparts of the proposed tandem mass spectral interpretations. Basic local alignment search tool (BLAST) analysis demonstrated that all N-glycosylation sites contained a N-linked glycosylation consensus sequence located in IgG variable domains (three in HC sequences, two in kappa and one in lambda LC sequences) (table 1). In contrast, neither consensus nor non-consensus glycosylation sites were identified in ACPA-IgG constant domains (CH1 and CL) of the F(ab′)2. In summary, these data show that ACPA-IgG Fab-glycans are linked to conventional N-glycosylation sites within the IgG variable region and indicate that these glycosylation sites have been introduced by somatic hypermutation.
ACPA-IgG N-glycosylation sites are located within variable regions of the F(ab′)2 fragments
ACPA-IgG Fab-glycans modulate avidity to citrullinated antigens
The predominant function of the antibody variable domain is binding of the cognate antigen. As we could locate the additional N-linked glycans in the variable region of the antibody, we reasoned that it could influence the binding of citrullinated antigens.31 ,37–40 Therefore, we next performed ELISA and iSPR experiments to measure the interaction of two ACPA-IgG monoclonal antibodies (mACPA-1 and mACPA-2), naturally glycosylated in their variable region (wild type (WT)),41 with the CCP2 antigen. These mAbs were also produced in a non-Fab-glycosylated form by mutating the Asn residue of the variable domain N-glycosylation sites back to the germline sequence (non-glycosylated (NG); figure 4A). The presence or absence of Fab-linked N-glycans was controlled by gel electrophoresis of the mACPA WT/NG under reducing conditions (figure 4B). As expected, both mAbs showed binding to CCP2 in ELISA and iSPR (figure 4C, D). When the variable domains were devoid of glycans, mACPA-1 NG exhibits a ∼1.9-fold higher binding to CCP2 in ELISA (figure 4C). This result was confirmed by the >2-fold decrease of dissociation constant (KD) (Kd/Ka) measured by iSPR (figure 4D and see online supplementary table S2). Likewise, the absence of the N-linked glycan in the variable domains of mACPA-2 also affected antigen binding. However, in this case, we observed a decrease of binding to CCP2 in ELISA and a >2-fold higher KD in iSPR (figure 4C, D and see online supplementary table S2). Together, these data clearly demonstrate that the variable domain glycosylation of ACPA-IgG can modulate the binding avidity to citrullinated antigens.
Variable domain glycosylation of anti-citrullinated protein antibodies (ACPA)-IgG modulates binding to CCP2. (A) Partial amino acid sequences of heavy (HC) and light (LC) chains of mACPA-1 and mACPA-2 monoclonal antibodies. Both the HC and the LC chains of wild type (WT) mACPA-1 contain an N-glycosylation site, whereas two sites are present within the HC of mACPA-2 WT. To remove these variable domain N-glycosylation sites and generate non-glycosylated (NG) mACPA, the asparagine residues of the N-glycosylation consensus sequences (Asn-X-Ser/Thr, with X≠Pro) were mutated back to germline-encoded sequences. CDR, complementarity determining region; FR, framework region. (B) Additional heavy (HC2-3) and/or light chains (LC2) observed by SDS-PAGE of mACPA antibodies under reducing conditions confirmed the presence of variable domain N-glycans in these autoantibodies. (C) ELISA experiments (n=2 in quadruplicate) were performed to assess the binding of mACPA-1 and mACPA-2, WT and NG, to CCP-2. ELISA results are reported as arbitrary units (AU) per ng/mL of IgG. (D) Dissociation constant (KD), measured by surface plasmon resonance (n=1 in quadruplicate), representing avidity of the mACPA WT and NG for CCP2. M, molar.
Discussion
Our data show a unique feature of ACPA as the vast majority of ACPA molecules in all RA-donor samples analysed thus far contain additional N-linked glycans in their variable domains. Moreover, we have obtained strong indications that the presence of these additional N-linked glycans results from the introduction of N-linked glycosylation sites as a result of somatic hypermutation and that the presence of these unusual sugars can modulate the binding to citrullinated antigens.
The presence of N-linked glycans in the variable domain of human antibodies has been described before. It has been estimated that up to 15–25% of polyclonal human serum IgG of healthy donors can carry Fab-linked N-glycans.28 ,34 ,42 Although further structural analyses will be required to more carefully quantify this percentage in ACPA, our results clearly show that ACPA-IgG are more Fab-glycosylated than other IgG molecules. In contrast, this feature could not be detected for a number of other autoantibodies tested. However, due to the limited number of autoantibodies tested so far, we cannot exclude that the unique mechanism of Fab-hyperglycosylation also occurs in other types of autoantibodies, and additional analyses will have to be performed to reveal the full picture of the relevance and consequences of introducing N-linked glycosylation sites during a (chronic) immune response. Nonetheless, our data indicate that the extensive Fab-glycosylation of ACPA-IgG is not a general hallmark of autoantibodies, but rather a specific feature of ACPA or of a particular type of autoimmune response that still needs to be defined. Importantly, the higher molecular weight of ACPA-IgG was also observed in RA patient samples that were collected at the time of diagnosis (see online supplementary table S1). These data indicate that the Fab-glycosylation of ACPA-IgG is already present in early RA patients (disease duration less than 1 year) and suggest that this feature could develop before the onset of RA.
Using high-end mass spectrometry, we found ACPA Fab-portion N-glycosylation sites only in the variable region (table 1). The relatively low number of N-glycosylation sites identified by mass spectrometry is in line with the highly polyclonal nature of ACPA-IgG molecules. Accordingly, protease digestion probably resulted in a high number of variable domain peptides below the detection limit of LC-MS/MS. Therefore, it is likely that the six N-glycosylation sites identified belong to more frequently used immunoglobulin variable domains. In agreement, two of the six N-glycosylation sites were identified in ACPA-IgG of more than one donor (see table 1). More importantly, as all N-glycosylation sites detected by our analysis were not germline-encoded, it is very likely that the high degree of ACPA variable domain glycosylation is the result of extensive somatic hypermutation. The latter finding is intriguing as it suggests that ACPA-producing B cells that have been able to introduce an N-linked glycosylation site during somatic hypermutation have a selective advantage over B cells that failed to do so. Such selective advantage rendered to the B cell could, conceivably, result from the acquired ability of ACPA to interact with lectins that, for example, provide a survival advantage upon cross-linking of the (hyperglycosylated) B cell receptor (BCR). A similar mechanism has been proposed in case of follicular lymphoma B cells where mannose-rich Fab-glycans on cell-surface BCRs create a functional bridge with lectins, thereby providing a survival signal.43 Thus, the introduction of an N-linked glycosylation site by ACPA-producing B cells could be involved in the ‘breach of tolerance’ and/or the outgrowth/expansion of these cells.
The thought that somatic hypermutation results in the formation of N-linked glycosylation consensus sequences is also sustained by the high rate of non-synonymous somatic hypermutation detected in ACPA sequences obtained by single cell PCR of synovial fluid B cells and by the generation of Fab-glycosylated ACPA mAbs from peripheral blood, as those used in this study.41 ,44 Finally, a recent study in mice further supports this notion by showing that antibodies which bind self-antigens may mask the antigen-binding site with N-glycans and, thereby, relieve the B cell from a continuous BCR signal that is otherwise leading to an ‘anergic’ state.40 If this would be the case, it is predicted that Fab-linked glycans modulate binding affinity of ACPA to citrullinated antigens. As we were unable to enzymatically remove the Fab-glycans from isolated polyclonal ACPA without denaturing them, we used monoclonal antibodies that do or do not carry Fab-glycans to study the effect of Fab-glycosylation on antigen-binding affinity. Our ELISA and iSPR experiments showed that the removal of Fab-glycosylation sites alters mACPA binding avidity to the CCP2 antigen. Although we cannot formally exclude that the mutation of amino acids themselves modifies the affinity of monoclonal antibodies for the antigen, our data are in line with the notion that ACPA Fab-glycosylation can influence binding to citrullinated antigens. Interestingly, the modulation of binding depends on the antibody tested and probably also on the nature of the antigen, suggesting that Fab-glycosylation also influences ACPA fine-specificity. In light of this reasoning, it is intriguing that the citrulline-specific immune response, despite signs of extensive isotype switching and somatic hypermutation, has been found to generate antibodies of remarkably low avidity.15 Moreover, patients harbouring ACPA of lowest avidity display the highest rate of erosive joint destruction.45 Thus, it is possible that ACPA Fab-glycosylation is responsible for the overall low avidity of the citrulline-specific immune response, which offers a novel perspective of autoantibody function in autoimmunity.
It will be important in future studies to determine whether the presence of an N-linked glycan in the variable domain offers a selective advantage to ACPA-producing B cells or whether other explanations form the basis of ACPA-hyperglycosylation. Likewise, it remains to be determined whether the accumulation of N-glycosylation sites is due to frequent antigen exposure, the diversity of citrullinated antigens or whether it results from other events such as the putative presence of abundant helper activity from CD4+ T cells.43 ,46
In summary, this work demonstrates that structurally different and larger antibodies characterise the RA-specific immune response to citrullinated antigens. This finding points to aberrations in the development of ACPA-specific B cells, reinforces and changes our understanding of basic disease mechanisms in RA and might help to understand how ACPA contribute to RA pathophysiology.
Acknowledgments
We are grateful for expert technical assistance from Simone Kruithof (Sanquin, Amsterdam), Ammar Muhammad, Gerrie Stoeken-Rijsbergen, Ellen van der Voort, Carolien Koeleman and Agnes Hipgrave Ederveen (LUMC, Leiden). We thank Rinse Klooster (LUMC, Leiden) and Kyra A Gelderman (VUmc, Amsterdam) for help in measuring anti-MuSK and anti-TGA antibodies. We thank Dr Jan Wouter Drijfhout for providing the CCP2 peptide (LUMC, Leiden).
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Footnotes
Handling editor Tore K Kvien
Contributors YR, TWJH, LAT, HUS and REMT designed the study. AW, JS, LvT, YR and LAT carried out the gel filtration chromatography and ELISA experiments. The glycosylation analysis was performed by YR with advice from AMD and MW. YR, PFK, GMCJ and PAvV collected and/or analysed LC-MS/MS data for N-glycosylation site identification. GW and TR provided ACPA monoclonal sequences. AZ and RCH produced ACPA monoclonals. JBCvB and GJMP carried out the SPR measurement and analysis. YR, HUS and REMT interpreted the data and wrote the paper.
Funding We acknowledge financial support from the Dutch Arthritis Foundation, The Netherlands Organization for Scientific Research (NWO; project number 435000033), the IMI funded project BeTheCure (contract no. 115142-2), the Netherlands Proteomics Center and the Center for Medical Systems Biology embedded in the Netherlands Genomics Initiative. YR was supported by a Boehringer Ingelheim funded project within BeTheCure. REMT is recipient of a NWO-ZonMW VICI grant (project number: 918.96.606). LT was supported by a NWO-ZonMW Vidi grant (project number: 016.126.334) and by a fellowship from Janssen Biologics. HUS is recipient of a NWO-ZonMW clinical fellowship (project number: 90714509). MW was supported by funding from the European Union's Seventh Framework Programme (FP7-Health-F5-2011) under Grant Agreement No. 278535 (HighGlycan).
Competing interests None.
Ethics approval Institutional review board of Leiden University Medical Center.
Provenance and peer review Not commissioned; externally peer reviewed.