Divergent and dominant anti-inflammatory effects of patient-derived anticitrullinated protein antibodies (ACPA) in arthritis development

About two thirds of rheumatoid arthritis (RA) patients develop anti-citrullinated protein antibodies (ACPA), which are a characteristic of the disease and sparsely observed in other clinical settings or in the healthy population. ACPA are used as a diagnostic criterion and often develop prior to diagnosis. Therefore, ACPA titers are important in the identification of individuals at risk of developing RA. Interestingly, the titers and target cross-reactivity of ACPA increase by time of diagnosis, suggesting a causality between anti-citrulline reactivity and the development of RA. However, only 50% of ACPA-positive at-risk individuals will progress to a clinical RA diagnosis. This observation suggests that there are different types of ACPA and that their relationship with RA development is more complex than presently assumed. To address the biological effect of ACPA in the establishment and development of inflammatory arthritis, we made use of the collagen antibody-induced arthritis (CAIA) model of passive arthritis. Using seven unique patient derived monoclonal ACPA, we observed that ACPA are predominantly anti-inflammatory, with some clones (C03 and BVCA1) completely inhibiting disease development. We have also identified a clone (C04) as having a disease-prone effect, by increasing and sustaining disease prevalence. This clone had previously been reported as being pro-inflammatory in a different model of joint inflammation. The anti-inflammatory effects of C03 were dependent on FcγR, since neither F(ab’)2 or a mutated Fc-null GRLR-C03 clone could mediate disease protection. Most importantly, mice receiving polyclonal ACPA enriched from RA sera were also partially protected from disease. Looking at the cell and tissue of origin of the ACPA clones, their V(D)J sequences, Fc glycosylation pattern and fine-specificities, we could not identify a common feature between anti-inflammatory ACPA distinguishing them from those without the properties. Together, our data demonstrates that circulating ACPA in RA patients are predominantly anti-inflammatory, emphasizing the need to study ACPA repertoires in order to determine their influence on the clinical outcome of the patient or at-risk individual.

patient using either flow cytometry sorting of CD138+ cells or 10X Genomic single cells transcriptomics.
Engineered human FcR null IgG1 construct with G236R/L328R mutations (GRLR) which has previously shown not to interact with any FcR 6 , was generated by gene synthesis (IDT) and megaprimer PCR of the 1 vector 7 . Unaffected citrulline reactivity was confirmed by CCP2 and cit-peptide ELISA.

Generation of F(ab')2 fragments
Preparation of F(ab')2 fragments from the C03 monoclonal antibody was performed with a Pierce F(ab')2 preparation kit according to the manufacturer instructions (ThermoFisher Scientific). Briefly, a solution of C03 monoclonal antibody was initially desalted using a Zeba Spin Desalting Column. Digestion was achieved after 6h incubation of the monoclonal BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) antibody with a freshly immobilized pepsin column. Undigested IgG and F(ab')2 were separated via a Nab protein A Plus Spin column, with confirmation of the purified F(ab')2 fragment by SDS-PAGE.

Patient-derived CCP-and flow-through-pools
The collection of patient-derived CCP-pool and their respective flow-through pools were prepared as previously described 8 . Briefly, plasma was obtained from peripheral blood samples of RA patients (n=37; ≈10 ml/patient) and centrifuged at 3000 g for 5 min. IgGs were purified on HiTrap Protein G HP columns (Cytiva), according to the manufacturer's instructions. After low pH elution and neutralization, IgG was dialyzed against PBS and applied to a CCP2 affinity
Antibody binding to full-length citrullined protein was evaluate by on-plate modification and ELISA. Briefly, antigen were diluted to 3 g/ml in 8M Urea and coated to high-bind half area plates (Corning). Plates were blocked with 5% low-fat milk in PBS and antigens citrullinated with 280mU/ml recombinant human PAD4 (Cayman chemicals) in 50 mM Tris 10 mM CaCl2 1 mM DTT for 3h at 37C. Binding was assayed at 5 g/ml hIgG1 to citrullinated vimentin, alpha enolase (in-house expressed human) and collagen II (bovine purified, chondrex) and at 0.5 g/ml hIgG1 to citrullinated fibrinogen (human purifed, Sigma Aldrich). The ACPA fine-specificity array is further described in 9,15 .

CCP-ELISA
The development of ACPA during the development of CAIA was tested through a CCP2-ELISA (Immunoscan CCPlus, Svar Life Science) with some modifications. Briefly, different serum dilutions from naïve mice or mice submitted to CAIA (day 15) were applied to a CCP pre-coated plate. Detection of murine anti-CCP-reactive antibodies was done with a peroxidase-conjugated F(ab')2 fragment goat anti-mouse IgG (Jackson ImmunoResearch) and using TMB as the enzymatic substrate. The arthritogenic cocktail of anti-CII antibodies used to induce CAIA was similarly tested in the same assay, in order to investigate any potential cross-reactivity of these antibodies with citrullinated antigens.

C1q-ELISA
The capacity of each ACPA clone in activating the classical pathway of complement was assessed by a C1q ELISA. A 96 well plate (Nunc, Maxisorp) was coated with each ACPA clone individually, starting at a concentration of 10μg/ml and followed by a stepwise two-fold dilution. Blocking was performed with a 2% ovalbumin solution. Purified mouse C1q protein (Complement Technologies Inc, USA) was then added to the wells at 1μg/ml followed by detected with a biotin-conjugated anti-mouse C1q antibody (ThermoFisher Scientific, clone JL-1). Streptavidin-HRP and TMB were used as enzyme and substrate of the reaction, respectively. The plate was washed 3-5 times between each step with PBS containing 0.05% Tween-20. All animal experimental settings were repeated at least twice, with a total number of 10 to 50 animals assessed per condition, depending on the ACPA clone used. A minimal of 5 animals per group was used at any given time given the prevalence of arthritis being generally 80-100%, thus allowing statistical calculations with a good degree of power, and assuming a nonnormal distribution of the data. Each cage contained no more than two animals of the same experimental group, with all experimental groups assigned to as many cages in order to minimize potential cage effects. Statistical analysis of arthritis development and prevalence between groups was conducted by non-parametric repeated-measures Friedman test with Dunn's multiple comparison test. When comparing two or more groups in single observation assays, a non-parametric Mann-Whitney or Kruskal-Wallis test was used. Prism GraphPad version 9.3.1 was used to calculate statistics and statistical significance was considered when p < 0·05 for a 95% confidence interval.