Study cohort
Patient plasma used in this study was collected from patients visiting the outpatient clinic of the Rheumatology Department at Leiden University Medical Center (LUMC) and selected based on ACPA levels. Healthy donor plasma was obtained from the biobank Rheumatic Diseases of the LUMC (protocol B14.011). All patients were diagnosed with ACPA-positive RA prior to sampling and met the 2010 American College of Rheumatology/European League against Rheumatism (ACR/EULAR) criteria for RA at the time of diagnosis. At the time of sampling, RA patients received different immunomodulatory treatments as listed in Supplementary Table 1. All donors gave written informed consent prior to sample acquisition. Permission for the conduct of the study was approved by the Ethical Review Board of the LUMC (protocols P13.171 and P17.151). Detailed patient information is provided in Supplementary Table 1.
Preparation of plasma
Plasma of patients 1, 2, 5 and 6 was collected from Ficoll-Paque-based isolation of peripheral blood mononuclear cells upon dilution of whole blood with phosphate buffered saline (PBS). Plasma of the remaining patients were collected upon centrifugation of whole blood for 10 minutes at 774 or 931 g and 20 °C. Plasma were stored at −20 °C until further use.
Peptide synthesis, protein modification, and column preparation
For affinity-based ACPA purification, we selected the citrullinated antigen cyclic citrullinated peptide 2 (CCP2; patent number: EP2071335), a “golden standard” antigen used in clinical practice for diagnostic and prognostic purposes. The non-modified version of CCP2 (CArgP2) served to remove antibodies binding to either the column material or the peptide backbone.
ACPA fine-specificity profiles were determined using the citrullinated peptides fibrinogen α 27-43 (FLAEGGGVCitGPRVVERH), fibrinogen β 36-52 (NEEGFFSACitGHRPLDKK), cyclic citrullinated peptide 1 (CCP1; HQCHQESTCitGRSRGRCGRSGS), vimentin 59-74 (VYATCitSSAVCitLCitSSVP) and enolase 5–20 (KIHACitEIFDSCitGNPTV)30,42, as well as citrullinated fibrinogen and fetal calf serum (FCS). N-terminally biotinylated CCP2 and CArgP2 as well as the N-terminally biotinylated peptides to determine ACPA fine-specificity profiles were prepared (kindly provided by Dr. J. W. Drijfhout, Department of Immunology, LUMC) as described previously42. Proteins were modified as described before43; however, citrullination of FCS was performed at 37 °C and additionally supplemented with 2.5 M NaCl and 0.5 µM EDTA.
CCP2 and CArgP2 columns were prepared by functionalizing 1 mL HiTrap® Streptavidin HP columns (GE Healthcare/Cytiva) with 1.0–1.5 mg N-terminally biotinylated CCP2 or CArgP2 peptide at a flow rate of 0.2 mL/min. Excess of peptide was removed by washing with PBS and glycine-HCl, pH 2.5. Columns were stored in 20% ethanol until usage.
Purification of ACPA from plasma
ACPA was purified from plasma using a tandem purification approach as depicted in Fig. 1a. In particular, CArgP2 and CCP2 columns were coupled in line to an ÄKTA pure purification system and equilibrated with PBS. Plasma was sterile filtered before respective volumes were manually injected into a 5 mL sample loop and applied to tandem purification at a flow rate of 0.5 mL/min. The unbound fraction was collected as ACPA-depleted plasma. After the sample application was completed, the sample loop and the columns were washed with PBS to remove the remaining unbound sample before ACPA was eluted from the CCP2 column with glycine-HCl, pH 2.5 at a flow rate of 0.5 mL/min. ACPA was collected in fractions of 0.2 mL. Immediately after fractionation, up to eight elution fractions starting from the beginning of the elution peak were pooled and desalted using 5 mL Zeba spin desalting columns (Thermo Fisher Scientific) according to the manufacturer’s instructions. Desalted ACPA was stored at 4 °C until further assessment in enzyme-linked immunosorbent assays (ELISAs) and IgG capturing. CArgP2 and CCP2 columns were washed with glycine-HCl, pH 2.5 to ensure removal of remaining bound material and equilibrated with PBS for subsequent purifications. For consistency, ACPA was purified from 2 mL plasma for patients 1–7. For patient 8, 4 mL plasma was applied due to its markedly lower ACPA level.
Enzyme-linked immunosorbent assays
ACPA IgG, tetanus toxoid (TT)-specific IgG and total IgG content of plasma, ACPA-depleted plasma and purified ACPA as well as ACPA fine-specificities were determined using in-house ELISAs. All samples were assessed with technical duplicates.
To determine ACPA IgG reactivity and levels, streptavidin-coated plates (Microcoat, standard capacity) were incubated for one hour at room temperature with 1 µg/mL N-terminally biotinylated CCP2 diluted in PBS/0.1% bovine serum albumin (BSA). Samples were diluted in PBS/1% BSA/0.05% Tween-20 (PBT) and applied for one hour at 37 °C. Bound ACPA IgG was detected by incubation with HRP-labelled rabbit anti-human IgG (DAKO, P0214; diluted 1:8000 in PBT) for one hour at 37 °C. ACPA fine-specificities were determined using the citrullinated peptides and proteins detailed above. Binding to citrullinated peptides was assessed by incubating streptavidin-coated plates with 10 µg/mL citrullinated peptide instead of 1 µg/mL CCP2. Binding to citrullinated proteins was assessed by coating Nunc Maxisorp plates (Thermo Fisher Scientific) with 10 µg/mL citrullinated protein diluted in coating buffer (0.1 M Na2CO3/0.1 M NaHCO3, pH 9.6) overnight at 4 °C, followed by blocking with PBS/2% BSA for at least six hours. Samples were diluted in PBT and applied overnight at 4 °C before bound ACPA IgG was detected by incubation with the HRP-labelled rabbit anti-human IgG (diluted 1:3000 in PBT) for 3.5 hours at 4 °C. Unspecific reactivity to all peptides and proteins was determined by applying the non-modified instead of the citrullinated peptide or protein.
The presence of TT-specific IgG was assessed by coating Nunc Maxisorp plates with 1.5 LF/mL TT in coating buffer for three hours at 37 °C. Plates were blocked with 1% BSA/coating buffer for one hour at 37 °C before samples diluted in PBT were applied for two hours at 37 °C. Bound TT-specific IgG was detected by incubation with HRP-labelled rabbit anti-human IgG (DAKO, P0214; diluted 1:10,000 in PBT) for one hour at 37 °C.
To determine the total IgG content, Nunc Maxisorp plates were coated with 10 µg/mL goat anti-human IgG-Fc (Bethyl, A80-104) for one hour at room temperature, followed by blocking with PBS/1% BSA/50 mM TRIS, pH 8.0 for 30 minutes at room temperature. Subsequently, samples diluted in PBS/1% BSA/50 mM TRIS/0.05% Tween-20 (PBTT, pH 8.0) were applied for one hour at room temperature. Bound total IgG was detected by incubation with HRP-labelled goat anti-human IgG (Bethyl, A80-104P; diluted 1:20,000 in PBTT) for one hour at room temperature.
All ELISAs were read out using ABTS and H2O2 at an iMark Microplate Absorbance Reader (BioRad) or Multiskan FC Microplate Photometer (Thermo Fisher Scientific). ACPA IgG levels were quantified based on an in-house standard of pooled RA patient plasma, total IgG based on IgG standard serum (Merck Millipore, S1). All quantifications were performed using the Microplate manager software MPM-6 (BioRad).
Monoclonal antibody production and purification
In-house produced monoclonal anti-TT IgG1 and ACPA IgG1 antibodies were expressed in FreestyleTM 293-F cells (Gibco) under glycoengineering conditions42. Briefly, cells were cultured in FreestyleTM 293 expression medium (Gibco) at 37 °C, 8% CO2 on a shaking platform. Medium was supplemented with D-galactose substrate prior to transfection and cells transfected with plasmids encoding the immunoglobulin heavy chain, the immunoglobulin light chain, the large T antigen of the SV40 virus (GeneArt), the cell cycle inhibitors p21 and p27 (Invivogen), β1,4-N-acetylglucosaminyltransferase III (GnTIII), α2,6-sialyltransferase 1 (ST6GalT) and β1,4-galactosyltransferase 1 (B4GalT1) using 293-FectinTM (Invitrogen). Transfection supernatants were harvested five to six days after transfection and purified using 1 mL HiTrap® Protein G HP affinity columns (GE Healthcare) according to the manufacturer’s instructions. Eluted antibodies were re-buffered to PBS and single heavy and light chains excluded using a 53 mL HiPrepTM 26/10 Desalting column (GE Healthcare). Purified monoclonal antibodies were concentrated using Amicon® Ultra centrifugal filter units with a molecular weight limit of 50 kDa (Merck Millipore) according to the manufacturer’s instructions. The purity and integrity of the monoclonal antibodies were assessed on SDS-PAGE. An overview of the monoclonal antibodies used throughout this study, including their theoretical mass is provided in Supplementary Table 4.
IgG capturing and subsequent IgG1 Fab generation
To capture IgG from purified ACPA and subsequently generate IgG1 Fab fragments, a similar procedure was used as described previously13,14. To this end, 20 µL CaptureSelect FcXL affinity matrix slurry (Thermo Fisher Scientific) was added to a Pierce Spin Column (Thermo Fisher Scientific) and washed three times with 150 mM phosphate buffer (PB). Purified ACPA was supplemented with 1% milk powder (Fantomalt, Nutricia) and 100 ng of an internal monoclonal antibody (mAb) standard (1:1 mixed trastuzumab: alemtuzumab). Purified and supplemented ACPA were incubated with the affinity matrix in fractions of up to 750 µL for each one hour head-over-head rotating at room temperature, and the unbound fraction was collected in between each incubation by centrifugation for one minute at 500 g. The affinity matrix with bound ACPA IgG was washed four times with PB after which ACPA IgG1 Fab fragments were generated by selective cleavage of captured IgG1 using the IgG1-specific protease immunoglobulin degrading enzyme (IgdE; branded FabALACTICA, Genovis), which cleaves human IgG1 just above the hinge region. For this purpose, the affinity matrix was resuspended in 50 µL PB containing 50 units of IgdE and incubated for at least 16 hours at 37 °C on a thermal shaker. The generated ACPA IgG1 Fab fragments were collected by centrifugation for one minute at 500 g.
To capture IgG from plasma, CaptureSelect FcXL affinity matrix was prepared in the same manner as described above. However, after washing, the affinity matrix was resuspended in 150 µL PB supplemented with 400 ng of internal mAb standard (1:1 mixed trastuzumab: alemtuzumab). A plasma volume containing an estimated 50 µg IgG1 was added and incubated with the affinity matrix for one hour shaking at room temperature. After four washes with PB, IgG1 Fab fragments were generated in the same manner as ACPA IgG1 Fab fragments.
LC-MS-based Fab profiling
To examine the intact Fab fragments that were released, a reversed-phase liquid chromatography-coupled mass spectrometry (LC-MS) method and data processing was used, which was described previously13,14. In short, the collected intact Fab fragments were separated using a Vanquish Flex UHPLC instrument (Thermo Fisher Scientific) equipped with a 1×150 mm MAbPac Reversed Phase HPLC column (Thermo Fisher Scientific) and directly coupled to an Orbitrap Fusion Lumos Tribrid or an Orbitrap Exploris 480 MS with BioPharma option (Thermo Fisher Scientific). During chromatographic separation, both the column pre-heater and the analytical column chamber were heated to 80 °C. The Fab fragments were separated over a 62 minute gradient at a flow rate of 150 µL/minute. Gradient elution was achieved using two mobile phases, A (0.1% HCOOH in MilliQ water) and B (0.1% HCOOH in CH3CN). At the start of the gradient, a mixture of 90% A and 10% B was used, ramping up from 10 to 25% B over one minute, from 25 to 40% B over 55 minutes, and from 40 to 95% B over the last one minute of the gradient. MS data were collected with the instrument operating in intact protein and low-pressure mode. Spray voltage was set at 3.5 kV, ion transfer tube temperature at 350 °C, vaporizer temperature at 100 °C, sheath gas flow at 15 arbitrary units, auxiliary gas flow at 5 arbitrary units and source induced dissociation (SID) at 15 V.
Spectra were recorded with a resolution setting of 7500 (at m/z 200) in MS1. Scans were acquired in the range of 500–4000 m/z using an automated gain control (AGC) target of 300% and a maximum injection time of 50 milliseconds. For each scan, 5 micro-scans were recorded. The raw spectra of the mass spectrometry data have been deposited to the MassIVE repository with identifier MSV000093196.
Fab profiling data analysis
To analyze the LC-MS results, the retention time and mass (in Dalton) of all intact Fab molecules were retrieved from the generated RAW files using BioPharmaFinder (BPF) 3.2 (Thermo Fisher Scientific), similar to the method described before13. In short, deconvolution was performed using the ReSpect algorithm (Thermo Fisher Scientific) between 5 and 57 minutes using 0.1 minute sliding windows with 25% offset and a merge tolerance of 30 parts per million (ppm). Noise rejection was set at 95% and the output range between 10,000 and 100,000 Da with a target mass of 48,000 Da and a mass tolerance of 30 ppm. Charge states between 10 and 60 were included and the Intact Protein Peak model was selected.
Further data analysis was performed using in-house Python 3.9.13 scripts using libraries: Pandas 1.4.444, Numpy 1.21.545, Scipy 1.9.146, Matplotlib 3.5.247 and Seaborn 0.11.2. Masses of the BPF identifications were recalculated using an intensity weighted mean, considering only the most intense peaks comprising 90% of the total intensity. Masses between 45.0 and 56.2 kDa with the most intense charge state above 1000 m/z and a BPF score ≥40 were considered to be Fab fragments of IgG1 antibodies. Fab fragments were considered to be identical within a defined mass and retention time window. The range of these windows was set to three times the standard deviation of the respective value observed for the internal mAb standard of the respective batch of samples that were processed with LC-MS. The windows were generally below 1.7 Da for the mass and below 1.4 minutes for the retention time. Concentrations of detected Fab molecules were determined by normalizing their sum intensity to the averaged sum intensity of the internal mAb standard and were corrected for the plasma volume applied to ACPA purification or total plasma IgG1 Fab profiling. The mAb standard was subtracted from the list of Fab fragments detected and the remaining Fab fragments were defined as unique Fab molecules based on their unique pair of mass and retention time. The degree of overlap between Fab profiles was determined by quantifying the relative abundance of overlapping Fab molecules and is shown as a percentage in the overlap heatmaps. Processed mass spectrometry data underlying donut plots, heatmaps, dot plots as well as raw data underlying depicted chromatography traces and the background-subtracted raw ELISA data are provided in the source data file.
Coupling of glycovariants
To couple glycovariants of the same antibody clone, we took advantage of the mass and retention time shifts detected upon IgG1 Fab profiling of a monoclonal ACPA IgG1 antibody (7E4WT; for more information, see Kissel et al.36), harboring two N-linked glycans in its variable domain (Supplementary Fig. 8). Accordingly, glycoforms differing by a galactose were determined based on a mass shift of 162.1 Da and a retention time tolerance of 0.1 minute, those differing by an N-acetylglucosamine were determined based on a mass shift of 203.2 Da and a retention time tolerance of 0.1 minute, and those differing by a sialic acid were determined based on a mass shift of 291.3 Da and a retention time shift of less than 1 minute.
Quantification of Fab molecules with zero, one, or two or more glycans
The proportion of Fab molecules harboring zero, one, or two or more glycans was defined using two key parameters: the median mass of Fab molecules in the ImMunoGeneTics information system (IMGT) database48 (47.8 kDa) and the estimated average mass of a single ACPA IgG Fab glycan ( ~2.4 kDa). The latter was calculated based on the highly prevalent ACPA IgG Fab glycans G2FBS1, G2FS2 and G2FBS2, which are routinely used to determine ACPA IgG Fab glycosylation through glycan release22,23,24,25. Consequently, a Fab molecule with one glycan was estimated to have a mass of ~50.2 kDa, while the mass of a Fab molecule with two glycans would be ~52.6 kDa, and the mass of a Fab molecule with three glycans would be ~54.9 kDa. We used these anticipated masses as the midpoint of each mass range and calculated the boundaries as follows: for Fab molecules with one glycan, the boundary was calculated as 50.2 ± 1.2 kDa (half the glycan); for Fab molecules with two or more glycans, the boundary was calculated as 52.6 − 1.2 kDa and 54.9 + 1.2 kDa. As a result, we established the following mass ranges: 46.0–49.0 kDa (Fab molecules without glycan), 49.0–51.4 kDa (Fab molecules with one glycan), and 51.4–56.1 kDa (Fab molecules with two or more glycans). The proportion of ACPA and plasma IgG1 repertoires harboring zero, one, or two or more Fab glycans was calculated as the relative abundance of Fab molecules within the respective mass range.
Statistics
Statistical analysis of the extent and level of Fab glycosylation was performed using a two-tailed Wilcoxon matched-pairs rank test and corrected for multiple testing using the Bonferroni-Dunn method. Significant differences are defined as follows: not significant (ns; P ≥ 0.05), *P < 0.05, **P < 0.01.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
